High frequency trading has been raining a lot of eyebrows recently – especially since the Flash Crash. Were they responsible for the market nose dive (either directly, or by virtue of their sudden withdrawal from the market) or were they simply at the scene of the crime? Were quantitative hedge funds – cousins on the family tree of market participants – also involved? In a comprehensive piece in Institutional Investor last week (“Inside the Machine: A Journey into the World of High-Frequency Trading“), Executive Editor Michael Peltz tells a tale of secrecy, massive trading volumes, and time measured in microseconds.
One of the central questions in Peltz’ article (and in a recent book by the Wall Street Journal’s Scott Patterson called “The Quants: How a New Breed of Math Whizzed Conquered Wall Street and nearly Destroyed It“) is whether quant hedge funds are just market makers like the High Velocity relatives. High velocity trading is all about market making. And Most hedge funds hate to be called market makers. As Peltz writes:
“Hedge funds don’t like to be called high frequency traders, as I quickly discovered after visiting with some of the biggest quantitative managers, including AQR Capital Management in Greenwich, Connecticut, and D.E. Shaw & Co. and Renaissance Technologies Corp. in New York…”
“…Most say they use many of the same tools as high frequency traders — employing high-speed computer programs and co-location services to generate, route and execute orders — but that their strategies have a longer time horizon.”
Likewise, regular readers may recall Professor Harry Kat writing on these pages in February 2007 that:
“Some of the largest market maker firms in the world disguise themselves as hedge funds these days.”
Kat was talking about brand name hedge funds. The similarity is that, regardless of the investment horizon (milliseconds or days), the effect is the same. These investors are thought to provide liquidity when it is needed and extract is back when it is in oversupply. In such a way, argues the hedge fund industry, these traders have a self-righting influence on markets.
But is this really true? And if so, is it true for all hedge fund strategies? A new paper by Petri Jylha, Kalle Rinne, and Matti Suominen of Aalto University (a.k.a. the Helsinki School of Economics) attempts to answer this critical question.
Their first challenge was to come up with a metric to measure the extent to which a hedge fund generated returns simply by providing liquidity to the market. They start by examining the degree of short-term mean-reversion in stock returns. If individual stock prices revert to their mean quickly, they surmise that there is less liquidity. If stock prices revert to the mean very slowly, then they conclude there is more liquidity. So they develop a metric to track the extent of mean reversion.
They construct a naive trading strategy that you might call the “no brainer liquidity provider strategy” (our term). Then they see which, if any, hedge funds’ returns are correlated with those strategies. The strategies with the high correlation to the “no brainer liquidity provider strategy” are seen to be suppliers of liquidity while those with a negative correlation are seen to be demanders of liquidity.
They find that in aggregate, hedge funds don’t have a significant exposure to this liquidity factor. However, they do find that for all strategies, some hedge funds are statistically-significant liquidity suppliers and some are statistically-significant liquidity demanders. That suggests that there are different styles within each strategy.
The balance of liquidity providers and demanders was different for each strategy (see table below from paper – first three columns – ignore the other for now):
You can see that 15.7% of convertible arb funds have a positive correlation to liquidity-provision factor while 7.2% have a negative correlation. This suggests to the authors that more convert arb funds are liquidity suppliers than are liquidity demander. The intuition, they say, is simple. Convert arb funds go long a less liquid assets (converts) and short a more liquid assets (equities). Joining convert arb managers as more likely be to liquidity suppliers than demanders: market neutral, fixed income arb, and long-short equity.
On the flip-side, several strategies seemed to be home to more liquidity demanders than suppliers. Dedicated short bias, emerging markets funds of funds and multi-strat. If you’re involved with any of those strategies right now, you may be nodding in agreement (or not, let us know).
The trio also examined the exposure of each hedge fund strategy to liquidity shocks (second set of columns in the table above). No surprisingly, funds of funds were particularly vulnerable to liquidity shocks. But for all strategies, the sensitivity to liquidity shocks depended on the absolute level of liquidity before the shock. In their words:
“The funds’ exposure to liquidity shocks also decreases in the level of liquidity. The average coefficient of the interaction between liquidity shocks and level of liquidity is negative, indicating that hedge fund returns are less (more) sensitive to liquidity shocks during periods of high (low) liquidity.”
This kind of non-linear relationship is exactly what pundits warned against prior to the financial crisis. Models of liquidity shocks based on high liquidity environments apparently don’t apply to low liquidity environments.
Finally, the authors examined the effect of the 2003 decision by the NYSE to introduce automated quote dissemination. They find that the May 2003 implementation of this system has a major effect on the proportion of hedge funds supplying and demanding liquidity:
“Hedge funds in all categories, except for dedicated short bias and global macro, decrease their exposure to liquidity provision returns after the implementation of Autoquote. Also, the average coefficient of liquidity provision returns is very significantly negative after the automation whereas the pre-Autoquote average coefficient is very significantly positive. The overall proportion of liquidity demanding funds increases from 2.0% to 6.8% while the proportion of liquidity providing funds decreases from 13.1% to 0.7%.”
As policymakers examine the aggregate effect of hedge funds on liquidity, this study suggests they should look beyond blanket statements like “hedge funds provide liquidity” and dig a little deeper.
The authors conclude with a footnote that echo’s Professor Kat’s 2007 comments:
“These findings are consistent with the idea that in recent years only the largest and most efficient players (hedge funds, but also other types of institutions) have been supplying liquidity to the markets.”