Institutional constraints and preferences: Implications for factors and performance evaluation

Abstract

Two perspectives on institutional ownership motivate this study. First, institutions display 'preferences' for stocks with certain characteristics that relate to asset pricing factors in a potentially causal way (Gompers and Metrick (2001)). Second, these `preferences' often reflect
constraints relating to agency considerations and liquidity [Del Guercio (1996); Berk and Green (2004)]. We provide a detailed empirical analysis to distinguish between the effects of preference vs. constraints on institutional ownership (IO), and find that a constraints interpretation is most suitable for the smallest institutions. Large institutions, where one expects costs of portfolio implementation to be most severe, seem to operate in an unconstrained investable universe. We then construct IO pricing factors - standard asset pricing factors restricted to stocks that satisfy IO constraints - to evaluate asset pricing effects from implicit market segmentation relating to IO. Contrary to both intuition regarding portfolio optimization and conventional wisdom regarding academic paper-portfolio benchmarks, we find that the market-factor premium is higher for stocks predicted to have high IO. As a consequence, institutional performance is worse when evaluated against factors culled from their investment universe than traditional global factors.