“There is nothing less practical than a bad theory,” wrote CEO Paul Shott Stevens of the Investment Company Institute (ICI) in a July 2016 blog. Mr. Stevens introduced a series of recent findings that the ICI suggests may present a rebuttal to their members of the “first-mover” hypothesis. What is that, one may ask, and why should I care?
Under the first-mover hypothesis, cited favorably in an update on the asset management review being conducted by the Financial Stability Oversight Council of the U.S. Treasury, sophisticated investors are said to redeem their shares in a poorly performing fund at full value before others can appreciate the increasing risks. According to the theory, the other investors suddenly follow, creating a cascade of redemptions and forced sales of fund assets. The Treasury is considering new rules to limit the risk of fire sales or prevent these first-mover-driven sell-offs.
However, in contravention to that theory, the ICI reported that, shortly after the UK’s shocking Brexit vote, investors redeemed just 1% of the more than $300 billion now held in U.S. investment-grade and high-yield bond funds. The first-mover theorists would have expected massive outflows from these funds after the shock. But that didn't happen.
According to ICI, the data suggests that the effect of redemptions to any one fund after Brexit may well have been offset by reinvestments in more balances or better protected funds. Simply put, sales were followed by buys, and that allowed the fund sector to reach equilibrium -- perhaps with lower asset values but without a selling panic at fire-sale prices.
ICI argues that academics whose fire-sale research focuses only on sales tend to miss that equilibrating dynamic. In fact, we would agree and add that it would not be the first time that academic studies with a distorted viewpoint were used to influence investor behavior.
In the late 1990s, the JPMorgan Chase Bank, in a series of collaborative studies with ASTEC Consulting, sought to respond to research theorists who alleged short sellers drove down prices for those investors who lent their securities to the shorts. Just as with the recent ICI analysis, the Chase-ASTEC rebuttal presented real world data in a multi-vector context to show that some preconceptions that drive academic theories, while intuitively appealing, are often quite wrong.
In the Chase-ASTEC study, any negative affect on prices from short-selling was shown to have been neutralized by the price support arising from reinvestment of the proceeds. For example, many short sellers immediately reinvested their sale proceeds into purchases of the second leg of a pairs trade. And even if the cash wasn't immediately reinvested, the short sellers eventually had to return the shares they had borrowed -- through purchases which offset the earlier downward pressure.
In a 2007 update to that study, the Center for the Study of Financial Market Evolution (CSFME) used global securities finance data to show attendees at an IMF-World Bank conference in Moscow that prices for Russian securities were actually being supported by short sellers. In other words, the dynamics held even for emerging markets.
In an even more recent empirical study, the CSFME employed data provided by members of the Risk Management Association in a robust rebuttal to allegations that hedge funds were borrowing equity shares in the securities finance markets for the express purpose of manipulating the outcome of corporate annual meetings. CSFME found that the spike in borrowings that academics had noted before the record date was not an attempt by hedge funds to take control of shares they didn’t own, and then manipulate the vote. Rather it was the result of loan recalls by the original fund investors in order to vote their own shares. The RMA-CSFME study used a much deeper pool of data to explain the spike, provided by eight lending agents, while the original academic study focused on borrowings, not returns, and used only two data providers.
In all of these cases, the academic studies did not go far enough, in part because their single-vector models did not have access to a sufficiently complete set of data upon which to base their conclusions. In many cases, similarly incomplete models are being used today to argue in favor of stringent reform regulations. It's not too surprising that the latest market data have shown that there is a growing tendency toward unintended adverse results from those regulations, as noted in my September, 2016 article in the RMA Journal.
Going forward, regulators may well find that multi-vector studies with robust data will conclude that, instead of the possibility that the market infrastructure is fragile, the opposite is true; that is, that markets are sufficiently resilient to absorb the sales by certain investor groups without damage to the overall market system. And perhaps academics will also find that the rebalancing of investor accounts within bond sectors is just one of many unanticipated beneficial consequences from come from otherwise stressed markets.
In any event, the take-away from all of these studies is that one vector of market activity, whether it's asset sales or security loans, is just not sufficient to model the real world. In each of the multi-vector cases, analysts used not just sales or borrows, but also buys and repayments to estimate “net flows” and show that the effect was not as bad as that anticipated by the single-vector theorists.
In his blog, the ICI's Mr. Stevens echoed Mr. William Dudley, the president of the New York Federal Reserve Bank, who, in September 2015, to invite new research into the dynamics of the market system. We hope that the results will be made available in the near term, before regulators have to conclude their final rules.