Commentary on "Discretion, Rules and Volatility

D iscussions of rules vs. discretion have occupied a central place in the analysis of macroeconomic policies in dynamic models. Costas Azariadis and Vincenzo Galasso have presented a new model that shows how a discretionary policy regime can lead to a more volatile fiscal policy than one pursued under a constitutional rule. They set up a simple fiscal policy story and examine the potential for discretionary policy to produce cyclical or chaotic dynamics that they regard as a volatile outcome. I discuss their results by reasoning with analogy to games of tacit cooperation and bargaining. In particular, I argue that some classical insights of Thomas Schelling can help us understand the reason discretionary policies in the Azariadis and Galasso model can generate multiple equilibria and the appearance of a random fiscal policy, whereas a regime of rules does not. The Azariadis and Galasso model is based on the simplest overlapping generations model. The only policy variable in their framework is the choice of a social security transfer payment from the young generation to the old generation. Their model of discretionary policy is the determination of the level of the transfer by a median voter or majoritarian voting arrangement. Because the median voter can costlessly change the transfer at each time, this is a discretionary regime. This model has multiple equilibria that reflects a fundamental indeterminacy problem. Some of these equilibria can also be dynamically inefficient, although they are all individually rational and satisfy a subgame perfection constraint. Volatility takes the form of a potentially cyclic or chaotic transfer sequence. Fiscal policy can appear to be random even though it is generated by a deterministic dynamic process. In contrast, Azariadis and Galasso show that the constitutional rule produces a different result. The indeterminacy problem disappears, and the resulting equilibrium profile is dynamically efficient.