Jeremy Siegel, Ph.D. The Future for Investors

A few weeks ago Robert Arnott, chairman of Research Affiliates, caused quite a stir by publishing a paper in the May/June edition of the 'Journal of Indexes' entitled "Bonds: Why Bother". In the paper Arnott concludes:

"For the long-term investor, stocks are supposed to add 5 percent per year over bonds. They don't. Indeed, for 10 years, 20 years, even 40 years, ordinary long-term Treasury bonds have outpaced the broad stock market."

Arnott maintains that the long-term excess return of stocks over government bonds is only 2.5 percent, not 5 percent. Although Arnott doesn't identify the source of the 5 percent premium, he claims that it is widely used by pension plans and other investment advisers.

His conclusions are clearly designed to challenge the thesis of my book, "Stocks for the Long Run", which strongly advocates a diversified portfolio of global stocks as the overwhelming choice for long-term investors.

But a careful look at Arnott's data should not lead investors to bonds. At today's prices, stocks are even more attractive, while bonds are considerably less attractive than their historical returns. Here's why.

The Evidence

Long-term financial data has never given a 5 percent advantage to stocks over bonds. The data that I use in "Stocks for the Long Run" from 1871 to the present are virtually identical to the data Arnott uses. We differ only on the stock returns in the early 19th century, where Arnott finds much lower returns. Unfortunately, the accuracy of this earlier stock data, as even Arnott admits, is highly questionable. The well-documented data from 1871 through 2008 gives stocks a 3.2 percent per year advantage over bonds.

Although not as high as 5 percent, 3.2 percent per year is not trivial. If the long-term real, after-inflation return on stocks is 6.2 percent, and the long-term real return on bonds is 3 percent (both very close to their historical averages), then, over 30 years, a portfolio in stocks, measured in terms of purchasing power, will appreciate to a sum of more than 150 percent above that in bonds. Over 40 years, investors more than triple their money in stocks compared to bonds. (By the way, the arithmetic average return -- rather than compound return -- of stocks over bonds from 1871 through 2008 is 4.6 percentage points. This is the number the finance profession uses in the capital asset pricing model and other analytical applications.)

The Past 40 Years

Arnott additionally claims that there are long periods when stock returns fall behind bonds, and he notes that, over the past 40 years, the returns on government bonds have outpaced "the broad stock market."

But this is not true. At the bottom of the recent bear market, in March 2009, government bonds did indeed outperform the S&P 500 Index over the past 40 years. But bonds never outperformed any index that tracked the "broad stock market" that includes large as well as small stocks.

But bonds have admittedly done extraordinarily well over the past four decades. Over that time, long-term Treasuries have returned nearly 12 percent per year before inflation, and the returns on stocks ending at the bottom of the bear market in March of this year were just a few basis points lower.

But 40 years ago treasury bonds were yielding over 6.3 percent, about twice their yield today. It is mathematically impossible for government bonds to come close to matching those 12 percent returns in future decades. Stocks, on the contrary, can easily repeat their returns over the past four decades, since those returns were near their historical average.

Returns on Stocks and Bonds

Should the fact that government bonds did so well in the recent past encourage investors to invest in bonds today?

Absolutely not. In fact, it is quite likely that bonds will not only underperform stocks in the future but will also give investors negative real returns. And it is clear from examining Arnott's graph on stock and bond returns that, after long periods when stocks have trailed bonds' performance, such as in 1900 and 1941, stocks subsequently strongly outperform bonds.

Furthermore, Arnott does not discuss the extremely long periods of time where bonds have not only underperformed stocks but have given investors negative after-inflation returns. For the 55-year period from December 1925, when the well-known Ibbotson stock and bond series begins, through January 1982, total real government bond returns were negative. This means that, by rolling over in long-term government bonds, reinvesting all the coupons, and thereby taking no income, investors' bond portfolios were sinking in value.

Most strikingly, for the 40-year period from 1941 through 1981, government bond investors lost a whopping 62 percent of their value after inflation. A loss in purchasing power over this long a period has never happened in stocks. There has never even been a 20-year period when real returns in stocks have been negative. In fact, the worst 30-year real return for stocks is plus 2.6 percent per year, just slightly below the average real return investors earn with government bonds.

Looking Forward

Looking at today's markets, the forward-looking prospects for government bonds are very poor. Yields on 30-year inflation-protected bonds are 2.3 percent, and yields are only 4 percent on 30-year Treasuries. In contrast, after stocks have fallen 50 percent from their previous high, as they did in March of this year, their subsequent 30-year real returns have always been in excess of 10 percent per year.

The 40-year outperformance of government bonds over large stocks has ended. Since the March 8 lows, stocks have rallied 30 percent, while government bonds prices have fallen sharply. By April 30, the 40-year stock accumulation in the S&P 500 Index is more than 15 percent ahead of bonds and even more for the broader stock market. Years from now, we will look back at Rob Arnott's article as a turning point that marked the end of both the long bull market in bonds and the bear market in stocks.