Farm size-performance relationship: A review

The report assesses the relationship between land size and performance in the developing world. Farm and plot performance data were gathered through an exhaustive review of mostly peer-reviewed publications over the last 22 years (1997-2018) in English, French and Spanish. Following the screening of the material, a selection of 472 papers was reviewed, creating a pool of over 1100 individual observations or cases. Both specific and general agricultural economics studies using land area as explaining variable in their performance estimates were explored. Three groups of indicators (i.e. gross output, net value and efficiency) were analysed according to area size in an effort to capture global indicators of performance, beyond the too often used partial indicators (e.g. yield or gross value per area). Analyses based on farm data show that there has been a revival of interest on the question particularly on sub-Saharan Africa (SSA) agriculture, given the increased rate of specific literature publications. The review looked for evidence documenting the various possible relationships that could relate the size of an agricultural holding to its performance (i.e. direct, inverse and non-monotonic). The main explanations shaping the size-performance relationship were explored, namely: the contextual rural input market (i.e. labour, land, input, etc.) imperfections but also methodological shortcomings of the existing literature. On the one hand, inverse relationship (IR) is clearly the dominant type of interaction between cropped land area and agricultural performance using the most common performance indicator group used (gross output mainly populated by studies using yield or total value). However, the economic literature has clearly demonstrated that the use of this type of indicator of performance is generally ill-advised in assessing the farm size performance relationship. On the other hand, the less frequent but more global productivity indicator group of "efficiency" and "net values" do not report such a clear-cut relationship. As a matter of fact, cases using "efficiency" performance indicators are more likely to record a direct relationship than IR. Moreover, the emergence of non-monotonic relationships needs to be highlighted showing that the relationship may not be constant. Tests conducted on the existing material clearly associate a number of rural factor market imperfections with the prevalence of the IR. Hence, IR is more likely to be a symptom of imperfections and lack of opportunities for rural labour than an advantage of a given type of farms. In turn, methodological reasons explored also indicate that narrower ranges of farm size in a given study increase the reporting of IR, particularly in SSA and when analysing partial performance indicators. From being an established stylised “fact” in development economics, IR may not be taken for granted because of empirical complexities in accurately assessing it but also because there is evidence that such a relationship depends on the performance indicator analysed. Hence, IR may not necessarily be considered systematic, continuous, stable through time, irreversible or universal. From a broader development intervention perspective, and based on the review results, the recommended performance indicators (i.e. net value and efficiency) show that larger farms tend to be more performant than the smaller farms. However, this does not suggest the abandonment of smallholders by policy as there are both critical economic and social justifications for the direct improvement of the living conditions of a large share of the population in most of the developing world. It rather advocates a revisited and expanded development role for medium sized ones.