In this paper, a model of interaction of formal and informal credit markets has been developed where the bank official (the ultimate provider of formal credit) faces a lending constraint. The bank official takes a bribe from the borrowers to disburse formal credit and he deliberately debars some potential borrowers from getting bank credit. Inadequate supply of formal credit and exclusion of a few borrowers by the official create a market for informal credit. The bank official and the moneylender (the supplier of informal credit) play a non-cooperative game in determining the bribing rate and the informal interest rate simultaneously. The central objective of the paper is two-fold. First, it shows that an agricultural credit subsidy policy may be counterproductive even when formal and informal credits are substitutes. This is contrary to the Gupta and Chaudhuri (1997) result that a credit subsidy policy is counterproductive only when the two types of credit are complementary to each other. Secondly, the paper considers two alternative ways of formulating a credit subsidy policy: (1) through an increase in the aggregate volume of formal credit supplied to the borrowers, keeping the formal sector interest rate at a reasonable level; and, (2) through a decrease in the rate of interest charged on this type of credit. The paper shows that if a credit subsidy policy is undertaken via the first path, it is actually able to lower the informal sector interest rate and improve both the agricultural productivity and welfare of the farmers. This result is crucial because all the earlier papers in this line have analyzed the effects of a credit subsidy policy through the second route and found it to be counterproductive in the presence of corruption in the distribution of formal credit.