Money/asset ratio as a predictor of inflation

Abstract

This paper modifies the quantity theory of money to forecast inflation, relating the latter to scale variables such as monetary aggregate M2 and government bonds that measure the money demand for asset transactions. The out-of-sample forecast results show that at least since the early 1990s, the money/asset model that uses the money supply/government debt ratio as a predictor has been significantly improved upon univariate and multivariate models, such as Phillips curve and term spread models, for forecasting U.S. inflation over one- to three-year horizons. In using real-time vintage data, I find that, since 2000Q1, the forecasts derived from the money/asset model have slightly improved upon those from the Greenbook in forecasting quarter-over-quarter CPI inflation at short horizons, from two- to four-quarter. These results imply that the Federal Reserve can use the money supply/government debt ratio to forecast and control the inflation rates, coordinating monetary policy with fiscal policy. Moreover, the money supply/government debt ratio can partly explain the U.S. inflation dynamics from the early 1960s until COVID-19.