Robert T wrote:.
Devout Indexer,
There's a difference between comparing performane of portfolios with very similar factors loads across US, Intl, EM, & Fixed Income allocations, and portfolios with similar long-term 'expected returns' but different factor loads across US, Intl, EM & Fixed Income. The short-term performance comparions of the former will be more similar than the latter (as I inidicated in that earlier thread - "There will likely be annual 'tracking' error between the two portfolios given the differences in their size and value exposure"). So you are answering a different question to the one posed in the OP. FWIW since inception of the two portfolios in November 2008 - the repective returns have been ETF=13.00, DFA Vector/Core=13.68, so a 0.68 percent difference - so even if the question related to 'expected return' rather than 'factor loads' my answer would still be no so far.
Just two points on expected returns: (i) There is a difference in the accuracy in portfolio contruction between using actual estimated factor loads and assumed factor loads. The Vector/Core portfolio used assumed factor loads. Personally I have more faith in the accuracy of comparisons using estimated factor loads. And you had suggested in some of your earlier posts that the value load for vector may have been closer to 0.45 suggesting higher expected portfolio returns, and (ii) the Vector/Core portfolio also assumes the EM size and value premiums are the same as the US and Intl markets. As you had indicated in an earlier thread, these premiums are likely higher in EM leading to a higher expected return of the Vector/Core portfolio, so perhaps another weak assumption in the contruction of the portfolio (given these two factors, the standard errors in the factor loads, factor premiums, and expected returns of the Vector/Core portfolio are likely higher). Fair assessment?
The most accurate comparison (acutal estimated factor loads using the same time period estimates, very similar size and value loads individually in US, Intl, and EM allocations, and similar expected tax efficiency) is between the ETF and DFA TM portfolios re: my answer to the question so far.
Robert
.
Robert,
In the 3 years since 12/08 (latest update on factors), HmL has been negative in every region of the world, so that wouldn't have meant anything. Further, despite your fear of a mismatched factor allocation between US Vector and the US ETF mix, I show 3 year #s (size/value) of 0.56/0.28 for the ETF and 0.57/0.26 for Vector--as close as close gets.
Seemingly, the only standout difference between the Vector/ETF mix is a duration mismatch between bond strategies (3-7yr t-note has greater term risk than 5YR Global), which certainly helped the ETF mix given that TERM was +9.3% per year over this period. Surprising, actually, that the DFA mix was ahead by ~1-2% depending on the time period given the 'term tailwind'.
To correct for this, I swapped 5YR Global for Int'd Govt in the Vector mix, and then compared this allocation rebalanced monthly to the ETF mix you came up with (using live data for BRSIX and IEI and index returns for MV, SV, ILV, ISC, and EM for ease of access) for the 3 years through November as I wait for Dec data. I came up with +15.6% for the Vector mix and +14.0% for the index data (probably 0.2% or so less net of ETF fees). So that 1-2% seems to hold over this period.
Both good portfolios, of course.