How Precise are Our Estimates of the Current Output Gap? New Evidence from Multivariate Estimates for the Euro-Zone

This paper investigates the precision of multivariate models of the output gap and considers their implications for the formulation of macroeconomic policy. Multivariate models identify the gap by including information from structural economic relationships, such as Okun's Law, the Phillips Curve or a production function. Several recent papers using data for the Euro zone have concluded that multivariate state-space models produce relatively precise estimates of the current output and unemployment gap, suggesting that policy makers can improve economic outcomes by reacting more strongly to estimated gaps. This paper attempts to reconcile these finding with more pessimistic results produced from a variety of other data sets and methodologies. I argue that some of the recent optimistic conclusions are due to misinterpretation of the evidence. Updated results from existing models also imply that the precision is substantially less than indicated in earlier work. I also relax some of the untested and atheoretical modeling assumptions imposed in earlier work to reduce model misspecification risk and thereby produces more meaningful measures of uncertainty. Overall, the results are broadly consistent with the "low precision"view, but confirm the apparent benefits of incorporating measures of capacity utilisation and highlight the role of model risk in production-function-based estimates.