Equity Risk Premiums: The 2012 Edition

As many of you who have been reading this blog for a while know, one of my obsessions is the equity risk premium. To me, it is the "number" that drives everything we do, while investing, and two events precipitated this post. The first was an article in the Economist on the topic, arguing that investors are expecting misreading the past and expecting higher returns from equities than they should. The second was the culmination of what has now become an annual ritual for me, which is updating my paper on equity risk premiums for the fifth year (I started in September 2008). You can download the paper by clicking here. For those of you who have no time for reading long tomes, I am going to try to summarize the paper in this post.

What is the equity risk premium? 
While it is always foolhardy to talk about "one" number encapsulating the stock market, I think the equity risk premium comes closest to meeting the requirements for such a number. The equity risk premium is the "extra return" that investors collectively demand for investing their money in stocks instead of holding it in a risk less or close to risk less investment. As a consequence, it reflects both their hopes and fears about stocks, rising as the fear factor increases.

Why does the equity risk premium matter?
The equity risk premium is used by almost everyone in finance, though it is often either taken as a given or used implicitly. Thus, a portfolio manager who decides to pull out of the stock market because she feels are stocks are over priced is telling you that she thinks that equity risk premiums will increase in the future. Investors who estimate the intrinsic value of assets or stocks are making explicit judgments about the equity risk premium (when they use DCF models) or implicit judgments (when they use book value or multiples). The costs of equity and capital that firms use to decide whether to invest in a project are built on equity risk premiums, as is all discounted cash flow valuation. Legislators and pension administrators decide how much to set aside to meet future pension obligations, based upon assessments of equity risk premiums.

What determines the equity risk premium?
Since the equity risk premium (ERP is a number for the entire stock market, it is determined by the overall characteristics of the investor population and macroeconomic factors. In particular:
a. The ERP should increase as investors become more risk averse and/or prefer current consumption more.
b. As uncertainty about economic growth, inflation and other macroeconomic variables increases, the ERP will rise.
c. Since investors are dependent upon the flow of information from firms (accounting or other), the ERP will rise as information becomes less reliable or less available.
d. The fear of catastrophe always hangs over equity investments and as that fear rises, the ERP will go up as well.
Since all of these factors can change over time, you should expect the equity risk premium to vary across time as well.

How do you measure the equity risk premium?
There are three broad approaches to estimating the equity risk premium and they can yield very different values:

1. Surveys: You can ask investors or analysts what they think stocks will generate as returns in the future and net out the risk free rate from this value to get to a equity risk premium. For instance, Merrill Lynch surveys global portfolio managers and reports a survey premium of 4.08% in early 2012, i.e., portfolio managers expect stocks to earn 4.08% more than the risk free rate. A survey of CFOs by Harvey and Graham yields a 3.50% equity risk premium but another one by Fernandez yields higher numbers (5-5.5% for the US and higher values for emerging markets). I distrust survey premiums because they often represent hopes (more than expectations) and are more reflective of the past than the future.
2. Historical premium: You can look at the past and estimate the premium you would have earned investing in stocks over a risk free investment. Thus, if you look at the 1928-2011 time period for the US, you would have earned an annual compounded return of 9.23% if you had invested in stocks, over this period, but an annual return of only 5.13%, investing in treasury bonds. The difference (4.10%) would be your historical risk premium. Even with historical data, you can get different numbers using different time periods, treasury bills instead of bonds as your risk free investment, and computing an arithmetic average instead of a compounded average. The values, for the US markets, range from 7.55% (arithmetic average premium for stocks over T.Bills from 1928-2011) to -3.61% (geometric average premiums for stocks versus T.Bonds from 2002-2011). Given the volatility in stock returns, you should be wary of equity risk premiums computed with less than 40 or 50 years of data (almost always the case with emerging markets) and be skeptical even when longer periods are used (the standard error, even with the 1928-2011 data, is about 2.36%). Implicitly, no matter which of these numbers you decide to use, you are assuming that the equity risk premium for the US market has not changed in any material fashion over the last century and that they will revert back to historical norms sooner or later.  If I had to use a historical risk premium, I would go with the 4.10%, since it is long term, a compounded average and over a long term risk free rate. However, I am much more uncomfortable with the assumption of mean reversion in the US market than I used to be. since, in my view, the structural shifts that have come out of globalization have changed the rules of the game. As a consequence, I no longer use historical premiums in either valuation or corporate finance.
3. Implied premium: Just as you can compute a yield to maturity (a forward looking value) for a bond, based upon the price you pay and the expected cash flows on the bond (coupons and face value), you can compute an expected return on stocks, based upon the price you pay and the expected cash flows on stocks (dividends and buybacks). On January 1, 2012, for instance, with the S&P 500 at 1257.60, I estimated an expected return on stocks of about 7.88%, which yielded an equity risk premium of 6.01% over the treasury bond rate of 1.87% on that day. It is true that this premium is a function of my assumptions about expected cash flows in the future, but there are two reasons why I trust it more than the historical premium. First, it is forward looking since it is based upon expected cash flows in the future. Second, there is real money backing up this number, since it is based on what investors are paying for stocks today (rather than what they are saying). Third, the error on your estimate (arising from your errors on expected cash flows) will be far lower than the standard error on a historical risk premium. Given the dynamic and shifting price of risk that characterizes markets today, I think it makes sense to compute and use an updated implied equity risk premium in valuation and corporate finance.

The range of estimates we obtain for the equity risk premium from the different approaches is large but they should be judged based upon how well they perform in forecasting the future (both of equity risk premiums and actual stock returns)

Looking at the correlations, the implied equity risk premium performs best, yielding the best predictor of not only next year's equity risk premium but also of actual returns on stocks over the next decade. The historical premium performs worst, often moving in the wrong direction.

How is the equity risk premium related to risk premiums in other markets (bonds and real estate, for instance)? 
In the corporate bond market, the price of risk is measured with the default spread, i.e., the difference between the yield to maturity on a risky bond and the risk free rate at the time. Even in the real estate market, the capitalization rate operates as a measure of expected return and the difference between that rate and the risk free rate is a measure of the risk premium in real estate. In the figure below, we graph all three numbers (the implied equity risk premium, the default spread on a Baa rated bond and the cap rate premium for the US from 1980 to 2011.

Note that real estate behaves like a very different asset class in the 1980s, with the cap rate premium often in negative territory. This was the basis for the advice that many of us got in that period that investing in a house or real estate provided diversification benefits, especially if the bulk of our wealth was tied up in financial assets. Starting in the 1990s, real estate has begun to look more like a financial asset, a finding that hit home with many in the last few years, as housing prices collapsed just as stock prices and corporate bond prices declined. If these trend lines continue to hold, we may need to find a new asset class to get the benefits of diversification in the future.

It is also worth noting that when the risk premiums in the three asset classes diverge, it is a sign that one market or the other is in a bubble. Note that in early 2000, the equity risk premium dropped to almost the level of the Baa default spread, reflecting the dot com bubble. In the 2004-207 period, default spreads and the cap rat premium plummeted, relative to the ERP, reflecting the housing and credit market bubble in that period.


What is the "right" equity risk premium to use in corporate finance and valuation?
So, what is the risk risk premium to use in today's markets? The answer depends upon what you are trying to do.

1. If you are making a judgment on asset allocation, i.e., the percent of your wealth that you want to invest in equities, bonds, real estate or other asset classes, you can bring your point of view into play. Thus, if you feel that the current implied premium of 6% is too high (low) and will thus come down (go up), you should invest more (less) in equities than you normally would (given your age, cash flow needs and risk aversion).
2. If you are valuing companies or assets, you generally should stick close to the current implied premium, notwithstanding your views in the asset allocation component. The reason is simple. Using an equity risk premium that is significantly different from the current implied premium brings in a market view into your valuation and thus confounds your final conclusion. To illustrate, if you use a 4% equity risk premium to value a stock in January 2012, you are effectively assuming that the S&P 500 is undervalued by about 25%. As a consequence, if you find your stock to be cheap, based on the 4% ERP, it is not clear whether you did so because the stock is in fact cheap or because of your market views.
3. If you are a business, using the ERP to estimate your costs of equity and capital, you have a little more leeway. You can use an average implied equity risk premium over time (it has been about 5% over the last decade) in your estimation, built on the premise that there is mean reversion even in implied premiums and that your projects are long term.
4. If you are a legislator or pension fund administrator, you also have some leeway. If you do not want your contributions to the fund to be volatile, you should use the average implied equity risk premium as well.

Back to the Economist

I like the Economist, as a news magazine and as a commentator on financial issues, but I think that this article does not quite hold together. First, it starts with a premise that I investors who look at historical data are getting an estimate of the premium that is too high and that a sustainable long term expected return on stocks should be a sum of the dividend yield and the expected long term growth in dividends (which would yield a lower value). Fundamentally, I don't disagree with that notion but I think that the use of the dividend yield is far too narrow a measure of cash flow. Incorporating the additional cash that firms are generating into the yield (either by adding in buybacks or computing a potential dividend) does provide a much higher expected return for stocks. Second, it implies that using a high risk premium is an aggressive assumption, i.e., it leads to investors paying more for stocks than they should, but the opposite is true. If you demand a higher return on stocks, you will pay less for them today, thus pushing down stock prices, making it the conservative assumption to use. In fact, if we take the article's suggestion and build in a lower equity risk premium, you would be be pushing up stock prices today dramatically. 

As a general rule, I find that discussions about the equity risk premium are rife with misunderstanding about what it is, why it changes over time and how it affects investing/valuation. It would be far healthier for all concerned if analysts and investors were more explicit about why they use the equity risk premiums they do and what market views are at their basis.