BANKING L ICENCES, B AILOUTS A N D REG U LATOR A B I L I T Y

I analyse a model in which it is socially optimal for banks to manage depositor funds but in which concerns about fraud discourage depositing and justify regulation. The regulator screens bankers and decides the level of charter value which they will receive as incentive to prevent fraud. She can also encourage deposits by insuring them. The optimal policy depends upon the regulator’s screening ability: high ability regulators rely upon charter value and low ability regulators rely upon deposit insurance. I relate these findings to the regulation of transition economy banks, to operational risk management, and to banking competition policy. I Introduction Charter value is the rent which accrues to the possessor of a banking licence. When charter value is high, bankers are more inclined to manage their business with sufficient care to retain their licences. Charter value is usually a consequence of deposit rate ceilings or anti-competitive behaviour and therefore comes at a social cost. Charter value can be affected by policy and the trade-off between the positive incentive effects of charter value and its negative efficiencyreducing effects is thus of obvious importance. In this paper I construct a partial equilibrium model which allows me to examine the way that this trade-off is affected by the regulator’s ability to distinguish good from bad bankers and by the level of deposit insurance provision. I also examine the optimal trade-off between charter value and deposit insurance. This work provides a conceptual framework which sheds some light upon competition policy discussions in the developed world, and which indicates how policy in developing countries should evolve as regulator ability increases. The model which I develop incorporates in reduced form insights from the pre-existing banking literature. I follow previous authors by assigning an economic role to banks in resolving informational asymmetries and enforcing first-best behaviour. 1 In contrast to other authors, I assume that monitoring is