Earnings-Day Volatility Is a Feature, Not a Bug

On October 28th, Alphabet announced earnings.[1] In response, its price fell nearly 13%, wiping approximately half a trillion dollars off its market capitalization. We saw similar price reactions to earnings announcements at the end of July. Should we be alarmed by these colossal moves in market prices? Does it mean we are in an AI frenzy?

No, huge price moves on earnings announcements are not evidence of market irrationality. Perhaps we’re in an AI frenzy, but if so, earnings-related volatility is not a reliable symptom. Stock prices reacting to earnings news is a desirable and rational phenomenon. Alphabet’s price should move when fundamental news is released. You should worry if Alphabet’s price fails to move in response to fundamental news, or if it moves in the absence of fundamental news. I’m not claiming that Alphabet’s giant price move was necessarily rational; I’m claiming that prices moving in response to information is a necessary but not sufficient condition for rationality.

You can’t have an efficient market unless prices move when new information arrives. If you don’t agree, then I must appeal to the highest authority in the land: the U.S. Supreme Court. You might be surprised to learn that the Court has weighed in on the Efficient Market Hypothesis (EMH), declaring that stock prices respond to public information and making the EMH central to securities fraud litigation.

Let me give you some background. Fama, Fisher, Jensen and Roll (1969) developed the event study, a statistical method for studying the reaction of stock prices to news. The basic idea is that in an efficient market, prices should react promptly to new information, a prediction that was largely confirmed in many academic studies since 1969. These results, among others, led Jensen (1978) to say that “I believe there is no other proposition in economics which has more solid empirical evidence supporting it than the Efficient Market Hypothesis.” The Supreme Court apparently agreed, because in 1988 they adopted (in Basic v. Levinson) the EMH as part of the “fraud-on-the-market” doctrine. After 1988, if you wanted to sue a publicly traded firm for securities fraud, you probably needed to hire an expert witness to do an event study. For example, in Enron’s class action securities litigation in 2006, the plaintiffs performed an event study showing that Enron’s stock price reacted to information released by the company.

In 2014, the Halliburton II case presented the Court with the opportunity to reject the EMH and instead embrace behavioral finance. The Court could have decided to go full behavioral, declaring that market prices do not properly reflect fundamental value. The Court instead chose to stick with the EMH. Chief Justice Roberts approvingly described event studies as “regression analyses that seek to show that the market price of the defendant’s stock tends to respond to pertinent publicly reported events.” Justice Thomas disagreed, displaying a fondness for behavioral finance and dismissing the EMH as outdated: “This view of market efficiency has since lost its luster.”

Is Alphabet’s decline of almost 13% obviously too big? No, not in the context of today’s levels of idiosyncratic volatility and dispersion, which remain high in the U.S. stock market, although somewhat lower than the extreme levels earlier this year.

Information arrives every day, not just on earnings announcement days. It’s just that more information arrives on earnings days, so that prices are more volatile on earnings days.[2] How much more volatile? About twice as much, judging by the standard deviation of abnormal returns. Figure 1 shows, for the largest and most liquid U.S. stocks, the cap-weighted ratio of earnings-day volatility to non-earnings-day volatility. Beaver, McNichols, and Wang (2018) show similar results.

The impact of earnings announcements on volatility has risen over time. In December 1998, volatility around earnings was 1.4x higher than normal, while in October 2026 it was 2.3x higher (Alphabet’s recent announcement is reflected in the October 2026 ratio). Beaver, McNichols, and Wang (2018) similarly find “a dramatic increase in information content at earnings dates from 2001 onward.” This increase perhaps reflects Regulation FD (2000) and Sarbanes-Oxley (2002).

Figure 1: Ratio of earnings-day volatility to non-earnings-day volatility—large liquid U.S. stocks

Let me comment on a few episodes. First, the ratio was not especially high during the tech stock bubble, a fact confirmed by Beaver, McNichols, and Wang (2018). Second, during times of great market stress, such as the GFC in 2008-2009 and COVID in 2020, the ratio falls. This pattern makes perfect sense: the earnings announcements immediately after March 2020 contained outdated information and much of the initial information about COVID did not arrive on earnings days. Third, you could argue that the low level of the ratio as of mid-2021 is consistent with a less rational market as the COVID bubble peaked. GameStop, for example, was a stock with very high volatility that was not concentrated around earnings dates.

Figure 2: Annualized volatility on earnings days versus other days—large liquid U.S. stocks

Figure 2 gives a different perspective by showing abnormal return volatility on earnings days and other days separately.[3] The darker line shows that idiosyncratic volatility around earnings days is currently extremely high, at 78% annualized. Only in the GFC and the aftermath of the tech stock bubble has earnings-related volatility been higher. So, yes, the volatility of Alphabet and other large tech stocks has been very high on earnings announcements.

On the other hand, the lighter line in Figure 2 shows that non-earnings volatility is also very high. Thus, what’s exceptional about today’s market is not the fact that volatility is high around earnings announcements; what’s exceptional is that volatility is high on both earnings days and non-earnings days.

So where does that leave us? The stock market is acting like it’s rationally assessing fundamental information. Of course, the market may be getting the answers wrong as it grapples with the implications of the AI economy. But at least it’s asking the right questions.

Endnotes

[1] References to this and other companies should not be interpreted as recommendations to buy or sell specific securities. Acadian and/or the author of this post may hold positions in one or more securities associated with these companies.

[2] Zdorovtsov (2009) provides related evidence that both public news arrivals and private information revealed through trading contribute to stock-return volatility.

[3] Since Figure 2 shows cap-weighted averages while Figure 1 shows cap-weighted ratios, the ratio of the lines in Figure 2 is not equal to the ratio shown in Figure 1.

References

Beaver, William H., Maureen F. McNichols, and Zach Z. Wang. “The information content of earnings announcements: new insights from intertemporal and cross-sectional behavior.” Review of Accounting Studies 23, no. 1 (2018): 95-135.

Fama, Eugene F., Lawrence Fisher, Michael C. Jensen, and Richard Roll. “The adjustment of stock prices to new information.” International Economic Review 10, no. 1 (1969): 1-21.

Jensen, Michael C. “Some anomalous evidence regarding market efficiency.” Journal of Financial Economics 6, no. 2/3 (1978): 95-101.

Zdorovtsov, Volodymyr M. News, trading, and stock return volatility. SSRN, 2009.