Forecasting Panel Data with Structural Breaks and Regime-Specific Grouped Heterogeneity

Abstract

Generating accurate forecasts in the presence of structural breaks requires careful management of bias-variance tradeoffs. Existing methods for forecasting time series under breaks reduce parameter estimation error by pooling estimates across pre- and post-break data necessarily inducing bias. Forecasting panel data under breaks offers the possibility to reduce parameter estimation error without inducing any bias if there exists a regime-specific pattern of grouped heterogeneity. We develop a new Bayesian methodology to estimate panel regression models in the presence of breaks and regime-specific grouped heterogeneity. Parameters are pooled within, but differ across, regime-specific groups. We develop a formal test of regime-specific patterns of grouped heterogeneity that integrates over all parameters and penalises model complexity. In an empirical application to forecasting inflation rates across 20 U.S. industries we find evidence of regime-specific grouped heterogeneity. Exploiting this information produces significantly improved forecasts relative to a range of popular methods.