Basis Trades and Treasury Market Illiquidity by Daniel Barth and Jay Kahn1

BRIEF SERIES OFFICE OF FINANCIAL RESEARCH 20-01 | July 16, 2020

Basis Trades and Treasury Market Illiquidity By Daniel Barth and Jay Kahn1

The Treasury futures basis trade seeks to exploit the price difference between cash Treasury securities and Treasury futures. This brief summarizes evidence on the size and extent of basis trading by hedge funds and assesses the possibility that the trade’s exposure to financing and liquidity risks contributed to Treasury market illiquidity in March 2020. This brief also highlights the potential for the trade to lead to further illiquidity. While we find that Treasury illiquidity in March placed stress on Treasury basis trades, the evidence casts doubt on the theory that stress in these trades amplified Treasury market illiquidity. Intervention by the Federal Reserve in the Treasury and repo markets may have limited spillovers that could affect financial stability.

asis trading is a form of near-arbitrage between the A variety of data sources and market commentaries Bcash and futures prices of Treasury securities — a have documented the rising popularity of basis trading usually small difference known as the basis. If this differ- among relative value hedge funds. Office of Financial ence is bigger than the cost of buying the Treasury and Research (OFR) internal data also provide evidence of financing that purchase in the repurchase agreement this. If the hedge funds undertaking these trades are (repo) market, then the trade is profitable. Traders rely large and highly leveraged, this may further magnify on repo markets for financing to achieve high leverage the associated potential systemic risks. Indeed, the for these trades. In early March 2020, stress in Treasury pervasive illiquidity in March 2020 led to large losses markets led to large basis trade losses for some relative at relative value hedge funds engaged in basis trades. value hedge funds. Basis trades tie together Treasury Such funds significantly unwound their positions as a markets, futures markets, and repo markets, all three of result. This has led regulators and market observers to which are crucial to price discovery and liquidity provi- speculate that stress in basis trades may have added to sion in the financial system. While the return on the stress in Treasury markets. However, the evidence casts basis trade is virtually guaranteed over the long term, doubt on this conclusion. Specifically, we find that the in the short term this trade is exposed to substantial Treasuries that were most likely to have been directly liquidity and margin risk. Stress on these trades there- affected by losses on the basis trade actually became fore presents a potential source of systemic risk. more valuable during March 2020.

Views and opinions are those of the authors and do not necessarily represent the views of the Office of Financial Research, the U.S. Department of the Treasury, or the Board of Governors of the Federal Reserve System. OFR briefs may be quoted without additional permission. In this brief, we first describe the mechanics of basis a bet prices will go down.) One immediate difference trades and the determinants of their returns. We also between the return on a basis trade and the return on a provide evidence on their size and possible impor- bill is the possible variation margin on the futures posi- tance in Treasury markets. We then detail how the tion. (Futures traders make variation margin payments returns to basis trades are compensation for exposure when the value of cash and collateral in their accounts to substantial margin and rollover risks, and highlight falls below set margin levels.) More importantly, basis the relationships that arise between the basis, the yield traders generally finance the long cash position in the on Treasury bills, the repo rate, and the federal funds repo market, which exposes the basis trade to rollover rate. We conclude with a discussion of the March 2020 and liquidity risks. The return on basis trade is thus Treasury illiquidity episode and a discussion of poten- equivalent to a synthetic bill plus a risk premium. tial risks associated with basis trades in the future.2 This risk premium is positive on average but can vary significantly and can turn negative during times of The analysis uses the OFR’s collection of cleared stress in funding markets. We discuss the sources of repo data, which began in October 2019. However, this premium below. remaining data gaps limit visibility into basis trading: specifically, we do not have high-frequency or precise In a simple form of the basis trade, at date 0, an investor data on hedge funds’ balance sheets or data on their purchases a Treasury note in the cash market for a substantial borrowing in the uncleared bilateral price P, and simultaneously opens a short position in portions of the repo market. Increasing transparency a Treasury note future at a price, F. This short futures would improve the abilities of financial regulators to position promises delivery of a Treasury note at date T. monitor the risks of this trade. Careful monitoring is Because the basis is typically narrow, investors leverage warranted to determine to what extent margin calls the trade by financing the purchase of the cash Treasury and rollover risk on basis trades may lead to market note through a repo loan at an interest rate r using the disruptions and potential hedge fund defaults. note as collateral. At date T, the investor takes the note returned from the repo contract and delivers it into the Structure of Basis Trades futures contract, receiving F, and using a portion of the futures payment to settle the repo debt. The profit from The basis trade relies on a relationship between the cash this trade is: Treasury market, where investors purchase Treasuries T today; the Treasury futures market, where investors F-P(1+r) . agree on a fixed price to pay for Treasuries they will Figure 1 presents a basic diagram of a basis trade, receive in the future; and the repo market, where tracking the flow of the underlying Treasury security investors borrow or lend Treasuries against cash today. over time and across the three markets. The profit- Theoretically, borrowing a Treasury today in the repo ability of the basis trade depends on the basis, that market, for which the investor pays interest at the repo is, the difference between cash and futures prices of rate, should cost the same amount as purchasing that a Treasury. Cash and futures prices converge as the Treasury today in the cash market with the agreement delivery date approaches (see Figure 2). This conver- to sell that Treasury in the futures market at a later date. gence is virtually guaranteed: at the delivery date, cash Very small variations from that ideal can be profitable and futures prices must be equal because on that date if the investment is leveraged using borrowed capital. a trader can buy a Treasury in the cash market and Basis trades are three-legged trades that span crucial immediately deliver it into the futures market. financial markets: cash Treasury markets, Treasury The profitability of basis trades is summarized by the futures markets, and repo markets. As we show, basis implied repo rate (IRR), which is the repo rate at which trades use long cash Treasury positions and short the profit on the basis trade would be zero. The IRR is futures positions to construct a payoff that, absent closely related to the yield on a Treasury bill because financing risks and other frictions, would be a net posi- the cash flows from the basis trade replicate those from tion similar to a Treasury bill. (In futures markets, long a Treasury bill maturing on the futures delivery date. positions are a bet prices will go up; short positions are

OFR Brief Series | 20-01 July 2020 | Page 2 Figure 1. Diagram of a Basis Trade

TreasuryTreasury market: market: PurchasePurchase Treasury Treasury

Deliver Treasury Treasury to to repo repo lender lender

t =0 t =1 t =2 t = T 1 t = T ... − RepoRepo market: market: OpenOpen repo repo tradetrade RollRoll over over repo repo RollRoll over over repo RollRoll over over repo repo CloseClose repo repo tradetrade

DeliverDeliver Treasury Treasury to futuresto futures

FuturesFutures market: market: ShortShort futures futures contract contract ReceiveReceive cash cash from from futuresfutures

Note: Arrows denote flow of Treasury security; cash moves in the opposite direction.

Source: Office of Financial Research

Figure 2. Convergence of Treasury Futures and Figure 3. Returns on the Basis Trade and Bill Cash Prices ($) FigureYields 3. Returns (percent) on the Basis Trade and Bill Yields (percent) Returns on the basis trade closely follow bill yields

102.00102.00 66 Cash bond bond 5-year5-year impliedimplied repo repo rate rate Futures invoice invoice 2-year2-year impliedimplied repo repo rate rate 101.75101.75 TreasuryTreasury billbill yieldyield 44

101.50101.50

101.25101.25 22

101.00101.00

100.75100.75

100.50100.50 −2-2

100.25100.25

−4-4 100.00100.00 100 80 60 40 20 0 100 80Days 60 to delivery 40 20 0 200620062008 2008 2010 2010 2012 2012 2014 2014 2016 2016 2018 20182020 2020 Days to delivery Note: One-week moving averages. Implied repo rates use the futures contract with the second-to-nearest delivery date.Note: One-week moving averages. Implied repo rates use the futures Note: Average over all five-year contracts since 2016. Sources:contract Bloomberg with Finance the L.P., second-to-nearest Center for Research in Securit deliveryy Prices/University date. of Chicago Booth School of Business, Office of Financial Research Sources: Bloomberg Finance L.P., Center for Research in Security Prices/ Sources: Bloomberg Finance L.P., Center for Research in Security Prices/ University of Chicago Booth School of Business, Office of Financial University of Chicago Booth School of Business, Office of Financial Research Research

The IRR therefore comprises the return to a bill plus a the IRR and the repo rate. Buying the basis while premium for risk, and closely tracks the return on an borrowing in the repo market amounts to a bet that 3 actual bill (see Figure 3). futures prices will converge to cash prices at a rate faster When the IRR is greater than the actual repo rate, than the repo rate. Traders can also “sell the basis,” by basis traders can profit by “buying the basis,” that is, selling a Treasury in the cash market, and opening a by buying the note and shorting the corresponding long position in the futures market, while securing Treasury futures, while borrowing in the repo market. the Treasury through repo lending. Selling the basis At delivery, the trader will earn the spread between amounts to a bet that futures prices will converge to cash prices at a rate slower than the repo rate. The bet

OFR Brief Series | 20-01 July 2020 | Page 3 is desirable when the actual repo rate is greater than the When Treasury cash prices also rise, these margin IRR, or when the trader is closing an existing position payments will be offset by the increased collateral from buying the basis. Because the futures price and value of the long Treasury note held. However, when the cash price of the Treasury are known to the basis Treasury note futures and cash prices diverge, as can trader, provided the basis trader also knows the repo occur in times of illiquidity, the increase in collateral rate, profits on these bets at delivery are guaranteed. value may not be enough to meet the variation margin requirement. The basis trader must make a sudden cash The basis trade does not, however, constitute true outlay to meet their margin call. arbitrage or offer risk-free profits. Several risks drive a wedge between the profitability of the basis trade and Leverage and default risk: The high leverage of many the yield on a Treasury bill and thus create potential basis traders compounds the risks they face. Repo consequences for financial stability. contracts with Treasury collateral allow borrowers to obtain extremely high leverage because haircuts are so Risks of Basis Trades small. A haircut is a discount on the value of an asset pledged as collateral. As a simple rule, the maximum If there were no margin requirements for the futures leverage obtainable for a given security as collateral contract and if there were no need to roll over repo is the inverse of the haircut. Treasuries typically have financing, then the basis trade would be pure riskless haircuts of around 2 percent. At this level, traders arbitrage, and there would be no deviation between could in principle achieve 50-to-1 leverage. For highly the IRR and the return on bills. In reality, the basis leveraged traders, small changes in margin require- trade is not a risk-free strategy. Among the risks basis ments or the cost of financing could lead to large cash traders face: outlays, and in the worst-case scenario could lead to Rollover risk: Financing costs are an important outright failure. source of risk for basis traders. The long position in Deviations of Basis Trade Returns From Bill the Treasury note is largely financed by borrowing in Returns the repo market. One option for this financing is to enter into short-term — possibly overnight — repo Deviations of the IRR from the return on Treasury trades. Interest rates on overnight repo are often bills reflect the risks basis trades face. In the absence lower than on term repo given the same collateral. of rollover risk and margin requirements, the first two However, rolling repo borrowing over daily exposes legs of the basis trade would be equivalent to a Treasury the trader to the possibility of rates rising before the bill. In this case, the return on the basis should be the trade is complete, which would require the trader to same as the return on that bill. Therefore, one way to post additional collateral to secure the same amount assess the extent of risks the basis trade faces at any of financing. If the trader is already highly leveraged, given time is to examine deviations of the return on the it may be difficult to raise more capital without selling basis trade from the return on the bill. assets, possibly at discounted or fire-sale prices. Such liquidity-driven sales could push prices down in other In particular, in times of relative illiquidity and high markets, instigating further margin calls and a liquidity balance sheet costs, the implied repo rate has devi- spiral. We note, however, that traders may manage this ated significantly from the rate of return on bills. One risk by financing through repo with a term matched example of these deviations occurred directly following Figure to the expiration of the futures contract, rather than the collapse of Lehman Brothers in 2008 (see 4 through overnight funding. More granular data would ). Immediately after that collapse, as liquidity dried be needed to determine the length of financing basis up in financial markets, implied repo rates fell across traders employ. contracts. The IRR decline reflected a flight to safety in Treasury markets. Treasury bill rates fell less, in Margin risk: When Treasury note futures prices rise, part due to the natural constraint of the zero lower the trader going long the basis will have to make vari- bound. This deviation persisted for several months. ation margin payments on the short futures contract. It reflected relatively slow-moving capital in the wake

OFR Brief Series | 20-01 July 2020 | Page 4 of hits to balance sheets of intermediaries during the crisis, as well as general concerns over counterparty Figure 4. Basis Trade Returns During the 2007-09 FigureFinancial 4. Basis Trade Crisis Returns (percent) During 2007-09 Financial Crisis (percent) risk. Note in this case that traders that were long the Liquidity events cause sharp deviations of the return on the basis trade from the yield on bills

basis would have profited from the basis narrowing and 6 6 turning negative. Lehman Brothers bankruptcy This pattern in 2008 is indicative of a more general rule 4 4 that returns on basis trades tend to depart from bill returns during times of stress. In general, deviations of

the return on basis trades from returns on Treasuries 2 2 have reflected the costs of liquidity provision, that is, the willingness of financial institutions to take on the margin risk and rollover risk that basis trades entail. 0 0 One way to summarize the costs of liquidity provision in Treasury repo markets is the difference between the -2−2

repo rate and the federal funds rate. Because Treasury 5-year implied repo rate repo rates reflect a willingness to accept Treasuries as 2-year5-year implied implied reporepo rate rate -4−4 2-year implied repo rate collateral, whereas lending in the federal funds market TreasuryTreasury bill bill yieldyield

is uncollateralized, differences in Treasury repo and JanJan 07 May May 07 Sep Sep 07 Jan Jan 08 May May 08 Sep Sep 08 Jan Jan 09 May May 09 Sep Sep 09 Jan Jan 10 07 07 07 08 08 08 09 09 09 10 federal funds rates will partly reflect balance sheet Note: Data are one-week moving averages. Implied repo rates use the futures contract with the second-to-nearest constraints that make accepting Treasuries more or less delivery date. Sources:Note: Bloomberg Data Finance are L.P.,one-week Center for Research moving in Securit averages.y Prices/University Implied of Chicago Booth repo School rates of Business, Office of costly. Figure 5 shows deviations of basis trade return Financialuse theResearch futures contract with the second-to-nearest delivery date. for five-year and two-year contracts from bill yields, Sources: Bloomberg Finance L.P., Center for Research in Security Prices/ University of Chicago Booth School of Business, Office of Financial along with the spread between the repo rate and federal Research funds rate. Deviations of the basis trade return from the

Figure 5. Deviations of the Basis Trade Return from the Bill Return and the Cost of Note Funding (percentage Spread of Implied Repo Rate over Bill Yield points) High spreads between sponsored borrower rates and the federal funds target 2.0 1.0 2.0 2-year basis deviation from bill (left axis) 1.0 2-year basis deviation from bill (left axis) 5-year basis deviation from bill (left axis) 5-year basis deviation from bill (left axis) GCF repo rate spread over fed funds (right axis) 0.80.8 1.5 GCF repo rate spread over fed funds (right axis)

0.60.6 1.01.0 0.40.4

0.5 0.20.2

0.00.0 0.00.0

-0.2−0.2 −0.5-0.5 20102010 2011 2011 2012 2012 2013 2013 2014 2014 2015 2016 2017 2018 2018 2019 2019 2020 Note: Spread between the IRR and bill yield for the second-to-deliver contract. Sources:Note: Bloomberg,Data are 14-day Center moving for averages Research of the on spread Security of the Prices, implied Officerepo rate of forFinancial futures contractResearch with the second-to-nearest delivery date over the yield for an equivalent maturity Treasury bill for two-year and five-year Treasury futures and the spread of the GCF Treasury repo index over the effective federal funds rate.

Sources: Federal Reserve Bank of New York Effective Federal Funds Rate, DTCC GCF Repo Index, Bloomberg Finance L.P., Center for Research in Security Prices/University of Chicago Booth School of Business, Office of Financial Research

OFR Brief Series | 20-01 July 2020 | Page 5 bill yield constitute a liquidity premium in basis trades, quarter, CBOT offers contracts for two-year, five-year, because basis traders will require compensation for the and 10-year Treasury notes as well as Treasury bonds.4 risks incurred in financing their Treasury holdings Only a certain basket of Treasuries can be delivered until the delivery date. These risks should be correlated into each of these futures contracts, with limits on this with the spread between the actual repo rate and the deliverable basket determined by the maturity of the federal funds rate. When Treasuries are more expensive note relative to the delivery date of the contract and the to finance, the excess returns on the basis must be high original maturity of the note at issuance.5 For instance, enough in equilibrium to induce basis traders to hold the deliverable basket for the two-year CBOT contract these Treasuries until delivery. Therefore, the excess contains Treasuries with an original maturity of less return on the basis closely follows the costs of financing than five years and three months and a residual matu- a note, as represented by the spread of the repo rate over rity on the last day of the delivery month between one the federal funds rate. year and eight months and two years. Otherwise-similar Treasuries will differ in whether they are deliverable Implementing Basis Trades into the futures contract. A conversion factor attached to the futures price is meant to account for the desir- Understanding sources of risk for basis trades and ability of individual Treasuries. A formula provided by where stress can manifest requires understanding the the CBOT determines these conversion factors. Due to technical details of implementing these trades. these conversion factors, only one Treasury note will be 6 In particular, only certain futures contracts and cheapest-to-deliver into each futures contract. Which Treasury notes are used in basis trading. On any given note is the cheapest-to-deliver can change over the life date, there is just one Treasury security that basis traders of a contract. want to own for each contract to make a particular deal Another complication is the structure of the repo market as profitable as possible, called the “cheapest-to-de- where basis traders finance their initial note purchase. liver” Treasury. Treasury futures are exchanged on The structure of this market can create difficulties the Chicago Board of Trade (CBOT) and cleared by in securing funding for basis trades and can amplify CME Clearing. Treasury note futures are offered by counterparty risk. Figure 6 describes the general CBOT quarterly and require physical delivery. Each structure of the repo market. Basis traders can borrow

Figure 6. Structure of the Bilateral Repo Market

DVPDVP InterdealerInterdealer

DVP DVPSponsored Sponsored Borrowing Borrowing DVP Sponsored Lending CashCash Lenders: Lenders: DVP Sponsored Lending

• MoneyMoney market market fundsfunds CashCash Borrowers: Borrowers: • Broker Pension and insurance funds Broker Dealers Hedge• fundsHedge Funds • • Pension and insurance funds Dealers • Sovereign wealth funds Others • • Sovereign wealth funds • • Others Banks • • Banks UnclearedUncleared Bilateral Bilateral Lending UnclearedUncleared Bilateral Bilateral Borrowing Borrowing

Federal Federal Reserve OvernightOvernight Reverse-Repo Reverse-Repo Reserve RepoRepo (RP) (RP) Facility Facility (ON-RRP)(ON-RRP) Facility Facility

Note: Arrows denote the direction of the flow of cash, from cash lenders to cash borrowers.

Source: Office of Financial Research

OFR Brief Series | 20-01 July 2020 | Page 6 cash from two venues in the bilateral repo market. The first is the sponsored segment within the Fixed Figure 7. Hedge Fund Treasury Futures Short FigurePositions 7. Hedge Fund($ billions)Treasury Futures Short Positions ($ billions) Income Clearing Corporation’s DVP Repo service. Short positions shot up after 2018

The second is the uncleared bilateral market between 900900 dealers and cash borrowers. Traders that are not clearing members of DVP can access the venue’s clearing 800800 services if a sponsoring member guarantees their trades. Specifically, sponsoring dealers are required to 700700 cover the clearing fund allocated to their counterparty. 600600 2-year For the sponsor, this offers the advantage of netting trades that provide cash to borrowers against trades 500500 that secure cash from money market funds and other 2-year cash lenders. However, sponsors are more exposed to 400400 the risk of counterparty default than in other sections 300300 5-year5-year of the market. As a result, sponsors generally charge borrowers a premium above rates in the interdealer 200200 market. 10-year10-year 100100 In the uncleared bilateral market, there are no central BondBond counterparties, so counterparty exposure cannot be 00 20082008 20102010 20122012 20142014 20162016 20182018 20202020 shared with other participants. Basis trades financed Note: Data are aggregate leveraged fund short positions in dollars of face value. Ultra bond futures and 10-year ultra note in uncleared bilateral repo markets therefore result in futuresNote: positions Data are are included aggregate with bond futures leveraged and 10-year fund note futures short respectively. positions in Sources:dollars Commodity of face Futures value. Trading Ultra Commission bond Commitment futures of and Traders, 10-year Office of Financialultra noteResearch direct counterparty risks for dealers. futures positions are included with bond futures and 10-year note futures respectively.

The parts of the repo market in which hedge funds Sources: Commodity Futures Trading Commission Commitment of borrow are somewhat divorced from the facilities Traders, Office of Financial Research through which the Federal Reserve controls repo rates. The Federal Reserve influences rates in the bilateral risen dramatically over recent years, and may have repo market through its facilities in the tri-party repo migrated away from large banks as a result of increased market, in which basis traders are unlikely to partic- balance sheet costs to those banks from post-finan- ipate. The Overnight Reverse-Repurchase Facility cial crisis regulations. One way to measure basis trade (ON-RRP) sets a floor on rates by offering an outside intensity among hedge funds is by looking at hedge option for certain money market funds to invest their fund short positions in Treasury futures (see Figure 7). cash. Since September 2019, the Federal Reserve has In early 2018, hedge funds began accumulating sizable also operated a Repo Facility (RP) that sets a ceiling on short positions in Treasury futures. The increase in rates by lending to primary dealers. However, because these futures positions was greatest in two-year and there is no direct link to the markets in which basis five-year notes. Smaller increases also occurred in traders borrow, pass-through of these facilities into repo 10-year note and Treasury bond futures contracts. At rates for basis traders depends on dealer intermediation. the peak of these short positions in 2019, the total face value outstanding for hedge funds was more than $800 Rise of Hedge Fund Basis Trades billion. Basis trading has existed since the late 1970s, though These short futures positions alone do not necessarily the rise of Treasury note basis trades is relatively recent: indicate basis trading, as futures may be used in many short-maturity note futures contracts were introduced kinds of trades. However, the OFR’s cleared repo only in 1989 for the five-year Treasury contract and in collection offers a clearer picture of hedge fund trades. 1990 for the two-year Treasury contract. Both contracts Because the OFR’s collection includes collateral infor- had little open interest until the last few years. Hedge mation for every repo contract in DVP, we can examine fund participation in the basis trade appears to have

OFR Brief Series | 20-01 July 2020 | Page 7 Figure 8. DVP Sponsored Reverse Repo by Participant Type ($ billions) Hedge funds make up the majority of sponsored DVP borrowing

Figure 8. DVP Sponsored Reverse Repo by Participant Type ($ billions)

350350

300300

250250

200200

150150 OtherOther

100100 Hedge Funds Hedge Funds

0 Nov2019 2019 DecDec 2019 JanJan 2020 FebFeb 2020 MarMar 2020 AprApr 2020 MayMay 2020

Note:Note: Data Data are aggregateare aggregate daily transaction daily volumes transaction. volumes. Sources: OFR Cleared Repo Collection, Office of Financial Research Sources: OFR Cleared Repo Collection, Office of Financial Research Figure 9. Money Market Fund Repo with FICC ($ billions) the specific collateral posted by hedge funds in support Dramatic expansion of sponsored lending, likely mirrored in Figure 9. Money Market Fund Repo with FICC of their sponsored borrowing in DVP. sponsored borrowing ($ billions) Hedge Fund Repo Positions 300300

While the majority of hedge fund repo borrowing 250250 likely occurs through bilateral uncleared trades, the sponsored DVP market offers insights not available 200200 in the uncleared bilateral market. Sponsored DVP is 150150 also increasingly important in its own right. At present, 100100 hedge funds make up the vast majority of sponsored borrowing (see Figure 8). The Depository Trust & 5050 Clearing Corporation first allowed hedge fund partic- 0 ipation in sponsored repo in 2017. Such participation Jan 2011Jan Jan 20132013Jan Jan 2015 Jan 20172017 Jan 2019 Note: Aggregate repo volume outstanding. FICC stands for Fixed increased dramatically after the expansion of sponsor- Note: Aggregate repo volume outstanding. FICC stands for Fixed IncomeIncome Clearing Clearing Corporation. Corporation. ship in April 2019. Recent repo data show that cash Sources: OFR Money Market Fund Monitor, Office of Financial Research provided by money market funds is mostly passed on Sources: OFR Money Market Fund Monitor,Office of Financial Research to hedge funds, so the expansion of sponsored lending provides a hint to the pace of growth in sponsored borrowing. Following the expansion of sponsorship, participation of money market funds in sponsored repo increased dramatically (see Figure 9). This may be associated with increased borrowing by hedge funds in DV P.

OFR Brief Series | 20-01 July 2020 | Page 8 In general, rates in the sponsored market for cash that sponsored borrowing rates are in general above borrowers are higher than interdealer rates, which interdealer rates highlights a potential for imperfect are again higher than the sponsored market for cash pass-through of liquidity-boosting interventions by the lenders. There are two primary reasons for this. First, Federal Reserve to the borrowing rates of hedge funds, sponsors are required to guarantee the trades of the adding an additional layer of liquidity risk to hedge entities they sponsor. This means that hedge funds that fund basis trades. trade in sponsored markets represent relatively high risk If hedge funds are actively basis trading, we would for their counterparties. Second, sponsored borrowers expect them to disproportionally hold the cheap- generally borrow early in the day relative to sponsored est-to-deliver Treasury notes. Figure 10 shows hedge lenders. Because much of sponsored borrowing uses fund positions in repo by security CUSIP prior to the cash from sponsored lenders, sponsored borrowing Dec. 1, 2019, futures delivery date and following that creates additional liquidity risk for sponsors. The fact

Figure 10. Hedge Fund DVP Repo in Treasuries by Maturity Date ($ billions, average daily transaction value) Figure 10. Hedge Fund DVP Repo in Treasuries by Maturity Date ($ billions, average daily transaction value) Large repo activity occurs within maturities eligiblePrior for toto Dec. delivDec. ery1, 1, 2019 2019 2 year 5 yearyear Kernel density Prior to Dec. 1, 2019 Kernel density Individual security 44 2 year 5 year Individual security KernelDeliverable density set 4 DeliverableIndividual security set 33 Deliverable set 3

22 2

11 1

00 2021-012021-012021-07 2021-072022-01 2022-012022-07 2022-07 2023-012023-01 2023-072022-07 2024-012024-01 2024-072024-07 2025-012025-01 2025-072025-07 0 2021-01 2021-07 2022-01 2022-07After 2023-01 and including 2023-07 Dec. 2024-01 1, 2019 2024-07 2025-01 2025-07 2 year AfterAfter andand includingincluding Dec. Dec. 1, 20191, 2019 5 year 2 year 5 yearyear 4 4

3 3

2 2

1 1

0 2021-010 2021-07 2022-01 2022-07 2023-01 2023-07 2024-01 2024-07 2025-01 2025-07 2021-01b 2021-07b 2022-01b 2022-07b 2023-01b 2023-07b 2024-01b 2024-07b 2025-01b 2025-07 2021-01b 2021-07b 2022-01b 2022-07b 2023-01bMaturity2022-07b date2024-01b 2024-07b 2025-01b 2025-07b Maturity date Maturity date Note: Dots are average daily outstanding positions in individual Treasuries; solid lines are smoothed fitted sums within maturity windows. Note:Above, Dots gray are areas average are deliverablesdaily outstanding for December positions 2019, in individual below forTreasuries; March 2020. solid linesThe green are smoothed dot denotes fitted positions sums within in the maturitycheapest-to -delivery windows. Above, gray areas are deliverables for December 2019, below for March 2020. The orange dot denotes positions in the cheapest-to-deliveryfor the two-year for Decemberthe two-year contract. December contract. Some positions have been excluded from the scatter for disclosure purposes. Sources: OFR Cleared Repo Collection, Office of Financial Research Sources: OFR Cleared Repo Collection, Office of Financial Research

OFR Brief Series | 20-01 July 2020 | Page 9 delivery date. The shaded windows in the top panel are 2. Regulatory costs of holding Treasuries for banks. the maturity dates of notes eligible for delivery into the Hedge funds and other proprietary traders are not December futures contract. Securities just inside this explicitly limited to maximum leverage ratios in the delivery window had significantly more hedge fund same way banks are. Standards imposed after the repo volume than securities just outside this window. 2007-09 financial crisis put direct leverage limits The largest position prior to December 1 was in the on banking institutions, typically requiring a capi- two-year window and was for the Treasury security tal-to-asset ratio of at least 5 percent, which implies that was cheapest-to-deliver for this contract. After a maximum leverage ratio of 20 to 1.7 Hedge fund December 1, the position in the cheapest-to-deliver leverage is constrained only by the haircuts on the for the two-year December contract had diminished collateral, and for Treasury securities haircuts are considerably, while positions had expanded for deliv- typically around 2 percent. This implies a max- erables for the March contract, highlighted in gray in imum leverage ratio for hedge funds of 50 to 1. the bottom panel. This is consistent with hedge funds Because the basis trade profits from tiny differences maintaining positions in the cheapest-to-deliver for in spreads, high leverage is necessary to make the contracts near to delivery. trade worthwhile. Basis trade activity may therefore have migrated from banks to hedge funds and other The decline in repo in the two-year cheapest-to-deliver less-regulated traders because leverage limits made Treasury occurs directly before the December futures the basis trade unprofitable for banks. Anecdotal contract’s first delivery date, with another smaller evidence supports this; the Financial Times has increase before its last delivery date. Even though only reported that many of the lead traders executing part of hedge funds’ basis positions are likely to be these trades at banks have left to join hedge funds.8 funded in DVP, the $4-$4.5 billion in repo outstanding in this Treasury note just before the delivery date is 3. Increasing segmentation in the bills market. equivalent to more than 10 percent of the outstanding Regulatory reforms may have made it more difficult notional value of the note. for unlevered funds to hold longer-term Treasuries and may have led to a natural source of demand for Reasons for Rising Hedge Fund Treasury long futures positions in Treasuries. As a result, seg- Positions mentation may have increased in the Treasury market, which would lead simultaneously to a higher Participation by hedge funds in basis trades requires disconnect between bond and bill returns and to those funds to hold securities until they roll. The demand for short Treasury note futures. primary source of demand for Treasury futures is from unlevered asset managers, such as mutual funds, It is difficult to determine which of these possible insurance funds, and pension funds. By holding these causes is most likely and, in equilibrium, they are all securities until they roll contracts, hedge funds provide reinforcing. However, all three explanations would liquidity by using repo leverage to purchase Treasuries. have resulted in increased costs associated with longer- While the exact reason for increases in volume in hedge term Treasuries for regulated or unlevered participants fund basis trades is unclear, possibilities include: relative to levered and unregulated participants such as hedge funds. This relative change in the costs of 1. Declining foreign demand for Treasuries. Since holding Treasuries for hedge funds and unlevered 2015, foreign holdings of Treasuries have been participants would in turn have made the basis trade declining as a share of total Treasuries outstand- more desirable for hedge funds and have led to a reallo- ing. Weakened demand for Treasury coupon cation of Treasuries from traditional holders to relative securities from foreign investors, combined with value hedge funds. This is consistent with the generally increased amounts of Treasuries outstanding, may higher return on basis trades relative to bills seen in the have pushed down the relative prices of notes and latter part of Figure 5. bonds, making the basis trade more attractive to hedge funds. The exposures implicit in Treasury basis trades are similar to those that have led to problems for hedge

OFR Brief Series | 20-01 July 2020 | Page 10 funds in the past. Long-Term Capital Management off-the-run securities. That so many of the risks of this (LTCM), for instance, was a highly leveraged relative trade materialized without substantial feedback into value hedge fund whose failure nearly led to a financial financial stability may have been a result of timely crisis in 1998 when the spreads they were betting on intervention by the Federal Reserve. In this section we diverged further than fund managers expected. LTCM review the March illiquidity episode, discussing possible suffered because it was highly exposed to liquidity and exacerbating factors, and then discuss the effects of this flight-to-safety risks for which it was poorly hedged. illiquidity on basis trades. We conclude by discussing While the introduction of central counterparties and the effects of Federal Reserve actions after March on cleared repo markets may have reduced some of the Treasury market liquidity and basis trades. risks that LTCM faced, modern relative value hedge funds are more plentiful, larger, and continue to be Onset of Treasury Market Illiquidity highly leveraged. In early March 2020, Treasury market liquidity decreased. As yields fell, volatility spiked, according Basis Trades and March 2020 to multiple option-implied indexes (see Figure 11). Treasury Market Illiquidity At the same time, bid-ask spreads began to increase, concentrated in off-the-run Treasuries (see Figure Treasury market illiquidity in March led to large losses 12). Standard spreads associated with liquidity risk, in hedge fund basis trades as the CBOT increased such as the on-the-run off-the-run spread, also spiked. margins and repo rates became extremely volatile. These indicators are consistent with a general flight to However, we are not aware of any hedge fund defaults liquidity, with investors selling off-the-run Treasuries on basis trades. While funds appear to have partially and either holding the proceeds as cash or purchasing exited these trades based on sales of the cheapest-to-de- more-liquid on-the-run Treasuries that could be more liver notes, it is not clear that these sales actually readily converted into cash. impaired Treasury market liquidity. Instead, the basis trade appears to have continued to provide net liquidity This illiquidity appears to have been reflected in an to underlying Treasuries relative to comparable inability of dealers that purchased Treasuries to offload

Figure 11. Treasury Volatility Indexes Figure 11. Treasury Volatility Indexes Treasury volatility indexes were at their highest level since the 2007-09 financial crisis 18 300

16 CME 10 10-year-year Treasury Treasury VIX VIX (left (left axis) axis) MOVE index index (right (right axis) axis) 250 14

12 200

10 150 8

6 100

4 50 2

0 0 Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar MaMar 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020202 Note: CME 10-year Treasury VIX and the MOVE Index are option implied Treasury volatility indexes. Sources: BloombergNote: CME Finance 10-year L.P., Treasury Office of FinancialVIX and Research the MOVE Index are option implied Treasury volatility indexes. Sources: Bloomberg Finance L.P., Office of Financial Research

OFR Brief Series | 20-01 July 2020 | Page 11 the securities on to ultimate buyers (see Figure 13). Figure 12. Bid-Ask Spreads for Off-the-run Making markets requires dealers to hold inventories Figure 12. Bid-Ask Spreads for Off-the-run Treasuries ($) of the Treasuries in which they make those markets, MarchTreasuries illiquidity was ($) concentrated in off-the-run securities with accompanying regulatory and balance sheet costs. 2-Year2-Year 3.0 Leading into March, primary dealers already had 3.0 5-Year5-Year elevated exposure to Treasuries, which had begun to 10-Year10-Year rise in late 2018. Sales in the Treasury market increased 30-Year30-Year 2.52.5 this exposure, especially to the shortest and longest residual maturity Treasuries. As the pandemic created demand for cash in the real economy, selling pressure 2.02.0 in the Treasury market came from multiple sources.

One notable source of sales was from foreign accounts. 1.51.5 Treasury International Capital System data show net decreases in foreign Treasury positions were around 1.0 $257 billion in the month of March, with a decrease of 1.0 $147 billion in foreign official accounts. Data from the Federal Reserve’s Factors Affecting Reserve Balances, 0.50.5 which provide a higher-frequency view of foreign official custody holdings with the Federal Reserve, 0.00.0 suggest these sales began in the last weeks of February.9 JanJan FebFeb MarMar AprApr MayMay JunJun Without immediate buyers, these Treasuries remained Note: Spreads are the difference between bid and ask prices for $100 notional in theNote: fourth-from-most-recent Spreads are Treasurythe difference issuance as ofbetween January 202 bid0. and ask prices for on dealers’ balance sheets and made the dealers hesitant Sources:$100 Bloomberg notional Finance in theL.P., Office fourth-from-most-recent of Financial Research Treasury issuance as to create markets in these off-the-run Treasuries. of January 2020. Sources: Bloomberg Finance L.P., Office of Financial Research Stress in Hedge Fund Basis Trades

Treasury market illiquidity had an immediate effect Figure 13. 13. Primary PrimaryDealer Dealer Treasury Treasury Net Exposure Net Exposure($ billions) on basis trades. First, rising volatility led to increasing ($Long billions)-term trends in dealer Treasury exposure were exacerbated margins on Treasury futures, and therefore margin by March sales calls on these positions (see Figure 14). From February 250250 28 to March 16, across Treasury note futures contracts margins rose by more than 30 percent, while margins < 22-years years 200200 on bond futures more than doubled, corresponding to 2-62-6 yearsyears their longer duration. Margin calls may have forced 6-116-11 years years > 11 years an accelerated decline of hedge fund short futures 150150 > 11 years positions (see Figure 15). In particular, total hedge fund shorts declined from $774 billion in face value to $684 billion between February 18 and March 17, 100100 with particularly large declines of around $71 billion in two-year Treasury shorts.10 Some portion of this decline preceded March, with total shorts declining by 5050 $25 billion between February 18 and March 3, which may have represented some foresight of the stress that 0 potential spread of COVID-19 to the United States JanJan JunJun NovNov AprApr SepSep FebFeb JulJul DecDec May OctOct Mar could put on Treasury markets. 20162016 20162016 20162016 20172017 20172017 20182018 20182018 20182018 2019 20192019 2020 Note: Data are net exposures by maturity. Note:Sources: Data Federal are Reserve net exposures Bank of New byYork maturity. Primary Dealer Statistics, Office of The variation margin payments on the futures contract Financial Research Sources: Federal Reserve Bank of New York Primary Dealer Statistics, were not fully offset by the increase in prices in the Office of Financial Research long note position. Futures prices rose more quickly

OFR Brief Series | 20-01 July 2020 | Page 12 than cash prices (see Figure 16). This widened the Figure 15. Primary Dealer Treasury Net Exposure basis, leading to further losses for basis traders who had Basis Return Deviations and Short-Rate Disconnect bought the basis. High($ billions) spreads between sponsored borrower rates and the federal funds target

The imperfect nature of pass-through within repo 2-year2-year short 5-year5-year short markets may also have contributed to the losses on 10-year short 10-year short basis positions. Figure 17 illustrates the pass-through 400400 2-year long 2-year5-year long of federal funds rate target changes onto the repo rates 5-year10-year long long for sections of the DVP market. Spreads between repo 10-year long rates and the federal funds rate target widened in the 300300 first two weeks of March 2020. At the same time, the spread between the sponsored borrowing rate — the rate at which hedge funds borrow in DVP — and 200200 interdealer rates was at the widest level seen since the OFR repo data collection began. While this may be in part a sign of sponsors’ concerns over intraday liquidity 100100 provision to their borrowers, the highest rates were concentrated among hedge funds, suggesting that their leverage may have been a concern. 0 0 2018-01Jan 2018-04Apr 2018-07Jul 2018-10Oct 2019-01Jan 2019-04Apr 2019-07Jul 2019-10Oct 2020-01Jan 2020-04Apr In combination, Treasury illiquidity and imperfect 2018 2018 2018 2018 2019 2019 2019 2019 2020 2020 Note: Absolute value of spread between the IRR and bill yield and the short-rate disconnect. repo pass-through led to a large disconnect between Sources: Bloomberg, Center for Research on Security Prices, Office of Financial Research the implied repo rate and Treasury bill yields across Note: Data are leveraged fund short and long positions in dollars of face value. contracts. The IRR followed the bill rate until early Sources: Commodity Futures Trading Commission Commitments of March, and then began to depart (see Figure 18). Traders, Office of Financial Research This departure began as the bill rate rapidly moved below the DVP sponsored borrowing rate and these Figure 16. 16. Futures Futuresand andCash CashPrices forPrices the Twofor -theYear Two-Year June 2020 ContractJune 2020($) Contract($) FigureFigure 14. 14.Maintenance Maintenance Margin Margin on Futures on ContractsFutures ($) Spread widened between cash and futures prices in March MarginsContracts increased ($) 30 percent during March 105105 Cheapest-to-deliverCheapest-to-deliver Price Price 800 FuturesFutures DeliveryDeliveryPrice Price

750 104104

700 103103 650

600 102102 550

500 101101 450

400 100100 10/24/19 12/09/1901/23/20 03/06/20 04/20/20 350 JanJan Mar May Jul SepSep NovNov JanJan Mar Note: Delivery Spreads price are isthe futures difference price multiplied betweenby bid the and conversion ask prices factor for 20192019 2019 2019 2019 20192019 20192019 20202020 2020 for$100 the notional cheapest in-to the-deliver. fourth-from-most-recent Prices are for $100 notional. Treasury issuance as of January 2020. Note: Data are for maintenance margins on $200,000 notional in two-year Note: Data are for maintenance margins on $200,000 notional in Sources: Bloomberg Finance L.P., Office of Financial Research Treasurytwo-year futures Treasury contracts. futures contracts. Sources: CME Group, Office of Financial Research Sources: CME Group, Office of Financial Research

OFR Brief Series | 20-01 July 2020 | Page 13 Figure 17. DVP Repo Rates (left axis, percentage point spread over fed funds target midpoint) and Federal Reserve Facility Participation (right axis, $ billions)

Interdealer Interdealer rate rate (left (left axis) axis) 300 300 Sponsored Sponsored lender lender rate rate(left axis)(left axis) Sponsored borrower rate (left axis) Sponsored borrower rate (left axis) 0.6 0.6 ON-RRP volume (right axis) 250 250 ON-RRP RP volume volume (right axis) (right axis) RP volume (right axis)

0.4 0.4 200 200

150 150 0.2 0.2

100 100 RP facility 0.0 0.0 RP facility

50 50 ON-RRP facility -0.2−0.2

0 0 JanJan 15 15 Feb Feb01 01 FebFeb 15 15 MarMar 01Mar 15 15 AprApr 0101Apr Apr 15 15May May 01 01

Note: Repo rates are average overnight Treasury rates for each market segment.. The two black lines represent the average rate offered by the Federal Reserve’s Overnight Reverse-Repurchase Facility (ON-RRP) and Repo Facility (RP). The two shaded areas are daily transaction volumes in these facilities.

Sources: Federal Reserve Bank of New York, OFR Cleared Repo Collection, Office of Financial Research

Figure 18. Basis Trade Return, Bills Rate, and DVP trades became less profitable. The spread between the Repo Rate (percent) IRR and bill yields then increased, rising well above 2.00 the average over the last two years, and falling around 2.00 Figure 19 2 percent per annum (see ). This disconnect 1.75 peaked around March 17, and then began to normalize, with the IRR falling back into line with both the bill 1.50 rate and actual repo rate. 1.25 All this evidence points to the conclusion that March illiquidity caused losses on the basis trade, which were 1.00 possibly compounded by large sales of the basis — with traders who had bought the basis now selling cash 0.75 Treasuries and buying futures in order to close their 2-Year implied repo rate 0.50 positions — under pressure from margin requirements. 5-Year2-Year implied implied repo repo rate rate 10-Year5-Year implied implied repo repo rate rate In this sense, while we are unaware of any hedge fund 0.25 0.25 10-Year implied repo rate defaults associated with the basis trade during this illi- DVP sponsored borrowing rate 2-MonthDVP sponsored Treasury bill borrowing yield rate 0.00 quidity episode, many of the risks of the basis trade 0.00 2-Month treasury yield appear to have materialized during March. JanJan 1515 FebFeb 01 01 FebFeb 15 MarMar 01 Mar Mar 15 15 AprApr 01

Note: DVP Repo rate is the average overnight rate for sponsored borrowers with Treasury collateral. Implied repo rates are for July contracts.

Sources: Bloomberg Finance L.P., Treasury Daily Yield Curve Rates, OFR Cleared Repo Collection, Office of Financial Research

OFR Brief Series | 20-01 July 2020 | Page 14 Signs of Liquidity in the Cheapest-to-Deliver Figure 19. Spread of Basis Return over Bill Yield These facts alone do not suggest any feedback effects (percentageFigure 19. Spread of Basispoints) Return over Bill Yield (percentage points) from stress in the basis trade to Treasury illiquidity. Spreads peaked to a sample high during March 2.02.0 Rather, there is some evidence that the basis trade 5-Year5-Year continued to provide liquidity through March. 2-Year2-Year

In particular, relative to comparable Treasuries, the 1.51.5 cheapest-to-deliver securities for the futures market traded at higher prices throughout March. Figure 20 shows spreads of Treasuries outside the deliverable set but 1.01.0 with similar maturity dates over the cheapest-to-deliver for the two-year and five-year note futures contract for June 2020. These deliverable spread increases coincided 0.50.5 with increases in bid-ask spreads across Treasuries, and had generally decreased by the beginning of April as stress in the Treasury market fell. 0.00.0 This premium for deliverable Treasuries runs counter

to the narrative that sales of the basis directly harmed -0.5−0.5 Treasury liquidity. If selling pressure from relative Jan19Jan 19 MarMar 19 19 MayMay 19 19 JulJul 19 19 SepSep 1919 NovNov 1919 JanJan 2020 MarMar 2020 value hedge funds had significantly harmed Treasury Note: Spread between the implied repo rate and an equivalent maturity bill yield for the futures Note:contract Spread with the second-to-nearest between the delivery implied date. repo rate and an equivalent liquidity, we might expect the price of the cheap- maturitySources: Bloomberg bill yield Finance for L.P., theCenter futures for Research contract in Security Prices/University with the second-to-of Chicago Booth School of Business, Office of Financial Research est-to-deliver securities to have fallen relative to nearest delivery date. Sources: Bloomberg Finance L.P., Center for Research in Security Prices/ comparable securities as dealers accumulated large net University of Chicago Booth School of Business, Office of Financial exposure to these specific Treasuries. That the premium Research rose suggests that any selling pressure was offset by the liquidity that the basis trade provides and the link it Figure 20. Spread on the Cheapest-to-Deliver 11 establishes to futures markets. This link may have Treasury (percentage points) become particularly valuable during the general flight to liquidity during March, and reduced pressure on 2-Year2-Year 5-Year5-Year dealers purchasing the cheapest-to-deliver. 10-Year 0.200.20 10-Year As a result, while the general evidence points to sales of the basis by hedge funds during March, we do not find conclusive evidence that these sales in turn caused 0.150.15 greater illiquidity in the Treasury market. While many of the risks of this trade seem to have materialized, evidence of spillovers into Treasury liquidity and short- 0.100.10 term funding disruptions are limited. However, it is worth noting that had liquidity not returned to the 0.050.05 Treasury market when it did, and had repo rates not fallen, the consequences for relative value hedge funds could have been much worse. 0.000.00

Effect of Federal Reserve Actions FebFeb 10 FebFeb 24 MarMar 09 09 MarMar 2323 AprApr 06 06 AprApr 20

Timely intervention by the Federal Reserve may have Note: Spread is the yield on a similar maturity nondeliverable been crucial for limiting the extent of hedge fund losses Treasury minus the yield on the cheapest-to-deliver. Cheapest-to- deliver is for June futures contracts. in the basis trade and in preventing broader spillovers. Sources: Bloomberg Finance L.P., Office of Financial Research

OFR Brief Series | 20-01 July 2020 | Page 15 Following March 16, returns on basis trades began to move back into line with the returns on Treasury bills, Figure 21. Cumulative Federal Reserve Purchases of and came closer to the cost of borrowing in the spon- the Cheapest-to-Deliver Securities ($ billions) sored repo market. Several Fed actions on March 16 and 2-Year CTD 2-Year CTD 5-Year CTD 17 may have contributed to this easing of pressure on 88 5-Year CTD hedge funds. In particular, Federal Reserve expansions 10-Year10-Year CTD CTD of Treasury purchases provided an additional source of BondBond CTD CTD demand for off-the-run Treasuries, while expansions 6 of the central bank’s repo facility reduced financing 6 risks associated with providing liquidity to Treasury markets. It is difficult to know exactly which of these 4 actions was most important, in particular because they 4 were mutually reinforcing. The Federal Reserve took the unusual action of including 2 the cheapest-to-deliver Treasury across contracts in its 2 purchases. The direct effect of these purchases may have been limited. Purchases of deliverables for longer 0 duration securities picked up almost immediately after 0 March 15. However, these longer-duration securi- JanJan 0101 JanJan 15 15FebFeb 01 01 FebFeb 15 MarMar 01 MarMar 15 AprApr 01 01 AprApr 15 15 MayMay 01 01 ties seem to have made up a relatively small portion of hedge fund short futures positions. Alternatively, Note: Cumulative purchases of the cheapest-to-deliver (CTD) Treasuries for June delivery. Fed purchases of the most popular two-year and five- Sources: Federal Reserve Bank of New York, Office of Financial Research year cheapest-to-deliver Treasuries were negligible until April. This is consistent with the basis trade still DVP repo rate across segments of the market, including providing liquidity to the market, as dealers may have in the sponsored borrowing segment. Expansion of been more comfortable holding Treasuries for which the repo facility likely reduced these rates by relieving they had a natural source of demand from basis traders. liquidity concerns among dealers. Following this However, the indirect effect of including cheap- expansion on March 16, the sponsored lending rate fell est-to-deliver Treasuries in Federal Reserve purchases to the zero lower bound defined by the Fed’s overnight may have been substantial. Even if purchases of short- reverse repurchase (ON-RRP) facility (see Figure 21). er-maturity cheapest-to-deliver Treasuries on March 16 The rate on sponsored borrowing largely fell in lockstep, were small, the knowledge that the Treasuries could reducing the cost of funding these Treasury positions be sold to the Federal Reserve in the future may have for hedge funds. made dealers more willing to hold these Treasuries, allowing hedge funds to gradually reduce their expo- Conclusion sure to the basis trade. The increase in Federal Reserve purchases in April may then represent concerns over Treasury market illiquidity in March placed severe stress the longer-term profitability of basis trades, which as on hedge fund basis traders. Media articles published in we discuss below is likely to be reduced. Without these the week of March 17 indicate that several relative value actions, dealers may not have been willing to hold the hedge funds experienced serious stress associated with 12 cheapest-to-deliver securities in order to accommodate the basis trade. These losses were driven by margin a gradual withdrawal from the basis trade. calls and repo rate volatility. Under this pressure, hedge funds unwound their positions, selling to dealers whose As these purchases may have made dealers more willing balance sheets were already swelling with Treasuries. to accept cheapest-to-deliver Treasuries, the Fed also lowered the costs of funding these Treasury holdings However, to our knowledge no defaults directly associ- for hedge funds. Fed actions succeeded in lowering the ated with the basis trade actually occurred, suggesting

OFR Brief Series | 20-01 July 2020 | Page 16 that these funds weathered the margin calls during Even if basis trades remain unpopular, it is worth this period. In addition, spreads for the cheapest-to-de- considering the vulnerabilities exposed by difficul- liver securities across contracts suggest that this trade ties in March. The transition of Treasury security continued to provide liquidity relative to comparable ownership toward highly leveraged funds exposes the off-the-run Treasuries. Rapid action by the Federal market to sell-offs because small changes in prices can Reserve in expanding its purchases of Treasury secu- trigger large losses. Expanded dealer Treasury exposure rities and its repo facilities may have stemmed much could exacerbate these sell-offs, hampering the ability of the potential impact of these trades on Treasury of those dealers to make markets. Finally, the role illiquidity. of repo markets in financing both dealers and leveraged Treasury holders, as well as the importance of Meanwhile, further reductions in hedge fund short Treasury collateral, tie short-term funding markets to positions suggest that the popularity of the basis the Treasury market, leading to potential spillovers into trade has decreased, with total short positions at $554 broader wholesale funding markets. Future episodes of billion as of April 28, a decline of almost 30 percent Treasury market instability therefore remain a possi- from mid-February. Basis trades remain unattractive bility and may trigger continued intervention by the for hedge funds. While the IRR has remained near Federal Reserve. zero, the spread that sponsors charge borrowers over their lending rates imposes an effective lower bound on hedge fund borrowing rates of around 12.5 basis points. (A basis point is one-hundredth of a percentage point.) At the same time, the implied repo rate has generally been in the range of 5 basis points for the two-year and five-year contracts, in line with bill rates, and has been below zero for the 10-year contract. As a result, net returns on new basis trades are likely to be negative in the immediate future. That a substantial portion of the outstanding of the cheapest-to-deliver securities has been purchased by the Federal Reserve without a substantial increase in implied repo rates further underscores the possibility that demand for basis trades will be tepid going forward. Nevertheless, risks of basis trades persist. Hedge funds still hold large short Treasury futures positions. Looking forward, relatively large amounts of short- term Treasury bill issuance may support returns on the basis trade, while other participants such as foreign official accounts appear to be at least temporarily satisfied with the liquidity buffers they established in March. However, relatively high costs of funding for hedge funds, combined with low returns on the basis trade, may lead to further sales of the basis and reduce the liquidity these trades provide to longer-term Treasury notes and bonds. The high leverage of hedge funds involved in basis trades continues to be a cause for concern. These funds remain exposed to sudden bouts of illiquidity in Treasury and repo markets.

OFR Brief Series | 20-01 July 2020 | Page 17 Endnotes

1 Daniel Barth, Senior Economist, Board of 5 For instance, the deliverable basket for the 10 There are two ways to reduce futures exposure: Governors of the Federal Reserve System two-year contract contains Treasuries with an for contracts maturing in March, hedge funds ([email protected]), and Jay Kahn, original maturity of less than five years and may have simply not rolled over into the June Research Economist, Office of Financial three months and a residual maturity on the last contract. For contracts maturing after March, Research ([email protected]). The day of the delivery month between one year and hedge funds would have to take on offsetting authors thank Amanda Buckley, Katherine eight months and two years. long positions. Gleason, Maryann Haggerty, Matthew McCormick, Sriram Rajan and Stacey Schreft 6 Specific collateral repo rates can vary for 11 It is possible that in order to keep basis trades for their comments and assistance. individual securities. If the repo rate on the open while meeting margin calls, hedge funds cheapest-to-deliver is particularly high, the may have sold Treasuries other than the cheap- 2 Previous research, in particular Andreas cheapest-to-deliver may not actually be the est-to-deliver, thus contributing to the lower Schrimpf, Hyun Song Shin, and Vladyslav most attractive Treasury to deliver into the price of other securities. It is difficult to reject Sushko “Leverage and margin spirals in fixed futures position. this possibility without more detailed data on income markets during the Covid-19 crisis,” hedge funds’ Treasury holdings. However, even BIS Bulletin No. 2, April 2, 2020 (https:// 7 These post-crisis limits stem mostly from the in this case the willingness of hedge funds to www.bis.org/publ/bisbull02.pdf), has discussed Dodd-Frank Wall Street Reform and Consumer sell other Treasuries to keep their basis trades the role of the basis trade in March Treasury Protection Act of 2010 and from the interna- open would still indicate excess demand for the illiquidity. While we confirm their results that tional Basel III standards. trade. losses on the trade occurred during March, we 8 See Joe Rennison, “Repo blame game moves provide evidence that feedback into Treasury 12 See for instance Stephen Spratt, “How a focus to hedge funds,” Financial Times, market illiquidity may have been limited. Little Known Trade Upended the U.S. Dec. 13, 2019. https://www.ft.com/content/ Treasury Market,” Bloomberg, March 17, 3 Throughout this brief, when we construct long 6427f16a-1d05-11ea-97df-cc63de1d73f4. 2020. https://www.bloomberg.com/news/ series of the returns on the basis trade, we use a articles/2020-03-17/treasury-futures-domino- 9 Sales from these foreign official accounts may roll date for the futures contract that avoids the that-helped-drive-fed-s-5-trillion-repo; Juliet have had particular importance for Treasury delivery month, so as to avoid any excess vola- Chung, “Hedge Funds Hit by Losses in Basis market illiquidity. Primary dealers are required tility caused by optionality in physical delivery. Trade,” The Wall Street Journal, March 19, to make “reasonable” markets for sales of In particular, we choose the futures contract 2020. https://www.wsj.com/articles/hedge- Treasuries by these accounts. Additionally, that on any given day has the second-nearest funds-hit-by-losses-in-basis-trade-11584661202; the funds from these sales seem to have been delivery date. Sonali Basak, Liz McCormick, Donal Griffin, invested in significant part in the Federal and Hema Parmar, “Before Fed Acted, Leverage Reserve’s foreign repo pool, which effectively 4 For 10-year note contracts and bond contracts, Burned Hedge Funds in Treasury Market,” removes reserves from the system, potentially the CBOT offers additional “ultra” futures, Bloomberg, March 19, 2020. https://www. making repo financing of Treasuries more with narrower maturity ranges eligible for bloomberg.com/news/articles/2020-03-19/ expensive delivery. before-fed-acted-leverage-burned-hedge-funds- in-treasury-trade.

OFR Brief Series | 20-01 July 2020 | Page 18