ABSTRACT
To this point, the models considered have presumed the regulator is able to commit himself to carry out the terms of any regulatory policy that is announced. This is indeed a powerful policy instrument for the regulator and one that may not be available in practice. For example, in the multiperiod Bayesian mechanism discussed in Section 6.1, the regulator commits to policies in future periods that are inefficient, given the information about the firm’s costs that can be inferred from the firm’s performance. In the extreme case where costs are the same in every period, the optimal regulatory policy has the regulator commit to establish a price in excess of marginal cost over the entire time horizon, even though the firm effectively reveals its actual production costs in the first period. If the current regulatory commission is unable to bind the actions of future regulatory commissions, it is unlikely that such a policy could be implemented, since future commissions would have an incentive to utilize the information revealed and to effect efficient prices.
