In this study, we use an extension of the heterogeneous autoregressive model to investigate the influence of time-varying risk aversion and macroeconomic, financial, and economic policy uncertainty measures on stock market volatility and correlation. Based on the findings, there is a stronger predictive ability of these variables at the monthly frequency than at the daily frequency. We also highlight the importance of risk aversion, which, alongside fundamental factors, reflects investor sentiment in predicting stock market volatility. Meanwhile, although uncertainty variables, such as economic uncertainty and financial uncertainty, are important, the widely used variable, economic policy uncertainty, is not helpful for predicting stock market volatility. Moreover, there is evidence of higher economic value and reduced portfolio risk when including risk aversion and economic uncertainty in international portfolio analysis.
Understanding how uncertainty and risk aversion affect the volatility and correlation of financial markets is important for investors when planning their risk management and portfolio selection strategies as well as for policymakers when devising their economic policies.1 Thus, we conduct an empirical study on the relative importance of time-varying risk aversion and different sources of uncertainty in order to predict stock market volatility and correlation as well as determine whether this information is useful for international portfolio analysis.
As emphasized in previous research, risk aversion and economic uncertainty are key determinants of financial returns and risk premiums. Many studies have assumed constant risk aversion and focused on the time variation in economic uncertainty (e.g., Kandel and Stambaugh, 1990; Bansal et al., 2005, 2014). The importance of time variation in risk aversion has been highlighted in Campbell and Cochrane's (1999) consumption-based asset pricing model. Recently, Bekaert et al. (2022) suggested an asset pricing model in which conditional volatility is driven by variation in both risk aversion and economic uncertainty.2 Economic uncertainty is primarily related to shocks from fundamental factors (e.g., consumption shocks in Bansal and Yaron, 2004), while risk aversion is driven by shocks from both fundamental and non-fundamental factors such as investor sentiment (e.g., Baker and Wurgler, 2006).3
This study contributes to the existing literature by comparing the predictive ability of risk aversion and various sources of economic uncertainty (macroeconomic, financial, and economic policy measures) for stock market volatility and correlation. Our results provide new insights into the relative importance of these predictor variables in stock market volatility and correlation.
In order to examine the influence of risk aversion and various uncertainty predictor variables on U.S. stock market volatility, we use the HAR model extended with exogenous predictor variables of Corsi (2009). As a benchmark model, we use the HAR model extended with the risk aversion and economic uncertainty measures of Bekaert et al. (2022). Our analysis shows the importance of the risk-aversion measure proposed by Bekaert et al. (2022). The results may indicate that non-fundamental factors, such as sentiment, are important for predicting variations in realized volatility. In addition, we find that the financial market uncertainty index from Ludvigson et al. (2021) is beneficial for predicting monthly realized volatility when added to the benchmark model.
In this study, the financial market uncertainty index is constructed from a large number of financial market variables, enabling it to capture the overall uncertainty of financial markets. In contrast to previous research, we find that EPU does not provide useful information for predicting stock market volatility once we account for RA and EU. Similar results are obtained for IP and CLI. Our analyses also show that exogenous predictor variables are more useful for modeling RV at the monthly frequency than at the daily frequency. This is partly because fluctuations in uncertainty follow a much smoother path than changes in stock market volatility, and partly because some of the predictor variables are not available at the daily frequency.