'DFA vs. Vanguard' Depends on Asset Class and Tilt in Question

Edward Tower, a professor at Duke University, and his masters student, Cheng-Ying Yang have recently published a new paper: “DFA Versus Vanguard: Has DFA Outperformed Vanguard by Enough to Justify its Advisor Fees?”

Needless to say, this is a subject about which reasonable people can and do disagree, so we read this with great interest. The analysis is based on data from 1999 to 2006. The authors perform their analysis by comparing the aggregate performance of DFA’s funds with the aggregate performance of what they claim to be a comparable portfolio of Vanguard funds. They find that the return of the DFA portfolio, “geometrically compounded, is 8.2% per year higher than Vanguard’s, and that DFA’s standard deviation of return is slightly higher.”

However, they note that, over the period in question, DFA funds (which are heavily weighted toward small and value stocks) “may have outperformed Vanguard because DFA funds were focused on the right styles.” To correct this, they construct a portfolio of Vanguard funds that mimic the asset class and style weights of the aggregate DFA portfolio. On the basis of this comparison, DFA’s compound return differential falls to 2.46% per year over the eight year period studied.

Another analysis used Fama French three factor analysis to construct the mimicking Vanguard portfolio, which reduces the advantage to 1.63% per year. Assuming a 2% per year reversal of the return differentials in favor of small and value tilts reduces the differential to 1.13% per year.

Thus far, the results of the analysis would seem to favor DFA. However, there are two caveats, one noted by the authors, and one not examined by them.

The first caveat is that the authors’ analysis does not take taxes into account. As DFA funds can only be obtained through an adviser to whom an investor must pay a fee (while Vanguard funds can be bought directly with no advisor fees), the level of the advisor’s fee and its tax treatment can have a significant impact on the DFA versus Vanguard comparison, and possibly tilt it in the latter’s favor.

The second caveat is that most investors do not see the problem in the same way as the authors’ analysis. Tower and Yang’s approach implies that investors use the aggregate performance of DFA versus Vanguard portfolios to determine which to use when implementing their asset allocation strategy. We disagree; in our experience, investors make fund by fund comparisons for different asset classes and tilts, and not aggregate comparisons of one family to another (see our November 2004 issue for an example of this type of analysis).

From this perspective, the answer to the question of “DFA or Vanguard” is not so clear cut; in our experience, it depends on the asset class and tilt in question.