Does money growth predict inflation in Sweden? Evidence from vector autoregressions using four centuries of data

Abstract In this paper, we add new evidence to a long-debated macroeconomic question, namely, whether money growth has predictive power for inflation or put differently, whether money growth Granger causes inflation. We use a historical dataset—consisting of annual Swedish data on money growth and inflation ranging from 1620 to 2021—and employ state-of-the-art Bayesian estimation methods. Specifically, we employ VAR models with drifting parameters and stochastic volatility which are used to conduct analysis both within- and out-of-sample. Our results indicate that the within-sample analysis—based on marginal likelihoods—provides strong evidence in favour of money growth Granger causing inflation. This strong evidence is, however, not reflected in our out-of-sample analysis, as it does not translate into a corresponding improvement in forecast accuracy.