Divining the future by dissecting the past is, like parachuting and wrestling crocodiles, a venture saddled with more than a little risk. Yet each is thrilling in its own way, and at times may even offer perspective–assuming you don’t lose an arm and a leg, figuratively or literally, depending on the sport.
Ours is a quantitative adventure in the land of investing, and among the various statistics we routinely review is the equity risk premium (ERP), and so the risk is limited to those little bits of paper with images of George, Abe, Alex, et. al. There is much debate about the underlying rationale for the ERP, or the excess return thrown off by stocks relative to the risk free rate, which we define here as the 12-month rolling total return on the S&P 500 less the same on 3-month T-bills. Nonetheless, the ERP is quite real, or at least it has been in the past. The question is whether it will continue in the future, and if so, by how much?
As our chart below illustrates, the ERP is hardly a static number. Back in July 1997, the trailing 12-month ERP was an extraordinary 47%; in September 2001, the ERP over the previous year had turned negative to the tune of -31%. The average ERP turns out to be around 7.9% since 1987 through the end of last year.
The ERP in recent years has tended to stay with in a range of 4-12%. Over time, that could deliver a tidy gain for the patient, long-term investor. Ah, but deciding if 4-12% will be high, low or middling is the asset allocator’s dilemma. Of course, there is no good answer, given the history of prognosticating investment returns and the myriad of factors that ultimately go in to determining how the future unfolds.
That said, one’s expectations of the future may be informed by the outlook for equity dividends. As Robert Arnott observed in a 2003 essay, most (nearly two-thirds) of the ~7.9% annualized total return of the U.S. stock market for the 200 years through 2002 came from dividends.
Further, dividend growth has averaged about 1% over the rate of inflation during the previous two centuries, points out Arnott, who’s chairman of Research Affiliates, a research shop that designed the “fundamental” indices that have become the basis for a suite of new ETFs, including PowerShares FTSE RAFI US 100 (NYSE: PRF). Alas, the inflation-adjusted per-share growth in earnings and dividend growth since 1965 has been zero, zilch, nada.
History, in short, suggests that the long-term return on equities, and thus the ERP, will depend heavily on dividends. Will dividends grow overall for equities? Will the rate of dividend growth lag, pace or exceed inflation? What will the risk-free rate be going forward? We don’t have answers, although we have some guesses. In any case, we know the right questions to ask.
Is it now de rigeur to use the T-bill as the risk free rate in these CAPMish calculations of the ERP?
Calculating the equity risk premium requires choosing a benchmark that represents the risk-free rate of return. As such, the risk-free rate implies zero risk. Of course, in the real world, no such thing exists. Even government bonds are subject to inflation risk. You could, of course, use inflation-indexed Treasuries, but then you’d be exposed to disinflation/deflation risk. Pick your poison.
Aside from that, one could debate what’s the proper maturity to use for government bonds of any stripe. Some say a 10-year Treasury better reflects the risk-free rate; others prefer the short end, the 3-month T-bill, for instance, which we use here. It all depends on the risk you’re focused on.
The bottom line: we choose to use the 3-month T-bill as a proxy for the risk-free rate. The rate of return on cash, in sum, seems to be a pretty good measure of what you can expect from a “risk-free” investment.