Predicting Asset Class Returns: Recommendations for Financial Planners

Developing reasonable estimates for stock and bond returns requires more than just historical data or the assumptions provided in financial software packages. Inappropriate assumptions can doom retirees to outliving their savings or forgoing a life style they could otherwise afford. There are better ways to forecast, and in this article I’ll suggest a few of them.

Larry Siegel addressed the issue of assumptions in a February 2012 article, "Jeremy Siegel, Rob Arnott and Other Experts Forecast Equity Returns." He reported on a 2011 gathering of experts, whose consensus was that equity returns over the next decade would be significantly lower than historical returns. In an August 2012 article, "The Profession's Faulty Assumptions: A Top Ten List," Bob Veres ranked "assumed U.S. equity returns" as the number one among "garbage-in" assumptions being used in financial planning.

I also looked into the assumptions used by a couple of the most popular financial planning packages. Both stock and bond returns were well above what I consider reasonable.

Historical view

Beginning in the 1970s, Roger Ibbotson and colleagues began publishing data on historical returns (going back to 1926) for various asset classes, research that evolved into the annual "Stocks, Bonds, Bills and Inflation Yearbook." Using historical returns from the Ibbotson Associates data became the most popular approach for estimating future returns. But, as time has passed, Roger Ibbotson and other researchers increasingly recognized that history provides only a partial guide to the future.

This chart shows compound annual returns for the 1926-2012 period:

Large company stocks


Intermediate-term government bonds


Historical Equity Risk Premium (ERP)


Source: 1926-2011 Ibbotson® Yearbook, 2012 Author's estimate

The historical difference between stock returns and government bond returns of 4.4% is slightly higher than the forward-looking estimates from the panel of experts cited by Larry Siegel, which mostly fell in the 3% to 4% range. The big difference, however, between history and the forward-looking outlook is in the bond category. The return one can expect from investing in bonds today is heavily influenced by current yields, and the yield on 10-year Treasury bonds as of year-end 2012 was 1.74%, a full 3.7% below the Ibbotson historical average.

This gap between historical and current bond returns raises questions. When predicting bond returns, how much weight should be given to historical returns as opposed to current yields? To estimate future stock returns, should the estimated ERP be added to the current government bond yield or to a future estimate of yields? I'll deal with these questions, among others, by looking at bonds and stocks separately.