Aswath Damodaran, Kerschner Family Chair in Finance Education and a Professor of Finance at New York University Stern School of Business, is well known for his books and articles in the fields of valuation, corporate finance, and investment management, philosophies, and strategies. On April 2, he treated the CFA Society Chicago to a tour de force through the foundations of risk premia, the macroeconomic determinants of equity risk, and how the risk premium can me misused.
Damodaran’s talk was followed by a panel which included himself, Michele Gambera, co-head of Strategic Asset Allocation Modeling at UBS Asset Management, and Bryant Matthews, global director research at HOLT. The panel discussion was moderated by Patricia Halper, CFA, co-chief investment officer at Chicago Equity Partners.
Damodaran pointed out that while the risk-premium is referred to as one number, it contains several various risk factors, such as political and economic risks, information opacity, and liquidity risks. Despite the underlying complexity, a common way to derive the risk premium is from the average volatility of some historical period. This, Damodaran warns, is a dangerous approach. By using historical data you can derive any risk premium you want by using the time horizon of your choosing. When you look at historical averages, you are also searching for a number that nobody has ever experienced. And even if they did, you should not believe that history will simply repeat itself. And even if history did repeat itself, you are still estimating a number with large error margins. In the end, the exercise is just not useful.
Damodaran has done a lot of work determining equity risk premia for different countries and makes his data available on his homepage. His approach is to derive an implied risk premium based on consensus forecasts of earnings and adding country risk premia for different countries. He cautions that there is no pure national premium thanks to our integrated world. Much of S&P earnings, for instance, are derived from abroad, and this must be taken into account.
For a person who has devoted so much time to estimating risk premia, it may come as a surprise that Damodaran thinks people should spend less time on it. His approach is that once you observe the market-implied risk premium, you should use this in your valuation model and devote your attention to estimating cash-flows. Right now, too many people are wasting too much time on valuing companies through finding the perfect risk-premia when cash-flows are ultimately going to determine whether they will get valuations right. Academic finance is another culprit here, which spends too much research time on discount rates.
Ask yourself this, are you working on your model’s risk premia because that is where you have superior knowledge, or because it is your comfort zone?
Damodaran is also critical of the use of the price-to-earnings ratio to assess valuation, since it looks at earnings only for the current period. In the US market the ratio may look high, but the pictures very different for current implied risk premia. Since 2008, risk-free rates have come down while expected stock returns have remained roughly the same. This actually implies a higher risk premium.
Michele Gambera shares Damodaran’s criticism of historically derived risk premia. He also pointed out that while the risk premium fluctuates a lot, we pretend in our models that it is constant. In effect, Gambera stressed, we are estimating a random-walk variable. A better approach for your valuations is to use a forward-looking covariance matrix with various factor loadings.
Should we therefore throw the historical data out the window? When asked the question, Bryant Matthews of HOLT pointed out that historical data are not all useless in a world where variables tend to mean-revert. But you may need to wait a long time for it to happen.
Is there a small-cap premium? Damodaran pointed out that if you estimate the historical premium since 1981, it is negative, which is clearly fictional. However, Matthews estimated a small cap premium of 0.6%, albeit with a standard error that makes it statistically zero. By slicing the equity market in other ways, he estimates that value stocks tend to have a 3.5% equity premium over growth stocks, while Fama and French’s quality stocks-factor enjoys a 2.1% premium over non-quality stocks.
Matthews has also calculated market implied risk premia for over 70 countries, and found it rising in the US from 0% in 2000 to 4% today. Such estimates, he pointed out, are often counterintuitive for clients. Surely, equities were riskier in 2000 when valuations were high. But precisely because valuations were so high, the implied risk premium, which was part of the discount rate, was low.
Can we make money by investing in high-risk premium stocks? After all, theory tells us returns are the reward for taking risk. Yet as Gambera pointed out, high-volatility stocks tend to be favored by investors in part as a way to leverage up according to the CAPM-models, as is done for instance in risk-parity models. At the same time, pointed out Matthews, low-volatility stocks are generally also high-quality stocks and therefore tend to have high return, despite their historically low risk.
Matthews argued that while profits are high for the US market as a whole, this really applies to only 100 companies. This concentration, he suggested, is due to lax regulations. Damodaran, however, suggested that antitrust measures cannot be relied on to change this fact. They may have been politically attractive in the time of Standard Oil, when that company’s dominated position allowed it to raise prices. The dominant firms of today are offering consumers very low prices. Break them apart and any politician will be met with discontent from voters.
Let us end with some historical perspective from Michele Gambera. Much of the early work on risk premia was made at a time of a very different market structure of industrialized countries. Steel and railroads ruled the day and many of today’s giants were not listed. The likes of Alphabet and Facebook pose new challenges in estimating risk premia. This suggests that now more than ever historical data will be misleading in estimating the risk premium, a modest number that means so much.