Even the St. Louis Model Now Believes in Fiscal Policy: Note

Abstract

Does fiscal policy matter for aggregate economic activity if it is unaccompanied by changes in the stock of money? This question, which would have attracted quick yes answers some years ago, but which has recently been the object of widespread research, came to the forefront of macroeconomic interest at least partly in response to the no anster implied in the work of monetarist economists. One key empirical element in the monetarist arsenal has been the result of the reducedform equation, initially published by Andersen and Jordan [3] and subsequently used as the cornerstone of the St. Louis model by Andersen and Carlson [1] . In particular, the St. Louis 1nding that monetary policy actions have a strong and lasting impact on nominal GNP, whereas the crowding out of E1scal policy actions is so rapid as to yield zero cumulative impact within only one year, attracted substantial attention not only among academic economists, but in the business and government communities as well. Many economists have criticized the methodology underlying the St. Louis equation. In addition, it is well known-at least among economists-that the equation's ex post predictive performance beyond the 1953-69 sample period used by Andersen and Carlson [1] has not been impressive (see [6]). Indeed, Andersen and Carlson [2] published the results of reestimating the equation through 1973 and reported a standard error of $5.2 billion, nearly half again greater than the $3.9 billion standard error that they had reported earlier. Nevertheless, the evidence provided by this equation has continued to influence discussions, at many levels, of the effectiveness (or lack thereof) of fiscal policy. What is not well known is that the St. Louis equation, reestimated to incorporate the most recent available data, now indicates that fiscal policy unaccompanied by changes in the stock of money does matter. Furthermore, the indicated f1rst-year cumulative multiplier of government spending on nominal GNP is about one and one-half-a value not surprising intuitively to believers in fiscal policy, and not very dissimilar to the values reported by familiar nonmonetarist models (see, e.g., [7] ). IShe purpose of this brief note is to present the results of reestimating the St. Louis equation and, in so doing, to shift whatever limited credibility one attaches to this simple piece of evidence away from denying fiscal impacts toward, on the contrary, supporting them.