Lesson 3 of 7
What the evidence on value says
Cheap stocks outperformed expensive ones across decades of published studies. See what an average finding hides from a portfolio of fifteen.
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
- Several studies, across different decades and data sets, found that stocks priced cheap relative to their assets or earnings outperformed stocks priced expensive.
- Researchers still argue over why: some call it compensation for real risk, others call it a mistake investors keep making.
- "On average, over long periods" describes many stocks and many years. Fifteen names carry far more of their own luck than an average across hundreds does.
Cheap stocks, measured against their own assets or earnings, beat expensive stocks across multiple studies spanning different decades and data sets. The finding recurs, but it is a statement about a long-run average across a large number of companies, not a schedule for any one of them.
A method, then the evidence
- 1934Security AnalysisGraham and Dodd set out a rule, not a study: buy below two thirds of net current asset value, a floor built on the balance sheet alone.
- 1992Fama and FrenchSorting US stocks by size and book-to-market explained average returns; market beta added little once those two were included.
- 1994Lakonishok, Shleifer and VishnyContrarian, low-multiple portfolios outperformed glamour stocks in their sample, which they attributed to investors overreacting to past growth.
- 2009Tweedy, Browne's collectionA revised booklet describing over 50 studies, most finding an edge in low price to assets, low price to earnings, insider buying and sharp declines.
What the studies found
Fama and French sorted the market by price-to-book into deciles. The cheapest decile beat the most expensive one by a wide margin, and the pattern ran through every decile in between.
Average annual return by price-to-book decile
- Decile 1, highest P/B · most expensive7.7%
- Decile 312.7%
- Decile 514.9%
- Decile 716.7%
- Decile 10, lowest P/B · cheapest19.6%
Lakonishok, Shleifer and Vishny compared contrarian, low-multiple stocks against glamour stocks bought on strong recent growth, and found the contrarian group did better in their sample. Tweedy, Browne's booklet, revised in 2009, collected more than fifty such studies, most pointing at low price to assets, low price to earnings, insider buying and stocks that had fallen sharply.
Two explanations, and no agreement
Researchers disagree on why cheap stocks did better. One view treats it as compensation for genuine risk: a cheap stock is often a distressed or out-of-favor one, and the extra return is the price of bearing that. The other view, closer to Lakonishok, Shleifer and Vishny's own reading, treats it as a mistake: investors extrapolate a company's recent growth too far into the future, pushing glamour stocks up and cheap stocks down beyond what the business itself justifies. Both readings are argued from the same data, and neither has settled the question.
What "on average" does not tell you
An average built across decades and thousands of companies can be true and still hide long stretches where the opposite happens. A method can be ahead on average and still trail in most of the years you actually hold it, depending on which years those are.
The studies behind this finding held hundreds of stocks. A portfolio of fifteen names carries far more of its own luck, so it can trail the average even in a year the average itself was positive.
Check yourself
Fama and French's 1992 study found that, once size and book-to-market were included, market beta...
- Was the strongest predictor of average returns
- Added little further relationship with average returns
- Could not be measured for most stocks
Show the answer
Answer: B.
Their well known finding is that beta carried little additional relationship with average returns once size and book-to-market were accounted for.Lakonishok, Shleifer and Vishny explained the edge of contrarian, low-multiple stocks mainly by...
- Lower financial leverage in the contrarian group
- Investors extrapolating past growth too far into the future
- Higher dividend payouts in the contrarian group
Show the answer
Answer: B.
Their reading was behavioral: they attributed glamour stocks' underperformance to investors overpaying for growth that then failed to continue at the same pace.A method beats the market on average across many years. Does that mean it beat the market in most of those years?
- Yes, beating the market on average requires beating it almost every year
- Not necessarily, a positive average can still include several losing years
- Yes, because an average cannot be positive with losing years in it
Show the answer
Answer: B.
Lakonishok, Shleifer and Vishny's own count shows six losing years out of twenty two. An average does not say how the individual years were spent.Sources
- Graham and Dodd, Security Analysis, 1934: buying below two thirds of net current asset value.
- Fama and French, Journal of Finance, 1992: size and book-to-market against beta.
- Lakonishok, Shleifer and Vishny, "Contrarian Investment, Extrapolation, and Risk," Journal of Finance, 1994.
- Tweedy, Browne, What Has Worked in Investing, 1992, revised 2009: low price to assets and earnings, insider buying, declines, small size.
What this leaves out
- These are findings about groups of stocks across long periods, not about any single company named here.
- Each study's own market, sample period and costs differ from today's US market and from each other.
- The bar chart and the sixteen-out-of-twenty-two count are both from named tables in Tweedy, Browne's collection, not this product's own computation.
- Nothing here is advice about any security, and none of it says a method will keep working.