Some Monetary Facts

This 1995 Federal Reserve Bank of Minneapolis Quarterly Review paper by George T. McCandless Jr. and Warren E. Weber examines three long-run monetary facts drawn from a cross-section of 110 countries over 1960-1990 (IMF International Financial Statistics data, up to 30 years of geometric-average annual growth rates per country), using three definitions of the money supply (M0, M1 and M2) and two robustness subsamples (21 OECD countries and 14 Latin American countries). They find, first, a high, almost-unity correlation between the growth rate of the money supply and the rate of inflation that holds for all three money definitions and across the full sample and both subsamples (correlation coefficients of 0.89 or higher, closest to one for the broader M1 and M2 aggregates); second, no correlation between the growth rates of money and real output in the full sample or the Latin American subsample, but a positive and economically non-trivial correlation (0.51 to 0.71, with a regression slope of roughly 0.1) in the OECD subsample alone; and third, essentially no correlation between inflation and real-output growth in the full sample once a single outlier (Nicaragua, which combined -12% output growth with 52% inflation during its civil war) is excluded, a finding that again holds up, though not always at conventional significance, across both subsamples. The authors argue that, to the extent these long-run correlations reflect causal relationships, central banks can expect to influence long-run inflation by adjusting money growth but cannot translate that into precise short-run inflation targets, since realized inflation also depends on real output and velocity growth that are difficult to predict, and that the absence of a long-run money-output correlation suggests monetary policy’s long-run effect on real activity is essentially nil, though the data are silent on any short-run real effects.

George T. McCandless Jr. | Macro Paper Warehouse

Some Monetary Facts