Title: Uncertainty, parameters' heterogeneity and cross-sectional dependence in fiscal reaction functions estimation
Authors: Roberto Golinelli - University of Bologna (Italy) [presenting]
Irene Mammi - University of Bologna (Italy)
Abstract: The estimation of fiscal reaction functions parameters, in order to assess the cyclicality of discretionary fiscal policies, has to tackle uncertainty over several dimensions. Uncertainty may stem from different sources including the timing and composition of fiscal measures implemented by policymakers; the difficulties in predicting variables such as potential GDP and output gap in real time, and data revisions; fiscal/volatility shocks and the increase in volatility of macroeconomic indicators. Additional empirical challenges are represented by cross-sectional dependence in fiscal shocks due to common latent factors or spatial spillovers and by the non-constancy of parameters across countries. An empirical estimation of fiscal reaction functions for the Euro area countries between 1996 and 2016 is provided. A dynamic fiscal rule is estimated for the cyclically-adjusted primary balance using dynamic-panel-data estimators, allowing for both parameters heterogeneity and common correlated random effects in the errors. The sensitivity of the estimates to the use of ex-post, real-time or forecast data is investigated and the impact of macroeconomic indicators revisions/forecast errors is evaluated. The role of interest-rate spreads is also analysed as determinants of both discretionary actions by policymakers and of fiscal policy volatility, measured from the residuals of estimated fiscal rules on real-time data.