14 B o o s t i n g
P r o d u c t i v i t y i n S u b - Sa h a r a n A f r i ca
likely overestimate or underestimate the degree of misallocation because they would reflect both the true share and a sampling error. The WBES-based measure of misallocation would tend to be overestimated if sectors with higher misallocation are overrepresented relative to their shares in the census. Evidence for African countries shows that most industries would have smaller misallocation in the WBES than their dispersion in the census data. Hence, the WBES might underestimate the true misallocation of each sector and therefore underestimate manufacturing productivity dispersion (Cirera, Fattal-Jaef, and Maemir 2018).
Limited Interpretation of Microeconomic Evidence One of the most widely used measures of firm-level productivity in the literature is total factor productivity revenue (TFPR)— typically defined as the ratio of firms’ sales (or revenues) to input costs (appropriately weighted by their production elasticities). It has been argued that TFPR is a measure of profitability (or firm performance) rather than productivity. Hence, differences in TFPR across firms may capture not only differences in physical efficiency but also differences in prices, which reflect product differentiation and markups in addition to costs (De Loecker and Goldberg 2014). The emergence of (output and less often input) price data and new techniques applied to databases with firm-level prices has enabled researchers to compute more accurate measures of physical efficiency. Evidence on the use of these techniques for emerging markets is presented in Cusolito and Maloney (2018) and references therein. Future work in Africa needs to distinguish productivity shocks (or technical efficiency) from demand shocks in the measures of TFPR among Sub-Saharan African production establishments. This requires the timely availability and recurrent production of high-quality data on output and input prices at the establishment level—a task that does not preclude improving the country coverage
as well as the methodology and periodicity of firm-level censuses. Such new and increased data impose other challenges: (a) wider availability of output price data rather than input price data at the establishment level; (b) reported output prices that are, in most cases, unit values; and (c) the need to undertake surveys at the product level if most manufacturing establishments in a specific sector are multiproduct. Having greater data availability on output and input prices does not prevent the need to impose more structure to identify the role played by demand shocks in the measured TFPR. Recent research using firm-level census with price data shows that there is still a larger dispersion of TFPR across manufacturing firms in Ethiopia, and this is mirrored by large differences in physical productivity. Prices tend to vary significantly less than productivity levels and do not constitute a major driving factor of TFPR differences (Söderbom 2018).
Plan of the Volume This volume documents the productivity trends in Sub-Saharan Africa in three different dimensions, assessing productivity at the aggregate level, the sectoral level, and the establishment level. It characterizes the evolution of productivity in the region relative to other countries and regions as well as country groups in Africa classified by their degree of natural-resource abundance and condition of fragility. The core of this volume rests upon the assessment of the implications for aggregate productivity of production decisions across agricultural farms and manufacturing firms in Sub-Saharan Africa. The next three chapters will present evidence on aggregate productivity from the perspective of production units, using recent household surveys for farmers and firm-level surveys for select African countries as well as frontier estimation techniques. The empirical work presented in this volume can provide further guidance for productivity analysis and the design of a policy agenda for the region.