Measures of macroeconomic uncertainty for the ECB’s survey of professional forecasters

This paper investigates to what extent different uncertainty measures commonly used in the SPF literature comply with a few reasonable properties. The measures published by the ECB in its quarterly report of the SPF results do not verify almost any of the properties. Unfortunately, the alternatives typically proposed in the literature do not perform much better under this metric. Instead, entropy-based measures and a new measure based on the Gini index seem more satisfactory in this regard. Independently of the measure chosen, the aggregation of the results from all the participants in each survey round may produce misleading results: they may compound true changes in uncertainty with artificial changes due to the variations in the panel of respondents to the survey. Using an aggregate measure of uncertainty from the subsets of forecasters that replied to two consecutive survey rounds, the paper finds significant increases in macroeconomic uncertainty in the euro area from 2001 to 2004, declines in uncertainty from the second half of 2004 to 2007, sharp increases from 2008 to mid-2009 and falls thereafter with the exception of the relatively more turbulent period between late 2011 and early 2012