I appreciate the opportunity to discuss this important topic with such accomplished panelists.1 I would like to share some of what the New York Fed's Open Market Trading Desk (the Desk) learned implementing market functioning purchases in the Treasury and agency MBS markets in the spring of 2020. Before going further, I would like to note that these remarks reflect my views and not necessarily those of the Federal Reserve Bank of New York or the Federal Reserve System.
Let me start with a brief background on the conditions that led the Federal Reserve to launch market functioning purchases. The emergence of the pandemic in early 2020 brought about an extreme demand for liquidity—a "dash for cash"—due to uncertainty in the economic outlook. U.S. Treasury securities are held by many investors for their money-like properties, and the Treasury market was the recipient of massive waves of selling as these investors sought to convert their holdings to cash. The sales quickly surpassed end-user purchase demand and the capacity of intermediaries to warehouse the imbalance, leading to the breakdown of two-way markets and strains that spilled over into other related markets, including agency MBS. In response, the Federal Open Market Committee (FOMC) took extraordinary actions to address the market dysfunction, first by directing the Desk to increase the daily offerings of overnight repurchase agreement transactions (repos) and offering new term repos. The temporary financing provided by these operations, while helpful, could not meet the permanent liquidity needs of all investors, and soon after the Desk was directed to purchase Treasury and agency MBS securities to support market functioning at what became an unprecedented pace.2
Today I will share three main lessons I took away from our experience implementing market functioning purchases in March 2020: first, the importance of being able to adapt purchases to changing conditions; second, the use of the purchases themselves to assess market pricing; and third, the importance of continuing to pursue regulatory and market structure initiatives that support resilience in government debt markets.
Real-time Radar: Adapting Purchases to Changing Indicators of Market Functioning
In periods of sharp market dysfunction, policymakers inevitably find themselves making rapid decisions with imperfect information. The first lesson I take from our experience is that adapting market functioning purchases in response to real-time indicators of market functioning, including market intelligence from participants and the signals from the operational results themselves, is essential to their effectiveness.
At the start of the pandemic, a variety of market indicators suggested markets were becoming increasingly illiquid. We observed selling pressure escalate dramatically through analysis of Treasury transaction data (TRACE), other flow data, and reports from market participants. Primary dealers, whose holdings were already at elevated levels, had limited capacity to intermediate the massive flows and were faced with increased risk as volatility spiked. Bid-ask spreads widened dramatically, particularly for deep off-the-run notes and bonds.3 However, our visibility into some of these important trends was limited, in part due to the lack of real-time and comprehensive cash market data on positions and leverage, in areas like the uncleared bilateral repo market.4
Given these factors, the Desk initially targeted a broad range of Treasury securities for purchase, announcing plans to spread all of our previously planned $80 billion in monthly purchases across the curve. Although the purchases were scheduled to occur over the course of mid-March to mid-April, it soon became apparent that the demand for immediate "cash liquidity" necessitated a more rapid response, and we subsequently front-loaded $37 billion of those purchases to one day (Friday the 13th). Those operations received over $100 billion in offers from primary dealers, indicating an enormous and immediate demand for liquidity. On March 15, the FOMC announced large-scale purchases of both Treasury securities and agency MBS, followed by an announcement on March 23 where the purchase amounts were expanded to "the amounts needed to support smooth market functioning and effective transmission of monetary policy to broader financial conditions." Daily Treasury purchases alone reached a peak pace of $75 billion, equivalent to the highest pace of purchases we had ever previously conducted on a monthly basis.5 By the end of the first four weeks of purchases, the Desk had bought over $1 trillion in Treasury securities and nearly half a trillion dollars in agency MBS.6
In addition to adjustments to the timing and amounts of purchases, the Desk frequently adapted the specific parameters of the purchase operations. At the end of each purchase day, the Desk assessed the outcome of the purchase operations, the status of market indicators for liquidity and relative value, data on positioning and flows, and the market intelligence we gathered, and used these variables to adjust our plans for coming purchases. For example, we typically do not purchase the cheapest-to-deliver securities for active Treasury futures contracts to avoid exacerbating scarcity in those issues. However, given the large unwinds that were occurring in the cash-futures basis trade at the time, we chose to include these securities in our operations to help absorb those sales. In the MBS market, we purchased securities for near-term settlement to help reduce pressure on dealer balance sheets, a deviation from our typical strategy of purchasing for forward settlement.
Of course, having access to better data would be beneficial in making these judgments, and there are many critical efforts underway to enhance data availability.7 But even with the best data, designing such purchases in a crisis will require judgment and adaptability, and use of real-time indicators, operational results, and market intelligence will remain critical for helping inform those decisions.
Making Sure the Price is Right: Using Operations to Assess Market Prices
Dysfunctional markets pose a particular problem for central banks as they evaluate the appropriate price for the securities they purchase: in the absence of liquid, two-way markets, market pricing sources may not be reliable. A second lesson from the Desk's pandemic purchases is that, to get the price right, the purchase operations themselves provide important information.
Historically, the Desk has chosen to purchase securities by conducting large-scale auctions, with offers on particular securities selected on the basis of their offered price relative to prevailing market prices and, in the case of Treasury purchases, measures of theoretical value according to our proprietary model. For Treasury purchases, this approach generally worked well during the pandemic, with market measures of prices tracking fairly closely to prices received in our operations. From March through June of 2020, during the height of the market dysfunction, nearly all of the Desk's Treasury purchases were executed cheap to market prices.
By contrast, agency MBS market price sources were more volatile. In the Treasury market, high-speed electronic trading is widely used for trading the most recently issued securities, while most trading in agency MBS is done via request-for-quote (RFQ) systems where dealers and customers transact at slower speeds. The major agency MBS trading platforms provide approximate price levels at which a trade is likely to occur. In normal conditions, these "indicative" prices closely approximate the actual bid and offer levels at which market trades are conducted. However, at the height of the dysfunction, the offered prices we received in our operations diverged significantly from these indicative prices —at times by more than one-quarter of a percentage point, either higher or lower than what might have been expected. Given the size of our operations relative to overall market volumes, Fed purchase operations became important opportunities to identify prevailing market prices for agency MBS.
As a result, we developed a variety of methods for using the prices offered to us in the operations to establish an effective market price for each security, which aided us in selecting which securities were being valued as cheap by the market and should be targeted for purchase. For example, one measure of pricing used was the second-best price we received on a security from a dealer other than the dealer that offered the best price (a measure commonly known in the market as a "cover").8 As two-way market activity picked up, we continued to adjust these approaches and integrated market price feeds back into the evaluation of security pricing.
Don't Forget the Dog Just Because It Didn't Bark
Finally, we should take some lessons from the "dogs that did not bark" during the pandemic. Despite the unprecedented flow of government securities in the market, we were fortunate to not see cracks in the market infrastructure—the financing, clearing, and settlement systems that undergird the markets for U.S. Treasuries and agency MBS. Some have used this as evidence that these processes need not be priorities in the work to enhance the resilience of the Treasury market. Let me be clear: that is not the lesson we should take.
We know that the financing, clearing, and settlement processes of the government securities markets can be more resilient.9 In recent years, the Treasury Market Practices Group (TMPG) has highlighted the risks in clearing and settlement in the Treasury market, finding that market participants may not be applying the same risk management rigor to the financing, clearing, and settlement of U.S. Treasury securities that they do in other markets.10 For example, the TMPG found that the share of the Treasury cash market that is centrally cleared has shrunk over the last two decades to under 25 percent , reducing the portion of the market that is subject to standard risk management processes like margin, default processes, and liquidity resource requirements.11 Similarly, the TMPG found that the large, uncleared bilateral repo market has inconsistent risk management practices, including zero haircuts. This is a practice that many counterparty risk management professionals would find questionable at best.12 More work should be done to ensure that these transactions effectively manage counterparty credit risk, and I am glad that both public and private sector groups have work planned in this area.13
Importantly, we do not know how the financing, clearing, and settlement of Treasuries and agency MBS would have fared had the Fed not offered unprecedented amounts of Treasury repurchase agreements and purchased extraordinary amounts of these securities. Had there been a default by a large market player, particularly in the uncleared market segment, it could have reverberated through the pipes of the system. It is essential to guard against both bilateral and systemic risks in financing, clearing and settlement processes, which can generate counterparty losses, fire sales, fails, and even possible risk to institutions that sit at the center of funding, clearing, and settlement for the U.S. Treasury and agency MBS markets.
The lesson is that we should not forget the dog simply because it did not bark. Efforts to bolster the resiliency of the U.S. Treasury and agency MBS markets—including the financing, clearing, and settlement mechanisms—are essential to reducing the frequency and severity of events like those of March 2020 and ensuring that the extraordinary interventions of central banks around the world remain extraordinarily rare. As part of the Inter-Agency Working Group on Treasury Market Surveillance, or IAWG, the New York Fed has been working steadily with counterparts across the official sector to advance policy proposals and research to support the fundamental soundness of the Treasury market.14 Finally, I invite you to join the IAWG's ninth annual U.S. Treasury Market Conference this fall, either virtually or in person. This forum has become an essential focal point for resiliency efforts.
Thank you again for welcoming me here today. I look forward to a productive panel discussion.
