Eugene Fama on Finance
What this covers
Eugene Fama of the University of Chicago speaks with Russ Roberts about the theoretical and empirical foundations of modern finance, the 2008 financial crisis, monetary policy, and the gap between what economists can claim to know and what they actually do. The conversation moves across market efficiency, active management, the origins of the crisis, and the limits of policy tools, with Fama pressing back against prevailing interpretations on nearly all fronts.
On markets and returns, Fama situates testing the Efficient Markets Hypothesis within the Joint Hypothesis Problem: you cannot isolate market efficiency from an underlying model of risk and return, so apparent anomalies and bubble narratives rest on unstated risk assumptions. He argues that mutual fund return distributions are symmetric with chance, making celebrated outperformance a matter of Luck Versus Skill rather than genuine stock-picking ability. The crisis itself, he contends, was a broad recession hitting all asset classes simultaneously—not a housing-specific bubble. He rejects leverage explanations via the Accounting Identity: savings equal lending globally, so net leverage is zero, and the puzzle lies on the demand side—why sophisticated worldwide investors bought mortgage securities in the first place. On monetary policy, Fama argues central banks have far less control than assumed. Endogenous Money aggregates like M1 and M2 cannot be commanded, and interest rates themselves are largely endogenous; the Fed can at most tweak rates through expectations. He also flags regulatory design: leverage rules favoring triple-A rated assets incentivized their manufacture, and failing to resolve investment banks left their risk-taking incentives intact.
Beyond these positions, Fama addresses Publication Bias and the Multiple Comparisons Problem in finance research, contends that behavioral findings about individual irrationality do not cleanly explain market-level outcomes, and expresses skepticism that stimulus effects are empirically estimable with real precision. He emphasizes the unpredictability of research paths and warns against overstating what current finance knowledge will mean in twenty years.
Fama argues that markets efficiently reflect available information so 'bubbles' and active-manager outperformance are largely illusions of hindsight and luck, and that the 2008 crisis is best explained by a large recession rather than housing irrationality, while monetary and fiscal policy levers are far weaker and more endogenous than mainstream economists assume.
- Testing market efficiency is inseparable from a risk-return model (the joint hypothesis problem), so apparent anomalies and bubbles are not clean evidence of irrationality.
- Mutual fund return tails are symmetric and consistent with chance, so active outperformance is mostly luck, not skill.
- The crisis hit all asset markets simultaneously, which fits a recession story better than a housing-specific bubble; leverage nets to zero globally so it can't be the whole story.
The financial crisis resulted from rational but mistaken pricing of tail risks by sophisticated investors, not from irrationality or simple greed.
- A bailout-and-greed story explains why issuers might create bad securities but cannot explain why sophisticated buyers worldwide who weren't being bailed out would buy them; if greed leads you to issue securities you know are worthless, you wouldn't hold them yourself, so the puzzle is on the demand side.
“if greed causes me to put out securities that I know are no good, why would I hold them?... My argument is it can't explain why people who weren't generating these things and weren't going to be bailed out by us, investors in Norway, whatever--why were they buying?”
- If there was a housing bubble, then all markets did the same thing at the same time—stocks, corporate bonds, and commodities all moved together—so to claim a housing bubble you must accept bubbles in everything simultaneously, which is hard to credit.
“the reality is, all markets did the same thing at the same time. So you have to really face that fact that if you think it was a housing bubble, it was a stock price bubble, it was a corporate bond bubble, it was a commodities bubble. Are economists really willing to live with a world where there are bubbles in everything at the same time?”
- The buyers of these securities were the world's most sophisticated investors, so the only coherent explanations are that they made a mistake after the fact or mispriced extreme tail-event probabilities ex ante, which are very difficult to estimate—story-telling about irrationality is entertaining but not convincing.
“if greed causes me to put out securities that I know are no good, why would I hold them?... My argument is it can't explain why people who weren't generating these things and weren't going to be bailed out by us, investors in Norway, whatever--why were they buying?”
- You cannot tell a story about the crisis based on leverage because savings must equal lending: for everyone who is short bonds someone is on the other side, so the net amount of leverage in the world is always zero, and the subprime securities were bought by sophisticated investors all over the world who did not borrow to do so.
“savings has to equal lending. For everyone that's short bonds, somebody is on the other side. The net amount of leverage in the world is always zero... So you can't tell a story based on leverage.”
- The term 'bubble' has lost its meaning; it used to denote something with a more-or-less predictable ending, but people now use it to label any big price swing that turns out after the fact to be wrong, which is meaningless because all price changes are 'wrong' after the fact as new information arrives.
“people see bubbles with 20-20 hindsight. The term has lost its meaning. It used to mean something that had a more or less predictable ending. Now people use it to mean a big swing in prices, that after the fact is wrong. But all prices changes after the fact are wrong.”
- Almost everything in the crisis can be explained simply by saying there was a big recession, with the housing collapse being a consequence rather than the cause; economists lack good theories of why recessions occur and still do not understand the Great Depression.
“My explanation is you had a big recession. I think you can explain almost everything just by saying you had a big recession.”
Market efficiency and risk-return models are logically inseparable and jointly untestable.
- Whenever you test market efficiency you are jointly testing it with some model of risk and return; the two are joined at the hip and cannot be separated, which means market efficiency is not testable on its own—but conversely a risk-return model is also untestable without assuming market efficiency.
“whenever you are testing market efficiency you are jointly testing efficiency with some story about risk and return. And the two are joined at the hip. You can't separate them... A risk-return model is untestable without market efficiency.”
- A random walk in prices means successive price changes are independent of one another, so you cannot predict future returns based on past returns.
“That means that expected changes are successive changes are independent of one another... It basically means that you can't predict future returns based on past returns.”
- A market is efficient when prices at any point in time reflect all available information.
“What you mean is that prices at any point in time reflect all available information.”
- Full adjustment of prices to information can occur without any trading taking place; if all information were available at very low cost, prices could adjust through bid-ask quotes alone, so it is not true that someone must trade to bring markets to equilibrium.
“Merton Miller always liked to emphasize that you could have full adjustment to information without trading. If all the information were available at very low cost, prices could adjust without any trading taking place. Just bid-ask prices.”
- Available information is costless or very cheap up to a point, after which the cost function gets very steep and rapidly goes to infinity; this makes prices very efficient on cheap information while making it not pay to exploit additional information, since perfectly predicting the future would be infinitely costly.
“information is available, available at very low cost, then the cost function gets very steep. Basically goes off to infinity very quickly... therefore prices are very efficient because the information that's available is costless.”
Apparent market inefficiencies like momentum and mutual fund outperformance are artifacts of luck and multiple comparisons rather than genuine predictability.
- Economics research suffers from an unstated multiple comparisons problem: the same data have been used by so many people in so many unreported ways that there is no real statistical basis to evaluate a reported result, and anomaly-chasing researchers don't report the tests where they found nothing because they 'weren't interesting' or weren't publishable.
“In all economics research, there is a multiple comparisons problem that never gets stated... the data have been used by so many other people and the people using it now use it in so many different ways that they don't report, that you have no real statistical basis to evaluate and come to a conclusion.”
- Although some mutual funds beat their benchmarks by 3-6% a year, only about 3% of funds do about as well as you would expect by pure chance, and with over 3000 funds you expect many to do extremely well by luck over their lifetimes, so the celebrated star managers about whom books are written are likely just the beneficiaries of sustained luck.
“They beat their benchmarks by 3-6% a year. Nevertheless, only 3% of them do about as well as you would expect by chance... with 3000-plus funds, you expect lots of them to do extremely well over their whole lifetime. So, these are the people that books get written about.”
- Contrary to expectation, the weak form of the efficient markets hypothesis has been subject to the most anomalies, the biggest current one being momentum—prices tend to move in the same direction for short periods, so last year's winners stay winners and losers stay losers for a few more months.
“the weak form one has been the one that has been subject to the most, what people call anomalies, in finance... The big one at the moment is what people call momentum--prices seem to move in the same direction for short periods of time.”
- There is an incredible demand for market inefficiency because the entire investment management business is premised on markets not being efficient, creating a bias among professional money managers who make more money by portraying themselves as skilled active managers.
“There's an incredible demand for market inefficiency. The whole investment management business is based on the idea that the market is not efficient... the bias comes from the fact that they make more money from portraying themselves as active managers.”
- Active management in trying to pick stocks must be a zero-sum game because winners can only win at the expense of losers; the industry as a whole holds the market portfolio, so before costs the extreme right and left tails of mutual fund returns offset each other, and after costs there is no evidence any investor reliably has enough information to cover their costs.
“active management in trying to pick stocks has to be a zero sum game, because the winners have to win at the expense of losers... the industry basically holds a market portfolio. That's all before costs. If you look at returns to investors then there is no evidence that anybody surely has information sufficient to cover their costs.”
Expected returns legitimately vary over time with risk appetite, making return predictability consistent with market efficiency.
- Predictability in returns can be consistent with efficient markets because there is an expected-return component reflecting what people require to hold securities, which has no reason to be constant through time and is in fact higher on stocks during recessions and lower in good times.
“there's an expected return component, what people would require in order to hold those securities; and there's no reason that that has to be independent through time... higher on stocks during recessions and lower during good times. So there can be predictability in returns that is consistency with efficient markets.”
- Empirical estimates of the effects of stimulus are subject to so much uncertainty that the evidence allows you to say almost anything you want, so neither stimulus advocates nor skeptics can legitimately point to the data to prove their case.
“If you look at the empirical evidence, it basically allows you to say anything you want, because the estimates of the effects of stimulus are subject to so much uncertainty.”
- It is mysterious how the Fed can control interest rates given that they are largely endogenous; central banks cannot have wanted 13-14% short rates in the early 1980s, and the simultaneous beliefs that central banks control rates and that international bond markets are open are mutually inconsistent—at most the Fed can tweak rates via inflationary expectations.
“if I look at the evolution of interest rates, is it credible that in the early 1980s the Fed wanted the short term interest rate to be 13-14%?”
- The Fed pays interest on reserves because if there were an opportunity cost to the massive reserves it injected, the system would experience hyperinflation; this turns the injected liquidity into interest-bearing debt, which is posturing that actually makes controlling inflation harder by removing reserves from the inflation calculus so that everything comes down to currency.
“Because they know that if there is an opportunity cost from these massive reserves they've injected into the system, we are going to have a hyperinflation.”
Bailouts and implicit guarantees create moral hazard and distort risk-taking incentives worse than allowing failures.
- The auto bailout did not just change incentives but reordered the precedence of claims in contracts, abrogating the rule of law, which is a dangerous long-term consequence that outweighs the narrow metric of jobs saved.
“they changed the ordering of precedence in contracts. That's something that's really dangerous.”
- Investment banks were not subject to the same disposition rules as ordinary commercial banks under the FDIC, which can arbitrarily come in and resolve a failed bank, so the ongoing problem is that the crisis response failed to kill investment banks' incentive to finance things the way they always have.
“the investment banks weren't subject to the same disposition rules that would face an ordinary commercial bank. They are not subject to the FDIC... The ongoing problem is that you haven't killed their incentive to finance things the way they always have.”
- Any probability that a firm will be bailed out distorts its decision-making and dulls its risk sensitivity, making it low-cost to hold risky securities, and bailing out failing investment banks was the worst consequence of the crisis episode.
“Any probability that you are going to be bailed out is going to distort your decision.”
- Whether the bailouts were a mistake is largely irrelevant because there is no political regime that will allow major financial firms to fail outright.
“It's irrelevant because there is no political regime that will let that happen.”
Stimulus faces fundamental resource constraints and implementation problems that undermine its theoretical case.
- The broad money aggregates M1 and M2 are basically endogenous and cannot be controlled by the central bank because you cannot force people to move demand deposits; only the monetary base is genuinely exogenous, as evidenced by massive free reserves throughout the Great Depression.
“from your own data, they show that there were massive free reserves throughout the Great Depression... we can't force people to move demand deposits... M1 and M2--those things are basically endogenous. The only thing that's sort of exogenous is the monetary base.”
- Fiscal stimulus is fundamentally constrained by an accounting/resource identity—the resources have to come from somewhere, so there is no free lunch; the spending must be financed now and ultimately paid back by future generations, often for things that turn out mostly useless.
“There is no free lunch.”
- Even granting a theoretical case that stimulus could use idle resources, the problem of implementation is horrendous—just as in regulation, implementation is always the killer that undermines the policy.
“that's the problem of implementation, which is horrendous. The same problem in regulation: implementation, which is always the killer.”
Serendipity and collegial feedback drive research breakthroughs; teaching should train thinking rather than current findings.
- Because today's finance findings may all be irrelevant in 20 years, the best a teacher can do is train students how to think about these problems so they can absorb future work that overturns current knowledge—and the possibility of being overturned is what makes the profession fun.
“in 20 years, it may all be irrelevant; so the best I can do is to train you about how to think about these things, so you can absorb stuff that comes along in the future that may overturn what's there now.”
- Nobody can work in a vacuum; you need colleagues around you to enrich your work, as illustrated by Merton Miller discarding 8 out of 10 of Fama's ideas and being right on all of them—you get the final credit but many inputs from others go into it.
“nobody can work in a vacuum. You really need colleagues around you to enrich your work. You get credit for it in the end, but there are a lot of inputs from other people that go into it in the meantime.”
- You cannot set out to write a great paper; research is full of serendipity, so the best work comes from picking a small problem that keeps spiraling into a bigger one, while many small problems end up as nothing—a process that is inherently unpredictable.
“My best stuff has always been--I didn't start thinking about writing a great paper. I started thinking about a little problem; it just kept working in circles into a bigger problem.”
- The Black-Scholes option pricing paper is the most important economics paper of the century because every economist read it and it created an entire industry in the applied financial domain—an impact no other paper can claim.
“The Black-Scholes option pricing paper in view is the most important economics paper of the century... every economist whether he went into finance or not, read that paper. And it created an industry.”
Behavioral microlevel irrationality does not reliably translate into measurable market-level inefficiency.
- Behavioral economists are very good at describing irrational-seeming microeconomic behavior of individuals, but the jump from individual irrationality to market-level inefficiency is much more shaky.
“the behavioral people are very good at describing microeconomic behavior--the behavior of individuals--that doesn't seem quite rational... The jump from there to markets is much more shaky.”