Are Empirical Measures of Macroeconomic Uncertainty Alike?

This paper considers a plethora of time-series measures of uncertainty for inflation and real output growth, which are widely used in empirical studies. Their relative performances are compared to a benchmark measure using the uncertainty measure reported by individual forecasters in the Survey of Professional Forecasters (SPF) for the period 1982-2008. The results show that the use of real-time data with fixedsample recursive estimation of an asymmetric bivariate GARCH model produces inflation uncertainty estimates which replicate the survey measure more closely than any other time-series models. There is, however, overwhelming evidence that many of the time series measures of growth uncertainty overestimate the level of benchmark survey measure. The implications of our results are discussed in the context of the extensive empirical studies on macroeconomic uncertainty. J.E.L Codes: E17; C32; C51