Lesson 2 of 8
Risk, or a mistake
The same premium has two readings: payment for bearing risk, or a recurring error by other investors. See what each side argues and what would settle it.
By Javier Sanz
Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.
Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)
In one minute
- One side reads the premium as payment for bearing a risk the market charges for; the other reads it as other investors repeatedly overpaying for recent growth.
- Fama, who documented the premium, argued for the risk reading and named the behavioural reading as its chief competitor in his own Nobel lecture.
- The two readings predict different things about the future, which is the only reason the argument matters to anyone holding shares.
A premium is a number. Why it exists is an argument, and it has not been settled by the people who found it. The two readings are not decoration: one says the extra return is the price of something unpleasant you have agreed to carry, the other says it is somebody else's recurring mistake.
Two readings of the same measured premium
Compensation for risk
- Cheap companies are cheap because something is wrong: weak profits, heavy debt, a shrinking market.
- The extra return is the market's charge for holding that, and it can arrive with a loss attached.
- If this is right, the premium can persist indefinitely and is not free.
A recurring error
- Investors extrapolate recent growth too far, paying up for winners and marking down losers past what the business justifies.
- The correction, when it comes, is the premium.
- If this is right, the premium can shrink once enough people know about it.
Both are argued from the same data by people who have read all of it.
What the man who measured it says
Fama used his Nobel lecture to set out both sides. He and French proposed their three-factor model as a version of Merton's intertemporal capital asset pricing model, which is the risk reading. He then names the alternative directly: the chief competitor to that risk story, he writes, is the overreaction hypothesis of DeBondt and Thaler and of Lakonishok, Shleifer and Vishny, in which prices overreact to the recent good times of growth companies and the recent bad times of value companies.
His two objections to it are worth carrying because they are testable. The behavioural account requires investors never to learn about their own bias. And Fama and French found in 1995 that the high average returns of value stocks persist for at least five years after the portfolios are formed, which he argues is a long time for a correction of irrational prices to take.
The awkward evidence for both sides
Novy-Marx measured profitability, defined as gross profits over assets, and found it had roughly the same power as book-to-market in predicting average returns. Profitable companies earned higher returns despite having higher valuation ratios, and controlling for profitability improved value strategies, especially among the largest and most liquid stocks.
He states plainly why that is a problem for the risk reading: profitable firms are less prone to distress, have longer cash flow durations and have lower operating leverage, which are the characteristics the distress-risk story requires the high-return group to have.
What would actually settle it
Three tests either side can be held to
-
Does the premium concentrate in companies that are genuinely fragile. The risk reading needs it to. If it lives in profitable, liquid companies, the reading is in trouble.
-
Does it survive publication. A mistake that everybody can read about should shrink. A risk premium has no reason to.
-
Does it survive the measurement choices. A result that needs equal weighting, one particular ratio and one particular period is neither risk nor mistake. It is a finding about a sample.
Two of those three are the subject of later lessons in this course. That is not an accident: the argument between the risk reading and the mistake reading is mostly settled, when it is settled at all, by evidence about decay and about construction rather than by theory.
Why this matters to a reader who holds shares
A premium that is payment for risk cannot be collected without the risk. A premium that is other people's error can vanish the year enough of them stop making it. The first calls for a horizon and a tolerance for looking wrong. The second calls for a reason to believe you are early rather than late.
Neither reading tells you to hold anything. Both tell you what you would need to believe.
Check yourself
What does the risk reading of the value premium claim?
- The extra return is the market's charge for bearing something genuinely unpleasant
- Cheap companies are always better businesses
- The premium is guaranteed over any ten-year period
Show the answer
Answer: A.
On the risk reading the return is compensation, not a free lunch, and it can arrive with real losses attached.Which finding did Fama cite against the overreaction explanation?
- That the premium disappeared after 1990
- That the high average returns of value stocks persist for at least five years after portfolio formation
- That value stocks have lower volatility
Show the answer
Answer: B.
He points to Fama and French's 1995 result, arguing that five years is a long time for a correction of irrational prices to take.Why is Novy-Marx's profitability result awkward for the distress-risk reading?
- Because it found no premium at all
- Because high returns showed up in profitable, liquid companies that are less prone to distress
- Because it used data from outside the United States
Show the answer
Answer: B.
The distress reading needs the high-return group to be fragile. His measured group was less prone to distress, had longer cash flow durations and lower operating leverage.Sources
- Fama, "Two Pillars of Asset Pricing", Prize Lecture, 8 December 2013, published 2014: the joint hypothesis problem, the ICAPM reading of the value premium, and the overreaction hypothesis named as its chief competitor.
- Novy-Marx, "The Other Side of Value: The Gross Profitability Premium", Journal of Financial Economics, 2013: gross profits to assets has roughly the same predictive power as book-to-market, and profitable firms are less prone to distress.
What this leaves out
- The lecture is one participant's account of an argument he is a party to. The papers he names make their own cases at greater length.
- DeBondt and Thaler and Lakonishok, Shleifer and Vishny are described here as Fama describes them. Their own papers are not quoted on this page.
- Nothing here resolves the question, and nobody has.
- None of it is advice about any security.