The empirical relation between idiosyncratic volatility (IVOL) and returns is mixed. Ang et al. (2006, 2009) report a negative relation using lagged realized IVOL, but Fu (2009) shows that this measure is a poor proxy for expectations and proposes using forecasts from EGARCH models which results in a positive relation. However, recent studies show that this positive relation disappears when using out-of-sample EGARCH models to generate the forecasts. We show that expected IVOL proxies used in the prior literature are noisy and propose using combinations of out-of-sample IVOL forecasts generated by different EGARCH models that pass basic diagnostic tests. For the sample for which these high-quality proxies can be created, we find a significant positive relation with returns. The magnitude is both statistically and economically meaningful as firms with expected idiosyncratic volatility of one standard deviation above the mean have expected returns that are approximately 3.4% to 4.1% higher per year.

特质波动率（IVOL）与收益率之间的经验关系是混合的。Ang等。（2006年，2009年）报告了使用滞后的已实现IVOL的负相关关系，但Fu（2009年）表明，该指标不能很好地代表期望，并建议使用EGARCH模型的预测来产生正相关。但是，最近的研究表明，使用样本外EGARCH模型生成预测时，这种正相关关系消失了。我们显示现有文献中使用的预期IVOL代理很嘈杂，并建议使用通过基本诊断测试的不同EGARCH模型生成的样本外IVOL预测的组合。对于可以创建这些高质量代理的样本，我们发现了显着的积极意义与收益的关系。该数量级在统计和经济上都具有重要意义，因为预期特质波动率比平均值高一个标准差的公司每年的预期收益大约高3.4％至4.1％。