Experimental Design Details
In order to assess the effect of short-term credit constraints on labour allocation decisions, we propose to conduct a cluster-randomized trial. As part of the study, farmers will be allocated to 3 groups: a control group, a maize loan group, and a cash credit group. In the first year of the trial, credit interventions will not be announced until January, which is the beginning of the “hungry season” and, with the final harvest only 3 months away, the beginning of the most labour-intensive part of the farming season. At this point in the season, farmers have made their cropping decisions. The main objective of this first phase of the project is to assess the labour supply impact of short-term credit constraints, i.e. to assess the degree to which additional resources affect household labour allocation. The primary goal is to 1) measure the degree to which the elimination of credit constraints lowers the provision of off-farm labour, 2) increases the hiring of outside labour in the short-run, and 3) affects final harvest outcomes.
Maize loans will be offered in early January, and will give farmers the option to obtain one 50kg bag of maize per month from January to March. Since zero-interest loans constitute a net transfer to farms, we will provide direct income transfers to a subset of farmers. The comparison of farmers in main treatment groups with the subsample of farmers in the control group receiving direct income transfers will allow us to separate the credit access mechanisms from the direct income effect of the maize program.
While one of the innovation of the proposed study lies in the provision of food loans, the fundamental mechanism addressed in this proposal is the impact of short-term credit constraints on agricultural labor supply. The comparison between the control and the cash credit arm will allow us to directly assess the degree to which farming outcomes are affected by credit constraints. While we conjecture that maize loans will have a larger impact on farm productivity than cash loans, the relative difficulty of enforcing cash repayment is likely to affect both uptake and repayment rates, and thus also the relative feasibility of both programs. The parallel introduction of cash credits will allow us to directly assess these claims empirically, i.e. to assess whether maize is used differently from cash credit and if yes, which loan program compares more favorably with respect to final agricultural production.
While short-term increases in production are of interest, the most important question addressed in the proposed study is whether short-term improvements in yields can have persistent positive effects on agricultural productivity. Higher yields resulting from the intervention in Year I should lower the likelihood of households facing food shortages in the subsequent year. A very successful short-term intervention therefore has the potential to permanently alter household productivity. These effects may be even larger if farmers are aware of future credit options at the beginning of the farming season. The availability of credit during the farming season should, as outlined in the theoretical model above, alter farmer’s plot size and crop choices at the beginning of the season, and lead to larger overall adjustments. To distinguish wealth from anticipated credit access mechanisms, 50% of treated farmers will be offered a continuation of the loan programs in Phase II of the program, while 50% of farmers will not be offered any further loan options. We expect farmers who received a loan in Year I to be on average wealthier at the beginning of the second season. By comparing farmers with continued programs to farmers who no longer get the programs we will be able to identify the additional effects (and extensive margin adjustments) achieved by allowing farmers to adjust their production plan to newly available credit options before starting field operations. We will measure wages within the study villages and on nearby large farms to help identify more general labor market impacts.