alpha myth

Bad Benchmarking and the Creation of False Manager Alpha

alpha myth

You don’t have to dig too deeply into IFA’s Website to realize that in our opinion, the pursuit of alpha is a fool’s errand that suckers investors into paying the high costs of active management while potentially causing them to miss out on the real source of returns, exposure to the compensated risk factors of the market (beta). When an active manager claims to have delivered alpha, the first question an investor should ask is, “alpha relative to which benchmark?” Usually, the benchmark is the one chosen by the fund manager, and therein lies a serious problem for would be manager pickers. Quite often, the manager-selected benchmark does not truly reflect the asset allocation and hence the risk exposure of the fund.

Using Morningstar Direct, we identified 103 funds representing $183.8 billion of assets where the primary prospectus (manager-selected) benchmark was the S&P 500 Index but the Morningstar Analyst-Assigned Benchmark was one of the following:

1)      Russell 1,000 Value Index (86 funds)

2)      Russell Mid-Cap Value Index (13 funds)

3)      Russell 2,000 Value Index (4 funds)

 

The table below shows how benchmark selection of the S&P 500 Index increases the reported alpha for these funds over a 15-year period.

Of the 103 funds, 44 had at least 15 years of returns data. The chart below shows the average alpha of these funds relative to the S&P 500 and relative to the Morningstar Analyst-Assigned benchmark.   

To see how this works for a single fund, please see the chart below for American Funds Washington Mutual A (AWSHX), which Morningstar considers to be a large cap value fund and assigns the Russell 1,000 Value Index as its benchmark. AWSHX also happens to be the largest of the 103 funds with $40.8 billion of average net assets.

As you can see, the alpha disappears once a proper benchmark is used. The moral of the story is that the “alpha” relative to the S&P 500 was actually beta in the form of increased size and value risk. The same risk exposure could have been achieved with index funds. One further test we ran of the 44 funds was a three factor regression, which provides a statistical analysis of alpha after accounting for exposure to the risk factors of market, size, and value. We found that not one of the 44 funds had significant alpha (a t-stat greater than 2) such that we could be at least 95% certain that luck was not the explanation.

The types of bad benchmarking we have seen are by no means limited to the example above. They include global funds that are benchmarked to purely domestic indexes as well as balanced funds that have a larger allocation to equities than their benchmark would suggest. Lately, we have even seen fixed income funds with a substantial allocation to equities.

So the next time somebody tells you that they have found a manager who delivered alpha, be sure to ask if it was truly alpha or beta disguised as alpha. The next question to ask is, “Was it statistically significant?”