Liquidity risk management and credit supply in the financial crisis $
Marcia Millon Cornett a ,Jamie John McNutt b ,Philip E.Strahan c ,Hassan Tehranian d ,n
a
Bentley University,USA梦见屋里漏雨
b
Southern Illinois University,Carbondale,USA c
Boston College and NBER,USA d
Boston College,USA
a r t i c l e i n f o
Article history:
Received 7May 2010Received in revid form 5October 2010
Accepted 2November 2010Available online 8March 2011JEL classification:G01G11G21
Keywords:
Financial institutions Liquidity risk Financial crisis
a b s t r a c t
小蛋壳的故事Liquidity dried up during the financial crisis of 2007–2009.Banks that relied more heavily on core deposit and equity capital financing,which are stable sources of financing,continued to lend relative to other banks.Banks that held more illiquid asts on their balance sheets,in contrast,incread ast liquidity and reduced lending.Off-balance sheet liquidity risk materialized on the balance sheet
and constrained new credit origination as incread takedown demand displaced lending capacity.We conclude that efforts to manage the liquidity crisis by banks led to a decline in credit supply.
&2011Elvier B.V.All rights rerved.
1.Introduction
In this paper,we study how banks managed the liquidity shock that occurred during the financial crisis of 2007–2009by adjusting their holdings of cash and other liquid asts,as well as how the efforts to weather the storm affected credit availability.Becau the Federal Rerve ts the aggregate supply of liquidity in the banking system,focusing on only time ries variation in liquidity merely illustrates choices made by the Fed
(that is,the aggregate supply of liquidity).Our strategy instead is to put a spotlight on within-bank variation in holdings of cash and other liquid asts,which allows for an understanding of why some banks cho to build up liquidity faster than others during the crisis.This approach helps explain why the Fed’s efforts to stimulate the economy with traditional tools of monetary policy were ineffective.
Our empirical model starts with the premi that banks hold cash and other liquid asts as part of their overall strategy to manage liquidity risk.In modern banks,liquidity risk stems more from exposure to undrawn loan commitments,the withdrawal of funds from wholesale deposits,and the loss of other sources of short-term financing than from the loss of demand deposits (e.g.,Diamond and Dybvig,1983).With both explicit and implicit government backing,deposits are unlikely to leave the banking system during cris.For example,Gatev and Strahan (2006)find inflows of depos-its during periods of low market liquidity,while
b350Contents lists available at ScienceDirect
journal homepage:/locate/jfec
Journal of Financial Economics
0304-405X/$-e front matter &2011Elvier B.V.All rights rerved.doi:10.1016/j.jfineco.2011.03.001
$
We are grateful to EffiBenmelech (the referee),G.William Schwert (the editor),Robert DeYoung,Jaso
n T.Greene,David Rakowski,and minar participants at Boston College,Boston University,Dartmouth University,the Federal Rerve Board of Governors and the Federal Rerve Banks of New York and Kansas City,Harvard University,South-ern Illinois University —Carbondale,and the Deutsche Bundesbank,European Banking Center,and European Business School’s Joint Con-ference on Liquidity and Liquidity Risk for their helpful comments.n
Corresponding author.
E-mail hranian@bc.edu (H.Tehranian).Journal of Financial Economics 101(2011)297–312
Pennacchi (2009)does not find such flows during the pre-Federal Deposit Insurance Corporation (FDIC)period.Together this suggests that deposits insulate banks from liquidity risk due to the advent of government guarantees.Liquidity risk from loan commitments,for example,was evident in aggregate data when the commercial paper markets froze following the September 2008failure of Lehman Brothers.Issuers responded by taking down funds from commercial paper backup lines issued by banks,leading to a decline in commercial paper out-standing and an increa in bank lending (Fig.1).At the same time,banks lost wholesale funds but gained retail deposits (Fig.2).1We show that banks more expod to
this liquidity risk incread their holdings of liquid asts,which in turn reduced their capacity to make new loans.
On the ast side of balance sheets,banks holding asts with low market liquidity expanded their cash buffers during the crisis.Specifically,banks that held more loans,mortgage-backed curities (MBS),and ast-backed curities (ABS)tended to increa holdings of liquid asts and decrea investments in loans and new commitments to lend.Becau of concerns about the liquidity of loans and curitized asts,the banks rationally protected themlves by hoarding liquidity,to the detriment of their customers and markets.Turning to
70
80
90100110120130J u n e 2007
= 100
Fig.1.Business lending ris as commercial paper moves back on the balance sheet.This figure shows the growth of commercial paper and bank business loans outstanding from June 2007through November 2008.Data are obtained from the website of the Board of Governors of the Federal Rerve (www.chicagofed ).
-15%
-10%-5%0%5%10%15%Wholesale Deposits
Core Deposits
Fig.2.Growth in deposits.This figure shows the weekly percentage change in core and wholesale deposits at commercial banks from September 10,2008through December 31,2008.Core deposits include transactions deposits plus fully insured (o $100,000)time deposits.Wholesale deposits include time deposits over $100,000.Data are obtained from the Federal Rerve’s H8weekly data on bank asts and liabilities.
1
Gorton (2009)and Gorton and Metrick (2009)draw parallels between the increa in haircuts in the repo markets and banking
(footnote continued )
panics and bank runs.The effects were greatest at large nonbank financial institutions.
M.M.Cornett et al./Journal of Financial Economics 101(2011)297–312
298
the right-hand side of the balance sheet,banks with stable sources offinancing were less constrained by the crisis and,thus,were able to continue to lend.Banks using more core deposits(all transactions deposits plus other insured deposits)and more equity capital tofinance their asts saw significant increas in lending,relative to banks that relied more on wholesale sources of debtfinancing.The results hold when we control for aggregate time effects, bankfixed effects,measures of loan demand,and the effects offinancial structure during normal market con-ditions.Moreover,the results are consistent across both large and small bank samples,although the economic impact is generally bigger for the large bank sample.
中国古代名画
We also test how banks managed shocks to loan demand stemming from preexisting unud loan commit-ments(held off the balance sheet).Unud commitments expo banks to liquidity risk,which became manifest when takedown demand incread following the collap of Lehman Brothers.Wefind that banks with higher levels of unud commitments incread their holdings of liquid ,their precautionary demand for liquidity incread)and also cut back on new credit origination (measured by summing on-balance sheet loans with off-balance sheet loan commitments).Loan commitment drawdowns thus displaced new credit origination during the crisis.
Our paper extends in three ways the empirical analysis of Ivashina and Scharfstein(2010),who u D
ealscan data to show that new bank lending growth fell less at banks funded with deposits and more at banks expod to unud credit lines.First,we show that liquidity risk exposure is not only negatively correlated with loan growth in the crisis,but it is also positively correlated with the growth in liquid asts.The parallel results support the interpretation that efforts to build up balance sheet liquidity displaced funding to support new lending. Second,we have a much larger and richer data t[drawn from the quarterly Federal Financial Institutions Exam-ination Council(FFIEC)Reports of Income and Condition (Call Reports)],which allows us to explore more dimen-sions of liquidity risk exposure and to quantify implica-tions of our results for overall credit supply.For example, we show that the market liquidity of bank asts nega-tively affected their accumulation of liquid asts and positively affected their loan growth.Also,we show that it is core deposits,not total deposits,which provided stable funding to banks.Third,we work to rule out loan demand explanations for our results by exploiting geographical exposure from the FDIC Summary of Deposits and loan account data available from Call Reports.
Becau we look at the whole banking system,our regressions can help draw out the macroeconomic impli-cations of our results.We quantify how much credit would have contracted if banks had entered the fall of 2008less expod to liquidity risk.This analysis suggests that the press
ure on bank balance sheets from takedowns on preexisting loan commitments and funding problems from wholesale markets account for most of the decline in new credit production.New credit production—that is, the sum of both on-balance sheet loans and undrawn commitments—fell by about$500billion in the fourth quarter of2008(out of a total of slightly more than$14 trillion of total loans plus undrawn commitments to lend at the end of2008).Had liquidity exposure been in the lower quartile across the whole banking system,our estimates suggest that new credit would have fallen by just$87billion,or almost90%less than the unadjusted figure.
lan是什么In the remainder of the paper,we provide in Section2 a brief chronology of thefinancial crisis to justify our identification strategy bad on time variation of the TED spread as a measure of liquidity strains on the banking system.After laying out the drivers of bank liquidity risk to motivate our empirical model,we describe the data and results in Section3.We conclude in Section4.
2.The TED spread during thefinancial crisis of2007–2009
Thefinancial crisis of2007–2009is the biggest shock to the US and worldwidefinancial system since the1930s and offers a unique challenge to bothfinancial institu-tions’and regulators’understanding of liquidity produc-tion and liquidity risk management.2Fig.3illustrates the time ries of new loan origin
ations to large business from Loan Pricing Corporation’s Dealscan databa from 2000to the end of2008.During the2001–2002recession, both lines of credit and term loans declined as would be expected during a mild recession.But,this earlier decline pales relative to the steep drop in new lending beginning in the middle of2007.
The crisis began in the summer of2007when the ast-backed commercial paper market began to unravel in the face of uncertainty about the value and liquidity of some mortgage-backed curities(Acharya and Schnabl, 2010).The brewing crisis can be en in the TED spread [the difference between the three-month London Inter-bank Offered Rate(LIBOR)and the three-month Treasury rate],which spiked above200basis points.From then until the spring of2009,the TED spread(as well as other similar indicators)remained both elevated and volatile. The TED spread is an indicator of perceived credit risk in the general economy.This is becau T-bills are consid-ered risk-free,while LIBOR reflects the credit risk of lending to commercial banks.An increa in the TED spread indicates that lenders believe the risk of default on interbank ,counterparty risk)is increasing. We plot the time ries variation of the TED spread from the beginning of2006to the end of the cond quarter of 2009in Fig.4.[Fig.4also shows(in the shaded area)the period we designate as the crisis period in our robustness test below.]
充电宝排名Time variation in the TED spread tracks the verity of the crisis cloly.For instance,the TED spread spiked in March2008as Bear Stearns failed.Conditions improved following the Bear Stearns bailout,and the TED spread subsided.In the summer of2008,however,concerns about mortgage foreclosures ro,further downgrades of mortgage-backed curities by the credit rating agencies 2See Brunnermeier(2009)for a detailed discussion of the events.
M.M.Cornett et al./Journal of Financial Economics101(2011)297–312299
occurred,and loss to holders of the curities mounted.Loss on mortgages and mortgage-backed curities eventually led to the failure of veral financial institutions,notably,Fannie Mae (Federal National Mort-gage Association)and Freddie Mac (Federal Home Loan Mortgage Corporation)and then American International Group,Inc.(AIG)and Lehman Brothers.The depth of the crisis dramatically expanded when financial markets were shocked by the collap of the institutions,along with the distresd sale of Merrill Lynch to Bank of America.The panic soon spread,leading to the expansion of insurance on deposits and interbank funds,first in Europe and then very quickly in the United States.The crisis truly abated only in the spring of 2009when the stress tests of the large US banks brought private capital back into the system.
3.Empirical strategy and results
In this ction,we first discuss the determinants of bank liquidity risk and then describe our empirical model,data,and results.
3.1.Liquidity risk management
Liquidity production is central to all theories of finan-cial intermediation.First,asymmetric information proces-sing allows banks to create liquidity through their ast transformation function (e Diamond and Dybvig,1983).Second,banks provide liquidity to borrowers in the form of credit lines and to depositors by making funds available on demand.The functions leave banks vulner-able to systemic increas in demand for liquidity from洋兰花语
Fig.3.Business loan originations collap.This figure shows the dollar value of new term loans and credit lines issued to large business from 2000(before the financial crisis)through 2008(at the height of the financial crisis).Data ud to construct the figure are obtained from the Loan Pricing Corporation’s Dealscan databa.
0.0%
0.5%1.0%1.5%2.0%2.5%3.0%3.5%4.0%4.5%5.0%Fig.4.The TED spread.This figure shows movements in the TED spread from 2006through the cond quarter of 2009.The TED spread is calculated as the difference between the three-month London Interbank Offered Rate (LIBOR)rate [obtained from the website of the Bulgarian National Bank (www.bnb.bg/#)]and the three-month Treasury rate [from the Federal Rerve Economic Data (FRED)website of the Federal Rerve Bank of St.Louis (rearch.stlouisfed/fred2/)].The shaded area includes the period we designate the crisis period in our empirical analysis.
M.M.Cornett et al./Journal of Financial Economics 101(2011)297–312
300
borrowers and,at the extreme,can result in runs on banks by depositors.In the traditional framework of banking, runs can be prevented,or at least mitigated,by insuring deposits and by requiring banks to issue equity and to hold cash ,Diamond and Dybvig,1983; Gorton and Pennacchi,1990).Systemic increas in demand for liquidity from borrowers,in contrast,depend on external market conditions and thus are harder for individual banks to manage internally.For example,when the supply of overall market liquidity falls,borrowers turn to banks en mas to draw funds from existing credit lines (Gatev and Strahan,2006).
Diamond and Rajan(2001b)note that while banks provide liquidity to borrowers,the loans themlves are relatively illiquid asts for banks.Subquently,when banks require liquidity,they could ll the ,ll and curitize mortgages to create mortgage-backed curities)or u the loans as ,mortgages rve as collateral for mortgage-backed bonds issued by the banks)(e Bhattacharya and Thakor(1993); Diamond and Rajan(2001b)).Such sales,however, become more difficult when market liquidity becomes scarce.Thus,Diamond and Rajan(2001b)also note that banks can ration credit if future liquidity needs are likely to be high.Diamond and Rajan(2001a)suggest banks can be fragile becau they must provide liquidity to deposi-tors on demand and becau they hold illiquid loans. Further,demands by depositors can occur at undesirable ,when loan
payments are uncertain and when there are negative aggregate liquidity shocks.In addition, Kashyap,Rajan,and Stein(2002)note similarities between some off-balance ,contingent)asts and on-balance sheet asts.In particular,an off-balance sheet loan commitment becomes an on-balance sheet loan when the borrower choos to draw on the commit-ment.Berger and Bouwman(2009)find that roughly half of the liquidity creation at commercial banks occurs through the off-balance sheet commitments.Thus, banks stand ready to supply liquidity to both borrowers and insured retail depositors and can enjoy synergies when depositors fund loan commitments.Recent evi-dence lends support to this notion.Gatev,Schuermann, and Strahan(2009)find deposits effectively hedge liquid-ity risk inherent in unud loan commitments and the effect is more pronounced during periods of tight liquidity.
The role of bank equity capital also plays a part in the liquidity provision function of commercial banks.Diamond and Rajan(2000)suggest equity capital can act as a buffer to protect depositors in times of distress.However,holding excessive equity capital can reduce liquidity creation and theflow of credit.Gorton and Winton(2000)conclude that regulators should be especially aware of the effects during recessionary ,periods when reg-ulators could want to increa capital standards to reduce the threat of bank failures.Recent evidence suggests bank size can affect which 感恩父母的话语简短
effect dominates.Berger and Bouwman(2010)find that higher capital levels crowd out depositors and decrea liquidity creation at smaller banks,but higher capital levels absorb risk and increa liquidity creation at larger banks.
Banks facilitate their operations with more than retail deposits and equity capital,most notably with uninsured wholesale deposits and subordinated notes and deben-tures.Rearchers and regulators have long been inter-ested in the alternate funding mechanisms and their role in imparting market discipline on bank behavior.3For example,Hannan and Hanweck(1988)find uninsured depositors require higher interest rates at riskier banks, and Maechler and McDill(2006)suggest uninsured depositors might not supply liquidity to weak banks at any price.Avery,Belton,and Goldberg(1988)find little evidence that holders of bank-issued subordinated notes and debentures effectively constrain bank risk.However, restrictive covenants have been found to be more com-mon in debt contracts when banks are riskier(e Goyal, 2005;Ashcraft,2008).
Size also matters.That is,the market’s perception of the risk of a bank can depend on the size of the bank.The Comptroller of the Currency’s statement before Congress on September19,1984that somefinancial institutions are too-big-to-fail(TBTF)was a positive wealth event for banks deemed TBTF(e O’Hara and Shaw,1990).Further evidence is provided by Black,Collins,Robinson,and Sch
weitzer(1997),who obrve aflight to quality as evidenced by changes in institutional ownership of TBTF bank equity shares.
3.2.Empirical specification
The discussion above suggests four key drivers of liquidity risk management for banks:(1)the composition of the ast ,the market liquidity of asts),
(2)core deposits as a fraction of totalfinancial structure,
(3)equity capital as a fraction offinancial structure,and
(4)funding liquidity exposure stemming from loan ,new loan originations via drawdowns). Ast size also likely relates to liquidity management,but it proxies for many other sources of heterogeneity.Hence, we include this variable in all of our regressions but refrain from interpreting its effect.
Our identification strategy is bad on the premi that tight liquidity conditions during thefinancial crisis,mea-sured by the TED spread,surprid banks and thus changed their management of liquidity risk exposure. That is,banks with high liquidity risk exposure would be expected to build up c
ash and other liquid asts and also to reduce new lending(particularly new commit-ments to lend)more than banks with low liquidity risk exposure when the TED spread spikes.We test this idea by interacting the TED spread with our four measures of liquidity exposure.
We build a quarterly panel data t from the beginning of2006through the cond quarter of2009that includes all commercial banks as described below.This sample has obrvations before and during thefinancial crisis,at least judging by movements in TED spreads.With the panel 3See Flannery(1998)for an overview of the role of market
discipline as it relates to regulatory supervision and Flannery(2001) for an overview of the notion of market discipline.
M.M.Cornett et al./Journal of Financial Economics101(2011)297–312301
