Although many hedge fund strategies have relatively low market correlations, it can be notoriously difficult for them to avoid getting sucked into a market downdraft. Last year’s negative performance across several strategies is proof. One strategy that has traditionally had a low correlation during normal times, but a propensity to sink during market storms is Merger Arbitrage. As we recently discovered, however, this asymmetric quality seems to be waning.

We were prompted by a new paper from California’s Research Affiliates that concludes merger arbitrage produces returns that are very similar to a short put position on the S&P 500. This relationship was first identified in this 2002 Journal of Finance article by Mark Mitchell Todd Pulvino and was further explored by hedge fund researchers Bill Fung and David Hsieh a couple of years later.

Mitchell and Pulvino, for example, found that returns from a passive merger arb strategy had a high market beta when the S&P500 monthly return was below a certain threshold and a much lower beta when it was above that threshold.

Similarly, Research Affiliates’ Shane Shepherd found that the market beta of a passive merger arb strategy was higher when the S&P was down than when it was flat or up (as the following table from the paper illustrates).

(Shepherd proposes a new “merger arbitrage risk factor” that can explain merger arb returns and reduce the market beta to basically zero in a two-variable regression.)

In a 2004 article in the Financial Analysts Journal, Fung & Hsieh examined the returns of merger arbitrage hedge *funds *to see if the same phenomenon existed. As the chart below from their paper shows, merger arb returns are less correlated with the S&P 500 when the market is up (click to large)…

This seminal study was based on merger arb index returns from January 1990 to December 2002. So we wondered what would happen if we updated the analysis to include data up to the present. Here’s what we found…

First, here is the relationship between the S&P500 and the HFRI Merger Arb Index. The S&P is on the horizontal axis and the HFRI MA Index is on the vertical. We include both a linear regression and a polynomial one to capture the non-linear dimension of this correlation. As you can see by the R-squareds, the polynomial regression has a better fit.

When we included the data from January 2003 to October 2009, this is what we got…

Note that the alpha (intercept) drops, the beta rises, and the R-squared rises in the linear regression. The polynomial regression loses some of its fit advantage.

If you’re starting to get the feeling the ’03-’09 data has a lot of linear characteristics, you’re not alone. When we looked at the January ’03-October ’09 data in isolation, here’s what we got…

The polynomial regression has lost its relative fit advantage, suggesting that there is little *non-linear-ness *to the correlation left between the S&P 500 and HFRI Merger Arb Index.

While merger arb funds seem to have lost their knack for down-side mitigation, they haven’t lost their alpha-producing abilities altogether. Alpha measures using both regressions have dropped from around 80 bps per month to 40 bps per month. A huge drop – but still an alpha of over 5% per annum. That’s enough to make a hero of a long-only equity manager.

The question remains, however, does this quality gleaned from merger arb hedge funds translate to a passive merger arb strategy – or are merger arb hedge fund managers trying to goose returns by hitching their wagons to the S&P?

## 2 Comments

December 9, 2009 at 11:45 am## MWC