pc_2014_05_

ISSUE 2014/05MAY 2014 ADDRESSING WEAK

INFLATION: THEEUROPEAN CENTRALBANK’S SHOPPING LIST

GRÉGORY CLAEYS, ZSOLT DARVAS, SILVIA MERLER ANDGUNTRAM B. WOLFF

Highlights• Euro-area inflation has been below 1 percent since October 2013, and medium-term

inflation expectations are well below 2 percent. Forecasts of the return to target infla-tion have proved wrong. The European Central Bank should act forcefully, but shouldundermine neither the major relative price adjustments between the euro-area coreand the periphery that are needed, nor the ongoing process of addressing weaknessesin Europe’s banking system.

• Reducing the deposit rate or introducing another long-term refinancing operation could bebeneficial, but would be unlikely to change substantially inflation expectations. Govern-ment bond purchases would be significantly beneficial, but in a monetary union with 18different treasuries, such purchases are difficult for economic, political and legal reasons.

• We recommend a monthly asset-purchase programme of €35 billion with a review ofthe amount after three months. EFSF/ESM/EU/EIB bonds, corporate bonds and asset-backed securities should be purchased, of which at least €490 billion, €900 billionand €330 billion respectively are suitables. Bonds of sound banks could be conside-red after the completion of the ECB’s assessment of bank balance sheets.

• While bond purchases distort incentives and make the ECB subject to private andpublic sector pressure, with potential consequences for inflation, such risks need to beweighed against the risk of persistently low inflation.

Grégory Claeys ([email protected]) is a Research Fellow at Bruegel. Zsolt Darvas([email protected]) is a Senior Fellow at Bruegel. Silvia Merler ([email protected])is a Bruegel Affiliate Fellow. Guntram B. Wolff ([email protected]) is Director of Bruegel.The authors thank Vitor Gaspar, Guonan Ma, Francesco Papadia, André Sapir, Shahin Vallée,Nicolas Véron and participants in a Bruegel seminar for comments and suggestions, and PiaHüttl for excellent research assistance.

Telephone+32 2 227 4210 [email protected]

www.bruegel.org

BRU EGE LPOLICYCONTRIBUTION

ADDRESSING WEAK INFLATION: THEEUROPEAN CENTRAL BANK’S SHOPPING LIST

GRÉGORY CLAEYS, ZSOLT DARVAS, SILVIA MERLER AND GUNTRAM B. WOLFF, MAY 2014

02

BR U EGE LPOLICYCONTRIBUTION

1. https://www.ecb.europa.eu/mopo/strategy/prices-

tab/html/index.en.html.

2. In an interview, ECB Exec-utive Board member Benoit

Coeuré said that the aca-demic definition of

‘medium term’ is 18months, but currently “it is

only normal that we seeinflation coming back moreslowly to the medium-term

objective”:http://www.ecb.europa.eu/press/inter/date/2014/html/

sp140116.en.html.

1 INTRODUCTION

There are clear benefits to price stability. Highinflation can distort corporate investment deci-sions and the consumption behaviour of house-holds. Changes to inflation redistribute real wealthand income between different segments of soci-ety, such as savers and borrowers, or young andold. Price stability is therefore a fundamentalpublic good and it became a fundamental principleof European Economic and Monetary Union. Butthe European Treaties do not define price stability.It was left to the Governing Council of the EuropeanCentral Bank (ECB) to quantify it: “Price stability isdefined as a year-on-year increase in the Har-monised Index of Consumer Prices (HICP) for theeuro area of below 2%”1. The Governing Council hasalso clarified that it aims to maintain inflationbelow, but close to, two percent over the mediumterm, though it has not quantified what ‘close-ness’means, nor has it given a precise definition ofthe ‘medium term’2. The clarification has beenwidely interpreted to mean that the actual target ofthe ECB is close to, but below, two percent infla-tion in the medium term.

In the current European circumstances, low over-all euro-area inflation implies that in some euro-area member states inflation has to be very low oreven negative in order to regain competitivenessrelative to the core. The lower the overall inflationrate, the more periphery inflation rates will haveto fall in order to achieve the same competitive-ness gains. Given that wages are often sticky andrarely decline, significant unemploymentincreases can result from the adjustment process.In addition, lower-than-anticipated inflation under-mines the sustainability of public and private debt

THE EUROPEAN CENTRAL BANK’S SHOPPING LIST Claeys, Darvas, Merler, Wolff

if the debt contracts are long-term nominal con-tracts. For governments, falling inflation ratesoften mean that nominal tax revenues fall, whichmakes the servicing or repayment of debt moredifficult.

Inflation in the euro area has been falling sincelate 2011 and has been below one percent sinceOctober 2013. Core inflation, a measure thatexcludes volatile energy and food price develop-ments, has developed similarly. Five of the 18euro-area member countries (Cyprus, Greece, Por-tugal, Slovakia and Spain) are already in deflation.Even in the countries that are not in a recession,such as Belgium, France and Germany, inflationrates are well below the euro-area target of close tobut below two percent. More worryingly, the ECB’sforecast suggests that inflation will not return toclose to two percent in the medium term.

Given the need to regain competitiveness, lower-than-target inflation in the euro-area periphery canbe expected and is even desirable. However, tofacilitate adjustment and achieve the overall ECBinflation objective, inflation in the euro area’s corecountries needs to stabilise and reach levelsabove two percent. A key question for policymak-ers is therefore why inflation rates are subdued incore countries despite very accommodative mon-etary policy conditions and the gradual revival ofeconomic growth. Policymakers must also con-sider which monetary policies are suitable forincreasing aggregate inflation in the euro area,while ensuring that the inflation differentialbetween the core and periphery remains. Finally,unresolved banking-sector problems are makingthe task of the ECB more difficult.

‘Given the need to regain competitiveness, lower-than-target inflation in the euro-area periphery

can be expected and is even desirable. To facilitate adjustment, inflation in the euro area’s core

countries needs to stabilise and reach levels above two percent.’

03

BR U EGE LPOLICYCONTRIBUTIONClaeys, Darvas, Merler, Wolff THE EUROPEAN CENTRAL BANK’S SHOPPING LIST

Is there a risk that the euro area as a whole willfall into outright deflation?

Inflation expectations have been falling since atleast mid-2012. Figure 2 on the next page pres-ents expectations from two sources (an ECBsurvey and a market-based indicator) and for twomaturities. The two-year-ahead expectations aresignificantly below two percent and even belowone percent according to the market-based indi-cator. In the period relevant for the ECB, inflationexpectations have thus become de-anchored fromtwo percent. Lack of ECB action when the ECB’sown medium-term inflation forecasts fell belowthe two percent threshold was a signal to marketsthat probably resulted in the downward revisionof longer-term inflation expectations. The ECB isnow less effective in anchoring longer-term expec-tations to, or close to, the two percent level.

There are four further reasons suggesting that theECB should already have adopted additional mon-etary stimulus:

1 The cost of deviations from the current inflationbaseline is asymmetric;

2 The track record of inflationary forecasts andexpectations suggests that significant changesin inflation are often unforeseen;

3 The Japanese experience suggests that long-term market expectations can be persistentlyupward-biased;

4 Earlier action can prevent being forced into

It is against this background that we discuss mon-etary policy options to address low inflation in theeuro area. Evidently, structural and banking poli-cies, wage-setting mechanisms and fiscal policiesalso need to play a role in addressing the reces-sion and the low-inflation problem. They are, how-ever, not discussed in this Policy Contribution.

2 HETEROGENEOUS INFLATION DEVELOPMENTSIN THE EURO AREA

Panel A of Figure 1 shows that the euro-area head-line inflation rate has been moving downwardssince late 2011, while Panel B indicates a similartrend for core inflation3. Panels A and B also high-light major differences between euro-area coun-tries. Countries in the euro-area periphery (whichwe define as Cyprus, Greece, Ireland, Italy, Spainand Portugal) had higher inflation rates than othereuro-area counties before the crisis, persistingwell into the crisis period. Only since 2013 hasinflation in the periphery clearly fallen below thatof the euro area as a whole4.

Several countries are already experiencing defla-tion. Cyprus, Greece, Portugal, Slovakia and Spainare in deflation, while the March 2014 inflationrates in the Netherlands (0.1 percent), Ireland (0.3percent), Italy (0.3 percent) and Latvia (0.2 per-cent) are rather close to zero when measured byheadline inflation. But even in Germany andFrance inflation has fallen below one percent.

3. This measure of coreinflation can be a proxy for

underlying price develop-ments. Core inflation was

less volatile than headlineinflation and on average

between 1999 and 2014,was 0.3 percentage pointsper year lower than head-

line inflation, which is a rel-atively small, though

non-negligible, difference.

4. Tax increases are partlyresponsible for the delayedfall in inflation rates during

the crisis: several of theperiphery countries

increased taxes, therebyincreasing inflation.

Euro area 18 Cyprus, Greece, Ireland, Italy, Spain, Portugal Other 12 (11 in Panel B)

-1

0

1

2

3

4

5

99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 140

1

2

3

4

99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Panel A: all items, headline inflation Panel B: core inflation

Figure 1: Inflationary developments in the euro area (% change compared to the same month of theprevious year), January 1999 to March 2014

Source: Bruegel calculation using data from Eurostat’s Harmonised Index of Consumer Prices dataset. Note: core inflation isdefined as the ‘Overall index excluding energy and unprocessed food’. Data for core inflation in Slovenia is not available forthe full period and therefore this country is not included.

Second, the ECB’s inflation forecasts and marketexpectations have been unable to predict signifi-cant deviations from the two percent threshold(Figure 3). When there was a sizeable deviation,ECB forecasts and market expectations both pre-dicted a gradual return to two percent, which hap-pened in some cases (see, for example, theDecember 2011 forecast of the ECB), but most ofthe time did not.

Third, the fact that long-term inflation expectationsin the euro area have so far not deviated too muchfrom two percent should not be taken as a guar-antee that inflation will return to the two percent

THE EUROPEAN CENTRAL BANK’S SHOPPING LIST Claeys, Darvas, Merler, Wolff BR U EGE LPOLICYCONTRIBUTION

04

much larger unconventional policy measureslater, when inflation falls so much that no otheroption remains.

First, at a low level of inflation, the costs of devia-tion from the ECB’s forecast inflation are highlyasymmetric. If inflation is higher than forecast, itwould mean that inflation would be closer to thetwo percent threshold – a benign development.But if inflation is lower than forecast, then coun-tries in the euro-area periphery would have tomaintain even lower inflation or higher deflation,with risks for the sustainability of public and pri-vate debt.

0.5

1

1.5

2

2.5

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

2 years ahead (SPF) Long-term ahead (SPF) 2 years ahead (market based) 10 year ahead (market based)

Figure 2: Inflation expectations: ECB’s Survey of Professional Forecasters and market-basedinflationary expectations in the euro area, 2002Q1-2014Q2

Source: ECB’s Survey of Professional Forecasters (SPF) and Datastream. Note: In the ECB’s survey the horizon of ‘Long term’is not specified. Market-based expectations refer to overnight inflation swaps, which can be used as a market-based proxy forfuture inflation expectations. The 2014Q2 values of market-based expectations are the average during 1-23 April 2014, whilethe latest available values for the SPF are end of March 2014.

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

HICP Dec-08 Dec-09 Dec-10Dec-11 Dec-12 Dec-13 Apr-14

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16

HICP Dec-06 Dec-07 Dec-08 Dec-09Dec-10 Dec-11 Dec-12 Dec-13

Panel A: ECB’s inflation forecasts Panel B: Market-based inflationary expectations

Figure 3: Vintages of inflation forecasts/expectations and actual inflation in the euro area

Source: Datastream, ECB. Note: The HICP is defined as a 12-month average rate of change; in panel A, the ECB Staff projectionsindicate a range referred to as “the projected average annual percentage changes” (see https://www.ecb.europa.eu/mopo/strat-egy/ecana/html/table.en.html). For simplicity, we take the average of the given range. In panel B, market-based expectationsrefer to overnight inflation swaps, which can be used as a proxy for future inflation expectations.

level without additional monetary policy meas-ures. In Japan, long-term inflation expectationsremained about one percent on average between1996 and 2013, though actual inflation wasslightly below zero (-0.1 percent, Figure 4). Theaverage forecast error for the 6-10 year inflationforecasts made in Japan between 1996 and 2003was 1.1 percentage points5.

At the same time, price developments in the euroarea are still significantly different from Japanduring the past two decades. In Japan, about halfof the items in the consumption basket fell in priceduring the period when the average inflation ratewas almost zero (Claeys, Hüttl and Merler, 2014).In the euro area there has been an increase in theshare of items in the HICP basket that are alreadyin deflation in recent months (to about 20 percentof the entire HICP basket), but this share is notvery high (and similar to shares observed in 2005when the inflation rate was close to two percent inthe euro area) and is still significantly lower thanin Japan.

Overall, inflation has been falling significantly andso have inflation expectations. Inflation forecastshave proved consistently too optimistic about thereturn of inflation to the two percent threshold inthe euro area and the one percent target in Japan.The ECB’s own forecast suggests that euro-areainflation will not return to close to two percent inthe medium term, and we see a substantial riskthat it will not return to this level even in the longerterm.

3 HOW TO ADDRESS LOW INFLATION IN AHETEROGENEOUS MONETARY UNION?

3.1 Key constraints

The ECB's task is complicated by two very specialcircumstances. First, the euro area is a heteroge-neous monetary union in which the process of rel-ative price adjustment between its different partsis ongoing. This adjustment is a consequence ofthe very substantial past divergence in prices. Tobetter understand the resulting problem for theECB, it is useful to resort to a simple example of atwo-country monetary union. In the monetaryunion, one region (say periphery) is depressedand runs a zero inflation rate, while the other

Claeys, Darvas, Merler, Wolff THE EUROPEAN CENTRAL BANK’S SHOPPING LISTBR U EGE LPOLICYCONTRIBUTION

05

5. We calculated theforecast error as the

difference between theinflation forecast made in a

certain year, minus theaverage inflation rate from

six to ten years later.Therefore, the most recent

forecast for which we couldcalculate the actual

forecast error was made in2003.

region (say core) has an inflation rate of one per-cent, still below the two percent target, eventhough there is almost full employment. The mon-etary stimulus should result in aggregate inflationin the monetary union increasing to the two-per-cent target. However, since there has to be a rela-tive price adjustment between the periphery andthe core, the monetary stimulus should ensurethat the inflation differential between the tworegions remains in place.

The stimulus must therefore increase inflation andactivity both in the core and the periphery. Thenecessary relative price adjustment implies thatinflation in the core should increase above thetarget, while periphery inflation has to stay belowit. If the stimulus would not have an impact on thecore, but only the periphery, then it would under-mine the necessary price-adjustment process.

The second problem for the ECB is that the processof bank balance-sheet repair is ongoing. Whenseveral banks have vulnerable capital and liquid-ity positions, a monetary stimulus aimed atincreasing bank lending to the private sector isless effective, similar to what happened after thethree-year longer-term refinancing operations(LTRO) in late 2011 and early 2012, when banksincreased lending to governments and accumu-lated reserves at the ECB. In addition, the ECB is

-2.5-2

-1.5-1

-0.50

0.51

1.52

2.5

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

Long-term (6-10 year) Consensus forecastsActual inflation (percent change compared tosame quarter of previous year)

Figure 4: Long-term inflation expectations andactual outcomes in Japan

Source: Consensus Economics (2014) (expectations) and IMF(actual inflation). Note: this figure is reproduced using our datasources from Figure 7 in Antolin-Diaz (2014). There are twoobservations per year, in April and October. For actual inflationwe plot the change in the all-items consumer price index com-pared to the same quarter of previous year in the quarterbefore the forecast was made.

6. http://www.bankofeng-land.co.uk/publications/Pag

es/news/2013/027.aspx.

THE EUROPEAN CENTRAL BANK’S SHOPPING LIST Claeys, Darvas, Merler, Wolff BR U EGE LPOLICYCONTRIBUTION

06

dealing with the quality of banks’ balance sheetsin the context of the asset quality review andstress test. Clearly, fixing the bank-lending chan-nel cannot be done by monetary policy butrequires action on the structural weaknesses ofbanks’ balance sheets. A key question is if possi-ble monetary policy measures (like new long-termliquidity provision to banks, asset purchase pro-grammes from banks and/or from other privatesector asset holders, or negative ECB deposit ratesfor banks) would be conducive to increased infla-tion under current circumstances. An equallyimportant question is whether such a monetarypolicy measure would remove the incentive to fixthe structural problems in the banking systemwhere necessary.

3.2 Policies to address low inflation in thespecial euro-area setting

Different policies could be deployed to increaseinflation and inflationary expectations:

• Reduce the Main Refinancing Operation (MRO)rate to zero percent;

• Negative rates for banks’ deposits at the ECB;• Ending the sterilisation of bond holdings from

the Securities Markets Programme (SMP);• New long-term (eg three years or longer) refi-

nancing operations, possibly made conditionalon net lending to the private sector;

• Asset purchases:– Purchase of euro-area or European debt

(debts of various European rescue fundsand the European Investment Bank);

– Purchase of sovereign debt of euro-areamember states;

– Purchase of non-sovereign debt such as thedebt of non-financial corporations, assetbacked securities (ABS) or debt of financialinstitutions;

– Foreign exchange intervention: purchase offoreign assets, such as non-euro area sov-ereign debt or corporate debt. Given the G7statement of February 2013 by central bankgovernors and finance ministers reaffirmingthe “longstanding commitment to marketdetermined exchange rates and to consultclosely in regard to actions in foreignexchange markets”, this policy measure isnot discussed further6.

Reducing ECB interest rates

The current 0.25 percent ECB main refinancing ratecould be marginally reduced, but the impact ofsuch a small reduction is unlikely to significantlychange inflation expectations. In addition, the ECBcould reduce the deposit rate, which banks receivewhen depositing liquidity at the ECB, from zero cur-rently to negative territory. Since currently bankscan hold excess reserves on their current accountat the ECB at zero interest, a negative deposit rateshould be accompanied by the same negativeinterest rate on excess reserves, to avoid the shift-ing of all deposits to excess reserves (Figure 5shows that banks shifted half of their deposits toexcess reserves when the deposit rate wasreduced to zero). A negative deposit rate wouldmean that banks pay interest for placing a depositat the central bank. This would reduce the incen-tive for banks to hold deposits and excess reservesat the central bank and should therefore promoteother uses by the banks of liquidity, such asgreater lending to the rest of the economy. How-ever, the sum of banks’ deposits and their excessreserves at the ECB is declining fast (Figure 5), andwith the normalisation of money markets they mayreturn to their pre-crisis close-to-zero values. Thisimplies that the direct impact of a negative depositrate, in terms of changing the incentives to holddeposits and excess reserves, would be minimal.

-1

0

1

2

3

4

5

-200

0

200

400

600

800

1000

2007 2008 2009 2010 2011 2012 2013 2014

Deposit facility interest rate (right axis)

Deposits, 30-day moving average (left axis)

Excess reserves, 30-day moving average(left axis)

Depo

sits

and

exc

ess

rese

rves

(€ b

illio

ns)

Deposit rate (%)

Figure 5: The ECB’s interest rate on the depositfacility, banks’ deposits at the ECB’s depositfacility and banks’ excess reserves at the ECB,January 2007 to April 2014

Source: Bruegel calculation based on ECB data. Note: banks’excess reserve is the reserves banks hold at their currentaccount with the ECB minus the minimum reserve requirement.

07


‘In normal times, central banks do not engage in long-term liquidity operations. But, when the

interbank market became dysfunctional during the crisis, several euro-area periphery countries

underwent a sudden stop in external financing. In response, the ECB provided ample liquidity.’

It is difficult to assess the quantitative impact of anegative deposit rate on credit and inflation, butthe example of Denmark does not suggest strongeffects. In July 2012, the Danish central bankreduced its deposit rate for banks to -0.2 percentand kept a negative rate until the 24 April 2014.The main motivation for the negative deposit ratewas to discourage the inflow of capital into Den-mark, because with the intensification of the eurocrisis, investors searched for safe assets. Themost direct effects were the reduction of Danishtreasury-bill yields below zero and a depreciationof the Danish Krona against the euro by about halfa percent from 7.43 to 7.46. This change was quitesizeable for Denmark, where the euro exchangerate is kept very stable. A negative ECB depositrate may lower treasury-bill yields especially ofcore euro-area countries and weaken theexchange rate of the euro, which would increaseinflation.

Some commentators (eg Papadia, 2013) haveargued that banks would in fact increase loaninterest rates in order to compensate for the lossfrom their deposits at the ECB. However, theDanish experience also showed that a negativedeposit rate does not necessarily have any impacton banks’ loan rates to their clients. Another con-cern is the impact of negative deposit rates onmoney-market activity. The ECB’s decision to cutthe deposit rate to zero has already led to the clo-sure of various money-market funds and coulddrain liquidity in the money markets7. In Denmark,however, money-market volumes decreased onlyslightly after the introduction of the negative cen-tral bank deposit rate. Investors exiting money-market funds would need to find otherinvestments, pushing liquidity to markets withcharacteristics similar to money markets.

Stopping the sterilisation of SMP holdings

Another possible measure would be stopping thesterilisation of the ECB’s Securities Market Pro-gramme (SMP) holdings. Under the SMP, the ECBbought about €220 billion of Greek, Irish, Por-

tuguese, Italian and Spanish government bonds.At present, there are €175.5 billion of SMP bondsleft, the maturities of which are not publicly dis-closed by the ECB. The bonds are held to maturityand the purchases are entirely sterilised. Stoppingtheir sterilisation would inject €175.5 billion intoeuro-area money markets.

However, the SMP was launched to address themalfunctioning of securities markets and restorean appropriate monetary policy transmissionmechanism, while not affecting the stance of mon-etary policy8. A key feature of the programme wassterilisation. Falling short of that commitment andchanging the objective at this point would be prob-lematic, because it might undermine trust in theECB’s other commitments. Importantly, the Out-right Monetary Transaction (OMT) programme isalso designed to be sterilised9. In the ongoing judi-cial discussions on the OMT10, stopping the steril-isation of SMP holdings would give a powerfulargument to the plaintiffs, who could say that theECB’s OMT commitments are unreliable.

New long-term (eg three years or longer)refinancing operations

In normal times, central banks do not engage inlong-term liquidity operations. One reason for thisis moral hazard: long-term central bank financingat rates below what banks could get from themarket might encourage excessive risk taking andkeep insolvent banks alive. However, when theinterbank market became dysfunctional duringthe crisis, several countries in the euro-areaperiphery underwent a sudden stop in externalfinancing. To address the problem, the ECB pro-vided ample liquidity. The maturity of the ECB’s liq-uidity operations were then extended from threemonths to six and twelve months. In December2011 and in February 2012 the ECB also con-ducted two extraordinary Longer Term Refinanc-ing Operations (LTROs) with maturities of threeyears, from which banks in the euro area borrowedalmost €1 trillion.

7. http://www.bloomberg.com/news/2012-07-06/jpmor-

gan-shuts-europe-money-market-funds-on-ecb-rate-c

ut.html.

8. http://www.ecb.europa.eu/press/pr/date/2010/html/pr

100510.en.html.

9. http://www.ecb.europa.eu/press/pr/date/2012/html/pr

120906_1.en.html.

10. See for instance in Wolff(2013).

08

BR U EGE LPOLICYCONTRIBUTION THE EUROPEAN CENTRAL BANK’S SHOPPING LIST Claeys, Darvas, Merler, Wolff

The ECB also introduced a policy of ‘full allotment’for all ECB liquidity operations. Under this proce-dure, the control of central bank liquidity is effec-tively moved from the central bank to the bankingsystem, because banks can access all the centralbank liquidity they need at a variable rate (if theyprovide sufficient eligible collateral).

These operations, along with the revised collateralpolicy (expanding and changing assets’ eligibilityrequirements in order to mitigate possible con-straints arising from collateral shortage) allowedliquidity-strained banks to refinance a large por-tion of their balance sheets through central banklending, available at a low interest rate and long-term maturity. In a heavily bank-based system,such as the euro area, these measures wereessential to avoid a financial and economic melt-down11.

However, these operations did little to trigger addi-tional lending to the private sector (even thoughthey may have helped to prevent a collapse ofexisting lending). To a great extent, banks eitherdeposited the cheap central bank funding at theECB for rainy days, or purchased higher yieldinggovernment bonds. Thereby, the LTROs in effectsupported liquidity, ensured stable long-term(three-year) financing of banks, subsidised thebanking system and helped to restore its prof-itability, and temporarily supported distressedgovernment bond markets. Considering the alter-native of a potentially escalating financial crisis,these developments were beneficial. However, theLTROs might have delayed bank restructuring andprolonged the existence of non-viable banks.

For two main reasons, the current situation is verydifferent from the situation when the two three-year LTROs were adopted.

First, one reason for the failure of the 2011-12LTROs to foster lending was the weak balancesheet of the banks and uncertainty about theintegrity of the euro area. With the ECB’s Compre-hensive Assessment, the structural weaknessesof the banking sector are gradually being mended.In addition, speculation about the break-up of theeuro area has become less relevant. Therefore, anew LTRO might be more effective, in particular ifECB financing is made conditional on banks

increasing their net lending to the non-financialprivate sector economy (similar to the Bank ofEngland’s Funding for Lending Scheme; seeDarvas, 2013; Wolff 2013). Such conditions, bydefinition, would exclude the use by banks of ECBliquidity to purchase government bonds. With col-lateralised lending to banks, the ECB exposure tocredit risk is minimal. In addition, the central bankwould not replace the banking system in supply-ing and allocating credit to the non-financial pri-vate sector. A new LTRO could therefore be a goodoption to foster credit growth.

Second, the current situation is different becausethere is no longer a liquidity crisis. In fact, somebanks are repaying their loans from the ECB early(Figure 6), even though they have to replace thatfunding at a higher cost from other sources. Thetake-up of LTRO liquidity might therefore be lim-ited and the programme could be ineffective intriggering lending and inflation.

4 ASSET PURCHASES

For any central bank, asset purchases alwaysinvolve difficult choices about what and how muchto buy. The central bank becomes an importantbuyer in financial markets and therefore can besubject to pressure from politicians and compa-

0

200

400

600

800

1000

1200

Mar

-04

Sep-

04M

ar-0

5Se

p-05

Mar

-06

Sep-

06M

ar-0

7Se

p-07

Mar

-08

Sep-

08M

ar-0

9Se

p-09

Mar

-10

Sep-

10M

ar-1

1Se

p-11

Mar

-12

Sep-

12M

ar-1

3Se

p-13

Mar

-14

Spain Greece Ireland

Italy Portugal Belgium

Germany Finland France

Cyprus Luxembourg Slovenia and Malta

Figure 6: Use of Eurosystem liquidity(€ billions), January 2003 to February 2014

Source/note: the ECB does not provide a country breakdownof the use of its facilities. Data come from National CentralBanks but the reporting standards differ. Therefore the lengthof the time series is not the same for all countries and forsome countries data does not seem to be publicly available.

11. Another crucial ECBmeasure during the crisiswas Emergency Liquidity

Assistance (ELA), an emer-gency liquidity line pro-

vided by national centralbanks (with the consent ofthe ECB’s Governing Coun-

cil) to solvent banks thatexceptionally and tem-

porarily do not have enough(or sufficiently high quality

collateral) to access normalEurosystem operations.

09


12. The expression crediteasing is also used when

non-government securitiesare purchased.

13. The potential impact ofQE on wealth inequality has

recently been in the spot-light, see Bank of England

(2012).

nies. This is a powerful argument for central banksto act early in order to avoid a low-inflation trap, inwhich standard monetary policy measuresbecome less effective. The longer an asset-pur-chase programme is delayed in a situation inwhich inflation is already very low, the greater therisk that an even larger purchase programme willultimately be needed.

In response to the global financial and economiccrisis, the Federal Reserve, Bank of England andBank of Japan engaged in large-scale asset pur-chase programmes, or quantitative easing (QE)12.From the beginning of 2009 to March 2014, theFederal Reserve purchased $1.9 trillion (11.9 per-cent of US GDP) of US long-term Treasury bondsand $1.6 trillion (9.6 percent of US GDP) of mort-gage-backed securities. Between January 2009and November 2012, the Bank of England pur-chased £375 billion (24 percent of GDP) of mostlymedium- and long-term government bonds. TheBank of Japan started a new round of asset pur-chases in March 2013 and plans to buy per year50 trillion yen of government bonds (10.4 percentof 2013 GDP), 1 trillion yen of exchange-tradedfunds (0.2 percent of GDP) and 50 billion yen ofJapanese real estate investment trusts (0.01 per-cent of GDP), in order to double the country’s mon-etary base in two years. In addition to such asset

purchases, these central banks also implementedprogrammes to support liquidity in various mar-kets. The ECB has made few asset purchases sofar but reacted to the crisis by providing liquidity tothe banking system. The size of the balancesheets of the central banks therefore increasedfor different reasons (Figure 7).

Asset purchases can be used if interest ratesreach the zero lower bound and refinancing oper-ations are ineffective, as discussed above. Thereare a number of channels through which assetpurchases can influence monetary conditions andthereby economic activity and prices:

• Money multiplier: if the money multiplier (theratio of broad monetary aggregates to the mon-etary base) is stable, then the asset-purchase-induced increase in the monetary base willincrease monetary aggregates, through morecredit to non-financial corporations and house-holds, which can boost demand.

• Altering yields: purchase by the central bank ofa particular asset will reduce the net supply ofthat asset to the private sector and increase itsprice and thereby reduce the return that ityields.

• Portfolio rebalancing: Unless the purchasedasset is a very close substitute for cash (suchas short-term treasury bills), investors whosold the asset will search for other investmentopportunities, pushing up prices and reducingyields in other markets too.

• Exchange rate: via portfolio rebalancing, previ-ous asset holders could invest in assetsdenominated in other currencies and therebydepreciate the home currency. This in turnmight increase import prices and thereby infla-tion, but could also boost export production andthereby economic activity.

• Wealth effect: the increase in asset prices canlead to a wealth effect for the asset holders,which can also increase consumption or invest-ment13.

• Signalling: asset purchases by the central bankwhen the zero lower bound on interest rates is

0

5

10

15

20

25

30

35

40

45

50

2006 2007 2008 2009 2010 2011 2012 2013 2014

Bank of JapanEuropean Central BankBank of EnglandFederal Reserve

Figure 7: Size of balance sheets of variouscentral banks (in % of GDP)

Source: FRED, IMF.

‘There is a powerful argument for central banks to act early in order to avoid a low-inflation trap.

The longer an asset-purchase programme is delayed when inflation is already very low, the

greater the risk that an even larger purchase programme will ultimately be needed.’

10


reached could signal to market participantsthat the central bank is serious about furthereasing monetary conditions. This can have animpact on inflation expectations and theexpected future path of policy rates, whichwould lead to a reduction in real interest ratestoday.

4.2 How would these channels work in the euroarea, and in particular, in core and peripherycountries?

The experience of the past few years in countriesthat have implemented asset purchases is thatthe money multiplier is unstable and fell signifi-cantly in parallel to the expansion of the monetarybase. Figure 8 shows that the in the US and the UK,M3 kept growing at about the same rate evenwhen the monetary base doubled, thus halvingthe money multiplier. Most likely the money mul-tiplier would be similarly unstable in the euro areaafter asset purchases, so the money multiplierchannel would not be effective.

The ECB’s balance sheet increased by 112 percentbetween September 2008 and June 2012, andhas decreased by 30 percent since then primarilybecause of the repayment by banks of the LTROs.The decline in the balance sheet as such is not anindication of tighter monetary conditions, butrather reflects the fact that liquidity conditions inthe inter-bank market have normalised.

The lowering of nominal yields is unlikely to be apowerful channel in the core, while it might have a

somewhat greater impact in the periphery. In coreeuro-area countries, both government bond yieldsand private-sector borrowing costs are currentlyvery low. The yield on the 10-year German bund isabout 1.5 percent per year, but even in Italy andSpain 10-year yields are about 3.1 percent, closeto yields of the US government. In terms of corpo-rate lending rates, nominal private sector borrow-ing rates are lower in the euro-area core than in theUK and just slightly higher than in the US – thesetwo countries that have already implementedlarge-scale asset purchases (Panel A of Figure 9).Since inflation is also expected to be lower in theeuro-area core than in the US, real lending ratesare slightly higher in the euro area than in the US,but still well below their pre-crisis values. There-fore, lowering real yields by shifting inflationexpectations could be somewhat more effectivein the core, but the decline in real rates is likely tobe limited.

For the periphery, nominal lending rates to non-financial corporations are higher than in the core,and because of even lower inflation expectationsthan in the core, real interest rates are significantlyhigher. To what extent the yield differentialbetween the lending rates in core and peripherycountries reflects financial fragmentation andgreater credit risk in the periphery remains anopen question. At the height of the euro crisis inthe summer of 2012, both factors likely playedmajor roles. Since then, fragmentation within theeuro area has eased. To the extent that financialfragmentation continues to play a significant role,ECB measures to limit fragmentation, such as

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000

11,000

12,000

1

2

3

4

5

6

7

8

9

10

11

12

13

90 95 00 05 100

400

800

1,200

1,600

2,000

2,400

2,800

3,200

3,600

4,000

4,400

4

6

8

10

12

14

16

18

20

22

24

26

2006 2008 2010 2012 20140

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000

5

6

7

8

9

10

11

12

13

14

15

2000 2005 2010

M3 (left axis) Monetary base (left axis) Multiplier (right axis)

United States United Kingdom Euro area

Figure 8: Monetary aggregates and money multipliers

Source: FRED, ECB, BOE.

11


asset purchases impacting either directly or indi-rectly the borrowing costs of non-financial corpo-rations, are justified. At the same time, assetpurchases can have an indirect impact on creditrisk via improved economic conditions.

A major question is the importance of changes toreal interest rates for the ongoing relative priceadjustment between the euro-area core andperiphery. Taylor (1999) estimated the semi-elas-ticity of consumption and investment with respectto the real interest rate in G7 countries. He foundthat interest sensitivity was significantly higher inFrance and Germany than in Italy. Therefore, thesame decline in real interest rates (either becauseof an increase in inflationary expectations or alower nominal yield) would be more expansionaryin core countries than in Italy (and probably inother periphery countries too). As we have argued,the scope for a decline in real rates is less in thecore than in the periphery. Therefore, cutting realrates to the private sector could be broadly neu-tral for the ongoing relative price adjustmentwithin the euro area, because in the core, limitedscope for reduction is accompanied by large inter-est rate sensitivity, while in the periphery, greaterscope for reduction is accompanied by small inter-est rate sensitivity.

Portfolio rebalancing would probably work both inthe core and the periphery. For example, investorsholding long-term German government bondsprobably have a preference for safe long-termassets. If the net supply of such assets to the pri-vate sector declines, previous asset owners wouldmost likely search for other fixed-income instru-ments with similar characteristics, such as bondsof major banks or non-financial corporations head-quartered in Germany or other core countries.Such a rebalancing would favour the financing ofthese corporations, which might have an impacton their investment decisions, in particular forcompanies that finance investment throughcredit.

A weaker euro exchange rate could directly helpto lift inflation because of its impact on importprices. The exchange-rate effect through exportswould likely favour both the core and periphery,but would have different impacts. Since core coun-tries with large trade surpluses have bigger trad-able sectors, their export performances wouldlikely be boosted more than the exports of periph-ery countries. Since labour markets are tighter incore countries, an export expansion would morelikely translate into wage increases, while this isless likely to happen in the periphery because of

Euro-area core Euro-area periphery United Kingdom United States

0

1

2

3

4

5

6

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 20140

1

2

3

4

5

6

7

8

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Panel A: nominal rates Panel B: real rates

Figure 9: Lending rates to non-financial corporations in the euro area, UK and US

Source: Bruegel based on ECB (lending rates in EU countries), IMF (inflation forecasts) and St. Louis FRED (US nominal inter-est rates). Note: Two data points per year are shown, one corresponding to the spring publication of the IMF’s World EconomicOutlook (WEO, published typically in April) and the other corresponds to the autumn WEO (published typically in October).The nominal interest rate is the average over six months before the publication of the WEO. The real lending rates were calcu-lated using a 2-year ahead inflation forecast from the WEO databases up until October 2007 (due to lack of forecasts for longerhorizons), while a 5-year ahead forecast average was used starting from April 2008. Inflation forecasts were relatively stablein pre-2008 WEOs but showed larger variations after 2008 and therefore our choice for considering different time horizons forinflation forecast before 2008 and from 2008 may not distort much the comparability of the two periods. The latest data forthe US is Q1 2014, and Feb 2014 for the Euro area and the UK. The Euro area core and periphery is calculated as a GDP weightedaverage with fixed weights.

12


high unemployment rates. A weaker euroexchange rate could thus help maintain the infla-tion differential between the core and periphery.However, we also acknowledge that a weakerexchange rate would mean that the euro area’scurrent account surplus would increase further. Itwould mean a sort of beggar-thy-neighbour policy,yet the mandate of the ECB is to maintain pricestability in the euro area and not to safeguardglobal imbalances, which have many othercauses.

Finally, asset purchases could also have a ‘sig-nalling’ impact on financing conditions in the euroarea. Buying assets would show the determina-tion to act, which would affect inflation expecta-tions and the anticipated path of policy rates. Howthis would work in different countries is uncertain.

4.2 Size of the asset purchase programme

Working out the appropriate size of asset pur-chases is far from easy. Some analysis consid-ered the total amount of asset purchases by theBank of England and the Fed and suggested simi-lar magnitudes for the euro area (20 to 25 percentof GDP, ie €1.9 trillion to €2.4 trillion). In our view,that is an inadequate benchmark, because a largeshare of asset purchases by the Fed and Bank ofEngland were crisis-response measures, and theassets were accumulated over five years. The ECBdealt with the crisis in a different way (using liq-uidity operations) and the situation in the euroarea is very different now.

A more relevant benchmark could be the amountpurchased by the Federal Reserve in its thirdround of quantitative easing (QE3). This round ofQE was announced in light of the weak economicsituation of the US economy at a time when theacute phase of the financial crisis was over – a sit-uation that is similar to the current euro-area situ-ation. In September 2012, the Fed announced itwould purchase $40 billion (€29 billion) of agencymortgage-backed securities per month, increasedto $85 billion (€61 billion) in December 2012 (byadding $45 billion per month of Treasuries). Giventhat the euro area’s economy is about 30 percentsmaller than the US economy, the same size, as ashare of GDP, would be between €20 and €40bil-lion per month in the euro area.

The ideal way to select the size of asset purchasesin the euro area would be through assessing itsexpected impact on inflation. However, it is ratherdifficult to measure this impact even in the US andthe UK, where large-scale asset purchases havebeen conducted, and it even more difficult toassess in the euro area.

Joyce et al (2012), Gagnon et al (2011) and Meier(2009) argued that asset purchase programmeshave had a strong direct effect by reducing long-term government bond yields by about 50-100basis points in the UK and US. Hancock and Pass-more (2011) and Krishnamurthy and Vissing-Jor-gensen (2013) reported similar findings for theFed's mortgage-backed security (MBS) purchaseprogramme in the US.

Conclusions on the impact on GDP and inflationdiffer in magnitude, though all research papersreport positive impacts. For the US for instance,Chung et al (2012) estimated that the combina-tion of QE1 and QE2 raised the level of real GDP bythree percent and inflation by one percent (animpact equivalent to a cut in the federal funds rateof around 300 basis points). Chen et al (2012)found that QE2 increased GDP growth by 0.4 per-cent, but had a minimal impact on inflation (equiv-alent to an effect of a 50-basis point cut in thefederal funds rate). In a recent paper Weale andWieladek (2014) estimated that asset purchasesequivalent to one percent of GDP led, respectivelyin the US and the UK, to a 0.36 and 0.18 percent-age-point increase in real GDP and to a 0.38 and0.3 percentage-point increase in CPI after five toeight quarters.

The share of capital markets is smaller in the euroarea than in the US and the UK, the health of euro-area banks has not been restored and nominalinterest rates are rather low in core euro-areacountries. It is therefore difficult to estimate theimpact of asset purchases on inflation in the euroarea, but most likely the effects are different fromthose in the US and UK. Assuming a 0.20 percent-age point inflation effect of a one percent of GDPasset purchase in the euro area, a yearly assetpurchase of about four percent of euro area GDP(about €400 billion) would lead to an inflationincrease of approximately 0.8 percentage pointsafter 18 months. Since core inflation in the euro

13


area is about 1.0 percent now, such an increasewould move it close to two percent. Since we donot know the exact size of the impact, we proposeto commence with €35 billion of asset purchasesper month, which would be close (as a share ofGDP) to the $85 billion per month purchase of theFed under QE3. Starting with a much lower volumecould be seen as too timid to have a substantialeffect. Due to the uncertainty in the transmission,we propose that the size of the purchases shouldbe reviewed after three months. The relevant cri-teria for the review should be the impact on actual(headline and core) inflation as well as on infla-tion expectations. Tapering should start only onceinflation and inflation expectations haveincreased substantially.

4.3 Design principles for an asset-purchaseprogramme

How could an ECB asset-purchase programme bedesigned and what would the purchase of differ-ent assets mean for monetary conditions in theperiphery and the core, given the potentially dif-ferent quality of bank balance sheets? What arethe limits of the different instruments? In our view,the ECB will have to choose which assets to buyusing five main criteria.

• First, the ECB should buy assets that will bemost effective in terms of influencing inflation,through the channels we have described.

• Second, there should be sufficient volumes ofassets available, to ensure that the ECB canpurchase enough while not buying up wholemarkets.

• Third, the ECB should try to minimise the impacton the private-sector financing process. WhileQE by definition changes relative prices, theECB should avoid buying in small markets inwhich its purchases would distort market pric-ing too much. The more the ECB becomes aplayer in a market, the more it can be subjectto political and private sector pressures whenit wants to reverse the purchases.

• Fourth, the ECB should buy only on the sec-ondary markets in order to allow the portfolio-rebalancing channel to work effectively.Purchasing on the primary market would implythe direct financing of entities, which should beavoided.

• Fifth, the assets should only originate from theeuro area and be denominated in euros,because of the 2013 G7 agreement noted insection 3.2.

In principle, the ECB could decide to buy anyasset, except government securities on the pri-mary market, which is clearly ruled out by theTreaty. In practice, the ECB’s task will be more com-plex than it has been for the Fed, the Bank of Eng-land and the Bank of Japan given the peculiaritiesof the euro area:

a Bank lending is much more important than inthe UK and the US;

b There is no euro-area wide sovereign asset(beyond the limited amounts of securitiesissued by the European Financial StabilityFacility (EFSF) and European Stability Mecha-nism (ESM) and the EU-wide bonds of the Euro-pean Investment Bank and EuropeanCommission); instead, each member stateissues sovereign debt and there are major dif-ferences in public debt levels (and thereby intheir perceived sustainability) in differentmember states;

c The outstanding stock of privately-issued debtsecurities is smaller (relative to GDP) than inthe US and the UK, and the roles of privately-issued debt securities vary widely in differenteuro-area member states.

4.4 Should there be a credit rating requirementfor the assets to be purchased?

An important question is to what degree the ECBshould care about risk. A number of points needto be considered:

• The risks in purchasing asset are fundamen-tally different from the risks inherent in collat-eralised central-bank lending. In the latter case,the risk to the ECB is well contained by the highhaircuts applied and the fact that the bank isthe counterparty, which remains liable forrepayment even if there is a default on the col-lateral. In the case of a purchase, the haircut tothe face value is determined by the marketsand the risk is taken directly onto the ECB’s bal-ance sheet.

• The ECB would take on board significant risk via

14


asset purchases under three circumstances:(a) systemic risk, ie risk when all asset classesare highly correlated, (b) when the purchasedportfolio is not diversified enough and concen-trated on a few assets in large volumes, and (c)when market prices are distorted and thereforedo not reflect well the riskiness of assets. – Systemic risk: it is the role of a central bank

to address systemic risk and the ECB wouldin any case be heavily exposed to it also vianormal central bank operations.

– Diversification: given the quantities thatwould be bought under an asset purchaseprogramme and assuming a proper diversi-fication of risk, the ECB could make a profiton its portfolio if it buys assets at non-dis-torted prices. Buying many high-risk assetsis therefore not problematic as such,because high returns would on averagecompensate for defaulting assets.

– Distorted market pricing: market prices canbe distorted because of market failures (egthe pricing of US subprime securities beforethe crisis) or because of central bank inter-vention in markets. To reduce the magnitudeof the latter, it is imperative that the ECBdoes not buy up whole markets, but limitsits purchase in each market to a small share.The less the ECB buys in any given market,the less risk it will take on board because themarket distortion would be kept to a mini-mum. If market-pricing mechanisms are fun-damentally wrong or if the ECB’s purchases(including the anticipation of such pur-chases) materially changes the market pric-ing of the asset, only then would the ECB risksignificant losses.

• The Treaty gives a mandate to the ECB to main-tain price stability, not to protect its balancesheet.

Given these considerations, we recommend areasonably low threshold for credit risk, butsuggest that some criteria on riskiness should beadopted, because the ECB should not turn itselfinto a high-risk investment fund. Restricting assetpurchases only to the eligible collateral (withoutany additional eligibility criterion) is anappropriate threshold and therefore this is ourrecommendation14. The pool of eligible collateralhas also the great advantage that the ECB already

has a well-defined list of eligible assets andtherefore the use of this list would limit lobbyingactivities for what the ECB should buy. It isimportant to highlight that our suggestion differsfrom the ECB’s revealed preference, becauseduring the 2009-12 Covered Bond PurchaseProgrammes (CBPPs), the only previous examplesof ECB unsterilised asset purchases, the criteriafor purchases was the eligibility of the assets ascollateral for refinancing operations with the ECBand a minimum rating of AA or equivalent, awardedby at least one of the major rating agencies15.

4.5 The pool of eligible assets

According to the ECB, total marketable assetseligible as collateral represented almost €14trillion at the end of 2013, equivalent to 146percent of euro-area GDP16. Figure 10 shows thatabout half of the Eurosystem’s eligible collateralpool at the end of 2013 comprised governmentbonds, with €6.37 trillion of central governmentsecurities and €0.42 trillion of regionalgovernment securities. The other half was splitbetween uncovered bank bonds (€2.28 trillion),covered bank bonds (€1.53 trillion), corporatebonds (€1.46 trillion), asset-backed securities(€0.76 trillion) and other marketable assets(€1.17 trillion). Other marketable assets includeEuropean debt (the debts of EU rescue funds andthe European Investment Bank). On top of thosemarketable assets, the ECB also accepts ascollateral non-marketable assets, mostly creditclaims17. Being non-marketable, such assetscannot be within the scope of an asset-purchaseprogramme, unless they are securitised.

Part of the eligible collateral has already beenpledged in the context of the ECB’s refinancingoperations and is therefore not available for pur-chase for the moment (Figure 10). With the repay-ment of the three-year LTRO, at the latest on itsterminal date of February 2015, a considerablepart of the currently-used collateral pool will befreed. It is difficult to compare the size of the totaleligible assets to the pool of assets already usedas collateral, because the former is available innominal terms whereas the latter is only availablein net terms, ie taking into account the haircutapplied by the ECB.

14. The minimum grade formarketable assets to be eli-

gible as collateral for mainECB operations is BBB- butABS under standard frame-

work require AAA/Aaa ratingat issuance and single A-

rating during the life of thesecurity:

http://www.ecb.europa.eu/pub/pdf/other/collateral-

frameworksen.pdf.

15. http://www.ecb.europa.eu/ecb/legal/pdf/l_17520090704en00180019.pdf??d74bb43a6071db357e77cc243

9415ebe.

16. In the permanent collat-eral framework, only euro-

denominated securities areaccepted, but under the

temporary collateral frame-work introduced during thecrisis, assets denominatedin USD, JPY ad GBP are also

accepted. For further detailssee:

http://www.ecb.europa.eu/pub/pdf/other/collateral-

frameworksen.pdf.

17. http://www.ecb.europa.eu/pub/pdf/scpops/ecbocp14

8.pdf.

15


4.6 What to buy? A shopping list for the ECB

European debt

A natural starting point for an ECB asset-purchaseprogramme would be euro-area wide governmentbonds, which do not exist. The closest existingasset, which could be bought without creating toomany distortions, would be bonds issued by theEFSF and the ESM. The bonds of the EuropeanUnion (issued by the European Commission) andthe European Investment Bank (EIB) representEU-wide supranational assets, but since the ECBis an EU institution, it could also consider EUassets. The total available euro-denominated poolof these bonds is around €490 billion (€230 bil-lion for EFSF/ESM, €60 billion for EU, €200 billionfor EIB).

Buying such pan-European assets would notaffect the relative yields of euro-area sovereigndebts and would not distort the market-allocationprocess within the private sector, which would beadvantages. While the transmission channelthrough lower yields may be weak, other channels(portfolio rebalancing, exchange rate, wealth andsignalling) would probably work well. We thereforerecommend that the ECB buys from this pool ofassets.

6371.6

428.5

2281.1

1531.6

1456

761.1

1167.3

331.3

95.7

282.2

402.4

119.9

324.8

117.5

536

0 1000 2000 3000 4000 5000 6000 7000

Central govmt. securities

Regional govmt. securities

Uncovered bank bonds

Covered bank bonds

Corporate bonds

Asset-backed securities

Other marketable assets

Non-marketable*

Used as ECB collateral (value after haircuts and valuation)

Eligible assets (nominal values)

Figure 10: Eligible assets and assets used asECB collateral (€ billions)

Source: ECB, http://www.ecb.europa.eu/paym/pdf/collat-eral/collateral_data.pdf?ba3bb0e0c2611c6740a278aa2ee7818a. Note: Eligible assets are in nominal values; assets usedas ECB collateral are after haircuts and valuation issues.Latest data available: 2013 Q4.

Government bonds

National sovereign debt offers the largest pool forECB purchases. The portfolio rebalancing effectwould work well, as would the exchange rate,wealth and signalling channels. The purchaseswould not distort the market allocation processwithin the private sector. In principle, the case fora government bond purchase programme is there-fore strong. However, the purchase of national gov-ernment debt would be more complicated for theECB as a supranational institution without a supra-national euro-area treasury as counterparty, thanit was for the Fed or the Bank of England. Severalrelevant issues should therefore be discussedcarefully to decide if government debt should bepurchased by the ECB.

The first issue is practical. Since there are 18 dif-ferent sovereign debt markets, the ECB wouldhave to decide which sovereign debt to buy. A pro-posal often made is to purchase government debtbased on the share of each national central bankin ECB capital (which reflects the size of the coun-tries in terms of GDP and population). However, tothe extent that debt-to-GDP ratios are different andthe demand for sovereign debt is different in dif-ferent countries, the ECB purchase would alter thespreads between countries and change the rela-tive price of sovereign debt. Even though probablyall ECB measures have different implications fordifferent euro-area members (eg the SMP directlybenefitted only five governments, the three-yearLTROs were primary used by euro-area peripherybanks), influencing relative yields may exposethe ECB to political pressure by individual coun-tries to increase or decrease the speed of pur-chases or change its portfolio. It could also lead tomoral hazard as market pressure for reformswould be altered.

Second, the Treaty on the Functioning of the Euro-pean Union prohibits extension of any kind of ECBcredit facility to public bodies, or the purchase ofgovernment securities on the primary markets bythe ECB (the same applies to national centralbanks). This treaty provision was agreed in orderto avoid the monetary financing of governmentdebt that could result in cross-border transfersbetween taxpayers. Therefore, a purchase of gov-ernment debt is allowed in the secondary bond

16


markets only if it is done for monetary policy pur-poses and without the risk that it would lead to thefinancing of government debt. Since the goal ofasset purchase will be to meet the ECB’s primaryobjective of price stability, purchase of govern-ment bonds would be allowed if the risk of mone-tary financing could be excluded.

Third, the ECB has a well-defined sovereign bondpurchase programme, the OMT programme, whichwe support (Darvas, 2012; Wolff, 2013). The basicidea of the OMT programme is to give the ECB a toolto buy government bonds in order to improvemonetary policy transmission in countries underfinancial assistance. It is debatable whether a QEprogramme based on ECB capital keys wouldundermine the logic of the OMT programme. How-ever, we note that a purchase based on capitalkeys would lead to small purchases relative towhat an OMT programme would require. For exam-ple, buying €17.5 billion of EU sovereign debt(one-half of our proposed €35 billion purchases)based on capital keys, would imply that the ECBbuys €3.1 billion Italian debt per month. Condi-tionality as required by the OMT programme maytherefore be less relevant in a QE programme. If acountry was under an OMT programme, its bondscould be excluded from the broader QE pro-gramme. On the other hand, buying governmentdebt of countries with uncertain debt dynamicswithout the political OMT agreement could exposethe ECB to greater political pressures when itwants to exit.

Fourth, experience shows that an ECB govern-ment-bond purchase programme would be politi-cally controversial. So far, the ECB has had twogovernment-bond purchasing programmes (SMPand OMT). Both were introduced under severestress and both were motivated by the goal ofrestoring the monetary transmission mechanism.Both programmes were and are highly controver-sial, not least because of different assessmentsof to what extent they constitute monetary financ-ing of government debt.

Overall, government-bond purchases would be anatural step because the bond market is very largeand the positive effects of such a QE would be sig-nificant. However, in a monetary union with 18 dif-ferent treasuries, such purchases are difficult for

the economic, political and legal reasons we haveoutlined. Purchases of private sector assets – ifwell designed – would achieve similarly benefi-cial effects on euro-area inflation and would pro-tect the ECB better from political pressure. Wetherefore do not recommend the purchase of gov-ernment bonds at this stage.

Bank bonds

The second largest asset class is bank bonds, with€3.8 trillion available in eligible covered anduncovered bonds. Purchasing bank bonds couldhave an effect through all the major channels wediscussed: portfolio-rebalancing, lowering yields,exchange rate, wealth and signalling. In particular,the previous holders of those bonds would haveto find other assets to buy, while the reduction inmarket yields would also reduce the yields onnewly issued bank bonds, thereby allowing banksto obtain non-ECB financing at a lower cost. Thiswould improve bank profitability and may improvethe willingness of banks to lend. However, bankbonds should be excluded from the ECB asset-pur-chase programme until the ECB’s ComprehensiveAssessment is concluded. Until then, any ECB pur-chases would lead to serious conflicts of interestat the ECB and would make a proper assessmentby the ECB more difficult. Moreover, those banksfor which the outcome of the Assessment will beunsatisfactory should continue to be excludedfrom the ECB’s asset purchases until they haveimplemented all the required changes in their bal-ance sheets. This might take several months afterthe completion of the Comprehensive Assess-ment.

Corporate bonds

Eligible corporate bonds comprise the third largestasset class with almost €1.5 trillion outstanding.However, this amount also includes non-euro areacorporate bonds and euro-area bonds issued inother currencies. European corporate bonds arebehaving well in terms of default: Moody’s defaultreport for February 2014 – which included a pre-diction of the forward trend for defaulters – showsthat the baseline expectation is on a downwardpath towards levels rarely seen since 2008, forEurope and globally, and that the pessimistic fore-cast is also less severe than it was in 2013.

17


0

1000

2000

3000

4000

5000

6000

200

400

600

800

1000

1200

99 00 01 02 03 04 05 06 07 08 09 10 11 12 13

Debt securities issued by NFCs (left scale)

MFI loans to NFCs (right scale)

€ billions

Figure 11: Outstanding stock of debt securitiesand loans of non-financial corporations (€ billions)

Source: ECB.

While there is no precise data on their magnitude,we estimate that the lower bound of eligible euro-area corporate bonds would be €900 billion. Inaddition, the supply of corporate bonds in the euroarea has grown considerably since 2009 (Figure11).

Figure 12 shows the heterogeneity of corporatebond markets in the euro area. The euro-areacorporate bond market is highly concentrated. Themain issuers of corporate bonds are Frenchcompanies, whose bonds make up 44 percent ofthe total outstanding (ie €466 billion). Germanand Italian corporate bonds follow at significantdistance with each about 12 percent (or around€126 billion) of the outstanding corporate bonds.The Netherlands comes fourth with 9.6 percent ofthe outstanding (€107 billion). But thanks to theportfolio rebalancing effect, the origin of thecorporate bonds is of less importance. Thebeneficial effect would come from the fact that thecurrent owners of the corporate bonds would selltheir bonds and use the cash for differentpurposes throughout the euro area. The origin ofthe bond says little about the owners of bonds,which are in some cases US funds. In any case,the ECB should not choose its purchasesaccording to geographic origin, similar to the wayit implemented the LTRO/MRO, which led to largeamounts of liquidity going to some countries only.In addition, the purchases would encourage newissuance of corporate bonds everywhere and leadto a diversification of the sources of funding (Sapirand Wolff 2013). Lower funding costs for

corporations should induce more corporateinvestment.

Asset-backed securities

Another class of assets that could be bought bythe ECB is asset-backed securities (ABS). YearlyEU securitisation issuance – which peaked in2008 – is much lower than in the US and has beendecreasing since 2008 (Figure 13).

The total outstanding stock of securitised productshas been stagnating at around €1.06 trillion for theeuro area compared to €2.5 trillion in the US

0

100

200

300

400

500

600

700

800

900

1000

Fran

ce

Germ

any

Italy

Neth

erla

nds

Aust

ria

Belg

ium

Portu

gal

Finl

and

Spai

n

Luxe

mbo

urg

Slov

akia

Gree

ce

Cypr

us

Irela

nd

Esto

nia

Slov

enia

Outstanding amounts of euro-denominated debtsecurities issued by non-financial corporations

Outstanding bank loans to non-financial corporations

Figure 12: Bonds vs. loans – financing of EU non-financial corporations (€billions)

Source: ECB. Note: The difference between the amount reported in this figure and the total eligi-ble corporate bonds shown on Figure 10 comes from the fact that here we only consider corpo-rate bonds issued by euro zone corporations, whereas eligible collateral include corporate bondsissued in the whole European Economic Area (EU countries and Iceland, Liechtenstein andNorway); see https://www.ecb.europa.eu/paym/coll/standards/marketable/html/index.en.html.

0

500

1000

1500

2000

2500

3000

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

European historical issuanceUS historical issuance

Figure 13: European and US historical issuance of securitised products (€billions)

Source: AFME – SIFMA.

18


(AFME, 2014). Products eligible as collateral forthe ECB amount to about €761 billion, but someof them originate from outside the euro area. Weestimate that the lower bound of eligible euro areaABS would be €330 billion.

It is worth highlighting that defaults on ABS inEurope have ranged between 0.6-1.5 percent onaverage, against 9.3-18.4 percent for US securiti-sations since the start of the 2007-08 financialcrisis18. The regulatory landscape for securitisedproducts has also changed considerably since thecrisis and made the products safer and moretransparent19.

Considering the total amount of European ABS,more than half (€612 billion) is based on resi-dential mortgages (see Table 1 below), which wereamong the best performing category of securitisedproducts (S&P, 2013) and would therefore be anatural target for ECB asset purchases. SME ABSconstitute a smaller part (€116 billion). The ABSstock outstanding is unequally distributed acrosscountries, with the main issuers being differentfrom the main issuers of corporate bonds. ABS pur-chases would be concentrated on the Nether-lands, Spain and Italy, and could therefore be agood geographical complement to corporate-bond

Table 1: Securitisation in Europe. outstanding stock in 2013Q4 (€ billions)

ABS CDO CMBS RMBS SME WBS TOTALAustria 0.3 0.2 1.8 2.3Belgium 0.1 0.2 63.3 17.8 81.4Finland 0.6 0.5 1.1France 22.6 2.0 10.2 1.9 0.5 37.3Germany 35.8 2.2 10.6 15.3 5.9 0.1 69.9Greece 14.3 1.8 4.3 7.2 27.6Ireland 0.3 0.2 0.4 37.6 38.5Italy 50.5 3.5 10.1 85.6 28.2 0.9 178.9Netherlands 2.3 0.9 2.5 249.7 8.0 263.5Portugal 4.2 26.2 5.3 35.7Spain 27.0 0.5 0.4 118.0 37.7 0.0 183.6Pan Europe* 2.0 33.9 13.0 0.2 3.2 0.2 52.4Multinational** 0.9 81.1 1.8 0.4 0.8 84.9Selected euro-area total 160.9 123.9 41.2 612.3 115.8 3.0 1057.2

Source: AMFE (2014). Note: All volumes in €. ABS: asset-backed securities for which collateral types include auto loans. creditcards. loans (consumer and student loans) and other. CDO: Collateralised Debt Obligations denominated in a European cur-rency, regardless of country of collateral. CMBS: Commercial Mortgage Backed Securities. RMBS: Residential Mortgage BackedSecurities. SME: Securities backed by small- and medium- sized enterprises. WBS: Whole Business Securitisation: a securiti-sation in which the cash flows derive from the whole operating revenues generated by an entire business or segmented partof a larger business. * Collateral from multiple European countries is categorised under ‘PanEurope’ unless collateral is pre-dominantly (over 90 percent) from one country. ** Multinational includes all deals in which assets originate from a variety ofjurisdictions. This includes the majority of euro-denominated CDOs.

purchases, which would be concentrated inFrance, Germany and Italy.

An ECB purchase could promote the developmentof securitisation in the euro area. The potential forsecuritisation is relevant, because many loanswould qualify for securitisation. In March 2014,the outstanding amount of loans in the EU to non-financial corporations stood at €4.2 trillion, and tohouseholds at €5.2 trillion20. From a monetarypolicy perspective, it would be very beneficial tocreate ABS that are based on a portfolio of Euro-pean assets. Ideally, the credit risk should bepooled at the level of the private sector, therebydeepening cross-border financial integration.However, the ECB should not wait for develop-ments in the ABS market before it starts buyingsecuritised products.

5 CONCLUSIONS

Low inflation in the euro area is particularly dan-gerous, given high private and public debt levels inseveral euro-area countries and the need for rela-tive price adjustment between the euro-area coreand periphery. According to recent ECB forecasts,average euro-area inflation is not expected toreturn to close to two percent in the medium term.

18. http://www.bis.org/review/r140407a.htm; S&Preports that the cumulative

downgrade rate in theperiod between mid-2007

and end-Q3 2013 was 33.8percent, meaning that the

S&P ratings on two-thirds ofEuropean structured

finance notes have eitherbeen stable or have risen

since mid-2007. S&P alsopoints out that default

trends vary substantiallyfor different asset classes,

with consumer transactions(ie RMBS, covered bonds

and consumer ABS) gener-ally outperforming corpo-

rate transactions (iecorporate securitisation,

CMBS and other ABS).

19. Retention requirements– which should lead sellers

of ABS to monitor carefullythe underlying collateral –

have been introduced in thecontext of the EU CapitalRequirements Directive,

and the EBA is working onthe technical details (ie 5percent retention require-

ment); seehttps://www.eba.europa.eu/-

/eba-publishes-final-draft-technical-standards-on-secur

itisation-retention-rules.

20. According to Darvas(2013), out of this €4.2 tril-lion, the stock of SME loans

in the EU in 2010 wasapproximately €1.7 trillion,

and the largest stock of SMEloans was in Spain

(€356bn), followed by Ger-many (€270bn), Italy(€206bn) and France

(€201bn).

19


The ECB’s commitment to its communicated objec-tive of keeping inflation “below but close to twopercent inflation in the medium term” has there-fore been undermined. In our view, it was a majormistake not to ease monetary conditions at thetime that the ECB’s own forecast signalled thatinflation will not return to two percent in themedium turn. Also, government policies, includingon bank restructuring and public investment,should have been implemented some time ago asa safeguard against the disinflationary process.

We also showed that inflation forecasts andexpectations about the return to normal inflationhave proved too optimistic, both in the euro arearecent years and in Japan for almost two decades.To effectively address the risk of persistently lowinflation, the ECB should act. Cutting ECB interestrates further and reducing the ECB’s deposit ratefor banks below zero would help but is unlikely tohave a sizeable impact. Designing a new verylong-term (eg three years or longer) refinancingoperation would not be very effective either,because liquidity conditions have normalised andbanks now have a preference for paying backthree-year LTROs earlier. Stopping the sterilisationof the government bond holdings from the Securi-ties Markets Programme (SMP) would be unwise,because that programme had a specific purposewith the stated feature of sterilisation.

The best option for the ECB, with the greatestpotential to sizeably influence inflation and infla-tionary expectations, is an asset-purchase pro-gramme. We recommend that the ECB starts anopen-ended programme of €35 billion per monthof asset purchases and reviews this amount afterthree months to see if its size needs to bechanged. Using empirical estimates from the liter-ature and our assessment, €35 billion per monthof purchases over the course of a year could liftinflation by 0.8-1.0 percentage points. The asset-purchase programme should only start to be cutback when inflation has increased and medium-term inflation expectations are anchored at twopercent. When inflation has stabilised close to butbelow two percent, the purchased assets should

be sold gradually at a pace that does not under-mine inflationary expectations. However, if infla-tion expectations rise significantly above twopercent, the sale of assets should be acceleratedand standard monetary policy tools should alsobe deployed to fight inflation.

In terms of available assets, we recommend thatthe ECB purchases privately-issued debt securi-ties and EFSF, ESM, EU and EIB bonds, but not gov-ernment bonds. The purchase of governmentbonds can be problematic if there is a governmentsolvency risk, and would be politically controver-sial. The combined stock of EFSF/ESM/EU/EIBbonds suitable for purchases is €490 billion. Interms of private debt instruments, we advise thatthe whole range of assets that are eligible as col-lateral at the ECB without further credit ratingrequirements should be considered. The total poolof such assets is €7 trillion. Of these, corporatebonds (for which we estimate that at least €900billion are suitable for purchases) and assetbacked securities (ABS, at least €330 billion suit-able for purchases) are preferable, while thebonds of sound banks could be considered onlyafter the ECB’s comprehensive assessment of thebanking system has been completed. We there-fore recommend that before the completion of thecomprehensive assessment, the ECB startsmonthly purchases of €15 billon of corporatebonds, €8 billon of ABS and €12 billion ofEFSF/ESM/EU/EIB bonds.

Given the relative size of the purchases comparedto the total size of the respective markets, the mis-pricing of risk would be limited and the ECB wouldnot take on board much risk if a sufficiently diver-sified portfolio of assets is purchased. We alsohighlight that the ECB’s Treaty-based primary man-date is maintaining price stability and there is noprohibition of monetary operations that exposesthe ECB to potential losses and profits. However, itis also clear that the ECB should avoid exposureto unchecked political and private sector pres-sures that could result in delays to the reversal ofasset purchases, which would undermine thebank’s price stability mandate.

20


REFERENCES

Antolin-Diaz, Juan (2014) ‘Deflation risk and the ECB’s communication strategy’, Fulcrum ResearchPapers, available at: http://www.fulcrumasset.com/files/frp201402.pdf

Bank of England (2012) ‘The Distributional Effects of Asset Purchases’, 12 July,http://www.bankofengland.co.uk/publications/Documents/news/2012/nr073.pdf

Chen, Han, Vasco Cúrdia and Andrea Ferrero (2012) ‘The macroeconomic effects of large-scale assetpurchase programs’, The Economic Journal vol. 122(564): F289–315

Chung, Hess, Jean-Philippe Laforte, David Reifschneider and John C. Williams (2012) ‘Estimating themacroeconomic effect of the FED’s asset purchases’, FRBSF Economic Letter 2001/03, January

Consensus Economics (2014)Claeys, Grégory, Pia Hüttl and Silvia Merler (2014) ‘Is there a risk of deflation in the euro area’,

Bruegel Blog, 3 AprilDarvas, Zsolt (2012) ‘The ECB’s magic wand’, Intereconomics: Review of European Economic Policy,

vol. 47(5): 266-267Darvas, Zsolt (2013) ‘Banking system soundness is key to more SME’s financing’, Policy Contribution

2013/10, BruegelECB (2009) ‘Decision of the European Central Bank of 2 July 2009 on the implementation of the cov-

ered bond purchase programme’, Official Journal of the European Union, ECB/2009/16ECB (2013) Collateral eligibility requirements – A comparative study across specific frameworks,

JulyGagnon, Joseph, Matthew Raskin, Julie Remache and Brian Sack (2011) ‘The Financial Market

Effects of the Federal Reserve’s Large-Scale Asset Purchases’, International Journal of CentralBanking 7, no. 1: 3–44

Hancock, Diana and Wayne Passmore (2011) ‘Did the Federal Reserve’s MBS purchase programlower mortgage rates?’ Journal of Monetary Economics, vol.58 no.5: 498-514

Krishnamurthy, Arvind and Annette Vissing-Jorgensen (2013) ‘The Ins and Outs of LSAPs’, availableat http://www.kansascityfed.org/publicat/sympos/2013/2013Krishnamurthy.pdf

Joyce, Michael, David Miles, Andrew Scott and Dimitri Vajanos (2012) ‘Quantitative easing andunconventional monetary policy – an introduction’, The Economic Journal, 122: F271-288

Meier, André (2009) ‘Panacea, Curse, or Nonevent? Unconventional Monetary Policy in the UnitedKingdom’, Working Paper 09/163, International Monetary Fund

Moody’s (2014) ‘EMEA Non-Financial Corporates: Positive Credit Trends Likely to Continue Through2014’, February

Papadia, Francesco (2013) ‘Should the European Central Bank do more and go negative?’ Author’sblog, http://moneymatters-monetarypolicy.blogspot.be/2013/11/should-ecb-go-negative.html

Sapir, André and Guntram B. Wolff (2013) ‘The neglected side of banking union: reshaping Europe’sfinancial system’, note presented at the informal ECOFIN, September

S&P (2014) Transition Study: Six Years On, Only 1.5% of European Structured Finance Has DefaultedS&P (2014) 2013 Annual European Corporate Default Study And Rating TransitionsTabakis, Evangelos and Kantaro Tamura (2013) ‘The use of credit claims as collateral for Eurosystem

credit operations’, Occasional Paper no. 148, European Central BankWeale, Martin and Tomasz Wieladek (2014) ‘What are the macroeconomic effects of asset pur-

chases?’ Discussion Paper no. 42, External MPC Unit, Bank of EnglandWolff, Guntram (2013) ‘The ECB’s OMT programme and German constitutional concerns’, in The G20

and Central Banks in the new world of unconventional monetary policy, Brookings