From James Picerno at the Capital Spectator blog comes one way to think about future expected returns. In investing, if one thing is true, it is that past performance is not a good predictor of future performance. But, the past is often prolog with good times following bad and vice-versa. Given the current valuation and normal volatility of an asset class, it should be possible to predict long run expected returns for an asset class.
In the Expected Equilibrium Risk Premiums, jp at the Capital Spectator blog, gives a brief introduction to one method of calculaing risk premiums across the major asset classes. According to jp (His blog handle),
"The basic idea is to reverse engineer expected risk premia by making assumptions
about the overall market’s price of risk, the volatility of each asset class and
the correlation matrix for all of the above.1 With that data in hand, we can
estimate the implied risk premia."
Picerno then does just that with the past 3 years and 10 years of data. Implied Long Term Risk Premia are calculated that range from a minimum of 0.3% for US Investment Grade Bonds to a maximum of 11.6% for Emerging Market Equities with a weighted average of 5.3% in his Global Market Index.
Addtionally, Picerno suggests a method for tactically over/under weighting asset classes based on an investors subjective outlook for an asset class over the shorter run. For instance, if "After crunching the numbers, you find that DDM tells you that the stock market’s expected performance will be considerably higher than the equilibrium-based estimate for the long run ...overweight equities relative to the market weight as a device to raise your portfolio’s expected return."
most investors, the relevant risk premia benchmark is probably not the riskfree interest rate but rather purchasing power (inflation). Finally, even the most sophisticated investors will have a hard time estimating all the factors for the large number of asset classes included in Picerno's study.
I think it is very disturbing to not even suggest using a valuation anchor in these calculations. If we have learned anything over the past 15 years, it is that markets will become unhinged from fundamentals over the medium term - and then snap back violently when the force driving them away from equilibrium is removed. Nasdaq 2000, the Housing Market, Solar panels, and energy prices are just a few examples of markets that were driven to unsustainable levels and then dropped like a rock when the stool was kicked out from under these markets.
As a quick aside, Picerno does not even state what he means by the long run. He uses ten years of historical data in his example, but doesn't state the length of the horizon on his forecast. In my experience, 10 years is too short to determine the long run.
For most investors, the relavant benchmark for calculating risk premia is probably inflation or some sort of purchasing power metric. Investors (savers) ultimately hope to spend their savings - or pass on their purchasing power to someone else. The three month T-Bill has been a very poor benchmark for purchasing power over many periods - especially inflationary periods like the 1970s and potentially the next ten years. While the three month t-bill is fine as a benchmark for cash over the long haul, I object to its use in any risk premia calculations. To paraphrase the voice out of the
audience of New York Fed President Dudley's speach, "people don't eat t-bills."
Finally, the range of asset classes that Picerno follows is awesome, but most investors should probably follow the rule laid down by Peter Lynch "to invest in what you know." If you don't understand the broader issues with REITS (use lots of leverage and have benefited from extremely low interest rates) or Emerging Market bonds (Venezuela is one of the largest issuers), then just looking at some historical return data is not enough to give a prominent place in your asset allocation to that asset class. For most people, sticking to stocks and bonds should be sufficient, and understanding the expected risk premia of those two, primary, asset classes is no easy job.
I suggest reading Picerno's piece in more detail, but approach it with caution as there are better ways to arrive at expected risk premia. One way I would suggest would be to test expected risk premia against the theory of risk parity. Risk parity simply states that all asset classes, when adjusted for risk, should return the same in the long run. In other words, there are no durable risk premia in the long run, just medium term pertubations from equilibrium that the savvy investor can over/under weight to beat the normal long run return to capital. This is the method that I have adopted and plan to write more about in the coming months.