Article
15 September 2026

5 Questions with Jessica Wachter: Pricing Risk, Reading Markets—From Rare Events to the AI Boom

We interview Jessica Wachter of the Wharton School, University of Pennsylvania, to gain her perspective on her research in asset pricing, portfolio choice, rare events, investor behavior, and artificial intelligence (AI) investment. Professor Wachter also discusses the relevance of her work to litigation and government investigations.

1. Your research spans asset pricing, risk premia, investor behavior, and portfolio choice. How might those research areas arise in government investigations or litigation involving investment-related issues?

Asset pricing is the science of how we think about securities prices, including risk and timing adjustments. More broadly, it provides a framework to explain how different eventualities find their way into the price of a security.

With a background in asset pricing, it is possible to analyze quite exotic securities from a set of core principles, rather than being an expert in one instrument versus another. That is the power of the discipline: asset pricing expertise provides an approach that applies to an almost infinite variety of securities. For example, once you know the contractual terms of a security, you have a methodology for pricing it.

That is useful in litigation, where this complexity often arises. A lot of litigation involves securities pricing: who sold what at what price, what did they know at the time, and if they had known something else, how would the price have changed? Asset pricing offers important perspectives on those questions.

In addition, asset pricing extends beyond prices themselves. It also helps to explain incentives—who benefits from what outcomes—and the difference between ex post, what actually happened, and ex ante, what people could possibly have understood at the time. The ex ante perspective connects to the issue of materiality, which fascinated me when I was at the Securities and Exchange Commission (SEC). Materiality involves the interplay between what is required to be disclosed and what is optimal for a company to disclose on its own and then how investors interpret, or potentially overinterpret, those disclosures.

A lot of litigation involves securities pricing: who sold what at what price, what did they know at the time, and if they had known something else, how would the price have changed? Asset pricing offers important perspectives on those questions.

Everything I have touched on so far concerns the first strand of my research: how securities are priced. Portfolio choice has been the other major component of my research from the very beginning: how an investor thinks about their stock and bond portfolios. Where asset pricing considers how securities are valued, asset allocation asks how they should be held. Questions about the appropriateness of investments, and whether a portfolio fits an investor’s circumstances, also arise in investigations and litigation, and they relate directly to how an individual thinks about their investments.

2. You developed influential work on “rare events,” unlikely but severe shocks, to explain stock market swings. In plain terms, what does that research show, and how does it inform disputes over risk, valuation, or claims that a loss was unforeseeable?

People may have heard of black swans, or five-sigma occurrences—these are the concepts at the heart of my research on rare events. In a statistical sense, the financial world is not normally distributed, yet for many years it was modeled as if it were. A realistic approach requires techniques that reflect this. One lesson of rare events research is that extreme price movements are to be expected. That is a feature of our markets and very different from other uncertainties people encounter in life.

What motivated my thinking about rare events is that prices can reflect outcomes that appear infrequently, or never, in the historical data. That is particularly true of risk premia: a bad outcome can play a more influential role in pricing than its probability alone might suggest. In effect, a bad outcome carries additional weight, because people do not want to hold an asset that falls in value just as they may also become poorer for other reasons. The stock market illustrates this dynamic. A decline for one company in a portfolio represents a small share of overall wealth, though it is, of course, meaningful for the people who work at the company. By contrast, a decline in the entire stock market affects everyone at once, so the pricing implications are very different.

My research shows that many assets “price in” rare events that do not necessarily appear in the time series. If you are evaluating asset prices, you must consider scenarios—I would not say worst-case, because often these are simply bad scenarios—that are not part of the observed data. That helps explain some of the pricing we see in markets. Put options are a good example: they build in a premium for a stock market crash. But even the equity market, I believe, is priced with a premium reflecting the possibility of a crash.

The central point here is that whenever you evaluate an outcome, you need to approach it with the perspective that very large price swings will sometimes happen in stocks in a way they do not happen in, say, the price of eggs.

3. Part of your work connects finance and cognitive science, examining how memory and past experience shape investor decisions. What does this tell us about how investors actually behave?

This is a relatively new line of research, though it preceded my time at the SEC. I came to this work with the view that every asset pricing model is actually a cognitive one, because investor expectations are involved. At its core, a model of asset prices represents how people think about the future, and that thinking affects the prices at which they buy and sell today. It is important to be realistic about how people form their expectations, rather than imagining something that does not exist.

I am not alone in this view. The efficient markets hypothesis is a powerful framework for asset pricing in general. People want to act in their self-interest, and even when knowledge and sophistication vary widely across investors, good information will often find its way into asset prices through the simple ability to buy and sell. Having said that, certain features of asset pricing are inconsistent with the efficient markets framework; those inconsistencies are what led me to model how investors form expectations.

At its core, a model of asset prices represents how people think about the future, and that thinking affects the prices at which they buy and sell today. It is important to be realistic about how people form their expectations, rather than imagining something that does not exist.

The efficient markets hypothesis was so influential that it produced its own counterpoint: behavioral finance, which documents where the rational model falls short. But behavioral finance largely catalogs those failures; it has not built an alternative of its own. Cognitive economics aims to fill that gap by developing a realistic account, one that researchers and practitioners can use, of how investors develop their expectations. The approach is still in its infancy, but that is its promise.

One thing this subject teaches you is the enormous heterogeneity among individuals; the work has deepened my appreciation of that fact. How individual behavior aggregates up to market pricing is a first-order question. My starting point is the perspective of the individual investor, as opposed to the market as a whole, and connecting the two remains an open challenge.

4. Your recent paper with Jonathan Wachter, “What Investment Data Implies about the AI Transition,” examines the current level of artificial intelligence (AI) investment under significant uncertainty. Setting predictions aside, when markets price and fund, for example, a new technology, what do those decisions reveal about investors’ expectations, and how can that analysis serve as evidence in a dispute?

The idea of this paper is that the investment in AI has to be profitable. If you are pouring a lot of money into something, your estimate of how productive that thing may be must have changed.

From that basic insight comes the paper. If investment rises from the tens of billions to the hundreds of billions over a period of a few years, the investors must be anticipating some payoff. Moreover, their expectation of that payoff will have risen. That expected payoff is the number we infer—or “back out”—from the investment data, because otherwise, why would the investment be worth making?

To back out that number, we use a standard condition from economic models of production. The intuition is that a company does not want to limitlessly scale up an investment: at a certain point, each additional dollar invested earns less than the one before, otherwise known as diminishing returns. The point at which this happens pinpoints the optimal scale of investment, and from that scale we can infer the payoff investors may be expecting.

One of the strengths of the U.S. financial system is that people make investments under uncertainty. Companies do not know what is going to happen, but if they do not make the investment, they will have lost a profitable opportunity for their shareholders. That is the paper’s framework: yes, the investment involves risk, but if risk always prevented companies from investing, the result would not be efficient either. So we work through the arguments. Could this be too much investment? Or could it be too little, with estimates that are too low?

Here, my rare events research ties in: sometimes it is not easy to imagine the future as a reality. Eventually there will be an outcome, and in hindsight, many will say it was obvious, but in the present moment, it does not feel that way. This happened in 2007, when I was motivated to write the rare events paper. A few people looked at the leverage in the housing market and forecasted a decline in house prices, but very few imagined anything like the 2008 financial crisis. Afterward, though, many thought it had been obvious all along.

As for how the analysis in the AI paper can serve as evidence in a dispute, I want to be careful. But broadly speaking, decisions made under uncertainty reveal the expectations behind them, and those expectations can be read from the data.

5. Across your research in asset pricing, derivatives, and risk, which topics map most directly onto expert work in securities litigation or investment-related investigations?

I will circle back to what I said at the beginning. One of the reasons I entered the finance field at the start of my career and later joined the SEC—besides finding the subject fascinating and enjoying learning about markets—is that finance is both theoretical and practical. The two elements cannot be separated. In my view, there is no such thing as theoretical asset pricing; the subject must stay close to the data and to the institutional features of real markets.

To offer a related example, when I began researching finance topics, I was drawn to the fact that modeling a derivative, or even the stock market itself, means working on many dimensions at once. A security is equations on a page. It is a contract between parties. It is a stream of cash flows. And in the case of a stock, it is also a claim on something physical: buildings and equipment. All of these dimensions exist simultaneously. Securities litigation interests me in part because it requires staying grounded in these practical matters, and that grounding deepens understanding of the subject.

One of the reasons I entered the finance field at the start of my career and later joined the SEC—besides finding the subject fascinating and enjoying learning about markets—is that finance is both theoretical and practical. The two elements cannot be separated.

The other thing I value about finance is the people involved. Something happened to somebody, or somebody is in trouble. It is interesting to be a part of that, and you can also be helpful to others. That is part of what drew me to the field.

Jessica A. Wachter

Dr. Bruce I. Jacobs Professor in Quantitative Finance,
The Wharton School,
University of Pennsylvania

5 Questions with Jessica Wachter: Pricing Risk, Reading Markets—From Rare Events to the AI Boom

1. Your research spans asset pricing, risk premia, investor behavior, and portfolio choice. How might those research areas arise in government investigations or litigation involving investment-related issues?

2. You developed influential work on “rare events,” unlikely but severe shocks, to explain stock market swings. In plain terms, what does that research show, and how does it inform disputes over risk, valuation, or claims that a loss was unforeseeable?

Inside DERA: How SEC Economists Turn Research Into Regulation

3. Part of your work connects finance and cognitive science, examining how memory and past experience shape investor decisions. What does this tell us about how investors actually behave?

5. Across your research in asset pricing, derivatives, and risk, which topics map most directly onto expert work in securities litigation or investment-related investigations?