The unusually high interest rates in many backward agricultural regions have been a source of some puzzlement to economists. Adding to this puzzle is the fact that these rates can take on a wide range of values, often within the same region. One important question is why arbitrage between sectors does not lead to more homogeneous and lower interest rates. The traditional answer to this is in terms of lender's risk. It is argued that moneylenders in backward regions face a great risk of default and, once this is taken into account, the effective interest rate turns out to be no higher than its counterpart in the organised sector. Consequently, there is no real room for arbitrage and the high interest rates are equilibrium prices. While this explanation may be of use in some situations, it is in general inadequate and can even be misleading. The reason (which novelists describing rural India have repeatedly written about) is that in these financial transactions the lender's risk frequently ranges from minimal to non-existent, because the personalised relation between the borrower and the lender enables the latter to extract the defaulted loan in the form of asset transfers from the borrower. In a somewhat stylised manner these deals could be thought of as follows. A peasant takes a loan from his employer, keeping some of his assets, like land, a standing crop or even the promise of labour services, as security or collateral. A collateral price is then fixed. This is the conversion price used to calculate the amount of collateral that has to be relinquished to the lender given a certain amount of default. Clearly the lower the collateral price the larger the quantity of collateral that the peasant has to forfeit. Hence the underpricing of collateral amounts to the charging of an implicit interest. A widespread feature of rural credit markets—reported in many diverse studies (Bardhan and Rudra, 1978; Myint, 1964; Raj, 1981; Roth, 1979; and Sivakumar, 1978, to mention just a few)—is the 'underpricing' of collateral. An interesting attempt to bring collateral explicitly into a model of usury was made by Bhaduri (1977; see also 1983) who argued that one of the reasons why landlords raise interest rates is, in fact, to encourage default and to claim ownership of the asset kept as collateral. A sharp contrast to the lender's risk model, this hypothesis has deservedly received much attention.1