The Effect of Monetary Policy on Real Commodity Prices: A Re-examination

In a recent paper, Jeffrey Frankel documents a negative and significant relationship between real interest rates and real commodity prices. This paper demonstrates that Frankel's results are open to question on methodological grounds. First, Frankel did not account for possible unit roots or cointegration in the data. When we estimate vector autoregressions and error-correction models on properly transformed data, our results show few significant responses of commodity price inflation to real interest rate changes. Further, Granger causality tests reveal potential problems of endogeneity. Third, we find evidence of at least one structural break in the majority of our estimated systems. Finally, Frankel's use of annual rather than monthly data results in an artificial small-sample problem, exacerbating issues of testing and inference. When we account for these issues, we find very little evidence of a relationship between monetary policy and commodity prices.