Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Sizing Up Repo
Arvind Krishnamurthy1
Stefan Nagel2
1 Northwestern
2 Stanford
University
University
March 2012
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Dmitry Orlov2
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Funding Flows in Shadow Banking System
!"&%
=&+12)
!&%7/+/(2#
-(./(#
"01.2#
!"&%
345#
!"&%
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
$6,+7(%+,#
8(9&#
$%&'(%)
*(+,(%#
:$;4#
<&1.067#
'()*+,(-.%
/0+1%
!!"#
!"#$%
Sizing Up Repo
>*(9&2672?#
@AB#C:5D#
;+2F#
3(<0%6E(2# ;&,,+7(%+,##
,(1.(%#
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Tri-Party Repo
2345#
2345#
$%&'()%*+#
,-.)%&/0#
10./*#
;%6<.%=
>.)-.%#
,6--)*.%)-#
76%*8
29::5#
!!"#
,6--)*.%)-#
76%*8
29::5#
Haircut: 5% in this example
Repo rate: Interest paid by borrower on loan amount ($95m)
Maturity: Varies from overnight to weeks or months.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
How did Repo contribute to the Crisis?
”Run on repo” by repo ”depositors” lead to collapse of
shadow banking (Gorton and Metrick)
Repo as the “deposit” in a shadow banking sector that holds
non-Agency MBS/ABS
Run of repo depositors created funding squeeze for shadow
banking system (GM: haircuts from 0% to 20% w/ $10 trillion
repo ≈ $2 trillion funding shortfall)
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
How did Repo contribute to the Crisis?
”Run on repo” by repo ”depositors” lead to collapse of
shadow banking (Gorton and Metrick)
Repo as the “deposit” in a shadow banking sector that holds
non-Agency MBS/ABS
Run of repo depositors created funding squeeze for shadow
banking system (GM: haircuts from 0% to 20% w/ $10 trillion
repo ≈ $2 trillion funding shortfall)
Data on repo
GM have haircut data, but for interdealer/dealer-hedge fund
repo, not for net cash investors.
Estimates of size (i.e. $10 tn) subject to serious double
counting problems.
We collect a new data set on repo, based on MMF and sec
lenders.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Findings and Lessons – Preview
Repo much less important than ABCP in terms of funding
non-Agency MBS/ABS pre-crisis and contraction during crisis
Replace word repo with “ABCP plus repo” in Gorton-Metrick.
Run on ABCP is a (solvency) problem for regulated banking
sector (w/ regular access to Fed), not for broker/dealer sector
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Findings and Lessons – Preview
Repo much less important than ABCP in terms of funding
non-Agency MBS/ABS pre-crisis and contraction during crisis
Replace word repo with “ABCP plus repo” in Gorton-Metrick.
Run on ABCP is a (solvency) problem for regulated banking
sector (w/ regular access to Fed), not for broker/dealer sector
Repo run limited to repo with non-govt. collateral
Funding problems more significant for subset of dealer banks
that are heavy users of non-govt. collateral
Haircuts of MMF-to-dealer repo increase much less than those
of the dealer-to-dealer repo in Gorton-Metrick
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Findings and Lessons – Preview
Repo much less important than ABCP in terms of funding
non-Agency MBS/ABS pre-crisis and contraction during crisis
Replace word repo with “ABCP plus repo” in Gorton-Metrick.
Run on ABCP is a (solvency) problem for regulated banking
sector (w/ regular access to Fed), not for broker/dealer sector
Repo run limited to repo with non-govt. collateral
Funding problems more significant for subset of dealer banks
that are heavy users of non-govt. collateral
Haircuts of MMF-to-dealer repo increase much less than those
of the dealer-to-dealer repo in Gorton-Metrick
Less a run of “depositors” than an “interbank” credit crunch
(due to capital concerns, liquidity hoarding?)
Here interbank means broker/dealers in making repo loans to
e.g., hedge funds, rather than commercial banks in making
loans to corporates.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Rehypothecation Problem in Existing Repo Data
*+,1-#
!23#
4%56%#
"785#
*+,,-#
*+,,-#
*+,,-#
$%&'%(#)#
$%&'%(#2#
./''&0%(&'#
./''&0%(&'#
!!"#
./''&0%(&'#
In this example, amount of funding for MBS purchase
provided to shadow banking system via repo is $100m, but
total volume of repo oustanding is $300m.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Rehypothecation Problem in Existing Repo Data
*+,1-#
!23#
4%56%#
"785#
*+,,-#
*+,,-#
*+,,-#
$%&'%(#)#
$%&'%(#2#
./''&0%(&'#
./''&0%(&'#
!!"#
./''&0%(&'#
In this example, amount of funding for MBS purchase
provided to shadow banking system via repo is $100m, but
total volume of repo oustanding is $300m.
Fed Primary Dealer Repo Data ($4.1 trillion Dec. 2007)
includes inter-dealer repo. Fed Tri-Party Repo Data ($2.5
trillion Dec. 2007) includes inter-dealer GCF repo.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Rehypothecation Problem in Existing Repo Data
*+,1-#
!23#
4%56%#
"785#
*+,,-#
*+,,-#
*+,,-#
$%&'%(#)#
$%&'%(#2#
./''&0%(&'#
./''&0%(&'#
!!"#
./''&0%(&'#
In this example, amount of funding for MBS purchase
provided to shadow banking system via repo is $100m, but
total volume of repo oustanding is $300m.
Fed Primary Dealer Repo Data ($4.1 trillion Dec. 2007)
includes inter-dealer repo. Fed Tri-Party Repo Data ($2.5
trillion Dec. 2007) includes inter-dealer GCF repo.
Analogy: Exclusion of inter-bank deposits in calculation of M2
money stock
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Data
Data (micro) on MMF repos from quarterly SEC filings of
MMF
We collect repos for 20 biggest MMF families from end of
2006 to mid-2010 (top 20 covers 80%+ of total MMF assets
under management)
≈ 16,000 repo agreements
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Data
Data (micro) on MMF repos from quarterly SEC filings of
MMF
We collect repos for 20 biggest MMF families from end of
2006 to mid-2010 (top 20 covers 80%+ of total MMF assets
under management)
≈ 16,000 repo agreements
Quarterly survey of major securities lenders conducted by Risk
Management Association (RMA), including the big securities
lenders (AIG, BNY Mellon, State Street, ...)
Data (aggregate) on cash collateral reinvestment, incl. repos,
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Data
Data (micro) on MMF repos from quarterly SEC filings of
MMF
We collect repos for 20 biggest MMF families from end of
2006 to mid-2010 (top 20 covers 80%+ of total MMF assets
under management)
≈ 16,000 repo agreements
Quarterly survey of major securities lenders conducted by Risk
Management Association (RMA), including the big securities
lenders (AIG, BNY Mellon, State Street, ...)
Data (aggregate) on cash collateral reinvestment, incl. repos,
Transactions data on emergency lending programs of the
Federal Reserve
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
MMF Data Example: Reserve Fund – Primary Fund
February 29, 2008 Repurchase Agreements
Notional
Counterparty
Rate
Init.
Rep.
Collateral
Coll. mkt.val.
1,000,000,000
450,000,000
500,000,000
140,000,000
1,000,000,000
...
Bear Stearns
Bear Stearns
Citigroup
Merrill Lynch
Morgan Stanley
3.28%
3.33%
3.23%
3.43%
3.29%
2/29/08,
2/29/08
2/29/08
2/29/08
2/29/08
3/3/08
3/3/08
3/3/08
3/3/08
3/3/08
ABS, CMO, TRR, TR3
ABS, CMO
MNI, TRR
WLR
WLR
1,048,922,871
472,500,201
556,131,379
146,599,193
1,020,794,540
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
Summary of MMF and Securities Lender Repo ($bn)
Quarter
2006Q4
2007Q1
2007Q2
2007Q3
2007Q4
2008Q1
2008Q2
2008Q3
2008Q4
2009Q1
2009Q2
2009Q3
2009Q4
2010Q1
1
2
3
Money Market Funds
Collected
Total
Total
Repo
Repo2
Assets2
2431
395
2312
324
387
2372
331
426
2466
412
528
2780
483
606
3033
501
592
3383
466
518
3318
433
592
3355
479
542
3757
546
562
3739
507
488
3585
495
495
3363
472
480
3259
427
440
2931
Securities Lenders
Cash
Repo
Collateral
431
1594
527
1834
504
1902
522
1754
478
1712
467
1537
509
1790
490
1519
228
954
212
779
257
882
244
865
229
850
263
837
Incomplete coverage of funds in MMF sample in 2006Q4.
Source: Flow of Funds Accounts.
Source: Federal Reserve Bank of New York
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Primary
Dealer
Repo3
3442
3619
3889
3886
4106
4278
4222
3989
3208
2743
2582
2499
2469
2477
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
How much of Total Repo Funding Do We Capture?
2007Q4, about $1.1 trillion repo lending in total from MMF
and securities lenders
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
How much of Total Repo Funding Do We Capture?
2007Q4, about $1.1 trillion repo lending in total from MMF
and securities lenders
Tri-party repo of $2.5tn, but includes interdealer repo
(∼$500bn?).
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
How much of Total Repo Funding Do We Capture?
2007Q4, about $1.1 trillion repo lending in total from MMF
and securities lenders
Tri-party repo of $2.5tn, but includes interdealer repo
(∼$500bn?).
Other repo lenders according to Flow of Funds Accounts,
2007Q4 (December 2010 release):
State and local governments $163.3bn
Government sponsored enterprises $142.7bn
Rest of the world $338.4bn
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
How much of Total Repo Funding Do We Capture?
2007Q4, about $1.1 trillion repo lending in total from MMF
and securities lenders
Tri-party repo of $2.5tn, but includes interdealer repo
(∼$500bn?).
Other repo lenders according to Flow of Funds Accounts,
2007Q4 (December 2010 release):
State and local governments $163.3bn
Government sponsored enterprises $142.7bn
Rest of the world $338.4bn
Who else: Corporations (appear to go through MMFs).
Foreign Central Banks (DVP, tri-party, $100 to $200bn).
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
Share
.5 .6
.7
.8
.9
1
MMF: Share of Collateral by Type (by value)
2007q1
2008q1
2009q1
2010q1
Quarter
U.S. Treasury
Corporate
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Agency
Other
Sizing Up Repo
Priv. ABS
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
0
.2
.4
Share
.6
.8
1
Securities Lenders: Share of Collateral by Type (by value)
2007q1
2008q1
2009q1
2010q1
Quarter
U.S. Treasury
Corporate
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Agency
Other
Sizing Up Repo
Priv. ABS
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
Repo versus ABCP
Repo of ABS
Bank holds ABS on balance sheet.
Issues short-term debt, overcollateralized, against ABS.
Typically overnight, so lenders can redeem the debt at par.
ABCP
Bank takes loans/securities, places them in SPV.
Issues short-term (ofter overnight) debt against loans.
Buyers of debt can sell the debt back to banks at par.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
Short-term Funding Pre-Crisis 2007Q2
Non-agency MBS/ABS
Amount
%
Total outstanding1
Short-term funding
ABCP2
Direct holdings3
MMF
Securities lenders
Repo4
MMF
Securities lenders
Total short-term
5275
100%
1173
22%
243
502
31
120
2069
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Corporate Bonds
Amount
%
5591
100%
5%
10%
179
369
3%
7%
1%
2%
40%
42
166
755
1%
3%
14%
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
Contraction in Short-term Funding
Non-Agency MBS/ABS
ABCP1
Direct holdings
MMF3
Securities lenders2
Repo
MMF
Securities lenders4
Total
Corporate bonds
Direct holdings
MMF3
Securities lenders
Repo
MMF
Securities lenders4
Total
2007Q2
2009Q1
Contraction
1173.2
511.0
-662.2
243.3
501.6
59.4
116.0
-183.9
-385.6
30.5
120.1
0.3
1.6
-30.2
-118.5
-1380.4
178.9
368.7
158.4
309.1
-20.5
-59.6
42.1
165.6
9.7
49.3
-32.4
-116.3
-228.8
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
0
300
50
100
150
Repo w/ non−agency MBS/ABS
ABCP outstanding
600
900
1200
200
1500
Comparison of repo and ABCP contraction
2006q3
2007q3
ABCP outstanding
2008q3
Quarter
2009q3
2010q3
Repo w/ non−agency MBS/ABS
ABCP outstanding net of amount funded through Fed CPFF
program
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Collateral in Tri-Party Repo
Short-term funding of private debt instruments
Demand or Supply?
Demand or Supply?
Repo demand contraction: MMF and securities lenders refuse
to lend against non-agency MBS/ABS collateral?
Repo supply contraction: Hedge funds and dealer banks scale
back investments in MBS/ABS and have lower MBS/ABS
funding needs?
Indication that largely demand driven
Quantity going to zero suggests demand effect
Price terms rise
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Maturity
Haircuts
Repo Rates
0
50
Maturity (business days)
100
150
200
250
Maturity Compression (vw.)
2006q3
2007q3
99th
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
2008q3
Quarter
98th
2009q3
95th
Sizing Up Repo
2010q3
90th
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Maturity
Haircuts
Repo Rates
0
Maturity (business days)
50
100
150
Maturity Compression (ew.)
2006q3
2007q3
90th
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
2008q3
Quarter
80th
2009q3
70th
Sizing Up Repo
2010q3
60th
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Maturity
Haircuts
Repo Rates
2
3
Percent
4
5
6
7
Haircuts by Collateral Type (vw.)
2006m7
2007m7
2008m7
Month
U.S. Treasury
Priv. ABS
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
2009m7
Agency
Corporate
Sizing Up Repo
2010m7
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Maturity
Haircuts
Repo Rates
0
2
Percent
4
6
Average Overnight Treasury Repo Rate (vw.) and Fed
Funds Rate
2006m7
2007m7
Fed Funds Rate
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
2008m7
Month
2009m7
Treasury Repo Rate (vw.)
Sizing Up Repo
2010m7
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Maturity
Haircuts
Repo Rates
0
1
2
Percent
3
4
5
Average Overnight Repo Rate (vw.) in Excess of Fed
Funds Rate
2006m7
2007m7
2008m7
Month
Agency
Corporate
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
2009m7
Priv. ABS
Sizing Up Repo
2010m7
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Maturity
Haircuts
Repo Rates
Limited Run on Repo
Increase in price terms on non-agency MBS/ABS (repo rate,
haircut, decrease in maturity) suggest “run on repo”.
Haircuts for Treasury and agency always stayed between 2-3%.
Tri-party repo haircuts increased much less during crisis than
the bi-lateral repo haircuts reported in Gorton and Metrick
(2011b)
Gorton and Metrick report haircuts > 50% for several
categories of securitized products
Why the difference?
Their data: dealer to dealer.
Credit crunch from dealers, given capital concerns? Defensive
actions of dealers, liquidity hoarding?
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Cross-Sectional Patterns by Repo Counterparty
Evidence for “run” on repo with private collateral, especially
non-agency MBS/ABS, but channel for collapse of shadow
banking is unclear.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Cross-Sectional Patterns by Repo Counterparty
Evidence for “run” on repo with private collateral, especially
non-agency MBS/ABS, but channel for collapse of shadow
banking is unclear.
We examine effects on dealer banks:
Dealer banks with higher exposure to private debt instruments?
“Run” on specific dealer banks irrespective of type of collateral
offered?
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Cross-Sectional Patterns by Repo Counterparty
Evidence for “run” on repo with private collateral, especially
non-agency MBS/ABS, but channel for collapse of shadow
banking is unclear.
We examine effects on dealer banks:
Dealer banks with higher exposure to private debt instruments?
“Run” on specific dealer banks irrespective of type of collateral
offered?
Analysis based on MMF repo data, focused on periods
Pre-Bear Stearns (BSC): Dec. ’07 - Feb ’08
Post-Lehman (LEH): Sep. ’08 - Nov. ’08
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Change Pre−BSC to Post−LEH
−.5 0 .5 1 1.5
Contraction/Expansion in Total MMF Repo
Total Repo
mfg
bcs
jpm bnpqy
hbc
bac
scgly
drb
ms
ing db
rbs
cs
ubs
mer c
gs
wb
0
.1 .2 .3 .4 .5 .6 .7 .8
Private Collateral Share Pre−BSC
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
.9
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Contraction Pre−BSC to Post−LEH
−.5
0
.5
1
1.5
Contraction/Expansion in MMF Repo w/ Private Collateral
Repo w/ Private Collateral
bcs
bnpqy
mfg
scgly
rbs ing
hbc
jpmdb
bac
cs
ubs
drb
ms
mer cgs
wb
0
.1 .2 .3 .4 .5 .6 .7 .8
Private Collateral Share Pre−BSC
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
.9
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
1500
Private Collateral Share and Post-Lehman CDS Rates
wb
Maximum 5−yr Senior CDS Rate
500
1000
ms
bsc
leh
hbc
gs
mer
rbs
bcs
bac cs
ing jpm db
drb
bnpqy
0
scgly
c
ubs
0
.2
.4
.6
Private Collateral Share Pre−BSC
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
.8
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Trading Assets (preliminary)
Change from pre-BSC to post-LEH:
Dealer banks with high private collateral shares:
GS = -32%
MS = -37%
MerLynch = -20%
Citi = -23%
Dealer banks with low private collateral share:
JPM = +10%
BAC = -10%
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Haircuts by Counterparty
1
2
Haircut
3
4
5
Haircuts
0
200
400
600
5−yr Senior CDS Rate 9/30/2008
Treasury
Agency
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Treasury/Agency
Private
Sizing Up Repo
800
1000
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Repo Rates by Counterparty
0
2
Repo Rate
4
6
8
Repo Rates
0
200
400
600
5−yr Senior CDS Rate 9/30/2008
Treasury
Agency
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Treasury/Agency
Private
Sizing Up Repo
800
1000
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Changes in Repo Quantities by Counterparty
Repo Terms by Counterparty on September 30, 2008
Cross-Sectional Patterns by Repo Counterparty
Dealer banks with highest private collateral funding needs
appear to run into trouble.
We do not see “runs” on high-risk dealer banks: financing
with high-quality collateral still available at normal terms.
But our data is not high frequency ...
Some money market investors stopped rolling over Lehman
Brothers repos, irrespective of collateral, only in the last few
days before bankruptcy (Copeland, Martin, and Walker (2010))
Concern existed that tri-party agents (JPMC and BNYM)
could cut off access to tri-party repo for high-risk dealer banks
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
Federal Reserve Programs
We focus on four principal programs:
1 PDCF (Primary Dealer Credit Facility), March 2008: Loan
facility that provided funding to primary dealers in exchange
for any tri-party-eligible collateral. Loans were overnight, and
made at the primary credit discount rate.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
Federal Reserve Programs
We focus on four principal programs:
1 PDCF (Primary Dealer Credit Facility), March 2008: Loan
facility that provided funding to primary dealers in exchange
for any tri-party-eligible collateral. Loans were overnight, and
made at the primary credit discount rate.
2 TSLF (Term Securities Lending Facility), March 2008: Facility
to loan Treasuries from the Fed’s portfolio in exchange for any
investment-grade collateral. Loans were 28-day, and rates
were set in an auction.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
Federal Reserve Programs
We focus on four principal programs:
1 PDCF (Primary Dealer Credit Facility), March 2008: Loan
facility that provided funding to primary dealers in exchange
for any tri-party-eligible collateral. Loans were overnight, and
made at the primary credit discount rate.
2 TSLF (Term Securities Lending Facility), March 2008: Facility
to loan Treasuries from the Fed’s portfolio in exchange for any
investment-grade collateral. Loans were 28-day, and rates
were set in an auction.
3 Maiden Lane, 2 facilities, various dates: Fed made loans to
SPVs that held non-agency ABS. Facilities were set up in
conjunction with interventions in Bear Stearns and AIG.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
Federal Reserve Programs
We focus on four principal programs:
1 PDCF (Primary Dealer Credit Facility), March 2008: Loan
facility that provided funding to primary dealers in exchange
for any tri-party-eligible collateral. Loans were overnight, and
made at the primary credit discount rate.
2 TSLF (Term Securities Lending Facility), March 2008: Facility
to loan Treasuries from the Fed’s portfolio in exchange for any
investment-grade collateral. Loans were 28-day, and rates
were set in an auction.
3 Maiden Lane, 2 facilities, various dates: Fed made loans to
SPVs that held non-agency ABS. Facilities were set up in
conjunction with interventions in Bear Stearns and AIG.
4 CPFF (Commercial Paper Funding Facility), October 2008:
Fed made loans to an SPV to purchase 3-month ABCP.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
0
50
Amount ($bn.)
100
150
200
Repo and Federal Reserve Funding of Non-Agency
MBS/ABS
2006q3
2007q3
2008q3
Quarter
Repo
TSLF
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
2009q3
Maiden Lane
PDCF
Sizing Up Repo
2010q3
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
50
Amount ($bn.)
100 150 200
250
300
Repo and Federal Reserve Funding of Corporate Debt
Securities
2006q3
2007q3
2008q3
Quarter
Repo
PDCF
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
2009q3
TSLF
Sizing Up Repo
2010q3
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
Amount ($bn.)
400
600
800
1000
1200
ABCP Oustanding (ex CPFF) and CPFF Funding
2006q3
2007q3
2008q3
Date
ABCP ex CPFF
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
2009q3
ABCP in CPFF
Sizing Up Repo
2010q3
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
Federal Reserve Programs: Terms
We focus on Fed Program Utilization around Sep. 30, 2008
TSLF looks attractive relative to market rates on Sep. 30,
2008
Schedule 2 auction on 10/1/08 yielded a (uniform) loan fee of
1.51%, compared private collateral repo rate spread to
Treasury repo on 9/30/08 of 7%.
Schedule 1 auction on 10/1/08 yielded a (uniform) loan fee of
0.42%, which is lower than many observed agency repo spreads
on 9/30/2008.
PDCF funding rates attractive for private collateral (2.25% on
9/30/2008).
Both must have carried stigma similar to discount window
borrowing for commercial banks.
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Quantity of Fed Funding
Fed Program Participation, by Dealer Bank
Fed Program Utilization by Primary Dealers in Sep ’08
maxout1
(1)
TSLF
maxout2
(2)
Total
(3)
PDCF
Total
(4)
Agency Share pre-BSC
1.43
(2.21)
0.77
(1.58)
46.16
(2.31)
9.34
(0.62)
Private Share pre-BSC
-0.36
(-0.48)
1.10
(1.92)
39.33
(1.66)
57.85
(3.26)
15
0.17
15
0.28
15
0.36
15
0.42
Observations
Adjusted R 2
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Discussion: Repo and contraction of shadow banking
Gorton-Metrick broadbrush picture: Run on repo by “depositors”
created big aggregate funding shortfall
Evaluation based on our evidence: Repo w/ non-Agency
MBS/ABS collateral too small to fit that picture
Run by repo “depositors” confined to risky/illiquid collateral,
relatively insignificant in terms of aggregate funding needs of
shadow banking system.
Contraction in ABCP and direct ABS/MBS investments of
short-term investors an order of magnitude bigger
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo
Quantity of Repo Funding
Repo Terms During the Financial Crisis
Cross-Sectional Patterns by Repo Counterparty
Federal Reserve Programs
Discussion: Repo and contraction of shadow banking
Emerging picture: Funding squeeze for some dealer banks
Run may have played a significant role for dealer banks with
high risky/illiquid collateral funding needs. Many of these
dealer banks almost failed (caveat: capital depletion, loss of
brokerage business, etc. could be correlated with collateral
composition).
Gorton-Metrick dealer-to-dealer repo haircut data indicates
dealer banks then raised cost of repo credit to each other and
to hedge funds in interbank repo (“credit crunch”)
Arvind Krishnamurthy, Stefan Nagel, Dmitry Orlov
Sizing Up Repo