I recently had the opportunity to participate at the 13th Colloquium on Financial Markets organized by the Centre for Financial Research in Cologne. It was a very interesting experience and I enjoyed it a lot. The program included research papers in Asset Pricing, Market Microstructure, and Asset Management. You may find the program here. In a future post, I will talk about my current research project on institutional trading activity and liquidity, which I presented at the Colloquium. In this post, I would like to summarize a great paper by Simon Gervais (Duke University) and Günter Strobl (Frankfurt School of Finance and Management), which I had the opportunity to discuss. I liked the paper very much and I think it is very necessary to have more theory guiding empirical research in the area of institutional investment.
Foto: Centre for Financial Research Cologne
In their paper, Transparency and Talent Allocation in Money Management, the authors investigate theoretically how different degrees of transparency arise endogenously in a context of competition among asset managers and investors. To do that, they propose a model where transparency can be used to signal quality.
The model’s ingredients are the following:
- There are three types of managers: high-skilled, h; medium skilled, m; and low-skilled, ℓ.
- Each manager observes her own type, sets up a fund, and chooses a level of transparency, t, which defines the probability that a ℓ-manager sends a bad signal to investors.
- Transparency increases the fund’s costs.
- Investors observe t and the history of returns each period and invest as long as net expected performance is positive.
- Diseconomies of scale and competition among investors imply that all funds deliver zero net expected performance in equilibrium (Berk and Green, 2004).
- Managers choose fee to maximize compensation.
Under these assumptions, the authors prove that in equilibrium, ℓ-managers and h-managers choose the same level of transparency. Why? ℓ-managers will always want to look like they are better than they really are. However, m-managers will never choose to pool with ℓ-managers and separate themselves from h-managers. A partial pooling equilibrium in which ℓ-managers and h-managers choose the same level of transparency and m-managers choose a different level of transparency exists as long as m-managers are sufficiently more transparent than the other two types and the number of periods is sufficiently large. The intuition for this result is that ℓ-managers do not want to incur the costs associated with the higher transparency of m-managers. Also, m-managers do not want to lower transparency in order to look like h-managers since in the long run, their lower returns will give them away. Finally, h-managers do not want to increase transparency to separate themselves from ℓ-managers, since in the long-run their higher returns will drive investors away from the latter. Therefore, in the partial pooling equilibrium both the most and the least skilled asset managers choose to be more opaque, while the average skilled managers choose to be more transparent.
The paper therefore provides a rationale for the existence of two levels of transparency in the asset management industry: mutual funds, subject to strong disclosure requirements, and hedge funds, which remain largely opaque to investors. Also, a number of new empirical predictions can be obtained from the model, which will provide the basis for future empirical research.
I think it would be interesting to further investigate some extensions in the future. For instance, how would the model’s results change if asset managers could signal their types through other choices, such as the decision to charge a performance-based compensation? How would entry of new funds every period affect the likelihood of the partial pooling equilibrium? What would happen if transparency enabled m-managers to send a signal to investors directly, just like ℓ-managers?
Taking the model to the data is not trivial. For instance, different degrees of transparency could attract investors with different degrees of sophistication, which in turn can impact their performance (see my previous entry). Also, hedge fund databases contain information only on those hedge funds that volunteer that information. Therefore, one would expect that the worst managers will be strategic about disclosing information, so even if hedge funds contain both the best and worst funds of the industry, as the model predicts, the hedge fund databases will be biased towards the best performers. The problem with opacity is that what’s opaque to investors is also opaque to researchers. These are some interesting challenges for empiricists.
The paper won the Best Paper Award at the Colloquium. Congratulations to the authors!