Sunday, January 20, 2013

The alpha dog don't hunt (maybe)

As a strong proponent of the semi-strong version of the Efficient Market Hypothesis (EMH), I have a bone to pick with the idea of alpha generation. First some terms explained.

Money managers such as mutual fund managers and hedge fund managers operate under the following premise: The attempt to grow investors' money over time. Some managers attempt to outperform the market (active managers) while others just attempt to replicate or stay with the market (passive managers). For this discussion we are only concerned with the investments of active managers. Whether successful or not (and success is a complicated determination as we will see), the investments will have some kind of results. There are two components of those results, alpha and beta.

Answering the last part first, beta is the component of a money manager's results (performance) that is attributable to or describable by the underlying market's results. That is to say someone who simply invests in the stock market will have results just from being invested in the stock market--regardless of the individual decisions that investor makes. A manager or investor who exactly mimicked the results* of the stock market would have a stock market beta equal to 1. One who exactly doubled the stock market's results* would have a beta equal to 2. Keep in mind that these figures apply up and down such that a portfolio with a beta of 2 would expect to have twice the up or down results of the underlying market. Comparing a manager with a beta of 1 to a manager with a beta of 2, we could say that the manger with beta equals 2 is expected to be twice as risky relative to the other manager considering the same underlying market for both.

Alpha on the other hand is the component of a money manager's results (performance) that is in excess or deficiency of the underlying market's results given the risk the manager took on. That is to say someone who simply invests in the stock market will have results just from the individual decisions that investor makes--regardless of the results from just being invested in the stock market. A manager or investor who exactly mimicked the results of the stock market would have a stock market alpha equal to 0% at any given level of beta.

The problems of moving from these theoretical constructs to the real world begin with the fact that if the market is really, really good at identifying, processing, and applying information (that is to say the market is highly efficient), then alpha shouldn't be possible (market participants shouldn't be able to out think the market). Our problems continue once we realize that it is very difficult to look at a money manager's results and separate out alpha and beta. The two get quite confoundedly interwoven for a number of reasons.

For one, when do we stop defining down the proper market comparison (benchmark) for a manager? Suppose he purchased 499 of the 500 stocks in the S&P 500 index in proportion to their weight in that index less the left out stock. Seems reasonable that the S&P 500 would be the proper benchmark. But the stock that gets excluded will make a big difference in how representative the S&P 500 is for his performance. Excluding Apple (currently ~4% of the index) versus excluding AutoNation (currently ~.01% of the index or 1/400 the value of Apple in the index) would have a significantly different effect. Suppose another manager purchased the same 499 stocks as the first but in quite different proportions. Suppose another only purchased 10 of the stocks. Describing all of these managers or any of them as compared to the  S&P 500 might not be a very meaningful description.

Another reason alpha and beta are difficult to separate is that manager process and performance is not always very transparent. This makes finding suitable benchmarks more difficult. We could go on, but the foundation of the problem is already well established. A manager's alpha might just be the beta of a better defined benchmark.

So what to make of the elusive search for alpha? My thought is that the only true alpha is good beta management--perhaps "meta beta"would be the clever term for it. The goal then for a money manager is to achieve a beta appropriate for the investor and opportunistic when and if possible. A beta of .5 would be great in relative down markets but would have an obvious cost of underperformance generally. A beta of 2.5 would be great most of the time but would have the extreme risk of permanently impairing capital some of the time. Opportunistic beta management is dangerously close to market timing, which almost certainly is a fools errand. Yet there may be some ability to capture gains once we consider the position of the individual investor we are attempting to match up with an appropriate beta (e.g., relatively high beta for an investor with no withdrawal needs, a high capital base, a long time horizon, and otherwise high willingness and ability to take on risk).

I look at this, my understanding of and belief (or disbelief) in alpha, as a work in process. I have to reconcile any degree of belief in alpha with any degree of belief in efficient markets. Starting with my assumptions and belief in efficient markets does have a flavor of receiving the answers in advance of asking the questions, but I'll try to keep an open mind. I'll keep thinking . . .



*Mimicking results, doubling results, etc. with regard to beta means both the end result (return) and the path to get there (volatility).