Long-Term Volatility Shapes the Stock Market’s Sensitivity to News

This article is a non-technical summary of the paper „Long-Term Volatility Shapes the Stock Market’s Sensitivity to News“, written by Christian Conrad, Julius Schölkopf, and Nikoleta Tushteva 

Macroeconomic announcements are scheduled releases of economic data and indicators by government agencies, for instance, the publication of consumer confidence or the Consumer Price Index (CPI). These pre-scheduled announcements are published at regular, e.g., at monthly frequency, and they provide information about the current or future stance of the business cycle. Research has shown that these announcements are responsible for a large share of return fluctuations in stock markets. In efficient markets, the primary driver of price changes following a macroeconomic announcement is the surprise component – the deviation of the actual announcement from the market expectation. However, stock markets react differently to the same announcement surprise at different times. Sometimes the news can have a big impact on the market, and other times it might not affect them much. The fundamental question of why the reaction of the stock market to the release of macroeconomic news varies over time has recently gained considerable attention. 

The literature suggests that the varying sensitivity of stock markets results from changes in the relative importance of cash flow versus discount rate news driven by expectations about monetary policy over time. Let us illustrate this by looking at the expected effects of a consumer confidence indicator release on returns. Suppose the economy is booming and the published data on consumer confidence is larger than expected. Good news about consumer confidence are a sign of strong consumer spending and positive news for future cash-flows. Thus, stock prices are expected to increase after the publication. However, the central bank is likely to increase interest rates when the economy is performing well and higher interest rates lead to an increase in the required returns of investors. Hence, the discount rate effect works in the opposite direction as the cash-flow effect. Overall, one would only expect a small response of stock markets to the good consumer confidence news. While in good states the central bank will respond to good news by tighter monetary policy, it will not respond to good news in bad states of the economy. Hence, the discount rate effect of good news will not weaken the positive cash flow effect in bad states of the economy. In short, expectations about future monetary policy affect the strength of how the discount rate effect drives the time-varying sensitivity of returns.   

We propose a complementary explanation for the time-varying sensitivity of the stock market that is based on the volatility feedback effect. In general, volatility measures how much returns fluctuate around its mean and, thus, volatility provides a measure of risk. When volatility increases, investors demand higher returns to justify the increased risk they are taking. Volatility feedback happens when an anticipated increase in volatility raises required returns, in turn necessitating an immediate stock-price decline. Furthermore, volatility feedback highlights the ‘no news is good news’ effect. When there is no cash flow news, expected future volatility and, hence, required returns are revised downwards and stock prices increase. Moreover, volatility feedback implies an asymmetric reaction of stock returns between good and bad news because, for large pieces of bad news, the discount rate effect will amplify the negative cash flow effect. 

Thus, our complementary explanation for time-varying sensitivity also focuses on discount rate news but is risk-based and not driven by expectations about monetary policy. To adequately measure risk and capture the volatility feedback effect, we draw on recent developments in the literature on volatility models and decompose stock market volatility into a short- and long-term component. Here we utilize the MF2-GARCH model, a generalized autoregressive conditional heteroskedasticity model for volatility recently proposed by Christian Conrad and Robert Engle.  While the short-term component captures day-to-day movements in volatility, the persistent long-term component is closely related to macroeconomic and financial conditions and behaves counter-cyclical.  In our paper, we aim to demonstrate that the level of the long-term volatility component primarily influences unexpected returns. Furthermore, we want to evaluate whether the asymmetric reaction to good and bad news is most pronounced during periods of high volatility. 

Using an event-study design, we investigate to what extent the volatility feedback effect contributes to explaining the time-varying sensitivity of stock markets to macroeconomic announcements using S&P 500 future returns and major U.S. macroeconomic announcements data over the 2001 to 2020 period. That is, we focus on explaining the return change in short windows of five minutes before and five minutes after macroeconomic announcements using the announcement surprises and the volatility component forecasts for the short- and long-term volatility on the day of the announcement. 

We show that the effect of the announcements is indeed the strongest when long-term volatility is at elevated levels. The short-term volatility component cannot explain variation in returns around macroeconomic announcements. Furthermore, we find strong evidence for an asymmetric response to good and bad news, which is again dependent on long-term volatility. When long-term volatility is high, the effect of good news is dampened by the discount rate effect while the effect of bad news is amplified by the discount rate effect. These findings are robust to controlling for other channels, such as monetary policy expectations or uncertainty. 

In conclusion, we can confirm that volatility feedback is an important complementary channel for explaining the time-varying sensitivity besides the expected monetary policy response. One of our contributions is to provide a measure of risk, i.e., the long-term volatility component, suitable for capturing the volatility feedback effect. Hence, our empirical evidence suggests that only long-term risks are priced in the risk-return relationship and not volatility in general.

References

Conrad, Schölkopf, Tushteva (2023): Long-Term Volatility Shapes the Stock Market’s Sensitivity to News. Available at SSRN.