No abstract

Policy Analysis withEconometric Models RECENTLY the rational expectations school has mounted an attack on the conventional use of simultaneous equations models for policy analysis. One might go further and say that among academic macroeconomists the conventional methods have not just been attacked, they have been discredited. The practice of using econometric models to project the likely effects of different policy choices, then choosing the best from among the projected outcomes, is widely believed to be unjustifiable or even the primary source of recent problems of combined high inflation and low economic activity. Instead, it is claimed, policy analysis should be formulated as choice among rules of behavior for the policy authorities and estimates should be made of the stochastic properties of the economy under each proposed rule to choose the best. This point of view has gained such wide acceptance in part because of its association with Lucas's theoretical demonstration that a Phillips curve could emerge in an economy in which such an association between inflation and real activity was not a usable menu for policy choice. Because users of conventional simultaneous equations models sometimes presented the Phillips curve as just such a menu, and because it became apparent in the 1970s that this menu was not helpful, an analysis that provided a cogent explanation for why the menu was chimerical had great appeal.As in most revolutions, the old regime toppled by the rational expectations revolution was corrupt and in some sense deserved its fate. However, as is often the case, the revolution itself has had its excesses, destroying or discarding much that was valuable in the name of utopian 0007-230318210002-0107 $1 .00/0 ?3 BrAookings Inistitlutioni 107 108 Brookings Papers on Economic Activity, 1:1982 ideology. This paper tries to assess where the revolution itself could use revision.I In this paper I argue that it is a mistake to think that decisions about policy can only be described, or even often be described, as choice among permanent rules of behavior for the policy authorities. A policy action is better portrayed as implementation of a fixed or slowly changing rule. I also argue that explicit identification of expectation-formation mechanisms is not necessary for policy analysis, concluding that the rational expectations critique of econometric policy analysis is a cautionary footnote to such analysis rather than a deep objection to its foundations. From this perspective, the conventional use of econometric models to aid in policy choice is neither self-contradictory nor meaningless. Applying Decision Theory to Economic PolicymakingFormal quantitative analysis of policy choice must begin with a model of the effects of policy. The model must describe the "outcome, " usually in the form of a probability distribution over future events in the economy, for each possible "setting of policy." To choose policy optimally, one evaluates the outcomes according to some objective function and chooses the best. Althou...

No abstract

No abstract

TheCase of t he Missing Money THE RELATION between the demand for money balances and its determinants is a fundamental building block in most theories of macroeconomic behavior. Since it is also a critical component in the formulation of monetary policy, it is not surprising that the money-demand function has been subjected to extensive empirical scrutiny. The evidence that emerged, at least prior to 1974, suggested that only a few factors (essentially income and interest rates, with due allowance for lags) were needed to explain adequately the quarterly movements in money demand. There were episodes that, during their course, gave the impression that the moneydemand function was shifting. On the whole, however, in the time allowed for final data revisions by a "wait and see" attitude, the apparent puzzles tended to clear up.' As has been widely documented,2 the U.S. economy is once again experiencing an apparent shift in the money-demand function. In particular, when money-demand functions that have been successfully fitted to pre-1974 data are extrapolated into the post-sample period, they consistently and significantly overpredict actual money demand. Furthermore, as the economy has moved into the upturn phase of the business cycle, the forecasting errors have mushroomed. While one might hope that subsequent data revisions could "solve" the present puzzle, this sanguine attitude seems unwarranted for a variety of reasons.First, the sheer magnitudes of the forecasting errors suggest that im-1. Such econometric "benign neglect" begs the real problems facing the monetary authorities, who are striving to make reasonable policy choices during these episodes.2. See, for example, Jared Enzler, Lewis Johnson, and John Paulus, "Some Problems of Money Demand," BPEA, 1:1976, pp. 261-80. 684Brookings Papers on Economic Activity, 3:1976 plausibly large data revisions would be required to explain current developments with equations of the sort I reported earlier. Second, the large forecasting errors for 1974-76 coincide with unusual conditions. Among other things, that period saw the most severe recession of the postwar era; an extended bout of double-digit inflation; the highest interest rates in many years; and many institutional changes in the financial structure. While the failure of an empirical macro relationship under such extreme conditions is perhaps not surprising, it should at least prompt the question of whether the specification was adequate to cope with them. In short, a reassessment of the current state of knowledge on the demand for money balances seems called for. OutlineThe plan of the paper is as follows. The next section reviews the forecasting experience with "conventional" money-demand equations, documenting the source and magnitude of the recent errors. It also considers whether the deterioration in the money-demand equation observed in the current cyclical episode had any counterpart in previous periods of recession and recovery. The second section reexamines the specification of the conventional equ...