The AMF Scientific Advisory Board dedicated a session to the issue of asset management.

Funds’ liquidity risk 

Serge Darolles (University Paris-Dauphine, member of the AMF Scientific Advisory Board) and his co-author ( (Guillaume Rousselet, McGill University) developed a structural model allowing for the stylized representation of the liquidity risks with which  investment funds are faced : indeed, on the asset side, fund managers can invest in illiquid assets which may be subject to large haircuts in case of forced sales  (market liquidity risk) ; on the liability side, clients can decide to redeem their fund units or shares (funding liquidity risk). This model points to an optimal cash holding, necessary to bear the liquidity risks, and determines the associated cost, especially in terms of performance.

The authors introduced a redemption limit parameter in their model (interpretable as gates) which is particularly relevant for regulators. What is more, they separated the market liquidity shock into two parts: the haircut corresponding to normal conditions on the secondary market, and the market impact which depends on the quantity of assets liquidated by the fund. This structure enables them to correlate asset side and liability side liquidity risks.

The main conclusions of the model are the following:

• The optimal quantity of cash holding increases with the likelihood of redemptions, but diminishes with market liquidity;

• In the straightforward case (no gates and no contagion), the default probability is not dependent upon redemptions, but decreases with market liquidity ;

• The market impact modeling implies contagion between funding liquidity and market liquidity, but gates tend to limit this contagion.

In a final stage, the authors calibrated the various parameters of their model for nine types of hedge fund strategies, from Lipper TASS data covering 2000-2015, and ranked the strategies according to their need for liquid assets.

During the discussion, the Scientific Board members stressed the relevance of such modeling of liquidity constraints in asset management. They recommended the model be refined (beyond a mere binary choice between totally liquid and totally illiquid assets via the introduction of assets of different maturities. They also suggested that the contagion mechanisms be explored further, through a general equilibrium model for instance. Eventually, they emphasized that the equal treatment principle among fund investors contradicts the fact that the fund should use its available cash to repay the clients that exit first.

Understanding the active / passive management debate

Thierry Roncalli summarized the definitions and empirical results that relate to stock pricing models. In particular he recalled the meaning of the alpha (idiosyncratic risk, or stock own risk) and that of beta (correlation with the market). He also addressed the development of the concept of factor investing (starting with the size, value and momentum factors). This introduction shed light on two types of asset investment strategies: the discretionary approach (on alpha) and the factor based approach (on the betas).

Secondly, he presented his understanding of the evolution of systematic (or rule-based) asset management, insisting specifically on the multiplication of strategies that are now offered by the market. The increasing share of systematic asset management could bear consequences in terms of systemic risk : indeed, the systematic approach relies on a limited number of mostly backward-looking quantitative models, and is by design prone to herding and amplification effects. The market for volatility was mentioned as a potential source of systemic risk. The brutal correction on the VIX at the beginning of February 2018 illustrates, notably by its step configuration, the existence of multiple triggers that were activated one after the other by the models. 

The third part of the presentation dealt with the notion of active share, which reflects the difference observed between the structure of a fund’s assets and the composition of its reference index. The active share is often associated with the tracking error to characterize management styles (even if the tradeoff between these two parameters is in fact constrained by the relationship that ties them together). The studies show that the active share’s value depends on the reference index as well as on the investment universe. Moreover, the active share is not an indicator of future performance.

Eventually, Thierry Roncalli stressed the link between performance and fees: if the chances of beating the reference index are 50 % when no charges are applied, this proportion falls to 40.1% for management fees of 1% and further to 30.9% for 2%. The author concluded by providing thoughts on the more appropriate performance indicator (ranking vs. benchmarking).

The Scientific Advisory Board members highlighted the importance of the questions raised by Thierry Roncalli. They were insistent on the need to assess active management’s share needed to maintain well-functioning markets and to establish equilibrium prices. They also noted the significant increase in the number of factors that are being discovered by operational and academic researchers, and stressed some overlap and ambiguity (despite strategies with the same name, two different asset managers can use factors that bear little correlation).