What if investors aren’t processing information as efficiently as we’ve assumed?
In a new paper, “Correlation Neglect in Asset Prices,” Wharton Professor Jessica Wachter and her co-author Hongye Guo reveal a striking pattern in U.S. stock market returns tied to the quarterly earnings cycle. Their research shows that investors systematically overreact to repetitive earnings news — and then correct that mistake in the months that follow. The cause is a behavioral phenomenon known as correlation neglect.
Below, Professor Wachter unpacks her findings, explores what they reveal about investor behavior, and considers the implications for market design, regulation, and the future of financial research.
Your paper challenges the long-held view that monthly stock returns aren’t predictable based on past returns. Can you walk us through the return pattern you discovered — and why it contradicts standard theories of market efficiency?
Jessica Wachter: For decades, one of the most robust facts in finance has been that monthly stock returns are essentially unpredictable from past returns — a cornerstone of the efficient market hypothesis.
When you align returns with the quarterly earnings cycle, a very clear and systematic pattern emerges. Returns in the first month of a quarter — what we call a newsy month, when investors first learn about aggregate earnings for the prior quarter — positively predict returns in the second month. Then, returns in that second, repetitive month negatively predict returns in the next quarter’s first month.
So monthly returns aren’t unpredictable at all; they alternate between continuation and reversal in a highly structured way. Not only is the pattern large and statistically significant, but it can also be turned into a profitable trading strategy. This directly contradicts the claim that markets incorporate all available information.
You attribute this pattern to “correlation neglect.” What is correlation neglect, and why is it so powerful that even seasoned investors seem to be susceptible to it?
Jessica Wachter: Correlation neglect is the tendency to treat information as independent when it is actually correlated. In our setting, investors correctly recognize that earnings announcements convey information about the economy — but they fail to fully appreciate that earnings announced later in the same quarter are partly repeating what they’ve already learned.
As a result, investors overreact in the second month of the quarter, treating familiar information as if it were new. That overreaction creates return continuation. When genuinely new information arrives in the following quarter, prices correct, generating reversal.
What makes correlation neglect so powerful is that it doesn’t rely on naïveté or lack of incentives. The signals are complex, diffuse, and spread across many firms. Even highly sophisticated investors are vulnerable because the mistake arises from basic cognitive processes — how humans aggregate and recall information.
What motivated you to investigate correlation neglect in asset prices, and what made this behavioral factor stand out to you as especially important?
Jessica Wachter: The starting point was really the structure of the earnings cycle itself. Earnings arrive in a very regular, staggered way within each quarter, and that immediately raises the question of how investors process information that is new versus information that is predictably related to what they have already seen. Once you look at returns through that lens, it becomes natural to ask whether investors fully undo those correlations when forming beliefs.
Correlation neglect was already in the back of our minds because it reflects something very basic about human cognition. Seeing the same underlying fact multiple times, even in slightly different contexts, tends to embed it more strongly in memory. Precisely discounting for the fact that two signals are correlated — rather than independent — is cognitively demanding, and in many realistic settings it may be difficult, if not impossible, to do perfectly.
What made correlation neglect especially compelling in this setting is that, when combined with the earnings cycle, it delivers a very sharp and falsifiable hypothesis. Investors should initially underreact when genuinely new information arrives, then overreact when similar information is repeated, and finally correct once truly new information appears again. That sequence — underreaction followed by overreaction — is exactly what we see in the data, both in returns and in expectations.
What are the broader economic implications of correlation neglect in asset prices for financial institutions, policymakers, and regulators?
Jessica Wachter: The first implication is a kind of epistemological humility. If correlation neglect is strong enough to show up in aggregate market predictability — which is remarkable given the size, liquidity, and sophistication of U.S. financial markets — then it is unlikely to be confined to investors alone. The same cognitive forces are likely present among institutions, policymakers, and regulators as well.
Second, while our results highlight systematic under- and over-reaction, they also show that markets do eventually correct. The inefficiency we document operates at a monthly frequency. On the longer horizons that matter most for aggregate investment and capital allocation, prices appear to incorporate information more accurately. While prices may not be efficient in the efficient market hypothesis sense, allocations may be efficient more broadly.
Finally, the results underscore just how important quarterly earnings are to investors. Quarterly reporting predates the SEC and originated as a New York Stock Exchange rule. Yet it is now so deeply embedded in market practice that it shapes return dynamics in a measurable way.
What excites you most about the future of this line of research? Are there related behavioral phenomena that financial scholars, practitioners, or regulators should be paying closer attention to? How could you see future research expanding on your work?
Jessica Wachter: What excites me most is that this work fits into a broader resurgence of what you might call cognitive economics. In some ways, it revives an idea from Keynes’s General Theory: that markets are not abstract machines, but social systems shaped by how people perceive, remember, and interpret information.
Correlation neglect is one example of this perspective. It highlights the central role of memory in financial markets — how repeated exposure to similar facts, when they appear in different contexts, can disproportionately influence beliefs and prices. This is not about irrationality in a narrow sense, but about the limits of real cognitive processes in complex environments.
Looking ahead, there is enormous scope for research that takes these mechanisms seriously. Memory, attention, and belief formation interact with institutional features like disclosure rules, information technology, and especially AI-generated summaries. Understanding those interactions — and when markets amplify or dampen cognitive biases — strikes me as one of the most promising directions for financial economics going forward.
Learn more about Professor Jessica Wachter.
View “Correlation neglect in asset prices” by Hongye Guo and Jessica Wachter on SSRN.
Read the Discoveries blog.