Saturday, May 10, 2008

Wal-Mart 401(k) Suit - Benchmark Bias

A Missouri employee has filed an ERISA fiduciary breach suit against Walmart (Braden vs. Wal-Mart Stores, Inc) for failing to help its retirement plan participants "Save Money Live Better".

The class action suit charges that retail share class mutual funds in the plan were imprudent and unnecessarily expensive since the plan could have qualified for institutional funds and that Merrill Lynch, the Plans Trustee, received undisclosed kickbacks in the form of 12b-1 revenue sharing payments. The issues of fee propriety and disclosure for servicing 401(k) Plans are being addressed by legislators and industry regulators. Whether general practice, as reflected in this case, has been imprudent in retrospect remains an issue for the courts to decide.

The suit also argues that most of the plans investment funds were actively managed funds which cost more and delivered less than either Vanguard index alternatives or a selected sample of low cost Vanguard actively managed funds. Both Vanguard's index funds(slected by fund category)and individually specified Vanguard actively managed funds were used in the suit as performance benchmarks by which to establish the range of participant losses.

Biased benchmarking is undoubtedly one of the largest sources of perceived or marketed "alpha" in the investment industry. We wondered if biased benchmarking might account for some of the alleged participant losses claimed in this case.

We created a composite portfolio using only the actively managed equity funds outlined in the suit(WMT Actual Funds). We weighted the funds based on their proportional assets at 12/31/2007 and assumed all portfolios were rebalanced annually. We were not interested in trying to replicate total portfolio returns using actual cashflows and other funds but simply wanted to compare the Plans actively managed equity fund returns to a similar weighted portfolio of independently selected benchmark indices (Index Benchmark)and to a similarly weighted portfolio using the specific Vanguard funds identified in the suit.

Historical returns for each of these portfolios are presented below:
Portfolio1Yr3 Yr6 Yr10Yr
WMT Actual Funds-3.3% 8.3%7.2%6.2%
Vanguard Alternative Funds-5.3%8.3%8.3%10.9%
Index Benchmarks-5.4%7.7%6.7%4.8%
Vanguard Composites-4.5%8.2%7.3%6.9%

WMT Funds Outperform Index
While the suit did not specify which specific Vanguard index funds were used as investment benchmarks (there were some categorical discrepancies), a portfolio using independently selected benchmark indices(shown below)for each fund, underperfomed both WMT's actual funds and the alternative Vanguard funds for the periods noted.

Vanguard Funds Outperform WMT Funds
This exercise supports the contention that the specified Vanguard active fund portfolio outperformed WMT's actual funds. However, fund selection bias (having the advantage of perfect hindsight in selecting Vanguards best past performing funds)and some significant style bias (categorical value bias across ½ the Vanguard Funds in comparison to the WMT funds)could account for some of the Vanguard outperformance.

To quantify this we compared the WMT funds to Vanguard funds using strictly identical style categories, to the extent possible (Vanguard does not have a foreign large growth fund). We noted a distinct value bias in the Vanguard active funds. Over the last 6 year performance period in the suit, value has outperfromed growth by a wide margin. In addition, in categories where Vanguard offers multiple actively managed funds, we used an equally weighted composite return of all those funds.

Selection & Style Bias Account for Vanguard Outperformance
These adjustments, reflected in the Vanguard Composite retruns, were intended to correct for style and lookback bias. Based on the results it looks like these biases account for virtually all of the selected Vanguard portfolio's excess returns over WMT's actual fund portfolio.

While there are obvious factual and interpretational limitations to this brief analytic, the point is that the calculation of relative investment gains or losses is unique to the yardstick or benchmark by which they are measured. Benchmarking bias, either intentional or not, can create negative alpha just as easily as it can positive alpha. Plan fiduciaries should not be too dogmatic in their benchmarking. Broadly evaluating absolute and relative benchmarks will provide additional insight into performance perceptions and fidcuiary exposures.

WMT FundsVanguard FundsVanguard Composite FundsIndex Benchmarks
AIM Intl Grth FLGVngd Intl Grth FLBVngd Intl Grth FLBMSCI EAFE Grth(net)
Amer Fnds EurpoPac FLBIntl Val FLVIntl Grth FLBMSCI EAFE
Ariel MCBStrategic Eq MCBStrategic Eq MCBR2500V
Davis NYV LCBWindsor LCVG&I & Div.IncomeR1000
Frkln SMID MCGCapital OppCap Opp & MC Grth R2500G
MFS Grth LCGPrimeCap LCGPrimeCap&