Crowded Spaces and Copycat Risk Management

Abstract

In recent years, a new form of risk has been recognized. This risk is the risk of crowded spaces. That is, how the saturation of a particular trading strategy can lead to a mis-measurement of the future expected returns and risks of the trading strategy. The primary focus of this risk has to do with copycat trading, whereby traders or investors engage in a similar trading strategy. As too many dollars chase the same strategy, the opportunity fades away but leaves the remaining traders at greater risk exposure. This paper focuses on an entirely neglected source of crowding. The crowding explicitly from the risk management process. In particular, this paper focuses on equity portfolio managers that use similar risk models and examines the extent to which these similar risk models can mis-measure risk and lead to crowding of the investment space, even when the investment models are completely unrelated to one another. Our empirical analysis with real-world professional portfolio management data finds that the risk modeling process leads to crowding and mis-measured risk. Interestingly, we find a rise in crowding due to portfolio construction immediately before the quant crisis of 2007. We also show how a simple method of altering the eigenvalue structure can reduce the average crowding from portfolio management by 20%, and in some cases as high as 61%. We also find that crowding in the financial system would be lower if the distribution of risk model usage amongst portfolio managers was more diversified.