New results in support of the fiscal policy ineffectiveness proposition.

New Results in Support of the Fiscal Policy Ineffectiveness Proposition

1. INTRODUCTION

THE "POLICY INEFFECTIVENESS" PROPOSITION of the new classical macroeconomics theory is well known. Under the joint hypotheses of rational expectations and flexible prices, anticipated money and fiscal policy should not influence real output in the short run; only surprises matter. (1)

Recent empirical work has not provided much support for this proposition. Early work by Barro (1977, 1978) appeared to support the proposition that only money surprises affect U.S. real output. However, later research by Mishkin (1982a, b), Gordon (1982), and Makin (1982), among others has cast substantial doubt on these early findings by suggesting that both anticipated and unanticipated changes in money influence output. Several recent studies (McElhattan 1982 and Laumas and McMillin 1984) that have conducted analogous tests of the effects of fiscal policy have found that anticipated and unanticipated fiscal policy changes affect U.S. real output in the short run as well. These independent results concerning the short-term nonneutrality of anticipated monetary policy, on one hand, and of anticipated fiscal policy on the other hand, appear to concur in rejecting the hypothesis of policy ineffectiveness.

There is reason to view this rejection of the policy ineffectiveness proposition with some skepticism, however, because none of these studies investigate the effects of monetary and fiscal policy simultaneously. Real output is affected by a large number of factors, but excluding either monetary or fiscal policy would seem to create the greatest potential shortcoming. Typically, real output is regressed either on anticipated and unanticipated money growth, or on anticipated and unanticipated fiscal stimulus, where the anticipated policy variables are the predicted values from policy equations. To the extent that either anticipated or unanticipated monetary and fiscal policy actions are correlated with each other, previous tests of money neutrality that exclude fiscal variables and tests of fiscal neutrality that exclude monetary variables suffer from an omitted variables problem leading to biased coefficient estimates.

Interactive monetary and fiscal actions create the potential for an omitted variable problem in previous tests of policy neutrality. Sargent and Wallace (1975, 1981) argue on theoretical grounds that government deficits lead to more rapid money growth either contemporaneously or in the future under a fairly general set of circumstances. Empirical support for this relationship using U.S. time series data is reported by Hamburger and Zwick (1981), Levy (1981), Allen and Smith (1983), and Grier and Neiman (1987). Regardless of direct causal linkages, interaction effects may also be induced by monetary and fiscal responses to common shocks. For example, systematic monetary and fiscal responses to unemployment changes through feedback rules may cause covariation in policies. Section 2 briefly discusses the extent of the bias that results from omitting variables when testing policy neutrality in the presence of such interactions.