Extract from a speech by Governor Powell at the Brookings Institution, Washington. The move to fully electronic trading raises important questions about the benefits of fully automated high-speed trading which may lead to industry concentration and liquidity fracturing as the arms-race continues. So it is a good time for market participants and regulators to collectively consider whether current market structures can be improved for the benefit of all.
Treasury markets have undergone important changes over the years. The footprints of the major dealers, who have long played the role of market makers, are in several respects smaller than they were in the pre-crisis period. Dealers cite a number of reasons for this change, including reductions in their own risk appetite and the effects of post-crisis regulations. At the same time, the Federal Reserve and foreign owners (about half of which are foreign central banks) have increased their ownership to over two-thirds of outstanding Treasuries (up from 61 percent in 2004). Banks have also increased their holdings of Treasuries to meet HQLA requirements. These holdings are less likely to turn over in secondary market trading, as the owners largely follow buy and hold strategies. Another change is the increased presence of asset managers, which now hold a bigger share of Treasuries as well. Mutual fund investors, who are accustomed to daily liquidity, now beneficially own a greater share of Treasuries.
Perhaps the most fundamental change in these markets is the move to electronic trading, which began in earnest about 15 years ago. It is hard to overstate the transformation in these markets. Only two decades ago, the dealers who participated in primary Treasury auctions had to send representatives, in person, to the offices of the Federal Reserve Bank of New York to submit their bids on auction days. They dropped their paper bids into a box. The secondary market was a bit more advanced. There were electronic systems for posting interdealer quotes in the cash market, and the Globex platform had been introduced for futures. Still, most interdealer trades were conducted over the phone and futures trading was primarily conducted in the open pit.
Today these markets are almost fully electronic. Interdealer trading in the cash Treasury market is conducted over electronic trading platforms. Thanks to advances in telecommunications and computing, the speed of trading has increased at least a million-fold. Advances in computing and faster access to trading platforms have also allowed new types of firms and trading strategies to enter the market. Algorithmic and high-frequency trading firms deploy a wide and diverse range of strategies. In particular, the technologies and strategies that people associate with high frequency trading are also regularly employed by broker-dealers, hedge funds, and even individual investors. Compared with the speed of trading 20 years ago, anyone can trade at high frequencies today, and so, to me, this transformation is more about technology than any one particular type of firm.
Given all these changes, we need to have a more nuanced discussion as to the state of the markets. Are there important market failures that are not likely to self-correct? If so, what are the causes, and what are the costs and benefits of potential market-led or regulatory responses?
Some observers point to post-crisis regulation as a key factor driving any decline or change in the nature of liquidity. Although regulation had little to do with the events of October 15, I would agree that it may be one factor driving recent changes in market making. Requiring that banks hold much higher capital and liquidity and rely less on wholesale short-term debt has raised funding costs. Regulation has also raised the cost of funding inventories through repurchase agreement (repo markets). Thus, regulation may have made market making less attractive to banks. But these same regulations have also materially lowered banks’ probabilities of default and the chances of another financial crisis like the last one, which severely constrained liquidity and did so much damage to our economy. These regulations are new, and we should be willing to learn from experience, but their basic goals–to make the core of the financial system safer and reduce systemic risk–are appropriate, and we should be prepared to accept some increase in the cost of market making in order to meet those goals.
Regulation is only one of the factors–and clearly not the dominant one–behind the evolution in market making. As we have seen, markets were undergoing dramatic change long before the financial crisis. Technological change has allowed new types of trading firms to act as market makers for a large and growing share of transactions, not just in equity and foreign exchange markets but also in Treasury markets. As traditional dealers have lost market share, one way they have sought to remain competitive is by attempting to internalize their customer trades–essentially trying to create their own markets by finding matches between their customers who are seeking to buy and sell. Internalization allows these firms to capture more of the bid-ask spread, but it may also reduce liquidity in the public market. At the same time it does not eliminate the need for a public market, where price discovery mainly occurs, as dealers must place the orders that they cannot internalize into that market.
While the changes I’ve just discussed are unlikely to go away, I believe that markets will adapt to them over time. In the meantime, we have a responsibility to make sure that market and regulatory incentives appropriately encourage an evolution that will sustain market liquidity and functioning.
In thinking about market incentives, one observer has noted that trading rules and structures have grown to matter crucially as trading speeds have increased–in her words, “At very fast speeds, only the [market] microstructure matters. Trading algorithms are, after all, simply a set of rules, and they will necessarily interact with and optimize against the rules of the trading platforms they operate on. If trading is at nanoseconds, there won’t be a lot of “fundamental” news to trade on or much time to formulate views about the long-run value of an asset; instead, trading at these speeds can become a game played against order books and the market rules. We can complain about certain trading practices in this new environment, but if the market is structured to incentivize those practices, then why should we be surprised if they occur?
The trading platforms in both the interdealer cash and futures markets are based on a central limit order book, in which quotes are executed based on price and the order they are posted. A central limit order book provides for continuous trading, but it also provides incentives to be the fastest. A trader that is faster than the others in the market will be able to post and remove orders in reaction to changes in the order book before others can do so, earning profits by hitting out-of-date quotes and avoiding losses by making sure that the trader’s own quotes are up to date.
Technology and greater competition have led to lower costs in many areas of our economy. At the same time, slower traders may be put at a disadvantage in this environment, which could cause them to withdraw from markets or seek other venues, thus fracturing liquidity. And one can certainly question how socially useful it is to build optic fiber or microwave networks just to trade at microseconds or nanoseconds rather than milliseconds. The cost of these technologies, among other factors, may also be driving greater concentration in markets, which could threaten their resilience. The type of internalization now done by dealers is only really profitable if done on a large scale, and that too has led to greater market concentration.
A number of observers have suggested reforms for consideration. For example, some recent commentators propose frequent batch auctions as an alternative to the central limit order book, and argue that this would lead to greater market liquidity. Others have argued that current market structures may lead to greater volatility, and suggested possible alterations designed to improve the situation. To be clear, I am not embracing any particular one of these ideas. Rather, I am suggesting that now is a good time for market participants and regulators to collectively consider whether current market structures can be improved for the benefit of all.
