Velocity and the Variability of Money Growth: Evidence from Granger-Causality Tests: Note

Abstract

Economists and policy makers were caught off guard by behavior of income velocity of money in 1982. The velocity of M1, which had grown at a 3.3 percent average annual rate from 1970-1981, feZl by 4.4 percent in 1982. In fourth quarter of 1982, decline was 12.6 percent at an annual rate. The 1982 decline in velocity significantly reduced nominal GNP growth and may have helped cause fall in real GNP. The 1982 episode has generated quite a bit of controversy. The decline in velocity is important not only for its possible relation to real GNP, but also because it has led some economists to seriously question Monetarists believe that there is a predictable link between money growth and nominal GNP growth. When velocity declined in 1982, that link appeared to have weakened, seemingly undermining monetarist beliefs. In fact, Robert Gordon (1983) has called 1982 episode the velocity recession and demise of monetarism. Milton Friedman (1984) has stated that decline in velocity does not undermine monetarism at all. He believes that velocity fell because of extreme volatility of money growth following announced change in Federal Reserve operating procedures in October, 1979. Figure 1 shows income velocity of M1 (V) and standard deviation of current and seven lags of M1 growth (S). A large increase in volatility of money growth in 1980 precedes decline in velocity in 1982. Based on this, Friedman states that 1982 episode strengthens, not weakens, case for monetarist proposition of a constant growth rule of money supply. This paper uses Granger-causality method to test Friedman's hypothesis that volatility of money growth causes velocity to change. We conclude that Friedman is correct, i.e., volatility of money, measured by standard deviation of money growth, causes velocity in Granger sense. In next section we review some of recent explanations offered by economists for 1982 decline in velocity. The concluding section presents our results of Granger tests of Friedman's hypothesis.