Cheap Stocks Won, and Everyone Ignored It for Fifteen Years

Benjamin Graham had been telling people to buy cheap stocks since the 1930s. Academics regarded this as folklore, the sort of thing practitioners believe because they have not read the theory. The efficient market hypothesis said clearly that a ratio as public and as easily computed as price-to-earnings could not possibly predict returns. Everyone can see it. It is already in the price.

In 1977, Sanjoy Basu tested it properly and found that Graham was right. And then, remarkably, the finding sat there for the better part of two decades before the field fully absorbed what it meant.

The problem: testing folklore against theory

The P/E ratio is the simplest valuation number there is: what you pay per dollar of the company's annual earnings. A stock at a P/E of 8 is cheap in the plainest possible sense. A stock at a P/E of 40 is expensive: you are paying forty dollars for each dollar of current profit, which only makes sense if you expect those profits to grow a great deal.

Basu's test, note the subtitle of his paper, was explicitly framed as a test of the efficient market hypothesis. That is the right framing, because the stakes were clear. If low-P/E stocks systematically beat high-P/E stocks even after adjusting for risk, then a piece of freely available public information predicts returns, and the efficient market hypothesis in its standard form is in trouble.

He sorted stocks into portfolios by P/E ratio and tracked what each group actually earned, adjusting for risk in the way the theory of the day demanded.

The key idea via analogy: expectations are a price you pay

The result was a fairly consistent pattern: lower P/E portfolios delivered higher risk-adjusted returns than higher P/E portfolios. Cheap beat expensive. And crucially, the gap was not explained by the low-P/E stocks being riskier in the CAPM sense.

Why would this happen? The most durable explanation is about expectations.

A high P/E is not just a number, it is an embedded forecast. When the market prices a company at forty times earnings, it is asserting that this company will grow spectacularly. When it prices another at eight times earnings, it is asserting that this company is going nowhere, or backwards.

Now, both of those forecasts might be perfectly reasonable. But the market tends to push them too far. Exciting growth companies are usually genuinely good, just not as good as the price assumes. Boring cheap companies are usually genuinely mediocre, just not as bad as the price assumes.

When a company priced for greatness merely does very well, the stock disappoints, because greatness was already paid for. When a company priced for disaster merely does poorly, the stock rallies, because disaster was already paid for. The return does not depend on how good the company is. It depends on how good it is relative to what the price assumed. And the price assumes too much at both ends.

That asymmetry is the whole value effect, and Basu found it sixteen years before the Fama-French three-factor model made it a household name.

Why it mattered

• It was one of the first serious empirical attacks on market efficiency. Along with Banz's size result a few years later, it forced the field to accept that publicly available characteristics really did predict returns.
• It gave value investing an academic footing. What Graham and Dodd had asserted from experience, Basu demonstrated with data and a risk adjustment. The gap between practitioner folklore and academic finance narrowed considerably.
• It is the ancestor of the value factor. The book-to-market ratio that Fama and French eventually used is a cousin of the earnings yield Basu studied. The underlying idea, buy the ones the market has priced pessimistically, is identical.
• It is a lesson in how slowly fields update. The result was published in the Journal of Finance in 1977. It took until the 1990s for the profession to fully digest it. Uncomfortable findings do not get accepted quickly, they get accepted eventually.

The honest limitations

• Earnings are a fragile denominator. A P/E ratio breaks when earnings are near zero or negative, and accounting earnings can be managed, distorted by one-off charges, or made meaningless by write-downs. This fragility is a large part of why later researchers preferred book value as the anchor.
• The risk adjustment relies on the CAPM. Basu adjusted for risk using the model of his time. If the CAPM measures risk badly, and we now believe it does, then "risk-adjusted" outperformance is a claim resting on a shaky instrument. The cheap stocks may have carried risks the model simply could not see.
• Low P/E and small size overlap. Cheap stocks are often small stocks. Untangling whether Basu found a value effect, a size effect, or a tangle of both was a debate that ran for years afterwards.
• Cheap can mean doomed. A low P/E can reflect a market that is exactly right about a company in terminal decline. The value strategy works on average across a portfolio; on any individual name, cheapness is often a warning rather than a bargain.

The one-line takeaway

Basu tested Wall Street's oldest piece of folk wisdom against the efficient market hypothesis and found the folklore won: stocks with low price-to-earnings ratios beat expensive ones on a risk-adjusted basis, because the market's expectations at both ends of the valuation scale are systematically too extreme.