Ever since the classic work of Arrow (1963) and Akerlof (1970), it has become well-known that when informational asymmetries exist, the market can behave in strange ways. In particular, bad products and bad clients may drive good products and good clients out of the market – a kind of market failure that has come to be known as adverse selection.1 In a couple of pioneering papers, Jaffe and Russell (1976) and Stiglitz and Weiss (1981) applied this idea to the credit market and showed that when lenders cannot distinguish between ‘good’ and ‘bad’ borrowers, the market-clearing interest rate might allow too many ‘bad’ borrowers and too few ‘good’ borrowers than is desirable from the lender’s point of view. In consequence, the lender might resort to credit rationing, leaving a part of the demand for credit unsatisfied.2 This would be socially inefficient, although it would be a rational response for the lender – a classic case of market failure.3