Discretionary Actions Research Articles

On the Desirability of Discretionary Policy

W HILE pointing out the risks inherent in discretionary stabilization policies, Milton Friedman (1953a, 1953b) also noted that these risks vary from situation to situation. Specifically, he said, The effect of action is clearly likely to be in the right direction more frequently if action is taken to counteract only movements in income than if it is taken to counteract mild movements as well (1953a, p. 131). Since the risk of taking action varies from situation to situation, it is quite clear that the risk of not acting also varies, and this fact naturally leads us to the two-part policy advocated by Bach (1949), among others, in which discretionary actions would be used to counter forecast deviations of variables from their target values, while built-in flexibility would be relied on to counter mild deviations. This paper seeks to give some formal content to the meaning of substantial and mild. In particular, I ask whether poor forecasting limits the number of instances in which discretionary actions should be used; and, in general, the answer turns out to be no. Since poor forecasting adds to the risk that discretionary actions will have perverse results, my analysis shows that it is extremely important to distinguish between the effectiveness of discretionary actions over the long run and their desirability at any particular time.' To illustrate these results, it will be necessary to derive formal decision criteria