Abstract

A Modified Federal Reserve of St. Louis Spending Equation for Canada France, Germany, Italy, the United Kingdom, and the United States

This paper replicates and modifies the Federal Reserve of St. Louis spending equation. The original estimates were based on quarterly U.S. data through 1969. The results tended to confirm the monetarist hypothesis that money (currency and demand deposits) affects GNP significantly over 5 quarters but government spending has only a transitory effect. Part 1 maintains the exact St. Louis specification but extends the sample to the mid-1970s and to data for Canada, France, Germany, Italy, and the United Kingdom. The results tend to confirm that money is a significant factor affecting GNP except in the United Kingdom. Unlike the original results government spending was also estimated to affect GNP significantly in the United States and in the other countries except France and Germany. Part 2 modifies the spending equation by not constraining ends of the lag distribution, including exports as another autonomous variable, searching for best-fit lag lengths, and transforming the variables into percent changes. Monetary effects were smaller in the revised common specification but generally significant even in the United Kingdom. Government spending and export effects were also generally significant. Results were qualitatively the same under either log or nonlog specifications with the exceptions that government spending in Germany and Italy was significant only under the log specification. Errors relative to the level of GNP were about 1 percent for every country except the United Kingdom for which the error was 2 percent. In no country was the estimated lag in the effect of money on GNP longer than the 5 quarters estimated in the original St. Louis model. Other lags were also short. For the period and the countries studied the results tend to confirm that current and lagged effects of changes in money and government spending account for major movements in GNP.