Tweet
I'm continuing my short series on asset allocation studies, with a report published by Ibbotson Associates in March 2006. This paper is a more rigorous study than the Christopher Moth paper (my Exhibit I).
Whereas Moth based his conclusions on a single optimization exercise over historical data, Ibbotson combines a similar optimization with three analyses of hypothetical, forward-looking returns for commodities. (That sounds implausible, but you should check the paper to see how they separate commodity investment yield into subcomponents, and model each one as a separate building-block of yield.) Moth's historical dataset is quarterly returns 1973–2000. Ibbotson's is annual returns 1970–2004. Moth's commodities benchmark is GSCI, while Ibbotson uses a blend of four benchmarks, and discusses the construction of each one. Moth was published in 2000, and Ibbotson's in 2006. Clearly in the years between, investors became more interested in alternative asset classes. The Ibbotson paper cites many other researchers who investigated commodities during the early 2000s. Needless to say, I think it is worth a few hours to read and re-read this great report.
Anyone investing in commodity-based ETFs would be wise to learn about the various factors in the design of these vehicles, and the Ibbotson paper is a good starting point.
There are a lot of results in the paper, and I was bewildered at first as to how to incorporate this great research into my own portfolio. Cutting to the chase, the key findings are in table 17 on page 44. This is a blend of the historical and forward-looking analyses. The researchers calculated an asset allocation policy for six investment styles, by keeping the risk (standard deviation) constant and optimizing the policy for the best returns. For example, the conservative portfolio's goal is to have a constant 5% standard deviation in returns, so what is the best expected return we can obtain for this amount of risk?
This is a nice presentation because you can compare side-by-side what happens if you include or exclude commodities as an asset class. For conservative and moderate-risk investors, you get a nice boost in expected returns (with no additional risk), by shifting some of your assets from bonds to commodities. The benefit of commodities is not as great for someone with an aggressive risk profile.
Now let's compare the Ibbotson portfolios to the Moth portfolio. The most striking difference is that Moth's optimization rules out international stocks from the portfolio. He argues that they are too highly correlated with U.S. stocks, and have worse return/risk. However, Ibbotson includes the years 2001–2004 in their study, and therefore do not reach the same conclusion. In fact, in Ibbotson's aggressive portfolio, over half is allocated to international stocks! I take this as a sign that the risky portfolios suggested by these studies are not reliable. Artifacts in the historical data create distortions, which are magnified in the two studies' riskiest portfolios.
In my own portfolio, I have decided to incorporate both studies' moderate portfolio recommendations. What better way to embrase a recommended diversification policy, than to diversify among several? In fact I have taken the average of four different studies, and will write about the last two here in the coming weeks.
Whereas Moth based his conclusions on a single optimization exercise over historical data, Ibbotson combines a similar optimization with three analyses of hypothetical, forward-looking returns for commodities. (That sounds implausible, but you should check the paper to see how they separate commodity investment yield into subcomponents, and model each one as a separate building-block of yield.) Moth's historical dataset is quarterly returns 1973–2000. Ibbotson's is annual returns 1970–2004. Moth's commodities benchmark is GSCI, while Ibbotson uses a blend of four benchmarks, and discusses the construction of each one. Moth was published in 2000, and Ibbotson's in 2006. Clearly in the years between, investors became more interested in alternative asset classes. The Ibbotson paper cites many other researchers who investigated commodities during the early 2000s. Needless to say, I think it is worth a few hours to read and re-read this great report.
Anyone investing in commodity-based ETFs would be wise to learn about the various factors in the design of these vehicles, and the Ibbotson paper is a good starting point.
There are a lot of results in the paper, and I was bewildered at first as to how to incorporate this great research into my own portfolio. Cutting to the chase, the key findings are in table 17 on page 44. This is a blend of the historical and forward-looking analyses. The researchers calculated an asset allocation policy for six investment styles, by keeping the risk (standard deviation) constant and optimizing the policy for the best returns. For example, the conservative portfolio's goal is to have a constant 5% standard deviation in returns, so what is the best expected return we can obtain for this amount of risk?
| Asset Class | Conservative | Moderate | Aggressive | |||
| With Commodities | Without Commodities | With Commodities | Without Commodities | With Commodities | Without Commodities | |
| Treasury Bills | 22.77% | 27.34% | 1.17% | 3.89% | 1.07% | 3.68% |
| TIPS | 10.24% | 17.64% | 7.21% | 17.80% | 2.29% | 5.93% |
| U.S. Bonds | 18.93% | 15.74% | 5.53% | 8.31% | 0.29% | 1.43% |
| International Bonds | 16.54% | 21.07% | 7.80% | 14.62% | 2.20% | 5.52% |
| U.S. Stocks | 15.48% | 9.69% | 30.65% | 36.79% | 18.13% | 39.57% |
| International Stocks | 6.11% | 8.52% | 22.05% | 18.58% | 57.05% | 43.88% |
| Commodities | 9.93% | 0.00% | 25.59% | 0.00% | 18.97% | 0.00% |
| Expected Return | 6.62% | 6.19% | 9.48% | 8.83% | 11.26% | 10.88% |
| Standard Deviation | 5.00% | 5.00% | 10.00% | 10.00% | 15.00% | 15.00% |
| Sharpe Ratio | 0.400 | 0.314 | 0.486 | 0.421 | 0.442 | 0.418 |
This is a nice presentation because you can compare side-by-side what happens if you include or exclude commodities as an asset class. For conservative and moderate-risk investors, you get a nice boost in expected returns (with no additional risk), by shifting some of your assets from bonds to commodities. The benefit of commodities is not as great for someone with an aggressive risk profile.
Now let's compare the Ibbotson portfolios to the Moth portfolio. The most striking difference is that Moth's optimization rules out international stocks from the portfolio. He argues that they are too highly correlated with U.S. stocks, and have worse return/risk. However, Ibbotson includes the years 2001–2004 in their study, and therefore do not reach the same conclusion. In fact, in Ibbotson's aggressive portfolio, over half is allocated to international stocks! I take this as a sign that the risky portfolios suggested by these studies are not reliable. Artifacts in the historical data create distortions, which are magnified in the two studies' riskiest portfolios.
In my own portfolio, I have decided to incorporate both studies' moderate portfolio recommendations. What better way to embrase a recommended diversification policy, than to diversify among several? In fact I have taken the average of four different studies, and will write about the last two here in the coming weeks.
Comment