The Global Dash for Cash: Why Sovereign Bond Market Functioning Varied across Jurisdictions in March 2020 Jordan Barone | Alain Chaboud | Adam Copeland | Cullen Kavoussi | Frank Keane | Seth Searls NO. 1 01 0 M ARCH 202 2 The Global Dash for Cash: Why Sovereign Bond Market Functioning Varied across Jurisdictions in March 2020 Jordan Barone, Alain Chaboud, Adam Copeland, Cullen Kavoussi, Frank Keane, and Seth Searls Federal Reserve Bank of New York Staff Reports, no. 1010 March 2022 JEL classification: G01, G12, E44, H63 Abstract As the economic disruptions associated with the COVID-19 pandemic increased in March 2020, there was a global dash-for-cash by investors. This selling pressure occurred across advanced sovereign bond markets and caused a deterioration in market functioning, leading to central bank interventions. We show that these market disruptions occurred disproportionately in the U.S. Treasury market and were due to investors selling pressures being far more pronounced and broad-based. Furthermore, we assess differences in key drivers of the market disruptions across sovereign bond markets, based on an analysis of the data as well as structured outreach to a range of market participants. Key words: sovereign bond markets, financial crisis, COVID-19 _ Barone, Copeland, Kavoussi, Keane, Searls: Federal Reserve Bank of New York (emails: jordan.baroneny.frb, adam.copelandny.frb, cullen.kavoussiny.frb, frank.keaneny.frb, seth.searlsny.frb). Chaboud: Federal Reserve Board (email: alain.p.chaboudfrb.gov). The authors thank Michael Fleming and Lorie Logan for comments and suggestions. This paper presents preliminary findings and is being distributed to economists and other interested readers solely to stimulate discussion and elicit comments. The views expressed in this paper are those of the author(s) and do not necessarily reflect the position of the Federal Reserve Bank of New York, the Board of Governors of the Federal Reserve System, or the Federal Reserve System. Any errors or omissions are the responsibility of the author(s). To view the authors disclosure statements, visit newyorkfed/research/staff_reports/sr1010.html. 2 In the early months of 2020, at the onset of the COVID-19 pandemic, heightened economic and market uncertainty led to a flight-to-quality as investors shifted their portfolios towards safe, sovereign bonds. As the shock intensified in the second week of March 2020, however, this flight-to-quality became a global dash-for-cash, as investors sought to sell sovereign bonds to meet redemptions and margin calls and to build cash buffers.2 These actions occurred across advanced economy sovereign bond markets, causing bond yields to spike and market functioning to deteriorate broadly and sharply, and prompting central banks to intervene via asset purchases to restore market functioning.3 This dash-for-cash occurred across advanced economy (AE) sovereign bond markets, though a range of metrics demonstrate that the March 2020 market disruptions occurred disproportionately in the U.S. Treasury market. To better understand this result, we assess differences in key drivers of the market disruptions across sovereign bond markets, based on an analysis of the data as well as structured outreach to a range of market participants. The first driver we consider is the depth and breadth of selling pressures across sovereign bond markets. The second driver is the difference in the buildup of leverage by investors leading up to the March 2020 shock. The third and final driver is the differences in market microstructure across sovereign bond markets, including market-maker obligations, the prevalence of central clearing, and reliance on electronic or voice trading platforms. We find that a main driver of the disproportionate disruptions to the Treasury market is that selling pressures were far more pronounced and broad-based in U.S. Treasuries than in other sovereign bond markets, reflecting the U.S. dollars role as the dominant global investment and funding currency. Differences in leverage dynamics also played a major role in explaining why the Treasury market faced larger disruptions to market functioning. Stronger pre-pandemic Treasury issuance, as well as supportive financing conditions and other factors, helped pave the way for a heavier build-up of leverage in the Treasury market than in other sovereign bond markets. As a result, the COVID-19 shock catalyzed more de-leveraging, and hence higher selling pressure, in the Treasury market. Finally, despite a number of differences in the market microstructure across sovereign bond markets, we conclude these differences were not primary drivers of the disproportionate disruptions to the Treasury market in March 2020. Although not the focus of this article, we note that in response to the global dash for cash, a number of central banks intervened in their respective sovereign bond markets by conducting asset purchases. In line with our results that disruptions to the Treasury market were more severe relative to other sovereign bond markets, the Federal Reserves response was larger and more front-loaded relative to 2 Fleming, Liu, Podjasek, and Schurmeier (2021) provides an analysis of this shift from flight-to-quality towards a demand for larger cash buffers in the U.S. 3 A burst of recent work has focused on analyzing Treasury market conditions during the COVID-19 shock, including Duffie (2020), Fleming and Ruele (2020), and Schrimpf, Shin and Sushko (2020). Further, Haddad, Moreira, and Muir (2020), and Kargar et al (2020) focus on how the COVID-19 shock effected the U.S. corporate bond market. 3 other central bank counterparties, such as the European Central Bank, the Bank of England, and the Bank of Japan (see Figure A1 in the Appendix).4 For a historical comparison, periods of heightened market volatility and uncertainty during the global financial crisis of 2007-08 (GFC) were also accompanied by short periods of rising Treasury yields and sharp strains in Treasury market functioning. However, Treasury selling pressures were not as strong during the GFC relative to the COVID-19 March 2020 event, likely reflecting concerns over bank creditworthiness during the GFC, which favored a shift by investors from bank deposits to Treasuries. Furthermore, the composition of Treasury investors in 2008-9 was different relative to 2020, with significantly lower participation from leveraged investors and open-ended mutual funds. The COVID-19 March 2020 event was also quite different from the disruptions in U.S. financial markets observed in September 2019.5 Rather than a global dash-for-cash, the adverse events of September 2019 were related to the low level of U.S. aggregate reserves (see, for example, Logan (2020a) and Copeland, Duffie, and Yang (2021). The rest of the article is organized as follows. The first section offers perspective on the performance of various measures of market functioning across jurisdictions during the shock. The second section explores drivers for the apparent outsized reaction in U.S. Treasury markets vis-a-vis foreign sovereign markets, including differences in: (1) the breadth and depth of selling pressures; (2) the expansion of sovereign supply and build-up of leverage; and (3) features of market microstructures. Section 1: Dislocations were Generally More Pronounced in U.S. Treasuries than in Foreign Markets At the start of the COVID-19 pandemic in late-February 2020, investors digested the economic repercussions of the spread of the virus and impending lock-down measures, and as is typical during periods of heightened economic uncertainty, began to demand higher-quality, safe assets. This behavior resulted in investors shifting their portfolios towards sovereign bonds, and the resulting buying pressure drove sovereign yields to broadly decline. As the crisis intensified in March 2020, however, investors demand for cash surged, leading to selling pressure on sovereign bonds and so increases in yields. This down-and-up pattern in yields is illustrated for 10-year U.S., German, U.K., and Japanese bonds in Figure 1.6 4 Logan (2020b), Clarida, Duygan-Bump, and Scotti (2021), Fleming, Liu, Podjasek, and Schurmeier (2021), and Vissing-Jorgensen (2021) detail the Federal Reserves actions in the Treasury market in response to the COVID-19 shock. Hutchinson and Mee (2022) and Hernandez de Cos (2021) describe the European Central Banks response to the COVID-19 shock, Kuroda (2020) details the Bank of Japans response and Tenreyro (2021) compares the Bank of England and Federal Reserves responses to the COVID-19 shock. 5 For details on the September 2019 disruption, see Afonso et al (2021). 6 In comparison to the other sovereign bond yields, overall declines in U.S. Treasury yields were larger in magnitude and sustained throughout the months after March 2020. This contrast largely reflected growing expectations at that time for the FOMC to reduce the Fed funds target range in response to the economic slowdown. Central banks in other jurisdictions were viewed as having comparably less room to lower respective policy rates. 4 In addition to the increase in yields in March 2020, there was an increase in the implied volatility of sovereign bond yields, reflecting in part investors uncertainty over the global economic repercussions of the pandemic. Figure 2 charts a measure of this volatility and illustrates how, across a number of sovereign bonds, this volatility started increasing in late February 2020 and peaked in March 2020. Alongside these changes in yields and volatility, sovereign bond liquidity deteriorated significantly in March 2020. A common measure of bond liquidity is the difference in prices that market makers offer to purchase and sell specific bonds, or the bid-ask spread. An increase in this bid-ask spread over late February and March 2020, for U.S., German, U.K., and Japan 10-year sovereign bonds is illustrated in Figure 3. This evidence, along with the aforementioned rise in volatility, suggests significant stress on trading conditions across sovereign bond markets. -150-130-110-90-70-50-30-10103001/01/20 01/22/20 02/12/20 03/04/20 03/25/20 04/15/20 05/06/20 05/27/20BpsU.S. GermanyU.K. JapanNote: This figure displays the cumulative yield changes for 10 year sovereign bonds, starting on January 1, 2020. U.S., Gernany, U.K., and Japan denote Treasury, Bund, Gilt, and Japanese Government Bond (JGB) securities, respectively.Source: Bloomberg Figure 1: Cumulative Yield Changes Across Sovereign Bond Markets3/183/10 5 -2024681012/31/19 02/19/20 04/09/20 05/29/20Z-scoreUSD EURGBP JPYNote: This figure displays a measure of implied volatility of soverign bonds using a 3 month - 10 year swaption. A swaption consists of an option on a forward-settling interest rate swap. The implied volatility measures the magnitude of expected futurefluctuations of the underlying swap rate, as priced into the option according to an option pricing model. For each soverign bond, the resulting measure is then normalized by its respective Z-score, where that Z-score is calculated on a 2017-2019 sample. USD is U.S. dollar, EUR is Euro, GBP is the British pound, and JPY is Yen.Source: Bloomberg Figure 2: Implied Volatility Across Sovereign Bond Markets3/10 3/1800.511.522.533.50.00.20.40.60.81.001/01/20 02/20/20 04/10/20 05/30/20BpsU.S. Germany U.K. Japan (right axis)Note: This figure shows the spread between bid and ask yields for 10 year soveriegn bonds on a ten-day backward-looking moving average. U.S., Germany, U.K, and Japan are Treasury, Bund, Gilt, and Japanese Governemt Bond securities, respectively. The U.S., Germany, and U.K. time-series are plotted against the left axis and the Japan time-series uses the right axis.Source: Bloomberg CBBT Figure 3: The Spread between Bid and Ask Yields Across Sovereign Bond Markets 6 -2024681001/01/20 02/20/20 04/10/20 05/30/20Z-scoreU.S. Germany U.K. JapanNote: This figure shows the spread between the ten-day backward-looking moving average of bid and ask yields for 10 year soveriegn bonds, normalized by their respective Z-score. Z scores are computed from a 10-day moving average of end-of-day bid-ask spreads using data from January 2017 to January 2019. U.S., Germany, U.K., and Japan are Treasury, Bund, Gilt, and Japanese Government Bond securities, respectively.Source: Bloomberg CBBT Figure 4: The Normalized Spread between Bid and Ask Yields Across Sovereign Bond Markets-4-20246801/01/20 02/20/20 04/10/20 05/30/20Z-scoreU.S. Germany U.K. Japan FranceNote: This figure displays the difference between investors expectations of the fair value yield of a sovereign bond and actual yields, using average spline errors normalized by a Z-score. In normal times, this difference should be close to zero; large deviations from zero could indicate stressed liquidity conditions or dislocations in price discovery. Average spline errors are the average of security yield errors from a BBG fitted curve. Z-scores are calculated on a 2017-2019 sample. U.S., German, U.K., Japan, and France are Treasury, Bund, Gilt, Japanese Governement Bond (JGB) and Obligations assimilables du Trsor (OAT)securities, respectively.Source: Bloomberg Figure 5: Deviations from Fair Value Pricing Across Soverign Bond Markets 7 Although selling pressures materialized across the board for sovereign bonds in March 2020, the impact on trading conditions for U.S. Treasuries was the largest. This can be seen when considering bid-ask yield spreads after they have been normalized by their historical averages. These normalized measures are illustrated in Figure 4 and demonstrate that the deterioration in sovereign bond liquidity was more pronounced in U.S. Treasuries, which during normal times have exhibited comparably lower and more stable bid-ask spreads, than in the German, U.K., and Japan sovereign bond markets. The disproportionate adverse impact on trading conditions in U.S. Treasury markets is also seen by comparing the differences between actual yield curves and model-implied fitted curves across sovereign bond markets. This comparison is informative, because differences between the two could indicate stressed liquidity conditions and/or dislocations in price discovery. As illustrated in Figure 5, the difference between fitted and actual yield curves is significantly larger for U.S Treasuries relative to German, U.K., Japanese, and French sovereign bonds. Section 2: Why Was the Deterioration More Pronounced in U.S. Treasuries? In this section, we explore the likely drivers behind the disproportionate deterioration in U.S. Treasury market functioning during mid-March. We begin by considering differences in investors selling pressures across sovereign bond markets. We then analyze differences in the pace of sovereign bond issuance leading up to the crisis and in the mix of investor types that were purchasing sovereign bonds at