Corporate credit markets have historically been especially prone to herding. The main drivers of herding have been past returns, rating changes and liquidity. Sell herding has been particularly strong and flows have been disproportionate after very large price moves. Herding can be persistent and lead to significant price distortions. Non-fundamental price overshooting is a valid basis for profitable contrarian trading strategies.
Cai, Fang, Song Han, Dan Li, and Yi Li (2016). “Institutional Herding and Its Price Impact: Evidence from the Corporate Bond Market,” Finance and Economics Discussion Series 2016-091. Washington: Board of Governors of the Federal Reserve System.
The post ties in with the subject of methods for detecting price distortions in the summary on price distortions.
The below are excerpts from the paper. Headings and some other cursive text has been added for context and convenience of reading.
What is herding and how strong is it?
“Intuitively, herding is measured as the tendency of funds to trade a given bond together and in the same direction (either buy or sell) more often than would be expected if they trade independently.”
“The level of institutional herding in corporate bonds is substantially higher than what has been documented for equities. Specifically, we estimate that the average bond herding levels of pension funds and mutual funds are each about 10 percent, significantly higher than the levels of about 3 percent for the respective type of equity funds. Intuitively, our estimates indicate that funds in each group are roughly 10 percent more likely to trade on the same side than one would expect if they made their trading decisions independently. Insurance companies, the largest investor group of corporate bonds, have an even greater tendency to herd than mutual funds and pension funds, boasting an average herding level of 13 percent.”
“We find that sell herding in corporate bonds is significantly stronger than buy herding, a result mostly driven by mutual funds, the most active traders and fastest-growing investors of corporate bonds. Over time, mutual funds also stand out by exhibiting unique trends, with buy herding levels declining and sell herding levels rising.”
“We further decompose intertemporal herding into an imitation component (institutional investors following others into and out of the same securities) and a habit component (investors following their own trades in the last quarter). We find that, strikingly, the positive intertemporal correlation in bond trading is mostly driven by institutions following others’ trades…The strong imitation-driven intertemporal herding in corporate bonds implies that herding in this market is more akin to run behaviors than in the equity market.”
What factors drive herding?
“We find that rating changes, bond liquidity, and past bond performance are key factors that drive herding by different types of investors. All institutional investors are found to herd more in lower-rated and smaller-sized bonds. Perhaps unsurprisingly, insurance companies react more to rating-change events, particularly downgrades, consistent with the fact that they are subject to rating-based regulatory constraints.”
“We also find that all investors herd to buy winning bonds and herd to sell losing bonds, with insurance companies’ herding behavior most sensitive to bonds’ past performance…Interestingly, this herding-to-performance relationship is nonlinear. Specifically, we find that extremely bad past performances are associated with disproportionally large selling herds, while top-performing bonds do not attract disproportionally large buying herds. Such asymmetry suggests that bonds’ extremely bad past performances may trigger a larger amount of simultaneous sells from institutional investors, a condition that could lead to further price declines and in turn result in more sells and a downward price spiral.”
“We also find that the herding level in trading lower-rated bonds, at 12 and 22 percent for high-yield and unrated bonds…is notably higher than that for investment-grade bonds, 9 percent.”
How long are herding episodes?
“We document strong persistence in herding, especially on the sell side. We show that bond investors not only herd within a quarter, but also herd over adjacent quarters. In fact, intertemporal correlation in corporate bond trading is much higher than that in equity trading, especially for insurance companies.”
What are the consequences of herding?
“We document a significant price-destabilizing effect of sell herding, suggesting that institutional sell herding could pose substantial risks to financial stability…the price-destabilizing effect is strongest for the most risky bonds during periods of market distress.”
“For corporate bonds…trading liquidity and price discovery rely on liquidity provision by securities dealers…Two recent trends have intensified…concerns…On the one hand, the U.S. corporate bond market has expanded rapidly since the crisis, boosted by significant increases in institutional holdings. On the other hand, over the same time, dealers have sharply shrunk their balance sheets, which may limit their market-making capacity…Institutional herding could drive asset prices away from their fundamentals, particularly on the downside, and dealers’ limited market-making capacity would only exacerbate this price distortion.”
What is the investment opportunity?
“If institutional investors herd based on non-fundamental factors such as reputation concerns, we should generally observe bond prices overshoot temporarily and reverse course in the long run. In contrast, if institutional herding is based on bond fundamentals, their collective trades should contribute to price discovery, and there should not be price reversal afterward.”
“We find that, while buy herding is associated with permanent price impact that facilitates price discovery, sell herding results in transitory yet significant price distortions and therefore excess price volatility.”
“The impact of institutional herding on long-term corporate bond returns is substantial. In particular, when investors herd to sell, bond prices fall substantially during the event period but reverse gradually over the following quarters. A contrarian portfolio that is long in bonds with the highest sell herding measures and short in bonds with the highest buy herding measures generates a cumulative abnormal return of 2.5 percent in six quarters after portfolio formation. Such an abnormal return is entirely driven by subsequent return reversals in bonds that experience heavy sell herding in the event period.”
What is the empirical evidence based on?
“We obtain data on institutional investors’ holdings of corporate bonds from Thomson Reuters Lipper eMAXX. This database contains quarter-end security-level corporate bond holdings of insurance companies, mutual funds, and pension funds.”
“Following the literature, we define “trades” as changes in funds’ quarter-end holdings. A limitation for this definition is that quarter-end portfolio snapshots cannot capture intra- quarter round-trip transactions.”
“We construct a “full sample” and a “herding sample,” both covering the period from 1998:Q3 to 2014:Q3.”
“The…herding measure…is by design estimated for a given group of institutions. In this paper, we first treat all investors as a single group and then treat each of the three types of investors, insurance companies, mutual funds, and pension funds, as an individual group.”