Mutual Fund Liquidity ransfoTrmation and Reverse Flight to ...

Mutual Fund Liquidity Transformation

and Reverse Flight to Liquidity

Yiming MaColumbia GSB

Kairong XiaoColumbia GSB

Yao ZengWharton

2020 Financial Stability ConferenceNovember 2020

Yiming Ma Kairong Xiao Yao Zeng Mutual Funds and Reverse Flight to Liquidity 1

Motivation

I Covid-19 crisis: traditionally liquid assets became strainedfrom unusually high selling pressuresI e.g. Haddad, Moreira, and Muir 20; Du�e 20; He, Nagel, and

Song 20; Kargar et al. 20

I Reverse �ight to liquidity

I Conventional wisdom: �ight to liquidity during marketdistressI high buying pressure for liquid assets


Heightened Selling Pressure in Treasury Market

-.4-.3

-.2-.1

0.1

US

Trea

sury

CD

S-ad

just

ed a

djus

ted

swap

spr

eads

(%)

Jan Feb Mar Apr May

Treasury discount ↓= Interest swap yield + Treasury CDS - Treasury yield ↑


Motivation

I The conventional negative correlation between Treasury yieldand VIX is reversed in Covid-19 crisis

Figure: Treasury Yields and VIX in Jan-Apr 2020

01-0201-0301-0601-07

01-0801-09

01-1001-1301-1401-1501-1601-1701-2101-2201-23

01-2401-27

01-2801-2901-30

01-3102-0302-04

02-0502-0602-0702-1002-11

02-1202-1302-1402-1802-1902-20

02-2102-24

02-2502-26 02-27

02-2803-0203-0303-04

03-05

03-06

03-09

03-10 03-11

03-1203-13

03-16

03-17

03-1803-19

03-20

03-2303-24

03-2503-26

03-2703-3003-31

04-0104-0204-0304-0604-07

04-0804-0904-1304-14

04-1504-16

04-1704-20

04-2104-22

04-2304-2404-2704-28

04-2904-30

11.

52

2.5

30-y

ear T

reas

ury

yiel

d (%

)

20 40 60 80VIX


Motivation

I Conventional wisdom: �ight to liquidity

Figure: Treasury Yields and VIX in Aug-Sep 2008

08-0108-04

08-05

08-06

08-0708-08

08-11

08-1208-13

08-14

08-1508-1808-19

08-2008-2108-22

08-2508-2608-2708-28

08-29

09-0209-03

09-0409-0509-08

09-0909-1009-11

09-12

09-1509-16 09-17

09-18

09-19

09-2209-23

09-2409-2509-26

09-29

09-30

44.

24.

44.

64.

830

-yea

r Tre

asur

y yi

eld

(%)

20 30 40 50VIX


Heightened Selling Pressure in IG Corp Bonds

-4-3

-2-1

0C

DS-

bond

bas

is (%

)

Jan Feb Mar Apr May

IG HY

CDS-bond basis ↓= corporate bond CDS spread ↑ - corporate bond spread ↑↑↑


Motivation

The market for U.S. Treasuries has long been viewed as the

world's most liquid �nancial market. That presumption was

questioned...

�Still the World's Safe Haven?� Darrel Du�e


What caused the reverse �ight to liquidity phenomenon?

I Bond mutual fundsI Liquidity transformation → large redemption in crisisI Pecking order of liquidation → concentrated sale of liquid

assets

I Implications of the reverse �ight to liquidity phenomenon

I With more liquidity transformation by bond fundsI Liquid assets may experience more volatilityI May steer central banks towards providing liquidity backstops

to bond funds


Literature

1. Disruptions in liquid asset market during Covid-19I Haddad, Moreira, and Muir 20 and Kargar et al. 20: document

debt market disruptionsI Du�e 20, He, Nagel, and Song 20: dealer constraintsI Falato, Goldstein and Hortacsu 20: fragility in fund �ows

I A new explanation for heightened selling pressure in liquid assets

through mutual fund liquidity transformation

2. Financial intermediation by banks versus non-banksI E.g. Diamond and Dybvig 83, Goldstein and Pozner 05, Hanson,

Shleifer, Stein, and Vishny 15, Ma, Xiao, and Zeng 20

I Asset-side implications of liquidity provision by mutual funds

versus banks

3. Mutual fund �ows and �nancial stabilityI E.g. Chen, Goldstein, and Jiang 10, Goldstein, Jiang, and Ng 17,

Chernenko and Sunderam 17, Jiang, Li, and Wang 20

I ↑ fund asset illiquidity → ↑ potential for liquid asset market

disruptions


Roadmap

1. Aggregate Trends

2. Empirical Evidence


Bond Fund Growth

I Bond mutual funds have become increasingly important...

Total Assets

010

0020

0030

0040

0050

00Fu

nd a

sset

s (b

illion

)

1995 2000 2005 2010 2015 2020

Fund Shares/Deposits

.15

.2.2

5.3

.35

Fixe

d-in

com

e fu

nd s

hare

s / B

ank

depo

sits

1995 2000 2005 2010 2015 2020

I AUM: $1 trillion to $4.5 trillion from 2007 to 2019

I Shares/Deposits: 0.22 to 0.35 from 2007 to 2019


Mutual Fund Liquidity Transformation

Bank

Illiquid assets

Liquid assets

Demandable debt

Bond funds

Illiquid assets

Liquid assets

Demandable equity

I Like banks, bond funds provide liquidity by issuing liquidclaims backed by (mostly) illiquid assets

I Liquidity transformation may lead to �rst-mover advantage,e.g. Market value drops from 100 to 80 but NAV is still at 90


Pronounced Out�ows at Bond Funds

I Bond funds su�ered an unprecedented $264 billion out�ows inMarch, 2020

Bond funds

-300

-200

-100

010

0Fu

nd fl

ows

(billi

on)

2000 2005 2010 2015 2020

Equity funds

-300

-200

-100

010

0Fu

nd fl

ows

(billi

on)

2000 2005 2010 2015 2020

I Equity funds su�ered much smaller out�ows → liquiditytransformation must matter...


Higher Asset Illiquidity, Larger Out�ows

I ...also evident from more pronounced out�ows at more illiquidbond funds

-20

-10

010

Cum

mul

ativ

e fu

nd fl

ows

(% A

UM

)

Jan2020 Feb2020 Mar2020 Apr2020 May2020

Government Investment gradeHigh yield Loan


Concentrated Sales of Liquid Assets

I Liquid assets were disproportionately liquidated

-10

-50

Posi

tion

chan

ge (%

)

<CC

C-

CC

C-

CC

CC

CC

+ B- B B+ BB-

BB BB+

BBB-

BBB

BBB+ A- A A+ AA

-AA AA

+AA

ATr

easu

ry


Large Aggregate Liquidations

I Aggregate Treasury liquidations by open-end funds amountedto $236 billion in 2020Q1

-300 -100 100 300Change in Treasuries (billion)

Money market fundsETFs

Insurance companiesNonfinancial business

GSEsBanks

Closed-end fundsABS

Credit unionsPension funds

Foreign banking offices in U.S.Broker dealers

HouseholdsMutual funds

Rest of the world


Empirical Evidence

Liquidity Transformation and Out�ows: Fund-level Evidence

Figure: Fund Out�ows versus Fund Liquidity Provision Index (LPI)

02

46

Fund

out

flow

s

0 2 4 6 8Liquidity Provision Index (LPI)

I LPI = per dollar expected contract payment - directliquidation value assets (Ma, Xiao and Zeng 19)


Cross-section Determinants of Out�ows

I High asset illiquidity → more out�ows

(1) (2) (3) (4)Out�ow Out�ow Out�ow Out�ow

Fund illiquidity 0.546∗∗∗ 0.545∗∗∗ 0.537∗∗∗ 0.564∗∗∗

[0.092] [0.092] [0.094] [0.096]

Institutional 0.137 0.135 0.283[0.203] [0.203] [0.204]

Index fund -0.259 -0.614[0.595] [0.602]

Fund objective F.E. No No No YesObservations 6,355 6,355 6,355 6,355Adj. R-squared 0.005 0.005 0.005 0.021

I Other controls: yield, return volatility, expense ratio, turnoverratio


What did funds sell when they face redemption?

Reverse Flight to Liquidity: Pecking Order

Liquidationsi,j = βFund Out�owj + γXi + θXj + εi,j

Figure: Liquidation to Redemption Sensitivity by Bond Rating-.5

0.5

11.

52

<CC

C-

CC

C-

CC

C

CC

C+ B- B B+ BB-

BB BB+

BBB-

BBB

BBB+ A- A A+ AA

-

AA AA+

AAA

Trea

sury


Pecking Order based on Relative Liquidity: Measure

Ranki,j =∑i′

Sharei′,j × 1 [Liquidityi > Liquidityi′ ]

I Ranki,j is de�ned as the sum of shares of assets held by fund jthat are less liquid than bond i


Pecking Order based on Relative Liquidity: An Example

I Compare two identical Treasury bonds held by (1) a Treasuryfund and (2) a corporate bond fund

I Treasury bond fund has more liquid assets → ranked lower inthe pecking order

I Corporate bond fund has more illiquid assets → ranked higherin the pecking order

I Corporate bond fund is more likely to sell the Treasury bondfor the same redemption


Reverse Flight to Liquidity: Pecking Order

Liquidationi,j = βOut�owsj × Ranki,j + γXi,j + εi,j

(1) (2) (3)Open-end

AllOpen-endActive

Open-endPassive

Out�ows 0.268∗∗∗ 0.208∗∗∗ 0.224∗∗∗

[0.016] [0.024] [0.035]

Out�ows*Rank 0.198∗∗∗ 0.386∗∗∗ 0.082[0.047] [0.060] [0.151]

Bond F.E. Yes Yes YesFund objective F.E. Yes Yes YesObservations 79,050 37,945 39,410Adj. R-squared 0.067 0.071 0.050


Did funds' selling pressure a�ect prices of liquid

assets?

Bond yields by liquidation rank

Figure: Pecking order of liquidation and bond yield

-40

-20

020

40Yi

eld

(bps

)

Feb Mar Apr May

Q1 Q2Q3 Q4


Impact on Asset Prices

Imputed Out�owi,t =∑j

Fund out�owj,t ×Holdingi,j,2019Q4∑k Holdingi,k,2019Q4

Returni,t = βImputed Out�owi,t × High ranki + γXi,t + εi,t

(1) (2) (3)Return Return Return

Out�ows -1.646 -1.741 -1.856[1.167] [1.174] [1.194]

Out�ows*High rank -3.443∗∗ -3.550∗∗ -3.506∗∗

[1.426] [1.433] [1.459]Rating-Time F.E. Yes No NoMaturity-Time F.E. Yes No NoIssuer F.E. No Yes YesBond F.E. No No YesObservations 398,678 398,678 398,677Adj. R-squared 0.174 0.175 0.166


What were the e�ects of Fed interventions ?

Fed Intervention

Unprecedented intervention by the Fed in �xed income markets:

I March 15: Announced purchase of Treasuries

I March 23: Announced purchase of IG corporate bonds

I April 9: Announced extension of corporate bond purchaseI Larger volumeI Include �fallen angels�


E�ect of Fed Interventions on Fund Flows

Outflowsi,t = βFedInterventiont + γXi,t + εi,t

(1) (2) (3) (4)Govt IG HY BL

Treasury bond purchase 0.014 0.043 0.093∗∗∗ 0.093∗∗∗

[0.032] [0.039] [0.026] [0.033]

Corp bond purchase 0.217∗∗∗ 0.366∗∗∗ -0.057 0.161∗∗∗

[0.040] [0.063] [0.035] [0.051]

Corp bond purchase expansion 0.057∗∗∗ -0.039∗∗ -0.069∗∗∗ -0.073∗∗∗

[0.017] [0.019] [0.008] [0.010]

Control Yes Yes Yes YesFund �xed e�ects Yes Yes Yes YesMonth �xed e�ects Yes Yes Yes YesObservations 13,573 6,452 17,819 9,364Adj. R-squared 0.058 0.071 0.067 0.096

I Announcements of corp bond purchase stopped out�ows, but Treasurybond purchase did not


E�ect of Fed Interventions on Fund NAV

∆NAVi,t = βFedInterventiont + γXi,t + εi,t

(1) (2) (3) (4)Govt IG HY BL

Treasury bond purchase 0.241∗∗ 0.188 -2.336∗∗∗ -2.618∗∗∗

[0.119] [0.236] [0.274] [0.290]

Corp bond purchase 0.901∗∗∗ 0.866∗∗ -1.261∗∗ -1.941∗∗∗

[0.178] [0.400] [0.486] [0.487]

Corp bond purchase expansion 0.399∗∗∗ 1.616∗∗∗ 1.621∗∗∗ 1.114∗∗∗

[0.039] [0.066] [0.111] [0.095]

Control Yes Yes Yes YesFund �xed e�ects Yes Yes Yes YesMonth �xed e�ects Yes Yes Yes YesObservations 13,565 6,451 17,816 9,357Adj. R-squared 0.180 0.427 0.323 0.377

I Announcements of corp bond purchase increased NAVs, but Treasurybond purchase did not


Why did reverse �ight to liquidity emerge this time?

I Increased liquidity transformation by funds since last crisis

I Bond fund AUM: $1 trillion to $4.5 trillion from 2007 to 2019

I With increased mutual fund liquidity transformation, goingforward:I Increased volatility in traditionally liquid assetsI "Are US Treasury Bonds Still a Safe Haven?" (He and

Krishnamurthy, 2020)I Mutual funds conduct liquidity transformation outside public

liquidity backstopsI More reliance on unconventional central bank asset purchases


Conclusion

1. What caused the reverse �ight to liquidity phenomenon?

Mutual funds transform liquidity by issuing demandable equityshares + using pecking order of liquidations

2. What are the policy e�ects?The purchase of illiquid assets directly alleviates fund out�owsand is more e�ective than the purchase of liquid assets

3. What was special about the Covid-19 crisis? Could everythinghappen again in the future?

Migration of liquidity transformation to mutual funds


Mutual Fund Liquidity ransfoTrmation and Reverse Flight to ...

Documents

Transcript of Mutual Fund Liquidity ransfoTrmation and Reverse Flight to ...