Good Companies at Fair Prices: Quality Minus Junk

Warren Buffett's most quoted line is that it is far better to buy a wonderful company at a fair price than a fair company at a wonderful price. It is a lovely sentiment, but for decades it was not a strategy, because nobody had pinned down what "wonderful company" means in a way you could compute.

Clifford Asness, Andrea Frazzini and Lasse Heje Pedersen set out to do exactly that. Their paper defines quality, measures it, prices it, and then asks the obvious question: if quality is so visible, why isn't it fully in the price already?

The problem: everyone says quality matters, nobody says what it is

Ask ten investors what a high-quality company is and you get ten answers: strong brand, good management, wide moat, clean balance sheet, reliable earnings. All plausible, none measurable.

Worse, there is a logical trap lurking. If quality is genuinely observable, then in an efficient market it should already be reflected in the price. A wonderful company should just be expensive, and expensive enough that its future return is perfectly ordinary. Buying quality should not make you money any more than buying a nicer house makes you money. So a paper claiming that quality earns extra returns has to explain why the market has not simply repriced it away.

The key idea via analogy: what would a rational buyer pay more for?

The authors start from first principles rather than from data mining. They ask: reading a standard valuation model, what characteristics should make a rational investor willing to pay a higher price per dollar of book value?

The answer falls out into three buckets, which become their definition of quality:

Profitability. How much profit the company generates per unit of its assets or sales. More profit for the same capital is worth more.
Growth. How fast that profitability is increasing. A business whose earnings power is expanding deserves a higher multiple than one standing still.
Safety. How little risk comes attached. Low volatility, low beta, low debt, stable earnings, low bankruptcy risk. A safe stream of cash flows is worth more than a risky one.

This is the elegant move: they do not go hunting through data for what worked. They reason out what a rational investor should pay up for, then go look at whether the market actually does.

Then they build a factor the standard way: buy a basket of the highest-quality stocks, short a basket of the lowest-quality ones. They call it QMJ, quality minus junk. (The word "junk" is deliberate and a little cheeky, and it is doing real work: the short leg is not just "average" companies, it is the genuinely unprofitable, unstable, indebted ones.)

The finding: the market pays up for quality, but not enough

Here is the result, and it is more subtle than "quality wins."

High-quality stocks are more expensive. The market is not blind. Investors do pay a premium for profitable, safe, growing companies, exactly as theory says they should.

But the premium they pay is surprisingly small, and it explains only a modest fraction of the variation in prices across companies. The market's willingness to pay up for quality is much weaker than a rational valuation model would imply.

And that under-payment is exactly where the return comes from. Because quality is underpriced relative to how good it is, quality stocks go on to deliver higher risk-adjusted returns. The QMJ factor earned significant risk-adjusted returns in the United States and, importantly, across 24 countries. Consistency across that many independent markets is the strongest argument the paper has that it found something real rather than a US-specific fluke.

There is also a neat kicker. Because quality and cheapness are somewhat opposed (good companies cost more), a QMJ signal and a value signal pull in different directions, and combining them works better than either alone. In fact, QMJ helps rescue the value strategy: it steers you away from the stocks that are cheap for the excellent reason that they are falling apart, which is the classic value trap.

Why it mattered

It turned a Buffett aphorism into a factor. After this paper, "buy wonderful companies at fair prices" is not folk wisdom, it is a portfolio you can construct, backtest and trade.
It supplied a defensible definition of quality. Profitability, growth and safety, derived from a valuation model rather than dredged from data, is now a common industry framing.
It made value investing more robust. The single most practical lesson is to screen out junk before you buy cheapness. Many value shops adopted precisely this.
It connects to a bigger theme. The safety component overlaps with the low-beta anomaly of the same authors, and the profitability component overlaps with Novy-Marx. QMJ is a synthesis: these apparently separate anomalies are facets of one phenomenon, the market underpaying for good businesses.

The honest limitations

Quality is a composite, and composites are flexible. The score bundles many accounting variables. Different reasonable bundles give different answers, and the more choices you make in constructing a signal, the more room there is for the historical result to be flattered by those choices.
The economic explanation is soft. The paper documents that the market underpays for quality but is honest that it does not fully explain why. It is a mispricing story without a fully satisfying mechanism, which always leaves a risk-based rival explanation lurking.
Quality is not a hedge for all seasons. Quality tends to do well in downturns, which is genuinely useful, but it can lag badly during speculative rallies when the market rewards precisely the junk it is short.
It has become crowded. Quality is now a mainstream smart-beta category with large assets tracking it. The premium available today may be smaller than the one measured over history.

The one-line takeaway

Asness, Frazzini and Pedersen defined quality as profitable, growing and safe, then showed that although the market does pay more for such companies, it does not pay nearly as much more as it rationally should, and that shortfall is a persistent source of return across two dozen countries.