MINYANVILLE ORIGINAL Over the last 60 years, the concept of risk premium has embedded itself so deeply in finance that it is hard to think of investing without relying upon it. Like the man who drinks water all his life and so thinks it has no taste, many people treat arguments based on risk premium as obvious because they have been ubiquitous in finance for so long. For example, many people wouldn’t think twice before agreeing to statements like:
Since conservative investors prefer low-risk portfolios, assets that add a lot of risk to portfolios have to carry a higher average return than assets that are either low-risk or uncorrelated with conservative portfolios.
Since levering an investment increases its risk, it should also increase its expected return.
If the risk of an investment goes up and there is no change to its expected future value, its price today should go down.
It’s important to separate this idea from merely keeping expected value constant. If a risk-free 10-year bond pays 5% interest, a bond with a 10% chance of default must pay approximately 6% (the exact number depends on the timing of potential defaults and the size of potential recoveries) just to have the same expected value as the risk-free bond. To have a risk premium, the risky bond would have to pay more than 6%, so that its expected return was higher than that of the risk-free bond. In other words, the fact that junk bonds sell for higher yields than investment grade bonds does not prove that there is a risk premium. We would need to show that portfolios of junk bonds had higher long-term average returns than portfolios of investment grade bonds.
Eric G. Falkenstein argues that there is no risk premium, and there never was, so conventional investing advice is deeply misguided. More important, he has developed a consistent and plausible alternative explanation. This is a very valuable argument, even if it is ultimately not correct. You cannot understand risk premium if you think it is obvious, you need to see why it might not exist to see how to look for it. And, of course, if the argument is correct, it is even more valuable.
The book also describes an investment approach, a version of what is generally called low-volatility investing. The author is among the pioneers in this area and he advocates a reasonable version of it. However you need not accept his argument to take advantage of the insights that led to the general development of low-volatility investments. There are different theories out there to explain why it works. Falkenstein, in my opinion, has the boldest plausible explanation, but even if he is correct, that doesn’t mean it leads to the best practical portfolio advice or the best investment products (of course, it also doesn’t mean the contrary). Theory is important, but implementation details like fees, expenses, taxes, execution quality, data quality, and dozens of others are more important. So just because you like this book doesn’t mean low-volatility investing is for you, and just because you like low-volatility investing doesn’t mean you have to like this book.
The Missing Risk Premium: Why Low Volatility Investing Works
rates four stars on each of the following four scales:
There aren't many quadruple four-star books out there, which automatically makes this interesting. It's also a bargain, at $14.95. When I reviewed the author's first book, Finding Alpha: The Search for Alpha When Risk and Return Break Down
, I complained about the $95 list price and suggested it should be $25 list to sell for $15 at Amazon. I don't know if he was paying attention, but if he was, he traded through my bid by a nickel. Unfortunately he didn't listen to my complaints about the copy editing and production values.
Prior to 1950, when the main historical portion of The Missing Return Premium
begins, there was not much consideration of risk premium in finance. There were a few exceptions, such as Louis Bachelier, but to mainstream financiers, a good stock was a stock that went up and a bad stock was a stock that went down. People knew it was hard to tell the difference, so even a good investor bought lots of bad stocks, but those were mistakes rather than the result of risk for which the investor would be compensated in the long run. Investment research meant studying investment fundamentals to pick sound securities, not estimating probability distributions to optimize portfolios. Randomness was noise that made investing more difficult, not the essential basis of investing.
In the conventional account, from 1950 to 1975, starting with Harry Markowitz and ending with Eugene Fama, modern finance discovered and verified a single, universal risk premium that explained everything, leading to a financial version of the Enlightenment. Old wisdom was thoroughly overturned and a new, rational, empirically validated theory explained everything. Even better, this theory fit seamlessly into the current views in economics and meshed with some quantitative researchers in other social sciences. It was beautiful, mathematically consistent and true.
Falkenstein retells this story armed with a skeptical mind and knowledge of subsequent discoveries. It is a selective account, as it has to be to fit into half of a small book, but it is not unfairly selective. That is, while he leaves out many nuances and simplifies much of the argument, he does confront the strongest points in favor of a risk premium. In this version, researchers were drawn to the idea of risk premium for theoretical reasons. Results that confirmed risk premium were cheered without skepticism; results that contradicted it were subjected to vigorous challenge that eventually overturned them.
This is an account everyone interested in quantitative finance should read because it gives a fresh and important look at the formative years of the field. Personally, I wish it had been higher-minded. There is no mention of the attacks on the field from economists and practitioners, and the tremendous practical good done by the research. It suggests a group of self-satisfied ideologues blindly creating evidence for an obviously false theory. In fact this was a group of brilliant, hard-working, skeptical people who questioned every assumption as rigorously as Falkenstein does, but without his benefit of hindsight, and who read and debated internal and external criticism. That the result has so many flaws by modern lights is not a criticism of the researchers but a testament to how hard it is to say anything at all about expected return.
At one point, Falkenstein mentions unnamed people who consider Eugene Fama to be a “lightweight.” When I first read the book, I thought it was smear, hiding behind unnamed sources to say something the author is afraid to say aloud. However, after an email correspondence with the author, I see that he meant the comment to be a criticism not of Fama, but of other academics who were obsessed with theoretical rigor rather than empirical tests. I mention it since it would be an incendiary claim, and other readers may draw the same conclusion as I initially did.
Next, Falkenstein surveys 31 asset classes and investment spreads and finds evidence for positive risk premium in only four (REITs, Equities versus risk-free debt, high-yield bonds versus high quality bonds and short-term treasuries versus longer term). Thirteen show zero risk premium, and 14 show a negative return premium (including high-risk stocks versus low-risk). This is a valuable study, informative and interesting to read. I disagree with some of the specifics, but the overall picture is worth considering.
The book moves on to a more general historical and philosophic context. Why should mere exposure to risk be rewarded? There are always lots of people willing to do it. Some may even like it, after all, people gamble apparently for pleasure in casinos and dangerous sports and bar fights and lots of other areas in life. But even if everyone found risk distasteful, if people got paid for doing distasteful work then people who clean toilets would make more than professional athletes and movie stars.
Even if people did get paid for taking risk, it would not be for well-understood statistical risk like buying a volatile stock. Real risk is walking into complicated and uncertain situations, for which probability distributions are unknown. And losing money (especially other people’s money) is nothing like the pain of being thought an idiot. Defying convention, underperforming the benchmark, being wrong while looking stupid; these are the unpopular forms of pain.
The single most important message in the book, which is also in Finding Alpha
, is that expected return is not something you get for passive exposure to known unpleasantness, it is a niche in which you have advantages over other people. Success does not come from combining well-known bets a little better, it comes from making your own bets.
Finally, Falkenstein works out the mathematics of a market in which risk consists of underperforming the average rather than poor gross returns. He justifies this with a variety of arguments, and shows that it results in a zero premium for risk. This is an important result, even if you disagree with the assumptions or think the empirical predictions are false.
Even more interesting, he addresses the flaw that sinks most models without risk premia, he shows that rational, risk-averse investors will not offset the actions of irrational or risk-loving investors. In conventional models, if some investors bid up the price of an asset beyond equilibrium values, smart investors will reduce holdings or short the asset, quickly restoring equilibrium. In Falkenstein’s model, arbitragers may reduce holdings in the overpriced assets, but will not short them, and will not reduce holdings enough to restore the equilibrium price. I believe this is an accurate description of financial markets, although I’m not sure that Falkenstein’s model for it is correct.
One interesting thing to me is Falkenstein’s equation 5.34, which shows the expected return investors demand on an asset increases with the fraction held by other investors. I come up with the identical formula in chapter 5 of Red-Blooded Risk
. Mine is stated inversely; I show the expected return decreases with the fraction the investor herself holds. The two are mathematically equivalent. Falkenstein’s version is a consequence of investors caring only about relative performance. My version comes from the assumption investors hold Kelly-optimal portfolios and ignore not only all other investors, but all other potential investments. I doubt there is a deep economic connection between our models, I suspect it is a case that certain simple relations pop up in different contexts. Like most applied mathematicians, however, I instinctively trust work when familiar relations appear naturally. It doesn’t mean you’re right, but when you are right, you usually get things with obvious parallels to other right things.
The Missing Risk Premium
is an important book. At worst, the effort you expend to refute its claims will deepen your understanding of conventional models. At best, it will be a breakthrough to a new understanding of financial markets and better investment results. I suspect you’ll end up somewhere in the middle, which is a good place to be, especially if you hate to deviate from the benchmark.