Volume 21 Issue 3

International Food and Agribusiness Management Review

Official Journal of the International Food and Agribusiness Management Association

Volume 21 Issue 3 2018

EVERY GENERATION NEEDS ITS LEADERS. As a farmer-owned cooperative, we help lead the agricultural industry by promoting the most innovative, responsible farming practices available today. And we want to make sure that the next generation of leaders is well trained for tomorrow. That’s why we proudly support youth leadership programs like FFA. It’s also why we’ve invested in scholarships and countless land-grant university initiatives across the country. At CHS, we know every generation needs its leaders. And every potential leader needs a little help becoming one.

Visit chsinc.com/stewardship to learn more about our ongoing commitment to leadership.

©2016 CHS Inc.

International Food and Agribusiness Management Review

Editorial Staff Executive Editor

Gerhard Schiefer, University of Bonn, Germany

Regional Managing Editors Asia, Australia, and New Zealand

Derek Baker, UNE, Australia Kim Bryceson, University of Queensland, Australia Kevin Chen, IFPRI-Bejing, China Jeff Jia, University of Exeter, United Kingdom Nicola M. Shadbolt, Massey University, New Zealand

Europe

Pegah Amani, Technical Institute of Sweden, Sweden Vera Bitsch, Technical University of Munich, Germany Laura Carraresi, University of Bonn, Germany Alessio Cavicchi, University of Macerata, Italy Hans De Steur, Ghent University, Belgium Loic Sauvee, UniLaSalle, Beauvais, France Cristina Santini, University San Raffaele, Italy Jacques Trienekens, Wageningen University, The Netherlands

North America

Ram Acharya, New Mexico State University, USA Yuliya Bolotova, Clemson University, USA Michael Gunderson, Purdue University, USA William Nganje, North Dakota State, USA R. Brent Ross, Michigan State University, USA Aleksan Shanoyan, Kansas State University, USA David Van Fleet, Arizona State University, USA Nicole Olynk Widmar, Purdue University, USA Cheryl Wachenheim, North Dakota State University, USA

South America

Aziz da Silva Júnior, Federal University of Vicosa, Brazil Jose Vincente Caixeta Filho, University of Sao Paulo, Brazil Emilio Morales, University of New England, Australia

Africa

Nick Vink, Stellenbosch University, South Africa

Editorial Board Filippo Arfini, Universita’ di Parma, Italy Stefano Boccaletti, Universita’ Cattolica, Italy Michael Boehlje, Purdue University, USA Dennis Conley, University of Nebraska - Lincoln, USA Francis Declerck, ESSEC Business School, France Jay Lillywhite, New Mexico State University, USA Woody Maijers, INHOLLAND University, The Netherlands

Marcos Fava Neves, FEA / USP / PENSA, Brazil Onno Omta, Wageningen University, The Netherlands Hernán Palau, Buenos Aires University, Argentina Christopher Peterson, Michigan State University, USA Thomas Reardon, Michigan State University, USA Mary Shelman, (Retired) Harvard Business School, USA Johan van Rooyen, University of Stellenbosch, South Africa

The IFAMR (ISSN #: 1559-2448) is published quarterly and the archived library is available at http://www.ifama.org. For copyright and publishing information, please contact: Marijn van der Gaag, Administrative Editor Wageningen Academic Publishers • P.O. Box 220 6700 AE Wageningen • The Netherlands • Tel: +31 317 476511 Fax: +31 317 453417 • Email: ifamr@wageningenacademic.com • Web: http://www.wageningenacademic.com/loi/ifamr.

International Food and Agribusiness Management Review Volume 21 Issue 3, 2018

TABLE OF CONTENTS 1.

Value adding in the agri-food value chain

293

2.

Cyber security on the farm: an assessment of cyber security practices in the United States agriculture industry

317

The effects of international price volatility on farmer prices and marketing margins in cattle markets

335

Technical efficiency and marketing channels among small-scale farmers: evidence for raspberry production in Chile

351

Defining U.S. consumers’ (mis)perceptions of pollinator friendly labels: an exploratory study

365

The role of members’ commitment on agri-food co-operatives’ capitalization, innovation and performance

379

Dual moral hazard and adverse selection in South African agribusiness: it takes two to tango

391

Fertilizer distribution flows and logistic costs in Brazil: changes and benefits arising from investments in port terminals

407

Maria Emilia Cucagna and Peter D. Goldsmith

Andrew Geil, Glen Sagers, Aslihan D. Spaulding, and James R. Wolf

3.

L. Emilio Morales

4.

Roberto Jara-Rojas, Boris E. Bravo-Ureta, Daniel Solís, and Daniela Martínez Arriagada

5.

Hayk Khachatryan and Alicia Rihn

6.

Gustavo Marcos-Matas, Arianna Ruggeri and Rino Ghelfi

7.

Thulasizwe Mkhabela

8.

Débora da Costa Simões, José Vicente Caixeta-Filho, and Udatta S. Palekar

International Food and Agribusiness Management Review

i

Table of contents

9.

Volume 21, Issue 3, 2018

Views on sustainability and the willingness to adopt an environmental management system in the Mexican vegetable sector Luz E. Padilla-Bernal, Alfredo Lara-Herrera, Alberto Vélez Rodríguez, and María L. Loureiro

10. Florida’s Natural® and the supply of Florida oranges Carlos Omar Trejo-Pech, Thomas H. Spreen, and Lisa A. House

International Food and Agribusiness Management Review

ii

423

437

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2017.0051 Received: 7 June 2017 / Accepted: 21 November 2017

Value adding in the agri-food value chain RESEARCH ARTICLE Maria Emilia Cucagnaa and Peter D. Goldsmith

b

aFormer

graduate student and bProfessor, Department of Agricultural and Consumer Economics, University of Illinois, 1301 West Gregory Drive, Urbana, IL 61801, USA

Abstract Global agricultural markets reflect the increasing complexity of modern consumer demand for food safety and quality. This demand has triggered changes throughout the food industry, and led to greater opportunities for product differentiation and the potential to add value to raw commodities. Greater differentiation and value adding over time has in turn dramatically changed the price spread or marketing bill between the farm value of products and the retail value. Thus a significantly greater percentage of the final price paid by consumers is now garnered down chain rather than up chain over the last 20 years. This apparent shifting of value creation or addition, as measured by the marketing margin, has invigorated empirical questions as to where, and how much value, is created along the agri-food value chain. First we define value creation/ adding, and then we estimate the economic value added for 454 firms. We validate our findings by creating and employing three additional value creation measures – the modified economic value added, the creation or destruction of value, and the persistence of value creation. Finally we estimate value creation at each node of the value chain, measure the relative differences among firms and nodes, and estimate a model measuring the drivers of value adding.

Keywords: value creation, value adding, agribusiness, food, agricultural, value chain JEL code: Q13, O31, Q16 Corresponding author: pgoldsmi@illinois.edu

Š 2018 Cucagna and Goldsmith

293

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

1. Introduction Global agricultural markets have become increasingly complex due to changes in consumer demand, the development of complex food standards mainly related to food safety and food quality, advances of technology, and changes in the industry structure along the value chain (Goldsmith et al., 2002; Humphrey and Memedovic, 2006). As a result over time, a significantly greater percentage of the final price paid by consumers is garnered down chain as opposed to up chain. The input and production stages of the food and agribusiness value chain now supply about 16% of the final food value, with the other 84% coming from post-farm gate stages (Figure 1). This change in the farm and marketing bill share suggests a shift over time in the loci of food and agribusiness value addition.

The Economic Research Service (ERS)’s Food Dollar Series (formerly the Marketing Bill), uses prices to measure the share of the consumer’s dollar received by each stage of the agrifood value chain and has been of great interest to researchers and policymakers since the 1940s (Canning, 2011). Corresponding with the Food Dollar analysis, policymakers at times feel farmers do create levels of value in excess of what they capture, and as a result suspect anti-competitive forces are at play (Canning et al, 2016; EU Council, 2016; U.S. Congress, 2002). But while value creation/value addition and capture along the value chain is a relevant concept in agribusiness, it is perhaps surprising that managers and academics ‘often do not know how to (explicitly) define value or measure it’ (Anderson and Narus, 1998; Lindgreen et al., 2012). Consistent with a lack of a definition, research to date has not measured the level nor assessed the loci of value addition along the agrifood value chain. The objectives of this manuscript are threefold; to provide a financial, and thus formal, definition of value added; measure value addition by firms across the four nodes/stages of the value chain; and third to test hypotheses as to the drivers of value creation that differ across the four nodes.

Specifically we are the first to measure the economic value added (EVA) for firms and stages along the agri-food value chain. We validate our findings by creating and employing three additional value creation measures; the modified economic value added (MEVA), the creation or destruction of economic value added (CEVA), and the persistence of economic value added (PEVA). We then use the EVA to test specific hypotheses as to how and where value creation occurs along the agri-food value chain. Finally, we discuss

90

% of US real food dollar

80

85.7%

80.7%

70 60 50 40 30

19.3%

14.3%

20

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

10

Year Farm share %

Marketing share %

Lineair (farm share %)

Lineair (marketing share %)

Figure 1. Food bill and farm share of the U.S. Real Food Dollar: 1993 to 2011 (adapted from USDA, 2017).

International Food and Agribusiness Management Review

294

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

how our results inform policymakers as to the causes of value creation and capture. In short, we provide empirical evidence that producers capture relatively less value because they create relatively less value, and they create relatively less value because the differentially employ capital. We posit that stages need not equally create value nor employ equal levels of value creating capital. Therefore the differential use of capital may not reflect anti-competitive forces but reflect a form of sustainable equilibrium when the unit of analysis is not the firm or stage, but the entire value chain.

2. Literature review Historically, the agribusiness sector has been a commodity-oriented industry, with a strong focus on maximum efficiency, homogenous products, and economies of scale (Grunert et al, 2005). In the past, firms may not necessarily have needed to focus on creating value since companies could be profitable as a consequence of controlled distribution channels, regulated markets, the acquisition of badly performing firms, or a scarcity of resources to produce (Doyle, 2000; Lindgreen et al., 2012). Although agri-food markets had been efficient at producing homogeneous products, the evolution of consumer demand to safe, high quality, and convenient products, as well as technological advances and increased competition, require the production of differentiated products. Therefore, a company focus on value adding activities allows firms to meet the expectations of consumers by offering high quality and differentiated products that reflect emergent agribusiness market opportunities (Alexander and Goodhue, 1999; Coltrain et al., 2000; Streeter et al., 1991). In recent years, value creation or value adding in agriculture, and its management, has emerged as a business survival strategy (Kampen, 2011).

The food and agribusiness literature defines value creation or value adding in agribusiness when a firm changes a product’s current place, time, and form to characteristics more preferred in the marketplace (Anderson and Hanselka, 2009; Coltrain et al., 2000). The concept of value creation historically has been synonymous with value adding, and the concept of value adding is helpful when analyzing the potential profitability of agriculture (Coltrain et al., 2000). Coltrain et al. (2000) provide a more specific definition by giving an example of value adding in agriculture: ‘to economically add value to an agricultural product (such as wheat) by processing it into a product (such as flour) desired by customers (such as bread bakers).’ The 2002 United States Farm Bill defines value adding in agriculture as a change in the physical state of any agricultural commodity through a production method or handling process by which the agricultural commodity or product is produced or segregated (USDA, 2013). The end result of changing value is that it expands the customer base for the product, or makes available a greater portion of revenue derived from the marketing, processing, or physical segregation of the agricultural commodity or product achieved by the producer (U.S. Congress, 2002). Since 2009, the USDA has awarded 863 value-added producer grants totaling $108 million to help firms transform their commodities or products into higher valued goods (Farm Futures, 2014). In sum, the Coltrain, USDA and U.S. Congressional definitions present value adding as encumbering, or internalizing, additional steps found in the value chain; thus their definition is an activity measure.

Amanor-Boadu (2003) integrates the activity definition with a profitability measure. He attempts to clarify the concept of value creation/adding by defining two conditions that an activity must satisfy. The first involves a firm performing an activity that traditionally has been performed at another stage farther down the value chain; or (2) if one is rewarded for performing an activity that has never been performed in the value chain (Amanor-Boadu, 2003; Evans 2006). Both are consistent with the above activity definition. He adds though that if the total profitability of the performing organization is not increased by the value-added activity, then the activity cannot be deemed to have contributed any value to the value chain or to customers, and thus it fails to qualify as a value-added activity.’ (Amanor-Boadu, 2003). Thus value adding must not only disintermediate an activity (i.e. Coltrain), but must be profitable.

The finance literature adds greater specificity to the concept of value adding by focusing on capital utilization and capital use efficiency. The popularity of a capital-based definition of value creation has increased since the 1990s, and its processes have been widely studied in the mainstream business literature (Anderson, 1995; International Food and Agribusiness Management Review

295

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Parasuraman, 1997; Walter et al., 2001). From an economic perspective, a company creates value by producing net benefits in excess of the cost of capital (Chmelikova, 2008). Traditional accounting information such as return on equity, earnings per share, net operating profit after taxes (NOPAT) and return on investment are common measures or proxies for value creation (Ibendahl and Fleming, 2003). Although widely used, these measures fail both to capture a firm’s value creation that results from management actions, and to account for the full cost of capital (Sharma and Kumar, 2010). Similarly, stock price may be an incomplete measure because it does not reflect the true value of the firm (Keef and Roush, 2003; Sparling and Turvey, 2003; Turvey et al., 2000). That is, accounting income or profitability may not be a good estimator of true value creation, and value creation efficiency (Sharma and Kumar, 2010). This becomes important later on in our analysis as we show value chain nodes can exhibit comparable rates of profitability but differing levels of value creation.

A more accurate measure of value creation, EVA, represents how well a company produces operating benefits, given the amount of capital invested (Chmelikova, 2008). An increasing number of companies have begun to employ EVA to measure the value created by each business unit, since this measure provides better information than traditional measures (Geyser and Liebenberg, 2003; Ibendahl and Fleming, 2003). EVA shifts the focus simply from profitability, which may be ethereal, to earning a return on capital above the opportunity cost of the capital (Walbert, 1994). Furthermore, the EVA metric combines operating efficiency and asset management from a managerial accounting perspective (Anderson et al., 2005).

The EVA of a company is defined as ‘a measure of the incremental return that the investment earns over the market rate of return’ (Sharma and Kumar, 2010). Formally, EVA is expressed as:

(

)

AdjNOPAT EVA = ___________ – Cost of Capital × NOA NOA

(1)

where, AdjNOPAT is the adjusted NOPAT that does not include non-operating revenues and expenses, such as training costs or restructuring expenditures. The net operating assets (NOA) represents the total capital employed by a firm via its main business activities. NOA involves an adjustment similar to that of NOPAT and removes accounting items that are not used to generate an operating profit of the core business, such as non-operating fixed assets.

The cost of capital is the weighted average of debt and equity cost (Equation 2). The EVA takes into account the cost of capital that managers must pay for the capital they employ. Using the cost of capital allows managers to determine the true value generated by the capital employed (Anderson et al., 2005; Sharma and Kumar, 2010).

( ) ( )

D E ______ rk = rd (1 – t) + re _____ (2) D + E D + E where, rk is cost of capital, rd is cost of debt, re is cost of equity, D is the debt, E is total capital and t is the corporate tax rate.

2.1 The value chain The value chain is a ‘series of value adding processes which flow across many companies and creates products and services which are suitable to fulfill the needs of customers (Martin and Jagadish, 2006; Noemi, 2012).

Each step in the chain, from basic inputs to consumer goods, serves as a link or stage in the value chain (Opara, 2003). It is a system of firms that interact to positively impact one another’s performance (Bigliardi and Bottani, 2010). The value chain framework emerges as a key aspect in the analysis of the drivers of International Food and Agribusiness Management Review

296

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

business success and value creation (Matopoulos et al., 2007). Within this context, firms within each stage seek to offer maximum value to their customers and to improve company performance, while obtaining a competitive advantage from being part of the value chain (Bourlakis et al., 2012; Ketchen et al., 2008).

Consequently, value creation does not occur in isolation but rather within a value chain framework where companies can operate most efficiently as members of the chain. In this position, a company may more easily establish collaborative relationships based on trust, commitment, and cooperation among the chain’s members, which helps to improve all the companies’ performance and thus create value (Lindgreen et al., 2012; Vanyi, 2012). Additionally, the chain framework connotes a level of complementarity along the chain.

Firms and stages need not all similarly employ capital or add value at the same levels. For example, some stages need not employ high levels of R&D, and thus need not create a lot of value. Yet they may still be sustainable and profitable, and comprise essential links in the food and agribusiness value chain.

For the purpose of this study, following Humphrey and Memedovic (2006), the agri-food value chain is split into four main stages or nodes: inputs, production, processing, and delivery to consumers (Figure 2). Stage 1, inputs, involves biotechnological, agro-chemical and fertilizer, animal health, animal breeding and farm equipment companies. Firms in this sector supply agricultural products and services to farmers as its primary customers. Stage 2, production, includes all of the activities involved in the production of raw food materials, such as crop and livestock commodities. Production serves mainly food processors and manufacturers, the next link down the chain. Stage 3, processing, comprises food processing and manufacturing, including beverage, breweries, wineries and packaged food companies. These firms convert raw materials into either branded or unbranded food products. These products are then marketed to the retail stage for distribution and sale to consumers, the terminal stage in the food and agribusiness value chain. Stage 4, retail, distributes, sells and markets food products to consumers. In the last stage of the agri-food value chain, we include firms involved with food distribution, grocery retail, and food service.

In general, we are curious as to value creation differences among the four chain nodes as well as to the underlying drivers of value creation. Specifically in this research we test hypotheses concerning both firm level value creation, and value adding across the various stages of the food and agribusiness value chain.

We test the following hypothesis.

Ho I: value addition levels differ across stages as each stage contributes differently to the value creation process This hypothesis implies that dummy variable coefficients pertaining to a value chain node be significantly different from zero when the reference group is Stage 2, the production node.

Stage 1. Empirical evidence suggests that the increase in private investment among Stage 1 firms in agricultural research has been driven by the establishment and strengthening of intellectual property (Fulton and Giannakas,

Stage 1

Stage 2

Stage 3

Stage 4

Inputs and equipment

Production

Processing and manufacturing

Distribution, retail, food service

Up chain

Down chain

Figure 2. The four stages of the food and agribusiness value chain.

International Food and Agribusiness Management Review

297

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

2002; Goldsmith, 2001; Moschini and Lapan, 1997). Effective intellectual property development in seed and equipment markets, for example, allows suppliers to establish internationally recognized brands reflecting highly differentiated products. Additionally, horizontal mergers through acquisition, and coordination through technology licensing, allow seed firms to achieve significant research and development, marketing, and production economies. Alternatively, though, Demont et al. (2007) conclude that agricultural producers capture most of the value created by upstream input suppliers, when analyzing the benefits literature, thus value addition upstream in Stage 1 may be relatively low. We test the following hypothesis.

Ho II: we expect the agricultural input Stage (1) to be a high value-adding node given the presence of input brands and the high level of agricultural research. The evidence of horizontal coordination among the companies of this stage also provides support to this hypothesis This implies that the coefficient for the Stage 1 dummy variable be significantly different from zero, and positive, when the reference group is Stage 2.

Stage 2 is the most commoditized sector of the value chain (Carlson, 2004; Phillips et al., 2007). The stage, which includes farm production, presents low product margins, high price dependence in transactions, and a low level of product differentiation. Companies often operate within competitive markets and strive to compete on cost management and economies of scale (Goldsmith and Bender, 2004). In recent years, an increasing number of structural changes have taken place within this stage, showing a decline in the number the firms but an increase in the size of those firms (Boehlje et al., 2011). Large firms that have been able to obtain the benefits of scale economies may be responsible for making this stage profitable (Humphrey and Memedovic, 2006; Lobao and Meyer, 2001). Alternatively, some firms within the production stage do perform value added activities that permit them to capture a greater proportion of the consumer dollar (Born and Bachmann, 2006; Kampen, 2011). Production firms also increasingly engage in greater levels of vertical coordination, such as contractual arrangements, or vertical integration, in order to both improve efficiency and to create more value (Hendrikse and Bijman, 2002; Sporleder, 2006). For example, shifting from commodity grain production to differentiated grain and oilseed allows producers to create and capture added value (Goldsmith and Silva, 2006; Hayenga and Kalaitzandonakes, 1999). We test the following hypothesis.

Ho III: the lack of coordination and the commercialization of undifferentiated products as well as the condition of being primarily a price taker makes the production Stage (2) exhibit the lowest levels of value addition among the four value chain nodes Thus, the coefficient estimates for all three value chain dummy variables corresponding to the node should be positive and significantly different from zero.

Stage 3. The process of turning raw agricultural outputs into food and beverage products ‘adds value’ to raw commodities in an economic sense, but these activities may also significantly alter the appearance, storage life, nutritional value, and content of the raw materials (Gopinath et al., 1996). A processor’s main activity is to transform commodities into food products, a process that adds economic value to the products.

Increasingly, consumers demand higher quality and premium food products with accompanying features as their incomes rise (Sexton, 2013). Modern food systems also involve long and complex supply chains thus processing and manufacturing firms must innovate to meet the needs of buyers in terms of ingredient traceability, new product development, extending shelf life, and efficiently promoting their products.

Companies at this stage may focus on producing higher-quality products, as well as manufacturing more branded and differentiated products, to achieve a competitive advantage and thus create value (Omidvar et al., 2006). Product differentiation including advertising to establish and maintain branded manufacturing goods, involves significant investment in R&D and intangible assets (Oustapassidis and Vlachvei, 1999; USDA, 2013). We test the following hypothesis.

International Food and Agribusiness Management Review

298

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Ho IV: food manufacturing (Stage 3) will exhibit a relatively high level of value addition, compared to upstream stages, due to strong product differentiation, coordination with retailers, and access to low cost inputs from the production stage We present no hypothesis ranking Food Manufacturing among the four nodes. However, the coefficient estimate for the Stage 3 dummy variable should be positive and significantly different from Stage 2, our reference stage.

Stage 4. The last stage of the agri-food value chain serves the consumer. One of the main drivers of innovation in this stage for food retailers, restaurants, and hospitality firms occurs by differentiating through service and retail brands to better meet consumer demand (Burch and Lawrence, 2005; Humphrey and Memedovic, 2006). Upstream integration though into brand food manufacturing and private label production changes the relationship between stages 3 and 4 as retail firms strive to support the innovations that better serve the needs of consumers (Burch and Lawrence, 2005). There is evidence of increasing market power at the retail end of the agri-food value chain as a consequence of increasing concentration and consolidation in the sector (Humphrey and Memedovic, 2006; Viaene and Gellynck, 1995). The influence of retailers over processors, manufacturers, and also the consumer allows the retailer sector to obtain a competitive advantage and capture more value created along the chain (Burch and Lawrence, 2005). The growing market power of downstream firms may limit up-chain firms from moving to high-value-added activities such as distribution, marketing and retailing (Farfan, 2005; Liu and Niemi, 2014). Greater market power down chain also allows for greater control over information flows and thus has a competitive advantage for innovation (Farfan, 2005; Humphrey and Memedovic, 2006). We test the following hypothesis.

Ho V: the retail firms (Stage 4) with their consumer orientation, need to innovate, and strategic position within the value chain, will be a relatively high value creator, compared to upstream stages Thus, the coefficient estimate for the dummy variable for Stage 4 should be positive and significantly different from zero.

Several researchers have identified different accounting categories and business activities that determine firm-level value adding (Kalafut and Low, 2001; Kale et al., 2001; Zéghal and Maaloul, 2010). Investment in innovation related activities drives value creation (Amanor-Boadu, 2003; Coltrain et al., 2000). ‘Innovation is the embodiment, combination, or synthesis of knowledge in original, relevant, valued new products, processes, or services,’ (Sawang and Unsworth, 2011). Innovation associated with value creation activities refers to the performance of activities that enhance ‘existing processes, procedures, products, and services or creating new ones’ using organizational structures (Amanor-Boadu, 2003). R&D expenditures (Artz et al., 2010; Cohen and Klepper, 1996; Damanpour, 2010; Meisel and Lin, 1983; Zona et al., 2013) and goodwill and intangible assets (Kramer et al., 2011) are key factors in determining firm level of innovation. Therefore, we propose the following hypotheses with respect to innovation.

Ho VI: firms with higher levels of intangible assets and goodwill create more value The coefficient estimate for the goodwill and intangible asset variable should be positive and significantly different from zero Ho VII: firms with higher levels of research and development expenditures create more value Similarly, the coefficient estimate for the R&D expenditure variable should be positive and significantly different from zero.

Product differentiation is also a key component of value creation and firm success. The ratio of the cost of goods sold (COGS) to sales serves as one measure of product differentiation (Balsam et al., 2011; Goldsmith International Food and Agribusiness Management Review

299

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

and Sporleder, 1998; Nair and Filer, 2003). The use of COGS as a measure of product differentiation results from the logic associated with raw input transformation. The more a purchased input remains the same, or passes through to the buyer’s customer, the less differentiated is the product and the less value has been added by the firm. Thus if the inbound input cost per unit, ceteris paribus, comprises a high percentage of the outbound product’s price, then the product has a relatively high COGS, and little additional value has been created and added. The expression of ‘turning a sow’s ear into a silk purse,’ captures the COGS-product differentiation logic. The buying firm adds significant value to the low cost input, a ‘sow’s ear’, and achieves a high valued product, such as a ‘silk purse.’ We test the following hypothesis.

Ho VIII: firms with higher degrees of product differentiation, a smaller ratio of COGS to sales, create more value Thus, the coefficient for the calculated variable COGS will be negative and significantly different from zero.

Previous studies suggest that firm size is a relevant characteristic that influences the intensity of a firm’s new product development (Damanpour, 2010; Hecker and Ganter, 2013; Zona et al., 2013). Large firms may be more creative because of the availability of financial resources, technological possibilities, access to highly skilled labor, as well as knowledge capability and economies of scope. Furthermore, large firms have a greater capability to afford the cost of new ventures as well as to manage the risk of unsuccessful efforts (Camisón-Zornoza et al., 2004; Damanpour, 2010; Hecker and Ganter, 2013). We test the following hypothesis.

Ho IX: there is a positive relationship between firm size and value creation The coefficient for the firm size variable should therefore be positive and significantly different from zero.

Finally, we test the relationship between the level of leverage and the ability to create value. Capital structure may not be significant in the process of value creation as debt level’s impact on value creation is not clear. On the one hand increasing leverage reduces the firm’s tax liability, and allows, ceteris paribus; the opportunity to more productively invest surplus cash flow, an urgency to perform well, and the sale of underperforming or unrelated businesses or assets (Houle, 2008). However, financial leverage is a risk factor and may effect risk taking (Ely, 1995; Hua and Templeton, 2010). Also standard finance theory posits that debt and equity may be direct substitutes (see Barclay and Smith, 1999). Therefore, we offer no hypothesis as to the relationship between the level of leverage and the level of value creation.

3. Method We estimate EVA (Equation 1) and populate the capital-asset pricing model (CAPM) model using the 10year Treasury bond return, which reflects the risk-free rate of return. The market risk premium is calculated from the average of the last 25 years as the difference between the return of the Standard & Poor’s 500 and the risk-free return. We assign the country risk premium using Moodys country rating (Moodys, 2014). The beta of each firm is estimated using the unleveraged beta by industry reflecting the average market debt/ equity ratio by industrial sector.

Cost of debt is the average cost of the debt for each firm (Equation 3):

Interest expenditures rd = ___________________ (3)

Total liabilities The variable tax rate (t) serves as the proxy for the official corporate tax rate of each country. Weights are calculated based on the book value mix of debt and equity. To estimate the cost of equity, the CAPM is used (Equation 4), and comprises three elements: risk-free bonds (Rf), the stock’s equity beta (β=1.0 is average

International Food and Agribusiness Management Review

300

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

risk), and the market risk premium (Rm-Rf) needed to attract investors to hold onto the market portfolio of risky assets (Bruner et al., 1998; Sharma and Kumar, 2010). re = Rf + β(Rm – Rf )

(4)

The EVA metric is a dollar scaled metric and thus is highly correlated with firm size. Therefore, the EVA metric is sensitive to size bias (Anderson et al., 2005; Ibendahl and Fleming 2003). That is, larger firms, ceteris paribus, create more nominal value. Thus, we also create and employ a second value creation metric, the MEVA, which is scale neutral, and is expressed as a net percentage rather than a nominal dollar value.

We define MEVA as follows:

(

)

AdjNOPAT MEVA = ____________ – Cost of Capital × 100 NOA

(5)

MEVA captures the real and relative value addition by food and agribusiness firms, as opposed to the nominal value reflected in the EVA. Firms can either create capital, that is, produce more capital than they use; or destroy capital, produce less capital than they use. A positive EVA or MEVA indicates that the firm or value chain stage adds value because the operating assets or working capital employed generate a return sufficient to cover the capital costs of those assets. On the contrary, a negative EVA or MEVA indicates that a company destroys value because the returns from its net operating assets fall short of the returns required to capitalize the company. The EVA/MEVA concepts reflect how efficiently or inefficiently firms use capital.

Both measures employ a theoretically robust conceptual framework that produces useful metrics for analyses across firms, industries, and time.

3.1 Data We employ an unbalanced panel of a ten-year time-series (2003-2012) of 454 agri-food companies (Table 1) from 25 countries (64% are U.S. companies, 3.26% United Kingdom, 3.89% Japan, 3.36% China, and 2.86% from Mexico and others). The data originate from income statement and balance sheets for each of the firms collected by Morningstar Inc. Placing firms into discreet value chain stage categories is obviously fraught with a lack of precision and potential bias because firms may operate in more than one stage. Yet our work requires some form of data structure that permits the data be statistical analyzed and formally tested via hypotheses. As a result, we use Morningstar’s classification of industry groups; which relies upon firms’ self-declaration of their principle activity areas.

The data encompass all firms that Morningstar classifies as belonging to the following sectors; agricultural inputs and chemicals, beverages, packaged food, farm construction, restaurants, food distribution, farmproducts, and grocery. We divide the firm-level data into stages or our nodes; inputs, production, processing, and delivery to consumers, using Morningstar’s identification of the firm’s stated main activity and industry.

There are 97, 65, 156, and 136 firms in Stage 1, 2, 3 and 4, respectively. All financial data are converted into U.S dollar units based on the World Bank’s official exchange rate.

This study, for validation purposes, creates and employs two additional value creation measures; the CEVA and PEVA. The CEVA takes a value 1 when the MEVA>0 for any given year and, 0 otherwise. This measure accounts for the firms that may have extreme EVA and MEVA values within the ten-year study period. The second validation measure, PEVA, takes a value equal to 1 if the firm uses its capital efficiently for at least five years, and 0, otherwise. In this way, PEVA identifies firms that are consistently efficient users of capital over time.

International Food and Agribusiness Management Review

301

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Table 1. Summary statistics.

Variable

Definition

NOA Net operating assets Std. error Relative standard error of the mean AdjNOPAT Adjusted NOPAT1 Std. error Relative standard error of the mean EVA Economic value added Std. error Relative standard error of the mean MEVA Modified EVA Std. error Relative standard error of the mean CEVA Proportion of yrs w/a pos. MEVA Std. error Relative standard error of the mean Size Total assets(intangible+goodwill) Std. error Relative Standard Error of the mean Leverage Total liabilities/total equity Std. error Relative standard error of the mean Innovation R&D+intangibles+ goodwill Std. Error Relative standard error of the mean COGS Cost of good sold Std. error Relative standard error of the mean WACC Cost of capital Std. error Relative standard error of the mean Companies Observations 1

Unit Stage1

Stage2

Stage3

Stage4

All

$

5,978.55 -10,832.23 1.81 680.456 -1,465.82 2.15 263.579

2,494.47 -4,926.71 1.98 210.218 -456.259 2.17 43.349

6,053.28 -13,993.36 2.31 639.48 -1,993.79 3.12 256.586

2,199.51 -5,188.15 2.36 245.093 -598.569 2.44 107.024

4,466.69 -10,534.10 2.36 479.824 -1,443.89 3.01 187.868

-829.093 3.15 -0.195 -17.652 90.52 0.608

-301.547 6.96 -0.806 -19.65 24.38 0.517

-1,360.62 5.30 0.833 -15.714 18.86 0.614

-424.762 3.97 3.434 -18.421 5.36 0.666

-944.42 5.03 1.162 -17.544 15.10 0.616

-0.5 0.97 3,067.04

-0.487 0.79 4,206.56

-0.472 0.71 3,444.28

-0.487 0.79 4,381.75

-11,896.44 1.79 1.449

-6,192.53 2.02 1.122

-8,814.95 2.10 1.099

-8,490.96 2.47 1.561

-9,321.51 2.13 1.315

-4.893 3.38 1,021.13

-2.603 2.32 282.837

-20.719 18.85 2,790.08

-11.458 7.34 720.667

-14.123 10.74 1,479.89

-2,595.57 2.54 72.716 -15.934 0.22 0.072 -0.067 0.93 97 770

-690.86 2.44 77.535 -26.812 0.35 0.063 -0.084 1.33 65 431

-8,359.22 3.00 65.204 -17.266 0.26 0.063 -0.052 0.83 156 1,251

-2,086.82 2.90 70.59 -23.124 0.33 0.064 -0.054 0.84 136 1,014

-5,389.71 3.64 69.96 -20.646 0.30 0.065 -0.061 0.94 454 3,466

$

$

$

$

-0.489 0.80 6,636.80

NOPAT = net operating profit after taxes.

3.2 Firm size Previous literature uses different proxy variables for firm size. Generally, firm size can be measured by using the total sales (Arundel and Kabla, 1998), total assets (Gopalakrishnan and Damanpour, 2000; Zehri et al., 2012) or number of employees (Ettlie et al., 1984; Sawang and Unsworth, 2011; Zona et al., 2013).

Following Goldsmith and Sporleder (1998), Camisรณn-Zornoza et al. (2004) and Damanpour (2010), we use total assets as a proxy for firm size in this study. From total assets the amount of intangibles and goodwill is subtracted to avoid double counting and to allow their use as a measure for the level of innovation assets.

International Food and Agribusiness Management Review

302

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

3.3 Leverage We represent leverage in this research as the ratio of total liabilities to total equity. The interpretation of the leverage variable is as follows: the higher the ratio, the more debt a company uses in its capital structure.

Capital structure may not be significant in the process of value creation as debt level’s impact on value creation is not theoretically clear. On the one hand increasing leverage reduces the firm’s tax liability, and allows, ceteris paribus; the opportunity to more productively invest surplus cash flow, an urgency to perform well, and the sale of underperforming or unrelated businesses or assets (Houle, 2008). However, financial leverage is a risk factor and may effect risk taking (Ely, 1995; Hua and Templeton, 2010).

3.4 Cost of good sold Cost of goods sold is the expense a company incurs in order to manufacture, and create a product. Cost of goods sold is expressed as a percentage of revenue. At high levels of COGS, ceteris paribus, suppliers contribute a high proportion of the value produced by the firm. In general, the higher the level of COGS, the lower the level of value adding by the firm.

3.5 Innovation Following previous studies, we explore the role of innovation on value adding by means of three variables: R&D expenditures (Zona et al., 2013); goodwill (Degryse et al., 2012); and intangible assets (Kramer et al., 2011). The ‘Innovation 1’ variable aggregates all three innovation types. The ‘Innovation 2’ variable encompasses only capital assets (intangible assets and goodwill) and the ‘Innovation 3’ variable captures only R&D expenditures.

4. Results 4.1 Summary statistics across the value chain ■■ Size of firms Stages 1 (inputs) and 3 (processing) are comparable in terms of the average size of firms with the mean value of NOA being $5.979 and $6.053 billion dollars per firm, respectively. Their NOA values are 34% and 36%, respectively, higher than the mean operating assets of the agri-food value chain as a whole. The average net operating assets for Stage 4 (retail) averages $2.200 billion dollars, 12% lower than for Stage 2 (production). Stage 1 firms present the most uniform distribution in terms of size with a coefficient of variation of 1.81 about 25% smaller than the average for the entire value chain (Figure 3). While Stage 4 maintains a coefficient of variation of 2.36, about 30% larger than Stage 1.

■■ Earnings Likewise, this study does not find significant differences between Stage 1 and Stage 3 in terms of earnings as measured by AdjNOPAT. Stage 1 produces on average an AdjNOPAT of $680 million dollars, and Stage 3 generates net operating profits of $639 million dollars, on average, per firm. Thus stages 1 and 3 are of comparable size and earnings.

Statistically stages 2 and 4 do not differ. Stage 2 has an average AdjNOPAT of $210 million dollars, which is 14% lower than the average net operating profits for Stage 4, which produces $245 million dollars. This statistical equivalency in terms profitability is interesting. While Stage 2 (production) has earnings comparable with retail (Stage 4), we show later that Stage 2 employs capital quite differently, and thus creates value quite differently from Stage 4.

International Food and Agribusiness Management Review

303

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Firm size per stage

80,000

Firm size (USD million dollars)

70,000 60,000 50,000 40,000 30,000 20,000 10,000 0

1

2

Stage

3

4

Figure 3. Firm size by stage. Firm size is measured as adjusted total assets. It is expressed in million dollars.

â– â– Levels of value creation Agricultural inputs, Stage 1, is the node that creates the most value, producing, on average, an added economic value of $264 million dollars per firm, which is 40% higher than the average EVA across the whole value chain (Table 1). The production stage, Stage 2, creates, on average, $43 million dollars per firm, significantly (23%) less than the average firm along the chain. Stage 1 mean EVA statistically differs between Stages 2 (production) and 4 (retail), while the null hypothesis for significant differences between Stage 1(inputs) and Stage 3 (processing) cannot be rejected (Table 2). As discussed above EVA, as a metric, is highly correlated with firm size.

Down-chain has the highest rates of value adding, as measured by MEVA. On average, Stage 3 (processing) and Stage 4 (retail) create additional operating capital at the rate of 0.83% and 3.43% points per year respectively above their cost of capital, while upstream the average rate of value creation within Stages 1 and 2 are negative. Stage 4 differences are statistically significant from the other three stages. Stage 4, closest to the consumer, has a mean value MEVA that is 196% higher than the mean MEVA value of the overall chain, thus is almost 3x more effective using capital compared to the average firm. Stage 4 firms also show much higher levels of MEVA homogeneity across the firms within the stage. The coefficient of variation is 1/3rd the level compared with the entire value chain, and only 5.5% the variation of Stage 1, the input stage. There are 136 companies in Stage 4, of which 71 (52%) have persistent positive levels (PEVA) of value adding; a minimum of five years of positive value creation. Stage 2, the production stage, for example comprises 65 firms, and only 32% of those firms are persistent value creators.

Up-chain nodes, on average, destroy value. Their rate of return is less than the cost of the capital they employ. Stage 1’s average return on capital is 0.195% percentage points (slightly) below its average cost of capital. While Stage 2, the production stage, has a return on capital almost 1% (0.806%) less than its cost of capital. On average, up chain firms do not generate enough operating profit to cover the opportunity cost of their capital. Stage 1 MEVA differences from Stage 4 are statistically significant, but this is not true for the differences between Stage 1 and Stages 2 and 3. It is interesting to note that 68% of Stage 2 firms persistently International Food and Agribusiness Management Review

304

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Table 2. Mean tests.1 Null hypothesis Variable

S1=S2

S1=S3

S1=S4

S2=S3

S2=S4

S3=S4

All equal

NOA AdjNOPAT EVA MEVA CEVA Size Leverage Innovation 1 Innovation 2 Innovation 3 COGS

0.00 0.00 0.00 0.56 0.00 0.00 0.70 0.02 0.06 0.00 0.00

0.88 0.53 0.87 0.20 0.78 0.00 0.59 0.00 0.00 0.00 0.00

0.00 0.00 0.00 0.00 0.01 0.00 0.87 0.24 0.52 0.00 0.03

0.00 0.00 0.00 0.09 0.00 0.03 0.98 0.00 0.00 0.14 0.00

0.62 0.67 0.24 0.00 0.00 0.48 0.59 0.07 0.15 0.99 0.00

0.00 0.00 0.00 0.00 0.01 0.05 0.44 0.00 0.00 0.05 0.00

0.00 0.00 0.00 0.00 0.00 0.00 0.86 0.00 0.00 0.00 0.00

1 The

table shows the P-values of the mean tests calculated to determine whether the value outcomes for each stage are statistically different from each other for each variable of interest. ‘S1’ means Stage 1, ‘S2’ means Stage 2, ‘S3’ means Stage 3 and ‘S4’ means Stage 4. The null hypothesis is that the mean of particular variable are equal between two stages. Rejecting the null hypothesis means that there are statistically significant differences between two particular stages (rejecting the null are shown in italics).

destroy value, as measured by the PEVA. It is the chain node with the highest percentage of firms having negative value creation levels across time.

■■ Innovation assets and R&D investment Stage 3, the manufacturing stage, is the most innovative chain member as measured by the summation of goodwill, intangible assets and research and development investment. Manufacturing firms on average invest 66% of their average adjusted total assets in innovation: R&D, intangibles, and goodwill. The differences with the other three stages are statistically significant. Stage 3 has the highest coefficient of variation among its firms, thus is the most heterogeneous in terms of employing innovation capital (Figure 4). Stage 4, the retail stage, is the second most innovative value chain node, having 21% of adjusted total assets in innovation activities, which is one-third the level of Stage 3. Stage 1 assigns only 15% of adjusted total assets to innovative activities, which is one-fifth the rate occurring among Stage 3 firms. The difference in the employment of innovation assets between Stage 3 and Stage 1 is statistically significant. As the most commoditized node, it is not surprising that Stage 2, the production stage, is the least innovative chain member, with on average 9% of total assets invested in innovative assets.

Intangibles and goodwill, the Innovation 2 variable, dominate innovation activity (99%) in all stages, except for agricultural inputs (Stage 1) where R&D comprises 16% of innovation activities. Average Stage 1 R&D expenditures amount to $141 million dollars per year per firm, or 34% of the average net income. This is eight times more than is spent in Stage 3, the next highest stage with investment in R&D. Again we point out to the reader this important statistic, whereby there are in general significant differences across the four nodes in terms of research and development investment, and specifically, differentially high levels of R&D among Stage 1 firms.

■■ Cost of goods sold The COGS is statistically different across the four value chain nodes (Figure 5). Consistent with the MEVA analysis, the down-chain nodes generate higher gross margins. The agricultural production stage, Stage 2, has the highest COGS at 78%, 17% above Stage 3, which has the lowest COGS at 65%. Product differentiation at Stage 1 with high performance seeds or branded equipment allow agricultural suppliers higher gross margins International Food and Agribusiness Management Review

305

Cucagna and Goldsmith

Volume 21, Issue 3, 2018 Innovation level per stage

Innovation 1 (USD million dollars)

80,000

60,000

40,000

20,000

0

1

2

3

Stage

4

Figure 4. Innovation level per stage. The graph shows the level of innovation as measured by the variable ‘Innovation 1’, which is composed of goodwill and intangible assets, and R&D expenditures. It is measured in million dollars.

COGS distribution by stage

0.04

Stage 1

0.03

Density (%)

Stage 4 Stage 2

0.02 Stage 3

0.01

0

0

50

COGS (%)

100

150

Figure 5. Differing behaviors among the four stages as measured by the density distribution of cost of goods sold (COGS). Notice the differences in the level of the mode (location of the peak) and the homogeneity (breadth) among the four stages. compared with their customers in Stage 2. Alternatively, Stage 2’s high COGS reflects the raw commodity nature and the competitive market structure commonly found at that node of the value chain.

International Food and Agribusiness Management Review

306

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

4.2 Empirical model An ordinary least squares regression model provides estimates of the drivers of EVA and MEVA across the agri-food value chain. We employ a second model for validation purposes that utilizes probit regression to estimate CEVA and PEVA. Formally, the models are as follows: Value Added Measure {EVAit, MEVAit, CEVAit, PEVAit } = α + ρStage1 + βStage3 + γStage4 + δlogSizeit + θleverageit + νlnn2it + τlnn3it + πCOGSit + CountryFE + YearFE + uit

(6)

The model includes country-fixed effects (CountryFE) and year-fixed effects (yearFE) to avoid endogeneity due to unobservable information in the error term. Stage 1, 3, and 4 are dummy variables that takes a value equal to one if, respectively, the firm belongs to Stage 1, 3, or 4, and zero otherwise. The values express the difference with respect to Stage 2, the production stage. The expectation is that Stages 1, 3, and 4 create greater value than Stage 2.

The model specifies the logarithm of adjusted total assets to address the effects of firm size on value creation.

The expected sign on firm size is positive. Total current liabilities over total equity serves to measure the role of debt on value creation. There is no expected sign on debt level. The model includes two innovation measures; ‘Inn2’ is the value of the firm’s intangible and goodwill assets and ‘Inn3’ reflects R&D expenditures.

Greater levels of innovation should produce higher levels of value creation. Finally, the variable COGS is included in the model to proxy for the level of product differentiation. The higher the level of COGS, the lower is the level of product differentiation. The expected sign is negative as it is expected that product differentiation and value adding are positively related.

Our panel involves numerous countries with differing tax policies. The NOPAT variable of the EVA estimate, employs NOPAT, thus may be biased. We validate the value creation estimate by replacing NOPAT with the net operating profits before taxes. The results are robust showing no change in the EVA coefficient estimates. Additionally, the four regression models control for the country fixed effects, which in part addresses differences in tax policies.

■■ H ypothesis I: value addition levels differ across stages as each stage contributes differently to the value creation process We confirm Hypothesis I as the value addition levels at Stages 1, 3, and 4 significantly differ from Stage 2, the reference stage. In terms of EVA, Stage 3 contributes the most value in the chain, adding $41 million per year per firm more than a Stage 2 firm, ceteris paribus (Table 3). The coefficient estimate is significant at the 0.10 level. Similarly, retail (Stage 4) delivers about $39 million more dollars per year per firm in terms of EVA than Stage 2. The coefficient estimate is significant at the 0.05 level.

■■ H ypotheses II: we expect the agricultural input Stage (1) to be a high value-adding node given the presence of input brands and the high level of agricultural research Consistent with Hypothesis II (and III), the up-chain Stages 1 and 2 attain low levels of value adding and do not statistically differ in terms of their level of value added. The superiority of the value creation levels of down-chain nodes in comparison with Stage 2 is consistent with the contemporary agribusiness literature that suggests that Stage 2 is the most commoditized stage (Van der Ploeg, 2000).

We cannot accept the null for Hypothesis II that Stage 1 firms effectively add value compared to Stage 2. In terms of EVA and MEVA, Stage 1 firms are statistically no different from Stage 2 firms. This empirically confirms the commonly held belief that more value creation occurs down chain, as opposed to up chain.

International Food and Agribusiness Management Review

307

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Table 3. Regression results.1 Variables

(1) EVA

(2) MEVA

(3) CEVA

(4) PEVA

Stage 1

-15.928 (22.532) 40.634* (24.768) 39.273** (19.805) 40.346*** (7.133) 3.645* (2.133) 0.045*** (0.011) 1.464*** (0.136) -27.550** (11.316) -8.626 (55.598) Yes Yes 0.027 0.010 0.952 0.505 3,466

-0.601 (1.017) 1.586* (0.931) 4.716*** (0.958) 2.544*** (0.126) 0.029 (0.019) 0.001*** (0.000) 0.001 (0.002) -0.113 (0.327) -1.744 (1.232) Yes Yes 0.005 0.000 0.000 0.186 3,466

0.055** (0.028) 0.088*** (0.025) 0.146*** (0.026) 0.059*** (0.005) 0.001 (0.001) 0.001*** (0.000) 0.001*** (0.000) -0.036*** (0.009) 0.417*** (0.054) Yes Yes 0.120 0.000 0.002 0.264 3,466

0.158*** (0.028) 0.216*** (0.024) 0.218*** (0.026) 0.088*** (0.007) 0.001 (0.001) 0.001 (0.001) 0.001 (0.001) -0.009 (0.009) -0.116* (0.064) Yes Yes 0.007 0.008 0.893 0.310 3,466

Stage 3 Stage 4 LN(Size) Leverage Innovation 2 Innovation 3 COGS Constant Country fixed effects Year fixed effects P-value Stage 1=Stage 3 P-value Stage 1=Stage 4 P-value Stage 3=Stage 4 R-squared Observations

1 Robust standard errors in parentheses; *** P<0.01, ** P<0.05, * P<0.1; in column 1 and 2, the appropriate econometric model is to

use a probit model. Both the probit and the ordinary least squares (OLS) econometric techniques were used: The two methods give essentially identical results, with insignificant differences in the variables. For simplicity, only the results of the OLS regression are shown; COGS = cost of goods sold; EVA = economic value added; MEVA = modified economic value added; CEVA = creation or destruction of value; PEVA = persistence of economic value added.

Both stages have low rates of value creation, with conditional means close to zero of -0.194 and -0.806 for Stage 1 and Stage 2 respectively.

The constant coefficient in the CEVA model though is statistically significant at the 0.01 level, indicating that Stage 2 firms do have a positive probability in any year to add value. However, the added value will be low, and will be swamped by large levels of value destruction in a given year. As hypothesized, the results show differences in the probability to create value across stages, and down-chain stages have higher probabilities for value creation than those up chain.

PEVA, the final value creation metric reflects the persistence of value creation. The dummy variable takes a value equal to 1 if a company uses its capital efficiently for at least five years (half of the period covered by the database) during the study period, and zero otherwise. This variable identifies the proportion of companies that persistently create value within each stage. Stage 4 firms, the retail stage, possess a probability of being persistent that is 21.8 percentage points higher than a Stage 2 firm (production), which has a negative 11.6% probability. The coefficient estimate for Stage 3 is 21.6% and for Stage 1, 15.8% greater than Stage 2. The constant is negative and statistically significant at the 0.10 level and thus indicates a negative probability that a Stage 2 firm will be a persistent value creator. The PEVA model results validate the hypotheses that International Food and Agribusiness Management Review

308

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

value creation levels along the food and agribusiness value chain differ across stages, and levels are higher down chain compared with up chain.

■■ H ypotheses III: the lack of coordination and the commercialization of undifferentiated products as well as the condition of being primarily a price taker makes the production stage exhibit the lowest levels of value addition among the four value chain nodes We accept the null for Hypothesis III that Stage 2 firms add relatively little value to the food and agribusiness chain, as measured by capital utilization efficiency. In a given year, 52% of Stage 2 firms do create value, which reflects the competitive nature of the stage. The average firm creates $43 million of added value.

Firm performance within Stage 2 is heterogeneous reflected by the high coefficient of variations for EVA (6.96) and MEVA (0.97). It is important to note that there are firm-year pairs in Stage 2 where significant value destruction occurs. Stage 2 firms also lack persistency where only 32% of the firms create value for at least five years of the study period. The low PEVA levels result in a negative and significant constant estimate for PEVA. ■■ H ypotheses IV: food manufacturing (Stage 3) will exhibit a relatively high level of value addition, compared to upstream stages, due to strong product differentiation, coordination with retailers, and access to low cost inputs from the production stage ■■ H ypotheses V: the retail firms (Stage 4) with their consumer orientation, need to innovate, and strategic position within the value chain, will be a relatively high value creator, compared to upstream stages We accept null hypotheses IV and V that Stages 3 and 4 create significant value. Stage 3 and Stage 4, respectively increase the rate of value adding in terms of returns on invested capital 1.59% and 4.71%, respectively, and the estimates are statistically significant at the 0.10 and 0.01 levels. The empirical analysis shows that Stage 4, the retail stage, most efficiently uses its capital; being the largest contributor to the value creation process along the food and agribusiness value chain. The MEVA metric, as opposed to EVA, accounts for a standardized value creation level without taking into account the nominal quantity of capital invested in the productive process.

CEVA is a dummy variable that takes value equal to 1 if the observation has a positive value creation level in a given year during the study period, and zero, otherwise. Thus the level of value adding, whether in nominal terms (EVA) or in real terms (MEVA) does not factor into the estimation. The coefficient value reflects the probability of creating positive value compared with Stage 2. All stages have a positive and significant probability that a firm creates more value than a firm in Stage 2. Positive probabilities range from the highest being Stage 4 (retail) at 15% to Stage 1 (inputs), at 6%. The probability for a retail firm is 65% higher than Stage 3 (manufacturing), which is 60% more likely than Stage 1.

■■ H ypothesis VI: firms with higher levels of intangible assets and goodwill create more value. The coefficient estimate for the goodwill and intangible asset variable should be positive and significantly different from zero ■■ H ypothesis VII: firms with higher levels of research and development expenditures create more value. Similarly, the coefficient estimate for the R&D expenditure variable should be positive and significantly different from zero All the regression models include two measures of innovation for testing hypotheses VI and VII: 1) intangible and goodwill assets as percentage of total assets, the Inn 2 variable level; 2) and R&D expenditures, the Inn3 variable. Both are hypothesized to carry a positive sign because they drive innovation, and thus, value creation.

International Food and Agribusiness Management Review

309

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

The results support Hypothesis VI. The coefficient estimates for intangibles and goodwill assets (Inn2) are positive for all four models, and significant at the 0.01 level for all but the PEVA model. In real terms, intangibles and goodwill raise MEVA, albeit slightly, and the result is significant at the 0.01 level. Intangible and goodwill assets though effect EVA significantly less than R&D expenditures. A 1% increase in the ratio of intangibles and goodwill to total assets, raises EVA $45,000 for a firm in a given year.

The coefficient estimates for R&D expenditures (Inn3) are also positive for all four models and are significant at the 0.01 level for all but the MEVA and PEVA models. We accept the null for Hypothesis VII. Increasing R&D expenditures raises annual EVA per firm $1.5M, which is over 30 times the impact of intangibles and goodwill assets. Both intangibles and goodwill, and R&D lead to persistent value addition. The coefficient estimates in the PEVA model are small but significant at the 0.01 level. The regression results provide empirical evidence of not only a positive relationship between innovation and value creation, but also a measure of their contribution to value creation along the food and agribusiness value chain.

■■ H ypothesis VII: firms with higher degrees of product differentiation, a smaller ratio of COGS to sales, create more value Hypothesis VIII asserts that the ratio of COGS to sales, which is a measure of product differentiation, drives value creation. The smaller this ratio, the larger is the gross margin. Increasing the ratio of COGS to sales by one unit reduces the firm’s EVA by $27.6 million dollars. The coefficient estimate is significant at the 0.05 level. Similarly, the probability of consistently creating value in any given year decreases by 3.6% when the COGS:sales ratio increases by one unit. Both coefficient estimates are significant at the 0.01 level and confirm Hypothesis VIII. For the CEVA and PEVA models, the regression results are negative but not statistically significant. The regression results for the four models show a positive relationship between product differentiation and value creation.

■■ Hypothesis IX: there is a positive relationship between firm size and value creation The logarithm of adjusted total assets as a measure of firm size has a positive and statistically significant coefficient estimate, at the 0.01 level, in all four of the regression models. Therefore, firm size is a significant determinant of value creation, which supports Hypothesis IX. An increase of 1% in size increases the EVA level by $40 million dollars. An increase of 1% in the size of a firm also increases the rate of value creation (MEVA) by 2.5 percentage points. The probability of CEVA increases by 6% for the average company when the firm size increases by 1%. Finally, the probability of persistently adding value increases by 9% when firm size increases by 1%. These results are consistent with previous literature where larger companies, ceteris paribus, create greater value because of their better access to resources, economies of scale and scope, and the greater ability to spread risk.

Finally, leverage is measured as total liabilities divided by total equity. This variable captures the type of capital employed by the firm. Theory is not clear on the effect of leverage on value creation, thus we present no hypothesis. The results were weak on the linkage between leverage levels and value adding. Only one (EVA) of the four models generates a statistically significant (0.10) coefficient for the leverage variable.

Firms with higher levels of debt do produce greater levels of EVA. The result is significant at the 0.10 level.

Increasing the ratio of debt to equity by one unit increases the level of value creation by $3.6 million dollars, ceteris paribus. These results though should be cautiously interpreted as none of the validation results support this finding.

4.3 Two short examples The following two example applications of the EVA methodology demonstrate that while on average up chain stages of the food and agribusiness value chain are relatively poor users of capital, there are exceptions. We compare two Stage 1 firms, a poor performer, Alamo, with a strong performer, Monsanto. The Alamo Group International Food and Agribusiness Management Review

310

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Inc., manufactures and sells agricultural and infrastructure maintenance equipment, and persistently creates value. However, on average, Alamo Group Inc. destroys $4.5 million dollars at the average rate of -1.12% below their cost of capital (Figure 6). Thus, value destruction can be extreme in a given year. On the other hand, another Stage 1 firm, Monsanto, annually creates $1,165 million dollars of EVA at an annual rate of 7.6 percentage points above their cost of capital. Monsanto invests 10% of sales in R&D expenditures and maintains 30% of the adjusted total assets in goodwill and intangible assets. The firm is also a consistent value creator having positive CEVA in all the years under analysis. Furthermore, Monsanto maintains an average COGS of 49.06%, which is 24% lower than the average firm in the input stage.

5. Conclusion The objectives of this manuscript were threefold: (1) to provide a financial, and thus formal, definition of value added; (2) to measure value addition by firms across the four nodes/stages of the value chain; and (3) to test hypotheses as to the drivers of value creation that differ across the four nodes.

First, the current understanding of value creation or value adding in food and agri-business presents a broad and qualitative definition. The current definition holds that firms create or add value when they introduce or modify a new method or product, or they perform an activity that was previously performed by another member of the chain. Furthermore, the current definition allows for greater coordination as a method for creating or adding value. We advance the definition of value adding by building from the work of Amanor-Boudu (2003) and provide a formal measure of value creation. By explicitly measuring the level of value creation our method effectively formalizes the thinking as to how and when a firm creates value.

Specifically, we follow the general finance literature and define value creation as a capital creation process; capital created in excess of the capital used to produce the new capital. Using our more formal definition of value creation adds clarity by showing that Stage 2, the production stage, is inferior across all four of our measures at creating value compared to others along the agri-food value chain. But interestingly, and we argue importantly, Stage 2â&#x20AC;&#x2122;s profitability levels, as measured in nominal terms, or per asset compares quite favorably with down chain firms, i.e. Stage 4. Though beyond the scope of this research, we show important

Alamo and Monsanto value creation levels: 2003-2012

MEVA: Percentage return on capital

25

Monsanto

20 15 10

Alamo

5 0 2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

-5 -10

Year Alamo

Monsanto

Figure 6. The value creation levels of two different firms within the same stage (1), Alamo and Monsanto: 2003-2012.

International Food and Agribusiness Management Review

311

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

differences between value creation and profitability. These differences may be material for explaining the sustainability of the Stage 2 business model that creates and thus captures low levels of value.

Second, extensive research using the Food Dollar or Marketing Bill approaches clearly shows the decline in the share of the consumer dollar accruing to producers. It is unclear what such findings specifically indicate about value creation and capture. But from a policy perspective, producers and policymakers assert that anti-competitive practices along the value chain explain the disproportionately low level of value capture by producers, i.e. see the EU Council (2016). Our analysis using all four metrics, EVA, MEVA, CEVA, and PEVA, adds some specificity to both issues; the linkages between the decline in the Food Dollar capture, and the assertion as to whether producers capture a disproportionate share of value created. In particular, of the 65 Stage 2 firms only 32 are persistent value creators, for example. Thus our findings do not state that Stage 2 firm cannot create value, but on average firms create less capital than they employ. Our findings do not support the premise that anti-competitive forces result in Stage 2 firmsâ&#x20AC;&#x2122; low level of value capture.

Instead we indicate that producers on average capture a level of value commensurate with the value they create, especially as compared to the other nodes along the value chain.

Third, our method also allows us to validate these findings by testing nine hypotheses derived from theoretical drivers of value creation. The results of the empirical tests confirm that Stage 2 (and Stage 1) differently employs capital and higher levels of COGS, which explains the causes for low levels of value creation.

Stage 2 holds significantly lower levels of innovation assets such as goodwill and an intangible assets, and engages in very little research and development, compared with other nodes along the value chain. One implication of the fact that Stage 2 holds differential levels of capital and innovation assets may reflect an unconventional form of risk sharing along the agri-food value chain. While producers (Stage 2) face many risks such as price and weather, they can rely on chain partners up and down chain to engage in the high risk and costly activities of research and development and intangible asset assembly. Analyses that focus on risk management at the farm gate alone may fail to capture the complete risk sharing that occurs along the agri-food value chain as a whole.

We posit from these findings that there exists intra-stage complementarities across the agri-food value chain, whereby the value chain as a whole employs capital well and creates sufficient levels of value. It is improper to assume that each node must similarly employ capital and create capital at the same rates. Maybe the relatively competitive and uncoordinated Stage 2 production stage supports value creation within other stages along the value chain by, for example, holding down the COGS for down chain firms. There may be tradeoffs across the chain whereby innovation occurs up and down chain away from production, which in turn reduces the burden for Stage 2 firms to deploy innovation capital. This logic suggests that firms in Stage 2 in general focus on productive activities, rather than riskier creative activities. The implication is that all firms along the value chain need not employ the same capital plans, for example involving like levels of goodwill and intangible assets, or research and development, with respect to value chain. Critical though, to Stage 2 sustainability, may be linkages and complementarities within a chain. The policy implication for those seeking to strengthen rural economies is that optimal chain performance may occur when some firms, say in Stage 2, are allowed to specialize, remain independent, achieve economies of scale, and serve a more complementary role, and by doing so, keep raw material prices low and the supply thick.

Finally, this study is restricted insofar as it only uses financial data. Thus neglects the important topic as to whether coordination, as a source of innovation, can also account for the differences across the four stages.

An area for further research would be an analysis of the impact of coordination (horizontal and vertical) on value creation. Coordination among value chain actors implies complementarities among the activities of different chain members. The food and agribusiness value chain, is a system whereby vertical linkages may impact the investment decisions of individual ďŹ rms within the chain. (Hendrikse and Bijman, 2002).

Coordination among members in a chain can reduce inventory holding costs, delivery times, transportation costs, levels of loss and damage, and improve customer service (Lee et al., 1997; Simatupang et al., 2002), which may simultaneously improve capital use efficiency and levels of value added. Further work is needed International Food and Agribusiness Management Review

312

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

to determine how best to parameterize coordination and collaboration within the EVA analytical approach.

Doing so would clarify the empirical relationship between coordination and value creation.

References Alexander, C. and R.E. Goodhue. 1999. Production system competition and the pricing of innovations: an application to biotechnology and seed corn. American Agricultural Economics Association. Available at: http://tinyurl.com/ya3rv4yr.

Amanor-Boadu, V. 2003. A conversation about value-added agriculture. Value-added business development program. Department of Agricultural Economics. Kansas State University. Available at: http://tinyurl. com/y8q7ajbm.

Anderson, J.C. 1995. Relationships in business markets: exchange episodes, value creation, and their empirical assessment. Journal of the Academy of Marketing Science 23(4): 346-350.

Anderson, A.M., R.P. Bey and S.C. Weaver. 2005. Economic value added adjustments: much to do about nothing. Available at: http://tinyurl.com/y7vat3j4.

Anderson, D.P. and D. Hanselka. 2009. Adding value to agricultural products. Available at: http://tinyurl. com/y86n69g3.

Anderson, J.C. and J.A. Narus. 1998. Business marketing: understand what customers value. Harvard Business Review 76: 53-67.

Artz, K.W., P.M. Norman, D.E. Hatfield and L.B. Cardinal. 2010. A longitudinal study of the impact of R&D, patents, and product innovation on firm performance. Journal of Product Innovation Management 27(5): 725-740.

Arundel, A. and I. Kabla. 1998. What percentage of innovations are patented? Empirical estimates for European firms. Research policy 27(2): 127-141.

Balsam, S., G.D. Fernando and A. Tripathy. 2011. The impact of firm strategy on performance measures used in executive compensation. Journal of Business Research 64(2): 187-193.

Barclay, M.J. and C.W. Smith. 1999. The capital structure puzzle: another look at the evidence. Journal of Applied Corporate Finance 12(1): 8-20.

Bigliardi, B. and E. Bottani. 2010. Performance measurement in the food supply chain: a balanced scorecard approach. Facilities 28(5/6): 249-260.

Boehlje, M., M. Roucan-Kane and S. Bröring. 2011. Future agribusiness challenges: strategic uncertainty, innovation and structural change. International Food and Agribusiness Management Review 14(5): 53-82.

Born, H. and J. Bachmann. 2006. Adding value to farm products: an overview. National Center For Appropriate Technology. Available at: https://tinyurl.com/ybmzy7ko.

Bourlakis, M., G. Maglaras and C. Fotopoulos. 2012. Creating a ‘best value supply chain’? Empirical evidence from the Greek food chain. The International Journal of Logistics Management 23(3): 360-382.

Bruner, R.F., K.M. Eades, R.S. Harris and R.C. Higgins. 1998. Best practices in estimating the cost of capital: survey and synthesis. Financial Practice and Education 8: 13-28.

Burch, D. and G. Lawrence. 2005. Supermarket own brands, supply chains and the transformation of the agri-food system. International Journal of Sociology of Agriculture and Food 13(1): 1-18.

Camisón-Zornoza, C., R. Lapiedra-Alcamí, M. Segarra-Ciprés and M. Boronat-Navarro. 2004. A metaanalysis of innovation and organizational size. Organization Studies 25(3): 331-361.

Canning, P. 2011. A revised and expanded food dollar series a better understanding of our food costs. USDA Economic Research Service Available at: http://tinyurl.com/ycp8nan2.

Canning, P., A. Weersink and J. Kelly. 2016. Farm share of the food dollar: an IO approach for the United States and Canada. Agricultural Economics 47: 505-512.

Carlson, W. 2004. Don’t produce a commodity. Greenhouse Grower 22(12): 20-23.

Chmelíková, G. 2008. Economic value added versus traditional performance measures by food-processing firms in the Czech Republic. International Food and Agribusiness Management Review 11(4): 49-66.

Cohen, W.M. and S. Klepper. 1996. Firm size and the nature of innovation within industries: the case of process and product R&D. The review of Economics and Statistics 78: 232-243.

International Food and Agribusiness Management Review

313

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Coltrain, D., D. Barton and M. Boland. 2000. Value added: opportunities and strategies. Available at: http:// tinyurl.com/ycoxv5m9.

Damanpour, F. 2010. An integration of research findings of effects of firm size and market competition on product and process innovations. British Journal of Management 21(4): 996-1010.

Degryse, H., P. de Goeij and P. Kappert. 2012. The impact of firm and industry characteristics on small firms’ capital structure. Small Business Economics 38(4): 431-447.

Demont, M., K. Dillen, E. Mathijs and E. Tollens. 2007. GM crops in Europe: how much value and for whom? EuroChoices 6(3): 46-53.

Doyle, P. 2000. Value-based marketing. Journal of Strategic Marketing 8(4): 299-311.

Ely, K.M. 1995. Operating lease accounting and the market’s assessment of equity risk. Journal of Accounting Research 33: 397-415.

Ettlie, J.E., W.P. Bridges and R.D. O’keefe. 1984. Organization strategy and structural differences for radical versus incremental innovation. Management science 30(6): 682-695.

EU Council. 2016. Strengthening farmers’ position in the food supply chain and tackling unfair trading practices. EU Council Outcome of Proceedings 15508. Available at: http://tinyurl.com/y9ztsew5.

Evans, E. 2006. Value added agriculture: is it right for me? EDIS document FE638, Florida Cooperative Extension Service, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL, USA. Available at: http://edis.ifas.ufl.edu/fe638.

Farfan, O. 2005. Understanding and escaping commodity-dependency: a global value chain perspective.

Available at: http://tinyurl.com/yd9wm6ky.

Farm Futures. 2014. USDA announces $25 million in value-added producer grants. Available at: http:// tinyurl.com/ya4j753n.

Fulton, M. and K. Giannakas. 2002. Agricultural biotechnology and industry structure. AgBioForum 4(2).

Available at: http://tinyurl.com/yalb93c5.

Geyser, M. and I.E. Liebenberg. 2003. Creating a new valuation tool for South African agricultural cooperatives. Agrekon 42(2): 106-115.

Goldsmith, P.D. 2001. Innovation, supply chain control, and the welfare of farmers: the economics of genetically modified seeds. American Behavioral Scientist 44(8): 1302-1326.

Goldsmith, P. and K. Bender. 2004. Ten conversations about identity preservation. Journal on Chain and Network Science 4(2): 111-123.

Gopalakrishnan, S. and F. Damanpour. 2000. The impact of organizational context on innovation adoption in commercial banks. IEEE Transactions on Engineering Management 47(1): 14-25.

Goldsmith, P., A. Salvador, D. Knipe and E. Kendall. 2002. Structural change or logical incrementalism? turbulence in the global meat system. Journal on Chain and Network Science 2(2): 101-115.

Goldsmith, P.D. and C. Silva. 2006. Specialty soybean and corn survey. National Soybean Research Laboratory.

Available at: http://tinyurl.com/y7jmcnaa.

Goldsmith, P.D. and T.L. Sporleder. 1998. Analyzing foreign direct investment decisions by food and beverage firms: an empirical model of transaction theory. Canadian Journal of Agricultural Economics/Revue canadienne d’agroeconomie 46(3): 329-346.

Gopinath, M., T.L. Roe and M.D. Shane. 1996. Competitiveness of US food processing: benefits from primary agriculture. American Journal of Agricultural Economics 78(4): 1044-1055.

Grunert, K.G., L.F. Jeppesen, K.R. Jespersen, A.M. Sonne, K. Hansen, T. Trondsen and J.A. Young. 2005.

Market orientation of value chains: a conceptual framework based on four case studies from the food industry. European Journal of Marketing 39(5/6): 428-455.

Hayenga, M. and N. Kalaitzandonakes. 1999. Structure and coordination system changes in the US biotech seed and valueadded grain market. Available at: http://tinyurl.com/y7ej4lbr.

Hecker, A. and A. Ganter. 2013. The influence of product market competition on technological and management innovation: firm-level evidence from a large-scale survey. European Management Review 10(1): 17-33.

Hendrikse, G. and J. Bijman. 2002. Ownership structure in agrifood chains: the marketing cooperative.

American Journal of Agricultural Economics 84(1): 104-119.

Houle, M.D. 2008. Economic value added. Senior Honors Papers 38. Available at: http://digitalcommons. liberty.edu/honors/38.

International Food and Agribusiness Management Review

314

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Hua, N. and A. Templeton. 2010. Forces driving the growth of the restaurant industry in the USA. International Journal of Contemporary Hospitality Management 22(1): 56-68.

Humphrey, J. and O. Memedovic. 2006. Global value chains in the agrifood sector (working paper). Available at: http://tinyurl.com/y8qh4obd.

Ibendahl, G.A. and R.A. Fleming. 2003. Using economic value added (EVA) to examine farm businesses.

In: Annual Meeting of the Southern Agricultural Economics Association. Available at: http://tinyurl. com/y7emtljn.

Kalafut, P.C. and J. Low. 2001. The value creation index: quantifying intangible value. Strategy and Leadership 29(5): 9-15.

Kale, P., J. Dyer and H. Singh. 2001. Value creation and success in strategic alliances: alliancing skills and the role of alliance structure and systems. European Management Journal 19(5): 463-471.

Kampen, K. 2011. Financial analysis of three value-added dairy enterprises in Vermont, Wisconsin, and New York. PhD thesis. California Polytechnic State University, San Luis Obispo, CA, USA. Keef, S.P. and M.L. Roush. 2003. The relationship between economic value added and stock market performance: a theoretical analysis. Agribusiness 19(2): 245-253.

Ketchen Jr, D.J., W. Rebarick, G.T.M. Hult and D. Meyer. 2008. Best value supply chains: a key competitive weapon for the 21st century. Business Horizons 51(3): 235-243.

Kramer, J.P., E. Marinelli, S. Iammarino and J.R. Diez. 2011. Intangible assets as drivers of innovation: empirical evidence on multinational enterprises in German and UK regional systems of innovation.

Technovation 31(9): 447-458.

Lee, H.L., V. Padmanabhan and S. Whang. 1997. The bullwhip effect in supply chains. Sloan management review 38(3): 93.

Lindgreen, A., M.K. Hingley, D.B. Grant and R.E. Morgan. 2012. Value in business and industrial marketing: past, present, and future. Industrial Marketing Management 41(1): 207-214.

Liu, X. and J. Niemi. 2014. A new balance of power between suppliers and retailers in Finland. Proceedings in Food System Dynamics 12-17. Available at: http://ageconsearch.umn.edu/record/199060.

Lobao, L. and K. Meyer. 2001. The great agricultural transition: crisis, change, and social consequences of twentieth century US farming. Annual review of sociology 27: 103-124.

Martin, S.K. and A. Jagadish. 2006. Agricultural marketing and agribusiness supply chain issues in developing economies: the case of fresh produce in Papua New Guinea. Available at: http://tinyurl.com/ybpjucv7.

Matopoulos, A., M. Vlachopoulou, V. Manthou and B. Manos. 2007. A conceptual framework for supply chain collaboration: empirical evidence from the agri-food industry. Supply Chain Management 12(3): 177-186.

Meisel, J.B. and S.A.Y. Lin. 1983. The impact of market structure on the firm’s allocation of resources to research and development. Quarterly Review of Economics and Business 23: 28-43.

Moody’s. 2014. Research and rating. Available at: http://tinyurl.com/kfepkkq.

Moschini, G. and H. Lapan. 1997. Intellectual property rights and the welfare effects of agricultural R&D. American Journal of Agricultural Economics 79(4): 1229-1242.

Nair, A. and L. Filer. 2003. Cointegration of firm strategies within groups: a long-run analysis of firm behavior in the Japanese steel industry. Strategic Management Journal 24(2): 145-159.

Noemi, V. 2012. Members of a supply chain and their relationships. Applied Studies in Agribusiness and Commerce. Agroinform Publishing House, Budapest, Hungary, pp. 131-134.

Omidvar, V., D.G. Brewin and J.G. Carlberg. 2006. Meat processing in North America: successes, failures and opportunities. In: Southern Agricultural Economics Association Annual Meeting. Orlando, FL, USA. Available at: http://tinyurl.com/y6w27vts.

Opara, L.U. 2003. Traceability in agriculture and food supply chain: a review of basic concepts, technological implications, and future prospects. Journal of Food Agriculture and Environment 1: 101-106.

Oustapassidis, K. and A. Vlachvei. 1999. Profitability and product differentiation in Greek food industries.

Applied economics 31(10): 1293-1298.

Parasuraman, A. 1997. Reflections on gaining competitive advantage through customer value. Journal of the Academy of marketing Science 25(2): 154-161.

International Food and Agribusiness Management Review

315

Cucagna and Goldsmith

Volume 21, Issue 3, 2018

Phillips, J., E.J. Holcomb and K. Kelley 2007. Determining interest in value-added planters: consumer preference and current grower and retailer perceptions. HortTechnology 17(2): 238-246.

Sawang, S. and K.L. Unsworth. 2011. Why adopt now? Multiple case studies and survey studies comparing small, medium and large firms. Technovation 31(10): 554-559.

Sexton, R.J. 2013. Market power, misconceptions, and modern agricultural markets. American Journal of Agricultural Economics 95(2): 209-219.

Sharma, A.K. and S. Kumar. 2010. Economic value added (EVA) – literature review and relevant issues.

International Journal of Economics & Finance 2(2): 200-220.

Simatupang, T.M., A.C. Wright and R. Sridharan. 2002. The knowledge of coordination for supply chain integration. Business process management journal 8(3): 289-308.

Sparling, D. and C.G. Turvey. 2003. Further thoughts on the relationship between economic value added and stock market performance. Agribusiness 19(2): 255-267.

Sporleder, T. 2006. Strategic alliances and networks in supply chains: knowledge management, learning and performance measurement. Quantifying the Agri-Food Supply Chain 15: 161-171.

Streeter, D.H., S.T. Sonka and M.A. Hudson. 1991. Information technology, coordination, and competitiveness in the food and agribusiness sector. American Journal of Agricultural Economics 73(5): 1465-1471.

Turvey, C.G., L. Lake, E. Van Duren and D. Sparling. 2000. The relationship between economic value added and the stock market performance of agribusiness firms. Agribusiness 16(4): 399-416.

U.S. Congress. 2002. Farm security and rural investment act of 2002. In: 107th Congress, 2nd Session. Public Law (No. 107-171) May. Available at: http://tinyurl.com/y93m5hfw.

USDA. 2013. Farm structure and organization. Available at: http://tinyurl.com/ybvoma9s.

USDA. 2017. Food Dollar Series. Available at: http://tinyurl.com/yb6d3wtp.

Van der Ploeg, J.D. 2000. Revitalizing agriculture: farming economically as starting ground for rural development. Sociologia ruralis 40(4): 497-511.

Vanyi, N. 2012. Members of a supply chain and their relationships. Applied Studies in Agribusiness and Commerce 6(5): 131-134.

Viaene, J. and X. Gellynck. 1995. Structure, conduct and performance of the European food sector. European Review of Agricultural Economics 22(3): 282-295.

Walbert, L. 1994. The Stern Stewart performance 1000: using Eva™ to build market value. Journal of Applied Corporate Finance 6(4): 109-112.

Walter, A., T. Ritter and H.G. Gemünden. 2001. Value creation in buyer-seller relationships: theoretical considerations and empirical results from a supplier’s perspective. Industrial Marketing Management 30(4): 365-377.

Zéghal, D. and A. Maaloul. 2010. Analysing value added as an indicator of intellectual capital and its consequences on company performance. Journal of Intellectual capital 11(1): 39-60.

Zehri, C., A. AbdelBaki and N. BouAbdellah. 2012. How intellectual capital affects a firm’s performance.

Australian Journal of Business and Management Research 2(8): 24-31.

Zona, F., A. Zattoni and A. Minichilli, A. 2013. A contingency model of boards of directors and firm innovation: the moderating role of firm size. British Journal of Management 24: 299-315.

International Food and Agribusiness Management Review

316

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2017.0045 Received: 17 May 2017 / Accepted: 7 September 2017

Cyber security on the farm: an assessment of cyber security practices in the United States agriculture industry RESEARCH ARTICLE Andrew Geila, Glen Sagersb, Aslihan D. Spauldingc, and James R. Wolf aSenior

d

Security Analyst, Compeer Financial, 2000 Jacobssen Drive, Normal, IL 61761,USA

bAssistant

Director and Professor of Security, and dProfessor of Information Systems, School of Information Technology, Illinois State University, P.O. Box 5150, Normal, IL 61790-5150, USA cProfessor

of Agribusiness, Department of Agriculture, Illinois State University, P.O. Box 5020, Normal, IL 61790-5020, USA

Abstract The goal of this study was to survey farmers and agribusiness owners about their perceptions of cyber security, and how age, gender, and education might affect those perceptions. Using the Health Belief Model as a framework, the survey measured the constructs of perceived susceptibility, severity, benefits, barriers, self-efficacy and cues to action. In addition to the framework, levels of previous cyber-crime victimization and technology implementation were measured. The results of this survey demonstrated that perceived susceptibility to cyber-attacks and the perceived benefits of protective technology are related to an individualâ&#x20AC;&#x2122;s choice to implement cyber security technology. Over half of the respondents had been victims of a computer security incident, demonstrating that even individuals working in agriculture can be impacted by computer crime incidents. This project deepens the understanding of how individuals react to known threats, and what motivates them to adopt protection technologies.

Keywords: cyber security, health belief model, agribusinesses threat perceptions, computer self-efficacy JEL code: Q16, O33, D83, I12 Corresponding author: jrwolf@ilstu.edu

Š 2017 Geil et al.

317

Geil et al.

Volume 21, Issue 3, 2018

1. Introduction The last place the average American might expect to observe advanced computer technology being used is in the tilled farm fields, but nothing could be further from the truth. According to the 2013 USDA Farm Computer Usage and Ownership report, 67% of farms in the United States have access to the Internet. Farmers and other agricultural operators are applying technology more than they ever have before (USDA, 2013).

While technological advances have created new opportunities, enhanced productivity, and promoted connectivity, they have also introduced new threats. Cyber threats currently pose some of the greatest threats to the United States economy. It is important to recognize that protecting the information technology assets of the agriculture industry is just as important as protecting those of our nation’s banks, hospitals, and retail industries.

Numerous incidents of cyber-attacks against individual and organizations occur on a daily basis. The U.S. Secretary of Defense Leon Panetta noted during a speech, ‘We are literally the target of thousands of cyber-attacks every day – every day!’ (Orr, 2013). Panda Security Labs (2013) revealed that nearly a third of all computers scanned around the world are infected with some kind of malware. Trojans were the most prevalent malware, followed by worms, viruses, and spyware. Some of the highest malware infection rates were found in China (59.36%), Turkey (46.58%), and Peru (42.55%). The United States was found to have a moderate infection rate (30.58%). As many anti-virus scanners have low detection rates, it is possible that actual infection rates are even higher.

As agriculture grows more connected to the global Internet, it has become a target of malicious actors.

According to the Verizon (2013) Data Breach Investigation Report, most of the known cyber-attacks against businesses occurred against organizations with less than one hundred employees. Of the 621 documented breaches listed in the report, eleven of them occurred against agricultural organizations (Verizon, 2013). It is imperative for agricultural organizations, businesses, and individuals working in the field to be aware of the potential threats against them.

Farmers, agribusiness owners, and other individuals employed or associated with agriculture provide a unique sampling population. The line between business and personal use of a computer may be blurred. Like other small business industries, the home office may also serve as the office for the farm. The same computer used for personal matters, social networking, and gaming, might also be used to do farm taxes, log onto the local farm’s co-op website, or complete USDA eForms.

Using the Health Belief Model (HBM), this study examined the views and perceptions of cyber security by individuals employed in the agriculture industry. Examining the self-efficacy of individuals involved in agriculture suggests the need for cyber security education. Agribusiness owners and farm operators do not need to be certified or educated to the same degree as information technology experts, no more than a nurse or doctor in a hospital needs to be. However, individuals and small business owners operating in a globally connected environment need to be aware of the cyber threats and risks that their operations face.

2. Literature review As technology continues to evolve at a rapid pace, farmers and agribusinesses face constant adoption choices.

It is important to examine the types of technology in use and their implementation in the industry beyond the personal computer.

International Food and Agribusiness Management Review

318

Geil et al.

Volume 21, Issue 3, 2018

2.1. Health care models In many ways, choosing to use computer security technology parallels the choice to obtain a vaccine to guard against a disease. The individual must weigh the value of the protective measure against costs, barriers, and benefits. Specific to the computer security, the individual’s self-efficacy also must be weighed in the choice.

Fear appeal models, which examine both the threat and self-efficacy, are more appropriate for examining computer security technology adoption. Two different models, the Protection Motivation Theory (PMT) and HBM, are discussed below. Each of these, although utilized mainly in the health care field, can be applied to the information security research.

■■ Protection Motivation Theory Rogers (1975) proposed the PMT to explain the processes involved with coping with a threat. The PMT explains adaptive and maladaptive coping with a health threat as a result of threat appraisal. It was later revised in 1985 to allow the theory to encompass persuasive communication, focusing on the cognitive processes that facilitate behavior change (Rogers, 1985).

Rogers stated four components of a fear appeal to which an audience would respond. Perceived susceptibility is the individual’s estimation of the probability with which they will contract a disease. Perceived severity is the individual’s estimation of the effect that the disease would have on them if they were to contract it.

Response efficacy is the degree to which an individual believes a preventive method or treatment will avert the threat. Self-efficacy, which was added by Maddux and Rogers (1983), is the individual’s belief in his/ her own ability to complete the treatment successfully.

The PMT has been widely used to explain the factors that influence and predict health behaviors, including adherence to prescribed medical treatments (Flynn et al., 1995; Searle et al., 2000), genetic testing (Helmes, 2002), skin cancer and tanning (Jones and Leary, 1994; McClendon and Prentice-Dunn, 2001), alcohol consumption (Murgraff et al., 1999), and smoking (MacDonell et al., 2013).

The PMT has been applied to several research projects on computer security technology, and it has been used in several studies on users’ intentions to adopt security software to protect against the threat of spyware (Chenoweth et al., 2009; Johnston and Warkentin, 2010). Vance et al. (2012) used the PMT as a framework to study habitual information systems security compliance within a Finnish municipal organization. Within the field of academia, the PMT has been applied to study the adoption of anti-plagiarism software (Lee, 2011).

■■ Health Belief Model The HBM was originally created to predict the behaviors of individuals related to their personal health activities. Created by Rosenstock (1966), the HBM argues that the belief in a threat, combined with the belief in the effectiveness of a protective behavior, predicts the likelihood of adopting that behavior. Originally, the HBM was developed in response to the failure of a tuberculosis health screening program. The researchers wanted to understand the factors that influence individuals’ choices to reject the screening. Further studies using the HBM model have attempted to explain the rejection of vaccines, elective surgery, and other medical treatments. In summary, the HBM proposes that individuals will accept a medical treatment if they believe they are susceptible to a disease, believe that the treatment will effectively prevent the threat, and that barriers to successfully completing the treatment are minimal.

The perceived susceptibility construct examines the individuals’ beliefs that a threat can affect them. In a health care context, this might be the individuals’ belief that they will contract a disease or virus. In the context of computer security, perceived susceptibility refers to the likelihood of individuals to believe that their computers can become infected with a computer virus or be ‘hacked.’ Several factors might affect a

International Food and Agribusiness Management Review

319

Geil et al.

Volume 21, Issue 3, 2018

respondent’s level, such as previous incidents of victimization, knowledge of other individuals’ victimization, or previous education on the subject.

In addition to the likelihood of an incident being perceived as important, the degree of influence is also critically important. Perceived severity examines individuals’ beliefs that event would have an effect on them. In a security context, this would refer to a perceived effect of a cyber-attack on an individual or his/ her business. If the perceived effect of an event is believed to be low, it is likely that an individual would take few, if any steps to prevent it from occurring.

The perceived benefits construct examines the role of the individual’s perception of the usefulness or utilitarian value of a new behavior or technology in decreasing the risk of an event. In health care, individuals are more likely to adopt a behavior (e.g. take a vaccine) if they believe that it will lessen the likelihood of them contracting the disease. In the context of security behavior, individuals would be more likely to install anti-virus, patch software, or take cyber-security training if they believe those measures would better protect them from cyber-attacks.

The perceived barriers construct examines the individual’s perception of the obstacles or obstructions preventing the adoption of a new behavior. Out of the original four HBM constructs, the perceived barriers may be the greatest factor in determining behavior change (Janz and Becker, 1984). Technology is a rapidly changing field, and individuals and organizations may not have the time, effort, or financial resources to invest in continually evolving security.

The HBM was modified in 1988 to include the self-efficacy construct (Rosenstock et al., 1988). Selfefficacy was defined by Bandura (1977) as the ‘personal judgments of one’s capabilities to organize and execute courses of action to attain designated goals’. Zimmerman (2000: 83) noted: ‘self-efficacy measures focus on performance capabilities rather than on personal qualities, such as one’s physical or psychological characteristics. The respondents judge their capabilities to fulfill given task demands, such as solving fraction problems in arithmetic, not who they are personally or how they feel about themselves in general’.

In essence, self-efficacy examines the individuals’ beliefs that they are able to make a decision regarding certain events or topics.

The cues-to-action construct assumes that previous events, interactions with other people, and other activities influence people’s behavior and motivate them to change their behavior. In health care, examples would include illness of family members, media reporting, signs of a disease outbreak, or advice from health care practitioners. Applied to the world of cyber security, examples of cues might include malware infection, media reporting of significant cyber-attacks, knowledge of attacks against one’s own industry, security notification pop-ups in a browser (Whalen and Inkpen, 2005), or even friends and colleagues having experienced a recent cyber-security incident.

Socio-demographic variables, including age, income, gender, race, and others, affect the core constructs of the HBM. Other variables, such as knowledge of the threat, prior threat interaction, and education levels, for instance, could influence the individuals’ responses to a potential threat.

The HBM has some limitations that could limit its utility in information systems research (or health research, for that matter). A core assumption of the HBM is that all individuals have the same of access to equal information about the disease. In reality, individual education and experiences can have profound effects on an individual’s understanding of a disease.

The HBM does not consider environmental or economic factors that prohibit or promote the recommended action. For example, although an individual might feel that he/she is highly susceptible to a disease, that vaccine is highly effective, and that he/she would obtain the vaccine if he/she saw peers contracting the disease, the cost of vaccine might prevent him/her from obtaining the vaccine. Social pressure might also International Food and Agribusiness Management Review

320

Geil et al.

Volume 21, Issue 3, 2018

affect the adoption of a behavior; an individual might avoid exercising in the gym because he/she is shy or nervous around other people. In the context of information systems research, although an individual might believe they are highly vulnerable to computer security threats, they may not be able to purchase protective technology due to high cost.

Finally, the HBM does not consider individuals’ behaviors that occur for reasons unrelated to protection.

For example, an individual might wear a seat belt while operating a car only to comply with the state law rather than out of concern for their health and well-being. Other activities might occur for aesthetic reasons, such as exercising for appearance or social interaction rather than health reasons.

■■ Health Belief Model based information security research Two previous studies have utilized the HBM to examine individuals’ adoption of security technology. The first was completed by Ng, Kankanhalli, and Xu (2009). In their study, the authors believed that most models used to study technology adoption focused primarily on the tangible benefits. Consequently, the authors opted to use the HBM to examine the adoption of technology to prevent a negative outcome rather than attain a tangible positive benefit. The sample in their study comprised two classes of IT students from a university and employees of three IT-related firms. Their study focused on email safety and organizational security awareness.

The study revealed that perceived susceptibility, perceived benefits, and self-efficacy significantly affected email security related behavior. From a theoretical standpoint, this study demonstrated the success in using a health based model to explain computer security behavior. Practically, the study demonstrated that if a person’s perceived susceptibility is higher, he/she is more likely to engage in good computer security behavior. Ng et al. (2009: 823) noted, ‘the importance of perceived severity (as a moderator), perceived susceptibility and perceived benefits instructs us on how to design the content for organizational security awareness messages’. The authors also noted that future research should focus on individuals who utilize IT resources but do not use them as the core of their business.

Claar (2011) completed the second information security study utilizing the HBM. As part of his doctoral dissertation, Claar explored the security habits of home computer users using the HBM as a model, noting ‘striking similarities in the beliefs and perceptions in protecting one’s health and in protecting one’s computer from infection and attack.’

Claar’s (2011) population of interest included all home computer users who were responsible for their home computer security. Snowball sampling was used to recruit the participants for this study. The first group of respondents was a group of undergraduate students at a university. In addition, several Google news groups were chosen randomly to advertise the survey and recruit respondents. Because snowball sampling was implemented, it was unknown how many potential respondents who received the invitation to take the survey declined to take it. Ultimately, the project recruited 186 responses for the analysis.

The results of Claar’s study revealed that the perceptions of vulnerability to an attack and prior experience with security incidents were the most significant contributors to the use of computer security (Claar, 2011).

Perceived barriers to the implementation of computer security technology and self-efficacy were also found to influence computer security usage. In essence, a high perceived vulnerability to cyber-attack, the belief that security technology was not obstructive, and the individuals’ self-confidence to implement computer security technology were significant factors in this study.

Both Ng and Claar’s studies used respondents who likely had received some level of IT-related security awareness training before participating in the research. For future research, Claar sought to have his research applied to non-technology savvy individuals (2011). Thus, this model framework for cyber security will be implemented for the first time in an industry that does not focus on IT, let alone IT security.

International Food and Agribusiness Management Review

321

Geil et al.

Volume 21, Issue 3, 2018

While both the PMT and the HBM offer frameworks for studying security technology adoption, the HBM appears to be more appropriate. The PMT focuses more on the behavioral response based on fear appeals, whereas the HBM is concerned with behavioral abilities. This study concentrated on the current perceptions and deployment of security technology, rather than fear appeals that might motivate the respondents to implement the protective technology. Using the HBM will better allow us to determine the behavioral activities, which affect the perceptions or use of protective computer technology.

2.2. Hypotheses The purpose of the project was to examine the implementation of cyber security and perceptions of threats, vulnerabilities, and self-efficacy among farmers, agribusinesses owners, agricultural industry employees, and other individuals involved in agriculture. The results provided descriptive information about the use of technology, previous computer crime victimization, credit and debit card use, and interest in cyber security education.

The survey instrument was developed based on the tool created by Claar (2011). Copyright permission to use a modified version of the instrument for the use in this project was obtained from the original author (C. Claar, personal communication, November 14, 2013). Although the original instrument questions targeted the generic individual, some items were modified to make them relevant to the individual farmer or agribusiness owner.

Perceived susceptibility refers to the individuals’ beliefs that they are vulnerable to a computer security incident. When individuals believe that their computer is likely to be a victim of a computer security incident, they are more likely to implement a security technology to prevent it. As such, the following hypothesis was established: H1: perceived susceptibility to computer security incidents is positively related to computer security usage.

The perceived severity construct is the individual’s belief that if a computer security incident were to occur, the event would have a negative effect on his/her lifestyle and financial health, would disrupt business activity, and the like. If a user believes that the loss of computing functionality due computer security incident is high, he/she would be more likely to implement technology to protect his/her computer. As such, the following hypothesis was established: H2: perceived severity of computer security incidents is positively related to computer security usage.

In the HBM, perceived benefits referred to an individual’s perceptions of the effectiveness of an action (like a receiving a vaccine) to reduce the probability of contracting a disease. Similar to computer security technology, anti-virus, anti-spyware, and other network protection technologies can reduce the risk of a computer becoming infected with malware. When an individual believes those technologies are beneficial to his/her computer’s health, he/she would be more likely to implement them. The following hypothesis follows: H3: perceived benefits of security technology are positively related to computer security usage.

Although an individual might feel an action is beneficial at reducing a threat, certain mitigating activities might be unpleasant, too costly, or inconvenient to implement. Computer security software often inconveniences the users, causes difficulty in completing tasks, and obstructs productivity while trying to secure a system.

If an individual feels that a protective technology is too obstructive for productivity, they are less likely to implement it. The following hypothesis follows: H4: perceived barriers to implementing computer security technology are negatively related to computer security usage.

International Food and Agribusiness Management Review

322

Geil et al.

Volume 21, Issue 3, 2018

Self-efficacy refers to the individualâ&#x20AC;&#x2122;s belief in his/her ability to perform an action. Individuals with greater confidence in their ability to perform an action are more likely to initiate and engage in that action. Information security self-efficacy refers to the individualâ&#x20AC;&#x2122;s ability to select, install, configure, and operate security technology, such as anti-virus, anti-spyware, and network firewalls on his/her computer. As such, the following hypothesis follows: H5: information security self-efficacy is positively related to computer security usage.

A cue to action refers to the knowledge of another individual, or information obtained from a reliable source, about the spread of computer viruses, computer vulnerabilities, or suspicious activity by the userâ&#x20AC;&#x2122;s computer. For example, an individual might be more likely to install anti-virus software if he/she sees news reports about a computer virus spreading across the internet, just like individuals might also be more likely to engage in a preventive activity if their peers, neighbors, or other affiliates are affected by a disease. As such, the following hypothesis follows: H6: cues to action are positively related to computer security usage.

With the ability to compare demographics of the respondents, three further hypotheses related to the age, gender, and education of the respondents will be tested against each of the primary hypotheses. Previous information systems research has demonstrated that age, gender, and other demographic moderators can have an effect on technology adoption (Liang and Xue, 2009). Thus, the moderating effects of age, gender, and education will be examined as follows: H7a: age significantly moderates the relationship between Perceived Susceptibility and computer security usage.

H7b: age significantly moderates the relationship between Perceived Severity and computer security usage.

H7c: age significantly moderates the relationship between Perceived Benefits and computer security usage.

H7d: age significantly moderates the relationship between Perceived Barriers and computer security usage.

H7e: age significantly moderates the relationship between Information Security Self-efficacy and computer security usage.

H8a: gender significantly moderates the relationship between Perceived Susceptibility and computer security usage.

H8b: gender significantly moderates the relationship between Perceived Severity and computer security usage.

H8c: gender significantly moderates the relationship between Perceived Benefits and computer security usage.

H8d: gender significantly moderates the relationship between Perceived Barriers and computer security usage.

H8e: gender significantly moderates the relationship between Information Security Self-efficacy and computer security usage.

H9a: education significantly moderates the relationship between Perceived Susceptibility and computer security usage.

H9b: education significantly moderates the relationship between Perceived Severity and computer security usage.

H9c: education significantly moderates the relationship between Perceived Benefits and computer security usage.

H9d: education significantly moderates the relationship between Perceived Barriers and computer security usage.

H9e: education significantly moderates the relationship between Information Security Self-efficacy and computer security usage. The hypotheses are shown in Figure 1.

International Food and Agribusiness Management Review

323

Geil et al.

Volume 21, Issue 3, 2018

Perceived susceptibility Perceived severity Perceived benefits

Cyber security behavior

Self-efficacy

Cues to action

Gender

Age

Education

Figure 1. Health Belief Model based survey model.

3. Material and methods The study population comprised farmers, producers, agribusiness owners, or other individuals who have participated in a USDA Farm Service Agency (FSA) program in three selected Illinois counties. Any farm operator, owner, or agribusiness owner in those counties who had participated in an FSA program and registered in the FSA database had a chance of being selected for this survey. Utilizing the Freedom of Information Act, a list of all participants in the three selected counties was obtained. The three counties were selected for their number one rankings for each of the top three agricultural commodities in Illinois: cash crops, cattle, and hogs. The statistics and rankings were obtained from the National Agricultural Statistical Service (USDA, 2007).

3.1. Data The sample from each county was drawn by random sampling. Because the population roster provided by the FSA contained many individuals in the agriculture industry in a specific county, a true representative sample could not be drawn.

Overall, 1,800 respondents, 600 from each county, were randomly selected. The sample was then split into two groups, ‘A’ and ‘B’, with 300 participants from each county in each group. The first group received the paper copy of the survey to complete, with the option to complete the survey online. The second group received an invitation letter to complete the survey online.

McLean County is ranked first in the state for the value of sales in grains, oilseeds, dry beans, and peas.

Although McLean ranks number two in the state in total acreage for these crops, it ranks number one in the total cash sales from them (U.S. Department of Agriculture). Of the 1,513 respondents in the 2007 Agricultural Census, 812 of them identified their primary occupation as ‘farming’ while the remaining 701 selected ‘other.’ The average net cash income of an agriculture operation was $117,050. Twin cities of Bloomington-Normal and Illinois State University are also located in McLean County. This geographically largest county in the state has a population in excess of 153,000, most residing within the boundaries of the Bloomington-Normal metropolitan areas.

Jo Daviess County is located in northwestern Illinois and boasts the highest inventory of cattle in the state.

It borders the states of Iowa and Wisconsin, both of which are considered major agricultural producers

International Food and Agribusiness Management Review

324

Geil et al.

Volume 21, Issue 3, 2018

themselves. The county had 1,016 farms, as of 2007, averaging 267 acres1. Overall, 443 out of 1,016 Census respondents identified their primary income as ‘farming.’ The average net cash income of an agriculture operation in Jo Daviess County was $44,710.

De Kalb County in north central Illinois ranks number one in hog inventory. The total head count of hogs and pigs, according to the Census, was 225,397. As of 2007, the county had 930 farms, averaging 440 acres (U.S. Department of Agriculture). The average net cash income for an agriculture operation in De Kalb County was $84,017.

These three counties represent three of the top agriculture industries in Illinois. Jo Daviess has the highest head count of ducks within the state while McLean County ranks seventh in the state in sheep and lamb count. Additionally, many silos, co-ops, grain storage sites, and other agribusinesses can be found throughout each of these counties.

3.2. Survey instrument The paper survey was mailed via the U.S. Postal system directly to each farm operator or owner, using his/her address registered with the FSA. The consent from the respondents was obtained utilizing an informational cover letter. The cover letter explained the purpose of the survey and provided contact information for the project. Implied consent was obtained from the respondents completing and returning the survey. This survey was not anonymous but confidentially was guaranteed to respondents in the cover letter. The survey included a pre-paid return postage envelope to encourage the return of the surveys to the researcher.

The respondents receiving the invitation to participate online were provided a similar cover letter, with instructions on how to access the website. The online survey matched the paper survey, allowing equal comparison between answers provided. The online site allowed respondents to complete the same survey using radio buttons and drop boxes. A link on the index page of the site provided a connection to the survey website hosted at SurveyShare.com.

The online survey required respondents to provide their survey identification number before being permitted to proceed. The survey identification number was included in the invitation. A review of all submissions to the online site revealed no attempts by intruders or cyber vandals to modify the survey results. A review of IP access logs to the website revealed no traffic above the response rates of the survey.

The first section of the survey instrument requested basic demographic information from the respondent.

Demographic variables include age, gender, length of time in agriculture, role of the respondents, income levels, and the types of agriculture in which they were involved.

The respondents were provided a series of security scenarios representing the current threats that internet users face (Table 1). The original scenarios in Claar’s instrument were derived from research by Boss (2007), who formulated them using the 2001 U.S. Department of Justice National Crime Victimization Survey.

Each item ‘assesses the degree to which individuals feel that is likely they will experience the scenario, and assesses the impact to them were it to happen’ (Boss, 2007: 75).

Based on the scenarios provided, the respondents were asked to respond to each of the first three constructs (perceived susceptibility, perceived severity, and perceived benefits) relative to each scenario. Perceived susceptibility was measured on a five point Likert scale ranging from ‘very likely’ to ‘very unlikely’. Perceived severity was measured on a five point Likert scale ranging from ‘no effect’ to ‘major effect’. Perceived benefits were measured on a five point Likert scale ranging from ‘poor’ to ‘excellent’.

1

1 acre = 0.40 ha.

International Food and Agribusiness Management Review

325

Geil et al.

Volume 21, Issue 3, 2018

Table 1. Security scenarios.

Scenario My computer system becoming corrupted by a virus.

My computer being taken over by a hacker.

My files becoming corrupted by a computer virus.

My personal identity being stolen (credit cards, social security number, etc.).

My agribusiness or farm identity being stolen (loan fraud, etc.).

Not being able to access the Internet due to a computer virus.

My computer becoming infected with a virus by visiting a website.

‘Perceived Barriers’ refer to the cost and difficulty in implementing computer security technology. The construct consisted of four questions measured on a Likert scale ranging from ‘highly agree’ to ‘highly disagree’ (BAR1, ‘the expense of computer security technology (anti-virus, firewalls, etc.) is a concern for me.’; BAR2, ‘the use of computer security software (anti-virus, firewalls, etc.) would make day-to-day tasks more difficult’; BAR3, ‘implementing and using computer security technology is time consuming’; BAR4, ‘implementing security technology on the computers I use would require an investment of effort other than time.’).

Information ‘Self-Efficacy’ refers to the individual’s belief that he/she can successfully implement security technology to protect his/her computer. The construct comprises four questions measured on a Likert scale ranging from ‘highly agree’ to ‘highly disagree’ (SEF1, ‘I can determine the appropriate computer security technology for my computer’; SEF2, ‘I can correctly install and manage computer security technology for my computer’; SEF3, ‘I know how to find information on how to respond to a computer security problem’; SEF4, ‘if my computer were to become infected by a virus, I would know how to fix it’).

The ‘Cues to Action’ construct refers to the previous experiences, triggers, or events that prompt a user to implement computer security technology. The construct comprises four questions measured on a Likert scale ranging from ‘highly agree’ to ‘highly disagree’ (CUE1, ‘if another agribusiness or farmer were to tell me of a recent experience with a computer virus, I would be more conscious of my computer’s chance of being attacked’; CUE2, ‘if my computer were to start behaving strangely, I would be concerned if I had been the victim of a computer virus’; CUE3, ‘if I saw a news report or read a newspaper or magazine about new computer security vulnerability, I would be more concerned about my computer’s chances of being attacked’; CUE4, ‘if I received an email from the maker of my computer’s operating system about new security vulnerability, I would be more concerned about my computer’s chances of being attacked’).

‘Computer Security Usage’ was the dependent variable in this research. Each respondent was asked about his/her use of after-market anti-virus, anti-spyware, and network firewall protection. If an individual has purchased after-market software, he/she might be concerned about computer security to the point they have made an investment in it. The measure of computer security usage consists of three questions measured on a binary scale, with ‘yes’ response being worth 1 point, ‘no’ and ‘don’t know’ being worth 0 points. If a user has implemented all three technologies, he/she received a ‘high’ rating; two technologies were listed ‘medium’; one technology was ‘low’; and no technology received a score of ‘none.’ This created a scale against which independent variables could be compared. Two questions were designed to elicit information about previous computer security incidents that had affected the respondent. The respondents could respond yes, no, or that they didn’t know. Those who did respond yes were asked to provide information on how long ago the incident had occurred.

International Food and Agribusiness Management Review

326

Geil et al.

Volume 21, Issue 3, 2018

3.3 Data analysis 136 mail surveys were returned, and thirty-two were completed electronically, for a total of 168 responses.

After removing unusable responses from the total due to missing data, blank returns, and other incomplete surveys, 138 surveys were deemed valid. This resulted in a valid response rate of 8.1% (Table 2). The largest number of returns was from DeKalb County (N=54), followed by McLean County (N=45) and Jo Daviess County (n=39).

The first mailing of surveys and invitation letters was sent on January 9th, 2014. After four weeks, a total of one hundred and two mailed surveys were returned. The website received 30 responses.

Table 2. Demographic data.

Variable Gender Male Female Age 18-20 21-30 31-40 41-50 51-60 61-70 71-80 Education Completed high school Some college Completed two-year degree Completed four-year degree Some graduate work Graduate degree Doctoral degree Professional degree Role in agriculture Farm operator Farm employee Landlord (non-farming) Agribusiness â&#x20AC;&#x201C; owner Agribusiness â&#x20AC;&#x201C; employee Gross income Less than $50,000 $50,001 to $100,000 $100,001 to $250,000 $250,001 to $500,000 $500,001 to $750,000 More than $750,000

Frequency (n)

Percentage

118 20

85.5 14.5

2 5 18 39 41 19 11

1.4 3.6 13.0 28.3 29.7 15.9 8.0

26 28 10 42 11 15 3 3

18.8 20.3 7.2 30.4 8.0 10.9 2.2 2.2

98 1 30 7 2

71.0 0.7 21.7 5.1 1.4

43 22 21 19 11 22

31.2 15.9 15.2 13.8 8.0 15.9

International Food and Agribusiness Management Review

327

Geil et al.

Volume 21, Issue 3, 2018

■■ Sample characteristics Half of the respondents reported that they had been previously affected by a computer security incident (Table 3). Of those who had been affected, almost half of the incidents had occurred within the past year (46.4%), or the respondent could not remember when the incident occurred (39.1%). There was a significant effect by previous computer incident victimization on computer security usage (F=4.437, P=0.014).

■■ Construct reliability analysis Cronbach’s alpha is a measure of internal consistency to how closely related a set of items are as a group. A high value is considered above 0.9, while in most social science research, above 0.7 is considered acceptable.

A reliability analysis for all of the model constructs was completed, revealing the Cronbach’s alpha level was above 0.7 (Table 4). The scale of reliability did not improve with the removal of any of the construct variables.

■■ Independent variables The items for each of the constructs were combined into a single factor score by means of aggregation. The range of the data covered the entire range of possibilities, with the exception of Cues to Action. The highest aggregated response for Cues to Action was 4.5. The skewedness for these constructs ranged from -0.705 to 0.049, indicating an acceptable distribution of values. The kurtosis of the values ranged from -0.668 to 2.840. The Cues to Action variable was the only construct outside of the acceptable range of ±2. Because the exceptionally high value is positive, this demonstrates that the distribution of values were located in the tail of distribution, rather than around the mean.

Table 3. Previous computer security incidents.

Previously affected by a computer security incident Yes No Don’t know If yes, how long ago?1 Year Month Week Don’t remember 1

Frequency

Percentage

69 58 11

50.4 42.3 7.2

32 7 3 27

46.4 10.1 4.3 39.1

n=69.

Table 4. Cronbach’s alpha scores.

Construct

Cronbach’s alpha

n of items

Perceived susceptibility Perceived severity Perceived benefits Perceived barriers Information self-efficacy Cues to action

0.940 0.933 0.951 0.794 0.869 0.786

7 7 7 4 4 4

International Food and Agribusiness Management Review

328

Geil et al.

Volume 21, Issue 3, 2018

■■ Dependent variables Computer security usage was the dependent variable for this study. Respondents were asked to provide information on their use of anti-virus, anti-spyware, and firewall technology for their computers. A majority of the respondents used anti-virus, while rates of anti-spyware and firewall usage were much less (Table 5).

The dependent variable was determined by assigning each respondent a score reflective of their overall implementation of security technology. The range of scores was from none to high, with a standard deviation of 0.880, a skewedness of 0.117, and a kurtosis of -0.659. Only 12.3% of the respondents indicated they do not use computer security technology.

4. Results To test the hypotheses established for this study, ordinal logit (PLUM) regression was conducted using SPSS (SPSS version 22.0, IBM Corporation, Armonk, NY, USA) (Table 6). This allowed the ordinal dependent variable, computer security usage, to be regressed on the independent variables of perceived susceptibility, severity, benefits, barriers, self-efficacy, and cues to action (Model 1). A second step was taken to regress the main modifying variables (age, gender, education), and the two way interactions between those moderating variables of susceptibility, severity, benefits, barriers, self-efficacy, and cues to action against the dependent variable (Model 2).

For Model 1, the research model has a Chi-Square factor of 13.555 and a significance of 0.038. The significance level below 0.05 demonstrates the model is significant with the predictors. When the moderating variables are introduced in Model 2, the model fitting information changed to a Chi-Square of 40.338 and a significance level of 0.048, thus indicating the model was still significant.

In Model 1 regression, the main effects of susceptibility, severity, benefits, barriers, self-efficacy, and cues to action, were tested (H1-H6). H1, which predicted that perceived susceptibility to computer security incidents would be positively related to computer security usage was supported (μ=0.367, P=0.033). H2, which predicted perceived severity of computer security incidents would be positively related to computer security usage, was not supported (μ=-0.050, P=0.770, n.s.). H3, which predicted the perceived benefits of computer security technology was positively related to computer security usage, was supported (μ=0.577, P=0.008). H4, which predicted the perceived barriers of implementing computer security technology would be negatively related to computer security usage was not supported (μ=-0.305, P=0.176, n.s.). H5, which Table 5. Dependent variable responses.

Variable

Yes

No/don’t know

Anti-virus Anti-spyware Firewall

112 55 20

26 83 118

Table 6. Regression information.1 Model 1 Model 2

-2 log likelihood

Chi-square

df

Sig.

333.736 306.753

13.355 40.338

6 27

0.038* 0.048*

1

Model 1 = regression of ordinal dependent variable on the independent variables of perceived susceptibility, severity, benefits, barriers, self-efficacy, and cues to action; Model 2 = regression of the main modifying variables and the two way interactions between moderating variables against the dependent variable; df = degrees of freedom; * = values differ significantly at the 0.05 level.

International Food and Agribusiness Management Review

329

Geil et al.

Volume 21, Issue 3, 2018

predicted information security self-efficacy would be positively related to computer security usage, was not supported (μ=0.241, P=0.169, n.s.). H6, which predicted cues to action would be positively related to computer security usage, was not supported (μ=0.075, P=0.786, n.s.).

In Model 2, the main effects of susceptibility, severity, benefits, barriers, self-efficacy, and cues to action, were tested, along with the modifying factors age and education, along with the interactions between those factors and the main factors. Due to the low number of female respondents, an accurate test of the modifying effect could not be completed.

Hypotheses H7a-f, which predicted that age would have a significant moderating effect with susceptibility, severity, benefits, barriers, self-efficacy, and cues to action, only H7a was supported (H7a: μ=0.034, P=0.040; H7b: μ=0.017, P=0.365, n.s.; H7c: μ=-0.009, P=0.661, n.s.; H7d: μ=0.003, P=0.908, n.s.; H7e: μ=-0.017, P=0.291, n.s.; H7f: μ=0.024, P=0.359, n.s.). The main effect of age on computer security usage was also not significant (μ=-0.199, P=0.181).

Hypotheses H9a-f, which predicted that education would have a significant moderating effect with susceptibility, severity, benefits, barriers, self-efficacy, and cues to action were not supported (H9a: μ=-0.214, P=0.065, n.s.; H9b: μ=-0.030, P=0.773, n.s.; H9c: μ=-0.113, P=0.399, n.s.; H9d: μ=0.200, P=0.147, n.s.; H9e: μ=-0.105, P=0.324, n.s.; H9f: μ=-0.245, P=0.191, n.s.). The main effect of education on computer security usage was also not significant (μ=1.852, P=0.053, n.s.).

Although not part of the original proposed model, two other major factors were also tested for their overall effect as modifying variables on the dependent variable. The effect of gross income was found to be notsignificant (μ=0.103, P=0.213). Total farm size (acres) was also determined to be not-significant (μ=0.041, P=0.232).

Several other significance tests were run to determine if there were correlations or differences amongst groups within the respondent pool. There was a significant correlation between the level of computer security usage and age, but the association was weak (X2=0.288, P=0.048) (Table 7). Levels of computer security usage did not differ for respondents with farms above the mean farm size (938), compared to those below it (F=0.688, P=0.408). The role of the respondent did not have a significant effect on computer security usage. (F=1.402, P=0.238).

5. Discussion Using the HBM, the project was able to evaluate the cyber security perceptions of individuals involved in agriculture. From descriptive measures, the means of perceived severity indicate that there is a moderate to high level of concern that a computer security incident would impact the individual. The benefits of computer security technology were viewed as moderately high, indicating that anti-virus and other technologies are viewed as having some value in protecting computers. Perceived barriers and perceived susceptibility were rated at moderate levels, demonstrating the implementation of computer security technology was not viewed as overly obstructive, but nor did the respondents consider themselves as overly susceptible to a cyber-attack.

Finally, self-efficacy averaged below the median score, indicative that respondents might not feel comfortable with selecting, configuring, and managing computer security technology.

The model revealed perceived susceptibility and perceived benefits are significant factors for members of the agricultural community when it comes to implementing computer security technology. Only perceived susceptibility was a modifying factor on age. Surprisingly, education as a modifying variable had no effect on the choice to implement protective measures.

Neither the size of the farm, nor the role of the respondent, appears to have an effect on the adoption of computer security technology. Significant differences in computer security usage were observed between International Food and Agribusiness Management Review

330

Geil et al.

Volume 21, Issue 3, 2018

Table 7. Regression parameters.1 Est. (Îź)

Model 1 Susceptibility Severity Benefits Barriers Self-efficacy Cue to action Model 2 Susceptibility Severity Benefits Barriers Self-efficacy Cue to Action Age Education Total farm size (acres) Gross income Susceptibility * Age Severity * Age Benefits * Age Barriers * Age Self-efficacy * Age Cue to Action * Age Susceptibility * Education Severity * Education Benefits * Education Barriers * Education Self-efficacy * Education Cue to action * Education 1*

Std. error Wald

df

Sig.

Interval Lower bound

Upper bound

0.367 -0.050 0.577 -0.305 0.241 0.075

0.172 0.172 0.216 0.225 0.175 0.277

4.544 0.085 7.147 1.831 1.896 0.073

1 1 1 1 1 1

0.033* 0.770 0.008* 0.176 0.169 0.786

-0.705 -0.388 0.154 -0.746 -0.102 -0.467

-0.030 0.288 1.000 0.137 0.584 0.617

-0.083 -0.794 2.168 -2.196 2.022 0.372 -0.199 1.852 0.041 0.103 0.034 0.017 -0.009 0.003 -0.017 0.024 -0.214 -0.030 -0.113 0.200 -0.105 -0.245

1.278 1.408 1.515 1.670 1.263 1.965 0.149 0.958 0.000 -0.128 0.017 0.018 0.020 0.023 0.016 0.026 0.116 0.105 0.134 0.138 0.106 0.187

0.004 0.318 2.047 1.729 2.560 0.036 1.788 3.739 2.947 1.552 4.212 0.822 0.192 0.013 1.115 0.841 3.402 0.083 0.712 2.102 0.972 1.709

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

0.948 0.573 0.152 0.189 0.110 0.850 0.181 0.053 0.232 0.213 0.040* 0.365 0.661 0.908 0.291 0.359 0.065 0.773 0.399 0.147 0.324 0.191

2.423 1.966 5.138 1.077 4.498 4.223 0.093 3.729 -3.207 -0.330 0.066 0.052 0.030 0.047 0.015 0.075 0.013 0.175 0.149 0.470 0.104 0.122

2.423 1.966 5.138 1.077 4.498 4.223 0.093 3.729 1.214 0.074 0.066 0.052 0.030 0.047 0.015 0.075 0.013 0.175 0.149 0.470 0.104 0.122

= P<0.05; 1 acre = 0.40 ha.

those previously affected by a computer security incident compared those who had not. Individuals who had been previously impacted by an incident were more likely to have higher levels of computer security. This possibly indicates that many security installations are done for reactive purposes after an incident, rather than before it occurs.

This study was among the first to use the HBM to study computer security implementations. Ng (2009) reported similar results amongst perceived susceptibility and perceived benefits. Claar (2011) found that perceived susceptibility and previous experience with a computer security incident affected the use of computer security technology. These studies shared a common theme that the perception of the threat and the benefits of protective technology are the most significant factors contributing to computer security implementation.

Similar to both of the previous studies, this study found that cues to action variable was not a significant factor in computer usage.

International Food and Agribusiness Management Review

331

Geil et al.

Volume 21, Issue 3, 2018

In the context of computer security, this study has deepened the understanding of human behavior in the face of threats. The dynamics of the agricultural population allows researchers to examine the implementation of computer security in an area where personal and business activities may be closely intertwined. Other similar studies have focused either on the individual computer user in the home environment or on employees in an organizational setting but not on both concurrently.

5.1. Limitations and future work As with any study, this research had some limitations. The high non-response rate of the survey population may be due to many unknown factors. It will never be known how many of the respondents chose not to respond precisely because they do not have computer technology implemented. Other respondents may not have wished to disclose sensitive information about their security implementation.

The online nature of the study likely limited the number of respondents. Some respondents might not have been able to complete the study due to a lack of internet access, while others may not have felt comfortable completing a survey about computer security online. Other researchers have noted that e-mail solicitation of respondents is a very poor method of recruiting participants in security studies (Claar, 2011). Perhaps the best way to engage participants continues to be through physically mailed surveys or direct interaction.

Another limitation of this study is the self-reported nature of the dependent variable. The self-report bias might have motivated the respondents to report a higher level of usage than would be determined through direct observation, which would have affected the results. This project also made the assumption that the respondent understood the differences between anti-virus, anti-spyware, and firewall technologies.

Finally, the list of participants was obtained from a Federal agency. The frequency with which the FSA program updates, audits, or purges its participant list is unknown. The high rate of returned surveys due to incorrect mailing addresses indicates that the list is poorly maintained, or at the very least, not updated.

Using the HBM for information security research is a relatively new method of researching technology related behavior. While other studies have focused on traditionally IT-savvy populations, future research could be applied to industries not typically considered IT-centric. Even with agriculture, certain segments of the industry, i.e. aviation, marketing, and others, regularly use technology; thus, they could be studied more extensively. Beyond the scope of security technology, further research on the factors that affect technology adoption in agriculture should continue to be conducted to deepen the understanding of consumer behavior.

The dependent variable in this study was the use of anti-virus, anti-spyware, and firewall technology implementation. Future research could further expand upon security behavior, such as browsing suspicious websites, opening unsolicited email attachments, file sharing, and online piracy activity, to determine risky behavior. Investigating the level of engagement of these activities by respondents might further explain the constructs like perceived susceptibility and self-efficacy.

5.2. Practical implications The low rates of self-efficacy amongst the respondents indicate there are gaps in security knowledge in the agricultural community. Half of the respondents reported they had been affected by a computer security incident. Over half the respondents in this study indicated the desire to participate in some form of cyber security training. One respondent noted the need for a computer security standardized practices, which could be made to parallel other industries.

The practical conclusion to be drawn is that there exists a need for computer security education within the industry. An opportunity exists for leading organizations in the industry, such as Farm Bureaus, Associations,

International Food and Agribusiness Management Review

332

Geil et al.

Volume 21, Issue 3, 2018

or the FSA to offer an awareness level education programs, or liaison with organizations which do. Security awareness programs should train users on the purpose, function and basic methodology of computer security.

5.3 Conclusions American agriculture faces risks on a regular basis. While traditional threats of weather and market instability continue to exist, the cyber-attack threats are now imminent. The very technology that helps connects farmers and others in agriculture to the world risks opening the door to new threats. The threat of disruption to personal or business functions as the result of a cyber-attack is high. Information security for a farm or agribusiness should not be neglected, and technology alone cannot provide sufficient protection.

Agriculture is one of the core pillars of the American economy. Protecting those involved in agriculture from cyber-attack is critical to ensuring the continuity of business activities, personal livelihood, and the American way of life. Technology can greatly enhance productivity, but only if that technology is safe, secure, and protected. As a nation, we have a vested interest in protecting our nationâ&#x20AC;&#x2122;s farmers from risk in the fields or online.

References Bandura, A. 1977. Self-efficacy: toward a unifying theory of behavioral change. Psychological Review 84 (2): 191-215.

Boss, S. 2007. Control, perceived risk and information security precautions: external and internal motivations for security behavior. PhD dissertation, University of Pittsburgh, Pittsburgh, PA, USA. Chenoweth, T., R. Minch and T. Gattiker. 2009. Application of protection motivation theory to adoption of protective technologies. In: Proceedings of the 42nd Hawaii international conference on system sciences. IEEE Computer Society, Washington, DC, USA. Claar, C.L. 2011. The adoption of computer security: an analysis of home personal computer user behavior using the health belief model. PhD dissertation, Utah State University, Logan, UT, USA. Flynn, M.F., R.D. Lyman and S. Prentice-Dunn. 1995. Protection motivation theory and adherence to medical treatment regimens for muscular dystrophy. Journal of Social and Clinical Psychology 14(1): 61-75.

Helmes, A.W. 2002. Application of the protection motivation theory to genetic testing for breast cancer risk.

Preventive Medicine 35(5): 453-462.

Janz, N.K. and M.H. Becker. 1984. The health belief model: a decade later. Health Education Quarterly 11(1): 1-47.

Johnston, A.C. and M. Warkentin. 2010. Fear appeals and information security behaviors: an empirical study.

MIS Quarterly 34(3): 549-66.

Jones, J.L. and M.R. Leary. 1994. Effects of appearance-based admonitions against sun exposure on tanning intentions in young adults. Health Psychology 13(1): 86-90.

Lee, Y. 2011. Understanding anti-plagiarism software adoption: an extended protection motivation theory perspective. Decision Support Systems 50(2): 361-69.

Liang, H., and Y. Xue. 2009. Avoidance of information technology threats: a theoretical perspective. MIS Quarterly 33(1): 71-90.

MacDonell, K., X. Chen, Y. Yan, F. Li, J. Gong, H. Sun, L. Xioaming and B. Stanton. 2013. A protection motivation theory-based scale for tobacco research among Chinese youth. Journal of Addiction Research and Therapy 4: 154-170.

Maddux, J. E. and R.W. Rogers. 1983. Protection motivation and self-efficacy: a revised theory of fear appeals and attitude change. Journal of experimental social psychology 19(5): 469-479.

McClendon, B.T. and S. Prentice-Dunn. 2001. Reducing skin cancer risk: an intervention based on protection motivation theory. Journal of Health Psychology 6(3): 321-328.

Murgraff, V., D. White and K. Phillips. 1999. An application of protection motivation theory to riskier singleoccasion drinking. Psychology and Health 14(2): 339-350.

International Food and Agribusiness Management Review

333

Geil et al.

Volume 21, Issue 3, 2018

Ng, B.Y., A. Kankanhalli and Y.C. Xu. 2009. Studying usersâ&#x20AC;&#x2122; computer security behavior: a health belief perspective. Decision Support Systems 46(4): 815-825.

Orr, B. 2013. Pentagon expands cyber defense amid daily attacks. CBS News. Available at: http://tinyurl. com/yb4vvnnk.

Panda Security Labs. 2013. Annual Report PandaLabs. Available at: http://tinyurl.com/ycd5vh6j.

Rogers, R.W. 1975. A protection motivation theory of fear appeals and attitude change. Journal of Psychology 91: 93-114.

Rogers, R.W. 1985. Attitude change and information integration in fear appeals. Psychological Reports 56(1): 179-182.

Rosenstock, I.M. 1966. Why people use health services. The Milbank Memorial Fund Quarterly 44(3): 94-124.

Rosenstock, I.M., V.J. Strecher and M.H. Becker. 1988. Social learning theory and the health belief model.

Health Education and Behavior 15(2): 175-183.

Searle, A., K. Vedhara, P. Norman, A. Frost and R. Harrad. 2000. Compliance with eye patching in children and its psychosocial effects: a qualitative application of protection motivation theory. Psychology, Health and Medicine 5(1): 43-54.

United States Department of Agriculture. 2007. Census publications. Available at: http://tinyurl.com/y8t3pam6.

United States Department of Agriculture National Agricultural Statistics Service. 2013. Farm computer usage and ownership. Available at: http://tinyurl.com/y9pmfee4.

United States Department of Justice. 2002. Bureau of justice statistic national crime victimization survey.

Available at: http://tinyurl.com/ydy8c52u.

Vance, A., M. Siponen and S. Pahnila. 2012. Motivating IS security compliance: insights from habit and protection motivation theory. Information and Management 49(3-4): 190-198.

Verizon. 2013. 2013 Data breach investigations report. Available at: http://tinyurl.com/crnzu89.

Whalen, T. and K.M. Inkpen. 2005. Gathering evidence: use of visual security cues in web browsers. Available at: http://tinyurl.com/y6uprdbv.

Zimmerman, B.J. 2000. Self-efficacy: an essential motive to learn. Contemporary Educational Psychology 25(1): 82-91.

International Food and Agribusiness Management Review

334

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2017.0020 Received: 15 February 2017 / Accepted: 7 November 2017

The effects of international price volatility on farmer prices and marketing margins in cattle markets RESEARCH ARTICLE L. Emilio Morales Lecturer, UNE Business School, University of New England, Armidale, NSW 2351, Australia

Abstract This study examines the effects of export price volatility in cattle markets using panel data from twelve countries between 1970 and 2013. Fixed-effects models with Driscoll and Kraay standard errors were estimated to control for cross-sectional dependence. Results indicate that price transmission depends on prices previously paid to farmers, variations in export prices and volatility of export prices, which reduces farmer prices in developed countries and it increases them in developing countries. In contrast, marketing margins are reduced by contemporaneous export price volatility and are increased by previous volatility.

Exporters in developing countries take more time to transmit shocks in international prices, pay lower prices to farmers and absorb a bigger proportion of price fluctuations. These price transmission imperfections affect investments, technology adoption, production level and quality across the chain in developing countries, which negatively impact farmers, input and service providers, traders and other actors of the beef cattle chain.

Keywords: price volatility, vertical price transmission, marketing margins, cattle markets, panel crosssectional dependence JEL code: M21, Q13, Q18 Corresponding author: emilio.morales@une.edu.au

Š 2017 Morales

335

Morales

Volume 21, Issue 3, 2018

1. Introduction Volatile prices in agricultural markets and their transmission through agro-food chains have been a focus of interest and public concern. Large unanticipated variations in prices create uncertainty, increase the price risk for producers, traders, consumers and governments, and lead to sub-optimal decision making (Minot, 2014). According to the Food and Agriculture Organization (FAO) (2011), food price volatility is likely to continue in the medium term and possibly increase, making farmers, input suppliers, processors and consumers more vulnerable to negative economic effects. Buguk et al. (2003) and Gilbert (2010) related volatility of food prices to increases in demand variability, especially in agricultural products used for food and bio-fuel purposes; increases in supply variability due to adverse weather resulting in poor production yields; low demand elasticity; low supply elasticity; and exchange rate variability. Other factors, including stockholding, speculation, and macroeconomic variables such as inflation and exchange rates, can also affect the volatility of food prices. The degree that each factor impacts on price volatility depends on supply and demand elasticities, which reflect the degree of response that producers and consumers have when prices vary (Gilbert, 2010; Nourou, 2015; Roache, 2010).

According to Shroeter and Azzam (1991), Tomek and Kaiser (2014) and Wohlgenant (2001), variability in the output price is positively related to marketing margins, and consequently, negatively affects vertical price transmission. In addition, market power, quality, risk, contracts, technical change, transport and transaction costs, public interventions, perishability of the product, and time lags in supply and demand have been indicated as factors that affect marketing margins and price transmission along agro-food chains (Brester and Marsh, 2001; Bunte, 2006; Conforti, 2004; Popovic et al., 2016; Rapsomanikis and Mugera, 2011; Wohlgenant, 2001). Even though vertical price transmission and marketing margins in agro-food chains have been studied by several researchers, including Ahn and Lee (2015), Facker and Goodwin (2001), Kaspersen and Foyn (2010), Newton (2016), Von Cramon-Taubadel (1998) and Wohlgenant (2001), there has been a lack of research to determine the effects of price volatility in international markets on farmer prices and marketing margins, which can help to highlight differences between groups of countries, providing valuable insights for policy interventions aiming to improve efficiency and competition in agro-food chains.

Prices signal consumer preferences back to producers, who can allocate more inputs to products preferred by consumers (Norwood and Lusk, 2008). Inefficiencies in price transmission compromise the performance of the whole chain; however, lags in price transmission could be positive for the performance of the value chain, as farmers experience reduced price risk and are able to make more efficient decisions in the short run (Swinnen and Vandeplas, 2013). Evidence of distortions in short-term price transmission, including price leveling1 and price averaging2, were found in the Australian beef market by Griffith et al., (1991), and Griffith and Piggott (1994). These distortions smooth or reduce the price volatility in an effort to keep real prices relatively stable, but these studies also found evidence of asymmetric transmission in the short run, with wholesalers and retailers more prone to pass on price falls than price rises. These findings are consistent with those reported by other researchers, such as Xia and Li (2010).

Given the current relevance of volatility and transmission of food prices in global policy discussion, the study of the effects of price volatility can inform producers, input suppliers, processors, retailers and policy makers about the efficiency of agro-food chains and the level of risk protection for farmers. Thus, the contribution of this article to the literature lies in the way it formally tests the effects of export price volatility on vertical price transmission and marketing margins, being, at the best of our knowledge, the first research that directly estimates these effects on an agro-food chain in a group of developed and developing countries.

Considering the differences in cattle production systems and characteristics of the chain between developed and developing countries, this research aims to identify main differences between these groups of countries that can contribute to policy recommendations.

1 2

Wholesalers or retailers apply price levelling when they hold their selling prices relative stable when farm prices vary.

Price averaging is a practice that sets a high spread on some meat types to compensate for other low price spreads set on other types.

International Food and Agribusiness Management Review

336

Morales

Volume 21, Issue 3, 2018

The remainder of this article is organized as follow. The second section presents the theory of price volatility, vertical price transmission, and marketing margins, and the models used to test the impacts of export price volatility. The third section describes the data used in this study and presents some descriptive statistics, and the fourth section provides the results of the panel estimation modeling. Finally, the fifth section presents the main conclusions and discusses policy recommendations derived from the main results obtained in this research.

2. Methodological approach According to Minot (2014), price instability is the variation over time in the prices of a product. This variation can be measured using the coefficient of variation (CV), with CV=s/Âľ, where s is the standard deviation of a variable over time and Âľ is the mean value over the period under analysis. To avoid potential bias related to the length of time under analysis when price series are non-stationary, a popular alternative measure used is standard deviation of returns or unconditional volatility. Returns (rt) are defined as the proportional change in price from one period to the following one. stdev(r) =

â&#x2C6;&#x161;[â&#x2C6;&#x2018;

n t=1 n

]

1 (r â&#x20AC;&#x201C; r)2 (1) â&#x20AC;&#x201C;1 t

where rt = ln(Pt) â&#x20AC;&#x201C; ln(Ptâ&#x20AC;&#x201C;1) and r = 1n â&#x2C6;&#x2018; t=1 rt. Based on this measure, this study uses a three-year standard deviation of proportional price changes as a measure of price volatility. n

Tomek and Kaiser (2014) define price transmission as the process of price adjustments in the marketing system. Zorya et al. (2012) indicated that price transmission is essential to incentivize production and to respond to global food scarcity or surplus. Hence, this study focuses on vertical transmission of shocks from export to domestic prices, which is used as an indicator of market integration (Rapsomanikis and Mugera, 2011). In contrast, increasing marketing margins could inflate retail prices and deflate farmer prices, with subsequent negative implications for agribusinesses, including farmers, service providers and input suppliers (Wohlgenant, 2001; Norwood and Lusk, 2008). Therefore, the proportion of the variation in prices that are transmitted is a key indicator for assessing the degree of market integration, efficiency and competition in agro-food chains. Wohlgenant and Mullen (1987) modeled the vertical spread of prices from wholesale or export (downstream) to farm (upstream) based on the derived demand for farm output as presented in Equation 2: Pft = f (Pet, Qft, Ct)

(2)

where Pft is the farmer price of a commodity at time t; Pet is the export price of a commodity at time t; Qft is the quantity of farm output at time t; and Ct is a vector of marketing input prices in time t, including transport costs and wage rates. When price variations are perfectly transmitted throughout the chain, price fluctuations at different levels will be fully and instantaneously transmitted (Ahn and Lee, 2015, Kaspersen and Foyn, 2010; Rapsomanikis and Mugera, 2011). In addition, Meyer and von Cramon-Taubadel (2004), Frey and Manera (2007), and Ahn and Lee (2015) defined the relationship between wholesale and farmer prices, including current and previous prices in a vertical marketing chain, as shown below: đ?&#x2018;&#x203A;đ?&#x2018;&#x203A;

đ?&#x2018;&#x203A;đ?&#x2018;&#x203A;

đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;=1

đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;=0

đ?&#x2018;&#x192;đ?&#x2018;&#x192;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C; = đ?&#x203A;źđ?&#x203A;ź0 + â&#x2C6;&#x2018; đ?&#x203A;źđ?&#x203A;źđ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x2018;&#x192;đ?&#x2018;&#x192;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;â&#x2C6;&#x2019;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + â&#x2C6;&#x2018; đ?&#x203A;˝đ?&#x203A;˝đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; đ?&#x2018;&#x192;đ?&#x2018;&#x192;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + đ?&#x153;&#x20AC;đ?&#x153;&#x20AC;đ?&#x2018;Ąđ?&#x2018;Ą

(3)

This specification is appropriate only when price series are stationary. In the case the series are non-stationary and integrated order one (I(1)), a model in first differences is more suitable for avoiding a potential spurious regression when there appears to be a significant relationship among variables trending over time (Granger and Newbold, 1974).

International Food and Agribusiness Management Review

337

Morales

Volume 21, Issue 3, 2018

Moreover, in agricultural markets some periods may pass before prices adjust to shocks in other markets due to contracts and transport delays. Thus, dynamic models that include lagged endogenous and exogenous variables have been used to assess price transmission (Conforti, 2004; Fackler and Goodwin, 2001). The empirical model used in this study is based on the approach to test asymmetric price transmission introduced

by Wolffram (1971) and Houck (1977), and further developed by Meyer and von Cramon-Taubadel (2004).

Additionally, the model includes a term to measure the indirect effects on farmer price of volatility transmitted through export prices, based on the effects reported by Shroeter and Azzam (1991), Tomek and Kaiser (2014) and Wohlgenant (2001), and a dummy variable that can detect a structural break after the change in trend of commodity prices. đ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C; = đ?&#x203A;źđ?&#x203A;ź + â&#x2C6;&#x2018;

đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;<đ?&#x2018;&#x2021;đ?&#x2018;&#x2021;

đ?&#x2018; đ?&#x2018; =1

đ?&#x153;&#x192;đ?&#x153;&#x192;đ?&#x2018; đ?&#x2018; đ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; + â&#x2C6;&#x2018;

+ đ?&#x203A;żđ?&#x203A;żđ?&#x203A;żđ?&#x203A;żđ?&#x203A;żđ?&#x203A;żđ?&#x2018;&#x201E;đ?&#x2018;&#x201E;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C; + â&#x2C6;&#x2018;

+â&#x2C6;&#x2018;

đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;<đ?&#x2018;&#x2021;đ?&#x2018;&#x2021; đ?&#x2018; đ?&#x2018; =0

đ?&#x2018;&#x20AC;đ?&#x2018;&#x20AC;

đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;<đ?&#x2018;&#x2021;đ?&#x2018;&#x2021; đ?&#x2018; đ?&#x2018; =0

â&#x2C6;&#x2019; đ?&#x203A;˝đ?&#x203A;˝đ?&#x2018; đ?&#x2018; â&#x2C6;&#x2019; đ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; +â&#x2C6;&#x2018;

đ?&#x153;&#x201D;đ?&#x153;&#x201D;â&#x201E;&#x17D; đ??żđ??żđ??żđ??żđ??żđ??żâ&#x201E;&#x17D;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; + â&#x2C6;&#x2018;

â&#x201E;&#x17D;=1

đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;<đ?&#x2018;&#x2021;đ?&#x2018;&#x2021;

đ?&#x2018; đ?&#x2018; =0

đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;<đ?&#x2018;&#x2021;đ?&#x2018;&#x2021; đ?&#x2018; đ?&#x2018; =0

+ đ?&#x203A;˝đ?&#x203A;˝đ?&#x2018; đ?&#x2018; + đ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018;

â&#x2C6;&#x2019; đ?&#x153;&#x152;đ?&#x153;&#x152;đ?&#x2018; đ?&#x2018; â&#x2C6;&#x2019; đ?&#x153;&#x17D;đ?&#x153;&#x17D;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; đ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018;

(4)

+ đ?&#x153;&#x152;đ?&#x153;&#x152;đ?&#x2018; đ?&#x2018; + đ?&#x153;&#x17D;đ?&#x153;&#x17D;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; đ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; đ??ˇđ??ˇđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; + đ?&#x153;&#x2018;đ?&#x153;&#x2018;đ??ˇđ??ˇ2005â&#x2C6;&#x2019;13 + đ?&#x2018;˘đ?&#x2018;˘đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;

where Pfit is the farmer price in country i at time t; Pfit-s is the farmer price in country i at time t-s; Peit-s is the export price in country i at time t-s; Dit-s and D+it-s are dummy variables with Dit-s=1 if Peit<Peit-1 and Dit-s=0 otherwise, and with D+it-s=1 if Peitâ&#x2030;ĽPeit-1 and D+it-s=0 otherwise; Qfit is the quantity of farm output in country i at time t; Chit is the matrix of prices of marketing input h in country i at time t; Ď&#x192;eit-s is the export price volatility estimated as a three-year standard deviation in country i at time t-s; D2005-13 is a dummy variable that covers the period since 2005 when commodity prices changed their declining trend; and uit is a vector of stochastic errors.

According to Tomek and Kaiser (2014) and Wholgenant (2001), marketing margins or price spreads are functions of the difference between downstream and upstream prices, which represents charges for processing, packaging, transporting, storing and any other activity that could add value to the raw farm product. In the study case of this research, marketing margins will be the difference between export and farmer prices.

Thus, the derived wholesale demand, farm supply, and marketing inputs prices affect marketing margins, along with other factors such as market power, risk, quality, technical change, and time lags in supply and demand. Lyon and Thompson (1993) defined the following four marketing margin models: MMt = f(Pet, Ct)

Mark-up model

(5)

MMt = f(Pet, PetQft, Ct)

Relative Price spread model

(6)

MMt = f(Qft, Ct)

Marketing cost model

(7)

MMt = f(Pft, Et(Pft+1), Ct)

Rational expectations model

(8)

where MMt is the marketing margin at time t and Et(Pft+1) is expected value of farm price at time t+1. The marketing margin model used in the present study encompasses the four models previously presented, considering naĂŻve farmer price expectations, with an expected farmer price at time t+1 equal to the one observed at time t. Additionally, in line with the price transmission model, the marketing margin model also includes dynamic lag components, the indirect effect of volatility transmitted through export prices, and a dummy variable that can detect a structural break after the change in trend of commodity prices. đ??żđ??żđ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013; = đ?&#x203A;źđ?&#x203A;ź + â&#x2C6;&#x2018;

đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;<đ?&#x2018;&#x2021;đ?&#x2018;&#x2021;

đ?&#x153;&#x192;đ?&#x153;&#x192;đ?&#x2018; đ?&#x2018; đ??żđ??żđ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; + â&#x2C6;&#x2018;

đ?&#x2018; đ?&#x2018; =1 đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;<đ?&#x2018;&#x2021;đ?&#x2018;&#x2021;

+â&#x2C6;&#x2018;

đ?&#x2018; đ?&#x2018; =0

đ?&#x2018;&#x2020;đ?&#x2018;&#x2020;<đ?&#x2018;&#x2021;đ?&#x2018;&#x2021; đ?&#x2018; đ?&#x2018; =0

đ?&#x203A;˝đ?&#x203A;˝đ?&#x2018; đ?&#x2018; đ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; + đ?&#x203A;żđ?&#x203A;żđ?&#x203A;żđ?&#x203A;żđ?&#x203A;żđ?&#x203A;żđ?&#x2018;&#x201E;đ?&#x2018;&#x201E;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C;đ?&#x2018;&#x201C; + â&#x2C6;&#x2018;

đ?&#x153;&#x152;đ?&#x153;&#x152;đ?&#x2018; đ?&#x2018; đ?&#x153;&#x17D;đ?&#x153;&#x17D;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;Ąđ?&#x2018;Ąâ&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; đ??żđ??żđ??żđ??żđ??żđ??żđ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;đ?&#x2018;&#x2019;â&#x2C6;&#x2019;đ?&#x2018; đ?&#x2018; + đ?&#x153;&#x2018;đ?&#x153;&#x2018;đ??ˇđ??ˇ2005â&#x2C6;&#x2019;13 + đ?&#x2018;˘đ?&#x2018;˘đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;

đ?&#x2018;&#x20AC;đ?&#x2018;&#x20AC;

International Food and Agribusiness Management Review

338

đ?&#x153;&#x201D;đ?&#x153;&#x201D;â&#x201E;&#x17D; đ??żđ??żđ??żđ??żđ??żđ??żâ&#x201E;&#x17D;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;đ?&#x2018;&#x2013;

â&#x201E;&#x17D;=1

(9)

Morales

Volume 21, Issue 3, 2018

where MMit is the marketing margin between wholesale or export and farmer prices in country i at time t; MMit-s is the marketing margin in country i at time t-s; Peit-s is the export price in country i at time t-s; Qfit is the quantity of farm output in country i at time t; Chit is the matrix of prices of marketing input h in country i at time t; Ď&#x192;eit-s is the export price volatility estimated as a three-year standard deviation in country i at time t-s; D2005-13 is a dummy variable for the period since 2005 when the declining trend in commodity prices changed; and uit is a vector of stochastic errors.

Considering the relevance of production contracts agreed in advance to achieve vertical coordination and the cyclical behavior of beef cattle prices, it is expected a recursive process of price adjustments over time, where the quantity produced and farmer prices in year t depend on prices paid to farmers and other influential variables during the previous year t-1, following naĂŻve expectations (Norwood and Lusk, 2008; Tomek and Kaiser, 2014). Hence, the price transmission and marketing margins empirical models estimated in this study include one lag of those variables that have expected lagged effects.

Empirical studies using panel data have become popular, as panels enable control for individual heterogeneity and they provide more data, more variability, less collinearity among the variables, more degrees of freedom, and more efficiency (Baltagi, 2013). However, when modeling cross-sections of countries, regions, states or counties, the aggregate units are likely to exhibit significant cross-sectional or spatial correlation among the groups included in the panel. This cross-sectional dependence can lead to biased statistical inference, which affects the validity of result estimation tests (Cameron and Trivedi, 2005; Pesaran, 2004). Therefore, panel estimation should properly test for cross-sectional dependence and, when necessary, overcome the negative effects of this issue by modifying the model to capture spatial correlation, or by adjusting the standard errors of the coefficient estimates as suggested by Driscoll and Kraay (1998).

3. Data Beef cattle export and farmer prices, herd size and farmer maize prices panel data of twelve countries from 1970 to 2013 was obtained from FAOSTAT (2017). This research focuses on this case study in developed and developing countries, given their data availability and simplicity, as it requires the comparison of cattle prices at two different levels of the chain. The countries included in this study were the developed countries of The Netherlands, Germany, Switzerland, United States, Australia and New Zealand, and the developing countries of Panama, Colombia, Hungary, Kenya, South Africa and Thailand. These countries were selected according to their cattle exports relevance and data availability for this study. Oil prices were obtained from British Petroleum (BP) (2016), and gross domestic product (GDP) per capita was obtained from the Economic Research Services of the United States Department of Agriculture (2016). Increases in oil price are expected to affect vertical price transmission by raising marketing margins, as oil is a critical input in the value-added process, which includes transportation costs. However, oil is also a relevant input in the production process at farm level, similarly to maize used for feeding purposes. GDP per capita is related to income per capita, price of labor, and the level of development of the country, which should affect the degree of market efficiency. Finally, variations in cattle herd size represent fluctuations in the quantity of cattle supplied to the market.

All prices and GDP were transformed into United States dollars and then deflated using the annual consumer price index of the United States, which was obtained from the Bureau of Labor Statistics of the United States Department of Labor (2016). The data used in the models are in United States dollars in values for the year 2000. Table 1 presents descriptive statistics of average values for the variables used in the three groups of countries studied in this research, including a group of developed countries, a group of developing countries, and a group of all selected countries together.

The summary statistics demonstrate that farmer and export prices, on average, are higher in developed countries compared to developing countries. In contrast, marketing margins are bigger in developing countries than in developed countries, and export price volatility is bigger and more variable in developing countries than International Food and Agribusiness Management Review

339

Morales

Volume 21, Issue 3, 2018

Table 1. Panel data average statistics.1

Farmer cattle prices (USD base 2,000/ton) Export cattle prices (USD base 2,000/ton) Marketing margins (USD base 2,000/ton) Cattle herd size (thousands of heads) Farmer maize prices (USD base 2,000/ton) Oil price (USD base 2,000/barrel) GDP per capita (USD base 2,000) Export price volatility 1

All countries

Developed countries

Developing countries

Mean

Std. dev.

Mean

Std. dev.

Mean

Std. dev.

2,082 2,631 578 18,485 284 39 17,313 0.28

777 1,062 501 944 118 22 3,558 0.06

2,835 3,105 425 27,270 331 39 30,720 0.23

1,251 1,489 622 2,381 175 22 6,275 0.08

1,274 2,158 875 9,700 237 39 3,906 0.34

254 777 700 800 66 22 877 0.12

Marketing margin values included in this study are positive or equal to zero; GDP = gross domestic product.

in developed countries. To graphically confirm the relationships under analysis, Figure 1 plots the average farmer cattle prices, export cattle prices, and volatility of export prices for all countries, and for developed and developing countries.

Figure 1 shows that farmer and export prices tend to move together in both the all countries group and the developed countries group, while the relationship of both prices appears to be weaker in the case of developing countries. Consequently, exporters in developing countries appear to absorb most of the variations in export prices and offer more stable prices for farmers, which are substantially lower than those offered, on average, to farmers in developed countries. In addition, there is no clear evidence in the three cases presented in Figure 1 that export price volatility and farmer price are directly related, which supports the model of indirect effects transmitted through export prices to test whether exporters absorb most of the price fluctuations. Moreover, export price volatility in developing countries is greater than that exhibited in developed countries, which demonstrates the existence of more unstable prices for exporters in developing countries, and it could be considered as a reason for the lower prices offered to farmers in these less developed countries. Similarly, comparisons between average marketing margins and volatility of export prices for all countries included in this study, and for developed and developing countries, are presented in Figure 2.

In the graphs for all, developed and developing countries groups, there appears to be a relationship between marketing margins and export price volatility. However, marketing margins in developing countries are greater than those exhibited in developed countries, which supports the idea that exporters in developing countries absorb most of the variations in export prices and offer more stable but lower prices to farmers, as the lower prices appear not linked to the volatility faced when the cattle is exported. Further statistical analysis will be conducted in this study to formally test for potential relationships between export price volatility, vertical price transmission, and marketing margins in the three groups of countries.

4. Empirical results Unit root tests were conducted on the series included in this study to test for non-stationarity. The series were initially tested using the test proposed by Im et al. (2003) with trend. Given the potential bias effects of cross-sectional dependence on this test, the series were also tested using the test proposed by Breitung (2000), and Breitung and Das (2005). The results of both tests are presented in Table 2.

The results of the unit root tests that considered the potential effects of cross-sectional dependence demonstrate that several series can be classified as stationary, with the exception of cattle herd size, farmer maize price, oil price and GDP per capita. These results corroborate the findings of previous studies, including those of Canova (2007) and Uhligh (2005), who demonstrated that several time series in economics are stationary International Food and Agribusiness Management Review

340

Morales

Volume 21, Issue 3, 2018 Farmer price

6,000

Export price

Volatility of export price

0.50

USD b 2,000/ton cattle live weight

5,000

0.40

4,000 0.30 3,000 0.20 2,000 0.10

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1970

0

1972

1000

B

3-year moving sd of returns

A

9,000

0.00 0.50

0.40

7,000 6,000

0.30

5,000 4,000

0.20

3,000 2,000

3-year moving sd of returns

USD b 2,000/ton cattle live weight

8,000

0.10

C

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

0

1970

1000 0.00 0.80

6,000

0.60 4.000

0.40

3,000

2,000 0.20

3-year moving sd of returns

USD b 2,000/ton cattle live weight

5,000

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

0

1970

1000

0.00

Figure 1. Average farmer and export cattle prices versus average volatility of export cattle prices. A = all countries; B = developed countries; C = developing countries.

International Food and Agribusiness Management Review

341

Morales

Volume 21, Issue 3, 2018 Volatility of Export Price

0.50

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

0.00

1988

0

1986

0.10

1984

500

1982

0.20

1980

1000

1978

0.30

1976

1,500

1974

0.40

1972

2,000

3-year moving sd of returns

Marketing Margin

2,500

1970

USD b 2,000/ton cattle live weight

A

0.50

3,500 B

0.40

2,500 0.30

2,000 1,500

0.20

1000

3-year moving sd of returns

USD b 2,000/ton cattle live weight

3,000

0.10

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1970

0

1972

500

4,000

0.80

3,500 3,000

0.60

2,500 2,000

0.40

1,500 1000

0.20

3-year moving sd of returns

USD b 2,000/ton cattle live weight

C

0.00

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

1980

1978

1976

1974

1972

0

1970

500 0.00

Figure 2. Average marketing margins versus average volatility of export cattle prices. A= all countries; B = developed countries; C = developing countries.

International Food and Agribusiness Management Review

342

Morales

Volume 21, Issue 3, 2018

Table 2. Unit root tests results of series.1

Natural logarithm of farmer cattle prices (USD base 2,000/ton) Natural logarithm of export cattle prices (USD base 2,000/ton) Natural logarithm of marketing margins (USD base 2,000/ton) Natural logarithm of cattle herd size Natural logarithm of farmer maize prices (USD base 2,000/ton) Natural logarithm of oil price (USD base 2,000/barrel) Natural logarithm of GDP per capita (USD base 2,000) Export price volatility

Zt-bar test statistic (Im et al., 2003)

Lambda test statistic (Breitung and Das, 2005)

Level

Level

First difference

First difference

-3.1387*** -11.9577***

-2.1773**

-3.7385***

-4.3843*** -14.6041***

-3.8548***

-8.5838***

-11.1528*** -16.4604***

-8.0453***

-7.0478***

-9.8657*** -12.5354***

0.4942 -0.7088

-5.5399*** -5.8936***

-2.5424*** -13.3666***

-0.6267

-5.9113***

-10.8220***

2.4343

-4.6680***

-2.0478** -1.8240**

0.1085

-6.2926*** -13.2735***

-4.5093***

-10.1597***

1 The null hypotheses are series has unit root; *** and ** indicate that the parameter is significant at the 1 and 5% levels, respectively;

the Breitung and Das tests for natural logarithm of farmer cattle and maize prices, and marketing margins, were conducted on data with imputed values given the unbalanced nature of the original panel data; GDP = gross domestic product.

and have been incorrectly classified as non-stationary. The test for pooled estimation versus fixed-effects demonstrated that it is not possible to reject that the constant coefficients are equal to zero, which supports the use of a single constant coefficient using fixed-effects estimation for the vertical price transmission and marketing margin models (Baltagi, 2013). In addition, a Hausman (1978) test supports the use of fixed-effects estimations instead of random-effects, while a Pesaran (2004) test found evidence of crosssectional dependence in the vertical price transmission modeling3. An alternative estimation was run using a nonparametric covariance matrix estimator developed by Driscoll and Kraay (1998). This alternative method produces heteroskedasticity and autocorrelation consistent standard errors that are robust to spatial and temporal dependence. The estimated vertical price transmission models for the three groups of countries using Driscoll and Kraay standard errors are presented in Table 3.

The results for the all countries group indicate that lagged cattle farmer prices and first differences of herd size, farmer maize prices and GDP per capita are significant variables that explain variations in the cattle price paid to farmers. Farmer price lagged one period and first differences of farmer maize prices and GDP per capita have a positive influence on prices paid to farmers, while the first difference of herd size has a negative effect. This outcome suggests that farmer prices tend to follow the changes exhibited in the price paid during the previous year, following expectations according to the cattle price cycle. This result is consistent with the analysis for the U.S. market provided by Tomek and Kayser (2014). In addition, higher feeding and labor costs represented in first differences of farmer maize prices and GDP per capita have a positive influence on farmer cattle prices, while increases in herd size raise cattle supply, decreasing cattle prices paid to farmers. In contrast to other models, there is no evidence of asymmetry in price transmission.

Finally, there is no evidence of a structural break in price transmission after 2005, and the volatility transmitted through export prices and the first difference of oil prices are non-significant.

3

In the case of marketing margin modelling, it was not possible to perform the Pesaran (2004) test due to the more restricted number of observations available in the panel, which contained a group of observations that reported values without a defined natural logarithm value.

International Food and Agribusiness Management Review

343

Morales

Volume 21, Issue 3, 2018

Table 3. Price transmission estimation results using fixed-effects with Driscoll and Kraay (1998) standard errors.1

Constant Log of farmer cattle prices lagged one period Decreases in log of export cattle prices Increases in log of export cattle prices Decreases in log of export cattle prices lagged one period Increases in log of export cattle prices lagged one period First difference of log of cattle herd size First difference of log of farmer maize prices First difference of log of oil price First difference of log of GDP per capita Export price volatility transmitted through decreases in log of export prices Export price volatility transmitted through increases in log of export prices Export price volatility transmitted through decreases in log of export prices lagged one period Export price volatility transmitted through increases in log of export prices lagged one period Dummy period 2005 to 2013 Within R2 Groups Observations 1

All countries

Developed countries

Developing countries

Coefficient Std. error

Coefficient Std. error

Coefficient Std. error

0.5358**

0.8825***

0.2011 0.0293

0.5751*** 0.9093***

0.1824 0.0372

0.7218** 0.3228 0.8323*** 0.0474

-0.0043

0.0269

0.1173

0.0733

-0.0322

0.0264

0.0050

0.0265

0.1269*

0.0722

-0.0283

0.0255

0.0385

0.0294

-0.1093

0.0683

0.0866**

0.0340

0.0395

0.0290

-0.1079

0.0689

0.0895**

0.0343

-0.3152*

0.1744

-0.6388

0.4319

-0.1559

0.1651

0.2798*** 0.0780

0.4167*** 0.1037

0.1324

0.0843

0.0480 0.0306 0.7367*** 0.2557

0.0482 0.3908

0.0430 0.3533

0.0423 0.7632**

0.0327 0.2968

0.0062

0.0056

-0.0016

0.0100

0.0093

0.0070

0.0017

0.0096

-0.0345*** 0.0118

0.0224*

0.0119

0.0036

0.0058

0.0259*** 0.0091

-0.0043

0.0088

0.0028

0.0047

0.0139*

0.0077

-0.0068

0.0058

0.0041 0.8412 12 482

0.0197

-0.0112 0.8924 6 243

0.0286

0.0216 0.7504 6 239

0.0211

Dependent variable is log of farmer prices; ***, respectively; GDP = gross domestic product.

**

and * indicate that the parameter is significant at the 1, 5 and 10% levels,

In the case of developed countries, lagged cattle farmer prices, raises in export prices, the first difference of farmer maize prices and volatility transmitted through increases and decreases in export prices lagged one period are significant and have a positive impact on cattle farmer prices. There is evidence of asymmetry in price transmission contemporaneously, where export price increases are transmitted, but not decreases, in contrast to the findings reported by Griffith and Piggott (1994), Griffith et al. (1991) and Xia and Li (2010). However, volatility transmitted through increases in export prices has a significant negative effect on cattle prices paid to farmers, which demonstrates a contemporaneous detrimental effect on price increases transmitted to farmer prices when export prices are more volatile. Similarly to the results obtained for the

International Food and Agribusiness Management Review

344

Morales

Volume 21, Issue 3, 2018

all countries group, there is no evidence of a structural break in price transmission after 2005 or effects of oil prices on cattle prices paid to farmers in the developed countries group.

The model for the developing countries group indicates that lagged cattle farmer prices, increases and decreases in lagged export prices, the first difference of GDP per capita and volatility transmitted through increases in export prices are significant and increase the cattle prices paid to farmers. Moreover, there are no signs of asymmetry in price transmission, and transmission take longer in comparison to the results observed for the developed countries group. Nevertheless, there is asymmetry in the effect of export price volatility, with significant positive effects on farmer prices for volatility transmitted through contemporaneous increases in export prices, in contrast to the results obtained for developed countries. This outcome suggests that only large contemporaneous export price increases are transmitted to farmer prices, resulting in a smaller influence on farmer prices respect to the effect observed in developed countries, which supports the theory that exporters tend to absorb most of the price variations in less developed countries. In contrast to the results for the all countries and developed countries group, the first difference of maize prices are non-significant, which suggests a less intensive use of this type of feeding in developing countries, linked to lower cattle prices paid to farmers. Likewise to the results for the other groups, there is no evidence of a structural break in price transmission after 2005 and the first difference of oil prices are non-significant for the developing countries group.

To test for the indirect effects of volatility of export prices on marketing margins on the three groups studied, models were estimated using Driscoll and Kraay standard errors, and the results are presented in Table 4.

Table 4. Marketing margin estimation results using fixed-effects with Driscoll and Kraay (1998) standard errors.

Constant Log of marketing margins lagged one period Log of export cattle prices Log of export cattle prices lagged one period First difference of log of cattle herd size First difference of log of farmer maize prices First difference of log of oil price First difference of log of GDP per capita Export price volatility transmitted through log of export prices Export price volatility transmitted through log of export prices lagged one period Dummy period 2005 to 2013 Within R2 Groups Observations

All countries

Developed countries

Developing countries

Coefficient Std. error

Coefficient Std. error

Coefficient Std. error

-5.7130*** 1.2867 0.3298*** 0.0924

-5.8760*** 1.4974 0.4040*** 0.0937

-6.0047** 0.2601**

2.5656*** 0.1647 -1.2576*** 0.2324

2.4441*** 0.3627 -1.1879*** 0.4029

2.6584*** 0.2426 -1.2403*** 0.2438

-0.2320

0.6687

-0.8671

0.9585

0.2962

0.9348

-0.1620

0.1833

0.0586

0.2541

-0.2735

0.2393

-0.0593 -1.2852*

0.0715 0.7107

-0.1894 -2.4196

0.1721 1.7666

0.0685 -1.2982

0.0758 1.3948

2.2122 0.1246

-0.0001*** 0.0000

-0.0002*** 0.0001

-0.0001*** 0.0000

0.0001*** 0.0000

0.0001*

0.0001

0.0001*** 0.0000

-0.0274 0.7062 11 301

0.0008 0.7430 5 141

0.0623

-0.0541 0.6848 6 160

0.0535

1

0.0997

Dependent variable is log of marketing margins; ***, ** and * indicate that the parameter is significant at the 1, 5 and 10% levels, respectively; Germany was excluded from these estimations due to a low number of positive marketing margins values; GDP = gross domestic product.

International Food and Agribusiness Management Review

345

Morales

Volume 21, Issue 3, 2018

The estimation results for all, developed and developing countries groups indicate that marketing margins lagged one period, contemporaneous export cattle prices and volatility transmitted through export prices lagged one period are significant and have positive effects on marketing margins. Conversely, export cattle prices lagged one period and contemporaneous volatility transmitted through export prices have a significant negative effect on marketing margins. These outcomes confirm the expected positive effect on marketing margins of lagged marketing margins, contemporaneous export prices and the volatility transmitted through export prices lagged one period, which are consistent with the results of Shroeter and Azzam (1991), Tomek and Kaiser (2014) and Wohlgenant (2001). In addition, for the model of the all countries group the first difference of GDP per capita has a negative effect on marketing margins. This latest result indicates that in less developed countries, with lower GDP per capita, cattle exporters capture bigger marketing margins in comparison to those obtained in developed countries.

5. Conclusions This article contributes to the literature by examining the relationships between volatility in international prices, farmer prices, and marketing margins in cattle markets using panel data from twelve countries, comprising both developed and developing countries. Consequently, this study addresses the question: does export price volatility affect prices paid to cattle farmers and marketing margins captured by exporters?

The empirical results of the test on the complete group of all countries using fixed-effects estimation with Driscoll and Kraay (1998) standard errors to control for cross-sectional dependence indicate that indirect effects of volatility transmitted through export prices affect farmer prices and marketing margins. Relevant differences arise in the results for developed countries and for those countries categorized as developing countries. Previous cattle farmer prices increase prices paid to farmers for all groups and maize prices raise farmer prices for all countries and developed countries groups, where this feeding alternative is used more intensively, results consistent with those described by Tomek and Kaiser (2014). In addition, GDP per capita increases farmer prices for all countries and developing countries groups. Previous export prices are only significant for the group of developing countries, which demonstrates the longer period required for prices to be transmitted in those countries respect to developed countries. There are significant asymmetries in price transmission in developed countries with increases in contemporaneous export prices positively influencing cattle farmer prices, but not price decreases. This result differs from the findings reported by Ahn and Lee (2015), Facker and Goodwin (2001), Kaspersen and Foyn (2010), Von Cramon-Taubadel (1998) and Xia and Li (2010). However, in those countries there is a reduction on farmer prices that is adjusted by contemporaneous volatility transmitted through increases in export prices. Similarly, there is asymmetry in volatility for the developing countries group, with only contemporaneous volatility transmitted through increases in export prices raising cattle prices paid to farmers. This outcome and the non-significance of contemporaneous export prices suggests that solely large increases in current export prices are transmitted to farmer prices in developing countries.

In the case of cattle marketing margins, previous margins, contemporaneous export prices and volatility transmitted through the previous year export prices increase marketing margins, which corroborates the findings reported by Shroeter and Azzam (1991), Tomek and Kaiser (2014) and Wohlgenant (2001). Conversely, previous export cattle prices and contemporaneous volatility transmitted through export prices decrease marketing margins. Therefore, export price volatility decreases marketing margins contemporaneously, but increases marketing margins when lagged one period. This detrimental effect that contemporaneous volatility of export prices have on cattle marketing margins could be explained by production contracts with prices previously agreed. In contrast to the price transmission models, the GDP per capita reduces the marketing margins in the model for all countries group, which corroborates the findings that in more developed countries cattle prices paid to farmers are higher. These results demonstrate that even though the export price volatility is higher in developing countries, it is not the reason that prices paid to farmers are lower in comparison to those paid in developed countries. Cattle exporters in developing countries tend to pay more stable but lower prices to farmers, thus absorbing most of the fluctuation in export prices. These findings are different from the expected outcomes according to Shroeter and Azzam (1991), Tomek and International Food and Agribusiness Management Review

346

Morales

Volume 21, Issue 3, 2018

Kaiser (2014), and Wohlgenant (2001), who indicated that variability in international prices affect price transmission and marketing margins in agro-food chains.

The highly fluctuating behavior of marketing margins in developing countries, the longer period required for price shocks on export prices to be transmitted to farmers, the similarities of the effects of export price volatility on marketing margins across countries, the positive influence of GDP per capita on prices paid to farmers and the negative effect of this variable on marketing margins, suggest that exporters pay lower prices to farmers in developing countries to maximize profits rather than to minimize price risk for farmers and guarantee a firm production volume. These imperfections in the price transmission reduce the incentives for farmers to invest and adopt new technologies to increase the quality and quantity produced, which contributes to exacerbate price volatility in international markets, as highlighted by FAO (2011), and Zorya et al. (2010).

What are the policy implications of these findings? Instead of implementing price stabilizing programs that reduce price transmission from international to domestic prices, policy makers in developing countries are encouraged to take actions that promote more competition at wholesale level to improve the prices paid to farmers and allow a higher proportion of variations in international cattle prices to be transmitted to farmers.

These actions could incentivize an increase in the level of farm investment and in the volume of cattle offered in each country, which would also benefit input and service providers, traders and other actors in the chain.

Future research that aims to further test the theory of market integration and provide useful insights for policy recommendations should study the effects of market structure and spillover effects among related agro-food chains on price transmission and marketing margins.

Acknowledgements I would like to express my gratitude to the referees and managing editor of this journal for their valuable time devoted to review this manuscript and for providing very helpful comments and suggestions.

References Ahn, B. and H. Lee. 2015. Vertical price transmission of perishable products: the case of fresh fruits in the Western United States. Journal of Agricultural and Resource Economics 40: 405-424.

Baltagi, B.H. 2013. Econometric analysis of panel data, 5th edition. John Wiley and Sons, West Sussex, UK. Breitung, J. 2000. The local power of some unit root tests for panel data. In: Advances in econometrics vol.

15: nonstationary panels, panel cointegration, and dynamic panels, edited by B.H. Baltagi. JAY Press, Amsterdam, the Netherlands, pp. 161-178.

Breitung, J. and S. Das. 2005. Panel unit root tests under cross-sectional dependence. Statistica Neerlandica 59: 414-433.

Brester, G.W. and J.M. Marsh. 2001. The effects of U.S. meat packing and livestock production technologies on marketing margins and prices. Journal of Agricultural and Resource Economics 26: 445-462.

British Petroleum (BP). 2016. Statistical review of world energy. Available at: http://tinyurl.com/y78qgtrm.

Buguk, C., D. Hudson and T. Hanson. 2003. Price volatility spillover in agricultural markets: an examination of U.S. catfish markets. Journal of Agricultural and Resource Economics 28: 86-99.

Bunte, F. 2006. Pricing and performance in agri-food supply chains. In: Quantifying the agri-food supply chain, edited by C.J.M. Ondersteijn, J.H.M. Wijnands, R.B.M. Huirne and O. van Kooten. Springer, Dordrecht, the Netherlands, pp. 37-45.

Cameron, A.C. and P.K. Trivedi. 2005. Microeconometrics: methods and applications. Cambridge University Press, New York, NY, USA. Canova, F. 2007. Methods for applied macroeconomic research. Princeton University Press, New Jersey, NJ, USA.

International Food and Agribusiness Management Review

347

Morales

Volume 21, Issue 3, 2018

Conforti, P. 2004. Price transmission in selected agricultural markets. FAO Commodity and Trade Policy Research Working Paper No. 7, Basic Foodstuffs Service (ESCB), Commodities and Trade Division, Food and Agriculture Organization of the United Nations (FAO), Rome, Italy. Available at: http:// tinyurl.com/y9ygvrzy.

Driscoll, J. and A.C. Kraay. 1998. Consistent covariance matrix estimation with spatially dependent data.

Review of Economics and Statistics 80: 549-560.

Economic Research Services. 2016. International macroeconomic data set. Available at: http://tinyurl.com/ y7s86vpu.

Fackler, P.L. and B.K. Goodwin. 2001. Spatial price analysis. In: Handbook of Agricultural Economics, edited

by B.L. Gardner and G.C. Rausser. Elsevier Science, Amsterdam, the Netherlands, pp. 971-1024.

FAOSTAT. 2017. FAO statistical database. Available at: http://faostat.fao.org.

Food and Agriculture Organization (FAO). The state of food insecurity. FAO Publishing Policy and Support Branch, Rome, Italy.

Frey, G. and M. Manera. 2007. Econometric models of asymmetric price transmission. Journal of Economic Surveys 21: 349-415.

Gilbert, C.L. 2010. How to understand high food prices. Journal of Agricultural Economics 61: 398-425.

Granger, C.W.J. and P. Newbold. 1974. Spurious regressions in econometrics. Journal of Econometrics 2: 111-120.

Griffith, G.R., W. Green and G.L. Duff. 1991. Another look at price levelling and price averaging in the sydney meat market. Review of Marketing and Agricultural Economics 59: 97-109.

Griffith, G.R. and N.E. Piggott. 1994. Asymmetry in beef, lamb and pork farm-retail price transmission in australia. Agricultural Economics 10: 307-316.

Hausman, J.A. 1978. Specification Tests in Econometrics. Econometrica 46: 1251-1271.

Houck, J.P. 1977. An approach to specifying and estimating nonreversible functions. American Journal of Agricultural Economics 59: 570-572.

Im, K.S., M.H. Pesaran and Y. Shin. 2003. Testing for unit roots in heterogeneous panels. Journal of Econometrics 115: 53-74.

Kaspersen, L.L. and T.H.Y. Foyn. 2010. Price transmission for agricultural commodities in uganda: an empirical vector autoregressive analysis. Uganda Strategy Support Program (USSP), Working Paper No. 6, International Food Policy Research Institute (IFPRI), Washington, WA, USA. Available at: http://tinyurl.com/ycuk7jmp.

Lyon, C.C. and G.D. Thompson. 1993. Temporal and spatial aggregation: alternative marketing margin models. American Journal of Agricultural Economics 75: 523-536.

Meyer, J. and S. von Cramon-Taubadel. 2004. Asymmetric price transmission: a survey. Journal of Agricultural Economics 55: 581-611.

Minot, N. 2014. Food price volatility in Sub-Saharan Africa: has it really increased? Food Policy 45: 45-56.

Newton, J. 2016. Price transmission in global dairy markets. International Food and Agribusiness Management Review 19: 57-72.

Norwood, F.B. and J.L. Lusk. 2008. Agricultural marketing and price analysis. Pearson, Prentice Hall, NJ, USA. Nourou, M. 2015. Can mastitis ‘contaminate’ poultry? Evidence on the transmission of volatility between poultry and other commodity prices. International Food and Agribusiness Management Review 18: 183-196.

Pesaran, M.H. 2004. General diagnostic test for cross-section dependence in panels. Working Paper. Trinity College, Cambridge, UK. Popovic, R., B. Radovanov and J.W. Dunn. 2016. Food scare crisis: the effect on Serbian diary market.

International Food and Agribusiness Management Review 20: 113-127.

Rapsomanikis, G. and H. Mugera. 2011. Price transmission and volatility spillovers in food markets of developing countries. In: Methods to analyse agricultural commodity price volatility, edited by I. Piot-Lepetit and R. M’Barek. Springer, Dordrecht, the Netherlands, pp. 165-179.

Roache, S.K. 2010. What explains the rise in food price volatility? International Monetary Fund Working Paper 10/129. Available at: http://tinyurl.com/ya28tjt6.

International Food and Agribusiness Management Review

348

Morales

Volume 21, Issue 3, 2018

Shroeter, J. and A. Azzam. 1991. Marketing margins, market power, and price uncertainty. American Journal of Agricultural Economics 73: 990-999.

Swinnen, J. and A. Vandeplas. 2013. Price transmission and market power in modern agricultural value chains. LICOS Centre for Institutions and Economic Performance, University of Leuven, Discussion Paper 347/2014. Available at: http://tinyurl.com/y7dynfeu.

Tomek, W.G. and H.M. Kaiser. 2014. Agricultural Product Prices, 5th edition. Cornell University Press, Ithaca, NY, USA. Uhlig, H. 2005. What are the effects of monetary policy on output? Results from an agnostic identification procedure. Journal of Monetary Economics 52: 381-419.

United States Department of Labor, Bureau of Labor Statistics (BLS). 2016. Consumer price index: all urban consumers (CPI-U). Available at: http://data.bls.gov/pdq/SurveyOutputServlet.

Von Cramon-Taubadel, S. 1998. Estimating asymmetric price transmission with the error correction representation: an application to the German pork market. European Review of Agricultural Economics 25: 1-18.

Wohlgenant, M. 2001. Marketing margins: empirical analysis. In: Handbook of Agricultural Economics, vol. 1, edited by B. Gardner and G. Rausser. North-Holland Publishing Company, Amsterdam, the Netherlands, pp. 933-970.

Wohlgenant, M. and J. Mullen. 1987. Modeling the farm-retail price spread for beef. Western Journal of Agricultural Economics 12: 119-125.

Wolffram, R. 1971. Positivistic measures of aggregate supply elasticities: some new approaches â&#x20AC;&#x201C; some critical notes. American Journal of Agricultural Economics 53: 356-359.

Xia, T. and X. Li. 2010. Consumption inertia and asymmetric price transmission. Journal of Agricultural and Resource Economics 35: 209-227.

Zorya, S., R. Townsend and C. Delgado. 2012. Transmission of global food prices to domestic prices in developing countries: why it matters, how it works, and why it should be enhanced. Working Paper 71268. World Bank, Washington, WA, USA. Available at: http://tinyurl.com/yb2lqtnn.

International Food and Agribusiness Management Review

349

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2016.0168 Received: 25 October 2016 / Accepted: 8 August 2017

Technical efficiency and marketing channels among smallscale farmers: evidence for raspberry production in Chile RESEARCH ARTICLE Roberto Jara-Rojas a, Boris E. Bravo-Uretab,c, Daniel Solísd, and Daniela Martínez Arriagadae aAssistant

Professor, bProfessor, and ePhD Student, Department of Agricultural Economics, Universidad de Talca, P.O. Box 747, Talca 3460000, Chile cProfessor,

Department of Agricultural and Resource Economics, University of Connecticut, Storrs, CT 06269-1182, USA

dAssistant

Professor, Agribusiness Program, College of Agriculture and Food Sciences, Florida A&M University, Tallahassee, FL 32307, USA

Abstract This study analyses technical efficiency (TE) levels among small-scale raspberry producers in Central Chile. Special attention is given in to investigate the impact of the marketing channel used by the farmers on their technical performance. The data used in this study were obtained from a farm-level survey of 139 small-scale raspberry farmers. A stochastic production frontier model was used to evaluate the association between TE, extension, training and farmers’ decisions to sell their production directly to the agro-industry or indirectly through an informal middleman. The empirical results show that the decision to sell raspberries using informal channels is negatively associated with farm productivity and revenues. The analysis also reveals a positive relationship between TE and income among experienced and trained farmers. Implementing food quality and safety standards was also found decisive in increasing farm income. Policy implications stemming from these results are also discussed.

Keywords: marketing channels, stochastic production frontiers, technical efficiency, small-scale farmers, raspberry, productivity JEL code: D24, F63, Q13, O13 Corresponding author: rjara@utalca.cl

© 2017 Jara-Rojas et al.

351

Jara-Rojas et al.

Volume 21, Issue 3, 2018

1. Introduction Since the 1960s, Chile has promoted non-traditional exports as part of a general outward-looking economic strategy (Barham et al., 1992; Melo et al., 2014). During the last 10 years, the value of fruit exports has increased at annual averages close to 13%. In the 2015 season, more than 50% of the fruit production was exported, representing over US $4.4 billion in sales (ODEPA, 2017a). Raspberry production represents approximately 3% of total fruit exports. However, this is an important economic enterprise for a large number of small-scale farmers, and thus has a substantial implication for the economic wellbeing of many rural families and their communities (DomĂ­nguez, 2012).

Raspberries have shown high volatility in real prices over time. Real prices can fluctuate by as much as 300% from one season to the next, making raspberry a very risky endeavor with the potential for high profits, but also for high losses (Challies and Murray, 2011). Figure 1 shows the average price per kilogram received by Chilean farmers. Lower prices during 2005 and 2012 can be explained by high levels of global production especially in Poland, Serbia and USA (the three top raspberry producers in the world, Chile ranks fourth). On the contrary, in 2008 and 2014 global output decreased due to adverse climatic conditions in many countries resulting in higher prices and revenues for Chilean producers (Fedefruta, 2016). This variability in revenues combined with high labor costs has forced medium-to-large producers in Chile to exit the market, and has allowed small-scale farmers, mostly family operations, to expand their participation in raspberry production.

Chile currently has over 21,000 farmers growing raspberries on 16,000 hectares, which results in an average farm size equal to 0.76 hectares. Raspberry production is concentrated in the Maule Region (67% of the total land and 77% of farmers) with an average farm size below of the national average at 0.66 hectares.

The Bio-Bio and Los Lagos regions account for 20 and 10% of the total land, respectively (SAG, 2016).

Raspberries are highly susceptible to physical damage and bruising; therefore, harvest and post-harvest grading and packing require intensive use of well-trained workers to handle these activities. Mechanical harvesting saves a significant amount of labor, an increasingly scarce resource in Chilean agriculture; however, the initial capital outlay and maintenance costs of mechanized systems are substantial, making them financially feasible only for large-scale operations. Moreover, the overall farm architecture (i.e. spacing and layout of

Raspberry prices

3

US$ per kilogram

2.5 2 1.5 1 0.5 0 Raspberry prices

2005 0.75

2006 0.9

2007 1.3

2008 2

2009 1.71

2010 1.19

2011 1.07

2012 0.97

2013 2

2014 2.57

2015 1.98

2016 1.8

Figure 1. Average prices received by farmers in US$ per kilogram, 2005-2016 (adapted from ODEPA, 2017b).

International Food and Agribusiness Management Review

352

Jara-Rojas et al.

Volume 21, Issue 3, 2018

hedgerows, trellises and irrigation systems) must be redesigned to accommodate mechanical harvesting and the initial capital required for such farm transformation is considerable (Strik, 2007). Cultivars suitable for mechanical harvesting are also required. Thus, the innovation process requires high levels of investment but lack of funds among small-scale farmers is a significant constraint for the Chilean raspberry sector (Challies and Murray, 2011).

In Chile, the National Institute for Agricultural Development (INDAP, its official acronym in Spanish) is the main agency providing support to small-scale agriculture, with the aim of improving the competitiveness and market orientation of small-scale farmers. INDAP also finances technical assistance and management programs, and implements general assistance for low-income farmers (OECD, 2008). The agricultural extension assistance for raspberry farmers is mainly focused on various aspects of production and on helping farmers to comply with regulations focusing on Good Agricultural Practices (GAP). A total of 280,000 smallscale farmers cultivate four million hectares, almost 25% of the agricultural land in Chile. These farmers produce roughly 45% of the annual crops (317,058 hectares), vegetables (31,342 hectares), wine grapes (56,250 hectares) and livestock in the country, and 29% (86,143 hectares) of the major fruit crops (apples, avocados, and table grapes). More importantly, in the context of this study, small-scale farmers account for 96% of all raspberries grown in the country (SAG, 2016).

Considering that raspberry production is an important cash crop for small-scale producers and that the lack of funds is one of the main barriers to improve production levels, Domínguez (2012) describes a set of priorities that must be addressed to increase the competitiveness and productivity of this sector. These priories include: (1) establishment of plant breeding programs to develop higher yielding varieties; (2) streamlining marketing channels; (3) greater focus on Individually Quick Frozen (IQF) products rather than block pack products to generate higher farm revenues; and (4) adaptation to climate fluctuations, particularly through the adoption of improved irrigation technologies. These are important priorities but the work required to develop and promote the adoption of innovations is a lengthy process especially for small-scale farmers.

Consequently, in the short run, it is critical that farmers make the best use of their current technologies in order to enhance their competitiveness. In this context, understanding the efficiency gaps that might exist in the utilization of the available technology is an important endeavor.

Numerous empirical studies had estimated productivity and efficiency gaps in agriculture, focusing mainly on annual crops, dairy or livestock sectors (Elasraag and Alarcón, 2015). However, empirical studies centering on the productivity of the fruit sector, especially among small-scale farmers, are scarce. The few exceptions are Plénet et al. (2009), who measured efficiency in peach and nectarine production in France, and Henriques et al. (2009) and Moreira et al. (2011) who studied the TE of vineyards for wine production. There is also some work on table grapes (Ma et al., 2012), olives (Lachaal et al., 2005), and citrus (Lambarraa et al., 2007); but, to our knowledge, the present article is the first to study farm level TE for raspberry production.

The main goal of this article is to describe the production technology of small-scale raspberry farmers and analyze prevailing TE levels in the Maule region of Chile. Studying the sources of efficiency in agriculture is important because it allows farmers and policy makers to identify and target private and public resources in the most appropriate manner to improve agricultural production, productivity and agricultural incomes (Bravo-Ureta et al. 2007; Ogundari 2014). Our study also adds to the literature by explicitly analyzing the relationship between technical efficiency and the marketing channel used.

2. Overview of the raspberry agricultural chain sector.

Around 80% of the Chilean raspberry agricultural chain is export-oriented, berries are processed as frozen and the rest are exported as pulp or juice (Challies and Murray, 2011). Exports of fresh raspberries were interrupted in 2010 when Serbia and Mexico started to provide fresh fruit to Europe and USA at lower prices than Chile (Domínguez, 2012) and, as a result, production is currently marketed almost exclusively as processed fruit (Fedefruta, 2016).

International Food and Agribusiness Management Review

353

Jara-Rojas et al.

Volume 21, Issue 3, 2018

The international market for fresh fruit is regulated by different standards and norms, some of which are mandatory and enforced by public entities. Other standards are voluntary, developed by global food distribution chains, such as GlobalGap (Neven and Reardon, 2004). In developed countries, mandatory private food safety and quality standards govern the importation of fresh fruits. These norms are also becoming increasingly important in the domestic markets of many non-OECD countries in Africa, Latin America and East Asia due to the expansion of supermarket chains (Henson and Humphrey, 2009).

The market exclusion of small-scale farmers due to lack of funds is a concern, and strategies that encourage the implementation of stringent requirements pose major challenges for policymakers in providing opportunities for small-scale farmers to upgrade their operations (Asfaw et al., 2009). In Chile, since 2000, there have been several initiatives to support the certification of private GAP such as GlobalGAP or TESCO, which are standards to facilitate access to the most competitive and demanding markets. The results have been successful especially for medium and large farmers (CofrĂŠ et al., 2012). In the case of small-scale farmers, INDAP has promoted GAP practices among small-scale famers since 2005 with mixed results. Handschuch et al. (2013) show that the main barriers to implementing GAP certification among raspberry farmers are low educational levels, limited volumes and poor quality of the fruit sold. However, once farmers adopt GAP certification, a positive effect has been observed on the quality of their fruit as well as on their net raspberry income. Small-scale production and low levels of formal education are major challenges to meet any type of certification process (Handschuch et al., 2013). Therefore, small-scale farmers are highly dependent on the technical support provided by extension agents contracted by INDAP, such as Technical Assistance Service (SAT) and Local Development Program (PRODESAL), to guide farm management and fruit marketing, and to help comply with standards. SAT includes extension support, as well as the design, financing, monitoring and evaluation of technical assistance projects that are implemented in the field by external contractors (Apey and Barril, 2006). The aim of SAT is to increase the competitiveness of peasant enterprises in national and international markets. In contrast, PRODESAL aims to build technical and productive capacity among low-income, small-scale farmers and their families, with the goal of increasing their share of value added along the production process. PRODESAL is implemented at the local level through agreements between INDAP and municipal governments (Challies and Murray, 2011). Likewise INDAP provides financial capital exclusively for small-scale farmers such as credits (short or long run) and investment projects. The Investment Development Program (PDI) is an initiative that co-finances investment projects that enable the modernization of production processes, and provides support for project design and implementation. The difference with credit is that PDI is a non-refundable benefit.

In Chile, raspberry farmers have two alternative marketing channels: direct sale to the agro-industry; and the use of an informal trader (Challies, 2010). Formal channels include sending the fruit to raspberry collection centers located near the raspberry fields, from where the fruit is transported to agro-industry firms. Formal channels also include the possibility of sending the fruit directly to agro-industry firms. Usually the fruit is sent to collection centers or agro-industry in trays (not pre-packed). Under both modalities, the payment conditions are 30 to 60 days from the date of the invoice. The agro-industry firms export the raspberries directly or sell them to domestic wholesalers.

An alternative trading system includes transient intermediary traders, known colloquially (and slightly derogatorily) as conchenchos, which are common players in the informal trade business (Challies and Murray, 2011). Conchenchos generally buy raspberries by the tray for as low a cash price as possible, and then transport the fruit and sell it to agro-industry firms. Also there is a strong connection among low prices and low quality of the fruit. Despite the low prices they pay, these informal traders solve several problems especially for disadvantaged farmers: (1) they provide transportation for those producers who have no private means and cannot deliver their produce directly to an agro-industrial market; (2) provide immediate cash for farmersâ&#x20AC;&#x2122; operational and living expenses; and (3) make it possible to have transactions without formal invoicing, thus avoiding tax payments.

International Food and Agribusiness Management Review

354

Jara-Rojas et al.

Volume 21, Issue 3, 2018

3. Material and methods 3.1 Methodological approach In this study, we employ Stochastic Production Frontier (SPF) methodology to measure farm level TE. Following Battese and Coelli (1995), the general model can be depicted as: Yi = exp(χiβ + νi – μi) (1) where Yi is the value of the raspberries produced by the ith farmer, x are inputs, β is a vector of unknown parameters, and ν–μ=ε is the composed error term. The term ν represents a two-sided random error with a normal distribution (v~N [0, σv2]) that captures stochastic factors beyond the farmer’s control (e.g. climate, luck, etc.) and statistical noise. The term μ is a one-sided non-negative component that captures the TE of the producer. In other words, μ measures the gap between observed and maximum output that could be produced if the farm operated on the frontier, given the technology, inputs and the production environment.

TE for the ith farm can be measured as: TEi = exp(-μi) (2) where μ is the efficiency term as defined above. TE for each farm is calculated using the conditional mean of exp(–μ), given the composed error term for the stochastic frontier model (Battese and Coelli, 1988). TE ranges between 0 and 100%, where a value of 100% denotes full efficiency.

The maximum-likelihood method developed by Battese and Coelli (1995) makes it possible to estimate the determinants of farm technical inefficiency (TI) in a one-step procedure. Thus, TI can be estimated by incorporating the following expression in the frontier model shown in Equation 1: k μj = δ0 + ∑ δn znj + ωj n=1

(3)

where μj is technical inefficiency, znj are variables that affect efficiency, δn are unknown parameters to be estimated, and ωj is an error term.

3.2 Data, study area and empirical model This study was undertaken in the North Maule Basin, Province of Curicó, in Central Chile (Figure 2). The data used were obtained from a farm-level survey of 139 small-scale farmers, carried out between July and September of 2011. The questionnaire was divided into the following six sections: (1) human capital; (2) crops and land use; (3) inputs and infrastructure; (4) credit and incentives; (5) social capital; and (6) perceptions1.

To estimate TE levels and its determinants we used the following Cobb-Douglas (CD) stochastic production frontier: ln Raspberryi = αi + β1 ln Landi + β2 ln Pinputsi + β3 ln Labori + β4 ln Channeli + β5 ln Plantsi + νi – μi [δ0 + δ1 Agei + δ2 Educationi + δ3 Experienciei + δ4 Extensioni + δ5 Trainingi + ωi] (4) where Raspberry represents the value of the raspberry production of the ith farm; Land is the number of hectares devoted to raspberry production; Pinputs represents expenses on purchased inputs used for raspberries 1

A copy of the questionnaire can be found in Supplementary materials S1.

International Food and Agribusiness Management Review

355

Jara-Rojas et al.

Volume 21, Issue 3, 2018

Distribution of production Maule region 10,850 hectares 16,325 farmers Average: 0.66 ha per farmer Province of Curicó Maule Biobío Araucanía Los Ríos and Los Lagos

Biobío region 3,203 hectares 3,420 farmers Average: 0.94 ha per farmer Araucanía region 621 hectares 468 farmers Average: 1.33 ha per farmer Los Ríos and Los Lagos regions 1,208 hectares 235 farmers Average: 5 ha per farmer

Figure 2. Study area and raspberry statistics. (new vegetative material, fertilizers, pesticides); Labor is the value of both unpaid (family) and hired labor.

The value of unpaid labor was computed as kilograms harvested by family members times the price paid to hired workers per kilogram; Plants is a continuous variable that specifies the age of the canes in years and controls for the productive potential of the raspberry plants (younger plants are expected to produce more than older ones). Channel is a dummy variable equal to 1 if the fruit is marketed directly and zero if a trader is used. It is important to indicate that the decision to sell the production directly or indirectly is made post hoc, i.e. at harvest time. Therefore, the type of trade system selected does not affect the production decisions implemented by the farmer. This issue is important because it avoids any potential endogeneity problems in our estimations shown in Section 3. We should also mention that the raspberry producers in our data do not have pre-production contracts.

The inefficiency term μi is explained by the following variables: Age and Education of the household head, both in years; Experience or knowledge of raspberry production of the household head; Extension (if the household head received extension); and Training (if the household head participated in training courses).

The βs and δs are the parameters to be estimated; νi is the stochastic noise; and ωj is an error term. Table 1 shows a definition of variables used included in Equation 4.

International Food and Agribusiness Management Review

356

Jara-Rojas et al.

Volume 21, Issue 3, 2018

Table 1. Definitions of variables used in the econometric model.

Variable

Type

Definition

Raspberry Land Inputs Labor Channel

continuous continuous continuous continuous dummy

Plants Age Education Experience Extension Training

continuous continuous continuous continuous dummy dummy

raspberry production value in US$ hectares worked expense in plants, fertilizers, and pesticides in US$ value of total labor in US$ 1 if the farmer sold his produce to an informal trader; 0 if the produce is sold in the agri-industry age of the raspberry plants, in years age of the farmer, in years education of the farmer, in years farmer’s experience in raspberry production, in years 1 if the producer has received technical assistance by INDAP; 0 otherwise 1 if the producer has participated in training courses in raspberry production; 0 otherwise

4. Results and discussion Table 2 presents descriptive statistics for the sample. On average, the annual value of raspberry production is US $3,760 per farm, with a standard deviation of US$ 3,286. On average, the amount of Land devoted to raspberry production in our sample is one hectare, and the average expenditure in fertilizers and pesticides (Pinputs) is US$ 148. The amount spent on new plants is nearly zero, despite the fact that young plants and improved varieties are crucial to increase the productivity and competitiveness of the sector. Labor represents the major expense in raspberry production and has an average value, including unpaid family labor, of US$ 2,246. Direct costs, the sum of the expenditures on Labor and Pinput, reach US$ 2,494 on average. There is no official statistics for small-scale raspberry producers in Chile, but for medium and large-scale farmers the average direct cost is approximately US$ 6,800 (ODEPA, 2016). This difference is not surprising given the higher labor and overhead costs that lager farmers encounter in comparison to family farms.

The average age of the head of household is 52 years, which is consistent with other studies focusing on small-scale farmers in Chile (Jara-Rojas et al., 2012a, 2013). The level of education of the household head in the sample is low with only 7.8 years of schooling. On average, household heads have 13.4 years of experience in raspberry production. Many (42.5%) of the farmers have contacts with extension from SAT or PRODESAL, and 40.3% have received training in topics related to raspberry production and GAP. Table 2. Descriptive statistics of the variables included in the econometric model.

Variable

Units

Mean

Std. dev.

Max.

Min.

Raspberry Land Pinputs Labor Channel Plants Age Education Experience Extension Training

US$ hectares US$ US$ % years years years years % %

3,760 1.0 148 2,246 74.1 5.7 51.5 7.8 13.4 42.5 40.3

3,286 0.7 182 1,941 – 2.6 8.7 3.2 5.6 – –

17,272 5.0 1,515 10,850 – 11.0 76.0 14.0 22.0 – –

136 0.1 5.3 60 – 2.0 24.0 2.0 2.0 – –

International Food and Agribusiness Management Review

357

Jara-Rojas et al.

Volume 21, Issue 3, 2018

Table 3 presents the maximum likelihood estimates for the SPF model. The null hypothesis that γ=0 is rejected at the 1% significance level, which lends support to the SPF model, i.e. the SPF model is superior to an average production function resulting from estimation using ordinary least squares. Moreover, the value for γ is statistically significant, with a value of 0.78, which indicates that inefficiency is an important contributor to observed output variability (Battese and Coelli, 1995). The function coefficient is 0.922, revealing decreasing returns to scale.

The parameters for the three inputs in the CD production frontier, which can be interpreted as partial production elasticities, are statistically significant at the 5% level or better. Typically, Land exhibits the largest elasticity in studies analyzing small-scale agriculture (Jaime and Salazar, 2011). However, our study suggests that Labor is the most significant input, with a partial elasticity equal to 0.62. This value indicates that a 10% increase in the value of Labor results in a 6.2% increase in the value of production revealing the importance of labor in raspberry farming. According to our survey data, harvest labor accounts for roughly 95% of labor costs and 93% of total operating costs. Other TE studies dealing with fruit production report higher levels of elasticity for land (Coelli and Sanders, 2013; Guesmi et al., 2012; Moreira et al., 2011).

The parameter of the variable Plants is negative and significant, which confirms the fact that raspberry plants produce less as they age. Raspberry plants have their best yields in the first six years, but many farmers keep their plants for more than 10 years. Although the Plants parameter does not capture the possible effect of different raspberry varieties, this is a matter that should be considered by farmers, consultants, and policy makers. In our study, 99% of the farmers grow the ‘Heritage’ variety, but it is likely that using improved varieties could increase yields and fruit quality. Figure 3 shows the decreasing association between yield and years of the plants of our sample. Usually small-scale farmers producing own plants and thus the potential yields are lower than certified nursery plants.

Of particular importance in this study is the parameter for the dichotomous variable Channel, which equals 1 if the fruit is marketed directly and zero if a trader is used. Our estimates show that the parameter for Channel is negative and significant with a value of -0.156. This result suggests that farmers who sell their Table 3. Stochastic production frontier results.1 Variable

Coefficients

Constant Land Purchased inputs Labor Channel Plants Inefficiency model Constant Age Education Experience Extension Training Returns to scale Log likelihood function σ2 = σv2 + σu2 γ = σu2 / σ2 Average TE

5.715*** 0.214*** 0.086** 0.622*** -0.156* -0.099***

0.905 0.069 0.047 0.061 0.100 0.020

4.925*** -0.050*** -0.057 -0.287*** -0.559 -1.569*** 0.922 80.47 0.690*** 0.780*** 81.0%

1.690 0.022 0.066 0.068 0.696 0.783

1 ***

Standard error

0.105 0.079

= significant at 1%; TE = technical efficiency.

International Food and Agribusiness Management Review

358

Jara-Rojas et al.

Volume 21, Issue 3, 2018

9,000 7,799

kg per hectare

8,000

7,567

7,275

7,000

5,985

6,000

5,223

5,477

5,000

4,498

4,958 4,631

4,000 3,000

0

2

4

6 Age of the plants (years)

8

3,930 10

12

Figure 3. Production in kilograms per hectare and age of the plants. production using an informal trader (or conchencho) have a value of output that is 25% lower (US$ 3,463 vs 4,613)2, ceteris paribus, than those who sell directly to the market. As mentioned, Channel is related to the quality of the fruit. Farmers who can meet high quality standards can sell to the markets for fresh or frozen IQF raspberries; thus, getting a higher price compared to those with lower quality fruit who must sell the berries to juice or marmalade factories or to an informal trader.

Our results also show a strong relationship between the level of GAP practices employed in the farm and the use of formal trade (Table 4). Farmers in our sample use 12 different GAP practices. Specifically, most of the farmers (82.6%) with lower level of GAP (1 to 3 practices) sell their production to informal traders.

This percentage decreases to less than 50% for those farmers with 6 GAP practices. On average, farmers in our sample have implemented 5 of the 12 recommended practices and the most adopted is ‘Fruit storage’, which is a place where farmers select, classify and pack the fruit prior to transportation to the agro-industry.

Table 4 also shows a positive trend between the number of GAPs implemented and income. Ten farmers had implemented 10-12 GAP practices and their average income was US$ 7,554, while the income for those farmers with 1-3 was US$ 1,288. Handschuch et al. (2013) show that Chilean small-scale raspberry farmers benefit from the implementation of food quality and safety standards through better farming and higher management skills. Our findings reveal a direct link between GAP practices, higher volume and better quality. Gains in fruit quality also facilitate the access to formal markets and thus higher income.

Figure 4 shows the distribution of TE levels among our studied sample. The average TE level is 81% indicating that, an average farm in the sample could, in principle, increase its level of production by 19% using the current input quantities and technology. Figure 4 also shows that more than 50% of producers attain TE in the 70-79% range; and 22% of farmers reach a TE of 90% or higher. The average TE value is consistent with other studies focused on Latin America. Jara-Rojas et al. (2012b) reported an average level of TE of 80%, Solís et al. (2009) 78%, and Bravo-Ureta et al. (2007) found an average TE of 78% in their meta-analysis.

The bottom of Table 3 presents the parameters of the determinants of inefficiency. Following the usual practice, the interpretation is in terms of TE (instead of inefficiency). Frequently, the variable Age is used as a proxy for household experience. However, the literature shows mixed results with respect to the relationship between Age and TE. For example, young farmers can be relatively more efficient because they are more educated (Mariano et al., 2010); yet, older farmers can exhibit higher efficient due to more experience (Jaime and Salazar, 2011). Following Bozoğlu and Ceyhan (2007), we include the variables Age and Experience separately. Our results indicate that younger farmers are more efficient, but the parameter is 2

In terms of value per hectare, the average production value for those farmers that sell directly to the market is US$ 4,075/ha, and for those farmers using an informal trader is US$ 3,017/ha.

International Food and Agribusiness Management Review

359

Jara-Rojas et al.

Volume 21, Issue 3, 2018

Table 4. Good Agricultural Practices (GAP) practices and raspberry income.

Type of GAP practices Input warehouse Harvest tools warehouse Packing Fruit storage Latrines Fence Signs Workers dining SAG1 records Input applications records Harvest records Formal business

Number of GAP N° of farmers % of total 78 56.1 84 60.4 7 5.0 114 82.0 94 67.6 42 30.2 38 27.3 18 12.9 57 41.0 124 89.2 50 36.0 48 34.5

Frequency 1 2 3 4 5 6 7 8 9 10 11 12

N° of farmers % of total 4 2.88 5 3.60 14 10.07 20 14.39 37 26.62 29 20.86 9 6.47 7 5.04 4 2.88 4 2.88 1 0.72 5 3.60

Informal channel sales?

Number of GAP practices 1-3 4-6 7-9 10-12

Yes n=19 (82.6%) n=73 (76.8%) n=6 (54.5%) n=5 (50.0%)

No n=4 (17.4%) n=22 (23.2%) n=5 (45.5%) n=5 (50.0%)

Raspberry income2 US$ 1,288a US$ 3,317b US$ 4,017b US$ 7,554c

1

The Chilean Agriculture and Livestock Service (SAG) is the institution in charge of record of raspberry farmers and other value chain participants (e.g. traders, packing). Also SAG is in charge of monitoring GAP norms for farmers (food safety norm 341).

2 Different letters indicate significant differences (Tukey’s test, P<0.05) in raspberry income among different groups of GAP practices.

80.0 70.0 60.0 50.0 Average TE: 81% 40.0 30.0 20.0 10.0 0.0

20-29

30-39

40-49

50-59

% of farm in intervals

60-69

70-79

Number of farms

Figure 4. Distribution of technical efficiency (TE) scores.

International Food and Agribusiness Management Review

360

80-89

>90

Jara-Rojas et al.

Volume 21, Issue 3, 2018

not significant, while Experience in raspberry production is positively and significantly associated with TE. This result is in line with Rao et al. (2012) that analyze the participation in supermarket channels among vegetable producers in Kenya.

The parameter for Education shows a positive but non-significant relationship with TE. Abdulai and Huffman (2000) found that education has a positive and significant effect on TE, and suggested that an appropriate response to changes in market prices requires management skills acquired through education and access to information. Also, Asfaw et al. (2009) identify the lack of human and physical capital as major factors that limit the adoption of safety standards by small-scale farmers. The same authors add that public investment designed to promote farmersâ&#x20AC;&#x2122; productivity and connectivity to markets, and the promotion of collaborative action among producers are crucial policies to build the technical capacity of farmers.

Consistent with Feder et al. (2004), the variable Extension shows a positive but non-significant relationship with TE. The Extension services provided by the PRODESAL and SAT programs focus on various aspects of production, such as fertilization and crop protection, but do not address issues related to marketing channels that could explain this result. Extension services showed a positive association with TE in Lindara et al.

(2006). The parameter for the variable Training is significant and reveals a positive effect on TE. Training is defined as short courses taken by farmers, usually related to raspberry production and GAP topics. This result suggests that training courses that help farmers develop GAP help boost TE and this finding is consistent with those of Li and Sicular (2013).

Table 5 shows mean values for TE, Raspberry Income (RI) and Gross Margin (GM) among farmers in the sample for several variables. For example, TE, RI and GM for farmers who sold their produce to an informal trader (captured by the variable Channel) are significantly lower than for those farmers who sold to the agro-industry. Similar significant differences are exhibited when comparing farmers with training who had an average TE of 84% and an average RI of US$ 4,676, while those without training had an average TE of 80% and an average RI of US$ 3,176. Challies (2010) also found that training courses are highly beneficial in helping small-scale farmers become successful raspberry producers. In addition, we include the variable Project, which captures the effect of the PDI program and the results show that participants reach significantly higher levels of TE (84%), RI (US$ 5,497) and GM (US$ 4,596) compared to farmers Table 5. Differences in technical efficiency (TE), income and gross margin.

Variable Channel Informal trade Formal trade Projects Without PDI3 With PDI Training No Yes Credit by INDAP No Yes

N (%)

TE (%)

Raspberry income1

Gross margin2

74 26

81.6 83.1

3,463 4,613*

1,059 2,247*

81 19

81.2 85.3*

3,361 5,497*

4,485 4,596

60 40

80.7 84.0*

3,176 4,676*

4,485 4,596*

57 43

81.5 82.6

3,994 3,573

4,441 4,743

1

Total raspberry income in US$.

The gross margin (GM) is computed as Raspberry Income (RI) less expenditures on Purchased Inputs (PI) and Labor Cost (LC): GM = RI â&#x20AC;&#x201C; (PI + LC) 3 PDI = Investment Development Program. * Indicates significant differences at 5% confidence level (t-test) 2

International Food and Agribusiness Management Review

361

Jara-Rojas et al.

Volume 21, Issue 3, 2018

without such projects (TE=81.2%; RI=US$ 3,361; and GM=US$ 4,485, respectively). By contrast, credit provided by INDAP exhibits no significant effects on TE, RI and GM. Usually credit is used to cover direct costs and not for investment.

5. Summary and conclusions This study analyzed the determinants of TE for a sample of small-scale raspberry farmers in central Chile.

The empirical results suggest that the marketing channel used by farmers to sell their production plays an important role on the productivity and TE estimates. The empirical results also showed that human capital, in terms of Age, Experience and Training, is a crucial factor associated with higher levels of TE, where the latter is a proxy for managerial performance.

The Chilean Government is directly involved in supporting small-scale raspberry producers and the overall agricultural chain through SAG and INDAP, two leading governmental agencies within the Ministry of Agriculture (Challies and Murray, 2011). While SAG has a regulatory function, INDAP is the main agency that provides support to small-scale farmers and its mission is to increase the competitiveness of such farmers.

Given that raspberry production is an important source of income for small-scale farmers in Chile, this article has some policy implications that can be of significance to this vulnerable sector of producers. First, to increase the profitability and farm income of the raspberry sector, it is imperative to improve the managerial ability of small-scale farmers. The ability to produce and market high-quality fruit has a major impact on farm profitability particularly when output prices are low as was the case in the 2011-2012 seasons (50% lower than the 2008-2010 period). Thus, training programs provided by INDAP should be designed to promote technical capabilities and compliance with required quality standards. Second, INDAP should improve the targeting of incentive programs that help to acquire new and better varieties of raspberries so as to enhance the productivity of the sector. Finally, now that small-scale farmers have been working on raspberry production for more than 15 years, it is important to strengthen technical assistance focusing on managerial topics in order to improve TE and to enhance farm income and profitability among poor rural households.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2016.0168.

Materials S1. Questionnaire.

6. References Abdulai, A. and W. Huffman. 2000. Structural adjustment and economic efficiency of rice farmers in Northern Ghana. Economic Development and Cultural Change 48(3): 503-520.

Apey, A. and A. Barril. 2006. Pequeña agricultura en Chile: rasgos socio productivos, institucionalidad y clasificación territorial para la innovación. INDAP, ODEPA, MUCECH, and IICA. Santiago, Chile.

Asfaw, S., D. Mithöfer and H. Waibel. 2009. EU food safety standards, pesticide use and farm-level productivity: the case of high-value crops in Kenya. Journal of Agricultural Economics 60(3): 645-667.

Barham, B., M. Clark, E. Katz and R. Schurman. 1992. Nontraditional agricultural exports in Latin America.

Latin American Research Review 27: 43-82.

Battese, G.E. and T.J. Coelli. 1988. Prediction of firm-level technical efficiencies with a generalized frontier production function and panel data. Journal of Econometrics 38: 387-399.

Battese, G.E. and T.J. Coelli. 1995. A model for technical inefficiency effects in stochastic frontier production function for panel data. Empirical Economics 20: 325-332.

Bozoğlu, M. and V. Ceyhan. 2007. Measuring the technical efficiency and exploring the inefficiency determinants of vegetable farms in Samsun province, Turkey. Agricultural Systems 94: 649-656.

International Food and Agribusiness Management Review

362

Jara-Rojas et al.

Volume 21, Issue 3, 2018

Bravo-Ureta, B., D. Solís, V. Moreira, J. Maripani, A. Thiam and T. Rivas. 2007. Technical efficiency in farming: a meta-regression analysis. Journal of Productivity Analysis 27(1): 57-72.

Challies, E. 2010. Agri-Food Globalisation and rural transformation in Chile: smallholder livelihoods in the global value chain for raspberries. PhD dissertation, Victoria University of Wellington, Wellington, New Zealand.

Challies, E. and W.E. Murray. 2011. The interaction of global value chains and rural livelihoods: the case of smallholder raspberry growers in Chile. Journal of Agrarian Change 11: 29-59.

Coelli, T.J. and O. Sanders. 2013. The technical efficiency of wine grape growers in the murray-darling basin in Australia. In: Chapter 11 wine economics: Quantitative studies and empirical applications, edited by E. Giraud-Heraud and M.-C. Pichery. Palgrave Macmillan, London, UK. Cofré, G., I. Riquelme, A. Engler and R. Jara-Rojas. 2012. Adopción de Buenas Prácticas Agrícolas (BPA): costo de cumplimiento y beneficios percibidos entre productores de fruta fresca. Idesia 30(3): 37-45.

Dominguez, A. 2012. Chilean raspberry industry. In: 8th World Conference International Raspberry Organization (IRO) Canada, Abbotsford, 3-7 June 2012. IRO, Santiago, Chile.

Elasraag, Y.H. S. and Alarcón. 2015. Efficiency of wheat production in Egypt. New Medit 14(4): 19-27 Feder, G., R. Murgai and J.B. Quizon. 2004. Sending farmers back to school: the impact of farmer field schools in Indonesia. Review of Agricultural Economics 26(1): 45-62.

Guesmi, B., T. Serra, Z. Kallas and J.M. Gil Roig. 2012. The productive efficiency of organic farming: the case of grape sector in Catalonia. Spanish Journal of Agricultural Research 10(3):552-566.

Handschuch, C., M. Wollni and P. Villalobos. 2013. Adoption of food safety and quality standards among Chilean raspberry producers – Do smallholders benefit? Food Policy 40: 64-73.

Henriques, P., M. Carvalho and R. Fragoso, 2009. Technical efficiency of Portuguese wine farms. New Medit 1: 4-9.

Henson, S. and J. Humphrey. 2009. The impacts of private food safety standards on the food chain and on public standard-setting processes. Available at: http://www.fao.org/3/a-i1132e.pdf.

Jaime, M. and C. Salazar. 2011. Participation in organizations, technical efficiency and territorial differences: a study of small wheat farmers in Chile. Chilean Journal of Agricultural Research 71: 104-113.

Jara-Rojas, R., B.E. Bravo-Ureta and J. Díaz. 2012a. Adoption of water conservation techniques: a socioeconomic analysis for small-scale farmers in Central Chile. Agricultural Systems 110: 54-62.

Jara-Rojas, R., B.E. Bravo-Ureta, A. Engler and J. Díaz. 2013. An analysis of the joint adoption of soil and water conservation practices in Central Chile. Land Use Policy 32: 292-301.

Jara-Rojas, R., B.E. Bravo-Ureta, V. Moreira and J. Díaz. 2012b. Technical Efficiency and Natural Resource Conservation: Empirical Evidence from Small-Scale Farmers in Central Chile. In: 28th International Conference of Agricultural Economists, Foz de Iguazú, August 18-24, 2012. International Association for Applied Econometrics, Cambridge, UK. Lachaal, L., B. Karray, B. Dhehibi and A. Chebil. 2005. Technical efficiency measures and its determinants for olive producing farms in Tunisia: a stochastic frontier analysis. African Development Review 17(3): 580-591.

Lambarraa, F., T. Serra and J.M. Gil. 2007. Technical efficiency analysis and decomposition of productivity growth of Spanish olive farms. Spanish Journal of Agricultural Research 5(3): 259-270.

Li, M. and T. Sicular. 2013. Aging of the labor force and technical efficiency in crop production: evidence from Liaoning province, China. China Agricultural Economic Review 5(3): 342-359.

Lindara, L., F.H. Johnsen and H.M. Gunatilake. 2006. Technical efficiency in the spice based agroforestry sector in Matale district, Sri Lanka. Agroforestry Systems 68(3): 221-230.

Ma, C., W. Mu, J. Feng and W. Jiao. 2012. Assessing the technical efficiency of grape production in open field cultivation in China. Journal of Food, Agriculture and Environment 10: 345-349.

Mariano, M.J., R. Villano and E. Fleming. 2010. Are irrigated farming ecosystems more productive than rainfed farming systems in rice production in the Philippines? Agriculture, Ecosystems and Environment 139(4): 603-610.

Melo, O., A. Engler, L. Nahuelhual, G. Cofre and J. Barrena. 2014. Do sanitary, phytosanitary, and qualityrelated standards affect international trade? Evidence from Chilean fruit exports. World Development 54: 350-359.

International Food and Agribusiness Management Review

363

Jara-Rojas et al.

Volume 21, Issue 3, 2018

Moreira, V.H., J.L. Troncoso, and B.E. Bravo-Ureta. 2011. Technical efficiency for a sample of Chilean wine grape producers : a stochastic production frontier analysis. Ciencia e Investigación Agraria 38: 321-329.

Neven, D. and T. Reardon. 2004. The rise of Kenyan supermarkets and the evolution of their horticulture product procurement systems. Development Policy Review 22(6): 669-699.

ODEPA. 2016. Costos de producción. Frambuesas. Oficina de Estudios y Políticas Agrarias. Ministerio de Agricultura. Available at: http://tinyurl.com/yc6muuvc.

ODEPA. 2017a. Boletín de frutas y hortalizas procesadas/Exportaciones de fruta fresca. Oficina de Estudios y Políticas Agrarias. Ministerio de Agricultura. Available at: http://tinyurl.com/yaewd523.

ODEPA. 2017b. Frutas Frescas. Oficina de Estudios y Politícas Agrarias, Ministerio de Agricultura. Available at: http://tinyurl.com/y9525uj9.

OECD. 2008. OECD Review of Agricultural Policies: Chile. OECD Publishing, Paris, France.

Ogundari, K. 2014. The paradigm of agricultural efficiency and its implication on food security in Africa: what does meta-analysis reveal? World Development 64: 690-702.

Plénet, D., P. Giauque, E. Navarro, M. Millan, C. Hilaire, E. Hostalnou, A. Lyoussoufi, and J.-F. Samie. 2009.

Using on-field data to develop the EFI© information system to characterise agronomic productivity and labour efficiency in peach (Prunus persica L. Batsch) orchards in France. Agricultural Systems 100(1-3): 1-10.

Rao, E., B. Brümmer and M. Qaim. 2012. Farmer participation in supermarket channels, production technology, and efficiency: the case of vegetable in Kenya. American Journal of Agricultural Economics 93(4): 891-912.

Rosas, F. 2016. Perspectivas temporada mundial de los berries 2016/2017, PMA Fruit trade Latin America, In: 4to Seminario de berries, Temuco-Chile, 27 October 2016. Produce Marketing Association (PMA), Newark, DE, USA. Servicio Agrícola y Ganadero (SAG). 2016. Estadística de superficie nacional de frambuesas. Available at: http://tinyurl.com/y7fl6vr4.

Solís, D., B. Bravo-Ureta and R. Quiroga. 2009. Determinants of household efficiency among small-scale hillside farmers in El Salvador and Honduras. Journal of Agricultural Economics 60(1): 202-219.

Strik, B.C. 2007. Berry crops: worldwide area and production systems. In: Berry fruit: value-added products for health promotion. Taylor and Francis, Boca Raton, FL, USA, pp. 3-50.

International Food and Agribusiness Management Review

364

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2017.0044 Received: 10 May 2017 / Accepted: 13 December 2017

Defining U.S. consumers’ (mis)perceptions of pollinator friendly labels: an exploratory study RESEARCH ARTICLE Hayk Khachatryan

a

and Alicia Rihnb

aAssistant

Professor, and bResearch Associate, Food and Resource Economics Department, Mid-Florida Research and Education Center, University of Florida, 2725 S. Binion Road, Apopka, FL 32703, USA

Abstract Declining pollinator insect populations is an important global concern due to potential negative environmental and economic consequences. However, research on consumer perceptions of pollinator friendly traits is limited.

Understanding consumer perceptions is important because they impact behavior and product selection. In turn, this affects the effectiveness of relevant policies and pollinator insects’ access to beneficial plants. This manuscript quantifies consumers’ perceptions of plant traits that aid pollinators. U.S. consumers (n=1,243) were surveyed to identify their perceptions of pollinator friendly traits. Binary logit models and marginal effects were estimated using 22 plant traits and consumers’ purchasing interest, existing knowledge, and demographic variables. Results imply consumers interested in purchasing pollinator friendly plants selected positive traits regardless of accuracy. Furthermore, consumers selected traits that aligned with their knowledge.

Older participants had more accurate perceptions of pollinator friendly traits. Results highlight the challenges facing regulatory efforts geared towards promoting pollinator friendly products/practices.

Keywords: binary logit model, consumer behavior, environment, promotions JEL code: M31, Q13 Corresponding author: hayk@ufl.edu

© 2018 Khachatryan and Rihn

365

Khachatryan and Rihn

Volume 21, Issue 3, 2018

1. Introduction Recently, pollinator insects have become an important environmental topic due to decreasing populations and their global significance (Hanley et al., 2015; Klein et al., 2007; Wratten et al., 2012). Estimates indicate 70% of the world’s food crops rely on insect pollination (Klein et al., 2007) worth a total value of €153 billion (~$194.7 billion; Gallai et al., 2009). Additionally, pollinators contribute to biodiversity, wildlife food availability, and prevention of soil erosion and water runoff (Hanley et al., 2015; Wratten et al., 2012). Thus, declining pollinator populations has potential to harm global markets, food availability, and the environment.

Contrary to recent trends, in 2017, the U.S. Department of Agriculture’s National Agricultural Statistics Service reported that honeybee populations were increasing; however, there are many questions that have yet to be addressed (National Agricultural Statistics Service, 2017).1 To date, pollinator-related research has focused on causes of declining populations (Fairbrother et al., 2014) and overall economic and production impacts (Figueiredo Jr et al., 2016; Gallai et al., 2009; Klein et al., 2007) but relatively few studies address consumer perceptions of ‘pollinator friendly’ products (Rihn and Khachatryan, 2016; Wollaeger et al., 2015).

Consumer perceptions are important because they influence behavior, purchasing intentions (Costanigro et al., 2015; Stranieri and Banterle, 2015), and (in this case) pollinator insects’ access to habitat and nutrient sources (Breeze et al., 2015; Fairbrother et al., 2014; McIntyre and Hostetler, 2001). Evidence suggests consumers are confused about pollinator-related claims which can influence behavior (Wollaeger et al., 2015). For example, consumer perceptions and their intrinsic definitions influence their purchasing choices for eco-friendly foods (Campbell et al., 2015; Stranieri and Banterle, 2015). This may be problematic since consumer perceptions may not align with the actual product characteristics which can impact marketing efforts, labeling strategies, promotional message clarity, and policy effectiveness (Campbell et al., 2013, 2015; Stranieri and Banterle, 2015). This issue is amplified by ‘pollinator friendly’ being a credence attribute, which is not searchable unless in-store promotions (e.g. labels) are used. But, with the wide variety of pollinator-related labels (Rihn and Khachatryan, 2016), how do consumers perceive and define ‘pollinator friendly’ plants? We do not know.

The present study’s objective is to better understand consumers’ definitions of ‘pollinator friendly’ products by investigating the relationship between consumer factors (i.e. purchase interest, knowledge, demographics) and perceptions of pollinator friendly product attributes. Section 2 provides a brief review of relevant literature summarizing pollinator friendly product attributes, policy implications, and the existing pollinator-related consumer behavior research. Section 3 outlines the research methodology while Section 4 presents the results.

Lastly, Section 5 provides a brief discussion and concluding remarks.

2. Background: definitions, policy, and consumer behavior research Several definitions of practices that aid pollinators are available; however, very few definitions exist that clearly identify product characteristics that aid pollinators. The U.S. Forest Service (2015) and Xerces Society (2015) indicate that providing habitat and/or nutrients to pollinators constitutes ‘pollinator friendly’ products.

Several studies have identified product-specific (plant) traits related to aesthetics (Kendal et al., 2012), production practices (Gabriel and Tscharntke, 2007; Kiester et al.,1984), and physiological characteristics (Kiester et al., 1984), including: integrated pest management (IPM) strategies, organic production, natural production, environmentally friendly production, native origins, fragrant flowers, reduced/no pesticide use, and (often) the production of fruit, nectar, flowers, and/or pollen. Thus, a ‘pollinator friendly’ label can imply many different traits which may result in consumer confusion and reduce the label’s effectiveness.

Policy implications associated with defining and labeling pollinator friendly products are related to mandatory labeling or restrictions on use. Currently, a relevant debate is the mandatory labeling of neonicotinoid 1

Honeybees are a frequently studied pollinator insect, partially because they are very economically important due to commercial use in many operations producing crops that require insect pollination (Klein et al., 2007; Mwebaze et al., 2010).

International Food and Agribusiness Management Review

366

Khachatryan and Rihn

Volume 21, Issue 3, 2018

(neonic) pesticides. Neonic pesticides are systemic pesticides used to protect crops from insect predation.

The systemic nature of the pesticide means it is present within the entire plant including parts utilized by pollinators (pollen, nectar). This means neonics may affect pollinator insects’ health and behavior (Blacquiére et al., 2012). Currently, the UK government and several U.S. retailers (e.g. Home Depot) have restricted the use of neonic pesticides (Environmental Protection Agency, 2013). However, existing scientific research of the risks of neonic pesticides to pollinators is inconclusive (Barbosa et al., 2015; Blacquiére et al., 2012; Fairbrother et al., 2014; Hanley et al., 2015). For instance, Pilling et al. (2013) studied the affect of neonics in pollen over 4 years and found no differences between neonic-treated and control hives’ health.

Blacquiére et al. (2012) determined that the lethal and sublethal effects of neonics on pollinator insects only occurred in lab experiments but not in field experiments. Another study (Fairbrother et al., 2014) reported that Varroa mites and disease are the primary cause of worldwide bee loss. This finding is supported by the USDA’s report on honeybee health (National Agricultural Statistics Service, 2017). Regarding consumer behavior research, research shows that not many consumers are aware of neonic pesticides and many are confused about what ‘neonic-free’ labeling means (Rihn and Khachatryan, 2016; Wollaeger et al., 2015).

This is problematic since in order for a policy to be effective, consumers must understand the key message being communicated to them (Brécard, 2014). Without a clear understanding of consumers’ perceptions of products that aid pollinators the marketing potential and policy effectiveness relative to using pollinatorrelated labels is limited.

The effectiveness of pollinator-related labels is especially important because evidence suggests consumers are interested in pollinator-benefiting policies and products. In 2008, UK consumers were willing to pay £1.77 billion/year (~$3.52 billion) to support bee protection policies (Mwebaze et al., 2010). Breeze et al. (2015) determined UK tax payers were willing to pay £13.4 per year (~$21.61/year) to conserve wildflowers for pollinators. In 2012, U.S. consumers were willing to pay $4.78-6.64 billion to purchase beneficial plants or donate to butterfly conservation programs (Diffendorfer et al., 2014). While these studies emphasize broad consumer awareness of the importance of conserving pollinators, consumer perception studies are needed to understand the motives behind this behavior. Currently, there are two relevant consumer perception studies. Wollaeger et al. (2015) demonstrate consumers are more likely to purchase plants produced using ‘bee friendly’ production methods when compared to traditional insect management practices. Consumers’ purchasing frequency positively affected their awareness and knowledge of ‘bee friendly’ production methods.

Similarly, Rihn and Khachatryan (2016) found consumer knowledge affects purchasing behavior and that broad pollinator labels (e.g. ‘pollinator friendly’) are preferred to species-specific labels (e.g. ‘bee friendly’).

However, neither of the studies delved into consumers’ underlying perceptions and their accuracy. In this study we address this gap.

3. Methodology 3.1 Survey design An online survey was used to assess consumer perceptions of ‘pollinator friendly’ traits. In the survey, participants indicated from a pre-determined list which traits they considered to be beneficial to pollinator insects. Ornamental plants (in general) were selected as the product because they are key nutrient and habitat sources for pollinator insects (U.S. Forest Service, 2015; Xerces Society, 2015). In order to capture participants’ overall perceptions of ‘pollinator friendly’ traits, specific ornamental plant examples were not included. The 22 listed traits were developed from consultations with green industry professionals and existing literature. The list also included an ‘other, please list’ option to insure all potential traits were covered. Product traits were randomized to eliminate any order effect and participants were asked to ‘select all that apply.’

Likert scales were used to measure participants’ purchase interest for products that aid pollinators (1=not at all interested; 7=very interested) and knowledge of pollinator-related topics (1=not at all knowledgeable; 7=very knowledgeable; similar to Campbell et al. (2013) and Wollaeger et al. (2015)). Lastly, participants completed a standard set of socio-demographic questions.

International Food and Agribusiness Management Review

367

Khachatryan and Rihn

Volume 21, Issue 3, 2018

3.2 Sample summary A sample of 1,243 U.S. participants was collected during January 2015 using an online survey conducted

by Qualtrics, LLC. Participants were recruited from Qualtrics’ online panel. Online surveys have previously been used to collect data from a wide variety of participants in consumer perception studies (Campbell et al., 2014, 2015; Wollaeger et al., 2015). The average age of participants was 52 years old (Table 1). Males comprised 42% of the sample. Most (54%) of participants had less than a 4 year college degree. Participants’ 2014 household income was in the $51,000-60,000 range and the average household size was 2.6 people.

86% of the sample classified themselves as Caucasian/white. U.S. population statistics are provided for comparison purposes (U.S. Census Bureau, 2014). Overall, the sample over-represented older consumers, females, higher education levels, higher income households, and Caucasian/white consumers. Some of these results may be attributed to the study product (plants) where older women are the core consumers (Mason et al., 2008).

3.3 Econometric model The empirical model focused on the following themes: (1) understanding consumers’ perceptions of traits that aid pollinators; (2) how their interest in purchasing products to aid pollinators affected those perceptions; and (3) how their existing knowledge of pollinators/related topics and their socio-demographics influenced those perceptions. Following Campbell et al. (2013), a set of binary logit models and marginal effect estimates were used to determine the impact of the explanatory variables (i.e. knowledge, purchase interest, and sociodemographic characteristics) on their perceptions of ‘pollinator friendly’ traits.

Table 1. Summary statistics of U.S. respondents in an online survey exploring consumer perceptions of ‘pollinator friendly’ plant traits (n=1,243).

Age Gender Education1 Education2 Education3 Income

Household Ethnicity/race

Description

Sample mean (std. err.) U.S.1 mean

Age (in years) of participant 1=male; 0=female 1= less than 4 year degree; 0=otherwise 1= Bachelor’s degree and/or some graduate courses; 0=otherwise 1= Graduate degree; 0=otherwise 2014 gross household income 1≤$20k 2=$21k-30k 3=$31k-40k 4=$41k-50k 5=$51k-60k 6=$61k-70k 7=$71k-80k 8=$81k-90k 9=$91k-100k 10≥$100k Number of people in household 1=Caucasian/white; 0=otherwise

51.605 (30.670)*** 0.421 (0.494)*** 0.540 (0.499)*** 0.298 (0.458)***

37.6*** 0.490*** 0.707*** 0.189***

0.162 (0.368)*** 5.397 (3.070)***

0.104*** $51,939***

2.599 (1.345) 0.859 (0.349)***

2.54 0.781***

1 Adapted

from U.S. Census Bureau (2014). and * indicate significance at P-values ≤0.001, 0.010, and 0.050, respectively. Significance was determined using singlesample t-tests. ***, **,

International Food and Agribusiness Management Review

368

Khachatryan and Rihn

Volume 21, Issue 3, 2018

To accommodate the binary logit model, the traits were coded to equal 1 if selected and 0 if they were not selected.2 A binary logit model was analyzed for each trait. Specifically, the probability (Pi) of the ith participant selecting each trait can be represented by Pi =

1 β (1) 1+e-x’i

where xi represents participant i’s purchasing likelihood, knowledge, and socio-demographic variables and β indicates the estimated coefficients. Marginal effects were then estimated.3 The marginal effects indicate ‘the percent change given a one-unit increase from the mean’ for continuous variables while the dummy explanatory variables specify ‘the percent change for a move from the base attribute level to the level of interest’ (Campbell et al., 2015). Alternative models were also ran to test for heterogeneity but the results were similar and available from the corresponding author upon request.

4. Results 4.1 Exploratory analysis of perceptions Participants’ perceptions of different ‘pollinator friendly’ traits varied (Table 2). Most participants selected traits associated with flowers (i.e. pollen producing, flower producing, nectar producing, bright colored flowers, fragrant, and produces fruit) as being beneficial. This is likely due to consumers realizing that flowers are a main source of nutrients for adult pollinator insects (Kiester et al., 1984). However, bright colored flowers were not always beneficial to pollinators since plant breeding efforts emphasizing aesthetic characteristics can reduce nutrient availability (Landry, 2010). The aesthetic results may also reflect that consumers associate bright colors with aiding pollinators since 31.9% selected bright colored foliage. Additionally, 35.9% of participants selected native as a beneficial trait. This is not surprising since native plants have coevolved to aid native pollinators and are often preferred by pollinator insects over exotic plant species (Frankie et al., 2005). Production methods were also frequently selected (including environmentally friendly, pesticide free, grown using natural practices, organic, and grown using IPM strategies). A small percentage (1.9%) of consumers viewed aiding pollinators as a marketing gimmick.

Many of these findings are consistent with previous literature on products that aid pollinators. However, there were some inconsistencies as well. 30% of consumers associated locally grown with aiding pollinators and 22% indicated that a product classified as ‘pollinator friendly’ meant it was safer for humans (Table 2). To date, neither of these traits has been shown to positively affect pollinators. Increasing consumer interest and demand for local and sustainable products is likely responsible for these misperceptions. Local production is popular due to product acclimation to the local environment and consumers’ perceptions of local community benefits (i.e. economy, jobs, etc.) (Campbell et al., 2014; Wehry et al., 2007). Interest in sustainably produced plants (i.e. ones perceived as ‘safer for humans’) is often due to human and environmental health concerns (Campbell et al., 2014). If consumers perceive ‘pollinator friendly’ positively, they may project additional positive traits (such as local and safe for humans) onto those products to enhance their benefits and attractiveness. Alternatively, consumers may not be knowledgeable about pollinator friendly products and therefore used their personal preferences and past experiences to shape their perceptions (Campbell et al., 2015; Wollaeger et al., 2015).

These results provide an overview of consumer perceptions of product traits that aid pollinators; however, additional quantitative results need to be considered in order to make inferences from the data. In the next section, the influence of purchase interest, knowledge, and socio-demographic variables on consumer

2

For instance, if a participant indicated ‘fragrant’ was a trait that aids pollinators then fragrant equals 1, conversely, if s/he did not select fragrant it equals 0.

3 Due to limited space, only the marginal effect estimates are provided in the manuscript. The binary logit estimates are available upon request from the corresponding author.

International Food and Agribusiness Management Review

369

Khachatryan and Rihn

Volume 21, Issue 3, 2018

Table 2. Percentage of respondents who selected accurate and inaccurate ‘pollinator friendly’ plant traits (n=1,243).

Trait definition

Percent selected1

Accurate trait2

Grown using integrated pest management strategies Organic Grown using natural practices Native Fragrant Pesticide free Environmentally friendly Nectar producing Flower producing Pollen producing Produces fruit Other Genetically modified Safer for humans Marketing gimmick More expensive Greenhouse grown Locally grown Bright colored foliage Pesticides were used Bright colored flowers None of the above

12.0 25.0 33.6 35.9 39.7 40.9 40.9 57.1 59.6 61.5 32.4 1.1 2.1 22.2 1.9 7.1 8.2 30.2 31.9 2.3 48.9 5.1

Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Varies3 – No No No No No No No No Varies4 No

1

Respondents were instructed to ‘select all that apply’; hence, the percentages do not sum to 100%.

A definition of product and/or traits that aid pollinator insects was not available. Therefore, green industry professionals and existing literature were used to identify beneficial traits. Traits that improve pollinator health include: integrated pest management strategies (Kiester et al., 1984), natives (Frankie et al., 2005), organic systems (Gabriel and Tscharntke, 2007; Morandin and Winston, 2005), environmentally friendly, and natural practices (Frankie et al., 2005). Additionally, plants have coevolved with pollinator species to attract specific pollinators through fragrance, flower morphology, and nutrient sources (i.e. pollen and nectar) (Kiester et al., 1984). Conversely, pesticides have been shown to negatively influence pollinator health (Fairbrother et al., 2014; Hanley et al., 2015; Pimentel, 2005).

3 Not all fruit producing crops require insect pollination; however, several fruit producing crops rely on insect pollination (Gallai et al., 2009; Klein et al., 2007) and 23% of fruits are highly economically vulnerable to pollinator population loss (Potts et al., 2010).

Therefore, the ‘fruit producing’ trait is listed as ‘varies’.

4 Although flowers are beneficial to pollinator insects (Kiester et al., 1984), bright colored, long-lasting flowers are often bred at the expense of the plant’s reproductive organs (including pollen and nectar) which can be detrimental to pollinators (Landry, 2010).

Therefore the ‘bright colored flowers’ trait is listed as ‘varies’ since it can vary between species and cultivars.

2

perceptions of products that aid pollinators using the marginal effect estimates from the binary logit models are discussed.

4.2 Marginal effects for accurate traits Marginal effect estimates provide insights on why consumers perceive certain traits as beneficial and not others. For ease of interpretation, accurate traits were divided into production method traits (Table 3) and product traits (Table 4). Consumers who were interested in purchasing products to aid pollinators had an increased probability of correctly identifying beneficial production methods (Table 3). Consumers who were knowledgeable about neonic pesticides were 9.7% more likely to select organic production methods as being International Food and Agribusiness Management Review

370

Khachatryan and Rihn

Volume 21, Issue 3, 2018

Table 3. Marginal Effect estimates from binary logit models exploring consumer perceptions of accurate â&#x20AC;&#x2DC;pollinator friendlyâ&#x20AC;&#x2122; production method traits (n=1,243).

Purchase interest1 Pollinator friendly plants Knowledge Neonicotinoid pesticides Landscape, garden, plants Environmental stewardship Pollinators (in general) Pollinator health Bee keeping Plants that improve pollinator health Pollinator friendly features Entomology Agriculture Socio-demographics Age Gender Income Household Education Ethnicity Log likelihood LR chi2 Prob>chi2 Pseudo R2

Integrated pest management

Organic

Natural practices

0.044 (0.011)***

0.021 (0.021) 0.014 (0.009) 0.009 (0.007) 0.011 (0.010) 0.006 (0.011) 0.006 (0.007) 0.013 (0.010)

0.097 (0.031)*** 0.022 (0.012) 0.017 (0.010) -0.021 (0.015) 0.017 (0.016) 0.007 (0.010) 0.015 (0.015)

0.024 (0.037) -0.000 (0.013) 0.009 (0.012) -0.001 (0.017) 0.001 (0.018) 0.018 (0.012) -0.006 (0.017)

0.032 (0.040) -0.002 (0.014) 0.028 (0.012)** 0.025 (0.017) 0.009 (0.020) 0.008 (0.013) 0.002 (0.018)

0.073 (0.040) -0.008 (0.014) 0.015 (0.013) -0.027 (0.018) 0.020 (0.020) -0.013 (0.013) 0.026 (0.018)

-0.016 (0.015) -0.013 (0.011) 0.003 (0.011)

-0.012 (0.018) -0.011 (0.013) -0.003 (0.012)

-0.029 (0.019) -0.026 (0.013) -0.008 (0.013)

-0.006 (0.019) -0.004 (0.013) 0.006 (0.013)

-0.001 (0.001) -0.050 (0.026) -0.003 (0.005) -0.016 (0.010) 0.018 (0.008)** -0.114 (0.034)*** -635.419 113.03 <0.001 0.0817

0.000 (0.000) -0.043 (0.029) -0.013 (0.005)** -0.012 (0.011) 0.008 (0.009) -0.053 (0.040) -748.108 74.27 <0.001 0.0473

0.000 (0.000) 0.026 (0.017) -0.000 (0.003) -0.010 (0.007) 0.006 (0.006) 0.036 (0.028) -422.835 58.30 <0.001 0.0645

***, **,

0.063 (0.013)***

Environmentally friendly

0.013 (0.008)

-0.020 (0.010)* -0.003 (0.007) -0.009 (0.008)

0.063 (0.012)***

Pesticide free

0.000 (0.000) -0.014 (0.030) -0.006 (0.005) -0.010 (0.012) -0.005 (0.010) -0.040 (0.043) -797.831 66.67 <0.001 0.0401

0.070 (0.013)***

0.002 (0.001)* -0.070 (0.031)* -0.018 (0.006)** 0.005 (0.012) 0.024 (0.010)* -0.027 (0.044) -783.438 95.45 <0.001 0.0574

and * indicate significance at P-values â&#x2030;¤0.001, 0.010, and 0.050, respectively; standard errors are presented in the parentheses.

Base categories include: not interested in purchasing products(s) to aid pollinators, not knowledgeable about neonicotinoid pesticides, not knowledgeable about landscape/garden/plants, not knowledgeable about environmental stewardship, not knowledgeable about pollinators (in general), not knowledgeable about pollinator health, not knowledgeable about bee keeping, not knowledgeable about plants that improve pollinator health, not knowledgeable about pollinator friendly features, not knowledgeable about entomology, not knowledgeable about agriculture, female, graduate degree, and other ethnicity.

1

International Food and Agribusiness Management Review

371

Khachatryan and Rihn

Volume 21, Issue 3, 2018

Table 4. Marginal effect estimates from binary logit models exploring consumer perceptions of accurate ‘pollinator friendly’ plant traits (n=1,243).

Purchase interest1 Pollinator friendly plants Knowledge Neonicotinoid pesticides Landscape, garden, plants Environmental stewardship Pollinators (in general) Pollinator health Bee keeping Plants that improve pollinator health Pollinator friendly features Entomology Agriculture Socio-demographics Age Gender Income Household Education Ethnicity Log likelihood LR chi2 Prob>chi2 Pseudo R2 ***, **,

Nectar producing

Flower producing

0.081 (0.013)***

0.078 (0.013)***

Pollen producing 0.068 (0.012)***

Fragrant 0.043 (0.013)***

Native 0.065 (0.013)***

Fruit producing2 0.073 (0.012)***

-0.079 (0.041) -0.010 (0.014) 0.026 (0.013)* 0.003 (0.018) -0.008 (0.020) 0.009 (0.013) -0.007 (0.019)

-0.059 (0.041) 0.047 (0.014)*** 0.016 (0.013) 0.028 (0.018) -0.014 (0.020) -0.007 (0.013) 0.003 (0.019)

-0.066 (0.041) 0.002 (0.014) 0.027 (0.013)* 0.031 (0.018) 0.000 (0.020) 0.006 (0.013) -0.006 (0.018)

0.013 (0.039) 0.010 (0.014) 0.012 (0.012) 0.011 (0.017) 0.002 (0.019) -0.006 (0.012) 0.013 (0.018)

-0.018 (0.038) -0.003 (0.014) 0.025 (0.012)* 0.013 (0.017) 0.003 (0.019) -0.018 (0.012) 0.008 (0.017)

0.038 (0.037) -0.002 (0.013) -0.012 (0.012) 0.09 (0.016) 0.017 (0.018) 0.003 (0.012) 0.002 (0.017)

-0.019 (0.019) -0.006 (0.014) 0.024 (0.013)

-0.021 (0.019) -0.025 (0.014) -0.008 (0.013)

-0.020 (0.019) -0.017 (0.014) 0.020 (0.013)*

-0.021 (0.018) -0.017 (0.013) 0.020 (0.013)

-0.013 (0.018) 0.013 (0.013) -0.008 (0.013)

-0.002 (0.017) -0.040 (0.013)** 0.027 (0.012)*

0.000 (0.001) -0.017 (0.031) -0.006 (0.006) -0.017 (0.012) 0.012 (0.010) 0.091 (0.043)* -792.311 93.55 <0.001 0.0557

0.001 (0.001) -0.022 (0.031) -0.010 (0.006) -0.003 (0.012) 0.008 (0.010) 0.134 (0.043)** -767.045 123.68 <0.001 0.0746

0.004 (0.001)*** -0.020 (0.031) -0.000 (0.005) 0.007 (0.012) 0.005 (0.010) 0.072 (0.042) -752.198 129.80 <0.001 0.0794

-0.000 (0.001) -0.019 (0.030) 0.003 (0.005) -0.019 (0.012) 0.009 (0.010) 0.076 (0.044) -798.698 54.74 <0.001 0.0331

0.000 (0.000) -0.050 (0.029) -0.002 (0.005) -0.019 (0.011) 0.019 (0.009)* 0.016 (0.042) -761.096 83.40 <0.001 0.0519

0.001 (0.001) 0.006 (0.029) -0.010 (0.005)* -0.000 (0.011) 0.015 (0.009) -0.072 (0.039) -728.389 92.57 <0.001 0.0597

and * indicate significance at P-values ≤0.001, 0.010, and 0.050, respectively; standard errors are presented in the parentheses.

1 Base categories include: not interested in purchasing product(s) to aid pollinators, not knowledgeable about neonicotinoid pesticides, not knowledgeable about landscape/garden/plants, not

knowledgeable about environmental stewardship, not knowledgeable about pollinators (in general), not knowledgeable about pollinator health, not knowledgeable about bee keeping, not knowledgeable about plants that improve pollinator health, not knowledgeable about pollinator friendly features, not knowledgeable about entomology, not knowledgeable about agriculture, female, graduate degree, and other ethnicity.

2 In this table, the ‘fruit producing’ trait is presented with the accurate traits due to the importance of honeybees in the production of economically important fruit crops (e.g. blueberries, citrus, strawberries, etc.; Gallai et al., 2009; Klein et al., 2007).

International Food and Agribusiness Management Review

372

Khachatryan and Rihn

Volume 21, Issue 3, 2018

‘pollinator friendly’. Additionally, consumers who were knowledgeable about environmental stewardship were 2.8% more likely to indicate pesticide free as a production practice that aids pollinators. Interestingly, being knowledgeable about pollinator friendly features reduced the likelihood of selecting IPM by 2.0%.

This may reflect low consumer knowledge about what constitutes IPM strategies. Regarding the influence of socio-demographic variables, older participants were 0.23% more likely to select environmentally friendly as a beneficial trait, while males and consumers with higher incomes were 7.0% and 1.8% less likely to select environmentally friendly. Higher income individuals were 1.3% less likely to select natural practices.

More educated respondents were 1.8% and 2.4% more likely to select organic and environmentally friendly production methods. Caucasian/white consumers were 11.4% less likely to indicate organic practices.

Consumers’ purchase interest also increased their probability of selecting accurate product traits (Table 4).

Consumers who were knowledgeable about landscapes, gardens, and plants were 4.7% more likely to select flower producing as a beneficial trait. Plant aesthetics were a primary attribute when making purchasing decisions (Kelley et al., 2001; Kendal et al., 2012). As a result, this group of consumers may have an increased interested in aesthetic characteristics. Consumers knowledgeable in environmental stewardship were 2.7% more likely to select pollen producing and 2.5% more likely to select native. Entomology knowledgeable consumers were 4.0% less likely to select fruit producing. Consumers knowledgeable in agriculture were more likely to select pollen (2.0%) and fruit producing (2.7%) traits as beneficial. Older participants were also more likely to select pollen producing (0.4%). Individuals with higher incomes were less likely to select fruit producing (-1.0%). Individuals who had obtained a higher education level were 1.9% more likely to select native as a beneficial trait. Caucasian/white consumers had a higher probability of selecting the nectar (9.1%) and flower producing (13.4%) traits.

4.3 Marginal effects for inaccurate traits Regarding inaccurate traits, consumers who were interested in purchasing pollinator friendly products did not perceive ‘pollinator friendly’ as a marketing gimmick (Table 5). This is intuitive because if consumers are interested in purchasing products that aid pollinators, they are more likely to actively seek out those products rather than discount the information as a marketing gimmick. Neonic pesticide knowledgeable consumers were 0.8% more likely to inaccurately select genetically modified. Consumers who were knowledgeable about pollinators were 3.0% less likely to inaccurately select safer for humans. Consumers knowledgeable about bee keeping were 1.2% more likely to select expensive. Consumers interested in purchasing pollinator friendly products were more likely to select bright colored foliage (4.3%) and flowers (7.2%) as traits that aid pollinators. Consumers knowledgeable about neonicotinoid pesticides were 9.2% less likely to select ‘bright colored flowers’. For socio-demographics, age negatively influenced the probability of selecting genetically modified and marketing gimmick. Males were less likely to select bright colored foliage (-7.7%) and flowers (-8.8%). Caucasian/white consumers were 8.6% less likely to select safer for humans but 8.7% more likely to select bright colored foliage and 9.4% more likely to select bright colored flowers.

Consumers’ increased purchase interest improves the probability of inaccurately selecting locally grown by 5.0% (Table 6). Knowledge about pollinator friendly features or agriculture increased consumers’ likelihood of selecting greenhouse grown by 2.9 and 1.4%, respectively. Purchase interest negatively impacted the probability of selecting ‘none of the above’. Age negatively affected the likelihood of selecting ‘pesticides were used’.

5. Discussion: emerging consumer perception patterns Cumulatively, when examining consumers’ accurate and inaccurate perceptions and how purchase interest, knowledge, and socio-demographics influence these perceptions, several interesting patterns emerge (Tables 3-6). First, increased interest in purchasing products to aid pollinators results in the consumer selecting more positive traits even if they are not accurate (e.g. locally grown). A potential explanation for this result is that if consumers perceive pollinator beneficial products positively (as indicated by increased purchase interest) International Food and Agribusiness Management Review

373

Khachatryan and Rihn

Volume 21, Issue 3, 2018

Table 5. Marginal effect estimates from binary logit models exploring consumer perceptions of inaccurate â&#x20AC;&#x2DC;pollinator friendlyâ&#x20AC;&#x2122; plant traits (n=1,243).

Genetically modified

Purchase interest1 Pollinator friendly plants -0.002 (0.001) Knowledge Neonicotinoid pesticides 0.008 (0.004)* Landscape, garden, plants 0.001 (0.002) Environmental stewardship -0.000 (0.001) Pollinators (in general) 0.001 (0.002) Pollinator health 0.003 (0.002) Bee keeping 0.001 (0.001) Plants that improve -0.001 (0.001) pollinator health Pollinator friendly features -0.002 (0.002) Entomology 0.000 (0.001) Agriculture 0.001 (0.001) Socio-demographics Age -0.000 (0.000)** Gender 0.004 (0.003) Income 0.001 (0.001) Household -0.001 (0.001) Education -0.001 (0.001) Ethnicity -0.002 (0.003) Log likelihood -92.005 LR chi2 59.94 2 Prob>chi <0.001 Pseudo R2 0.2379

Safer for humans

Marketing gimmick Expensive

Bright colored foliage

-0.005 (0.002)**

-0.002 (0.006)

0.054 (0.031) -0.019 (0.011) 0.001 (0.010) -0.030 (0.015)* 0.004 (0.016) 0.010 (0.010) 0.026 (0.015)

0.006 (0.005) 0.002 (0.002) -0.000 (0.002) -0.002 (0.003) 0.002 (0.003) 0.002 (0.002) -0.002 (0.002)

0.031 (0.018) 0.003 (0.007) -0.001 (0.006) -0.012 (0.008) 0.005 (0.009) 0.012 (0.006)* -0.014 (0.008)

-0.006 (0.037) 0.015 (0.013) -0.001 (0.011) 0.005 (0.016) 0.002 (0.018) 0.010 (0.012) -0.005 (0.017)

-0.092 (0.042)* 0.013 (0.014) 0.003 (0.013) 0.034 (0.018) -0.008 (0.020) -0.011 (0.013) 0.005 (0.019)

0.017 (0.015) -0.019 (0.011) 0.015 (0.011)

0.002 (0.003) 0.001 (0.002) -0.002 (0.002)

0.012 (0.008) -0.005 (0.006) 0.001 (0.006)

-0.006 (0.017) -0.013 (0.012) 0.020 (0.012)

-0.005 (0.019) -0.008 (0.014) 0.001 (0.013)

0.000 (0.000) -0.004 (0.025) -0.006 (0.005) 0.007 (0.009) -0.002 (0.008) -0.086 (0.032)** -629.953 41.83 <0.001 0.0321

-0.000 (0.000)* 0.000 (0.004) 0.001 (0.001) 0.001 (0.002) 0.000 (0.001) 0.010 (0.007) -99.110 30.36 0.0239 0.1328

0.000 (0.000) -0.077 (0.028)** -0.001 (0.005) -0.005 (0.011) -0.006 (0.009) 0.087 (0.042)* -741.059 58.30 <0.001 0.0378

0.001 (0.001) -0.088 (0.031)** 0.005 (0.006) -0.008 (0.012) 0.006 (0.010) 0.094 (0.045)* -801.910 99.58 <0.001 0.0585

0.015 (0.010)

0.000 (0.000) -0.001 (0.014) 0.001 (0.003) 0.008 (0.005) 0.003 (0.005) 0.025 (0.022) -304.100 20.25 0.262 0.0322

***, **,

0.043 (0.012)***

Bright colored flowers 0.072 (0.013)***

and * indicate significance at P-values â&#x2030;¤0.001, 0.010, and 0.050, respectively; standard errors are presented in the parentheses. categories include: not interested in purchasing product(s) to aid pollinators, not knowledgeable about neonicotinoid pesticides, not knowledgeable about landscape/garden/plants, not knowledgeable about environmental stewardship, not knowledgeable about pollinators (in general), not knowledgeable about pollinator health, not knowledgeable about bee keeping, not knowledgeable about plants that improve pollinator health, not knowledgeable about pollinator friendly features, not knowledgeable about entomology, not knowledgeable about agriculture, female, graduate degree, and other ethnicity.

1Base

International Food and Agribusiness Management Review

374

Khachatryan and Rihn

Volume 21, Issue 3, 2018

Table 6. Marginal effect estimates from binary logit models exploring consumer perceptions of inaccurate ‘pollinator friendly’ production method and plant traits (n=1,243).

Greenhouse grown interest1

Purchase Pollinator friendly plants 0.005 (0.006) Knowledge Neonicotinoid pesticides 0.014 (0.019) Landscape, garden, plants 0.002 (0.007) Environmental stewardship -0.004 (0.006) Pollinators (in general) -0.016 (0.009) Pollinator health -0.001 (0.009) Bee keeping 0.004 (0.006) Plants that improve -0.013 (0.009) pollinator health Pollinator friendly features 0.029 (0.009)*** Entomology -0.003 (0.006) Agriculture 0.014 (0.006)* Socio-demographics Age -0.000 (0.001) Gender 0.000 (0.015) Income -0.001 (0.003) Household 0.002 (0.006) Education -0.002 (0.005) Ethnicity -0.021 (0.019) Log likelihood -331.988 LR chi2 34.25 Prob>chi2 0.0078 Pseudo R2 0.0491 ***, **,

Locally grown

Pesticides were used

0.050 (0.012)***

Other2

None of the above

-0.001 (0.002)

-0.002 (0.001)

-0.008 (0.002)***

-0.018 (0.036) -0.008 (0.013) 0.015 (0.011) 0.009 (0.016) -0.014 (0.018) -0.013 (0.012) 0.018 (0.016)

0.003 (0.005) 0.003 (0.002) 0.003 (0.002) -0.004 (0.003) 0.001 (0.003) 0.001 (0.002) -0.000 (0.002)

-0.004 (0.006) 0.001 (0.002) 0.001 (0.002) 0.001 (0.002) 0.003 (0.002) 0.000 (0.002) -0.002 (0.002)

-0.000 (0.011) -0.001 (0.002) -0.005 (0.003) -0.006 (0.004) 0.001 (0.005) 0.003 (0.003) -0.002 (0.004)

0.015 (0.017) -0.003 (0.012) -0.003 (0.012)

0.003 (0.003) 0.001 (0.002) -0.000 (0.002)

-0.001 (0.002) 0.001 (0.002) 0.000 (0.002)

-0.003 (0.004) 0.002 (0.003) -0.001 (0.003)

-0.000 (0.000)* -0.002 (0.004) -0.001 (0.001) 0.000 (0.001) 0.004 (0.006) 0.003 (0.007) -113.680 39.77 0.001 0.1489

0.000 (0.000) 0.005 (0.004) -0.001 (0.001) 0.002 (0.001) 0.002 (0.001) 0.007 (0.008) -69.943 13.25 0.719 0.0865

-0.000 (0.000) -0.000 (0.006) 0.001 (0.001) 0.001 (0.002) -0.000 (0.002) -0.001 (0.008) -186.773 111.75 <0.001 0.2303

0.000 (0.000) -0.035 (0.028) -0.005 (0.005) -0.021 (0.011) 0.006 (0.009) -0.018 (0.039) -723.742 59.57 <0.001 0.0395

and * indicate significance at P-values ≤0.001, 0.010, and 0.050, respectively; standard errors are presented in the parentheses.

1 Base categories include: not interested in purchasing product(s) to aid pollinators, not knowledgeable about neonicotinoid pesticides, not knowledgeable about landscape/garden/plants, not

knowledgeable about environmental stewardship, not knowledgeable about pollinators (in general), not knowledgeable about pollinator health, not knowledgeable about bee keeping, not knowledgeable about plants that improve pollinator health, not knowledgeable about pollinator friendly features, not knowledgeable about entomology, not knowledgeable about agriculture, female, graduate degree, and other ethnicity.

2 The ‘other’ trait allowed participants to note traits that were not included in the provided list. Other traits were inconsistent with being solely in the accurate or inaccurate categories.

Participants’ other list included: do not know (n=6), pet safe (n=1), larvae food plants (i.e. herbs; n=1), or blank (n=6).

International Food and Agribusiness Management Review

375

Khachatryan and Rihn

Volume 21, Issue 3, 2018

they associate it with other positive traits (much like the ‘halo effect’ discussed by Wu and Petroshuis (1987)).

Thus they are more likely to have positive opinions regardless of accuracy, which sequentially influences their product choices.

There are advantages and disadvantages to this phenomenon. Advantages include the opportunity to promote products that aid pollinators which increases product availability and can be leveraged to generate consumer interest in those products. In turn, this may lead to increased profits and greater abundance of pollinator friendly products in the environment which may have substantial long-term impacts on pollinator insect populations (Frankie et al., 2005; Hanley et al., 2015). However, if consumers’ obtain greater satisfaction from bright colored foliage and flowers (depending on species/cultivar) than from pollinator friendly traits, the non-beneficial traits may outweigh the beneficial traits. This may be problematic since plant aesthetics are a primary purchase driver but do not always benefit pollinators (Kelley et al., 2001; Kendal et al., 2012; Landry, 2010). Pollinator-related labels may be able to overcome this issue; however, to what extent is unknown and outside the scope of this study.

Consumers’ existing knowledge also influences their perceptions of what constitutes a product that aids pollinators. Results imply that existing knowledge and interests strongly affect consumer perceptions which, in turn, influence their choices (Campbell et al., 2013; Wollaeger et al., 2015). For instance, consumers knowledgeable in landscaping, gardens, and plants select flower producing (an important aesthetic trait).

Environmental stewardship knowledgeable consumers primarily select environment friendly attributes (pesticide free, pollen producing). Similarly, neonic pesticide knowledgeable consumers avoid selecting pesticide containing options an (as reflected through the selection of organic) which is consistent with Wollaeger et al. (2015). These patterns provide insights into how consumers’ existing knowledge influences their perceptions which can be used to increase awareness of traits that positively affect pollinator health.

Regarding socio-demographic variables, age appeared to have the most impact with older participants having a more accurate perception of traits that aid pollinators. This is not surprising considering older consumers are the core consumers of plants (Mason et al., 2008), meaning they are likely more familiar with the products and their impact on pollinators. Education also appeared to increase the accuracy of participants’ selection of traits that benefit pollinators.

In conclusion, research has shown consumers are interested in pollinator conservation measures but, to date, very few studies investigate consumer perceptions of products that aid pollinators. We found consumers’ interest in purchasing pollinator friendly products, existing knowledge, and socio-demographics all contribute to their perceptions of beneficial traits. Overall, findings indicate some confusion exists about what traits are actually beneficial to pollinator insects. However, results should be interpreted cautiously since there are unobserved individual/consumer characteristics that (due to data limitations) were not included in the analyses.

Though the study results are consist with previous studies addressing the impact of consumer knowledge on behavior (Campbell et al., 2013; Rihn and Khachatryan, 2016) and consumer behavior toward traits that benefit pollinators (Wollaeger et al., 2015) indicating robustness of the present results. Future studies incorporating additional variables and experimental methods (e.g. incorporation of live plants, exposure to pollinator-related news in mass media, treatment groups, etc.) could further test the robustness of results.

There is an opportunity for researchers to further quantify how difference consumer characteristics influences their definitions of ‘pollinator friendly’ products. Furthermore, policy makers and industry stakeholders could benefit from educating consumers about pollinator beneficial traits and use in-store promotions to influence consumer behavior toward those items. Ultimately, this could positively influence demand for pollinator beneficial products and improve pollinator health through increased availability of beneficial products.

International Food and Agribusiness Management Review

376

Khachatryan and Rihn

Volume 21, Issue 3, 2018

References Barbosa, W.F., G. Smagghe and R.N.C. Guedes. 2015. Pesticides and reduced-risk insecticides, native bees and pantropical stingless bees: pitfalls and perspectives. Pest Management Science 71: 1049-1053.

Blacquiére, T., G. Smagghe, C.A.M. van Gestel and V. Mommaerts. 2012. Neonicotinoids in bees: a review on concentrations, side-effects and risk assessment. Ecotoxicology 4: 973-992.

Brécard, D. 2014. Consumer confusion over the profusion of eco-labels: lessons from a double differentiation model. Resource and Energy Economics 37: 64-84.

Breeze, T.D., A.P. Bailey, S.G. Potts and K.G. Balcombe. 2015. A stated preference valuation of the nonmarket benefits of pollination services in the UK. Ecological Economics 111: 76-85.

Campbell, B., H. Khachatryan, B.K. Behe, J.H. Dennis and C.R. Hall. 2014. U.S. and Canadian consumer perceptions of local and organic terminology. International Food and Agribusiness Management Review 17: 21-40.

Campbell, B., H. Khachatryan, B.K. Behe, J.H. Dennis and C.R. Hall. 2015. Consumer perceptions of ecofriendly and sustainable terms. Agricultural and Resource Economics Review 44: 21-34.

Campbell, B., S. Mhlanga and I. Lesschaeve. 2013. Perception versus reality: Canadian consumer views of local and organic. Canadian Journal of Agricultural Economics 61: 531-558.

Costanigro, M., O. Deselnicu and S. Kroll. 2015. Food beliefs: elicitation, estimation and implications for labeling policy. Journal of Agricultural Economics 66: 108-128.

Diffendorfer, J.E., J.B. Loomis, L. Ries, K. Oberhauser, L. Lopez-Hoffman, D. Semmens, B. Semmens, B. Butterfield, K. Bagstad, J. Goldstein, R. Widerholt, B. Mattsson and W.E. Thogmartin. 2014. National valuation of monarch butterflies indicates an untapped potential for incentive-based conservation.

Conservation Letters 7: 253-262.

Environmental Protection Agency. 2013. Colony collapse disorder: European bans on neonicotinoid pesticides.

EPA, U.S. Department of Agriculture, Washington, D.C., USA. Fairbrother, A., J. Purdy, T. Anderson and R. Fell. 2014. Risks of neonicotinoid insecticides to honeybees.

Environmental Toxicology and Chemistry 33: 719-731.

Figueiredo Jr, H.S.D., M.P.M. Meuwissen, J.D.A. Filho and A.G.J.M.O. Lansink. 2016. Evaluating strategies for honey value chains in Brazil using a value chain structure-conduct-performance (SCP) framework.

International Food and Agribusiness Management Review 19: 225-250.

Frankie, G.W., R.W. Thorp, M. Schindler, J. Hernandez, B. Ertter and M. Rizzardi. 2005. Ecological patterns of bees and their host ornamental flowers in two northern California cities. Journal of Kansas Entomological Society 78: 227-246.

Gabriel, D. and T. Tscharntke. 2007. Insect pollinated plants benefit from organic farming. Agriculture, Ecosystems and Environment 118: 43-48.

Gallai, N., J.-M.Salles, J. Settele and B.E. Vaissiere. 2009. Economic valuation of the volunerability of world agriculture confronted with pollinator decline. Ecological Economics 68: 810-821.

Hanley, N., T.D. Breeze, C. Ellis and D. Goulson. 2015. Measuring the economic value of pollination services: principles, evidence and knowledge gaps. Ecosystem Services 142: 137-143.

Kelley, K.M., B.K. Behe, J.K. Biernbaum and K.L. Poff. 2001. Consumer preference for edible-flower color, container size and price. HortScience 36: 801-804.

Kendal, D., K.J.H. Williams and N.S.G. Williams. 2012. Plant traits link people’s plant preferences to the composition of their gardens. Landscape and Urban Planning 105: 34-42.

Kiester, A.R., R. Lande and D.W. Schemske. 1984. Models of coevolution and speciation in plants and their pollinators. An American Naturalist 124: 220-243.

Klein, A.-M., B.E. Vaissiere, J.H. Cane, I. Steffan-Dewenter, S.A. Cunningham, C. Kremen and T. Tscharntke.

2007. Importance of pollinators in changing landscapes for world crops. Proceedings of the Royal Society B: Biological Sciences 274: 303-313.

Landry, J.L. 2010. Identifying ‘pollinator-friendly’ cultivars for gardens and greenroofs. MSc thesis, The Pennsylvania State University, State College, PA, USA. Mason, S.C., T.W. Starman, R.D. Lineberger and B.K. Behe. 2008. Consumer preference for price, color harmony and care information of container gardens. HortScience 43: 380-384.

International Food and Agribusiness Management Review

377

Khachatryan and Rihn

Volume 21, Issue 3, 2018

McIntyre, N.E. and M.E. Hostetler. 2001. Effects of urban land use on pollinator (Hymenoptera: Apoidea) communities in a desert metropolis. Basic Applied Ecology 2: 209-218.

Morandin, L.A. and M.L. Winston. 2005. Wild bee abundance and seed production in conventional, organic and genetically modified canola. Ecological Applications 15: 871-881.

Mwebaze, P., G.C. Marris, G.E. Budge, M. Brown, S.G. Potts, T.D. Breeze and A. Macleod. 2010. Quantifying the value of ecosystem services: a case study of honeybee population in the UK. In: 12th Annual BIOECON conference â&#x20AC;&#x2DC;from the wealth of nations to the wealth of nature: rethinking economic growthâ&#x20AC;&#x2122;, Venice, Italy.

National Agricultural Statistics Service. 2017. Honey Bee Colonies. U.S. Department of Agriculture, Washington D.C., USA. Pilling, E., P. Campbell, M. Coulson, N. Ruddle and I. Tornier. 2013. A four-year field program investigating long-term effects of repeated exposure of honey bee colonies to flowering crops treated with thiamethoxam. PLoS ONE 8: e77193.

Pimentel, D. 2005. Environmental and economic costs of the application of pesticides primarily in the United States. Environment, Development and Sustainability 7: 229-252.

Potts, S.G., J.C. Biesmeijer, C. Kremen, P. Neumann, O. Schweiger and W.E. Kunin. 2010. Global pollinator declines: trends, impacts and drivers. Trends in Ecology and Evolution 25(6): 345-353.

Rihn, A. and H. Khachatryan. 2016. Does consumer awareness of neonicotinoid pesticides influence their preferences for plants? HortScience 51: 388-393.

Stranieri, S. and A. Banterle. 2015. Consumer interest in meat labelled attributes: who cares? International Food and Agribusiness Management Review 18: 21-38.

U.S. Census Bureau. 2014. Topics. U.S. Department of Commerce, Suitland, MA, USA. U.S. Forest Service. 2015. Pollinator friendly practices. U.S. Department of Agriculture, Washington D.C., USA. Wehry, R.H., K.M. Kelley, R.D. Berghage and J.C. Sellmer. 2007. Caputring consumer preferences and interest in developing a state plant promotional program. HortScience 42: 574-580.

Wollaeger, H.M., K.L. Getter and B.K. Behe. 2015. Consumer preferences for traditional, neonicotinoidfree, bee-friendly, or biological control pest management practices on floriculture crops. HortScience 50: 721-732.

Wratten, S.D., M. Gillespie, A. Decortye, E. Mader and N. Desneux. 2012. Pollinator habitat enhancement: benefits to other ecosystem services. Agriculture, Ecosystems and Environment 159: 112-122.

Wu, B.T.W. and S.M. Petroshuis. 1987. The halo effect in store image measurement. Journal of the Academy of Marketing Science 15: 44-51.

Xerces Society. 2015. Gardens. The Xerces Society, Portland, OR, UDA. Available at http://www.xerces. org/pollinator-conservation/gardens.

International Food and Agribusiness Management Review

378

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2016.0163 Received: 30 September 2016 / Accepted: 28 December 2017

The role of members’ commitment on agri-food co-operatives’ capitalization, innovation and performance RESEARCH ARTICLE Gustavo Marcos-Matas a, Arianna Ruggerib and Rino Ghelfic aAssistant

Professor, Public University of Navarra. Department of Business Management, Edificio Departamental Los Madroños, Campus Arrosadía, 31006 Pamplona, Spain

bResearch

Associate and cProfessor, University of Bologna, Department of Agricultural Sciences, Facoltà di Agraria, Viale Fanin 50, 40127 Bologna, Italy

Abstract Undercapitalization has been recognized as a problem affecting Italian co-operatives to perform in modern agri-food markets. An empirical study on 50 Italian agri-food co-operatives was carried out to investigate co-operatives members’ commitment capability to impact on the level of capitalization. The level of capitalization was also investigated as a mean to influence co-operatives’ innovation and subsequently their performance. The results show how co-operatives with more committed memberships display higher levels of capitalization and that capitalization positively relates with innovation levels. The latter is also confirmed to enhance co-operatives’ performance. These results are relevant from a managerial point of view as they reveal the importance of members’ commitment in this particular type of organizations and provide new insights to improve co-operatives’ innovation and performance.

Keywords: co-operative, capitalization, members’ commitment, innovation, performance JEL code: Q13, M10 Corresponding author: gustavo.marcos@unavarra.es

© 2018 Marcos-Matas et al.

379

Marcos-Matas et al.

Volume 21, Issue 3, 2018

1. Introduction According to the report of World Cooperative Monitor (2015), at world level the co-operative movement is particularly active within the agri-food sector. This is also confirmed at Italian level where, according to the last data provided by the national report on agri-food co-operatives by Osservatorio della Cooperazione Agricola Italiana (Di Tullio and Gandini, 2015), co-operatives play a key role both in relation to the national co-operation system and to the agricultural production. Considering the latest available data (Di Tullio and Gandini, 2015), in 2013 Italy accounted 9,795 agri-food co-operatives, about 13% of total agri-food enterprises, they involve 37% of the total agricultural gross sealable production and 25% of national income from agrifood sector. Still, according to the report, Italian co-operation is one of the most advanced at European level, considering that, among co-operatives, 79.5% of production input is provided by their members. The Italian co-operative movement is also particularly active in terms of exportations, by contributing at 13% of total national exports. Focusing on their geographical distribution, the highest concentration of co-operatives is located in the south of Italy, by involving the highest number of agricultural co-operatives, followed by those located in the north. However, co-operatives are quite fragmented in relation to the capability of generating income, as only 13% of them are able to address 87% of the total income generated by all co-operatives in Italy, still they are mainly concentrated in the northern areas (82% of total income has been generated in the north). This distribution is similar also considering the number of employed persons (Di Tullio and Gandini, 2015).

Despite the relevant role played, co-operatives have been showing a decreasing trend, due both to changes in the firm structure and the economic crisis. Excessive leverage is one of the most important problems agricultural co-operatives face (Russo et al., 2000). The undercapitalization of co-operatives in Italy has been hypothesized by many Italian authors (Bono, 2011, 2012; Iannello, 1994; Mazzoli and Rocchi, 1996; Russo et al., 2000). Fiorentini (1995) compared the capital structure of a sample of Italian agricultural co-operatives with a sample of investor owned firms and found that co-operatives were undercapitalized.

The performance of the leverage ratio is heterogeneous and therefore difficult to interpret. In some cases, its growth seems to indicate a conscious choice by the co-operative to take on more debt in order to take advantage of a higher profitability that is superior to the cost of the debt. However, the same indicator may also be influenced heavily by the low level of capitalization of many Italian co-operatives (Bono, 2011).

The co-operative formula, compared with investor owned firms, has been assumed to be a model with difficulties in gathering the needed financial resources and therefore less innovative (Bono, 2012; Harte, 1997; James and Sykuta, 2005). As a result, co-operatives insufficiently capitalized might risk not to be able to remain a viable organizational form (Chaddad et al., 2005; Cook, 1995; Vitaliano, 1983).

Nonetheless, many co-operative firms survive and, in many cases, show high performances (Feng and Hendrikse, 2012). There is a wide consensus that members’ commitment is one major prerequisite for co-operative’s success (Fulton, 1999; Österberg and Nilsson, 2009; Zeuli and Betancor, 2005). In order to acquire the required capital to implement growth-related strategies and remain competitive, agricultural co-operatives need to encourage their members to participate on equity raisings. To this extent, the study provided by Euricse (Prandi, 2014), focusing on a sample of 4,451 Italian co-operatives operating in different sectors between, emphasizes that a minority of co-operatives (9.1%) has been able to reach an adequate capitalization with strong members’ participation. In addition they recorded optimal performances of growth of the share capital and showed a greater use of social equipment to strengthen the self-financing and the link between members and the co-operative. In this sense, also the use of the instrument of social lending has increased with a positive relation to the increasing size of the co-operative and to the higher level of capitalization. In fact, the instrument of loan capital was used not only out of necessity (self) funding, but probably as a fiduciary service social co-operative, intended also in terms of remuneration, deemed safe and convenient, the savings of the shareholder. The study confirms this approach also for the recent crisis’ years (2008-2011), as these co-operatives have been able to guarantee the enhancement of members through the

International Food and Agribusiness Management Review

380

Marcos-Matas et al.

Volume 21, Issue 3, 2018

compliance with the requirement of mutualistic prevalence, the effective participation of the members to the co-operative life (e.g. assemblies) and their engagement in growth capital.

Given the framework provided, the present study aims at shedding light on co-operatives’ financial constraint hypotheses. In particular, at first, it aims at investigating to what extent capitalization might be influenced by members’ commitment and whether this is a relevant aspect for the success of this kind of organization as stated by several authors (Fulton, 1999; Österberg and Nilsson, 2009). At second, the study aims at highlighting the possible effects of capitalization in relation to the acquisition of sufficient risk capital to finance profitable investment opportunities and to improve their performance (Bono, 2011, 2012; Iannello, 1994; Mazzoli and Rocchi, 1996; Russo et al., 2000).

2. Theoretical framework 2.1 Members’ commitment Investor owned firms raise capital in order to achieve financial return for their investors. They benefits proportionally to their capital share in terms of income from dividends and the increased market value of the shares (Mills and Davies, 2013). Instead, co-operative capital is different owing to equally distributed profits according to patronage, the restrictions on transferability of shares and the lack of liquidity through a secondary market for the transfer of such shares (Cook, 1995). Hence, when compared with investor owned firms, co-operative capital does not offer to investors comparable economic benefits. Nevertheless co-operatives members do not only seek for profit, but other relational aspects.

The term of social capital has been used to address such features of a relationship as trust, norms and networks. This concept characterizes the quality of relationships and can help to improve the effectiveness of economic relationships (Spear, 2000).

In the context of co-operatives, it is possible to consider the co-operative values and principles as such social capital. This social capital is shared among members, replacing others incentives such as price, or the formal authority. Hence, this social capital constitutes the unifying force of this organizational form (Valentinov, 2007).

One of the key features of co-operatives which favors the creation and utilization of social capital is members’ commitment. This commitment has been recognized as the differential factor from the investor owned firms (Spear, 2000). Without members’ commitment co-operatives would have problems to operate and even to establish (Fulton, 1999). Fulton (1999) argued that commitment is a key aspect that make farmers choosing this model even if they could obtain higher prices from an investor owned firm. This reasoning leads to think that members’ commitment is driving their willingness to invest and therefore the capitalization of the co-operative.

Hypothesis 1: Members’ commitment has a positive influence in co-operative’s capitalization.

2.2 Effects of capitalization on innovation Investments in R&D in Italian co-operatives are usually a less relevant part of the strategies adopted by agri-food co-operatives (Bono, 2012). Agri-food marketing co-operatives have to compete with other forms of organization in a market characterized by a concentrated and powerful demand, overproduction in many produces and changing consumers’ demands on products (Bijman and Hendrikse, 2003; HernándezEspallardo et al., 2013). These challenges force co-operatives to adopt strategies of adaptation (Bijman and Ruben, 2005). Adaptation and innovation have become critical elements of business conduct affecting the competitiveness of the co-operatives (Gianakas and Fulton, 2005). In general, within the agri-food industry, the processes of innovation are primarily of an incremental nature and they do not have their main source of International Food and Agribusiness Management Review

381

Marcos-Matas et al.

Volume 21, Issue 3, 2018

investment in R&D as much as in formal and informal economic learning (Christensen, 1996; Galizzi and Venturini, 1996). Competitive strategies pursued by agricultural co-operatives in response to environmental and structural changes in the food system include value-added processing, brand name development, and entry into international markets. All of them require substantial investments (Chaddad and Cook, 2004).

Hypothesis 2: Co-operativeâ&#x20AC;&#x2122;s capitalization has a positive effect on its innovation.

2.3 Effects of innovation on performance Roberts and Amit (2003) describe the importance of innovation as a leading asset to reach a competitive advantage and superior profitability. As revealed in many studies, innovation and firm performance have a positive relationship (Calantone et al., 1995, Capon et al., 1990; Han et al., 1998; Zahra and Das, 1993).

Innovation would appear in product, process, market, factor and organization (Kao, 1989), but the first three dimensions are more familiar in the innovation literature (Johne and Davies, 2000; Najib and Kiminabi, 2011; Otero-Neira et al., 2009; Rosli and Sidek, 2013). Many economists have accepted innovation as a key condition for business performance, competitiveness, and economic wealth (Caird, 1994). A study by Deshpande et al. (1993) indicated that innovativeness is positively related to organizational performance in terms of relative profitability, market share, and growth. Baldwin and Johnson (1996) showed the significant impact of innovation on a wide variety of business performance measures, including market share and return on investment. Further, Salavou (2002) also found that product innovation was a significant determinant of business performance based on return on assets.

Hypothesis 3: Co-operativeâ&#x20AC;&#x2122;s innovation has a positive impact on performance.

3. Materials and methods 3.1 Data collection According to Osservatorio della Cooperazione Agricola Italiana (Di Tullio and Gandini, 2015) and Bertagnoni (2015), Emilia Romagna is the leader region of the co-operative movement in Italy, including 701 agri-food co-operatives, representing 14% of the national amount of co-operatives, and producing an overall income of over 13 billion euro, corresponding to one third of the total income generated at national level by the cooperative agri-food sector in 2013. In details, about 7% (51) of co-operatives in Emilia Romagna reached an income over 40M euro; 14% (101) performed between 7M and 40M euro; 29% (200) between 2M and 7M euro; and 60% (349) below 2M euro. The main important sectors refer to meat, horticulture and wine productions. The questionnaire targeted managers of co-operatives located in Emilia Romagna region. The co-operatives involved in the study belong to two out of 4 national associations of co-operatives that are Legacoop and Confcooperative. The list of co-operatives contacted included small medium and large firms, and every typology of agri-food sector.

Online questionnaire were submitted between July and October 2015. Out of 100 questionnaires, 58 questionnaires were successfully completed, although only 52 were the valid ones.

3.2 Questionnaire design The instrument used was a questionnaire created ad hoc. The measurement scales are based on the literature and were adapted when necessary with the feedback from prior interviews and pre-tests.

International Food and Agribusiness Management Review

382

Marcos-Matas et al.

Volume 21, Issue 3, 2018

■■ Capitalization About capitalization, considering the peculiar structure and mission of co-operatives organizations, the scale was developed according to the available literature investigating capitalization needs for co-operatives (Fiorentini, 1995; Iannello, 1994; Mazzoli and Rocchi, 1996; Russo et al. 2000). The pre-tests feedbacks were relevant to refine the aspects that mainly involve agri-food co-operatives. The items referred to the availability of capital resources, the difficulties to access external funds, and the difficulties to fund new investment projects, with particular regard at the needed support of ancillary services undertakings.

■■ Innovation The innovation inventory of Oslo Manual of OECD/Eurostat (2005) was adopted to include the most common innovation aspects regarding agri-food co-operatives, following they were validated through pretests’ feedbacks.

■■ Performance As performance, there were considered those outcomes that show the fulfilment of the organizations goals (Kumar et al., 1992). Thus, Quinn and Rorbaugh (1993) proposal were adopted to define its measurement.

They used four models that simultaneously describe the organization performance. That is, the human relations model regarding the development and satisfaction of the participants, the internal process model referring to the development of the activities and operational processes, the open system model in regard to the adaptation to the market, and the rational goal model that refers to achievement of productivity and efficiency goals. To this extent these models were adapted to the co-operatives’ scenario following the pretests feedbacks to validate the items. The assessment of the measurement model let with only three items of the models. The item regarding to the human relations did not showed to be statistically significant on this scale. Nevertheless, the item of the rational goal model is represented as the achievement of better prices for farmers and it can also act as a proxy variable of the human relations model as the prices have been proven to be a good predictor of the farmers’ satisfaction (Hernández-Espallardo et al., 2013).

■■ Commitment Members’ commitment was measured with the scale developed by Gundlach et al. (1995) and Kumar et al.

(1995). This concept consider members’ willingness to make short-term efforts (Gundlach et al., 1995) and investments (Kumar et al., 1995).

■■ Control variables Control variables contributed to assess the validity of the model, as with them, the core effect of concepts is better isolated. Two variables were included in the analyses: co-operative’s self-declared turnover and number of members.

■■ Validity tests The scales used in the measurement model are made from formative indicators. In this type of scales, each item, referring to the different dimensions of the concepts, contributes or adds to the latent construct (Fornell and Larcker, 1981). According to the methodology described by Rossiter (2002) for formative scales, the most relevant components of each of the concepts supported by literature were included, in addition to the comments and feedback from the prior interviews and pre-tests. For a proper model specification with partial least squares (PLS) methodology, these formative constructs should address at least major part of their domains (Hair et al., 2012).

International Food and Agribusiness Management Review

383

Marcos-Matas et al.

Volume 21, Issue 3, 2018

In order to evaluate the measurement model, it was needed to test the validity of the scales used. To assess the validity of the formative constructs, it was used the weight of each indicator for each latent construct according to criteria of significance regardless of size, as formative items are viewed as multidimensional and not similar measures reflecting the same underlying construct (Chin, 1998; Vinci et al., 2010).

The estimation of this validity was performed with the PLS approach with the bootstrapping technique, which calculates the weights of the items on the construct and their significance. At this regard, several items of some scales were removed because their weights were not statistically significant. The validity of the scales was confirmed as the weights of the items were sufficiently significant (Table 1). Despite the use of weights to assess constructs’ validity, as Marcoulides et al. (2009: 173) pointed, the ‘latent’ PLS constructs are not really latent variables since they are not explained by their covariations, just approximately’. In PLS sums of the indicators are not weighted, they can only be estimated by weighted sums of their indicators, being loading vector proportional to the weight vector (Scheeweiss, 1993). However, this aspect was not a critical issue to the model application since in this investigation there were used formative constructs.

3.3 Data analysis To test the hypotheses, the structural equation modelling was used. The analyses were conducted following the PLS methodology, which is especially useful to test theories when formative measures, non-normal data1, many latent variables and limited sample size are involved (Chin, 1998; Henseler et al., 2009). Both measurement and structural models were estimated with SmartPLS 2.0 M3 software2 (Ringle et al., 2005).

1

None of the variables shows a normal distribution in Kolmogorv-Smirnov normality test.

Settings. PLS algorithm: no missing value algorithm, path weighting scheme, 500 max. iterations, 1.0E-5 abort criterion; bootstrapping algorithm: no sign changes, 52 cases, 3,000 samples.

2

Table 1. Weights and t-values.1 Construct

Item

Weights

t-value

Capitalization

CAP1 CAP3 CAP4 CAP5 INN2 INN3 INN4 INN5 INN7 INN8 INN9 INN10 PER2 PER3 PER4 COMM2 COMM5 COMM6 COMM8

0.24 0.39 0.25 0.44 0.24 0.11 0.18 0.24 0.16 0.18 0.20 0.15 0.24 0.44 0.44 0.32 0.34 0.31 0.42

2.35 6.36 2.40 4.83 3.61 1.81 3.32 5.11 3.17 3.79 3.58 2.92 1.84 5.22 3.59 2.57 2.51 1.86 3.79

Innovation

Performance

Commitment

1

Probability of critical t-values: P<0.10 for t>1.65; P<0.05 for t>1.96 and P<0.01 for t>2.58.

International Food and Agribusiness Management Review

384

Marcos-Matas et al.

Volume 21, Issue 3, 2018

4. Results To assess the structural model by PLS, the variance of the dependent latent variables explained by the independent variables that predict them (R2) must be considered and, in particular, the significance of the coefficients associated with the proposed relationships must be assessed (Cepeda and Roldán, 2008).

In this case, the dependent variables are performance, innovation and capitalization. The value of R2 expresses the proportion of variance of the dependent variable that is explained by the independent variable, so the closer it is to one, the more explanatory power the model will have, always bearing in mind that R2 must not be less than 0.1 (Falk and Miller, 1992).

The model had sufficient predictive power to explain the variable performance (R2=0.20) innovation (R2=0.27) and capitalization (R2=0.19) as they are above 0.1 (Falk and Miller, 1992).

As a complement when analyzing the size of R2 as a criterion of predictive significance, the technique of reusing the sample proposed by Stone (1974) and Geisser (1975) can be applied through the blindfolding process. This technique consists of omitting part of the data when estimating a dependent variable from other independent variables, and then attempts to estimate those data by using previously estimated parameters. This process is repeated until each datum has been omitted and estimated. This technique calculates construct crossvalidated redundancy index Q2 that represents a measure of how well the observed values are reconstructed from the estimated parameters. Hair et al. (2012) recommends assessing cross-validated redundancy besides the effect sizes. For the model to have predictive validity, Q2 must be greater than zero. The model shows an adequate predictive capacity, as three dependent variables, namely performance (Q2=0.15), innovation (Q2=0.12) and capitalization (Q2=0.10), Q2 is greater than zero (Geisser, 1975).

After analyzing these statistics, it was conducted the structural model test. The analysis generated the estimates of the standardized coefficients for the proposed relations and their t-values obtained through bootstrapping process (Table 2) (Chin, 1998).

In relation to the theoretical model applied, all estimations have shown significant coefficients as predicted, confirming the predicted relations (Figure 1). In fact, the results show a positive relation between capitalization measures and innovation in the co-operative as predicted in H1. Likewise members’ commitment show to have a positive influence in co-operative’s capitalization measures (H2). And finally, as stated in H3, cooperative’s innovation measures are positively related with its performance.

As for the control variables, all of them had non-significant influence on dependent variables, except the relation of co-operative’s turnover with innovation, which shows a positive effect.

Table 2. Standardized coefficients (β) and t-values.

Hypotheses

β

t-value1

H1: Capitalization → Innovation H2: Innovation → Performance H3: Commitment → Capitalization

0.46 0.42 0.44

3.91*** 2.89*** 3.87***

1 ***

= P<0.01.

International Food and Agribusiness Management Review

385

Marcos-Matas et al.

Commitment

Volume 21, Issue 3, 2018

0.44***

Capitalisation R2=0.19

0.46***

0.42***

Innovation R2=0.27

Performance R2=0.20

Control variables Turnover

Number of members

Figure 1. Theoretical model and results (standardized path coefficients and R2).* = P<0.10; ** = P<0.05; *** = P<0.01; n.s. = non-significant.

5. Conclusions The aim of the study was to assess the role of capitalization of the co-operatives on their innovation and performance since they are theorized and probed to be an undercapitalized formula (Prandi, 2014; Russo et al., 2000).

Therefore at first, it was verified whether members’ commitment has a positive influence on co-operatives’ capitalization as it has been theorized to be the factor that lead co-operatives to remain competitive in relation with investor owned firms (Österberg and Nilsson, 2009). The results of the analyses confirm the hypothesis that members’ commitment is positively related with co-operative levels of capitalization. Members’ commitment is the reflection of the co-operatives’ values and principles (Spear, 2000) and they are the main capital suppliers (Mills and Davies, 2013; Nilsson et al., 2012), hence this leads to think that trust and relational norms can be drivers of co-operatives’ investments. As commented by Chaddad et al. (2005), it has been suggested that financial constraints are the weaknesses of co-operatives in an increasingly concentrated, tightly coordinated and capital-intensive food system (Cook, 1995; Vitaliano, 1983). Furthermore, risk capital acquisition in the traditional co-operative firm is limited by number, wealth, and risk-bearing capacity of its current members (Chaddad et al., 2005). Thus, in order to acquire the required capital to implement these growth-related strategies and remain competitive, agricultural co-operatives need to encourage their members to participate on equity raisings.

Results also confirm the second hypothesis about the positive effect of the capitalization level on co-operatives’ innovation. In fact, more capitalized co-operatives are more prone to carry out innovative projects. Capitalized co-operative can be able to provide the sufficient capital to invest and/or to achieve adequate standards to access to financial markets so as to fund innovation. As confirmed by the evidence and the literature, own members’ capital input is considered a precondition for financing entities to approve loans and support risky financial decision. Similarly, some authors stated that highly committed memberships are more able to engage differentiated product development (Fulton, 1999) and also demonstrated that suppliers’ (i.e. members in this case) trust and commitment determines their involvement in customers’ (i.e. co-operatives) new product development (Walter, 2003).

With regards to the third hypothesis, referring to the innovation effect on co-operatives’ performance, it has been also confirmed that investing on innovation induces better performances. This inquiry is in line with those on literature (Calantone et al., 1995; Capon et al., 1990; Han et al. 1998; Zahra and Das, 1993). In addition, the results would apply to a market oriented perspective to the extent that innovation is considered a key driver to match customers’ preferences and to adapt to the market needs, by reflecting on improved performances (Costa and Jongen, 2006; Hult et al., 2005).

International Food and Agribusiness Management Review

386

Marcos-Matas et al.

Volume 21, Issue 3, 2018

6. Managerial implications From a managerial point of view, these findings highlight that it is essential to create the conditions to improve co-operatives members’ commitment since this is a mean to increase capitalization. To this extent, co-operatives shall favor communication and information sharing among their members, to enhance members’ participation in the co-operative (Barraud-Didier et al., 2012). In fact, the literature points out that informationsharing is a lead agent to improve commitment and motivation (Guerrero and Barraud-Didier, 2004). Thus, the co-operative must not forget to focus and invest on its social relationships with members on account to achieve good capitalization levels through a better members’ commitment.

Furthermore, the interest of the study is undoubtedly relevant to explain capitalization in relation to competitiveness related variables (innovation and performance). As for innovation, the results point out that it is related with the level of capitalization. This effect can overcome the natural tendency of agri-food cooperatives to act with a short-term view mainly focused towards farmers’ needs rather than the co-operative as a firm, driven by market and innovation views (Baamonde, 2009; Kyriakopoulos et al., 2004). The findings also backed up the idea that those co-operatives that leads more innovative actions are those which show better performance. In fact, they perform better to develop the activities and operational processes, to adapt to the market, and to achieve productivity and efficiency goals.

These implications might be also of interest to develop policy recommendations to improve the general competitiveness of the co-operative formula and its maintenance in the long run, with positive implications on rural areas.

7. Limits of the study and further research Given these insights, it shall be underlined that the study focused only on a limited sample, with a circumscribed geographical extent. Additionally, it is limited to one sector and type of organizations that are agri-food cooperatives. These aspects make findings not able to be ascribed to this context and generalizations might be taken with caution. Future research could aim at testing the hypotheses adopted within a bigger sample.

In addition, another limitation comes from the measurement method as it is based on respondent’s subjective perceptions. The use of this type of measures is common in the literature, although triangulation with archival or secondary data would be advisable in further researches.

Future research should focus on confirming the hypothesis adopted on the influence of commitment on capitalization through other empirical methods as panel groups technique or dyad data from farmers and managers, helping to reinforce and validate the theoretical model. Besides, they could address the testing of this model on investor owned firms in order to find out how, in a value chain perspective, suppliers’ commitment might affect similarly.

References Baamonde, E. 2009. El cooperativismo agroalimentario. In: Mediterráneo económico. El nuevo sistema agroalimentario en una crisis global 15, edited by J. Lamo de Espinosa. Fundación Cajamar, Almería, Spain, pp. 229-246.

Baldwin, J.R. and J. Johnson. 1996. Business strategies in more-and less-innovative firms in Canada.

Research policy 25(5): 785-804.

Barraud-Diedier, V., M.-C. Henninger and A. El Akremi. 2012. The relationship between members’ trust and participation in the governance of co-operatives: the role of organizational commitment. International Food and Agribusiness Management Review 15(1): 1-24.

International Food and Agribusiness Management Review

387

Marcos-Matas et al.

Volume 21, Issue 3, 2018

Bertagnoni, G. 2015. Cooperazione agroalimentare in Emilia-Romagna: una storia di successo, un futuro di sfide. Assessorato Agricoltura Caccia e Pesca, Regione Emilia-Romagna. Available at: http:// tinyurl.com/yaq99qpe.

Bijman, J. and G. Hendrikse. 2003. Co-operatives in chains: institutional restructuring in the Dutch fruit and vegetables industry. ERIM Report Series Research in Management, ERS-2003-089-ORG. Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=474429.

Bijman, J. and R. Ruben. 2005. Repositioning agricultural co-operatives in the North and the South: where do the Twain meet? Paper presented in 2nd Agri-ProFocus Expert Meeting, Deventer, November.

Available at: http://edepot.wur.nl/18623.

Bono, P. 2011. Le performance delle cooperative agroalimentari. In: Osservatorio della Cooperazione Agricola Italiana, Rapporto 2008-2009. Agrirete Service, Rome, Italy.

Bono, P. 2012. La specificità dell’impresa cooperativa. In: Osservatorio della Cooperazione Agricola Italiana, Credito e cooperazione. Report 2012. Agrirete Service, Rome, Italy.

Caird, S. 1994. How do award winners come up with innovative ideas? Creativity and Innovation Management 3(1): 3-10.

Calantone, R.J., S.K. Vickery and C. Dröge. 1995. Business performance and strategic new product development activities: an empirical investigation. Journal of Product Innovation Management 12(3): 214-223.

Capon, N., J.U. Farley and S. Hoenig. 1990. determinants of financial performance: a meta-analysis.

Management Science 36(10): 1143-1159.

Cepeda, G. and J. Roldán. 2008. Aplicando en la práctica la técnica PLS en la administración de empresas [Applying the PLS technique in business management practice]. University of Seville. Available at: http://goo.gl/Xoi9oM. Chaddad, F. and M.L. Cook. 2004. Understanding new co-operative models: an ownership-control rights typology. Review of Agricultural Economics 26(3): 348-360.

Chaddad, F.R., M.L. Cook and T. Hecklelei. 2005. Testing for the presence of financial constrains in US agricultural co-operatives: an investment behaviour approach. Journal of Agricultural Economics 56(3): 385-397.

Chin, W.W. 1998. The partial least squares approach to structural equation modelling. In: Modern methods for business research, edited by G.A. Marcoulides. Lawrence Erlbaum Associates, Mahwah, NJ, USA, pp. 295-358.

Christensen, J.F.S. 1996. Innovative assets and inter-asset linkages – a resource-based approach to innovation.

Economics of Innovation and New Technology 4(3): 193-210.

Cook, M.L. 1995. The future of U.S. agricultural co-operatives: a neo-institutional approach. American Journal of Agricultural Economics 77(5): 1153-1159.

Costa, A.I. and W.M.F. Jongen. 2006. New insights into consumer-led food product development. Trends in Food Science and Technology 17(8): 457-465.

Deshpande, R., J.U. Farley and F.E. Webster Jr. 1993. Corporate culture, customer orientation, and innovativeness in Japanese firms: a quadrad analysis. The journal of Marketing 57(1): 23-37.

Di Tullio, E. and E. Gandini. 2015. I numeri della cooperazione agroalimentare italiana. In: Osservatorio della cooperazione agricola italiana, Rapporto 2015. Report 2012. Agrirete Service, Rome, Italy.

Falk, R.F. and N.B. Miller. 1992. A primer for soft modeling. University of Akron Press, Akron, OH, USA. Feng, L. and W.J. Hendrikse. 2012. Chain interdependencies, measurement problems and efficient governance structure: co-operatives versus publicly listed firms. European Review of Agricultural Economics 39(2): 241-255.

Fiorentini, G. 1995. Forza e debolezza delle cooperative: aspetti economici e finanziari. In: Proceedings of the ‘Seconda giornata di studi economici sulla cooperazione’. Censcoop, Bologna, Italy.

Fornell, C. and D.F. Larcker. 1981. Structural equation models with unobservable variables and measurement error: algebra and statistics. Journal of Marketing Research 18(3): 328-388.

Fulton, M. 1999. Co-operatives and member commitment. Finnish Journal of Business Economics 48(4): 418-437.

International Food and Agribusiness Management Review

388

Marcos-Matas et al.

Volume 21, Issue 3, 2018

Galizzi, G. and L. Venturini. 1996. Product innovation in the food industry: nature, characteristics and determinants. In: Economics of innovation: the case of food industry. Physica-Verlag, Heidelberg, Germany, pp. 133-153.

Geisser, S. 1975. The predictive sample reuse method with applications. Journal of the American Statistical Association 70(350): 320-328.

Giannakas, K. and M. Fulton. 2005. Process innovation activity in a mixed oligopoly: the role of co-operatives.

American Journal of Agricultural Economics 87(2): 406-422.

Guerrero, S. and V. Barraud-Didier. 2004. High-involvement practices and performance of French firms.

International Journal of Human Resource Management 15(8): 1410-1425.

Gundlach, G.T., Achrol, R.S. and J. Mentzer 1995. The structure of commitment in exchange. Journal of Marketing 59(1): 78-92.

Hair J.F., M. Sarstedt, T.M. Pieper and C.M. Ringle. 2012. The use of partial least squares structural equation modeling in strategic management research: a review of past practices and recommendations for future applications. Long Range Planning 45(5-6): 320-340.

Han, J.K., N. Kim and R.K. Srivastava. 1998. Market orientation and organizational performance: is innovation a missing link? The Journal of marketing 62(4): 30-45.

Harte, L.N. 1997. Creeping privatization of Irish co-operatives: a transaction costs explanation. In: Strategies and structures in the agro-food industries, edited by J. Nilsson and G. Van Dijk. Van Gorcum and Comp, Assen, the Netherlands, pp. 41-49.

Henseler, J., C.M. Ringle and R.R. Sinkovics. 2009. The use of partial least squares path modeling in international marketing. In: New challenges to international marketing (Advances in International Marketing, Volume 20), edited by R.R. Sinkovics and P.N. Ghauri. Emerald Group Publishing Limited, Bingley, UK, pp. 277-319.

HernĂĄndez-Espallardo, M., N. Arcas-Lario and G. Marcos-MatĂĄs. 2013. Farmersâ&#x20AC;&#x2122; satisfaction and intention to continue their membership in agricultural marketing co-operatives: neoclassical versus transaction costs considerations. European Review of Agricultural Economics 40(2): 239-260.

Hult, G.T.M., D.J. Ketchen and S.F. Slater. 2005. Market orientation and performance: an integration of disparate approaches. Strategic Management Journal 26(12): 1173-1181.

Ianniello, G. 1994. Impresa cooperativa: caratteristiche strutturali e nuove prospettive di finanziamento.

Cedam, Padua, Italy.

James, B.S.Jr and M.E. Sykuta. 2005. Property right and organizational characteristics of producer-owned firms and organizational trust. Annals of Public and Co-operative Economics 76(4): 545-580.

Johne, A. and R. Davies. 2000. Innovation in medium-sized insurance companies: how marketing adds value.

International Journal of Bank Marketing 18(1): 6-14.

Kao, J.J. 1989. Entrepreneurship, creativity and organization. Prentice Hall, New Jersey, NJ, USA. Kumar, N., L.K. Scheer and J.-B.E.M. Steenkamp. 1995. The effects of perceived interdependence on dealer attitudes. Journal of Marketing Research 32(3): 348-356.

Kumar, N., L.W. Stern and R.S. Achrol. 1992. Assessing reseller performance from the perspective of the supplier. Journal of Marketing Research 29(2): 238-253.

Kyriakopoulos, K., M. Meulenberg and J. Nilsson. 2004. The impact of co-operative structure and firm culture on market orientation and performance. Agribusiness 20(4): 379-396.

Marcoulides, G.A., W.W. Chin and C. Saunders. 2009. A critical look at partial least squares modeling. MIS Quarterly 33(1): 171-175.

Mazzoli, M. and E. Rocchi. 1996. La finanza delle cooperative. Liocorno Editore, Rome, Italy.

Mills, C. and W. Davies. 2013. Blueprint for a co-operative decade. International Co-operative Alliance.

Available at: https://goo.gl/6jBIq7.

Najib, M. and A. Kiminami. 2011. Innovation, cooperation and business performance: some evidence from Indonesian small food processing cluster. Journal of Agribusiness in Developing and Emerging Economies 1(1): 75-96.

Nilsson, J., G.L.H. Svendsen and G.T. Svendsen. 2012. Are large and complex co-operatives losing their social capital? Agribusiness 28(2): 187-204.

International Food and Agribusiness Management Review

389

Marcos-Matas et al.

Volume 21, Issue 3, 2018

OECD/Eurostat. 2005. Oslo manual: guidelines for collecting and interpreting innovation data, 3rd Edition, OECD Publishing, Paris, France. Österberg, P. and J. Nilsson 2009. Members’ perception of their participation in the governance of co-operatives: the key to trust and commitment in agricultural co-operatives. Agribusiness 25(2): 181-197.

Otero-Neira, C., M.T. Lindman and M.J. Fernández. 2009. Innovation and performance in SME furniture industries: an international comparative case study. Marketing Intelligence and Planning 27(2): 216-232.

Prandi, P. 2014. Capitalizzazione mutualità e partecipazione. In: La cooperazione italiana negli anni della crisi, Secondo Rapporto Euricse, a cura di Carlo Borzaga. Euricse – Istituto Europeo di Ricerca sull’Impresa Cooperativa e Sociale, Trento, Italy, pp. 190-201.

Quinn, R.E. and J. Rohrbaugh. 1983. A spatial model of effectiveness criteria: towards a competing values approach to organizational analysis. Management Science 29(3): 363-377.

Ringle, C.M., S. Wende and A. Will. 2005. SmartPLS 2.0.M3. SmartPLS, Hamburg, Germany. Available at: http://www.smartpls.com.

Roberts, P.W. and R. Amit. 2003. The dynamics of innovative activity and competitive advantage: the case of Australian retail banking, 1981 to 1995. Organization Science 14(2): 107-122.

Rosli, M.M. and S. Sidek. 2013. The impact of innovation on the performance of small and medium manufacturing enterprises: evidence from Malaysia. Journal of Innovation Management in Small and Medium Enterprise 2013: 885666.

Rossiter, J. 2002. The C-OAR-SE procedure for scale development in marketing. International Journal of Research in Marketing 19(4): 305-335.

Russo, C., D. Weatherspoon, C. Peterson and M. Sabbatini. 2000. Effects of managers’ power on capital structure: a study of Italian agricultural co-operatives. International Food and Agribusiness Management Review 3(1): 27-39.

Salavou, H. 2002. Profitability in market-oriented SMEs: does product innovation matter? European journal of innovation management 5(3): 164-171.

Schneeweiss, H. 1993. Consistency at large in models with latent variables. In: Statistical modelling and latent variables, edited by K. Haagen, D.J. Bartholomew and M. Deistler. Elsevier, Amsterdam, the Netherlands, pp. 299-320.

Spear, R. 2000. The co-operative advantage. Annals of Public and Co-operative Economics 71(4): 507-523.

Stone, M. 1974. Cross-validatory choice and assessment of statistical predictions. Journal of the Royal Statistical Society 36(2): 111-47.

Valentinov, V. 2007. Why are co-operatives important in agriculture? An organizational economics perspective.

Journal of Institutional Economics 3(1): 55-69.

Vinci, V.E., W.W. Chin, J. Henseler and H. Wang. 2010. Handbook of partial least squares: concepts, methods and applications. Springer, Berlin, Germany.

Vitaliano, P. 1983. Co-operative enterprise: an alternative conceptual basis for analyzing a complex institution.

American Journal of Agricultural Economics 65(5): 1078-1083.

Walter, A. 2003. Relationships-specific factors influencing supplier involvement in customer new product development. Journal of Business Research 56(9): 721-733.

World Cooperative Monitor 2015. Exploring the cooperative economy, report 2015. Available at: http:// tinyurl.com/ydxv9bmy.

Zahra, S.A. and S.R. Das. 1993. Innovation strategy and financial performance in manufacturing companies: an empirical study. Production and operations management 2(1): 15-37.

Zeuli, K. and A. Betancor. 2005. The effects of co-operative competition on member loyalty. Paper presented at 2005 Annual Meeting, Minneapolis, November. Available at: http://tinyurl.com/ya6u46yu.

International Food and Agribusiness Management Review

390

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2016.0177x Received: 23 November 2016 / Accepted: 21 November 2017

Dual moral hazard and adverse selection in South African agribusiness: it takes two to tango RESEARCH ARTICLE Thulasizwe Mkhabela Chief Operations Officer, Agribusiness Development Agency, Pietermaritzburg, KwaZulu-Natal 3202, South Africa; Senior Lecturer Extraordinaire, Department of Agricultural Economics, Stellenbosch University, Matieland 7602, South Africa

Abstract The paper employs a dual moral hazard and adverse selection model to analyse partnerships in agribusiness under joint venture contracts with asymmetric information and imperfect quality measurement by the agent and principal both of which contribute to the final quality of the product in terms of production effort and marketing (offtake) effort, respectively. A salient feature of this paper is the analysis of the ramifications of joint venture contract for quantity and quality, which is often deficient in most previous analyses of moral hazard. The research found that contracts that have rewards based on the quantity produced weakened the agentâ&#x20AC;&#x2122;s incentive to make effort in ensuring quality. This finding could explain why most contracts in agriculture for products with differentiated markets rarely use retail-price conditioned contracts.

Keywords: joint venture, agribusiness, moral hazard, Africa JEL code: C65, Q13, Q15, Q18 Corresponding author: mkhabelat@ada-kzn.co.za

Š 2018 Mkhabela

391

Mkhabela

Volume 21, Issue 3, 2018

1. Introduction The agribusiness (agrifood) system has been undergoing fundamental change that is transforming traditional marketing relationships in response to changing consumer needs. The agribusiness supply chains and networks which were once characterised by autonomous and independent actors are rapidly becoming more globally interconnected with a wide array of complex relationships (Cook and Chaddad, 2000; Machethe et al., 1997; Ruben et al., 2006).

In Africa, agribusiness is a burgeoning sector of local economies and continental economy. The sector has been receiving unprecedented attention from policy makers and private sector investors alike with increasing favourable policy interventions and private capital investments from both large multinational food and agribusiness companies and up and coming local food companies (World Bank, 2013). Governments in Africa have been devoting much effort and resources to creating an environment conducive to the take-off and success of agribusiness sector. The World Bank (2013) further purports that if the increased attention received by African agriculture and agribusiness were to be matched with more electricity, irrigation, smart business and trade policies and a dynamic private sector that works hand in hand with governments to link small-scale farmers with consumers in an ever-urbanizing Africa, the sector could contribute US$1 trillion by 2030.

In economics, moral hazard occurs when one person takes more risks because someone else bears the cost of those risks. A moral hazard may occur where the actions of one party may change to the detriment of another after a financial transaction has taken place. Moral hazard and adverse selection are often used interchangeably although, strictly speaking, they are not synonymous. On the one hand, adverse selection occurs when there is lack of symmetric information prior to a deal between a buyer and a seller. On the other hand, moral hazard occurs when there is asymmetric information between two parties and when change in the behaviour of one party occurs after a deal is struck. Both expressions are used to describe situations where one party is at a disadvantage compared to the other.

There is no doubt that trade liberalization and globalization have brought about stringent sanitary and phytosanitary requirements and this is attested to by both experience and an ever-growing body of research and published work (Henson and Loader, 2001; Otsuki et al., 2001; Ruben et al., 2006). These stringent quality requirements present complications and non-tariff barriers to entry for farmers and agribusinesses in fresh produce both for local and export markets. These hindrances have been described as emerging barriers to agricultural and food trade (Ruben et al., 2006). The afore-mentioned developments exert additional demands on producers and processors of fresh produce to meet high and uniform quality standards and frequent delivery requirements (Reardon et al., 1999). Sourcing of perishable produce to secure all-year round supply, under private label, can be guaranteed through partnerships and long-term contracts between primary producers (farmers) and processors.

Joint ventures in the agrifood chains are seen as an innovative response to emerging developments in the industry. These trends are brought about by the ever-evolving consumer demands and the concomitant quality requirements as identified through supply-chain analyses to understand market structure and performance.

Supply chains are understood as transformation processes from inputs through primary production, processing and marketing to the final consumer (Porter, 1990). These transformation processes involve three dimensions: (1) organizational systems for the coordination amongst agents; (2) knowledge systems for combining information, skills and technologies; and (3) economic mechanisms for product and technology selection and for providing market access. Thus, supply chain performance can be assessed with efficiency parameters, searching for specialization according to comparative advantage and towards integration for reducing transaction costs. However, trust is an indispensable ingredient for joint ventures and other forms of partnership to succeed. According to Newman and Biggeman (2016), trust is an integral part of maintaining any successful business relationship, especially within agriculture and trust is required for any transaction to

International Food and Agribusiness Management Review

392

Mkhabela

Volume 21, Issue 3, 2018

take place. Trust is the cohesion in agricultural transactions which creates the value of relationships between transacting parties (Wilson, 2000).

In 2001 Lazzarini et al. (2001) introduced the concept of netchains at the interface of vertical supply chains and horizontal networks. According to Lazzarini et al. (2001), netchains can be viewed as a multi-layer hierarchy between suppliers, processors and retailers where horizontal coordination between reciprocal agents is embedded in a framework of vertical deliveries as indicated in Figure 1. It is worth reiterating that the agricultural and food sector is facing ever-evolving and stringent quality requirements. Horizontal cooperation (such as in farmers cooperatives) may present one of the best approaches to better cope with the stringent quality criteria and changing quantity demands emerging from chain partners. Figure 1 illustrates a typical supply chain in the agrifood industry. It can clearly be seen from the Figure 1 that relationships are indispensable in the agrifood sector if one is to survive and remain in business. Furthermore, the relationships are interconnected and multi-facet in nature thus information is shared amongst more than two stakeholders.

This web of relationships further buttresses the need for partnerships in order to be integrated along the value chain. By implication, messing relationships with one partner or role-player could have telling repercussions for one within the sector.

The Lazzarini et al. (2001) netchains provide linkages between horizontal networks of suppliers and vertical supply chains. Netchains involve different types of interdependencies (nested) amongst agents, for example: (1) reciprocal cooperation based on mutual exchange between suppliers; (2) sequential delivery systems based on planning along the supply chain; and (3) pooled interdependencies at business level to guarantee standardisation and harmonisation of produce and processes.

Finally, the role of formalisation of relationships amongst the various role-players cannot be ignored. Thus, contracts have a vital function in the relationships between chain and networks partners. Contracts define the rules and obligations of engagement for establishing cooperation between both network and chain agents.

Contracts can be viewed as a cost-reducing mechanism in the case of repeated transactions between agents.

Self-enforcing contracts that involve trust and loyalty are preferred for transactions that involve the delivery of high quality products to reduce monitoring and enforcement costs (Ruben et al., 2006:7; Shaban, 1987).

Suppliers

Processors

Distributors

Consumers

. . . . . . . . .

.

. . . . .. . . .. .

Figure 1. An illustration of supply chain in the agrifood sector (adapted from Lazzarini et al., 2001).

International Food and Agribusiness Management Review

393

Mkhabela

Volume 21, Issue 3, 2018

There is a plethora of available options for integrating horizontal networks and vertical chain contracts in order to guarantee risk-sharing and ensuring trust relationships (Masuku et al., 2007). Due to the inherent high risks, complexities and difficulties of monitoring numerous heterogeneous agents, entire-channel process control is increasingly preferred (Goodhue, 1999; Janssen and Van Tilburg, 1997; Van der Laan, 1993). Most of these entire-channel process control approaches are forms of self-enforcing contracts in order to obviate the need for stringent monitoring by the principal (Hardaker et al., 2015; Kvaløy, 2006).

On the one hand, market access for farmers is one of the most binding constraints to agricultural development and commercialisation of small- to medium-scale farmers the world over and South Africa is no exception.

Farmers are increasingly interested in being vertically integrated along the value chain in order to maximise profit through beneficiation of their produce beyond the farm gate. It should also be noted that farmers, being rational business people, are always looking for better business opportunities that maximize their utility while balancing their risk portfolios. On the other hand, agro-processors and other agribusinesses are always looking for agricultural produce (raw products) in sufficient quantities and of the required quality to be supplied consistently. The desire by agribusinesses to control the supply of raw materials has led to increased vertical coordination in the food market. Thus, the food market might be thought of as an orchestra with each level of the market contributing its utility to the final product (Kohls and Uhl, 1998:211). The problem then becomes how to marry the objectives of these two players in the agribusiness value chain and create a mutually beneficial working arrangement.

In South Africa, as in the rest of the African continent, commercial agricultural production has been dominated by few large-scale commercial farmers who are vertically integrated into the food and input (agribusiness) value chain through their membership in cooperatives. Thus, the bulk of agricultural product, especially in South Africa, is from the few large-scale commercial farmers in the face of a large number of small-scale farmers who contribute disproportionately small quantities of agricultural product into the market. This has led to the dualistic nature of the South African agricultural sector. Policy makers and the South African government have attempted to address this dichotomous sector through instituting a land reform programme which seeks to redress imbalances of the past through land restitution and land redistribution1. The land reform programme has brought about a number of new commercial farmers who had been alienated from commercial agriculture thus have limited connections within the market. Furthermore, most these new farmers have a paucity of technical agriculture skills, little or no capital and limited access to credit due lack of collateral and credit track record required by financial institutions.

A number of partnerships between farmers and agribusinesses have been established in South Africa, especially since the land reform programme started in the 1990s. The partnerships that exist are varied in their nature but mostly would be categorized as either strategic partnerships, joint venture partnerships or contract farming. Strategic partnership refers to an arrangement where an experienced farmer is assigned to assist relatively inexperienced farmers (mostly land reform beneficiaries) as a mentor. This model of partnership is often referred to as ‘the mentorship programme’ in South African agricultural and land reform parlance. There is growing evidence that there is no ideal model as each case is different thus there is no one size fits. The success of partnerships between farmers and agribusinesses has also been somewhat mixed (Terblanche et al., 2014). This study focuses on joint venture partnership between land reform beneficiaries and agribusiness in South Africa, using the province of KwaZulu-Natal as a case study.

The analysis presented in this research work explains an apparent anomaly frequently observed on many agricultural contracts which manifest itself in the principal’s use of seemingly uniform contracts for the purposes of governing the relationships with heterogeneous agents. Previous work on adverse selection in a similar environment to the study being reported in this paper has been done by Leegomonchai and Vukina (2005). In their work they tested whether chicken companies allocated production inputs of varying quality 1

Land restitution is aimed at restoring land ownership to indigenous communities who were dispossessed of their land during colonialization and apartheid. Land redistribution seeks to give more land ownership to previously disadvantaged people through the distribution of state land and purchase of private land through a willing seller willing buyer principle.

International Food and Agribusiness Management Review

394

Mkhabela

Volume 21, Issue 3, 2018

by providing high ability agents with high quality inputs or by providing low ability agents with high quality inputs. The first strategy of providing high quality inputs to high ability agents is meant to stimulate the career concerns type of response on the part of the growers (agents), whereas the second strategy would generate a ratchet effect. Leegomonchai and Vikuna (2005) found no significant input discrimination based on grower abilities that would lead to either career concerns or ratchet effect type of dynamic incentives.

Empirical tests of contract theory are typically done with either cross-industry and cross-firm data or intra-firm data. The first approach of using cross-industry or cross-firm data can provide more generalizable empirical results but faces the econometric problem of unobserved heterogeneity (Chiappori and Salanié, 2003). The second approach of using intra-firm data will generate specific results that are difficult to generalize but has the advantage of dealing with agents that operate in the same environment, thus drastically reducing the potential of unobserved heterogeneity that is rife under the first approach. The research being reported in this paper belongs to the second category of studies. The data come from records of companies that contract the production of fresh produce (vegetables) with independent farmers.

The main objective of the research was to identify and quantify moral hazards and adverse selection in selected farmer/agribusiness partnerships in KwaZulu-Natal. Subsidiary objectives were to develop and refine methodology for measuring dual moral hazard and to recommend approaches to ameliorating moral hazard and adverse selection in joint venture partnerships in agribusiness.

2. Conceptual framework The development of the theory of incentives has been a major advance of economics in the last forty years.

Conflicting objectives and decentralized information are the two basic ingredients of incentive theory. That economic agents pursue at least to some extent their private interests is the essential paradigm for the analysis of market behaviour by economists. What is proposed by incentive theory is to maintain this major assumption in the analysis of organizations, small number of markets and any other kinds of collective decision.

The tenant of incentive theory therefore is tantamount to the problem of delegation of a task to an agent with private information. This private information can be of two types: either the agent can take an action unobserved by the principal, the case of moral hazard or hidden action; or the agent has some private knowledge about his/her cost or valuation that is ignored by the principal, the case of adverse selection or hidden knowledge.

The theory studies when this private information is a problem or the principal, and what is the optimal way for the principal to cope with it. Another type of information problem has also been raised in the literature, the case of non-verifiability where the principal and the agent share ex post the same information but no third party and, in particular, no Court of Law can observe this information (Aghion and Bolton, 1987). One can study to which extent the non-verifiability of information is also problematic for contractual design.

3. Methodology Relationships between agribusiness firms and farmers often focus on the fundamental asymmetry of information between principals (agribusinesses), and agents (farmers). The key variable that is only partially revealed to agribusinesses by primary producers is the amount of effort the farmers put forth in trying to achieve production objectives. Agribusinesses’ net income, or, in the case of cooperatives, farming trusts or community property associations, total net benefit rises in farmers’ effort, but falls in farmers compensation.

Farmers’ rewards are assumed to be proportional to the total amount of benefits (effort) that they create, so the problem from the agribusinesses’ perspective is to choose an incentive plan that implements the optimal amount of effort from farmers in order to maximize total net benefits (Grossman and Hart, 1983). −

−

−

Q = f (S, P) = λ1S−λ2P with λ1 > 0, λ2 > 0

(1)

From (1), the inverse demand function facing the processor i is given by: International Food and Agribusiness Management Review

395

Mkhabela

Volume 21, Issue 3, 2018 −

−

−

−

P = f (S, Q) = aS − b1Q with a = λ1 / λ2, b = 1 / λ2

(2)

The cost function: In economics, any choice between two or more activities comes at a cost and this is normally referred to as opportunity cost. The opportunity cost principle holds true for the demand function facing the processor. There is a cost associated with each effort because it is unpleasant and forgoes the opportunity to undertake other activities. Input production costs, c, are a function of the efforts in quantity and quality. As is traditionally the case in models of this kind, we assume marginal production costs regardless of quantity effort volume and quadratic in line with the given level of quality effort (Equation 7):

c = c1

qs 2 with c1 > 0 2

(3)

And similar to the models of franchise, we assume that the private cost of effort for the processor is the same as for the grower. Then, the processor´s cost, C, will be: −

Qs 2 with c1 > 0 C = c1 2

(4)

3.1 The double moral hazard model The principal-agent theory postulates that when the risk-averse agent faces a trade-off between the provision of incentives and risk sharing, an outcome-conditioned sharing contract can be a second-best pay scheme (Holmström and Milgrom, 1987; Stiglitz, 1974). This tenant of the principal-agent theory implies a dual compensation scheme, w, consisting of (1) a fixed payment, α, that is independent of the observed outcome, and (2) an incentive payment that equates to a positive share, β, of the publicly observable outcome: w = α + βxP (5) In this model, revenue is the performance indicator as it has been shown by (Rubin, 1978) that when the principal has a greater potential to impact on retail demand through branding revenue sharing contracts are better instruments to provide appropriate incentives than profit sharing contracts.

In most cases the principal also provides some effort which invariably affects the outcome thus the incentive provision for both the agent’s action and the principal’s own effort level must be recognized in designing the agent’s incentive scheme. In this regard, the processor chooses the parameters of the incentive scheme, α and β, to maximize her/his expected profit subject to the constraints that both processor and agent choose individually their efforts to maximize their certainty equivalent and the grower attains at least his/her – reservation utility, U­i, such that:

MaxCE

Pr ocessor

α ,β

st

MaxCE 1

= (1 − β )QP − C − α

Pr ocessor

q,s

= α + β QP − c −

(6)

ρ 2 σ βQP 2

(7)

(Grower’s incentive compatibility constraint)

α + β QP − c −

ρ 2

σ β2QP ≥ Umin

(8)

(Grower’s reservation constraint)

International Food and Agribusiness Management Review

396

Mkhabela

Volume 21, Issue 3, 2018

MaxCE b

Pr ocessor

= (1 − β )QP − C − α

(9)

(Processor’s incentive compatibility constraint) The optimization problem is sequentially in Equation 6-9. First, the effort choice made by the processor is determined:

MaxCE Pr ocessor Qb 2 = (1 − β )Q(a (θ b + s ) − c1 − α (10) b 2 The optimal solution to the processor’s decision of effort Equation 10 is:

ACE Pr ocessor (1 − β )aθ = (1 − β )Qaθ − c1Qb = 0 → b = ∂b c1

(11)

Second, given the processor’s choice, the efforts in quantity and quality that maximize the grower’s certainty equivalent are determined:

MaxCEq ,1s st Pr oducer = α + β Q (a(θ b + s) − b1Q) − c1

(12)

Qs 2 ρ 2 2 2 2 2 − β Q a s σs 2 2

The first-order necessary conditions for maximizing Equation 12 with respect to q and s yield:

s2 − ρβ 2 qa 2 s 2σ s2 = 0 (13) ∂q 2 ∂CE 1st Pr oducer = β qa − c1 qs − ρβ 2 q 2 a 2 sσ s2 = 0 (14)

∂CE 1st Pr oducer = β (a (θ b + s ) − 2b1 q ) − c1 ∂s

The reaction functions derived from the maximization problems thus defined are:

q=

β (a (θ b + s )) − c1

s2 2

(15)

β (2b1 + ρβa 2 s 2σ s2 βa (16) s= c1 + ρβ 2 qa 2σ s2

Substituting the previous expressions, q=f(β) and s-f(β), into Equation 6 and 8 and choosing α and β, KuhnTucker conditions reveal a boundary solution with CE1st Producer=Umin, in Equation 8 implying:

α = − βQP + c +

ρ 2

2 + U min σ BQP

(17)

Finally substituting Equation 17, b=f(β), q=f(β) and s=f(β) into Equation 6 and maximizing with respect to β, the value of β optimal may be obtained.

It should be stated ab initio that the principal-agent model defined in Equation 1 has been widely used to analyse numerous issues in economics, including in agriculture (Just and Pope, 2002; Richards et al., 1998; Viaggi et al., 2009). Despite the widespread use of the principal-agent model in analysing partnerships such as share-cropping contract, contract farming and joint venture, explicitly solving the first conditions that define the decision variables in the contract remains largely elusive. The afore-mentioned difficulty and paucity of empirical studies on principal-agent model notwithstanding, the importance of quantitative approaches to

International Food and Agribusiness Management Review

397

Mkhabela

Volume 21, Issue 3, 2018

analyse the phenomena of moral hazard and adverse selection in real life cases remains indispensable thus the ability to generate numerical data is a key step in solving this conundrum.

This section attempts to model double moral hazard using a range of observations as scenarios from 12 selected cases of joint ventures and contract farming contracts in KwaZulu-Natal, South Africa. The observations were derived from experiences of working with these 12 cases over a period of four years and continuous engagements with principals and agents involved. The engagements included prolonged facilitation of negotiations between the principals and agents on issues of the pricing mechanisms for the products and the accrual of dividends. The qualitative data derived from field were analysed through Mathematica (Wolfram Research, Inc., 2016) to solve the model as outlined in the methodology section and to produce quantitative data. The data produced by Mathematica were then used in Matlab (The MathWorks, Inc., 2012) to draw the planes depicting the most suitable representation of the behaviours of both the principal and agent. After several iterations of running the model through Mathematica, the following parameters were chosen: a=1, b1=0.00001 and c1=1. It should be noted that it is important to calibrate the model first before settling on the parameters to be used based on the observations collected from the cases being studied.

A vital feature of this research work is capturing the effect of the principal’s effort on quality of produce. Thus the effort – quality relationship was –amended to include both the grower’s effort (agent) and the processor’s – effort (principal) as follows: S=θb=s, where parameter θ>0 is a proxy for the importance of the principal’s effort. Therefore, there are three free parameters in the model, i.e. the processor’s efficiency factor (θ); the grower’s coefficient of absolute risk aversion (ρ); and the variance of input quality (σ2s). The latter two parameters can be jointly identified as ρσ2s, since both parameters act on the producer’s risk premium in a similar manner. A free parameter is a variable in a mathematical model which cannot be predicted precisely or constrained by the model and must be estimated experimentally or theoretically and is a variable that can be adjusted to make the model fit the data (Calvert et al., 2004; Kline, 2015).

Scenarios were used in the analyses and presentation of the results. The first to be considered is a scenario in which the agent is risk-neutral.

4. Results 4.1 Scenario 1: risk-neutral agent and principal The first step of the analysis was to compute the solution to the agency problem assuming risk neutral agents.

A wide range of 0 to 0.9 in intervals of 0.1 for the efficiency factor θ was considered.

Figure 2 shows how the share of the outcome, β, varies as a function of the importance of the principal’s effort, θ. The results are consistent with the postulations of the agency theory (Bamberg and Spremann, 1989; Eisenhardt, 1989; Fama, 1980), the value of β is at its maximum at when the processor has no interest in the quality of the produce and its value is 1, implying that the agent receives all the revenue accruing from the sale of the produce. Similarly, and congruent to the predictions of franchise models, the share of the outcome β is decreasing in θ. This finding resonates with the findings of other previous studies elsewhere such as Lafontaine (1992) and Holmström and Milgrom (1991). The finding which is also collaborated by others is that when the franchisor inputs are more important, less vertical separation is observed, as predicted.

The results pertaining to the importance of the grower’s effort are shown in Figure 2. Figure 3 clearly shows a positive relationship between the grower’s effort in quantity and the principal’s effort. Thus, the grower’s effort in quantity varies as the importance of the principal’s effort increases. The intervals of θ range from 0 to 0.9. The quantity input curve appears smooth, gradual and somewhat concave with a minimum of θ=0.5.

International Food and Agribusiness Management Review

398

Mkhabela

Volume 21, Issue 3, 2018 1.2 1 0.8 β

0.6 0.4 0.2 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0.5

0.6

0.7

0.8

0.9

0.5

0.6

0.7

0.8

0.9

θ

Figure 2. The share of the outcome (β).

100,000 80,000 60,000 q

40,000 20,000 0 0

0.1

0.2

0.3

0.4 θ

Figure 3. Quantity input (q).

3 2.5 2 s

1.5 1 0.5 0 0

0.1

0.2

0.3

0.4 θ

Figure 4. Expected input quality (s).

The results of the quality efforts of both the agent and the principal are shown in Figure 4 and 5, respectively.

Considering the shapes of the efforts curves as a function of the efficiency factor, θ, gives useful insights into the relationship between the quality efforts of both the agent and principal. Interestingly, the results show that when the processor’s puts more importance on the quality of the produce, the grower expends less effort. This finding is counter intuitive but is indicative of the problem of moral hazard and invokes the International Food and Agribusiness Management Review

399

Mkhabela

Volume 21, Issue 3, 2018

1.2 1 0.8 0.6 b

0.4 0.2 0 -0.2 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

θ

Figure 5. Processor’s effort. principle of free-riding as the grower now depends more on the effort of the processor. Contrariwise, the processor’s effort in ensuring acceptable quality is increasing in θ, as it can be deducted from Equation 10.

4.2 Scenario 2: risk-averse agent and risk-neutral principal In order to model this risk-averse agent and risk-neutral principal, the parameter estimates were obtained by searching over an equi-distance spaced grid of 100 values for each and every parameter ranging from 0 to 0.9 for θ and 0 to 0.00009 for ρσ2s.

Figure 6 shows that as the producer’s risk premium (ρσ2s), that is risk aversion or quality variance, increases, given the value of the θ, the share of the outcome, β, decreases.

This finding is collaborated by the postulation of the principal-agent framework with risk-averse agents as discussed by Holmström and Milgrom (1991). The Holmström and Milgrom (1991) model states that: If one individual has more than two tasks to perform and performance of a task is not well measurable, the

Figure 6. The share of the outcome (β).

International Food and Agribusiness Management Review

400

Mkhabela

Volume 21, Issue 3, 2018

implementation of incentives lead to a misallocation of effort. The individual will concentrate on that task for s/he receives the strongest incentives.

The principal-agent theory as discussed in Holmström (1979) offers the canonical model including a hidden one-shot action taken by an agent contracted to provide effort. Furthermore, Holmström and Milgron (1987) were able to provide proof of the optimality of linear reward schemes2. The sequence-of-actions model has a corresponding (static), highly tractable companion (Lundesgaard, 2001).

In general, an increase in the importance of the principal’s effort, given the value of the agent’s risk premium, will decrease the incidence of β. However, this result is not robust when the efficiency factor converges to zero, in which case the value of β increases. Similarly, Figure 7 depicts the behaviour of the input quantity as a function of the efficiency factor, θ, and risk premium, ρσ2s. When the efficiency factor (θ) converges to zero, an increase in the risk premium (ρσ2s) will decrease the grower’s effort in quantity. Conversely, when the efficiency factor (θ) increases above zero, the input quantity increases and furthermore the negative incidence of an increase in the risk premium (ρσ2s) on input quantity declines.

The quality efforts of both the agent and principal are represented in Figure 8 and 9, respectively. Taking a cursory look at both figures leads one to clearly deduce that in most cases, as represented by the observations in this study, the quality efforts of the agent and principal are divergent in that they vary in opposite directions.

Simply put, when the efficiency factor (θ) and/or the risk premium (ρσ2s) increase, the processor’s effort increases; on the contrary, the primary producer’s effort in quality decreases. A departure from the general trend observed was when the efficiency factor (θ) converged to zero, in which case both the agent and principal’s efforts increased as the risk premium (ρσ2s) increased.

2

In the one-shot model, reward schemes are never linear.

Figure 7. Quantity input (q).

International Food and Agribusiness Management Review

401

Mkhabela

Volume 21, Issue 3, 2018

Figure 8. Expected input quality (s).

Figure 9. Processor’s effort (b).

5. Discussion Most previous studies in moral hazard ignore the fact that the agent makes both efforts in quantity and quality. The results presented in this paper shed some light on the importance and value of including in a moral hazard analysis model that in reality the agent expends effort in both quantity and quality.

The results of this analysis give some pointers as to why contracts different from share cropping contracts are often used for products where quality is a key competitive attribute and a distinguishing factor as it is currently the case in South Africa with the advent of large multinational and local agribusiness firms in the agricultural sector in partnership with land reform beneficiaries. It is common knowledge that in many highly differentiated products markets the consumers do not automatically know the quality of the product nor the accuracy of the information supplied about the characteristics of the products and this referred to as ‘lemon problem’ (Akerlof, 1970). The ‘lemon problem’ is the issue of information asymmetry between the International Food and Agribusiness Management Review

402

Mkhabela

Volume 21, Issue 3, 2018

buyer and seller of an investment or product as popularized by a 1970 research paper by economist Goerge Akerlof. This observed asymmetric information between the processors and consumers has the potential to negatively impact the functioning of the product market (Akerlof, 1970).

The findings of this study should be viewed in the context of sub-Saharan Africa. Sub-Saharan Africa (SSA) has the lowest adult literacy rate. Huebler (2007) observed that SSA had an adult literacy rate of 63% in 2008. The relationship between a literate society and economic development is well documented (Jogwu, 2010). The ability of market participants to engage in effective and informed interaction in the market place is critical since, especially given that countries in SSA are transitioning from command economies to marketled development where illiterates interact with literates (African Progress Panel, 2015). The literate-illiterate interaction may engender information asymmetry in the contracting, that is, a situation where one of the parties to a contract, the literate party, has information that is not available to the other party, the illiterate party (Boadu, 2016).

The reality of the interactions at the market place gives rise to the two common problems associated with information asymmetry namely adverse selection and moral hazard. In adverse selection, the illiterate party may negotiate a contract that actually hurts their interests. For example, in a number of the cases studies in KwaZulu-Natal, South Africa, primary producers of fresh produce in partnership (joint ventures) with established agribusinesses tend to accept lower payment for their produce thinking they are getting a good deal believing that their produce is of inferior quality when in fact it meets the required quality standards.

With moral hazard, the illiterate party lacks the ability to enforce terms of the contract. Contract that face these problems are also susceptible to the problem of the tragedy of commons. The tragedy of the commons is an economic theory of a situation within a shared-resource system where individual users areacting independently according to their own self-interest behave contrary to the common good of all users by depleting that resource through their collective action (Hardin, 1968). Such contracts cannot be valuemaximising and often third party intervention is required to fill the gap thus levelling the playing field. The challenge is to define contract rules that do not destroy incentives on the part of the literate parties to interact with illiterate parties in the market place.

Moral hazard and adverse selection in big public-private sector partnerships and large scale land reform projects are skewing benefits in favour of the privileged elite who are more powerful due to their political connectedness, higher than average education and economic standing in society. The foregoing notwithstanding, there is a general consensus that there is a prominent role for large scale commercial agriculture and private sector (agribusinesses) to play in contributing towards the attainment of national and development goals in Africa. There is a school of thought that postulates that moral hazard can be prevented by a combination of incentives and constraints (Xion et al., 2013). Monetary incentives such as profit, dividends and bonus payment for meeting and exceeding quality and quantity requirements are known to minimize the risk of moral hazard. Devolving more decision-making and handling of finances by the farmer as opposed to strict management by the processor were also found to incentivize the producer to be more involved thus negatively affecting moral hazard. Constraints factors that were found to have a negative bearing on moral hazard included production environmental supervision and supervision and enforcement of general good agricultural practice.

6. Conclusions A number of useful conclusions can be drawn from the gleanings and nuances obtained from the results of this study. Of uppermost importance to investors in agribusiness, managers, policy makers and implementers of agricultural development programmes is an understanding of what makes partnerships work and what leads to premature termination of such partnership schemes in the South African agricultural sector environment.

Moral hazard and adverse selection are major limitations for joint ventures in agribusiness in Africa, in general, and South Africa, in particular. There is general information asymmetry between the principal (agribusiness) and the agent (farmers/primary producers) which leads to an unequal relationship which generates mistrust International Food and Agribusiness Management Review

403

Mkhabela

Volume 21, Issue 3, 2018

and suspicious between the partners. Furthermore, the recent history of South Africa, where land has been predominantly owned by one particular race group (whites) engenders mistrust now that substantial amount of land has been transferred to the black population group and the latter is being assisted by the government to make their land productive.

The conclusions point to several areas of intervention and investment opportunities for the public sector and agribusiness entrepreneurs, respectively. The provision of and investment in basic adult literacy, including and business management literacy and numeracy, for emerging commercial farmers is desirable. The study found that there is unequal yoking of agents and principals in agribusiness with the scales tilted in favour of the principal who are often educated, thus sophisticated, at the detriment of the agents (farmers) who are, in the main, illiterate or semi-literate. The investment in literacy improvement would enable the farmers to access and better synthesize available information to make better business decisions and better engage with the agribusiness partners.

The provision and availability of more market information such available markets, real time prevalent product prices, quality and applicable standards (such as phyto-sanitary, accreditation and traceability) to narrow the information gap between the farmers and agribusiness partners presents another opportunity for both investors and policy makers. Investment in information and communication technology such mobile phone apps and similar platforms to provide farmers with both marketing and agronomic information is one avenue that could be explored. South Africa has relatively good cell phone network coverage and most farmers have smart phones that could be used optimally to better inform the farmers thus empowering them to make informed choices.

It is generally accepted that public sector provided agricultural extension services have virtually collapsed in Africa leading to a paucity of technical information for farmers, especially emerging commercial farmers.

The gap created by the insufficiency of government provided extension services presents an opportunity the private sector to provide technical production support to farmers and agribusiness sector. Input supply and commodity linked extension services and cost recovery from the farmers either directly or through cession on the proceeds from the sale of produce presents a viable option that has worked elsewhere in the world.

Joint ventures and other forms of partnership providing real equity and shares to farmers in the agribusiness are a viable option. Governments in Africa, and this is particularly true in South Africa, are willing to acquire shares in agribusinesses on behalf of the farming community through investing public pension funds and other related investments. This approach makes business sense in that it increases a sense ownership of the agribusiness by the farmers thus militating against moral hazard and incentivises performance and adherence to contractual obligations.

Some of the nuances from the study point towards the need for a neutral third-party mediator to solve contractual disputes and provide adjudication between contracting parties while providing an assurance of fairness and a favourable environment to re-negotiate terms when the need arises. This arbitration and mediation role naturally lends itself to be the domain of government and policy space. However, there is room for the private sector to support the capacity of the public sector through providing transaction advisory services and providing international best practice. Transaction advisory and mediation services are indispensable if joint ventures and similar business models are to successfully take-off and thrive in Africa. The existence of relative safety (protection and safe-guarding of interests) and stability for both contracting parties and the provision of a mutually-agreed arbitration process are necessary pre-conditions. Courts of law can provide enforcement of contractual obligations but this is often a tedious and expensive options thus the need for a more expedient and affordable option.

International Food and Agribusiness Management Review

404

Mkhabela

Volume 21, Issue 3, 2018

References African Progress Panel. 2015. Power, people and planet: seizing Africa’s energy and climate opportunities.

African progress report 2015. Available at: http://tinyurl.com/prvpwvm.

Aghion, P. and P. Bolton. 1987. Contracts as a barrier to entry. American Economic Review 77: 388-401.

Akerlof, G.A. 1970. The market for ‘lemons’: quality uncertainty and the market mechanism. Quarterly Journal of Economics 84(3): 488-500.

Bamberg, G. and K. Spremann (eds.). 1989. Agency theory, information and incentives. Springer-Verlag,.

Berlin, Germany.

Boadu, F.O. 2016. Agricultural law and economics in sub-Saharan Africa: cases and comments. Elsevier, London, UK. Calvert, G., C. Spence and B.E. Stein. 2004. Handbook of multisensory processes. MIT Press, Massachusetts, MA, USA. Chiappori, P.A. and B. Salanié. 2003. Testing contract theory: a survey of some recent work. In: Advances in economics and econometrics – theory and applications, edited by M. Dewatripout, L. Hansen and P. Turnovsky (eds.). 8th World Congress, Econometric Society Monographs. Cambridge University Press, Cambridge, UK, pp 115-149.

Cook, M.L. and F.R. Chaddad. 2000. Agroindustrialization of the global agrifood economy: bridging development economics and agribusiness research. Agricultural Economics 23: 207-218.

Eisenhardt, K.M. 1989. Agency theory: an assessment and review. The Academy of Management Review 14(1): 57-74.

Fama, E.F. 1980. Agency problems and the theory of the firm. The Journal of Political Economy 88(2): 288-307.

Goodhue, R.E. 1999. Input control in agricultural production contracts. American Journal of Agricultural Economics 81(3): 616-620.

Grossman, S. and O. Hart. 1983. An analysis of the principal-agent problem. Econometrica 51: 7-45.

Hardaker, J.B., L. Gudbrand, J.R. Anderson and R.B.M. Huirne. 2015. Coping with risk in agriculture: applied risk analysis (3rd ed.). CABI, Boston, MA, USA. Hardin, G. 1968. The tragedy of the commons. Science, New Series 162(3859): 1243-1248.

Henson, S. and R. Loader. 2001. Barriers to agricultural exports from developing countries: the role of sanitary and phyto-sanitary requirements. World Development 29(1): 85-102.

Holmström, B. 1979. Moral hazard and observability. The Bell Journal of Economics 10(1): 79-91.

Holmström, B. and P. Milgrom. 1987. Aggregation and linearity in the provision of intertemporal incentives.

Econometrica 55(2): 303-328.

Holmström, B. and P. Milgrom. 1991. Multi-task principal agent analysis: Incentive contract, asset ownership and job design. Journal for Law, Economics and Organisation 7: 26-52.

Huebler, F. 2007. International education statistics: official school ages: primary, secondary and compulsory education. Available at: http://tinyurl.com/y8m9hrlv.

Janssen, W.G. and A. Van Tilburg. 1997. Marketing analysis for agricultural development: suggestions for a new research agenda. In: Agricultural marketing and consumer behavior in a changing world, edited by B. Wierenga, A. Van Tilburg, K. Grunert, J.B.E.M. Steenkamp and M. Wedel. Kluwer, Boston, MA, USA, pp. 57-74.

Jogwu, C.N.O. 2010. Adult illiteracy: the root of African underdevelopment. Education 130(3): 490-498.

Just R.E. and R.D. Pope. 2002. A comprehensive assessment of the role of risk in U.S.A. agriculture. Kluwer Academic Publishers, New York, NY, USA. Kline, R.B. 2015. Principles and practice of structural equation modeling. Guilford Publications, New York, NY, USA. Kohls, R.L. and J.N. Uhl. 1998. Marketing of agricultural products. 8th Edition. Simon & Schuster/A Viacom Company, Upper Saddle River, NJ, USA. Kvaløy, O. 2006. Self-enforcing contracts in agriculture. European Review of Agricultural Economics 33(1): 73-92.

International Food and Agribusiness Management Review

405

Mkhabela

Volume 21, Issue 3, 2018

Lafontaine, F. 1992. Agency theory and franchising: some empirical results. The RAND Journal of Economics 23(2): 263-283.

Lazzarini, S.G., F.R. Chaddad and M.L. Cook. 2001. Integrating supply chain and network analyses: the study of netchains. Journal of Chain and Network Science 1(1): 7-22.

Leegomonchai, P. and T. Vukina. 2005. Dynamic incentives and agent discrimination in broiler production tournaments. Journal of Economics and Management Strategy 14(4): 849-877.

Lundesgaard, J. 2001. The Holmström-Milgrom model: a simplified and illustrated version. Scandanavian Journal of Management 17(3): 287-303.

Machethe, C., T. Reardon, and D. Mead. 1997. Promoting farm – nonfarm linkages for empowerment of the poor in South Africa: a research agenda focused on small-scale farms and agroindustry. Development Southern Africa 14: 377-394.

Masuku, M.B., J.F. Kirsten and R. Owen. 2007. A conceptual analysis of relational contracts in agribusiness supply chains: the case of the sugar industry in Swaziland. Agrekon 46(1): 94-115.

Newman, C. and B.C. Briggeman. 2016. Farmers’ perceptions of building trust. International Food and Agribusiness Management Review 19(3): 57-76.

Otsuki, T., J.S. Wilson and M. Sewadeh. 2001. Saving two in a billion: quantifying the trade effect of European Food Safety Standards on African exports. Food Policy 26(5): 495-514.

Porter, M.E. 1990. The comparative advantage of nations. MacMillan, Basingstoke, UK. Reardon, T., J-M. Codron, L. Busch, J. Bingen and C. Harris. 1999. Global changes in agrifood grades and standards: agribusiness strategic responses in developing countries. International Food and Agribusiness Management Review 2(3/4): 421-435.

Richards, T.J., K.K. Klein and A.M. Walburger. 1998. Principal-agent relationships in agricultural cooperatives: an empirical analysis from rural Alberta. Journal of Cooperatives 13: 21-34.

Ruben, R., M. Slingerland and H. Nijhoff. 2006. Agro-food chains and networks for development: issues, approaches and strategies. In: Agro-food chains and networks for development, edited by R. Ruben, M. Slingerland and H. Nijhoff. Springer, Dordrecht, the Netherlands, pp. 1-25.

Rubin, P. 1978. The theory of the firm and the structure of the franchise contract. Journal of Law and Economics 21: 223-233.

Shaban, R.A. 1987. Testing between competing models of sharecropping. The Journal of Political Economy 95(5): 893-920.

Stiglitz, J. 1974. Incentives and risk sharing in sharecropping. Review of Economic Studies 41: 219-255.

Terblanche, S.E., J.B. Stevensand and M.G. Sekgota. 2014. A comparative analysis of two land reform models: the Mashishimale farm management model and the Nkumbuleni strategic partnership model, South Africa. South African Journal of Agriculture Extension 42(2): 81-102.

The MathWorks Inc. 2012. Matlab and statistics toolbox release. Natick, Massachusetts, MA, USA. Van der Laan, H.L. 1993. Boosting agricultural exports: a marketing channel perspective on an African dilemma. African Affairs 92: 173-201.

Viaggi, D., F. Bartilini and M. Raggi. 2009. Combining linear programming and principal-agent models: an example from environmental regulation in agriculture. Environmental Modelling and Software 24(6): 703-710.

Wilson, P.N. 2000. Social capital, trust, and the agribusiness of economics. Journal of Agricultural and Resource Economics 25(1): 1-13.

Wolfram Research Inc. 2016. Mathematica, Version 10.4. Wolfram Research, Champaign, IL, USA. World Bank. 2013. Growing Africa: unlocking the potential of agribusiness. AFTFP/AFTAI. The World Bank, Washington, WA, USA. Available at: http://tinyurl.com/ldpb9cv.

Xion, Y., Z.C. Lu and Y. Ding. 2013. Prevention of farmers’ moral hazard in safe farming in China: by incentives or constraints? Journal of Agribusiness in Developing and Emerging Economies 3(2): 131-150.

International Food and Agribusiness Management Review

406

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2017.0037 Received: 1 May 2017 / Accepted: 27 December 2017

Fertilizer distribution flows and logistic costs in Brazil: changes and benefits arising from investments in port terminals RESEARCH ARTICLE Débora da Costa Simões a, José Vicente Caixeta-Filhob, and Udatta S. Palekarc aConsultant,

Agroconsult, Rua Cônego Vicente Miguel Marino, 275 #234 C, Barra Funda 01135-020, São Paulo-SP, Brazil

bProfessor,

“Luiz de Queiroz” Agricultural College (ESALQ), University of São Paulo (USP), Av. Pádua Dias 11, São Dimas 13418-900, Piracicaba-SP, Brazil

cProfessor,

University of Illinois at Urbana-Champaign, College of Business, 4016 BIF, 515 East Gregory Drive, 61820 Champaign, IL, USA

Abstract This study analyzes the impact of anticipated investments to alter Brazilian port infrastructure on fertilizer flows and fertilizer transportation logistics costs using a linear programming model designed for the task.

The most notable among these investments are directed toward accelerating port development in Brazil’s “Northern-Arc”, thereby increasing fertilizer supply to new markets opening throughout the country’s expanding agricultural frontier, particularly in northern Mato Grosso state, while increasing supply to existing markets. Results from model runs show that the anticipated port infrastructure investments should ensure nationwide fertilizer logistics savings of US$ 845 million over the 2017 through 2025 period. Although these estimated benefits are outstanding, the study indicates that further expansion of Brazil’s port system, particularly in the Northern-Arc, presents additional opportunities. Model projections were that in 2025, after all planned infrastructure improvements are operational, port terminals will be near full capacity, which should make planning for future projects a current priority.

Keywords: fertilizer market, logistic flows, mathematical programming, Northern-Arc JEL code: C61, R41, R42 Corresponding author: deborasimoes@gmail.com

© 2018 Da Costa Simões et al.

407

Da Costa Simões et al.

Volume 21, Issue 3, 2018

1. Introduction Primarily due to rapid agricultural production growth, the Brazilian fertilizer market grew from 16.4 million metric tons in 2000 to 32.2 million metric tons in 2014 (National Association of Fertilizer Diffusers – ANDA, 2015). This impressive growth rate calls for logistical investments to increase the country’s limited capacity to supply the fertilizer needed to meet increasing future demand. The challenge has been to demonstrate this reality to both public and private investors.

Over coming years, Brazilian agribusiness is expected to continue growing. Agroconsult (personal communication) estimates that by 2025 Brazilian grain production may surpass 310 million metric tons, an increase of more than 100 million metric tons from the 2015 figure. Over those ten years, the country will have been responsible for a 56% increase in the world’s supply of soy and a 16% increase in the world’s supply of corn. Given that Brazil has the largest amount of unexplored arable land in the world, a supportive government, presents favorable climatic and soil conditions for agriculture, and can competitively produce a wide range of products, these estimates are reasonable. Even with these advantages, Brazilian agricultural producers still need to prepare and fertilize the poor, acidic soils that predominate in the country. Consequently, the foreseen growth of Brazilian agricultural production, both through expansion and yield improvement, will certainly lead to a proportional increase in fertilizer demand, making the country’s fertilizer supply chain crucial to food security.

In 2015, Brazil was the world’s fourth largest fertilizer consumer by volume, behind China, India and the United States (IFA, 2015). Agroconsult (personal communication) estimates that Brazil will demand 43.6 million metric tons of fertilizer by 2025, an additional volume of 11.4 million metric tons over the 2014 value.

The majority of Brazil’s fertilizer demand is satisfied through importation and it is estimated that demand for imported fertilizer will reach 33.1 million metric tons in 2025 (therefore, 76% of the country´s total demand), an increase of 38% over the volume imported in 2014 (Agroconsult, personal communication). In the coming years, fertilizer imports will continue to be needed to sustain Brazilian agricultural production growth.

The reliance on imports to satisfy the internal fertilizer market adds steps to the supply chain, making logistics planning crucial to transportation and storage company profits and to help insure that farmers can negotiate attractive pricing. The planning involved to economically and efficiently transport fertilizer in Brazil is both complex and often unrewarding due to the country’s degraded, muddled, and problematic logistics infrastructure. The costs to import fertilizer represent a significant percentage of the end consumer’s price and the Brazilian port complex adds an implicit cost: demurrage.

Besides the normal charges involved in the import process (maritime shipping, insurance, port expenditures and the Brazilian additional freight for the renovation of the merchant marin tax – AFRMM), the fertilizer industry imports often incur supplementary demurrage costs, i.e. fines for delays that occur when vessels are prevented from berthing and discharging cargo within the stipulated lay days and time. In Brazil, demurrage charges are always a possibility. Some Brazilian port delays are predictable, occurring due to seasonal port congestion (most often between June and September), others arise as the unpredictable aftereffect of a set of operational problems at ports, such as a lack of space for berthing, insufficient storage capacity, lack of equipment, or bureaucratic impediments.

Using the average prices for the main fertilizer products imported into Brazil in 2014 as a basis for comparison, Brazilian overland shipping expenditures alone add an additional 21.1% to the price of urea; 15.4% to the price of monoammonium phosphate (MAP), and 22.2% to the price of potassium chloride (KCl). After adding the other import logistics costs, such as maritime shipping charges, port costs, demurrage and the AFRMM tax, these percentages reach 40.1, 29.3 and 42.2%, respectively. This group of logistics chain factors penalizes fertilizer consumers as it dramatically raises costs associated with the international movement of products; and the penalty added to Brazilian import costs is larger than most, exceeding that of China and India.

International Food and Agribusiness Management Review

408

Da Costa Simões et al.

Volume 21, Issue 3, 2018

In the 2016 Logistics Performance Index survey from the World Bank (2016), Brazil occupies the 55th position in a ranking of 160 countries, having scored 3.09 out of 5.00 points. In this same year, China scored 3.66 (46th place) and India scored 3.42 (35th position). When compared with the scores from other years, this score shows that Brazil’s performance is, at best, erratic, having fluctuated between 41st (2012) and 65th (2014). Moreover, of the six criteria that each country was judged by, aspects directly related to imports (customs and international shipment) had the lowest Brazilian scores in 2016.

Taking into consideration the recent policies driven by the Brazilian government, this study embodies three main measures that may reduce the Brazilian fertilizer industry’s total logistics costs: (1) improvements in port operations and processes intended 7to decrease demurrage expenses; (2) greater usage of the country’s railways and waterways – which in 2015 account for less than 5% of total fertilizer flows within Brazil; and (3) investment in new structures and port capacity, especially focusing on the development of the “NorthernArc”1 corridors that will help reduce bottlenecks at Brazil’s southern ports and may loosen supply in some markets.

This study will analyze the effect of new investments in Brazilian port logistics infrastructure on the internal flows and final consumer cost of imported fertilizers from 2014 through 2025. It also contains an estimate of the potential volume of fertilizer that can arrive through the Northern-Arc ports alone and measures potential gains to Brazilian farmers, especially in Mato Grosso state, should they begin receiving imported fertilizer from these northern ports rather than southern ports.

2. Regional demand for imported fertilizer The volume of imported fertilizer brought into each Brazilian mesoregion can be estimated based on the difference between the amount of fertilizer demanded in each market and the corresponding production or consumption of domestically produced fertilizer in that market. Demand was determined at the specific product level (i.e. urea, MAP, KCl, etc.) by mesoregion rather than at the aggregated fertilizer level to minimize estimation errors and produce more robust and accurate results. It was assumed that domestic production is preferred over imports; therefore, all types of fertilizers produced domestically are distributed among the demanding markets, preferably from the nearest site producing the specified type of fertilizer, with the remaining mesoregional demand for that type of fertilizer being supplied by imports.

This study used national fertilizer production data by product and Brazilian administrative region published

by ANDA (2015).2 State level production was then estimated according to the production capacity available in each state using information found in ANDA’s 2014 Yearbook. ANDA also publishes data on total fertilizer raw material deliveries at the country level. The raw material market in each state is estimated by using main nutrient requirement formulas for each crop in a particular mesoregion and ANDA’s historical data regarding nutrient and raw material deliveries.

ANDA’s state, regional, and national data were extrapolated to the 46 mesoregions based on area under cultivation statistics provided by the Brazilian Institute of Geography and Statistics’ (IBGE) database (2015) for 2014 and Agroconsult’s projections for the years from 2015 through 2025 (Agroconsult, personal communication).3 The quantities of imported fertilizer needed to supply demand in each mesoregion through the year 2025 were then calculated under the assumption that imported fertilizers would be used to meet any shortfalls in domestic supply.

1

In this paper, ports in the “Northern-Arc” are those that are operational or planned and located on the Amazon River, one of its tributaries, or the northern Atlantic Coast. They are the ports of Itacoatiara (AM), Santarém (PA), Vila do Conde (PA), Itaqui (MA) and Miritituba (PA) unless otherwise noted.

2 There are five Brazilian administrative regions: North, Northeast, Central-West, Southeast and South.

3 The 46 mesoregions are noted in Supplementary Table S1.

International Food and Agribusiness Management Review

409

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Brazil was also separated into 12 alternative regions. These regions were not purposely aligned with the official Brazilian administrative regions but were determined in consonance with current imported fertilizer flows and state borders.

3. Port capacity, freight and other logistics costs This study uses imported fertilizer data acquired from 14 port complexes that handle 100% of the fertilizer imported into Brazil. Included among these port complexes are those in country’s Northern-Arc (Itacoatiara (AM), Santarém (PA), Vila do Conde (PA), Itaqui (MA) and Miritituba (PA)), which despite current insignificance, is an area expected to gain prominence as an import/export hub over the coming years.

The initial capacity of each port has been defined as the maximum annual volume of fertilizer imported through it from 2011 through 2014, in agreement with information collected from ANDA (2014) and SECEX (2015). The maximum volume of fertilizer coming into Brazil through all these ports in any of those years was 24.4 million metric tons (Table 1). These figures do not include fertilizer raw materials, such as sulfur, phosphate rock and ammonia.

For the purpose of this paper, logistics costs encompass international shipping charges, port direct charges, demurrage, AFRMM tax and internal freight. International maritime shipping prices from the main fertilizer exporting countries to Brazil in 2014 were obtained from Independent Chemical Information Service weekly reports (ICIS) (ICIS, 2014). As ICIS data do not state the individual ports of destination, one price for maritime transport of fertilizers to Brazilian ports has been adopted for all ports other than the inland ports of Itacoatiara and Santarém in Brazil’s Northern-Arc. In these two cases, a charge based on additional time in transit has been added.

Table 1. Estimated port capacities and logistics costs (2014 base).1 Ports

Santos Paranaguá Vitória Rio Grande2 S.F do Sul Imbituba Recife Maceió Aratu Itaqui Vila do Conde Itacoatiara Santarém Miritituba3

Estimated capacity (×1000 t)

Logistics costs (US$/t)

3,747 8,780 1,964 4,620 1,694 89 308 283 977 1,607 130 121 116 0

25.00 25.00 25.00 25.00 25.00 25.00 25.00 25.00 25.00 25.00 25.00 25.00 25.00

Sea freight

Added to river ports

2.25 1.37

Port costs

Demurrage AFRMM4 (25%)

Total costs

28.46 20.00 27.61 14.87 15.76 15.76 14.87 14.87 28.04 12.74 23.36 23.36 23.36

6.00 7.00 4.00 2.00 4.00 4.00 2.00 2.00 8.00 12.00 4.00 0.00 2.00

74.33 65.00 70.77 52.34 55.95 55.95 52.34 52.34 76.30 62.18 65.46 63.27 64.66

1

14.87 13.00 14.15 10.47 11.19 11.19 10.47 10.47 15.26 12.44 13.09 12.65 12.93

Table is calculated with information from databases ICIS, Vessel distance, ANDA, Secex, and Agroconsult.

Rio Grande Complex includes the port of Porto Alegre.

3 The port of Miritituba is mostly attended by barges that follow to or come from the port of Vila do Conde or port of Santarém.

Therefore, it can have both ports as a reference for logistics costs regarding fertilizer clearance. The waterway freight to reach Miritituba terminal has been added as internal freight cost, which will be detailed.

4 AFRMM = Additional freight for the renovation of the merchant marin tax.

2

International Food and Agribusiness Management Review

410

Da Costa Simões et al.

Volume 21, Issue 3, 2018

The average maritime freight rate for fertilizer shipped from foreign ports to Brazilian ports has been estimated to be US$ 25/mt, which is the average freight rate from two relevant fertilizer exporting origins, the ports of Yuznhy, Ukraine and Vancouver, Canada, to the Northern-Arc port of Vila do Conde. The duration of the maritime portion of the import process is also relevant as it is used to estimate the additional cost for fertilizer shipment via waterway from Vila do Conde to the inland Brazilian ports of Santarém and Itacoatiara. The voyage to Santarém takes 17 additional hours, which results in a supplementary charge of US$ 1.37/mt; the voyage to Itacoatiara takes an additional 28 hours, which results in a supplementary charge of US$ 2.25/ mt. The journey time data came from Vessel Distance (2015) and considers the shortest sea route between selected ports.

A 25% AFRMM tax has also been applied to the maritime shipping cost and to overall port costs, including demurrage. All the costs by port of arrival in Brazil and these ports’ fertilizer import capacities are presented in Table 1.

Internal freight costs between Brazilian ports and the study’s 12 fertilizer demanding regions have been calculated based on the following equation developed by ESALQ-LOG (personal communication): Fij = 13.54590 + 0.026130 × Rij + 0.01307 × Wij + 5.00000 × MIj

(1)

where: Fij = total cost to move freight from Brazilian port i to destination j; Rj = total road distance between Brazilian port i, to destination j; Wj = total waterway distance from Brazilian port i, to destination j; MIj = dummy variable associated with additional shipping costs from Brazilian port i, to inland ports (equal to 1 for the ports of Miritituba or Itacoatiara, 0 otherwise).

The waterway mode has been taken into account only for the routes beginning at the port of Itacoatiara and transshipped to the Porto Velho-RO terminal and from the ports of Vila do Conde and Santarém and transshipped to the river port of Miritituba. In these cases, additional shipping costs have been added to the total freight shipping costs shown in Equation (1).

In absence of any official truck road network for trucks or for trucks carrying fertilizer, the distance between a Brazilian port and the fertilizer’s destination was determined using data obtained from Google Maps. As a criterion, for sake of simplicity, it is assumed that quickest routes are sufficient, i.e. the quickest routes have been selected since shipping costs are usually lower in these cases. River distances are based on information from the Brazilian National Water Transport Agency (ANTAQ, 2014).

4. Mathematical model The mathematical structure of linear programming adopted in this study follows the main features of the “transportation problem” optimization model. According to Jensen (1999) and Sharma et al. (2012), the transportation problem seeks to minimize the costs involved in moving products from a number of origins or sources to a set of destinations while satisfying some constraints.

As mentioned by Sharma et al. (2012), the transportation problem was firstly presented by Hitchcock (1941) and further discussed by Koopman (1949) in his reference work about the utilization of the transportation system. Since then, this method evolved in order to include multi-objective problems such as minimizing costs and duration of the flow concomitantly. The same technique might be applied to evaluate the competitiveness and feasibility of existing or new logistic projects. The results obtained can help both private and public sectors to better allocate their scarce resources and improve returns by identifying opportunities, choosing the best locations and planning the most suitable size of the investment, among other benefits derived from measuring and analyzing such flows and costs.

International Food and Agribusiness Management Review

411

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Milanez et al. (2010), for instance, developed a mathematical model to estimate the logistics infrastructure required to support future demand for ethanol in Brazil. Frias et al. (2013) planned the best logistic network for a company from the petrochemical industry whereas Hamad and Gualda (2014) did the same for agribusiness companies but also measuring the impact of seasonality and taxes. Besides, this method is quite often adopted in a number of studies and reports aiming at guiding public policies and new investments in infrastructure, especially focusing on Brazilian grains export flows (ANTT, 2014; Caixeta-Filho, 2010; MAPA, 2013).

There is no tradition, however, in developing optimization models for the fertilizer industry in Brazil. Some few examples come from Carvalho (2009), who developed a model to estimate the best location for fertilizer blending units in the Center-West region, as well as Pereira et al. (2016), who design and implemented a model to adjust and plan the logistic network for nitrogen fertilizers produced by Petrobras due to the company’s new investments in the country.

To achieve the purpose of this study, the model developed is comprehensive. Regarding supply, the model considers 14 points of Brazilian origins. Those 14 origins are comprised by the main Brazilian ports used for current fertilizer importation and the Northern-Arc port complexes, which have the potential to play an important role in the future. The demand side is represented by 46 demanding mesoregions defined by their agricultural profile and current logistic characteristics. For each mesoregion, a municipality of reference has been selected according to its relevance in the mesoregion’s agricultural production. The state of Mato Grosso, for example, was broken into four mesoregions with the following reference cities: North – Sorriso, South – Rondonópolis, East – Canarana, and West – Campo Novo do Parecis. The complete list mesoregions, their corresponding reference municipalities, the states they are located in, and the ports that supply their imported fertilizer can be found in Supplementary Figure S1.

In order to make uncluttered graphic illustration of the study’s results possible and to clarify our tables and discussions, the outcomes from analysis of each of the 46 mesoregions’ data were ascribed to one of this study’s 12 regions depending on the mesoregions location. The regions were defined respecting state borders and the fertilizer supply chain’s current logistics structure. The 12 regions’ boundaries are graphically depicted in Supplementary Figure S2.

The general structure of the imported fertilizer transportation model’s objective function (2 and 3) and its constraints (4 and 5) are presented below: 14 46

MinZ = ∑ ∑ Cij Xij

(2)

i=1 j=1

where: Cij = Si + Pi + Ei + Ti + Fij

(3)

subject to: 14

∑ Xij = Dj , for all j (4)

i=1 46

∑ ≤ 0i , for all i (5) j=1

in which: Z = value of the objective function; Xij = volume of fertilizer transported between Brazilian port i and destination mesoregion j; Cij = total logistic cost from Brazilian port i to destination mesoregion j; Si = maritime cost of transport to Brazilian port i; Pi = port costs at Brazilian port i; Ei = demurrage expenses at Brazilian port i; International Food and Agribusiness Management Review

412

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Ti = AFRMM tax at Brazilian port i; Fij = freight cost from Brazilian port i to destination mesoregion j; Dj = volume of imported fertilizer demanded by destination mesoregion j; Oi = Brazilian port j capacity to receive imported fertilizer.

The fertilizer mathematical model was solved using the Simplex method for linear programming problems and run with the help of the Solver Premium Platform (FrontLine Solvers version 2015, Frontline Systems Inc, Incline Village, NV, USA).

5. New investments in port infrastructure: possible scenarios A number of companies intend to invest in the Brazilian fertilizer sector as the sector’s outlook is very positive.

Two projects to increase domestic production should be in operation within the time-frame considered in this paper. One is a Petrobrás’ facility planned for Três Lagoas in the state of Mato Grosso do Sul. That facility will add approximately 1 million metric tons of urea to the market beginning in 2020. The other project is the expansion of KCl production undertaken by Vale do Doce in the state of Sergipe that will more than double its 2015 level of production by 2020 (ANDA, 2013).

Attention is being given to the improvement of Brazil’s road, railway, and port transportation infrastructure.

In June 2015, the Brazilian government announced the second phase of the Brazilian Logistic Investments Program 2015-2018, which will continue modernization of the country’s transport infrastructure. The Program relies primarily on private sector initiatives. Investments in Brazil’s port sector that directly impact imported fertilizer flows are expected to reach approximately US$ 15.9 billion (Brasil, 2015).

The Program proposes that four new port terminals dedicated to fertilizer handling be constructed: two at the Port of Santos, one at the Port of Santarém, and one at the Port of Itaqui (Brasil, 2015; Government of Maranhão, 2015). Program proposals also include construction of a new berth at the port of São Francisco do Sul that will increase the port’s current fertilizer import capacity (A Notícia, 2015). New investment is also planned for the Port of Paranaguá to expand the terminal at Antonina and for the Port of Rio Grande.

Moreover, the Paranaguá and Antonina Port Administration (APPA) has announced that it may dedicate an extra berth for fertilizer shipments to keep pace with internal demand growth (APPA, 2015; Global Fert, 2015; Zero Hora, 2015).

Demurrage expenses should drop when these port investments become operational. Since this potential impact is difficult to measure, the effects of similar measures adopted by APPA at Paranaguá in 2012 and 2013 have been adopted as a proxy. Using this proxy, a demurrage cost reduction of 30% is expected for all ports receiving investments except for Itaqui, where a 40% decrease was set due to current critical storage levels (Vargas, 2015). All other costs were maintained at 2014 prices.

All investments taken into consideration in this study’s Base scenario are summarized in Table 2 as are their impacts on import capacity or domestic fertilizer production.

In order to enrich the analysis, three alternative scenarios were constructed. Scenario 2 considers that no investments in the Northern-Arc ports will have resulted in operational facilities by 2025; consequently, additional fertilizer demand will be entirely supplied through southern ports. In this case, demurrage costs were maintained at 2014 levels. Scenario 3 does not restrict the capacities of Northern-Arc ports, simulating that maximum optimal volume can be imported through these terminals as all projected facilities are operational and have been upgraded. Finally, since new fertilizer terminals are already planned for the Northern-Arc ports of Santarém and Itaqui, Scenario 4 will focus the analysis on the other Northern-Arc ports: Vila do Conde, Miritituba, and Itacoatiara.

International Food and Agribusiness Management Review

413

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Table 2. Investments in the Brazilian fertilizer sector accounted for in the Base scenario: logistics and national production (2016-2025).

Port/facility

Investment

Paranaguá (PR)

Additional berth dedicated to fertilizers shipments and construction of a new terminal in Antonina.

Rio Grande Improvements in port operations and (RS) dredging as well as investments in fertilizer blending.

São Francisco Build up of a new berth with a do Sul (SC)1 total capacity of 3,000 Kt for all merchandises.

Santos (SP) Improvements in Tiplam Terminal and construction of Outeirinhos Terminal for fertilizer, which is included in the Plan of Logistic Investments (PLI) 20152018.

1 Itaqui (MA) Construction of a fertilizer terminal with total capacity of 4,300 Kt as stated in PLI 2015-2018.

Santarém Construction of a fertilizer terminal with 1 (PA) total capacity of 1,600 Kt as stated in PLI 2015-2018.

Três Lagoas Nitrogen Fertilizer Unit – UFN (MS) III to produce ammonia and urea.

Construction has already started but works are currently stalled.

Carnalita Expand current mining activities in (SE) Sergipe state.

1

Timing

Extra-capacity (thousand mt)

Source

2016-2020

2,950 (500 per year plus 450 in 2018)

2016-2020

1000 (200 per year)

APPA (2015); Global Fert (2015) Zero Hora (2015)

2021

342 (maintaining fertilizer share in total movement through the port) 1,845 (990 Tiplan, 855 Outeirinhos)

A Notícia (2015)

2018-2020

2,250 (750 per year)

2020-2022

1,440 (half in 2020 and half in 2022)

Government of Maranhão (2015) Brasil (2015)

2020

1000 (urea only)

ANDA (2013)

2020

~500 KCl resulting in a total production of 1000

ANDA (2013)

2017-2020

Brasil (2013); Personal communication (A. Leal)

Considering a usage rate of 90%.

6. Country-level results Increasing import capacity is of utmost importance to the country’s fertilizer sector. Present infrastructure was sufficient to meet fertilizer demand levels in 2015 and 2016 but will have reached capacity by 2017 if nothing is done. Figure 1 illustrates that the additional capacity of 10 million tons resulting from the planned investments addressed in the Base scenario should temporarily alleviate capacity pressure on existing terminals. But even with these infrastructure improvements, the ports will have neared their capacity limit by the end of 2025 with a 96.8% utility, highlighting the importance of long term logistics planning for the fertilizer industry.

Beyond the development of additional fertilizer import capacity, both public and private sectors must carefully identify additional opportunities and accurately estimate their potential benefits so that future projects will provide the best results for farmers and fertilizer companies. As a comparison of the average logistic costs per ton of imported fertilizer resulting from each scenario shows (Figure 2), allocating resources to the best port investment alternatives should lead to cost savings that could reach US$ 7.60/mt in 2025.

In a commodity market, lower logistic cost is expected to enhance competitiveness. Study results indicate that there are opportunities to lower these costs through the expansion of both southern and northern Brazilian International Food and Agribusiness Management Review

414

Da Costa Simões et al.

Volume 21, Issue 3, 2018 Imports

40.0

Port capacity

Current (2014) capacity 34.3

Million metric tons

35.0 30.0 25.0

24.4

20.0 15.0 10.0

22.9

22.5

23.7

24.9

2014

2015

2016

2017

26.0

26.9

26.0

27.8

29.4

30.6

31.9

33.2

2018

2019

2020

2021

2022

2023

2024

2025

5.0 0.0

Figure 1. Port capacity vs demand for fertilizer imports in Brazil, MMT (2014-2025).

US$ per metric ton

Scenario 1 108 106 104 102 100 98 96 94 92 90

Scenario 2

Scenario 3

Scenario 4

Investments in Northern ports start to become operational 2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025

Figure 2. Average logistic costs (US$/mt) from 2014 through 2025 for each scenario. Scenario 1: takes into account all the announced investments in infrastructure. Scenario 2: does not consider new investments in Northern-Arc ports. Scenario 3: considers maximum optimal volume that can be imported through all NortherArc ports. Scenario 4: considers maximum optimal volume that can be imported through the terminals of Vila do Conde, Miritituba and Itacoatiara. ports (Table 3); however, due to the location of latent and potential fertilizer demand, Brazil’s Northern-Arc ports are the focus of investors’ attention. The output from the Base scenario corroborates this optimism: from 2014 to 2025, while national demand for imported fertilizer is projected to increase at an annual rate of 3.5%, annual arrivals at Northern-Arc ports is projected to grow by 10.1% (Table 3). Moreover, all the additional capacity resulting from planned investments in the Northern-Arc ports of Itaqui and Santarém is expected to be fully utilized the moment it becomes available. In line with these results, Northern-Arc ports would account for 17.1% of total fertilizer imports by 2025, against their 8.7% share in 2014.

It is important to highlight that, despite the noted infrastructure investments directed toward Brazil’s Northern-Arc ports, Scenario 3 results indicate that their potential is far from being exhausted. These ports could handle more than 33% of the total amount of fertilizer forecast to be imported into the country in 2025 if expansion was unconstrained, but current and planned investments in these terminals would need to double to reach this level (Figure 3).

International Food and Agribusiness Management Review

415

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Table 3. Base scenario: fertilizer imports and share by group of ports (2014-2025).

Year

2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 CAGR4

Volume of fertilizer (× 1000 mt)

Share (%)

Southern ports1

Northeastern ports2

Northern-Arc ports3

Southern ports1

Northeastern ports2

Northern-Arc ports3

19,345 19,013 20,076 21,263 21,659 21,725 20,126 21,987 22,764 23,288 24,542 25,846 2.7%

1,450 1,442 1,568 1,568 1,568 1,568 791 794 860 1,568 1,568 1,568 0.7%

1,974 1,974 1,974 1,974 2,724 3,474 4,944 4,944 5,664 5,664 5,664 5,664 10.1%

85 85 85 86 83 81 78 79 78 76 77 78

6 6 7 6 6 6 3 3 3 5 5 5

9 9 8 8 10 13 19 18 19 19 18 17

1

Santos, Paranaguá, Vitória, Rio Grande and São Francisco do Sul. Recife, Maceió and Aratu.

3 Itaqui, Vila do Conde, Itacoatiara, Santarém and Miritituba.

4 CAGR = Compound annual growth rate.

2

Scenario 1 – Estimated volume 35%

31%

29%

37%

45%

Million metric tons

35%

Scenario 3 – Potential volume

2.0

2.0

2.0

2.0

2.7

3.5

2014

2015

2016

2017

2018

2019

Estimated/potential (%)

64%

58%

62%

58%

54%

51%

4.9

4.9

5.7

5.7

5.7

5.7

2020

2021

2022

2023

2024

2025

Figure 3. Fertilizer imports through Northern-Arc ports: estimated vs potential volume, MMT (2014-2025).

Scenario 1: takes into account all the announced investments in infrastructure. Scenario 3: considers maximum optimal volume that can be imported through all Norther-Arc ports.

7. Regional fertilizer flows and port’s area of influence The regional approach to fertilizer output from port to consumer offers a deeper understanding of how each of the 46 fertilizer demanding mesoregions is currently supplied and of how the new investments listed in Table 2 will change distribution flows. Table 4 presents a summary of the actual 2014 fertilizer flows between ports and the regions designed for this study and those projected for 2025 in the Base scenario. Relative increases in share registered in 2025 are highlighted in green, while relative decreases are shown in red.

The expected investments to expand import capacity are needed to meet additional, future fertilizer demand by those mesoregions already under the influence of a specific port. With the exception of the port in Vitoria, Table 4 shows that existing flows will be intensified rather than redirected.

International Food and Agribusiness Management Review

416

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Table 4. Fertilizer flows from ports to demanding regions, base scenario (×1000 mt).

Region of Year destination1

Ports

RS/SC

4.063 (100%) 4.494 (100%) 557 (17%) 1.126 (30%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 4.620 5.620

2014 2025

PR

2014 2025

SP

2014 2025

MG

2014 2025

ES/RJ

2014 2025

GO/DF

2014 2025

MS

2014 2025

MT

2014 2025

BA

2014 2025

MAPITO

2014 2025

Northeast

2014 2025

North

2014 2025

Total 1

2014 2025

Rio Grande

Total Paranaguá, SF Sul and Imbituba 0 (0%) 0 (0%) 2.732 (83%) 2.689 (70%) 1.528 (61%) 1.734 (61%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 814 (60%) 0 (0%) 1.201 (100%) 2.508 (100%) 4.276 (100%) 6.710 (80%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 12 (2%) 214 (17%) 10.563 13.855

Santos

Vitória

0 (0%) 0 (0%) 0 (0%) 0 (0%) 960 (39%) 1.095 (39%) 1.766 (74%) 2.159 (75%) 0 (0%) 0 (0%) 399 (29%) 2.027 (100%) 0 (0%) 0 (0%) 0 (0%) 101 (1%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 3.125 5.382

0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 636 (26%) 713 (25%) 334 (100%) 365 (100%) 156 (11%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 1.126 1.078

Maceió, Recife and Aratu 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 1.052 (100%) 1.094 (81%) 0 (0%) 0 (0%) 398 (95%) 474 (90%) 0 (0%) 0 (0%) 1.450 1.568

Itaqui 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 15 (0%) 455 (5%) 0 (0%) 256 (19%) 1.410 (100%) 2.706 (100%) 22 (5%) 52 (10%) 160 (30%) 387 (31%) 1.607 3.857

Vl Conde, Itaco. and Santarém 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 1.149 (14%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 367 (68%) 658 (52%) 367 1.807

4.063 4.494 3.289 3.815 2.488 2.830 2.402 2.871 334 365 1.369 2.027 1.201 2.508 4.291 8.415 1.052 1.350 1.410 2.706 420 527 539 1.259 22.858 33.167

RS = Rio Grande do Sul; SC = Santa Catarina; PR = Paraná; SP = São Paulo; MG = Minas Gerais; ES = Espírito Santo; RJ = Rio de Janeiro; GO = Goiás; DF = Distrito Federal; MS = Mato Grosso do Sul; MT = Mato Grosso; BA = Bahia; MAPITO is an acronym that stands for a geographical region in Brazil encompassing the states of Maranhão (MA), Piauí (PI) and Tocantins (TO).

International Food and Agribusiness Management Review

417

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Figure 4 illustrates the actual and future areas of influence of each group port or group of ports considered in this study. The Figure identifies the general areas in Brazil the ports supplied in 2014 and those that are projected to supply in 2025. A comparison of the two maps indicates that the announced port investments will modify logistics in central and west-central Brazil the most, specifically in northern areas of the states of Goiás and Mato Grosso. The west-central state of Rondônia will also be affected, as Northern-Arc ports will be unable to continue to supply fertilizer to this market in the future. Consequently, by 2025, the port of Paranaguá’s area of influence will expand northwest to meet fertilizer demand from the state of Rondônia.

An opportunity will also arise to increase flows from the port of Rio Grande, in Brazil’s far south, to western Paraná in 2025.

A direct result of the abovementioned rearrangement of fertilizer distribution routes is that total logistics costs shift. Results from the Base scenario, shown in Table 4, indicate that seven of the study’s 12 regions, which account for more than 80% of Brazil’s total fertilizer imports, would benefit from the new infrastructure.

Cost reductions would vary from US$ 0.57 to 4.94 per metric ton, with shipments to the Mato Grosso (MT) and MAPITO4 regions seeing the largest reductions. On the other hand, four regions’ markets would bear an increase in their average fertilizer supply expenditures since future constraints will force them to change their current port of origin to a more expensive alternative. For instance, by 2025, the state of Goiás will no longer be able to receive fertilizer originated at the port of São Francisco do Sul but will completely rely on the port of Santos, a much more expensive alternative.

Under the Base scenario, average imported fertilizer logistics costs for the country as a whole are estimated to increase by 1.5% between 2014 and 2025 caused by increased demand. Nearly 40% of that additional demand is concentrated in the state of Mato Grosso, the state with the highest logistic costs in 2014. Results for that state would be worse if no investments in the Northern-Arc ports were made (Table 5; Scenario 2).

Under Scenario 2, all regional logistics costs in 2025 would be higher that those estimated in the Base scenario except in the Northeast, with the Brazilian average price for imported fertilizer increasing 6.3% over the 2014 average, reaching US$ 106.03/mt.

Scenario 3 also brings important outputs (Table 5) that show that more extensive investment in Northern-Arc ports would lead to a reduction in average Brazilian imported fertilizer logistic costs. In this scenario, the 4

MAPITO is an acronym that stands for a geographical region in Brazil encompassing the states of Maranhão (MA), Piauí (PI) and Tocantins (TO).

Vila do Conde

Vila do Conde Itaqui Itacoatiara

Itaqui Itacoatiara

Santarém Recife

Miritituba

Santarém Recife

Miritituba

Maceió

Maceió Aratu

Aratu Legend: Northern ports Itaqui Paranaguá & SF do Sul Santos Rio Grande Vitória Recife, Aratu & Maceió

Vitória

Vitória Santos Paranaguá & SF do Sul

Santos Paranaguá & SF do Sul Rio Grande

Rio Grande

Figure 4. Port area of influence in 2014 and 2025, base scenario. Map elaborated with Philcarto (version 5.75, Philcarto, France (http://philcarto.free.fr)).

International Food and Agribusiness Management Review

418

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Table 5. Average logistic costs to supply each region, 2014 and 2025, US$/mt.

Region1 BA GO/DF MAPITO MG MS MT PR RS/SC SP ES/RJ Northeast North Brazil

Base scenario

Scenario 2

Scenario 3

2014 (a)

2025 (b)

∆ (b-a)

% (b/a)

2025 (c)

% (c/b)

2025 (d)

% (d/b)

103.38 108.69 97.71 98.61 98.47 131.40 90.78 79.18 91.90 89.84 70.88 118.28 99.75

104.51 113.02 93.77 97.56 96.54 126.47 89.08 78.60 89.93 89.89 71.79 131.52 101.30

1.13 4.34 -3.94 -1.04 -1.93 -4.94 -1.70 -0.57 -1.97 0.05 0.91 13.25 1.55

1.1 4.0 -4.0 -1.1 -2.0 -3.8 -1.9 -0.7 -2.1 0.1 1.3 11.2 1.5

111.07 114.53 113.51 98.92 97.26 131.43 90.55 79.20 91.91 89.89 71.14 132.78 106.03

6.3 1.3 21.1 1.4 0.8 3.9 1.6 0.8 2.2 0.0 -0.9 1.0 4.7

103.40 105.58 93.77 96.82 101.58 117.57 88.23 78.60 88.76 89.89 71.79 114.60 98.47

-1.1 -6.6 0.0 -0.8 5.2 -7.0 -1.0 0.0 -1.3 0.0 0.0 -12.9 -2.8

1 BA = Bahia; GO = Goiás; DF = Distrito Federal; MAPITO is an acronym that stands for a geographical region in Brazil encompassing

the states of Maranhão (MA), Piauí (PI) and Tocantins (TO); MG = Minas Gerais; MS = Mato Grosso do Sul; MT = Mato Grosso; PR = Paraná; RS = Rio Grande do Sul; SC = Santa Catarina; SP = São Paulo; ES = Espírito Santo; RJ = Rio de Janeiro.

Northern-Arc ports of Vila do Conde, Santarém, Itacoatiara, Miritituba and Itaqui face no limits on expansion to supply fertilizer markets. Most other opportunities for increased logistics savings are likely to continue to be in northern Brazil and in the more centrally located states of Mato Grosso and Goiás. Fertilizer logistics costs in Mato Grosso and Goiás would be even lower if transport bottlenecks at the port of Paranaguá, the states’ major source of imported fertilizer, were removed.

8. Results for the state of Mato Grosso Regarding the state of Mato Grosso, that has a very large territorial size, central location and complex logistics, it was divided into four separate mesoregions: North, South, East, and West. The port of Paranaguá is currently the only port supplying the state’s fertilizer demand, as shown in Figure 5. In the future, planned projects will create new flows that will economically supply the state’s North mesoregion from the Northern-Arc port of Santarém, and East mesoregion from the ports of Itaqui and Santos. Considering the constraints on Northern-Arc port capacity, one of the main results of this study is the estimate that the port of Paranaguá will continue to supply 70.7% of the North mesoregion’s fertilizer demand, 34.8% of the East mesoregion’s demand, and 79.7% of the state’s total demand in 2025. That port will remain the supplier of 100% of the South and West mesoregions imported fertilizer.

The evolution of logistics costs to import fertilizer into Mato Grosso, shown below in Table 6, is a reflection of the flow changes noted in the previous paragraph. While savings opportunities will occur countrywide, Mato Grosso’s North region (MT3) will receive the greatest cost benefits under the Base scenario, with a potential supply cost reduction of 5.9%, or US$ 8.05 per metric ton in 2025 from the 2014 level. Results from this scenario show that farmers and blenders in MT3-North will save up to US$ 164.4 million between 2015 and 2025 if projects are concluded within the expected timelines, which is equivalent to the 2014 cost of 500 thousand metric tons of product. One should remark that, for the entire state of Mato Grosso, savings under the Base scenario would reach US$ 212.4 million between 2014 and 2025, or the 2014 cost of 640.4 thousand metric tons of fertilizer. On the other hand, if there were no investments in Northern-Arc ports and import supply continued to rely solely on the port of Paranaguá (Scenario 2; Table 6) and logistic costs would remain stable.

International Food and Agribusiness Management Review

419

Da Costa Simões et al.

Volume 21, Issue 3, 2018 Santarém (1.2 million t; 13.7%) Vila do Conde

Vila do Conde Itaqui (0.5 million t; 5.4%)

Itaqui Itacoatiara

Itacoatiara

Santarém Miritituba

Porto Velho

Porto Velho

Miritituba

Legend: Ports River terminals

Santos (0.1 million t; 1.2%) Paranaguá (6.7 million t; 79.7%)

Paranaguá (4.3 million t; 100%)

Current flows New flows Fertilizer demand

Figure 5. Fertilizer flows to supply Mato Grosso’s mesoregions, 2014 and 2025. Map elaborated with Philcarto.

Table 6. Average logistic costs to supply Mato Grosso regions (MT), 2014 and 2025, US$/mt.

Base scenario MT1 – east MT2 – south MT3 – north MT4 – west MT average

Scenario 2

Scenario 3

2014 (a)

2025 (b)

∆ (b-a)

% (b/a)

2025 (c)

% (c/b)

2025 (d)

% (d/b)

127.61 120.08 135.93 135.83 131.40

124.68 117.98 127.88 133.73 126.47

-2.93 -2.10 -8.05 -2.10 -4.94

-2.3 -1.7 -5.9 -1.5 -3.8

127.60 120.08 135.93 135.83 131.43

2.3 1.8 6.3 1.6 3.9

122.79 117.98 113.51 123.78 117.57

-1.5 0.0 -11.2 -7.4 -7.0

9. Concluding remarks In general, logistic investments are of utmost importance if future Brazilian fertilizer demand is to be met.

This study demonstrated that there are very clear rationales that support an increase in fertilizer flows through ports located in Brazil’s Northern-Arc, if the capacity to increase flows existed. Planned investments to develop Northern-Arc port infrastructure to expand capacity are needed to meet growing Brazilian fertilizer demand and reduce logistic bottlenecks affecting southern Brazilian ports. The investments should also lead to more competitive imported fertilizer prices to farmers and blenders. Even with the anticipated investments, southern ports will still account for the majority of Brazil’s future supply of imported fertilizer in 2025; that is, while southern ports’ absolute trade will continue to increase, the trade will be more evenly distributed across all ports.

The outcomes from the alternative scenarios presented in this study show that there is a benefit to be gained from additional investments in Northern-Arc ports, such as the ports of Vila do Conde, Santarém, Itacoatiara and Miritituba. The planning needed to make future investments must be on the sector’s current agenda as projections show that with current and planned projects in full operation port terminals will be running at full capacity by 2025.

It is important to emphasize that although the model specifications used in this paper help to show the very clear attractiveness and competitive advantages of some ports, such as Santarém, Itacoatiara and Miritituba, it does not show the actual flows that could be observed from such ports. First, the linear programming International Food and Agribusiness Management Review

420

Da Costa Simões et al.

Volume 21, Issue 3, 2018

technique solves the mathematical model considering that all the agents are working together in a stable, well-coordinated market with perfect information, which is unlikely. Despite trading companies exerting some form of oligopoly power, they do not know in advance which port a competitor may use. The port with the cheapest available product at one time may not remain the optimum choice a very short time later.

Secondly, the model was not designed to adjust for the possibility that trading companies may choose to load or off-load product at a sub-optimal port, which often occurs to avoid congestion and delay when using the otherwise optimal choice. Northern-Arc ports are frequently the sub-optimal option of choice.

It is recommended when traders select the optimal fertilizer supply port, they include proxies for capacity values in their sensitivity analysis, which would normally be carried out using linear programming software.

As Brazilian markets expand, the type of modeling applied in this study can be used to help convince potential investors – public and private – that the application of resources and innovative talent to address Brazilian logistical issues would be to their and society’s benefit.

Acknowledgement The authors would like to acknowledge the support from the following entities when developing this article: Agroconsult, ESALQ-LOG, PECEGE and the ADM Institute for the Prevention of Post-Harvest Loss.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2017.0037.

Table S1. The 46 mesoregions based on area under cultivation statistics provided by the IBGE database (2015) for 2014 and Agroconsult.

Figure S1. The 46 mesoregions and port terminals.

Figure S2. The 12 regions in Tables 4 and 5.

References A Notícia. 2015. Infraestrutura: Novo berço no Porto de São Francisco do Sul e BR-280 entram no pacote de concessões do governo federal. Joinville, June 9, 2015. Available at: http://tinyurl.com/yazdxrul.

Brasil. 2013. 1ª Consulta Pública dos Arrendamentos Portuários: Santos e Pará. Available at: http://tinyurl. com/yb223afc.

Brasil. 2015. Logistic investments program 2015-2018: ports. Available at: http://tinyurl.com/ycataemm.

Brazilian Institute of Geography and Statistics (IBGE). 2015. Produção Agrícola Municipal. Available at: http://tinyurl.com/yco4pbdu.

Bureau of Foreign Trade (SECEX). 2015. The system of analysis of foreign trade information – Aliceweb.

Available at: http://tinyurl.com/ycudcuca.

Caixeta-Filho, J.V. 2010. Desafios para a Otimização da Logística Agroindustrial Brasileira. In: 48° Congresso da Sociedade Brasileira de Economia, Administração e Sociologia Rural (SOBER), Campo Grande, Tecnologias, Desenvolvimento e Integração Social. Available at: http://tinyurl.com/y7hoenr7.

Carvalho, L. 2009. Estudo de Localização de Fábricas Misturadoras de Adubo na região Centro-oeste Brasileira Utilizando um Modelo de Programação Linear. Dissertação (Mestrado – Programa de PósGraduação em Engenharia de Produção / Processos e Gestão de Operações) – Escola de Engenharia de São Carlos da Universidade de São Paulo, São Carlos, Brazil.

Frias, L.F., I. Farias and P. Wanke. 2013. Planejamento de redes logísticas: um estudo de caso na indústria petroquímica brasileira. Revista de Administração Mackenzie 14(4): 222-250.

Global Fert. 2015. Entrevista: Diretor da APPA destaca retração da importação de fertilizantes em 2015.

Campinas, September. 30, 2015. Available at: http://tinyurl.com/y7d8h7jg.

International Food and Agribusiness Management Review

421

Da Costa Simões et al.

Volume 21, Issue 3, 2018

Government of Maranhão. 2015. Investimentos anunciados devem elevar em 30% o volume de cargas do Porto do Itaqui. Available at: http://tinyurl.com/ybz5nmtb.

Hamad, R., and N. Gualda. 2014. Modelagem de redes logísticas com demandas sazonais: influência do custo de estoque e do crédito de ICMS. Journal of Transport Literature 8(2): 295-324.

Hitchcock, F.L. 1941. The distribution of product from several source to numerous localities. Journal of Mathematics and Physics 20: 224-230.

ICIS. 2015. The market weekly world. Weekly freight rate indicators. All 2014 reports. Database available at https://www.icis.com.

International Fertilizer Association (IFA). 2015. Statistics database: IFADATA. Available from: http://ifadata. fertilizer.org/ucSearch.aspx.

Jensen, P. 1999. Network flow programming methods. In: Operations Research Methods. University of Texas. Chapter 5. Available at: http://tinyurl.com/ya52flbn.

Koopman, T.C. 1949. Optimum utilization of transportation system. Econometrica 17: 136-146.

Milanez, A., D.Nyko, J. Garcia and C.O. Xavier. 2010. Logística para o etanol: situação atual e desafios futuros, BNDES Setorial n. 31, p. 49-98, Banco Nacional de Desenvolvimento Econômico e Social, Rio de Janeiro, Brazil.

Ministry of Agriculture and Livestock (MAPA) and Ministry of Transportation; Special Ports Secretariat (MT). 2013. Análise das Rotas Alternativas para Escoamento da Safra Agrícola. Porto de Santos: ações prioritárias. Relatório Final. Brazilian Federal Government, Brasília, DF, Brazil.

National Agency of Land Transportation (ANTT). 2014. Estudo de Demanda. Ferrovia Norte-Sul: Trecho Porto Nacional/TO – Estrela d’Oeste/SP. Relatório 1. Available at: http://tinyurl.com/ybo3pvfa.

National Association of Fertilizer Diffusors (ANDA). 2013. Indústria Nacional de Matérias-Primas para Fertilizantes: Investimentos 2013-2018. 3° Congresso Brasileiro de Fertilizantes. São Paulo, August 26, 2015. Available at: http://tinyurl.com/yc3oq3jw.

National Association of Fertilizer Diffusors (ANDA). 2014. Anuário Estatístico do Setor de Fertilizantes 2014. ANDA, São Paulo, Brazil.

National Association of Fertilizer Diffusors (ANDA). 2015. Statistics spreadsheets (1998-2015). Available at: http://tinyurl.com/yac3f8u9.

National Water Transport Agency (ANTAQ). 2014. Hidrovias Brasileiras. Indicadores de Transporte de Cargas: Tonelada Útil Transportada (t) e Tonelada Quilômetro Útil (TKU) 2013. Available at: http:// tinyurl.com/y89y7d4d.

Pereira, A.A., M.A. Oliveira, I.C. Leal Júnior. 2016. Custo de transporte e alocação da demanda: análise da rede logística de uma produtora brasileira de fertilizantes nitrogenados. Journal of Transport Literature 10(4): 5-9.

Port Administration of Paranaguá and Antonina (APPA). 2015. Notícias. Empresa russa de fertilizante poderá ampliar investimentos em Antonina. Available at: http://tinyurl.com/ya72gte9.

Sharma, G., S.H. Abbas and V.K. Gupta. 2012. Solving transportation problem with the various method of linear programming problem. Asian Journal of Current Engeneering and Maths 1(3): 81-83.

Vargas, C. 2015. Mercado de Fertilizante and Logística no Brasil. Fórum ABAG Estadão 2015. São Paulo, SP. Available at: http://tinyurl.com/y9kxuams.

Vessel Distance. 2015. Port distance calculation: routing. Available at: https://www.vesseltracker.com/en/ Routing.html.

World Bank. 2016. Logistics performance index – LPI. Available at: http://lpi.worldbank.org/international/ global.

Zero Hora. 2015. Campo Aberto com Gisele Loeblein: Como é o projeto de expansão da Yara no Rio Grande do Sul. Porto Alegre. Available at: http://tinyurl.com/yc9wseow.

International Food and Agribusiness Management Review

422

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2017.0015 Received: 8 February 2017 / Accepted: 13 December 2017

Views on sustainability and the willingness to adopt an environmental management system in the Mexican vegetable sector REVIEW ARTICLE Luz E. Padilla-Bernal a, Alfredo Lara-Herrerab, Alberto Vélez Rodríguezc, and María L. Loureirod aProfessor-Researcher,

Academic Unit of Accounting and Administration, Autonomous University of Zacatecas, Comercio y Administracion s/n. Col. Progreso, Zacatecas, Zac., Mexico bProfessor-Researcher,

Academic Unit of Agronomy, Autonomous University of Zacatecas, Carr. Zacatecas-Guadalajara Km 15.5 Cieneguitas, Zac., Mexico

cProfessor-Researcher,

Academic Unit of Electrical Engineering, Autonomous University of Zacatecas, Av. Ramón López Velarde 801, Zacatecas, Zac., Mexico

dProfessor-Researcher,

Department of Economic Theory, University of Santiago de Compostela, Avda. das Ciencias s/n, 15782 Santiago de Compostela, Spain

Abstract In Mexico, agriculture’s main environmental problems are related to water resources, deforestation and soil erosion. These problems are more acute in arid or semiarid regions such as in the state of Zacatecas.

Environmental management systems (EMS) can be an alternate means for improving environmental conditions. In this study, factors that determine willingness to adopt an EMS in the vegetable production units of the state of Zacatecas were identified. We also analysed views on sustainability and production practices oriented toward environmental management and care of natural resources, as well as drivers and barriers to EMS adoption. Factors determining EMS adoption were level of education, awareness of the importance of caring for and protecting natural resources, application of agricultural practices oriented toward protecting the environment, and ignorance of environmental problems.

Keywords: environmental management systems, natural resource management, vegetable sector JEL code: Q15, Q56 Corresponding author: luze@uaz.edu.mx

© 2018 Padilla-Bernal et al.

423

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

1. Introduction The agricultural sector greatly depends on natural resources for production (Aigner et al., 2003; FAO, 2012a) and can both harm and benefit the environment. Today’s agricultural practices contribute around a third of greenhouse effect gases, but the use of good practices can neutralize carbon as well as create environmental services and generate renewable energy while, at the same time, contribute to food security (FAO, 2012a).

In this context, actions that help to guarantee the sustainability of the agricultural sector are of growing importance. The agricultural production units (PUs) that form part of the food production supply chain, in terms of sustainability, face great challenges and opportunities. This involves maintaining a supply of safe, healthy products, confronting environmental impacts generated by agriculture and putting into practice standards of fair labour.

In the case of Mexico, the main agro-environmental problems are associated with water resources and deforestation. Moreover, the growing importance of soil pollution through misuse of chemical products, emission of greenhouse effect gases and soil erosion are also highlighted (OCDE, 2008; SEMARNAT, 2006). According to data from INEGI (2009), the technology used in most of the agricultural area of Mexico is based on the use of chemical fertilizers, herbicides and insecticides, while organic fertilizers are used to on a much smaller scale (García-Gutiérrez and Rodríguez-Meza, 2012). However, these problems can be reduced by implementing an integral program of care and protection of the environment that involves adopting improved agricultural practices, conservation measures and management of environmental impacts.

In arid or semiarid regions, as in the state of Zacatecas, these problems are much more acute. In 2014 in Zacatecas, 87.7% of the agricultural area was rainfed and 12.3% was irrigated (SIAP-SAGARPA, 2015).

Irrigation water is extracted from 34 aquifers, of which 14 (41%) is overexploited, according to CNA (2015).

Ground water is used to irrigate nearly 150,000 h, 12% of the cultivated area of the state (SIAP-SGARPA, 2015). On 32.8% of this area, seven vegetables are cultivated (chilis, onions, tomatillos, potatoes, garlic, tomatoes and lettuce), generating 62.3% of the value of the production on this area.

Sustainability in this context has become a huge challenge for PUs of the agricultural sector. One step toward handling this problem is transparency in terms of the impact of the production processes of organizations on their ecological and social environment. Consumers, civil society and other agents involved in the supply chain increasingly require that enterprises consider the social and environmental consequences of their activities (Hartmann, 2011:299-304). As an alternative means for improving environmental conditions, independent of normative aspects (Darnall and Sides, 2008:95-97), a series of mechanisms of voluntary adoption have emerged to contribute to the process (Grolleau et al., 2007; Khanna, 2001; Segerson, 2013). Environmental management systems (EMS) and several production protocols that can be used for certification and/or labelling are found among these mechanisms (FAO 2012b, 2014). Although these mechanisms are voluntary, their application with verifiable criteria and standards can be a prior condition for entering some markets.

Although EMSs are widely applied in different industrial sectors, they are less used in the agricultural sector, especially in developing countries (Raymond, 2012). A growing number of publications have studied, theoretically and empirically, what determines adoption of voluntary measures of environmental management in non-agricultural sectors (e.g. Arora and Carson, 1995; Blackman, 2008; Del Brío and Junquera, 2003; Merli et al., 2016; Rezessy and Bertoldi, 2011; Videras and Alberini, 2000). Application of EMSs mainly in the industrial sector has made agricultural producers perceive them as not applicable for them, too complex, difficult and costly (Carruthers, 2005). One of the main barriers to adopting an EMS is the scepticism about the results in terms of improving the environment and the benefits of applying one to their PUs. According to Segerson (2013) there are serious doubts about the effectiveness and efficiency of these mechanisms for achieving the proposed goals. The lack of information on applying EMSs in agriculture generates uncertainty regarding the benefits and costs of their implementation (Carruthers, 2005; Carruthers and Vanclay, 2012; Williams, 2009).

International Food and Agribusiness Management Review

424

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

According to SEMARNAT (2006), most of the agricultural PUs in Mexico report a lack of information on natural resources management and on care and protection of the environment. Likewise, few studies report growers’ views of the impact agriculture has on the environment, the importance they give to the sustainability of their PUs, the environmental performance of their PUs or their views on the drivers and barriers to adopting an EMS in this sector. The aim of this study was to help fill the information gap existing in the situation of adoption of voluntary measures of environmental management in the PUs of the agricultural sector. This would allow the different agents involved in the sector’s development to have information to aid in planning their activities in the short and medium term and contribute to the economic, social and ecological sustainability of this productive sector.

The objective of this study is to identify the factors that determine the probability that a vegetable PU in the state of Zacatecas adopt an EMS. Also, views on sustainability and production practices oriented toward environmental management and care of natural resources in the PUs were determined as were the drivers and barriers to EMS adoption. Elements are provided that contribute to development of an EMS for the agricultural sector in Mexico. The study adapted the ‘perceptions’ on sustainability proposed by Rankin et al. (2011), who designed a structure for establishing levels of sustainability in agribusinesses in terms of perceptions, actions and performance criteria. Likewise, we followed the proposal of Hauschildt and Schulze-Ehlers (2014), who connected sustainability views, drivers and barriers to concrete acquisition of practices for sustainability in the industry of food services. It should be mentioned that, although there currently exist diverse tools, measurements and standards that cover different dimensions of sustainability in the PUs and the supply chain (FAO, 2014; ISS, 2009), in our study only environmental sustainability and its management are addressed.

The research questions were the following. What are the views on sustainability of the growers or technicians of the vegetable PUs? What is the level of environmental performance of the vegetable PUs? What are the drivers and barriers to EMS adoption in the vegetable sector? What are the factors that determine the probability of the willingness to adopt an EMS in the vegetable PUs in the state of Zacatecas?

2. Factors that affect adoption of sustainable agricultural practices As we have indicated, there is little literature on application and adoption of an EMS in agriculture. There is, however, a large body of studies related to the adoption of innovations and practices designed to reduce the environmental impact of agriculture (e.g. Ajayi, 2007; Blesh and Barrett, 2006; Kontogeorgos et al., 2015; Welsh and Rivers, 2011). These papers include debates over how views of the world and sustainability, or attitudinal variables, as well as the characteristics of the grower, influence decision-making and the environmental performance of the production unit (Welsh and Rivers, 2011).

To determine which variables correlate with adoption or acquisition of more awareness in the application of sustainable agricultural practices in a statistically significant way, based on previous theorization and statistical tests, researchers have regularly applied logistic regression models or probit (Feder et al., 1985; Knowler and Bradshaw, 2007). For example, Arellanes and Lee (2003) analysed the determiners of adopting minimum tillage in Honduras. Likewise, Ma et al. (2009), to explore factors that underlie the attitude of producers toward agricultural production and their environmental awareness, applied a model of logistic regression. The results revealed that age, education and agricultural economic efficiency were the three factors that affect the attitude of local producers toward agricultural production. Moreover, Kontogeorgos et al. (2015), using logistic regression, determined that farming experience, consumption of their own produce and recycling agrochemical containers defined the level of environmental conscientiousness of young producers in Central Macedonia, Greece.

Studies that seek the reasons for adopting or not adopting sustainable agricultural practices show that there are numerous independent variables that regularly explain adoption. However, as Knowler and Bradshaw (2007) state, there is not enough empirical evidence to conclude that there are universally significant variables that International Food and Agribusiness Management Review

425

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

explain adoption of sustainable agricultural practices. To promote adoption of practices aimed to reduce the environmental impact of agriculture, focalized effort must be made in accord with the particular conditions of the growers and the region.

3. Materials and methods To determine the probability that a PU adopt an integral program of care and protection of the environment, denominated EMS, a logistic regression model was used. The views of sustainability of the surveyed personnel and the drivers and barriers to adopting an EMS were evaluated with a principal components analysis (PCA). The production practices of the PUs oriented toward environmental management and care of natural resources were determined by calculating the environmental performance index (EPI), following the proposal of Carruthers (2005). The statistical analysis was done with SPSS v23 software.

3.1 Sources of information and determination of the sample Using interviews with four technicians from vegetable PUs to identify their production and environmental protection practices, as well as a review of literature on sustainability and environmental management, a questionnaire was designed. The questionnaire was divided into five sections (Supplementary Table S1). The first poses questions relative to their views on sustainability. The second section enquires into the orientation of the PU toward environmental management and care of natural resources. The third section requests information on their strategies aimed at adoption of an EMS (barriers and drivers). In the first three sections, the responses are measured with a five-point Likert scale. Section four is called certification and adoption of an EMS. In this section there is one dichotomous question about the existence of certifications related to quality, food safety and environmental management and one open question about the type of certifications obtained by the PU. Also in this section, a dichotomous question is asked about the willingness of the PU to adopt an integral program of care and protection of the environment. Finally, section five contains questions about the PU and socioeconomic information of the interviewee. The questionnaire was given to technicians or owners of the PUs. The questionnaire was piloted in February 2015.

The number of producers or technicians to be interviewed was determined by obtaining a representative sample of the PUs that cultivate vegetables. The universe from which the sample was selected was the census of vegetable growers (chili, tomatoes and garlic) available in SAGARPA, Zacatecas Delegation, and associations of growers such as the Cluster de Agricultura Protegida, A.C. and the Sistema Producto Tomate.

The sampling method was simple random (Mendenhal and Reinmuth, 1981). The sample was determined with a confidence level of 95%, a level of variability of 0.25 (considering a criterion of maximum variance and a level of precision of 7%). Considering a universe of 2,204 growers registered in the censuses (chili=1,749, tomato=278 and garlic=177) and adjusting with a finite population correction factor, the minimum sample was 180 questionnaires. During the period of March through July 2015, 207 questionnaires were applied, and of these, 202 were completely answered and useful for the study. The criteria for selection of the PUs to be surveyed were the following: (1) an area of â&#x2030;Ľ10 hectares of vegetable crops cultivated in the field or one hectare under protected agriculture; (2) agricultural activity reported in 2014; and (3) willingness of the owner or technician to respond to questions. The characteristics of the PUs surveyed are presented in Table 1.

3.2 Information processing To determine the probability of adopting an EMS, a logistic regression model was used, applying the technique of maximum likelihood. The dependent variable, WEMS, is a binary variable equal to 1 if the PU is willing to adopt an EMS and 0 otherwise.

The independent, or explanatory, variables were classified in the following way: characteristics of the PU and interviewee, views on sustainability, environmental performance of the PU, and drivers and barriers to adopting an EMS. Some of the variables are observed and others were constructed through the following International Food and Agribusiness Management Review

426

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

Table 1. Characteristics of the production units (PUs) surveyed.

Characteristic

Variable

Quantity

Area cultivated in all of the PUs and method Field (ha) 8,568.8 of cultivation Protected agriculture (ha) 415.5 Total (ha) 8,984.2 Method of cultivation in PU Only field cultivation 123 Only protected agriculture 36 Both field and protected agriculture 43 Type of market where they sell their Local 76 produce National 64 Local, national 54 Local, national and international 8 Certifications of the PU Some type of certification 37 No certification 165 Interviewee level of education Elementary 38 Secondary 44 High school 31 Bachelor´s degree 76 Graduate degree 13 Gender of the interviewees Female 17 Male 185

Percentage 95.4 4.6 100.0 60.9 17.8 21.3 37.6 31.7 26.7 3.5 18.3 81.7 18.8 21.8 15.3 37.6 6.4 8.4 91.6

methodology. The characteristics of the PUs and the interviewees considered were market and level of education. These were processed as dichotomous variables (Table 2); the variable market took the value of 1 if the PU commercialized its produce on national and/or international markets and 0 if sold only in the local market. In the case of the variable education, 1 was given if the interviewee had a high school or higher (bachelorâ&#x20AC;&#x2122;s or graduate) education and 0 otherwise.

Table 2. Definition of variables and sample statistics.

Factors

Variable

Definition

Mean

Standard deviation

Willingness to adopt an Environmental management systems (EMS) Characteristics of PU and interviewees

WEMS

0.866

0.341

0.624

0.486

0.594

0.492

Views on sustainability

F1_RIO

1 if the production unit (PU) is willing to adopt a program of care and protection of the environment 1 if PU sells its produce in domestic or international markets 1 if interviewee has high school or higher education Driven by profitability; innovative and organizational Social Normative Environmental performance index Internal motivators Barriers availability of resources and information Barriers availability of time

3.619

0.788

3.917 4.010 6.528 4.066 3.883

0.791 0.803 1.691 0.787 0.812

3.252

1.067

Market Education

F2_societal F3_compliant PU environmental performance EPI Motivators and barriers to EMS M1_internal adoption B1_informa B2_time

International Food and Agribusiness Management Review

427

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

Views on sustainability were determined by the responses to sixteen questions, adapting the proposals of Rankin et al. (2011) and Hauschildt and Schulze-Ehlers (2014). Interviewees were asked to respond to statements on a scale of 1 to 5: 1=don’t agree to 5=totally agree. For greater clarity and to reduce the variety of views on sustainability, the responses were subjected to a PCA with Varimax rotation. The PCA revealed three statistically significant factors. The criterion for preserving an indicator in a factor was factorial loads equal to or greater than 0.5, and for a factor, a Cronbach’s alpha coefficient greater than or equal to 0.70.

The Cronbach’s alpha coefficient is a widely used measure of reliability. According to Curkovic (2003), this coefficient should have a value of 0.70 or higher for more precise constructs and of 0.55 to 0.69 for moderately broad constructs. Following the five levels of sustainability proposed by Rankin et al. (2011), the factors were denominated ‘driven by profitability, innovative and organizational’, ‘Social’ and ‘Normative’.

These factors were converted into the variables F1_RIO, F2_societal and F3_compliant (Table 2).

In the same way, to determine the perception of drivers and barriers to EMS adoption, interviewees were asked to respond to seven questions in each case on a scale of 1 to 5, in which 1=not important to 5=highly important. To reduce the variety of drivers and barriers, the responses were also subjected to a PCA with Varimax rotation. The analysed indicators were obtained based on the proposals of Carruthers (2005); Carruthers and Vanclay (2012); Corbett et al. (2003); Grolleau et al. (2007); Massoud et al. (2010) and Merli et al. (2016). The PCA of the motivators resulted in two factors: ‘internal drivers’ and ‘competitiveness and compliance’. Because of Cronbach’s alpha coefficient obtained in the latter factor, it was excluded from the logistic regression model and only the variable M1_internal was retained (Table 2).

The PCA of the barriers to EMS adoption revealed two significant factors: ‘availability of resources and information’ and ‘availability of time’. The two factors were converted into the variables B1_informa and B2_ time (Table 2).

The level of environmental performance of the PU was obtained by calculating the variable EPI, based on Carruthers (2005). To this end, the interviewees self-evaluated by answering groups of questions (indicators) referring to the following variables: water, soil, biodiversity, agrochemicals, pollution, waste management, environmental management of the business. The response scale given was 1 to 5, where 1=does not apply or not done and 5=always applies or always done. The EPI by variable represents the relationship between the score of the level studied relative to the highest possible score. To make all of the scores comparable, the highest score considered was ten. The EPI was obtained as averages of the considered variables.

The linear model proposed for determining the factors that influence willingness to adopt an EMS is the following: Y* = α + Σ4j=1βjXji + μi, i=1, 2, 3, …, N

(1)

where X1i represents a vector of variables with characteristics of the PU and interviewees (dichotomous variables Market, Schooling), X2i presents the views on sustainability (F1_RIO, F2_societal, F3_compliant), X3i environmental performance of the PU (EPI) and X4i drivers and barriers to EMS adoption (M1_interno, B1_informa, B2_time). β1-β4 are the coefficients (vectors or scalers) to be estimated, α and μ are the intercept and error term, respectively. The interpretation of the latent variable in models of this type are the potential advantages of adopting an EMS, that is, the perceived difference between adopting and not adopting an EMS. The model is of discrete election with a dichotomous variable indicating the willingness to accept EMS adoption, as the dependent variable Yi: Y=1

yes

Y* > 0,

Y=0

otherwise

International Food and Agribusiness Management Review

428

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

4. Results and discussion 4.1 Views on sustainability in the production units Table 3 presents the views on sustainability considered in the study, mean, standard deviation and factorial loads of the PCA. The factors that were statistically significant with eigenvalues above one explain 59.55% of the total variance. Indicators with factorial loads below 0.5 were not included in the factors. Fourteen of the indicators have factorial loads above 0.5. The indicators ‘strategy for reducing costs’ and ‘strategy to improve the UP’s position in the market’ have factorial loads below 0.5, and therefore were not included with the factors. Likewise, the indicator ‘belief that leads to caring for and protecting the environment’ was excluded from the analysis because it had factorial loads above 0.5 in more than one factor.

Cronbach’s alpha coefficients for evaluating internal consistency and reliability were high in the three factors, with values of 0.890, 0.760 and 0.725, respectively. According to Tavakol and Denick (2011), the alpha coefficient can take on values from 0 to 1, considering a value above 0.70 to be acceptable, suggesting that the indicators with high factorial loads are highly correlated within the factors.

Table 3. Mean, standard deviation and factorial loads of the principle component analysis of views on sustainability.1 Views on sustainability

Mean2 Standard Factor 1 driven Factor 2 deviation by profitability; social innovative and organizational

An integral part of the production unit (PU) core A set of values that guide the work of the PU A way to strengthen the PU’s image A way to improve the work environment A way to reduce risks in the PU A value integrated into the PU A strategy for improving the PU’s position in the long run An opportunity to improve income A way to reduce impacts on the environment to preserve it for the future A way to reduce hunger and increase social wellbeing A way to express solidarity with other growers A belief that leads to caring for and protecting the environment Compliance with laws and standards of environmental protection Production of safe products for consumers A strategy for reducing costs A strategy for improving the PU’s position on the market Eigenvalue Cronbach’s Alpha

3.30 3.38 3.76 3.53 3.64 3.43 4.04

1.080 1.092 1.048 1.116 1.018 1.011 0.951

0.794 0.776 0.709 0.706 0.659 0.599 0.587

0.124 0.137 0.255 0.389 0.308 0.099 0.496

0.125 0.230 0.188 0.053 0.070 0.432 -0.023

3.84 4.32

1.045 0.798

0.560 0.221

0.430 0.807

0.216 0.146

3.72

1.015

0.258

0.687

0.132

3.71 4.18

1.068 0.891

0.266 0.058

0.579 0.533

0.315 0.531

3.95

0.926

0.168

0.110

0.848

4.07 3.58 3.85

0.886 1.100 1.066

0.097 0.471 0.471

0.294 0.025 0.422

0.714 0.491 0.361

6.978 0.890

1.455 0.760

1.094 0.725

1 2

Total observations = 202; Kaiser-Mayer-Olkin = 0.898; values ≥0.5 (bold) are considered in the factor.

Scale: 1 = not applicable or don’t agree to 5 = in total agreement.

International Food and Agribusiness Management Review

429

Factor 3 normative

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

The PCA found similarities among the levels of organization sustainability proposed by Rankin et al.

(2011). A factor was identified in which indicators are found that show that the PUs only perform actions in favour of the environment or society to comply with established laws and industrial standards (Level 1).

Another factor includes indicators that comprise the aspects relative to what the authors define as ‘Driven by profitability; innovative and organizational’ (Levels 2, 3, and 4) and a third factor that considers indicators whose social aspects are of prime importance. It should be pointed out when asked the question about the importance of sustainability in the PU, 64.8% of the interviewees answered that it was important or very important. However, 12.4% said it was not important or had little importance.

4.2 Production practices oriented toward environmental management and care of natural resources Regarding practices aimed toward production and care of the environment and carried out in the PUs studied, it was found that the activities that are most neglected are those related to the business’ environmental management (5.54), which include aspects such as designing formal environmental contingency plans, program for reducing solid and green waste, training program in environmental aspects for workers, use of technology that save water and energy, that is, actions that would lead to environmental management. Water is another of the variables that had one of the lowest values (5.99). To evaluate this variable, the following aspects were analysed: care to avoid polluting the aquifers, analysis of irrigation water, irrigation application efficiency and rainwater harvesting.

The variables biodiversity and agrochemicals, which represent conservation of the region’s flora and fauna, and use and handling of agrochemicals applied to the crops have higher average indexes (7.33 and 7.18) (Table 4). In the case of agrochemicals, the following was included: suitable storage of agrochemicals, efficient application of agrochemicals to crops, integrated pest management and use of only those agrochemicals authorized by regulating organisms. The high value obtained in the variable biodiversity is attributed to some growers’ awareness that prevents them from harming animals of the region and of the region’s scarce vegetation. The value obtained in the case of agrochemicals is mostly attributed to government divulgation programs that promote good use and handling of agrochemicals (BUMA) (SENASICA, 2014) and reducing pollution during their application – Sistema de Reducción de Riesgos de Contaminación de Producción Primaria de Vegetales (SRRC) (SENASICA, 2010).

By groups of PUs, classified by market type, the lowest index was obtained by the PUs that sell their produce only on the local market (5.78), while the highest (6.98) was for the group that sell their produce mostly on the national and/or international markets. The t test for equality of means showed that there are significant differences in the seven averages of the evaluated variables of environmental performance (water, soil, biodiversity, agrochemicals, pollution, waste management and business environmental management), as Table 4. Environmental performance in the vegetable production units by market type.

Variable

Local

Water Soil Biodiversity Agrochemicals Pollution Waste management Business environmental management Environmental performance index

Local, domestic and/ or international

Index

Mean

Standard deviation

Mean

Standard deviation

Mean

Standard deviation

5.44 5.92 6.91 6.06 6.04 5.13 4.89 5.78

1.72 1.58 1.68 2.19 2.00 2.41 2.30 1.61

6.32 7.15 7.58 7.85 7.30 6.74 5.93 6.98

1.64 1.88 1.69 1.91 1.92 2.53 2.09 1.58

5.99 6.69 7.33 7.18 6.82 6.13 5.54 6.53

1.72 1.87 1.71 2.19 2.04 2.60 2.22 1.69

International Food and Agribusiness Management Review

430

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

well as in the EPI, between the PUs that are mostly oriented toward local markets and those that are oriented toward national and international markets (P<0.05), suggesting that the market in which the PUs sell their produce marks differences in their production practices and care of the environment.

It should be highlighted that 37 (18.3%) of the PUs studied have some type of certification related to safety or quality of their produce, environmental protection and/or well-being of their workers, such as PrimusLabs, Buenas Prácticas Agrícolas and BUMA. The other 165 (81.7%) have never been certified. In the group of certified PU there are eight that ship their produce to international markets.

4.3 Drivers and barriers to adopting an environmental management system According to Carruthers and Vanclay (2012), EMS adoption is multi-dimensional, encompassing different factors, such as pressure from consumers and other actors of the value chain and the desire to improve performance and implement an integrated management system (Grolleau et al., 2007; Carruthers and Vanclay, 2012), among other aspects. Although it should be pointed out that factors change over time can vary in accord with individual circumstances and context.

Table 5 presents the factorial loads of the PCA, the mean and standard deviation of the seven drivers of EMS adoption analysed. Two factors resulted in eigenvalues above one: ‘Internal drivers’ and ‘competitiveness and compliance’, explaining 64.6% of the total variance. The seven indicators showed factorial loads above 0.5.

The factor denominated ‘internal drivers’ had a Cronhach’s alpha coefficient above 0.7, while in the factor ‘competitiveness and compliance’, this coefficient was 0.629, and so, was excluded from later analyses.

The drivers the interviewees considered most important for adopting an EMS were that it facilitates access to national and international markets (4.51) and reduces production costs (4.38), followed by PU sustainability (4.21). These responses are congruent with the findings of Carruthers (2005); Carruthers and Vanclay (2012) and SAI Platform (2015) in that the market is a determining factor for application of good agricultural practices.

In terms of barriers to EMS adoption (Carruthers, 2005; Carruthers and Vanclay, 2012; Grolleau et al., 2007; Massoud et al., 2010; Merli et al., 2016; Williams et al., 2000), Table 6 presents the mean, standard deviation and two factors resulting from the PCA. These were denominated ‘availability of resources and information’ and ‘availability of time’ and had eigenvalues above one and Cronbach’s alpha coefficients above 0.7. They explained 62.2% of the variance. The factorial loads of all the indicators surpassed 0.5.

Table 5. Drivers to adopting an environmental management system.1 Drivers

Mean2

Standard deviation

Factor 1 internal

Factor 2 competitiveness and compliance

Consistency with personal principles Prevent harm to workers Improve the production unit’s image Improve production unit sustainability Reduce production costs Facilitate access to national and international markets Compliance with norms for environmental protection Eigenvalue Cronbach’s alpha

3.92 4.01 4.11 4.21 4.38 4.51 3.76

0.989 0.990 0.978 0.886 0.902 0.806 0.963

0.896 0.863 0.656 0.646 0.041 0.251 0.406 3.515 0.835

0.116 0.117 0.404 0.436 0.807 0.746 0.560 1.011 0.629

1 2

Total observations = 202; Kaiser-Mayer-Olkin = 0.839; values ≥0.5 (bold) are considered in the factor.

Scale: 1 = not important to 5 = highly important.

International Food and Agribusiness Management Review

431

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

Table 6. Barriers to adopting an environmental management system.1 Barriers

Mean2 Standard deviation

Factor 1 Availability of information and resources

Factor 2 Availability of time

Ignorance of environmental aspects and their management Not knowing how to quantify environmental damage Additional expenses of monitoring, training, infrastructure and equipment No personnel in the production unit trained to carry out actions that contribute to protecting the environment No government support Too much paperwork and documented requirements Lack of time for planning and implementing environmental actions Eigenvalue Cronbach’s alpha

3.78

1.122

0.820

0.161

3.84 3.91

1.086 1.168

0.775 0.735

0.018 -0.076

3.71

1.054

0.657

0.202

4.17 3.26 3.24

1.090 1.187 1.199

0.649 0.105 0.104

0.209 0.885 0.877

2.922 0.787

1.435 0.749

1 2

Total observations = 202; Kaiser-Mayer-Olkin = 0.723; values ≥0.5 (bold) are considered in the factor.

Scale: 1 = not important to 5 = highly important.

Lack of government support (4.17) and additional expenses for monitoring, training, infrastructure and equipment (3.91) were the two indicators that, on average, the interviewees declared to be the most important barriers to EMS adoption. Another two barriers that had values above 3.70 were not knowing how to quantify environmental damage (3.84) and ignorance of environmental aspects and their management (3.78), similar to results found by Massoud et al. (2012). Regarding the indicators relative to information on environmental management, Carruthers and Vanclay (2012) point out that EMS users know more about their enterprise’s impacts and look for solutions more actively than non-users. The factor denominated ‘availability of time’ comprises the indicators ‘too much paperwork and documented requirements (3.26) and ‘lack of time for planning and implementing environmental actions’ (3.24).

The results obtained show that lack of financial resources and ignorance of environmental problems are the most important barriers to EMS adoption in the vegetable PUs of Zacatecas. Interviewees assigned more importance to these aspects than to time and work required by the application of activities related to environmental management. These results agree with Carruthers and Vanclay (2012), who consider that the costs, concerns for necessary skills, time and availability of resources are common barriers to EMS adoption and other agricultural practices.

4.4 Factors determining the probability of willingness of adopting an environmental management system Table 7 presents the estimation results of the logistic regression model. To interpret the sensitivity of the willingness to accept EMS adoption relative to the explanatory variables, marginal effects are also reported.

For the continuous independent variables, the marginal effects measure change in the estimated probability after an increase of one unit in the explanatory variable. In the case of discrete variables, the marginal effect is calculated as the difference between estimated probabilities and the sample means when the dummy variables take the value 1 and 0, respectively. The percentage of correct predictions was 95.0%. However, the percentage of correct predictions obtained with a model with only one intercept is 86.6%, which means that what is gained in the model conditioned by the explanatory variables is limited. The sensitivity, the proportion of observations correctly predicted as 1, and the specificity, the proportion of observations correctly

International Food and Agribusiness Management Review

432

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

Table 7. Estimation of the logistic regression model.

Parameter

Estimator

Wald statistic Marginal effects

Intercept -15.0392 11.89 Market 0.4978 0.39 Education 1.7782 4.30 F1_RIO 0.0996 0.01 F2_societal 2.2432 7.84 F3_compliant 1.1864 3.16 EPI 1.2974 5.81 M1_internal 1.0110 1.84 B1_informa -1.8341 4.78 B2_time 0.1973 0.20 Nagelkerke R2 Cox and Snell R2 Log likelihood Proof of likelihood χ2 (9) Correct predictions (percentage) Correct predictions with intercept Sensitivity (proportion of observations correctly predicted as 1) Specificity (proportion of observations correctly predicted as 0) Number of observations Number of observations that accept adoption of an integral program of care and protection of the environment

0.0195 0.0696** 0.0039 0.0877*** 0.0464 0.0507** 0.0395 -0.0717** 0.0077 0.754 0.411 -52.01 106.880 95.0 86.600 97.7 77.8 202.0 175

**, *** = level of significance of the parameter at 5 and 1%, respectively. Marginal effects are calculated based on the sample means.

predicted as 0, are acceptable. The Cox and Snell R2 indicates that 41.1% of the variance of the dependent variable (WEMS) is explained by the explanatory variables.

The results of the model estimation showed that the statistically significant variables in WEMS were interviewee level of education (education), the social factor of views on sustainability (F2_societal), the EPI, and the barriers factor lack of resources and information (B1_informa). Significance of the variable Education shows that there is 6.96% higher probability that a PU accepts adopting an EMS if the technician or owner (interviewee) have high school or higher (bachelor’s or master’s) education than in PUs whose owners have lower levels of schooling. In a PU where there are technicians or owners with an opinion more oriented toward social aspects of sustainability, there is an increase of 8.77% in the probability of adopting an EMS for each unit of change in this factor (F2_societal). The likelihood that a PU adopt an EMS increases with improvements in production practices and care and protection of the environment. This is supported by the result of the variable EPI coefficient, which is positive and significant. The variable B1_informa (Barrier lack of availability of resources and information) resulted negative and statistically significant, meaning that the less the lack of financial resources and the less the ignorance of the environmental problems, the greater the probability that the PU adopt an EMS.

5. Conclusions and recommendations Three factors summarize the growers’ views on sustainability: (1) profitability and innovative and organizational aspects of the PU; (2) social aspects; and (3) compliance with the norms relative to care and protection of the environment. Most of the growers believe that sustainability is important for their PU. However, when they self-evaluated the application of production practices oriented to care of natural resources and environmental management, the EPI showed that there is still much to improve in production and commercialization practices to achieve a more sustainable sector. This is more visible in the activities relative to care of water International Food and Agribusiness Management Review

433

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

and establishment of formal strategies for environmental protection. However, it should be pointed out that when the PUs are more oriented toward national and international markets, there is a marked difference in EPI values, showing that the market is a determining factor in application of actions that lead to adoption of better agricultural practices and environmental performance of the PU. Growers exhibited lack of information regarding what caring for and protecting the environment involves and how an EMS would be applied to agriculture. They are uncertain in terms of what the benefits and costs of its implementation would contribute to their PU. Aspects were reaffirmed in determining factors that impact the probability of willingness of adopting an EMS in vegetable PUs. Those aspects were level of education of the technician or owner of the PU, clear conviction of the importance of caring for and protecting natural resources for future generations and of increasing social well-being, application of agricultural practices oriented toward protection and care of the environment, ignorance of environmental problems, as well as lack of sufficient financial resources to deal with the situation. For this reason, we recommend promoting information on agriculture’s impacts on the environment and the benefits of the PU’s adopting more environment-friendly production practices. This can take two directions. On the one hand, adopting sustainable production practices and managing environmental impact should be promoted by providing information on the benefits of natural resource sustainability, and on the other, by divulging the advantages for the market and for the PU of having ‘greener’ production and contributing to care and protection of the environment.

Overcoming the barriers to EMS adoption is an important step prior to deciding its implementation. Having clear knowledge of the advantages of improving natural resource management to care for and protect the environment could contribute to reducing producers’ perceived barriers.

For a mechanism of voluntary adoption to be effective as an EMS in the agricultural sector, suitable incentives for participation should be considered in the design and implementation. That is, at the government level, a strategy could be developed around three fundamental axes. First, the design of an integral program of care and protection of the environment should consider management of the agricultural PUs’ impacts, recognize the heterogeneity of the producers, and prescribe the manner in which the program will be monitored. Second, economic incentives should be provided to the PUs that are involved in the program and show effective results. Third, extension services should be offered to facilitate the producers’ adoption of the program.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2017.0015.

Table S1. Sections and questions included in the questionnaire.

References Aigner, D.J., J. Hopkins and R. Johansson. 2003. Beyond compliance: sustainable business practices and the bottom line. American Journal of Agricultural Economics 85(5): 1126-1139.

Ajayi, O.C. 2007. User acceptability of sustainable soil fertility technologies: lessons from farmers’ knowledge, attitude and practice in Southern Africa. Journal of Sustainable Agriculture 28(3): 121-143.

Arellanes, P. and D. Lee. 2003. The determinants of adoption of sustainable agriculture technologies: evidence from the hillsides of Honduras. Proceedings of the 25th International Conference of Agricultural Economists. Available at: http://purl.umn.edu/25826.

Arora, S. and T. Cason. 1995. An experiment in voluntary environmental regulation: participation in EPA’s 33/50 program. Journal of Environmental Economics and Management 28(3): 271-286.

Blackman, A. 2008. Can voluntary environmental regulation work in developing countries? Lessons from case studies. Policy Studies Journal 36(1): 119-141.

Blesh, J.M. and G.W. Barrett. 2006. Farmers attitudes regarding agro landscape ecology: a regional comparison.

Journal of Sustainable Agriculture 28(3): 121-143.

International Food and Agribusiness Management Review

434

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

Carruthers, G. 2005. Adoption of environmental management systems in agriculture. An analysis of 40 case studies. No. 05/032. RIRDC Australian Government, Canberra, Australia.

Carruthers, G. and F. Vanclay. 2012. The intrinsic features of environmental management systems that facilitate adoption and encourage innovation in primary industries. Journal of Environmental Management 110: 125-134.

Comisión Nacional del Agua (CNA). 2015. Disponibilidad de agua subterránea (DOF 20 de abril de 2015).

Available at: https://tinyurl.com/y7kht24o.

Corbett, C., A. Luca and J.N. Pan. 2003. Global perspectives on global standards: a 15-economy survey of ISO 9000 and ISO 14000. ISO Management Systems, pp. 31-40. Available at: https://tinyurl.com/ y7ze7jqc.

Curkovic, S. 2003. Environmentally responsible manufacturing: the development and validation of a measurement model. European Journal of Operational Research 146: 130-155.

Darnall, N. and S. Sides. 2008. Assessing the performance of voluntary environmental programs: does certification matter? The Policy Studies Journal 36(1): 95-117.

Del Brío, J.A. and B. Junquera. 2003. Influence of the perception of the external environmental pressures on obtaining the ISO 14001 standard in Spanish industrial companies. International Journal of Production Research 41(2): 337-348.

Feder, G., R. Just and D. Zilberman. 1985. Adoption of agricultural innovations in developing countries: a survey. Economic Development and Cultural Change 33(2): 255-298.

Food and Agriculture Organization of the United Nations (FAO). 2012a. Greening the economy with agriculture. In: Greening the economy with agriculture, edited by N. El-Hage Scialabba. FAO, Rome, Italy, pp. 2-6.

Food and Agriculture Organization of the United Nations (FAO). 2012b. Sustainability assessment of food and agriculture systems. Guidelines. (Test Version 1.0). Available at: https://tinyurl.com/y8szzrw4.

Food and Agriculture Organization of the United Nations (FAO). 2014. Sustainability assessment of food and agriculture systems. Guidelines. (Version 3). Available at: https://tinyurl.com/pl5jqqs.

García-Gutiérrez, C. and G.D. Rodríguez-Meza. 2012. Problemática y riesgo ambiental por el uso de plaguicidas en Sinaloa. Ra Ximhai 8(3): 1-10.

Grolleau, G., N. Mzoughi and A. Thomas. 2007. What drives agrifood firms to register for an Environmental Management System?. European Review of Agricultural Economics 34(2): 233-255.

Hartmann, M. 2011. Corporate social responsibility in the food sector. European Review of Agricultural Economics 38(3): 297-324.

Hauschildt, V. and B. Schulze-Ehlers. 2014. An empirical investigation into the adoption of green procurement practices in the German Food Service Industry. International Food and Agribusiness Management Review 17(3): 1-32.

Instituto Nacional de Estadística y Geografía (INEGI). 2009. VIII Censo agrícola, ganadero y forestal.

Estados Unidos Mexicanos. Censo Agropecuario 2007, INEGI, Aguascalientes, Mexico.

Intertek Sustainability Solutions (ISS). 2009. Agriculture standards benchmark study 2009. Sustainable Agriculture Initiative (SAI) Platform. Available at: https://tinyurl.com/ycktl588.

Khanna, M. 2001. Non-mandatory approaches to environmental protection. Journal of Economic Surveys 15: 291-324 Knowler, D. and B. Bradshaw. 2007. Farmers’ adoption of conservation agriculture: a review and synthesis of recent research. Food policy 32: 25-48.

Kontogeorgos, A., M. Tsampra and F. Chatzitheodoridis. 2015. agricultural policy and the environment protection through the eyes of new farmers: evidence from a country of southeast Europe. Procedia Economics and Finance 19: 296-303.

Ma, Y., L. Chen, X. Zhao, H. Zheng and Y. Lü. 2009. What motivates farmers to participate in sustainable agriculture? Evidence and policy implications. International Journal of Sustainable Development and World Ecology 16(6): 374-380.

Massoud, M.A., R. Fayad, M. EI-Fadel and R. Kamleh. 2010. Drivers, barriers and incentives to implementing environmental management systems in the food industry: a case of Lebanon. Journal of Cleaner Production 18(3): 200-209.

International Food and Agribusiness Management Review

435

Padilla-Bernal et al.

Volume 21, Issue 3, 2018

Mendenhall, W. and J. Reinmuth. 1981. Estadística para administración y economía. Wadsworth International/ Iberoamericana, San Francisco, CA, USA. Merli, R., M. Preziosi and C. Ippolito. 2016. Promoting sustainability through EMS application: a survey examining the critical factors about EMAS registration in Italian organizations. Sustainability 8(197): 1-14.

OCDE. 2008. Desempeño Ambiental la Agricultura desde 1990, OCDE, Paris, France.

Rankin, A., A.W. Gray, M.D. Boehlje and C. Alexander. 2011. Sustainability strategies in U.S. agribusiness: understanding key drivers, objectives, and actions. International Food and Agribusiness Management Review 14(4):1-20.

Raymond, R. 2012. Improving food systems for sustainable diets in a green economy, In: Greening the economy with agriculture, edited by N. El-Hage Scialabba. FAO, Rome, Italy, pp. 132-184.

Rezessy, S. and P. Bertoldi. 2011. Voluntary agreements in the field of energy efficiency and emission reduction: review and analysis of experiences in the European Union. Energy Policy 39(11): 7121-7129.

SAI Platform. 2015. Partnering with farmers towards sustainable agriculture: overcoming the hurdles and leveraging the drivers. Practitioners’ guide 2.0. Available at: https://tinyurl.com/ycycd7xq.

Secretaría de Medio Ambiente y Recursos Naturales (SEMARNAT). 2006. La gestión ambiental en México.

SEMARNAT, Ciudad de México. C.P, Mexico.

Segerson, K. 2013. When is reliance on voluntary approaches in agriculture likely to be effective? Applied Economic Perspectives and Policy 35(4): 565-592.

Servicio Nacional de Sanidad, Inocuidad y Calidad Alimentaria (SENASICA). 2014. Buen Uso y Manejo de Agroquímicos. Available at: https://tinyurl.com/ybruosfx.

Servicio Nacional de Sanidad, Inocuidad y Calidad Alimentaria (SENASICA). 2010. Lineamientos generales para la operación y certificación de sistemas de reducción de riesgos de contaminación en la producción primaria de alimentos de origen agrícola. DGIAAP-SAGARPA, Ciudad de México, Mexico.

Servicios de Información Agroalimentaria y Pesquera (SIAP-SAGARPA). 2015. Producción anual. Available at: https://tinyurl.com/y8yfpbvc.

Tavakol, M and R. Dennick. 2011. Making sense of Cronbach’s alpha. International Journal of Medical Education 2: 53-55.

Videras, J. and A. Alberini. 2000. The appeal of voluntary environmental programs: which firms participate and why? Contemporary Economic Policy 18(4): 449-461.

Welsh, R. and R.Y. Rivers. 2011. Environmental management strategies in agriculture. Agriculture and Human Values 28(3): 297-302.

Williams, T. 2009. Environmental management in agriculture and the rural industries: voluntary approaches to sustainability and globalization imperatives. RIRDC. No. 09/023. Australian Government. Union Offset Printing, Canberra, Australia.

Williams, H., A. van Hooydonk. P. Dingle and D. Annandale. 2000. Developing tailored environmental management systems for small businesses. Eco-management and Auditing 7(3): 106-113.

International Food and Agribusiness Management Review

436

OPEN ACCESS International Food and Agribusiness Management Review Volume 21 Issue 3, 2018; DOI: 10.22434/IFAMR2017.0041 Received: 3 May 2017 / Accepted: 1 December 2017

Florida’s Natural® and the supply of Florida oranges CASE STUDY Carlos Omar Trejo-Pech a, Thomas H. Spreenb, and Lisa A. Housec aAssistant

Professor of Agribusiness Finance, Department of Agricultural & Resource Economics, The University of Tennessee, Institute of Agriculture, 2621 Morgan Circle Drive, 308D Morgan Hall, Knoxville, TN 37996, USA; Researcher, Escuela de Ciencias Económicas y Empresariales, Universidad Panamericana at Guadalajara, Calz. Circ. Pte. 49, Zapopan, 45010 Jalisco, Mexico bProfessor

Emeritus, Food and Resource Economics, University of Florida, P.O. Box 110240, Gainesville, FL 32611, USA

cProfessor,

Food and Resource Economics, and Director of the Florida Agricultural Market Research Center, University of Florida, 1083 MCCB, P.O. Box 110240, Gainesville, FL 32611-0240, USA

Abstract This case study provides a thorough description of the U.S. orange juice industry, and focuses on Florida’s Natural, a cooperative of citrus growers and owner of the Florida’s Natural® brand. Florida’s Natural® competes mainly with Tropicana, owned by PepsiCo, and with Minute Maid and Simply Orange, brands of The Coca-Cola Company. The objective of the case is to evaluate the orange juice industry, assess the position of Florida’s Natural within the industry, and propose business actions for the cooperative. By the end of 2016, the orange juice industry was in the midst of a severe crisis, threatened by decreasing supply and changing consumption preferences. Total orange production in the State of Florida during the 2015-16 season was the smallest crop since the 1960s due mainly to a disease known as citrus greening. Marketers were also facing consumers’ concerns regarding high levels of calories and sugar in some juice categories.

Furthermore, on May 2016, the Food and Drug Administration mandated a change in the nutrition facts label on packaged food, becoming effective in the summer of 2018, which may impact marketers in the sector.

Keywords: agribusiness, orange juice industry, citrus greening, Florida’s Natural JEL code: Q13, M2, M3 Corresponding author: ctrejope@utk.edu

A teaching note has been prepared for this case study. Interested instructors at educational institutions may request the teaching note by contacting the author or IFAMA. © 2018 Trejo-Pech et al.

437

Trejo-Pech et al.

Volume 21, Issue 3, 2018

1. Introduction Late in 2015, Dr Robert Behr was appointed CEO of Florida’s Natural Growers (Florida’s Natural hereafter), the leading cooperative commercializing the Florida’s Natural® brand of citrus juices. With a Ph.D. in Food and Resource Economics from the University of Florida, Dr Behr had plenty of experience in the citrus industry after serving as director of economics and marketing at the Florida Department of Citrus and holding different management positions at Florida’s Natural. Yet, managing any citrus company posed a tremendous challenge. One issue was the scarce supply of oranges. Orange production in the State of Florida, the home of the cooperative, reached its lowest level in decades during the 2015-16 season due mainly to the disease Huanglongbing (also known as citrus greening – HLB). Demand for orange juice was also a concern. Customers’ tastes and preferences toward beverages were changing rapidly and juice companies were offering an array of innovative products. Furthermore, some consumers were moving away from juices due mainly to the perception that they contained high levels of sugar. By the end of 2016, Dr Behr and his team needed to revisit the situation of the cooperative and the industry in order to implement a sustainable business strategy. Questions under consideration included the following. What efforts should Florida’s Natural put forward to better position its Florida’s Natural® brand? What strategies should the firm pursue in order to assure adequate supply of high quality inputs?

2. Juice and juice drinks 2.1 Recent trends Juices are defined as packaged beverages containing 100% fruit or vegetable juice. Juice drinks are packaged beverages with less than 100% fruit juice or vegetable content.1 Off-trade (e.g. in stores) annual sales revenue of juices and juice drinks (J&JD) in the U.S. was estimated at about $17.3 billion with 2,261.5 million gallons sold in 2016 (Euromonitor, 2017).2 Table 1 shows that J&JD volumes decreased in recent years while prices increased. J&JD industry analyses by Euromonitor and others (Brown and Washton, 2013; Euromonitor, 2016, 2017) suggested the following. The drop in sales during recent years was caused, in part, by consumers’ concerns regarding high sugar content in some beverages. This concern affected juice drinks the most since, unlike juices, juice drinks contain added sugar and artificial ingredients. The prices increase is attributed to the emergence of a premium segment, which includes exotic juices and/or blends (e.g. coconut water, aloe vera juice, pomegranate juice, and prune juice), smoothies (e.g. Naked and Odwalla brands), and unpasteurized, cold-pressed juices. In addition, the continuing declining supply of oranges in the last decade caused by disease has contributed to the decline in sales volume and increase in prices observed in Table 1. Orange juice and orange drinks together had an estimate share of around 36% of the J&JD industry in 2016.

2.2 Attributes and prospects of juices and juice drinks By the end of 2016, market research specialists forecast volume of J&JD to continue to decline in 2017 and 2018 by 1 to 2%.3 However, the juice category (100% juice), which had positive growth in previous years, was forecast to continue growing into the future. In recent years, the highest growth in beverages has come 1 2 3

Euromonitor (2017) separates juice drinks (up to 24% juice), nectars (25 to 99% juice content), and coconut and other plant waters.

Unless otherwise stated, statistics in this study refer to off-trade figures.

Statistics obtained by authors from Passport (formerly Euromonitor International).

Table 1. Off-trade sales growth of juice and juice drinks in the United States (Euromonitor, 2017).

Year to year change of volume Year to year change of value

2013

2014

2015

2016

-1.7% -1.3%

-1.9% -1.7%

-1.9% 1.5%

-0.8% 1.4%

International Food and Agribusiness Management Review

438

Trejo-Pech et al.

Volume 21, Issue 3, 2018

from premium, high profit, low calorie, ‘better-for-you’ juices that attracted health conscious millennials and Boomers (Brown and Washton, 2013). A study by Tetra Pak in 2016 (Crawford, 2016) found potential for growth of juices in particular areas: (1) 67% consumers rate ‘all natural’ as the most interesting juice attribute; (2) vegetable blends is a growing category, attracting customers probably because they have lower natural sugar and offer different nutrients; and (3) 60% of consumers look for juices with proven medical benefits (i.e. functional products).

Despite its high sugar content, fruit beverages are an important source of vitamins and minerals and represent a cost-effective way to meet consumers’ daily fruit intake recommendations (Leschewski et al., 2016). This has allowed manufacturers to successfully launch fortified juices with vitamins and minerals, and campaigns that emphasize naturally occurring nutrients in juices. Table 2 provides nutrients in selected fruit juices.

The same year, 2016, researchers from Michigan State University and the Economic Research Service of the USDA (Leschewski et al., 2016) published an article addressing whether specific nutrients in J&JD garner price premiums. Categories other than nutrients, in their hedonic price model, include product categories (e.g. flavor, brand name, private label, etc.), packaging, and attributes of the acquisition (such as store type, region, etc.). The study provides results across the two categories, juices and juice drinks. Overall, the study finds that: (1) all nutrients garner premium prices in juices while only selected nutrients garner premium prices in juice drinks; and (2) sugar garners premium prices in juice drinks only. The model estimates, for instance, that adding one additional mg of antioxidants in juices leads to a 5% increase in the price per ounce. As expected, sugar is not associated with price premiums in juices. Price premiums, according to the researchers, may reflect both manufacturers’ costs and consumers’ willingness-to-pay.4

4

Interestingly, in the juice drinks category (and more specifically, for the non-diet subcategory), one additional gram of sugar is related to a 1% price premium. The authors argue that consumers in this category prefer the taste of sweeter drinks, are willing to pay for additional sugar, and manufacturers price the cost accordingly.

Table 2. Percent of the daily value of nutrients in eight ounces of assorted fruit juices. Figures based on a 2,000 calorie diet (Leschewski et al., 2016).

Nutrient

Apple

Cranberry Grape cocktail (purple)

Grapefruit Orange (white)

Pineapple

Prune

Energy, kcal Protein, g Total sugars, g Dietary fiber, g Total fat, g Vitamin A, RAE Vitamin E, mg Vitamin C, mg Calcium, mg Phosphorous, mg Magnesium, mg Iron, mg Sodium, mg Potassium, mg

6% 0% 76% 1% 0% 0% 0% 4% 2% 2% 2% 5% 0% 8%

7% 0% 94% 0% 0% 0% 3% 100% 1% 0% 1% 1% 0% 1%

5% 2% 49% 1% 0% 4% 1% 156% 2% 4% 8% 3% 0% 11%

7% 3% 109% 2% 0% 0% 0% 42% 3% 2% 8% 4% 0% 9%

9% 3% 109% 10% 0% 0% 2% 18% 3% 6% 9% 17% 0% 20%

8% 3% 119% 0% 0% 0% 0% 0% 0% 3% 6% 3% 0% 10%

5% 3% 63% 3% 1% 1% 2% 143% 2% 4% 7% 6% 0% 12%

International Food and Agribusiness Management Review

439

Trejo-Pech et al.

Volume 21, Issue 3, 2018

3. Orange juice and orange drinks 3.1 Flavors shares Orange has the highest share among available flavors in the J&JD industry. In 2016, sales volume of orange juice (100% juice content) in the U.S. represented around one half of total juices (Euromonitor, 2017; Statistica, 2017). Orange drinks also had the highest-in-category share with about 20% volume of total drinks (Euromonitor, 2017). Volume of orange juice and orange drinks together represented 36.4% of J&JD consumed in the U.S. in 2016. Figure 1 provides sales of juice and juice drinks by flavors over time.

100

80

2 3.3 4 9.4

11.8

11.7

11.2

11.2

9.3

9.4

12.5

13.8

59.8

59.4

57.6

56.3

2013

2014

2015

2016

%

60

2.4 3.1 4.2 9

3.5 3 4.3 8.7

3.5 3 4.3 8.3

40

20

0

Orange

Mixed fruits

Apple

Other flavous

Tomato

Cranberry

Grape 100

0.7 3.7 6

80

%

60

4.2

0.7 6

4.2

0.73.5

3.7

5.9

4.1

0.7 3.4 5.8

8.9

9.5

9.7

9.8

10.5

10.4

10.2

10

11.7

11.3

11

10.7

15.2

14.6

16

17

17

17.6

18

18.4

22.1

22

20.9

20.1

2013

2014

2015

2016

4.1

40

20

0

Orange

Mixed fruits

Other flavors

Lemon

Berry

Fruit punch

Cranberry

Grape

Strawberry

Grapefruit

Figure 1. Flavors of juices (above) and juice drinks (below) in the U.S. by shares (Euromonitor, 2017).

International Food and Agribusiness Management Review

440

Trejo-Pech et al.

Volume 21, Issue 3, 2018

3.2 Orange juice consumption According to estimates by the Florida Department of Citrus (FDOC), presumed consumption of orange juice in the U.S. decreased from 1,084 million single strength equivalent (SSE) gallons in the 2012-13 crop season to 883 million SSE gallons in 2015-16 (Figure 2). This is equivalent to a decline of per capita consumption from 3.4 to 2.7 gallons during that period. Presumed consumption is an approximate, disappearance measure estimated by the FDOC as season-beginning inventory plus Florida and other states production plus U.S. imports minus U.S. exports minus season-ending inventory (FDOC, 2017). Presumed consumption in the U.S. was forecast by the FDOC to continue decreasing to 779 million SSE million gallons in the coming 2016-17 season.

The A.C. Nielsen Report provides more precise measures of orange consumption but for a reduced market coverage (e.g. selected grocery stores). Figure 3 shows historical orange juice sales volume (million SSE gallons) and prices (USD/gallon) disaggregated by two type of juices: refrigerated not-from-concentrate and refrigerated reconstituted.5 Figure 3 also gives projections by the FDOC for the 2016-17 season.6 Notfrom-concentrate orange juice, the premium sub-category in 100% juices, gained share at the expense of reconstituted orange juice, moving from 57% in 2012-13 to 67% in 2016-17. Price increases were also slightly higher for not-from-concentrate.

3.3 A new policy with potential to affect orange juice consumption The decline of orange juice consumption in the U.S. (Figure 2 and 3) was due to the combination of scarce orange juice supply and changing consumersâ&#x20AC;&#x2122; preferences regarding juices and juice drinks in general, as discussed. The concern regarding the connection between consumption of fruit beverages and increased risk for health problems due to high levels of calories and sugar, has recently prompted policy changes in the United States. In 2015, the USDA revised the Dietary Guidelines for Americans recommending abstaining 5

Frozen juice and shelf stable juice are other categories in the A.C. Nielsen Report. Together they represent less than 5% of total volume. We do not present those statistics in this document, but are readily available in FDOC (2017).

6 The FDOC provides projections for three price scenarios (high, mid, and low). Projections shown in this document are those for mid-level prices.

1,084 966

Volume (million SSE gallon)

1000 800

912

821

883 779

623

600

547 447

400

421

418

30

55

458

387

392

409

51

39

200 0

2012-2013

2013-2014

52 2014-2015

2015-2016

2016-2017 F

Crop season Presumed consumption

Florida production

Other U.S. Production

U.S. imports

Figure 2. U.S. orange supply and presumed consumption (million single strength equivalent gallons).

International Food and Agribusiness Management Review

441

Trejo-Pech et al.

Volume 21, Issue 3, 2018

Refrigerated RECON 7.93

600

560.76 524.84 490.03 464.23

7.62 7.59 Volume (million SSE gallon)

Total orange juice

7.25 7.27

430.2 7 6.8 6.55 6.54

400 310.32 320.03

291.59 281.16 269.4

200

6.2

5

20 17 12 -1 3 20 13 -1 4 20 14 -1 5 20 15 -1 6 20 16 -1 20 7 12 -1 3 20 13 -1 4 20 14 -1 5 20 15 -1 6 20 16 -1 7

6

143.6 5.14 4.96 4.93

20

-1

6

16

5

15

4

-1

20

14

3

-1

20

13

-1

20

12

6.3

190.49 176.64 163.52 212.49

4.79 4.87

0

20

8

Price (USD/gallon)

Refrigerated NFC

Crop season Volume

Price

Figure 3. Volume (million single strength equivalent gallons) and prices (USD per gallon) of 100% orange juice at the retail level (Nielsen Topline Sales Annual Reports, generated by the Florida Department of Citrus, available at: http://tinyurl.com/y9kdapqg). Data up to the third week of October in the corresponding season. Original source takes data for U.S. grocery stores doing $2 million and greater annual sales, drug stores doing $1 million and greater annual sales, mass merchandisers (like Target), Walmart, club (Sam’s and BJ’s), dollar stores (Dollar General, Family Dollar and Fred’s), and military/DECA. Forecast for the 2016-17 season according to FDOC (2017). NFC = Not-from-concentrate; RECON = reconstituted. from fruit drink consumption and limiting fruit juice consumption (Leschewski et al., 2016). Furthermore, on May 2016 the Food and Drug Administration released a new design for the nutrition facts label on packaged foods, which would become effective in the summer of 2018. Figure 4 provides the new label design of a hypothetical product highlighting the changes.

Figure 4. New label for packaged foods (FDA website).

International Food and Agribusiness Management Review

442

Trejo-Pech et al.

Volume 21, Issue 3, 2018

Changes in the new label include: (1) contents of vitamin D and potassium will be declared in the label, replacing vitamin A and C, which may continue to be labeled but will not be mandatory; (2) a new line, indicating the amount of ‘added sugar’ will be required in new labeling; and (3) total calories, in larger font, will be added. Orange juice analysts believe that the new label policy might positively impact orange juice consumption (Euromonitor, 2016; Trejo-Pech et al., 2017) if marketers appropriately communicate to consumers the high level of potassium (which is more likely lacking in the U.S. population compared to other nutrients) inherent in oranges. Orange juice has the second highest level of this nutrient across commercially available juices (Table 2). The fact that 100% orange juice has no added sugar is another element juice companies may emphasize.

4. The Florida orange juice industry The Florida citrus industry traces its origin to Spanish settlers who first brought citrus fruit to the New World in the 16th century. With its subtropical climate and sandy soils, which are highly suitable for citrus cultivation, the industry flourished in northeast Florida along the St. John’s River before being pushed farther south as freezes periodically visited north and central Florida.

An important innovation in the industry was the development of frozen concentrated orange juice (FCOJ) soon after the end of World War II. FCOJ, developed by Florida Department of Citrus scientists (Pearcy and Goldberg, 2000), offered the means to allow long-term storage of orange juice, thereby extending the marketing season year-round. Concentrate orange juice was stable when stored in frozen form and when reconstituted (recon, as known in the industry) with water, it had a good taste. The next important development was aseptic storage, which allowed long-term storage of single strength orange juice. Not-from-concentrate (NFC) orange, a non-concentrated juice that is pasteurized before packing and chilling, was introduced by Tropicana in the 1950s (Pearcy and Goldberg, 2000), becoming the fastest growing orange juice product in the following decades. The better taste of NFC orange juice over FCOJ positioned this product in the premium category.

By the 1970s, Florida-based companies such as Tropicana and Minute Maid (owned by Coca-Cola) dominated the orange juice market in the United States and Canada. Orange juice had become a part of the American breakfast; per capita consumption had reached six gallons of single strength equivalent annually.

In the 1980s, a series of freezes destroyed a large portion of the productive citrus acreage in Florida. Florida had also suffered a freeze in 1962 that had prompted one large citrus company to establish a processed orange industry in Saõ Paulo state in southeast Brazil. The Florida freezes of the 1980s prompted a major expansion of the citrus industry in Brazil. In the 1985-86 season, the United States imported more than one-half of its orange juice supply from Brazil. The Florida industry recovered, however, and by the 1997-98 season produced over 240 million 90-pound boxes of oranges, a record crop. By this time, NFC orange juice had claimed about 30% of the U.S. market. Tropicana, with its Tropicana Pure Premium brand, held the largest share of the NFC market. Another company, Florida’s Natural, claimed the number two position in the U.S. NFC market.

Historically, Florida has been an iconic place for orange production and processing of orange juice by bringing innovations to the market place. By 2016, it was one of the most important orange producing regions in the world (the orange industry in Brazil had flourished over time, producing around three times the total volume produced in Florida in the 2015-16 season, according to FDOC statistics) and the main producer in the U.S. (Figure 2 shows that Florida accounted for approximately 90% of total U.S. production in the 2015-16 season).

Orange production in Florida has been severely affected during the 2000s by citrus greening, which was first found in Florida in 2005. By the end of 2016, no effective cure for citrus greening was foreseen in the near future. Figure 5 shows orange production during the last decade. Oranges produced in Florida represented 50.6% of presumed consumption in the U.S. in 2016 (Figure 2), and were very valuable for orange juice companies in the U.S. For instance, Florida’s Natural’s management believed that selling orange juice with 100% content of oranges produced in Florida was critical to provide high quality orange juice to American consumers.

International Food and Agribusiness Management Review

443

Trejo-Pech et al.

Volume 21, Issue 3, 2018

240 220 Million 90 pound boxes

200 180 160 140 120 100 80 60 40 20

-1 6

15

-1 5

20

14

-1 4 13

20

20

-1 3

12

-1 2

20

11

-1 1

10 20

20

-1 0

09

-0 9

20

08

-0 8

20

07

-0 7

20

06

-0 6

20

05

-0 5

20

04

-0 4

20

03

20

20

02

-0 3

0

Crop seasons Total Orange NFC

FCOJ Fresh

Figure 5. Florida orange production, 2002-2003 through 2015-16 (Florida Department of Citrus, Outlook Report 2016). The chard provides total round oranges production, in million 90 pound boxes, by utilization: FCOJ refers to oranges utilized for from concentrate orange juice, NFC for not-from-concentrate chilled orange juice, and fresh refers to oranges going to the fresh market.

5. Orange juice market players 5.1 Florida’s Natural By 2016, Florida’s Natural Growers, better known as Florida’s Natural, was a division of Citrus World Inc. Florida’s Natural was a cooperative of citrus growers with headquarters in Lake Wales, Florida, a small city located between Tampa and Orlando. Florida’s Natural was a federated cooperative because it was owned by fresh fruit cooperatives.7 An individual grower became a member by first joining one of the fresh fruit cooperatives. The cooperative was initially organized in 1933 ‘by a few growers with a passion for producing the best possible citrus products on Earth.’8 In 2016, the group had around 1000 farmer/members across twelve grower associations harvesting 60,000 plus acres of citrus groves for processing. Florida’s Natural members planted, raised and cared for the trees; grew and cultivated the fruit; and processed and mainly packaged pure, not from concentrate citrus juice.

Florida’s Natural brands portfolio included Florida’s Natural, Donald Duck, Bluebird, and Growers Pride.

Florida’s Natural was a successful premium brand ranked among the two giant global brands: PepsiCo’s Tropicana and Coca-Cola’s Minute Maid and Simply Orange. Florida’s Natural products included orange juices, grapefruit juices, lemonades, related juices, and juice blends. Figure 6 provides a list of products under the Florida’s Natural umbrella by the end of 2016. The core product line of Florida’s Natural was the not-from-concentrate 100% premium orange juice. In an effort to serve the emerging reduced-sugar, reduced-calorie market segment, Florida’s Natural introduced in 2015 the ‘Fit and Delicious’ product line, 7 8

A grower may choose to grow for the fresh market, processed market or both.

Florida’s Natural website. Retrieved on June 26, 2016 from https://www.floridasnatural.com/who-we-are.

International Food and Agribusiness Management Review

444

Trejo-Pech et al.

Volume 21, Issue 3, 2018 Orange and grapefruit juices

Other juices

Fit and delicious and juice blends

Figure 6. Florida’s Natural portfolio of products. which comprises products with 45% less sugar and calories.9 Florida’s Natural employed about 650 people in its processing plant and 100 people in its packaging plant.

Florida’s Natural was a consolidated brand. Indeed, a survey conducted during 2016 by MSW-ARS Research/ The Brand Strength Monitor showed that Florida’s Natural market shares were increasing through the year (Figure 7). Furthermore, the importance of the Florida’s Natural brand went beyond the orange juice segment.

Florida’s Natural was one of the top 20 brands in the juice and juice drinks industry (all flavors included) in the U.S., as shown in Table 3.10 In 2016, virtually all fruit processed by Florida’s Natural was received from its cooperative members, but the processing plant was open to receive high quality fruit from other Florida producers as well, as shown in Table 4. The cooperative had a healthy financial position with a very low level of debt, capital expenditures investment representing around 20% of total property, plant and equipment during 2011 to 2013; with this portion reduced in 2014 and 2015, reflecting the excess capacity/scarce supply of citrus crops in recent 9

A video related to “Fit and Delicious” is available at: http://tinyurl.com/y862ce49.

In addition, Florida’s Natural was a top brand in the premium grapefruit juice market (Bouffard, 2014a). According to Euromonitor (2017), grapefruit juice had a 2% share in the J&JD industry in 2016.

10

International Food and Agribusiness Management Review

445

Trejo-Pech et al.

Volume 21, Issue 3, 2018

22 20

(%)

18 16 14 12 10 Jan/Feb

Mar/Apr

May/Jun

Jul/Aug

Men

Sep/Oct

Women

Nov/Dec Total

Figure 7. Brand preferences for Florida’s Natural orange juice by gender (MSW-ARS Research/The Brand Strength Monitor, 2017). 15,738 respondents, aged 18-74, interested in orange juice. Shares up to December 2016 only shown in this figure for the purpose of this study.

Table 3. Top 20 brand shares of juice and juice drinks (all flavors) in the United States (Euromonitor, 2016).1 Brand

Company name (global brand owner)

%

Minute Maid Private label Capri-Sonne Tropicana Ocean Spray Simply Snapple Simply Orange Sunny Delight Tampico Kool-Aid Florida’s Natural Welch’s Dole Hawaiian Punch V8 V8 Splash Brisk Mott’s Naked Others Total

Coca-Cola Co, The Private Label Deutsche SiSi-Werke GmbH & Co KG PepsiCo Inc Ocean Spray Cranberries Inc Coca-Cola Co, The Dr Pepper Snapple Group Inc Coca-Cola Co, The Sunny Delight Beverages Co Houchens Industries Inc Kraft Heinz Co Florida’s Natural Growers National Grape Co-operative Association Inc Dole Food Co Inc Dr Pepper Snapple Group Inc Campbell Soup Co Campbell Soup Co Unilever Group Dr Pepper Snapple Group Inc PepsiCo Inc Others

9.0 7.4 6.9 6.3 5.5 4.7 4.3 3.9 3.3 3.3 3.0 2.6 1.5 1.4 1.4 1.4 1.2 1.1 1.0 1.0 29.7 100.0

1 Data extracted by authors from Brands/Soft Drinks/Juice/USA categories. Sorted according to off-trade sales volume. Brands with

shares lower than 1% as of 2015 were grouped as ‘Others’.

International Food and Agribusiness Management Review

446

Trejo-Pech et al.

Volume 21, Issue 3, 2018

years due to citrus greening, which was affecting production not only in Florida, but in nearly all major citrus production regions in the world. Florida’s Natural had kept its efficiency at high levels even during the difficult times the industry was facing, according to the sales turnover ratios, measured as sales to total assets. Table 4 provides selected financial items of Florida’s Natural. As part of a cooperative, farmer/ members received a proportional share of the cooperative’s profits after it deducts the costs of processing the fruit and marketing the products.

5.2 Tropicana Tropicana Products, Inc., along with Gatorade and Quaker, was an important part of PepsiCo’s fast-growing nutrition segment in 2016.11 Tropicana procured its oranges from the State of Florida, supplementing its fruit needs with imports from Brazil when Florida supplies were insufficient to meet its production needs (Esterl, 2012). In 2016, Tropicana’s portfolio of products marketed in the U.S. included seven product lines: Tropicana Pure Premium, Trop50, Tropicana Drinks, Tropicana Twister, Premium Lemonades and Drinks, Farmstand, and Tropics. By then, only the Tropicana Pure Premium product line was advertised by Tropicana as 100% Florida orange juice. Trop50 was a juice drink with 50% less sugar and calories and no artificial sweeteners. Introduced in 2009, Trop50 was considered a breakthrough category innovation, containing 50 calories per each 8 fl oz serving, and naturally sweetened from the stevia plant (PR Newswire, 2009). Other than orange, flavors offered by Tropicana included cranberry, strawberry, mixed, peach, apple, and grape, among others. Fruits were processed by Tropicana in its plants located in Bradenton and Ft. Pierce, Florida.

According to an article published in December 2016 by The Washington Post (Heat, 2016), Tropicana had recently been active trying to gain market share by targeting millennials, a 75 million segment with room for growth. Tropicana was advertising its orange juices in Ashton Kutchers’ A Plus digital news site, a channel that delivers its content through a website and a mobile app, with a focus on positive journalism, and that has 11.5 million unique monthly visitors. Videos of orange juice (http://tinyurl.com/y9vgmcdg) as a ‘feel good morning beverage’ have started to be shared in Facebook by celebrities.

5.3 Minute maid and simply orange By 2016, the Coca-Cola Company participated in the orange juice industry with its Minute Maid and Simply Orange brands. Minute Maid was one of the first brands to sell reconstituted orange juice (recon) in the late 1970s (Goldberg and Hogan, 2004), and continued selling recon juices for decades. Only recently, in 2011, Minute Maid entered the premium not-from-concentrate juice segment (Euromonitor, 2011). Minute Maid’s orange juice product line included both from concentrate and not-from-concentrate juices, and some products labeled as functional drinks (e.g. with vitamins and minerals like calcium). Other product lines included Lemonade & Punch, Light Juice Drinks, Variety Juice & Other (e.g. apple and grape juices, as well as sparkling punch), and Kids’ Juice & Juice Drinks (e.g. Coolers and juice boxes). Simply Orange,

11

Tropicana was acquired by PepsiCo on August, 1998 from Seagram as part of the strategy of the firm to focus on the nutrition market (Miller, 2016).

Table 4. Florida’s Natural selected financials (Florida’s Natural records).

Boxes received from cooperative members Capital expenditure to PP&E Sales to assets Working capital to assets Long-term debt as a % of equity

2011

2012

2013

2014

2015

92% 25% 1.4 23% 0.07

94% 22% 1.4 22% 0.06

94% 18% 1.2 21% 0.07

92% 13% 1.2 25% 0.05

91% 10% 1.3 32% 0.05

International Food and Agribusiness Management Review

447

Trejo-Pech et al.

Volume 21, Issue 3, 2018

in contrast, offered not-from-concentrate juice only; stand-alone orange juice and orange juice blends with fruits such as banana, mango and tangerine.

5.4 Orange juice market shares Table 5 provides market shares of refrigerated orange juice sales by top players in the premium orange juice segment, as estimated by ‘Dairy Foods’ (Kennedy, 2015). Private labels captured an important portion of the premium market, representing strong competition for national brands as supermarkets and hypermarkets had a significant share in the distribution system (Table 6).

6. Supply – citrus greening and planting programs The main challenge the Florida citrus industry faced in 2016 was from two diseases: citrus canker and HLB, the latter also known as citrus greening. There is no cure for either disease although the negative effects of citrus canker can be mitigated. Greening, however, was a much more challenging disease. Research efforts were underway on several fronts, but by the end of 2016 no effective means of reducing the impact of the disease had been found.12 Figure 5 shows Florida round orange production for the past 14 years. The 201516 crop of 81.4 million boxes was the smallest crop in decades and was about one-third the size of the crop produced in 2003-04. Furthermore, according to the Commercial Citrus Tree Inventory, new plantings in Florida had been less than tree removals lately. New planting incentive programs were needed to encourage 12 The

US Environmental Protection Agency approved the use of the bactericides in early 2016 which are believed to suppress the bacteria that causes HLB.

Table 5. Shares of refrigerated premium orange juices (based on Kennedy, 2015).1 Tropicana Pure Premium Simply Orange Private label Florida’s Natural Minute Maid Premium Others Total category

Shares of USD value

Shares of volume sold

29% 22% 16% 11% 7% 14% 100%

28% 20% 19% 10% 7% 16% 100%

1 Shares, estimated for the 52 Weeks Ending February 22, 2015, are for individual brand line listings, not total brand listings. Original

data is from Information Resources Inc. for the 52 weeks ended February 22, 2015, covering total U.S. multi-outlet (supermarkets, drugstores, mass market retailers, military commissaries and selected club and dollar retail chains).

Table 6. Distribution channels off-trade sales juice and juice drinks (%) as of 2015 (adapted from Euromonitor, 2016).

Convenience stores Discounters Forecourt retailers Hypermarkets Supermarkets Independent small grocers Other grocery retailers Non-grocery specialists Non-store retailing Vending Internet retailing

2.8% 5.1% 11.9% 28.9% 32.7% 4.9% 5.2% 1.4% 1.3% 0.8% 0.6% International Food and Agribusiness Management Review

448

Trejo-Pech et al.

Volume 21, Issue 3, 2018

growers to invest in new tree planting or the industry could disappear. The challenge for both public and private entities was to develop programs in the face of production risks posed by citrus greening.

6.1 New planting programs Several planting programs were available in 2016 to help Florida citrus growers to reduce production risk and to protect leading citrus juice brands. The Florida Department of Citrus (Spreen and Zansler, 2016a) estimated in 2016 that 47 million additional trees were needed over the next ten years in Florida to recover the roughly 32% decline of citrus groves since 2004. Spreen and Zansler (2016b) feature the planting programs outlined next.

Announced in 2014, the USDA funded a program known as the Tree Assistance Program (TAP). The TAP reimbursed growers for replanting trees/acreage lost to the effects of citrus greening. The program reimbursed about 60% of the cost of the trees, planting cost, and land preparation costs. Based on an estimated TAP contribution of $1,230 per acre, it was expected that the program would cover up to 6.0 million trees. In 2016, according to Florida Citrus Mutual, a growers’ trade organization, the TAP had paid out $6.8 million to cover the planting of more than 1 million trees (Bouffard, 2016). A number of eligibility provisions and payment constraints limited the widespread use of this program, which was beneficial mainly to small and medium-sized growers.

Minute Maid, the subsidiary of the Coca-Cola company, had implemented a program to incentivize production as well. The program consisted on a long-term contract offered to growers. The Minute Maid contract offered price floors and price ceilings in exchange for long-term growers’ commitments to provide fruit to the firm. The price paid would be subject to the Florida Department of Citrus post estimate price, which captures orange prices within the State of Florida. ‘A thriving Florida citrus industry is critical to helping us build our “Simply” and “Minute Maid” juice brands’ commented Steve Cahillane, President of Coca-Cola Americas (Giles, 2014: 22).

In 2014, Florida’s Natural announced major commitments to incentivize citrus plantings in Florida. Steve Caruso, CEO of Florida’s Natural by then,13 stated: ‘We can’t think of a better way to invest than with our growers... It demonstrates [our] belief in the long-term sustainability of the Florida citrus grower and will help enable the “Florida’s Natural” brand to continue to grow.’14 6.2 Florida’s Natural planting incentive program The Florida’s Natural PIP, worth $10 million to growers, was initially offered to cooperative members only, but Florida’s Natural was considering the possibility to accept applications by non-members (Bouffard, 2014b). Under the Florida’s Natural program, growers are given an interest free loan of $10 per tree for each orange tree planted. Growers then are asked for a 13-year commitment, three years for the non-bearing portion and 10 years of bearing life. Each year that fruit delivery is made, one-tenth of the loan is written off. After 10 years of fruit delivery, the entire loan is forgiven. Typically, healthy orange trees have a useful life of at least 30 years. The goal of Florida’s Natural was to provide support for up to one million new trees planted. Due to the PIP, the cooperative expected to receive about 2 million orange boxes per year after the lag period between planting and production (orange trees start bearing fruit in the third or fourth year). This amount is roughly equivalent to the decrease of Florida’s Natural processed production in 2013-14 (i.e. the Florida’s Natural plant processed about 15 million boxes in 2013, down from more than 17 million in 2012-13 (Bouffard, 2014b)). To ensure the success of the program, Florida’s Natural would support growers with technical service in terms of planting density, grove caretaking, fertilization, and pest control practices.

Table 7 provides selected characteristics of the PIP. 13

Stephen Caruso served as CEO for the cooperative during 22 years. His retirement as CEO was announced at the 2014 annual meeting held on November 11, 2014 (Bouffard, 2014a).

14 Cited in Giles (2014: 22).

International Food and Agribusiness Management Review

449

Trejo-Pech et al.

Volume 21, Issue 3, 2018

Table 7. Main characteristics of the planting incentive program (PIP) by Florida’s Natural (based on Bouffard, 2014b).

Total amount for the PIP Expected number of new orange trees Expected additional production Growers’ commitments

$10 million 1 million 2 million orange boxes per year 1. Sell all oranges from new trees to Florida’s Natural 2. Adhere to certain standards on grove caretaking, including fertilization and pest control practices 3. Plant 270 to 350 trees per acre

Agricultural practices, as encouraged by Florida’s Natural, were considered relevant for groves to remain financially viable. Prior to greening, the standard planting density was much lower than the recommendation in the PIP. Higher density was desirable particularly for the potential partial loss of harvestable oranges due to greening. Indeed, by 2016, high density planting was becoming a common practice in the overall Florida industry; Spreen and Zansler (2016a) showed that the rate of new tree plantings had slightly outpaced the rate of new acreage plantings between 2013-14 and 2014-15 seasons, suggesting higher tree densities. Minimum standards on caretaking, fertilization, and pest control practices were also important since adding fertilizers and other nutrients to the trees’ roots helped to fight citrus greening (Semuels, 2015). It was estimated that the $10 Florida’s Natural incentive per tree represented 75% or more of total replanting costs, including the tree, irrigation, and labor.15 6.3 Prices and costs at the grove level A recent publication by the Institute of Food and Agricultural Sciences at the University of Florida (Singerman, 2016) provides the cost structure of orange production in southwest Florida. By surveying growers, this study reported that total cost production was $2,328 per acre on average, during the 2015-16 season (including management cost and investment financial costs). Furthermore, particularly horticultural practices costs have increased lately due mainly to efforts by growers to mitigate the negative effects of citrus greening. Costs in 2016 implied a break-even on-tree price varying from $6.21 to $13.30 per box depending on estimated planting densities. While on-tree prices received by growers for processed oranges had also increased lately – as reported by the Florida Department of Citrus, costs had increased faster than on-tree prices, reducing profitability for growers.

7. Florida oranges It has been claimed that Florida produced the highest quality oranges in the world (Herndon et al., 1994).

Oranges from Florida were also highly valued due to the fact that were more productive for juicing compared to oranges from California, which were more appealing for the fresh market (Pearcy and Goldberg, 2000).

Others preferred oranges produced in the U.S. over oranges of the leading producer Brazil, or other countries, due to safety concerns. Some chemicals used in the production process in those countries (e.g. Carbendazim, a product to control fungus and mold) were not approved for use on orange crops in the United States.

In any case, Florida oranges were highly valued by consumers, and 100% juice from Florida products were positioned in the premium segment of the U.S. market. By 2016 some, but not all juices, in the Tropicana premium product line had the tag line ‘Squeezed from Fresh Florida Oranges’ in its logo. The 100% juice from Florida marketing strategy by Tropicana was re-introduced in 2012 for its Pure Premium line, years after the firm started to mix in oranges from Brazil in 2007 (Brown and Washton, 2013). Florida’s Natural had focused marketing and communication efforts to show customers its quality proposition by using 100% 15

Estimated by authors assuming 225 planting density, $400 irrigation cost per acre, and $10.5 cost per tree plus planting cost.

International Food and Agribusiness Management Review

450

Trejo-Pech et al.

Volume 21, Issue 3, 2018

Florida fruit as well. Florida’s Natural stated in its website: ‘There is one national orange juice brand that can proudly claim to use only 100% American oranges. It is not “Simply Orange” and it is not “Tropicana”.

Only “Florida’s Natural” never imports orange juice from overseas. Great Taste. Naturally.’16 Figure 8 shows Florida’s Natural pride on being a Florida orange producer.

8. Concluding comments According to USDA production estimates compiled by the Florida Department of Citrus, during the 20152016 season orange juice supply from Florida represented about 90% of domestic production. The continuous drop of orange production due to citrus greening, however, was challenging the orange industry. Indeed, analysts were predicting that some small to medium size juice processors would shut down their operations in the near future. As one industry participant interviewed in a 2014 article commented: ‘When you got an industry infrastructure that can handle 200 million boxes of oranges and you’re currently producing 100 million boxes, everybody does not get to play.’ (Bouffard, 2014c). In the fall of 2016, the National Agriculture Statistics Service of the U.S. Department of Agriculture announced its first projection for the 2016-17 season: Florida orange crop was expected to further decline by 12% to 72 million boxes.

Florida’s Natural top management recognized the need to revisit the firm’s strategy. Dr Behr and his team were concerned about how to help the cooperative’s growers and to better serve Florida’s Natural customers.

On the supply side, the team was about to discuss possible adjustments, if any, to the Planting Incentive Program. A potential issue for the discussion was whether it was worth to expand the PIP given its costs, risk, and potential benefits. On the market side, the team knew the cooperative needed to use wisely their resources to continue competing closely against the two giant beverage firms PepsiCo and Coca Cola, which presumably had higher marketing budgets. What actions could Florida’s Natural take to leverage its branding position? According to the article in The Washington Post referred before (Heat, 2016): ‘juice companies that thrive with this generation [millennials] have packaging that makes it look local, tout the health benefits and understand that it is about what the brand represents, than what the product actually is.’ Did Florida’s Natural products have the attributes to reach the millennials niche? These and other questions were relevant for Florida’s Natural and other orange juice companies by the end of 2016.

16

“Florida’s Natural commercial: Flag” video available in the company’s website. Retrieved on June 26, 2016 from https://www.floridasnatural. com/who-we-are/videos.php.

Figure 8. Picture posted in Florida’s Natural Facebook page.

International Food and Agribusiness Management Review

451

Trejo-Pech et al.

Volume 21, Issue 3, 2018

Acknowledgment Carlos Trejo-Pech acknowledges that this work was partially supported by the USDA National Institute of Food and Agriculture, Hatch Multi-State project 1012420.

References Bouffard, K. 2014a. Lake Wales enjoying a good year for OJ. The Ledger, November 11. Available at: http:// tinyurl.com/y79db8v8.

Bouffard, K. 2014b. Cooperative’s new program offering growers cash to plan more trees. The Ledger, November 16. Available at: http://tinyurl.com/y8x5ka84.

Bouffard, K. 2014c. At least one juice processor expected to close. The Ledger, July 06. Available at: http:// tinyurl.com/yde963bv.

Bouffard, K. 2016. State offers up to $250,000 incentive to growers. Herald-Tribune, August 25. Available at: http://tinyurl.com/y9jeekrg.

Brown, R. and R. Washton. 2013. Fruit and vegetable juices: U.S. market trends. Packaged Facts. Essential Insights on Consumer Markets, April 2013, 1-112. Available at: http://tinyurl.com/ya6zfeg8.

Crawford, E. 2016. Three trends driving sales of 100% juice even as sugar debate rages. Food navigatorusa.com. Available at: http://tinyurl.com/y8ubvz7e.

Esterl, M. 2012. Pepsi finds fungicide traces in Tropicana OJ. The Wall Street Journal (Online), January 14.

Available at: http://tinyurl.com/ybz62ml7.

Euromonitor. 2011. Minute maid – creating a global brand. Euromonitor International. November 3. Database available at: https://www.euromonitor.com/sign-in.

Euromonitor. 2016. Juices in the U.S. Passport Euromonitor International, March 2016. Database available at: https://www.euromonitor.com/sign-in.

Euromonitor. 2017. Juices in the U.S. Passport Euromonitor International, February 2017. Available at: http://tinyurl.com/y8czauct.

FDOC. 2017. Florida citrus season retrospective: 2016-17 season. Final season outlook update. Florida Department of Citrus, Economic and Marketing Research Department Working Papers. September 2017. Available at: http://tinyurl.com/yc8zw7km.

Giles, F. 2014. Programs promote plantings, citrus processors and USDA step up to support growers who are putting new trees in the ground. Florida Grower, December 2014: 107(12): 22-24.

Goldberg, R. and H. Hogan. 2004. Can Florida orange growers survive globalization? Harvard Business School Publishing, Case 9-904-415, 1-25. Harvard Business School Publishing, Boston, MA, USA. Heat, T. 2016. Can millennials learn to love orange juice? The Washington Post, December 22. Available at: http://tinyurl.com/yd7fng94.

Herndon, P., K. Morris and R. Goldberg. 1994. Alcoma: The strategic use of frozen concentrated orange juice futures. Harvard Business School Publishing, Case 9-595-029. Harvard Business School Publishing, Boston, MA, USA. Kennedy, S. 2015. Sales rise for veggie juices, juice smoothies. Dairy Foods – Market Trends, June 8.

Available at: http://tinyurl.com/ydgcybke.

Leschewski, A., D. Weatherspoon and A. Kuhns. 2016. A segmented hedonic analysis of the nutritional composition of fruit beverages. International Food and Agribusiness Management Review 10(3): 119-140.

Miller, S. 2016. CEO helped Pepsi become ‘Choice of New Generation.’ The Washington Post, June 2.

Available at: http://tinyurl.com/ybp3hjmm.

MSW-ARS Research/The Brand Strength Monitor. 2017. United States: brand preferences for Florida’s Natural orange juice from January/February 2016 through July/August 2017, by age. In Statista – The statistics portal. Available at: http://tinyurl.com/y73fjveu.

Pearcy, B. and R. Goldberg. 2000. Florida Department of Citrus. Harvard Business School Publishing, Case 9-900-009. Harvard Business School Publishing, Boston, MA, USA.

International Food and Agribusiness Management Review

452

Trejo-Pech et al.

Volume 21, Issue 3, 2018

PR Newswire. 2009. Tropicana first to introduce the all-natural sweetness of stevia to the orange juice aisle with the launch of Trop50, March 30. Available at: http://tinyurl.com/y8uvlmcr.

Semuels, A. 2015. Florida without oranges. The Atlantic, January 27. Available at: http://tinyurl.com/ybz2eap7.

Singerman, A. 2016. Cost of production for processed orange in Southwest Florida, 2015/16. Citrus Research and Education Center (CREC) at IFAS-UF. Available at: http://tinyurl.com/yd4pkq4r.

Spreen, T. and M. Zansler. 2016a. Florida round orange production trends. Economic & Market Research Department, Florida Department of Citrus Working Paper Series, September 2016, Working paper 2016-1, 1-26. Available at: http://tinyurl.com/y8wlbtqo.

Spreen, T. and M. Zansler. 2016b. Economic analysis of incentives to plant citrus trees in Florida. HorTechnology 26(6): 720-726.

STATISTICA. 2017. Outlook report for juices in the U.S. Available at: http://tinyurl.com/yc88uasn.

Trejo-Pech, C., T. Spreen and M. Zansler. 2017. Is growing oranges in Florida a good investment? Proceedings of the Agricultural and Applied Economics Association 2017 Annual Meeting. Available at: http:// tinyurl.com/yc7zyh3t.

International Food and Agribusiness Management Review

453