Commentary a Calibrated Classical Model

Robert J. Barro’s article, “Inflation and Growth,” presents the results of regressions of per capita output growth on inflation across countries and over time. He finds that there is a statistically significant negative slope coefficient estimate in these regressions; the estimate usually lies between 0.02 and 0.04. As far as I can tell, Barro wants to interpret these results as follows. Suppose a central bank can choose between two different monetary policies. The first monetary policy induces an average inflation rate of 5 percent per year, whereas the second monetary policy induces an average inflation rate of 15 percent per year. Barro argues that his regressions imply that growth will be on average 0.3 percent higher per year in the first monetary policy regime. Of course, the problem with this interpretation is obvious: Growth and inflation are both endogenous variables. Given this endogeneity, can we really view this negative coefficient as being an opportunity for the monetary authorities to influence growth rates? My discussion has three parts. In the first part I set up and calibrate a simple macroeconomic model. In the model, monetary policy has no effect on growth. Yet, because of endogeneity bias, regressions of growth on inflation produce coefficient estimates on the order of 0.024. This indicates that the quantitative magnitude of Barro’s estimates are entirely consistent with classical models of money in which monetary policy has no effect on long-run growth. In the second part of my discussion, I assess Barro’s attempts to eliminate this endogeneity bias with instrumental variables. I conclude that these attempts are not very convincing. I summarize my conclusions in the third part. A CALIBRATED CLASSICAL MODEL