This paper examines the survivorship bias and investment styles in surviving and defunct funds operated by commodity trading advisors (CTAs).
The motivation for studying CTA funds is based on the fact that CTA funds can potentially provide diversification from global equities. It is generally known the CTA funds generate returns which have low correlation to global equities. What is less well known, but even more interesting, is that CTA fund returns tend to be positive and large during declines in global equities. The goal of this paper is to investigate the consistency of this behavior across both surviving and defunct funds, as well as the amount of bias inherent in using surviving funds to estimate the returns of all (surviving and defunct) funds.
The data set is supplied by Tass Asset Management. It consists of 901 CTA funds, 304 of which survived until the end of the sample in 1996, while the remaining 597 ceased operation prior to that date.
The key findings in this paper are as follows.
One, there is a 20% chance that a CTA fund will cease operation in any given year. This is much higher than the dissolution rates in mutual fudns.
Two, the survivorship bias is 3.4% per year. That means the returns of surviving CTA funds, on average, are 3.4% higher than the return of all CTA (surviving and dissolved) funds.
Three, there is one dominant investment style in CTA funds, which is not affected by the entry or dissolution of CTA funds. This style is called "trend following" based on the description of CTAs.
Four, the option-like payout feature (i.e. positive and large returns during down months in global equities) is consistently exhibited in dissolved as well as surviving CTA funds.
Five, the performance of dissolved funds operated by multi-fund CTAs was better than dissolved funds operated by single-fund CTAs. We attribute this to "reputational" effects. Given the incentive compensation scheme in CTA funds, there is a tendency for CTAs to take more risk when their funds' net asset values are substantially between the high water mark. Multi-fund CTAs have less incentive to do so, because they need to protect their reputation to the investors in other funds.
Two important issues remain open for future research: how do "trend following" CTAs generate the option-like payout feature in their returns? and will they continue to do so in the future?