Print
E-MailItzhak Ben-David is Associate Professor of Finance at The Ohio State University. This post is based on a recent paper authored by Professor Ben-David; Francesco Franzoni, Professor of Finance at the University of Lugano; Rabih Moussawi, Assistant Professor of Finance at Villanova University; and John Sedunov, Assistant Professor of Finance at Villanova University.
The U.S. asset management industry has become increasingly concentrated in recent times. Over the last 35 years, the largest institutional investors have quadrupled their holdings in the equity market. As of September 2015, the largest asset manager oversaw 5.1% of the total equity assets in SEC 13F filings, and the largest 10 managers managed 23.4% of these assets, which represents about 35% of overall institution ownership.
Such concentration of assets held by the largest asset managers is arguably associated with increased risk concentration, which can have a systematic impact. In other words, the risk that the largest institutions pose is that idiosyncratic shocks to any large individual player in an economy is hardly diversifiable, due to their sheer size (Gabaix, Gopikrishnan, Plerou, and Stanley, 2006; Gabaix, 2011). In this vein, large institutional investors are not equivalent to a collection of smaller independent entities; rather, they have an incompressible institutional identity that leaves a large footprint in the market. That is, they are “granular,” and idiosyncratic shocks to those large institutions are also “granular” in that they cannot be diversified away or absorbed by the market. The asset management space indeed has experienced many examples of idiosyncratic events at the institutional investor level that have had widespread repercussions in the financial system. Recent examples include the bankruptcy of Lehman Brothers in 2008, JP Morgan’s trader (the “London Whale”) who built a large short position in credit default swaps that led to trading losses exceeding $6 billion within weeks, and the sudden departure of co-founder Bill Gross from Pimco in September 2014 which caused unprecedented large withdrawals and consequent drastic liquidations. It is important to note that idiosyncratic events need not be rare or extreme to have an impact on asset prices. A large institution that initiates trades to accommodate investor flows, or for portfolio rebalancing, or for risk-management reasons may cause price dislocations. Regulators have expressed similar concerns about systemic risks that could result from the high concentration of assets under a single large manager.
Given that the evidence on the effect of large firms is so far anecdotal, in our NBER working paper, The Granular Nature of Large Institutional Investors, we provide a large-sample study on the impact of large institutional investors on price stability. Using more than 35 years of ownership data from 13F filings, we measure the effect of large institutional ownership on stock volatility.
Our empirical results show that ownership by the largest institutional investors increases the volatility of the underlying securities. We use two distinct identification strategies to address potential endogeneity concerns. The first relies on “local bias,” that is, the prior finding that asset managers overweight firms that are located closer to the investor’s headquarters. Using this identification technique, we find that the economic magnitude is large: a 1% increase in stock ownership causes an increase in stock volatility of about 12 to 18 basis points, relative to a daily average of 3.5%. Our second identification strategy relies on the merger between two large institutional investors, Blackrock and Barclays Global Investors, that took place at the end of 2009 and spawned the top institution in the market. The granularity theory in this context suggests that the shocks to one large consolidated organization (the merged firm) have a greater impact than the shocks to separate entities (the pre-merger organizations). Consistent with this hypothesis, stocks owned by the combined entity exhibit higher volatility than stocks owned by the pre-merger firms and that this effect persists well after the merger event.
The question remains, however, of whether large institutions’ trades are different from those of a collection of small investors that add up to the same total size. If they are not, then unbundling a large institution into smaller pieces would not eliminate the potential price distortions induced by large institutions. We address this question in multiple fashions. First, using trade-level data, we find that the trades by the largest institutions are associated with larger price impact than those of smaller institutions. Second, we compare the trades of those large institutions to those of a random collection of smaller independent institutions with the same total amount of portfolio holdings. The goal is to build a synthetic institution representing the counterfactual world in which large institutions are broken up into smaller entities. Our results show that large investors trade in a more concentrated portfolio of stocks and in bigger sizes than the synthetic institutions. For example, after 2000, the 10 largest firms trade in just 51 % of the available stocks, while the synthetic organizations trade in 72% of the universe. This concentrated trading is likely to exacerbate price impact. Furthermore, the size of the trades of the large investors is substantially bigger than that of the synthetic institutions and therefore are more likely to impact prices. For example, 16% of the trades of large institutional investors are above the 90th percentile of the distribution of trades of the synthetic institutions, and 4% of the trades of the large firms exceed the 99th percentile of the same distribution. Finally, we explore the role of investor flows into large asset managers as potentially one of the causes for the large trades we document. We execute this test using the monthly flow data of US mutual funds, because mutual fund flows are measured more accurately and on a relatively higher frequency. We document that the correlation between mutual fund flows is higher among funds under the same management company than among independent funds.
We further investigate whether the increase in volatility is a desirable outcome of institutional ownership, e.g., large institutions leading to faster price discovery. We provide two tests along these lines. First, we document that ownership by large institutions is associated with stronger daily return autocorrelation, indicating reduced price efficiency. Second, we show that the returns of stocks that are owned by large institutional investors co-move with the returns of the rest of these institutions’ portfolios proportionally to their ownership level, suggesting that the underlying securities are exposed to the same shocks, presumably spilling over from the large institutional investor.
Overall, our study shows that ownership by large institutional investors increases the volatility in the prices of the portfolio securities. Not only does our analysis document such risk creation by the largest institutions, but it also establishes a causal link between large institutional ownership and the increase in volatility of stocks through large trades, which translate into substantial price pressure. Large institutions have a “granular” nature that leads them to trade in a less diversified way than a random collection of independent entities. Finally, our analysis suggests that the increase in volatility is associated, at least in part, with an increase in noise.
The full paper is available for download here.
