2. Borrowing by Businesses and Households
Historically high levels of business debt and the weakening in household finances could pose a significant medium-run vulnerability for the financial system
Vulnerabilities arising from business debt, which were already elevated at the start of the pandemic, have grown further. Business debt levels increased notably earlier this year as businesses borrowed heavily to weather the pandemic-related shutdowns. However, some of that debt was extended through the PPP and may be eligible for forgiveness, and the low level of interest rates means that businesses can carry more debt. In contrast to business borrowing, household borrowing advanced more slowly than overall economic activity before the COVID-19 crisis and remained heavily concentrated among borrowers with high credit scores. As a result, vulnerabilities arising from household debt were at more modest levels on the eve of the shock; nonetheless, a substantial number of households are facing increasing financial distress.
Table 2 shows the amounts outstanding and recent historical growth rates of forms of debt owed by nonfinancial businesses and households as of the end of the second quarter of 2020. Total outstanding private credit was split about evenly between businesses and households, with businesses owing $17.6 trillion and households owing about $16.1 trillion.
Table 2. Outstanding Amounts of Nonfinancial Business and Household Credit
| Item | Outstanding (billions of dollars) |
Growth, 2019:Q2–2020:Q2 (percent) |
Average annual growth, 1997–2020:Q2 (percent) |
|---|---|---|---|
| Total private nonfinancial credit | 33,669 | 6.7 | 5.7 |
| Total nonfinancial business credit | 17,604 | 10.5 | 6.2 |
| Corporate business credit | 11,041 | 11.2 | 5.6 |
| Bonds and commercial paper | 7,126 | 9.4 | 6.1 |
| Bank lending | 1,684 | 25.7 | 4.5 |
| Leveraged loans * | 1,126 | -.6 | 14.3 |
| Noncorporate business credit | 6,564 | 9.3 | 7.6 |
| Commercial real estate | 2,551 | 5.7 | 6.2 |
| Total household credit | 16,065 | 2.8 | 5.2 |
| Mortgages | 10,592 | 3.0 | 5.4 |
| Consumer credit | 4,079 | 0.8 | 4.8 |
| Student loans | 1,677 | 4.0 | 8.9 |
| Auto loans | 1,198 | 3.1 | 4.9 |
| Credit cards | 954 | -8.0 | 2.3 |
| Nominal GDP | 19,487 | -6.0 | 3.7 |
Note: The data extend through 2020:Q2. Growth rates are measured from Q2 of the year immediately preceding the period through Q2 of the final year of the period. The table reports the main components of corporate business credit, total household credit, and consumer credit. Other, smaller components are not reported. The commercial real estate (CRE) row shows CREdebt owed by both corporate and noncorporate businesses. The total household-sector credit includes debt owed by other entities, such as nonprofit organizations. GDP is gross domestic product.
* Leveraged loans included in this table are an estimate of the leveraged loans that are made to nonfinancial businesses only and do not include the small amount of leveraged loans outstanding for financial businesses. The amount outstanding shows institutional leveraged loans and generally excludes loan commitments held by banks. For example, lines of credit are generally excluded from this measure. The average annual growth rate shown for leveraged loans is computed from 2000 to 2020:Q2, as this market was fairly small before 2000.
Source: For leveraged loans, S&P Global, Leveraged Commentary & Data; for GDP, Bureau of Economic Analysis, national income and product accounts; for all other items, Federal Reserve Board, Statistical Release Z.1, "Financial Accounts of the United States."
Accelerating debt growth and the decline in gross domestic product in the second quarter led to a sharp increase in the ratio of credit to gross domestic product
Before the onset of the pandemic, the combined total debt owed by businesses and households expanded at a pace similar to that of nominal GDP for several years. Between the end of 2019 and June, credit growth accelerated and reached about 9 percent in annualized terms, mostly reflecting strong business borrowing. The precipitous drop in GDP following the outbreak and the increase in business borrowing have caused a dramatic rise in the credit-to-GDP ratio to historical highs (figure 2-1). The household debt-to-GDP ratio had fallen steadily over the long expansion but has jumped recently, returning to levels last seen in 2012 (figure 2-2).
2-1. Private Nonfinancial-Sector Credit-to-GDP Ratio
Note: The shaded bars with top caps indicate periods of business recession as defined by the National Bureau of Economic Research: January 1980–July 1980, July 1981–November 1982, July 1990–March 1991, March 2001–November 2001, December 2007–June 2009, and February 2020–June 2020. GDP is gross domestic product.
Source: Federal Reserve Board staff calculations based on Bureau of Economic Analysis, national income and product accounts, and Federal Reserve Board, Statistical Release Z.1, "Financial Accounts of the United States."
2-2. Nonfinancial Business- and Household-Sector Credit-to-GDP Ratios
Note: The shaded bars with top caps indicate periods of business recession as defined bythe National Bureau of Economic Research: January 1980–July 1980, July 1981–November 1982, July 1990–March 1991, March 2001–November 2001, December 2007–June 2009, and February 2020–June 2020. GDP is gross domestic product.
Source: Federal Reserve Board staff calculations based on Bureau of Economic Analysis, national income and product accounts, and Federal Reserve Board, Statistical Release Z.1, "Financial Accounts of the United States."
Business debt outstanding has grown rapidly so far in 2020 as companies take advantage of low interest rates to bolster cash reserves needed to manage through the pandemic
Borrowing by businesses, likely seeking to bridge pandemic-related interruptions to revenues, was extremely high in the first half of 2020 (figure 2-3). Most of the growth in the first quarter was driven by a surge in bank credit-line draws in March, while the growth in the second quarter was driven by very strong corporate bond issuance and by approximately $500 billion in loans extended under the PPP. However, a significant fraction of the PPP loans may be eligible for forgiveness, as noted earlier, and banks' loan extensions not tied to the PPP declined in the second and third quarters as some firms started to repay their credit lines and as the loan supply tightened. Firms' liquid assets increased notably in the first and second quarters, suggesting that firms were keeping their borrowed funds largely as a buffer. Moreover, historically low interest rates continue to somewhat mitigate investor concerns about default risks arising from high leverage. The net issuance of riskier forms of business debt—high-yield bonds and institutional leveraged loans—had remained high overall through 2019 but slowed during the acute market strains earlier this year. In the second quarter, net issuance of high-yield bonds rebounded, while leveraged loan net issuance contracted. In the third quarter, both high-yield bond and leveraged loan net issuances returned to roughly average historical levels (figure 2-4).
2-3. Growth of Real Aggregate Debt of the Business Sector
Note: Nominal debt growth is seasonally adjusted and is translated into real terms after subtracting the growth rate of the price deflator for core personal consumption expenditure price.
Source: Federal Reserve Board, Statistical Release Z.1, "Financial Accounts of the United States."
2-4. Net Issuance of Risky Business Debt
Note: Institutional leveraged loans generally exclude loan commitments held by banks. Key identifies bars in order from top to bottom.
Source: Mergent, Fixed Income Securities Database; S&P Global, Leveraged Commentary & Data.
After significantly increasing in the aftermath of the pandemic, business debt vulnerabilities have moderated more recently but appear high relative to their historical range
An indicator of the leverage of large businesses—the ratio of debt to assets for all publicly traded nonfinancial firms—was at its highest level in 20 years at the end of the second quarter (figure 2.5).7 An alternative indicator of business leverage that subtracts cash holdings from debt—net leverage—also remains near 20-year highs, but it ticked down in the second quarter as firms' cash position improved.
2-5. Gross Balance Sheet Leverage of Public Nonfinancial Businesses
Note: Gross leverage is an asset-weighted average of the ratio of firms' book value of total debt to book value of total assets. The 75th percentile is calculated from a sample of the 2,500 largest firms by assets. The dashed sections of the lines in the first quarter of 2019 reflect the structural break in the series due to the 2019 compliance deadline for Financial Accounting Standards Board rule Accounting Standards Update 2016-02.
Source: Federal Reserve Board staff calculations based on S&P Global, Compustat.
Despite lower interest rates, the ratio of earnings to interest expenses (the interest coverage ratio) dropped sharply in the second quarter. The decrease was driven by the significant earnings declines as a result of the COVID-19 outbreak. The interest coverage ratio for the median firm is now down to its historical median, and the ratio is negative for many firms because of negative earnings (figure 2-6).
2-6. Interest Coverage Ratios for Public Nonfinancial Businesses
Note: The interest coverage ratio is earnings before interest and taxes over interest payments. Firms with leverage less than 5 percent and interest payments less than $500,000 are excluded.
Source: Federal Reserve Board staff calculations based on S&P Global, Compustat.
In part reflecting the declines in earnings, credit quality deteriorated notably after the onset of the pandemic but showed signs of stabilization, particularly among large firms, in the third quarter. The pace of corporate bond downgrades was elevated through the spring but slowed considerably in the summer. At the end of the third quarter, about half of nonfinancial investment-grade debt outstanding was rated in the lowest category of the investment-grade range (triple-B)—near an all-time high. As has been mentioned in previous Financial Stability Reports, widespread downgrades of investment-grade bonds to speculative-grade ratings could lead investors to accelerate the sale of downgraded bonds, possibly generating market dislocations and downward price pressures in a segment of the corporate bond market known to exhibit relatively low liquidity. Expected bond defaults rose sharply in March to the highest post-crisis levels and, while they have moved down since then, remain above their long-term medians. However, only about 5 percent of outstanding bonds are due within one year, and less than 20 percent of outstanding bonds are due within three years.
Vulnerabilities in the leveraged loan market appear to have lessened somewhat since May, especially for sectors less affected by COVID-19 and for large firms. The share of newly issued loans to large corporations with high leverage—defined as those with ratios of debt to earnings before interest, taxes, depreciation, and amortization greater than 6—dropped in the first quarter but returned in the second and third quarters to the historical highs reached in recent years (figure 2-7). While realized defaults have increased since May, there is some evidence that expected future defaults have decreased over this time frame (figure 2-8). Moreover, downgrades of leveraged loans, which rose sharply in the second quarter, slowed significantly in the third quarter and have returned to pre-pandemic levels. This evidence suggests a more stable outlook for future defaults than in May.
2-7. Distribution of Large Institutional Leveraged Loan Volumes, by Debt-to-EBITDA Ratio
Note: Volumes are for large corporations with earnings before interest, taxes, depreciation, and amortization (EBITDA) greater than $50 million and exclude existing tranches of add-ons and amendments as well as restatements with no new money. Key identifies bars in order from top to bottom.
Source: S&P Global, Leveraged Commentary & Data.
2-8. Default Rates of Leveraged Loans
Note: The data begin in December 1998. The default rate is calculated as the amount in default over the past 12 months divided by the total outstanding volume at the beginning of the 12-month period. The shaded bars with top caps indicate periods of business recession as defined by the National Bureau of Economic Research: March 2001–November 2001, December 2007–September 2009, and February 2020–June 2020.
Source: S&P Global, Leveraged Commentary & Data.
Small businesses have been substantially more affected to date by the effects of COVID-19, and strains associated with the performance of small business debt may worsen significantly
Credit quality for small businesses has worsened notably since the COVID-19 outbreak and has not yet stabilized, with many small businesses closing or scaling back operations significantly during the crisis. Short- and long-term delinquencies at small businesses in August were at elevated levels last seen in 2011. Many small businesses relied on PPP loans to weather the pandemic-related period of low earnings, but the PPP ceased extending loans in August. Survey evidence suggests that credit availability has tightened for small businesses. Moreover, many small businesses report having scarce cash on hand and anticipate financial strains in coming months as they exhaust PPP funds and as accommodation measures expire.
While stresses on households have grown, credit quality has been supported by new and expanded government programs that have lifted household incomes
Although households were generally in sound financial condition before the pandemic, they have experienced a significant loss in earnings due to the spike in unemployment and business closures. Moreover, job losses were heavily concentrated among the most financially vulnerable, including lower-wage workers, young people, women, and minorities. The deterioration in household credit quality to date was mitigated by new and enlarged government programs that have supported household incomes—including expanded unemployment insurance and direct stimulus payments in the CARES Act—and by a moderate improvement in economic activity. However, most COVID-related support for households has already expired or will expire in the coming months, which risks increasing financial stress for many low- to moderate-income households. Strains associated with the performance of household debt may worsen significantly and affect lenders throughout the financial system.
Borrowing by households continued rising at a modest pace in the first half of 2020, with new extensions of credit skewed toward prime-rated borrowers...
Through September of this year, household debt (after an adjustment for general price inflation) edged higher, on net, with debt owed by households with prime credit scores continuing to account for most of the growth. By contrast, inflation-adjusted loan balances for the remaining one-half of borrowers with near-prime and subprime credit scores have changed little since 2014 (figure 2-9).
2-9. Total Household Loan Balances
Note: Subprime are those with an Equifax Risk Score below 620; near prime are from 620 to 719; prime are greater than 719. Scores are measured contemporaneously. Student loan balances before 2004 are estimated using average growth from 2004 to 2007, by risk score. The data are converted to constant 2019 dollars using the consumer price index.
Source: Federal Reserve Bank of New York Consumer Credit Panel/Equifax; Bureau of Labor Statistics, consumer price index via Haver Analytics.
. . . but the sudden increase in unemployment in the spring and sharp decline in earnings have led to a sharp rise in the share of mortgages that are either delinquent or in loss mitigation...
Mortgage debt accounts for roughly two-thirds of total household credit, with mortgage extensions skewed toward prime borrowers in recent years (figure 2-10). Although many households are facing substantial losses in earnings, widespread loss-mitigation measures have helped damp the effect of COVID-19 on mortgage delinquencies (figure 2-11).8 The percentage of mortgages that are either delinquent or in loss mitigation stood at 7 percent in August, below the 2007–09 financial crisis peak of 9 percent, with the caveat that, unlike during the Great Recession, most are currently in a loss-mitigation program rather than being delinquent. Note also that some borrowers in loss mitigation have kept making mortgage payments. Although the severe decline in economic activity and tightening of lending standards originating from the COVID-19 shock might put downward pressure on house prices, at the end of the second quarter, the estimated share of outstanding mortgages with negative equity is very low (figure 2-12). The ratio of outstanding mortgage debt to home values at the end of the second quarter remains at the level seen in the relatively calm housing market of the late 1990s (figure 2-13). Higher levels of homeowner equity generally reduce the likelihood of borrower defaults and provide lenders with a degree of protection against credit losses even as borrowers take advantage of loss-mitigation measures. These considerations lessen concerns that a deterioration in lenders' balance sheets might impede future credit issuance and further worsen the economic outlook.
2-10. Estimate of New Mortgage Volume to Households
Note: Year-over-year change in balances for the second quarter of each year among those households whose balance increased over this window. Subprime are those with an Equifax Risk Score below 620; near prime are from 620 to 719; prime are greater than 719. Scores were measured a year ago. The data are converted to constant 2019 dollars using the consumer price index. Key identifies bars in order from left to right.
Source: Federal Reserve Bank of New York Consumer Credit Panel/Equifax; Bureau of Labor Statistics, consumer price index via Haver Analytics.
2-11. Mortgage Loss Mitigation and Delinquency
Note: Loss mitigation includes tradelines that have a narrative code of forbearance, natural disaster, payment deferral (including partial), loan modification (including federal government plans), and loans with zero scheduled payment and a nonzero balance. Delinquent includes loans reported to the credit bureau at least 30 days past due. The line break represents the data transitioning from quarterly to monthly beginning January 2020.
Source: Federal Reserve Bank of New York Consumer Credit Panel/Equifax.
2-12. Estimate of Mortgages with Negative Equity
Note: Estimated share of mortgages with negative equity according to CoreLogic and Zillow. For CoreLogic, the data are monthly. For Zillow, the data are quarterly and, for 2017, are available only for the first and fourth quarters.
Source: CoreLogic; Zillow.
2-13. Estimates of Housing Leverage
Note: Housing leverage is estimated as the ratio of the average outstanding mortgage loan balance for owner-occupied homes with a mortgage to (1) current home values using the CoreLogic national house price index and (2) model-implied house prices estimated by a staff model based on rents, interest rates, and a time trend.
Source: Federal Reserve Bank of New York Consumer Credit Panel/Equifax; CoreLogic; Bureau of Labor Statistics via Haver Analytics.
. . . and some households are struggling to make debt payments
The remaining one-third of total debt owed by households, commonly referred to as consumer credit, consists mainly of student loans, auto loans, and credit card debt (figure 2-14). Table 2 shows that consumer credit rose 0.8 percent over the year ending in the second quarter and currently stands at a little more than $4 trillion.
2-14. Consumer Credit Balances
Note: The data are converted to constant 2019 dollars using the consumer price index.
Source: Federal Reserve Bank of New York Consumer Credit Panel/Equifax; Bureau of Labor Statistics, consumer price index via Haver Analytics.
Borrowers with subprime credit scores accounted for about one-fifth of outstanding auto loan balances as of the end of the third quarter (figure 2-15). Despite the long economic expansion and low interest rates, delinquency rates on auto loans for subprime borrowers were elevated during the past several years. In response to the COVID-19 outbreak, the share of auto loans that were either delinquent or in loss mitigation jumped and, by August, was about 50 percent higher than the share observed in January, although this share is notably down from May and June (figure 2-16). Similar to mortgage borrowers, many—but not all—auto loan borrowers in loss mitigation have stopped making payments. As of August, 4.5 percent of all auto loan borrowers had not made a payment since at least April.
2-15. Auto Loan Balances
Note: Subprime are those with an Equifax Risk Score below 620; near prime are from 620 to 719; prime are greater than 719. Scores are measured contemporaneously. The data are converted to constant 2019 dollars using the consumer price index.
Source: Federal Reserve Bank of New York Consumer Credit Panel/Equifax; Bureau of Labor Statistics, consumer price index via Haver Analytics.
2-16. Auto Loss Mitigation and Delinquency
Note: Loss mitigation includes tradelines that have a narrative code of forbearance, natural disaster, payment deferral (including partial), loan modification (including federal government plans), and loans with zero scheduled payment and a nonzero balance. Delinquent includes loans reported to the credit bureau as at least 30 days past due. The line break represents the data transitioning from quarterly to monthly beginning January 2020.
Source: Federal Reserve Bank of New York Consumer Credit Panel/Equifax.
Consumer credit card balances contracted sharply in response to depressed consumer spending and declines in credit card utilization rates (figure 2-17). Subprime and near-prime borrowers, taken together, account for about half of consumer credit card balances. The share of credit card balances in delinquency fell since May (figure 2-18). This improved performance is likely driven in part by COVID-related accommodations provided by lenders.
2-17. Credit Card Balances
Note: Subprime are those with an Equifax Risk Score below 620; near prime are from 620 to 719; prime are greater than 719. Scores are measured contemporaneously. The data are converted to constant 2019 dollars using the consumer price index.
Source: Federal Reserve Board of New York Consumer Credit Panel/Equifax; Bureau of Labor Statistics, consumer price index via Haver Analytics.
2-18. Credit Card Delinquency Rates
Note: Delinquency is at least 30 days past due, excluding severe derogatory loans. The data are four-quarter moving averages. Subprime are those with an Equifax Risk Score below 620; near prime are from 620 to 719; prime are greater than 719. Credit scores are lagged four quarters.
Source: Federal Reserve Board of New York Consumer Credit Panel/Equifax.
Finally, the already elevated delinquency rates on student loans highlight the state of finances for some households going into the COVID-19 outbreak. The risk that student loan debt poses to the financial system appears limited at this time; the majority of loans were issued through government programs, and protections originally introduced in the CARES Act, which were later extended, guarantee payment forbearance and stop interest accrual through December 31.
A Retrospective on the March 2020 Turmoil in Treasury and Mortgage-Backed Securities Markets
The U.S. Treasury market and the market for agency residential mortgage-backed securities (RMBS) are among the most liquid securities markets in the world. These markets are critical to the overall functioning of the financial system and to the effective transmission of monetary policy to the broader economy. Many companies and investors treat Treasury securities as risk-free assets, almost as cash, and expect to be able to quickly sell them to raise money to meet any need for liquidity. In mid-March, however, as the effects of the COVID-19 pandemic on financial markets intensified, Treasury and RMBS markets experienced severe dislocations and market functioning became unusually strained. Intense and widespread selling pressures in a context of unprecedented uncertainty appeared to strain dealers' intermediation capacity or willingness to absorb further sales and intermediate in both markets. The Federal Reserve responded through a series of policy actions aimed at supporting smooth market functioning. Following these actions, the acute stresses receded, and market functioning has since largely been restored. We look back at the March events and examine the roles of some institutional participants—in particular, foreign institutions, hedge funds, mortgage REITs (mREITs), principal trading firms (PTFs), and dealers—which have all been reported as having contributed significantly to the March turmoil. A range of other institutional investors, such as mutual funds, may also have contributed to the turmoil and is the subject of ongoing analysis.
The March turmoil
In late February and early March, as fears about the economic effects of the coronavirus intensified and investors moved into the safety of U.S. Treasury securities, Treasury yields fell sharply, and agency RMBS spreads widened (figure A). Daily trading volumes in both on- and off-the-run securities in the Treasury market increased substantially, and daily volumes in the agency RMBS market also spiked.1 Consistent with previous episodes of heightened volatility, measures of trading costs increased notably; for example, indicative bid-ask spreads for the 10-year on-the-run and first and second off-the-run Treasury securities began to rise sharply (figure B).
A. Yields on Nominal Treasury Securities and Agency RMBS Spread
Note: Smoothed yield curve estimated from off-the-run Treasury coupon securities. Agency residential mortgage-backed security (RMBS) spread is the difference between the yield on the 30-year 2.5 percent coupon uniform mortgage-backed security and the duration-matched Treasury yield.
Source: Federal Reserve Bank of New York; Federal Reserve Board staff calculations; JPMorgan Chase & Co.
B. 10-Year Indicative Bid-Ask Spreads
Source: Federal Reserve Bank of New York, New Price Quote System.
Trading conditions deteriorated rapidly in the second week of March as a range of investors sought to sell Treasury securities, particularly those viewed as less liquid, in order to raise cash. Amid unusually poor market functioning and extreme volatility, Treasury yields increased (as shown in figure A), while agency RMBS spreads widened sharply. While trading volumes of both securities remained robust, bid-ask spreads widened dramatically, particularly for off-the-run Treasury securities (as shown in figure B). Stresses soon spilled over into the more liquid on-the-run segment of the Treasury market as well as into the Treasury futures market. Stresses were also evident in a breakdown of the usually tight link between Treasury cash and futures prices, with the Treasury cash–futures basis—defined as the difference between prices of Treasury futures contracts and prices of Treasury cash securities eligible for delivery into those futures contracts—widening notably.
The Federal Reserve's actions to support market functioning
The Federal Reserve took a number of steps to support smooth market functioning. First, between March 9 and March 17, the Federal Reserve expanded its overnight and term repo operations to address disruptions in Treasury financing markets and ensure that the supply of reserves remained ample. Second, on March 15, the Federal Open Market Committee (FOMC) authorized the purchase of at least $500 billion and $200 billion of Treasury securities and agency RMBS securities, respectively, and on March 23, the FOMC announced it would expand the size of asset purchases in the amounts needed to support smooth market functioning. Third, on March 17, the Federal Reserve established the PDCF to support dealer-intermediated markets by expanding primary dealers' access to term funding against a wide range of collateral. Fourth, on March 31, the Federal Reserve established the FIMA (Foreign and International Monetary Authorities) Repo Facility to give central banks and other international monetary authorities the ability to access dollars for liquidity purposes without having to sell their Treasury securities outright. This facility complemented the additional provision of dollar funding through the expansion and enhancement of dollar liquidity swap lines announced by the Federal Reserve and several other central banks during the third week of March. Finally, on the regulatory front, on April 1, the Federal Reserve announced a temporary change in its supplementary leverage ratio rule by excluding U.S. Treasury securities and reserve balances from the calculation of the ratio for BHCs.2
While these actions helped support smooth market functioning, it is important to identify and better understand the drivers behind the March turmoil. To do so, we next provide a (necessarily preliminary) discussion of the likely roles played by several important groups of market participants as the March events unfolded.
The role of foreign institutions
Large-scale sales of U.S. Treasury securities by foreign investors likely contributed to the March turmoil. Indeed, based on Treasury International Capital data, foreign investors are estimated to have sold a record amount of more than $400 billion of Treasury securities in March.3 More than half of this decline reflected liquidations by foreign official institutions, as foreign central banks sought to raise U.S. dollar cash in order to hold precautionary liquidity and to intervene in foreign exchange (FX) markets. The precautionary demand was reflected in a sizable March increase in deposits in the Federal Reserve's foreign repo pool. The introduction of the temporary FIMA Repo Facility helped broaden the reach of the Federal Reserve's provision of U.S. dollar liquidity overseas beyond its dollar swap lines and contributed to the stabilization in the U.S. Treasury market.
The role of hedge funds
Treasury market functioning over this period may also have been affected by the activities of hedge funds, particularly those engaged in relative value (RV) trades. These trades, which generally involve trading to take advantage of small price differences between Treasury cash securities and futures, between on-the-run and off-the-run Treasury securities, and between Treasury securities and MBS, align the relative prices of these assets and typically promote market functioning. They usually entail investors being long in one instrument and short in the other.4 For instance, the Treasury cash–futures basis trade consists of a long cash Treasury position and a short position in a Treasury futures contract with a similar maturity. These trades often involve significant leverage, which is obtained by financing the cash Treasury position in repo markets. Under normal circumstances, this type of trading activity acts to keep the cash–futures basis narrow.
In late February and early March, as Treasury market volatility increased, repo rates rose, and the Treasury cash–futures basis began to widen, many RV hedge funds reportedly reduced their Treasury positions as they unwound their basis trades, which may have contributed to further basis widening.5 Indeed, market commentary has pointed to the sale of Treasury positions by RV funds exiting their basis trades and their sudden role reversal from net buyers of less liquid Treasury securities to net sellers as principal factors contributing to the Treasury market dislocations in mid-March. However, due to a lack of comprehensive data on hedge funds' Treasury cash and derivatives positions, it is unclear what the actual volume of Treasury sales by RV hedge funds was in March and how large a role RV funds played in amplifying the March Treasury market illiquidity.
That said, a set of proxy indicators can be used to shed some light on hedge funds' arbitrage activity in March. For instance, a measure of U.S. hedge funds' holdings of Treasury securities published in the Enhanced Financial Accounts of the United States indicates that hedge funds reduced their cash Treasury positions by about $35 billion (or 3 percent) in the first quarter of the year, consistent with the narrative that hedge fund selling contributed to the Treasury market selloff.6
On the futures side of the basis trade, data from the Commodity Futures Trading Commission show that leveraged funds, including hedge funds, had sizable net short positions in Treasury futures contracts before the COVID-19 outbreak. In March, leveraged funds reduced their net short futures positions by about $80 billion, suggesting that hedge funds unwound some of their Treasury cash–futures basis trades.7
In addition, most RV funds' basis trades require funding, which is typically provided by dealers through repo. Supervisory data indicate that hedge fund Treasury financing from dealers increased in late February through late March, suggesting that dealers continued to finance hedge fund Treasury activities even as market volatility spiked. Similarly, in the June Senior Credit Officer Opinion Survey on Dealer Financing Terms (SCOOS), dealers reported no changes in funding volume collateralized by Treasury securities to RV funds in mid-March relative to mid-February volumes. However, about one-fourth of dealers reported that demand for Treasury financing by these funds increased during the selloff, while a similar fraction of dealers indicated that the availability of such financing decreased. Together, supervisory data and the SCOOS results suggest that the decline in hedge funds' Treasury holdings, in the aggregate, was likely not driven by their inability to finance these positions.
In sum, the reduction in hedge fund Treasury positions may have contributed notably to Treasury market volatility in mid-March amid a massive repositioning by a wide range of investors. However, so far, the evidence that large-scale deleveraging of hedge fund Treasury positions was the primary driver of the turmoil remains weak.
The role of mortgage real estate investment trusts
Similar to the Treasury market, liquidity in the agency RMBS market deteriorated significantly in March. Reportedly contributing to the March turmoil were mREITs, which are leveraged investment companies that invest in pools of agency RMBS and other mortgage-backed assets.8 Such firms primarily fund their holdings of these long-maturity assets using short-term borrowing in the agency RMBS repo market through dealers. At the same time, they typically hedge their interest rate risk by taking short positions in Treasury securities or swaps (or both). As a result, mREITs have leveraged exposures to RMBS–Treasury spreads and RMBS–swap spreads, financed by repo loans that can be rapidly withdrawn.
As RMBS spreads began to widen in late February and spread volatility rose, mREITs' hedges fell in value, triggering margin calls. Because mREITs generally maintain low levels of unencumbered assets that can be used to satisfy the increased margin requirements, the margin calls likely precipitated the unwinding of some of the mREITs' agency RMBS positions. The substantial forced selling and rapid deleveraging intensified stresses in the agency RMBS market and contributed to a further widening of spreads, creating a feedback loop between spread widening and forced deleveraging of mREITs' portfolios. However, following the Federal Reserve's March 23 announcement of increased purchases of Treasury securities and agency RMBS, agency RMBS market functioning improved considerably, with agency RMBS spreads rapidly tightening and spread volatility slowly diminishing. By late March, the pace of agency RMBS selling from mREITs and other levered investors had slowed substantially.
The role of principal trading firms
PTFs are active in the electronic segments of the Treasury cash and futures markets, where trading takes place using a central limit order book (CLOB).9 In the cash market, PTFs predominantly transact in the most liquid on-the-run Treasury securities. PTFs, along with some dealers, are known to adopt high-speed automated trading strategies that account for a significant share of trading and liquidity provision.10 The 2015 Joint Staff Report showed that in previous periods of market stress, PTFs have contributed to keeping quoted bid-ask spreads in these parts of the Treasury market relatively tight.11 More precisely, PTFs—and dealers employing high-speed trading technology—are able to keep spreads tight by reducing posted depth and replenishing the order book faster to manage their exposure to volatility.
During the March turmoil, however, unprecedented strains were also witnessed in the on-the-run segment of the Treasury market, with market depth plummeting and quoted bid-ask spreads widening sharply, raising questions about the role of PTFs in the turmoil (figure C). Some research suggests that the observed widening in bid-ask spreads for on-the-run Treasury securities during the March events points to an unusual reduction in the speed with which high-speed trading entities were replenishing quotes on the order book in response to trades.12 Order book replenishment was not sufficiently fast to avoid significantly heightened bid-ask spread levels and increased volatility as PTFs and other high-speed trading entities scaled down Treasury market-making activity, in aggregate. The reduction in high-speed market-making activity appears to have contributed to the spread of pandemic-related stresses to even the most liquid segments of financial markets. The reasons behind this reduction merit further analysis but would most likely include the considerably elevated economic uncertainty in mid-March, the exceptionally high volatility of Treasury yields, and the breakdown in typical correlations within the Treasury market as well as between the prices of Treasury securities and other assets.
C. On-the-Run Market Depth
Note: Market depth is defined as the average top three bid and ask quote sizes.
Source: Repo interdealer broker community.
The role of dealers
Dealers play a central role in U.S. Treasury and agency RMBS markets by participating in primary markets, buying and selling securities from clients, and providing Treasury and agency RMBS financing to other market participants. Dealers were holding unusually high levels of these securities even before the pandemic, reflecting in part strong Treasury issuance over recent years. Beginning in late February, as a wide range of investors, both domestic and foreign, rushed to obtain liquidity or to rebalance their portfolios in the face of the pandemic, dealers absorbed large amounts of less liquid securities, including off-the-run Treasury securities and agency RMBS, onto their balance sheets. By the second week of March, amid expanding inventories, imbalanced client trading flows, and heightened volatility, some dealers reportedly reached their intermediation capacity or became increasingly unwilling to absorb further sales.13 At the same time, investor demand for repo financing rose sharply, in particular against Treasury collateral, putting further pressure on dealer balance sheets and pushing up dealer funding costs.14 Market commentary pointed to dealer balance sheet constraints and their reluctance to intermediate as important factors behind the deterioration in the functioning of Treasury and agency RMBS markets in early March.
Following the expansion of the Federal Reserve's asset purchases, dealer balance sheet pressures eased in late March as dealers were able to offload some of their inventories, and Treasury and agency RMBS market functioning improved. Indeed, dealer inventory holdings of U.S. Treasury securities declined from their mid-March peak as the Federal Reserve's Treasury purchases picked up, suggesting that asset purchases absorbed some of the Treasury securities that might have otherwise been held on dealer balance sheets (figure D). Similarly, the increase in agency RMBS purchases following the March 23 announcement coincided with some reduction in dealer inventories of agency RMBS securities toward the end of March, indicating that the agency RMBS purchases also helped alleviate balance sheet constraints related to agency RMBS.
D. Fed UST Purchases and Dealer UST Inventory
Note: UST stands for U.S. Treasury. The volume of dealers' non-rehypothecated Treasury repurchase agreements serves as a proxy for the total dealer securities inventory.
Source: Federal Reserve Board, Form FR 2052a, Complex Institution Liquidity Monitoring Report.
Overall, asset purchases were effective in freeing up dealer balance sheet capacity and improving dealers' willingness to intermediate these markets. Funding conditions for dealers also gradually improved following the expansion of repo operations and the announcement of the PDCF, with borrowing rates for dealers declining notably. In addition, dealer Treasury financing volumes increased in late March, indicating that dealers were able to use their spare balance sheet capacity to support their clients' activities.
Looking ahead
The functioning of Treasury markets, including the capacity of dealer balance sheets to absorb extraordinary flows, was discussed extensively at a recent conference sponsored by the members of the interagency working group (the Board of Governors of the Federal Reserve System, the Commodity Futures Trading Commission, the Federal Reserve Bank of New York, the Securities and Exchange Commission, and the U.S. Treasury Department).15 Participants discussed a number of proposals to ensure that dealer capacity could be used effectively in an environment of growing Treasury supply, including the possibility of wider clearing in Treasury cash and repo markets and the potential for use of "all to all" trading platforms that allow buyers and sellers to trade without a dealer intermediary. Meanwhile, the Securities and Exchange Commission has proposed changes to Treasury market regulation that could encourage wider access to Treasury market trading platforms and thereby promote forms of all-to-all trading.16
1. In general, the most recently issued Treasury securities are the most frequently traded and thus the most liquid. These securities are known as "on the run" securities, while less recent issues are called "off the run" securities. Return to text
2. For an overview of these and other Federal Reserve actions to mitigate the economic effects of the COVID-19 pandemic, see the box "The Federal Reserve's Monetary Policy Actions and Facilities to Support the Economy since the COVID-19 Outbreak" in Board of Governors of the Federal Reserve System (2020), Financial Stability Report (Washington: Board of Governors, May), pp. 9–15, https://www.federalreserve.gov/publications/files/financial-stability-report-20200515.pdf. Return to text
3. See also Carol Bertaut and Ruth Judson (2014), "Estimating U.S. Cross-Border Securities Positions: New Data and New Methods," International Finance Discussion Papers 1113 (Washington: Board of Governors of the Federal Reserve System, August), https://www.federalreserve.gov/pubs/ifdp/2014/1113/ifdp1113.pdf. Return to text
4. If investors have a long position, they have bought and own the asset, while if they have a short position, they have sold the asset but do not yet own it. Return to text
5. Hedge fund deleveraging was reportedly due in part to a combination of factors, including increased margin requirements on futures positions, margin calls on losing trades, and compliance with internal risk-management practices. Return to text
6. Quarterly hedge fund balance sheet estimates in the Enhanced Financial Accounts are based on the Securities and Exchange Commission Private Funds Statistics form and reflect hedge funds' positions in cash Treasury securities. See Board of Governors of the Federal Reserve System (2020), "Enhanced Financial Accounts: Hedge Funds," webpage, https://www.federalreserve.gov/releases/efa/efa-hedge-funds.htm. Return to text
7. Of note, although the decline in leveraged funds' Treasury futures positions was significant, it was not outsized relative to the volatility of the positions in previous months. Return to text
8. MBS holdings of mREITs predominantly consist of agency securities. Return to text
9. On a CLOB, participants can post quotes for buying and selling securities, with incoming orders matched to outstanding quotes using an electronic matching engine. Return to text
10. PTFs have become increasingly important in electronic Treasury markets over the past several years. They now account for the majority of traded volumes on electronic interdealer broker platforms in the Treasury market, playing an important role in the provision of liquidity by posting quotes and replenishing those quotes quickly. See Doug Brain, Michiel De Pooter, Dobrislav Dobrev, Michael Fleming, Pete Johansson, Collin Jones, Frank Keane, Michael Puglia, Liza Reiderman, Tony Rodrigues, and Or Shachar (2018), "Unlocking the Treasury Market through TRACE," FEDS Notes (Washington: Board of Governors of the Federal Reserve System, September 18), https://www.federalreserve.gov/econres/notes/feds-notes/unlocking-the-treasury-market-through-trace-20180928.htm; and James Collin Harkrader and Michael Puglia (2020), "Principal Trading Firm Activity in Treasury Cash Markets," FEDS Notes (Washington: Board of Governors of the Federal Reserve System, August 4), https://www.federalreserve.gov/econres/notes/feds-notes/principal-trading-firm-activity-in-treasury-cash-markets-20200804.htm. Return to text
11. For further details, see U.S. Department of the Treasury, Board of Governors of the Federal Reserve System, Federal Reserve Bank of New York, U.S. Securities and Exchange Commission, and U.S. Commodity Futures Trading Commission (2015), Joint Staff Report: The U.S. Treasury Market on October 15, 2014 (Washington: Treasury, Board of Governors, FRBNY, SEC, and CFTC, July), https://www.treasury.gov/press-center/press-releases/Documents/Joint_Staff_Report_Treasury_10-15-2015.pdf. Return to text
12. See Dobrislav Dobrev and Andrew Meldrum (2020), "What Do Quoted Spreads Tell Us about Machine Trading at Times of Market Stress? Evidence from Treasury and FX Markets during the COVID-19-Related Market Turmoil in March 2020," FEDS Notes (Washington: Board of Governors of the Federal Reserve System, September 25), https://www.federalreserve.gov/econres/notes/feds-notes/what-do-quoted-spreads-tell-us-about-machine-trading-market-stress-march-2020-20200925.htm. Return to text
13. Limits on dealers' intermediation capacity may be driven by their internal capital, liquidity, and risk-management practices; their compliance with regulations; or concerns over their profit and loss statements. Return to text
14. Dealers typically use repo to fund both their cash Treasury positions and their lending to clients through Treasury reverse repos. Thus, the ability and willingness to engage in repo, which increases the size of dealers' balance sheets, will affect their willingness to take on additional inventories and provide lending through reverse repos. Return to text
15. See Federal Reserve Bank of New York (2020), "The 2020 U.S. Treasury Market Conference," press release, September 29, https://www.newyorkfed.org/newsevents/events/markets/2020/0929-2020. Return to text
16. See Securities and Exchange Commission (2020), "SEC Proposes Rules to Extend Regulations ATS and SCI to Treasuries and Other Government Securities Markets," press release, September 28, https://www.sec.gov/news/press-release/2020-227. Return to text
References
7. The dashed sections in the series in the first quarter of 2019 reflect a structural break due to a new accounting standard that requires operating leases, previously considered off-balance-sheet activities, to be included in measures of debt and assets. Return to text
8. Loss mitigation is a broad term that describes a variety of loan relief programs implemented by banks to help borrowers cope with payments, including the loan forbearance programs described in the May Financial Stability Report, payment deferrals (including partial payment deferrals), loan modifications (including federal government plans), and loans with zero scheduled payments and a nonzero balance. Return to text