Posted by: Aaron Pressman on February 21, 2007
Mebane Faber, a money manager at Cambria Investment Management in Los Angeles, has been doing some interesting thought experiments on his blog about how ordinary investors can try to replicate the returns of Harvard University’s amazing endowment. The goal is to create a simple portfolio from ETFs or index mutual funds that captures as much of the diversification benefits and returns of the endowment’s fund without taking on too much risk.
Back in November, Faber compared Harvard’s returns to a simple equal-weighted mix of five indexes: the S&P 500, the MSCI EAFE Index, 10-year US Treasuries, the National Association of Real Estate Investment Trusts Index, and the Goldman Sachs Commodity Index. Since there are no obvious proxies for hedge funds and private equity that ordinary investors can buy, he left them out. Then he compared returns for the period 1983 through 2004 after deducting 0.4% a year for hypothetical management fees from the all-index portfolio. The results weren’t too surprising. Harvard gained 15.7% a year, the S&P rose 14.1% annually and the five-way mix returned 12.4%. The index mix had the lowest volatility, however, with a standard deviation of 9.7% a year versus 11.9% for Harvard and 16.4% for the S&P.
It got a little more interesting when Faber added leverage to the index-based portfolio, reckoning that the hedge and private equity funds that Harvard uses are also leveraged. With 50% of borrowed funds added to the mix, the all-index portfolio gained 15.3% with a standard deviation of 14.5%.
This week, Faber took another look comparing Harvard and Yale’s asset allocations and performance to an all-index mix that tried to time markets based on a simple moving average strategy. In this test, if the index of any of the five asset classes dropped below its 10-month moving average, the portfolio moved that portion of its assets to cash until the index crossed back above the 10-month moving average. The timing and indexing strategy had a slightly lower return than the previous equal-weighting index play of 11.7% a year but its standard deviation dropped to 6.8% and it never had a down year. A double-leveraged version of the index and timing strategy performed best of all, gaining 18.1% a year but with a standard deviation of 14%. More info about the timing and indexing strategy can be found in Faber’s paper here.