Contract farming is considered by most authors to be a positive development for agricultural innovation in developing countries. However, there is serious concern whether smaller farmers can benefit from these arrangements. This systematic review analysed the evidence in the literature on income effects for smallholders.'
The objective of the review was to understand the effect of contract farming on income, and under which enabling, or limiting, conditions contractual arrangements are more effective.
The review included all studies on contract farming in low- and middle-income countries that had an econometric design to reduce selection bias in effect estimates. Contract farming was defined as a sales arrangement between a farmer and a firm, agreed before production begins, which provides the farmer with resources or services. The search (1 October 2015) covered all relevant electronic libraries and yielded 8,529 hits, 195 of these were related to contract farming, 75 studied effectiveness and 23 had a proper econometric design to be included in the meta-analysis. Double-screening was used throughout.
Headline Findings: a summary statement
The literature suffers from publication and survivor bias; programmes with non-significant effects are systematically underreported and all the studies assessed the effectiveness of the contractual arrangement after they had survived the start-up phase (which may have a very high attrition rate).'The included studies are, therefore, really only representative of enduring contract farming arrangements. The average income effect was found to be 38%, when controlling for the publication bias,'but this is still likely to be an overly optimistc result and the true programme effect of contract farming is likely to be much lower. The authors also found that contracted farmers have'significantly larger holdings, and more assets,'than the average farmers in the studies which suggests that the poorest famers are being excluded from the programmes.
The meta-analysis was based on data from 22 studies, covering 26 empirical instances of contract farming, covering 13 countries (including Indonesia, Nigeria, Madagascar, and Peru).
Implications for policy and practice
The biases identified in the literature mean that strong conclusions can not be drawn. However, the large income effect does suggest that for contracting to be successful it must offer large benefits to outweigh the transaction costs and loss of autonomy. Additionally, as the contracted farmers had significantly above average assets, this type of intervention may be more appealing to farmers who can both invest more and accept more risk.
The authors also found tentative evidence for how to increase program effectiveness:
- For perennial crops, a price premium seems to be highly effective and it is better if the famers pay cash for the inputs than receive credit from the contracting firm (though the firm can facilitate access to, and quality of, the inputs).
- For annual crops, a price premium seems to be necessary for the contract to work and if the local market pays competitive prices then farmer organisations can facilitate the start of farmer-firm relations, however, if these organisations don't exist then contracts with higher than market prices are significantly more effective.
- For animal husbandry, packages that provide both inputs and credit are highly effective.
It is worth noting that contract farming was only run in certain regions of countries, likely those with good access to infrastructure or nearby processing facilities, and the results may not be applicable to other areas.
Implications for further research
Research should take care to cover the performance and dynamics of contract farming over time, to capture variations in income effects and drop-out rates, and also publish non-significant results.
Overall, there can be medium confidence in the results of this systematic review; it is a comprehensive and statistically rigorous exploration of contract farming in low- and middle-income countries that offers measured conclusions based on the best available evidence. However, the review has the following limitations: the authors did not report independent data extraction by at least two reviewers and the authors did not report results by risk of bias.