EQUITIES just aren't what they used to be. It was once gospel that equities out-perform low-risk bonds. But if you invested in the stock market around 1999 the balance of your portfolio probably suggests otherwise. During the post-war era equity returns have been positive. Enough so that the equity-risk premium, the return equities generate in excess of the risk-free rate (which is normally short-term Treasuries), is often assumed to be between 5% to 8%.
Now, Aleph Blog and others are probably much more suited to discuss the finer points of the Equity Risk Premium, other than to say David Merkel usually uses longer term bonds in order to decide whether there is a risk premium for debt versus equity in general. A.S. here refers to the risk free rate, generally 3 month treasuries.
The first thing that strikes me is this theoretical person who dumped a boatload of money into the stock market in 1999, held it for 11 years, and is thus our layman's proof of the lack of an equity risk premium. Theoretical set and forget investors like this are wonderful rhetorical points, but would have most likely sold out in 2003 and not jumped back into the market until, say 2006. The classic "buy high/sell low" investor. But individual investor irrationality and inexperience doesn't prove or disprove the equity risk premium. Additionally, the problem with this type of data point, like Felix's advice and belief in zero equity risk premium, is that the way it is said begs interpretation from the lay person to stay away from equities entirely. Equities are risky, there is no equity risk premium, and oh look if I had just kept my money in cash I would be doing so much better than I would be doing now. Investing in hindsight is a wonderful way to lose sleep and otherwise confuse your strategies for the future planning.
Moreover, is A.S. taking into account just the price level of the index, or also including dividends? A $10,000 investment in the S&P 500 would have, with dividends reinvested, yielded $11,656 by April 1st 2010. A similar investment in T-bills would have grown to $13,674.
So, if we are going to assume an investor, why does A.S. assume a fully invested, maximum risk taking investor, taking no dividends? Well, convenience, surely. I spent most of the morning looking for monthly returns for the Lehman Aggregate Bond Index fruitlessly, and while Robert Shiller and Kenneth French keeps their data nicely updated, William Sharpe does not.
So, putting aside the likelihood that "real" investors are more likely to be bad trend followers, get nervous and try to market time, and otherwise engage in a whole host of poor investing decisions, what does a "real" investors portfolio look like?
First, lets assume that this investor is investing on a regular basis, through something like a 401k. Thus, a newly minted grad in 1999 (which A.S. may have been, who knows with the Economist's anonymous blogger policy) gets their first job in Jan 1 1990, and puts aside $1000 per month into their company sponsored Defined Contribution Plan. Next, let us also assume that this investor does not invest entirely in stocks, but also elects to invest some money in bonds (I used VBMFX as a proxy for the bond market, see above for the dearth of monthly LABI returns online). What do the different portfolios look like for an investor who is "in the market" versus one who puts their money into the three month t-bill each month?
From January 1, 1999, an investor who invested regularly on the first of month, every month, invested 100% in equities (represented here by the S&P 500), would have invested, over time, $136,000 and on April 1, 2010 have $155,864.79. This is a simple return of 14.61%, and a yearly geometric return of 1.37%.
For the same period, an investor who invested regularly on the first of month, every month, invested 100% in 3 month treasuries, would have invested, over time, $136,000 and on April 1, 2010 have $155,301. This is a simple return of 14.19%, and a yearly geometric return of 2.81%.
Geometric returns doesn't take into account the amount invested during the time period. If we look at the Dollar Weighted Returns (using IRR), we see the that the Market portfolio yields 2.51% versus a 2.45%.
If we add bonds into the portfolio, the returns of the portfolio during the period increase even more (3.98% dollar weighted average for the time period for a 60/40 stock/bond portfolio).
All of this is to say that the academic question of the risk premium may or may not be related to the way that portfolios are (or should) be constructed in real life, and that even cherry picking the sample size of recent data can mask the differences in the way that returns are quoted. Obviously how well each investor slept with each portfolio is different, but that is a different post.
Alternately, past performance is no guarantee of future returns.
But seriously, if anybody has monthly bond index returns going back as far as Shiller's data, please email me.