This paper investigates the effect of oil shocks on U.S. stock market volatility based on a new hybrid model that combines the least absolute shrinkage and selection operator (LASSO) with the Markov regime-switching model (MS-LASSO). Considering five oil shocks, the results show that the LASSO method containing Markov regime-switching improves forecasting accuracy from the statistical and economic perspectives. These results are confirmed in robustness checks of alternative evaluation method, alternative forecasting horizons, and alternative historical years. Moreover, we find that the net price increase indicator (NPI2) is an effective oil shock, while the large price increase (LPI) has nearly no influence during the sample period. Furthermore, we find that oil shocks have time-varying performance, which highlights the importance of considering regime switching.

本文基于将最小绝对收缩和选择算子 (LASSO) 与马尔可夫政权转换模型 (MS-LASSO) 相结合的新混合模型，研究石油冲击对美国股市波动的影响。考虑到五次石油冲击，结果表明，包含马尔可夫状态转换的 LASSO 方法从统计和经济角度提高了预测精度。这些结果在替代评估方法、替代预测范围和替代历史年份的稳健性检查中得到证实。此外，我们发现净价格上涨指标（NPI2）是一个有效的石油冲击，而大幅价格上涨（LPI）在样本期间几乎没有影响。此外，我们发现石油冲击具有随时间变化的性能，