Public Statements & Remarks

Remarks of Chairman J. Christopher Giancarlo to the Federal Reserve Board of New York Third Annual Conference on the Evolving Structure of the U.S. Treasury Market

“The Importance of Large Trade Size Liquidity in U.S. Financial Markets”

November 28, 2017

Thank you. Good afternoon.

Before I begin, I want to thank President Bill Dudley, for his service, his counsel and his strong leadership of the New York Fed.

Thanks, Bill. We all have been fortunate to have you at the helm during these challenging times. We owe you a debt of gratitude.

Introduction

Speaking of captains at the helm, not far from here in 1524, Giovanni da Verrazzano became the first European to discover New York harbor. Significantly, he recognized the possibilities in exporting minerals, wood and furs from the new world to the old.

Approximately one hundred years later, Henry Hudson rediscovered this area, and settlers followed. Many came from the Netherlands seeking to profit from the same minerals wood and furs – the trading of commodities. They initiated a chain of events that bring us here today.

Why New York? Why here? One reason is people were willing to supply trading liquidity. That liquidity underpinned markets and established trade. One early Dutch resident of the city wrote about those who chose the “New Netherlands” over other colonies because, in his words, “…many more commodities were easier to obtain here than there.”1 Liquidity of supply here in Manhattan made trade possible back in Europe.

This part of New York’s history could be told through liquidity and the trading of commodity and financial contracts, which expanded at the turn of the twentieth century to include the hedging markets that the CFTC supervises.

For clarity, let me define an important term I just used. I define “liquidity” in the trading markets as the degree to which financial assets may be easily bought or sold with minimal price impact by ready and willing buyers and sellers. Just as such trading liquidity enabled Lower Manhattan to quickly become the epicenter of New World finance, it remains today the most important gauge of the efficiency in markets.

Yet even though trading liquidity is a timeless characteristic of healthy market activity, its nature is ever-changing. Today’s liquidity is vastly different than the liquidity of yesteryear, in large part thanks to technology.

It is also different than the liquidity of recent experience. Ten years ago was the onset of the financial crisis of 2007-2009, and five years ago marked the beginning of the implementation of the major banking and trading reforms in response to the crisis. Liquidity has changed qualitatively since those two reference points, for both better and worse.

Many observe that higher risk-based capital requirements have made it less desirable for banks to hold inventory and make markets in riskier asset classes, like corporate bonds. Some observe that the leverage ratio has made it less desirable to hold inventory and make markets in less risky asset classes, such as Treasury bonds. Still others are concerned over the extent to which regulatory changes have hurt liquidity.

And yet it is probably time to end the somewhat stale debate over whether trading liquidity is better or worse than during the years leading up to the crisis. Who can say whether those conditions were appropriate in light of the systemic instability that followed?

The debate we should have now is whether trading liquidity is optimal for today’s markets. We should ask whether we have achieved a policy mix that provides contemporary markets with the optimal balance of healthy trading liquidity and systemic risk reduction.

To help us with that analysis, today I would like to explore the current quality of liquidity in one important area that the CFTC regulates: the Treasury futures market. I will address current liquidity in the Treasury Futures and its measurement, rather than look at changes in that liquidity over time.

The trading liquidity of U.S. Treasuries and their associated futures contracts is of particular interest: Treasuries are a benchmark for other U.S. and world markets, and the U.S. government’s ability to fund its debt efficiently and consistently is of vital national interest.

The structure of the US treasury futures market is highly sophisticated. It encompasses participants that play important and specific roles as liquidity providers. Broker-dealers traditionally serve as market-makers to their customers. They help buyers and sellers transact by trading continuously with both sides of the market. Principal trading firms do not serve customers, but are significant traders in the Treasury markets with their reliance on technology to underpin their trading practices. We must appreciate the mix of participants in these markets in order to assess the changing nature of trading liquidity, its qualify and quantity.

My presentation will concentrate on one particular issue with respect to the definition and measurement of liquidity in Treasury futures contracts: large-trade size.

Why large trade size? Two reasons:

First, one of the main complaints about current liquidity conditions is that it is difficult to trade large positions in various fixed income markets.

Second, large-trade size is of particular concern to Treasury market structure because that is the way primary dealers have traditionally participated in the Treasury markets.

As I have already mentioned, dealers have always been an important part of the modern diverse trading market. Yet their influence has ebbed since the crisis and the ensuring regulatory reforms.

Markets with limited institutional dealer participation, like many in China and other developing economies, are often characterized by small trade size and high levels of price and trading volatility. And yet, well-developed and sophisticated financial markets, such as those in the US, have traditionally afforded dealers and other institutional participants the ability to efficiently execute large size transactions.

Our goal must be to avoid a one-size-fits-all approach to complex trading markets for financial and derivatives products.

Too often in the wake of the financial crisis, financial market policy makers here and abroad have promoted policy prescriptions that support retail market participation at the expense of institutional participation.  Undoubtedly, retail participation in US financial markets is important.  Yet, if the US financial and derivatives markets are to remain the world’s deepest and most liquid, they must clearly reflect a broad range of trading interests, market position and trading sophistication.  That includes assuring their ability to efficiently serve large institutional participants trading large size order without unnecessarily increased cost and/or volatility.

Today, I will provide some empirical evidence on the current quality of trading liquidity and then give a short review on developments in the relevant academic literature. I’ll then introduce some work-in-progress at the CFTC on measuring what I’ll call “large-trade liquidity” in the U.S. Treasury futures markets.

Background Reviews

By way of background, with respect to whether liquidity has been deteriorating since the financial crisis, the main empirical findings can be summarized as follows:

  • Bid-ask spreads, trading volume, and issuance do not indicate significant deterioration of trading liquidity;2
  • Market depth and turnover provide some evidence of deteriorating trading liquidity, particularly in less liquid markets (e.g., off-the-run U.S. Treasuries, corporate bonds);3
  • The flash rally in U.S. Treasuries in October, 2014, has raised questions about the potential fragility of liquidity. Market participants were able to trade throughout that brief episode, but only in limited quantities, while interest rate swaps experienced outsized price swings on no consequential news.4
  • Anecdotal evidence from market participants describes increasing difficulty to conduct large trades.5
  • Dealers have reported declines in corporate and foreign bond inventories and a reduced share of the single-name credit default swap market, suggesting less supply for large trades.6

These findings rely on the most common measures of liquidity, which characterize market averages rather than individual trades. The bid-ask spread, for example, is clearly indicative of the cost of trading, but the realized cost of particular trades or sequences of trades might be quite different. More precisely, a set of trades might move the mid-point of the market, might widen or narrow the bid-ask spread, or both.7

Academic Papers

A number of academic papers, therefore, rightly consider liquidity in terms of price impact and try to incorporate in trading costs the extent to which transactions or sequences of transactions move the prevailing market price.8

The price impact literature, however, typically does not address the execution of large trades,9 which is an important part of the debate on current liquidity conditions. But it is widely known and understood that traders now break down large orders, or “parent orders,” into “child orders,” which, in turn, are executed through numerous transactions with different counterparties.10 And there is a developed science about how to efficiently break down parent orders.

In one strategy, for example, execution begins with a large chunk of a total order so as to attract market interest. Execution then proceeds in evenly-sized pieces, but culminates in a final, large chunk so as to take advantage of the liquidity that has appeared.11 There are other similarly elaborate large trade execution tactics.

The reason that most studies relying on publicly available data cannot say anything about the liquidity of large trades is that their data sets have details only on individual executions: there is little information about which executions are tied to which child and which parent orders.

Data at the CFTC on Treasury futures, however, when combined with some strategic assumptions, does allow for analysis of large transaction liquidity. I want to share that information with you.

The Study

The CFTC has Treasury futures market data on child orders – all executions are connected with various identifiers to a single child order. While we do not have data on how child orders connect to a large parent order, we do have data connecting each child order to a customer. Therefore, we make the following assumption: all child orders entered by the same customer within one hour, exclusively to buy or exclusively to sell, are deemed to be part of the same parent order.12

With that assumption, we can identify parent orders and calculate the price impact cost of executing each parent order. More precisely, the price impact cost of the parent order is defined as the volume-weighted execution price of the entire parent order minus the average trade price over the minute before the submission of the first child order.

Since we are interested in the price impact of large orders, we can compare the average price impact for parent orders of relatively large size with those of relatively small size. While perfectly reasonable, this comparison alone wouldn’t fully capture an important aspect of trading in large size, namely, the volatility of the price impact.

For example, perhaps a trader wants to sell 5000 Treasury note contracts, which is a large order. The trader will certainly break this trade up into several smaller child orders, as mentioned earlier, but would then also need to decide how quickly to execute those child orders. Relatively quick execution would cost more in terms of price impact, but would run less risk that the market price would move against the trader, which in this case, is down. Relatively slow execution of the child orders, on the other hand, would minimize price impact costs but would subject the trader to significant risk that prices will fall over the course of the relatively slow execution of the child orders. This is the very definition of the volatility of price impact.

In general terms, then, the quality of liquidity of large trades is a function of both the average price impact and the volatility of the price impact. In a very liquid market, executing large parent orders do not cost a lot in terms of price impact nor do they risk volatile results in terms of price impact. By contrast, in a less liquid market, large trades have both large and volatile price impact costs.13

Preliminary Findings

The point is that standard measures of trading liquidity, like bid-ask spreads and market depth, are not very useful in considering an issue of contemporary concern, namely, “large-trade liquidity,” the ability to buy or sell in large size without moving markets.

Now, here are some of the findings for the front ten-year note futures contract from July to August, 2017:14

Here are two graphs based on data from July and August of 201715. (The asterisks are data points for size-buckets of parent orders and the lines/curves are rough numerical fits through the data.)

Graph 1. Data on 10-Year Treasury Note Futures contracts. Average price impact increases with parent order size, as expected (blue line). The standard deviation of the price impact also increases with size (red line), and is also quite large: execution is risky.

In other words, in thinking about large-trade liquidity, parent order size matters both with respect to average price impact and the risk around that average.

As commonly used in the literature, just saying that half the bid-ask spread for the note contract is between 1 and 1.5 cents16 does not tell the full story. Saying that market depth (top three price levels) is easily over 5,000 contracts17 similarly does not tell the full story. Rather, the story is more fully told in looking at both the average price impact and the volatility of the prince impact incurred by the execution of a large trade.

Graph 2 makes this point in another way, with data on both the 10-Year Treasury Note Futures and the Treasury Bond Futures contracts. As the size of the parent order increases, the time to execution increases, which means the price risk of the execution increases.

The large-trade liquidity of the bond contract is clearly worse than of the note contract.

Next steps are to continue the analysis and potentially be able to say something about how large-trade liquidity has evolved over time.

To repeat:

    Average price impact increases with order size.

    The time required to complete an order increases with order size.

    The standard deviation of the price impact across individual orders also increases with order size.

    The volatilities of price impact are very large relative to the average price impact. For orders less than 500 contracts, the volatility is about 13 times the absolute average price impact, while for orders greater than 3,500 contracts, the volatility is close to 4 times the average. In plain terms, execution risk has likely increased, and this, in turn, implies transaction costs have also gone up.

In short, these numbers allow us to quantify how the transaction cost, measured in terms of average price impact, increases as a function of order size.

Conclusion

In drawing to a close, it is clear that large trade liquidity is different than small trade or retail trade liquidity. And, it is quite clear that large trade liquidity matters for a markets that are as large, deep and sophisticated as U.S. Treasury futures.

Preliminary analysis in the context of U.S. Treasury futures indicates that large-trade liquidity—measured by the average price impact and the standard deviation of the price impact of large trades—is not as abundant as what bid-ask spreads and market depth alone suggest. It is deserving of further study. I thank the CFTC Office of Chief Economist, especially, Bruce Tuckman, Esen Onur and Eleni Gousgounis, for their excellent initial work.

The next step in this research program is to complete the relevant time series analysis and draw inferences about trends in large-trade liquidity in U.S. Treasury futures. Data available at the CFTC can be used to conduct similar analyses with respect to other futures and swap markets.

It is true that worries about large trade liquidity are not centered solely on U.S. Treasury futures markets or even other futures markets. Furthermore, particular impact of the leverage ratio on low-risk cash and repo markets may have pushed trading towards toward derivatives markets and away from cash markets.18 But given data availability, these futures markets are a good place to develop measurements of large-trade liquidity.

I recommend that market participants consider engaging in such studies. The more we collectively learn about trading liquidity – especially liquidity supporting large trade sizes - the better job we as regulators and policymakers can do of keeping up with evolving market structures and making sure that they serve the economic needs of society.

The ability to efficiently execute large size transactions is critically important to US financial markets.  If the US financial and derivatives markets are to remain the world’s deepest and most liquid, they must clearly reflect a broad range of trading interests, market position and trading sophistication.  That includes assuring the ability of US markets to efficiently serve large institutional participants trading large size order without unnecessarily increased cost and/or volatility.

The time has come to move beyond the debate over whether trading liquidity is better or worse than during the years leading up to the financial crisis. The debate should be about whether trading liquidity is optimal for today’s markets. It is time to ask whether we have achieved a policy mix that provides a proper balance of systemic risk reduction and healthy trading liquidity for the most diverse range of market participants in support of strong economic growth and broad-based prosperity.

I began by discussing the early history of this great city, built layer by layer. In the last century, the Second World War was a turning point. After the European conflict many people thought of New York City as “the capital of the world.” The City had clearly become the international capital of finance, fashion, and the arts. I agree with one writer of that time, who said “New York is … the promised land.”

That is why we have gathered here. In our new world of technological finance, in a world of ever-changing trading liquidity, this moment is another notable turning point. We see history written with every day, new challenges and opportunities propelling us forward. In many ways, we are like Captain Verrazzano, the first discoverer of New York. We encounter a new world, a world of challenges, but also of rich opportunity and vast potential. Thank you for inviting me to share my view of it.

Thank you.

References

Adrian, T., Fleming, M., Shachar, O., and Vogt, E. (2015), “Has U.S. Corporate Bond Market Liquidity Deteriorated?” Liberty Street Economics, October 5.

Adrian, T., Fleming, M., Shachar, O., and Vogt, E. (2017), “Market Liquidity after the Financial Crisis,” Annual Reviews of Financial Economics, Volume 9, pp. 43-83, November.

Adrian, T., Fleming, M., and Vogt, E. (2017), “An Index of Treasury Market Liquidity,” Federal Reserve Bank of New York Staff Reports, No 827, October 2017.

Adrian, T., Fleming, M., Vogt, E., and Wojtowicz, Z. (2016), “Corporate Bond Market Liquidity Redux: More Price-Based Evidence,” Liberty Street Economics, February 9.

Almgren, R., Thum, C., Hauptmann, E., and Li, H. (2005), “Direct Estimation of Equity Market Impact,” May 10, working paper.

Anbinder, T. (2016). City of Dreams: The 400-Year epic History of Immigrant New York, New York: Houghton Mifflin Harcourt.

Bao, J., O’Hara, M., and Zhou, X. (2016), “The Volcker Rule and Market-Making in Times of Stress,” working paper, September.

Bessembinder, H., Jacobsen, S., Maxwell, W., and Venkataraman, K. (2016), “Capital Commitment and Illiquidity in Corporate Bonds,” working paper, July.

Bessembinder, H., and Venkataraman, K. (2010), “Bid-Ask Spreads: Measuring Trade Execution Costs in Financial Markets, Encyclopedia of Quantitative Finance, Wiley.

BIS (2016), “Fixed Income Market Liquidity,” January.

Blackrock (2015), “Addressing Market Liquidity,” July.

Blackrock (2016), “Addressing Market Liquidity: A Broader Perspective on Today’s Bond Markets,” February.

Committee on Capital Markets Regulation (2015), “Nothing but the Facts: U.S. Bond Market Liquidity,” December 2014.

Conrad, J., and Wahal, S. (2017), “The Term Structure of Liquidity Provision,” working paper, August 9.

CME Group (2016), “The New Treasury Market Paradigm,” June.

Degryse, H., de Jong, F., and van Kervel, V. (2015), “The Impact of Dark Trading and Visible Fragmentation on Market Quality,” Review of Finance, Volume 19(4), pp. 1587-1622, July.

Deutsche Bank (2016), “Searching for Liquidity,” March.

Dick-Nielsen, J., and Rossi, M. (2016), “The Cost of Immediacy for Corporate Bonds,” working paper, August 21.

DTCC (2016), “Trends and Risks in Bond Market Liquidity,” September.

Duffie, Darrell (2016), “Why Are Big Banks Offering Less Liquidity To Bond Markets?,” Forbes, March 11, 2016, available at https://www.forbes.com/sites/lbsbusinessstrategyreview/2016/03/11/why-are-big-banks-offering-less-liquidity-to-bond-markets/#4a169f7a29de.

Engle, R., Ferstenberg, R., and Russell, J. (2012), “Measuring and Modeling Execution Cost and Risk,” The Journal of Portfolio Management, Volume 38(2), pp. 14-28, Winter.

Fett, N., and Haynes, R., “Liquidity in Select Futures Markets,” CFTC, February 1.

Gabrielsen, A., Marzo, M., and Zagaglia, P. (2011), “Measuring Market Liquidity: An Introductory Survey,” working paper, December 30.

Gousgounis, E., and Onur E. (2017), “Is Pit Closure Costly for Customers? A Case of Livestock Futures,” working paper, 2017.

Goyenko, R., Holden, C., and Trzcinka, C. (2009), “Do Liquidity Measures Measure Liquidity?” Journal of Financial Economics, Volume 92, pp. 153-181.

International Monetary Fund (2015), “Global Financial Stability Report: Vulnerabilities, Legacies, and Policy Challenges,” October.

Joint Staff Report (2015), “The U.S. Treasury Market on October 15, 2014,” Department of the Treasury, Board of Governors of the Federal Reserve System, Federal Reserve Bank of New York, Securities and Exchange Commission, Commodity Futures Exchange Commission, July 13.

Marshall, B., Nguyen, N, and Visaltanachoti, N. (2012), “Commodity Liquidity Measurement and Transaction Costs,” Review of Financial Studies, Volume 25(2), pp. 599-638.

Mizrach, B. (2015), “Analysis of Corporate Bond Liquidity,” FINRA Office of the Chief Economist, Research Note.

Obizhaeva, A., and Wang J. (2013), “Optimal Trading Strategy and Supply/Demand Dynamics,” Journal of Financial Markets, Volume 16(1), pp. 1-32.

Oehmke, M., and Zawadowski, A. (2015), “The Anatomy of the CDS Market,” Columbia University.

Papanyan, S. (2015), “Heightened Bond Liquidity Risk is the New Normal,” BBVA Research, U.S. Economic Watch, September 3.

PWC (2015), “Global Financial Markets Liquidity Study,” August.

TBAC (2013), “Assessing Fixed Income Market Liquidity,” Presentation to Treasury Borrowing Advisory Committee (TBAC), July.

Trebbi, F., and Xiao, K. (2015), “Regulation and Market Liquidity,” NBER Working Paper No. 21739, November.

Van Kervel, V., and Menkveld, A. (2017), “High-Frequency Trading Around Large Institutional Orders,” working paper, September 27.

Whalen, R., and Scott, J. (2015), “Can the Credit Default Swap Market be Salvaged? Issues for Borrowers and Investors,” Kroll Bond Rating Agency.

Wood, D. (2015), “GFMA, IIF, ISDA Plan Liquidity Lobbying Push,” Risk.net, July 10.

1 Anbinder. City of Dreams: The 400-Year Epic History of Immigrant New York. New York: Houghton Mifflin Harcourt, 2017, 27.

2 Adrian, Fleming, Shachar, and Vogt (2015, 2017), Adrian, Fleming, Vogt, and Wojtowicz (2016), Adrian, Fleming, and Vogt (2017), Bessembinder, Jacobsen, Maxwell, and Venkataraman (2016), DTCC (2016), Joint Staff Report (2015), and Trebbi and Xiao (2015). TBAC (2013) notes that bid-ask spreads are “spiky,” so that liquidity as measured by bid-ask spreads may not be consistently available.

3 DTCC (2016), Papanyan (2015), Oehmke and Zawadowski (2015), PWC (2015), TBAC (2013), Whalen and Scott (2015).

4 See Joint Staff Report (2015).

5 Bao, O’Hara, and Zhou (2016), BIS (2016), Blackrock (2015,2016), Committee on Capital Markets Regulation (2015), Deutsche Bank (2016), Dick-Nielsen and Rossi (2016), Papanyan (2015), PWC (2015), and Wood (2015).

6 See Board of Governors of the Federal Reserve System. “Monetary Policy Report,” p. 27. 7 July 2017 & Staff of the Division of Economic Risk Analysis of the US Securities and Exchange Commission. “Report to Congress: Access to Capital and Market Liquidity,” p. 225. August 2017.

7 For a review of liquidity measurements in financial markets see Gabrielsen, Marzo, and Zagaglia (2011), Goyenko, Holden, and Trzcinka (2009), and Marshall, Nguyen, and Visaltanachoti (2012). Also, Fett and Haynes (2017) apply these various liquidity measures to futures markets.

8 See, for example, Bessembinder and Venkataraman (2010), Conrad and Wahal (2017), and Gousgounis and Onur (2017).

9 Notable exceptions are Almgren et al (2005) and Engle, Ferstenberg, and Russell (2012), which use proprietary data on parent orders and are closest in spirit to the work described in this talk.

10 See, for example, DTCC (2016), Mizrach (2015), Van Kervel and Menkveld (2017).

11 See, for example, Obizhaeva and Wang (2013) and Degryse et al. (2012).

12 The hour interval was chosen after feedback from traders. Nevertheless, this assumption might very well exclude some long-lasting orders that should have been included in the study.

13 Unfortunately, we do not have data on the execution strategies of individual customers. This means the average and standard deviation of realized price impacts in our sample is a blend across all customers, those that chose relatively fast executions (high average cost / low volatility of cost) and those that chose relatively slow executions (low average cost / high volatility of cost).

14 Spread trades were excluded.

15 We analyzed outright transactions on the lead-month contract only.

16 See Fett and Haynes (2017).

17 See Fett and Haynes (2017).

18 See, for example, CME Group (2016) and DTCC (2016). See also Duffie (2016).

Last Updated: November 28, 2017