| 讲座简介： | This paper proposes a predictive quantile regression with multiple thresholds to capture the underlying regime switching mechanism in the prediction of stock returns. The predictability of each predictor is allowed to switch from one regime to another according to the value of a threshold variable and could vary aross quantiles, and the predictors could possess different degrees of persistence. A sequential estimation procedure, joint with an adaptive group Lasso refinement, is proposed to efficiently and consistently estimate the unknown multiple thresholds. To remove the impact of model misspecification from the sequential estimation on subsequent inference for the thresholds, a partitioned estimation of the thresholds is further considered. The resulting threshold estimators do not depend on parameters from other regimes asymptotically and have symmetric limiting distributions. The adaptive Lasso is finally adopted to identify the important predictors in each regime to improve prediction accuracy at each quantile, and is shown to achieve the oracle property. Monte Carlo simulations demonstrate the nice performance of our procedure in finite samples. The empirical analysis for the U.S. stock returns shows that the return predictability changes with the economy policy uncertainty across the quantiles. |
