Economics and Business Review, vol. 1(15), no. 3, 2015 by EBReview

Economics and Business Review

ISSN 2392-1641

Volume 1 1(15) 2 2015 Volume (15)Number Number 3 2015 CONTENTS Introduction Piotr Manikowski, W. Jean Kwon ARTICLES The changing architecture of the safety net in insurance worldwide: post-crisis developments Jan Monkiewicz, Lech Gąsiorkiewicz, Marek Monkiewicz The determinants of nonlife insurance penetration in selected countries from South Eastern Europe Klime Poposki, Jordan Kjosevski, Zoran Stojanovski Microeconomic and macroeconomic determinants of the profitability of the insurance sector in Macedonia Tanja Drvoshanova-Eliskovska Policyholder and insurance policy features as determinants of life insurance lapse – evidence from Croatia Marijana Ćurak, Doris Podrug, Klime Poposki Longevity risk and the design of the Polish pension system Marek Szczepański Polish farmers’ perception of spring frost and the use of crop insurance against this phenomenon in Poland Monika Kaczała, Dorota Wiśniewska Insurance and risk management systems in Russia Nadezda Kirillova BOOK REVIEWS Jeremy Rifkin, Zero Marginal Cost Society. The Internet of Things, the Collaborative Commons, and the Eclipse of Capitalism, Palgrave Macmillan, New York 2014 (Jan Polowczyk) Andrzej Rzońca, Kryzys banków centralnych. Skutki stopy procentowej bliskiej zera [Central Banks Crisis. The Impact of Interest Rates Close to Zero], Wydawnictwo C.H. Beck, Warszawa 2014 (Tadeusz Kowalski)

Poznań University of Economics Press

Editorial Board Ryszard Barczyk Witold Jurek Cezary Kochalski Tadeusz Kowalski (Editor-in-Chief) Henryk Mruk Ida Musiałkowska Jerzy Schroeder Jacek Wallusch Maciej Żukowski International Editorial Advisory Board Udo Broll – School of International Studies (ZIS), Technische Universität, Dresden Wojciech Florkowski – University of Georgia, Griffin Binam Ghimire – Northumbria University, Newcastle upon Tyne Christopher J. Green – Loughborough University John Hogan – Georgia State University, Atlanta Bruce E. Kaufman – Georgia State University, Atlanta Steve Letza – Corporate Governance Business School Bournemouth University Victor Murinde – University of Birmingham Hugh Scullion – National University of Ireland, Galway Yochanan Shachmurove – The City College, City University of New York Richard Sweeney – The McDonough School of Business, Georgetown University, Washington D.C. Thomas Taylor – School of Business and Accountancy, Wake Forest University, Winston-Salem Clas Wihlborg – Argyros School of Business and Economics, Chapman University, Orange Jan Winiecki – University of Information Technology and Management in Rzeszów Habte G. Woldu – School of Management, The University of Texas at Dallas Thematic Editors Economics: Ryszard Barczyk, Tadeusz Kowalski, Ida Musiałkowska, Jacek Wallusch, Maciej Żukowski • Econometrics: Witold Jurek, Jacek Wallusch • Finance: Witold Jurek, Cezary Kochalski • Management and Marketing: Henryk Mruk, Cezary Kochalski, Ida Musiałkowska, Jerzy Schroeder • Statistics: Elżbieta Gołata, Krzysztof Szwarc Language Editor: Owen Easteal • IT Editor: Piotr Stolarski

ISSN 2392-1641

POZNAŃ UNIVERSITY OF ECONOMICS PRESS ul. Powstańców Wielkopolskich 16, 61-895 Poznań, Poland phone +48 61 854 31 54, +48 61 854 31 55, fax +48 61 854 31 59 www.wydawnictwo-ue.pl, e-mail: wydawnictwo@ue.poznan.pl postal address: al. Niepodległości 10, 61-875 Poznań, Poland Printed and bound in Poland by: Poznań University of Economics Print Shop Circulation: 300 copies

Volume 1 (15) Number 3 2015 CONTENTS Introduction...................................................................................................................................... 3 Piotr Manikowski, W. Jean Kwon ARTICLES The changing architecture of the safety net in insurance worldwide: post-crisis developments Jan Monkiewicz, Lech Gąsiorkiewicz, Marek Monkiewicz............................................................ 5 The determinants of nonlife insurance penetration in selected countries from South Eastern Europe Klime Poposki, Jordan Kjosevski, Zoran Stojanovski...................................................................... 20 Microeconomic and macroeconomic determinants of the profitability of the insurance sector in Macedonia Tanja Drvoshanova-Eliskovska......................................................................................................... 38 Policyholder and insurance policy features as determinants of life insurance lapse – evidence from Croatia Marijana Ćurak, Doris Podrug, Klime Poposki.............................................................................. 58 Longevity risk and the design of the Polish pension system Marek Szczepański............................................................................................................................. 78 Polish farmers’ perception of spring frost and the use of crop insurance against this phenomenon in Poland Monika Kaczała, Dorota Wiśniewska........................................................................................... 90 Insurance and risk management systems in Russia Nadezda Kirillova.......................................................................................................................... 112 BOOK REVIEWS Jeremy Rifkin, Zero Marginal Cost Society. The Internet of Things, the Collaborative Commons, and the Eclipse of Capitalism, Palgrave Macmillan, New York 2014 (Jan Polowczyk).............. 121 Andrzej Rzońca, Kryzys banków centralnych. Skutki stopy procentowej bliskiej zera [Central Banks Crisis. The Impact of Interest Rates Close to Zero], Wydawnictwo C.H. Beck, Warszawa 2014 (Tadeusz Kowalski)................................................................................................................... 125

Economics and Business Review, Vol. 1 (15), No. 3, 2015: 3–4 DOI: 10.18559/ebr.2015.3.1

Introduction The insurance industry continues to play an important role in economic development in every domestic and regional market as it does for the global economy as well for as societal well-being. As its importance rises it is subject to the ever-increasing oversight from regulatory authorities – not only within the industry but also in the financial services sector. We observe changes in legislation, economic fluctuations and many more processes that are arising in every corner of the world. At the same time insurance markets are not free from various disturbances or recession, let alone the impact of the recent financial crisis on the markets. Therefore we should strive to better understand the relationship between insurance market development, economic growth and societal well-being, especially in countries with fast growing insurance markets. This issue of the Economics and Business Review on Risk Management and Insurance includes a number of studies that examine topics related to the functioning of modern insurance markets and in particular in selected Southern and Northern European markets. As there are differences in market development and the availability of research sources there are some differences in the depth and scope amongst the papers in this issue. We consider that such differences should be regarded as a starting point for the further development of the national markets covered in this issue. The first paper entitled “The changing architecture of the safety net in insurance worldwide: post-crisis developments” provides a review and analysis of the direction of the evolution of the structure of the safety net in insurance as compared to that of banking. It also discusses the influence of recent regulatory initiatives in insurance markets. Special attention is paid to macroprudential supervision which has been revived as a major regulatory factor in the aftermath of the recent global financial crisis. Matters related to Global Systemically Important Insurers are also covered. All the remaining papers analyze issues in domestic insurance markets. We list them in geographical sequence from Southern to Northern Europe. All together these papers cover ten countries: Albania, Belarus, Bosnia and Herzegovina, Croatia, Macedonia, Moldova, Poland, Russia, Serbia and Ukraine. The first paper entitled “The determinants of nonlife insurance penetration in selected countries from South Eastern Europe” examines the factors influencing non-life insurance penetration in South Eastern European markets during 1995–2011. By applying a panel vector error correction model the authors find a significant relationship amongst non-life insurance penetration and GDP per capita, the number of passenger cars per 1,000 people and local regulation.

Economics and Business Review, Vol. 1 (15), No. 3, 2015

The article titled “Microeconomic and macroeconomic determinants of the profitability of the insurance sector in Macedonia” identifies the micro- and macroeconomic factors affecting the profitability of the Macedonian insurance market. These determinants include, but are not limited to, the value of assets and interest rates. The author suggests recommendations from microperspectives (e.g., productive use of resources, innovative products, portfolio enhancement and recapitalization) and from macro-perspectives (e.g., structural reforms, extension of savings and investments, new financial instruments, mutual projects between the insurance and banking sectors). “Policyholder and insurance policy features as determinants of life insurance lapse – evidence from Croatia” deals with issues related to life insurance lapse. The authors investigate the drivers behind such lapses in Croatia using survey data. They find that the main determinants include the number of children, income level, the financial status of policyholder as well as duration of the life insurance contract. The study titled “Longevity risk and the design of the Polish pension system” focuses on the management of longevity risk in the reformed pension system in Poland. The author examine whether the current Polish pension scheme is more resistant to the risk of longevity than the previous one, especially the impact of pension account changes from defined benefit to defined contribution. “Polish farmers’ perception of spring frost and the use of crop insurance against this phenomenon in Poland” presents the findings from a March 2012 survey using a focus group of 750 farmers across Poland. The authors describe how farmers assess spring frost in the context of other sources of risk and investigate if there are any interdependencies between the perception of spring frost and the use of crop insurance to cover this peril. The analyses indicate that spring frost perception primarily depends on a farmer’s experience in terms of most natural perils. That is, any kind of loss, regardless of its cause and scale, is conducive to ranking spring frost risk. The final article entitled “Insurance and risk management systems in Russia” covers issues related to corporate insurance in the Russian Federation, the method of formation of corporate insurance systems, the assessment tools used to assess the financial condition of insurers and insurance service users. The author adds recommendations for improvements to the insurance systems in the Russian Federation. This study offers a summary of statistical data on the Russian insurance market in recent years. We, the editors of this issue of the Economics and Business Review, would like to express our sincere thanks to the anonymous reviewers for comments and suggestions as well as to the Language Editor for making this issue possible. Piotr Manikowski, W. Jean Kwon

Economics and Business Review, Vol. 1 (15), No. 3, 2015: 5–19 DOI: 10.18559/ebr.2015.3.2

The changing architecture of the safety net in insurance worldwide: post crisis developments1 Jan Monkiewicz2, Lech Gąsiorkiewicz2, Marek Monkiewicz3

Abstract : This aim of this paper is to explain the safety net of the insurance sectorunderstood as the total means which ensure the safety of the insurance markets and their customers and how it has been heavily affected by recent regulatory initiatives. The paper then provides a review and analysis of the directions of the evolution of the architecture safety net in insurance as compared to banking. Special attention is paid to macroprudential supervision which is believed to constitute a major regulatory innovation in the aftermath of the recent global financial crisis. Additionally new restructuring and resolution concepts and tools are discussed. Particular attention is focused on Global Systemically Important Insurers (G-SII’s) which are the focus of safety net regulation and which provide a regulatory impetus for the remaining part of the industry. Keywords : safety net in insurance, globally systemically important insurance institutions, systemic risk, macroprudential supervision. JEL codes : G15, G22, K23.

Introduction The recent global financial crisis has revealed a need for the substantial rearrangement of the existing financial safety net to address new challenges coming both from within the financial sector at large as well as its individual components. Its final goal is a better identification of the risks of the financial system and a better delivery of the crisis management tools addressing them. Most of the new initiatives in this regard originated within the G20 and Financial Stability Board – de facto financial reform secretariat of G20. It underlines the importance of 1

Article received 21 December 2014, accepted 3 August 2015. Warsaw University of Technology, Faculty of Management, Narbutta 85, 02-524 Warsaw, Poland; corresponding author: j.monkiewicz@wz.pw.edu.pl. 3 Warsaw University, Faculty of Management, Szturmowa 1/3, 02-678 Warsaw, Poland. 2

Economics and Business Review, Vol. 1(15), No. 3, 2015

the topic in the current political agenda and provides the necessary weight behind the reform proposals. Overwhelmingly the political and research focus so far is concentrated on the banking sector and relevant solutions for banks. In the aftermath this “banking perspective” has been increasingly applied to other financial sectors. This is particularly well observed in the insurance area. The focus of our analysis is the insurance sector and insurance relevant responses to the reform initiatives. Its purpose is to provide an analysis of the direction of the evolution of the safety net structure in insurance as compared to banking and assess its specificity and consequences. The paper is split into four sections. In the first we discuss the context of the whole process which we believe is the emerging new regulatory paradigm of the financial sector which essentially calls for a macroprudential approach and public management of the risks of the financial system. In the next section we discuss the conventional safety net arrangements in insurance which dominated throughout the world prior to the recent global financial crisis. Finally we review major new developments in the insurance safety net such as the emergence of macroprudential supervision, introduction of multilayer regulatory standards, the development of special resolution regimes, systemic crisis back stops within government structures and provide an assessment. Section four contains our conclusions.

1. The context – the emerging new regulatory paradigm of the financial system As a result of the developments during the last global financial crisis and the lessons learned the entire previous regulatory and supervisory paradigm has been placed in question. The whole of this “Washington consensus”, supported by the IMF and surrounding institutions’ recommendations and policies, was based on efficient market orthodoxy. It has dominated the financial regulatory domain over the last 25 years or so and has fallen apart as an effect of the last global financial crisis [Helleiner 2010]. Its essence relied on unconditional faith in the efficiency and rationality of the financial markets. It has been assumed that financial markets are, in principle, efficient though with a tendency to short term volatility. Their proper functioning required basically only adequate access to market information and market discipline. These markets should not have been overburdened with regulatory discipline but should have been left to their own devices. The”Washington consensus” has basically assumed that the financial system is safe with private risk management executed at the level of individual financial institutions. Consequently it believed that financial innovations such as securitisation or derivatives are generically good at providing more opportunities for private risk management in financial systems hence making them safer.

J. Monkiewicz, L. Gąsiorkiewicz, M. Monkiewicz, The changing architecture of the safety net

The “Washington consensus” focused its attention on the safety and stability of individual financial institutions without paying much attention to their interconnectedness and common exposure and even possible contagion channels (see Table 1). Its principal centre of supervision has therefore been the microprudential bodies. Generally speaking the heart of this old paradigm was based on a “regulatory trilogy” which encompassed greater transparency, more disclosure and more effective risk management by individual financial firms [Eatwell 2009]. Table 1. The Washington and Basel consensus in comparison Features

View of financial markets

Instruments applied

Supervision in place

Washington ––Largely efficient, ratio- ––First hand role of consensus nal and self repairing market discipline ––Prone to short term ––Enhanced transpardisruption ency and disclosure ––Financial innovations ––Private risk managecontribute to financial ment (VaR models) stability and safety within the financial ––Requires better more institutions timely information but should be left to their own devices

––Formal and superficial ––VaR models and microprudential supervision – the route to stability ––System made safe by allowing individual institutions to manage risk ––Supervision isolated from politics

Basel consensus

––Material, penetrating and profound ––Macrosystem – wide perspective ––Safety of the financial system becomes a public preoccupation ––Excessive complexity and financial innovation put under strict control ––Supervision infiltrated by politics

––Financial markets are inherently procyclical and prone to herding ––Financial innovation and increasing complexity can make the system less stable ––Governance and business models should be subject to public control

––First hand role of the regulatory discipline ––Enhanced regulatory and supervisory powers ––Public management of the financial system risk ––Leverage limits and countercyclical capital buffers

Sources: [Baker 2013: 117] and authors’ own additions.

The focus of the new “Basel consensus” is a “macroprudential” approach and with this a demand for public management of the risks of the financial system. Its important feature is the introduction of special regulatory standards for the systemically important financial institutions, both at global and domestic levels. The new consensus also promotes a new supervisory parameter for micro prudential supervisors which should focus their attention not

Economics and Business Review, Vol. 1(15), No. 3, 2015

only on “solo” companies but entire financial groups and their internal risk management systems. The new consensus clearly elevates the role of regulatory discipline which should take precedence over the market. This includes even questioning of fundamental property rights and the granting of special resolution powers to public bodies. Management boards and shareholders under the new rules clearly much less trusted to control market excesses than as in the past. The emerging regulatory and supervisory model in insurance is characterised additionally by the accelerating globalisation of regulatory choices. Recent decisions of the Financial Stability Board regarding the designation in 2013 of Globally Systemic Important Insurers (G-SIIs) and of the International Association of Insurance Supervisors, regarding accelerated development of the Comframe, including its Insurance Capital Standard (ICS) to be implemented by 2019 by all Internationally Active Insurance Groups (IAIG) operating globally, are the most profound indications of this new, qualitative development. The upcoming new regulatory system is substantially growing in complexity in comparison to the one we know. It introduces a multilayer regulatory architecture which is composed of at least of four layers – the “ordinary” companies’ layer, IAIGs layer, G-SIIs layer and a “specials” layer (mutuals, captives). It is additionally possible to have a Domestically Systemic Insurance Institutions (DSII) layer. Finally the new regulatory model clearly expands the regulatory parameter by introducing a macroprudential pillar to control, mitigate and possibly prevent systemic risk. On the supervisory front we are confronted with the development of enhanced supervisory penetration and the emergence of a multipolar supervisory system. Classical microprudential bodies are complemented with macroprudential authorities and enhanced consumer protection agencies. Moreover special crisis management arrangements are becoming an important part of the new regulatory and supervisory framework. Their role is to limit potentially negative spillovers and secure crisis management plans of action in advance to avoid improvisation [Claessens and Kodes 2014]. Supervisors are expected to be more holistic and penetrative in their approach to their oversight and assessment. A good example is the development of the group wide supervision concept. Moreover the new supervisory models take into account the need for increased transborder coordination with MMOUs, supervisory colleges and Crisis Management Groups as available toolkits. New supervisory models need to additionally recognize the increased role of shared supervision and supervisory co-decisions. It is also characterised by the implementation of new, forward looking supervisory tools – early warning indicators, scenarios and stress testing. Finally, the upcoming supervisory model is placed under the growing role of central banks.

J. Monkiewicz, L. Gąsiorkiewicz, M. Monkiewicz, The changing architecture of the safety net

2. Safety net arrangements in insurance prior to the crisis Historically safety nets are the main by-product of the Great Depression 1929–1933. Once mainly implicit and ad hoc they have now become more and more explicit and permanent in nature. Initially the concept of the safety net was a narrowly defined feature relevant to banking. For this reason its role in the stability of the financial system and the combat of the systemic risk has been enhanced. With time this strict limitation began to wane. The concept of the financial safety net came into the ownership of the entire financial system. It also started to be expressis verbis recognised in academic literature not only as the tool for addressing macroeconomic and macroprudential concerns but also to assist in accomplishing the microeconomic issues. It thus reflects the growing convergence of financial markets and the growing interconnectedness of financial institutions. It also reflects our better understanding of the changes that have taken place in the financial system. Without going into detail the financial safety net may be defined as all devices which ensure the protection of the safety of the financial markets and their customers [Solarz 2008; Monkiewicz 2013]. These devices may include both private and public elements, both regulations and institutions. The safety net is a public construct and is shaped predominantly by the State. Private elements become a part of the safety net only once they are authorized or “accredited” by the State. Contemporary safety nets are focused primarily on the prudential protection of financial intermediaries and their customers with little attention given to financial products. The aims of this protection may vary in different jurisdictions and market segments. In some (e. g. banks) macroeconomic and stability concerns prevail, in others (e. g. insurance) more microeconomic targets are targeted. At national level the net is an aggregate of the individual segments of the financial sector with various links and dependencies. At the international level it is an aggregate of national constructions and international layers. There are two sets of functions that may be allocated to the safety nets: –– preventive, which protects financial systems against the financial shock –– mitigating, (crisis management) which aims at limiting the cost of the failures of financial systems. Overall the safety of the financial system may be viewed as an aggregate of the safety nets of the individual financial sectors embracing inter alia banking, insurance and securities. These individual parts of the sector specific safety nets exist both in competitive as well as in cooperative relationships. In either case these may lead to the convergence of some of the elements of the nets. A good example is the default compensation pay out cap which in most cases today at a similar level across different financial sectors in various

Economics and Business Review, Vol. 1(15), No. 3, 2015

jurisdictions. These may also lead to some distortions due to the effect of the regulatory interpretation Under current circumstances there is a clear danger of extending relevant banking safety net elements to other sectoral nets and the entire financial market. It may result in the replacement of standards which best account for the specificity of non banking financial institutions and non banking sectors. The safety nets of the insurance sector have their industry specific institutional and structural peculiarities. This reflects firstly the different risk profile of insurance companies compared to banks and other financial institutions. In deposit taking institutions up to 80% of their overall risk is represented by the credit risk whereas in insurance the major class of risk is market risk (40%) and insurance risk (30%) [SwissRe 2010: 6]. Moreover the latter results from the losses incurred and hence is unrelated to the business cycle in contrast to credit or market risk. Additionally insurance contracts are frequently over a long period with a long settlement time. It takes on average more than 10 years to settle claims on general and motor third party liability insurance. A substantial part of life contracts terminate only after twenty to forty years. An important characteristic of insurance compared to banking is the low exposure of insurers to liquidity risk which is an effect of the specificity of the insurance funding model. It makes the central bank in contrast with banking largely irrelevant for the safety of the insurance sector. Additionally insurers sectoral interconnectedness remains, unlike the banks, relatively low with no intense trading amongst individual agents which scales down sectoral contagion and the domino effect. Considering the design of the insurance safety net prior to the recent global financial crisis three principal building blocks could be identified worldwide: prudential regulation, public oversight and insurance guarantee systems (market oversight). On the face of it, it seems quite similar to the architecture existing in the banking sector. A major difference vis-a-vis banking is the absence of the central bank and its lender of last resort function, crucial at times of distress. Prudential regulations have come to the forefront of financial regulation quite recently – only at the beginning of the 90’s in the XXth century [Vittas 1991]. They regulate concurrently a growing area of insurance activities. They define amongst others the principles of undertaking insurance activities, their pursuit, the principles of their financing and sound management as well as the principles of the safe wind down and market exit. They are evolving over time in response to the evolution of the safety perspective and safety models. Prudential regulations are de facto impinging on owners’ competences and oversight. This is a reflection of the lack of reliability of the latter from the public point of view. The less trusted owners’ oversight is, the more public prudential regulations become necessary. This is well grounded theoretically in the agency theory and potential conflicts amongst the owners (principals)

J. Monkiewicz, L. Gąsiorkiewicz, M. Monkiewicz, The changing architecture of the safety net

and management (agents). Additionally it is reinforced by trends in ownership structure and nature which becomes increasingly diluted and concentrated on short term strategic goals. As a result instead of strong owners’ oversight increasingly “quasi owners’ oversight“ emerges. It is concentrated in the hands of management and thus relieved of the owners’ incentive structures. This potential for conflict between agents and principals is particularly big in insurance due to the complex insurance finance and business model and the long transaction settlement time. Both of these factors increase the danger of manipulation in financial reporting and overall performance and helps by hiding the real situation from the stakeholders. Hence we come today to the situation in which owners’ oversight is performed in principle on the basis of binding public prudential standards. The next major pillar of the insurance safety net is public prudential supervision. This pillar is essentially responsible for the daily monitoring of insurance companies, taking remedial action and ensuring their adequate compliance with regulatory rules and principles. The theoretical basis for the existence of public supervision in insurance is formulated in the theory of representation by Dewatripont and Tirole [1994] and developed further by Plantin and Rochet [2007]. According to this theory management of financial intermediaries such as banks and insurers which finance themselves by debt issuance to their customers are under pressure from their shareholders to take risky actions in order to accomplish extraordinary profits. This is fully rational from their perspective as equity (shareholder capital) in these institutions represents only small part of their overall financing. The leverage ratio (i.e. assets to equity) which illustrates this phenomenon is nowadays on average 10 for commercial banks and 3(P/C)-10(Life) for insurance companies [SwissRe 2010: 6]. Possible losses for the shareholders therefore are small and shared to large extent with their creditors: depositors and policyholders. On the other hand eventual extraordinary gains become fully appropriated by the shareholders. This natural moral hazard cannot be tempered by their creditors in the case of these institutions. Both dispersed depositors and policyholders have neither adequate technical knowledge nor the necessary information and competences to perform creditors’ oversight. They are therefore neither in a position to control their principals nor their agents (management). This role is essentially taken by prudential supervision which, according to the theory, becomes a trustee of these small lenders. Apart from the protection of the individual policyholders prudential supervisors have been frequently tasked with other duties including a contribution to market integrity, its efficiency and financial stability. The dominant supervisory model prior to the crisis was the one focusing on the financial safety of the individual insurance companies and their risk management systems that was believed to create a basis for this safety. The third and the last pillar of the safety net in insurance prior to the crisis was composed of insurance guarantee schemes exercising a kind of a market

Economics and Business Review, Vol. 1(15), No. 3, 2015

supervision. They are an element of crisis management in case of liquidity or solvency failure of the insurance company. They have been in most cases special purpose funds created and financed by insurance companies at the public request. They may be pre-funded before the failure happens or post funded when need arises after the failure. They become active once specified criteria are met. These specified criteria are normally either default or loss of liquidity. This is not a rare phenomena. In 1988–2008 in the US there were on average 33 insurers’ insolvencies annually in the non life and 21 insolvencies in the life sector. In contrast to banking they are relatively recent innovations applied to insurance. They came in existence in the 70’s and 80’s during the last century in the US and subsequently spread to other countries. Since these institutions are most often privately financed and managed by insurance industry players they may have a general inclination to discipline insurance companies and control their risk policies and behaviour to minimise collective expenditures in case of default. These collective guarantee schemes may, in case of insurance industry, similarly to banking, play two different roles – pay box or risk minimiser. The essence of the pay box role is to pay out to eligible persons a guaranteed compensation amount notwithstanding the available resources of the failed company. The difference between available assets and liabilities is financed collectively by the remaining healthy insurance companies in proportion to the established criteria, most often the premium income. The kind of eligible persons and size of the compensation may vary depending on the specific rules adopted in a given jurisdiction. Some policyholders such as large corporations or managers of the failed entity and large shareholders are often excluded from the compensation offer. There are also frequently maximum limits of compensation offered. Risk minimising role is an innovative and more complex function of insurance guarantee schemes. It is also much less popular with in the schemes. Its essence is to mitigate the default risk to the insured and to mitigate the adverse effects of the materialisation of default risk. The guarantee schemes in this role focus their attention on the prevention of failure by monitoring the risk taking by their sponsors and offering some financial assistance to overcome transitional difficulties if the need arrives. Additionally guarantee schemes may offer one off solutions or be involved in finding portfolio acquirers to protect the interests of the insured by continuing existing insurance contracts. In their role as risk minimisers guarantee schemes approach the activities normally restricted to supervisors and become important partners. There are good arguments for having assigned this role to guarantee schemes but there are also weighty counter arguments. The major bonus offered is increased flexibility in selecting the best possible solution to a specific problem. The most important malus is the additional provision of the moral hazard incentive

J. Monkiewicz, L. Gąsiorkiewicz, M. Monkiewicz, The changing architecture of the safety net

3. Safety net arrangements in insurance – post-crisis As indicated before the recent global financial crisis has produced an ideological shift in the existing regulatory and supervisory paradigm. The micro prudential regulation and supervision dominated the scene in the pre crisis era and focused primarily on the safety of individual insurance companies has been considered ineffective and inadequate. In the aftermath of the financial crisis a need for a different perspective – a macroprudential supervision – was recognized and generally accepted [Nier et al. 2011; Osiński, Seal and Hoogduin 2013; Houben 2013] though its application to the non monetary sectors is still poorly developed. Its focus is on a market wide perspective and the safety of all market participants. Its role is the mitigation of systemic risk and the maintenance of financial stability [IAIS 2013]. Therefore its special task is detecting financial market interlinkages, identifying common exposures of insurance companies and the possible contagion effect that may be of relevance. As a result the emergence of the two pillar supervisory system in insurance as is the case in banking adds to the complexity of supervisory arrangements (see Table 2). Table 2. Micro- and macroprudential supervision in comparison Features

Microprudential supervision

Macroprudential supervision

Principal goal

Protection against default of individual institutions

Protection of the stability of the entire financial sector

Final goal

Protection of the customers and investors

Limiting macroeconomic costs of financial crises

Supervisory perimeter

Individual financial institutions

Financial system as a whole

Interlinkages and common exposures

Of little importance

Very important

Tools

Applied to individual entity

Applied to all or some specific groups of entities

Source: [Szpunar 2012].

These two pillars evidently need to cooperate and reinforce each other, however there may be areas where their perspectives differ and decisions are not so obviously arrived at. Take the case where the microprudential supervisor urges individual institutions to improve their balance structures and produces similar balance profiles in all entities. Their common exposures as a result increases, which is precisely what the macroprudential supervisor would like to avoid. This fallacy of the composition effect is the most relevant but not the only conflicting issue amongst the two supervisory perspectives.

Economics and Business Review, Vol. 1(15), No. 3, 2015

The leading role in this new supervisory pillar is given to central banks because of their involvement hitherto in financial stability issues and the vast analytical resources that they possess. This is in itself an additional concern of the insurance industry which suspects a lack of insurance related knowledge in this new supervisory agent. Additionally two more interesting and important developments with regard to prudential supervision in the aftermath of the recent financial crisis arise. This refers to the fact that supervision has begun to apply increasingly prospective supervisory tools in particular scenario analyses and prudential stress tests. This reinforces the risk preventive function of supervisors which may react well in advance of possible failures. Also supervisory authorities have begun, more often than before, to apply discretionary powers and a principles based approach. It also received many more tasks associated primarily with the development of consolidated or group wide supervision. Prudential regulations in the insurance sector have been, until recently, addressed as a matter of principle to all insurance companies only with some exceptions in the case of mutuals and captives. With the designation of an initial list of Global Systemically Important Insurers (G-SIIs) by FSB on July 18th, 2013 a new layer of prudential regulation, based on the adds on principle, is emerging. The same is true with regard to banking which set the scene and seems to be the master cook (see Table 3). The said G-SIIs, including in 2013, Allianz SE, AIG, Assicurazioni Generali S.p.A, Aviva plc, AXA S.A., MetLife Inc., Ping An Insurance (Group) Company of China, Ltd., Prudential Financial, Inc. and Prudential plc., and all subsequently designated entities will be subject to a range of additional regulations. They include the recovery and resolution planning requirements as defined by FSB’s Key Attributes of Effective Resolution Regimes, including the requirement for the establishment for each entity of a Crisis Management Group (CMG) and the setting up of a recovery and resolution plan (RRP), enhanced group-wide supervision, including a group wide supervisor to oversee the development and implementation of a Systemic Risk Management Plan and higher loss absorbency requirements (HLA). Of the measures outlined enhanced supervision was with immediate effect, the Crisis Management Groups should have been established by July 2014 and recovery and resolution plans, including a liquidity risk management plan should have been ready by the end of 2014. Implementation details for the HLA are to be finalized by 2015 and to be applied in 2019 by all G-SIIs identified in November 2017. Before that the IAIS is supposed to work out the “ordinary” loss absorbency capacity for the insurance world to have been ready by the G20 Summit in 2014. In 2014 FSB will have additionally designated G-SIIs and appropriate risk mitigating measures for major reinsurers. As in the case for banking, national jurisdictions will be expected to designate important insurers in their domestic market and to assign proper risk mitigating measures. This may add to the ex-

J. Monkiewicz, L. Gąsiorkiewicz, M. Monkiewicz, The changing architecture of the safety net

Table 3. FSB framework for systemic banks and insurers in comparison Framework for banks

Framework for insurers

First designation date

November 2011

July 2013

No of institutions

Overall justification

Size, global activity, interconnectedness, complexity, substitutability

Size, global activities, interconnectedness, non traditional and non insurance activities, substitutability

Measures: ––enhanced supervision More intense and effective supervision, including stronger supervisory mandates, resources and powers ––effective resolution planning ––higher loss absorbency (HLA)

Timeline

More intense and effective supervision with direct regulatory powers over holding companies, and oversight of the Systemic Risk Management Plan (SRMP) Establishment of recovery and resolution plans (RRP) including liquidity management plans

Establishment of recovery and resolution plans (RRP) including liquidity risk management plans Capital surcharge ranging from Capital surcharge to be de1 to 3.5% of risk weighted assets veloped with Basic Capital Requirement (BCR) and HLA for total balance or some activities ––Enhanced supervision: FSB ––Enhanced supervision: SRMP framework in 2010 as of mid 2014 ––Effective resolution: resolution ––Effective resolution: recovery planning requirement by 2012 and resolution plans by end ––Capital surcharge: in 2016– 2014 2019 ––Capital surcharge: BCR as of 2015, HLA as of 2016

Source: [Thimann 2014: 6].

isting batch of the prudential regulations across the globe [IAIS 2013]. This may additionally create new institutions complementing the existing safety net. This may be the case with regard to the resolution authorities which may be necessary to mitigate risk emanating from systemically important insurers – global or domestic. The recovery and the resolution arrangements initially developed by the FSB post crisis for the banking sector seem to have finally found their way into insurance and form another innovation in the insurance sector safety net. It is based on the assumption that run-off and portfolio transfer tools traditionally used to resolve financial failures of the insurance companies may not be sufficient to mitigate the systemic impact of a large, complex insurance group. In principle, according to FSB recommendations, all insurers that

Economics and Business Review, Vol. 1(15), No. 3, 2015

could be systemically significant if they fail should be subject to a resolution regime consistent with the “Key attributes to effective resolution regimes for financial institutions”. This of includes all G-SIIs at a minimum [FSB 2014a]. As in the case of banking their implementation is supposed to govern the insurance institutions in an orderly manner without taxpayers’ exposure to loss from solvency support. An effective resolution regime should include [FSB 2014a] “stabilisation options” which provide for the continuity of systemically important functions and “ liquidation options” that provide the mechanism for the orderly closure and wind down of all or parts of the business whilst protecting the insurance policyholders. Such a regime should meet at least the following criteria: “– ensure continuity of systemically important financial services, “– protect insurance policyholders, allocate losses to shareholders and uninsured creditors, “– seek to minimise the overall costs of resolution in home and host jurisdictions, “– enhance market discipline and provide incentives for market based solutions “– not to rely on public solvency support” [FSB 2014a]. To achieve its goals and objectives the resolution authority should coordinate, on the one hand with the relevant policyholder protection schemes and on the other with the relevant supervisory authority. Table 4. Actions and timelines of G-SIIs recovery and resolution planning Action

Responsible

Finalise guidance on the identification of criti- FSB with participation cal functions and critical shared services in the of IAIS insurance sector

Completed by Mid 2015

Report on the status of resolution strategies G-SII Crises Management End 2015 and plans for all G-SIIs and possible challenges Groups (CMG) Develop a proposal for draft guidance on the development of effective resolution strategies for G-SII

FSB with participation of IAIS

End 2015

Report on results of Resolvability Assessment Process(RAP)

G-SII Crises Management End 2016 Groups

Source: [FSB 2014b: 16].

FSB expects that all insurers that could be systemically significant will be subject to regular resolvability assessments and the whole process is carefully planned (see table 4). Under such assessments resolution authorities should assess whether the resolution strategy and operational resolution plans ensure the continuity of critical functions by the insurer concerned without negative externalities. The FSB acting together with IAIS has developed guidance to

J. Monkiewicz, L. Gąsiorkiewicz, M. Monkiewicz, The changing architecture of the safety net

assist authorities in their evaluation of the importance of the functions that G-SIIs provide to financial markets and the real economy. Additionally all insurers that could be systemically important should be subject to ongoing process of recovery and resolution planning. The Recovery and Resolution Plans (RRP) devised should be reviewed and assessed by the microprudential supervisory authorities in cooperation with policyholders protection schemes as well as with the resolution authorities concerned. FSB recommends granting resolution authorities extremely strong mandates and powers. The resolution authority should have inter alia the power to restructure, limit or write down all liabilities of the distressed insurer, including insurance and reinsurance. It could inter alia terminate future benefits and guarantees, reduce the value of contracts upon surrender, terminate or restructure options provided to policyholders, reduce the value or restructuring reinsurance contracts and the like [FSB 2014a]. The suspension of policyholders rights in resolution could continue until the temporary suspension or withdrawal from their insurance contracts with an insurer. It should also have power directly or indirectly over the insurer in resolution. These global developments are subsequently to be introduced in individual jurisdictions. Currently at EU level the existing EU directive on the reorganisation and winding up of insurance undertakings assumes that each individual member state is responsible for its own procedures in this regard. To align however with G20 and FSB recommendations the EU Commission has prepared a draft legislatory proposal in the relevant EU framework. In December 2013 the EU Parliament adopted a report on the said framework in which FSB recommendations were largely supported and thus made the Commission responsible. Finally, apart from the changes highlighted before, we should also mention the setting up on May 15th, 2013 of The International Forum of Insurance Guarantee Schemes with the intention of facilitating the sharing of experience of the leading insurance schemes in providing protection to policyholders in the event of the failure of an insurance company. Currently IFIGS membership includes Australia, Canada, Taiwan, France, Germany, Greece, Korea, Malaysia, Norway, Poland, Romania, Singapore, Spain, UK and the United States of America. Undoubtedly they will soon start to deal also with crossborder insolvency claims. It will result in a new role for the IGS in insurance relevant safety nets.

Conclusions As our analysis indicates the safety net of the current insurance market and its operators is a complex matter which has multiple determinants and which tends to evolve over time reflecting current values and beliefs. It is built from many interrelated elements which must be properly balanced and coordinated. It is

Economics and Business Review, Vol. 1(15), No. 3, 2015

a part of the broader category of financial system safety and its safety net with which many interconnections exist. This existence also allows a possibility for the appearance of negative externalities i. e. contagion effect. In the course of the recent financial crisis existing safety net arrangements have been severely tested and new life has been injected into the concept of the safety net. As a result new concepts and ideas in relation to the safety net have arisen. The most important of these new elements include: –– creation of a new supervisory pillar – macroprudential supervision – for the benefit of financial stability and to mitigate systemic risk, –– differentiation of the spectrum of available regulatory standards and the creation of special regulatory regimes for systemically important institutions, –– creation of special rules for the purpose of the restructuring and the resolution of systemically important institutions, –– creation of systemic crisis back stops within government structures, –– assigning a special role to group wide regulation and supervision. Many of these innovative ideas are borrowed from the banking sector and cannot be easily implanted into insurance with its specific business model. All of these are taking the insurance industry into uncharted waters and require proper responses both from the regulatory, supervisory and business communities. It also requires much additional empirical research.

References Baker, A., 2013, The New Political Economy of the Macroprudential Ideational Shift, New Political Economy, no 18(1): 112–139. Claessens, S., Kodres, L., 2014, The Regulatory Responses to the Global Financial Crisis: Some Uncomfortable Questions, IMF Working Paper, March. Dewatripont, M., Tirole, J., 1994 The Prudential Regulation of Banks, MIT Press, Cambridge, Mass. Eatwell, J., 2009, Practical Proposals for Regulatory Reform, in: Subacchi, P., Monsarrat, R. (eds.), New Ideas for the London Summit: Recommendations to the G20 Leaders, Royal Institute for International Affairs, The Atlantic Council, Chatham: 11–15. FSB, 2014a, Key Attributes of Effective Resolution Regimes for Financial Institutions, 15 October. FSB, 2014b, Towards Full Implementation of the Key Attributes of Effective Resolution Regimes for Financial Institutions, Report to The G20 on Progress in Reform of Resolution Planning for Globally Systemically Important Financial Institutions (G-SISIs), 12 November. Helleiner, E., 2010, A Bretton Wood Moment? The 2007–2008 Crisis and the Future of Global Finance, International Affairs, May: 619–636. Houben, A., 2013, Aligning Macro- and Microprudential Supervision, in: Kellerman, J.A. et al. (eds.), Financial Supervision in the 21st Century, Springer, Berlin: 201–220. IAIS, 2013, Global Systemically Important Insurers: Policy Measures, July.

J. Monkiewicz, L. Gąsiorkiewicz, M. Monkiewicz, The changing architecture of the safety net

Monkiewicz, M., 2013, Bezpieczeństwo rynku ubezpieczeniowego UE a systemy gwarancyjne pewności ochrony ubezpieczeniowej. Teoria i praktyka, Poltext, Warszawa. Monkiewicz, J., Małecki, M. (eds.), 2014, Macroprudential Supervision in Insurance. Theoretical and Practical Aspects, Palgrave Mcmillan, London. Nier, E.W., Osiński, J., Jacome, L.I., Madrid, P., 2011, Institutional Models for Macroprudential Policies, IMF Staff Discussion Note, I/18. Osiński, J., Seal, K., Hoogduin, L., 2013, Macroprudential and Microprudential Policies: Toward Cohabitation, IMF Staff Discussion Note, June. Plantin, G., Rochet, J.Ch., 2007, When Insurers Go Bust. An Economic Analysis of the Role and Design of Prudential Regulation, Princeton University Press. SwissRe, 2010, Regulatory Issues in Insurance, Sigma, no. 3. Solarz, J.K., 2008, Zarządzanie ryzykiem systemu finansowego, PWN, Warszawa. Szpunar, P., 2012, Rola polityki makroostrożnościowej w zapobieganiu kryzysom finansowym, NBP, Materiały i Studia, nr 278, Warszawa. Thimann, Ch., 2014, How Insurers Differ from Banks: A Primer for Systemic Regulation, July, Axa Group and Paris School of Economics: 1–23. Vittas, D., 1991, The Impact of Regulation on Financial Intermediation, The World Bank, WPS, August.

Economics and Business Review, Vol. 1 (15), No. 3, 2015: 20–37 DOI: 10.18559/ebr.2015.3.3

The determinants of non-life insurance penetration in selected countries from South Eastern Europe1 Klime Poposki2, Jordan Kjosevski3, Zoran Stojanovski4

Abstract : This study examines the determinants of non-life insurance penetration in 8 countries from South Eastern Europe (SEE), during the period 1995–2011, applying a panel vector error correction model (PVECM). This model will help us to identify the most important determinants of non-life insurance penetration in selected SEE countries. As a measure for non-life insurance demand we used non-life insurance penetration. Empirical results provided the evidence that the number of passenger cars per 1,000 people, GDP per capita and rule of law positively and significantly influence the non-life insurance penetration. The results also indicate that when the non-life insurance penetration deviates from its long-run equilibrium the speed of adjustment will subsequently bring it back to the equilibrium level, which in our case will take almost 1 year. Keywords : non-life insurance penetration, South Eastern Europe, PVECM. JEL codes : C39; G22; O16.

Introduction The non-life insurance markets in almost all transition countries in Central and Eastern Europe started to grow rapidly in 1990’s due to improved economic conditions and introduced reforms, which had to be conducted prior to EU entry. By introducing risk pooling and reducing the impact of large losses on the corporate sector and households, the insurance industry reduces the amount of capital that would be needed to cover these losses individually, encouraging additional output, investment, innovation and competition. Furthermore, using risk-based pricing for insurance protection, the insurance industry can change the behaviour of economic agents, contributing to the prevention of accidents, improved health outcomes and efficiency gains. Finally, insurance can also im 1

Article received 23 January 2015, accepted 3 August 2015. University St. Kliment Ohridski, 1 Maj 66, Bitola, Republic of Macedonia; Insurance Supervision Agency of the Republic of Macedonia; corresponding author: klime.poposki@aso.mk. 3 Independent researcher. 4 Insurance Supervision Agency of the Republic of Macedonia. 2

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

prove the efficiency of other segments of the financial sector, such as banking and financial markets (e.g., by enhancing the value of collateral through property insurance and reducing losses at default through credit guarantees and enhancements). Nevertheless this growth did not rise evenly. For instance in 1999 the non-life insurance penetration in SEE countries was 2.97% and it reached 3.20% in 2011, whilst in Central and Eastern Europe it was 1.24% in 1999 and increased to 2.03% in 2011. The large disparity across countries in the use of non-life insurance raises questions about what causes this variation and, thus, what determines non-life insurance penetration. Some authors have proposed a variety of different socio-economic and institutional factors as possible determinants of non-life insurance penetration.The contribution of this paper is to understand what drives the non-life insurance consumption within a sample of 8 countries from SEE (Albania, Belarus, Bosnia and Hercegovina, Croatia, Macedonia, Moldova Serbia and Ukraine) for the period 1995–2011. As a measurers of non-life insurance demand we will follow Feyen, Lester, and Rocha [2011] and will use non-life insurance penetration (non-life insurance premiums in relation to GDP)5. We apply the Kao panel cointegration test and panel vector error correction model to estimate the relationship between the variables. The paper is organized as follows. Section 2 highlights literature on theoretical research and empirical findings relevant to the demand for non-life insurance. Section 3 presents methodology and data which we incorporate in the analysis. The results of the empirical research are given in Section 4. The paper finishes with some concluding remarks and suggestions for the future work that are outlined in section 5.

1. Literature review In this section we present the theoretical research and highlight the most relevant findings. The theoretical frameworks are usually followed by the empirical investigation of the developed models. Then we proceed to the empirical studies which for the most part evaluate the impact on non-life insurance demand in and across particular countries.

1.1. Theoretical studies Theoretical models of non-life insurance demand, starting from the seminal papers of [Pratt 1964; Arrow 1971; Mossin 1968], predict that for a given level 5

Penetration indicates the level of development of insurance sector in a country. Penetration is measured as the ratio of premium underwritten in a particular year to GDP. Within insurance there is life insurance penetration which considers premiums from life insurance policies only as a percentage of GDP and non life insurance penetration which considers premiums from other than life insurance policies such as auto insurance, health insurance, etc.

Economics and Business Review, Vol. 1(15), No. 3, 2015

of risk exposure and a given price, insurance demand is increasing with risk aversion, probability of loss and total wealth [Sweeney and Beard 1992; Szpiro 1985]. Whether the propensity to insure – i.e., the desired coverage as a percentage of the wealth at stake – should increase or not, depends on the behaviour of risk aversion: Arrow [1971] shows that it increases if people are characterized by increasing relative risk aversion. Most of the above authors have commented on the elasticity of insurance consumption with respect to income and wealth in the light of the long-standing debate on insurance as an inferior good. Mossin [1968] first delineated the conditions for this to happen: the intuition is that if the utility function is characterized by decreasing absolute risk aversion, then a higher endowment of wealth reduces risk aversion and therefore the demand for insurance). Moreover whilst by Mossin’s Theorem full coverage is optimal under the fair actuarial price, the degree of coverage decreases with the loadings – Schlesinger [2000]. The so-called “inverted economic cycle” of insurance in which one pays first then, in the event of loss, receives his dues, suggests that the financial rate of return, seen as an opportunity cost for those who allocate funds in an insurance policy, should be inversely related to demand. That is self-insuring gives an opportunity-gain to invest the amount of the premium saved on financial markets, which increases along with the prevailing rate of return. However, Falciglia [1980] shows that higher market interest rates should lower insurance demand only if consumers have a decreasing risk aversion and are net savers; although these conditions seem reasonable, the relationship between interest rates and insurance demand nevertheless remains an empirical question.

1.2. Review of the empirical evidence Despite the critical role that the insurance sector plays for financial and economic development and reasonable evidence that the sector has promoted economic growth, there have been few studies examining the factors that drive the development of the insurance sector. Moreover the bulk of the existing empirical research focuses on the growth of the life insurance sector, using the most frequently cited papers [Beck and Webb 2003; Browne and Kim 1993; Outreville 1996; Li et al. 2007]. The dependent variables for the vast majority of models was the life insurance density (number of US Dollars spent annually on life insurance per capita) and the life insurance penetration (total life premium volume divided by GDP). Explanatory variables that have been shown to significantly impact life insurance demand are GDP per capita, inflation (real, anticipated or feared), development of the banking sector, institutional indicators (such as investor protection, contract enforcement, and political stability). Variables that appear to have a borderline impact include education, old and/or young dependency ratio (ratio of the population above the age of 65, or below 15, to

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

the number of persons aged 15 to 64), urbanization, size of the social security system, life expectancy, and market structure. Sherden [1984] was first to focus on the sensitivity of non life insurance purchase. In a cross-sectional analysis of consumption patterns limited to automobile insurance in 359 townships in the state of Massachusetts in 1979, Sherden [1984] finds that the demand for motor insurance is generally inelastic with respect to price and income and that the demand for comprehensive and collision coverage increases substantially with increased population density. Beenstock, Dickinson, and Khajuria [1988] using an international dataset (12 countries over a period of 12 years) to examine the relationship between property liability insurance premiums and income, found that marginal propensity to insure i.e., increase in insurance spending when income rises by 1$, differs from country to country and premiums vary directly with real rates of interest. Again the decision of consumer and his/her initial wealth status are significant factors also when shortrun or longrun consumption of insurance is considered. Based on a cross-sectional logarithmic model of non-life insurance penetration of 55 developing countries, Outreville [1990] confirms the Beenstock, Dickinson, and Khajuria [1988] main result of an income elasticity greater than unity. The level of financial development is the only other factor found to significantly impact non-life insurance consumption. Browne, Chung, and Frees [2000] study 22 OECD countries from 1987 through 1993 and focus on the premium density of two lines of insurance: motor vehicle (usually purchased by households) and general liability (normally bought by businesses). Panel data analysis demonstrates that income (GDP per capita), wealth, foreign firms’ market share, and the form of legal system (civil law or common law) are significant factors to explain the purchase of the two types of insurance. Per capita income has a much greater impact on motor insurance than on general liability. Esho, Kirievsky, and Zurbruegg [2004] expand the work of Browne, Chung, and Frees [2000] by using a larger set of countries and by introducing the origin of the legal system and a measure of property rights in their model. Dummy variables, characterizing the English, French, German, and Scandinavian legal systems’ origins, are found to have an insignificant effect. Results show a robust relationship between the protection of property rights and insurance consumption as well as a significant effect of loss probability and income. Esho et al. [2004] also include one of Hofstede’s dimensions, Uncertainty Avoidance, as a proxy for risk aversion. They find a marginally positive relationship and conclude that culture does not seem to play an important role in non-life insurance demand. Based on a analysis of 5 countries (Bosnia and Herzegovina, Croatia, Macedonia, Serbia and Slovenia) Njegomir, Stojić, and Marković [2011] analysed the performance in the non-life insurance industry for the period

Economics and Business Review, Vol. 1(15), No. 3, 2015

2004–2008. They used three models for capturing influences of market structure and liberalisation on market profitability. Firstly, market structure, liberalisation and performance are put in relation to the strength of economy and corresponding rate of return, model 2 connects the former with the number of competitors and their dominant line of insurance, whilst in model 3 they used the threat of substitutes as a control variable. The research results of all three models show support for the S-C-P hypothesis. Their results are important for governments that wish to achieve affordable and available insurance for all. Governments interfere in insurance markets by pro-competitive and pro-liberalising policies. Their research results could provide insurance companies with a useful comparison across different national markets throughout the ex-Yugoslavia region, thus enabling them to formulate optimal competitive strategies. The research of Njegomir and Stojić [2012] examines factors that affect the attractiveness of the Eastern European non-life insurance market for foreign insurers for the period 2004–2009. The region encompasses non-life insurance industries in 15 countries: Bosnia and Herzegovina, Croatia, Macedonia, Montenegro, Serbia, Slovenia, Hungary, Czech Republic, Slovakia, Romania, Bulgaria, Poland, Lithuania, Latvia and Estonia. The research results indicate that the main forces affecting market attractiveness are insurance demand, entry barriers, market concentration and the return on investment and only market concentration has negative impact. Poposki and Kjosevski [2013] used an international dataset (16 countries from Central and South-Eastern Europe (CSEE) over the period of 1998–2010 years) to identify determinants of the demand of non-life insurance. They used a fixed-effects panel model. As a measure for demand for non-life insurance they used non-life insurance penetration and non- life insurance density. Their results show that GDP per capita, number of passenger cars, gini coefficients, level of education and rule of law are the most robust predictors of the use of non-life insurance. Private credit, inflation, trade, population density, control of corruption and government effectiveness do not appear to be strongly associated with non-life insurance demand.

2. Data In our study we use an unbalanced panel for 8 countries from SEE (Albania, Belarus, Bosnia and Hercegovina, Croatia, Macedonia, Moldova Serbia and Ukraine), over the period 1995–2011. In order to obtain more information we used annual panel data. The choice of the time period in this paper was contingent upon the availability of data. Following a similar approach nearly every international comparative study uses insurance density and penetration as dependent variables. These variables

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

have the advantage of being easily available, annually, for a large number of countries. A disadvantage of density and penetration is that they combine premiums across various lines of insurance. In some countries motor insurance is the dominant line of business whilst in others the focus is on the liability line of insurance. Aggregate premiums result in a loss of information reducing the likelihood that significant explanatory variables will be discovered. Density and penetration measure slightly different effects. Penetration measures nonlife insurance consumption relative to the size of the economy, whil density compares non life insurance purchases across countries without adjusting for income. High GDP countries will spend more on insurance, in absolute terms, as they have more assets to protect. Therefore we expect a very high correlation between insurance density and GDP – indeed one of the reasons for the paucity of research in determinants of non-life insurance may have been a belief that purchases are driven by wealth and little else. Penetration measures relative insurance consumption, as the overall wealth effect has been removed through division by GDP per capita. It measures how wealth is allocated to insurance in relative terms: two countries with similar GDP per capita may exhibit different insurance consumption patterns, an effect captured by penetration and not by density. For this reason we use nonlife insurance penetration – NLIP to be our primary variable, and we do not use density in our research. Factors that we use as control variables, which may explain the consumption of non-life insurance, include the following: –– Economic: GDP per capita – GDPPC; number of passenger cars per 1,000 people – NPV; ratio of quasi-money – RQM; inflation annual percentage – INF; –– Demographic: population density – PD; level of education-EDU; –– Institutional: rule of law-RL. Table 1 shows the descriptive statistics for the variables used in our main regression. We observe a large variation in levels of non-life insurance penetration Table 1. Descriptive statistics NLIP

GDPPC

INF

NPV

RQM

EDU

Mean

1.467951 3,235.130 20.99215 146.8917 34.61879 85.98519 40.95094 –0.571447

Median

1.515000 2,294.356 8.200000 132.0000 28.80000 81.00000 39.40245 –0.571275

Maximum

4.770000 15,889.35 415.8000 372.0000 276.0000 127.0000 85.69712 0.534200

Minimum

0.330000 321.0268 –1.700000 19.00000 –8.300000 47.00000 9.091730 –1.935360

Std. Dev.

0.867376 3,022.882 49.02067 79.55874 37.10290 22.47021 17.87228 0.437886

Jarque-Bera

9.013120 247.3972 9639.323 14.05565 1969.573 3.216263 4.835785 0.053405

Probability

0.011036 0.000000 0.000000 0.000887 0.000000 0.200261 0.089109 0.973651

Observations

122

134

135

120

132

135

113

110

Economics and Business Review, Vol. 1(15), No. 3, 2015

across countries, from 0.33 to 4.77 of GDP. We also present the Jarque-Bera (JB) test of normality distribution. From this we can conclude that just two of the variables do not satisfy the assumption for normal distribution (GDP and INF). Data are obtained from various sources. Non-life insurance penetration is obtained from Sigma, Swiss Re Economic Research & Consulting, Swiss Re, Zurich and national insurance associations. Education is obtained from EdStats, World Bank. GDP per capita, inflation, number of passenger cars per 1,000 people, ratio of quasi-money, trade and population density are obtained from the World Development Indicators (WDI) database. Rule of law is obtained from the Worldwide Governance Indicators.

2.1. Economic factors All previous studies, whether on life or non-life insurance, conclude that income, measured as GDP per capita, is the most important factor affecting purchasing decisions [Fortune 1973; Campbell 1980; Beenstock, Dickinson, and Khajuria 1986; Lewis 1989; Outreville 1990]. Beck and Webb [2003], Ward and Zurbruegg [2000], Beenstock, Dickinson, and Khajuria [1988], point out a positive relationship in industrialized countries between national income and non-life insurance spending. Browne, Chung, and Frees [2000], analyzed general liability and motor vehicle insurance in OECD countries and found a significantly positive relationship between premium density and GNP per capita. Additionally [Esho, Kirievsky, and Zurbruegg 2004] examined developed and developing countries between 1984 and 1998 and found a strong positive relationship between national income and the nonlife insurance premium. Outreville [1990] and Ward and Zurbruegg [2000] strongly emphasized that the insurance industry, through risk transfer, financial intermediation and employment can generate externalities and economic growth. The higher level of income creates a greater demand for non-life insurance to safeguard acquired property. We expect income to have a strong, positive impact on non-life insurance consumption. We include the number of passenger cars per 1,000 inhabitants because most countries require mandatory third party liability insurance (comprehensive car insurance is usually voluntary but also common in many countries). Financial development is associated with the widespread securitization of cash flows, which enables households to secure future income through the ownership of financial assets. By offering similar benefits, life insurance is expected to generate higher sales in countries with a high level of financial development. The measurement of financial development is very controversial [Jung 1986], but two alternative proxies are usually employed. One is the ratio of quasi-money (M2-M1) to the broad definition of money (M2) – it shows the complexity of the financial structure (a higher ratio indicates a higher level of financial development) and another is the ratio of M2 to the nominal GDP – financial

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

deepening (demand for money per unit of output). Broad money M2 is often taken as an adequate measure of the financial sector in developing countries in view of the predominance of the banking sector due to the lack of data on other financial assets [Hemming and Manson 1988; Liu and Woo 1994]. Following the previous studies we use the ratio of quasi-money (M2-M1) as a measure of financial development. We hypothesize a positive correlation with non-life insurance demand. The next economic variable that we used in our research is the inflation rate. It is used to account for monetary discipline. It is expressed by the GDP deflator (annual percentage). For non-life insurers unanticipated inflation leads to higher claims costs, thereby eroding profitability. Inflation is often accompanied by rising interest rates, which reduce the value of guarantees of return. Rising inflation can have a negative effect on demand and may lead to policyholders cancelling their policies as well as increasing costs for insurers. In the case of deflation, or if very low inflation persists, interest rates tend to fall. With this variable we expect a negative correlation with non-life insurance consumption.

2.2. Demographic factors Feyen, Lester, and Rocha [2011] explained that the size of population determines the operating background, that is to say, the size of market, for the nonlife insurance industry. We, therefore, include the population density (people per sq. km. of land area) for each country into our regressions and assume that its effect on the non-life insurance consumption is positive. A primary determinant for purchasing insurance is to minimise the damage from an adverse event. Unfortunately measuring attitudes to risk is difficult and in the past most insurance studies have used education to proxy risk aversion. Schlesinger [1981], demonstrates that an individual with a higher loss probability, a higher degree of risk aversion, or a lower level of initial wealth, will purchase more insurance. According to the discussion of Browne and Kim [1993], in general a higher level of education may lead to a greater degree of risk aversion and greater awareness of the necessity of insurance. Nonetheless Szpiro [1985] proved the negative correlation between the level of education and risk aversion. It was deemed that higher education leads to lower risk aversion, and that, in turn, leads to more risk-taking by skilled and well-educated people. When [Browne, Chung, and Frees 2000; Esho, Kirievsky, and Zurbruegg 2004] were discussing non-life insurance; they also took the level of education as a proxy for risk aversion. Therefore education is hypothesized to be ambiguous in relation to non-life insurance demand. As an indicator of the level of education across countries we use the tertiary gross enrollment ratio defined by the UNESCO Institute of Statistics as the total enrolment in tertiary education, regardless of age, expressed as a proportion of the eligible school-age population.

Economics and Business Review, Vol. 1(15), No. 3, 2015

2.3. Institutional factors Legal stability is important for a vibrant and growing non- life insurance market. The more stable the legal system in the country the higher the willingness of contracting parties to initiate business relationships. To measure property rights’ protection we use the rule of law index provided by the The Worldwide Governance Indicators. This index reflects perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular, the quality of contract enforcement, property rights, the police and the courts, as well as the likelihood of crime and violence. The legal system in force in a country may impact the development of insurance as it specifies the liabilities of those responsible for damage, and defines the business environment of insurers [Browne, Chung, and Frees 2000]. The United States is the world leader in per capita consumption of liability insurance. The American legal system may be a contributing factor by encouraging Americans to over consume property-liability insurance [Syverud et al. 1994]. Browne, Chung, and Frees [2000] find the legal system to be a significant factor in the development of non-life insurance. Esho, Kirievsky, and Zurbruegg [2004] also investigate the impact of the legal system but find it insignificant after checking income and property rights. Recently Park, Lemaire, and Chua [2010] showed that the use of a Common Law legal system is the most important determinant of toughness of bonus-malus systems in automobile insurance. Therefore it is hypothesized that there is a positive relationship with nonlife insurance consumption.

3. Methodology Given the hypotheses specified above we employ co-integration and error correction techniques to capture the long-run relationship and short-run dynamics between the dependent and independent variables. We specify the model for the determinants of non-life insurance penetration (NLIP) in the Western Balkans with an expected sign for each variable, as follows: NLIP = f(GDPPC(+), NPC(+), RQM(+), INF(–), PD(+), EDU(+), RL(+).

(1)

The most common specification is the log-linear form used by [Outreville 1996; Browne and Kim 1993; Feyen, Lester, and Rocha 2011]. The log-linear form is used for demand functions specified on macroeconomic variables which tend to display exponential growth. The above model is hereby written in log-linear form as:

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

L(non-life insurance penetration)it = β0 + β1L(GDP per capita)it + β2L(number of passenger cars per 1,000 people)it + β3(ratio of quasimoney)it + β4(inflation)it + β5L(population density)it + β6L(education (2) level)it + β7(rule of law)it + uit, where: β – a coefficient that should be an estimate, uit – a scalar disturbance term, i – indexes a country in a cross section, t – indexes time measured in years. Based on the established model we will estimate the determinants that affect the demand for non-life insurance in the SEE countries

3.1. Panel unit root test To formulate an econometrics model it is important to know whether the data generating process of variables is based on a stationary process or not. In the presence of non-stationary properties of standard estimation are not valid. In addition it might cause problem of spurious regression Verbeek [2004]. To avoid the problem which may arise because of the existence of non stationary variables one might have to identify the order of integration of variables. Although several methods have been proposed by considering different assumptions there is no uniformly powerful test for unit root. However, it has been widely acknowledged that standard unit root tests can have a low value against stationary alternatives for important cases [Campbell and Perron 1991]. As an alternative the recently developed panel unit root is applied. In this paper, we test for stationarity of the panel using a Maddala and Wu panel unit root test for unbalanced panels. Maddala and Wu [1999] proposed a Fisher-type test which combines the p-values from unit root tests for each cross-section i. The test is non-parametric and has a chi-square distribution with 2n degrees of freedom where n is the number of countries in the panel. They state that not only does this test perform best compared to other tests for unit roots in panel data but it also has the advantage that it does not require a balanced panel, as do most tests.

3.2. Panel cointegration test The concept of cointegration has been widely used in literature to test the presence of long-run relationships amongst variables. Similar to individual unit root tests, cointegration tests in time series literature suffer from low value when the time horizon is short. Panel techniques may be better in detecting cointegration relationships since a pooled levels regression combines cross-sectional and time series information in the data when estimating cointegrating coefficients.

Economics and Business Review, Vol. 1(15), No. 3, 2015

Kao [1999], proposed panel cointegration tests similar to the Engle and Granger [1987] framework which include testing the stationarity of the residuals from a levels regression.

3.3. Panel vector error correction model According to Engle and Granger [1987] if two series are cointegrated they can be characterized as being generated by an error correction mechanism. However the presence of a cointegration relationship cannot explain the direction of causality among the variables. In order to analyze the direction of causality, a panel-based vector error correction model (PVECM) should be performed. The PVECM is a restricted panel vector autoregression (PVAR) model with a cointegration built into its specification. The cointegration term is known as the correction term since deviations from the long-run equilibrium are corrected gradually through a series of partial short-run adjustments. The PVECM is shown as follows: p

ΔL(NLIP )t = Ci + ∑ γ 'ΔFi ,t −k + αECMt −1,

(3)

k =1

where: i – r epresents the panel identity or cross-country identifier k represents the lag length, p – r epresents the optimal lag length selected in accordance with the Schwarz Criterion (SC), F – a vector of the stationary forms for seven variables related to GDP per capita, number of passenger cars per 1,000 people, ratio of quasi-money, inflation, population density, level of education and the rule of law. The error-correction-term ECMt–1 is defined as the difference between the actual non-life insurance penetration at time t–1 and its estimate from the long-run equation in the same period. The presence of ECMt–1 in this equation demonstrates the dynamic short-run adjustment. When the non-life insurance penetration deviates from its long-run equilibrium the ECM term will subsequently work to bring it back to the equilibrium level. Therefore its coefficient α is expected to be negative.

4. Empirical results Table 2 shows the unit root tests results. The ADF and PP Fisher-type test were performed using 95% critical values (in parenthesis after each statistic). The table shows that rate of inflation (INF) and the ratio of quasi-money RQM are

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

stationary at levels I (0). The immediate conclusion from this analysis is that any dynamic specification of the model in the levels of the series is likely to be inappropriate and may be plagued by problems of spurious regression [Adam 1992]. It is also argued that econometric results of the model in the levels of the series may not be ideal for policy making. This proposition thus lends credence to the earlier doubts cast over the efficacy of past studies in policy decisions. Lastly the above mentioned variables were not included in the co-integration analysis because, by definition, an I (0), or I (2) is not expected to have a longrun relationship with I (1) series [Adam 1992]. Table 2. Unit root tests ADF – Fisher Chi-square LLP

PP – Fisher Chi-square IPS

Level

First Difference

Level

First Difference

LNLIP

20.8071 (0.1860)

41.7717 (0.0004)

14.2309 (0.5816)

81.4438 (0.0000)

LGDPPC

2.08823 (0.8636)

40.1109 (0.0008)

5.51594 (0.9925)

52.1942 (0.0000)

LNPV

24.6933 (0.1316)

22.3708 (0.0754)

35.5491 (0.0033)

50.8299 (0.0000)

RQM

38.8550 (0.0011)

35.5023 (0.0001)

INF

41.036 (0.0005)

381.381 (0.0000)

LPD

28.1975 (0.0299)

38.4756 (0.0013)

18.0005 (0.3239)

127.298 (0.0000)

LEDU

14.8111 (0.3219)

25.0422 (0.0342)

2.27838 (0.9998)

52.4518 (0.0000)

30.2203 (0.1132)

59.8468 (0.0000)

59.7285 (0.0653)

97.7486 (0.0004)

But according to Juselius [2003], if the time perspective of the studies has macroeconomic behaviour in the medium and long- run then most macroeconomic variables exhibit considerable inertia, consistent with no stationary rather than stationary behaviour. Because inflation would not appear to be statistically different from a non-stationary variable, treating it as a stationary variable is likely to invalidate the statistical analysis and lead to incorrect economic conclusions. On the other hand, treating inflation as a non-stationary variable gives us the opportunity to find out which other variable(s) has/have exhibited a similar stochastic trend by exploiting the cointegration property. Because the time perspective of our study are the long historical macroeco-

Economics and Business Review, Vol. 1(15), No. 3, 2015

nomic movements, the inflation ratio of quasi-money, we will include this in our model and will treat it as a non-stationary variable at their levels. Since it has been determined that the variables under examination are integrated in order I (1), the co-integration test is performed. Cointegration analysis addresses the problem of spurious regressions amongst non-stationary time series. Estimation in a system context may shed light on important interrelationships amongst series whilst reducing the risk of endogeneity bias – Banerjee et al. [1993]. The most important application of cointegration in economic estimations is that it shows that there is a long-run relationship between variables which are cointegrated. The results from the Kao test indicate that the null hypothesis of no cointegration is rejected at the 1 percent level of significance which implies that there exists a cointegration relation between direct non-life insurance penetration and selected variables. Table 3. Kao residual cointegration test Series D(NLIP) RQM D(LNPV) D(LGDPPC) INF D(LEDU) D(RL) D(LPD)

ADF t-Statistic

Probability

–4.859517

0.0000

Null Hypothesis: No cointegration.

In Table 4 the parameters α have an expected negative sign in all four groups, which determines the speed of adjustment towards equilibrium. The speed of adjustment parameter is –0.55. These results indicate that when the non-life insurance penetration deviates from its long-run equilibrium the speed of adjustment will subsequently work to bring it back to the equilibrium level, which in our case will take almost 1 year. Next, the regression results indicate a positive association between the number of passenger vehicles per 1,000 people and non-life insurance penetration. This finding confirms the empirical result in literature that a high number of passenger cars per 1,000 people impacts positively on non-life insurance consumption Feyen, Lester, and Rocha [2011]. This result suggests that motor third party liability insurance takes a dominant place in the insurance market in SEE countries and confirms that car penetration is a driver of insurance development in SEE countries. The reason is that people in these countries are not yet sufficiently informed and have not yet acquired an insurance culture and mainly use car insurance or compulsory motor third party liability insurance (comprehensive car insurance is usually voluntary but also common in many countries). The positive effect of GDP per capita in non-life insurance penetration as demonstrated in development literature is confirmed by the results of this study. GDP per capita has a positive

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

impact on non-life insurance penetration during the period under investigation. Obviously increased income allows for higher consumption in general, makes insurance more affordable and creates a greater demand for non-life insurance to safeguard acquired property. The positive impact of macroeconomic conditions on purchasing decisions of non-life insurance indicates that the good shape of the domestic economy in countries from SEE is a source of the growth of operations of the real sector and other customers of insurance companies and creates higher demand for new insurance (i.e. property insurance and protection against financial risk). In this study inflation appears to have a negative influence on non-life insurance penetration. Therefore macroeconomic stability plays an important role in the development of the non-life insurance market. An n unstable economic environment can result not only in lower disposable incomes, but is also associated with higher inflation, increased uncertainty for the insurer and the insured. Inflation leads to higher claims’ costs thereby eroding profitability. It has the greatest effect on long-tail lines: the longer the tail, the greater the impact. If inflation rises in the short term it is less harmful if premium rates can be adjusted. But this is not always possible if regulations or the competitive environment do not allow it. Longer periods of high inflation are very costly for non-life insurers. Table 4. PVECM results Coefficient

Standard errors

t-statistics

Probability

–0.554412c

0.09772

–2.60336

0.0154

0.190013

Variables

0.024814

0.224548

0.8233

LNPV

2.314518

0.43935

5.26802

0.0777

RQM

0.019273

0.00185

10.4145

0.5334

LGDPPC

0.498557

0.28703

1.73695

0.0538

INF

–0.042795

0.00710

6.02467

0.0600

LPD

–4.722229

7.03927

–0.67084

0.7783

LEDU

–0.482487

0.69590

0.69333

0.9792

0.823282

0.34365

2.39569

0.9953

Coefficient of determination R2

52.03

White Heteroskedasticity Tests Lagrange Multiplier p-value

0.1457

The Jarque berra normallity test

0.6192

a b

, and indicates test statistic is significant at the 10%, 5% and 1% level.

Economics and Business Review, Vol. 1(15), No. 3, 2015

The models are also checked for the explanatory power of the coefficient of determination, the important influence of dependent variables, heteroscedasticity, serial correlation and normality of the parameters of the equations. The coefficient of determination R2 in the model presented is 52.03. This means that the dependent variable of 52.03% is found to be appropriate by the independent variables. The residual white heteroscedasticity test (p-value 0.7549) indicates no heteroscedasticity in the models. The Lagrange Multiplier (LM) test showed that there is no serial correlation between residuals at any lag (p-value 0.1457). The Jarque berra test is used for testing whether the series is normally distributed. As can be seen from Table 4 we cannot reject the null hypothesis of a normal distribution and can therefore conclude that these residuals have normal distribution.

Conclusions This paper ascertains empirically the determinants of non-life insurance consumption in 8 countries from SEE using time series data from 1995 to 2011 by applying the cointegration and panel vector error correction model. We find proof of the existence of a relationship amongst several of the variables under consideration. Specifically we discover that the number of passenger cars per 1,000 people, GDPPC and inflation are significant predictors of non-life insurance penetration. The results show that the parameter for the speed of adjustment(ECMt–1) indicates that short-term deviation from long-term balance corrected at rate of 55% takes almost 1 year. In general, Croatia as a member of EU, has a more developed insurance (life and non-life) sector than the other seven countries included in the research (Albania, Bosnia and Hercegovina, Belarus Macedonia, Moldova, Serbia and Ukraine). Better regulation and supervision in Croatia were partly motivated by the European integration process and the need to adopt EU standards. Thus many of the insurance sector weaknesses traditionally characterising emerging markets have gradually been eliminated. The membership of the EU increases consumers’ confidence in the stability of the market, thus stimulating the demand for non-life insurance products. Prior to becoming a member of the EU new entrant countries had to conduct a number of reforms in order to improve their economic environment and measure up to EU standards. Therefore we can emphasise the importance of working at joining the EU by the non-member countries included in the research. In future research, when more data become available, it would be useful to take a much bigger sample in terms of countries and periods, which will lead to a greater understanding and knowledge of determinants of non-life insurance demand. Also in the future more attention should be placed on the sup-

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

ply side of insurance industries by analyzing and identifying factors that cause different degrees of cost and profit efficiencies across countries. This may further highlight factors that promote sound insurance growth.

References Adam, C.S., 1992, Recent Developments in Econometric Methods: An Application to the Demand for Money in Kenya, Economic Research Consortium Special Paper, no. 15. Arrow, K., 1971, Essays in the Theory of Risk Bearing, Markham, Chicago des Assurances CE (2007) Non-life insurance in Europe, Tech. rep. Beck, T., Webb, I., 2003, Determinants of Life Insurance Consumption across Countries, World Bank and International Insurance Foundation. Banerjee, A., Dolado, J.J., Galbraith, J.W., Hendry, D., 1993, Co-integration, Error Correction, and the Econometric Analysis of Non-Stationary Data. Advanced Texts in Econometrics, Oxford University Press, Oxford. Beenstock, M., Dickinson, G., Khajuria, S., 1986, The Determination of Life Premiums: An International Cross-Section Analysis Insurance, Mathematics and Economics, no. 5: 261–270. Beenstock, M., Dickinson, G., Khajuria, S., 1988, The Relationship between PropertyLiability Insurance and Income: An International Analysis, Journal of Risk and Insurance 55: 259–272. Beck, T., Webb, I., 2003, Determinants of Life Insurance Consumption across Countries, World Bank and International Insurance Foundation. Browne, M., Chung, J., Frees, E., 2000, International Property-Liability Insurance Consumption, Journal of Risk and Insurance, no. 67: 73–90. Browne, M.J., Kim, K., 1993, An International Analysis of Life Insurance Demand, Journal of Risk and Insurance, vol. 60, no. 4: 616–634. Campbell, R.A., 1980, The Demand for Life Insurance: An Application of the Economics of Uncertainty,. Journal o f Finance, no. 35: 1155–1172. Campbell, J.Y., Perron, P., 1991, Pitfalls and Opportunities: What Macroeconomists Should Know about Unit Roots, Macroeconomics Annual, National Bureau of Economic Research: 141–201. Engle, R.F., Granger, C.W.T., 1987, Co-integration and Error Correction Representation, Estimation and Testing, Econometrica, vol. 55: 215–276. Esho, N., Kirievsky, D., Zurbruegg, W.R., 2004, Law and the Determinants of PropertyCasualty Insurance, Journal of Risk and Insurance, 71: 265–283. Falciglia, A., 1980, The Demand for Non-life Insurance: A Consumption-oriented Model, GP, 17: 45–53. Feyen, E., Lester, R., Rocha, R., 2011, What Drives the Development of the Insurance Sector? An Empirical Analysis Based on a Panel of Developed and Developing Countries, Policy Research Working Paper, no. 5572. Fortune, P., 1973, A Theory of Optimal Life Insurance: Development and Tests, Journal of Finance, no. 28: 587–600. Hemming, R., Manson, A.,1988, Is Privatization the Answer? Finance and Development, 25(3): 31–33.

Economics and Business Review, Vol. 1(15), No. 3, 2015

Jung, W.S.,1986, Financial Development and Economic Growth: International Evidence, Economic Development and Cultural Change, no. 34: 333–346. Juselius, K., 2003, The Cointegrated VAR Model: Econometric Methodology and Macroeconomic Applications. Kao, C., 1999, Spurious Regression and Residual-Based Tests for Cointegration in Panel Data, Journal of Econometrics, no. 90: 1–44. Lewis, F.D., 1989, Dependents and the Demand for Life Insurance, American Economic Review, no. 79: 452–466. Li, D., Moshirian, F., Nguyen, P., Wee, T., 2007, The Demand for Life Insurance in OECD, The Journal of Risk and Insurance, no. 74: 637–652. Liu, L.Y., Woo, W.T., 1994, Saving Behaviour under Imperfect Financial Markets and the Current Account Consequences, Economic Journal, no. 104: 512–527. Maddala, G.S., Wu, S., 1999, A Comparative Study of Unit Root Tests with Panel Data and a New Simple Test, Oxford Bulletin of Economics and Statistics, no. 61: 631–652. Mayers, D., Smith, C.W., 1990, On the Corporate Demand for Insurance: Evidence from the Reinsurance, Journal of Business, vol. 63: 19–40. Mossin, J., 1968, Aspects of Rational Insurance Purchasing, Journal of Political Economy, no. 79: 553–568. Njegomir, V., Stojić, D., Marković, D., 2011, Liberatisation, Market Concentration and Performance in the Non-Life Insurance Industry of Ex-Yugoslavia, Ekonomski Misao Praksa DBK BR.1 GOD XX: 21–40. Njegomir, V., Stojić, D., 2012, Determinants of Non-life Insurance Market Attractiveness for Foreign Investments: Eastern European Evidence, Ekonomska Istraživanja, vol. 25, no. 2: 297–310. Outreville, J.F., 1990, The Economic Significance of Insurance Markets in Developing Countries, Journal of Risk and Insurance, no. 57: 487–498. Outreville, J.F., 1996, Life Insurance Markets in Developing Countries, Journal of Risk and Insurance, no. 63. Park, S., Lemaire, J., Chua, C.T., 2010. Is the Design of Bonus-Malus Systems Influenced by Insurance Maturity or National Culture?, The Geneva Papers Issues and Practice, no. 35: S7-S27. Poposki, K., Kjosevski, J., 2013, The Determinants of Non-life Insurance Demand in Central and South Eastern Europe. An Empirical Panel Investigation, XII International Conference, Оhrid, 301505, Оctober. Pratt, J., 1964, Risk Aversion in the Small and in the Large, Econometrica, no. 32: 122–136. Schlesinger, H., 1981, The Optimal Level of Deductibility in Insurance Contracts, Journal of Risk and Insurance, vol. 48: 465–481. Sherden, W., 1984, An Analysis of the Determinants of the Demand for Automobile Insurance, Journal of Risk and Insurance, no. 51: 49–62. Schlesinger, H., 2000, The Theory of Insurance Demand, in: Dionneed, G., Handbook of Insurance, Kluwer, Boston, MA. Syverud, K.D., Bovbjerg, R.R., Pottier, S.W., Will, R.W., 1994, On the Demand for Liability Insurance: Comments, Texas Law Review, no. 72: 1629–1702. Szpiro, G.G., 1985, Optimal Insurance Coverage, Journal of Risk and Insurance, vol. 52, no. 88: 705–710.

K. Poposki, J. Kjosevski, Z. Stojanovski, The determinants of non-life insurance penetration

Sweeney, G., Beard, T., 1992, The Comparative Statics of Self-protection, The Journal of Risk and Insurance, no. 59(2): 301–309. Vadas, G., 2007, Wealth Portfolio of Hungarian Households – Urban Legends and Facts, MNB Working Paper. Verbeek, M., 2004, A Guide to Modern Econometrics, John Wiley and Sons ltd Publishers. Ward, D., Zurbruegg, R., 2000, Does Insurance Promote Economic Growth?, Evidence from OECD Countries, The Journal of Risk and Insurance, vol. 67, no. 4: 489–506.

Economics and Business Review, Vol. 1 (15), No. 3, 2015: 38–57 DOI: 10.18559/ebr.2015.3.4

Microeconomic and macroeconomic determinants of the profitability of the insurance sector in Macedonia1 Tanja Drvoshanova-Eliskovska2

Abstract : The aim of this paper is to investigate the impact of the most representative microeconomic and macroeconomic determinants on the profitability of the insurance sector in Macedonia. The Johansen cointegration technique has been applied to the regression model with quarterly data for the period of time from 2006 to 2011. The results confirm the theoretical suggestions that the assets have a statistically significant positive impact on ROE, from the micro perspective. The interest rate on denar deposits without a currency clause for enterprises has a statistically significant positive impact on ROE and ROA, whilst the rate on deposits of non-financial entities in terms of GDP has a statistically significant negative impact on ROE and ROA, from the macro perspective. Recommendations for increasing the profitability of insurance companies: more productive use of their resources, launching innovative products, enlarging their portfolio, promotions to investors for recapitalization. Recommendations from the macro aspect: structural reforms, extension of savings investments in banks, implementation of new financial instruments, mutual projects amongst the insurance and banking sectors in order that they become complementary. Keywords : microeconomic and macroeconomic determinants, cointegration, profitability, insurance. JEL codes : C32, G22.

Introduction Increased insurance activities enlarge the number of insurance companies as the main provider, which increases the chance of making a profit. Profitability is one of the most important goals of financial management, with a single priority – maximizing the wealth of the owner [Al-Shami 2008]. Special emphasis is placed on achieving profit under sudden and unexpected changes in eco 1 Article received 13 January 2015, accepted 3 August 2015. The opinions expressed in this l research are those of the author only. 2 Senior Credit Analyst at Stopanska Banka AD Skopje, 11 Oktomvri 7, 1000 Skopje, Republic of Macedonia, drvosanova.tanja@yahoo.com.

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

nomic circumstances. From a microeconomic perspective, wrong decisions in insurance companies’ asset management generate bad loans that lead to: deterioration of the quality of their portfolio, increased risk and jeopardizing the liquidity in the insurance sector. Moreover with all side effects taken into account, this encourages negative macroeconomic implications which would have a negative impact on macroeconomic aggregates such as investment and gross domestic product. We conducted this study: to examine the relationship between the determinants and the profitability of the insurance sector; to identify the microeconomic and macroeconomic determinants that influence the profitability of the insurance sector in Macedonia; and to conduct a systematic and detailed econometric analysis of the profitability of the insurance sector in Macedonia in the period 2006–2011 and its microeconomic and macroeconomic determinants. The motives for identifying and exploring the determinants of the profitability of the insurance sector arise from their possible impact on the economy as a whole. Thorough knowledge of them enables more control of the driving trends ensuring better risk management. Insurance companies will be able to take the necessary actions to improve their profitability. The Insurance Supervision Agency of Macedonia (ISA) and all other relevant supervisory bodies can react anticipatively in moments of crisis and bankruptcies. Investors will have the opportunity to protect their investment and focus on the most cost-effective projects for the insurance companies. Insurance users will be able to make the best choice based on the results of the research. In order to create a systematic review of the effect of the current insurance activities on the economic fundamentals, regular analysis of the determinants of the profitability of the insurance sector is required. The remainder of this paper is structured as follows: a review of the theoretical and empirical literature concerning the microeconomic and macroeconomic determinants of the profitability of the insurance sector is presented in Section 1. In Section 2 we present a brief review of the insurance sector in Macedonia, followed by Section 3 that demonstrates the empirical testing and analysis of the determinants of the profitability of the insurance sector in Macedonia. The final section offers conclusions and recommendations.

1. Theoretical and empirical literature 1.1. Microeconomic determinants of profitability in the insurance sector A number of existing studies focus on analyzing the determinants of the profitability in the context of the banking sector. We find, however, no exhaustive empirical work for the insurance sector, especially about the economic transi-

Economics and Business Review, Vol. 1(15), No. 3, 2015

tion in countries in Central and Eastern European (CEE). Some of the relevant, albeit general, studies are discussed in this section. The profit rate is defined as a financial measure that is used to assess the ability of a company (industry) to generate gains compared to its total cost over a period of time. According to Al-Shami [2008], there are many different ways to measure profitability, such as the rate of return on assets – ROA and the return on equity – ROE. ROA is the indicator of the profitability of the company in terms of total assets. ROA indicates how efficiently management uses the funds to make a profit. ROE indicates the amount of profit the company realizes from the invested funds of shareholders. The use of the single accounting system by life insurance companies [Wright 1992], makes it difficult to measure the profitability compared to other financial institutions or companies. As for insurance companies profitability depends on many factors, including the actual mortality rate, investment income, capital gains or losses, policy distribution of state dividends fees and taxes. The difference in profit between insurance companies from the same geographic region suggests the existence of internal factors or features of the insurance companies themselves. Ćurak, Pepur, and Poposki [2011] researched the determinants of the financial performance of Croatian composite insurers, between 2004 and 2009. The determinants of profitability, selected as explanatory variables, include both internal factors, specific to insurance companies and external factors, specific to the economic environment. The results of the panel data show that company size, underwriting risk, inflation and return on equity have a significant influence on insurers’ profitability (ROA). This survey indicates that the Croatian insurance market has a low level of development, but it is very dynamic. Hrechaniuk, Lutz, and Talavera [2007] pointed out that the size of the insurer is important determinant of its profitability. In this context it is much harder for smaller companies to write insurance policies than for bigger ones, since smaller companies cannot secure their clients in the case of aggregate uncertainty or a big catastrophe. It is interesting to note that there are different results shown on the impact of the size of the insurance companies on profitability in Spain and Ukraine. Thus the influence in Ukraine is positive and negative in Spain. Most likely the negative relationship in Spain is due to high administrative costs, typical for the large insurance companies. According to the survey of Kashish and Kasharma [1998], conducted for insurance companies in Jordan, profitability is treated as a dependent variable and is calculated as the rate of return on assets. A positive and statistically significant relationship has been found between the age of the company and its profitability for the year of 1994, whilst the results for 1995 are of lesser significance. The expectations for a positive relationship between the age and the profitability of the company are confirmed in the Vijayakumar and Kadirvelu [2004] study. The older the company is the greater will be the opportunities to

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

increase profitability, because the experience and efficiency in the manufacturing process can reduce costs. It was concluded that age is the strongest determinant of profitability. The capital of a company represent its own funds which provide an opportunity to take on broader activities and achieve higher profits, on the one hand, but on the other, funds include their own costs. The relationship between the volume of capital and profitability in the banking sector has been analyzed by Buser et. al. [1981]. It was concluded that banks that have a relatively large volume of capital impose invisible barriers to the entry of competition in the banking industry. Actually these banks can financially serve more customers and can take higher risks, which will secure profitability, whilst other banks with lower levels of capital would be prevented from competing in the banking sector due to the increased costs. Empirical research on this was made by Berger [1995] in analyzing the US banking system. He identifies a positive relationship between the profitability of banks and their capitalization. He highlights that well-capitalized banks in case of a bankruptcy threat would face lower costs to overcome the situation, due to having a reduced cost of borrowing. The choice of the appropriate rate of borrowing for the companies’ management is not easy due to its vague effect on profitability, more precisely; sometimes the effect can be negative or positive. The theoretical findings show that companies choose the borrowing rate that best suits their capital structure and fit the characteristics and performance of the company. In this regard the study of Harrington [2005] supports the theory of capital structure in respect of the relationship between the rate of borrowing and the rate of profitability. He explains that when a company does well, then borrowing can contribute to the achievement of a higher rate of return on equity – ROE, assuming the fixed costs of the company remain unchanged or increase with low dynamics. In this way a financial leverage will be created, whereby the additional revenue will be distributed just amongst the equity holders and thus will increase profitability expressed by ROE. But this financial leverage also has its effect when the company operates with a negative financial result (negative ROA), thus the loss multiplies the decrease of invested capital, Petrevski [2008]. Hence it is important to determine the optimum level (the border line) of financial leverage and to take advantage of its positive effects. However a generally accepted opinion is that the company with a lower rate of indebtedness, i.e. a higher rate of own funds, is in better position to protect itself against various risks. Hurdle [1974] also points out that the company with an increased rate of borrowing is exposed to greater financial risk than the company with a low rate of borrowing. Relevant in this context is the study of Vijayakumar and Kadirvelu [2004] with their theoretical assumptions about the negative relationship between debt and profitability. Although the estimated coefficient of indebtedness did not confirm their theory, (namely they received positive signs of the coefficient), still there is an empirical argument for the expect-

Economics and Business Review, Vol. 1(15), No. 3, 2015

ed positive relationship between leverage and profitability in certain cases. The reason for such a result is the low level of indebtedness of the companies that had been taken in their sample as operated in the energy industry which had a high risk and required a high degree of capitalization. According to Panayiotis, Athanasoglou, and Delis [2008] survey that banks with lower rate of indebtedness (higher capital) will generally achieve a higher rate of ROA, but a lower rate of ROE. This study shows that ROA is valid as the main index by which to measure profitability, because the analysis of the link between debt and ROE shows that not enough attention is paid to the risk which is incurred through high indebtedness, which is often determined by the requirements of the legislation on the minimum capital of banks. Hutchison and Cox [2006] examined the relationship between financial leverage and ROE for the banking sector in the US. They found a negative relationship between debt (expressed as the ratio of capital and assets) and profitability of banks which was not relevant for the top banks. The rate of loss is the ratio between the annual damages paid by insurance companies and collected premiums, Al-Shami [2008]. In insurance companies the annual damages paid tend to be lower than the collected premiums. Thus the rate of loss will be lower. Hrechaniuk, Lutz, and Talavera [2007] examined the performance and the determinants of profitability of the insurance sector in Spain, Lithuania and Ukraine in specific years. Their theoretical model anticipates that the rate of loss will affect adversely on the insurance companies’ financial results. The results show that the rate of loss positively affects the financial results of companies in Lithuania, whilst it negatively affects profitability in Ukraine. The estimated coefficient for the rate of loss of insurance companies in Ukraine supports their hypothesis of an inverse relationship between the rate of loss and profitability of insurance companies which is statistically significant.

1.2. Macroeconomic determinants of profitability in the insurance sector In the context of macroeconomic determinants only a few theoretical explanations for their impact on personal observations are found. We summarize them below. Gross domestic product – GDP is the measure of overall economic activity in a country. When increasing economic factors work more and there are more opportunities for achieving positive financial results. From that perspective GDP growth is expected to have a positive impact on the profitability of the insurance sector. Insurance companies, such as financial institutions, mobilize financial resources and have the opportunity to place them in the banks in the form of deposits or financial instruments in the stock market. Thus an increase in the

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

interest rate would lead to the expectation that the insurance sector would achieve higher interest income and increased profitability. The banking and insurance sectors are structural elements of the wider financial system in the economy. As both sectors offer financial services they are unavoidably influenced by the nature of their business and have the ability to cooperate and they can be complementary. Furthermore these two sectors can act as competitors or substitutes in the fight for attracting customers wishing to save. Depending on the development of the financial system the effect of the activity of the banking sector can be either positive or negative with regard to the profitability of the insurance sector.

2. Brief review of insurance market in Macedonia The insurance sector in Macedonia is the third segment in the financial system representing only 3.3% of the total assets in the financial market. It consists of 15 insurance companies, 26 insurance brokerage companies, 9 companies of insurance representation and 1 bank – acting in the field of life insurance. It is characterized by a moderate market concentration, a growth trend, especially in life insurance, which is dominantly in foreign ownership, in conformity with the regulatory framework and enhanced supervision.3 The basic indicators for the insurance sector in Macedonia are presented in Table 1 for the period of 2006 to 2011. Table 1. Key indicators of the insurance sector in Macedonia Descriptive statistics Gross Written Insurance Insurance Premiums Pene Density (GWP) in tration Rate in MKD Rate (%) MKD

Gross Paid Claims in MKD

Profit/loss – earnings before tax in MKD

ROA (%)

ROE (%)

2006

5,445,239.00

1.70

2,669.00

2,797,124.00

311,710,863.00

1.86

8.58

2007

6,108,839.00

1.80

2,988.00

2,865,555.00

310,660,678.00

2.05

6.77

2008

6,421,435.00

1.60

3,135.00

3,182,341.00

275,818,962.00

1.66

4.65

2009

6,182,401.00

1.53

3,012.00

2,962,250.00

–100,848,992.00 –1.40

–3.99

2010

6,480,874.00

1.53

3,151.00

2,988,373.00

102,127,970.00

0.61

1.75

2011

6,808,264.00

1.50

3,304.00

3,006,170.00

–57,238,407.00 –0.57

–1.75

Source: Annual Reports of the insurance market in Macedonia from ISA, www.aso.mk.

Source: National Bank Of the Republic of Macedonia, Financial Report for stability in Macedonia for 2013.

Economics and Business Review, Vol. 1(15), No. 3, 2015

According to the degree of development of the insurance market it can be noted that the level is appropriate to the level of the related group of countries in South East Europe emphasizing the potential for growth. Generally the main characteristics of the group of countries are: equal participation of the insurance sector in the structure of the financial system, common structure of the portfolio of the insurance products, where the most important product is the compulsory third party liability4 cover for motor vehicles with a potential for covering more catastrophic risks.

3. Empirical testing and analysis of the determinants of profitability of the insurance sector in Macedonia In order to review the theoretical suggestions and compare them with the results from the other surveys presented an empirical analysis is made of the determinants of profitability of the insurance sector. Research of this kind has not been conducted for Macedonia so far. The most representative variables are taken in the regression model. These variables cover the main activities of the insurance sector from the micro and macro aspects which determine the profitability in the best manner. Variables are taken on an aggregate level in order to examine their impact on the whole insurance sector. The following theoretical hypotheses are checked in this study as follows: 1. There is a positive relationship between the size of the insurance sector and the profitability of the insurance sector in Macedonia 2. There is a positive relationship between the volume of capital of the insurance sector and the profitability of the insurance sector in Macedonia 3. There is a negative relationship between the leverage of the insurance sector and the profitability of the insurance sector in Macedonia 4. There is a positive relationship between the economic activity in the country and the profitability of the insurance sector in Macedonia 5. There is a positive relationship between the investment level of the insurance sector and the profitability of the insurance sector in Macedonia 6. There is a negative relationship between the growth of the banking sector and the profitability of the insurance sector in Macedonia. For these tests quarterly data for the period 2006 to 2011 are used. The data are taken from the websites of the ISA and the National Bank of the Republic of Macedonia (NBRM) websites to which a linear interpolation5 is applied. 4 Source: National Bank Of the Republic of Macedonia, Financial Report for stability in Macedonia for 2013. 5 Linear interpolation was performed using the standardized formula y = y0 + (x – x0) × (y1 – y0)/(x1 – x0), to t allow the interpolation of annual data for the variable of the insurance

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

The basic equation for the regression model is: yi = β0 + β1x1i + β2x2i + … + βnxni + ε1,

(1)

where: yi – the dependent variable, x1i, x2i, …, xni – the independent microeconomic variables, β0, β1, β2, …, βn – the coefficients to be calculated, ε1 – the rated error which includes all the other factors that affect the dependent variable, but are not taken into the independent variables analyzed. The applied variables in this model are as follows: –– Dependent variables (as a measure of the profitability of the insurance sector) {{ ROE – Rate of Return on Equity in the insurance sector (calculated as the ratio between net income and equity), expressed as a percentage; {{ ROA – Return On Assets of the insurance sector (calculated as the ratio between net income and assets), expressed as a percentage. –– Independent microeconomic variables {{ LNASSETS – Natural logarithm of the assets of the insurance sector, where the assets are expressed in million denars (as a measure of the size of the insurance sector); {{ LNEQUITY – Natural logarithm of equity in the insurance sector, where the capital is expressed in millions of denars (as a measure of the funds of the insurance sector); {{ LEVERAGE – Rate of equity in relation to the assets of the insurance sector, expressed as a percentage (as a measure of leverage of the insurance sector). –– Independent macroeconomic variables {{ GDPGROWTH – Growth rate of real gross domestic product, expressed as a percentage (as a measure of overall economic activity); {{ INTEREST – Interest rate on denar deposits without a currency clause of enterprises expressed as a percentage (as savings which the insurance sector receives from the investment of funds in banks); {{ DEPTOGDP – Rate on deposits of non-financial entities in terms of gross domestic product, expressed as a percentage (as a measure of the development of the banking sector and the major competitive sector of the insurance sector). sector on a quarterly basis by using data (annual and quarterly) for the same variable from the banking sector, which is the most appropriate sector, characterized by most of the similarities. So, y is the corresponding value of the quarterly interpolated net profit, equity and assets of the insurance sector, y0 and y1 are the annual value of net profits, equity and assets of the insurance sector, whilst x0 and x1 are the corresponding values of quarterly net profit, equity and assets of the banking sector taken from the website of the NBRM.

Economics and Business Review, Vol. 1(15), No. 3, 2015

–– Dummy variables6 7 {{ DUM1 – The variable which covers the impact of the global economic crisis; 8 {{ DUM2 – The variable which covers the impact of the increased cost of value adjustment of the claims against insurance premiums, as a result of the application of the regulation on the valuation of the items in the balance sheet (Annual Report, ISA 2012, p. 25). The basic equation applied and adapted to the research of the relationship between determinants and profitability of insurance companies in Macedonia is presented as following: yi is the profitability of insurance companies, ROE and ROA, x1i, x2i, …, xni are LNASSETS, LEVERAGE, LNEQUITY, GDPGROWTH, INTEREST, DEPTOGDP, β0, β1, β2, …, βn, are coefficients, the parameters to be calculated that determine the direction and intensity of the impact of the determinants on the profitability of insurance companies in Macedonia. In order to establish the regression model, it is first necessary to determine the integration features of the time series, which include the examination of the (non)stationary or the variables. By using the two most popular tests, Augmented Dickey Fuller – ADP and Phillips Peron – PP one, the hypothesis that the time series has a single root (Unit Root), or that the time series is non-stationary was examined. Only variables integrated in the same order are progressed in the research process. The results of both tests are shown in the following tables. As presented in the results the variable Leverage is excluded because it is undoubtedly integrated in the different level I (2). The regression model developed can be shown in 4 specifications: ROEt = β0 + β1* LNASSETSt + β2* LNEQUITYt + εt,

(2)

ROAt = β0+ β1* LNASSETSt + β2* LNEQUITYt + εt,

(3)

ROEt = β0 + β1* GDPGROWTHt + β2* INTERESTt + β3* DEPTOGDPt + εt, (4) ROAt = β0 + β1* GDPGROWTHt + β2* INTERESTt + β3* DEPTOGDPt + εt. (5) 6

The presence of dummy variables should provide a stability to the estimated ratios especially in situations where exogenous factors affect the dependent variables, such as the economic crisis, whose greatest impact on the Macedonian economy as a whole was reflected in 2009, when it inevitably affected the results of the profitability of the insurance sector of Macedonia, as well as the effects of the application of the regulation on the valuation of the items in the balance sheet, which is an administrative measure and which had an effect of caunsing loss-making in the insurance sector in 2011. 7 It takes the value 1 for the whole of the year of 2009 and 0 for all other periods. 8 It takes the value 1 for the whole of the year of 2011 and 0 for all other periods.

[47]

–1.99 –2.00 –2.39 –1.84 –2.11 –1.95 –1.42 –1.60

t-statistic

In level

Test critical Test critical values: 5% values: 10% –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.03 –2.66 –5.62 –5.17 –2.66 –3.52 –1.93 –5.38 –7.11 –1.64

t-statistic

Test critical Test critical values: 5% values: 10% –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.03 –2.66

1st difference

Integrative characteristics of series ADF-test

–21.83

–4.52

t-statistic

Test critical Test critical Conclusion values: 5% values: 10% I(1)* I(1)* I(1)** I(1)* –3.00 –2.65 I(2)* I(1)* I(1)* –3.03 –2.66 I(2)*

2nd difference

–1.96 –2.00 –1.74 –1.84 –1.79 –2.04 –1.81 –1.53

t-statistic

In level

Test critical Test critical values: 5% values: 10% –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –7.37 –5.43 –2.66 –3.51 –1.99 –5.36 –6.33 –12.44

t-statistic

Test critical Test critical values: 5% values: 10% –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64 –3.00 –2.64

1st difference

Integrative characteristics of series PP-test

–4.52

t-statistic

Test critical Test critical Conclusion values: 5% values: 10% I(1)* I(1)* I(1)** I(1)* –3.00 –2.65 I(2)* I(1)* I(1)* I(1)*

2nd difference

* and ** means rejection of the Null Hypothesis: the appropriate variable has a unit root (is non-stationary) at 5% and 10% level of significance.

ROE ROA LNASSETS LNEQUITY LEVERAGE GDPGROWTH INTEREST DEPTOGDP

Variable

Table 3. Unit root test PP

* and ** means rejection of the Null Hypothesis: the appropriate variable has a unit root (is non-stationary) at 5% and 10% level of significance.

ROE ROA LNASSETS LNEQUITY LEVERAGE GDPGROWTH INTEREST DEPTOGDP

Variable

Table 2. Unit root test ADF

Economics and Business Review, Vol. 1(15), No. 3, 2015

Regression equations (2) and (3) are used to assess the impact of microeconomic determinants to be examined, whilst regression equations (4) and (5) investigate the effect of macroeconomic determinants on the profitability of the insurance sector of Macedonia. This division is made in order to avoid increasing the parameterization of the model. Accordingly possible endogeneity between variables and integration features of the time series, Johansen co-integration technique9 are used. The order is defined as Vector Auto Regression – VAR which determines the number of past values of the variables or time delays (Lags). The results indicate that the most appropriate order of the VAR-model in the first two specifications is VAR 2, and in the third and the fourth it is VAR 1, meaning the inclusion of 2 or 1 lags in the model which ensures correction Table 4. VAR lag order selection criteria for the regression model ROE = f (LNASSETS, LNEQUITY & DUM1, DUM2) Lag

FPE

AIC

0.000320

0.461078

0.907414

0.566221

64.69490

1.32e–05

–2.764.171

–1.871.500

–2.553.885

28.90730*

3.61e–06*

–4.169628*

–2.830621*

–3.854198*

ROА = f (LNASSETS, LNEQUITY & DUM1, DUM2) Lag

FPE

AIC

2.53e–05

–2.076.467

–1.630.131

–1.971.324

67.33993

8.85e–07

–5.467.030

–4.574.359

–5.256.744

29.92534*

2.24e–07*

–6.950797*

–5.611791*

–6.635368*

ROE = f (GDPGROWTH, INTEREST, DEPTOGDP & DUM1, DUM2) Lag

FPE

AIC

3439.835

19.48785

20.08296

19.62804

42.43729*

946.9640*

18.11324

19.50184*

18.44035*

16.24436

1253.842

18.09103*

20.27311

18.60506

ROA = f (GDPGROWTH, INTEREST, DEPTOGDP & DUM1, DUM2) Lag

FPE

AIC

320.6339

17.11497

17.71008

17.25516

45.80115*

70.53614*

15.51611*

16.90470*

15.84322*

12.03932

136.8805

15.87617

18.05825

16.39020

* indicates the order of VAR according to each criterion. 9

It allows mult-ivariate assessment based on the method of maximum likelihood.

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

for the possible endogeneity in the regression model. In determining the cointegration and the number of cointegrating relationships, i.e. cointegrating vectors amongst variables in the equations, we also, examined if there is a stationary linear combination, i.e. vector with I (0) integration process, amongst the variables that are not stationary. For this we employed Maximal Eigen value of the Stochastic Matrix – λmax and Trace of the Stochastic Matrix – λtrace.10 In economic practice the second, third and fourth option are most often used as evidenced in Johansen [1992] and Harris and Solis [2003]. Hence the results of the tests for the cointegration in the first and second regression specification clearly distinguish option 4 and in the third and fourth equations option 2 has been chosen as the optimum. Table 5. Pantula-principle for determining the number of cointegration vectors in the model ROE = f (LNASSETS, LNEQUITY & DUM1, DUM2) No Intercept

Intercept

No Trend

Trend

Λtrace

Λmax

Test type

ROА = f (LNASSETS, LNEQUITY & DUM1, DUM2) Test type

No Intercept

Intercept

No Trend

Trend

Λtrace

Λmax

ROE = f (GDPGROWTH, INTEREST, DEPTOGDP & DUM1, DUM2) Test type

No Intercept

Intercept

No Trend

Trend

Λtrace

Λmax

ROA = f (GDPGROWTH, INTEREST, DEPTOGDP & DUM1, DUM2) Test type

No Intercept

Intercept

No Trend

Trend

Λtrace

Λmax

Both tests test the null hypothesis, according to which it is claimed that there is no cointegration between variables, i.e. r = 0.

Economics and Business Review, Vol. 1(15), No. 3, 2015

After determining the order of the VAR and cointegrating vector specifications the VAR-model is transformed into a method of vector error correction – VECM.11 From all the econometric results for the developed regression model only those coefficients of the variables that established long-term equilibrium and which are statistically significant are interpreted. Table 6. Estimated coefficients for the first specification Dependent variable ROE; ROE = f (LNASSETS, LNEQUITY & DUM1, DUM2)

Variable

Stan dard error

t-statistic

Critical values at 1% significance level

LNASSETS

0.24

0.12

1.89

2.82

2.07

1.72

LNEQUITY

–0.02

0.14

–0.13

–2.82

–2.07

–1.72

TREND

–0.15

0.43

–0.35

–2.82

–2.07

–1.72

Error correction mechanism (ECM)

–0.69

0.21

–3.31

–2.82

–2.07

–1.72

Approximate time of adjustment R2

Critical values at 5% significance level

Critical values at 10% significance level

Con clusion

***

1.45 quarters 58.85%

* and *** means rejection of the Null Hypothesis: the coefficient is not statistically different from zero at the10% and 1% level of significance.

The results from Table 5 indicate that if the variable LNASSETS increases by 1 percent then the variable ROE increases by an average of 0.24 percentage points, assuming other variables remain unchanged. The coefficient to the variable is statistically significant at the 10% level of significance. The coefficient before the variable LNEQUITY is negative and does not have a statistically significant impact on ROE taken as a dependent variable. The coefficient before the TREND is negative and does not have a statistically significant impact on ROE. The timing of adjustment from the short-term imbalance to the long-term equilibrium is 1.45 quarters and is statistically significant at all levels of importance, whilst the coefficient of determination R2 indicates that 58.85% of the 11 This model enables the separation of long-term relationships between the variables from short-term relationships. Also it can calculate the adjustment from short-term imbalance to long-term equilibrium.

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

variance in the profitability of the insurance sector expressed through ROE is determined by the variances of the microeconomic determinants, assets and equity of the insurance sector. Table 7. Diagnostic tests for first regression Diagnostic tests for regression ROE = f (LNASSETS, LNEQUITY & DUM1, DUM2) Calculated statistics

Critical values at 1% significance level

H0: No serial correlation in the residuals

11.40

21.67

H0: Normality in the residuals

81.81

16.81

H0: Homoscedastic residuals

81.97

88.38

Conclusion

***

*** indicates rejection of the null hypothesis at 1% significance level.

The results from Table 6 indicate that the econometric results are relevant and unbiased in terms of the first and third test whilst the second test shows that residuals do not follow a normal distribution pattern and cannot be properly distributed logically since only a small sample of data is analyzed. Table 8. Estimated coefficients for the second specification Dependent variable ROA; ROA = f (LNASSETS, LNEQUITY & DUM1, DUM2)

Variable

Stan dard error

t-statistic

Critical values at 1% significance level

LNASSETS

0.03

0.91

2.82

2.07

1.72

LNEQUITY

0.005

0.04

0.12

2.82

2.07

1.72

TREND

–0.05

0.11

–0.45

–2.82

–2.07

–1.72

Error correction mechanism (ECM)

–0.72

0.15

–4.80

–2.82

–2.07

–1.72

Approximate time of adjustment R2

Critical values at 5% significance level

Critical values at 10% significance level

1.39 quarters 77.97%

*** indicates rejection of the null hypothesis at 1% significance level.

Con clusion

***

Economics and Business Review, Vol. 1(15), No. 3, 2015

The results from Table 7 show that the coefficients before the variables LNASSETS and LNEQUITY are positive and do not have a statistically significant impact on the profitability of the insurance sector when ROA is taken as a dependent variable. The coefficient before the TREND is negative and it does not have a statistically significant impact on ROA, taken as a dependent variable. Long-term coefficients are not statistically significant in this regression. The results of diagnostic tests for this specification are similar to the results from the first specification. Table 9. Diagnostic tests for second regression Diagnostic tests for regression ROA = f (LNASSETS, LNEQUITY & DUM1, DUM2) Calculated statistics

Critical values at 1% significance level

H0: No serial correlation in the residuals

11.99

21.67

H0: Normality in the residuals

59.30

16.81

H0: Homoscedastic residuals

78.10

88.38

Conclusion

***

*** indicates rejection of the null hypothesis at 1% significance level.

Table 10. Estimated coefficients for the third specification Dependent variable ROE; ROE = f (GDPGROWTH, INTEREST, DEPTOGDP & DUM1, DUM2)

Variable

Stan dard error

t-statistic

Critical values at 1% significance level

0.09

0.24

0.38

2.83

INTEREST

2.79

0.69

4.04

2.83

2.08

1.72

***

DEPTOGDP

–0.13

0.02

–6.50

–2.83

–2.08

–1.72

***

Intercept

13.67

4.17

3.28

–2.83

–2.08

–1.72

***

Error correction mechanism (ECM)

–0.40

0.13

–3.08

–2.83

–2.08

–1.72

***

GDPGROWTH

Approximate time of adjustment R2

Critical values at 5% significance level

Critical values at 10% significance level

2.08

1.72

2.50 quarters 38.44%

*** indicates rejection of the null hypothesis at 1% significance level.

Con clusion

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

The results from Table 9 reveal that if the variable INTEREST increases by 1 percentage point, the variable ROE increases by an average of 2.79 percentage points, assuming other variables remain unchanged. The coefficient before this variable is statistically significant at all levels of significance. If the variable DEPTOGDP increases by 1 percentage point, the variable ROE on average reduces by 0.13 percentage points, assuming other variables remain unchanged. The coefficient before the variable in question is statistically significant at all levels of importance. If the independent macroeconomic variables have value zero the intercept indicates that ROE will be 13.67%. The coefficient before this variable is statistically significant at all levels of importance. The timing of the adjustment from short-term imbalance to long-term equilibrium is 2.5 quarters and it is statistically significant at all level of importance. The coefficient of determination R2 indicates that 38.44% of the variance in the profitability of the insurance sector expressed through ROE is determined by the variances of these macroeconomic determinants. The results indicate that the econometric results are relevant and unbiased in terms of all three tests. Table 11. Diagnostic tests for third regression Diagnostic tests for regression ROE = f (GDPGROWTH, INTEREST, DEPTOGDP &Â DUM1, DUM2) Calculated statistics

Critical values at 1% significance level

H0: No serial correlation in the residuals

15.57

32.00

H0: Normality in the residuals

12.62

20.09

H0: Homoscedastic residuals

45.79

63.69

Conclusion

***

*** indicates rejection of the the null hypothesis at 1% significance level.

The results from Table 11 indicate that if the variable INTEREST increases by 1 percentage point the variable ROA increases by 0.69 percentage points on average, assuming other variables remain unchanged. The coefficient before the variable in question is statistically significant at all levels of importance. If the variable DEPTOGDP increases by 1 percentage point, the variable ROA on average reduces by 0.02 percentage points, assuming other variables remain unchanged. The coefficient before the variable in question is statistically significant at all levels of importance. If the independent macroeconomic variables have value zero the intercept indicates that ROA will be 1.65%. The coefficient before the variable in question is statistically significant at 10% at all levels of importance. Time adjustment of short-term imbalance to the long run equilibrium is 2.22 quarters and it is statistically significant at all levels of importance whilst

Economics and Business Review, Vol. 1(15), No. 3, 2015

Table 12. Estimated coefficients for the fourth specification Dependent variable ROA; ROA = f (GDPGROWTH, INTEREST, DEPTOGDP & DUM1, DUM2)

Variable

Stan dard error

t-statistic

Critical values at 1% significance level

GDPGROWTH

0.06

1.00

2.83

2.08

1.72

INTEREST

0.69

0.16

4.31

2.83

2.08

1.72

***

DEPTOGDP

–0.02

0.004

–5.00

–2.83

–2.08

–1.72

***

Intercept

1.65

0.96

1.72

–2.83

–2.08

–1.72

Error correction mechanism (ECM)

–0.45

0.12

–3.75

–2.83

–2.08

–1.72

***

Approximate time of adjustment

Critical values at 5% significance level

Critical values at 10% significance level

Con clusion

2.22 quarters

52.77%

* and *** means rejection of the Null Hypothesis: the coefficient is not statistically different from zero at 10%, 5% and 1% significance level.

the coefficient of determination R2 indicates that 52.77% of the variance in the profitability of the insurance sector expressed through ROA is determined by the variances of these macroeconomic determinants. The results indicate that the econometric results are relevant and unbiased in terms of all three tests. Table 13. Diagnostic tests for fourth regression Diagnostic tests for regression ROA = f (GDPGROWTH, INTEREST, DEPTOGDP & DUM1, DUM2) Calculated statistics

Critical values at 1% significance level

H0: No serial correlation in the residuals

11.47

32.00

H0: Normality in the residuals

8.66

20.09

H0: Homoscedastic residuals

50.65

63.69

Conclusion

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

Conclusions and recommendations The results for the specifications with microeconomic determinants indicate that only assets positively affect the profitability of the insurance sector expressed by ROE whilst none of the variables considered affects ROA. Moreover the specification using macroeconomic determinants was better because two macroeconomic variables affect the dependent variable. Namely, the interest rate on deposits of enterprises positively affects both measures of profitability whilst the deposits of non-financial entities adversely affect ROE and ROA, which indicates that the banking sector is more competitive than the insurance sector and it fullfils the function of a substitute for the insurance sector. The most probable reason for such partially illogical results obtained from the specifications with microeconomic determinants arise from certain limitations such as the small sample taken for analysis and the fact that the annual data were interpolated to quarterly levels. Also the results of the third and fourth regression equation suggest that the GDPGROWTH does not affect profitability which is probably also due to the analysis for a short period of time and that gross domestic product may not be suitable as a variable in examining the profitability of the insurance sector. In order to be more precise it should be noted that this may be a consequence of the fact that the insurance sector has a small share in the overall financial sector and in general throughout the Macedonian economy. Specifically the assets of the insurance companies are only a 3.4% share of the total assets of the financial sector as of 2011 (FSR, NBRM, 2012) and from that point of view, due to the large discrepancy between these two variables, it can be concluded that the growth rate of real GDP does not affect the profitability of the insurance sector. Taking into consideration the results obtained from the four regression equations, appropriate recommendations can be made to the planners of economic policies in order to increase the profitability of the insurance sector in Macedonia and implement more successful risk management. Based on the results of the regression equations with microeconomic determinants recommendations are directed at the managers of insurance companies and investors: 1. The creation of conditions to increase the assets and equity of the insurance companies through more effective and efficient use of their resources, especially human resources, through the creation of ideas, projects and the launch of innovative products with lower prices in order to increase the profitability of the insurance sector. 2. An active promotion of the insurance industry to investors to raise capital which will allow expansion of the range of insurance products. Dialogue with the banks about investment projects, for loans or the exchange of securities for the purpose of recapitalization and the implementation of projects.

Economics and Business Review, Vol. 1(15), No. 3, 2015

Based on the results of the regression equations with macroeconomic determinants, recommendations need to be made to the insurance companies and also to other entities whose decisions have a stake in macroeconomic movements such as the state, the NBRM, ISA and banks. In this respect we suggest the following recommendations: 1. In the process of the implementation of structural reforms in order to boost GDP in the form of infrastructure investment it is desirable to use a wider range of insurance products from domestic insurance companies to protect against possible risks to be used regularly with the aim of a greater stimulation of profitability. 2. In the context of the interest rate it would be a desirable extension of investment from insurance companies in the banks in the form of deposits in order to increase profitability. Better planning of investments is implemented in terms of stable interest rates. To maintain a stable monetary system – a fiscal mix is recommended. 3. In respect of the coefficient in front of the variable that represents the development of the banking sector, i.e. the share of deposits of non-financial entities in GDP, which indicates a substitutable effect, it is necessary for the insurance sector to enter into greater cooperation with the banking sector in order to become complementary, not substitutive. It would be worthwhile if announcements for the sale of life insurance are implemented by the raising of loans from the banks on mandatory basis for all types of loan on offer. In addition it would be wise to introduce new mutual products or projects in the banking and insurance sectors, by which means banks would ensure their investments in insurance companies. This product could increase the profitability of the insurance sector and improve the process of risk management in the banking sector. However care should be taken in introducing this product as it requires detailed analysis and the involvement of experts. The risk of the introduction of this product could mean a possible spillover of the risks from the banking sector into the insurance sector. To avoid this it is necessary for the insurance companies and banks to regularly update and strengthen their risk management policies as well as having a detailed involvement and cooperation with ISA institutions and the prudent supervision of the NBRM of such products, each in its own domain.

References Al-Shami, 2008, Determinants of Insurance Companies’ Profitability in UAE, College of Business, University Utara Malaysia. Berger, A.N., 1995, The Relationship between Capital and Earnings in Banking, Journal of Money, Credit and Banking, no. 27.

T. Drvoshanova-Eliskovska, Microeconomic and macroeconomic determinants

Buser, S.A., Chen, A.H., Kane, E.J., 1981, Federal Deposit Insurance, Regulatory Policy and Optimal Bank Capital, Journal of Finance, 35. Ćurak, M., Pepur, S., Poposki, K., 2011, Firm and Economic Factors and Performance: Croatian Composite Insurers, The Business Review Cambridge, vol. 19, no. 1: 136–142. Davidovska-Stojanova, B., Jovanovich, B., Kadievska-Vojnovik, M., Ramadani, G., Petrovska, M., 2008, Real Estate Prices in the Republic of Macedonia, National Bank of the Republic of Macedonia. Harrington, C., 2005, The Effect of Competitive Structure on the Relationship between Leverage and Profitability. Harris, R., Sollis, R., 2003, Applied Time Series Modeling and Forecasting, John Wiley and Sons. Hrechaniuk, B., Lutz, S., Talavera, A., 2007, Do Determinants of Insurers’ Performance Differ in Old EU, the New EU and Outside?, University of Bonn and University of Manchester. Hurdle, G.J., 1974, Leverage, Risk, Market Structure and Profitability, Review of Economics & Statistics, 56 (4). Hutchison, Cox, 2006, The Casual Relationship between Capital and Profitability, Proceedings of the Annals of Financial Economics Westwood Development Group, University of Ontario, Institute of Technology, Ontario. ISA, 2010, Annual report of the insurance market in Republic of Macedonia in 2009, Insurance Supervision Agency. ISA, 2011, Annual report of the insurance market in Republic of Macedonia in 2010, Insurance Supervision Agency. ISA, 2012, Annual report of the insurance market in Republic of Macedonia in 2011, Insurance Supervision Agency. Johansen, S., 1992, Determination of Cointegration Rank in the Presence of a Linear Trend, Oxford Bulletin of Economics and Statistics, vol. 54, no. 3: 383–397. Kasharma, M.K., 1998, Actors Affecting the Profitability of Insurance Companies in Jordan, Alyarmouk University. Kashish, K., 1998, Factors Affecting the Profitability of Insurance Companies in Jordan, Working Paper, Alyarmouk University. Panayiotis, P., Athanasoglou, S., Delis, M., 2008, Bank Specific, Industry Specific and Macroeconomic Determinants of Bank Profitability, International Financial Markets, Institutions & Money, vol. 18. Petrevski, G., 2008, Management Banks, Faculty of Economics – Скопје. Vijayakumar, A., Kadirvelu, S., 2004, Determinants of Profitability: The Case of Indian Public Sector Power Industries, Management Accountant – Calcutta, vol. 39; part 2, Institute of cost and works accountants of India: 118–132. Wright, K.M., 1992, The Life Insurance Industry in the United States an Analysis of Economic and Regulatory Issues, Country Economics Department the World Bank policy research working paper, no. 857.

Economics and Business Review, Vol. 1 (15), No. 3, 2015: 58–77 DOI: 10.18559/ebr.2015.3.5

Policyholder and insurance policy features as determinants of life insurance lapse – evidence from Croatia1 Marijana Ćurak2, Doris Podrug2, Klime Poposki3

Abstract : According to theoretical and empirical literature, life insurance lapse is determined by macroeconomic and insurer-specific factors, demographic and economic features of the policyholder as well as specific characteristics of the life insurance policy. Existing empirical literature on determinants of life insurance options is dominantly focused on developed insurance markets. The aim of this paper is to investigate the drivers of lapse in Croatia using survey data. The research encompasses both policyholder and life insurance contract features. The results of the research show that the number of children, income level, and the financial status of policyholder as well as the duration of the life insurance policies influence policyholders’ decisions to lapse life insurance. Keywords : life insurance, lapse, Croatia. JEL codes : G22, C42.

Introduction Taking into consideration the specific characteristics of an emerging insurance market and the fact that existing literature on factors influencing life insurance options is focused on developed insurance markets, this paper analyses policyholder and insurance policy features as determinants of using of life insurance options in Croatia. The analysis shows that the most important influential factor is change in the financial status of the policyholder. Additionally, income level, number of children and duration of the life insurance policy also have the importance. Life insurance options mitigate the liquidity constraints of a life in 1

Article received 15 April 2015, accepted 3 August 2015. University of Split, Faculty of Economics, Cvite Fiskovića 5, 21000 Split, Croatia; corresponding author, e-mail: marijana.curak@efst.hr. 3 University St. Kliment Ohridski, Faculty of Tourism and Organizational Sciences, M. Tito 95, 6000 Ohrid and Insurance Supervisory Agency, Vasil Glavinov 12, Skopje, Republic of Macedonia. 2

M. Ćurak, D. Podrug, K. Poposki, Policyholder and insurance policy features

surance policy for the policyholder. On the other hand, with the lapse of a life insurance policy, the policyholder could suffer financial loss as well as loss of insurance coverage. However there are various options that could be embedded in a life insurance policy. Precisely, according to the broad definition of life insurance lapse, it refers to “all legal or contractual policyholder options which can significantly change the value of future cash-flows. This includes options to fully or partly terminate, decrease, restrict or suspend the insurance cover as well as options which allow the full or partial establishment, renewal, increase, extension or resumption of insurance cover” [CEIOPS 2010: 155]. Besides the policyholders, life insurance lapse influences the life insurance companies as well. The options impose risk for life insurers in terms of “loss or change in liabilities due to a change in the expected exercise rate of policyholder options” [CEIOPS 2010: 155], affecting liquidity, profitability as well as the solvency of life insurers. Thus it is important to analyse which factors determine policyholders’ decisions to use life insurance options. The existing empirical studies are mainly focused on macroeconomic and insurance-company level factors of life insurance lapse and less on policyholder and insurance policy features. The key reason is the limited availability of data. The main sources of data on life insurance lapse are life insurance associations and their surveys, national supervisory authorities and insurance companies [Elling and Kohanski 2013]. The current empirical research on life insurance lapse is focused on developed countries [Renshaw and Haberman 1986; Kagraoka 2005; Milhaud, Loisel, and Maume-Deschamps 2010; Eling and Kiesenbauer 2013; Canadian Institute of Actuaries 2014]. Although the literature on factors influencing life insurance demand in emerging countries has been increasing [most recently studies are: Elango and Jones 2011; Śliwiński, Michalski, and Roszkiewicz 2013; Dragos 2014], according to our best knowledge, besides two papers on the general characteristics on life insurance lapse in India [Kumar 2009; Surana and Gaur 2013], there is no study of policyholder and insurance contract factors influencing the exercise of life insurance options in emerging life insurance markets. Since the emerging markets have specific insurance market features as well as economic, demographic and social characteristics, it is valuable to investigate if the determinants of the use of life insurance options in developed insurance markets could be confirmed in emerging markets as well. Consequently the main purpose of this paper is to examine the determinants of life insurance options in Croatia. The main indicators of the life insurance market development in Croatia (penetration of 0.8% of GDP and density of $104 per capita [Swiss Re 2014]) indicate a low level of development. The market was growing very fast, at double-digit premium growth rate, until 2009. However since then it has been declining [Croatian 2014]. There are no official statistics on life insurance lapse in Croatia. However according to the data of the insurance companies’ representatives, life insurance lapse started to increase with the financial and economic crisis. In 2009 the

Economics and Business Review, Vol. 1(15), No. 3, 2015

number of lapsed policies was 20 to 30 percent higher in comparison to previous year [Banka 2010]. This empirical research is based on the survey data collected through questionnaires addressed to a sample of 113 respondents, applying Chi-square, the Mann-Whitney U test, as well as a t-test, depending on the characteristics of variables and logistic regression. The results of the research show that amongst the demographic features of policyholders, an important determinant of using life insurance options is the number of children. Influential economic factors are the income level and financial status of policyholders whilst amongst life insurance characteristics, the duration of the policy has significance. The paper will contribute to the literature of life insurance options, analysing the determinants of lapse in a less developed life insurance market and it is the first paper that investigates life insurance options in Croatia. Additionally and differently from the existing empirical studies, this research is based on survey data. The rest of the paper is structured as follows. Section 2 discusses the effects and determinants of the exercise of life insurance options. The review of empirical literature is given in Section 3. Data and methodology follows in Section 4. The results are presented in Section 5, whilst Section 6 concludes.

1. Theoretical considerations on the effects and determinants of life insurance lapse As an embedded option, lapse has implications on many aspects of life insurers’ business, from liquidity and management of underwriting risk to profitability and solvency. In order to provide cash flow to policyholders who request lapse, the insurers have to sell the assets. In cases of unfavourable market conditions the assets can be liquidated at low value, which could produce loss for the insurers. An additional effect on the profitability of life insurers results from a more conservative investment strategy that might be used in order to ensure an adequate level of liquidity, which will reduce the return [Elling and Kiesenbauer 2013]. The insurers will not receive the cash flow from lapsed life insurance policies and that additionally affects their liquidity and profitability. According to the results of the research of long-term insurance lapse of Pinquet, Guillén, and Ayuso [2011], policyholders who lapsed policies have better health histories in comparison to those who did not cancel the contracts. The lapse could affect the underwriting risk of insurer. With lapse the structure of the policyholders becomes unfavourable for insurers since those who have a worse health record will have a higher share in the portfolio of insured risks. Thus the problem of adverse selection could occur. Additionally, refer-

M. Ćurak, D. Podrug, K. Poposki, Policyholder and insurance policy features

ring to EIOPA and CEA (Insurance Europe) Eling and Kohanski [2013], they emphasize the effect of lapse on solvency. Selling assets at lower value reduces the net value of the insurance companies. Although, according to the theoretical and empirical literature on life insurance lapse (for comprehensive review of the literature, see Eling and Kohanski [2013]), macroeconomic and insurance-specific factors could determine lapse [Dar and Dodds 1989; Outreville 1990; Kuo, Tsai, and Chen 2000; Kim 2005a; 2005b; Cox and Lin 2006; Kiesenbauer 2012] this paper is focused on the policyholder and features of the life insurance contract encompassing demographic characteristics and the health status of policyholders, the economic characteristics of policyholders as well as specific features of life insurance policies. Demographic factors include age, gender and number of children of the policyholder. The age could have both direct and indirect influence on lapse. The younger individuals are usually new policyholders. Since lapse in the early years of a policy results in higher loss, the policyholders abstain from exercising policy options. Supposing that those who are older contracted the life insurance much earlier, they have less to lose with the lapse. Consequently, they take more options. The indirect effect of age on lapse results from differences in income during the life cycle. Because at the beginning of working life and fixed-term work contracts, young people often have a less stable income and they are not able to pay insurance premium regularly. In contrast, older individuals have a more constant income stream which is the result of work experience and a long-term job with the same employer. Consequently they use lapse less in comparison to younger individuals. Gender affects risk aversion and consequently could be a determinant of life insurance lapse. Due to social and economic factors females are usually more risk averse in comparison to males. They usually demand life insurance if they are sure of their ability of pay insurance premium. Therefore women use lapse less than men. Considering the number of children, families with a higher number of children usually have less income at their disposal. In times of financial problems they will cut the burden of life insurance premiums and take the opportunity of lapsing more often in comparison to families with a lesser number of children. Amongst other factors related to policyholder characteristics, their health condition could determine life insurance lapse. Health condition influences health care expenditure and in case of bad health the policyholder will be forced to lapse a life insurance policy. This factor could gain in importance in the situation of an aging population, increasing medical treatment expenditure and the reducing coverage of publicly financed health care. When taking into consideration the income level, policyholders with lower and an unstable level of income are more exposed to financial troubles which

Economics and Business Review, Vol. 1(15), No. 3, 2015

make them unable to pay an insurance premium, in comparison to those with a higher and stable income. The financial burden could be result of being fired, problems with loan repayment or other liquidity problems. Thus a change in the financial status of the policyholder could result in opting for lapse (emergency fund hypothesis) as a way of alleviating financial problems. Since annuity insurance provides periodically an amount of money during the specific period or for whole life, which makes this type of insurance important source of liquidity in the case of the bad financial status of policyholder, it is expected that policyholders of annuity insurance will exercise lapse less in comparison with policyholders of other types of life insurance policies. Amongst distribution channels, agents usually cause higher lapses. Motivation for high intermediate commission leads agents to sell life insurance using an aggressive selling strategy and malpractice to individuals who often do not have adequate knowledge of life insurance. The agents do not take into account the individual’s needs but their own benefits. Since the commission for the first year is much higher in comparison to renewal commission, the agents are more focused on acquiring a new policyholder rather than then taking care of existing one [Surana and Gaur 2013]. This could results in lapses of life insurance policies. A longer duration of policy implies a higher probability of liquidity problems of the policyholder which could affect his/her ability to pay insurance premium and then opting for lapse. Thus the policy duration could be an influential factor in life insurance lapse. Features of a life insurance policy related to the frequency of premium payment, insurance premium size and the value of the policy are additional potential factors that could influence life insurance lapse. A higher frequency of premium payment could cause an increase in lapse. A more frequent payment of insurance premium, such as monthly payment may impose financial difficulties for the policyholder in comparison with a yearly payment. The same is true for the amount of insurance premium. A higher insurance premium could cause financial troubles for policyholders making them to opt for lapse more often than in the case of a lower life insurance premium. As the value of the life insurance policy increases the policyholder could lose a greater amount of money in the case of lapse. Thus it is to be expected that the value of the insurance policy and lapse rate are negatively correlated.

2. Review of empirical studies The first empirical study of the influence of policyholder and contract features on life insurance lapse is made on a sample of the life insurance contracts of seven life insurers in the United Kingdom in 1976. Applying logistic regres-

M. Ćurak, D. Podrug, K. Poposki, Policyholder and insurance policy features

sion and binomial model, Renshaw and Haberman [1986] find that the policyholder age, duration of the contract, type of life insurance policy and insurance companies influence lapse. Kagraoka [2005] researches lapse of annuity contracts of a single insurance company in Japan in the period 1993 to 2001. The author uses the Poisson model and a negative binomial model. The results confirm the importance of change in unemployment rates and the time elapsed from the contract date. The lapse and surrender of life insurance contracts of the large Italian bancassurance company from 1991 to 2007 with a Poisson modelling approach is analysed by Cerchiara, Edwards, and Gambini [2009]. According to the results the use of life insurance options is determined by the contract duration and type, the age of policyholder and the calendar year of exposure. Milhaud, Loisel, and Maume-Deschamps [2010] analyse endowment policies of a single insurance company in Spain. The study covers the period 1999 to 2007 and it is based on the classification and regression tree (CART) and logistic regression models. The policyholder and life insurance contracts characteristics investigated are policyholder age, contract duration, type of life insurance, sum insured, risk premium and saving premium. The results show a statistical significance of contract duration and profit benefit option. The determinants of lapse in German insurance market are investigated by Eling and Kiesenbauer [2011] covering endowment, annuity and term life insurance policies of a single insurance company in the period between 2000 and 2010. The authors apply Poisson, binominal and negative binominal models. The research shows that product type and policy duration are contract features important for lapse. Amongst the policyholder characteristics that influence the exercise of life insurance options, age and gender are significant. Canadian Institute of Actuaries [2014] analyses lapse for fully-guaranteed individual renewable and convertible 10-year term insurance policies. The analysis is based on data of ten companies in the period 2005–2010 and on percentage calculation and comparison. The analysis is done according to both the number of policies and their face value. The research encompasses the following factors: age, gender, amount, policy duration, smoking status, mortality rating, preferred underwriting classification, joint/single, policy structure (stand-alone or rider), payment frequency, mode of payment (pre-authorized or other) and province. For some of the analysed factors there is a variation of the lapse rates amongst the number and value of the policies.4 The empirical research of policyholder and life insurance contract characteristics is limited to the studies reviewed above. All the studies are based on a sample of insurance companies, mostly on a single company and all of 4

Because of the many factors that are analysed in this study, as well as the differences in the results amongst the number and amount of policies, we do not present detailed results.

Economics and Business Review, Vol. 1(15), No. 3, 2015

the studies refer to developed life insurance markets. This paper applies a different methodology and it encompasses a less well-developed life insurance market.

3. Data and methodology The analysis is based on the survey data collected through questionnaires on a sample of 113 respondents in Croatia. The sample was formed randomly. We used an indirect approach to the respondents. Specifically we used a web questionnaire. The information on the questionnaire was sent by e-mail to many randomly selected e-mail addresses. The questionnaire was anonymous. The sample consisted of both individuals who used life insurance options and those who did not. All participants are separated in two groups, depending on their usage of the options. There are 45 respondents who exercised life insurance options whilst the number of those who did not lapse is 68. The respondents that used life insurance options were placed in the first group. Their questionnaire consisted of 19 questions referring to the influence of demographic and economic factors as well as policy features on life insurance options. The first part of the questionnaire was related to questions on policyholder age, gender, number of children, health status, income level and change in his/her financial status. The second section was focused on questions about life insurance options and policy features, such as policy type, distribution channel, duration of contract, frequency of premium payment, premium amount and sum insured. The second group of respondents are policyholders who didn’t use life insurance options. Their questionnaire had 13 questions about policyholder characteristics and policy features. Respondents were also asked to answer questions about the change in their health and financial situation from the time of buying a life insurance policy. The first analysis of the survey data is based on statistical tests. Depending on the variable characteristics, Chi-square, Mann-Whitney U test, as well as ttest are used. The Mann-Whitney U test is used to analyze ordinal scaled variables, whilst the t-test is applied in analysing quantitative variables that are not ordinally scaled. Chi-square is used to analyze whether distributions of categorical variables differ from one another. In order to check all factors that could influence the life insurance options, the data are additionally analysed using binary logistic regression and the backward: Wald method. To identify key determinants of life insurance options a dichotomous variable is computed, indicating whether the life insurance option is used or not. That is, life insurance options = 1, if life insurance option is used; 0, if life insurance options is not used. On the basis of Pearson’s Chi-square statistic, the predictors are determined as to whether they are associated with life insurance options.

M. Ćurak, D. Podrug, K. Poposki, Policyholder and insurance policy features

4. Results The first set of analysed factors referred to the demographic characteristics of policyholders. The relationship between the age and life insurance options is presented in Table 1. Table 1. Relationship between age and life insurance options Count Age

Life insurance options

Total

Yes

18–30

31–40

41–50

51–60

61–70

113

Total

Table 2. Results of the Mann-Whitney U test for variable age Age Mann-Whitney U

1167.000

Wilcoxon W

3513.000

Z Asymp. Sig. (2-tailed)

–2.243 0.025

In the questionnaire the question considering the age variable was formulated in a way that we proposed 5 classes and the respondent had to choose in which class he/she belongs. The largest number of respondents (43) belongs to the age group 31–40. The smallest number of respondents (2) is in the age group 61 and older. The most numerous users of life insurance options are individuals in the age group 41–50 years whilst the number of those who use and do not use life insurance options in the age group 51–60 is almost the same. In order to determine the link between age and life insurance options, the Mann-Whitney U test was conducted and the results obtained are presented in Table 2. An empirical significance of 0.025 indicates that there is a correlation between age and life insurance options. The result is in the accordance with the result of the empirical study of life insurance options in Italy done by Cerchiara, Edwards, and Gambini [2009] as well as the study of Eling and Kiesenbauer [2013] based on the German insurance market.

Economics and Business Review, Vol. 1(15), No. 3, 2015

Table 3. Cross-tabulation for gender Life insurance options Gender

Total

Yes

Male

Female

113

Total

Table 4. Chi-Square test for gender Value

Asymp. Sig. (2-sided)

0.332

0.564

Continuity Correction

0.146

0.702

Likelihood Ratio

0.332

0.565

0.329

0.566

Pearson Chi-Square b

Fisher’s Exact Test Linear-by-Linear Association N of Valid Cases

113

The nexus between gender and life insurance lapse is presented in Table 3. The database consisted of 56.64% male and 43.36% female respondents. Approximately 60% of female respondents stated that they did not use life insurance options. An almost equal number of men and women use life insurance options. The correlation between gender and the usage of options is presented in Table 4. The empirical significance is 0.564. Thus the results of the Chi-Square test indicate that there is no correlation between gender and life insurance options. The question considering the number of children in the questionnaire was set in such a way that we proposed 4 categories: without children, 1, 2, and 3 Table 5. Cross-tabulation for the number of children Life insurance options Number of children

113

3 or more Total

Total

Yes

M. Ćurak, D. Podrug, K. Poposki, Policyholder and insurance policy features

Table 6. Results of Mann-Whitney U test for variable number of children Number of children Mann-Whitney U

1202.000

Wilcoxon W

3548.000

–2.055

Asymp. Sig. (2-tailed)

0.040

or more children. Table 5 shows the relationship between number of children and life insurance options. The largest number of the respondents in this survey has 2 children. Only 7 out of 31 individuals with no children used life insurance options. An equal number of respondents with 3 or more children used/ did not use life insurance options. Considering those who used lapse, 33.33% of them have no children or have 1 child, whilst 66.67% are those who have 2 children. The results of Mann-Whitney U test of link between number of children and life insurance options are shown in Table 6. The result shows that the number of children is a significant determinant of life insurance lapse. This Table 7. Cross-tabulation for health status Life insurance options Health status

Significant changes Without changes

Total

Yes

108

113

Table 8. Chi-Square test for health status Value

Asymp. Sig. (2-sided)

0.889

0.346

Continuity Correction

0.226

0.634

Likelihood Ratio

0.865

0.352

0.881

0.348

Pearson Chi-Square b

Fisher’s Exact Test Linear-by-Linear Association N of Valid Cases

113

Economics and Business Review, Vol. 1(15), No. 3, 2015

confirms that individuals with more children use more life insurance options which is in accordance to theoretical considerations. Health status is the next factor whose influence on life insurance lapse is analyzed with the results presented in Table 7. Variable health status was tested in a way that respondents who did not use options had to state whether their health status has changed significantly. For respondents who have used options, the question was adapted in such a way that they were asked whether their health status had changed significantly by the time of using the options. There are 108 respondents who didn’t have any significant changes in health status. Only 5 respondents claimed that they had changes in health status and the number is almost equal for those who used and those who did not exercise life insurance options. The results of the Chi-Square test are shown in Table 8, indicating that there is no relationship between health status and life insurance options. In the questionnaire income was divided into 7 categories: without income; to 3,000 kn; 3001–5000 kn; 5001–7000 kn; 7001–10000 kn; 10001–12000 kn; more than 12 000 kn. The link between the life insurance options and income level is presented in Table 9. The majority of most respondents have an income Table 9. Cross-tabulation for income level Count

Life insurance options

Total

Yes

Income no income level (HRK) to 3000

3001–5000

5001–7000

7001–10000

10001–12000

more than 12000

113

Total

Table 10. Mann-Whitney U test for income level Income level Mann-Whitney U

1266.500

Wilcoxon W

3612.500

Z Asymp. Sig. (2-tailed)

–1.583 0.014

M. Ćurak, D. Podrug, K. Poposki, Policyholder and insurance policy features

of 7001–10000 HRK and the smallest number of respondents has no income. The respondents who use options mostly belong to the lower income groups. When considering 45 respondents who used options, 22 of them have monthly incomes of 3001–5000 HRK. There is almost no user of life insurance options in the higher income group. This is consistent with the hypothesis that individuals belonging to higher level income groups are more able to pay premium regularly and less forced to exercise options in comparison with those who have lower income. Based on the results of the Mann-Whitney U test shown in Table 10 there is a significant relationship between income level and life insurance options. Table 11. Cross-tabulation for change of financial status Life insurance options Change of financial status

Total

Yes

113

Total

Table 12. Chi-Square test for change of financial status Value

Asymp. Sig. (2-sided)

Pearson Chi-Square

13.826

0.001

Continuity Correctionb

12.238

0.002

Likelihood Ratio

13.729

0.001

13.703

0.001

Fisher’s Exact Test Linear-by-Linear Association N of Valid Cases

113

The question related to the change in the financial situation was set in such a way that the respondents were asked if they had any problems with liquidity. The question was adapted for those respondents who have used options who then had to state if they had any problems with liquidity at the time of using the options. Table 12 shows the results of the Chi-Square test for the change of financial status. The empirical significance is 0.001 confirming that there is a statistically significant correlation between life insurance options and change of financial status. Analyzing the respondents who used life insurance options, 62.22% of them experienced changes in their financial situation from the time

Economics and Business Review, Vol. 1(15), No. 3, 2015

Table 13. Cross-tabulation for policy type Life insurance options Policy type

Term insurance Endowment assurance Annuity insurance

Total

Yes

113

Table 14. Chi-Square test for policy type Value

Asymp. Sig. (2-sided)

Pearson Chi-Square

0.051

0.975

Likelihood Ratio

0.050

0.975

Linear-by-Linear Association

0.034

0.855

N of Valid Cases

113

they bought life insurance. Amongst those who did not use options even 94.11% of them did not experience any change of their financial condition. In respect of the type of policy the respondents own they were asked to choose amongst term insurance, endowment insurance or annuity insurance. The majority of respondents (89) have an endowment assurance policy, which is the most popular type of life insurance in Croatia (its share of total life insurance premiums is 75% [Croatian Insurance Bureau 2014]). Those who have aÂ term insurance policy are 17.5% of the total number of respondents. Only 5 respondents have annuity insurance policy. The correlation between the policy type and life insurance options is presented in Table 14. The empirical significance of 0.975 indicates that policy type does not have an influence on life insurance options. For the variable of distribution channel the respondents were asked through which channels they had bought their life insurance policy, broker, agent or bancassurance. The data on the relationship between the distribution channel and life insurance options is presented in Table 15. The largest number of respondents (94) purchased life insurance through an agent. The smallest number of respondents (7) used aÂ broker. The number of respondents who bought life insurance policy through bancassurance and who used life insurance options is twice the number of those who used the same distribution channel and

M. Ä&#x2020;urak, D. Podrug, K. Poposki, Policyholder and insurance policy features

Table 15. Cross-tabulation for the distribution channel Life insurance options DistriÂ bution channel

Total

Yes

Broker

Agent

113

Bancassurance

Total

Table 16. Chi-Square test for the distribution channel Value

Asymp. Sig. (2-sided)

Pearson Chi-Square

4.159

0.125

Likelihood Ratio

4.074

0.130

Linear-by-Linear Association

1.994

0.158

N of Valid Cases

113

did not exercise life insurance options. Table 16 presents the results of the ChiSquare test for variable distribution channels. Based on the result the variable distribution channel is not statistically significant. The next question the respondents were asked was related to the frequency of premium payment (monthly, quarterly, half-yearly annually). The largest number of respondents (48) pays monthly premium, whilst the smallest number of respondents (11) pays their life insurance premium half-yearly. Table 18 presents the results of the Chi-Square test for frequency of payment. The value of empirical significance is 0.389 and it shows that frequency of payment is not statistically significant linked to life insurance lapse.

Table 17. Cross-tabulation for frequency of premium payment Life insurance options Frequency monthly of payquarterly ment half-yearly yearly Total

Total

Yes

113

Economics and Business Review, Vol. 1(15), No. 3, 2015

Table 18. Chi-Square test for frequency of premium payment Value

Asymp. Sig. (2-sided)

Pearson Chi-Square

3.016

0.389

Likelihood Ratio

3.173

0.366

Linear-by-Linear Association

0.705

0.401

N of Valid Cases

113

Table 19. Group statistics for insurance value Life insurance options

Mean

Std. Deviation

Median

Minimum

Maximum

Yes

118414.91

101976.363

87126.00

7600

456000

101255.92

72269.925

87400.00

4500

311600

Total

108089.14

113

85328.884

87126.00

4500

456000

Table 20. Test statistics for insurance value

Insurance value

Equal variances assumed

Sig.

Sig. (2-tailed)

1.513

0.222

1.403

101

0.164

Table 21. Group statistics for premium amount Life insurance options

Mean

Std. Deviation

Median

Minimum

Maximum

Yes

4030.67

4507.706

3000.00

120

17741

2881.19

3425.947

1862.00

100

18000

Total

3338.94

113

3914.512

2158.00

100

18000

Table 22. Test statistics for premium amount

Premium amount

Equal variances assumed

Sig.

Sig. (2-tailed)

2.563

0.112

1.537

111

0.127

M. Ä&#x2020;urak, D. Podrug, K. Poposki, Policyholder and insurance policy features

In the questionnaire respondents were asked to provide information on the value of their life insurance. The relationship between the insurance value and life insurance options is presented in Table 19. The average insurance value for respondents who used life insurance options is 118,414.91 kn whilst the average insurance value for respondents who didnâ&#x20AC;&#x2122;t use life insurance options is 101,255.92 kn. Thus the respondents who use life insurance options have aÂ policy of higher value. However, according to the test, there is no correlation between the insurance value and life insurance options. Additionally the respondents were asked about the amount of their premiums. The average premium for respondents who used life insurance options Table 23. Cross-tabulation for the duration of contract Life insurance options

Number Duration of the contract

Total

Yes

113

Economics and Business Review, Vol. 1(15), No. 3, 2015

is 4,030.67 kn, and the average premium amount for respondents who did not use life insurance options is 2,881.19 kn. Therefore respondents who used life insurance options paid a higher premium than respondents who did not use life insurance options. Table 22 presents the results of the t-test to analyse the link between premium amount and life insurance options. The empirical significance is 0.127. Thus it can be concluded that there is no statistically significant difference in premium amount between those who used and those who didn’t use life insurance options. The final question in the questionnaire was related to the contract duration. According to the data in Table 23, 31.11% of respondents who used the options of life insurance had a life insurance policy of 20 years duration, whilst the number of those respondents with shorter contracts who used options is insignificant. The results of the t-test referring to the link between the duration of the contract and life insurance options are shown in Table 24. Since the empirical significance is 0.029, the duration of the contract has an influence on life insurance options. A longer duration of the insurance contract may increase the probability that the owners of life insurance policies will experience financial difficulties and consequently use more life insurance options. The result is in accordance with the theoretical assumption. The same results are obtained by Cerciara et al. [2009], Milhaud et al. [2010] and Eling and Kiesenbauer [2011]. In table 25 original variables are crossed with predictions. In 62.2% of cases respondents who used life insurance options the predictions correspond to the actual situation. In 82.4% of cases respondents who did not use life insurance options, the predictions correspond to the actual situation. Table 24. Test statistics for the duration of the contract

Duration of the contract

Equal variances assumed

Sig.

Sig. (2-tailed)

4.179

0.043

–2.211

111

0.029

Table 25. Classification table Predicted Observed Life insurance options

Life insurance options Yes

Percentage Correct

yes

62,2

82,4

Overall Percentage

74,3

M. Ćurak, D. Podrug, K. Poposki, Policyholder and insurance policy features

Table 26. Results of logistic regression B

S.E.

Wald

Sig.

Exp(B)

Number of children

–0.373

0.224

2.764

0.096

0.689

Income level

–0.304

0.150

4.088

0.043

0.738

Change of financial status

1.642

0.494

11.055

0.001

5.167

Duration of the contract

0.059

0.029

4.077

0.043

1.060

–1.240

1.149

1.166

0.280

0.289

Constant

According to the result of the logistic regression the most influential factor of life insurance options is a change in financial status. Financial difficulties related to loss of job, problems with repayment of loans, etc., lead to problem of liquidity for policyholders when, the payment of insurance premiums becomes a burden forcing them to use life insurance options. The number of children is a significant determinant of life insurance lapse. Families with a greater number of children usually have less disposable income and using life insurance options is a way of reducing the burden of life insurance premiums. The income level is confirmed as an important factor of life insurance options. Individuals in higher income groups are more able to pay insurance premiums regularly and are less obliged to exercise an option in comparison with those who have a lower income. The duration of the contract also has an influence on life insurance options. A longer duration of the insurance contract may increase the probability that the policyholders will experience financial difficulties and consequently use more life insurance options. The results of the logistic regression analysis almost confirm the results of the test. A comparison of the results of the multivariate regression analysis and the results of the tests, show that there is only a variable of policyholder age, having a significant influence on life insurance options according to the analysis which is not of significance when checking all factors. In summary, amongst the characteristics of the policyholders analysed as well as the features of life insurance contracts, the number of children, income level, financial status and duration of the life insurance policy are confirmed as statistically significant factors of life insurance lapse. Although the research is limited by the size of the sample due to the unavailability of data, the research is the first analysis of the drivers of life insurance options in Croatia and the attempt to emphasize the importance of the management of embedded options in a life insurance contract as well as the importance of making data on life insurance options publicly available. This is an important prerequisite for further research.

Economics and Business Review, Vol. 1(15), No. 3, 2015

Conclusions This research provides an understanding of the key lapse determinants in Croatian life insurance markets. Based on survey data of 113 respondents the analysis shows that amongst the demographic features of policyholders an important determinant of using life insurance options is the number of children. Influential economic factors are the income level and financial status of policyholders whilst amongst the characteristics of life insurance the duration of the policy has significance. The results of the research have implications for both insurance companies and regulators. Since insurance options affect the liquidity and profitability of the life insurer it is important for decision makers at insurance companies to manage the risk of the options, taking into consideration the factors influencing lapse. Since life insurance options are an important risk for insurance companies which affects solvency, regulators should consider the determinants of life insurance lapse in the context of risk models. Additionally, in order to reduce the problem of information asymmetry between policyholders and insurance companies, policyholders should be provided with adequate information of the detrimental consequences of lapse at the moment of purchasing insurance. This would be helped by a greater understanding of insurance policy features amongst potential policyholders. For further research it is important to have publicly available data at industry and company level that would be provided by the supervisory authority or insurers’ association. The data would provide a basis for the analysis of the macroeconomic and company-specific drivers of insurance options.

References Banka, 2010, Životno osiguranje – ne isplati se ni otkup police ni zajam, http://www. banka.hr/hrvatska/zivotno-osiguranje-ne-isplati-se-ni-otkup-police-ni-zajam [access: 31.03.2015]. Canadian Institute of Actuaries, 2014, Lapse Experience Study for 10-Year Term Insurance, http://www.cia-ica.ca/publications/publication-details/214011 [access: 25.03.2015]. CEIOPS, 2010, QIS5 Technical Specifications, Annex to Call for Advice from CEIOPS on QIS5, http://ec.europa.eu/internal_market/insurance/docs/solvency/qis5/201007/ technical_specifications_en.pdf [access: 27.02.2015]. Cerchiara, R.R., Edwards, M., Gambini, A., 2009, Generalized Linear Models in Life Insurance: Decrements and Risk Factor Analysis under Solvency II, Working Paper, AFIR Colloquium Rome, http://30923.vws.magma.ca/AFIR/Colloquia/Rome2/ Cerchiara_Edwards_Gambini.pdf [access: 27.02.2015]. Cox, S.H., Lin, Y., 2006, Annuity Lapse Rate Modeling: Tobit or not Tobit?, Society of Actuaries, http://library.soa.org/files/pdf/Cox%20Linn%20paper%2011-15-06.pdf [access: 27.02.2015].

M. Ćurak, D. Podrug, K. Poposki, Policyholder and insurance policy features

Croatian Financial Service Supervisory Agency, 2014, Insurance Statistics in 2013, http://www.hanfa.hr/EN/nav/106/statistics.html#section0 [access: 15.03.2015]. Croatian Insurance Bureau, 2014, Croatian Insurance Market 2013, http://www.huo. hr/eng/publications/50/ [access: 30.03.2015]. Dar, A., Dodds, C., 1989, Interest Rates, the Emergency Fund Hypothesis and Saving through Endowment Policies: Some Empirical Evidence for the U.K., Journal of Risk and Insurance, vol. 56, no. 3: 415–433. Dragos, S.L., 2014, Life and Non-life Insurance Demand: The Different Effects of Influence Factors in Emerging Countries from Europe and Asia, Economic Research – Ekonomska istraživanja, vol. 27, no. 1: 169–180. Elango, B., Jones, J., 2011, Drivers of Insurance Demand in Emerging Markets, Journal of Service Science Research, vol. 3, no. 2: 185–204. Eling, M., Kiesenbauer, D., 2013, What Policy Features Determine Life Insurance Lapse: An Analysis of the German Market, Journal of Risk and Insurance, vol. 81, no. 2: 241–269. Eling, M., Kochanski, M., 2013, Research on Lapse in Life Insurance – What Has Been Done and What Needs to Be Done?, The Journal of Risk Finance, vol. 14, no. 4: 392–413. Kagraoka, Y, 2005, Modeling Insurance Surrender by the Negative Binomial Model, Working Paper, http://www.musashi.jp/~kagraoka/research/NBM_017.pdf [access: 27.02.2015]. Kiesenbauer, D., 2012, Main Determinants of Lapse in the German Life Insurance Industry, North American Actuarial Journal, vol. 16, no. 1: 52–73. Kim, C., 2005a, Modelling Surrender and Lapse Rates with Economic Variables, North American Actuarial Journal, vol. 9, no. 4: 56–70. Kim, C., 2005b, Report to the Policyholder Behavior in the Tail Subgroups Project, Society of Actuaries, https://www.soa.org/Files/../06-Kim [access: 27.02.2015]. Kumar, J., 2009, Lapsation of a Life Insurance Policy, Bimaquest, vol. IX, no. 2: 38–44. Kuo, W., Tsai, C., Chen, W.-K., 2003, An Empirical Study on the Lapse Rate: The Cointegration Approach, Journal of Risk and Insurance, vol. 70, no. 3: 489–508. Milhaud, X., Loisel, S., Maume-Deschamps, V., 2010, Surrender Triggers in Life Insurance: Classification and Risk Predictions, Working Paper, http://isfa.univ-lyon1. fr/vmaume/sites/default/files/documents/lapse.pdf [access: 27.02.2015]. Outreville, J.F., 1990, Whole-life Insurance Lapse Rates and the Emergency Fund Hypothesis, Insurance: Mathematics and Economics, vol. 9, no. 4: 249–255. Pinquet, J., Guillén, M., Ayuso, M., 2011, Commitment and Lapse Behaviour in Long Term Insurance: A Case Study, Journal of Risk and Insurance, vol. 78, no. 4: 983–1002. Renshaw, A.E., Haberman, S., 1986, Statistical Analysis of Life Assurance Lapses, Journal of the Institute of Actuaries, no. 113, vol. 3: 459–497. Sliwinski, A., Michalski, T, Roszkiewicz, M., 2013, Demand for Life Insurance – An Empirical Analysis in the Case of Poland, The Geneve Papers on Risk and Insurance – Issues and Practice, vol. 38, no. 1: 62–87. Surana, S.S., Gaur, A.K., 2013, Lapsation of Policy: A Threat or Course for Life Insurance Industry, International Journal of Social Science & Interdisciplinary Research, vol. 2, no. 4: 91–94. Swiss Re, 2014, World Insurance in 2013: Steering towards Recovery, Sigma, no. 3/4.

Economics and Business Review, Vol. 1 (15), No. 3, 2015: 78–89 DOI: 10.18559/ebr.2015.3.6

Longevity risk and the design of the Polish pension system1 Marek Szczepański2

Abstract : The aim of this article is an analysis of the design of the Polish pension system in its benefits phase (decumulation of the collected pension capital) in the context of longevity risk management. The object of the research involves both the public pension system (including the latest legal amendments related to the collection and payment of pension capital from pension funds) and supplementary pension schemes: occupational pension schemes, individual retirement accounts and individual accounts for retirement security. The main question which the author addresses is whether the current Polish pension system is more resistant to the risk of longevity in its present legal and institutional design than the previous Polish public pension scheme, based on PAYG financing method and a defined benefit formula for the calculation of pension benefits. Keywords : the risk of a longer than expected life expectancy (longevity risk), average life expectancy, pension systems. JEL codes : J11, J14.

Introduction Over the last decades of the XXth and the first decade of the XXI century there have been unprecedented, and to some extent unexpected, increases in life expectancy [Cocco and Gomes 2001: 2]. The length of time people are expected to live in most OECD countries has increased by 25 to 30 years during the XXth century. This increase in life expectancy is the result of the progress of civilisation, improved working conditions, better medical care and lifestyle changes. But “improvements in mortality and life expectancy are uncertain. In this regard, longevity risk is associated with the risk that future mortality and life expectancy outcomes turn out differently from those expected” [Anatolin 1

Article received 22 October 2014, accepted 3 August 2015. Article prepared within the statutory research at the Faculty of Management Engineering of Poznań University of Technology. 2 Poznań Univeristy of Technology, Faculty of Engineering Management, Strzelecka 11, 60‑965 Poznań, Poland, marek.szczepanski@put.poznan.pl.

M. Szczepański, Longevity risk and the design of the Polish pension system

2007: 3]. Managing longevity risk became a global issue, approaches used in the EU countries and the USA are confronted with new, sometimes innovative ideas, implemented also in Asian countries [Roy 2012]. Scientists conducting research in the field of pension economics but also public institutions and providers of private financial service (banks, insurance companies, occupational pension schemes, asset management companies) are interested in institutional solutions (pension scheme design) as well as financial instruments which could help to manage the longevity risk, such as annuities, longevity-linked instruments, risk-sharing in Defined Contribution pension schemes, de-risking in Defined Benefit Schemes, etc. This article focuses on the issues of managing longevity risk in the reformed pension system in Poland and in particular – the construction of public and supplementary pension systems and their ability to adapt to the challenges associated with longevity risk. The basis for further consideration is the proper definition of longevity risk, which is not the same as the demographic risk related to the ageing of the population. What is this longevity risk, also defined as the risk of age length, Hull [2011: 82–88], and what is its essence? Longevity risk can be defined at both individual and aggregate levels. Individual longevity risk (sometimes referred to as specific longevity risk) is based on the fact that a person lives longer than expected. Such a risk may be associated with the premature exhaustion of savings or improper distribution of investments in time [Stallard 2006; Pitacco et al. 2009]. Individual longevity risk, the presence of which can carry severe negative consequences for individuals, does not present any danger to the financial stability of pension systems. There is also an aggregate longevity risk, sometimes called the risk of trend, that affects the entire population. It consists of the fact that in a given population, the average life expectancy will be longer than expected. In other words it is the risk of incorrect estimates of future trends in mortality rate. Together both specific and aggregate longevity risks form total longevity risk [Blake Burrows 2001]. For pension systems the aggregate longevity risk is particularly important. The risk of longevity, which refers to the phase of paying out pension benefits (pension capital decumulation) affects both public pension systems as well as the supplementary pension schemes (occupational or individual) that provide benefits for life (annuities). The degree of vulnerability of pension systems to longevity risk depends on their structure, and especially the methods of financing and the pension formula utilised (method of calculating benefits). Today in economically developed countries demography risk is associated with the process of demographic aging (population growth of old people in relation to the working generation and is associated inter alia with decreasing average fertility rates and increases in life expectancy). The demographic risk in the pension systems results from the need to finance current benefits paid

Economics and Business Review, Vol. 1(15), No. 3, 2015

by the current income. Demographic risk particularly affects those pension systems which are wholly or predominantly based on generational contract (PAYG), or the financing of current pension benefits from current contributions or taxes obtained from the working generation. But also funded pension schemes are not completely immune to demographic risk. Generally it can been stated that demographic risk arises from the fact that successive generations are living longer. Longevity risk relates to those individuals or demographic groups, which live longer than expected. Without going into further details regarding the issues of the financing of pension systems, which already C do not provide self-financing and require subsidies from the state budget in most economically developed countries, it is worth drawing attention to the fact that social security systems in relation to the increased demographic risk have and will continue to have serious problems in meeting their obligations to people who will continue to live longer on average, but who will not exceed the average life expectancy for a given age group in demographic forecasts. Meanwhile, in each demographic group a proportion of people live unusually long. This group is mostly affected by the risk of longevity.

1. Longevity risk in pension systems – a model approach In an attempt to answer the question whether the structure of the pension system in Poland – with regard to recent statutory changes3 – provides adequate and sufficient protection against longevity risk, it is worth recalling the definition of social risk. The risk of old age, like some other types of social risks (e.g. linked to childbirth) can be recognized not only as a threat, but also as an opportunity. The very fact of living up to the statutory retirement age and a long lifetime after passing that milestone is of course a good thing, like the birth of a child in a family. The risk of old age can be considered as a personal risk in addition to other types of social risks – such as the death of a breadwinner, an illness, disability, unemployment, maternity (and more precisely – its financial implications), accidents at work and occupational diseases [Mierzejewska 2005: 210]. Social risk associated with old age refers to the possibility of a decreasing household income after the contractual threshold of old age, which is determined by the statutory age of retirement. In English literature the risk of old age is de 3

An Act of 6 December 2013 amending certain acts in relation to the definition of principles for the payment of pensions from funds collected in open pension funds, Journal of Laws 2013, pos. 1717 introduced far-reaching changes in the structure of the public (base) pension system in Poland and drastically limited the participation of the second pillar (private pension funds) in total pension security in old age.

M. Szczepański, Longevity risk and the design of the Polish pension system

fined as “the risk of insufficient income during retirement” [Rejda 2001: 8] or “the risk of financial dependency in old age” [Vaughan and Vaughan 2001: 10]. In Poland, as in most European countries, the pension system involves the use of the insurance risk management method. The payment of benefits depends on pension contributions made equally by employees and employers during their working lives. This applies to the occupational (public) pension system. In separate pension systems for the armed forces, judges and prosecutors, pensions are financed by a supply method from general taxes. For the management of longevity risk it is particularly important to properly define the risks of old age covered by pension security. In Polish literature this is aptly illustrated by a model of Tadeusz Szumlicz (see Figure). SAVINGS O = PR + C + W A

INSURANCE

retirement age

Longevity risk average further life expectancy

A – adoption of occupational activity O – total retirement savings PR + C + W – the possible forms of retirement savings (pension rights + pension capital + wealth)

Old age risk characteristics

Source: Based on: [Szumlicz 2005: 242]

Using a model approach to the risk of old age, the longevity risk can be placed in an individual’s third cycle of life. Considering unitary and individual terms (microeconomic level), the risk of old age in the first phase (accumulation) lies in the fact that a person does not gather sufficient retirement savings, and in the second phase (from the age of retirement until the end of the average life expectancy) that the accumulated savings provide too little income. In the third phase, for people living longer than expected, in addition to the risk of low income (e.g. low level of pension benefits offered by the public pension system) there still exists the risk of the partial or total exhaustion of any additional accumulated resources (e.g. in an individual or occupational pension plan, in other forms of savings, etc.), namely the implementation of individual longevity risk. Individual longevity risk does not occur in pension systems providing benefits in the form of an annuity. In the case of life annuity, the funds in a participant’s account are converted to a series of lifetime benefits. If a participant dies, there is no inheritance, because at the moment of retiring the individual account is liquidated and the individual equity feeds the insurance fund [Otto

Economics and Business Review, Vol. 1(15), No. 3, 2015

and Wiśniewski 2013: 23]. In contrast, the aggregate longevity risk (the risk of trend) begins to affect institutions paying out benefits for life and which also ensure the indexation of benefits. If a public pension system, in addition to providing pensions in the form of an annuity made available another retirement product: guaranteed payment of benefits after retirement – decapitalizing an individual retirement account within a specified period such as 15 years (with the possibility of inheritance when a given person dies earlier), also the public system would cover the risk of the exhaustion of accumulated pension capital for people living longer than expected. Thus it would be possible to realize the individual longevity risk. As a matter of fact the implementation of social risks is easier to cope with for wealthy and very wealthy individuals rather than people with lower incomes. Others, however, cannot be left alone with this type of threat posed to the resources of their households. This applies particularly to the elderly. Therefore most countries of the world have utilized ( for more than a hundred years) social security systems which include pensions. The problem is that these can operate more or less effectively, provide full or only partial and insufficient protection against social risk and in this case – the risk against longevity associated with the risk of old age. This depends largely on the structure (legal and institutional solutions and methods of funding) of public supplementary pension systems as well as methods of forecasting further life expectancy, which form the basis for the calculation of the amount of pension benefit.

2. Longevity risk in the Polish pension system The year 1999 marked the introduction of a comprehensive, systemic pension reform, one of its main objectives being the division of risk between the financial and the labour markets by introducing a three-pillar structure, and in particular, a second capital funded pillar and private pension funds (called “OFE”) operating within it. A mixed PAYG-funded scheme was created. In this scheme, the first pillar (administered by the Social Insurance Institution) is financed by current pension contributions of the working generation (the so-called generational contract) and the second pillar is composed of the pension savings of the working generations invested in the financial market. Contributions for a pension in Poland are relatively high and amount to 19.52% of gross income, of which the employer pays 9.76% and the remaining 9.76% by the employee. In the accumulation phase of pension capital the accumulated capital is recorded in the form of pension benefits (the first individual retirement account in the Social Insurance Institution confirming the state’s commitment to pay future benefits, which will be financed by successive generations of workers) and financial capital (assets accumulated in pension funds, recorded in the

M. Szczepański, Longevity risk and the design of the Polish pension system

second individual account, with coverage in financial instruments of a specified market value purchased with part of the contributions transferred to the second pillar). Diversifying sources of future pension funding was supposed (according to the reform’s creators) to reduce the risk to the long-term stability of the pension system. Whilst the first pillar (PAYG) is more sensitive to the risk of demographics which increases with the ageing of the population (an increase in the number of people receiving pension benefits at retirement age in relation to the contributors of working age) the financed pillar is subject to different (demographically non-correlated) kinds of risk (including investment risk). Both the PAYG and the funded pillar are not immune to aggregate longevity risk. A more comprehensive summary of the different types of risk in PAYG and fully funded pension schemes are presented in Table. Comparison of the pay-as-you-go model and the fully funded one in terms of common types of risk Threats ––demographic risk ––low level of occupational activity and high unemployment ––moral risk manifesting itself in a tendency of a premature occupational deactivation in order to gain benefits ––political risks associated with the unjustified redistribution of resources in order to gain political support of advantaged groups ––longevity risk

––demographic risk ––(less than that for PAYG) ––capital market crisis ––inadequate investment policy ––high inflation ––longevity risk

Source: Based on: [Jurek 2011: 7].

The second significant change was the replacement of the formula for calculating benefits – a transition from a defined benefit (DB) to a defined contribution system (DC), leading to the individualization of benefits (equivalence of benefits in relation to the contributions, a departure from the redistribution of income in a given generation of retirees). Diversification, however, was not only to apply to the phase of capital accumulation, but also the phase of its consumption (decumulation), which carries the risk of longevity. According to the initial assumptions of the pension reform of 1999 the payment of benefits from capital accumulated in the second pillar of the pension system was to be dealt with by pension institutes (created especially for this purpose), which would not only pay benefits under the second pillar but also multiply the accumulated capital and invest it in the low risk financial instruments. However such pension institutions never came to existence. For 15 years the pension reform has not been completed because there was no legislation regarding the payment of pensions from the second

Economics and Business Review, Vol. 1(15), No. 3, 2015

pillar. Only recently has enacted legislation been enacted (Act of 26 December 2013) to finally regulated this important issue. The payment of the total pension funds accumulated in the first and second pillar will lie in the hands of the Social Insurance Institution. A lifetime pension (annuity) will remain the only available product. What is more the funds accumulated in the second pillar in the phase of capital accumulation will be gradually transferred to the Social Insurance Institution 10 years before a person retires (these funds will cover the current payment of benefits for the previous generation of retirees) in exchange for pension rights recorded in a participant’s account on a special sub-account valorized according to different regulations than the rights recorded in the first pillar. Without going into any detailed evaluation of the new solution for the gradual transfer of pension capital from private pensions funds to the Social Insurance Institution (the so-called safety slide), which has been subjected to critical analysis in a separate study Szczepański [2013, p.168–170], it is worth formulating a few questions regarding the management of longevity risk: –– What are the advantages and disadvantages of entrusting all pension payments to one state institution (the Social Insurance Institution) from the point of view of managing longevity risk? Will it increase or rather decrease the safety of payments? Will an alternative solution – to create a competitive solution on the market of pension payments – be more risky and more expensive for retirees? –– Is the selected retirement product – a lifetime pension correct from the point of view of managing longevity risk? –– Is the use of tables later of life expectancy common to both men and women (so called unisex tables) for the calculation of pension benefits in the new, reformed pension system the right solution? –– What were the alternative solutions? Entrusting all pension payments from the first and the second pillar to a state institution (the Social Insurance Institution) means the elimination of the competition mechanism, which could occur in the phase of benefit payments if the benefits of the second pillar were to be paid out by specialized institutions (pension institutes) or non-specialized financial institutions, which would deal with this in addition to other tasks (private pension funds, life insurance companies, investment funds and other entities). However is competition necessary in the phase of payments? Would the maintenance of the process of benefit payments from the second pillar by competing private financial institutions result in an excessive cost of such maintenance? Undoubtedly the advantage of ceding all payments to one institution (in this case the Social Insurance Institution having years of experience and an appropriate basis for the payment of pension benefits) makes it possible to link payments from the first and the second pillar and utilize the economies of scale to reduce the unit cost of pension payments. Finally, it is the state that remains responsible for

M. Szczepański, Longevity risk and the design of the Polish pension system

the security and continuity of pension benefits, so using a specialized state institution for this purpose seems to be the right solution. Government experts, in order to justify the recent changes in the public pension system, explicitly state that only the state is able to take on the demographic risk, including – longevity risk: “as the only entity able to deal with the demographic risk is the state. Thus, the issue of payment of benefits accumulated in private pension funds should also be linked to the interests of public finances” [MPiPS 2013: 5]. However the examples of insurance companies that pay annuities and have already developed a method of spreading risk within the insurance community risk, demonstrates that the state monopoly regarding the payment of pensions, although still present in most countries, does not have to be the only acceptable solution. Merging pension payments from the first and the second pillar into one state institution does not increase the aggregate longevity risk, as the risk is spread over a much larger number of participants in the pension system than would be in the case of entrusting payments to multiple operators serving smaller groups of retirees. To a large extent longevity risk affecting the people of a given year of beneficiaries (the demographic cohort) reduces the risk of a shorter than expected life span of other retirees receiving pensions. It is known that in every age group there are people living less than the average life expectancy, as well as people living longer than expected. It is difficult to assume that these two groups will always balance one another. The risk of longevity cannot be completely eliminated and the state (directly or indirectly) must take responsibility for the elderly for whom the benefits of the public pension system are often the main or sole source of income. An alternative solution worth considering would be to create a separate state pension institution, which would take over the assets of pension funds and invest them in the financial market and pay out benefits from the second pillar. According to the author it would be a better solution and it would retain the diversification of risk and comply with the key idea of the pension reform of 1999 – “security through diversity”. The original concept of the payments from the second pillar was proposed by W. Otto and M. Wiśniewski who believe that this task should be delegated to common pension associations (private financial institutions managing private pension funds) that would create two sub-funds: of Lifetime Capital Pensions and of Guaranteed Lifetime Capital Pensions. The retirees would have the choice of a PTE – General Pension Society and one of the two pension products: pension annuity or pension annuity with a guaranteed payment period (for example, until they reach the age of 77 years). This second pension would be lower, but it would be possible to inherit it, if the retiree’s death occurred within the warranty period [Otto and Wiśniewski 2013: 24–25]. As for the additional voluntary pension systems functioning under the third pillar, neither in the system of group savings for additional pension in the work-

Economics and Business Review, Vol. 1(15), No. 3, 2015

place (occupational pension systems – PPE, available since 1999), nor in individual systems (individual retirement accounts – IKE, operating since 2004, or individual accounts of retirement security – IKZE, since 2011 onwards) is there any product offered in the form of a retirement annuity. Legal regulations for occupational pension plans (Law of 20 April 2004, Art. 42), IKE and IKZE (The Act of 20 April 2004, Art. 34) provide that the payment of money may take place at once or in instalments after a retiree reaches the age of 60 years (occupational pension plans or individual retirement accounts) or 65 years (in individual accounts of retirement security). Any payment of instalments will last until the depletion of savings accumulated in occupational pension plans, IRA or in individual accounts of retirement security and not in the form of benefits payable for life. There is quite a realistic scenario according to which a person saving for retirement will receive an additional one-time payment at the age of 60 or 65 years of age and by living unusually long this person will deplete this additional fund and in the last phase of life his or her standard of living (based solely on funding from the public pension system) will be significantly lower. Such a structure of payments from the third pillar of the pension system in Poland does not protect against longevity risk. During the presidential campaign, and after the presidential election in Poland (May 2015), there are projects of far-reaching changes in the public pension system in Poland, for example – to allow retirement after 40 years of payment of pension contributions for men and 35 years for women as well as changes in the formula for calculating pensions from the defined benefit formula (defined contribution, DC) to the formula with a defined benefit (defined benefit, DB). Such solutions can significantly increase the aggregate longevity risk in the public pension system. For example a large group could retire at the age of 58 (workers who began their professional career at the age of 18) and then receive pension benefits over 25 years or longer. Ensuring continuity of payments without increasing pension contributions would mean a significant reduction in pensions. Taking into account the low level of participation in additional pension schemes Poland (only approximately 2.2% of employees are covered by occupational pension schemes, only 5.2% of people of working age saves on the IKE and 3.2% saves for IKZE – at the end of 2014 r., according to Komisja Nadzoru Finansowego [Polish Financial Supervision Authority]). This could lead to poverty in the last decades of their lives because of an expected decrease of pension benefits paid out from public pension system.

Conclusions and recommendations Both solutions with multiple entities or with one entity paying out pension benefits from the public pension system in Poland have their advantages and disadvantages. The author agrees that “there is no simple answer to the ques-

M. Szczepański, Longevity risk and the design of the Polish pension system

tion whether the payment of benefits should be addressed by one centralized institution or by many competing entities. Both solutions have their pros and cons”[KNUiFE 2004: 8]. Also the fact that the latest statutory amendments have selected only one pension product: a lifetime annuity benefit in the Polish public pension scheme seems satisfactory. In contrast the obvious drawback of previously introduced changes to the public pension system in Poland is the occurrence of actuarial risk which may directly cause an increase in longevity risk. Public pensions schemes with a defined benefit formula are more sensitive to longevity risk. Reverting to the pension system prior to 1999, based almost exclusively on the PAYG financing method and the DB formula would mean a significant increase in longevity risk and increase risks of damage to long-term sustainability of the pension system. This would mean an increase in systemic risk of the whole pension system in Poland – in both the short and long term. The purpose of the payment of benefits should be to ensure an optimal level of life for beneficiaries continued throughout the duration of life. The right solution to this problem requires the development of an algorithm and parameters to determine the optimal value of benefits. Actuarial risk is associated with the adoption of poorly estimated parameters (e.g. the longer life expectancy in terms of months for a given demographic age group as the basis for the calculation of benefits in the new pension scheme). When pension payments are made directly from the accumulated capital the pensioner begins to bear a risk. Above the minimum guaranteed by the state the level of benefits is determined by the amount of capital held and by a legally defined algorithm to determine the scope of the provision. The adoption of the algorithm, which in the sphere of assumptions departs from reality, can cause two kinds of results. A too slow decumulation of capital in the population reduces the beneficiaries’ level of consumption and causes the transfer of non-consumed pension capital to the next generation. On the other hand, a too high payout level may end up with prematurely depleted capital and result in the realization of longevity risk. The problem then is a decline in living standards of pensioners and a burden for the state because of payments of minimal guaranteed pensions [KNUiFE 2004: 78]. Therefore the necessary missing link in the pension scheme is to create an institution of national actuary which will be properly able to accurately forecast demographic trends and an to make an appropriate calculation of base pension benefits on the basis of further life expectancy. This will enable a more effective management of both demographic and longevity risks. Additional pension systems (occupational pension systems, individual retirement accounts, individual accounts of retirement security) do not protect the savers against longevity risk as they do not offer annuities. In many countries a widely used solution is to buy an annuity at the commencement of the

Economics and Business Review, Vol. 1(15), No. 3, 2015

withdrawal of accumulated additional pension capital. However in Poland a life insurance with perpetuity payments is very poorly developed and their availability is limited. Nevertheless accumulating additional pension capital may be useful for those people who live longer than expected – at least some of them will be able to take advantage of this capital during old age if the size of the additional savings is significant enough and does not get consumed before. As there are no additional systems in Poland with defined benefits such as the occupational pension schemes in Western Europe or the U.S., aggregate longevity risk does not affect those employers who offer pension schemes.

References Anatolin, P., 2007, Longevity Risk and Private Pensions, OECD Working Paper on Insurance and Private Pensions, no. 3. Blake, D., Burrows, W., 2001, Tha Case of Longevity Bonds: Helping to Hedge Mortality Risk, Journal of Risk and Insurance, no. 68 (2): 339–348. Cocco, J.F., Gomes, F.J., 2011, Longevity Risk, Retirement Savings, and Financial Innovation, Netspar Discussion Papers, http://faculty.london.edu/fgomes/cg.pdf [access: 15.06.2015]. Hull, J.C., 2011, Zarządzanie ryzykiem instytucji finansowych, Wydawnictwa Profesjonalne PWN, Warszawa. Jurek, Ł., 2011, Rekonstrukcja wieku emerytalnego w dobie demograficznego starzenia, Polityka Społeczna, nr specjalny, cz. I: 22–25. KNUiFE, 2004, Wypłata emerytur z II filara nowego system emerytalnego, Warszawa. Mierzejewska, M., 2005, Egzemplifikacja ryzyk gospodarstwa domowego, w: Szumlicz, T. (red.), Społeczne aspekty ubezpieczenia, Szkoła Główna Handlowa w Warszawie: 209–221. MPiPS, 2013, Uzasadnienie do projektu ustawy o zmianie niektórych ustaw w związku z określeniem zasad wypłaty emerytur ze środków zgromadzonych w otwartych funduszach emerytalnych, Ministerstwo Pracy i Polityki Społecznej, Warszawa, 10 października, http://www.mpips.gov.pl [access: 20.10.2013]. OECD Statistics, http://stats.oecd.org [access: 16.01.2014]. Otto, W., Wiśniewski, M., 2013, Emerytury kapitałowe: mechanism ekonomiczny, w: Chybalski, F., Marcinkiewicz, E. (red.), Współczesne zabezpieczenie emerytalne. Wybrane aspekty ekonomiczne, finansowe i demograficzne, Wydawnictwo Politechniki Łódzkiej, Łódź: 23–41. Pitacco, E., Denuit, M., Haberman, S., Olivieri, A., 2009, Modelling Longevity Dynamics for Pensions and Annuity Business, Oxford University Press, New York. Rejda, G.E., 2001, Principles of Risk Management and Insurance, Addison WesleyLongman, New York. Stallard, E.L., 2006, Demographic Issues in Longevity Risk Analysis, The Journal of Risk and Insurance, vol. 73, no. 4: 575–609. Szczepański, M., 2013, „Bezpieczeństwo dzięki zrównoważeniu” – wstępna ocena proponowanych zmian, in: Szczepański, M. (ed.), Reformowanie systemów emerytalnych

M. Szczepański, Longevity risk and the design of the Polish pension system

– porównania i oceny. Pension Reforms – Comparison and Evaluation, Publishing House of Poznań University of Technology, Poznań: 147–175. Szumlicz, T., 2005, Ubezpieczenie społeczne. Teoria dla praktyki, Oficyna Wydawnicza Branta, Bydgoszcz–Warszawa. UNFE, 2004, Wypłata emerytur z II filara nowego system emerytalnego, Urząd Komisji Nadzoru Ubezpieczeń i Funduszy Emerytalnych, Warszawa. Ustawa z dnia 20 kwietnia 2004 r. o indywidualnych kontach emerytalnych oraz indywidualnych kontach zabezpieczenia emerytalnego, Dz.U., nr 116, poz. 1205 z późn. zm. Ustawa z dnia 20 kwietnia 2004 r. o pracowniczych programach emerytalnych, Dz.U., poz. 1207. Ustawa z dnia 6 grudnia 2013 r. o zmianie niektórych ustaw w związku z określeniem zasad wypłaty emerytur ze środków zgromadzonych w otwartych funduszach emerytalnych, Dz.U., poz. 1717. Vaughan, E.J., Vaughan, T.M., 2001, Essentials of Risk Management and Insurance, Wiley, New York. Word Bank Statistic, http:// www.data.worldbank.org [access: 16.01.2014].

Economics and Business Review, Vol. 1 (15), No. 3, 2015: 90–111 DOI: 10.18559/ebr.2015.3.7

Polish farmers’ perception of spring frost and the use of crop insurance against this phenomenon in Poland1 Monika Kaczała2, Dorota Wiśniewska3

Abstract : According to Polish farmers spring frost is one of the most dangerous natural perils which a farm may face. The aim of the paper is to describe how farmers assess spring frost in the context of other sources of risk and to investigate if there are any interdependencies between the perception of spring frost and the use of crop insurance to cover this peril. The factors affecting the perception of spring frost were identified. The identified determinants of spring frost assessment were then used to construct an ordered response logit model that enables a classification of the farmer according to his assessment of spring frost. Keywords : agriculture, spring frost, risk perception, crop insurance, ordered response logit model. JEL codes : Q120, G220, D81.

Introduction “Risk perception is the subjective assessment of the probability of a specified type of accident happening and how concerned we are with the consequences” [Sjöberg, Moen, and Rundmo 2004: 8]. Farmers’ perception of sources of risk has been researched in different countries [Harwood et al. 1999; Coble et al. 1999; Chiotti et al. 1997; Meuwissen, Huirne, and Hardaker 1999; Tucker, Eakin, and Castellanos 2010; Assefa, Meuwissen, and Oude Lansink 2014], mainly in the USA. Some studies have shown that personal risk perception influences the type of risk management strategy undertaken by a farmer [Beal 1996; Tucker, Eakin, and Castellanos 2010]. It also affects demand for insurance [Ogurtsov, van Asseldonk, and Huirne 2009; Sherrick et al. 2004]. Risk perception could vary depending on the country in which farmers operate [Boholm 2003; Dessai 1

Article received 6 March 2015, accepted 3 August 2015. Poznań University of Economics, Department of Insurance, al. Niepodległości 10, 61-875 Poznań, Poland; corresponding author, e-mail: m.kaczala@ue.poznan.pl. 3 Poznań University of Economics, Department of Econometrics, al. Niepodległości 10, 61‑875 Poznań, Poland. 2

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

et al. 2004; Eakin 2006; Fraser-Mackenzie, Sung, and Johnson 2014]. However almost no research has been conducted so far to determine farmer’s risk perception in Eastern and Central Europe, especially in the post-communist countries. Poland as an example of a Central European post-communist country has been selected for this study because it is one of the largest areas and has the most people in the EU employed in agriculture (it accounts for over 10% of EU arable land and over 25% of the EU agricultural population) [Statistical Yearbook of Agriculture 2013: 394]. According to Polish farmers spring frost is the most dangerous natural peril which a farm may face, followed by winterkill and drought [Kaczała and Wiśniewska 2015: 100]. Spring frost is usually defined as a drop in the air temperature to 0°C and below at the times when the mean temperature for 24 hours remains above 0°C [Chromow 1977: 138–139]. The definition of spring frost used in subsidised crop insurance refers to „damage caused by a drop in temperature below 0°C between 15th Apr and 30th June which have caused full or partial plant damage or full or partial crop loss”.4 Spring frost is inherent to Poland, although its severity and geographical distribution is varied [cf. e.g. Koźmiński and Michalska 2001: 75; IMGW 2013: 60–63]. There are areas in Poland where the spring frost-free period has shrunk (the north-east) [Kalbarczyk 2010; Grabowski 2010] and where it has lengthened (the region of Bydgoszcz) [Dudek, Żarski, and Kuśmierek-Tomaszewska 2012]. However according to estimates the pessimistic scenario assumes an increase in the frequency and severity of spring frost in Poland [Klimkowski 2002; Kundzewicz 2012: 21]. The aim of the paper is to investigate the factors affecting the Polish farmers’ perception of spring frost. Firstly, we describe how farmers assess spring frost in the context of crop insurance and the appraisal of other sources of risk. Secondly, we investigate the factors affecting the perception of spring frost amongst arable farmers in Poland. First, the objective features of the farmers and their farms will be considered. Next, the significance of past experience concerning different adverse events (weather phenomena, agricultural policy, changes in market prices, changes in crop technology, health problems, etc.). The following hypotheses will be tested: H1: F armers who perceived spring frost as dangerous are more likely to use crop insurance covering spring frost than farmers who are not afraid or who have a neutral attitude towards this peril. H2: Th ere are factors differentiating farmers between those who assess spring frost as either a dangerous, neutral or not dangerous peril and as a consequence it is possible to construct a practically applicable tool to identify individuals with one of the above perceptions of spring frost. 4 This is the final definition that was originally formulated in Art. 3 Section 2 point 11 of the Act of July 7th 2005 and then it was altered twice – by art. 1 point. 4c of the Act of 2nd March 2007 and by art. 1 point. 1a of the Act 25th July 2008.

Economics and Business Review, Vol. 1(15), No. 3, 2015

The paper is divided into three sections. The first section is devoted to data and methodology, the second contains empirical results and the last presents interpretations of the results and conclusions.

1. Scope and methods of the study 1.1. Data Primary data was gathered on the basis of a survey conducted in March 2012 by means of the CATI method, with the use of a structured questionnaire, on a focus group of 750 farmers across Poland who grow crops. A representative sample was selected on the basis of the farm location and size. Answer variants and respondents’ profiles were expressed by means of different qualitative variables: binary variables, polynomial variables – both nominal and ordinal ones. The data about the farmers and the characteristics of their farms was collected. Farm managers assessed 13 perils in the scale from 1 to 7, where 1 denoted a negligible peril, whilst 7 represented a definitely dangerous phenomenon. The list included moveable perils, such as hail, flood, winterkill, spring frost, drought, hurricane, plant pests and diseases, the farmer’s health problems, increase in agricultural input prices, price volatility on the crop markets, political changes, property damage and sudden changes in agricultural technology. The data on acceptable losses in crops and losses in crops leading to a farm’s bankruptcy were obtained according to declarations made by farmers, as well as data about loss experience and insured perils.

1.2. Methods applied in the subsequent stages of study In order to describe the structure of responses to the question about the perception of spring frost risk and evaluate these responses with regard to the assessment of other risks, some frequency tables were created and the distribution of responses analysed. Due to the fact that an ordinal scale was used for the evaluation of particular risks (scored from 1 to 7), a non-parametric KruskalWallis test could be applied to determine significant differences in the evaluation of each risk, with particular focus on whether spring frost risk is considered as the most dangerous risk of all. Apart from the non-parametric analysis of variance a Spearman’s rank correlation analysis was conducted in order to determine the correlation between the perception of spring frost risk and the evaluation of other risks. The findings of this analysis encouraged the extension of the study using cluster analysis, which would help to establish homogeneous groups of respondents depending on their risk perception. First of all, in order to determine the number of clusters, an agglomerative method was used. This was followed by the application of the k-means method in order to classify the respondents into specified clusters.

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

In order to verify the hypothesis concerning the correlation between the spring frost risk perception and various qualitative features, a number of contingency tables (cross-tabulation) was produced and the Pearson’s test of independence was conducted. As some of the features considered had quite a few variants, a problem appeared with regard to the appropriate sample size in each cell of the contingency table. Therefore, spring frost risk perception was categorised into three classes: –– Low level of risk, if it was evaluated 1 or 2, –– Medium level of risk, if evaluated 3 to 5, –– High level of risk, if evaluated 6 to 7. Reduction of the number of variants of the given variable made it easier to interpret the way in which different variants of the considered qualitative features affect the perception of spring frost risk. Cramer’s coefficient, based on chi-squared statistics was used as a measure of strength of this correlation. In the cases when the considered determinants of spring frost risk perception were quantitative (e.g. how many times in the previous ten years a given risk had occurred), the non-parametric Kruskal-Wallis test was applied in order to find out if the three classes of risk differed in terms of the qualitative feature value. The potential determinants of risk perception researched can be put into three groups: a) objective features of the respondents and their farms: –– sex, age, educational background, –– farm size, production purpose, dominant soil quality class, the use and character of additional, non-farming sources of income, dominant production, –– types of crops, –– province where the farm is located; b) subjective opinions of the respondents, i.e.: –– the degree of crop loss which does not jeopardise the farm operation, –– the degree of crop loss leading to bankruptcy; c) experience related to different perils: –– the frequency of various adverse occurrences in the previous 10 years, –– the scope of adverse occurrence affliction, i.e. the evaluation of the influence the adverse phenomenon had on the farm’s income from crops (in the scale of 1 to 4, where 1 denotes lack of influence on the income, and 4 denotes a very big influence). In the last stage of the research two ordered categories logit models were constructed in order to produce a tool to permit the respondents’ classification into one of the three determined risk classes. In the first model the potential exogenous variables were assumed to be only the objective features of the respondents and their farms, which were identified by means of the correlation analysis of their qualitative features. In the other model, the potential exogenous variables also included those which reflected the respondents’ subjective

Economics and Business Review, Vol. 1(15), No. 3, 2015

features and their experience regarding adverse occurrences. The qualitative (nominal) features were introduced into the model through a number of binary variables; hence, if a given variable had i-variants, one of them was assumed to be the base and i-1 of the variables were introduced into the model. Selection of the variables for both models was carried out by means of stepwise regression. It was assumed that the variables which remained in the model would be significant at the confidence level of 95 percent. Unfortunately none of the available respondents’ features could be directly used to measure their risk aversion. In order to make the best possible use of the survey findings, additionally composite (synthetic) variables were established which could reflect risk aversion. The first one being the mean of diagnostic variables was constructed on the basis of opinions about a brand new insurance product, i.e. index insurance. The opinions were expressed as the answers to two questions as to whether the respondents liked the product (on a scale from 1 to 6); in the first question variant the price of the product was not revealed. This was supposed to illustrate the respondents’ propensity for accepting novelties. The second composite variable was to reflect the degree of the respondents’ trust toward insurance companies. It was identified on the basis of 6 questions regarding the degree of agreement with a particular opinion about the operations of insurance companies. The responses were given in the ordinal scale of 1 to 5, where 1 represented lack of agreement, and 5 reflected a high degree of agreement. Both the measures were introduced into the aforementioned logit models in order to attempt to improve the classification quality. GRETL and Statistica10PL software was used for all the calculations.

2. Empirical results 2.1. The structure of spring frost risk assessment and the use of crop insurance As has already been mentioned the respondents assessed the degree of spring frost risk on a scale of 1 to 7, where 7 denotes the highest degree of risk. As a result of 750 observations it became clear how some of the respondents rated this risk. Their ratings are presented in Figure 1. I Importantly the graph additionally presents a separate distribution of sub-groups of those who declared that they had insured their crops against spring frost and those who claimed to have no insurance of this kind. The analysis of this graph shows clearly that respondents most often gave high grades to spring frost: 5 in 29 percent of the cases and 6 in 24 percent. It is also quite obvious that the highest grades were given by the people who had insured their crops (6 and 7) and the lowest grades (1, 2, 3 and 4) were given by those who had not insured their crops.

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

30 25 20 15 10 5 0

3 Uninsured

Insured

Total

Figure 1. Distribution of spring frost risk rating in the whole focus group and in sub-groups of those who had insured their crops and those who had not Table 1. Percentage of respondents who declared a low, medium or high spring frost risk rating vs. insurance or lack of it against spring frost Scope

Spring frost risk assessment (%)

Independence test results

Low

Medium

High

Total

Chi sq.

12.103

Insured

p-value

0.002

Uninsured

Cramer coeff.

0.127

In order to confirm the statistical significance of the response distribution an independence test of qualitative features was conducted assuming three classes of risk assessment. The findings presented in Table 1 show that there is a correlation between the risk assessment and a decision to buy a crop insurance policy at a confidence level of 99.8 percent. This correlation is not very strong. Table 2 shows the structure of various risk assessments and a mean rating for each peril. The findings of the Kruskal-Wallis test prove significant differences between the perceptions of those risks. Rather importantly spring frost constitutes the most highly rated risk amongst the occurrences connected with adverse weather phenomena and plant diseases. This risk perception is not significantly lower than the similarly assessed drought and winterkill. Amongst all other risks the most highly rated ones were the risks connected with increases

[96]

Kruskal-Wallis test

0.99

1.742

Std.dev.

18%

4.624

15%

Average grade

23%

12%

17%

Grade given

Drought

Risk assessed

Flood

0.00

2.001

3.081

12%

11%

10%

20%

31%

Hail

24%

11%

1.59

4.704

Winterkill 1.695

4.517

12%

18%

26%

19%

10%

Hurricane 1.883

2.96

14%

12%

21%

31%

1.716

4.169

13%

25%

20%

12%

Health problems 2.081

1.616

5.245

28%

10% 3.607

21%

25%

14%

Rising prices of agricultural input

13%

15%

25%

Agricultural market volatility 1.775

4.959

23%

21%

24%

14%

1.977

4.208

13%

17%

24%

15%

16%

Political changes

0.00

0.99

0.00

0.64

0.00

H(12; N = 9750) = 1269.807, p-value = 0.000 p-value obtained from pair comparisons (grades given to spring frost risk and other risks)

1.812

3.511

11%

29%

16% 6%

16%

Spring frost

17%

14%

20%

16%

Plant diseases

Table 2. Findings of the examination of spring frost risk vs. other risks assessment Property damage 0.00

2.109

3.564

13%

20%

12%

29%

0.00

1.954

3.181

11%

15%

17%

31%

Technological changes

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

in prices of agricultural input and risk concerning agricultural market volatility. In the first case the assessment is even significantly higher than the rating given to the spring frost risk. Another issue that was subject to research was the correlation between the growth in the negative perception of spring frost risk and the potential growth or drop in assessment of other risks. Rank correlation coefficients which are presented in the second column of Table 3 enable evaluation of the direction and strength of these correlations. Subsequent risks are listed in a specific order: the first to be listed are the ones whose assessment is most tightly correlated to the evaluation of spring frost risk. The strongest positive correlation can be seen between spring frost and winterkill, hurricane and hail, whilst the weakest one are in respect of flood and health problems. Table 3. Measures of the strength of the correlation between spring frost perception and assessment of other risks Spearman’s rank correlation coefficients Type of risk

Calculation Calculation based on excluding the all observations farmers who rated all risks very highly

Winterkill

0.5608

0.5553

Hurricane

0.3296

0.2668

Hail

0.2922

0.2301

Plant diseases and pest

0.2805

0.2650

Political changes relating to agriculture

0.2705

0.2455

Drought

0.2386

0.1993

Rising prices of agricultural input

0.2300

0.2137

Crop prices fluctuations

0.2217

0.2008

Dramatic changes in cultivation technology

0.2007

0.1664

Property damage

0.1907

0.1457

Health problems

0.1613

0.1374

Flood

0.0868

–0.0198a

Statistically insignificant correlation at the significance level of 0.05.

Although all the coefficients of this correlation do not achieve very high values it may be surprising that they are all positive. This may mean that there is quite a large group of people who ranked one of the risks high (low) and at the same time was prone to evaluate all the other risks in the same way. Therefore, in order to identify more precisely the way in which respondents evaluated the

Economics and Business Review, Vol. 1(15), No. 3, 2015

7 6 5 4 3 2 1

Property damage Changes in technology

Political changes

Adverse crops’ price changes

Increase of input prices

Farmers’ health problems

Plant diseases and pests

Hurricane

Winterkill

Spring frost

Hail

Flood

Drought

Cluster 1

Cluster 2

Cluster 3

Cluster 4

Cluster 5

Cluster 6

Cluster 7

Cluster 8

Figure 2. Average ratings given to particular risks in the identified clusters Source: Own calculations with the use of Statistica 10PL

risks, it was decided that they should be divided into groups which similarly assessed particular types of risks. As a result of cluster analysis the respondents were divided into eight clusters. Average ratings of particular perils by the people grouped in a given cluster are presented in Figure 2. Cluster analysis made it possible to establish the possible cause of positive correlation between the ratings given to particular risks: there is quite a large group of respondents (approximately 100 people) who rated all risks highly, regardless of their provenance; these people constituted cluster 1. This manner of risk perception can be rather puzzling – it might result from the fact that either all these people are highly risk-averse, which is quite difficult to verify, or they have abundant negative experience (which is very unlikely, as occurrences such as adverse price changes or agricultural policy shifts do not affect such limited groups), or it may be the effect of carelessness or impatience in offering responses. Considering the latter potential cause of establishing cluster 1 it was decided that examination of the correlation between spring frost risk perception and other features of the respondents would additionally involve checking if these regularities also occurred in the case of all 750 observations and in the subgroup out of which cluster 1 respondents had been excluded. Consequently Table 3 also presents the rank correlation coefficients obtained in the case of the reduced group. It turned out that a large majority of the correlations was

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

confirmed, albeit they were a little lower (as had been expected). What is important is that the negative sign of the correlation between spring frost risk and flood risk evaluation obtained was more factually correct.

2.2. Factors affecting the perception of spring frost 2.2.1. Respondents’ objective features The results of the chi-square independence test clearly point to the fact that risk perception is not affected by features such as sex, educational background, farm size, production purpose or sources of income. Age and its influence on risk perception was placed on the verge of significance and the confidence level regarding the occurrence of particular regularities depends on the scope of the group. Distribution of spring frost perception depending on the respondents’ age is shown in Table 4. It is worth mentioning that persons aged over 60 rated spring frost highly much more often than younger people (40.18 percent of the oldest group, while 34 and less percent was reported in the younger groups). At the same time, the rating distribution in the other age groups is similar enough to make it impossible to reject the zero hypothesis concerning lack of correlation between the respondent’s age and their evaluation of risk. Table 4. Percentage of respondents evaluating the spring frost risk as high, medium or low depending on their age Response distribution for all observations

Independence test findings

Risk evaluation

<40 (%)

41–50 (%)

51–60 (%)

>60 (%)

All Excluding observacluster 1 tions

High

33.94

33.99

31.84

40.18

Chi sq.

8.353

10.462

Medium

51.52

50.74

58.80

46.43

p-value

0.213

0.106

Low

14.55

15.27

9.36

13.39

C. coeff.

0.075

0.090

The factor which significantly affects spring frost perception is the dominant production. As Table 5 indicates, the highest ratings were given relatively more often if the dominant production involved pigs (more than 42 percent of these respondents gave it the highest rating) and crops (39.3 percent respectively). However the correlations are not strong, despite being statistically significant. It might seem likely that the type of cultivated crop has a substantial influence on spring frost perception. It turns out, however, that the above assumption is valid only for winter barley and rape. Although their cultivation, as expected, slightly increases the likelihood of giving the risk a very high rating, Table 6 shows that this influence is not that strong and out of the remaining

100

Economics and Business Review, Vol. 1(15), No. 3, 2015

Table 5. Spring frost rating distribution according to the dominant production Risk evaluation

Crops (%)

Milk (%)

No dominant production (%)

Pigs (%)

Independence test findings

High

39.30

25.00

26.37

42.55

Chi sq.

15.044

Medium

49.73

60.00

60.44

46.81

p-value

0.0199

Low

10.96

15.00

13.19

10.64

C. coeff.

0.102

Table 6. Comparison of plants whose cultivation has the largest influence on spring frost risk perception, in the light of the Chi-square test of independence Type of plant

Winter barley

Rape

Winter triticale a

Chi-sq. stat. (p-value)

Relationship

11.7172 (0.003)

45.3 percent of farmers cultivating winter barley gave the highest rating to spring frost risk whilst 8.18 percent of these farmers gave it the lowest rating. In the cases of farmers who do not cultivate this crop the percentages are 31.5 percent and 13.9 percent respectively

9.94 (0.007)

47.44 percent of farmers cultivating rape gave the highest rating to spring frost risk while 3.85 percent of these farmers gave it the lowest rating. In the cases of farmers who do not cultivate this crop the percentages are 32.9 percent 13.7 percent respectively

4.33 (0.115a)

36.9 percent of farmers cultivating winter triticale gave the highest rating to spring frost risk while 11.21 percent of these farmers gave it the lowest rating. In the cases of farmers who do not cultivate this crop the percentages are 30.5 percent 14.9 percent respectively

Statistically insignificant correlation at the significance level of 0.05.

winter crops only triticale cultivation affects the spring frost perception at the verge of significance. Another feature of the farm that seems to have a significant influence on risk perception is the farm’s location. In the cases of the lower number of responses from Opole, Pomerania, Warmia-Masuria and Silesia provinces, they had to be removed from some of the cells of the contingency tables. However amongst other provinces it was possible to identify the ones where the highest ratings were relatively more often given to spring frost risk (Table 7). It refers to the provinces of Lublin, Wielkopolska, Kujawy-Pomerania, Świętokrzyskie and Łódź. Unfortunately location despite its statistical significance is also not a very strong determinant in risk perception.

101

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

Risk assessment

Lower Silesia

Lublin

Małopolska

Mazovia

Subcarpathisn

Podlasie

Świętokrzyskie

Lubuskie

West Pomerania

KujawyPomerania

Wielkopolska

Łódź

Table 7. Farm location vs. the structure of spring frost risk evaluation (%)

High

Medium

Low

Test of independence results: Chi sq. = 35.59, p-value = 0.033; Cramer coeff.= 0.15

2.2.2. Respondents’ subjective features and farming loss experience The results of the independence test indicate that the risk assessment distribution is not contingent on acceptable or non-acceptable loss of crops. On the other hand a strong correlation was noticed between various risk assessments and different variables denoting the farmer’s experience with them. As the respondents were divided into three groups depending on their risk perception, whilst the occurrence frequency was indicated on a ratio scale, a classical analysis of variance and non-parametric Kruskal-Wallis test was conducted in order to identify the significant differences in the frequency of occurrence of particular risk connected with natural phenomena which the three groups had experienced. Table 8 presents the results which indicate that the higher the given phenomenon’s frequency of occurrence, the larger the propensity to rate spring frost risk as highly dangerous. A reverse correlation can only be seen in the case of flood frequency (in the areas often struck by floods the propensity for spring frost assessment as highly dangerous is smaller). This may be due to the fact that the shores of lakes, large ponds and river banks are conducive to the cultivation of crops which are vulnerable to spring frost. Adverse phenomena connected with low temperatures (winterkill) have the largest influence on spring frost risk perception. Apart from the influence of adverse occurrences the effect these phenomena had on the farmers’ income from crops was also examined. The survey identified this effect within an ordinal scale (with four feature variants) regarding not only the natural phenomena but also entirely different occurrences. The results of the examination of the relationship between the degree to which the occurrence struck the respondent and their risk perception category are presented in Table 9. In order to interpret this table properly one has to keep it in mind that within the whole focus group 34 percent of respondents rated spring frost risk the highest. If a visibly large percentage of respondents who were struck by a given

102

Economics and Business Review, Vol. 1(15), No. 3, 2015

Table 8. Frequency of the phenomena vs. spring frost risk perception – results of the classical analysis of variance and the non-parametric Kruskal-Wallis test Phenomenon Drought Flood Hail Spring frost Winterkill Hurricane

Statistics F and H

p-values

Type of relationship

5.384

0.005

The more frequently drought occurred, the higher the risk was rated

11.36

0.003

4.242

0.015

9.016

0.011

9.327

0.000

20.55

0.000

29.391

0.000

72.14

0.000

33.372

0.000

75.897

0.000

4.687

0.009

5.44

0.066

Fire, Animal attacks, Plant diseases

The more frequently flood occurred, the lower the risk was rated The more frequently hail occurred, the higher the risk was rated The more frequently spring frost occurred, the higher the risk was rated The more frequently winterkill occurred, the higher the risk was rated The more frequently hurricanes occurred, the higher the risk was rated Lack of significant relationship

Table 9. Assessment of the degree to which a given occurrence affected the farm’s income vs. the frequency of giving the spring frost risk a low, medium or high rating – the Chi-square test of independence results Occurrence

Chi-sq. statistics (p-value) V-Cramer coeff. 13.780

Drought

(0.032) 0.09585 30.730

Flood

(0.000) 0.14313 9.260

Hail

(0.160)

0.07857 79.570 Spring frost

(0.000) 0.23032

Type of relationship 48 percent of the people most severely struck by drought (whilst only 27.4 percent of those who did not make losses because of drought) rated spring frost risk as the most dangerous 41 percent of the people most severely struck by flood (whilst only 23.81 percent of those who did not make losses because of flood) rated spring frost risk as the most dangerous 53.13 percent of the people most severely struck by hail (whilst only 33.3 percent of those who did not make losses because of hail) rated spring frost risk as the most dangerous – unfortunately, in the other response variants there are no visible regularities 76 percent of the people most severely struck by spring frost (whilst only 29.27 percent of those who did not make losses because of spring frost) rated it as the most dangerous risk

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost 244.530 Winterkill

(0.000) 0.40376 57.622

Hurricane

(0.000) 0.19600

Plant diseases

42.447 (0.000) 0.16822 34.396

Health problems

(0.000)

Rising prices of agricultural input

24.040

Fluctu ations of crop prices Political changes

Property damage

0.15143

(0.002) 0.12660 19.640 (0.012) 0.11443 37.370 (0.000) 0.15784 38.810 (0.000) 0.16085 62.460

Technology

(0.000) 0.20406

103

64 percent of the people most severely struck by winterkill (whilst only 25 percent of those who did not make losses because of winterkill) rated spring frost risk as the most dangerous 61 percent of the people most severely struck by hurricane (whilst only 35.77 percent of those who did not make losses because of hurricanes) rated spring frost risk as the most dangerous 42 percent of the people most severely struck by plant diseases (whilst only 26 percent of those who did not make losses because of plant diseases) rated spring frost risk as the most dangerous 38 percent of the people most severely struck by health problems (whilst only 33 percent of those who did not make losses because of health problems) rated spring frost risk as the most dangerous.- at the same time a positive relationship is seen in all the feature variants 38 percent of the people most severely struck by this occurrence (whilst only 24 percent of those who did not make losses because of it) rated spring frost risk as the most dangerous 36 percent of the people most severely struck by price fluctuations and 30 percent of those who did not make losses because of price fluctuations rated spring frost risk as the most dangerous. Weak relationship 50 percent of the people most severely struck by political changes (whilst only 28.9 percent of those who did not make losses because of political changes) rated spring frost risk as the most dangerous 47 percent of the people most severely struck by property damage and 28.6 percent of those who did not make losses because of property damage rated spring frost risk as the most dangerous.. Weak relationship 45 percent of the people most severely struck by technological changes (whilst only 26.7 percent of those who did not make losses because of technological changes) rated spring frost risk as the most dangerous

occurrence rate spring frost risk as the most dangerous it proves a strong positive relationship. It is rather obvious that the highest percentage of respondents who rated spring frost risk as the most dangerous related to people who had incurred severe losses because of spring frost (76 percent), winterkill (64 percent), hurricane (61 percent) and hail (53 percent). What is also essential, in the case of all the adverse phenomena, the percentage of respondents who

104

Economics and Business Review, Vol. 1(15), No. 3, 2015

gave spring frost risk the highest grade is significantly higher when the adverse occurrence had a serious influence on the farmer’s income.

2.3. The logit model in respondent classification according to their spring frost risk perception class Considering the fact that it was possible to identify several features of the respondents which affect their spring frost risk perception a decision was made to evaluate their diagnostic power by means of constructing a logit model for ordered categories. This model would make it possible to obtain a correct hit ratio for a person with particular characteristics, classifying them into one of the three categories: those who gave spring frost risk low, medium and high rating. For practical purposes it is advisable to obtain correct classification only on the basis of objective, easily identifiable features of the farmer and his/her farm. Therefore the first thing was to find significant variables among the objective features of the respondents. Table 10 presents such significant variables for this model along with its assessment of the parameters: Table 10. Significant variables and logit model parameter assessments – objective features (model 1) Variables and cut off points of the model

Coefficients

Standard deviation

p-value

Is plant production dominant

0.43571

0.1452

0.0027

Is rape cultivated

0.61021

0.2417

0.0116

Is winter barley cultivated

0.53065

0.1771

0.0027

Wielkopolska Province

0.69714

0.2085

0.0008

Kujawy-Pomerania Province

0.59759

0.2643

0.0238

Łódź Province

0.56519

0.1976

0.0042

Świętokrzyskie Province

0.67596

0.3407

0.0473

Cut1

–1.32877

0.1444

0.0000

Cut2

1.36698

0.1422

0.0000

Confidence ratio test: Chi-sq.(7) = 156.374 [0.0000]

By looking at the model above (model 1) it can be seen that the set of the significant variables and the parameter signs that accompany them are not surprising. The findings presented here are in accord with the results of the statistical analysis of the relationship between the respondent’s objective features and his/her risk perception. Unfortunately whilst the relationships were statistically significant, they were not strong. This results in a very low hit ratio obtained from the model established - it amounts to only 55 percent (Table 11).

105

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

When evaluating the model quality it has to be kept in mind that the results obtained should be compared to the minimum hit ratio obtained as a result of random classification. In the case of unequally sized groups, when the researcher aims to obtain the best possible classification quality, the minimum hit ratio in each of the defined groups is established in accordance with proportional chance criterion [Wiśniewska 2012: 112]. In the case analysed it is barely 41.34 percent. Q-Press statistics confirm with quite a high level of confidence that the achieved hit ratio for this research is significantly higher than the assumed minimum. On the other hand it has to be remembered that the hit ratio was established for the estimation group – in the separate validation group the classification quality usually decreases. Table 11. Classification matrix and hit ratios for model 1 Actual assessment

Classification Low

Medium

High

Hit ratio (%)

Low

Medium

374

High

221

Hit ratio (total)

Due to the unsatisfactory classification quality on the basis of model 1, the set of exogenous variables was extended to include the variables which characterise the frequency of adverse occurrences and the degree of their influence on income from crops. Table 12 presents the variables in the established model and assessments of significant parameters. Due to the fact that experiences relating to various adverse occurrences affected spring frost perception to a much larger extent than the objective features, the obtained hit ratio was much more accurate (Table 13). The hit ratio is not only significantly higher than the minimum established on the basis of the proportional chance criterion (41.34 percent), but it would probably exceed the hit ratio based on the maximum chance criterion – it is equal to the observation percentage of the largest class, i.e. it amounts to 53 percent. The hit ratio for farmers with medium and low levels of risk perception could be considered satisfactory. Unfortunately only slightly more than 50 percent of the persons who presented a high level of risk perception were accurately classified in this category. In an attempt to seek a better classification method, a binary variable was added, which equalled 1 if a person was classified in cluster one (which consisted of people who rated all risks as highly dangerous). Although this variable proved to be statistically significant it only improved the hit ratio accuracy in groups other than “high”. Subsequently other variables

106

Economics and Business Review, Vol. 1(15), No. 3, 2015

Table 12. Significant variables and logit model parameter assessments – objective features and experiences regarding risks (model 2) Coefficients

Standard deviation

p-value

Is plant production dominant

0.37203

0.1674

0.0262

The number of winterkill problems

0.11543

0.0398

0.0037

Influence of drought on income

0.21106

0.0702

0.0027

Influence of spring frost on income

1.76175

0.1123

0.0000

Influence of hurricanes on income

0.24975

0.0741

0.0007

–0.24791

0.0703

0.0004

Farm located in Lublin Province

0.78787

0.3722

0.0343

Cut1

1.42225

0.2834

0.0000

Cut2

6.26592

0.4036

0.0000

Variables and cut off points of the model

Influence of crop price fluctuations on income

Confidence ratio test: Chi-sq. (7) = 621.265 [0.0000]

Table 13. Classification matrix and hit ratios for model 2 Actual assessment Low

Classification Low

Medium

High

Hit ratio (%)

Medium

334

High

121

137

Hit ratio (total)

were introduced which were substituted for features relating to the propensity for risk taking and insuring oneself against risk (described in Section 2.2.). Unfortunately these variables were not significant.

Conclusions On the basis of the hit ratio matrix one can say that model 2 very well identifies the people who rate spring frost risk perception as low or medium but it undervalues these ratings for people in the “high” group. This means that in order to identify the people who rate spring frost as dangerous additional information would have to be introduced into the model. One of the options to achieve this aim is to use the psychometric paradigm [Fischoff et al. 2000], although its scope in explaining the differences in perception of particular risks

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

107

is quite limited (up to 20 percent of the variation [Sjöberg, Moen, and Rundmo 2004: 17, 20 and the literature cited there]), and likewise, the cultural theory [Oltedal et al. and the literature cited there]. Furthermore one should examine the possibility of explaining the rating variations by means of introducing (a) different variable(s) concerning trust rather than the composite variable used in the model, which refers to confidence in insurance companies. Unfortunately the available data does not permit such an extension of the study. The propensity for a given degree of spring frost risk perception is closely related to individual experience concerning the amount and value of the damage caused by some natural perils. In 2011, which was one year before the research was conducted, there was massive damage caused by spring frost, which in turn had been preceded by even more severe losses caused by winterkill. In the course of the following year, i.e. 2012, just before the survey was carried out, catastrophic losses caused by winterkill occurred again and the spring frost season was just about to begin. The value of the losses which were caused by both these occurrences is shown in Table 14 which includes the data relating to compensations paid from subsidised crop insurance policies. What is important is that the data in question is considerably undervalued in comparison with the actual amount of loss in agriculture caused by spring frost. First of all, the data almost exclusively refers to losses of crops, and in the years 2011 and 2012 fewer than 25 percent of crops were insured, including as few as 20 percent of crops being insured against the risks of spring frost and winterkill [justification of the change in the 2014 Act: 9–10]. Secondly, the most vulnerable vegetables and fruits are hardly ever insured as subsidised products (due to exceeding the amount which makes them eligible for obtaining the state subsidy for insurance premium) [justification of the change in the 2014 Act: 2–3, 6], whilst losses in horticulture and fruit farming caused by spring frost were as high as 80 percent in comparison with the long-term mean [Doroszewski et al. 2013: 278]. This sequence of events can explain the strong correlation between the amount of loss caused by winterkill and its effect on income from farming and spring frost perception, as both events were at a similar time, both were connected with freezing weather and both resulted in huge losses for farmers. Additionally this frequency of frost-related occurrences and the scope of loss they had caused could result in overestimating spring frost risk perception. These presumptions are corroborated by other study findings, according to which negative experiences exacerbate the given risk perception [i.a. Riad, Norris, and Ruback 1999; Norris, Smith, and Kaniasty 1999; Keller, Siegrist, and Gutscher 2006 and the literature cited there]. It might be possible that this is the very reason why the model was not exactly suited to people who rated spring frost as highly dangerous. Assuming this one has to point at the effects of the so-called hedonic adaptation [Fredrick and Loewenstein 1999], which appears in response to unfavourable circumstances. Research carried out by Burns, Peters, and Slovic

108

Economics and Business Review, Vol. 1(15), No. 3, 2015

Table 14. Compensation paid from subsidised crop insurance in the years 2008–2012 (in PLN) 2008 Drought

2009

2010

2011

2012

157 832 109

1 357 150

604 312

2 313 371

116 227

Flood

304 347

1 567 676

4 362 625

1 894 610

1 081 410

Winterkill

898 860

2 268 298

31 487 705

160 644 322

587 776 226

2 910 069

32 922 210

8 049 651

137 249 546

7 386 628

31 445 204

82 688 525

53 144 004

59 736 981

122 345 441

193 390 589

120 803 859

97 648 297

361 838 830

718 705 932

Spring Frost Hail and others Total

Source: Justification of the change in the 2014 Act: 9.

indicates that after the initial growth in perception of a given risk as dangerous, its negative evaluation decreases with time and becomes relatively stable [Burns, Peters, and Slovic 2012]. Without panel data, however, it is hard to state whether this situation took place with regard to spring frost within the studied period. Judging by the short period of time which elapsed between the occurrences and the survey, it seems highly doubtful. By comparing the hit ratios in models 1 and 2 one can state that in order to identify a given farmer’s propensity for a particular spring frost perception, knowledge about his/her prior experience is indispensable. This causes difficulty in the application of model 2 by insurance companies with reference to new customers if their damage record is unknown. The analyses carried out also indicate that spring frost perception primarily depends on a farmer’s experience in terms of most natural perils as well as others (price-related in particular). Any kind of loss, regardless of its cause, is conducive to ranking spring frost risk as more dangerous. Simultaneously the assessment is not contingent on the level of loss in crops, which may either be perceived as normal or may lead to farm’s bankruptcy. Identifying a farmer’s perception of sources of risk enables adjustment of the products offered and their prices as well as cost cutting in marketing and distribution. From a product analysist’ s point of view it is very useful to answer two questions concerning insurance cover – the range of the perils covered and the level of integral franchise. As has been demonstrated spring frost perception is not reliant on an acceptable or catastrophic level of loss – and vice versa. Perception of spring frost as dangerous is, on the other hand, correlated with a similar perception of winterkill and to a slightly smaller extent, hail and hurricanes. This means that one of the products offered at present, which involves a 10 percent level of integral franchise and a package covering perils such as spring frost, winterkill, hail or hurricane can be viewed as an appropriate market solution.

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

109

Moreover, there is a correlation, albeit a weak one, between the respondents’ perception of spring frost risk and the fact that they were insured against it. Therefore there are by no means any grounds on which H1 could be rejected. It has to be emphasised that this poor correlation is quite likely to result from a generally low level of crop insurance, as only 30 percent of the farmers had any crop insurance [Kaczała and Wiśniewska 2015: 104]. In years 2008–2013 the average losses in insured crops caused by spring frost ranged from 40 to 1964 PLN, as far as subsidized insurance is concerned. In 2014 the premium rates for this insurance, for a single risk of spring frost, ranged from 0,5% to 10% depending on the type of crop. Unfortunately the available data does not allow to calculate the average loss per hectare and to compare it with hypothetical cost of insurance, therefore an assessment of rationality of choices made by growers with regards to purchase of insurance is not possible.

References Act of 7 July 2005 on livestock and crop insurance subsidies, Journal of Laws, no. 150, item 1249, incl. subsequent amendments. Act of 7 March 2007 on changing the act on crop and livestock insurance subsidies and some other acts, Journal of Laws, no. 49, item 328. Act of 25 July 2008 on altering the crop and livestock insurance act and the national producers’ register, national farms register and subsidy applications register, Journal of Laws, no. 145, item 918. Assefa, T.T., Meuwissen, M.P.M., Oude Lansink, A.G.J.M., 2014, Price Volatility Perceptions and Management Strategies in European Food Supply Chains, Working paper Ulysses, http://www.fp7ulysses.eu/publications/ULYSSES%20Scientific%20 Paper%206_Price%20volatilty%20peceptions%20and%20management%20strategies%20in%20European%20food%20supply%20chains.pdf. Beal, D.J., 1996, Emerging Issues in Risk Management in Farm Firms, Review of Marketing and Agricultural Economics, no. 64 (3): 336–347. Boholm, Å., 2003, The Cultural Nature of Risk: Can There Be an Anthropology of Uncertainty?, Ethnos: Journal of Anthropology, no. 68(2): 159–178. Burns, W.J., Peters, E., Slovic, P., 2012, Risk Perception and the Economic Crisis: A Longitudinal Study of the Trajectory of Perceived Risk, Risk Analysis, no. 32 (4): 659–677. Chiotti, Q., Johnston, T., Smit, B., Ebel, B., 1997, Agricultural Response to Climate Change: A Preliminary Investigation of Farm-level Adaptation in Southern Alberta, in: Ilbery, B., Chiotti, Q., Rickard, T. (eds.), Agricultural Restructuring and Sustainability: A Geographical Perspective, CAB International, Wallingford, UK: 167–183. Chromow, S.P., 1977, Meteorologia i Klimatologia [Meteorology and Climatology], PWN, Warszawa. Coble, K.H., Knight, T.O., Patrick, G.F., Baquet A.E., 1999, Crop Producer Risk Management Survey: A Preliminary Summary of Selected Data, Agricultural Economics Information Report 99–001, Department of Agricultural Economics, Mississippi State University.

110

Economics and Business Review, Vol. 1(15), No. 3, 2015

Dessai, S., Adger, W.N., Hulme, M., Turnpenny, J., Köhler, J., Warren, R., 2004, Deﬁning and Experiencing Dangerous Climate Change, Climatic Change, no. 64: 11–25. Doroszewski, A., Wróblewska, E., Jóźwicki, T., Mizak, K., 2013, Ocena szkód w roślinach sadowniczych i ogrodniczych spowodowanych przez przymrozki w maju 2011 roku [Damage assessment in fruit farming and horticulture caused by spring frost in May 2011], Acta Agrophysica, no. 20(2): 269–281. Dudek, S., Żarski, J., Kuśmierek-Tomaszewska, R., 2012, Tendencje zmian występowania przymrozków przygruntowych w rejonie Bydgoszczy [Trends in the occurrence of ground frosts in the region of Bydgoszcz], Woda-Środowisko-Obszary Wiejskie, vol. 12, no. 2 (38): 93–106. Eakin, H., 2006, Weathering Risk in Rural Mexico: Climatic, Institutional, and Economic Change, University of Arizona Press, Tucson. Fischoff, B., Slovic, P., Lichtenstein, S., Read, S., Combs, B., 2000, How Safe Is Safe Enough? A Psychometric Study of Attitudes toward Technological Risks and Benefits, in: Slovic, P. (ed.)., The Perception of Risk, Earthscan, London: 80–104. Fraser-Mackenzie, P., Sung, MCH., Johnson. J.E.V., 2014, Toward an Understanding of the Inﬂuence of Cultural Background and Domain Experience on the Effects of Risk-Pricing Formats on Risk Perception, Risk Analysis, vol. 34, no. 10: 1846–1869. Fredrick, S., Loewenstein, G., 1999, Hedonic Adaptation, in: Kahneman, D., Diener, E., Schwarz, N. (eds.), Well-being: The Foundations of a Hedonic Psychology, Russell Sage Foundation, New York: 302–329. Grabowski, J., 2010, The Occurrence of Ground Frost in the Mazurskie Lakeland between the Years 1966 and 2005, Acta Agrophysica, no. 185, Rozprawy i Monografie (6): 99–110. GUS, Statistical Yearbook of Agriculture 2013, http://stat.gov.pl/en/topics/statisticalyearbooks/statistical-yearbooks/statistical-yearbook-of-agriculture-2013,6,8.html [access: 10.02.2015]. Harwood, J., Heifner, R., Coble, K., Perry, J., Somwaru, A., 1999, Managing Risk in Farming: Concepts, Research, and Analysis, Agricultural Economics Report, no. 774. U.S. Department of Agriculture, Washington. IMGW 2013, Niebezpieczne zjawiska meteorologiczne – geneza, skutki, częstość wystę powania [Dangerous meteorological phenomena – their origin, effects and frequency of occurrence], p. 1, wiosna, lato, Warszawa. Justification of the bill on changing the act on crop and livestock insurance subsidies, 2014, http://legislacja.rcl.gov.pl/lista/2/projekt/56568/katalog/56606 [access: 30.09.2014]. Kaczała, M., Wiśniewska, D., 2015, Risks in the Farms in Poland and Their Financing – Research Findings, Research Papers of Wrocław University of Economics (Prace Naukowe Uniwersytetu Ekonomicznego we Wrocławiu), iss. 381, www.ceeol.com. Kalbarczyk, R., 2010, Spatial and Temporal Variability of the Occurrence of Ground Frost in Poland and Its Effect on Growth, Development and Yield of Pickling Cucumber (Cucumis Sativus L.), 1966–2005, Acta Sci. Pol., Hortorum Cultus, no 9 (3): 3–26. Keller, C., Siegrist, M., Gutscher, H., 2006, The Role of the Affect and Availability Heuristics in Risk Communication, Risk Anal., vol. 26, no. 3: 631–639. Klimkowski, C., 2002, Istota, skutki i zarzaądzanie ryzykiem katastroficznym w rolnictwie polskim [The nature, effects and management of catastrophic risk in

M. Kaczała, D. Wiśniewska, Polish farmers’ perception of spring frost

111

agriculture in Poland], Inst. Ekonomiki Rolnictwa i Gospodarki Żywnościowej, Warszawa. Koźmiński, C., Michalska, B., 2001, Atlas klimatycznego ryzyka uprawy roślin w Polsce, [Atlas of climatic risk to crop cultivation in Poland], Akademia Rolnicza w Szczecinie, Uniwersytet Szczeciński, Szczecin. Kundzewicz, Z., 2012, Zmiany klimatu, ich przyczyny i skutki – możliwości przeciwdziałania i adaptacji [Climate changes, their causes and effects – potential precautions and adaptation possibilities], Studia BAS, no. 1 (29): 9–30. Meuwissen, M., Huirne, R., Hardaker, B.,1999, Perceptions of Risk and Risk Management Strategies: An Analysis of Dutch Livestock Farmers, American Journal of Agricultural Economics, no. 81: 1284–1285. Norris, F.H., Smith, T., Kaniasty, K., 1999, Revisiting the Experience-behavior Hypothesis: The Effects of Hurricane Hugo on Hazard Preparedness and Other Self-protective Acts, Basic Appl. Soc. Psychol., no. 21(1): 37–47. Ogurtsov, V.A., van Asseldonk, M.A.P.M., Huirne, R.B.M., 2009, Purchase of Catastrophe Insurance by Dutch Dairy and Arable Farmers, Appl. Econ. Perspect. Pol., no. 31 (1): 143–162. Oltedal, S., Moen, B.E., Klempe, H., Rundmo, T., 2004, Explaining Risk Perception. An Evaluation of Cultural Theory, Rotunde publikasjoner Rotunde, no. 85. Peacock, W.G., Brody, S.D., Highfield, W., 2005, Hurricane Risk Perceptions among Florida’s Single Family Homeowners, Landscape Urban Plann., no. 73 (2–3): 120–135. Riad, J.K., Norris, F.H., Ruback, R.B., 1999, Predicting Evacuation in Two Major Disasters: Risk Perception, Social Influence and Access to Resources, J. Appl. Soc. Psychol., no. 29 (5): 918–934. Sherrick, B.J., Barry, P.J., Ellinger, P.N., Schnitkey, G.D., 2004, Factors Influencing Farmers’ Crop Insurance Decisions, Am. J. Agr. Econ., no. 86 (1): 103–114. Sjöberg, L., Moen, B.E., Rundmo, T., 2004, Explaining Risk Perception. An Evaluation of the Psychometric Paradigm in Risk Perception Research, Rotunde publikasjoner Rotunde, no. 84. Slovic, P., Flynn, J.H., Layman, M., 1991, Perceived Risk, Trust, and the Politics of Nuclear Waste, Science, no. 254: 1603–1607. Tucker, C.M., Eakin, H., Castellanos, E.J., 2010, Perceptions of Risk and Adaptation: Coffee Producers, Market Shocks, and Extreme Weather in Central America and Mexico, Global Envi-ronmental Change-Human and Policy Dimensions, no. 20 (1): 23–32. Tucker, C.M., Eakin, H., Castellanos, E.J., 2010, Perceptions of Risk and Adaptation: Coffee Producers, Market Shocks, and Extreme Weather in Central America and Mexico, Global Envi-ronmental Change-Human and Policy Dimensions, no. 20 (1): 23–32. Wiśniewska, D., 2012, Analiza dyskryminacyjna w prognozowaniu zmian cen akcji. Nowa koncepcja konstruowania prognoz jakościowych, Wydawnictwo Uniwersytetu Ekonomicznego w Poznaniu, Poznań.

Economics and Business Review, Vol. 1 (15), No. 3, 2015: 112–120 DOI: 10.18559/ebr.2015.3.8

Insurance and risk management systems in Russia1 Nadezda Kirillova2

Abstract : The article presents the results of a study of Russian insurance risk management systems. In particular it covers issues related to corporate insurance in the Russian Federation, the method of formation of corporate insurance systems, financial condition assessment tools for insurers and insurance service users and a recommendation for improvements to the insurance enterprise systems in the Russian Federation. This study offers a summary statistical data on the insurance market. Keywords : insurance, risk management, Russian Federation, technical risk, financial stability, policyholder protection. JEL codes : K43.

Introduction An open competitive environment is a requisite for the improvement in productivity, an increase in private investment activity and support of the national economy. Insurance plays an important role in this regard. Insurance is also one of the most efficient, stable and clear mechanisms in managing risks arising from businesses with different needs. The benefit of insurance can be found throughout the last 20 years of capital market development in the Russian Federation. Currently risk management, including risk financing through insurance, of large industrial complexes is no longer just to manage individual risks one at a time but to manage all of them with a built-in, integrated risk management system. For the development of financially and operationally sound insurance systems we need a transparent infrastructure within the systems. This includes but is not limited to a definition and differentiation of insurable risks, identification of risk retention processes, creation of a private insurance system, definition of insurance companies and the proper evaluation of their financial con 1

Article received 21 November 2014, accepted 3 August 2015. Financial University under the Government of the Russian Federation, Financial and Economic Faculty, Department “Insurance Business”, 49 Leningradsky Prospekt, Moscow, Russia, 125993, nvk_66@mail.ru. 2

N. Kirillova, Insurance and risk management systems in Russia

113

ditions. Standardization of the insurance systems, markets, insurers and other critical elements is thus required. This study attempts to address these issues. The article focuses on the organization of corporate insurance in the Russian Federation. The paper is organized as follows. Section 1 describes the general characteristics of corporate insurance in Russia. Section 2 presents the methodology of formation and functioning of corporate insurance programmes. Another part is devoted to the identification of the financial condition of insurers (Section 3). The author’s conclusions and a systematization of corporate insurance systems are shown in Section 4 and are based on the current data and the real experiences of modern corporate industrial insurers.

1. Corporate Insurance in the Russian insurance system The modern private insurance system in Russia can be said to have developed with the adoption of the new law of Insurance in 1992 (or a little earlier with the appearance of the first insurance cooperatives). Historically the driver of this development has been corporate and mostly industrial structure-related. This trend still continues. The consumption of insurance in the 1990’s was, however, often through cases of so-called “grey schemes,” created for the for minimization of income tax liabilities (and encashment in times of crisis). Then came insurance contracts initiated by banks as part of their process of granting loans, as were insurance contracts for foreign partners in international projects. Insurance on property risks as well as (quasi-) social insurance for employees (for example, packages for voluntary accident insurance and voluntary health insurance) began to form in Russia. This reflects the development of a meaningful insurance system, thus protecting consumers in general against the negative consequences of various risks. The share of property insurance – in terms of premiums and mainly for legal entities amounted to 393,82 billion roubles or 43,5% of the total insurance premiums in 2013, and to 420.4 billion roubles or 42.6% of the total insurance premiums in 2014 [The Central Bank 2015]. In 2012 the total volume of property insurance premiums insurance of legal entities amounted to 57% of the market [Federal], see Table 1 and 2 for a summary of data for 2013 and 2014.

2. Methods of corporate insurance programmes A comprehensive insurance coverage for the corporate insured involves the following procedures: –– detection of risks and identification of insurable risks; –– integrated approaches for the assessment of insurable risks;

[114] 21.95

739.13

134.25 31.48

Business and financial risks

Total voluntary insurance

Compulsory insurance of civil liability for motor vehicle owners

Other than medical and motorcycle

Source: http://www.cbr.ru.

Сurrency rate as at 01.01.2014: 1 euro = 45 roubles.

904.86

29.74

Liability insurance

Total voluntary and compulsory

393.82

Property insurance

165.73

208.73

Personal insurance (non‑life)

Total compulsory

84.89

Bln. r.

Life insurance

Type of insurance

20.1

3.7

0.7

3.0

16.4

0.5

0.7

8.8

4.6

1.9

Bln. EUR

100.0

18.3

3.5

14.8

81.7

2.4

3.3

43.5

23.1

9.4

% of total

Premiums

Table 1. Premiums and Claims payments in the Russian Federation 2013

111.1

110.2

109.4

110.3

111.4

110.2

99.4

104.6

113.2

160.5

% (the previous year)

420.77

94.77

17.4

77.37

326.0

1.66

7.14

201.73

103.14

12.33

Bln. r.

9.4

2.1

0.4

1.7

7.2

0.04

0.2

4.5

2.3

0.3

Bln. EUR

100.0

22.5

4.1

18.4

77.5

0.4

1.7

48.0

24.5

2.9

% of total

Claims

112.9

123.1

144.9

119.0

110.3

79.4

133.0

111.0

110.9

92.4

% (the previous year)

[115]

Source: http://www.cbr.ru.

Сurrency rate as at 01.01.2015: 1 euro = 68 roubles.

987.77

27.93

Other than medical and motorcycle

Total voluntary and compulsory

150.92

Compulsory insurance of civil liability for motor vehicle owners

178.85

808.92

Total voluntary insurance

Total compulsory

22.56

Business and financial risks

420.4

Property insurance 37.85

219.58

Personal insurance (non‑life)

Liability insurance

108.53

Bln. r.

Life insurance

Type of insurance

14.53

2.63

0.41

2.22

11.9

0.33

0.56

6.18

3.23

1.6

Bln. EUR

100.0

18.1

2.8

15.3

81.9

2.3

3.8

42.6

22.2

11.0

% of total

Premiums

Table 2. Premiums and Claims payments in the Russian Federation 2014

108.5

107.5

88.7

111.9

108.7

101.2

126.3

105.9

104.5

127.9

% (the previous year)

472.27

109.2

18.89

90.31

363.07

3.54

10.3

224.51

110.49

14.23

Bln. r.

6.28

1.61

0.28

1.33

5.34

0.05

0.15

3.3

1.62

0.21

Bln. EUR

100.0

23.1

4.0

19.1

76.9

0.8

2.2

47.5

23.4

3.0

% of total

Claims

111.4

114.6

108.5

116.0

110.5

208.2

142.9

110.3

106.5

115.3

% (the previous year)

116

Economics and Business Review, Vol. 1(15), No. 3, 2015

–– development of a methodology of forming and using risk maps of industrial and financial risks, liability risks, personal risk; –– self-evaluation of relations and security; estimation of insurance costs and the economic effect; –– creating technical specifications of insurance cover; –– examination of the financial condition of insurers – contractors insurance programmes; monitoring and evaluation of insurance programmes from the standpoint of industrial insurers, as illustrated in Figure 1.

Identification and classification of risks of industrial complexes Industrial audit Survey Consulting

Complex risks of structural subdivisions Register of insurance risks

Quantitative assessment of insurance risks Subdivisions of the enterprise (property management, financial management, risk management department, etc.), technology specialists, appraisers, survey

Definition and differentiation of actual damage, probability of losses by type of risk, the severity of the damage, the maximum amount of damage by type of technology

Formation and portfolio monitoring of industrial complexes Insurer risk management framework brokers, insurance companies, reinsurance companies

Differentiated cost, insurance sums, franchises, limits of liability, levels of coverage

Figure 1. Formation of an insurance system industrial complex

Key elements in the formation of insurance programmes are: –– the limits of liability insurance; –– availability of reinsurance cover; –– the assessment self-insurance – determination of the size of franchises.

117

N. Kirillova, Insurance and risk management systems in Russia

The search for optimal mechanisms to increase the working capital and capital reinvestment requirements lead to the use of new mechanisms for interaction with insurers, reinsurers and reinsurance brokers; requirements cover industrial risks, risks which encourage staff to create captive insurance companies to enter the international reinsurance market, to optimize the methods for identifying and the evaluation of the financial state of insurance companies, Figure 2 [Kirillova 2007, 2008]. Identification of the financial condition of insurance companies and their ranking according to financial stability:

Ranking of insurance companies' base rates for each type of insurance

compliance with the requirements of the financial condition of the insured

Accreditation insurance companies to participate in the insurance programmes

compliance with regulatory requirements

rating assessment by activity

Overall rating of insurance companies for certain types of insurance programmes

Distribution of aggregate own deduction limit insurance companies

Figure 2. Distribution of the aggregate limit of own deduction of insurance companies in the insurance programmes

3. Identification of the financial condition of insurance companies In accordance with Russian legislation the guarantees of the financial stability of the insurer are: –– economically reasonable insurance tarifs; –– insurance reserves, sufficient for the fulfillment of obligations under contracts of insurance, coinsurance, reinsurance, mutual insurance; –– equity; –– reinsurance (minimum capital for Russian companies 120 million roubles on insurance other than life; 240 million roubles for life insurance and 480 million roubles for reinsurance); formation and placement of insurance reserves, covering equity assets.

118

Economics and Business Review, Vol. 1(15), No. 3, 2015

Legislation also defines the basic requirements of the financial condition of insurers: –– solvency margin; –– placement of insurance reserves; –– assets covering equity; –– net assets (for joint stock companies). Along with the regulatory requirements for the financial condition of insurers those insured made their demands. Significant factors in optimal insurance protection can be combined in the following areas: –– history of the insurance company relationships with the company; –– management of the insurance company and its business reputation in the insurance market; –– tarif policy, the promotion and adaptation of insurance products; –– reinsurance; –– financial condition of insurers (set of economic indicators) – Figure 3. Complex analytical indicators of the financial condition of insurance companies

Cash flow

Capital

Asset quality

Premiums and payments

Reinsurance

Profitability; solvency

Investment

Management, goodwill

Figure 3. The main directions of financial state evaluation of insurance companies

If all the legislative requirements are normal, the performance analysis unit can be assessed in four areas: –– insurance, –– reinsurance, –– investment, –– management and goodwill of the insurer.

N. Kirillova, Insurance and risk management systems in Russia

119

4. Modern approaches to the formation of corporate insurance systems in the Russian Federation Currently corporate systems of industrial enterprises in Russia are developing very differently. A description of integrated approaches to the formation of risk management systems with an insurance basis and separate requirements for property insurance in respect of mortgages can easily be identified. For example, Magnitogorsk Iron and Steel Works was amongst the first in the steel industry to have developed and implemented a comprehensive risk management system: a risk management division, an approved risk management policy and a corporate standard for risk management. In the annual report the following risk management policy is stated: –– identification and evaluation of the principal risks faced by MMK group; –– completion of the introduction of a comprehensive risk management system; –– introduction of a multi-faceted approach to the evaluation of risks of building and structural failure; –– improvement of risk management procedures regarding non-fulfilment of payment obligations for steel products; –– to mitigate the risk of production stoppages, the CEO approved the list of plants considered production bottlenecks [OJSC]. In another big energy company, Rossetti, along with compulsory health insurance and compulsory insurance against accidents at work, the company declared support and development of additional corporate voluntary medical insurance, voluntary insurance against accidents and illnesses, private pension provision [Rosseti]. In the consolidated annual accounts it is indicated that the group did not have full coverage for its plant facilities, business interruption, or any third party liability in respect of damage caused to the estate or the environment as a result of accidents or for the Group. As long as the Group obtains adequate insurance coverage, there is a risk that the loss or destruction of certain assets could have a materially adverse effect on the financial position of the Group. In the requirements for the head of risk management Irkutsk Oil Company stated following: –– development of risk management methodology (concepts, risk management policy, risk and specifications, etc.); –– planning for the creation of risk management systems and the achievement of the stated objectives of risk management; –– carrying out purposeful work to identify threats of loss and to identify the sources of risk (analysis reporting, business audits and audit, analysis of management decisions, analysis of the causes of deviations, etc.); –– risk assessment, quality and effectiveness of the existing risk management system;

120

Economics and Business Review, Vol. 1(15), No. 3, 2015

–– establishing and maintaining up to date risk maps for each of these risks; –– software development, coordination and organization of goals set for the detailed risk, being in the responsibility of risk managers and owners; –– ensuring the implementation and application of the necessary organizational and managerial forms of work to achieve the objectives of risk management in their area of r esponsibility, and the active integration in their work; –– control of the accuracy, adequacy or completeness of the approved control procedures and risk management; –– advising on business process optimization and improvement of economic activity of the internal control of the Company and subsidiaries and affiliates; –– controlling the goals enshrined in risk retention within the prescribed limits. –– conducting periodic reporting on the effectiveness of risk management in its area of responsibility for the management unit and the company; logging of data related to sales and any potential risks in their area of responsibility: [OOO Irkutsk]. Thus we see that the understanding of the necessity and value of insurance arises from different sources. Nevertheless there is no common approach as the methodological base is so diverse that it creates certain difficulties in the formation and development of systems for insurance enterprises.

References Federal State Statistics Service, Official website, http://www.gks.ru [access: 13.06.2015]. Kirillova, N.V., 2007, The Mechanism of Formation of Corporate Insurance Coverage of Large Industrial Enterprises, Bashkir University Bulletin, no. 12. Kirillova, N.V., 2008, Insurance Industry: Theory, Methodology, Practice. Abstract of the Thesis for the Degree of Doctor of Economics, Finance University under the Government of the Russian Federation, Moscow. Kirillova, N.V., Bellucci, A., 2014, Key Insurers Indicators in the Reports of Insurance Companies: Russian and Italian Experience, Review of Business and Economics Studies, no. 4. The Central Bank of the Russian Federation, http://www.cbr.ru [access: 10.08.2015]. OJSC Magnitogorsk Iron and Steel Works, http://www.mmk.ru/annualreport [access: 13.06.2015]. OOO Irkutsk Oil Company (INK), http://irkutskoil.com [access: 13.06.2015]. Rosseti, Open Joint Stock Company (JSC ROSSETI), http://www.rosseti.ru [access: 13.06.2015].

Economics and Business Review, Vol. 1 (15), No. 1, 2015: 121–124 DOI: 10.18559/ebr.2015.3.9

BOOK REVIEWS

Jeremy Rifkin, Zero Marginal Cost Society. The Internet of Things, The Collaborative Commons, and the Eclipse of Capitalism, Palgrave Macmillan, New York, 2014: 356, ISBN 978-1-137-27846-3 Jeremy Rifkin is an American economic and social theorist, lecturer, and political advisor. He is the author of many books about the impact of technological changes on the economy and society. He has written above 20 books since 1973. Rifkin’s work has also been controversial. Opponents have attacked the lack of scientific rigor in his claims. His most recent book is The Zero Marginal Cost Society. The last Rifkin’s book consists of five parts and their titles can be treated as the shortest summary: The untold history of capitalism, The near zero marginal cost society, The rise of the collaborative commons, Social capital and the sharing economy, and The economy of abundance. The last part presents the summary of the Rifkin’s post-capitalist age vision. Rifkin’s reasoning is based on historical comparisons and present technology trajectories. The initial infrastructure of both the first and second industrial revolutions in America and Europe was put in place in 30 years, and matured in another 20 years. The first industrial revolution was peaking in the last two decades of 19th century. The second industrial revolution was being born in that time in America and Europe. The discovery of oil, the invention of the combustion engine, and the introduction of the telephone started a new communi-

cation/energy complex that would dominate the 20th century. According to Rifkin, the second industrial revolution peaked and crashed in July 2008, when the price of crude oil hit a record $147 a barrel on world markets. The foundation of the first and second industrial revolutions was a high concentration of economic power. Vertically integrated corporates were the most efficient means of organizing the production and distribution of mass produced goods and services. They reduced transaction costs, increased productivity, lowered the marginal costs of production and distribution, and lowered the price of goods and services, allowing the economy to flourish. While those at the top of the corporate pyramid benefited the most from the increasing efficiency, the living standard of the masses also improved. According to Rifkin (pp. 70–71) we meet the ultimate contradiction at the heart of capitalist system. The driving force of the system is greater productivity. The competitors race to introduce new technologies lowers their production costs and the price of their products and services. The finish line of the race is where the marginal cost of producing each additional unit is nearly zero. When the race is finished, goods and services become nearly free, profits dry up, the exchange

122

Economics and Business Review, Vol. 1(15), No. 3, 2015

of property in markets shuts down, and the capitalist system dies. The leap in productivity is possible because the emerging Internet of Things (IoT) is the first smart infrastructure in history. In the future it will connect every machine, vehicle, residence and business in an intelligent network comprised of a Communications Internet, Energy Internet, and Logistic Internet. According to Rifkin, before 2040 most of the energy to heat houses, power plants and drive vehicles, will be nearly free. A new manufacturing model called 3D printing is developing exponentially along with the other components of the IoT infrastructure. More and more companies are now producing physical products the way software produces information in the form of video, audio, and text. The Energy Internet (a merger of Internet technology and renewable energies) will change the way power is generated and distributed. In the next decades hundreds of millions of people will produce their own renewable energy in their homes, offices, and factories, and share green electricity with each other on an Energy Internet, just as we now generate and share information online. It will allow billions of people to share energy at near zero marginal cost in an IoT net. The IoT will stimulate developments of collaborative commons. For a long time economists regarded the commons as a unique economic model connected to feudal society. Over the past 25 years, a younger generation of scholars and practitioners has begun to rediscover the commons as a governing model. Cooperative is an autonomous association of persons united voluntarily to meet their common economic, social, and cultural needs and aspirations through a jointly-owned and democratically controlled enterprise. Cooperatives are based

on the values of self-responsibility, democracy, and equality. Cooperative members believe in the ethical values of honesty, openness, and social responsibility (p. 211). The first and the second industrial revolutions required huge sums of capital, and therefore had to be organized in vertically integrated enterprises under centralized control to achieve economic of scale. The third industrial revolution based on IoT requires less finance capital and more social capital, scales laterally rather than vertically, and is best implemented by a commons management rather than by capitalist market mechanism. The traditional capitalistic system is defined by private property. According to Rifkin (p. 225), the privately owned car can be the signature item. A car reflects the desire to be free. But for Internet generation freedom means the ability to optimize one’s life by the diversity of experiences. Freedom is measured more by access to networks sharing products and services than ownership of property in markets. Freedom for Internet generation is the ability to collaborate with others, in a peer-to-peer world via Internet. There are two kinds of a sharing economy benefits: rational and emotional (p. 252). The rational benefits given by respondents are the following: saving money, impact on the environment, lifestyle flexibility, practicality of sharing, easy access to goods and services. As for emotional benefits, respondents mentioned above all generosity, followed by a feeling of being a valued part of a community, being smart, being more responsible, and being a part of a movement. When marginal costs drop to near zero, profits disappear. Goods and services will be essentially free. When most things become nearly free, the operat-

Book reviews

ing rationale of capitalism as an organizing mechanism to produce and distribute goods and services becomes meaningless. That is because capitalism’s dynamism feeds off scarcity. If goods and services are scarce, they have exchange value and can be priced in the markets. “Free in price” implies “free from scarcity” (p. 273). According to Rifkin’s theory, it means that, if marginal costs of producing additional units of a good or service is nearly zero, scarcity has been replaced by abundance. The third industrial revolution is following faster. The World Wide Web went on line in 1990 and matured by 2014, connecting much of the human race across a communications medium that operates at near zero marginal costs. The same exponential pattern of Communications Internet growth is moving the Energy Internet forward on a similar timeline, with the prospect of approaching near universal generation of green electricity in many countries at near zero marginal cost in 25 years. The Logistics Internet, although still in its infancy, is likely to run apace. As for 3D printing, it is already experiencing a faster growth trajectory than the Communnication Internet at a comparable stage of development. Rifkin’s vision of the future is quite consistent and comprehensive. Technological progress will reduce costs and prices of products and services. The exponential growth of computers power according to the Moore’s law will push the similar changes in renewable energy and logistics. We have experienced these phenomena during two last decades. Rifkin predicts that this trend will lead to a significant reduction in marginal costs, and that they will be close to zero. As a result, it will lead abundance of goods and the general welfare. And it is the weakest element of his forecast.

123 The crucial concept of the Rivkin’s theory is a marginal cost. Marginal cost is the term used in economics and business to refer to the increase in total production costs resulting from producing one additional unit of the item. Rifikin did not present theoretical background and academic discussion connected with marginal costs, although he presented many other economics concept like for example: relations between technological innovations and productivity (by O. Lange and J.M Keynes), collaborative commons (by E. Ostrom), transactional costs (by R. Coase), or individual happiness (by R. Layard). Textbooks of economics theory teach that marginal costs first go down, but then bottom out and begin to rise again as quantity increases. In the Rifkin’s concept marginal cost does not include initial investment and overhead costs (p. 273). Rifkin describes a world of free fixed costs (e.g. free wi-fi for everyone). To some extent this is happening right now. It is very difficult to defend the thesis that scarcity will disappear. Each new product or service innovation at the beginning is a scarce resource and is expensive. Step by step, following the cumulative experience, the costs and prices are dropping. And prices, as a basic indicator of scarcity will not disappear. The progress we experience in the last two centuries is stimulated by a constant tension of competition an cooperation. Collaborative commons can disseminate by IoT support and enrich the diversity of our societies, but it is doubtful that they will replace competitive markets. Rifkin’s book is regarded as the counterpoint to T.Piketty’s Capital in the Twenty-first Century. Rifkin argues for an optimistic outlook for the rest of this century while Piketty argues for a pessimistic vision. The real future

124

Economics and Business Review, Vol. 1(15), No. 3, 2015

will probably be a combination of both views. And the additional remark for readers: the last book by Rifkin is not very friendly for reading. It consists of above 300 pages with high density of rows, without tables

or figures, which could improve perception of the author’s concepts. Jan Polowczyk

Poznań University of Economics Faculty of International Business and Economics

Economics and Business Review, Vol. 1 (15), No. 2, 2015: 125–126 DOI: 10.18559/ebr.2015.3.10

Andrzej Rzońca, Kryzys banków centralnych. Skutki stopy procentowej bliskiej zera [Central Banks Crisis. The Impact of Interest Rates Close to Zero], Wydawnictwo C.H. Beck, Warszawa 2014: 542, ISBN 978-83-255-5544-3 The book under review deserves attention, both because of its subject matter and because its author is a member of the Monetary Policy Council. The work is aimed at readers who are interested in central banking as well as those professionally involved in the issues of monetary policy and in particular the current interest rates close to zero. The book is also an important academic source on contemporary economic policy. In 2015 90% of the world’s developed economies have a zero interest rate. Shortterm interest rates as low as they are today are unusual. In the economic history of the world only two periods (around 1895 and in the second half of 1930s) were marked by short-term interest rates remaining very close to zero. These conditions undoubtedly pose a challenge for theoreticians and, above all, practitioners, who must take decisions which potentially and actually influence nominal and real variables as well as for those observing the open economy macroeconomics in the Polish and global contexts. The question of modern interest rate volatility and its real value at close to zero or even a negative value should be considered as one of the main effects of the global crisis that require careful analysis. This is also caused by a high degree of liquidity in the global economy which, in conjunction with budget deficits, significantly reduces the influence of monetary and fiscal policy instruments on key macroeconomic variables. The author is a recognized expert in the field of central banking and monetary policy and has previously published works on

interest rates and the said policy. Andrzej Rzońca’s works, mainly co-authored, with a few exceptions are the only Polish titles used in the bibliography of the reviewed book. On the one hand this reflects the status of the author and on the other it expresses his assessment of Polish specialist literature on the subject. This extensive work of 542 pages has an eloquent style. The author aims at presenting existing literature published in English and thus brings this element of modern economics closer to the Polish reader. In the introduction the author declares that the book will endeavour to answer a number of questions of which (p. 7) the most important is: “[…] why, despite the period of six years since the outbreak of the crisis, economic recovery has been so weak” (reviewer’s note: in the USA and in the eurozone). Other questions posed by Andrzej Rzońca are (p. 7): “Why do so few entities start […] business activity, even when – as in the United States today – the net profits of enterprises are the highest in history in relation to GDP? Why is staff turnover so low? Why is the variability of macroeconomic indicators similar to the historical low and does not reflect this uncertainty?” In my opinion the common denominator is that they relate to objective phenomena and processes which have their primary sources in technology and structural changes in the global economy including its real sector. The author is certainly aware of these variables but he does not attribute them their proper meaning in his choice of factors.

126

Economics and Business Review, Vol. 1(15), No. 3, 2015

The work opens with a chapter entitled Introduction to the analysis. This provides an overview and a presentation of the basic concepts and the approach to the analysis. It refers to previous research on the impact of interest rates close to zero. Chapter Two – The impact of the potential sustainability of the natural interest rate at a low level – includes a discussion of the new Keynesian analytical framework for monetary policy, the relationship between central bank policy and the natural interest rate, as well as the potential impact of central bank operations on the expectations of economic entities. The following chapters, are devoted to the impact of an interest rate close to zero on: restructuring and uncertainty after the crisis, credit and other forms of external financing, the ability of the central bank to increase money supply and finally the state of public finances. Chapter Seven – Observations from Japan, compared to Sweden and South Korea – is empirical. The author uses his findings from the previous chapters and based on reports and literature gives an account of the course of events in these three countries. Chapter Eight – Conclusions from the analysis – recommendations for monetary policy – closes the work. The recommendations split into three main areas: “interest rate policy after the outbreak of the financial crisis”, the issue of “quantitative easing and the monetary policy in the countries that do not suffer from the financial crisis due to the unconventional measures taken by major central banks”. The book uses and brings the current global state of knowledge on the subject to the Polish readers. The work is a valuable educational tool for all those interested in the problems of contemporary

central banking. The author’s own analytic diagrams that visualize his outlook on the transmission mechanisms of interest rates close to zero on the financial and the real sector are particularly valuable. The author’s book raises expectations with regard to the totality of the contents and literature. Whilst the first chapter is an introduction and serves as a foundation for further considerations its overall scope leaves the reader unsatisfied. In my opinion the author missed the key factors underlying the current situation concerning the interest rates of central banks probably in order to ensure conciseness and to focus on transmission channels. The attempt to understand the current challenges faced by central banks today will be one-sided, if the following factors are not taken into account: changes in technology, structural changes in the global economy related to the supply and demand shocks created after China entered the global economy, the mistakes of central banks (mainly those of FED after 2001). These factors significantly changed the central banks’ operational environment as the banks faced the problem of a drastic decrease in inflation and expectations of inflation (the loss of directional indicators in the form of inflation) and a drop in real market interest rates. Without a brief reference to that situation it is impossible to correctly answer the questions that the author himself posed in the introduction to his book. The subject matter of the book is important and timely. The work will serve as an inspiration for further research including empirical studies. Tadeusz Kowalski

Poznań University of Economics Faculty of International Business and Economics

Aims and Scope Aims and Scope Economics and Business Economics Review andisBusiness the successor Review toisthe thePoznań successor University to the Poznań of Economics University Review of Economics which wasReview which was published by the Poznań published University by the Poznań of Economics University Press ofin Economics 2001–2014. Press TheinEconomics 2001–2014. and ThBusiness e Economics Review and Business Review is a quarterly journal is a quarterly focusing on journal theoretical focusing andon applied theoretical research andwork applied in the research fields work of economics, in the fields manof economics, management and finance. agement The Review and finance. welcomes The Review the submission welcomes of the articles submission for publication of articles dealing for publication with micro, dealing with micro, mezzo and macro mezzo issues. and All macro texts are issues. double-blind All textsassessed are double-blind by independent assessedreviewers by independent prior to reviewers acceptance. prior to acceptance.

Notes for Contributors Notes for Contributors 1. Articles submitted 1. Articles for publication submittedinfor thepublication Economicsinand theBusiness Economics Review and Business should contain Revieworiginal, should contain original, unpublished workunpublished not submitted work for not publication submitted elsewhere. for publication elsewhere. 2. Manuscripts intended 2. Manuscripts for publication intended should for publication be writtenshould in English be written and edited in English in Word andand edited sentin to:Word and sent to: review@ue.poznan.pl. review@ue.poznan.pl. Authors should upload Authors twoshould versions upload of their twomanuscript. versions of One theirshould manuscript. be a comOne should be a complete text, while inplete the second text, while all document in the second information all document identifying information the author(s) identifying should thebe author(s) removed should be removed from files to allowfrom themfito lesbe tosent allowtothem anonymous to be sent referees. to anonymous referees. 3. The manuscripts 3. are Thetomanuscripts be typewritten arein to12’ be typewritten font in A4 paper in 12’ format font inand A4 be paper left-aligned. format and Pages be left should -aligned. Pages should be numbered. be numbered. 4. The papers submitted 4. The should papers submitted have an abstract shouldofhave not an more abstract than 100 of not words, morekeywords than 100and words, the keywords Journal and the Journal of Economic Literature of Economic classification Literature code.classification code. 5. Acknowledgements 5. Acknowledgements and references to grants, and references affiliation, to postal grants,and affiliation, e-mail addresses, postal andetc. e-mail should addresses, appearetc. should appear b, etc a, b, etc as a separate footnote as a to separate the author’s footnote name toa,the author’s name and should not be and included should innot the be main included list of footnotes. in the main list of footnotes. 6. Footnotes should 6. Footnotes be listed consecutively should be listed throughout consecutively the text throughout in Arabicthe numerals. text in Arabic Cross-references numerals. Cross-references should refer to particular should refer section to particular numbers: e.g.: section Seenumbers: Section 1.4. e.g.: See Section 1.4. 7. Quoted texts of7.more Quoted thantexts 40 words of more should thanbe 40separated words should frombe theseparated main body from by the a four-spaced main bodyindenby a four-spaced indentation of the margin tation as aof block. the margin as a block. 8. Mathematical notations 8. Mathematical should meet notations the following should meet guidelines: the following guidelines: – symbols representing – symbols variables representing should be variables italicized, should be italicized, – avoid symbols above – avoid letters symbols and use above acceptable letters and alternatives use acceptable (Y*) where alternatives possible, (Y*) where possible, – where mathematical – where formulae mathematical are set out formulae and numbered are set out these andnumbers numbered should thesebe numbers placed against should be placed against the right margin as... the (1), right margin as... (1), – before submitting – before the final submitting manuscript, the ficheck nal manuscript, the layout of check all mathematical the layout of formulae all mathematical carefullyformulae carefully (including alignments, (including centring alignments, length of centring fraction lines lengthand of fraction type, sizelines and and closure type,ofsize brackets, and closure etc.), of brackets, etc.), – where it would –assist where referees it would authors assistshould refereesprovide authorssupplementary should provide mathematical supplementary notes mathematical on the notes on the derivation of equations. derivation of equations. 9. References in the 9. text References should be in the indicated text should by thebeauthor’s indicated name, by the date author’s of publication name, date andofthe publication page num-and the page number where appropriate, ber where e.g. Acemoglu appropriate, ande.g. Robinson Acemoglu [2012], and Robinson Hicks [1965a, [2012], 1965b]. HicksReferences [1965a, 1965b]. shouldReferences should be listed at the endbeoflisted the article at the in end theofstyle the article of the in following the styleexamples: of the following examples: Acemoglu, D., Robinson, Acemoglu, J.A.,D., 2012, Robinson, Why Nations J.A., 2012, Fail.Why The Origins Nations of Fail. Power, The Prosperity Origins of Power, and Poverty, Prosperity and Poverty, Profile Books, London. Profile Books, London. Kalecki, M., 1943, Kalecki, Political M., Aspects 1943, ofPolitical Full Employment, Aspects of Th Full e Political Employment, Quarterly, The Political vol. XIV,Quarterly, no. 4: 322–331. vol. XIV, no. 4: 322–331. Simon, H.A., 1976,Simon, From Substantive H.A., 1976,toFrom Procedural Substantive Rationality, to Procedural in: Latsis, Rationality, S.J. (ed.),in: Method Latsis,and S.J. Appraisal (ed.), Method and Appraisal in Economics, Cambridge in Economics, University Cambridge Press, Cambridge: University Press, 15–30. Cambridge: 15–30. 10. Copyrights will 10.beCopyrights establishedwill in the be established name of theinE&BR the name publisher, of the namely E&BR publisher, the Poznań namely University the Poznań of University of Economics Press. Economics Press. More informationMore and advice information on the and suitability advice and on the formats suitability of manuscripts and formats canofbe manuscripts obtained from: can be obtained from: Economics and Business Economics Review and Business Review al. Niepodległościal. 10Niepodległości 10 61-875 Poznań 61-875 Poznań Poland Poland e-mail: review@ue.poznan.pl e-mail: review@ue.poznan.pl www.puereview.ue.poznan.pl www.puereview.ue.poznan.pl

Subscription Economics and Business Review (E&BR) is published quarterly and is the successor to the Poznań University of Economics Review. The E&BR is published by the Poznań University of Economics Press. E&BR is listed in ProQuest, EBSCO, and BazEkon. Subscription rates for the print version of the E&BR: institutions: 1 year – €50.00; individuals: 1 year – €25.00. Single copies: institutions – €15.00; individuals – €10.00. The E&BR on-line edition is free of charge. Correspondence with regard to subscriptions should be addressed to: Księgarnia Uniwersytetu Ekonomicznego w Poznaniu, ul. Powstańców Wielkopolskich 16, 61-895 Poznań, Poland, fax: +48 61 8543147; e-mail: info@ksiegarnia-ue.pl. Payments for subscriptions or single copies should be made in Euros to Księgarnia Uniwersytetu Ekonomicznego w Poznaniu by bank transfer to account No.: 96 1090 1476 0000 0000 4703 1245.