Mean Reversion - Value vs Growth Stocks and the Overreaction Debate
Benjamin Graham is probably the most famous contrarian investor who ever lived. Together with David Dodd, he invented what we now call value investing. The whole idea is simple. Buy stocks that other people don’t like. Stocks with low prices compared to their earnings or book value. Cheap stocks. Unpopular stocks.
But here is the thing. Is buying cheap stocks the same as buying stocks that recently crashed? Chapter 15 of Burton and Shah’s book asks exactly this question. And the answer matters because it separates two very different views of how markets work.
The Two Camps
On one side you have Fama and French. They showed that stocks with high book-to-market ratios tend to outperform. Their explanation? These stocks carry some hidden risk factor. The market is still efficient, you are just being compensated for taking on more risk.
On the other side you have De Bondt and Thaler. They found that stocks which performed terribly over the past three years tend to bounce back and outperform in the next three years. Their explanation? Markets overreact. People get too pessimistic about losers and too optimistic about winners. Eventually prices correct.
Both approaches give you a simple rule to beat the market using publicly available data. And that is a problem for anyone who believes markets are perfectly efficient.
The Weird January Effect
Here is something strange that both groups found buried in their data. A huge chunk of the outperformance happens in January.
Fama and French noticed that the effect of book-to-market on returns was roughly twice as strong in January compared to the rest of the year. The effect still existed year-round, but January was clearly special.
De Bondt and Thaler found the same pattern, but even more extreme. Their “loser” stocks earned most of their excess returns in January. Without January, the mean reversion in their data basically disappears. And this effect persisted even five years after they formed their portfolios. They themselves said they were surprised by it.
So what is going on with January? Nobody had a complete answer. But it raised uncomfortable questions about both sets of findings.
Is It Just About Price?
When a stock’s returns are terrible for three years, the stock price itself does most of the dropping. Dividends might get cut, but that is secondary. Book values are stickier. They don’t move nearly as fast as stock prices.
So think about it this way. If book value stays roughly constant and the stock price drops a lot, you end up with a high book-to-market ratio. That same stock also shows up in De Bondt and Thaler’s “loser” portfolio because of those bad returns.
In other words, the Fama-French value stocks and the De Bondt-Thaler loser stocks might be largely the same stocks. The two approaches could be picking up the same thing from different angles.
The Overreaction Hypothesis
De Bondt and Thaler saw their findings through a behavioral lens. Markets tend to overshoot. When good news keeps coming for a stock, investors get euphoric. They start ignoring occasional bad news. The stock gets overpriced.
Then when the tide turns and bad news piles up, the same thing happens in reverse. Investors overreact to the downside. The stock gets hammered below its real value. Eventually the market corrects, and prices revert to something more reasonable.
This is the overreaction hypothesis. And it is basically saying that investors are naive.
Lakonishok, Shleifer, and Vishny Take It Further
In 1994, three researchers named Lakonishok, Shleifer, and Vishny (people call them LSV) published a paper that bundled all these value strategies together. Their argument was bold. Fama-French, De Bondt-Thaler, and classic Graham-Dodd value investing are all basically doing the same thing. Buy stocks that are out of favor. Avoid the glamorous ones everybody loves.
LSV called growth stocks “glamour stocks.” These are the sexy companies that trade at high prices because everyone expects amazing future earnings. Value stocks are the boring, beaten-down names nobody wants to touch.
Their key claim was that value stocks are not just better performers. They are also less risky. LSV looked at bad market periods and found that value stocks still outperformed glamour stocks even when the overall market was falling. If value stocks outperform even in downturns, it is hard to argue that their higher returns are compensation for extra risk.
LSV also put their money where their research was. They founded a money management firm that by 2011 managed over $65 billion. And they weren’t the only ones. Dimensional Fund Advisors, started by students of Fama and French, grew to over $250 billion. Contrarian research turned into serious business.
The Small Stock Problem
Critics pushed back hard. One major criticism was that these effects were concentrated in small-cap stocks. And small stocks come with messy data.
Think about it. A small stock might not trade for a whole month. During that time, the last recorded price just sits there. But the actual value of the company could be changing. When it finally trades again, it looks like the price jumped in one day. But it didn’t really. The price was changing all along, just nobody was recording it.
Conrad and Kaul pointed out another technical problem. De Bondt and Thaler calculated returns by averaging monthly returns and adding them up. This works correctly only if you rebalance your portfolio to equal weights every single month. That means buying and selling every month, which costs money. When Conrad and Kaul recalculated using three-year holding periods without monthly rebalancing, the results mostly went away. Except for January again.
Ball, Kothari, and Shanken piled on. They showed that the “loser” stocks in De Bondt and Thaler’s data had really low absolute prices. When a stock trades at $2, even a tiny price fluctuation of 12.5 cents creates a huge percentage swing. They also found that switching from December-end data to June-end data made much of the outperformance disappear.
So the critics said: your results are fragile. They depend on small stocks with bad data, constant rebalancing that is impractical, and specific timing choices.
The Strange Double Standard
Here is what is interesting. Most of these criticisms apply equally well to Fama and French. Fama-French averaged returns in a similar way. Small stocks were a core part of their model too. The same data problems should have bitten them just as hard.
But somehow, the academic world gave Fama and French a pass while hammering De Bondt and Thaler. Book-to-market became a standard factor used in asset pricing models everywhere. But “past three-year performance” never made it into those same models. Maybe it was timing. Maybe it was politics. The book doesn’t fully explain why, but it points out the inconsistency.
Daniel and Titman Look for the Missing Risk
If Fama and French are right that book-to-market represents a hidden risk factor, someone should be able to find that risk factor. In 1997, Daniel and Titman went looking.
They examined high book-to-market stocks to find what common risk linked them together. Their conclusion? High book-to-market stocks do move together. But not because they share some mysterious distress risk. They move together because they are in similar industries, similar regions, similar types of businesses. They have similar characteristics, not similar risk exposures.
This was a problem for the efficient market explanation. If there is no identifiable risk factor, then the higher returns of value stocks cannot be explained as compensation for risk. It looks more like the market is just mispricing them.
What Does This All Mean?
The contrarian debate in the 1990s essentially revived Benjamin Graham’s old idea. Markets often price things wrong. They overshoot on the upside for popular stocks and overshoot on the downside for unpopular ones.
Whether you measure “out of favor” by book-to-market ratios or by past poor returns, you tend to find the same thing. Unloved stocks outperform over time. The explanations differ. Fama and French say it is risk. De Bondt and Thaler say it is overreaction. LSV says it is naive investors chasing glamour.
The evidence tilts slightly toward the behavioral side. But it is not a slam dunk. Data problems with small stocks are real. The January effect is unexplained. And future data could always shift the balance.
For you as an investor, the practical takeaway is this. When everyone hates a stock, that might be exactly the time to look at it more carefully. And when everyone loves a stock, be a bit skeptical. Crowds overshoot. They always have. The academics just spent a few decades proving what Benjamin Graham already knew.
Previous: Fama-French and Predicting Stock Prices