Effects of Idiosyncratic Shocks on Macroeconomic Time Series

Abstract

A factor structure for VAR model error terms is adopted to examine the dynamic relationships of major macroeconomic time series. The structure, which is testable, is used to trace the consequences of a contemporaneously “ceteris paribus” (or idiosyncratic) change in each variable in the VAR model. The impulse responses to idiosyncratic shocks are shown to be a dynamic representation of the Granger causality. In the analyses of the U.S. monthly data from 1954 to 2011 for four key variables, inflation is found to respond negatively (positively) to an increase in unemployment (the federal funds rate), holding other variables contemporaneously fixed. The real variables (output and unemployment) appear unresponsive to idiosyncratic changes in the nominal variables (the federal funds rate and inflation). A common factor is observed to have a positive effect on unemployment and negative effects on output, inflation and the federal funds rate.

Keywords: vector autoregression, error factor, identification, Granger causality, impulse responses, Phillips curve, monetary neutrality

JEL Classification: C32, E30