3. Quality and low volatility

Level 9: Factors

The two factors in the previous module are the veterans. This one deals with two others that came later and that raise different problems: quality, which is hard to define without cheating, and low volatility, which is easy to measure and which theory should not allow to exist.

Quality: businesses that do not come apart

There is no single definition of quality, and that is its first weak point. The definitions used seriously converge on four measurable traits: high and above all stable profitability over the years, little debt, moderate asset growth rather than expansion by buying things, and profits that genuinely convert into cash instead of remaining accounting entries.

The trap is plain to see and needs saying early: quality is not the same as a good investment. A magnificent company at an absurd price is a bad purchase, and the market knows it: excellent businesses trade dear. If the factor pays anything, it is because the market does not charge enough for that excellence, not because the excellence is free.

That said, there is one way of measuring quality that has held up better than the rest. Robert Novy-Marx contributed a specific piece at the start of the last decade. Measuring profitability with gross margin relative to assets, that is, with a figure sitting very high up the income statement and therefore less easy to manipulate than the bottom line, he found that this measure sorted the future returns of shares with a power comparable to that of the value ratio itself.

The interesting part comes next. Profitable companies tend to trade dear, so a quality criterion frequently selects the opposite of what a value criterion selects. That, which looks like a problem, is the advantage: two criteria that fail at different moments make the journey less demoralizing. And their combination has a simple reading, buying good businesses at reasonable prices, which is exactly what the investing tradition has spent a century saying without being able to measure it. Fama and French ended up incorporating profitability and investment policy into their own model.

Low volatility: the uncomfortable anomaly

Here things get serious. The whole of theory rests on an idea that looks like common sense: whoever takes on more risk must expect more return, because otherwise nobody would take it on. The shares that swing most should pay more than the calm ones.

The data have been saying otherwise for half a century. As early as 1972 several independent studies found that the relationship between risk and return was far flatter than the model required. And something worse than flat was subsequently documented: in many periods and markets, the most volatile shares returned less than the calm ones, not only per unit of risk, but in absolute terms.

The most cited result compares quintiles of US shares ranked by volatility from the late 1960s to the 2008 crisis: a dollar invested in the calmest quintile ended up multiplied several dozen times, while the same dollar in the most agitated quintile did not even recover what was put in. More risk, less money, over forty years. That is not a premium that fails to show up: it is a premium with the sign reversed.

First explanation: those who cannot use leverage buy beta

The explanation that has caught on most was developed by Andrea Frazzini and Lasse Pedersen. Many investors want returns above the market's but cannot or will not borrow: funds with an express prohibition in their prospectus, pension plans with regulatory limits, private investors with no access to credit or no stomach for it.

Those people have only one route left for raising their expected return: buying the racier shares, which carry the leverage built into the security itself. That demand makes the volatile expensive and the calm cheap, and since price and future return run in opposite directions, the result is the one observed.

Second explanation: people like lottery tickets

The other explanation is psychological and probably complementary rather than rival. People overpay for small probabilities of enormous prizes, just as they buy lottery tickets knowing the odds are against them. Highly volatile shares are lottery tickets: almost all of them disappoint and a few multiply twenty-fold, and it is those few that occupy the mind of whoever is buying.

On top of this comes a professional incentive that pushes the same way. A manager judged against an index knows that a calm portfolio will lag in rising markets, and lagging is what gets you sacked. Even if he knows the volatile stuff is expensive, proving it is not worth his while.

What this implies for your portfolio, and what it does not

Harvesting this anomaly in full requires leverage: buying the calm stuff and borrowing until you match the market's risk. And there it is worth recalling what this academy's Risk course sets out calmly:

• Leverage does not scale the average, it scales the path. It changes the shape of the trajectory, not just its slope.
• It introduces a risk the unleveraged portfolio does not have: forced liquidation at the worst moment, decided by whoever lends to you rather than by you.
• Its cost is not zero, and in the stretches where money gets more expensive it eats precisely the difference you were going to capture.

Without leverage, a low-volatility portfolio does not give you more return: it gives you a similar return with fewer shocks. For many people that is worth a good deal, but it is a different promise from the one usually sold.

Real case: the calm quintile and the year 2008

The year 2008 illustrates the asymmetry with round numbers. While the US index fell around 37%, low-volatility portfolios fell considerably less, in the region of twenty-odd points, and quality portfolios behaved similarly for a related reason: when credit is cut off, having little debt and stable profits stops being boring and becomes the only thing that matters. The lesson is not that these factors always protect, because they do not, but that their return is concentrated in the years that hurt most. Hold on to that idea for the next module.