Let Data, Not Politics, Guide Regulation
Several years after the global financial crisis, the fierce debate over regulation continues to be driven by strong beliefs -- largely uninformed ones -- rather than hard facts.
Some believe more regulation is necessary, others that it would cause the downfall of our markets. No one, however, seems to be talking about the evidence for or against regulation.
This is partly because very little evidence exists on the effects of regulation or its efficacy. It is extremely difficult to isolate the impact of any proposed regulatory change from everything else that is occurring simultaneously within financial markets. Studies claiming to analyze those effects are almost always confounded by other factors, such as the environment that led to a given regulation; the response of market participants to regulation or anticipated regulation; the selection of securities targeted for regulation; and the timing of regulation.
To overcome these problems and identify an effect unrelated to other possible effects, it would be useful to run a controlled experiment, as medical research does with clinical trials. Is that feasible in financial markets? My colleagues Steven Kaplan, Berk Sensoy and I recently conducted such an experiment to get a better understanding of one type of financial regulation: bans on short selling certain stocks, a very simple measure that was put in effect worldwide for financial stocks during the crisis.
The prevailing view at the time was that allowing short selling would further distort markets and adversely affect prices, moving them further away from fundamentals. In theory, short selling would depress prices further and “cause” excess volatility, thus destabilizing markets even more. Another theory, however, is that short selling improves price discovery, and so any ban on shorting would itself distort markets.
The data show that, empirically, the effect on prices of short-sale restrictions has been mixed because of the difficulty in isolating the regulation from the effects of supply-and-demand responses in the market. For example, after the short-sale ban, many researchers quickly gathered the data to analyze its effects on prices. The difficulty, though, was that regulatory intervention and its application to certain securities were almost surely related to other factors in the market -- such as the conditions that spawned the regulation debate in the first place -- that also influence security prices. In fact, the aftermath of the regulatory changes themselves may be exactly the wrong time to try to identify their impact, since they typically occur at these most extreme times and are applied to a highly selected group of securities.
Working with a large (more than $15 billion in assets), unidentified money manager, my co-authors and I were able to conduct an experiment on short selling. We randomly increased and restricted the supply of shares available for lending that were held by the money manager. These random “supply shocks” were unknown to any outside investors, were unannounced and weren’t motivated by market conditions. In effect, we moved the supply of lendable shares in the market in a way that wasn’t confounded by any anticipated demand response.
Just like a medical trial, we had a randomized group of “treated” stocks (whose supply of lendable shares changed) and a randomized group of “control” stocks, whose lendable shares remained the same. We focused on the stocks with high loan fees (of at least 25 basis points per year; the average was more than 4 percent per year), which tend to be those that have high shorting demand relative to their supply. These are the set of stocks where a supply change would be most effective.
We ran the experiment twice: once during the turmoil of the financial crisis -- Sept. 5 to Oct. 3, 2008 -- and from June 5 to Oct. 1, 2009, a calmer period. For the treated stocks, the experiment was able to substantially increase the supply of shares available for lending. On average, we moved enough of the supply of shares of these stocks to comprise 214 percent of daily trading volume, almost 37 percent of total short interest, and 6.9 percent of total institutional ownership. At its peak, more than $580 million of securities were lent out from the experiment.
What did we find? First, actual loan fees on the stocks we treated declined significantly, about 2 percent to 3 percent per year. That’s good news: Increasing loan supply, holding everything else constant, should lower loan fees and prices.
Second, what happened to the underlying stock prices? Well, despite focusing on high-demand stocks and producing sizeable changes to supply, we didn’t find any adverse effects on stock prices, volatility or even bid-ask spreads. Positive and negative shocks to shorting supply had no price consequences for the underlying stock, even for the least liquid ones with the highest shorting demand. In short (pun intended), we found no evidence that shifting supply mattered. And we found identical results in both experiments.
For the investment manager, this was good news, since it meant earning additional revenue from lending his shares without adversely affecting the prices of his stock holdings. For policy makers, these results don’t provide any support for the view that regulation on shorting can prop up stock prices. But, as with any experiment, caution must be taken when extrapolating beyond the confines of our laboratory. We cannot say for certain whether higher levels of restricted lending than we could provide in our experiment, or restrictions for other types of stocks not held by this particular money manager, would have yielded different results. However, as in medical trials, if we think these effects apply to a broader group than the subjects of our experiment, then perhaps we do learn something about its efficacy in general.
While it is difficult for individuals to conduct experiments on their own, especially on a wide scale, governments and their agencies are in a unique position to do so. If more such analyses were done, policy debates on financial regulation would be better informed and guided. And, policies themselves could be better structured and focused.
Unfortunately, it’s not just a matter of running experiments. Getting policy makers to confront data is probably a bigger issue. Consider the Securities and Exchange Commission, which had conducted a related experiment on shorting -- the SHO program -- three years before the financial crisis and found evidence consistent with our conclusions that shorting restrictions are ineffective. If, instead of focusing on politics, policy makers had paid attention (or chosen not to ignore) the data, they would have created much better and more useful policies. Of course, hoping politicians will abandon politics in favor of data is like hoping my 5-year-old will abandon his toys in favor of bed when he’s tired -- both would be better off, but it isn’t going to happen anytime soon.
(Tobias J. Moskowitz is Fama Family professor of finance at the University of Chicago Booth School of Business, and is a contributor to Business Class. He also serves as a consultant to AQR Capital Management LLC. The opinions expressed are his own.)
Read more opinion online from Bloomberg View. Subscribe to receive a daily e-mail highlighting new View columns, editorials and op-ed articles.
Today’s highlights: the View editors on how Germany gained from the euro and the problems with the Facebook IPO; Clive Crook on Europe at the brink; Jonathan Alter on political substance and slander; Ezra Klein on the fight over Bain; Caroline Baum on overregulating banks; Panagis Vourloumis on Greek shock therapy; Junheng Li on China’s economic misinformation.
To contact the writer of this article: Tobias J. Moskowitz at email@example.com
To contact the editor responsible for this article: Max Berley at firstname.lastname@example.org