Research output (as part of the European FOC project):
High-speed computerized trading, often called high-frequency trading (hereafter HFT), has increased dramatically in financial markets over the last decade. It currently accounts for 55% of trading volume in US equity markets, 40% in European equity markets, and is quickly growing in Asian, fixed income, commodity, foreign exchange, and nearly every other market (see chart 5 here).
Although evidence suggests that HFT increases market efficiency, there are serious concerns that HFT firms (1) contribute to market instability (see the flash crash and also Nanex research), (2) possess an unfair speed advantage over other investors, and (3) syphon money from markets with no added social benefit. Because of these concerns, policy makers worldwide are spending considerable effort deciding if and how to regulate HFT.
My research addresses several important, unanswered questions regarding HFT:
What purpose does HFT serve in markets? How does it increase market efficiency, and why does it lead to market instability? Why is HFT so prevalent? Why does it concentrate in certain securities yet ignore others? How can HFT simultaneously make profits yet reduce transaction costs, and are microsecond speeds really necessary in markets?
First, what purpose does HFT serve, i.e., what role does it play in markets? To answer this question, it is important to first point out the dramatic changes in markets over the last 15 years. Whereas before most trading was human mediated, it is now almost completely automated. Nearly all of the influential trading pits and trading floors of the 20th century have closed and are now replaced by fully electronic markets. In these electronic markets, many HFT firms play a very important role as the new, automated liquidity providers -- replacing traditional liquidity providers who previously "made markets."
Second, why does HFT increase market efficiency but decrease market stability? When a process is automated, we normally expect there to be gains in efficiency, and automating liquidity provision in markets should be no different (see below for a more precise argument based on price synchronization). We also should not be surprised to observe decreased stability. Machines are notoriously bad at decision-making when faced with novel circumstances, and therefore can exacerbate conditions during times of market stress. During the flash crash, many HFT firms simply shut down their systems.
Third, why is HFT so prevalent ... why has it replaced traditional market makers operating in trading pits and on trading floors? The quick answer is that HFT firms can process more information than humans when determining the values of securities. As a result, they set better prices in markets and transact the majority of investor order flow. Traditional market makers simply cannot compete and are squeezed out of the market (see Gerig and Michayluk (2013) for full details).
Fourth, why does HFT concentrate in certain securities yet ignore others, how does it make profits yet reduce transaction costs, and are microsecond speeds really necessary?
All of these questions can be answered once it is realized that HFT's main contribution to markets -- over-and-above what liquidity providers in trading pits and on trading floors could accomplish -- is a proper synchronization of prices (see the video on the right).
Financial securities are often very closely related to one another, and their prices simply cannot be kept aligned without the use of machines (there are over 1000 transactions per second in US equities alone during the trading day). Indeed, a decade ago it took several minutes for prices to fully synchronize, whereas today they do so almost instantaneously (see Fig. 2A in Gerig (2013)).
To understand the effects of price synchronization in markets, it is useful to draw an analogy to animal groups. Animals that move together in groups scan their environment with "many eyes", which allows them to quickly evade threats or find potential food sources. Financial markets are similar. In markets, the state of the economy is monitored by a large number of investors who quickly broadcast any changes to each other and the rest of society via price movements. In both settings, synchronization is necessary for "many eyes" to function properly.
Below, the video on the left shows the price movements of 40 large-cap US stocks in 1-minute intervals on Feb. 24, 2010. Movements to the right are increases in prices and to the left are decreases in prices. Notice how price fluctuations strongly resemble the motions of schooling fish (see Fig. 5 in Gerig (2013)).
By synchronizing prices, HFT allows the "many eyes" of different investors to function as one coherent group. Analogous to the effects observed in animal groups, synchronization in financial markets facilitates information transfer between investors, increases efficiency, and allows fewer resources to be spent on informed individuals (see Gerig (2013) for details). Animals, in fact, increase synchronization when threatened so that these effects are enhanced.
Because HFT keeps the prices of related securities aligned, it is not surprising that HFT is very active in those securities that have strong economic overlap and that are closely related to one another -- e.g., the stock of large corporations, ETF's, futures contracts on indices, etc. -- and is relatively inactive in the stock of small corporations. Price synchronization, therefore, explains why HFT concentrates in certain securities and ignores others (see Fig. 4A in Gerig (2013)).
How can HFT lower transaction costs yet still make profits? First, HFT might simply extract less money from markets than the human liquidity providers they have replaced. Second, when prices are synchronized, information diffuses rapidly from security to security and informed investors who trade different but related securities are forced to compete with one another. They make less profit as a result. HFT is able to keep a portion of this extra money, but because HFT is competitive, the rest is redistributed to average investors in the form of reduced transaction costs (see Gerig (2013) for details).
Finally, are microsecond speeds really necessary in markets? Although not necessary, a strong case can be made for markets to operate at these speeds (see Fricke and Gerig (2013)). In order for prices to be as accurate as possible, they must update at the same speed that information arrives in the market. Because investors, in aggregate, transact thousands of times per second in markets (with each transaction potentially containing information about every security in the market), the prices of securities become
stale on the order of microseconds to milliseconds and must be updated at these speeds for transactions to be priced correctly.
Although synchronization has several beneficial effects, it is not a panacea for markets. When prices are tightly connected to each other, errors will quickly propagate through the financial system if safeguards are not in place. In addition, if shared misconceptions exist among investors, they are amplified by synchronization so that prices are less accurate overall (see lemmings on right). Finally, synchronization, if implemented incorrectly, can create spurious structure in markets so that prices are more (or less) correlated than economic fundamentals warrant.