Non-Linear Predictability in Stock and Bond Returns:When and Where Is It Exploitable?

We systematically examine the comparative predictive performance of a number of alternativelinear and non-linear models for stock and bond returns in the G7 countries. Besides Markovswitching, threshold autoregressive (TAR), and smooth transition autoregressive (STAR) regimeswitching (predictive) regression models, we also estimate univariate models in which conditionalheteroskedasticity is captured through GARCH, TARCH and EGARCH models and ARCH-in meaneffects appear in the conditional mean. Although we fail to find a consistent winner/out-performeracross all countries and asset markets, it turns out that capturing non-linear effects is of extremeimportance to improve forecasting performance. U.S. and U.K. asset return data are “special” in thesense that good predictive performance seems to loudly ask for models that capture non-lineardynamics, especially of the Markov switching type....