Hedge funds and the Treasury cash-futures disconnect

Hedge funds and the Treasury cash-futures disconnect

Jay Kahn 1 Daniel Barth 2

1Office of Financial Research, U.S. Department of the Treasury2Board of Governors of the Federal Reserve System

2021 Financial Stability Conference

Disclaimer: The views expressed in this presentation are those of the speaker and do not necessarily representthe views of the Office of Financial Research, the U.S. Department of the Treasury or the Board of Governors of theFederal Reserve System.

Introduction Cash-futures disconnect Hedge fund basis trades Limits to arbitrage March 2020 illiquidity Conclusion

March 2020 disruption in Treasury markets

2006 2008 2010 2012 2014 2016 2018 2020

4

6

8

10

12

14

16 Treasury VIX (left axis)MOVE Index (right axis)

50

100

150

200

250

Jan Feb Mar Apr May Jun0.0

0.5

1.0

1.5

2.0

2.5

3.02-Year5-Year10-Year30-Year

I Most dramatic increase in Treasury volatility since at least 2009.


During March, hedge fund Treasury exposure decreased by >$200BDocument a major shift in Treasury markets:I HF Treasury exposure increased by $960B between 2017 and 2019.

2016-01 2016-07 2017-01 2017-07 2018-01 2018-07 2019-01 2019-07 2020-01

−200

0

200

400

600

800

1000

1200

1400Yo

Y ch

ange

($ b

illion

s)Banks and dealersPension and insuranceAsset managersHedge funds

In this paper we ask three questions:1. Why were hedge funds holding so many Treasuries?

2. Why did they sell so many in March 2020?

3. What, if any, are the implications for financial stability?


Why were hedge funds holding so many Treasuries?

1. >70% associated with disconnect of cash and futures prices.I Treasury futures overvalued to replicating cash portfolio.I Disconnect made arbitrage through the repo market profitable.I Rely on mixture of public aggregates and regulatory micro data to link to

hedge funds.

2. Model shows trade facilitates risk-sharing and economizes onbalance sheet costs between dealers and asset managers.I Basis traders act as warehouses for Treasuries.

I Exposed to both margin and rollover risk.I Present empirical evidence consistent with model.I Present evidence of effect on underlying cash securities.

3. Discuss how trade contributed to March illiquidty.I Roughly $100B in sales associated with basis trade.I However, prices of Treasuries associated with the trade rose during March.I Timely intervention by the Fed may have prevented further spillovers.


Cash and futures prices of Treasuries related through arbitrage

255075100Days to delivery

−2.0

−1.5

−1.0

−0.5

0.0

0.5

Spre

ad o

ver l

ast i

nvoi

ce FuturesCash

0

50

100

150

Open

Inte

rest

($ b

illion

s)

5-Year

2011 2012 2013 2014 2015 2016 2017 2018 2019 2020−0.5

0.0

0.5

1.0

Spre

ad

2-Year arbitrage deviations

2011 2012 2013 2014 2015 2016 2017 2018 2019 2020−0.5

0.0

0.5

1.0

Spre

ad

5-Year arbitrage deviations

0

100

200

300

400

Open inte

rest

0

100

200

300

400

Open inte

rest

In a frictionless market:

Pt,τ︸︷︷︸Note price

=T∑

s=t

Bt,scs︸︷︷︸Intervening coupons

+ Bt,T Ft,τ,T︸︷︷︸Discounted futures price

.

Convergence only occurs for cheapest-to-deliver (CTD) Treasuries.

Establish empirical counterpart: cash prices from CRSP, futures pricesfrom Bloomberg.I Show that this relationship does not hold in recent years.

Non-convergence Deliverability premium Past crises


The cash-futures basis trade

Treasury market: Purchase Treasury

Deliver Treasury to repo lender

...

Futures market:

Repo market:

Short futures contract

Open repo trade

t = 0

Roll over repo

t = 1

Roll over repo

t = T − 1

Close repo trade

t = T

Deliver Treasury to futures

Receive cash from futures

In reality, arbitrageurs rely on repo financing to arbitrage this spread.I Short futures, long cash, funded using the cash note as collateral.I Borrowing appears to be largely overnight.

Trade is exposed to two sources of risk:1. Rollover risk on repo borrowing.

2. Margins on short futures position.

Risks compounded by leverage on the trade, in principle only limited byTreasury haircut.


The basis trade and hedge fund Treasury exposure0

500

1000

1500

Trea

sury

Exp

osu

res

($ B

illio

ns)

2013 2014 2015 2016 2017 2018 2019 2020

Total Short (Form PF)

Short Futures (CFTC)

Total Long (Form PF)

Long Futures (CFTC)

-.2

0.2

.4.6

Spre

ad (%

)

-200

020

040

060

0A

ggre

gat

e N

et R

epo E

xposu

re ($

Bill

ions

)

2014 2015 2016 2017 2018 2019 2020

Net Repo BorrowingArbitrage Spread

Form-PF data consistent with large basis positions.I In 2017, gross HF Treasury exposures were $1.06T and net repo

borrowing was -$49B.I In 2019, gross HF Treasury exposures were $2.02T and net repo

borrowing was $535B.Classify funds as “large basis traders” using their net repo borrowingand Treasury exposures.I In 2019, >$714B (60%) of HF gross Treasury exposure and $505B

(94%) of net repo borrowing was attributable to basis traders.I 30% of primary dealer repo lending against Treasuries.

Basis traders have median leverage of 17.6 and average leverage of 21.Repo detail Futures


Evidence from hedge fund repo

Intermediaries:I Broker dealersI Large banks

Cash Lenders:I Money market fundsI Pension and insurance fundsI Small banks

Cash Borrowers:I Hedge fundsI Others

DVP Interdealer

DVP Sponsored Lending

Uncleared Bilateral Lending Uncleared Bilateral Borrowing

DVP Sponsored Borrowing

Average rate: 0.44

Average rate: 0.39

Average rate: 0.33

Hedge funds borrow in two repo markets:I DVP sponsored borrowing (cleared bilateral).I Uncleared bilateral repo.

Use OFR DVP repo data for additional evidence.I All trades since October 2019.I Provides detail on collateral, rates, counterparties.I Hand-mapped to hedge funds.

Data on DVP important since repo markets are segmented.I Compare to tri-party repo rate of 0.34.

Sponsored repo


Data from DVP repo also consistent with large basis positions

2021-01 2022-01 2023-01 2024-01 2025-010

1

2

3

4

2 year 5 yearPrior to Dec. 1, 2019

Kernel density

Deliverable set

2021-01 2022-01 2023-01 2024-01 2025-01Maturity date

0

1

2

3

4

2 year 5 yearAfter and including Dec. 1, 2019

All outstanding hedge fund Treasury position by collateral maturity.I Deliverable maturities shaded gray.I Borrowings using the cheapest-to-deliver are particularly high:

I ≈ $4.5B, or 10% of outstanding. Reduced directly prior to first delivery date.

Especially large given only fraction of trades likely to be funded in DVP.


Model of the cash-futures disconnect

Show disconnect results from limits to arbitrage & segmented markets.

Hedge Funds

Repo dealer Treasury dealer

Money market fund Treasury seller

Clearing house

Asset manager

Repo market Treasury market

Futures marketI Emphasizes HF role in risk sharing

between asset managers anddealers.I Hedge funds act as warehouses

for Treasuries.I Limits to arbitrage mean

warehousing is imperfectI Dealers face more risk than optimal.

I But basis trades also exposeTreasury market to potential risks.I Margin constraints and repo

market tightening→ suddendecreases in prices.

I Can be triggered by sales by noisetraders.

Model detail


Empirical evidence on limits to arbitrage

Univariate regressionsCF Disconnect 2-Year 5-Year 10-Year BondGCF - IOER 0.30∗∗∗ 0.36∗∗∗ 0.52∗∗∗ 0.30∗∗

Dealer exposure 1.13∗∗∗ 1.74∗∗∗ 2.23∗∗∗ 1.44∗∗∗

Multivariate regressions|CF Disconnect| 2-Year 5-Year 10-Year BondMaitenance margins 0.051 0.450∗∗∗ 0.214∗∗ 0.070∗∗∗

VIX 0.665∗∗∗ 0.271 0.640∗∗∗ 1.555∗∗∗

Regressions use second-to-deliver.I GCF-IOER as proxy for repo spread.I Net Treasury exposure from primary dealer statistics.I Use VIX because Treasury volatility measures tied to futures.I Maintenance margins from CME group.I Fixed effects to control for distance to delivery.


Onset of March illiquidity

May 2018

Sep 2018

Jan 2019

May 2019

Sep 2019

Jan 2020

May 2020

Sep 2020

Jan 2021

−200

−100

0

100

200

Diff

erence from

2

/2

6/2

0 ($

billions)

March, 2020Foreign official Treasuries

Foreign repo pool

I Dealer balance sheets saturated withTreasuries. (top)

I Large sales from foreign officialaccounts and mutual funds inresponse to COVID-19. (bottom)

I Cost of making markets rose, leadingto rising volatility.


Basis trade unwind

Jan 01 Feb 01 Mar 01 Apr 01 May 01−1000

−750

−500

−250

0

250

500

750

1000Futures value

5-year futures margin95% of price changes 2018-2019

Jan 15 Feb 01 Feb 15 Mar 01 Mar 15 Apr 01

0.00

0.25

0.50

0.75

1.00

1.25

1.50

1.75

2.00

2-year futures-implied yield

5-year futures implied yield

DVP sponsored borrowing rate

2-month Treasury yield

As volatility rose, margins were breached in early March. (left)I Moves large relative to 95% of movements 2018-2019 (green area).I CBOT responded by raising margins, increasing cost of basis trade.

Hedge funds exited the basis trade:I $105B decrease in hedge fund short futures exposure.I $104B decrease in large basis trader long Treasury exposures.

As a result, cash-futures disconnect increased. (right)Introduction Cash-futures disconnect Hedge fund basis trades Limits to arbitrage March 2020 illiquidity Conclusion

Deliverability premium during MarchPremium on CTDs suggests trade continued to provide liquidity tounderlying.

Jan 01 Feb 01 Mar 01 Apr 01 May 01 Jun 01 Jul 01

−0.025

0.000

0.025

0.050

0.075

0.100

0.125

0.150

Delivera

bilit

y p

rem

ium

2-year

5-year

Bond

10-year

2022 2023 2024 2025 2026 2027 2028 2029 2030 2031Maturity date

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Yie

ld

Fitted yieldActual yieldDeliverable yields

Cheapest-to-deliver

I Despite selling pressure, CTD price particularly high during March. (left)I Especially high for the more popular securities (2-year, 5-year).I Concentrated in the CTD in a snapshot on March 11. (right)

Suggests dealers more willing to hold CTDs during March.


Federal reserve may have intervened at exactly the right time

Jan 15 Feb 01 Feb 15 Mar 01 Mar 15 Apr 01 Apr 15 May 01−0.2

0.0

0.2

0.4

0.6

0.8

Spre

ad

over

FF m

idpoin

t

ON-RRP facility

InterdealerSponsored lenderSponsored borrower

Jan 15 Feb 01 Feb 15 Mar 01 Mar 15 Apr 01 Apr 15 May 010

50100150200250300350

Facilit

y v

olu

me

RRP FacilityRP Facility

Days RP capacity exceeded

RP facility

Date Jan 01 Feb 01 Mar 01 Apr 01 May 01

0

2

4

6

82-Year CTD5-Year CTD10-Year CTDBond CTD

Support was provided to dealers on two fronts:I Expansion of the repo facility lowered borrowing costs and made

carrying Treasuries less expensive.I Inclusion of CTD in purchasese made it easier for dealers to wait out

stress.


Conclusion

In the wake of March stress, focus on hedge fund holdings ofTreasuries.I A sizeable amount of these holdings are associated with the

cash-futures basis trade.I Show this trade became profitable because of a disconnect between

Treasury cash and futures prices.

We emphasize this trade is popular because of broad changes inTreasury market structure.I Hedge funds act as warehouses for Treasuries when dealers and asset

managers find them costly to hold.I Consistent with empirical evidence on the behavior of the cash-futures

disconnect.

The model also emphasizes the risk that the trade amplifies dashes forcash.I While basis trading appears to have decreased, this risk is still present.


Appendix


Size of hedge fund sales over March

Combine data figures from OFR annual report derived from PF and from theCFTC Traders in Financial Futures:

Long Short Feb-MarFeb Mar Feb Mar Long Short

Exposure (OFR AR) 1,406 1,164 905 735 242 170Less derivatives (TFF) 311 275 721 618 36 102

Cash estimate 1,095 889 184 116 206 68Price adjustment (VBILX) 30 5

Cash sales 236 73

Compare to:

I $35 billion estimated decrease in flow of funds.I $250 billion decrease from Cayman islands in TIC data.

Of a similar size to sales by foreign official accounts and mutual funds.


Non-convergence of cash and futures prices for other deliverable Treasuries

20406080100120−2.0

−1.5

−1.0

−0.5

0.0

0.5

1.0

Spre

ad o

ver l

ast i

nvoi

ce

1st cheapest-to-deliver

FuturesCash

20406080100120−2.0

−1.5

−1.0

−0.5

0.0

0.5

1.02nd cheapest-to-deliver


−2.0

−1.5

−1.0

−0.5

0.0

0.5

1.0

Spre

ad o

ver l

ast i

nvoi

ce

3rd cheapest-to-deliver


−2.0

−1.5

−1.0

−0.5

0.0

0.5

1.04th cheapest-to-deliver

Arbitrage requires that cash and futures prices convergeI Show deviations from final invoice price for 5-year deliverables:

P̃t,τ,T − FT ,τ,T Ft,τ,T − FT ,τ,T

Back Introduction Cash-futures disconnect Hedge fund basis trades Limits to arbitrage March 2020 illiquidity Conclusion

Deliverability premium

Convergence solely for the CTD provides a close link between arbitragetrades and the CTD.I How does this affect cash prices of the CTD?

2008 2010 2012 2014 2016 2018 2020

0.0

0.2

Yield

spre

ad (

%)

Cheapest-to-deliverOn-the-run

2008 2010 2012 2014 2016 2018 2020

−0.10

−0.05

0.00

0.05

Yield

spre

ad (

%)

2008 2010 2012 2014 2016 2018 2020

0.0

0.2

0.4

Yield

spre

ad (

%)

2008 2010 2012 2014 2016 2018 2020

0.0

0.1

Yield

spre

ad (

%)

I Calculate error from a fittedyield curve spline.

I Compare to the error for asimilar maturity on-the-run.

CTD premium2-year 5-year

OTR prem. 1.167∗∗∗ 0.492∗∗∗

R2 0.832 0.532

10-year BondOTR prem. 0.115∗∗∗ 0.631∗∗∗

R2 0.044 0.109Back


Cash-futures disconnect: past crises

1998-01 1998-04 1998-07 1998-10 1999-01 1999-04 1999-07 1999-10 2000-01−1.5

−1.0

−0.5

0.0

0.5

1.0

1.5

2.0

Spre

ad

Ruble devaluation LTCM bailout

0

5

10

15

20

25

30

35

40

Open

inte

rest

2008-01 2008-04 2008-07 2008-10 2009-01 2009-04 2009-07 2009-10 2010-01−5

−4

−3

−2

−1

0

1

2

Spre

ad

Lehman Brothers' Bankruptcy

0

20

40

60

80

100

Open

inte

rest

Cash-futures disconnect responds to stress in financial marketsI LTCM: Price of futures rose relative to cash (left).

I Liquidity crisis, similar to March 2020.I LTCM had non-trivial bond basis positions.

I Lehman Brothers: Price of cash rose relative to futures (right).I Dash to safety rather than cash.

Both suggest the importance of limits to arbitrage in the disconnect.Back


Hedge fund net repo40

060

080

010

0012

0014

00A

ggre

gat

e Rep

o E

xposu

re ($

Bill

ions

)

2013 2014 2015 2016 2017 2018 2019 2020

Total Repo BorrowingTotal Repo Lending

Back


Large Treasury futures positions0

200

400

600

800

$ Bi

llions

2013 2014 2015 2016 2017 2018 2019 2020

Back


Hedge fund participation in sponsored repoI Classification of hedge funds part of broader OFR effort on mapping

cleared repo.I Hedge funds represent the major source of sponsored borrowing:

I Over 2020, sponsored lending made up around 20% of all DVPtransactions, sponsored borrowing around 13%.

Back


Model of the cash-futures disconnect

Show disconnect results from limits to arbitrage & segmented markets.

Hedge Funds

Repo dealer Treasury dealer

Money market fund Treasury seller

Clearing house

Asset manager

Repo market Treasury market

Futures market

Four assets: bills (PN,t ), notes (PN,t ),repo (BRR,t ) and futures (Ft ).

Four agents:I Dealers: Mean-variance preference,

choose portfolio of bills and notes.I Speculators: Mean-variance

preference, hold futures to delivery.I Money market funds: Liquidity

tradeoff between bills and repo.I Hedge funds: Risk neutral, subject

to margin constraints, fund basispositions in the repo market.

Futures deliver at time 2, at time 1 the note market has net supply, S.


Model equilibrium

Margin E [Futures return]

E [Cash return]

No arbitrageNo arbitrageArbitrage capacity

Risk sharing

More risk to speculators

I Given a supply of notes, there is a fixed amount of short-rate risk att = 2 that must be split between dealers and speculators.

I As more risk is allocated towards speculators, futures prices fall whilecash prices rise (mean-variance preferences).


Model equilibrium


E [Cash return]


Risk sharing


I In absence of limits to arbitrage, HFs act as perfect warehouses forTreasuries:

E [Cash return] = E [Futures return]

I Risk shared optimally between dealers and speculators, no disconnectbetween cash and futures prices.


Model equilibrium


E [Cash return]


Risk sharing


I In presence of limits to arbitrage, HFs act as imperfect warehouses forTreasuries:

E [Cash return] > E [Futures return] (cash undervalued)

I Dealers end up with more risk than optimal, disconnect results betweencash and futures prices.


Risks of the trade can impact the cash market

Increase in net supply

Margin E [Fut return]

E [Cash return]

Arbitrage capacity

Risk sharing

Risk sharing′′

Risk sharing′

Increase in margins

Marg E [Fut return]Marg′

E [Cash return]

Arbitrage capacity

Risk sharing

I Increase in net supply can lead to large increase in prices.I Increases in margins shift in arbitrage capacity, force sales by HFs.I Repo illiquidity increases disconnect and leads to sales by HFs.

Back


Hedge funds and the Treasury cash-futures disconnect

Documents

Transcript of Hedge funds and the Treasury cash-futures disconnect