Confidence and information usage: Evidence from soil testing in India

Abstract Informational barriers are often considered to be a major constraint to the adoption of improved farming practices, inputs, and technologies by smallholder farmers. In the Indian context, it is widely believed that farmers misapply chemical fertilizers because they lack scientific information on soil conditions and corresponding fertilizer recommendations, thus resulting in imbalanced and potentially detrimental fertilizer application. Policymakers are frequently interested in providing farmers with various streams of information to overcome these informational barriers to optimize farming activities. However, such informational interventions frequently fail either because generic recommendations may be ill‐suited for decision makers in highly heterogeneous agricultural environments or because farmers' beliefs may be so entrenched as to make them unresponsive to new information. We implemented a field experiment in Bihar, India to test whether plot‐specific fertilizer recommendations affect farmers' fertilizer use. We find little evidence for sizable impacts on fertilizer use in general, though impacts are more apparent for low cost or costless recommendations such as increasing the use of highly subsidized fertilizers or shifting the timing of application. Despite modest evidence of such effects, even those fall short of their potential magnitude. We show that treated farmers who are less confident in their subjective beliefs about optimal fertilizer application rates (i.e., with more disperse priors) are more responsive to the recommendations and have a higher ex ante willingness to pay for soil testing. These results suggest that heterogeneity in beliefs may constrain the overall effectiveness of information provision, even when the information is tailored to individual farms.