Skip to navigation
Skip to main content
Abstract for "Let's Do It Again: Bagging Equity Premium Predictors" by Tae-Hwy Lee
The literature on excess return prediction has considered a wide array of models. We consider imposing restrictions in a regression forecast model. We use bootstrap aggregation (bagging) as a means of imposing restrictions on parameters and/or forecasts itself, following Ian Gordon and Peter Hall (2008). We show (i) that bagging yields a combined forecast of the two competing models with and without restriction and thereby it introduces shrinkage, and (ii) that the resulting bagging forecast has smaller aymptotic MSE (despite a larger asymptotic bias) than the forecast that results from a simple restricted estimator. Monte Carlo simulation is conducted to show its superior out-of-sample forecasting performance under various situations. In an empirical application using the same data set as used in John Campbell and Samuel Thompson (2008, Review of Financial Studies), "Predicting the Equity Premium Out of Sample: Can Anything Beat the Historical Average?", we find a substantial forecast improvement from the bagging restricted forecasts when it is compared with the simple restricted model of Campbell and Thompson.