Money Growth and Inflation—How to Account for the Differences in Empirical Results

Empirical analyses have presented different results on the long-run relationship between money growth and inflation with some pointing to a stable relationship with a slope coefficient of close to one and others suggesting instability or a weakening of the relationship over time. Using the example case of the United States and nearly 150 years of data, we provide a systematic investigation into and comparison of the results from time series-based empirical evidence on the relationship between money growth and inflation. We use the results from a wavelet analysis as a benchmark as it offers a flexible framework that provides information on the relationship both across different frequencies and different points in time. We relate these results to those in the literature obtained from other empirical approaches and investigate the underlying causes of differences in the results. We argue that it is possible to arrive at a consistent conclusion of a stable correlation between money growth and inflation in the United States at cycles of 30 to 60 years with a declining trend in the slope relationship even though the empirical results in the literature appear to be at odds. We show that in some analyses, the evidence on the “long-run” relationship is distorted by unintentionally including higher frequencies or that results are dominated by outliers at very low frequencies for which the data do not contain much information. Furthermore, the way in which different analyses account for time variation also can influence the results.