Central Bank Policies and Systemic Risks

7/28/2019 Central Bank Policies and Systemic Risks

1/34

Summy

Major central banks have taken unprecedented policy actions ollowing the nancial crisis. Inaddition to keeping interest rates low or a prolonged period, they have taken a host o uncon-ventional measures, including long-term liquidity provision to banks in support o lending, aswell as asset purchases to lower long-term interest rates and to stabilize specic markets, suchas those or mortgages.

Although the objectives dier somewhat across central banks, these policies have generally aimed to supportthe macroeconomy (by avoiding deation and depression) and address short-term nancial stability risks.Using econometric and other evidence, this chapter nds that the interest rate and unconventional policies

conducted by the central banks o our major regions (the euro area, Japan, the United Kingdom, and theUnited States) appear indeed to have lessened vulnerabilities in the domestic banking sector and contributedto nancial stability in the short term. Te prolonged period o low interest rates and central bank asset pur-chases has improved some indicators o bank soundness. Central bank intervention mitigated dysunction intargeted markets, and large-scale purchases o government bonds have in general not harmed market liquidity.Policymakers should be alert to the possibility, however, that nancial stability risks may be shiting to otherparts o the nancial system, such as shadow banks, pension unds, and insurance companies. Te centralbank policy actions also carry the risk that their eects will spill over to other economies.

Despite their positive short-term eects or banks, these central bank policies are associated with nancialrisks that are likely to increase the longer the policies are maintained. Te current environment shows signs odelaying balance sheet repair in banks and could raise credit risk over the medium term. Markets may be alertto these medium-term risks, as central bank policy announcements have been associated with declines in some

bank stocks and increases in yield spreads between bank bonds and government bonds. Central banks alsoace challenges in eventually exiting markets in which they have intervened heavily, including the interbankmarket; policy missteps during an exit could aect participants expectations and market unctioning, possiblyleading to sharp price changes.

Even though monetary policies should remain very accommodative until the recovery is well established,policymakers need to exercise vigilant supervision to assess the existence o potential and emerging nancialstability threats, and they should use targeted micro- and macroprudential policies where possible to mitigatesuch threats to allow greater leeway or monetary policy to support the macroeconomy. Macroprudentialpolicieswhich may include robust capital standards; improved liquidity requirements; and well-designed,dynamic, orward-looking provisioningshould be implemented in a measured manner, as needed. Te crisishas shown that corrective policies enacted ater the risks materialize may be too late to contain damage tonancial stability. As the experience with some macroprudential policies is relatively limited, their eectiveness

should be careully monitored. In the meantime, the unconventional monetary policy actions should continue,as they have, to keep nancial stability goals in mind.

International Monetary Fund | April 2013 1

3chapter

DO ceNtraL BaNK pOLIcIeS SINce the crISIS carrY rISKS tOFINaNcIaL StaBILItY?


2/34

GLOBAL FINANCIAL STABILITY REPORT

2 International Monetary Fund | April 2013

he central banks o the largest advancedeconomies have taken unprecedentedmeasures to combat the deepest andmost prolonged period o recession and

nancial instability since the 1930s. Tese measuresinclude an extended period o very low interest ratesas well as so-called unconventional policiesprovid-ing long-term liquidity to banks to support the owo credit, lowering long-term rates through bondpurchases, and stabilizing specic markets such asmortgage lending.1 Central banks have also issuedorward guidance, in which they announce anintention to maintain an accommodative stance oran extended period. We will reer to the combina-tion o exceptionally low policy interest rates andunconventional policy measures as MP-plus to

indicate that these policies go beyond conventionalmonetary policy in terms o tools and objectives.

Te objectives o MP-plus are to benet not onlythe macroeconomy but also nancial stability. Byproviding liquidity to banks and buying specicassets, MP-plus directly mitigates short-term insta-bility in nancial markets and vulnerabilities in thedomestic banking sector. In addition, MP-plus alsoindirectly limits stress in the nancial sector to theextent that it succeeds in preventing a sharper eco-nomic downturn. By encouraging economic activitythrough its easing o credit conditions, MP-plus canhelp strengthen private and public balance sheets andthus make a more durable contribution to nancialstability. Such benets may result, or instance, irms take advantage o lower longer-term rates byextending the maturity prole o their debt.

However, MP-plus may have undesirable sideeects, including some that may put nancial stabil-ity at risk. Ample bank liquidity may raise credit riskat banks by compromising underwriting and loan

Note: Tis chapter was written by S. Erik Oppers (team

leader), Ken Chikada, Frederic Lambert, ommaso Mancini-Grioli, Kenichi Ueda, and Nico Valckx. Research support wasprovided by Oksana Khadarina.

1Examples o the unconventional policies are quantitativeeasing by the Federal Reserve, the Funding or Lending Schemeby the Bank o England, and the announcement o the OutrightMonetary ransactions o the European Central Bank. Te Banko Japan implemented a program o quantitative easing in theearly 2000s andalong with other unconventional policy mea-suresagain in the atermath o the global nancial crisis.

quality standards, and it may encourage a delay innecessary balance sheet repair and bank restructur-ing. Likewise, low interest rates encourage othernancial institutions, including pension unds, insur-

ance companies, and money market mutual unds,to increase risk by searching or yield. A search oryield can help push the market value o some assetsbeyond their undamental value (bubbles) or drivean excessive increase in balance sheet leverage. Insome cases, risks may stem not rom the unconven-tional policies themselves but rom the difculties inexiting rom them. Where central banks intervenedin markets to mitigate instability, their presencemay aect market unctioning or mask continuingvulnerabilities, complicating exit and raising thepotential or policy missteps.

Tis chapter aims to bring empirical evidence tobear on some o the nancial stability eects o MP-plus. It denes and quanties the MP-plus policies oour major central banksthe Federal Reserve, theEuropean Central Bank (ECB), the Bank o Japan(BOJ), and the Bank o England (BOE)and thenidenties possible risks to domestic nancial stabilityand to the nancial health o banks. Banks are theocal point o the chapter because they are naturallyleveraged and, as a whole, they are the most systemi-cally important nancial institutions in the advancedeconomies that are actively using MP-plus policies.Te potential eects on pension unds and insurancecompanies and evidence o emergent bubbles arecovered in Chapter 1. Te risk that central bank mea-sures will have macroeconomic and nancial stabilityeects abroad is an important topic that deservescareul analysis; to keep the scope o this chaptermanageable, it is not covered here, but it is examinedin Chapter 1 and in an IMF paper on unconventionalmonetary policy (IMF, orthcoming).2

In the areas it examines, the chapter nds ewimmediate nancial stability concerns associated with

MP-plus. So ar, it appears to have increased some mea-sures o bank soundness; and in markets where centralbanks have become major players, their interventioneither has not appreciably aected market liquidity or ithas corrected market dysunction. However, the longer

2Also see previous IMF publications or the eect on pensionsand insurance (or example, Chapter 2 o the September 2011GFSR) and spillovers (Chapter 4 o the April 2010 GFSR).


3/34

c ha p t e r 3 DO CENTRAL BANk POLIC IES SINCE ThE CRISIS CARRY RISkS TO FINANCIAL STABILITY?


that MP-plus policies remain in place, a number opotential uture risks are likely to increase, includingheightened credit risk or banks, delays in balance sheetrepair, difculties in restarting private interbank und-

ing markets, and challenges in exiting rom markets inwhich central banks have intervened. Te markets maybe alert to these medium-term risks, since the analysisnds evidence o an increase in the medium-term risko bank deault ater MP-plus announcements.

Policymakers should use micro- and macropruden-tial policies where possible to counter the nancialstability risks that may be emerging over the mediumterm. Implementing such policies in a measured man-ner, as needed, would allow MP-plus greater leewayto support price stability and growth while protect-ing medium-term nancial stability. However, the

exceptional nature o current monetary policies andthe relatively untested macroprudential tools in manycountries make this uncharted territory or policy-makers, and the eectiveness o the policy mix shouldbe careully monitored.

With a ocus on nancial stability, the chapterwill not address the timing or modalities o the exitrom MP-plus, although Box 3.1 notes some nancialstability risks that may arise with exit. Te chapterwill also not assess the current and uture economiceectiveness o unconventional monetary policies.Tese topics are covered in IMF (2010a) and IMF(orthcoming) respectively.

Mp-plus: an Ovviw

Ater the start o the nancial crisis in 2007, cen-tral banks in major advanced economies undertooka number o MP-plus measures.3 Tese measurescan be classied into our groups (with some overlapbetween groups): Prolonged periods o very low interest rates,sometimes

combined with orward guidance on the length o

time or which rates are expected to remain low; Quantitative easing(QE), which involvesdirect

purchases in government bond markets to reduceyield levels or term spreads when the policy rate isat or close to the lower bound;

3Annex 3.1 lists the various announcements o MP-plus mea-sures since the start o the nancial crisis.

Indirect credit easing(ICE), in which central banksprovide long-term liquidity to banks (sometimeswith a relaxation in access conditions), with theobjective o promoting bank lending; and

Direct credit easing(DCE), when central banksdirectly intervene in credit marketssuch asthrough purchases o corporate bonds or mort-gage-backed securitiesto lower interest rates andease inancing conditions (and possibly mitigatedysunction) in these markets.

MP-plus measures were taken with both macro-economic and nancial stability objectives in mind,with the mix depending, in part, on the mandates ospecic central banks. Te nancial stability objectivesare the subject o this chapter. Box 3.2 summarizes

IMF (orthcoming), which looks at the macroeco-nomic eects o unconventional monetary policies.

Tese operations have led to a undamentalchange in the size and composition o central bankbalance sheets. otal assets have increased signi-cantly, mostly in the orm o government securities,bank loans, equities, and mortgage-backed securi-ties (able 3.1 and Figure 3.1). Tese shits entailedspecic (and new) risks or central banks, includingcredit and market risks. Unless they are adequatelymanaged, including through enhanced loss-absorb-ing capacity, these risks (or perceptions about them)may aect the ability o central banks to perormtheir mandated roles and their credibility. I balancesheet assets are managed poorly, they could aectnancial stability, as discussed later in this chapter.

Outlined below are some risks that are, or mightbecome, associated with MP-plusnot all o themare currently evidentalong with recommendationsor corresponding policy responses. Te next sectionswill examine the extent to which some o these risksare emerging todayin specic nancial markets aswell as in nancial institutionsand which o them

may become more pronounced over the mediumterm. Te descriptions below are meant to providethe ull scope o potential channels through whichnancial stability could be aectedsome o thesechannels are examined below, others in Chapter 1.Tese eects ocus on domestic institutions and mar-kets; as noted above, other IMF publications addressthe important potential spillovers to other economies.


4/34



Prolonged periods o low interest ratescan aect theproitability and solvency o inancial institutions. Alattening o the yield curve puts pressure on banksinterest margins, and low interest rates increase thenet present value o liabilities o pension unds andlie insurance companies. Low-yielding assets mayinduce excessive risk taking in a search or yield,

which may maniest itsel in asset price bubbles.he low opportunity cost o unds and reduced netinterest margins may also give banks incentives todelay the cleanup o their balance sheets and reducepressure on authorities to demand vigorous bankrestructuring. Low interest rates could also encouragepockets o excessive releveragingin banks, which

tbl 3.1. ass holdings o Mjo cnl Bnks rld o Mp-plus, 200812

Dec-08 Dec-09 Dec-10 Dec-11 Oct-12 Reasons

Bank of England (in billions of pounds)Liquidity (longer term)1 170 24 17 10 11 Provide adequate bank refinancingAsset Purchase Facility

Gilts . . . 188 198 249 375 Raise nominal spending in order to meetinflation target by affecting level and shape

of yield curveCorporate bonds . . . 1.55 1.12 0.65 0.03 Improve liquidity in corporate creditCommercial paper . . . 0.43 0.00 0.00 . . .

Funding for lending . . . . . . . . . . . . 4.42 Encourage lending to the real economyMemorandum items:Total assets

GDP

238 238 247 290 4141,441 1,402 1,467 1,516 1,548

Bank of Japan (in trillions of yen)Liquidity (new stimulus) ... ... 24.8 32.0 29.0 Ease financing conditionsOther outstanding loans and repo 39.9 42.3 18.8 7.5 3.7Asset purchases

Commercial paper ... ... 0.1 2.0 1.5 Reduce market rates and risk premiumsCorporate bonds ... ... 0.1 1.5 2.9 across various types of financial assetsGovernment bonds and bills ... ... 1.2 5.6 28.4 and combat deflation risks

ETFs, REITs ... ... 0.02 0.9 1.6Memorandum items:Total assets 123 123 129 143 150Total sovereign holdings 63.1 72.0 76.7 90.2 107.6GDP 501 471 482 471 477

European Central Bank (in billions of euros)Short-term liquidity 226 81 249 160 117 Maintain sufficient bank intermediationLong-term liquidity 617 669 298 704 1059 and provide longer-term bank financingAsset purchases

Covered bonds (CBPP) ... 29 61 62 70 Sustain key bank funding channelGovernment bonds (SMP) ... ... 75 213 208 Maintain/restore European Central Bank policy

rate transmissionMemorandum items:Total assets

GDP

2,043 1,852 2,004 2,736 3,0479,242 8,922 9,176 9,421 9,503

Federal Reserve (in billions of U.S. dollars)Short-term liquidity

Loans and repo 2743 86 45 9 1.2 Provide adequate short-term bank funding

U.S. dollar swaps 554 10 0.08 100 12.5 Provide adequate funding for foreign exchangeoperations

Long-term liquidity Provide adequate long-term bank fundingTALF . . . 0.30 0.67 0.81 0.86 against MBS and ABS collateral

Asset purchasesAgency MBS . . . 908 992 837 852 Support housing financeAgency debt 20 160 147 104 82 Support GSEsTreasury securities 476 777 1016 1672 1651 Affect level and shape of yield curve

Memorandum items:Total assets

GDP

2,241 2,237 2,423 2,928 2,83214,292 13,974 14,499 15,076 15,653

Sources: Central banks websites; Haver Analytics; and IMF sta estimates.

Note: ABS = asset-backed securities; CBPP = Covered Bond Purchase Programme; ETFs = exchange traded unds; GSEs = government-sponsored enterprises; MBS = mortgage-backed securities;

REITs = real estate investment trusts; SMP = Securities Market Programme; TALF = Term Asset-Backed Securities Loan Facility.

1Zero short-term liquidity provision over the sample period outstanding at end-December 2008.

2Includes use o Extended Collateral Term Repo and Long-Term Repos.

3

Includes 28-day transactions under the TSLF (Treasury Securities Lending Facility) o about $190 billion.


5/34



are naturally leveraged, but also in the noninancialcorporate or household sectors.Banks will require

vigilant risk-based supervision, capital requirementsshould be adjusted to account or the true riski-ness o loan portolios and other assets, and well-designed dynamic and orward-looking provisioningshould be implemented (see Wezel, Chan-Lau, andColumba, 2012).

Quantitative easingcould exacerbate shortages osae assets (although the policy intention is, inpart, to encourage investment in riskier, more

productive assets).4 As with indirect credit eas-ing, the large increases in bank liquidity associ-

ated with QE could make inancial institutionsaddicted to central bank inancing (since centralbank intermediation o interbank unds shits

4Te availability o sae assets could decline through increasedcentral bank holdings (as a result o QE purchases) and throughthe increased encumbrance o assets, as banks post more collateralat central banks to obtain unding. Te latter is encouraged ascentral banks relax collateral rules. See also Chapter 3 o the April2012 GFSR.

6000

4000

2000

0

2000

4000

6000

500

400

300

200

100

0

100

200

300

400

500

Other assets

Lending to banks(short-term)Lending to banks(long-term)

Government securitiesBanknotes

Bank depositsCapital and reserves

Other liabilities

250

200

150

100

50

0

50

100

150

200

250Other assets Other assets

Lending to banksGovernment securities

Banknotes

Bank deposits

Capital and reserves

Other liabilities

4000

3000

2000

1000

0

1000

2000

3000

4000Other liabilitiesCapital and reservesBank deposits

BanknotesCovered bonds

Government securities

072006 0908 11 1210

072006 0908 11 1210 072006 0908 11 1210

072006 0908 11 1210

Figure 3.1. Changes in Central Bank Balance Sheets, 200612

Sources: Haver Analytics; national central banks; and IMF staff estimates.

Note: Government bonds purchased under the Bank of Englands quantitative easing (QE) program are held by a separate subsidiary, which is financed by loans from the Bank ofEngland (under other assets). Reported here are the amounts purchased under the asset purchase facility (the corresponding loan amount is subtracted from other assets).

1Including agency securities.2Special-purpose vehicles, commercial paper, and money-market-related assets.

Bank of England(In billions of pounds)

Federal Reserve(In billions of U.S. dollars)

European Central Bank(In billions of euros)

Bank of Japan(In billions of yen)

Lending to banks(short-term)Lending to banks(long-term)

Other assets

Lending to banks

Government securities 1

Mortgage-backedsecurities

Crisis-related facilities 2

Swaps

Banknotes

Bank depositsCapital and reserves


6/34



In considering the risks to nancial stability o exit rom

MP-plus, it is useul to distinguish between two aspects,namely, an exit rom low policy rates and the sale ocentral banks accumulated inventory o assets, most o

which are debt securities.

In the current cycle, as in previous ones, the centralbank will need to raise interest rates at some point tosaeguard price stability. But the need to sell assets totighten policy is less evidentcentral banks could sim-ply hold them to maturity and use other policy tools;but other concerns, including political considerations,may still prompt asset sales. Hence, the challenges andrisks o both types o exit must be anticipated and

managed, especially since the use o MP-plus policies isuncharted territory or policymakers.1

Te main nancial stability risks o exit areassociated with an unexpected or more-rapid-than-expected increase in interest rates, especially at thelonger end o the yield curve. Hence, when the timecomes to tighten nancing conditions or banks andthe economy, central banks would likely aim or ananticipated and gradual increase in interest rates,giving economic agents time to adjust. A disorderlyincrease or an overshootingperhaps as a result oshits in market sentimentwould make adjust-ment to the new nancial environment much more

difcult, heightening the risks listed below.Many MP-plus policies are unprecedented, and

they have now been in place or a relatively longtime. It is thereore even more important than dur-ing a normal tightening cycle that exit strategies are

well communicated to the general public as well asto markets, nancial institutions, and other centralbanks. Te risks below also underline the impor-tance o eorts to ensure that bank soundness andmarket liquidity are restored as soon as possible tominimize the nancial stability threats o a utureexit rom MP-plus.

Risks associated with increasing interest rates

include the ollowing:

Note: Prepared by S. Erik Oppers and Nico Valckx.1See IMF (2010a) or a description o the principles under-

lying exit strategies; IMF (orthcoming) presents some urtherthoughts on the topic.

Banks and other inancial institutions may incur

capital losses on ixed-rate securities. While theevidence suggests that a rise in interest ratesincreases net interest margins or banks, improv-ing their proitability over time, losses on ixed-rate securities available or sale are immediate.In the short term, thereore, weakly capitalizedbanks could suer. For inancial institutions

with long-term liabilities, such as pension unds,capital losses may be oset by a decrease in thenet present value o liabilities.

Credit risk or banks may increase. Higher interestrates could weaken loan perormance, especially ithe rise is in response to an inlation threat rather

than improved economic circumstances. Spillovers to other countries or markets may occur. Shiting expectations o the path o uture inter-est rates can lead to inancial lows betweenmarkets and countries that could be sudden andpotentially disruptive, especially i the timing otightening diers across central banks.

Risks associated with asset sales include the ollowing: Shits in market sentiment may lead to sharp

increases in yields. Uncertainty about the necessityor willingness o central banks to sell their largeportolios o government bonds and other assets

could lead to shits in market sentiment whencentral bank asset sales materialize.

Policy missteps may disrupt markets. I centralbanks sell assets beore underlying market vulner-abilities are addressed, dysunction could resur-ace. his risk is heightened in markets wherecentral banks hold a large share o outstandingsecurities or played an important market-makingrole, especially i ongoing market dysunction isnow masked by central bank intervention.

Banks may ace unding challenges. Just as thecounterpart o purchases o assets by central banks

was an increase in banks excess reserves, the

counterpart o asset sales would likely be a declinein banks excess reserves. his disintermediationo interbank liquidity by the central bank wouldhave to be oset by a revival o private interbankmarkets. I this market is not ully restored, somebanks could ace unding challenges.

Box 3.1. Finnil Sbiliy risks assoid wi exi om Mp-plus poliis


7/34



credit risk away rom the private parties), delay-ing balance sheet repair and the restoration o aninterbank market. Improved liquidity risk man-agement in banks and implementation o Basel IIIliquidity requirements can help ease some o theserisks (see Chapter 2 o the April 2011 GFSR).

Indirect credit easingcould make inancial institu-tions dependent on long-term central bank (thatis, public sector) inancing, delaying the restora-

tion o private sources o unding and providingincentives to allocate bank credit toward bor-rowers that qualiy or the associated lendingprogram. Some o these borrowers might nototherwise qualiy or loans, thereby weakeningunderwriting standards, with potential adverseeects on longer-term loan perormance and

hence on the uture health o banks. hese risksto loan perormance should be acknowledgedby banks and their supervisors, and appropriateorward-looking provisions should be made.

Direct credit easingcouldintroduce distortions toprices and market unctioning i central banksbecome the dominant buyer in markets in whichthey intervene. hese distortions could emergewith rising expectations o an imminent central

bank exit and could under certain circumstanceslead to large price swings and other dysunction.Banks may be hurt by these price swings i theyhold large volumes o securities traded in thesemarkets. Supervisors should be cognizant o thesepotential risks, which banks should be required toaddress.

Central banks have deployed a variety o unconventional

measures during the crisis. But is there a limit to theirefectiveness in case o a potentially prolonged downturn?

A orthcoming IMF publication, Unconven-tional Monetary Policies: Recent Experience andProspects, addresses three questions about uncon-

ventional monetary policies. First, what policieswere tried, and with what objectives? Second, werepolicies eective? And third, what role might thesepolicies continue to play in the uture?

Central banks in key advanced economiesadopted a series o unconventional monetary poli-cies with two broad goals. Te rst was to restore

the unctioning o nancial markets and intermedia-tion. Te second was to provide urther monetarypolicy accommodation at the zero lower boundo policy interest rates. Tese two goals are clearlyrelated, as both ultimately aim to ensure macroeco-nomic stability. But each relies on dierent instru-ments: the rst on targeted liquidity provision andprivate asset purchases, and the second on orwardguidance and bond purchases.

Tese policies largely succeeded in achieving theirdomestic goals, and were especially eective at the

time o greatest nancial turmoil. Market unction-ing was broadly restored, and tail risks declinedsignicantly. Policies also decreased long-term bondyields, and in some cases credit spreads. Someevidence also suggests that these policies encour-aged growth and prevented deation, althoughthis conclusion is less clear-cut, given the long lagsand unstable relations between variables, and theunresolved question o what would have happened

without central bank policy intervention.Unconventional monetary policies had a mixed

eect on the rest o the world. Early policyannouncements buoyed asset prices globally, andlikely beneted trade. Later announcements had

smaller eects and increased capital ows to emerg-ing markets, with a shit to Latin America and Asia.Sound macroeconomic policies can help managethese capital ows. Yet, when ows become exces-sive, with the risk o sudden reversals, they can giverise to policy strains in recipient countries.

Looking ahead, unconventional monetary policiesmay continue to be warranted i economic condi-tions do not improve or i they worsen. Yet, bondpurchases in particular seem to exhibit diminishingeectiveness, and their growing scale raises risks. Akey concern is that monetary policy is called on todo too much, and that needed scal, structural, and

nancial sector reorms are delayed.

Box 3.2. t Moonomi eivnss o Mp-plus

Note: Prepared by ommaso Mancini-Grioli.


8/34



Some o these risks are closely connected to theintended policy objectives. For example, althoughcentral bank intervention may distort marketdynamics or unctioning in a way that may have

negative implications or nancial stability, draw-ing investors (back) into intervened markets may inact have been the intended goal o the policy. Tishighlights the care with which the potential threatsto nancial stability need to be evaluated.

es o Mp-plus on Mks

Mony nd Inbnk Mks

Te prolonged period o low interest ratesincreases risks in money markets, includingthrough developments in money market mutual

unds (MMMFs). With interest rates remainingnear zero in the maturities at which MMMFs arepermitted to invest, these institutions are experi-encing very low (in some cases zero or negative)returns that in many cases ail to cover the costso und management. As a consequence, U.S.MMMFs have raised credit risk modestly (withinthe connes o regulatory restrictions), engaged inmore overnight securities lending, granted ee waiv-ers, and turned away new money.

Te undamental problem is that to become pro-itable the MMMF industry needs to shrink urther,and the risk is that it may do so in a disorderlyashion. For example, another run on MMMFs mayoccur i downside credit risks materialize or securi-ties lending suddenly halts, ueling investors earo MMMFs breaking the buck (that is, ailingto maintain the expected stable net asset value).Once started, a run may accelerate because inves-tor guarantees that were established in the wakeo the Lehman Brothers bankruptcy have beenremoved, and the Dodd-Frank Act precludes theFederal Reserve rom unilaterally stepping in to

provide liquidity to the sector.5

Although the assets5Te U.S. reasury Department introduced the emporary

Guarantee Program, which covered certain investments inMMMFs that chose to participate in the program and has nowexpired. Te Federal Reserve created an Asset-Backed Com-mercial Paper Money Market Mutual Fund Liquidity Facility,through which it extended credit to U.S. banks and bank holdingcompanies to nance their purchases o high-quality asset-backedcommercial paper rom MMMFs.

o MMMFs are already shrinking in the low interestrate environment as investors seek higher returnselsewhere, an outright run would be undesirable andcould have systemic consequences i the unding that

these institutions provide to banksdirectly andthrough overnight securities lendingdries up.

Central bank interventions in the interbankmarkets were a response to a signicant reductionin interbank lending activity that mostly resultedrom increased sensitivity to counterparty risk.With indirect credit easing policies, central banksmade longer-term unds available at xed lowrates and sotened collateral rules, aiming to avoida severe credit contraction. Tis orm o crediteasing lowered interbank spreads during the crisis,especially in the euro area and Japan. By partially

replacing the interbank market, central banks playa crucial role in the distribution o bank unding insome areas.

From a money-market perspective, risks stem notso much rom central bank intervention itsel asrom a misstep in the eventual withdrawal rom themarket.I central banks exit rom interbank marketsbeore underlying conditions are addressed andthe private bank unding market is ully restored,renewed strains could resurace, with the costs oshort-term bank nancing turning signicantlyhigher or some banks.Tese risks are difcult toquantiy because central bank intervention maymask the dysunction it was designed to address. Adecomposition o interbank spreads may oer someinsights (Figure 3.2).Central bank liquidity no lon-ger appears to signicantly aect interbank marketspreads in the United States and the United King-dom. Tis could indicate that uture central bankexit rom these markets would not aect interbankspreads there. In the euro area and Japan, however,central bank intervention (Central bank liquidityin Figure 3.2) appears to continue to mask more

elevated interbank market spreads due to increasedsensitivity to counterparty risk (Bank risk inFigure 3.2). Tis could be an indication that spreadscould increase i and when central banks withdrawbank liquidity, although the gradual decline o suchliquidity in Japan over the past year (see Figure 3.1)does not appear to have led to signicantly increasedyield spreads.


9/34



Mogg nd coo Suiis Mks

Direct credit easing by the major central banksthrough interventions in mortgage and corporatebond markets have attempted to improve liquid-ity and lower interest rates or borrowers in thesemarkets. During 2009 and the rst hal o 2010,the Federal Reserve purchased close to $1 trillion in

mortgage-backed securities (MBS) to support the U.S.housing market and alleviate pressures on the balancesheets o U.S. banks. It made a new commitment tobuy MBS in September 2012 in an eort to lowermortgage interest rates urther and spur credit exten-sion (Figure 3.3). In two purchase programs, the ECBbought a total nominal amount o 76.4 billion ocovered bonds, and the BOE bought up to 1.5 bil-

lion in corporate bonds. Te BOJ also maintains alimited program to purchase corporate bonds, realestate investment trusts (J-REIs), and exchange-traded unds (corporate stocks).

Some central banks have made extensive purchasesin these markets. While geared toward clear objectives,these programs may mask continuing underlying dis-tortions, and their removal may pose policy challenges.

Te programs o the Federal Reserve and ECB appearto have reduced yields as intended (see Figure 3.3; andIMF, orthcoming).6 In particular, the purchases o the

6In addition, an analysis (not reported here) o Federal Reserveinterventions in MBS markets and ECB interventions in euroarea covered bond markets (controlling or other risk actors)conrms the signicant eect on yields o these MP-plus policies.

200

100

0

100

200

300

400 Residual

Central bank liquidityBank riskGDP riskLIBOR-OIS spread

200

100

0

100

200

300

400

200

100

0

100

200

300

400

2007

200

100

0

100

200

300

400

0908 11 1210

2007 0908 11 1210

2007 0908 11 1210

2007 0908 11 1210

Figure 3.2. OIS Counterparty Spread Decompositions(Three-month LIBOR-OIS spread, in basis points)

Sources: Bloomberg L.P.; Datastream; JPMorgan Chase; and IMF staff estimates.

Note: CDS = credit default swaps; LIBOR = London interbank offered rate; OIS = overnight indexed swap; PMI = Purchasing Managers Index. Decomposition based on least-squaresregressions of weekly LIBOR-OIS spreads on a constant (not shown), indicators of growth risk (PMI-based GDP tracker), sovereign risk (changes in sovereign CDS spread; the SovxWestern Europe Index for the euro area; sovereign CDS spreads for the United Kingdom and the United States), bank risk (major bank equity index historic 90-day volatility), and central

bank liquidity (liquidity provision to banks as a percent of banking sector assets). Higher growth, increased central bank liquidity, lower bank volatility, and sovereign risk all tend toreduce LIBOR-OIS spreads. The sovereign risk contribution is not shown, as it is very small relative to the other factors in the regression.

British PoundU.S. Dollar

Japanese YenEuro


10/34



Federal Reserve have made it a major market player,holding 20 percent o outstanding MBS.7 Central bankintervention in these markets does not in itsel threatennancial stability (indeed, it was designed to saeguardit), but it does raise policy risk surrounding a utureexit. While the presumption may be that central banksshould not and would not exit beore underlying con-ditions permit, the large current role o central banksmay mask underlying vulnerabilities in the privatemarket that may be difcult to assess. An inadvertentlypremature exit could have an adverse impact on market

liquidity and prices i it turns out that underlying mar-ket conditions have not improved.

Govnmn Bond Mks

Te Federal Reserve, BOE, and BOJ bought govern-ment bonds in quantitative easing programs with themain goal o lowering long-term interest rates. Teanalysis in IMF (orthcoming) ound that these poli-cies were broadly eective in reducing interest rates inthese markets. Forward guidance has also kept yields

on government bonds low. Te longer the guidance isin place, however, the more complacent markets may

7In the euro area, although the ECB holds only 5 percento outstanding covered bonds, it also played a large role in theprimary market, purchasing about 10 percent o covered bondissuance in 2009, 5.5 percent in 2010, and nearly 4 percent in2012. Covered bonds are also increasingly issued and retained bybanks or use as a high-quality collateral source or accessing ECBlending acilities.

become about the implicit promise o intervention. Soar, studies have suggested that the Federal Reservesinterventions have not impaired market unctioning(Fleming and Mizrach, 2009; Engle and others, 2012).Market indicators appear to support this conclusion:overall, in the United Kingdom and the United States,the price impact o trade was relatively stable duringperiods o central bank asset purchases, and in Japan itappears to have allen (Figure 3.4). With the possibleexception o the rst round o QE by the FederalReserve, correlations between central bank purchases

o government bonds and liquidity indicators such asprice volatility, turnover, and the price impact o tradeare generally small (Figure 3.5).

Trough its Securities Market Programme (SMP),the ECB temporarily sought to support sovereign bondmarkets in periphery euro area countries that showedsigns o dysunction. Te Outright Monetary ransac-tions program (OM), announced in September 2012,also aims at supporting targeted sovereign bond marketsby reducing risk premiums on these targeted securities.8Yields on periphery sovereign bonds have declined

signicantly since the announcement o the OM, eventhough the program has not yet been activated.Te increasing share o government bonds held by

central banks may present risks to nancial stability.

8Te ECBs indirect credit easing through three-year liquid-ity operations in late 2011 and early 2012 are also seen to haveimproved liquidity conditions in some euro area sovereign bondmarkets.

Figure 3.3. Central Bank Intervention in Real Estate Securities Markets

30-year MBS index (left scale)15-year MBS index (left scale)

Federal MBS holdings (right scale)

Federal Reserve Purchases of Mortgage-Backed Securities (MBS)35 years (left scale)710 years (left scale)

10+ years (left scale)CBPP holdings (right scale)

Euro Area Covered Bond Yields and CBPP Purchases 1

Sources: European Central Bank; Federal Reserve; JPMorgan Chase; and IMF staff estimates.Note: DCE = direct credit easing; CBPP = Covered Bond Purchase Programme. Shaded areas show different periods of DCE and CBPP purchases.1Covered bond yields refer to euro area Pfandbriefe indices.

0

200

400

600

800

1000

1200

0

1

2

3

4

5

2009 2010 2011 2012 2013

BillionsofU.

S.

dollars

Percent

Billionsofeuros

Percent

DCE1 DCE2

0

20

40

60

80

100

0

2

4

6

8

10

2009 2010 2011 2012 2013

CBPP1 CBPP2


11/34



Figure 3.4. Central Bank Holdings of Domestic Government Securities and Market Liquidity, by

Maturity

Sources: Bank of England; Bank of Japan; Bloomberg L.P.; Datastream; Federal Reserve Bank of New York; Japan, Ministry of Finance; Japan SecuritiesDealers Association (JSDA); JPMorgan Chase; U.K. Debt Management Office; U.S. Treasury; and IMF staff estimates.

Notes: APF = Asset Purchase Facility (Bank of England); APP = Asset Purchase Program (Bank of Japan); JGBs = Japanese government bonds; QE =quantitative easing (Federal Reserve). Left panels are central banks' holdings of domestic government securities. QE1, MarchOctober 2009; QE2, August2010June 2011; QE3, October 2011present. APF1, March 2009January 2010; APF2, October 2011October 2012. APP, November 2010present. Right

panels show the price impact of trade, an indicator of market liquidity, defined as the weekly percentage p rice change (in absolute terms) d ivided by the weeklytrading volume. Impact data are weekly for the United States and the United Kingdom, and at a 10-day frequency for Japan, interpolated from JSDA monthly

data.

United States

0

5

10

15

20

25

30

35

40

4550

03 years36 years611 years11+ years

Treasury Securities(In percent of outstanding debt)

QE1 QE2 Twist/QE3

0.00

0.01

0.02

0.03

0.04

0

0.02

0.04

0.06

0.08

03 years

36 years

611 years

11+ years

(right scale)

Price Impact of Trade(4week moving average)

QE1 QE2 QE3

0

10

20

30

40

50

60

70

07 years

715 years

15+ years

APF Holdings

(In percent of outstanding debt)

APF1 APF2

0.00

0.02

0.04

0.06

0.08

0.10

0.12

07 years

715 years

Price Impact of Trade

(5week moving average)

APF1 APF2

0

10

20

30

Bills (up to 1 year)

JGBs (2+ years)

APP

0.000

0.001

0.002

0.003

0.004

0.005

1 year

All traded

Government Securities(In percent of outstanding debt)

Price Impact of Trade(4week moving average)

APP

Japan

United Kingdom

2006 0807 09 10 11 12 13 2006 0807 09 10 11 12 13

2006 0807 09 10 11 12 13 2006 0807 09 10 11 12 13

2006 0807 09 10 11 12 13 2006 0807 09 10 11 12 13


12/34



Te Federal Reserve and the BOJ now each holdsome 10 percent o their respective governmentsdebt, the BOE holds 25 percent, and the ECB holdsan estimated 5 percent to 6 percent o the outstand-

ing sovereign debt o Italy and Spain. Te shares oFederal Reserve and BOE holdings o longer-datedsovereign bonds are even higher at more than 30percent. Te central banks large holdings couldaect market expectations. Once economic condi-tions warrant the withdrawal o monetary stimulus,markets may anticipate that central banks will switchrom buying government bonds to actively sellingthem, and political pressure may be exerted to movethe monetary authorities in that direction. Suchexpectations could sharply drive up yields.9 Tere-ore, it will be important that, well in advance o the

need or tightening, central banks communicate thecircumstances in which a tightening may occur andclariy that tightening need not imply outright sell-ing o bonds rom the central banks balance sheet.10o the extent that large holdings o governmentbonds could result in large implicit or explicit lossesor central banks (i the securities are marked tomarket or sold beore maturity), it will be importantto have arrangements in place that ensure adequatecapital or indemnication or losses (Box 3.3).

es on O Mks

Markets that are not directly targeted by MP-pluspolicies may nonetheless be aected. Credit easing,quantitative easing, and commitments to prolongedlow policy interest rates may trigger ows into othermature asset markets (corporate bonds, equities, com-modities, secondary currencies, and even housing).While encouraging a certain degree o risk taking isindeed the purpose o many MP-plus policies, theycould unintentionally lead to pockets o excessivesearch or yield by investors and to exuberant price

developments in certain markets, with the potential

9In 1994, the Federal Reserve caught market participants oguard by suddenly raising policy rates, causing turmoil in bondmarkets and especially in the agency MBS market, where investorsinsufciently understood prepayment risks.

10Te implications o government bond holdings on commer-cial banks balance sheets are discussed in the nal section o thechapter.

0.50

0.25

0.00

0.25

0.5003 years

36 years

611 years

11+ years

United States

0.50

0.25

0.00

0.25

0.50

QE1 QE2 QE1 QE2 QE1 QE2 QE1 QE2

PVol Turnover Daily Trading Price Impact

PVol Turnover Trading Price Impact

QE1 QE2 Twist

/QE3

QE1 QE2 Twist

/QE3

QE1 QE2 Twist

/QE3

QE1 QE2 Twist

/QE3

PVol Turnover Daily Trading Price Impact

07 years

715 years

15+ years

United Kingdom

0.50

0.25

0.00

0.25

0.50Bills (up to 1 year)

All traded JGBs

Japan

Figure 3.5. Correlations between Central Bank Holdings of Government

Securities and Market Liquidity, by Maturity of Holdings

Sources: Bank of England; Bank of Japan; Bloomberg L.P.; Datastream; Federal Reserve Bank of NewYork; Japan, Ministry of Finance; Japan Securities Dealers Association (JSDA); JPMorgan Chase; U.K.Debt Management Office; U.S. Treasury; and IMF staff estimates.

Note: JGBs = Japanese government bonds; P-Vol = conditional bond return volatility (see discussion

below); QE = quantitative easing. Figures show correlations between central bank holdings ofgovernment securities (as a percent of outstanding debt by maturity segment) and four indicators ofliquidity in the government bond market during periods of active quantitative easing . P-Vol is estimated

from daily data (log first differences), with an exponential Garch(1,1) process, allowing for asymmetricleverage effects. Trading is the average daily trading volume during a particular week. Tu rnover is

weekly trading volume divided by the outstanding stock of debt (by segment). Price impact is the weeklypercentage price change (in absolute terms) divided by the weekly trading volume. For Japan, turnover

and trading data are interpolated from JSDA monthly volumes to tri-monthly periods.


13/34



Risks on balance sheets o central banks have increased

since the start o the crisis, with potential negative conse-quences or their nancial strength and independence.

Enhanced liquidity provision, relaxation o col-lateral rules, and sizable asset purchases have led toincreases in the absolute size o central bank balancesheets, an increase in the duration and diversity oassets, and a decline in asset quality. Tese changespose risks, including: Implicit or explicit valuation losses as a result o a

rise in interest rates; Declines in operating income when central banks

increase their holdings o long-dated securities

with low coupon interest rates; and Possible impairment losses on assets with credit risk.Te extent to which the various central banks are

exposed to these risks diers, depending on the scopeand nature o their unconventional policies (whichthemselves may be inuenced by a central banks risktolerance). Te Federal Reserve, Bank o England(BOE), and Bank o Japan (BOJ) purchased largequantities o bonds to lower long-term yields andsupport economic activity, whereas the EuropeanCentral Bank (ECB) mainly expanded the provisiono liquidity to support bank unding (see able 3.1). he Federal Reserve holds a large portolio o rea-

sury securities and mortgage-backed securities (16percent o GDP at end-2012), and it has extendedthe maturity o its holdings o reasury securitiesconsiderably over time: its modiied durationa measure o interest rate sensitivityincreasedrom about 2 beore the crisis to nearly 8 mostrecently. his means that a 1 percent increase ininterest rates would reduce the portolios market

value by 8 percent; and taking into account bondprice convexity, the drop in market value wouldcorrespond to a capital loss o about 4 percent othe Federal Reserves total assets.

he BOJ and BOE are also subject to interest

rate risk given their sizable government bondholdings (about 24 percent o GDP each atend-2012). A 1 percent increase in interest rates

could result in a loss o about 1 percent o totalassets or the BOJ and 6 percent or the BOE.1For the BOJ, this igure could increase on urtherimplementation o its Asset Purchase Program. Inaddition, the BOJ is also subject to market riskrom its holdings o private assets.2

he ECB increased its lending exposure to banksin euro area periphery countries rom 20 percento total reinancing operations in 2006 to abouttwo-thirds in 2012, which raised its credit riskproile. hese risks are mitigated to a considerableextent by collateral requirements. he ECB isalso exposed, but to a lesser extent, to credit andinterest rate risks arising rom holdings o covered

bonds and periphery sovereign bonds.Central banks can mitigate these risks in various ways. Shorten asset duration so that seigniorage

income matches central bank policy expense (orexample, central banks could negotiate an assetswap with national treasuries to boost income).

Increase the share o higher-yielding assetsthiswould most easily be accomplished by purchasingsuch assets during exit rom MP-plus.

Increase capital buers to cover potential losses,through proit retention or capital injection. Forexample, even beore most o its interventions,the ECB doubled its subscribed capital to 10.8

billion at end-2010. Similarly, in 2011, the BOJretained proits in excess o legal requirements tobuild up capital reserves.

Adjust haircut requirements to relect changes inthe quality o collateral.

Secure a ull indemnity rom national treasuriesor losses associated with MP-plus. For example,the BOEs Asset Purchase Facility is ully indem-niied by its reasury, and thereore the BOEdoes not ace associated inancial risks.

1Te BOEs exposures are kept o-balance-sheet in theBOE Asset Purchase Facility Fund.

2Te BOJs holdings o private sector securities are smalland thus pose relatively limited balance sheet risk despiteoccasional unrealized losses. Te BOJ does not ace substantialcredit risk on its lending acility, as it requires pooled col-lateral. Te BOEs Funding or Lending Scheme also entailssome credit risk, albeit only a limited amount given the smallsize o the program.

Box 3.3. Bln S risks o Unonvnionl poliy in Mjo cnl Bnks

Note: Prepared by Kotaro Ishi, Raphael Lam, Kenneth Sul-livan, and Nico Valckx.


14/34



or bubbles. Chapter 1 evaluates various potentialtransmission mechanisms. Te sharp rise in investordemand or credit products, combined with con-strained supply, is supporting a substantial declinein corporate borrowing costs. In turn, investorsare accommodating higher corporate leverage andweaker underwriting standards to enhance yield.Some components o the credit market, such as loanswith relaxed covenants, are experiencing more robustgrowth than in the last credit cycle (see Chapter 1).11

Although not analyzed here, the potential spillovereects o MP-plus to other economies are important.MP-plus could aect nancial stability in liquidity-receiving economies via three main channels: excessivecurrency movements, domestic asset price bubbles, andsudden stops once the global liquidity is unwound.

IMF (orthcoming) explores actual and potential spill-over eects rom MP-plus. Early MP-plus announce-ments, which strengthened market and nancialstability in the advanced economies, buoyed asset pricesglobally and led to the appreciation o currencies o

11Tese eects are covered in the September 2011 GFSR andin the orthcoming IMF paper. See also BIS (2012b).

emerging market economies. Tese announcementsmostly drew money back to the United States, whilelater announcements sent money to emerging markets,though with more muted eects on asset prices. Morebroadly, aggregate capital inows to emerging marketeconomies have mostly returned to their ample precrisislevels. Nevertheless, Chapter 1 nds that pockets opotential risk in some countries with more persistentcapital inows are raising the possibility o excesses insome important segments o emerging market econo-mies. For example, a unique eature o the current cycleis that corporations in such economies have increasedoreign-currency debt nancing in place o local-currency equity. While these debt levels are not yetthreatening, conditions are in place or a less avorableoutcome i the trend continues.12

es o Mp-plus on Finnil Insiuions

o quantiy the eects o MP-plus on the soundnesso domestic nancial institutions, the analysis here will

12Spillovers are also discussed in the April 2010 GFSR, as wellas in IMF (2012b) and BIS (2012a).

Te extent to which these dierent measures canbe used by central banks diers, depending on riskexposure and tolerance, institutional setup, andeconomic and nancial circumstances.

In addition, the extent to which these holdingsrepresent risks and are being recognized dependson accounting rules and how central banks intendto use the securities. I they intend to hold thesecurities to maturity, potential capital losses willnot be realized i interest rates rise (although inter-est income would be below markets rates untilmaturity). Te Federal Reserve, the ECB, and theBOJ value their holdings o securities at amortizedcosts, although in certain circumstances they are

required to take on impairments i values dropsubstantially. In contrast, the BOE uses mark-to-market accounting or government bonds and othersecurities. Te current ECB portolios are held tomaturity (and thereore not subject to marking tomarket) but a possible uture Outright Monetary

ransactions portolio would be marked to market.However, in all cases, market participants will likelyimpute the values o central bank holdings o securi-ties to evaluate their overall saety and soundness. Itbehooves central banks, thereore, to manage theirrisks in a transparent and consistent ashion.

Experience in some jurisdictions (mostly emerg-ing market economies) has shown that central bankscan execute their monetary policy unctions whileexperiencing large losses (or even while having nega-tive net worth), but such situations may neverthelessthreaten their independence and credibility. Histori-cal evidence shows that nancially weak centralbanks are prone to government intererence (Stella,

2008; and Stella and Lnnberg, 2008), therebypotentially undermining their policy perormance.Te extent to which independence is compromisedby nancial weakness would depend crucially onother saeguards or independence that are in placeor a particular central bank.

Box 3.3 (continued)


15/34



ocus on banks. Healthy banks are critical to nancialstability and to eective monetary policy transmission,as the recent nancial crisis has shown. Risks in banksare also potentially heightened because leverage is part

o their business model. MP-plus aects banks directlythrough various channels, including by providingliquidity, lowering bank unding costs (through lowinterest rates), and supporting asset prices (throughcentral bank asset purchases).13 MP-plus also hasimportant indirect benets or banks: by supportingeconomic activity, it increases the demand or loans andlowers credit risk in bank loan portolios.

Te eect o MP-plus on bank risk and its relationto nancial stability should be evaluated careully.One o the macroeconomic goals o MP-plus is argu-ably to encourage banks to contribute to economic

growth by clearing troubled assets rom their balancesheets and making more loans to sound borrow-ers (a risky activity). Financial stability would bethreatened only i risk taking by banks was exces-sive and worsened their nancial health. o evaluatenancial stability eects, it is thereore necessary tolook beyond narrow measures o bank risk to broadmeasures that would indicate a weakening o banksoundness, such as the z-score and bank deault risk.14

Te analysis uses three complementary approachesto assess the eects o MP-plus on banks. Te rst isan event study, which is based on the idea that anyeects o MP-plus policy initiatives on bank sound-ness (including bank deault risk and perormance)should immediately be reected in changes in bankstock prices, since the stock price is a risk-adjusteddiscounted value o uture bank income streams.Similarly, any eects o MP-plus on bank deaultrisk should immediately be reected in bank bondspreads. Relating a measure o MP-plus policy actionsto these market indicators at the time o an MP-pluspolicy announcement can thereore oer some insightinto market participants current view o their impact.

Te second approach urthers the understanding othe channels o impact on banks by using bank-leveldata. It relates indicators o monetary policy to mea-

13For a more thorough treatment o the various channels otransmission o MP-plus, see IMF (orthcoming).

14Te z-score is a standard measure o bank soundness that isinversely related to a banks probability o insolvency; see Laevenand Levine (2008) as well as the notes to able 3.7 in Annex 3.2.

sures o banks nancial health, including protability,risk taking, and the status o balance sheet repair.

Te third approach ocuses on a possible rise ininterest rate risk in banksa potential consequence

o the prolonged period o low interest rates. Itexamines two main channels through which banksare aected by increases in interest rates: net interestincome and the value o xed-rate securities (mainlygovernment bonds).

evn Sudy

Te event study analyzes the eect o MP-pluspolicy announcements on domestic bank stockprices and bank bond spreads. A complication isthat announcements may be partly expected andpriced into the markets beore the actual announce-

ment. Any measured eect on bank stock pricesand bank bond spreads may thereore seem mutedwhen compared with the announced measures.Tese prices would react only to new inormation,that is, the unexpected or surprise element o theannouncement. Bernanke and Kuttner (2005) andGrkaynak, Sack, and Swanson (2005) show thatthe surprise element o monetary policy announce-ments can be measured by changes in orward ratesat the time o announcement.15 Tese changes,representing the surprise element o the announcedpolicies, could then be related to changes in bankstock prices and bank bond spreads to gauge theirperceived impact on bank health.

Te event study used here gives an indication o themarket perception o the eects on banks equity o theannounced policies. Regressions o bank stock returnson the policy surprise measurethe change in interestrate uturesyield the ollowing results (able 3.2): Bank stock prices are not aected by a surprise

easing o monetary policy in the United States; butin the United Kingdom, bank stocks all 6.6 basis

15

Te one-year-ahead utures rate is used to measure the mon-etary policy surprise (see notes to able 3.2 or details) to captureboth the contemporaneous part o monetary policy announce-ments (the target policy rate) and any expected near-term uturedevelopments (or example, orward guidance). With the short-term interest rate approaching zero in later years, the movementsin the one-year-ahead utures rate may be limited and thus mayaect the coefcients in the regressions or the MP-plus period.Partly or this reason, surprises are allowed to have dierentialeects between the conventional and MP-plus periods.


16/34



points per basis point o surprise monetary eas-ing. hese eects are the same or conventionaleasing and or MP-plus easing. In the euro area,bank stocks all 5.6 basis points per basis point o

surprise conventional easing and an additional12.9basis points per basis point o MP-plus easing. he markets see the risk o uture bank deault

rising as a result o a surprise monetary easing,indicated by an increase in the spread betweenmedium-term bank bonds and governmentbonds over various maturities. Each basis point osurprise easing increases these spreads by between

0.071 and 0.154 basis point, depending on thecountry and the speciic maturity o the bonds.his eect is the same or conventional easingand or MP-plus in most cases, although there is

weaker evidence o an additional rise in the spreado 0.156 basis point or a surprise 1 basis pointMP-plus easing in the euro area.

In sum, the market perceives monetary easing ingeneral as neutral or negative or bank health (asmeasured by bank stock prices), and considers it asincreasing bank deault risk in the medium term.

tbl 3.2. rsuls om evn Sudy rgssions1

United States

Effect on Bank Stock ReturnMSCI Bank Stock Index

(Daily returns, in percent)

Effect on Financial Sector Credit RiskFinancial Sector BondGovernment Bond Spread2

(Daily changes, in basis points)

13 year 35 year 57 year

Effect of a surprise monetary easing, per basis point 0.078*** 0.087*** 0.075**Additional effect of MP-plus easing, per basis point Constant Change in constant, MP-plus events

Number of observations 103 103 103 103R-squared 0.085 0.066 0.090 0.044

Euro Area

Effect on Bank Stock ReturnMSCI Bank Stock Index

(Daily returns, in percent)



13 year 35 year 57 year

Effect of a surprise monetary easing, per basis point 0.056** 0.126*** 0.154*** 0.130***Additional effect of MP-plus easing, per basis point 0.129** 0.156*

Constant Change in constant, MP-plus events

Number of observations 156 156 156 156R-squared 0.187 0.212 0.215 0.121

United Kingdom

Effect on Bank Stock ReturnFTSE All Share (Bank) Index(Daily returns, in percent)



Effect of a surprise monetary easing, per basis point 0.066*** 0.071***Additional effect of MP-plus easing, per basis point Constant Change in constant, MP-plus events

Number of observations 138 138R-squared 0.089 0.033

Sources: Bank o America Merrill Lynch; Bloomberg L.P.; and IMF sta estimates.

Note: ***, ** and * indicate that estimated coecients are signicant at the 1 percent, 5 percent, and 10 percent level, respectively. indicates that the coecient was not signicant

at the 10 percent level; these coecients are not reported in the table. The conventional policy period is rom January 2000 through July 2007, and the MP-plus period is restricted to

events ater the Lehman Brothers collapse through October 2012. For the United States, the sample excludes September 12, 2001. A surprise monetary easing is measured by the change

in the one-year-ahead three-month Eurodollar utures rate or the United States, the equivalent Euribor utures rate or the euro area, and the equivalent Sterling utures rate or the United

Kingdom.

1For ease o interpretation, coecients are reported so that a positive coecient indicates a rise in returns or the bond spread as a result o monetary easing.

2All maturities are used or the United Kingdom because short-term spreads are not available. Adjusted or any options o corporate bonds, such as early retirement.


17/34



Te perceptions or conventional easing are gener-ally not dierent rom those or MP-plus measures.Tis nding is surprising in that it runs counter tothe nancial stability objectives o policymakers.

It may be an indication that even though policieshave aimed at supporting the macroeconomy andostering nancial stability in the short term, theymay nevertheless carry risks or bank soundnessover the medium term. Moreover, i the marketbelieves that central banks have superior inorma-tion on economic conditions, a surprise easing maybe seen as signaling that the central bank believesthat conditions are worse than the market perceived,leading to a all in bank stocks immediately ater theannouncement.16

Bnk-Lvl D anlysisTe second approach to investigating the eects o

MP-plus on bank soundness is to use bank-level datato measure nancial health. Whereas the event studylooked at market perceptions o bank soundness andrisk, this approach uses a panel regression methodol-ogy that directly relates indicators o monetary policyto various measures o bank soundnessbank prot-ability, risk taking, and eorts toward balance sheetrepair. Te required data are available or relativelyew banks in the euro area, Japan, and the UnitedKingdom, making a conclusive analysis or themmore difcult. Te analysis thereore ocuses on theUnited States. Te monetary policies considered coverconventional as well as unconventional measures.17

Te results rom bank-level data analysis need tobe interpreted with caution. Te analysis uses themonetary policy variables as independent variables,assuming they cause the changes in the banksoundness indicators. However, the central bankactions since 2007 have been partlyin response toproblems in banks, so they may not be truly inde-

16

In Japan (not included in our event study), the January22, 2013, Joint Statement by the government and the BOJ hasbeen associated with increases in bank stock prices. While thesedevelopments are too recent or a ull analysis, the explanation orthis opposite result may be that the announced policies have beenseen as increasing the likelihood o ending deation and improv-ing economic prospects in general, beneting banks and therebybuoying bank stocks.

17See Annex 3.2 or details on the estimation methodology andable 3.7 or detailed results.

pendent. Te analysis made adjustments to workaround this problem and to better capture the eectso MP-plus on bank soundness (see Annex 3.2). Inaddition, by using data only or the United States,

the analysis covers the banks where improvements insoundness have been most evident.

Another issue is that, besides the inuence omonetary policy, bank balance sheets have beenaected by scal, nancial, and other actors overthe period. Te regressions thereore also includevariables controlling or output growth, scal poli-cies, and stress in the nancial system (see Annex3.2). Still, the analysis may not be able to ully cap-ture the direct eects o MP-plus policies on banksi those policies manage to raise economic growthand thereby indirectly benet the nancial health

and riskiness o banks.Te estimated eects o MP-plus on banks

income statements and balance sheets are mixed.Te analysis calculates the eects o (1) monetaryeasing itsel, (2) a sustained period o easing, and(3) an expansion o the central bank balance sheet(able 3.3).18 Te analysis suggests that over thesample period, MP-plus has not appreciably aectedthe protability o banks and may reduce somemeasures o risk in banks over the medium term;but it also suggests that MP-plus may be delayingbalance sheet repair by banks, thereby potentiallyosetting the risk reduction eects. Specically: On risk taking, the analysis shows that MP-plus

policies appear to be achieving their intendedeects, with banks increasing their risky assets inresponse to the prolonged period o low inter-est rates (an indicator o MP-plus shown in thesecond group o rows in able 3.3).19 he lowinterest rates have also tended to decrease leverage(increase equity over total assets), but although itis statistically signiicant, the eect is so small asto be economically insigniicant.

18Te calculation o the eects reported in able 3.3 uses thestatistically signicant estimated coefcients reported in Annex3.2, able 3.7.

19Te rst result is consistent with ndings in previous empiri-cal studies on the precrisis period, which showed a signicantassociation between low interest rates and bank risk taking (DeNicol and others, 2010; Altunbas, Gambacorta, and Marqus-Ibaez, 2010; and DellAriccia, Laeven, and Suarez, 2013).


18/34


19/34



On proitability, low policy rates and the increasein central bank assets have had a negative eecton banks net interest margin, but the eect isagain so small as to be economically insignii-

cant.20 his eect is the result o two opposingeects: low rates reduce unding costs or banks;but over time, revenues rom new loans andixed income securities also decline, osetting thedecline in unding costs.

he benign developments in bank proitabilityare conirmed by the eect o MP-plus on bankz-scores. he z-score is an indicator o soundnessthat combines a banks proitability and capitaliza-tion, and it appears to have increased as a resulto the prolonged period o low interest rates andthe expansion o central bank balance sheets.

Although these developments in proitability andcapitalization show no immediate deterioration inbank soundness, these measures do not relect allcomponents o bank health.

A measure o loan perormance suggests that someaspects o MP-plus may be delaying balance sheetrepair by banks. Increased central bank assets (anindicator o MP-plus shown in the third group orows in able 3.3) tend to reduce loan-loss provi-sions. his may point to the risk that the ampleliquidity provided by central banks is giving banksan incentive to evergreen (roll over) nonperorm-ing loans instead o recording losses in theirproit and loss accounts. An alternative view isthat with MP-plus supporting economic activity,these loans are more viable and hence need ewerprovisions.21 A delay in balance sheet repair couldbe one reason or the market expectations o anincrease in bank deault risk over time that wasound in the event study.

he analysis does not ind evidence that MP-plusaects dierent kinds o banks dierently. heeects o MP-plus do not appear to depend on

bank asset size, or the ratio o equity to total assets,

20Te sign o the eect is in line, however, with other evidencethat has ound a positive relationship between the level o interestrates and net interest margins, as discussed in the next section.

21While the analysis or the United States would support bothexplanations, previous studies have ound evidence or delaysin balance sheet repair in Japan starting in the 1990s (Peek andRosengren, 2003; Caballero, Hoshi, and Kashyap, 2008).

or whether they are global systemically importantbanks.22

Ins r risk in Bnks

Banks are aected by an increase in interest ratesmainly through the interest rate spread between theirlending and borrowing (the net interest margin) andthrough their holdings o securities and derivatives.Indirect eects on loan perormance also play a role.Tese eects can work in opposite directions, and thenet eect o an increase in interest rates can be posi-tive or negative or banks, depending on the maturitystructure o their balance sheets and other actors.

Estimates rom a variety o sources suggestthatother things equalan increase in interestrates would have a positive eect on the net inter-

est income o banks. An analysis in BIS (2012a,Chapter 4) shows a positive relationship between theshort-term interest rate and the net interest margino banks in 14 major advanced economies. Te slopeo the yield curve also has a positive eect. Researchby Federal Reserve economists comes to a similarconclusion or U.S. banks (English, Van den Heu-vel, and Zakrajek, 2012). U.S. banks themselvesestimate that a rise in interest rates would increasetheir net interest income (Figure 3.6).

Interest rate increases can, however, also exposebanks to losses since they reduce the market value oxed-income assets (including government bonds),particularly i rates rise suddenly and unexpectedly.Such losses on government bonds are larger in alow-interest environment (see able 1.4 in Chap-ter 1).23 A hypothetical increase in interest rates rom2 percent to 4 percent would generate losses o 16percent on the market value o a 10-year bond (able3.4). A Value-at-Risk analysis assesses banks expo-sure to interest rate shocks on their trading portolio.For U.S. banks, such an analysis shows a decline in

22

In the regressions, interaction terms between these variablesand the MP-plus variables were generally insignicant. Te regres-sion results including these interaction terms are not reported inable 3.7.

23Bonds held in the available or sale category on a banksbalance sheet would suer mark-to-market losses, but i they arein the held to maturity category, the losses would be unreal-ized and not recognized in the prot and loss statements. Marketparticipants typically see through this accounting convention toestimate such losses.


20/34



capital losses at Italian and Spanish banks is the actthat rates on their domestic sovereign bonds havebeen high recently because o elevated risk premiumson these bonds, and the premiums have recently beendeclining; a continued decline could oset to someextent the eects o a rise in policy interest rates.

Corporate bond holdings could also generatelosses i interest rates rise, especially given the com-pressed yield spreads witnessed recently. However,bank holdings o corporate bonds are relativelylow. In the ourth quarter o 2012, U.S. depositoryinstitutions held only 5.3 percent o their assets in

corporate and oreign bonds (Board o Governors othe Federal Reserve System, 2013). Data rom theECB show that euro area banks hold 4.9 percento assets in bonds issued by nonnancial corpora-tions and other nonbanks (excluding sovereign debt)and only 1 percent o total assets in bonds issuedby nonnancial corporations alone. Banks in theUnited Kingdom hold 4.1 percent and 0.3 percent,respectively, o their assets in securities associatedwith these same categories. Banks in Japan holdbonds in industrial corporations amounting to only1.7 percent o assets. Given these small holdings, the

associated interest rate risk is likely limited.Eects o interest rate increases could also be elt

indirectly through loan perormance. Customersthat have borrowed rom banks at variable ratesmay nd it more difcult to adjust: a sharp rise ininterest rates could thereore raise nonperormingloan rates and the credit risk o banks. Te extent

interest rate risk in their trading books, although thatrisk remains above its precrisis level (Figure 3.6).Banks in Japan have a larger exposure to domestic

sovereign debt than those in any other advanced econ-omy (Figure 3.7; see also IMF (2012a) and the Octo-ber 2012 GFSR). Te BOJ (2012) notes that regionalbanks in Japan in particular are especially vulnerableto the risks o these large holdings: according to theBOJ, a 100-basis-point increase in interest rates acrossthe yield curve would lead to mark-to-market losses o20 percent o ier 1 capital or regional banks and 10percent or the major banks.

Holdings o sovereign debt by banks in Italy andSpain are also relatively high and have risen substan-tially since the beginning o the crisis. Te Bank oItaly (2012) reports that a 200-basis-point increasein interest rates would cost Italian banks 7.7 percento their capital through a combination o increases innet interest earnings and a all in the value o theirgovernment bond holdings. Mitigating the risk o

2.0

1.5

1.0

0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

2004 05 06 07 08 09 10 11 12 2004 05 06 07 08 09 10 11

0

20

40

60

80

100

120

140

160

180

Figure 3.6. Interest Rate Risk as Reported by U.S. Banks

Eect on Banks' Net Interest Income of a Gradual Rise

in Interest Rate of 200 Basis Points

(Weighted average, in percent of baseline

forecast)

Interest Rate ValueatRisk

(Weighted average, in millions of U.S. dollars)

Sources: Bloomberg L.P.; SNL Financial; and IMF staff estimates.

tbl 3.4. cluld Losss on 10-Y Bond s rsul

o ris in Ins rs

Coupon Yield on Bond

2 percent 4 percent 6 percent

Interest Rate Increases by Final Bond Price

1 percent 91 92 932 percent 84 85 87

Source: IMF sta estimates.

Note: Numerical example is based on a 10-year bond. Initial bond price is 100.


21/34



to which banks are aected by these losses alsodepends on the rationale that is moving the centralbank to increase interest rates. For instance, i thecause is related to adverse supply shocks, the eecton banks may be larger than i it is related to animproving economic situation; banks and their bor-

rowers would perorm better in the latter case andthus be in a better position to absorb losses.24

Te potential or capital losses on holdings oxed-rate securities and loans in the short term canbe signicant, even though the net eect o inter-est rate increases would be positive or banks overthe medium term. Te positive eect o higher netinterest income accumulates over time, osettingthe more immediate capital losses incurred predomi-nately by banks with signicant trading operations.25

24Te eect o MP-plus on ination is discussed in Chapter 3

o the April 2013 World Economic Outlook.25Recent stress tests perormed by the Federal Reserve on par-

ticipating bank holding companies (BHCs) in compliance withthe Dodd-Frank Act showed that trading and counterparty creditlosses o the 6 BHCs with signicant trading activities amountedto $97 billion, 21 percent o total losses o all 18 BHCs and 27percent o the total losses o the 6. Te severely adverse scenariocomprised adverse changes to several actors and included anincrease in the 10-year reasury yield o 100 basis points. Tese

Also, the positive eect on the net interest margin isimportant, since interest makes up well over hal obank income (some 80 percent in the United Statesand about two-thirds in the euro area, or example).Indeed, English, Van den Heuvel, and Zakrajek(2012) report that interest rate changes aect bank

protability mainly through the eect on net interestincome. Tis is in line with the nding summarizedin Figure 3.6 that U.S. banks have decreased theirinterest rate risk since the peak o crisis.

conlusions nd poliy Imliions

MP-plus has involved the unprecedented inter-vention o major central banks in various assetmarkets, including sovereign and corporate bondmarkets, markets or asset backed securities, andindirectlymoney and interbank markets. Banks

have been aected by the prolonged period o verylow nominal and real interest rates, by central bankasset purchases (through liquidity and price eects),and by direct liquidity support.

actors were imposed over the course o nine quarters. See able4 in Dodd-Frank Act Stress est 2013" (Federal Reserve Board,2013) or more details.

0

5

10

15

20

25

Norway

Switzerland

UnitedKingdom

Denmark

Finland

Sweden

Australia

1

Korea

Ireland

Netherlands

France

Austria

New

Zealand

Canada

Greece

2

Germany

Portugal

Belgium

Spain

Slovenia

UnitedStates

Italy

CzechRepublic

Japan

2006:Q1

2012:Q3

Figure 3.7. Bank Holdings of Government Debt in Selected Economies(In percent of banking sector assets)

Sources: Central banks' and national r egulators' websites; IMF, International Financial Statistics(IFS); and IMF staff

estimates.Note: Data on quarterly government debt holdings of domestic banks are taken from Arslanalp and Tsuda (2012).

Government debt is defined as general government gross debt on a consolidated basis and includes securities other than

shares, loans and other shortterm debt (not included elsewhere). Bank assets refer to IFS' other depository corporations.1Australia data refer to 2012:Q2.2The value of government debt holdings of Greek banks fell from 12.4 percent in 2011:Q4 to 8 percent in 2012:Q1, as a

result of an official debt restructuring.


22/34



Te analysis nds little evidence that MP-plushas given rise to a serious immediate degradationo nancial stability (able 3.5). Overall, the eectso MP-plus are associated with improved bank

soundness in the short term, a result in line withthe nancial stability objectives o policymakers. Inaddition, in some markets where central banks playa large role (including in interbank markets and insome sovereign bond markets in the euro area), MP-plus has been carried out in response to dysunction;in those cases, central bank actions can be seen aspreventing a worsening o market unctioning.

Over the medium term, however, MP-plus may begenerating risks that have not yet become evident inbanks. Forward-looking indicators may be showingthat the market is alert to these risks, with MP-plus

(and conventional monetary easing) hurting bankstocks in some countries and increasing market per-ceptions o bank deault risk. Te main risks associ-ated with MP-plus over the medium term are that: Balance sheet repair in banks may be delayed. here

is some evidence that unconventional central bankmeasures may be supporting a delay in balancesheet cleanup in some banks, with MP-plus havinga negative eect on loan provisioning. he currentenvironment may also be encouraging banks toevergreen loans rather than recognize them as non-perorming, as noted in Bank o England (2012),with banks providing borrowers with lexibility tomeet their obligations during periods o stress untileconomic conditions improve. But it is diicultto identiy weak but ultimately viable borrowers,and such evergreening may be keeping nonviableirms alive; their demise when rates rise could aectthe quality o the loan portolio over the mediumterm. Indeed, the Bank o England (2012) suspectsthat loan orbearance partly explains the recent lowcorporate insolvency rate in the United Kingdom.

An eventual rise in interest rates may hurt some

banks. Banks in several countries are holding largeamounts o government bonds. A rise in inter-est rates upon exit rom MP-plus could lead toactual losses on banks bond holdings held in theavailable-or-sale category.

Exit rom markets where central banks still hold

substantial amounts o securities may be challeng-

ing. Central banks are holding large amounts

o certain assets, particularly government bondsand securities linked to real estate. Expecta-tions o central bank sales o these large holdingscould lead to market disruptions, especially i

the desired policy stance shits quickly. he rapidrepricing o bonds can result in losses or bondholders (both banks and central banks). hesechallenges highlight the importance o a well-planned and clearly articulated communicationsstrategy or central bank exit rom such markets.

The volume and eiciency o interbank lending may

adjust to new, lower levels based partly on a reevalu-

ation o counterparty risks. With many banks nowrelying to a signiicant extent on central bankliquidity and banks withdrawing resources andskills rom interbank lending activities, it may be

diicult to restart these markets.

As the recovery proceeds and banking system risksbegin to rise, MP-plus measures should be accom-panied by micro- and macroprudential policieswhere needed, supported by robust data provisionby nancial institutions and vigorous risk-basedsupervision.26 Tese risks are slow moving and maybe masked by the near-term benets o crisis-relatedmeasures, making it crucial that they be addressedpromptly with prudential measures. Te precrisisperiod has shown that corrective policies imple-mented ater the risks reveal themselves may be toolate to contain nancial stability challenges.

Policies should be implemented in a measured man-ner, ocused on areas showing rising vulnerabilities.Tereore, authorities should assess where pockets ovulnerabilities exist and quantiy their systemic impor-tance. For this, more robust data encompassing a largershare o the nancial system are key. For example, morecomprehensive bank-level data would allow the aboveassessment o the impact o various MP-plus measuresto be replicated or countries besides the United States.

Tese analyses should help identiy which prudentialmeasures are most suitable to deal with those risks. othe extent that risks are identied in specic nancialinstitutions, these measures would have a micropruden-tial ocus. I the risks are aecting the nancial systemmore broadly (systemic risks), the measures would

26For essential elements o good supervision, see IMF (2010b).


23/34



tbl3.5.risksomM

p-plusndMiigingpoliis

TypeofMP-P

lusPo

Central Bank Policies and Systemic Risks

Documents

Transcript of Central Bank Policies and Systemic Risks