Production efficiency and policy impact of heterogeneous farm households in developing countries

Abstract

Agricultural development is an essential factor in the economic development of much of the developing world and comprises a significant element of foreign assistance portfolios. Over the last decade, there has seen a renewed interest in more credible estimates of the economic impacts of development programs, such as assistance to extension programs. We compare the estimation of technical efficiency to farm output and income as an outcome variable to evaluate the impact of development programs such as farm education and extension programs. We develop a simple theoretical model which shows that using technical efficiency as an outcome variable could be a viable alternative to more traditionally used outcome variables such as farm output and farm profit. We note that when farmers are capital constrained, extension programs can theoretically have a large efficiency effect despite a small or zero change in farm profits.

If farm technical efficiency is used as an outcome variable, then it must be estimated correctly. Mismeasurement of farm technical efficiency leads to misleading extension program evaluations. Farm households face heterogeneous infrastructural constraints (Suri 2011; Ojiem et al. 2006), credit constraints, information barriers and other input market constraints (Duflo, Kremer and Robinson 2011; Jack 201; Suri 2011and Stifel and Minten 2008), labor markets constraints (Henning and Henningsen 2007), socio-economical (Ojiem et al. 2006) and non-farm income opportunities (Chang et al. 2012) and thus have different access to agricultural inputs and outputs. These constraints have a substantial impact on agricultural production decisions of farm households. A key production decision of farm households is the allocation of resource to cash and food crops. Production of cash crops requires relatively higher market involvement in both the purchase of inputs and the selling of output than home-consumed food crops. The heterogeneous constraints across farm households leads to a substantial imbalance in the transaction costs associated with the production of each crop. Moreover, home-consumed crops may have quality attributes (e.g. color, taste, softness of dough, and suitability for certain dishes) not reflected in market prices. Factors such as transaction costs, crop quality attributes, and other factors such as household characteristics are farmer specific and drive a heterogeneous price wedge between the market prices for household’s crop production and the economic value of these crops for the household. These distinctions have important implication for farm productivity analysis, such as technical efficiency measurement.

The standard approach to productivity analysis, such as efficiency estimation, assume that farm households face homogenous price wedges that leads to homogenous set of production and profit functions. However, the price gap created by transaction costs, crop quality attributes, and other factors such as household characteristics generally vary among subsistence, semi-subsistence and commercial farmers and leads to a heterogeneous set of profit and production frontiers. Subsistence and semi-subsistence farmers who produce largely home consumed crops have potentially greater price wedges than commercial farmers. Failing to account for the heterogeneity in price wedges that drive heterogeneity profit and production frontiers is likely to lead to underestimation of the efficiency of subsistence and semi-subsistence farmers. We test if traditional productivity analysis indeed underestimates the efficiency of subsistence and semi-subsistence farmers by employing a conditional Data Envelopment Analysis (DEA) model for household survey data in Uganda. Results confirm that naïve estimates of efficiency understate the efficiency scores of subsistence and semi-subsistence farmers. The results cast doubt on policies, such as extension programs or other information treatments, that interpret low efficiency scores for subsistence and semi-subsistence farmers as a management shortfall.

We demonstrate the use of farm technical efficiency as an outcome measure by analyzing data from 2008-2012 for farm training program in Armenia. In this program, farmers received technical guidance on modern farm techniques. Two previous evaluations (Schwab and Shanoyan 2016; Fortson et al. 2012) find ambiguous evidence that farm profits increased. The measurement or potential gain from an extension program is captured using farm technical efficiency measures. We find evidence that the program in Armenia increased farm technical efficiency from 2008 to 2012.