Does “Sentiment Beta” beget “Sentimental Alpha”?

CAPM / Alpha Theory 14 May 2008

Pensions & Investments published a special report earlier this week on the increasingly important role played by academics in today’s world of investment management.  It contains a series of articles on the plethora of professors who augment their modest academic salaries with (lucrative) consulting gigs for asset managers.  One of the articles in the report that caught our eye was about Malcolm Baker of Harvard and Jeffrey Wurgler of NYU.  The duo has been writing about behavioral finance for several years.

Behavioral finance has often been touted as the successor to the CAPM since it aims to explain how the grand old model doesn’t hold up under empirical analysis.  Unfortunately, behavioral finance has so far lacked a unifying theory of its own capable of galvanizing the field of finance.  Still, Baker and Wurgler borrow from the lexicon of the CAPM to propose a measure they call “Sentiment Beta”.

As P&I points out, some big names have taken notice.  Bruce Jacobs of Jacobs, Levy tells the newspaper:

“This type of work is important especially in today’s markets, which has been characterized by wave after wave of investor sentiment — the tech bubble, the bursting of the tech bubble, the housing bubble, the bursting of the housing bubble, the credit bubble and the bursting of the credit bubble…

At first blush, “sentiment beta” sounds kind of redundant.  After all, doesn’t beta itself capture market sentiment?  If sentiment rises, the market rises.  And if the market rises, high-beta stocks rise anyway, right?

In fact, the relationship between high volatility stocks and investor sentiment is almost linear.  The higher the stocks volatility, the more its price becomes driven by “sentiment” (i.e. the higher the “sentiment beta”).  The chart below divides stocks into buckets depending on their “speculative” nature with the most highly speculative bucket on the right.

But when it comes to future returns, things aren’t quite as intuitive.  As the duo say in their 2007 paper on the topic, periods of high investor sentiment actually foreshadow lower future returns – a finding sure to warm the cockles of a contrarian investor’s heart.

The authors point out the irony in this finding:

“…the fact that riskier stocks (at least, stocks that are riskier by all outward appearances) sometimes have lower expected returns is inconsistent with classical asset pricing in which investors bear risk because they are compensated by higher expected return.”

As above, they divided stocks into deciles ranging from what amount to value stocks (“safe, easy to arbitrage”) to what amount to growth stocks (“speculative, hard to arbitrage”).  As the following chart from the paper shows, speculative stocks (on the right) had a much lower excess return in the month after one with high investor sentiment.

Not surprisingly “safe, easy to arbitrage” stocks reacted to market sentiment in the opposite manner.  They were more likely to outperform the market in a month after one with low investor sentiment (as, for example, sentiment may have crested and some sort of mean reversion might be at play).

Behavioral finance is often held out as a way to analyse (or at least temper the analysis of) individual stocks.  But how does this relate to broader issues of portfolio construction and manager selection (our interest on this website)?  As Baker and Wurgler warn us, sentiment beta is very much intertwined with traditional market beta:

“…the standard methodology for estimating fundamental market betas, an input to long-term capital budgeting and other important financial decisions, does not account for sentiment; doing so might improve estimates and clarify their interpretation…”

This is important to us because any time you modify or extend the definition of beta, you get a new alpha – in this case, a sentimental alpha.

Be Sociable, Share!

One Comment

Leave A Reply

← Report: "Exposure yardsticks may provide little insight about a fund's alpha potential" False Discoveries in Mutual Fund Performance: Measuring Luck in Estimated Alphas →