This study investigates the effect of news sentiment on stock market volatility using the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model and measures the asymmetric effect with the GJR-GARCH model. We adopt patented linguistic analysis that considers the semantic orientation process to quantify financial news that may attract investor attention. This study distinguishes between unclassified market news sentiment and macroeconomic-related news effects. The evidence suggests that both contemporaneous and lagged news are determinants of market volatility. The effect is especially strong with the market aggregate news sentiment index (ANSI) and the negative ANSI, particularly during the 2008–2009 financial crisis period. This analysis of news sentiment improves the accuracy of in-sample and out-of-sample volatility forecasting.

本研究使用广义自回归条件异方差（GARCH）模型调查新闻情绪对股市波动的影响，并使用GJR-GARCH模型衡量不对称效应。我们采用了获得专利的语言分析，该分析考虑了语义定向过程来量化可能引起投资者注意的金融新闻。这项研究区分了未分类的市场新闻情绪和与宏观经济相关的新闻影响。有证据表明，同期新闻和滞后新闻都是市场波动的决定因素。在市场总体新闻情绪指数（ANSI）和ANSI负的情况下，效果尤其明显，尤其是在2008-2009年金融危机期间。对新闻情绪的这种分析提高了样本内和样本外波动率预测的准确性。