Naturally, a disagreement between two high profile respected researchers makes for good press and lively debate that can also be highly instructive. One such debate took place in Toronto in 2004 at the conference of the National Association of Personal Financial Advisors, the transcript for which Prof Bodie posted on his website. Bodie also makes his case in the 1994 paper On the Risk of Stocks in the Long Run, available from SSRN and in his mass-audience 2007 book Worry-Free Investing.
The debate is about holding stocks in general, a whole diversified basket of stocks, such as the S&P 500 in the USA, or the TSX Composite in Canada (such as many ETFs and mutual funds allow investors to do), not about holding individual stocks, which are obviously forever exposed to financial catastrophe and bankruptcy. There may be a few casualties amongst companies within a broad index but as a whole there is little question of its survival, at least in countries like Canada and the USA.
Stocks are risky arguments
- The "conventional wisdom that if you hold stocks long enough they are bound to outperform all other asset classes" is wrong! Bodie proves this theoretically in the SSRN paper with an option-pricing example. In the same paper he also cites research by his mentor Samuelson and others like Robert Merton, another Nobel winner, that rely on expected utility maximization to analytically prove why such statements are wrong.
- The idea is false that eventually, after many years holding stocks, there will be a positive and high return, more or less in line with long term averages (e.g. those in the Credit Suisse Global Investment Returns Yearbook 2012 which showed annual real returns in Canada over the years 1900 to 2011 of 5.7% for equities vs 2.2% for government bonds and 1.7% for T-bills). With the aid of Monte Carlo simulation, Bodie calculates (Chapter 6, Worry-Free Investing) that due to the volatility of stocks, though the probability is low of such a bad outcome, after even 30 years of holdings stocks, their value could be half in real terms what it was at the start. In fact, he says, though the probability keeps going down, the really bad outcomes just get worse and worse. Meanwhile a risk-free security like a T-Bill wouldn't make much money compared to big majority good equity outcomes, but it would not lose money.
- Even in the historical record, highly successful countries like the USA and Canada have been the outliers as the Credit Suisse Yearbook shows. There is always the chance that the floundering of the Japanese stock market for the last two decades may be the direction we are headed too - reversion to a lower mean.
The essence of these counter-arguments relies on the historical record.
- "It is very significant that stocks, in contrast to bonds or bills, have never delivered to investors a negative real return over periods of 17 years or more" (using data from 1802 to 2006). Meanwhile both bonds and T-Bills have had real negative returns, even over periods as long as 30 years. Inflation takes its toll on low returns. Siegel did not even apply any estimate of taxes that would push bond and T-Bills further into negative territory.
- Siegel does concede in the debate transcript that such past performance does not mean he believes long-held stocks are absolutely safe or offer a guaranteed good return or the best return. In other words the future may not be like the past. That statement (in 2004) seems to have been a wise position since the Credit Suisse charts for Canada show that in the 27 years from 1984 to 2011, bonds outperformed equities by 1.5% per year.
- Valuation at the time of purchase of stocks has been shown to make an enormous difference in subsequent returns. It is obvious in retrospect that buying at a market low such as in 1932 or 1981 produced tremendous gains - see details in Evanson Asset Management's Stocks for the Long Run?. The question is whether any indicators can successfully predict when is a good time to buy, when the market is cheaply valued. The answer seems to be yes, valuation indicators such as dividend yield, trailing P/E ratio, Q-ratio do give worthwhile over- or under-valuation signals. Read Evanson for some examples. A blog called Laguna Beach Bikini (quite a name for a site with some substantial investing content but hey, that's California!) in Recent Ideas in Modern Finance and How Expected Return Varies delves into a paper on the topic which emphasizes the importance of dividend yields - a low current dividend yield presages low returns, high dividend yield foretells of higher future returns. When valuations are low, there are much better chances of making higher returns with equities.
- Simply putting all your money into equities on the expectation of higher returns isn't a wise plan - the long run for outperformance may be very long and who knows, you may need to cash out before you plan to, or the severe market gyrations may cause a panic reaction. Diversification between bonds and equities makes sense.
- The percentage allocation between equities and bonds (or other asset classes which we have not discussed today) should depend more on your goals and risk capacity not on the length of your time horizon - see our previous posts on Setting Investment Objectives, Risk: What Can You Afford and What Can You Put Up With? and Asset Allocation: the Most Important Investing Decision You Will Make. Bodie and Siegel seem to agree on this aspect of the issue at least!