The following is in response to Larry Siegel’s article, Jeremy Siegel, Rob Arnott and Other Experts Forecast Equity Returns, which was published last week:
Mr. Siegel:
Good article. There were a couple of areas for which I need clarification. What role do savings and asset allocation, in aggregate, play in determining the realized ERP? For example, if adults saved on average 10% of their earnings from 1900 to 1980, but now save on average 5% of their earnings, how does this affect the ERP? If interest rates normalize, because of government spending and realized inflation, how does this affect the ERP? What is the breakeven time period for achieving a given ERP? For many investors, time is relevant, and the sequence of returns is important. How does this fit into your discussions?
One last point. Academics are generally extremely overconfident in their perceived knowledge of the economy and financial markets. This impacts their discussions of concepts such as ERP, and often misleads financial advisors. It’s not what you know that hurts you in investing, it’s what you think you know that just ain’t so.
Mark Stuart
Stuart Investments
Scottsdale, AZ
Larry Siegel responds:
The realized ERP is the outcome of a large number of forces, the supply and demand for capital being only one. Among the others are the growth rate of the economy, the profitability of corporations, the amount of risk in the market, and investor sentiment.
Remember that the ERP is not a total return, but the difference in two rates of return. All other things equal, if households save more, there is probably no net effect on the ERP since they buy both more stocks (raising the past ERP and lowering the future one) and more bonds (doing exactly the opposite). However, there is some reason to think that as households become wealthier, i.e. save more, they are less risk averse (more risk tolerant) so they buy more stocks. Thus a higher savings rate might be associated with a higher past ERP and a lower future one. The recent dearth of savings, therefore, may have led to lower equity prices (two crashes in 6 years) and a higher forecast ERP.
In regard to your next question, when interest rates go up to historically average levels, our experience tells us that this would be temporarily bad for equities, lowering the historical or past ERP and raising the future ERP. There is, however, no “good” theory that forecasts this. The ERP should be the same in a high interest rate environment as in a low one. (By a “good” theory I mean one that is not an obvious retrofit.)
The ERP is a statistical concept called “expected value,” and not a time period-specific forecast, so a breakeven period cannot be calculated for it. If the ERP is 4%, then in any given year the realization could be +20% or -20% and we would not say that the forecast was wrong, only that the actual return deviated materially from the expected return. Over long enough periods, however, forecasts should be in some sense right, or they’re not worth making. My paper with Grinold and Kroner assumed that the forecast of the ERP we were making was for 10 years. However, we would be very surprised if the actual 10-year realized ERP fit the forecast tightly. It is very difficult to say what is a long enough period to evaluate a forecast like this, but if after 30 years the realized ERP was nowhere near 4%, on either side, I hope I’d have the modesty to admit that my forecast had been wrong.
