Are Specialists ‘Special’? The Case Of Institutional Investors
University of Oxford – Said Business School; Institute for New Economic Thinking at the Oxford Martin School; Kiel Institute for the World Economy
May 27, 2016
Are specialists ‘special’? We answer this question using data on the stock portfolios of U.S. institutional investors for the period 1990 – 2014. Our main findings are as follows: first, we show that many investors tend to focus their economically meaningful investments on a relatively small subset of stocks. This translates itself into a strong and significant portfolio overlap. Finally, specialized investors (those with concentrated portfolios) tend to emphasize certain stock-specific characteristics, but we find no evidence that their returns are significantly different from those of diversified investors. Overall, our findings suggest that specialists are not that special.
Are Specialists ‘Special’? The Case Of Institutional Investors – Introduction
Whether investors should hold concentrated or diversified portfolios is a long-standing research question. In this paper, we show empirically that the cross-sectional distribution of diversification strategies among institutional investors is very broad, such that highly diversified investors coexist alongside with much more specialized investors. In a world where specialized investors had superior information, they should (1) hold ‘special’ portfolios, i.e., focus on a different set of stocks compared to other investors, and (2) earn superior (risk- adjusted) returns. The main finding of this paper is that both of these hypotheses are rejected in empirical data for U.S. institutional investors during the period 1990 – 2014. Rather we find significant overlap in investors’ portfolios, meaning that many investors tend to cluster large parts of their investments on a similar set of stocks. This would suggest that specialists are unlikely to be very special, since they hold those stocks that most other investors tend to hold as well. In our main analysis we explore the implications of this finding in more detail. We show that specialists indeed put different weights on certain stock-specific characteristics, but they do not generate superior returns compared with more diversified investors. Overall, specialists are not that special.
Why are these findings important? First, our paper adds to the literature on optimal diversification. The standard result of modern portfolio theory a l´a Markowitz (1952), is that investors should diversify as much as possible. This, however, is at odds with mounting empirical evidence that many investors hold concentrated portfolios (see Heaton and Lucas (2000); Goetzmann and Kumar (2008)). Several recent theoretical contributions have shown why investors might rationally choose different levels of diversification in the presence of frictions, such as costs of information acquisition, borrowing constraints, and market illiquidity (see Van Niewerburgh and Veldkamp (2010); Roche, Tompaidis, and Yang (2013); Wagner (2011)). For example, in Van Niewerburgh and Veldkamp (2010) investors jointly optimize their investment and information choices, which in many cases leads investors to specialize in those stocks for which they possess more accurate (i.e., superior) information. The authors also show that an investor without any information should hold a diversified portfolio.
Having established that some investors might find it optimal to hold concentrated portfolios, a natural follow-up question is whether these specialists will focus on the same set of stocks or try to hold special portfolios, i.e., occupy niches. The theoretical literature does not offer a clear prediction in this regard: on the one hand, Wagner (2011) shows that, if investors care about price impact and systemic liquidation risk, no two investors will hold the exact same portfolio in equilibrium. In other words, in a world where liquidation risk matters, investors would aim to be as special as possible. On the other hand, there is a well-established literature on rational herding which shows that investors will rationally choose to hold very similar portfolios under certain circumstances, most importantly when the remuneration of investment managers depends on their relative performance. The results presented in this paper suggest that the latter factor seems to be more relevant for the set of institutional investors.
Our paper also adds to the empirical literature on investor performance (e.g., Jensen (1968); Grinblatt and Titman (1989); Carhart (1997); Coval and Moskowitz (2001); Kacperczyk, Sialm, and Zheng (2005); Ivkovi´c, Sialm, and Weisbrenner (2008); Cremers and Petajisto (2009); Lewellen (2011)). The broad consensus in this literature is that it is very difficult to detect investors with superior abnormal returns, which is in line with existing theories (Van Niewerburgh and Veldkamp (2010); Wagner (2011)): in equilibrium, the (expected) returns of diversified versus focused portfolios should be the same. In this paper we show that most investors tend to focus most of their economically important investments on a relatively small set of stocks. From this perspective it comes as no surprise that the vast majority of investors tend to generate similar returns. What is surprising, however, is that investors with intermediate diversification levels tend to perform poorly.
Lastly, a recent literature highlights the importance of common ownership on stock prices and the possibility of systemic events via overlapping portfolios (e.g., Coval and Stafford (2007); Greenwood and Thesmar (2011); Ant´on and Polk (2014); Greenwood, Landier, and Thesmar (2015)). For example, according to Greenwood and Thesmar (2011) an asset can be fragile due to concentrated ownership or because its owners face correlated or volatile liquidity shocks, i.e., they buy or sell at the same time. They present empirical evidence that the extent of common ownership indeed predicts price volatility. Ant´on and Polk (2014) show that the degree of mutual funds’ shared stock ownership forecasts cross-sectional variation in return correlation.1 Their main conclusion is that shared ownership causes excess comovement, with asset liquidations being a key mechanism behind this finding. Greenwood, Landier, and Thesmar (2015) propose a simple measure of vulnerability which quantifies the risk due to investors’ overlapping portfolios. In sum, the literature predicts that higher overlap can destabilize the system as a whole, but, to the best of our knowledge, there is no empirical study quantifying how overlapping actual investors’ portfolios are. In this paper, we show that the observed overlap varies over time and is much stronger than we would expect by chance.
The remainder of this paper is structured as follows: section 2 defines the holdings network and related variables. Section 3 explains the underlying dataset and section 4 presents the empirical results. Section 6 discusses the main findings and concludes.
See full PDF below.