Stock market volatility and jumps in times of uncertainty

Highlights

• Latent uncertainty is a significant predictor of stock market volatility and jumps.

• Higher uncertainty predicts higher return volatility of the S&P500 constituents.

• The impact of uncertainty shocks is higher in the post-2007 Great recession period.

• The predictive power of uncertainty is higher for financial stocks.

Abstract

In this paper we examine the predictive power of latent macroeconomic uncertainty on US stock market volatility and jump tail risk. We find that increasing macroeconomic uncertainty predicts a subsequent rise in volatility and price jumps in the US equity market. Our analysis shows that the latent macroeconomic uncertainty measure of Jurado et al. (2015) has the most significant and long-lasting impact on US stock market volatility and jumps in the equity market when compared to the respective impact of the VIX and other popular observable uncertainty proxies. Our study is the first to show that the latent macroeconomic uncertainty factor outperforms the VIX when forecasting volatility and jumps after the 2007 US Great Recession. We additionally find that latent macroeconomic uncertainty is a common forecasting factor of volatility and jumps of the intraday returns of S&P 500 constituents and has higher predictive power on the volatility and jumps of the equities which belong to the financial sector. Overall, our empirical analysis shows that stock market volatility is significantly affected by the rising degree of unpredictability in the macroeconomy, while it is relatively immune to shocks in observable uncertainty proxies.

Introduction

What are the key drivers of volatility and jumps in stock market prices? Historically, stock prices exhibit large swings during periods of heightened uncertainty in the economy. For example, the S&P500 index lost approximately 20% of its market value during the first quarter of 2020, while the VIX index jumped from 12.5% on 2nd January 2020 to 82.7% on 16th March 2020, in response to the COVID-19 pandemic uncertainty episode. The recent history contains many examples of rising stock price volatility and jumps in times of significant macro-oriented uncertainty shocks like the COVID-19 pandemic, the Great Recession and the Euro Area debt crisis. Despite the wealth of descriptive evidence, there is only limited empirical evidence in the literature showing the impact of economic uncertainty shocks on stock market volatility and jumps.¹ Moreover, while some recent empirical studies show that stock price volatility is positively correlated with several different measures of financial and macroeconomic uncertainty (Baker et al., 2020, Bloom, 2009, Bekaert and Hoerova, 2014; among others), little attention has been given to the dynamic impact and the predictive power of macroeconomic uncertainty shocks on stock market volatility and price jumps.² In this study, we fill this gap in the literature by empirically examining the impact and the predictive power of macroeconomic uncertainty on stock market volatility and jumps.

The extant empirical literature suggests that short-term volatility and jumps in the equity market are predictable to a degree using variables such as lagged realized volatility and implied volatility (Andersen et al., 2007, Bekaert and Hoerova, 2014, Corsi, 2009, Canina and Figlewski, 1993, Christensen and Prabhala, 1998, Fleming et al., 1995). Moreover, another strand of the literature shows that a large part of the time variation of equity market volatility can be explained by a single common factor. For example, Engle and Susmel (1993) demonstrate that the international stock markets have the same time varying volatility, while Anderson and Vahid (2007) show that a common factor which is constructed using the lagged volatility series of equity prices explains a large part of the aggregate time varying stock market volatility. A third strand of the literature shows that equity market volatility is related to business cycle fluctuations (Engle et al., 2013, Hamilton and Lin, 1996, Paye, 2012, Schwert, 1989; among others). For instance, Schwert (1989) finds that the yearly volatility of industrial production and interest rates forecasts aggregate stock market volatility, while Wachter (2013) shows that the time-varying probability of rare-disaster risk in the macroeconomy is an important early warning signal of rising volatility in the equity market. Other studies concentrate on equity price jumps instead of volatility and examine their relationship to macroeconomic news (Evans, 2011, Faust and Wright, 2018, Lahaye et al., 2011, Miao et al., 2014).

Motivated by the empirical findings that identify the significant impact of macroeconomic news releases and economic policy uncertainty on stock market volatility (Amengual and Xiu, 2018, Brenner et al., 2009, Engle et al., 2013, Kaminska and Roberts-Sklar, 2018, Liu and Zhang, 2015; among others), we investigate the stock market effect of unobservable (latent) macroeconomic uncertainty which captures the unforecastable (by economic agents) variations in key macroeconomic indicators. We base our analysis on a discounted cash-flow model in which we attribute the unexplained part of stock price volatility (the non-fundamental driven volatility) to macroeconomic uncertainty. As a proxy for macroeconomic uncertainty, we use the unobservable Macroeconomic Uncertainty measure of Jurado et al. (2015) (MU henceforth), which captures the time variation in the degree of unpredictability of US macroeconomic fluctuations. MU is defined as the squared forecast error of a multivariate factor model used for forecasting US business cycles.³ The results presented in the paper clearly show that latent macroeconomic uncertainty has significant predictive power on US stock market volatility and contains information which is different to the predictive information content of the VIX and other uncertainty proxies based on observable macroeconomic news. The fact that the MU factor has incremental predictive power when included into a multivariate forecasting regression model which includes the VIX, US Industrial Production and the Baa corporate default spread, shows that the MU factor indeed explains the part of stock market volatility which cannot be attributed to changes in fundamentals. Moreover, our VAR analysis reveals that a positive latent macroeconomic uncertainty shock has larger and more long-lasting positive effect on stock market volatility compared with the respective impact of VIX shocks and shocks to other popular observable economic uncertainty proxies. For example, the response of stock market volatility to MU shocks is more than 3 times larger in magnitude and persistence when compared with the respective response of stock market volatility to VIX or Economic Policy Uncertainty (EPU) shocks. Hence, our second and more significant contribution in the literature is that we show for the first time that the latent macroeconomic uncertainty outperforms the VIX and EPU when forecasting volatility in the US equity market.

When we decompose the realized variance of equity returns into its continuous and discontinuous part, we find that the latent MU factor does not perform well in forecasting equity price discontinuities (jumps). This result is puzzling, as previous literature (see Akhtar et al., 2017) has successfully linked unanticipated macroeconomic news and stock market jumps. Motivated by a strand in the literature that identifies tighter linkages between the macroeconomy and financial markets during the post-2007 crisis era (Abbate et al., 2016, Caldara et al., 2016), we split our sample to before and after the 2007 US recession period and re-estimate our models. Our econometric analysis identifies a spectacular rise in the forecasting performance of MU on both stock market volatility and jumps in the post-crisis period. Moreover, when estimating our VAR model for the post-2007 period, we find that the dynamic effect of MU shocks on stock market volatility and price jumps increases tremendously in magnitude. Importantly, our post-crisis VAR analysis identifies the MU shock as the most significant (in terms of magnitude and persistence) type of uncertainty shock affecting the time varying volatility and jump tail risk in the US equity market. Our findings provide further empirical insights to the findings of Abbate et al., 2016, Caldara et al., 2016, Ellington et al., 2017 who investigate the time variation in macro-financial linkages and find that the impact of financial shocks to US real business cycles has exponentially increased after the Great Recession. Our results are in line with this strand of literature since we also show that the impact of macroeconomic uncertainty shocks on US stock market volatility has exponentially increased during the post-2007 crisis period. Our analysis identifies an increasing effect of all macroeconomic uncertainty shocks (e.g. macro-uncertainty and monetary policy uncertainty) on stock market volatility and jumps after the 2007 US recession. Nevertheless, it is the latent MU factor that has the highest predictive power in the post-crisis period, when compared to that of observable economic uncertainty proxies like Economic Policy Uncertainty (EPU) and Monetary Policy Uncertainty (MPU).

Our findings are also broadly in line with those of Akhtar et al., 2017, Bernanke and Kuttner, 2005, Rangel, 2011 who find that the unanticipated component of Fed fund’s rate and of macroeconomic announcements has the most significant effect on stock market price jumps and jump intensities. While the relevant literature so far shows that jumps and co-jumps in stock market prices are attributed to scheduled releases of macroeconomic news (Bollerslev et al., 2008, Evans, 2011, Lahaye et al., 2011, Miao et al., 2014), our contribution in this strand of macro-finance literature is that we show that the key driver of stock market price volatility and jumps is the rising uncertainty about the future state of the economy, and not the uncertainty about economic policy which is based on macroeconomic news.⁴ Hence, the economic interpretation of our findings, is that, what matters most for equity price stability, is not the numerous large fluctuations in macroeconomic indicators which are relatively more predictable by financial market participants, but the relatively fewer unanticipated (difficult to be predicted ex ante) changes in macroeconomic outcomes.

In order to gain further insights on our results at the aggregate market level, we also examine the predictive power of the MU factor on the volatility and price jumps of individual US equities (S&P500 constituents), so as to identify the market sectors that have the highest sensitivity to macroeconomic uncertainty. Our forecasting regressions show that the latent macroeconomic uncertainty factor constitutes a common volatility and jump tail risk forecasting factor in the equity market (it enters significantly in predictive regressions on volatility and jumps of the S&P 500 constituents). Moreover, we empirically show for the first time in the literature that the MU factor outperforms the VIX when used as predictor of volatility and price jumps of individual stocks. Interestingly, we find that, although the MU factor performs well as a predictor of the volatility and price jumps of stocks belonging to many sectors of US stock market, it performs the best when predicting the volatility and price jumps of financial firms (with the weakest performance exhibited on the Technology and Healthcare sectors). It appears that the instability and turbulence in the US financial sector is, to a significant extent, driven by the rising uncertainty about the future state of the US economy.

The rest of the paper is organized as follows: Section 2 discusses the theoretical stock price volatility model and the channels linking macroeconomic uncertainty with stock market volatility. Section 3 describes the data and outlines the empirical methodology. Section 4 presents the empirical results and Section 5 reports the various robustness checks. Finally, Section 6 concludes.