This Campbell systematic review examines the impact of contract farming on income and food security of smallholder farmers in low‐ and middle‐income countries. The review summarises findings from 75 reports, of which 22 (covering 26 contract farming interventions) were used for meta‐analysis Contract farming may substantially increase farmer income with an average effect in the range of 23 to 54 per cent. There is upward bias in the estimate because of survivor bias in individual studies (no data on farmers who drop out of schemes) and in the body of evidence (no studies on contract farming arrangements that collapsed in their initial years), and publication bias in the literature (under‐reporting of insignificant outcomes). Therefore, some caution is needed in interpreting the findings. For farmers to give up their autonomy in marketing and prevent side‐selling, substantial income gains need to be offered. This is especially so for annual crops and when firms have contracts directly with farmers rather than through a cooperative. Poorer farmers are not usually part of contract farming schemes. In 61% of the cases, contract farmers had significantly larger landholdings or more assets than the average farmers in the region. Plain language summary The effectiveness of contract farming for raising income of smallholder farmers in low‐ and middle‐income countries The evidence suggests that contract farming arrangements need to offer clear incentives to farmers in order to survive over time. The review in brief Contract farming, a sales arrangement between a farmer and a firm, is popular with government and donors. Contract farming can produce substantial income gains for farmers. Moreover, these benefits may well be required for contract farming schemes to survive. Better‐off farmers are most likely to participate in contract farming schemes. What is this review about? Contract farming is a sales arrangement agreed before production begins, which provides the farmer with resources or services. The service package provided by the firm varies per location, and can include transport, certification, input provisioning and credit. This systematic review summarises evidence on income effects for smallholders to assess average effects and explore combinations of factors that increase these effects. What studies are included? Included studies had to examine the impact of contract farming on income and food security of smallholder farmers in low‐ and middle‐income countries. Studies had to use a comparison group with appropriate statistical methods to allow for selection effects. Seventy‐five studies were identified with quantitative estimates of the impact of contract farming of which 22 studies, covering 7,471 respondents, were of sufficient rigour to include in the meta‐analysis of income effects. The meta‐analysis covers 26 empirical instances of contract farming in 13 developing countries. What is the aim of this review? This Campbell systematic review examines the impact of contract farming on income and food security of smallholder farmers in low‐ and middle‐income countries. The review summarises findings from 75 reports, of which 22 (covering 26 contract farming interventions) were used for meta‐analysis. What are the main findings of this review? Contract farming may substantially increase farmer income with an average effect in the range of 23 to 54 per cent. There is upward bias in the estimate because of survivor bias in individual studies (no data on farmers who drop out of schemes) and in the body of evidence (no studies on contract farming arrangements that collapsed in their initial years), and publication bias in the literature (under‐reporting of insignificant outcomes). Therefore, some caution is needed in interpreting the findings. For farmers to give up their autonomy in marketing and prevent side‐selling, substantial income gains need to be offered. This is especially so for annual crops and when firms have contracts directly with farmers rather than through a cooperative. Poorer farmers are not usually part of contract farming schemes. In 61% of the cases, contract farmers had significantly larger landholdings or more assets than the average farmers in the region. What do the findings of this review mean? Contract farming covers a wide range of contractual arrangements. This heterogeneity makes it difcult to draw general conclusions from the literature published on this topic. The lack of studies on ‘failed treatments’ leads to an overestimation of the effectiveness of contract farming. Moreover, the analysis suggests a marked publication bias; all studies report on at least one case of contract farming that has a positive and statistical significant income effect. Relatively larger or richer farmers can cope better with these risks and are, therefore, more likely to take part in a contractual arrangement. This implies that contract farming is more suited to the relatively better‐off segment of the farming population. Further research should: (1) improve reporting of the intervention; (2) document the less‐successful instances of contract farming, and report inconclusive results (insignificant effects); and (3) capture other outcomes of contract farming such as (sector‐wide) innovation, and livelihood resilience. How up‐to‐date is this review? The review authors searched for studies published up to October 2015. This Campbell systematic review was published in December 2017. Executive summary/Abstract Background Contract farming is used by an increasing number of firms as a preferred modality to source products from smallholder farmers in low and middle‐income countries. Quality requirements of consumers, economies of scale in production or land ownership rights are common incentives for firms to offer contractual arrangements to farmers. Prices and access to key technology, key inputs or support services are the main incentives for farmers to enter into these contracts. There is great heterogeneity in contract farming, with differences in contracts, farmers, products, buyers, and institutional environments. The focus of this review lies on contract farming, defined as: “a contractual arrangement for a fixed term between a farmer and a firm, agreed verbally or in writing before production begins, which provides material or financial resources to the farmer and specifies one or more product or process requirements for agricultural production on land owned or controlled by the farmer, which gives the firm legal title to (most of) the crop or livestock” (Adapted from Prowse, 2012:12). The last decade shows a rapid increase in studies that use quasi‐experimental research designs to assess the effects of specific empirical instances of contract farming on smallholders. The objective of this systematic review was to distill generalised inferences from this rapidly growing body of evidence. Objectives The review synthesised the studies in order to answer two questions: • Question 1: What is known about the effect size of contract farming on income and food security of smallholder farmers in low‐ and middle‐income countries? • Question 2: Under which enabling or limiting conditions are contract farming arrangements effective for improving income and food security of smallholders Search methods A comprehensive electronic search was applied to Scopus, CAB Abstracts, Econlit, Web of Science, Tropag & Rural, and Agricola between 30 September and 21 October 2015. Snowballing the reference list in review articles and other repositories of research (e.g. worldwidescience.org, FAO, World Bank, Google Scholar) added more studies to the review. The search results were uploaded in EPPI Reviewer 4 and screened for relevance and the rigour of analysis of the effect estimates, in order to combine these results in a meta‐analysis of effectiveness. The main terms used to identify the pool of studies within which we expected to find studies that covered the effectiveness of contract farming arrangements were: contract farming, nucleus estate, cooperative, producer organisation, pre‐harvest agreement, value chain, farm‐firm, outgrower, and vertical integration. Selection criteria Each study selected for the meta‐analysis was required to resolve the counterfactual, that is, to use a comparison group to mimic the expected situation of farmers not having a contract. When assessing net‐effects, the characteristics of groups with or without a contract needed to be fairly similar. Ideally, the only difference was the condition of having a contract or not. Because firms tend to offer contracts to farmers having certain characteristics and farmers self‐select when they accept or reject the offer, econometric methods are required to credibly assess the net‐effects of contract farming. To be included in the review, studies needed to analyse the impact of the intervention on income or food security of smallholder farmers. However, only one study was found with food security as an outcome variable (Bellemare & Novak, 2017); all other studies included focused on income effects. The review, therefore, has a focus on the income effects of contract farming and a meta‐analysis explored this outcome. Data collection and analysis The electronic search retrieved 8,529 unique studies. After the full‐text screening, 195 studies were found to present research on contract farming. We excluded all papers that did not study the effectiveness of contract farming. The remaining set of papers was referred to as the core set and consisted of 75 studies that presented quantitative outcomes on smallholder farmers. Of the 75 studies in the core set, most did not meet the criteria for methodological and econometric rigour and had to be excluded from the meta‐analysis. The meta‐analysis was based on data from 22 studies, covering 28 empirical instances of contract farming, two of which had insufficient data to use in the meta‐analysis. The studies covered 7,471 respondents. Results We applied meta‐analysis on the studies that reported income effects. Based on the significance levels and effect sizes, we showed that the set of studies selected for meta‐analysis suffered from publication bias. All studies reported at least one empirical instance with a statistically significant positive income effect. Test results suggested that studies with non‐significant effects of contract farming are likely to exist but are not reported in the academic literature. Studies also suffer from survivor bias. All studies are cross‐sectional studies that assess the effectiveness of the contractual arrangement at one moment in time, but only after the contractual arrangements had been in place for some years when these had already survived the start‐up problems. This implies that contractual arrangements that had ceased to function are absent in the literature. The publication and survivor bias detected in this review preclude strong conclusions on the income effects of contract farming arrangements. The one study that analysed effects on food security reported a positive result (the duration of the hungry season was 8% [95% confidence interval= 0%,15%] lower for farmers having a contract). The meta‐analysis of the 22 studies showed that the average income effect of participation in these contractual arrangement is highly heterogeneous. The (uncorrected) pooled average effect‐size on the proxy for income used in each study, computed in the meta‐analysis, indicated a 62 percent increase (95% confidence interval=40%, 87%) in income for contract farmers over incomes of non‐contract farmers. However, strong evidence for publication bias suggests that the true effect of contract farming is likely to be much lower, although still substantially higher than non‐contract farming. When we consider these studies representative for enduring contract farming arrangements in general, the pooled average income effect is estimated in 38% (95% confidence interval=23%, 54%). In almost two‐thirds of the studies, the contracted farmers proved to have significantly larger holdings or to be richer than the average farmers in the area. A plausible explanation for this phenomenon, as mentioned in the studies, is that there are lower transaction costs with increasing farm scale and the capacity of the better‐endowed farmers to bear the production and post‐harvest quality risks inherent to contract farming arrangements. In the four studies in which contract farmers had relatively smaller than average plot sizes, the income effects are relatively low. Authors’ conclusions Contract farming is a container concept that covers a wide range of contractual arrangements. This heterogeneity makes it difcult to draw general conclusions from the literature published on this topic. The studies have a marked publication bias. All studies report at least one case of contract farming that has a positive and statistical significant income effect. Moreover, due to limits inherent to the (cross‐sectional) study designs used in these investigations, the estimated effect size is upward biased. The lack of studies on ‘failed treatments’ leads to an overestimation of the effectiveness of contract farming. Nevertheless, the results of the meta‐analysis suggest that contract farming arrangements need to offer clear incentives to farmers in order to survive over time in the context of free entry and exit of farmers. We generated the hypothesis that relatively large positive effects on income may be a precondition for farmers to continue the contractual arrangements with the firm and give up their autonomy in marketing, production and quality control. High benefits are needed to keep an arrangement attractive and to prevent farmers from dropping out. Modest expectations and careful planning are needed for contract farming to be effective and sustainable. The practitioner‐oriented literature indicated the high risk of failure in the first years and stressed the need for adaptive management and mechanisms to settle disputes. Whereas it is unlikely that contract farming arrangements will on average result in the income effects that we derived from the meta‐analysis, it shows the need for substantial income effects for contract farming arrangements to survive over time. If farmers may opt out of a contract ‐ which was the case in all empirical instances covered by the studies except oil palm in Indonesia ‐ those contractual arrangements having low effects are likely to disappear or be amended, and negative effects on smallholder well‐being are unlikely. Contract farming is an institutional arrangement that may be attractive for farmers who want to get access to services or inputs that they cannot obtain in the traditional (spot) market, or reach markets that are more remunerative. Farmers who are able to enter a contract farming arrangement tend not to be the poorest farmers in their region. Both firms and farmers face risks of non‐compliance. Relatively larger or richer farmers can cope better with these risks and are, therefore, more likely to take part in a contractual arrangement. This implies that contract farming is more suited to the relatively better‐off segment of the farming population. For annual crops, a price premium seems to be a necessary component of the service package in order to result in high income effects for farmers, especially in situations where no cooperative is involved as an intermediary between the firm and the farmers.