There is a broad consensus that bankers in LDCs engage in related (insider) lending. There is also a consensus that the practice should be discouraged through close supervision and regulation. We argue that such policy recommendations, in the context of high default risk, will not produce arm's length lending by LDC bankers. Rather, it will produce little lending at all. In fact, related lending exists precisely because bankers in LDCs face high levels of default risk. This does not mean, however, that related lending is economically efficient. Our empirical analysis indicates that related lending produces a second best allocation of capital compared to that which would be obtained in an efficient capital market. Bankers choose borrowers based on pre-existing family or business relationships, not their entrepreneurial or managerial talent. We operationalize this argument by looking at the lending strategies of Mexican banks during the period 1880-1913, and then analyze the effect of related lending on a large and finance-dependent industry—cotton textile manufacturing—by constructing a panel of mills that we follow over a 25 year period.