Contract farming as partial insurance

•We study the relationship between participation in contract farming and income variability.•We use cross-sectional data on 1,200 households in Madagascar.•We find a systematic association between participation in contract farming and lower income variability.•This suggests that contracts can be used as partial insurance mechanisms in places where insurance markets fail. A core result of contract theory is that contracts can help transfer risk from one party to another, the latter insuring the former. We test this prediction and explore the mechanism behind it in the context of contract farming, the economic institution wherein a processor contracts the production of a commodity to a grower. Specifically, we look at whether participation in contract farming is associated with lower levels of income variability in a sample of 1,200 households in Madagascar. Relying on a framed field experiment aimed at eliciting respondent marginal utility of participation in contract farming for identification in a selection-on-observables design, we find that participation in contract farming is associated with a 0.20-standard deviation decrease in income variability. Using mediation analysis to look at the mechanism behind this finding, we find support for the hypothesis that fixed-price contracts—which transfer all price risk from the grower to the processor—explain the reduction in income variability associated with contract farming. Because the assumption that makes our selection-on-observables design possible also satisfies the conditional independence assumption, we estimate propensity score matching and doubly robust weighted regression estimators, the results of which show that our core results are robust and that participation in contract farming would likely be more beneficial for those households that do not participate than for those who do. Our findings thus support the notion that, in a context where formal insurance markets fail, contracts can serve as partial insurance mechanisms.