Is there a trade-off between procyclicality and revisions in EC trend TFP estimations?
Potential output constitutes a central measure to determine compliance of the member countries with the EU fiscal rules. The EU uses a production function approach to estimate potential output. In a Kalman filter model together with a Bayesian approach TFP is decomposed into a trend and a cycle. The aim of this paper is to examine the relationship between two widely discussed issues of the EC estimate of potential output, procyclicality and the extent of revisions. Procyclicality of the TFP trend depends on the prior assumptions for the residual variance of the TFP cycle equation. Exploiting this, simulations over increasing values of the priors of the residual variance of the TFP cycle equation are run for eight EU countries, leading to decreasing procyclicality of TFP trend estimates. Procyclicality of the estimated TFP trend reduces the standard error of revisions for half of the countries considered, while it implies an increase for the other countries or has no effect. Thus there is a trade-off between procyclicality of the TFP trend and the revision error, but it is not so clear cut. The standard errors of revisions of real-time estimates of the TFP trend as a criterion of model selection could improve forecasts additionally to the marginal likelihood value employed by the EC.