Purpose

This paper examines the relationship between volatility, sentiment and returns in terms of levels and changes for both lower and higher data frequencies using quantile regression (QR) method.

Design/methodology/approach

In the first step, the study applies the Granger causality test to understand the causal relationship between realized volatility, returns and sentiment as levels and changes. In the second step, the study employs a QR method to investigate whether investor sentiment and returns can predict realized volatility. This regression method gives robust results irrespective of distributional assumptions and to outliers in the dependent variable.

Findings

Empirical results show that the VIX volatility index is a better fear gauge of market-wide investors' sentiments and has a predictive power for future realized volatility in terms of levels and changes for both higher and lower data frequencies. This study provides evidence that the relationship between realized volatility, investor sentiment and returns, respectively, is not symmetric for all quantiles of QR, as opposed to OLS regression. Furthermore, this work supports the behavioral theory beyond leverage hypothesis in explaining the asymmetric relation between returns and volatility at higher and lower data frequencies.

Originality/value

This paper adds to the limited understanding of investor sentiment’s impact on volatility by proposing a QR model which provides a more complete picture of the relationship at all parts of the volatility distribution for both higher and lower data frequencies and in terms of levels and changes. To the author knowledge, this is the first paper to study the volatility responses to positive and negative sentiment changes for developed market and to use both lower and higher data frequencies as well as data in terms of levels and changes.

中文翻译：

---

投资者情绪、实际波动率和股票收益

目的

本文使用分位数回归 (QR) 方法检验了波动率、情绪和回报在较低和较高数据频率的水平和变化方面的关系。

设计/方法/途径

在第一步中，该研究应用格兰杰因果检验来了解已实现波动率、回报和情绪之间的因果关系作为水平和变化。在第二步中，研究采用 QR 方法来研究投资者情绪和回报是否可以预测已实现的波动率。无论分布假设和因变量中的异常值如何，这种回归方法都能提供稳健的结果。

发现

实证结果表明，VIX 波动率指数是衡量整个市场投资者情绪的更好的恐惧指标，并且在较高和较低数据频率的水平和变化方面具有预测未来已实现波动率的能力。这项研究提供的证据表明，与 OLS 回归相反，已实现波动率、投资者情绪和回报之间的关系对于 QR 的所有分位数都不是对称的。此外，这项工作支持超越杠杆假设的行为理论来解释较高和较低数据频率下回报和波动率之间的不对称关系。

原创性/价值

本文提出了一个 QR 模型，增加了对投资者情绪对波动率影响的有限理解，该模型为较高和较低数据频率以及水平和变化方面的波动率分布的所有部分提供了更完整的关系图。据作者所知，这是第一篇研究发达市场对积极和消极情绪变化的波动性响应并使用较低和较高数据频率以及水平和变化方面的数据的论文。