April 21, 2009
Our analysis is admittedly crude, and Arnott’s data makes some imperfect assumptions – using 20-year Treasury bonds as the risk-free rate, for example. More appropriately, the risk-free rate should be represented by T-bills or by stripped-coupon Treasury bonds matching the time period measured (e.g., 20-year stripped coupon bonds when measuring 20-year equity returns).
Data imperfections notwithstanding, are long-term investors, with 40-year time horizons, justified in expecting the generous 6.27% premium, or is a more conservative 2.57% premium realistic?
Part of the answer lies in using 40-year averages over an 82-year period (from 1926 to 2008), which over-weight the time periods in the middle, and under-weight those at the ends. Those end periods include some of the worst equity performance on record – equity performed abysmally during the Great Depression and is down precipitously again today. Thus, as the time periods shrink from 40 years to one in the above table, the worst-performing decades are given progressively more weight.
Studies attempting to measure the equity risk premium can be found throughout the academic literature, and a consensus value has yet to emerge. We will not settle that debate here. Our view is consistent with Arnott’s, which is that a 2.5% premium is a reasonable expectation, and that this incremental return, compounded annually over long periods of time, adds significantly to one’s wealth.
Our only point of contention is Arnott’s claim that a 5% equity risk premium is a commonplace belief – or as he puts it, a “birthright” – since we know of no evidence to support the view that this is a widespread idea.
Bond index construction
Arnott takes aim at bond index funds, warning that “conventional bond indexes will load up on the most aggressive borrowers’ bonds” and will overweight those issuers with the highest outstanding debt levels. We wholeheartedly agree with this insight and have seen research at least as far back as 2004 making exactly the same point.
For most bond investors, though, this insight is the least of their concerns.
In 2008, the Barclay’s AGG index outperformed 92% of actively managed taxable bond funds and, more importantly, outperformed 98% of the assets in those funds. Last year was not an anomaly. Bond indices consistently outperform an overwhelming majority of actively managed funds, year after year, across many sectors.
Bond fund performance is closely (and inversely) tied to expenses, explaining why low-cost funds, like AGG with an expense ratio of .20, are able to achieve superior performance.
We are highly skeptical that a fundamentally-weighted version of the AGG would outperform the AGG after expenses, and Arnott’s data seems to validate our skepticism. His data shows that investment-grade fundamentally-weighted indices performed almost identically to the AGG.
Arnott’s results for high-yield and emerging market bonds show some support for the superiority of fundamentally-weighted indices. Whether fundamentally-weighted indices in these asset classes can be translated into low-cost products that outperform existing products remains to be seen.
If there is a problem in bond investing, it is among the actively managed funds, not in the way bond indexes are constructed.
Display article as PDF for printing.
Would you like to send this article to a friend?
Remember, if you have a question or comment, send it to .
