Frequency Dependence in a Real-Time Monetary Policy Rule
We estimate a monetary policy rule for the United States allowing for possible frequency dependence—i.e., allowing the central bank to respond differently to more persistent innovations than to more transitory innovations, in both the unemployment rate and the inflation rate. Our estimation method uses real-time data in these rates—as did the FOMC—and requires no a prior assumptions on the pattern of frequency dependence or on the nature of the processes generating either the data or the natural rate of unemployment. Unlike other approaches, our estimation method allows for possible feedback in the relationship. Our results convincingly reject linearity in the monetary policy rule, in the sense that we find strong evidence for frequency dependence in the key coefficients of the central bank’s policy rule: i.e., the central bank’s federal funds rate response to a fluctuation in either the unemployment or the inflation rate depended strongly on the persistence of this fluctuation in the recently observed (real-time) data. These results also provide useful insights into how the central bank’s monetary policy rule has varied between the Martin-Burns-Miller and the Volcker-Greenspan time periods.
JEL Codes: E52, C22, C32.
Keywords: Taylor rule, frequency dependence, spectral regression, real-time data.
Suggested citation: Ashley, Richard, Kwok Ping Tsang, and Randal J. Verbrugge, 2014. “Frequency Dependence in a Real-Time Monetary Policy Rule,” Federal Reserve Bank of Cleveland, Working Paper no. 14-30.