Moral hazards in credit relations

Finding evidence for the presence of moral hazard is difficult because measures of 'unwillingness' are not readily available. For this reason, many theoretical models born out of Information Theory remain untested. In this dissertation I develop two theoretical models and derive empirical tests for the presence of a specific form of moral hazard in credit transactions: diversion of borrowed funds. In the first model, I assess a household's demographic structure as a source of this specific risk, while in the second model, I explore the possibility that lenders schedule the release of funds to farmers to avoid default risks due to capital diversion. Summaries of these models and test results are included below.;Model I. While the fungibility of funds has been recognized as a problem facing agricultural credit programs, it is not clear how or where funds are diverted by borrowers. Household composition is likely to be an important determinant of the allocation of resources between consumption and production. In particular, given equal resources, households who have to support many dependents relative to others may be more likely to divert funds toward consumption. Evidence from rural South India supports the hypothesis that household composition is one of the determinants of equilibrium input demand, and farmer output. Accordingly, moneylenders consider both wealth and household composition when deciding on the size of loans to offer households.;Model II. The existing literature on informal rural credit ignores the timing of loan disbursements as a choice variable in lenders' decisions. This model introduces delayed disbursement as an alternative mechanism lenders use to cope with the most likely type of information asymmetry faced by rural lenders: moral hazard. A model is introduced in which a lender-faced with a limited liability constraint and moral hazard--must schedule the disbursement of loans to (a) avoid willful default, and (b) obtain the necessary current production information to assess the farmer's ability to repay. Empirical findings based on data from South India confirm the model's predictions that poor farmers with a bad credit history are the most likely targets of delay.