Quarterly Bulletin2015 Q3 | Volume 55 No. 3

Quarterly Bulletin2015 Q3 | Volume 55 No. 3

Topical articles

How has cash usage evolved in recent decades? What might drive demand in the future? 216Box UK coin and Scottish and Northern Ireland banknotes 218Box The Bank’s approach to payments 219Box The demand for cash: international comparisons 222

Bank failure and bail-in: an introduction 228Box The regulatory reform package 231Box Open and closed bank bail-in 234

Insurance and financial stability 242Box Managing firm failures 246Box Potential fragilities that can make insurers more likely to fail 248Box Potential sources of procyclical behaviour 253

How much do UK market interest rates respond to macroeconomic data news? 259Box The sensitivity of other asset prices to data news 266

Estimating market expectations of changes in Bank Rate 273Box The wedge between SONIA and Bank Rate 275

Over-the-counter (OTC) derivatives, central clearing and financial stability 283

Recent economic and financial developments

Markets and operations 298Box Equity and foreign exchange market volatility 302

Report

Monetary Policy Roundtable 308

Appendices

Contents of recent Quarterly Bulletins 314

Bank of England publications 316

Contents

The contents page, with links to the articles in PDF, is available atwww.bankofengland.co.uk/publications/Pages/quarterlybulletin/default.aspx

Author of articles can be contacted atforename.surname@bankofengland.co.uk

Except where otherwise stated, the source of the data used in charts and tables is the Bank of England or the Office for National Statistics (ONS). All data, apart from financialmarkets data, are seasonally adjusted.

Research work published by the Bank is intended to contribute to debate, and does notnecessarily reflect the views of the Bank or members of the MPC, FPC or the PRA Board.

Topical articles

Quarterly Bulletin Topical articles 215

216 Quarterly Bulletin 2015 Q3

• Cash is often in the media spotlight. While some predict its impending demise, most peoplecontinue to use banknotes in their day-to-day lives. Although consumers are less likely to usecash for transactions than they were in the past, aggregate demand for banknotes is increasing.

• This article considers how cash usage has evolved in recent decades, how cash is used today, andwhy the Bank of England needs to prepare for a future where cash remains important.

How has cash usage evolved in recentdecades? What might drive demand inthe future?By Tom Fish and Roy Whymark of the Bank’s Notes Directorate.(1)

Overview

The issuance of banknotes is probably the most recognisablefunction of the Bank of England. People use banknotes as astore of value and as a medium of exchange when buying orselling goods and services. The Bank of England seeks toensure that demand for its banknotes is met, and that thepublic retains confidence in those banknotes.

The payments landscape has changed considerably in recentdecades. People can now make payments using debit andcredit cards, internet banking, mobile ‘wallets’ andsmartphone apps. Yet cash continues to be important in theUnited Kingdom, with demand for Bank of England notesgrowing faster than nominal GDP (see summary chart).There is now the equivalent of around £1,000 in banknotes incirculation for each person in the United Kingdom.

The growth in demand has been driven by three differentmarkets. The evidence available indicates that no more than half of Bank of England notes in circulation are likely tobe held for use within the domestic economy fortransactions and for ‘hoarding’. The remainder is likely to beheld overseas or for use in the shadow economy. However,given the untraceable nature of cash, it is not possible todetermine precisely how much is held in each market.

The future rate of growth in demand for cash is uncertainand will depend on a number of factors including alternativepayment technologies, retailer and financial institutionpreferences, government intervention, and socio-economicdevelopments. Finally — and probably most importantly —it will depend on the public’s attitude towards cash.

Over the next few years, consumers are likely to use cashfor a smaller proportion of the payments they make. Evenso, overall demand is likely to remain resilient. Cash is notlikely to die out any time soon.

As such, the Bank continues to invest in banknotes. Withinthe next few years, the Bank will be launching new banknotesfor the £5, £10 and £20 denominations. The new notes willbe made of polymer — a cleaner and more durable material— and will incorporate leading-edge security features thatwill strengthen their resilience against the threat ofcounterfeiting.

£50

£20

£10

£5

£1

Notes in circulation (NIC)/GDP (left-hand scale)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

0

10

20

30

40

50

60

70NIC as a percentage of GDP Value of Bank of England NIC, £ billions

1975 80 85 90 95 2000 05 10 15

Summary chart Value of Bank of England notes incirculation: 1975 to present(a)

(1) The authors would like to thank Kevin Finan for his help in producing this article.

Sources: Bank of England and ONS.

(a) See footnote (a) of Figure 1 on page 221. Based on annual data. For 2015, chart uses dataup to end-July for NIC and data up to 2015 Q2 for GDP.

Click here for a short video on cash usage filmed in the Bank of England’s cash vaults.

Topical articles How has cash usage evolved in recent decades? 217

Most people in the United Kingdom use physical currency(‘cash’) in their daily lives.(1) Yet today, consumers havegreater choice than ever in how they pay for goods andservices. This choice has prompted some reports that cashis in the final stages of its life.

Although there are many ways to pay, cash remainsimportant, and the aggregate demand for banknotes continuesto grow faster than nominal GDP. This article outlines howthe use of cash has evolved over time and considers whatfactors might influence cash usage in the future.

The first section reviews what money is and theBank of England’s responsibilities with respect to banknotes.The second section explores the evolution of paymenttechnologies. The third section examines the different sourcesof demand for cash. Finally, the article considers differentfactors that might influence the future demand for cash andhow cash-issuing authorities should prepare. Throughout, thearticle draws on a range of data and research, including thefindings of a survey commissioned by the Bank in 2014 intothe public’s attitudes towards cash. A short video discussessome of the key topics covered in this article.(2)

Money and the Bank of England’s role inissuing banknotes

What is money?A common way of defining money is through the functions itperforms. As explained in a previous Bulletin article,(3) moneyserves as:

(i) a medium of exchange — money is used to facilitatetransactions, and is exchanged for goods and services;

(ii) a store of value — money retains value over a period oftime, thereby allowing people to transfer purchasingpower from today to some future date; and

(iii) a unit of account — money allows goods and services tobe priced in a comparable manner.

There are two types of money held by firms and households:cash and bank deposits.(4) Ninety-seven per cent of themoney belonging to people in the United Kingdom is heldelectronically as deposits, with the remainder held in physicalform. When a consumer withdraws cash from his or her bankaccount, they are exchanging electronic deposits for physicalcurrency — the total amount of money in circulation remainsunchanged.

Money is principally created through commercial banksmaking loans. Whenever a bank makes a loan, it creates adeposit in the borrower’s bank account. The borrower is thenable to access and spend this money. This can be done

electronically (for example, using a debit card) or bywithdrawing and then spending cash. To provide this cash tocustomers, banks need to purchase banknotes. For instance,commercial banks buy banknotes from the Bank of England inexchange for central bank reserves.(5) Effectively, commercialbanks exchange electronic funds for banknotes in the sameway that households draw down on their bank deposits whenthey take out cash from an ATM.

Commercial banks need enough notes to meet anticipatedcustomer demand. While the Bank, through its monetarypolicy, plays a key role in influencing overall demand formoney, it does not seek to influence the proportion of moneythat the public demands as banknotes. This is driven largelyby public preference.(6)

What is a Bank of England note?A banknote represents an IOU from the Bank of England to theholder of the note. Bank of England notes are able to circulatebetween consumers and businesses because of the promissoryclause that appears on them, which states that the Bank willpay the bearer the sum of the banknote on demand. Thisenables the holder to trust that it is worth the value printedon it. As a bearer instrument, no ownership information isrecorded for a banknote — the holder is simply assumed to bethe owner. This means that when a banknote changes hands,a transfer of value takes place with immediate settlementfinality.

Historically, the value of currency in the United Kingdom wastied to gold. In practice, this meant that the Bank wouldhonour its promise by exchanging its banknotes for anequivalent value in gold. But since the United Kingdom leftthe gold standard in 1931, Bank of England notes have been‘fiat money’ — money by government decree that is notconvertible to gold (or any other commodity). The promissoryclause still exists today, and banknotes will be repaid withnew Bank of England notes, or via electronic paymentor cheque.

As discussed above, a banknote that is sold by the Bank to thecommercial sector is paid for via an electronic funds transferto the Bank. The Bank invests the funds received into aninterest-bearing asset, such as a government bond. The Banktherefore receives interest income from the assets, yet pays nointerest on the corresponding liability (the banknote). Oncethe costs of banknote production and issuance have been

(1) Throughout the article, use of the word ‘cash’ refers to both banknotes and coin.(2) https://youtu.be/dXBLoAWzulU.(3) See McLeay, Radia and Thomas (2014a) for a detailed introduction to money.(4) A third type of money is central bank reserves — deposits placed by commercial

banks with the Bank of England. Throughout this article ‘banks’ and ‘commercialbanks’ are used to refer to banks and building societies together.

(5) Only a limited number of commercial banks buy banknotes directly from theBank of England. See the section below on the Bank’s note issuance objectives fordetail.

(6) See McLeay, Radia and Thomas (2014b) for information on how money is created.

218 Quarterly Bulletin 2015 Q3

UK coin and Scottish and Northern Irelandbanknotes

Although cash in circulation in the United Kingdompredominantly comprises Bank of England notes, there aretwo other components (Chart A).

The United Kingdom’s coin policy is managed by HM Treasury,with coins produced by the Royal Mint. At the end ofMarch 2015, there was £4.2 billion of coin in circulation.(1)

The Government also authorises seven commercial banks toissue banknotes in Scotland and Northern Ireland (S&NIbanknotes). The Scottish issuers are: Bank of Scotland plc;Clydesdale Bank plc; and The Royal Bank of Scotland plc. TheNorthern Ireland issuers are: AIB Group (UK) plc (trades asFirst Trust Bank); Bank of Ireland (UK) plc; Northern Bank

Limited (trades as Danske Bank); and Ulster Bank Limited.The aggregate value of S&NI banknotes in circulation totalled£6.6 billion at the end of February 2015.(2)

The Banking Act 2009 stipulates that the commercial issuersmust fully back their notes in circulation with ring-fencedassets, which can take the form of Bank of England notes,UK coin and funds held in an interest-bearing account at theBank. This gives holders of S&NI banknotes a similar level ofprotection to holders of Bank of England notes. If one of theauthorised banks were to fail, there would be sufficient fundsto reimburse all noteholders. The Bank monitors the sevenbanks’ compliance with this legislative regime.

What is legal tender?S&NI banknotes are not legal tender in the United Kingdom,and Bank of England notes are only classified as legal tender inEngland and Wales. However, legal tender has a narrowtechnical meaning in relation to the settlement of debt. If adebtor pays in legal tender the exact amount they owe underthe terms of a contract, and the contract does not specifyanother means of payment, the debtor has a good defence inlaw if he/she is subsequently sued for non-payment of thedebt. This has very little practical application to everydaytransactions. The acceptability of any payment method isentirely at the discretion of the parties involved in thetransaction.

deducted from this income, the net income earned — referredto as ‘seigniorage’ — is passed on to HM Treasury. Thisamounted to £506 million in 2014–15.(1)

At the end of July 2015, there were nearly three and a halfbillion Bank of England notes in circulation, across the Bank’sfour denominations, totalling around £63 billion (Table A).

The Bank is not the only issuer of sterling cash in theUnited Kingdom. The box above briefly summarisesarrangements for UK coin and banknotes in Scotland andNorthern Ireland.

The Bank’s note issuance objectivesIn delivering monetary stability, the Bank maintains thepublic’s confidence in the currency, achieved throughmonetary policy which is implemented through sterlingmarket operations. The Bank also needs to make sure that thepublic remains confident in the banknotes it issues.Noteholders should be confident in the integrity of theirbanknotes — these should be of good quality, easy toauthenticate and resilient against the threat of counterfeiting.Banknotes should also be readily available, and in thedenominations that the public requires.

The Bank has a specialist research team that assessesdevelopments in banknote security and technology. Basedon this research, the Bank will periodically launch newstate-of-the-art banknotes. Printing of banknotes has beentendered to the commercial sector since 2003. The currentprinter is De La Rue plc.

Table A Bank of England notes by denomination

£5 £10 £20 £50 Total

£1.7 billion £7.8 billion £40.5 billion £12.6 billion £62.6 billion

340 million 780 million 2,025 million 252 million 3,397 million notes notes notes notes notes

0

10

20

30

40

50

60

70

Bank of England notes S&NI notes UK coin

£ billions

Chart A Value of UK cash by component, 2015

(1) Source: Royal Mint.(2) See Bank of England; www.bankofengland.co.uk/banknotes/Documents/about/

scottish_northernireland_annualreport2015.pdf. There are also a number ofnon-UK sterling banknotes that are issued in the Isle of Man, Channel Islands andGibraltar. These are issued under currency board arrangements, with banknotesbacked one-for-one with UK pound sterling. However, unlike for the S&NIbanknotes, the Bank of England has no involvement or oversight of thesearrangements.

(1) See Bank of England (2015a).

Sources: Bank of England and Royal Mint.

Topical articles How has cash usage evolved in recent decades? 219

In order to meet demand for banknotes, the Bank worksalongside the four members of the Note Circulation Scheme(NCS) — G4S Cash Solutions (UK) Ltd, Post Office Ltd,Royal Bank of Scotland plc and Vaultex UK Ltd (a joint venturebetween Barclays plc and HSBC plc).(1) NCS membersundertake the wholesale processing and distribution activitiesrequired for effective circulation (including the identificationand removal of counterfeit and ‘unfit’ banknotes). Newbanknotes are sold to NCS members by the Bank, and are onlyreturned when they are unfit and ready for destruction. TheBank works with the NCS members to anticipate and meetnear-term changes in demand.

Through a tiered system, NCS members supply commercialbanks and large retailers with banknotes. These will, in turn,supply banknotes to smaller banks and retailers. The Bankalso works with the broader cash industry to ensure robustauthentication of banknotes at all stages of the note cycle.(2)

The evolution of payment technologies

To support the banknote distribution process, the Bank needsto anticipate how demand for cash may change over the longterm. This article goes on to discuss the factors that may bedriving demand for cash, of which there are many, followed byan assessment of how these may develop in the future. As astarting point, it is useful to review the developments inpayment technology that have occurred to date, since they

directly affect the public’s reliance on cash for makingpayments, which is an important component of aggregatedemand for cash.

A brief history of payment methodsIn order for money to function as a medium of exchange, thereneeds to be a secure way of transferring it. This is the corefunction that different payment methods provide.

People have been using various forms of physical money forthousands of years. Although paper money dates back to the7th century in China, banknotes did not become widely used inthe United Kingdom until almost a thousand years later. Infact, they were pre-dated by cheques, which were introducedto the United Kingdom in the 14th century.

The Bank first issued banknotes shortly after it was establishedin 1694. At the time, private banks and goldsmiths also issuedtheir own banknotes as receipts for deposits of gold. Overtime, legislation was introduced to erode the note-issuingpowers of private entities, and since 1921 the Bank has beenthe sole issuer of banknotes in England and Wales.

From the 17th century until the 1960s, cheques, banknotesand coin were the predominant ways in which transactionswere settled. However, innovation in the past five decades hasdelivered great change in how we pay, with electronicexchanges of value becoming increasingly used alongside thephysical exchange of cash.

In 1966, the credit card was introduced to the United Kingdomgiving consumers more choice at the point of sale. This wasquickly followed by the launch of automated electronicpayments in 1968, enabling direct debits and credits. Fromthis point on, regular payments (for example, wages and utilitybills) no longer needed to be made in person, saving time andeffort for both consumers and businesses. The introduction ofdebit cards did not take place until 1987.

More recently, the growth in e-commerce has led to furtherchanges in how consumers pay. And the growth in online andmobile banking, alongside the introduction of the FasterPayments Service in 2008, has enabled individuals to transferfunds electronically almost instantly. See the box on this pagefor an overview of the Bank of England’s approach topayments.

Payment habits todayConsumers need to make both regular and spontaneouspayments. For simplicity, ‘regular’ transactions are defined as

(1) See Allen and Dent (2010) for information on how the Bank works with the industryto manage the circulation of banknotes.

(2) See ‘The code of conduct for the authentication of machine-dispensed banknotes’;www.cashservices.org.uk/sites/default/files/the_code_of_conduct_for_the_authentication_of_machine-dispensed_banknotes.pdf.

The Bank’s approach to payments

The Bank does not seek to promote any one paymentmechanism over another, but in supporting its mission,aims to protect and enhance the stability of the financialsystem.

The Bank operates the United Kingdom’s Real-Time GrossSettlement (RTGS) infrastructure which settles interbankobligations arising from payments made through a numberof payment schemes: Bacs; CHAPS; Cheque and CreditClearing; the Faster Payments Service; LINK; and VisaEurope.(1)

The Bank also supervises those payment systems that arerecognised as systemically important under the BankingAct 2009.(2)

Payment scheme supervision and the operation of theRTGS infrastructure are carried out by separate areas ofthe Bank.

(1) See Dent and Dison (2012) for a detailed explanation of the Bank’s Real-TimeGross Settlement infrastructure.

(2) See Bank of England (2015b).

220 Quarterly Bulletin 2015 Q3

those that are made periodically (for example, utility bills) andare agreed in advance. ‘Spontaneous’ payments are made forgoods and services that are purchased at the point of sale (forexample, from retailers, either face-to-face, online, or viatelephone).

The technological developments discussed above havereduced consumers’ physical exchanges of money, particularlyin the case of regular payments. The majority of regularpayments are now facilitated by standing orders and directdebits, with only 10% paid for with cash in 2014. In valueterms, the proportion is even lower.(1)

For spontaneous payments, cash has been more resilient: itremains the most commonly used payment method,accounting for 52% of spontaneous transactions in 2014. Thevalue of spontaneous cash-based payments has, on the whole,been quite stable over the past fifteen years, as shown by theblue line in Chart 1.

That said, as a share of total payments for spontaneousconsumption, cash usage has been in gradual decline. AsChart 1 shows, this largely reflects the steady increase in theuse of debit cards over the past fifteen years. In 2014, byvalue, debit cards facilitated £362 billion of transactions,compared to £166 billion for cash — despite the latter beingused in a greater number of transactions.

What factors drive demand for cash?

The use of cash for regular and spontaneous paymentsdoes not tell the whole story. Despite consumersincreasingly moving to other payment methods, the valueof Bank of England notes in circulation continues to grow,and has trebled over the past two decades. This is shown inthe top panel of Figure 1. This growth, which has beenconcentrated in the two highest denominations — £20 and£50 notes — has outpaced growth in aggregate spending in

the economy: the blue line on Figure 1 shows that the ratio ofbanknotes in circulation to nominal GDP has been on anupward trend since the mid-1990s. And this trend is notunique to the United Kingdom. The box on page 222 considersinternational comparisons in cash usage in more detail.

Together with the findings noted above on the evolution ofpayments habits, this suggests that growth in demand forbanknotes has not been driven solely by domestictransactions. However, as untraceable bearer instruments, it isnot possible to locate where banknotes are being held at anyone time.

In order to simplify the issue, this article considers six broadsources of demand for banknotes, based on two uses acrossthree markets (Table B). Cash can be used either as a mediumof exchange or as a store of value in: (i) the domesticeconomy — all activity within the United Kingdom that couldbe observed by the authorities, as well as legal informaltransactions such as second-hand sales; (ii) the overseasmarket; and (iii) the shadow economy — in other words, allillegal activities, as well as any legitimate activities that areunlawfully concealed from the authorities.

In reality, there will be regular movements between eachcategory. For example, if a tourist brings cash from overseas and spends it in a shop in London, these funds willmove from the overseas market to the domestic economy.Nonetheless, these categories are useful for illustrativepurposes when considering the stock of cash at a snapshot intime.

The remainder of this section examines each of these sourcesof demand and considers which may have contributed togrowth in demand for cash in recent years.

The evidence available indicates that no more than half of thestock of notes in circulation is likely to be held within thedomestic economy, either for transactional purposes or as astore of value. The remainder is likely to be held eitheroverseas or for use in the shadow economy, primarily as astore of value in both cases. No single source of demand islikely to have been behind the sustained growth. Indeed,the stock of notes held within each of the three markets forcash is likely to have grown.

(1) Based on data from Payments UK — 2015 UK cash and cash machines and 2015 UK consumer payments survey, available at www.paymentsuk.org.uk.

01 03 05 07 09 11 130

100

200

300

400

500

600

2000 02 04 06 08 10 12 14

Cash

Cheque

Debit card

Credit card

Indices: 2000 = 100

Chart 1 Value of payment methods used forspontaneous payments

Market Use

Domestic economy

Medium of exchange(transactional holdings)

Store of value(hoarded funds)Overseas

Shadow economy

Table B Different components of cash holdings

Sources: Payments UK and Bank calculations.

0.0

1.0

2.0

3.0

4.0

5.0

6.0

0

10

20

30

40

50

60

70NIC as a percentage of GDP Value of Bank of England NIC,(a) £ billions

1975 80 85 90 95 2000 05 10 15

£50£20

£10£5

£1

Notes in circulation (NIC)/ GDP (left-hand scale) The value of notes

in circulation has increased threefold over the past 20 years.

As at end-July 2015, the total value of Bank of England notes in circulation stood at

£1,000per person in the UK

equivalent to around

£62.6bnAverage annual growth in NIC over the past decade:

Euro Australian dollar

Canadiandollar

Swedishkrona

US dollar Sterling(c)

8.5%

6.0% 6.0% 5.6%4.7%

-2.4%

The growth in notes in circulation has outpaced growth in aggregate spending in the economy since the mid-1990s.

…to reach record levels. This trend is common across many countries, but is not universal.

Bank of England notes in circulation have risen substantially over recent decades…

Topical articles How has cash usage evolved in recent decades? 221

Domestic economy — transactional holdingsCash is used in the domestic economy to facilitatetransactions for goods and services. This transactionalcycle involves cash moving between banks (in branchesand ATMs), consumers (held in purses and wallets), andretailers (stored in tills or safes waiting to be banked).A bottom-up approach is used to estimate the value ofbanknotes that may be held within each of these three atany one time.

Data reported to the Bank indicate that £10 billion of cash washeld on average by banks in 2014. This figure has grown by a

third in the space of a decade. This growth partly results fromhistorically low interest rates which have reduced theopportunity cost of holding non interest bearing assets.An additional factor is likely to be the expansion in theUnited Kingdom’s ATM estate, which has grown by a fifthto 70,000 machines in the past decade.

The value of consumer transactional holdings is estimated intwo stages. First, there is the average value of cash thathouseholds will hold at any one time to facilitateconsumption (for both regular and spontaneous purchases).Second, a ‘buffer’ is added to account for the residual amount

Figure 1 The evolution of notes in circulation in the United Kingdom and overseas

Sources: Bank of England, ONS, other central banks and Bank calculations.

(a) Banknotes are deemed ‘in circulation’ once they have been acquired by the commercial sector from the Bank for their face value. Notes in circulation (NIC) are a liability of the Bank. Stocks of notes that are held inBank of England vaults or by members of the Note Circulation Scheme (under certain conditions) are not ‘in circulation’. Based on annual data. For 2015, chart uses data up to end-July for NIC and data up to 2015 Q2 for GDP.

(b) As of the end of March 2015. (c) Based on Bank of England notes in circulation. The box on page 222 discusses the other components of cash in the United Kingdom.

222 Quarterly Bulletin 2015 Q3

that individuals hold in their wallet for precautionaryreasons.(1) Taking these together suggests that householdsheld between £3 billion and £4 billion in transactional cash atany one time during 2014 (a range that has changed little overthe past decade).

The average cash holdings of merchants are a function of cashtakings, how frequently takings are banked, and the value ofthe ‘float’ and ‘petty cash’ that is retained at any one time.Based on anecdotal feedback from the retail sector, UKmerchants are estimated to hold between £2 billion and£5 billion in cash in 2014.

These three components are aggregated in Table C, giving arange of £15 billion to £19 billion that may be held in thetransactional cycle. This suggests that between 21% and 27%of total UK cash was held within the domestic transactionalcycle at any one time in 2014.(2) Applying the samemethodology to data from previous years suggests that thevalue held in the transactional cycle has risen, largely due to

(1) The value of cash held on aggregate at the point of withdrawals is estimated bydividing total consumer cash spending in 2014 (£191 billion) by the average numberof cash withdrawals per person. Dividing this by two gives the average value heldbetween withdrawals. Added to this is a buffer of £20 to £40 per adult, based onresponses to the Bank-commissioned survey.

(2) Bank of England notes, S&NI banknotes and UK coin.

The demand for cash: internationalcomparisons

The trend of growing demand for cash is common across manycountries, but it is not universal.

Growth in banknote demand has outstripped growth in GDP inrecent years in Australia, Canada and the United States, aswell as in the euro area. At around £1,000 per person, UK cashholdings are slightly higher than in Canada (£970), somewhatlower than in Australia (£1,220), and notably behind both theeuro area (£2,130) and the United States (£2,500).

There are a number of factors behind these internationalvariations: some countries will be quicker to embracealternative payments than others; some nations’ citizens willhave a higher propensity to hoard cash as a store of value;some currencies see greater overseas demand; and higherincome levels may support greater demand.

The significantly higher per capita holdings for the US dollarare not surprising given its common use outside of theUnited States. The dollar is often considered a ‘safe haven’currency and held as a store of value, particularly by citizens incountries with less stable economies. It is also used as amedium of exchange in countries that have undergone‘dollarisation’ — the process by which citizens of a country usea foreign currency in parallel with, or as an alternative to, adomestic currency. For example, Ecuador uses the US dollar,and adopted it as its official currency in 2000.

Given the difficulties in observation, direct measurements forthe proportion of US currency held abroad is not available.Nonetheless, the Federal Reserve has estimated (using indirectmethods) that around a half of US dollars in circulation wereheld abroad in 2012.(1) The euro performs a similar role withsome estimates suggesting that a quarter of euro currency wascirculating outside the euro area at the end of 2013.(2)

In Sweden, cash use has been falling. In recent years the valueof Swedish krona banknotes in circulation has declined by

around a fifth (Chart A). This is largely due to a very highpenetration of alternative payment methods, with the publicvery engaged in embracing new technologies — indeed, it iscommon for children to be given pocket money usingelectronic payment methods.

This trend has been supported by the ability of Sweden’sresidents to buy almost any good or service with alternativepayment methods. For example, homeless street vendorsselling Situation Stockholm (akin to the United Kingdom’sBig Issuemagazine) are able to accept payment via debit orcredit card. Increasingly, Swedish banks’ branches andretailers have also moved towards cashless arrangements.

Even so, cash remains important in Sweden. In 2014, therewas almost SEK 7,800 (around £590) in circulation for eachperson, and cash still accounted for around 20% of allpayments. Indeed, a new series of krona banknotes will beissued from this autumn.

0

50

100

150

200

250

300

350

400

450

2002 03 04 05 06 07 08 09 10 11 12 13 14

US dollar

Canadian dollar

Swedish krona

Australian dollarEuro

Sterling(a)

Indices: 2002 = 100

Chart A Growth in notes in circulation across nations

(1) See Judson (2012).(2) European Central Bank (2011), ‘The international role of the euro’;

www.ecb.europa.eu/pub/pdf/other/euro-international-role201107en.pdf.

Sources: Bank of England, Bank of Canada, European Central Bank, Reserve Bank of Australia,Riksbank, US Federal Reserve System and Bank calculations.

(a) Based on Bank of England banknotes.

Topical articles How has cash usage evolved in recent decades? 223

increased commercial bank holdings. However, as aproportion of UK cash, transactional holdings have been ingradual decline, down from between 34% and 45%since 2000.

These calculations are only illustrative and are impossible tovalidate given the untraceable nature of cash, yet the numbersare not inconsistent with the few comparator studiesavailable. For example, Gresvik and Kaloudis (2001) estimatethat between 37% and 47% of cash holdings in Norway couldbe explained by transactional use in 2000. Guibourg andSegendorf (2007) estimated the equivalent figure for Swedento be 33% in 2004.

These indicative findings raise the question: if the majority ofBank of England notes are not being used for everydaytransactions in the domestic economy, what are they beingused for?

Domestic economy — hoarded fundsIn the domestic economy, consumers may choose to hold cashas a store of value (and therefore are not intending to spend itin the short term). We refer to this as ‘hoarding’. People maychoose to save their money in a safety deposit box, or underthe mattress, or even buried in the garden, rather than placingit in a bank account.

Economic theory suggests that hoarding should not take placeunless there is a risk of a negative return to the individual ofholding bank deposits, either due to negative interest rates,the possibility of financial loss due to the potential failure ofthe bank, or due to the inconvenience of depositing funds.

Interest rates in the United Kingdom have not turned negative,and the Financial Services Compensation Scheme (FSCS)offers full insurance against deposit losses of up to £85,000.(1)

However, hoarding still occurs and did so long before thecurrent period of record low interest rates begun. For suchaction to be considered as rational, those that are hoardingcash must be gaining a non-financial benefit that they valuegreater than the financial returns and physical security offeredby a deposit account.

In 2014, the Bank commissioned a survey exploring consumerbehaviour in relation to cash usage and hoarding.(2) Responses

suggested that 18% of people hoarded cash, and those thatdid suggested the primary reason was to provide comfortagainst potential emergencies. Extrapolating the resultsindicated that a minimum of £3 billion was hoardeddomestically — around £345 per hoarder. In 2012, a separatesurvey commissioned by the FSCS estimated that householdswere holding around £5 billion in cash at home.(3)

However, these surveys are likely to produce underestimates.Given the sensitivity of the subject, many are likely to haveunder-reported their cash holdings (and 7% of respondents tothe Bank-commissioned survey preferred not to say). Limitedsamples also mean that surveys may fail to capture the effectof ‘super hoarders’ — based on anecdotal evidence, a smallnumber of people are thought to hoard large values of cash.As an illustrative example, if one in every thousand adults inthe United Kingdom were to hoard as much as £100,000, thiswould account for around £5 billion (nearly 10% of notes incirculation).

Overseas — transactional holdingsCash can exit and enter the United Kingdom via a number ofchannels. Large-value movements will be facilitated by banksand foreign exchange wholesalers before filtering through avery fragmented market. Smaller transfers will primarily resultfrom tourism and worker remittances.

Much of the cash located abroad — relating to its use as amedium of exchange — is the travel money purchased byinbound visitors to the United Kingdom, or the residual theystore when they return home. Beyond this, there will beBank of England notes held abroad in bureaux de changeawaiting purchase. Market intelligence indicates that demandfrom Europe and North America is primarily for this purpose.

Overseas — hoarded fundsOverseas residents may also demand Bank of England notes ashard currency — as a safe store of value. Although no firmevidence is available, market intelligence gathered by the Banksuggests that our highest denomination (the £50 note —which makes up 20% of Bank of England notes in circulation,by value) is primarily demanded by foreign exchangewholesalers abroad, which would be consistent with overseashoarding. The box on international comparisons on page 222highlights the significance of this source of demand for othermajor currencies.

Shadow economy — transactional holdingsThe shadow economy is defined broadly as covering ‘thoseeconomic activities and the income derived from them thatcircumvent or otherwise avoid government regulation,

(1) From January 2016, the FSCS deposit limit will be £75,000. See www.fscs.org.uk/news/2015/july/new-deposit-protection-limit-coming-on-1-january/.

(2) The survey was conducted by GfK NOP and involved 1,000 respondents.(3) See www.fscs.org.uk/uploaded_files/Press_Releases/press-release-2012-02-28.pdf.

Table C Estimated stock of UK cash in the domestic transactionalcycle

Lower bound Upper bound

Financial institutions £10 billion £10 billion

Consumers £3 billion £4 billion

Retailers £2 billion £5 billion

Total £15 billion £19 billion

Source: Bank of England estimates.

224 Quarterly Bulletin 2015 Q3

taxation, or observation’.(1) This comprises of two elements:(i) legitimate activities that are concealed from the authorities;and (ii) illegal activities and transactions.

An Institute of Economic Affairs report estimated the size ofconcealed legitimate activities to be equivalent to 10.3% ofGDP in 2012, with it having declined relative to GDP since thelate 1990s. This estimate was one of the lowest of all OECDcountries.(2) HM Revenue and Customs estimates that thevalue of uncollected tax, which includes estimates for hiddenactivity and tax evasion, has been broadly stable over the pastdecade.(3) In relation to illegal activities, data from theHome Office indicate that incidents of crime may have fallenin each year for the past decade, with significant reductions inrobbery, theft and domestic burglary.(4)

Together, this evidence suggests that activity in the shadoweconomy has not seen significant growth, so transactionalholdings of cash in the shadow economy are unlikely to havebeen the primary driver of the growth in banknote demand inrecent years.

Shadow economy — hoarded fundsUnless money is filtered back into the domestic economy, or isrecycled for transactions in the shadow economy, criminalsand tax evaders will hoard funds. Therefore, even if the flowof cash moving into the shadow economy is stable (or evenshrinking), the stock of cash held could be rising. Articles inthe media covering the large seizures of cash during policeraids suggest the values involved may be sizable. However,given its bearer-instrument status, there is limited research toconfirm the extent of cash held for use in the shadoweconomy, which will vary significantly across nations.

Lost, destroyed or forgottenA small amount of cash is also likely to have beeninadvertently lost or destroyed. Some will also have beenretained indefinitely by collectors. Data from euro-areacountries that imposed a deadline for the exchange of theirlegacy domestic currencies with euros suggest this mayaccount for around 1% of cash in circulation.

What might influence the demand for cash inthe future?

The future rate of growth in demand for cash is uncertain andthere are many variables that could influence the use of cash(as both a medium of exchange and as a store of value)domestically, overseas, or within the shadow economy. Thesecan be grouped into six broad themes: (i) alternative ways topay; (ii) alternative currencies; (iii) retailer and financialinstitution preferences; (iv) government intervention;(v) socio-economic developments; and (vi) the public’sattitude to cash.

Alternative ways to payOver the coming years, the number of alternative ways ofpaying is expected to grow, and is likely to reach greaternumbers of consumers. Most of these payment methods are‘wrapper’ services — payment methods that improve the userinterface while relying on existing infrastructure. For example,card issuers continue to add contactless technology, with thefunctionality now available on around 69 million cards.(5)

Mobile wallets have also entered the market. These can storepayment cards digitally, allowing contactless payments to bemade using a smartphone. They can also provide additionalsecurity — for example, the new Apple Pay wallet (launched inJuly 2015) requires fingerprint authentication via an iPhone.

Of course, the availability of technology is not enough byitself. Retailers need to be in a position to accept the newtechnologies. Most larger retailers have accepted plastic cardsfor many years, and an increasing number now accept cardpayments using contactless terminals.

It is also becoming easier for alternative payments to beaccepted by smaller businesses and by those that do notoperate from fixed premises. For example, technologies suchas iZettle allow small businesses, that typically rely on cashpayments (for example, market stalls, window cleaners andtaxi drivers) to accept card payments. Furthermore, anincreasing number of services can be paid for usingsmartphones without needing to have a card present — forexample, the Uber app allows users to pay for a private-hirevehicle automatically using pre-stored card details.

Alternatives are also making it easier to makeperson-to-person payments. For example, PayM (using theFaster Payments Service infrastructure) links users’ mobilephone numbers to their bank accounts, and enables paymentsvia text message.

Alternative currenciesBeyond alternative payment methods, some consumers cannow use alternative currencies.

In recent years, several local currency schemes have beenintroduced in towns and cities across the United Kingdom.(6)

For example, in Brixton, people can use Brixton Poundvouchers to buy goods in the local area. Local currencies areestablished to support local sustainability by incentivising

(1) See Dell’Anno and Schneider (2003).(2) See Schneider and Williams (2013).(3) HM Revenue and Customs (2014), ‘Tax gap estimates for 2012–13’;

www.gov.uk/government/uploads/system/uploads/attachment_data/file/364009/4382_Measuring_Tax_Gaps_2014_IW_v4B_accessible_20141014.pdf.

(4) www.ons.gov.uk/ons/publications/re-reference-tables.html?edition=tcm%3A77-373433.

(5) According to the UK Card Association. See www.theukcardsassociation.org.uk/news/contactlesslimitrise2015news.asp.

(6) Naqvi and Southgate (2013) consider local currencies in detail.

Topical articles How has cash usage evolved in recent decades? 225

spending at, and between, participants of the scheme. Theyrely on users trusting a new currency as a unit of account andas a medium of exchange. This is usually achieved by the localcurrency being backed by, and remaining on a fixed exchangerate with, sterling.

These schemes remain small (both individually and inaggregate) relative to the value of Bank of England notes incirculation and are unlikely to have a significant impact ondemand for cash in the long term.

Technological innovation has also led to the creation of digitalcurrency schemes such as Bitcoin.(1) A digital currency schemeincorporates both a new currency and a decentralisedpayment system — payments can be made without relying onintermediaries such as banks. Most of the schemes use apublicly visible ledger which is shared across a computingnetwork.

Digital currencies remain a nascent technology, and are stillused for only a tiny proportion of transactions, but theunderlying technology has the potential to be more widelyused in the financial system. Since they can enable theanonymous (or pseudonymous) exchange of value, digitalcurrencies could become more widely used by those that areconcerned about the privacy of transactions.

Retailer and commercial bank preferencesChanges in the future demand for cash will also be influencedby consumers’ ability to access it, and where they can spend it.Today, it is easy to access cash in the United Kingdom, witharound 70,000 ATMs, of which over 50,000 are free to use(accounting for 98% of all ATM withdrawals in 2014). Thereare also thousands of bank branches providing deposit andwithdrawal facilities, as well as retailers that offer a‘cash-back’ service.

A change in these distribution channels could deter peoplefrom using cash. For example, a reduction in the number ofATMs, or a move to charging for withdrawals would increasethe cost and effort incurred. Alternatively, a move to cashlessbank branches (as seen in Denmark and Sweden) would makeit harder for businesses to deposit cash takings, and mayinfluence them to phase out accepting cash.

While possible, there is no evidence to suggest thesedevelopments are likely in the United Kingdom in the nearterm. Banks continue to serve different payment needs, andthe recent growth in ATMs suggests that the public’s demandfor cash will continue to be met by the financial sector.

Most retailers still accept cash, with many smaller businessesrelying on it. If the proportion of retailers’ takings received incash declines (given consumer choice), the average cost ofprocessing cash transactions would increase, and in time, it

could become cost effective for retailers to steer customerstowards alternatives. However, there are no signs that thisoutcome will occur in the near term. According to the BritishRetail Consortium’s 2014 payments survey, the cost for aretailer of accepting cash is eight times cheaper than a debitcard transaction, and over 30 times cheaper than a credit cardtransaction.(2)

As long as cash remains useful as a medium of exchange and iseasy to acquire, it will also retain utility as a store of value.Cash hoarded by the public is held on the belief that it can beconverted for goods and services, or deposited into a bankaccount, at some future point. This logic applies equally tocash held for use within the shadow economy.

Government interventionIn the United Kingdom, neither the Bank nor the Governmenthave a specific objective to influence the proportion of moneythat is held as cash. However, a definitive shift away fromcash could be the by-product of a wide range of potentialpolicy changes.

Moves by the UK Government to transition welfare paymentsaway from cash (to deliver efficiency savings), coupled withthe rollout of basic bank accounts to address financialinclusion, will open up the opportunity for a sizable minority ofpeople to use payment cards for the first time.

In addition, any government intervention that impactsalternative payment methods could indirectly influence thefuture demand for cash. For instance, the EU Interchange FeeRegulations will place restrictions on the charges paid byretailers when processing card payments, and could lead tomore widespread acceptance of card payments.

In the United Kingdom, there are few policies directly limitingthe use of cash (although people are not permitted to use cashto buy scrap metal in England and Wales). Yet elsewhere,much more comprehensive limits have been placed on cashusage. For example, in many European countries, includingFrance and Italy, there are prohibitions on all cash transactionsabove a certain value in order to deter tax evasion and moneylaundering activities. More broadly, any policy targeted atreducing crime or tax evasion could lead to a smaller shadoweconomy, and consequently reduce demand for cash.

Socio-economic and geopolitical developmentsDemand for cash relating to each of the three markets couldalso be influenced by, among other things: changes (oranticipated changes) in macroeconomic indicators such as

(1) Ali et al (2014a) consider the emergence of digital currencies. Ali et al (2014b)consider the economics of digital currencies.

(2) See British Retail Consortium (2014), Payments Survey 2014;www.brc.org.uk/brc_policy_detail.asp?id=307&iCat=563&sPolicy=Payments&sSubPolicy=Retail+Payments+Survey.

226 Quarterly Bulletin 2015 Q3

GDP, interest rates, exchange rates and inflation; the demandfor international tourism; long-run demographic trends; andgeopolitical developments such as wars or changes ingovernments. Such external factors might be particularlypertinent for overseas demand, where the alternatives aregreater, and the decision to hold Bank of England notes islargely financial rather than emotional or habitual.

The public’s attitude towards cashOver the coming years, it is likely that alternative paymentmethods will become cheaper, easier to access for consumersand more widely accepted. Yet, alongside thesedevelopments, the underlying preferences of consumers willcontinue to be a key determinant in how much cash isdemanded.

Some change in payment preferences might be expected asdemographics evolve. For example, as people become moreconfident in using (and trusting) new technologies, they willfeel more comfortable in using alternative payment methods.

Even so, many consumers will continue to use cash. Despitethe advancements in technology, cash continues to offer aunique set of attributes as a medium of exchange that setsit apart from the alternatives. The Bank-commissioned2014 survey found that people like cash because it is fast andconvenient, universally accepted, and helpful for budgetmanagement. Using cash for transactions also providesimmediate final settlement at no visible cost to the consumer,and with no reliance on technology and central infrastructure.Finally, cash is entirely anonymous, enabling transfers of valuewithout any record being kept.

Thus far, no single payment method has offered or improvedupon this complete set of attributes. As such, cash remains akey part of the payment toolkit for large sections of society.

Less tangibly, but equally as important, the public retains anattachment to cash. Around a quarter of respondents to theBank-commissioned survey revealed that they would continueto use cash for transactions regardless of the alternativesavailable, or the incentives that may come with them.

Conclusion

Over the next few years, consumers will enjoy even greaterchoice when paying for goods and services and for paying eachother. As a result, cash consumption as a proportion of overallspending in the domestic economy will continue to decline.However, given consumer preferences, recent historical trends,and the absence of significant initiatives on the part ofretailers, banks or government to push people away from cash,the absolute amount of cash used for transactions is likely toremain resilient.

But this is only part of the story. The available evidencesuggests that there are multiple factors driving thenon-transactional growth in demand for cash. For theUnited Kingdom, as with many other countries, this hasresulted in demand for cash rising relative to GDP.

Of course, uncertainty remains over the future path ofdemand for cash. The Bank will therefore continue to monitordevelopments and respond as necessary. Based on existingevidence, it is likely that demand for cash will continue tobe substantial in the years ahead. As such, cash-issuingauthorities need to work with the cash industry and continueto invest in banknotes and coin.

In the United Kingdom, the Royal Mint is introducing a newand more secure £1 coin in 2017.(1) For banknotes, theBank of England will, following many years of research, launchthe next £5, £10 and £20 banknotes on polymer. Polymerprovides a strong foundation for increasingly sophisticatedsecurity features, and the new notes will provide a stepchange in counterfeit resilience. The new £5, featuringSir Winston Churchill, will be launched in Autumn 2016,followed by the new £10 featuring Jane Austen a year later.The new £20 will be launched in the next three to five yearsand will feature a historical character from the visual arts,which will be chosen following a public consultation.(2)

Cash remains important in the United Kingdom, and asVictoria Cleland, the Bank’s Chief Cashier, recentlyremarked, ‘because there is a lot of life left in cash [the Bankand the cash industry] need to work together to keep ithealthy and fit for purpose’.(3) The investment being madewill ensure the public can remain confident in it over themany years ahead.

(1) See www.royalmint.com/aboutus/news/the-new-1-pound-coin.(2) From May to July 2015, for the first time, the Bank asked the public for nominations,

from the field of visual arts, for who should appear on the next £20 note. In total29,701 nominations were made, covering 589 eligible visual artists. AtSeptember 2015, the Bank of England’s Banknote Character Advisory Committee wasconsidering all eligible nominations, and together with input from public focusgroups, was to produce a shortlist of three to five names from which the Governorwould make a final decision. The chosen character is due to be announced inSpring 2016, alongside a concept image showing the portrait as it will appear on thenote. See www.bankofengland.co.uk/banknotes/Pages/characters/nexttwenty.aspx.

(3) See Cleland (2015).

Topical articles How has cash usage evolved in recent decades? 227

References

Ali, R, Barrdear, J, Clews, R and Southgate, J (2014a), ‘Innovations in payment technologies and the emergence of digital currencies’, Bank of England Quarterly Bulletin, Vol. 54, No. 3, pages 262–75, available atwww.bankofengland.co.uk/publications/Documents/quarterlybulletin/2014/qb14q301.pdf.

Ali, R, Barrdear, J, Clews, R and Southgate, J (2014b), ‘The economics of digital currencies’, Bank of England Quarterly Bulletin, Vol. 54, No. 3,pages 276–86, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2014/qb14q302.pdf.

Allen, H and Dent, A (2010), ‘Managing the circulation of banknotes’, Bank of England Quarterly Bulletin, Vol. 50, No. 4, pages 302–10,available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/qb100405.pdf.

Bank of England (2015a), Annual Report 2015, available atwww.bankofengland.co.uk/publications/Documents/annualreport/2015/boereport.pdf.

Bank of England (2015b), The Bank of England’s supervision of financial market infrastructures — Annual Report, March, available atwww.bankofengland.co.uk/publications/Documents/fmi/annualreport2015.pdf.

Cleland, V (2015), ‘Working together to deliver banknotes for the modern economy’, available atwww.bankofengland.co.uk/publications/Documents/speeches/2015/speech838.pdf.

Dell’Anno, R and Schneider, F (2003), ‘The shadow economy of Italy and other OECD countries: what do we know?’, Journal of Public Financeand Public Choice 21, pages 223–45.

Dent, A and Dison, W (2012), ‘The Bank of England’s Real-Time Gross Settlement infrastructure’, Bank of England Quarterly Bulletin, Vol. 52,No. 3, pages 234–43, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/qb120304.pdf.

Gresvik, O and Kaloudis, A (2001), ‘Increased cash holdings — reduced use of cash: a paradox?’, Norges Bank Economic Bulletin Q4,pages 134–39.

Guibourg, G and Segendorf, B (2007), ‘The use of cash and the size of the shadow economy in Sweden 2007’, Sveriges Riksbank WorkingPaper Series No. 204.

Judson, R (2012), ‘Crisis and calm: demand for US currency at home and abroad from the fall of the Berlin Wall to 2011’, Board of Governors ofthe Federal Reserve System, International Finance Discussion Paper No. 1058.

McLeay, M, Radia, A and Thomas, R (2014a), ‘Money in the modern economy: an introduction’, Bank of England Quarterly Bulletin, Vol. 54,No. 1, pages 4–13, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2014/qb14q101.pdf.

McLeay, M, Radia, A and Thomas, R (2014b), ‘Money creation in the modern economy’, Bank of England Quarterly Bulletin, Vol. 54, No. 1,pages 14–27, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2014/qb14q102.pdf.

Naqvi, M and Southgate, J (2013), ‘Banknotes, local currencies and central bank objectives’, Bank of England Quarterly Bulletin, Vol. 53, No. 4,pages 317–25, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2013/qb130403.pdf.

Schneider, F and Williams, C C (2013), The shadow economy, Institute of Economic Affairs.

228 Quarterly Bulletin 2015 Q3

• During the financial crisis, several governments bailed out failing financial institutions becauseletting the firms fail and enter insolvency would have caused excessive disruption to the criticalservices that these institutions provide and to the wider financial system.

• Following the crisis, the framework for managing the failure of financial firms was reformed and anew tool, known as bail-in, was developed. Bail-in allows the authorities to make sure thatshareholders and creditors of a firm bear the costs of failure, without recourse to public funds.

Bank failure and bail-in: an introduction

By Lucy Chennells and Venetia Wingfield of the Bank’s Resolution Directorate.(1)

(1) The authors would like to thank Oliver Dearie and Andrew Gimber for their help inproducing this article.

Overview

During the financial crisis a number of governmentsintervened to support their largest banks, including by bailingthem out, in order to allow the financial system to continueto function. This was necessary because households,businesses and governments rely on the services that banksprovide and authorities did not have an effective means ofdealing with their failure without the use of public funds.

Bailing out large banks is undesirable. It is costly and it isalso likely to undermine the incentives for firms to be run in aprudent manner and for investors to monitor the activities ofthe firm to prevent excessive risk-taking from jeopardisingtheir investment. The cost of funding the firm is artificiallylowered, because the consequences of failure are at leastpartly borne by the public sector.

Following the financial crisis, the Financial Stability Boarddeveloped a set of principles for managing the failure ofsystemically important financial institutions. These ‘KeyAttributes’ of effective resolution regimes sought to ensurethat firms could fail without disrupting the financial system,without interrupting the critical services they provide and,importantly, without requiring public sector support. One ofthe tools included in the Key Attributes was bail-in.

In a bail-in, the claims of shareholders and unsecuredcreditors of the failed firm are written down and/orconverted into equity in order to absorb the losses andrecapitalise the firm or its successor. A bail-in is notnegotiated — it is imposed upon the firm and its creditors bythe authority responsible for resolution. It is designed tostabilise the firm, providing time to enable it to berestructured in order to address the underlying causes of its

failure. The aim is that the firm, or its successor, is able tooperate without public support.

The resolution of a large and complex firm is likely to involvebail-in. There are different approaches to bail-in and themechanisms that would be used to achieve it in differentjurisdictions vary. This article describes the mechanism thatis likely to be used in the United Kingdom and sets out theother elements that need to be in place for a bail-in to besuccessful. These include an appropriate legal frameworkwhich ensures that the resolution authority has thenecessary powers to act and, crucially, that firms havesufficient capacity — liabilities that can be bailed in — toabsorb the losses and be recapitalised.

There are a number of misconceptions about bail-in. Thearticle concludes by explaining how bail-in:

• is not an alternative term for contingent capitalinstruments;

• does not interfere unduly with shareholder and creditorproperty rights;

• is unlikely to be a cause of contagion to the widerfinancial system; and

• is not, by itself, the silver bullet that ends ‘too big tofail’.

Although bail-in remains untested in the United Kingdom, itsexistence and credibility appears to have already reduced theimplicit subsidy from the public sector to large banks. This islikely to correspond to a shift of risks from the governmentto private creditors and suggests that firms’ funding costshave also — appropriately — become more risk-sensitive.

Topical articles Bank failure and bail-in: an introduction 229

During the recent financial crisis, it was necessary for theauthorities to intervene to limit the disruption from failingbanks and other financial firms. This included providing publicfunds to recapitalise some banks. Although this successfullystabilised the financial system, the cost of doing so was borneby the public. As part of the package of measures adopted inthe aftermath of the crisis, a new tool — known as bail-in —was introduced to help ensure that shareholders and creditorsbear the costs of firm failure.(1) Its development forms a keypart of the efforts under way to remove the need for publicfunds to be used in this way again.

Bail-in is not a silver bullet. By itself it cannot guarantee thatthe resolution of a failed firm will be orderly. However it is anessential component of a wider framework that, takentogether, will allow authorities to intervene to manage thefailure of large, complex firms in an orderly way. This processis known as resolution. The Bank of England is the resolutionauthority in the United Kingdom.

This article provides an overview of bail-in. It sets out thebackground to the development of bail-in, as an additionalresolution tool. It explains how bail-in can be used to stabilisethe balance sheet of a failing firm until the firm can berestructured, and it describes some of the other elements thatneed to be in place, alongside the bail-in tool, in order toresolve a large complex firm successfully. The final sectionseeks to dispel some misconceptions relating to bail-in.

Setting the scene: the financial crisis, bailouts and the regulatory response

The global financial crisis that began in 2007 saw widespreadand severe disruption to the financial system. The bankingsector faced significant losses and in many cases banks’ accessto liquidity and funding was heavily restricted. This had asignificant impact on the real economy because it constrainedbanks’ ability to operate. For example, Chart 1 shows thesubstantial fall in UK GDP following the crisis. Governmentsand central banks in several developed countries intervened toprevent their banking systems from collapsing. Since thecrisis, a number of reforms have been adopted to reduce therisk that this type of intervention will be required in the future.This section explains the background to the development ofthe bail-in tool.

Interventions to support the banking systemThe measures taken to allow the banking system to continueto function included both traditional and non-traditionalmeasures: central banks provided emergency funding(liquidity injections) to financial markets and individualfinancial institutions to ensure that they could continue tomeet their obligations as they fell due; and they boughtcertain types of assets, such as corporate bonds, from financial

institutions (asset purchases) to improve the liquidity of creditmarkets. Public authorities also guaranteed some liabilities,such as deposits or new/existing debts (liability guarantees) toshore up confidence in the financial system.

In addition — and of particular relevance to this article — insome cases governments provided capital injections or ‘bailouts’ in exchange for full or partial ownership of individualfirms. This was necessary because the losses experienced bythese firms had reduced the amount of capital on theirbalance sheets to the point that they could no longer continueoperating.(2) The bailouts recapitalised firms to allow them tocontinue to operate.

The interventions were designed to allow the financial systemto continue to function, since the day-to-day activities ofhouseholds, businesses and governments rely on the servicesof banks and other financial institutions. Banks in particularplay a number of crucial roles in the economy. For example,they provide payment services to households and companies,so that they can make and receive payments; they extendcredit, in the form of mortgages and loans; and they offerother services, such as savings accounts and financialinsurance, that help households and companies manage thevarious risks that they face.(3) These services are sometimesknown as critical economic functions.

Without these interventions it is highly likely that a number offirms would have failed. In the absence of resolution regimesthat could credibly be applied to the largest banks andfinancial institutions, the only alternative to bailouts of thesefirms would have been to put them into a normal corporateinsolvency process. This process involves handing over themanagement of the firm to an insolvency practitioner oradministrator appointed by the court and, in all likelihood, the

(1) The term ‘firm’ is used as shorthand for financial firm. (2) For an introduction to banks’ balance sheets and the role of capital, see Farag,

Harland and Nixon (2013).(3) See Freixas and Rochet (2008) for a complete discussion of the role of banks.

95

100

105

110

115

120

125

130

135 Index: 2000 Q1 = 100

09 102000 01 02 03 04 05 06 07 08 11 1312 14 15

Chart 1 UK GDP, 2000–15(a)

(a) Chained-volume measure, at market prices.

230 Quarterly Bulletin 2015 Q3

day-to-day activities of the firm — including the criticalfunctions and services that it provides — would be interruptedfor a prolonged period or cease altogether.(1)

The interruption to the critical services that banks provide, andthe likely loss of confidence in the system that would havefollowed, would have led to widespread disruption to thefinancial system. Government intervention prevented suchdisruption, but the scale and costs of these interventions weresubstantial. Between October 2008 and October 2011, theEuropean Commission approved €4.5 trillion of state aidmeasures to financial institutions (equivalent to 37% of EU GDP) and the International Monetary Fund (IMF) estimatesthat the average increase in public debt associated with thecrisis was around 18% of GDP.(2)(3)

Public sector support and ‘moral hazard’The capital support that firms received from governments —the public sector bailouts of private firms — represented apublic subsidy to such firms. One of the adverse effects of theprospect of bailouts is the effect this has on the behaviour ofthe managers and owners because the cost of risk-taking isreduced for the firm. In particular, the expectation prior to thecrisis that bailouts would occur, at least for a set of firms thatwere perceived to be ‘too big to fail’, effectively created animplicit subsidy that reduced the cost of funding for thosefirms.(4) Investors expected governments to step in andprevent large banks from defaulting on their debts even if they failed. So the price that they charged for lending to large banks (for example, when buying bonds issued by thosebanks) was lower than it would have been without thisexpectation of government support. This lower price offunding the firms’ activities is likely to have encouraged themanagers of those firms to take more risk than was ideal forsociety as a whole.(5)

The managers and owners received the benefits of profitsearned from banking activities carried out in the run-up to thecrisis. But when the firms got into difficulty, losses wereshared with the public sector rather than being absorbedentirely by those who were, in theory, responsible forabsorbing the losses, that is the owners and unsecuredcreditors of the firms. This lack of an appropriate incentive toprotect against risk crystallising — because of being protectedfrom some or all of the adverse consequences — is known asmoral hazard.

Following the bailouts, the cost of funding increasedsubstantially for some of those governments that providedthem, particularly in the immediate aftermath of the crisis.(6)

The public support provided by governments to financialinstitutions increased the impression that governments were‘on the hook’, or responsible for, the losses of their financialsectors.

Post-crisis: the regulatory reform agendaThe financial crisis set in train a broad package of regulatoryreforms.(7) The Financial Stability Board (FSB), an internationalbody that monitors and makes recommendations about theglobal financial system, played a leading role in shaping thereform agenda. The FSB’s work was complemented by that ofthe Basel Committee on Banking Supervision (BCBS), whichsets international standards for the prudential regulation ofbanks. In response to the risks posed by having a set of globalbanks that are perceived to be ‘too big to fail’, G20 leadersendorsed a specific package of measures in November 2011.(8)

Some of these measures are summarised in the box on page 231. The most important of these, from the perspectiveof this article, was a new international standard for resolution regimes in order to manage the failure ofsystemically important institutions effectively. Bail-in wasincluded in the toolkit that should be available to resolutionauthorities.

Bail-in explained

The previous section introduced the background to thedevelopment of a framework for resolving failed financialfirms. This section outlines the tools that are available toresolution authorities to ensure that firm failure is managed inan orderly fashion, focusing on bail-in. It explains how bail-inworks, using a stylised example to describe how a failingbank’s balance sheet would look before and after the bail-in.It also describes the other elements that are needed for a bail-in to be effective.

Resolution tools available to authorities Once it has been determined that a firm is failing, or likely tofail, and that the other conditions for resolution have beenmet, the authorities responsible for resolution will try tomanage the failure in an orderly fashion and with a view tomeeting the objectives for resolution. These objectivesinclude preserving financial stability, ensuring the continuity ofcritical economic functions, and protecting insured depositorsand public funds.(9)

The first phase of a resolution is to stabilise the firm so thatthe critical functions that it provides can continue.Stabilisation is achieved either by transferring those critical

(1) For a more detailed discussion of the drawbacks of using ordinary corporateinsolvency techniques for banks, see Brierley (2009).

(2) See European Commission (2012).(3) See IMF (2014).(4) For a comprehensive review of the cost of ‘too big to fail’, see Siegert and Willison

(2015).(5) See, for example, Acharya, Anginer and Warburton (2015), Afonso, Santos and Traina

(2014) or Vazquez and Federico (2012).(6) See, for example, Acharya, Drechsler and Schnabl (2014).(7) See, for example, Davies (2012).(8) See Financial Stability Board (2011).(9) The Bank of England’s approach to resolution is set out in Bank of England (2014a).

This provides more details on the conditions for using resolution tools, the objectivesof resolution that need to be taken into account and the tools themselves.

Topical articles Bank failure and bail-in: an introduction 231

The regulatory reform package

The package of measures outlined by the FSB to address therisks posed by systemically important financial institutions hadthree main components. The first two of these are designedto reduce the probability that such firms fail. First, firmsidentified by the BCBS as global systemically important banks(G-SIBs) must have additional capital, beyond what is requiredfor other firms, given the greater impact of the failure of thesefirms on the financial system.(1) Second, the FSB agreed asignificant strengthening of the supervision of systemicallyimportant firms. This was intended to address the fact thatsome firms that had been assessed by supervisors and foundto be well-capitalised and highly liquid with strong riskmanagement systems in the run-up to the crisis subsequentlyneeded support.(2) A more robust approach to supervisionencourages supervisors to intervene early, to address problemsin firms before they become insurmountable.

The third component of the reform agenda is designed toreduce the impact of firm failure. The FSB proposed a newinternational standard for resolution regimes, to address thefact that the authorities in many jurisdictions did not have acomplete set of tools to allow them to intervene to managethe failure of financial institutions that could be systemicallyimportant. The resulting set of ‘Key Attributes’ of effectiveresolution regimes sought to make sure that firms could failwithout disrupting the financial system, without interruptingthe critical services provided by these firms, and withoutrequiring public sector support.(3)

In the European Union and other jurisdictions, many aspects ofthese reforms also apply to institutions that are notconsidered to be globally systemic. As a result banks are, onthe whole, required to have more capital and of a higherquality than was previously the case. Approaches tosupervision have also been enhanced. And, crucially, crediblealternatives to ordinary corporate insolvency — that is, bankresolution regimes — have been developed in jurisdictionswhere they did not previously exist.(4)

The shape of the new resolution regimesThe Key Attributes built on existing examples of internationalgood practice in resolution.(5) They set out the arrangementsthat should be in place to handle the failure of financialinstitutions that are considered systemic. They cover thefundamental elements of resolution regimes, including the

scope of firms to be covered, the objectives of resolution andthe nature of the tools and powers that should be available.The Key Attributes include bail-in within the set of tools thatshould be available to authorities.

The Key Attributes also set out a framework for handling cross-border co-operation, where firms operate in severaldifferent jurisdictions. And they contain requirements forpreparations that need to be undertaken before firms get intodifficulties:

• preparation of ‘recovery plans’ by firms, which set outactions the management intend to take to return the firmto a stable footing, should it get into difficulties;

• regular assessments by the authorities of howstraightforward (or otherwise) it would be to resolve thefirms using the tools available (known as ‘resolvabilityassessments’); and

• preparation of ‘resolution plans’ by the authorities, whichdetail the approach that the authorities are likely to takeshould the firm need to be resolved under the resolutionregime.

In addition, the Key Attributes set out certain safeguardsrelated to the use of resolution tools. These are necessarybecause use of the tools affects individual property rights. Thesafeguards are designed to achieve a balance betweenproviding certainty to creditors about how they would betreated in resolution and giving the authorities sufficientflexibility to carry out an orderly resolution. One of the keysafeguards is that where a shareholder or creditor of a firmthat is put into resolution is left worse off than would havebeen the case had the whole firm been placed into insolvency,that creditor should be entitled to compensation.(6)

(1) See BCBS (2011).(2) See FSB (2010).(3) See FSB (2014a). (4) In Europe the Bank Recovery and Resolution Directive established a European

framework for the recovery and resolution of banks and investment firms. It alsoexplicitly limits the use of bailouts and other forms of state support.

(5) For example, extensive revisions had been made to the existing US framework forintervening with failing banks by the ‘Dodd-Frank’ package adopted in Summer 2010.A permanent ‘special resolution regime’ for managing the failure of banks andbuilding societies was enacted in the United Kingdom in February 2009 (the BankingAct 2009) to replace temporary arrangements adopted during 2008, and theEuropean Union was in the process of developing a ‘crisis management framework’for banks and investment banks. The latter became the Bank Recovery andResolution Directive which was published in the Official Journal in June 2014.

(6) See Davies and Dobler (2011).

functions to another, financially sound institution, or througha bail-in that restores the capital position of the original firm.To carry out a transfer, resolution regimes include options totransfer all or parts of a firm’s business to a private sectorpurchaser or, pending an onward sale or share issuance, to a

temporary (or ‘bridge’) bank. The parts that do not need to bemaintained permanently would be wound down in a measuredway, for example in an asset management vehicle or as part ofa special administration procedure.

232 Quarterly Bulletin 2015 Q3

The second phase is restructuring, when additional changesthat need to be made to the structure of the firm and to itsbusiness model are carried out. If a transfer has taken place, itmay be that no further restructuring is needed, but with a bail-in, restructuring will be required to address the causes offailure and restore confidence in the firm. The structure of thefirm and its business model are modified including, wherenecessary, through disposals and winding down of businesslines. Under the new resolution arrangements in the European Union, a resolution that involves a bail-in of anexisting firm must be accompanied by a restructuring plan.The existing management of the firm may also be replaced aspart of the process.

When the restructuring is complete, the firm exits resolution.The resolution authority’s involvement in the firm (or in any entities that emerge from the resolution) comes to anend.

The key features of bail-inIn a bail-in, the claims of shareholders and unsecured creditors are written down and/or converted into equity toabsorb the losses of the failed firm and recapitalise the firm orits successor. This is done in a manner that respects thehierarchy of claims prescribed in insolvency law. A bail-inallows the firm to continue to operate and to meetsupervisory requirements so that the critical functions the firmprovides can be maintained immediately after entry intoresolution.

Unlike a debt-for-equity swap, where the terms of anyexchange of debt for shares are negotiated by the relevantprivate parties, a bail-in would be imposed upon the firm andits creditors by the resolution authority. There would be norequirement to get the consent of shareholders, creditors orthe existing management of the firm. And there is norequirement for court approval of the bail-in.

The concept of bail-in evolved in the aftermath of the failureof Lehman Brothers in 2008. Some of those who had beeninvolved in the discussions on how to handle the fallout fromthe failure of a large, cross-border investment bank set out amethod for using the firm’s own resources, rather than publicfunds, to restore the balance sheet of the firm. Calello andErvin (2010) proposed that the holders of the firm’s bondswould have their investments in the firm written down andconverted into shares. This would be an alternative to a publicbailout or a disorderly insolvency and would provide thecapital that the firm was required to hold in order to beallowed to operate by its regulator. This in turn would givethe authorities, and the firm’s new management, the timethey needed to find a permanent solution to the problems ofthe failing firm, which would involve a major restructuring ofits activities and the adoption of a new business plan.

Bail-in has the considerable advantage that it does notdepend on the authorities finding a willing and ablepurchaser for all or part of the business in a short period oftime. Nor does it require the firm to be broken upimmediately. Bail-in is therefore the tool most likely to beused for the largest, most complex firms where theprospect of finding a willing private sector purchaser forsignificant parts of the business is low and where thecomplexities of effecting a full or partial transfer would besubstantial. There is more than one way to carry out a bail-in (as set out in the box on page 234).

Bail-in can also be used on smaller firms. Accordingly, theBank Recovery and Resolution Directive gives resolutionauthorities the power to use the bail-in tool in respect of anyfirm that meets the conditions for use of the stabilisationtools. Resolution authorities will choose the tool which bestmeets the resolution objectives in the circumstances.

How does a bail-in work?In order to understand how bail-in works, it is helpful toconsider the key components of a balance sheet.

Figure 1 shows a stylised balance sheet of a bank. The bank’s‘sources of funds’ (its liabilities and capital) are shown on theright-hand side, and its ‘use of funds’ (its assets) are shown onthe left-hand side. The bank’s assets include loans that it hasmade to households and businesses; lending to other financialinstitutions; and holdings of securities such as governmentand corporate bonds, as well as its holdings of cash. Thebank’s liabilities are what the bank owes to others. Theyinclude deposits from households and firms as well as fundsthat the bank has borrowed, for example from institutionalinvestors such as pension funds, by issuing debt in the form ofbonds. Liabilities are shown in order of ‘seniority’ in thehierarchy of creditors, with the most senior liabilities at thetop and the most junior, which are first to absorb losses, at the

Assets

Subordinated debt

Unsecuredsenior

liabilities

Equity

Securedliabilities

Deposits

Assets(‘Use of funds’)

Liabilities and capital(‘Sources of funds’)

Order inwhich

losses areabsorbed

Figure 1 Simplified bank balance sheet

Topical articles Bank failure and bail-in: an introduction 233

Note: Block sizes are not to scale.

D

Assets£290

A

Loss on assets£10

Assets

Sub-debt £3

Unsecuredsenior liabilities

£10

Equity£9

Securedliabilities

£158

Deposits£120

Liabilities and capital

Loss on assets£10

£1 of the subordinateddebt layer is written down

The original£9 equity iswiped out

B

Assets£290

Liabilities and capitalLiabilities and capital Assets Liabilities and capital

Securedliabilities

£158

Securedliabilities

£158

Securedliabilities

£158

Deposits£120

Deposits£120

After bail-in

Unsecuredsenior liabilities

£10

Unsecuredsenior liabilities

£10

Unsecured £3

Equity£9

Sub-debt £2 Sub-debt £2

£2 subordinateddebt and

£7 unsecured senior debtconverted to equity

Unsecured senior debt layer is

smaller than itwas before the

bail-in

Deposits£120

Firm’s capitalposition isrestored

Before bail-in Step 2:recapitalisation

Step 1: write-down to absorb loss

Panel A Panel C Panel DPanel B

bottom. Senior liabilities include, for example, liabilities thathave been secured against assets on the other side of thebalance sheet, cash deposits and high-quality debt (such assenior unsecured debt). Junior liabilities include lower-quality,or subordinated, debt. Equity, which is fully loss absorbing, isthe firm’s capital.

In common with the other resolution tools, bail-in allows afailing firm to be stabilised prior to a restructuring. There aretwo distinct steps to the stabilisation phase:

(i) Absorbing losses.

The first step is to estimate the outstanding losses of the firm,which is achieved through an initial valuation of its asset andliabilities. This is necessary prior to any resolution, in order toestablish whether the firm is failing, or likely to fail. Losseswhich have not already been fully recognised are absorbed bywriting down the value of assets. The losses may or may notwipe out the existing equity in the firm, but they are likely topush the firm’s capital level below that which is required bythe firm’s prudential supervisor. If the losses exceed theexisting equity, each layer of unsecured creditors in thecreditor hierarchy will be written down, in the order of theirranking in insolvency,(1) until the amount necessary torecognise the outstanding losses is covered.

(ii) Recapitalising the balance sheet.

The second step is to restore the capital the firm needs tosupport its activities, to ensure that the market has confidencein the firm, and to meet the requirements of the prudentialsupervisor through the subsequent restructuring phase.(2) Thebulk of the recapitalisation is likely to be achieved byconverting the claims of creditors into equity.

Figure 2 illustrates how a bail-in can be used to absorb thelosses of a failing bank and recapitalise it. In panel A the bank,which has a total balance sheet of £300, suffers a £10 loss.This could be because some of the loans it has made are notrepaid. As a result of the loss, the bank will have insufficientcapital to continue to operate, so it enters resolution. Forsimplicity, it is assumed that the bank will need to berecapitalised to the same amount as before the firm enteredresolution. In this example, this means equity of £9.(3)

Figure 2 Stylised example of loss absorption and recapitalisation in a bail-in

(1) Holders of regulatory capital instruments will bear losses before other creditors. TheBank Recovery and Resolution Directive introduced a mandatory write-down ofregulatory capital instruments at the point of non-viability (PONV). This means thatcommon equity (CET1), additional Tier 1 (AT1) and Tier 2 (T2) that qualify asregulatory capital instruments must absorb losses, up to the extent required to meetthe resolution objectives, before or together with the use of any of the resolutiontools. Although it is possible that a PONV write-down alone could restore a firm toviability, where the losses are limited and the business model remains sound, theexpectation is that it would generally be applied at the same time as the relevantresolution tools.

(2) The capital requirement set by the supervisor will depend, among other things, on theexpected nature of the business in the future, after any restructuring has taken place,and the costs of restructuring.

(3) In practice, the capital required is proportional to the size and complexity of the firm’sbalance sheet.

234 Quarterly Bulletin 2015 Q3

Open and closed bank bail-in

There is more than one way to carry out a bail-in. In particularit can be done either through the use of powers to write downand convert liabilities into forms of ownership in the restoredfirm, or using a bridge bank. The economic effect is largely thesame in each case, although the legal processes followed arelikely to differ. In each case the bail-in allows the resolutionauthority to stabilise a firm. It provides time that will allowfor an orderly reorganisation of all or part of a failing firm, inorder to address the underlying cause of the failure. And ineach case losses are absorbed and the firm or a successorentity is recapitalised.

In an ‘open bank’ bail-in liabilities are written down orconverted into equity in the existing firm. The firm remainsopen for business throughout the process. In a ‘closed bank’bail-in, the liabilities that are to absorb losses remain in theoriginal legal entity that is put into an insolvency processand/or bailed in while the activity that is to be continued istransferred to a newly created entity. Since the original firm isclosed this is known as a closed bank bail-in. Both options areavailable to European resolution authorities under the BankRecovery and Resolution Directive.

Open bank bail-inIn the United Kingdom, the most likely approach is to applythe open bank bail-in power, as described in this article. Thepractical mechanism expected to be used to carry out a bail-inis described in the annex.

Closed bank bail-inBy contrast, in the United States the economic equivalent of aclosed bank bail-in would be carried out under Title II of the

Dodd-Frank Act. The firm in question would be resolved bytaking action on a non-operating holding company. Once ithas been determined that an institution meets the conditionsfor resolution, it is likely that the Federal Deposit InsuranceCorporation (FDIC) would be appointed as receiver (insolvencypractitioner) to the parent holding company. The expectationis that the assets of the holding company, which are largelymade up of equity and investments in operating subsidiaries,would be transferred to a bridge holding company controlledby the FDIC. The day-to-day operations of the bridge holdingcompany would be carried out by newly appointed directorsand officers. The operating entities that carry out criticalservices would remain open, preventing disruption to thewider economy; and the claims (equity interests) of theshareholders and the claims of the subordinated andunsecured debt holders would likely remain in the receivershipwhere they would be effectively bailed in.

This means that the liabilities left in the bridge holdingcompany would be materially less than the assets. The bridgeholding company would therefore be well-capitalised. Theparts of the firm that are not in receivership may berestructured to address the causes of failure and — potentially— split into one or more smaller firms that could be resolvedthrough a bankruptcy proceeding (consistent with therequirements under Title I of the Dodd-Frank Act in the United States).

Finally, the ownership and control of the bridge holdingcompany and surviving entities would be transferred to privatehands. This may be done by converting the claims of creditorsin the receivership into equity in the bridge holding company.

Panels B and C show the liability side of the balance sheet andbreak down the bail-in transaction into the two phases. Panel B shows how the loss is absorbed. All of the bank’sequity (£9) is written down and some of the subordinateddebt (‘sub-debt’) (£1) is written down to cover the loss of £10.The losses have been absorbed without reaching the seniordebt layer.

Panel C shows how the firm is recapitalised. The remainingsubordinated debt (£2) and some of the senior unsecured debt(£7) is converted into equity to restore the firm’s capitalposition. The revised balance sheet of the firm, once the bail-in has taken place, is shown in Panel D. The unsecuredsenior debt layer is smaller than it was before the bail-in, butthe firm now has enough capital to satisfy the regulator and tooperate.

The resolution authority must generally respect the insolvencycreditor hierarchy when applying the bail-in tool. But it will

not always be necessary for existing shareholders’ claims to be written down to zero (‘wiped out’) before creditors canhave part of their claims converted into equity. This is because the losses may not completely wipe out the originalequity and shareholders have a claim on the equity thatremains once the losses have been absorbed. (For example, ifthe loss in the example had been £8, rather than £10.) Oncethese losses have been recognised on the balance sheet,additional capital will still be needed to replenish what hasbeen lost, to restore market confidence in the firm and tomeet the supervisor’s prudential requirements. This isachieved by converting some or all of the claims of those next in line in the creditor hierarchy into equity (or otherinstruments) — holders of subordinated bonds and, if the scale of the bail-in requires it, also the holdings of unsecuredsenior bonds.

Despite the fact that the original shareholders may not havebeen wiped out completely, their interest in the firm will be

Topical articles Bank failure and bail-in: an introduction 235

severely diluted, in recognition of their position at the front ofthe queue for absorbing losses. Furthermore, some liabilities(such as insured deposits and fully secured creditor claims) areexempt from bail-in and it may be necessary for certain otherliabilities to be excluded from a particular bail-in, for examplewhen bailing them in would be impractical, or inconsistentwith the objective of ensuring continuity of critical functions,or would destroy overall value. Although some liabilities maybe excluded from bail-in even if they rank alongside othersthat do bear loss, the outcome for those bearing loss must notbe worse than would be the case if the whole firm were placedinto insolvency. This respects the overall creditor hierarchy ininsolvency.

In practice a bail-in will be more complex than this examplesuggests. The scale of the effect of a bail-in on shareholders,and for those creditors whose claims are converted intoequity, will depend on a number of factors that will vary fromcase to case. These include the estimated scale of the lossesthat must be absorbed; the nature of the firm’s liabilitystructure (that is, how much loss-absorbing capacity there isand in what form); and the nature of the firm that is expectedto emerge from the resolution and therefore the capitalrequirement that the supervisor imposes. The mechanism forcarrying out a bail-in, and the valuation processes whichunderpin it, are described in the annex.

Bail-in is explicitly a prelude to the reorganisation of thebusiness of a failing firm. Once the firm has been stabilised,the causes of the firm’s failure must be addressed. This mayinvolve restructuring, selling or winding down parts of the firmthat are no longer viable. This should ensure that, following abail-in, the firm can operate normally, including accessingmarket funding.

Additional requirements for a successful resolution The FSB Key Attributes require resolution authorities todevelop strategies for resolving firms within scope of theresolution regime. These ‘resolution strategies’ set out theauthorities’ preferred approach for resolving the failing firm ina way that protects the critical functions carried out by thefirm, financial stability and public funds. Where necessary,resolution authorities may require firms to take steps toremove barriers to their resolvability. This ensures that mostof the planning for a resolution happens before a firm gets intodifficulties.

Irrespective of what the preferred resolution strategy may be,a number of elements must be in place in order for aresolution to be successful. These include:

(i) A broad set of resolution powers and a supporting legalframework.

Resolution authorities must have appropriate powers tosupport the use of resolution tools. These should include, for

example, powers to transfer some or all of the shares, assetsand liabilities of the failing firm to another firm or to a bridgebank; powers to carry out a bail-in; and powers to wind downany parts of the balance sheet that are not critical — eitherdirectly or by transferring them to an asset managementvehicle. The resolution authority also needs to be able torequire the firm itself, or any entity created from theresolution, to adopt a new business plan, to overhaul theinternal governance of the firm and, in particular, to removemembers of the senior management.

To be confident that the resolution of a complex, cross-borderfirm is enforceable, the statutory resolution framework in eachaffected jurisdiction must recognise — in a legal sense — theresolution actions of other jurisdictions. This includes bothrecognition of the resolution itself (the bail-in, for example)and recognition that the fact that a resolution has taken placedoes not give other parties the right to terminate any financialcontracts that they have entered into with the failing firm orwith other firms, particularly those in the same group.(1) In the absence of a statutory framework for recognising cross-border resolution, enforceability may be facilitated bycontractual arrangements.(2)

(ii) Loss-absorbing capacity in the right amounts, right formand right place within the group.

The goal of stabilising the balance sheet of a firm, andrecapitalising it, using bail-in can only be achieved if there issomething for the resolution authority to bail in. In otherwords, there must be liabilities on the balance sheet that canbe written down or converted into equity — or otherwiseexposed to loss in resolution — without disrupting the day-to-day functioning of the firm or causing wider financialinstability.(3) This is known as loss-absorbing capacity. TheFSB proposal for a common international standard on ‘totalloss-absorbing capacity’ (TLAC) is designed to set a minimumlevel of loss-absorbing capacity for G-SIBs.(4) Within theEuropean Union, resolution authorities must set a minimumrequirement for own funds and eligible liabilities (or MREL) foreach firm covered by the Bank Recovery and ResolutionDirective. Like TLAC, MREL is designed to ensure that all firmshave enough loss-absorbing capacity, including liabilities thatcould be credibly exposed to loss in resolution.

These requirements will have two key effects. First, they mayrequire certain firms to change the legal structure of their

(1) Contracts commonly contain provisions which enable a party to terminate thecontract upon reorganisation or other changes impacting the counterparty or amember of the counterparty’s group. If these termination rights are immediatelyinvoked upon resolution this is likely to exacerbate the troubles facing the firm andfrustrate the resolution.

(2) See FSB (2014b).(3) These liabilities may also be exposed to loss if the bail-in tool is not used. For

example they may be ‘left behind’ in an insolvency to capitalise a bridge bank or theymay be used to facilitate the transfer of assets to a private sector purchaser.

(4) See FSB (2014c).

236 Quarterly Bulletin 2015 Q3

existing wholesale funding to make sure that, when imposinglosses, following the creditor hierarchy does not underminethe objectives of resolution. This may involve making certainliabilities that are eligible to be bailed in subordinate to certain operating liabilities (for example, writing into theircontracts that they are lower in the creditor hierarchy) ormoving them to a holding company that does not carry outany of the normal activities of the firm. Second, they willensure that firms are not motivated to change their existingfunding structure in a way that makes it harder to imposelosses on creditors in a bail-in, for example by replacing long-term wholesale funding with short-term corporatedeposits.

(iii) Authorities in different jurisdictions must co-ordinatearrangements to resolve cross-border firms.

In order to prepare for, and facilitate the resolution of,complex cross-border firms, resolution strategies are discussedin crisis management groups (CMGs), made up of home andkey host financial sector authorities. Within the EuropeanUnion, ‘resolution colleges’ will facilitate co-operationbetween home and host resolution authorities for firms thatoperate in more than one Member State. International co-operation will also be facilitated by ensuring that all partsof a group have sufficient loss-absorbing capacity in place, togive host authorities confidence that material localsubsidiaries can be recapitalised as necessary.

(iv) A clear method of providing temporary liquidity to thefirm in resolution.

Bail-in is designed to ensure that a firm is recapitalised up to alevel that restores market confidence, while the restructuringthat follows a bail-in should ensure the firm’s long-termviability. This should mean that the firm has access to marketfunding. However it may be the case that, in the short term, afirm requires liquidity as a temporary backstop if marketparticipants are not immediately willing to lend to it. Thisliquidity could come from the central bank’s publishedschemes, or be provided on a bilateral, individually tailoredbasis. Within the European Union, any liquidity provided frompublic sources would need to comply with the EuropeanCommission’s state aid framework. And the firm would beexpected to make a gradual return to using private sectorsources of funding, as confidence in it returned.

Some misconceptions about bail-in

As might be expected following the development of a new andpowerful tool, there are a number of misconceptions aboutbail-in. It is important to understand what bail-in adds to thetoolkit for resolving failing banks and equally important tounderstand what bail-in is not. Some key misconceptions arediscussed below and summarised in Figure 3.

‘Bail-in is an alternative term for CoCos’Bail-in has sometimes been used as if it were a term that isinterchangeable with CoCos (contingent convertible capitalinstruments). These are a form of financial instrument thathas features of debt, such as coupons paid at particular datesand in specified amounts, but which can be written down orconverted into equity automatically when a particular triggerpoint is met, for example when the firm’s core capital level hasfallen to a particular point.(1) They are financial instrumentsissued by firms which may be used to meet regulatory capitalrequirements and typically constitute a small fraction of afirm’s overall funding.

As this article explains, bail-in is not a form of financialinstrument, but a resolution tool which can be used byauthorities if that is the appropriate way to manage the failureof a firm. Bail-in is a statutory resolution tool, and can beapplied to a range of different liabilities, including, but notlimited to, subordinated debt and unsecured senior liabilities.It is not subject to any permission or consent from themanagement of the firm, its shareholders or creditors, or acourt. Rather, it can be used by resolution authorities topursue their resolution objectives when the trigger forresolution is met, and its use is constrained by certainsafeguards. It is also accompanied by a wide-rangingrestructuring of the firm, which would not be the casefollowing the triggering of a CoCo.

‘Bail-in unduly interferes with shareholder andcreditor property rights’ Bail-in has been criticised for being unduly invasive ofshareholder and creditor property rights. Bail-in involvesreducing and/or altering shareholder interests in and creditorclaims on the firm without the consent of those shareholdersand creditors. As with other stabilisation tools, bail-in can beused by resolution authorities without court approval.

This certainly affects the property rights of those concerned.But the legislative framework is very clear: action by theauthorities to resolve a firm can only take place if this isnecessary, having regard to the public interest in theobjectives of resolution. So, in common with the otherresolution tools, bail-in may only be used where it is in thepublic interest to do so.

And as already noted, the use of resolution tools is subject tocertain safeguards. In particular, the legislation is explicit thatno creditor will be worse off after the application of the bail-inthan would have been the case had the whole firm been putinto insolvency. If the resolution fails to respect thissafeguard, creditors are entitled to financial compensation. Itmay be the case that property rights are better supported

(1) See Bank of England (2014b).

Topical articles Bank failure and bail-in: an introduction 237

through a bail-in which allows the firm to continue in someform than they would be if all or part of the firm were put intosome form of insolvency. And, given the difficulty of splittingup large and complex firms so that their critical economicfunctions could be maintained while other activities arewound down, bail-in may be the only option which meets theresolution objectives.

‘Bail-in is a cause of contagion, not a remedy for it’Because financial institutions tend to hold each other’s debt,bail-in has been described as a potential cause of contagion.For instance, if a bail-in occurred at Bank A, other firmsholding debt or equity issued by Bank A would experience aloss, and possibly a change in the nature of their investment(from debt to equity).

The wealth shock that investors experience when firms fail isnot a new risk — indeed it is the norm when a non-bank firmfails. What bail-in does change is the fact that, when a bankfails, investors in that bank incur the loss rather than transmitthat wealth shock to the wider population of taxpayersthrough a government bailout. The BCBS is developing aregime that restricts banks’ investments in the loss-absorbingcapacity of systemically important banks, precisely so that ashock experienced by one G-SIB is not immediatelytransmitted to others.

It is less clear that other financial institutions, such as assetmanagement firms and pension funds, should be prevented orrestricted from investing in the loss-absorbing capacity of G-SIBs. These firms may not be exposed to the same set ofrisks as banks, and in a banking crisis they may be in a strongerposition to absorb losses than other banks would be.Moreover they are typically less highly leveraged than banks.

Financial institutions invest in the equity and the debt of manydifferent types of firm and some of these will fail. Butproviding that the usual measures that prevent investmentsfrom being concentrated in any single firm or sector areadhered to, it is not immediately obvious that any otherrestrictions should be imposed.

‘Bail-in is the silver bullet that solves too big to fail’Some commentators appear to believe that bail-in is the onlychange necessary to solve the problem of ‘too big to fail’, andthat once the tool has been incorporated into resolutionregimes, the job will be complete. By itself, the bail-in tool isnot sufficient to solve the difficulties posed by the failure of alarge complex international bank. Bail-in is a necessaryaddition to the authorities’ resolution toolkit, but needs to beaccompanied by a set of measures that ensure a bail-in can becarried out without causing unacceptable adverseconsequences for the financial system and the wider economy.

Figure 3 Dispelling some myths about bail-in

Myth 2: Bail-in unduly interferes with shareholder and creditor property rightsBail-in does affect shareholders’ and creditors’

property rights. But resolution authorities may only use resolution tools — including bail-in — when this is necessary, with regard to the public interest. And no

creditor will be worse off following a bail-in than would have been the case had the whole firm been

put into insolvency.

Myth 1: Bail-in is just another term for CoCos

CoCos(a) are financial instruments that can be written down or converted into equity when a trigger point is met (such as a bank’s capital falling below a certain level). Bail-in is not a financial instrument. It is a resolution tool which, under certain conditions, can be used by resolution authorities to impose losses on shareholders and creditors

of a failing firm.

Myth 3: Bail-in is a cause of contagion, not a remedy for it

It is true that if a bail-in occurred at one bank then other banks holding the failing bank’s debt could

experience losses. This is justified: bail-in is designed to ensure that shareholders and creditors of a failed firm bear the

cost of its failure. And the size of banks’ holdings of other banks’ debt is likely to be limited by regulation.

Myth 4:Bail-in is the silver

bullet that solves ‘too big to fail’

Bail-in is an essential addition to the resolution

toolkit. But the resolution of a large and complex international firm depends on a broad resolution framework being in place, not just a bail-in tool.

(a) Contingent convertible capital instruments.

238 Quarterly Bulletin 2015 Q3

As described above, these include measures such as ensuringthat firms have enough loss-absorbing capacity, in the rightparts of the firm and in the right form, that can be bailed in ifthat becomes necessary. And they include ensuringappropriate co-ordination between authorities in differentjurisdictions.

Conclusion

A key lesson from the recent financial crisis was thatauthorities in many jurisdictions had insufficient tools to dealwith the failure of their largest, most complex banks. As aresult governments were forced to inject capital into, or bail out, some firms in order to prevent widespread disruption.This article has set out how the new resolution tool known asbail-in can help to prevent the need for bailouts in the future.

Practical mechanisms for carrying out a bail-in continue toevolve. However the legal framework to support it is nowextensive, the planning and preparation for its use is robust

and the market has begun to understand — and to price in —its effects.

This means that despite being untested, bail-in has greatlyreduced rating agencies’ expectations that governmentsupport will be provided. They have adjusted theirmethodologies for rating banks, substantially removing theformer weight placed on the likelihood of systemic banksreceiving public support, and as a consequence the ratings formany firms have been downgraded.(1) This appears to havereduced the funding cost advantages that large bankspreviously received due to expectations that they would betoo big to fail.(2)

A reduced expectation that large, systemically important firmswill be bailed out should not only reduce the comparativeadvantage that they enjoy — their cost of funding shouldreflect the cost of risk more accurately — it should also reducetheir incentives to take excessive risks. Bail-in is therefore avitally important piece in the jigsaw of measures that arebeing taken to remove the problem of ‘too big to fail’.

(1) See Moody’s (2015) and DBRS (2015).(2) See, for example, Bank of England (2014c) or US Government Accountability Office

(2014).

Topical articles Bank failure and bail-in: an introduction 239

Annex Mechanism for carrying out a bail-in — the UKexample

Although the concept of a bail-in can be succinctlysummarised — it is a two-step process to absorb losses andrecapitalise a failing firm or its successor — in practice, it alsorequires a significant restructuring of liabilities and of thefailing firm. Resolution authorities therefore need practicalmechanisms and processes to carry out a bail-in. Althoughbail-in has not yet been used in the United Kingdom, and itsmechanics continue to be developed, the approach currentlyenvisaged is set out here.(1)

Prerequisites In order for bail-in to work it must be possible to identifythose liabilities that will potentially be within scope of thebail-in. Firms are therefore required to be able to provideaccurate information to facilitate this, and firms’ systemsmust be able to support the valuation exercises that underpinan orderly resolution process.

Immediately prior to bail-inPrior to a resolution, the Bank will appoint an independentvaluer who will need to conduct a number of ‘pre-resolution’valuations. These help to inform the determination that a firm is failing or likely to fail, inform the choice of resolutiontool, and may also support more detailed resolution planning.For example, pre-resolution valuations may be conducted toestimate the amount of liabilities that would be bailed in anddetermine a provisional rate of exchange for the liabilities that will be converted into equity. The pre-resolutionvaluations may also estimate the potential outcome fordifferent types of creditor if the firm were placed intoinsolvency in order to assess the risk of ‘no creditor worse off’compensation claims that may arise from the proposedresolution strategy.

‘Resolution weekend’In most complex cases, it will be advantageous for theauthorities to have up to 48 hours outside normal markethours to conduct the initial transactions. This is known as a‘resolution weekend’. (It will not always be essential to havean actual weekend — the amount of time required will dependupon the amount of planning that has been carried out andthe speed of the firm’s failure.)

In the run-up to resolution the firm will be in private hands.During the ‘resolution weekend’, the resolution authority mayinstruct the settlement systems to block the trading of thesecurities subject to bail-in (if trading in the firm’s securitieshas not already been suspended). A resolution administratormay be appointed. The administrator will control the votingrights of all shares in the firm. The shares will be transferredto a third-party depositary bank.

The Bank of England, as resolution authority, expects toprovide holders of liabilities that may be bailed in with a proxyfor the compensation that they may receive following the bail-in. This may be achieved through ‘certificates ofentitlement’ that can be issued by the failing firm to holders ofliabilities that are potentially within scope of the bail-in.These certificates would represent a potential right tocompensation. For example, where the associated liability isconverted into equity, the certificate represents a claim to ashare of that equity. The precise terms or value of theexchange would be calculated during the resolution process.

Bail-in period Following the resolution weekend, the independent valuer willconduct (or finalise) an asset and liability valuation todetermine the scale of losses that must be absorbed, and toinform the firm’s restructuring plan following the bail-in. Anequity valuation is also conducted to estimate the value of thefirm’s equity following the resolution. This should include theexpected costs of the proposed restructuring and the firm’snew capital requirement after resolution. The equity valuationrepresents the value that is available to compensate theaffected shareholders and creditors — that is holders of theliabilities that have been bailed in.

Once these valuations are completed and the restructuringplan has been drawn up the bail-in terms are set andannounced. The bail-in terms determine the extent of thewrite-down to be applied and the ratio of shares (if any) thatwill be due to each class of creditor as compensation.

Once at least 51% of the equity has been transferred to thenew holders, or after a set time period, voting rights in thefirm are likely to be transferred from the resolutionadministrator to the new equity holders.

Post-resolution valuation Following the resolution, an independent valuer is appointedto assess whether any additional compensation may be due tothe shareholders and creditors affected by the bail-in. Thisvaluation compares the estimated insolvency outcome foreach class of creditor, based on a hypothetical insolvency ofthe firm on the date the decision to take resolution action wasmade. An assessment is made of the actual outcome of theresolution for each class of shareholder and creditor. If theindependent valuer concludes that insolvency would haveprovided a better outcome, additional compensation may bedue. This aims to ensure that the bail-in has not left anycreditor worse off than would have been the case had thewhole firm entered normal insolvency proceedings.

(1) The Bank of England retains the ability to exercise its discretion when deciding howbest to resolve a firm in pursuit of the objectives of the resolution regime, based onthe facts at the time.

240 Quarterly Bulletin 2015 Q3

References

Acharya, V, Drechsler, I and Schnabl, P (2014), ‘A Pyrrhic victory? Bank bailouts and sovereign credit risk’, The Journal of Finance, Vol. LXIX,No. 6, pages 2,689–739.

Acharya, V, Anginer, D and Warburton, A J (2015), ‘The end of market discipline? Investor expectations of implicit government guarantees’,mimeo, available at http://risk.econ.queensu.ca/wp-content/uploads/2013/10/Acharya-et-al-End-of-Market-Discipline-2014.pdf.

Afonso, G, Santos, J A C and Traina, J (2014), ‘Do ‘too-big-to-fail’ banks take on more risk?’, Federal Reserve Bank of New York EconomicPolicy Review, Vol. 20(2), pages 41–58.

Bank of England (2014a), The Bank of England’s approach to resolution, October, available at www.bankofengland.co.uk/financialstability/Documents/resolution/apr231014.pdf.

Bank of England (2014b), Financial Stability Report, June, ‘Additional Tier 1 capital’, pages 33–35, available at www.bankofengland.co.uk/publications/Documents/fsr/2014/fsr35sec2.pdf.

Bank of England (2014c), Financial Stability Report, December, ‘Ending ‘too big to fail’’, pages 35–38, available atwww.bankofengland.co.uk/publications/Documents/fsr/2014/fsr36sec3.pdf.

Basel Committee on Banking Supervision (2011), ‘Global systemically important banks: assessment methodology and the additional lossabsorbency requirement’, revised 2013, available at www.bis.org/publ/bcbs207.pdf.

Brierley, P (2009), ‘The UK Special Resolution Regime for failing banks in an international context’, Bank of England Financial StabilityPaper No. 5, available at www.bankofengland.co.uk/financialstability/Documents/fpc/fspapers/fs_paper05.pdf.

Calello, P and Ervin, W (2010), ‘Economics focus: from bail-out to bail-in’, The Economist, 28 January 2010.

Davies, G and Dobler, M (2011), ‘Bank resolution and safeguarding the creditors left behind’, Bank of England Quarterly Bulletin, Vol. 51, No. 3,pages 213–23, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/qb110302.pdf.

Davies, H (2012), ‘The financial crisis and the regulatory response: an interim assessment’, International Journal of Disclosure and Governance,No. 9, pages 206–16.

DBRS (2015), ‘DBRS places 38 European banking groups under review negative due to systemic support’, May, available atwww.db.com/ir/en/download/DBRS_on_European_Banks_20_May_2015.pdf.

European Commission (2012), ‘New crisis management measures to avoid future bank bail-outs’, press release, available athttp://europa.eu/rapid/press-release_IP-12-570_en.htm?locale=en.

Farag, M, Harland, D and Nixon, D (2013), ‘Bank capital and liquidity’, Bank of England Quarterly Bulletin, Vol. 53, No. 3, pages 201–15,available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2013/qb130302.pdf.

Financial Stability Board (2010), ‘Intensity and effectiveness of SIFI supervision: recommendations for enhanced supervision’, November,available at www.financialstabilityboard.org/wp-content/uploads/r_101101.pdf?page_moved=1.

Financial Stability Board (2011), ‘Policy measures to address systemically important financial institutions’, November, available atwww.financialstabilityboard.org/wp-content/uploads/r_111104bb.pdf?page_moved=1.

Financial Stability Board (2014a), ‘Key attributes of effective resolution regimes for financial institutions’, October, available atwww.financialstabilityboard.org/2014/10/r_141015/.

Financial Stability Board (2014b), ‘Cross-border recognition of resolution action: consultative document’, September, available atwww.financialstabilityboard.org/wp-content/uploads/c_140929.pdf.

Financial Stability Board (2014c), ‘Adequacy of loss-absorbing capacity of global systemically important banks in resolution: consultativedocument’, November, available at www.financialstabilityboard.org/wp-content/uploads/TLAC-Condoc-6-Nov-2014-FINAL.pdf.

Freixas, X and Rochet, J C (2008), Microeconomics of banking, MIT Press Books, second edition.

Topical articles Bank failure and bail-in: an introduction 241

International Monetary Fund (2014), ‘From banking to sovereign stress: implications for public debt’, December, available atwww.imf.org/external/np/pp/eng/2014/122214.pdf.

Moody’s Investors Service (2015), ‘Moody’s publishes its new bank rating methodology’, March, available atwww.moodys.com/research/Moodys-publishes-its-new-bank-rating-methodology--PR_320662.

Siegert, C and Willison, M (2015), ‘Estimating the extent of the ‘too big to fail’ problem — a review of existing approaches’, Bank of EnglandFinancial Stability Paper No. 32, available at www.bankofengland.co.uk/financialstability/Documents/fpc/fspapers/fs_paper32.pdf.

US Government Accountability Office (2014), ‘Large bank holding companies: expectations of government support’, GAO-14-621, available at www.gao.gov/products/GAO-14-621.

Vazquez, F and Federico, P (2012), ‘Bank funding structures and risk: evidence from the global financial crisis’, IMF Working Paper 12/29,January, available at www.imf.org/external/pubs/ft/wp/2012/wp1229.pdf.

242 Quarterly Bulletin 2015 Q3

• Insurance companies play an important role in supporting economic activity. But insurers areexposed to a number of risks and can become distressed or fail.

• This article considers a number of channels through which insurance companies could haveadverse effects on financial stability, including: how insurer distress or failure might disrupt theprovision of critical services to the real economy; and how their behaviours can propagatesystemic risk in the financial system. The Financial Policy Committee has an ongoing workplan toassess the extent of risks to financial stability from insurance companies’ activities.

Insurance and financial stabilityBy Andrea French and Mathieu Vital of the Banking and Insurance Analysis Division, and Dean Minot of theInsurance Policy Division.(1)

(1) The authors would like to thank Rupal Patel and Hamid Riaz for their help in producing this article.

Overview

In the United Kingdom, insurance companies operate on alarge scale: their investment holdings stood at about£1.9 trillion at the end of last year — a figure broadlyequivalent to nominal UK GDP. Insurance companiessupport the real economy, by enabling households and firmsto transfer the risks they face as well as by helping to channelsavings into investment.

By the nature of their business, insurance companies areexposed to risks. An insurance company’s distress or failurecan arise through inadequate provisions for claims orinsufficient capital to withstand unexpected losses eitherfrom insured events or volatility in the assets they hold. Thelikelihood of distress or failure may be heightened throughcertain activities insurance companies can engage in. Forinstance, AIG nearly failed in 2008 as a result of lossesarising from the sale of credit default swaps by one of itsnon-insurance affiliated entities and its securities lendingbusiness.

This article sets out two main ways in which insurancecompanies could have adverse effects on financial stabilityand focuses on the key channels for transmission of risks tofinancial stability rather than conjunctural risk assessment.

The first transmission channel relates to the risk that privateand commercial policyholders face disruption in the criticalfinancial services provided by insurance companies if one orseveral insurers fail. This might occur if resolutionarrangements and guarantee protection schemes did notprovide sufficient protection for policyholders againstinterruption to critical financial services, or if in the absenceof alternative providers, the failure of an insurance companydisrupted the provision of critical services to prospective

policyholders. This risk is more likely to arise when a smallnumber of insurance companies dominate supply in a sector.The second source of systemic risk stemming from theinsurance sector involves activities that propagate or amplifyshocks to financial counterparties or markets. There areseveral channels through which this could occur. Insurancecompanies could affect the resilience of their financialcounterparties if, on a significant scale, they stopped fundingthem, stopped lending them securities, or were unable tomeet claims following the occurrence of insured events.Evidence suggests that insurers can also behave in aprocyclical way, exacerbating the credit cycle (for instance byextending guarantees) or the volatility of financial markets.In practice, it is the second source of systemic risk that isconsidered more important for the UK insurance sector.

In the United Kingdom, insurance companies are regulatedby the Prudential Regulation Authority (PRA) and theFinancial Conduct Authority. The PRA’s objectives includepromoting the safety and soundness of insurance companiesand securing an appropriate degree of protection for thosewho are, or may become, insurance policyholders. Theinsurance sector is also an important consideration for theFinancial Policy Committee, whose objectives includeprotecting and enhancing the resilience of the UK financialsystem. There are a number of regulatory measures in placeto mitigate the risks discussed in this article and the Bank willcontinue to monitor systemic risks emanating from theinsurance sector. The Bank also supports internationalefforts to strengthen the resilience of insurance companiesoperating cross-border. This includes ongoing work towardsthe development of a global Insurance Capital Standard andwork to identify and address the potential risks posed byglobal systemically important insurers.

Topical articles Insurance and financial stability 243

The insurance sector plays an important role in the provisionof critical financial services. First, insurance cover allowshouseholds, corporations and public sector entities to transferrisks. For example, general insurance companies help firmsand households limit the financial costs associated with theoccurrence of various risks to their physical property, legalliability and miscellaneous financial loss. Second, insurancecompanies channel savings into investment. Life insurancecompanies, for example, help individuals to cover risks arisingfrom uncertainty about their health and lifespan, and one waythat they do this is by gathering funds from policyholders andinvesting these in debt, equity and other assets.

To perform these services, insurance companies operatebusiness models that differ from other financial institutions,including banks.(1) For instance, insurers collect premiumsupfront from policyholders for services which might only bedelivered several years later — the so-called ‘invertedproduction cycle’.

Insurance companies’ business models are exposed to a varietyof risks on both sides of the balance sheet.(2) The value of theassets of an insurance company could decrease following adeterioration in financial market conditions. Meanwhile, thevalue of the liabilities of an insurance company could increasesharply if it had underestimated the losses that might arisefrom the occurrence of an insured event such as a naturaldisaster. The failure of an insurer can occur when theincidence of risks reduces its financial resources below a viablelevel and it is subsequently unable to recover its position.

In the United Kingdom, the Prudential Regulation Authority’s(PRA’s) objectives include promoting the safety and soundnessof insurance companies and contributing to the securing of anappropriate degree of protection for those who are, or maybecome, insurance policyholders. The Financial PolicyCommittee’s (FPC’s) primary objectives are to identify,monitor and take action to remove or reduce systemic riskswith a view to protecting and enhancing the resilience of theUK financial system.(3)

Although the UK insurance sector is not as large as theUK banking sector, it is still large in relation to overalleconomic activity. There are currently about 600 insurancecompanies authorised by the PRA in the United Kingdom.(4)

The investment holdings of UK insurance companies werevalued at about £1.9 trillion at the end of 2014: this is about40% of the value of the assets held by UK banks and is broadlyequivalent to nominal UK GDP. The three largest insurancecompanies domiciled in the United Kingdom held total assetsestimated at about £1 trillion at the end of 2014 (compared to£4 trillion for the three largest UK-domiciled banks).

This article describes the ways in which insurance companiescould affect financial stability and contribute to systemic risk.

The aim is to explain the key channels, rather than to provide aconjunctural assessment of systemic risks posed by theUK insurance sector today. The article draws on varioussources of data to consider trends in insurance markets andthe size of certain exposures and interconnections. It also useshistorical examples to illustrate how some of these channelshave operated in the past. The framework guiding the analysisof financial stability — based on sources of fragilities andchannels of transmission of shocks — is consistent with theanalytical framework previously developed by the FinancialStability Board (FSB).(5)

Figure 1 summarises some of the key channels explained inthis article. The distress or failure of one or several insurancecompanies could affect financial stability if it led to adisruption to the critical services that insurance companiesprovide (Figure 1, column 1). This applies both to existing andprospective policyholders and is discussed in the first sectionof this article.

(1) See Thimann (2014).(2) See Breckenridge, Farquharson and Hendon (2014).(3) The Financial Conduct Authority (FCA), meanwhile, has responsibility for protecting

customers, enhancing market integrity, promoting effective competition, enforcingrules and fighting financial crime.

(4) This includes life and general insurance companies as well as Lloyd’s Syndicates. Inaddition, as of July 2015, there were about 800 European Economic Area (EEA)authorised insurers operating in the United Kingdom as a branch and/or on afreedom of services basis (Bank of England data, available atwww.bankofengland.co.uk/pra/Pages/authorisations/fscs/insurance.aspx).

(5) See Bank of England (2015a), in particular Box 5.

!

Risks to the real economy

Provision ofcritical services

Risks to the financial system

Risks tosystemicallyimportant

financial markets

Risks tosystemicallyimportant

counterparties

Insurancecover

Channellingsavings intoinvestments

Procyclicalreactions to

shocks

Exacerbation ofthe boom-bustcycle of assetprices in the

medium term

Via fundingtransactions

Via securitieslending

activities

Viareinsurancecontracts

Figure 1 Potential transmission channels of risks frominsurers to the real economy and financial stability

244 Quarterly Bulletin 2015 Q3

Insurers could also contribute to systemic risks(1) if theybehaved in ways which propagated shocks to the financialsystem.(2) This could occur if their actions threatened theresilience of a sufficient number of, or systemically important,financial counterparties (Figure 1, column 2) or if their actionscontributed to the disruption of the functioning ofsystemically important financial markets (Figure 1, column 3).This channel is discussed in the second section of this article.The final section of the article provides an overview of currentefforts by authorities to mitigate these risks.

Interruptions to the provision of criticalservices

The Bank of England’s FPC has identified three critical financialservices to the real economy performed by the financialsystem. These are: (i) payment services; (ii) channellingsavings into investments; and (iii) insuring against anddispersing risk.(3) Disruptions to these critical services canhave an adverse effect on financial stability.

The critical financial services provided by insurance companiessupport real economic activity. As a result, disruptions to theirprovision could affect output growth and financial stability.For instance, in the wake of the terrorist attacks of11 September 2001, insurance companies stopped offeringterrorism insurance and third-party insurance (protectingagainst damages caused to other parties) to airline carriers andbusinesses. Commercial aeroplanes cannot fly without theprovision of third-party insurance — but air transport plays acritical role in facilitating economic activity. So in response,the UK Government set up a replacement insurance schemecalled Troika, which operated for a year and preventedsignificant disruption both to the aviation industry and otherbusinesses.

Existing policyholders might be at risk of disruption to theircover in the event of an insurance company’s failure. As aresult, UK authorities have at their disposal a number ofoptions to manage the failure of a firm and protect existingpolicyholders. The measures available include placing a failedfirm into ‘run off’, encouraging an administrator to transferparts of the business to another insurer or, where necessary,safeguarding policyholders that are eligible for protectionguarantees via the Financial Services Compensation Scheme(FSCS). These options are discussed in more detail in the boxon pages 246–47.

The failure of an insurance company could also disrupt theprovision of critical services to prospective policyholders in theabsence of alternative insurance providers. This issue is morelikely to arise when one or a few insurance companiesdominate supply in a sector. This section discusses how highlevels of market concentration and low substitutability

between insurance providers could prevent prospectivepolicyholders from accessing critical services. It focuses firston the provision of insurance cover and then on channellingsavings into investment. The box on pages 248–50 discussesthe main sources of fragility for insurance companies,including leverage and underreserving, imperfect risk transfer,maturity mismatches and liquidity mismatches.

Interruption to the provision of insurance coverSeveral categories of general insurance cover offered byinsurers (also referred to as lines of business) play animportant role in supporting economic activity. Some formsof insurance, including motor and employer’s liabilityinsurance, are compulsory. Others, such as life, marine,aviation, goods in transit and property insurance are oftennecessary conditions to contracts underpinning economictransactions. When a bank offers a mortgage, for instance, ittypically requires the homebuyer to have insurance on theproperty. So for many insurance markets, a lack of sufficientsubstitutability between providers due to high levels ofconcentration could amplify the effect of an insurancecompany’s distress or failure on the real economy.

Concentration among insurance providers is one of the mainfactors that would affect prospective policyholders’ ability tofind alternative providers of insurance cover in the event offailure of an insurance company. Based on insurancepremiums received in 2014, several lines of business managedby general insurers collectively displayed some signs ofconcentration in the United Kingdom. Chart 1 shows that thethree largest insurance companies per class of business had amarket share of about 50% in several of those lines ofbusiness.

Another metric to assess concentration is the Hirsch-Herfindahl Index (HHI). The HHI is bounded between 0 (in highly competitive markets) and 10,000 (in thecase of a monopoly).(4) If the HHI is between 1,500 and 2,500,concentration is thought to be moderately elevated but is nottypically thought of as a source of major concern.(5) In 2014,the HHI was close to 1,500 for only three lines of businessoffered by UK general insurers: accident and health, aviationand goods in transit.(6)

These figures suggest that in the United Kingdom there is onlya moderate degree of concentration in a few lines of general

(1) Systemic risk in the financial system refers to chains of failures due tointerconnectedness between participants, either bilaterally or via financial markets.See Systemic Risk Centre, www.systemicrisk.ac.uk/systemic-risk.

(2) See Debbage and Dickinson (2013).(3) See Bank of England (2013).(4) The HHI is calculated by summing the squared value of market shares held by

individual firms active in a market. (5) See US Department of Justice and Federal Trade Commission (2010).(6) Similarly, concentration is also relatively low in the market for life insurance

companies’ protection products. This includes products which provide cover againstcritical illnesses and loss of income, for instance.

Topical articles Insurance and financial stability 245

insurance business at present. The absence of concentrationshould reduce the risk of discontinuity of cover in the event offailure of a single insurance company. The extent of anydisruption will also depend on other insurance companiesbeing willing and able to use their expertise to step in. In someof the more competitive segments (such as the UK motorinsurance market) some insurance companies might be willingto increase their market share and therefore dedicate morecapital to the activity. In addition, underwriting capacity in aspecific line of business might also bounce back if acontraction in supply led to increases in premiums. This mightalso encourage new firms to enter the market as expectedprofits increase.

Although there seems to be sufficient availability of generalinsurance providers in the United Kingdom to deliver adequatesubstitutability in the sector, this issue has been a problem inother countries. For example, the failure of HIH InsuranceGroup in March 2001 severely disrupted the provision ofmandatory builders’ warranty insurance for a period of almosta year in Australia. HIH failed as a result of underpricing,underreserving and poor corporate governance.(1) BecauseHIH underpriced risks, concentration in the builders’ warrantyinsurance market increased substantially.(2) Competitorsexited the market and potential new market entrants werediscouraged from entering the sector at the prevailing pricelevel. As a result, there were few alternative providers ofbuilders’ warranty insurance left. At the time of its failure HIHwas the second largest Australian insurer with the equivalentof nearly £3 billion of assets. This illustrates how underpricingof risk by an insurer can have spillover effects on the widerinsurance market.

Although the risk of a lack of substitutability in the event of apotential distress or failure by a single UK general insurance

company does not appear high, there is a possibility ofcontagion risk between firms with similar business models.For instance, recent research has found evidence that when aninsurance company announces losses related to operationalrisks, this has a negative spillover effect on the value of theshares issued by other insurance companies.(3) This could havean adverse effect on financial stability if failure by severalinsurance companies led to disruptions to the provision ofcritical financial services to the real economy.

Interruption to the channelling of savings intoinvestmentLife insurance companies channel savings into investmentprimarily by offering savings products to households andcorporates and typically investing the funds they collect intodebt or equity issued by firms and governments.(4) PRA dataindicate that approximately 90% of all insurers’ assets wereheld by life insurers and the remaining 10% by general insurersin 2014 — reflecting their distinct business models.

Life insurers offer ‘accumulation’ products, which enablepolicyholders to save, and ‘decumulation’ products, whichprovide a steady stream of income. Accumulation productsinclude pensions and other savings funds. Decumulationproducts include annuities, which provide a pre-definedincome stream for an agreed term. The sale of any of theseproducts can create three broad categories of liabilities for lifeinsurers: unit-linked liabilities (where the value of the liabilitytracks exactly the returns of the funds invested); ‘with-profits’liabilities (where profits are added periodically to the fund andguaranteed by the insurer, so that liabilities are linked to butcan differ from returns on invested funds); and non-participating liabilities (where the value of the liabilitydoes not relate to the performance of investments). Table Aprovides an overview of UK life insurance companies’ liabilitiesby selected categories of accumulation and decumulationproducts in 2014.

(1) See Royal Commission (2003).(2) HIH held a market share of 90% meaning the Hirsch-Herfindahl Index in the builders’

warranty market reached at least 8,100. (3) See Cummins, Wei and Xie (2012).(4) Both life and general insurers invest in a broad range of assets. But general insurers

do not offer savings products and tend to invest in shorter-term assets than lifeinsurers.

0

20

40

60Market share, per cent

Largest provider

Top three providers

Acci

dent

and

heal

th

Avia

tion

Goo

ds in

tran

sit

Com

mer

cial

mot

orCo

mm

erci

alfin

anci

al lo

ssPe

rson

alfin

anci

al lo

ssPe

rson

alm

otor

Com

mer

cial

liabi

lity

Hou

seho

ld a

nddo

mes

tic ri

sks

Com

mer

cial

prop

erty

Mar

ine

Chart 1 Concentration of the main lines of businessoffered by general insurers

Source: PRA regulatory returns (December 2014).

Table A UK life insurance liabilities (in £ billions) by selectedcategories of products

Unit- Non-Products linked participating With-profits Total

Corporate pensions 572 0 20 592

Individual pensions 284 1 89 374

Savings and investments 183 4 84 271

Annuities 0 240 29 269

Total 1,040 246 221 1,506

Source: PRA regulatory returns (December 2014).

246 Quarterly Bulletin 2015 Q3

Managing firm failures

This box provides an overview of some of the options availableto UK authorities to mitigate the risk of disruption of cover forpolicyholders when an insurance company is in distress or fails.It also discusses some challenges which could complicate themanagement of a failed insurance company. Policies toreduce the likelihood of insurance companies failing, includingSolvency II, are discussed in the final section of this article.

What happens when a firm fails? When an insurance company operating in the United Kingdomis no longer viable, authorities can intervene swiftly byremoving its permission to enter into new contracts (thecompany cannot continue to write new business in suchcircumstances). This involves so-called ‘run-off’. Run-off ispart of a suite of possible recovery and resolution tools aimingat securing finality for policyholders with actual or potentialclaims.

There are two types of run-off: solvent run-off and insolventrun-off. Whether a run-off is solvent or insolvent will dependon whether an insurance company’s reserves and capital aresufficient to pay expected claims to policyholders (typically,this will only be known ex post). If reserves and/or capital aresufficient, the run-off will be solvent and the aim of the run-off will be to settle all actual and potential policyholders’claims. Firms or individuals seeking new cover can do so,where available, through alternative insurance providers. Asdiscussed in the main text of the article, this is easier whenconcentration is lower and alternative products or providersare readily available. There are currently over 100 insurancecompanies in solvent run-off in the United Kingdom.

Part VII transfer arrangements can also be utilised to transferbusiness to any UK or other European Economic Area (EEA)insurer. So if it is possible to find an insurance companywilling to take on a failed insurer’s business, existingpolicyholders might not face disruption to their cover.However there might be circumstances under which thismight be difficult to achieve, for instance when the value ofassets and liabilities of a failed insurer are highly uncertain.

What happens if the money runs out?If a failed insurance company’s liabilities exceed its availableassets — or it is unable to meet policyholders’ claims as theyfall due — it may be placed into provisional liquidation oradministration or may propose an insolvent scheme ofarrangement under the Companies Act 2006 and enter‘insolvent run-off’. In the United Kingdom, there are currently26 general insurers and two small life assurance firms ininsolvent run-off with protected claims.

The Financial Services Compensation Scheme (FSCS) is thecompensation fund of last resort for customers of authorisedfinancial services firms in the United Kingdom. The FSCS willsafeguard eligible policyholders if they have a protected claimunder a contract of insurance issued by an authorised insurerthrough an establishment in the United Kingdom, another EEAstate, the Channel Islands or the Isle of Man.

For claims relating to general insurance contracts, eligiblepolicyholders include most private individuals and smallbusinesses. For claims relating to long-term insurancecontracts (for instance life insurance), eligible policyholdersinclude most private individuals and businesses of all sizes.Provided the claim is made under a protected contract ofinsurance, there are no exclusions from FSCS eligibility forclaims under a compulsory insurance contract (such as third-party motor and employers’ liability for instance).

The FSCS compensates 100% of claims that arise under long-term insurance and compulsory insurance (motor andemployers’ liability) contracts, and since July 2015 hasincreased the compensation limit for professional indemnityclaims (from 90% to 100%) and introduced a new category ofclaims covered at 100% (death and incapacity claims due toinjury, sickness or infirmity). It compensates 90% for claimsarising from other types of general insurance policies.(1) TheFSCS provides an important safeguard to ensure continuity ofinsurance provision or compensation to eligible policyholdersand, thereby, reduces risks to financial stability and to the realeconomy.

FSCS protection for insurance contracts is funded by levies onfirms authorised by the Prudential Regulation Authority (PRA).Each insurer’s contribution is calculated on a tariff basis andinsurers contribute predominantly in proportion to theirrelevant net premium income. But there is an upper limit (setby the PRA) of funds available each year, so it is possible theFSCS could run out of funds in extreme circumstances. It mayfor example be a challenge for the FSCS to fully absorb thelosses arising from the failure of a large life insurer if no otherinsurer were willing to take over its assets and liabilities. Inthese circumstances, there are provisions for the FSCS toborrow commercially or from the UK Government (under theNational Loans Fund) and for the insurance industry (throughthe FSCS) to repay the loan over a long period.

Following the failure of HIH (see main text), the Australiangovernment introduced a claims support scheme broadlyequivalent to the United Kingdom’s FSCS. The Australianscheme paid out claims worth more than £86 million between2001 and 2003.(2) This illustrates the importance of such

(1) See FSCS webpage ‘What we cover’: www.fscs.org.uk/what-we-cover/eligibility-rules/compensation-limits/insurance-limits/.

(2) See http://archive.treasury.gov.au/content/hih_claims.asp?titl=HIH&ContentID=689.

Topical articles Insurance and financial stability 247

schemes in order to prevent discontinuity in critical servicesfor existing policyholders.

Are there outstanding challenges?Unlike for deposit-takers, there is no statutory resolutionregime in the United Kingdom for failed insurance companiesas an alternative to ordinary insolvency. Even though thecurrent regime to manage insurance companies’ failures in theUnited Kingdom is generally robust, there remain somepotential challenges.

For example, even though the FSCS covers compulsoryinsurance cover such as third-party motor and employers’liability, it does not cover reinsurance, marine, aviation,transport business and credit insurance. These are importantlines of business for the real economy (see main text) and thefailure of a large provider could therefore lead to disruptions

for existing policyholders if no competitor were willing toacquire the portfolio of policies underwritten.

Also, a court-led insolvency process retains the risk that legalchallenges may lead to disruption to the continuity ofpayments made to policyholders. Finally, entry intoadministration could trigger contractual rights for derivativecounterparties to exercise early termination rights and close out contracts (including cross-default clauses affectingother group companies), which could lead to losses andrequire the insurer to find alternative ways to hedge itsportfolios.

The Bank of England and other UK authorities are workingwith international partners to ascertain whether the currentframework for dealing with insolvent insurance companiesprovides adequate protection or needs to be reviewed.

At the end of 2013, UK life insurers held £1.6 trillion of assets,or about 12% of the total assets held by all financialinstitutions operating in the United Kingdom. As a result, lifeinsurance companies provide important intermediationservices to the wider economy. By way of comparisonUK banks held about £5 trillion of assets at the end of 2013,and pension funds held £1.4 trillion.(1)(2) In 2014, the PRAcollected data on a subset of assets held by insurancecompanies; Chart 2 shows these were invested in bonds,equities and other asset classes.(3)

Data from the Association of British Insurers on premiumincome indicate that concentration in the UK life insurancesector is relatively low in aggregate. Using data on netpremium income across all accumulation and decumulationproducts by the top 20 life insurance companies, the

estimated Hirsch-Herfindahl Index reached only about 800 atend-2013 (down from 850 the year before).(4)

Given the relatively low levels of concentration, the distress orfailure of a life insurer is not likely to prevent potential newpolicyholders from finding alternative suppliers of savings ordecumulation products within the insurance industry. Inaddition life insurers have retreated away from ‘with-profits’products since the early 1990s and instead have focused on unit-linked products where policyholders own shares (or so-called ‘units’) of funds and bear the risks arising from theinvestment. The proportion of UK life insurers’ long-termliabilities related to with-profits products decreased from 46%in 1991 to about 15% in 2014 (Table A).

The implication of this shift towards unit-linked products isthat the savings products offered by life insurers are now moresubstitutable: they resemble products offered by otherfinancial institutions including shares of funds managed byasset managers. Asset managers can compete with insurancecompanies in the savings products market and they offer unit-linked products which have attracted strong inflows.Between 2005 and 2014, asset managers’ total assets under

0

10

20

30

40

50

60

70

80

90

100

2005 06 07 08 09 10 11 12 13 14

Equities

Corporate bondsGovernment bonds

Property

Other

Per cent

Chart 2 UK life insurers’ asset allocation(a)

Source: PRA Life Assets data collection (December 2014).

(a) This excludes assets related to unit-linked products.

(1) See Burrows, Cumming and Low (2015).(2) In addition, UK life insurance companies held a larger proportion of their assets than

banks in government and corporate bonds. For instance, corporate bondsrepresented 23% of insurance companies’ investment holdings at the end of 2013,compared to 4% for major UK banks at the end of 2014 (PRA data). As a result,insurance companies are relatively more important investors in corporate bondmarkets.

(3) The data collection included assets valued at about £572 billion but did not includeassets related to unit-linked products.

(4) This is the case despite insurance companies remaining the main providers of ‘with-profits’ products and annuities. Only life insurers can sell annuities, soconcentration in this product is relatively higher, with Aviva, Legal & General andPrudential collectively holding a market share of about 45% in 2012. But‘with-profits’ products represent a relatively small and decreasing proportion of lifeinsurance companies’ businesses.

248 Quarterly Bulletin 2015 Q3

Potential fragilities that can make insurersmore likely to fail

This box identifies some key sources of fragility for insurancecompanies’ resilience to shocks: leverage and underreserving;imperfect risk transfer; and maturity and liquiditymismatches. The Financial Policy Committee has asked Bankand Financial Conduct Authority staff to further assess therisks from a number of activities which give rise to some of thefragilities discussed in this box.

Leverage, underpricing and underreservingInsurance companies typically issue little debt relative to thevalue of the assets on their balance sheet and their equity. Forexample, the debt to equity ratio of UK insurance companies,a measure of financial leverage, reached about 0.4 at the endof 2013.(1) For comparison, the same ratio for major UK banksdecreased from about 3.1 at the end of 2010 to about 1.7 atthe end of 2014.

However, insurance companies can achieve an effectanalogous to leverage by underreserving for future liabilities.This enables insurers to write a greater volume of contractsthan would otherwise be possible. But it reduces the amountof capital per unit of risk on the balance sheet and thereforerenders insurance companies more vulnerable to shocks.

There have been several instances in the past where reserveswere insufficient to withstand shocks and insurancecompanies failed. This issue typically arose when: insurancecompanies competed intensively on price and premiumsbecame too compressed to accumulate reserves; or shockswere much larger than expected and existing reserves wereeroded as a result of claims.

Fire, Auto and Marine (FAM) and Drake Insurance areexamples of UK insurance companies which failed mainly as aresult of underpricing leading to insufficient capital. TheJapanese insurance company Taisei Marine and Fire failed dueto a much greater-than-expected loss, underreserving,inadequate reinsurance and poor returns. These firms wererelatively small and so their failure had a limited impact onfinancial stability or the aggregate level of real economicactivity. But these examples show that underpricing andunderreserving can cause insurance company failure.

Risks arising from imperfect risk transferImperfect risk transfer refers to a situation where risks believedto have been transferred to another party still lead to lossesfor the party initially transferring the risks.

Exposures to intragroup entitiesInsurance companies could be exposed to the risks supposedlyborne by affiliated entities. This risk crystallises when theresources of the affiliated entities are insufficient to meet theirliabilities. For instance, in the years prior to the financial crisis,the non-insurance subsidiaries of both AIG(2) and Swiss Re soldsubstantial amounts of protection on collateralised debtobligations. Following the defaults of many issuers ofstructured debt securities, both subsidiaries were unable tomeet the obligations arising from the protection they sold andthe losses then became a threat to the wider groupmembership. The US authorities saw the potential disorderlyfailure of AIG as a material threat to financial stability, in partbecause it was deeply interconnected with financial markets.(3)

AIG had to be rescued and it benefited from almost £49 billionfrom the US Treasury (via the Troubled Asset Relief Program)and £78 billion committed by the Federal Reserve Bank ofNew York. Separately, Berkshire Hathaway acquired Swiss Reshares worth about £1.6 billion.

Risks posed by intragroup exposures are more complex andlikely to be greater when subsidiaries and branches aredomiciled across multiple jurisdictions, or when insurancecompanies are parts of larger groups such as bank-assurancegroups.(4) For example insurance companies might enter intointragroup transactions such as reinsurance arrangements orparental guarantees, sometimes in order to improve capitaland tax efficiency.(5)

Risks arising from maturity mismatchesThe average expected maturity of insurance companies’ assetsmay be shorter than their liabilities, giving rise to a maturitymismatch and the so-called positive ‘duration gap’.(6) Ifinsurance companies do match the duration of their assets andexpected liabilities, both the stream of income yielded by theassets held and the maturing assets will become availablewhen liabilities arise, which would also be consistent withSolvency II’s Prudent Person Principle.(7) Table 1 provides anoverview of the average duration gap in the insurance sectorof selected European countries.

(1) Annual reports data based on a sample of nine large firms.(2) See McDonald and Paulson (2014).(3) See www.treasury.gov/initiatives/financial-stability/TARP-Programs/aig/Pages/

default.aspx. (4) Losses could spill over and complex group structures could also make resolution

more difficult. (5) Insurers hold capital against intragroup exposures and counterparty risk, both under

ICAS and Solvency II regimes, potentially mitigating these risks to some extent.(6) There is a maturity mismatch when the maturity of assets is different from the

maturity of liabilities. If the maturity of liabilities is greater than the maturity ofassets, the duration gap is positive. If the maturity of liabilities is lower than thematurity of assets, the duration gap is negative. A negative duration gap can alsogive rise to risks, for instance liquidity risks.

(7) For instance, the pool of available long-dated assets may not be suitable from arisk/return perspective.

Topical articles Insurance and financial stability 249

Reinvestment risk in the presence of guaranteed returnsThe presence of a positive ‘duration gap’ may not be an issuein isolation, but it means that insurance companies with long-term liabilities need to reinvest in new assets wheninvestments mature. Threats to an insurance company’ssolvency can therefore arise if it has issued guaranteed returnson liabilities that are higher than the yield on assets availablewhen the former investments mature. Equitable Life’s failurepartially illustrates this point.

The International Monetary Fund (2015) and the Bank ofEngland (2015a) have highlighted that there are a number ofcountries where the domestic insurance sector as a whole isexposed to reinvestment risks due to a duration gap,guaranteed returns and low current yields, potentially givingrise to financial stability concerns.(1) For example, estimatesfrom the Deutsche Bundesbank have shown that 12 out of85 German life insurance companies would not be able tomeet Solvency I ‘own funds’ requirements if returns on assetsremained around 2.5% between 2015 and 2023. And 32would be at risk of defaulting if yields progressively trendedtowards 1.5% over the same period.(2)

Risks arising from liquidity mismatches Several of the activities insurance companies undertake —including securities lending activities, the sale of life insuranceproducts with flexible redemption options, or reliance onshort-term funding — can create liquidity mismatches and socould increase risks. A liquidity mismatch arises when theliquidity of assets differs from the liquidity of liabilities.Liquidity mismatches can give rise to illiquidity risk, which isthe risk that a company cannot meet its short-term liabilities.

Securities lending activitiesInsurance companies hold large portfolios of assets, often forlong periods of time. In order to boost their revenues,insurance companies sometimes lend, for a fee, securities such

as shares or bonds to selected counterparties. Securitieslending transactions are typically collateralised (against cashor other securities) and agreed upon with open maturities.This means that the lender has the right to recall the securitieson demand while the borrower has the right to return thesecurities borrowed at will.

These contractual features expose insurance companies toliquidity risks if borrowers return securities unexpectedly andin large amounts, for instance because they doubt the lender’screditworthiness.(3) Insurance companies would then have toreturn the cash collateral posted by securities borrowers atshort notice, and might struggle to meet this demand forliquidity if the cash collateral had been invested in less liquidassets.(4) This situation reflects AIG’s experience in 2008: itwas unable to meet collateral calls arising from its nearly£40 billion securities lending program and experienced amaterial cash drawdown, totalling around £17 billion duringthe second half of September 2008.(5)

Flexible redemption options on life insurance productsLife insurance companies sometimes sell policies embeddingflexible redemption options. For instance, some savingsproducts enable policyholders to access their funds at shortnotice.

These products give policyholders some control over thematurity of insurance companies’ liabilities, which increasesliquidity risks.(6) For instance when the Asian currency crisis of1997–98 affected the Republic of Korea, interest ratesincreased from 12% to 30% in December 1997. Holders ofpolicies indexed on lower rates redeemed their policies so theycould earn higher returns. These lapses forced Korean lifeinsurance companies to fire sale assets and many firms facedshortages in liquidity.(7) And a recent paper has estimated thatsome German life insurance companies could be exposed toruns by policyholders following a sudden increase in interestrates of around 2 percentage points.(8)

(1) The International Association of Insurance Supervisors (IAIS) has recently describedthe low-yield environment insurance companies currently operate in. See IAIS(2014).

(2) See Kablau and Weiss (2014).(3) Borrowers have an incentive to return the securities borrowed at short notice if they

doubt the viability of counterparties. This is because borrowers provide collateraland the value of this collateral exceeds the value of the securities on loan. This givesrise to counterparty credit risk.

(4) Raising cash by lending securities and reinvesting this cash in other financialinstruments also creates leverage risks.

(5) See Congressional Oversight Panel (2010).(6) For a discussion of triggers of lapses, see Kiesenbauer (2011). Although an insurance

company can charge a fee to policyholders redeeming their policies, this may not besufficient to deter policyholders from early redemptions under extremecircumstances.

(7) Gross written premiums fell from US$47 billion in 1996 to US$35 billion in 1998.(8) See Förstemann and Feodoria (2015).

Table 1 Duration gap in selected major EU life insurance markets

Country Duration gap(a)

Germany >10 years

Sweden >10 years

Austria >10 years

Netherlands 5½ years

France 4¾ years

Denmark 4¾ years

Spain <1 years

Italy <1 years

Ireland <0 years

United Kingdom <0 years

Sources: European Insurance and Occupational Pensions Authority, Moody’s Investors Service,Standard & Poor’s Ratings Services (May 2014) and Bank calculations.

(a) Duration gap is the difference between the average duration of liabilities and assets.

250 Quarterly Bulletin 2015 Q3

Reliance on short-term funding Like other financial institutions, insurance companies issuedebt securities to fund part of their activities.(1) Supervisoryintelligence suggests UK insurance companies typically issuelong-term securities. Nevertheless, some insurancecompanies use short-term debt financing, including for thepurpose of financing long-dated and illiquid assets. Forinstance, US life insurers relied upon the equivalent of£11.5 billion of ‘extendible funding agreement’ backed notes in2007 (a type of security that converts into short-term paperon selected ‘election dates’ if investors opt not to extend thematurity of their holding). And a study conducted by the

Financial Stability Board in 2013 showed that global insuranceand pension firms had repo exposures totalling approximately£27 billion to 17 financial institutions as of year-end 2012.(2)

As a result insurance companies could in theory face liquidityissues if short-term funding markets were to becomeimpaired.

(1) At the end of 2014, the proportion of debt securities issued or loans represented9.5% of the liabilities of UK insurance companies in aggregate, based on SNLFinancials data for a sample of 42 firms. So this does not appear to be a material riskfor UK insurers.

(2) Financial Stability Board (2013). This value is estimated based on 4% ofUS$1.1 trillion of reverse repo.

management increased from £350 billion to almost£850 billion.(1) And other financial institutions in theUnited Kingdom offer a range of substitutes for savingsaccumulation products.(2)

Nevertheless, existing policyholders could be at risk ofdiscontinuities in cover or disruption in payments if a lifeinsurer fails. The risk of disruption to payments of annuities isof particular concern. As an example, no firm acquiredEquitable Life when it faced unexpected claims arising fromguarantees attached to certain annuity products.(3) As aconsequence Equitable Life closed to new business.(4) Thevalues of certain policies were adjusted through acourt-sanctioned compromise scheme and policyholderswishing to surrender their policies early faced administrativedelays. The Government acknowledged that somepolicyholders had suffered financial losses as a result ofGovernment maladministration and agreed to paycompensation as a consequence.(5)

Amplification of shocks to the financialsystem and systemic risk

To perform the critical services discussed in the previoussection, insurance companies frequently engage intransactions with counterparties such as banks and otherfinancial institutions. There are many types of financialcontracts that insurance companies can enter into andsecurities they trade. Some, including securities lendingcontracts, usually give rise to short-term obligations. Others,such as investments in debt securities or some categories ofderivatives contracts (for example long-dated interest rateswaps) can give rise to much longer exposures.

This means that insurers could propagate or amplify shocks tocounterparties or markets through their individual andcollective actions (columns 2 and 3 on Figure 1). This sectiondescribes the channels through which shocks could transmitfrom the insurance sector to systemically important financialcounterparties, and then to markets.

Direct disruptions to systemically important financialcounterpartiesInsurance companies could have the potential to directlyaffect the resilience of systemically important financialcounterparties if, on a significant scale: they stopped fundingthem; stopped lending them securities; or were unable tomeet claims to them following the occurrence of insuredevents. In particular, banks are at the core of the financialsystem and so exposures between insurance companies andbanks could be particularly important. Recent analysis by theEuropean Systemic Risk Board has shown that there is asignificant degree of interconnectedness between the bankingsector and the insurance sector in Europe.(6)

Transmission via funding transactionsBanks and other major financial institutions typically raisefunds by issuing shares and debt securities. They can also doso by entering into collateralised funding agreements such asrepurchase agreements transactions (‘repo’). In a repo,borrowers sell a security or basket of securities against cashand agree to repurchase these at an agreed date and price.

Insurance companies hold large amounts of debt securitiesand shares issued by financial institutions. For instance, theInternational Monetary Fund estimates that European insurers

(1) See Investment Association statistics. These figures can include the assetmanagement arm of insurance companies: www.theinvestmentassociation.org/investment-industry-information/fund-statistics/funds-under-management.html?what=graph&show=3.

(2) Banks and building societies also offer savings accounts (including ISAs). A study bythe FCA showed outstanding balances in these accounts reached over £700 billion atthe end of 2013. See Financial Conduct Authority (2015).

(3) See Roberts (2012).(4) See Parliamentary and Health Service Ombudsman (2008).(5) As discussed in the box on pages 246–47, the FSCS is the compensation fund of last

resort for customers of authorised financial services firms in the United Kingdom.The FSCS will safeguard eligible policyholders if they have a protected claim under acontract of insurance issued by an authorised insurer through an establishment in theUnited Kingdom, another EEA state, the Channel Islands or the Isle of Man. Thepolicyholder protection rules place an obligation on FSCS to initially seek continuityof cover and to minimise the disruption in payments for eligible policyholders. As of3 July 2015 protection is 100% for claims relating to long-term (life) insurance. Inthe case of Equitable Life, FSCS protection was not triggered because of the specificcircumstances at the time leading to Equitable Life not being declared in default.

(6) See European Systemic Risk Board (2015).

Topical articles Insurance and financial stability 251

hold around 30%–35% of bonds issued by European financialfirms.(1) And in the United States, insurance companies heldabout £195 billion of corporate bonds issued by the financialsector, about 50% of which were issued by banks (Chart 3).Insurance companies also place deposits with banks and someinsurers lend cash to banks via repo.(2)

So there is a risk that if insurance companies allocated theirequity and debt financing portfolios away from the financialsector, financial institutions could face significant fundingdifficulties. A quick reallocation would be more likely toreflect counterparty credit risk fears. In this case, banks wouldalready be in difficulty and actions by insurers mightexacerbate the problem. Insurance companies could alsoreallocate their assets away from the financial sector if theyhad concerns about the concentration of their exposure tofinancial counterparties. But this would more likely be agradual transition process. Data collected by the PRA showthat at year-end 2013 in the United Kingdom, the share ofsecurities issued by banks represented 16% of life insurers’corporate bond portfolios and 10% of their equity portfolios.Importantly, these figures exclude investments through unit-linked investments funds, which represent about two thirds of life insurance companies’ balance sheets(Table A), and so constitute a lower bound estimate.

Transmission via securities lending activitiesLending securities such as shares to counterparties for a fee isa valuable service, but it also significantly increasesinterconnectedness between financial counterparties.(3)

Frequent borrowers of securities include banks, broker-dealersand investors such as hedge funds.

UK insurance companies are very significant participants inUK securities lending markets. They also have the capacity toincrease the volume of securities they lend: in May 2015 only

10% of securities eligible for lending were on loan.(4) As aresult, UK insurance companies could potentially propagateshocks to borrowers of securities if they collectively recalledsecurities on loan unexpectedly.(5)

Transmission via reinsurance contractsReinsurance companies are an important part of the insurancesector. They typically operate globally, with the total marketfor reinsurance reaching £367 billion in 2014 according toSwiss Re Sigma data. In comparison, total premium incomewas £3 trillion for insurance companies globally.(6)

Reinsurers offer insurance to general and life insurancecompanies. This helps insurance companies manage the risksthat they underwrite and spreads their liabilities, for instanceby sharing risks geographically or across product lines. Forexample, an insurance company selling storm protection in agiven region can mitigate the concentration risk it is exposedto by purchasing reinsurance.

When an insurance company purchases reinsurance, it retainsthe obligation to pay claims on the contracts it has written.But the purchase of reinsurance also creates ‘reinsuranceassets’ which pay-off when the insured event occurs. Thiscompensates the original insurer for the loss and helps it meetits claims obligations. As a result the original insurancecompany is therefore exposed to the counterparty credit riskof its reinsurance companies.

The distress or failure of a large reinsurer could have materialconsequences for insurance firms. For instance, generalinsurance companies transferred about 20%–25% of the riskthey underwrote to reinsurance providers in 2014. This couldcreate risks if these exposures were concentrated. Goodpractice for insurance companies is to spread their reinsuranceacross several providers, and research indicates that theinsurance sector in advanced economies appears to be resilientto the risk of reinsurance companies defaulting on theirobligations.(7)

An alternative way to obtain reinsurance is by issuinginsurance-linked securities (ILS). ILS are instruments whichenable the transfer of specific catastrophe insurance risks via

(1) See International Monetary Fund (2015).(2) Based on supervisory data, UK life insurance companies held deposits worth over

£6 billion with banking counterparties included within their top ten counterparties in2014. Large exposure data collected from banks show that a few insurancecompanies lent large enough amounts through repo markets to appear in banks top20 counterparties. However insurers active in the Unites States were not typicallylenders in repo markets: see National Association of Insurance CommissionersCapital Markets Bureau (2014a).

(3) See Financial Stability Board (2012).(4) Source: Markit Group Limited.(5) For instance, if recalling securities on loan forced securities borrowers to liquidate

their positions, this could force the latter to recognise losses, thereby potentiallyaffecting their resilience.

(6) See Swiss Re (2015).(7) See van Lelyveld, Liedorp and Kampman (2009) and Park and Xie (2014).

0

20

40

60

80

100

120

140

Bank Non-bank financial Insurance

Equity holdings

Long-term corporate bonds

£ billions

Chart 3 US insurers’ holdings of equities and bondsissued by financial institutions

Sources: National Association of Insurance Commissioners Capital Markets Bureau andBank calculations (December 2013).

252 Quarterly Bulletin 2015 Q3

capital markets.(1) Providers of so-called ‘alternative capital’(because it is a source of funding from outside theconventional reinsurance market) include participants such aspension funds, hedge funds and institutional investors. Theseinvestors have been attracted to the potentially higher returnsoffered by ILS instruments as a consequence of limited claimsand losses arising from severe natural catastrophe events inrecent years. For instance around £40 billion of ILS wereoutstanding globally at the end of June 2015, distributing riskto other parts of the financial system.(2) This increasesinterconnectedness between the insurance sector and otherfinancial intermediaries because investors holding ILS will besubject to the same risk of loss from catastrophic events asinsurers. Insurance companies might also progressivelybecome more dependent on alternative capital providers forreinsurance. The FPC has asked Bank and FCA staff to assessthese risks in more detail over the course of 2015 and intoearly 2016.

Disruptions to systemically important financialmarketsOne of the main channels through which insurance companies— and other large investors — could potentially disruptsystemically important financial markets is procyclicalbehaviour. This refers to the tendency to act in a way thatexacerbates observed changes in the prices of financial assets.Another channel includes providing services which mightexacerbate the credit cycle in the medium to long run, forinstance by extending financial guarantees.

Procyclical behaviourInvestors behave procyclically when they increase theirexposure to an asset class when its value is rising. Inaggregate, such behaviour might contribute to the formationof asset price ‘bubbles’ where prices of financial assets nolonger accurately reflect their risks. As a consequence, asudden reappraisal of risks could lead to sharp declines in assetprices. Reciprocally, investors behave procyclically if they sellassets when the prices of these assets are declining. Thismight occur because investors want to sell and invest in saferassets rather than risk further losses, or they may face liquidityconstraints, forcing them to sell their assets (so-called ‘fire sales’).(3) Although not a behaviour that is unique toinsurance and while such actions may reflect prudent riskmanagement on the part of individual insurers, when asignificant proportion of investors behave in such a way, it canamplify shocks.(4)

The procyclical behaviours described above could exacerbatethe tendency for financial markets to experience ‘booms’ and‘busts’. For instance, marking-to-market of instruments onfinancial institutions’ balance sheets can affect firms’ capitalpositions; fluctuations in the value of collateral posted insecurities financing and derivatives transactions can lead tocollateral calls and propagate liquidity shocks; and

‘flight-to-safety’ behaviour (whereby market participantsinvest only in low-risk assets) can decrease the interest rate onrisk-free assets used to value liabilities — thereby affecting thesolvency of firms with long-term insurance liabilities.

Insurance companies typically hold long-term and mostlyilliquid liabilities.(5) In principle this should enable them tolook through short-term fluctuations in asset prices. Onemight therefore expect their behaviour to dampen — or atleast not to amplify — shocks.(6) For instance, should assetprices deviate substantially from fundamentals as a result offire sales by other investors, long-term, value-driven investorssuch as insurance companies could in theory seize the long-run profit opportunities that arise from such deviations,or at least not feel the pressure to sell.

There is some tentative evidence of procyclical behaviour byinsurance companies, both internationally and in theUnited Kingdom.(7) In the United Kingdom, there wasevidence of procyclical shifts in asset allocation away fromequities and into fixed-income assets following the dotcomcrash of the early 2000s. In the United States and France,there was some evidence of insurance companies changingtheir allocations to equities in correlation with theperformance of the S&P 500. And the Nederlandsche Bankhas found evidence of Dutch insurers selling their holdings ofsovereign debt issued by ‘peripheral’ European governmentsafter their yields increased sharply in 2012–13 due toincreasing risks of default.(8)

There are a number of potential sources of procyclicalbehaviour and the box on page 253 discusses these further,focusing in turn on asset allocation strategies, riskmanagement practices, potential ‘runs’ on insurers, the use ofderivatives and some valuation techniques.

Regulators have taken a range of measures in the past to curbthe risks of fire sales of assets by insurers and reduce thelikelihood of procyclical behaviour. This might explain therelatively limited evidence available despite the number ofpotential sources of procyclical behaviour. Among others,these measures included: relaxing capital requirements (sothat firms would not have to hold increasing amounts ofcapital when volatility increased); dampening volatility invaluation (enabling firms to temporarily value assets orliabilities differently to the market price); and introducing

(1) There are three main structures of ILS: collateralised reinsurance, catastrophe bonds(including sidecars) and industry loss warranties.

(2) Data source: Aon Benfield, Reinsurance market outlook (June and July 2015 update).(3) See Shleifer and Vishny (2011).(4) See Brunnermeier and Pedersen (2009).(5) See, for instance, Box 1 in Breckenridge, Farquharson and Hendon (2014).(6) Governor Carney recently highlighted the positive role insurance companies can play.

See Carney (2014).(7) See Bank of England and Procyclicality Working Group (2014).(8) See Bijlsma and Vermeulen (2015).

Topical articles Insurance and financial stability 253

Potential sources of procyclical behaviour

There are a number of factors that could incentivise insurers tobehave procyclically. These include their own risk appetiteand risk management practices, potential ‘runs’ on insurers,following benchmark-driven investment strategies, the use ofderivatives as well as some valuation techniques.

Some risk management practices could lead insurers tobehave procyclically. For instance, there are some variableannuity products that give insurers the right to modify thecomposition of their portfolios from equities to bondsautomatically when the prices of equities fall substantially andvolatility increases. If insurers with similar portfolios allimplement such risk management practices simultaneously, inaggregate this could further contribute to declines in assetprices. The extent to which this channel is likely to deliversizable asset reallocations is difficult to estimate due tolimited data, but supervisory intelligence suggests that there isthe potential for this risk to crystallise.

Meanwhile, ‘runs’ on insurers can in principle arise whenpolicyholders are able to cash out their policies at their owndiscretion (as is for instance the case for some variable annuitycontracts even if this sometimes involves withdrawal fees).(1)

This might especially be the case if holders of savings productscollectively reacted to market stresses by withdrawing theirfunds. There have been instances of ‘runs’ on insurance firmsinduced by broader market stresses: Ethias, a Belgian insurer,required a capital injection of the equivalent of £1.1 billionfrom the Belgian government following a high number ofcancellations of policies and withdrawals of savings during themarket turmoil in 2008.(2) This did not affect UK insurers asthey had not sold similar products, including in theUnited Kingdom.

If the insurance sector widely uses common investmentstrategies based on common benchmarks or indices, this couldaffect the value of the securities within the benchmarks or thecomponents of indices. For instance, the InternationalMonetary Fund has shown that there was increasing evidenceof correlated trading (or ‘herding’) among equity and bondfunds investing in emerging markets.(3) This behaviour wasnot directly linked to insurance companies but as discussed inthe second section insurers are significant investors.

Finally, the insurance industry can use derivatives instrumentsin order to hedge risks, match assets and liabilities andmanage their portfolios efficiently. For instance, foreignexchange derivatives enable insurance companies to protectagainst variations in exchange rates. The gross notional valueof derivatives held by insurance entities with operations in theUnited Kingdom reached £400 billion at the end of 2014. The

comparable figure for US insurance companies reached£1.12 trillion at the end of 2013.(4) Chart A shows thedistribution of UK insurers’ derivatives portfolios by type ofderivatives instruments.

Although using derivatives for risk reduction or efficientportfolio management purposes can be effective and prudent,it also creates a dependency on being able to access derivativemarkets. Insurers’ ability to access derivatives could beimpaired in times of wider market stress, for instance as aresult of increasing concerns about counterparty credit risk.(5)

And so a decline in the availability of derivatives used forhedging purposes could force insurers to reallocate some oftheir assets in times of market stress.

Solvency II seeks to mitigate the risk of insurance companiesbehaving procyclically via dampening mechanisms such as theequity symmetric adjustment, the matching adjustment andthe volatility adjustment.(6) A recent study has suggested thatthe equity symmetric adjustment should be effective atreducing procyclicality of the standardised equity risk capitalcharge.(7) But further work and analysis will be necessary tojudge the effectiveness of these mechanisms.

Interest rates (53%)

Equity (9%)

Foreign exchange (22%)

Inflation (5%)

Credit (4%)Other (5%)

Chart A Types of derivatives instruments used byUK insurers

Source: PRA regulatory returns (December 2014).

(1) The US Financial Stability Oversight Council designated MetLife and PrudentialFinancial as domestically systemic institutions partly based on this risk factor. See US Treasury (2013, 2014).

(2) See the European Commission press release, ‘State aid: commission approvesrestructuring of Belgian insurance company Ethias’: http://europa.eu/rapid/press-release_IP-10-592_en.htm?locale=en.

(3) See International Monetary Fund (2014).(4) See National Association of Insurance Commissioners Capital Markets Bureau

(2014b).(5) As an illustration of this risk, the liquidity of equity derivatives declined following the

2007–08 crisis, especially for European and emerging market equities.(6) See Swain and Swallow (2015).(7) See Eling and Pankoke (2014).

254 Quarterly Bulletin 2015 Q3

floors to discount rates (thereby preventing liabilities fromincreasing sharply as would happen if lower discount ratesreflecting lower risk-free rates in times of stress were applied).These examples provide some evidence of potentialprocyclical behaviour, but further analysis would be required inorder to be more conclusive.

Activities which might exacerbate the credit cycle in themedium to long runInsurance companies can potentially contribute to swings inthe cost and quantity of credit available in the medium to longrun, the ‘credit cycle’. The channels through which this canoccur include direct lending, acting as financial guarantors,selling credit default swaps and purchasing some debtsecurities. This subsection discusses these in turn.

Like banks, life insurers can originate loans directly tohouseholds or the corporate sector. For example since 2005insurance companies increased their exposure to the propertysector, and at the end of 2014, UK life insurers had around£29 billion of loans secured by mortgages on their balancesheet. PRA data show that the loans secured by mortgages onUK life insurers’ balance sheets predominantly included loansto the commercial real estate (CRE) sector (approximately75%). In comparison, since 2008, banks have been reducingthe size of their CRE book and at the end of 2014 held loans ofabout £129 billion to the CRE sector in aggregate (Chart 4).

Direct lending by insurance companies should contributepositively to economic growth and may not be an issue per se.But it could also exacerbate the credit cycle if increasing creditsupply by insurance companies led to an increase incompetition between lenders, and if this in turn led to acompression in the rates charged to borrowers that did notsuitably reflect risks.

Monoline insurers(1) enable borrowers to raise funds where theborrower would otherwise be facing difficulties in doing so.They achieve this by issuing a financial guarantee to theborrower. This increases a security’s creditworthiness — sincethe monoline insurer provides a guarantee making theborrower more attractive to a wider pool of investors. Thiscan increase the supply of finance to the real economy.(2)

Issuing guarantees, however, could exacerbate the creditcycle if the guarantees were priced at a level that did notreflect the credit risk of the borrowers insured.(3) Whenmonoline insurers were downgraded(4) in early 2008 in theUnited States and the EU, because of concerns about theircapital adequacy, the loans they insured were alsodowngraded, contributing to large sell-offs of credit and loanassets. Empirical studies have shown that downgrades ofmonoline insurers had a significant negative impact on thevalue of the instruments insured.(5)

Insurance companies can also generate income by sellingcredit default swaps (CDS). Buyers of CDS purchaseprotection against the default of one or several referenceborrowers, for an agreed term, in exchange for regularpremium payments. CDS markets support the supply offinance by enabling buyers of CDS to better manage theircredit risk. But if the premia charged by CDS sellers do notreflect the credit risk borne, this activity can exacerbate thecredit cycle (by contributing to lower costs of funding inunderlying credit markets).(6) Insurance and financialguarantee companies remain active sellers of CDS globally(Chart 5). The scale of this activity is currently small: at theend of December 2014, the notional value of the protectioninsurance and financial guarantee companies had sold reached£47.5 billion, relative to £7.62 trillion of protection sold inCDS markets in aggregate. But as AIG’s near failure hasshown, both the scale and the concentration of sold CDS canmatter from a financial stability perspective.

Finally, insurance companies could also exacerbate the creditcycle if they purchased large volumes of securitised productssuch as asset-backed securities (ABS) backed by loansincluding mortgages, auto-loans or other receivables. Simple,transparent, high-quality ABS can have many benefits for boththe real economy and financial market participants.(7) Butdi Iasio and Pozsar (2015) have recently shown that in a lowinterest rate yield environment, insurance companies and

0

5

10

15

20

25

30

35

0

50

100

150

200

250

2005 08 11 14

Loans from UK banks to the CRE sector (left-hand scale)

Loans from UK insurers secured by mortgages (right-hand scale)

£ billions £ billions

06 07 09 10 12 13

Chart 4 UK insurers’ loans secured by mortgages

Sources: PRA regulatory returns and Bank of England (December 2014).

(1) Financial guarantors are often referred to as monolines because they tend tospecialise in their guarantee business.

(2) See Bergstresser, Cohen and Shenai (2010).(3) This could occur either by mistake as a result of the complexity of the instruments

insured, because of inadequate prudential standards, or due to competition betweenparticipants.

(4) See Geneva Association Systemic Risk Working Group (2010).(5) See Chen et al (2013).(6) See Shimy and Zhuz (2010).(7) See Bank of England and European Central Bank (2014).

Topical articles Insurance and financial stability 255

pension funds’ demand for higher-yielding assets canincrease.(1) This creates incentives for banks to ‘manufacture’greater volumes of more complex, higher-yielding creditinstruments, leaving the economy more prone to deepdeleveraging in response to shocks.

Mitigating systemic risk and preservingfinancial stability

This section considers a number of regulatory initiatives thatare currently under way that should help mitigate some of therisks to financial stability discussed in this article. It alsohighlights some outstanding issues. In addition, the box onpages 246–47 discusses some remaining challenges aroundthe resolution of failed insurance companies.

First, Solvency II is a harmonised risk-based regime which willbe implemented from 1 January 2016. The regime willenhance the resilience of insurance companies in the EuropeanUnion, by: introducing improved governance and riskmanagement requirements; enhancing the quality of capital;and introducing a rigorous approach to group supervision.(2)

These measures should contribute to mitigating some of thefragilities discussed in the box on pages 248–50, increase theresilience of the European insurance sector to shocks andreduce insurers’ probability of distress or failure. ThoughSolvency II contains measures to mitigate procyclicality,further work will be necessary to judge their effectiveness inthe face of adverse market shocks (see the box on page 253).

Other jurisdictions across the world use different solvencyregimes. In order to address the risks arising from inconsistentregulatory standards while encouraging the removal ofbarriers to cross-border activities of international groups, theInternational Association of Insurance Supervisors (IAIS) is

developing a common framework for supervisinginternationally active insurance groups, known as ComFrame.The quantitative aspect of ComFrame — the Insurance CapitalStandard (ICS) — will consist of a consolidated group-widestandard and a globally comparable risk-based measure ofcapital adequacy.

The Bank fully supports and contributes to the development ofboth ComFrame and the ICS. Their implementation will helpsupervisors collectively evaluate and if necessary address therisks arising from group-wide activities, so leading to morerobust insurance groups. Both ComFrame and the ICS couldsupport ‘real markets’(3) by facilitating open capital flows,cross-border activities and enabling insurers to invest. Thisalso encourages insurers to perform their stabilising role ofproviding long-term benefits to the real economy andpolicyholders.

The IAIS is also leading work to address the potential risksposed by global systemically important insurers (G-SIIs). First,it runs an annual exercise to identify G-SIIs, based on criteriaincluding size and substitutability (as discussed in the firstsection of this article), interconnectedness with othersignificant financial institutions and markets (as discussed inthe second section) as well as the extent of potentiallysystemically risky activities undertaken. And to mitigate therisks G-SIIs can pose, the IAIS has adopted and adapted thepolicy framework developed by the Financial Stability Boardfor addressing the risks posed by systemically importantfinancial institutions.

This policy framework comprises three elements: enhancedsupervision; removing barriers to recovery and resolution inthe event of a G-SII’s distress or failure; and higher loss-absorbency capacity (higher capital requirements).Together, these should reduce both the probability and theimpact of failure of G-SIIs. The IAIS is in the process ofreviewing its methodology for identifying G-SIIs and willpublish its final specification for higher loss absorbency in duecourse. The Bank of England takes an active interest in thiswork, which is evolving in line with the changes under way inthe insurance industry and the wider financial and regulatoryenvironment. The Bank endorses and is involved in the IAIS’saim of improving global consolidated supervision, increasingtransparency and harmonising the current disparate nationalapproaches to valuation and solvency requirements.

0

20

40

60

80

100

120

2003 05 07 09 11 13

Sovereigns

Financial firmsNon-financial firms

Securitised and multiple sectors

Aggregated

£ billions

04 06 08 10 12 14

Chart 5 CDS sold globally by insurance and financialguarantee companies (by reference entity)

Source: Bank for International Settlements, detailed tables on semi-annual over-the-counter derivatives statistics (December 2014).

(1) See di Iasio and Pozsar (2015).(2) See Bulley (2015) and Swain and Swallow (2015).(3) Real markets must be fair, resilient and effective, and have social licence to operate.

See Bank of England (2015b).

256 Quarterly Bulletin 2015 Q3

Conclusion

Insurance companies are important providers of criticalservices to the real economy. They support output growth byhelping businesses and households to manage their risks, andin the process channel savings into investments. But as thisarticle has shown, the insurance sector can adversely affectfinancial stability.

The activities insurance companies engage in could contributeto the formation of fragilities, and if one or several firms failedor were in distress, this could lead to disruptions to criticalservices to the real economy. Furthermore, insurers’ activitiesand behaviour could amplify shocks to the financial system,either via financial markets or directly through financialcounterparties. Given that concentration of insurance servicesis not particularly high in the United Kingdom, it is this secondsource of systemic risk that it is more important to monitor inpractice.

As part of its remit to identify, monitor and take action toreduce systemic risks and enhance the resilience of theUK financial system, the FPC will review some of the risksdiscussed in this article, including those arising from activitiessuch as providing financial guarantees. The FPC also askedBank and FCA staff to further assess the risks from activitiessuch as cash collateral reinvestment programmes associatedwith securities lending, or writing credit default swaps, asthese activities can increase fragilities through leverage ormaturity transformation, and contribute to the propagation ofshocks through interconnectedness.(1)

(1) Bank of England (2015c).

Topical articles Insurance and financial stability 257

References

Bank of England (2013), ‘The strategy for the Bank’s Financial Stability mission 2013/14’, available atwww.bankofengland.co.uk/about/Documents/strategy1314.pdf.

Bank of England (2015a), Financial Stability Report, July, available atwww.bankofengland.co.uk/publications/Documents/fsr/2015/fsrfull1507.pdf.

Bank of England (2015b), Open Forum: building real markets for the good of the people, available atwww.bankofengland.co.uk/markets/Documents/openforum.pdf.

Bank of England (2015c), ‘Record of the Financial Policy Committee Meeting: 24 June 2015’, available atwww.bankofengland.co.uk/publications/Documents/records/fpc/pdf/2015/record1507.pdf.

Bank of England and European Central Bank (2014), ‘The case for a better functioning securitisation market in the European Union:A Discussion Paper’, available at www.bankofengland.co.uk/publications/Documents/news/2014/paper300514.pdf.

Bank of England and Procyclicality Working Group (2014), ‘Procyclicality and structural trends in investment allocation by insurancecompanies and pension funds: A Discussion Paper’, available at www.bankofengland.co.uk/publications/Documents/news/2014/dp310714.pdf.

Bergstresser, D, Cohen, R and Shenai, S (2010), ‘Financial guarantors and the 2007–2009 credit crisis’, Harvard Business School WorkingPaper 11-051.

Bijlsma, M and Vermeulen, R (2015), ‘Insurance companies’ trading behaviour during the European sovereign debt crisis: flight home or flightto quality?’, De Nederlandsche Bank Working Paper No. 468.

Breckenridge, J, Farquharson, J and Hendon, R (2014), ‘The role of business model analysis in the supervision of insurers’, Bank of EnglandQuarterly Bulletin, Vol. 54, No. 1, pages 49–57, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2014/qb14q105.pdf.

Brunnermeier, M and Pedersen, L (2009), ‘Market liquidity and funding liquidity’, The Review of Financial Studies, Vol. 22, No. 6,pages 2,201–38.

Bulley, A (2015), ‘Reflecting on Solvency II: continuity and change’, available atwww.bankofengland.co.uk/publications/Documents/speeches/2015/speech819.pdf.

Burrows, O, Cumming, F and Low, K (2015), ‘Mapping the UK financial system’, Bank of England Quarterly Bulletin, Vol. 55, No. 2,pages 114–29, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2015/q201.pdf.

Carney, M (2014), ‘Regulating the insurance industry to support the real economy’, available at www.bankofengland.co.uk/publications/Documents/speeches/2014/speech730.pdf.

Chen, F, Chen, X, Sun, Z, Yu, T and Zhong, M (2013), ‘Systemic risk, financial crisis, and credit risk insurance’, The Financial Review, Vol. 48,Issue 3, pages 417–42.

Congressional Oversight Panel (2010), ‘The AIG rescue, its impact on markets, and the government’s exit strategy’, June Oversight Report,page 25.

Cummins, D, Wei, R and Xie, X (2012), ‘Financial sector integration and information spillovers: Effects of operational risk events on US banksand insurers’, Working Paper, Temple University, Philadelphia, PA.

Debbage, S and Dickinson, S (2013), ‘The rationale for the prudential regulation and supervision of insurers’, Bank of England QuarterlyBulletin, Vol. 53, No. 3, pages 216–22, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2013/qb130303.pdf.

Eling, M and Pankoke, D (2014), ‘Basis risk, procyclicality, and systemic risk in the Solvency II equity risk module’, Journal of InsuranceRegulation, Vol. 33, No. 1, pages 1–39.

European Systemic Risk Board (2015), ‘Network analysis of the EU insurance sector’, European Systemic Risk Board Occasional Paper No. 7.

Financial Conduct Authority (2015), ‘Cash savings market study report: Part I: final findings, Part II: proposed remedies’, Market StudyMS14/2.3.

Financial Stability Board (2012), ‘Report on securities lending and repos: market overview and financial stability issues’.

258 Quarterly Bulletin 2015 Q3

Financial Stability Board (2013), ‘Strengthening oversight and regulation of shadow banking policy framework for addressing shadow bankingrisks in securities lending and repos’.

Förstemann, T and Feodoria, M (2015), ‘Lethal lapses — how a positive interest rate shock might stress German life insurers’,Deutsche Bundesbank Discussion Paper No. 12/2015.

Geneva Association Systemic Risk Working Group (2010), ‘Systemic risk in insurance, an analysis of insurance and financial stability’.

di Iasio, G and Pozsar, Z (2015), ‘A model of shadow banking: crises, central banks and regulation’.

International Association of Insurance Supervisors (2014), Global Insurance Market Report (GIMAR).

International Monetary Fund (2014), Global Financial Stability Report: Risk taking, liquidity, and shadow banking: curbing excess while promotinggrowth, October.

International Monetary Fund (2015), Global Financial Stability Report: Navigating monetary policy challenges and managing risks, April.

Kablau, A and Weiss, M (2014), ‘How is the low-interest-rate environment affecting the solvency of German life insurers?’, DeutscheBundesbank Discussion Paper No. 27/2014.

Kiesenbauer, D (2011), ‘Main determinants of lapse in the German life insurance industry’, North American Actuarial Journal, Vol. 16, No. 1,pages 52–73.

van Lelyveld, I, Liedorp, F and Kampman, M (2009), ‘An empirical assessment of reinsurance risk’, Netherlands Central Bank DNB Workingpaper No. 201.

McDonald, R and Paulson, A (2014), ‘AIG in hindsight’, Federal Reserve Bank of Chicago Working Paper No. 2014-07.

National Association of Insurance Commissioners Capital Markets Bureau (2014a), ‘Update on US insurance industry exposure to securitieslending and repurchase agreements’.

National Association of Insurance Commissioners Capital Markets Bureau (2014b), ‘2013 year-end update to the insurance industry’sderivatives exposure’.

Park, S and Xie, X (2014), ‘Reinsurance and systemic risk: the impact of reinsurer downgrading on property-casualty insurers’, Journal of Riskand Insurance, Vol. 81, No. 3, pages 587–622.

Parliamentary and Health Service Ombudsman (2008), ‘Equitable Life: a decade of regulatory failure’.

Roberts, R (2012), ‘Did anyone learn anything from the Equitable Life? Lessons and learning from the financial crises’, Kings College.

Royal Commission (2003), ‘Report of the HIH Royal Commission’.

Shimy, I and Zhuz, H (2010), ‘The impact of CDS trading on the bond market: evidence from Asia’, BIS Working Paper No. 332.

Shleifer, A and Vishny, R (2011), ‘Fire sales in finance and macroeconomics’, Journal of Economic Perspectives, Vol. 25, No. 1, Winter, pages 29–48.

Swain, R and Swallow, D (2015), ‘The prudential regulation of insurers under Solvency II’, Bank of England Quarterly Bulletin, Vol. 55, No. 2,pages 139–52, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2015/q203.pdf.

Swiss Re (2015), ‘Sigma: world insurance in 2014: back to life’.

Thimann, C (2014), ‘How insurers differ from banks: a primer on systemic regulation’, LSE Research Online Documents on Economics 61218,London School of Economics and Political Science, LSE Library.

US Department of Justice and Federal Trade Commission (2010), ‘Horizontal integration guidelines’.

US Treasury (2013), ‘Basis for the Financial Stability Oversight Council’s final determination regarding Prudential Financial’.

US Treasury (2014), ‘Basis for the Financial Stability Oversight Council’s final determination regarding MetLife, Inc’.

Topical articles UK market interest rates and macroeconomic data news 259

• Macroeconomic data releases receive considerable attention in financial market commentary,suggesting they are an important source of information that investors use to form their viewsabout the economic outlook. This article quantifies the role played by these data releases inexplaining observed changes in market interest rates in the United Kingdom.

• It looks at which data releases — both domestic and foreign — tend to have most effect onUK interest rates. It also examines how the importance of data releases can change over time andconsiders the factors that may explain changes in the sensitivity of interest rates to data news.

How much do UK market interest ratesrespond to macroeconomic data news?By Fernando Eguren-Martin and Nick McLaren of the Bank’s Macro Financial Analysis Division.(1)

(1) The authors would like to thank Lu Liu and Sheheryar Malik for their help in producingthis article.

Overview

When new information comes to light, investors mayreassess their views about the future, causing interest rateson government bonds to change. The signal central bankpolicymakers and market participants take from thesechanges in market interest rates will depend on what hasdriven them. One important source of new information ismacroeconomic data releases, such as official estimates ofGDP and inflation. In order to better understand the roleplayed by these releases, this article explores theirrelationship with changes in market interest rates.

In this article new information or ‘news’ is measured as thedifference between the actual release of official or surveydata and market participants’ expectations prior to therelease. As data for any one macroeconomic variable is onlyreleased monthly or quarterly, the approach used in thisarticle is to combine news from a number of differentreleases into an aggregate economic surprise index (ESI).

Both short and long-term interest rates in theUnited Kingdom are found to respond to changes in this ESI,although their sensitivity to data news varies considerablyover time. In part reflecting these changes in sensitivity, theimportance of data news in explaining changes in interestrates varies. During certain periods it can explain up to40% of the variation in short and long-term interest rates.But on average data news only accounts for a relativelysmall proportion of the total variation (summary chart).This is consistent with previous research which suggests that

the volatility of asset prices cannot be explained by changesin underlying economic fundamentals alone.

There appears to be an important role for both domestic andforeign data news, from the United States and the euro area.Indeed, for both short and long-term interest rates, foreignnews is as important as domestic news. This is consistentwith the United Kingdom being a small open economy andwith an important role for spillovers from internationalfinancial markets to UK asset prices.

0

5

10

15

20

25

30

35

40

45

Three-year spot Ten-year spot

Per cent

May2003

May05

May07

May09

May11

May13

May15

Summary chart Estimated proportion of the variation inshort-term and long-term UK interest rates explained bya historically weighted ESI(a)

Sources: Bloomberg and Bank calculations.

(a) The economic surprise index (ESI) is constructed by aggregating the news component of severalmacroeconomic data releases. The plot depicts the R-squared statistic from 90-day rollingregressions of daily changes in UK interest rates on this ESI. See main text for more details.

260 Quarterly Bulletin 2015 Q3

Financial assets are held because of the returns they yield inthe future. Therefore financial investors are forward looking,and asset prices change as investors reassess their views aboutthe future. This can occur when new information affectsinvestors’ expectations or, since the future is uncertain, theextent to which they wish to insure themselves againstunexpected events. Asset prices therefore respond to a rangeof news events, including economic data releases, policycommunications and geopolitical developments.

This article examines how one particular type of news —macroeconomic data releases — affects asset prices, and inparticular market interest rates on UK government bonds.This provides an insight into which data releases tend to havethe largest impact on UK interest rates, considering bothdomestic and foreign data releases. It is also possible to assesswhether the sensitivity of interest rates to data news variesover time. Relatedly, one can explore how much of theobserved variation in interest rates can be attributed tomacroeconomic data news. Although data news is found tobe an important driver of interest rates during certain periods,it only explains a relatively small proportion of the variation inUK interest rates on average over time.

The article first sets out why macroeconomic data releasesand interest rates are expected to be related, and brieflyoutlines past research conducted on these issues. It thenpresents the analytical approach, which is based on analysingthe relationship between changes in interest rates and theaggregation of news across a wide range of different datareleases. Next, the findings on the role of macroeconomicdata news in driving interest rates are summarised, includinghow this has varied over time. A box applies the same methodto investigate the sensitivity of sterling exchange rates to datanews. The article then illustrates how the approach can beapplied to help central bank policymakers and marketcommentators to understand the drivers of changes in interestrates, using market reaction to data news during the summerof 2013 as an example. Finally, the potential drivers of thechanging sensitivity of interest rates to macroeconomic datanews over time are discussed briefly.

The link between macroeconomic data newsand market interest rates

Market interest rates on assets such as government bonds area natural focus for this article since there is expected to be aclose link between investors’ outlook for the economy andtheir expectations of future interest rates. In a UK context,one reason for this is because the Monetary Policy Committee(MPC) sets a short-term nominal interest rate (‘Bank Rate’) aspart of its toolkit for conducting policy in order to achieve itsinflation target.(1) Therefore if data news affects marketparticipants’ views about the prospects for future growth andinflation then this may also alter their expectations for the

path of Bank Rate. Market interest rates, in part, reflect thefuture path of Bank Rate that investors expect, and so wouldalso be expected to change.

Beyond a certain horizon, when short-term shocks are likely tohave died away, it is unlikely that data news would affectexpectations of Bank Rate. Therefore, for longer-term interestrates the effect of data news is more likely to come throughthe ‘term premium’. This is the difference between marketinterest rates of a given maturity and the expected future pathof interest rates over that horizon. This difference reflects thefact that the future is uncertain, and so investors must becompensated for the risk that interest rates turn out to bedifferent from their central expectation. Previous research hasfound that estimates of the term premium tend to be relatedto domestic and foreign macroeconomic developments,suggesting they may be responsive to data news.

In addition, between 2009 and 2012 the MPC purchasedassets such as government bonds to achieve their inflationtarget. These asset purchases, often referred to asquantitative easing (QE), affected the term premium andhence the interest rates on government bonds at a range ofmaturities.(2) Therefore, if macroeconomic data news affectsthe perceived probability of future changes in the stock ofassets held by the Bank, this could lead to changes in marketinterest rates.

As well as interest rates, macroeconomic data news could alsoaffect the prices of a variety of other assets. The box onpages 266–67 provides one example, examining the sensitivityof exchange rates to macroeconomic data news.

Understanding what is driving changes in asset prices, and therole played by macroeconomic data news, is important forpolicymakers such as the MPC for a number of reasons. Theextent to which interest rates respond to data news can shedlight on how macroeconomic developments have affectedfinancial market participants’ expectations of the future.These expectations can have important implications forhousehold and business behaviour, and so affect the outlookand hence the required stance of policy. In addition, changesin asset prices themselves can affect the economy, byaffecting the cost of borrowing faced by companies andhouseholds. But the overall impact these changes have on theeconomy will depend, in part, on what has driven thosechanges. For instance, if interest rates have fallen due to aweaker outlook for the future, then there may be less of a

(1) Monetary policy in the United Kingdom is set in order to meet the MPC’s objective todeliver price stability — a target of 2% consumer price inflation set by theGovernment — and, subject to that, to support the Government’s economicobjectives including those for growth and employment.

(2) There are a number of ways in which central bank asset purchases could affect marketinterest rates. For instance purchasing large quantities of government bonds maypush up the price of the bonds remaining in the private sector, reducing their interestrates. For more details on the Bank’s asset purchases and evidence on their impact,see Joyce, Tong and Woods (2011).

Topical articles UK market interest rates and macroeconomic data news 261

reaction in people’s desire to increase their borrowing andspending. Quantifying the role of data news can, therefore,help policymakers to understand what signal to take fromobserved changes in interest rates.

Variation in the sensitivity of interest rates to data news overtime could also be informative for the Bank’s Financial PolicyCommittee. For instance, it could help them to understandwhether changes in market microstructure have affected thesensitivity of interest rates to shocks, such as macroeconomicdata news. That might be particularly important to the extentthat there are financial stability risks from sharp increases ininterest rates. And these techniques could also be of moregeneral interest to private sector economists and marketcommentators who seek to understand the links betweenmacroeconomic news and financial markets.

Measuring the sensitivity of interest rates to datanewsTo measure the ‘news’ or ‘surprise’ contained in each datarelease, surveys of financial market participants can be used tocompare the release with what was expected beforehand. Forinstance, as an arbitrary example, on 23 July 2010, the medianexpectation for the annualised quarterly growth rate ofUK GDP in Q2 was 1.1%, as measured by the Bloombergsurvey of market participants prior to the data release. Theactual growth rate announced that day was 1.6%. That newsof 0.5 percentage points was equivalent to 1.9 standarddeviations relative to the distribution of UK GDP news.

The change in market interest rates on the day of the datarelease can be used to estimate the reaction of interest ratesto this data news. For instance, on the day of that positiveGDP news, the interest rate of three-year maturityUK government bonds rose by 8 basis points.

In principle, this reaction can be compared to the change ininterest rates on the days of subsequent GDP releases todetermine whether the sensitivity of interest rates to datanews has changed. As releases of most data series aremonthly/quarterly in frequency, however, it is difficult toassess whether the reaction to a particular type of data releasehas changed over relatively short periods of time.

Therefore, the approach used in this article is to combine newsfrom over a hundred different types of official and surveyreleases into an aggregate ‘economic surprise index’ (ESI). Thefact that the ESI has many more observations than individualdata releases allows for a better understanding of therelationship between data news and changes in interest rates.The construction of the ESI is described in more detail in thenext section.

Existing researchA number of studies have investigated the role ofmacroeconomic data releases in driving changes in assetprices. Initially these papers — such as Ederington and Lee(1993), ap Gwilym et al (1998) and Clare and Courtenay(2001) — did not measure market expectations directly, butlooked simply at changes in asset prices on the day of the datarelease. As a long backrun of surveys of financial marketparticipants became available for a wide set of releases, anumber of papers began to quantify the news component ofdata releases and to assess the response of asset prices.Brooke, Danton and Moessner (1999) and Joyce and Read(1999) were among the first ones to do this exercise focusingon sterling asset prices, while Balduzzi, Elton and Clifton Green(2001), Andersen et al (2003) and Gürkaynak, Sack andSwanson (2005) look at a much wider set of data releases andfocus on US dollar asset prices.

More recently a number of studies, such as Faust et al (2007),have begun to examine how these effects have varied overtime, using a range of alternative techniques. Goldberg andGrisse (2013) investigate the causes of this variation (this isdiscussed briefly in the final section of the article). Theapproach used in this article is based on the methoddeveloped and outlined in Swanson and Williams (2014a,2014b), which accounts for the historical importance ofdifferent macroeconomic data news when constructing an ESI.Where appropriate the results discussed in this article arecompared to the findings of the existing literature.

The approach: using an ‘economic surpriseindex’

In principle, there are a number of different ways ofaggregating the news component across different datareleases. This article compares the results from two differentapproaches: first an ‘equally weighted’ ESI that treats everydata series equally; and second a ‘historically weighted’ ESIthat weights releases from each data series differently basedon the extent to which interest rates have responded to thenews component in each release on average over the sample.

The ‘equally weighted’ ESI is constructed as the sum of all datanews on a given day. These include activity indicators (such asGDP or retail sales), inflation estimates (such as CPI inflation),labour market indicators (such as the unemployment rate) andbusiness surveys (such as the purchasing managers’ index(PMI)). Over a hundred different types of releases from theUnited Kingdom, United States and euro area are consideredeach quarter. The sample period is from January 2003 toMay 2015.

To allow comparison across different data series, each piece ofnews is measured relative to the standard deviation of news

262 Quarterly Bulletin 2015 Q3

for that data series over the full sample period.(1) Chart 1shows how this measure has evolved since 2013 for UK, USand euro-area data releases.(2) In the analysis below, the datanews from each country is combined into one aggregate ESI.

Some types of data releases, however, receive more attentionin financial market commentary than others. It may not,therefore, be appropriate to treat all data releases as equallyimportant in affecting interest rates.

The historically weighted ESI, developed by Swanson andWilliams (2014a, 2014b), tries to account for this by weightingdifferent data series according to the magnitude of the effectnews has had on market interest rates over the sample. Dataseries that do not have a statistically significant effect oninterest rates are excluded. This approach is described in moredetail in the appendix.

This weighting scheme has some shortcomings too. Theweights attached to each series in the historically weighted ESIare fixed throughout the sample period. This means that theESI includes only those types of data releases that areconsistently important across the entire sample period. Inreality, it may be that particular types of releases becomemore or less important at different points in time.

We use a regression to estimate how changes in the ESI affectmarket interest rates, and how this has changed over time.This analysis is based on equation (1) below.

(1)

where ∆it represents the daily change in market interest ratesand ESIt is the ESI index described above. β is the ‘sensitivity’coefficient which is the focus of the analysis.

The effect of data news on both short and long-term interestrates is examined.(3) The appendix contains more details onthe data used and the econometric approach.

The sensitivity of market interest rates todata news

This section starts by outlining the weights used in thehistorically weighted ESI. These weights themselves areinformative, as they provide an indication of which datareleases market participants pay most attention to. Thesection then turns to consider, for both approaches, theestimated sensitivity of interest rates to the ESI measures, andhow this has changed over time. The section ends byexamining how much of the observed variation in interestrates can be explained by macroeconomic data news.

On average it is found that data news only explains a relativelysmall proportion of the variation in UK interest rates.Although this may seem somewhat surprising, it is consistentwith previous research which suggests that the volatility ofasset prices cannot be explained by changes in underlyingeconomic fundamentals alone. This would be consistent witha large share of the variation in asset prices being explained bychanges in the risk premia required by investors, which mayrespond to a wide set of factors besides macro data news.

Which data releases have been associated withchanges in market interest rates?A range of both domestic and foreign data releases areimportant in explaining changes in short-term interest ratesover our sample period (Table A). This might be because theUnited Kingdom is a small open economy, so the outlook —and thus the prospects for interest rates –— is stronglyaffected by foreign as well as domestic shocks. It could alsoreflect a role for spillovers from international financial marketsto UK asset prices. Swanson and Williams (2014b), Goldbergand Grisse (2013) and Brooke, Danton and Moessner (1999)also find a significant role for foreign data news in drivingUK short-term interest rates.

The statistically significant data releases are those that oftenreceive most attention in financial market commentary — forinstance UK GDP, UK PMI and the change in US non-farmpayrolls.

20

10

0

10

20

30

40

50

60

Jan. May Sep. Jan. May Sep. Jan. May

United Kingdom

United States

Euro area

Cumulative standard deviations

2013 14 15

+

–

Sources: Bloomberg and Bank calculations.

(a) The economic surprise indices are constructed by aggregating the news component ofseveral macroeconomic data releases. See main text for more details.

Chart 1 Equally weighted ESIs since 2013(a)

(1) If, for example, on a given day there is a one standard deviation positive GDP newsand a one standard deviation positive retail sales news, the ‘equally weighted’approach assigns a value of two to the ESI. If the retail sales news was negative andthe GDP news was positive the ‘equally weighted’ approach assumes that the news inthese releases ‘cancels out’ and assigns a value of zero to the ESI.

(2) A historically weighted version of the ESI is not shown as these are specific to theasset considered.

(3) We focus on changes in three-year spot interest rates for the short term (the interestrate to borrow for a period of three years from today), and the ten-year spot interestrates for the long term.

Topical articles UK market interest rates and macroeconomic data news 263

It is perhaps surprising that there are no significant euro-areareleases, given that the euro area is the United Kingdom’slargest trading partner.(1) One explanation could be the way inwhich euro-area data tends to be released. For many dataseries, such as GDP, individual releases for each country areoften made in the days prior to the aggregate euro-arearelease. Therefore even if the joint news across each of thereleases is important for UK interest rates, it may be that theindividual country releases are not significant. And by thetime the aggregate euro-area measure is released, there isoften little additional news relative to the individual countryreleases.

International data releases also have an important effect onlong-term interest rates (Table B). Some of the data releasesthat are significant are similar to those for short-term interestrates and in line with what might be intuitively expected — forinstance the change in US non-farm payrolls. Other datareleases are less intuitive. For instance, it is not clear why theItalian manufacturing PMI would be particularly important forthe United Kingdom. Previous research — such as that bySwanson and Williams (2014b) — also finds somecounterintuitive data releases to be significant.

As discussed above, part of the reason for the sensitivity oflong-term interest rates to data news is likely to be because ofthe effect it has on the term premium. Consistent with this,past research has found that the term premium tends to varywith the economic cycle (Malik and Meldrum (2014)). Thispast research also finds an important role for commoninternational factors in driving changes in the term premium,which result in a strong correlation between long-terminterest rates across countries. That may be consistent withthe finding in this article of the role of foreign data releases inexplaining past movements in long-term UK interest rates.

How has the sensitivity of interest rates to datareleases varied over time?Both of the ESI approaches can be used to consider how thesensitivity of interest rates to data news has changed overtime, by regressing interest rate changes on the ESIs over arolling 90-day window.

Short-term interest ratesChart 2 and Chart 3 illustrate how the sensitivity ofUK short-term interest rates to data news has varied overtime, for the equally weighted and historically weighted ESIsrespectively. A regression coefficient close to one implies thatthe sensitivity of yields to the ESI is near the average over thebenchmark period (2003–08). A coefficient greater than oneindicates that yields are more sensitive to data news than inthe benchmark period; while a coefficient close to zero meansthat yields have become unresponsive to data news.(2)

Generally the estimated sensitivity has a positive value, whichimplies that interest rates increase in the face of positive newsin variables such as GDP or inflation for example. This isconsistent with the way expectations of monetary policywould be anticipated to change in response to such changes inthe economic outlook.(3)

Both methods suggest that, rather than being stable, theextent to which interest rates respond to data news variesconsiderably over time.

Table A Data releases included in the historically weighted ESI forshort-term UK interest rates (three-year)(a)

Estimated impacton short rates (basispoints per standard Estimated

Data release deviation news) standard error

UK GDP — first estimate 2.6 0.6US change in non-farm payrolls 2.5 0.4UK retail sales including fuel 1.9 0.5US ISM non-manufacturing 1.8 0.5UK services PMI 1.4 0.5UK net lending secured on dwellings 1.4 0.7US ISM manufacturing 1.3 0.3UK manufacturing production 1.2 0.2US existing home sales 1.2 0.4

Sources: Bloomberg and Bank calculations.

(a) UK releases in blue, US releases in magenta. Data releases are ordered according to their impact on UKthree-year spot interest rates. The impact is estimated from regressions of daily changes in three-year rateson macro news over January 2003–May 2015, and standard errors are robust to heteroskedasticity. Seeappendix for more details.

Table B Data releases included in the historically weighted ESI forlong-term UK interest rates (ten-year)(a)

Estimated impacton long rates (basispoints per standard Estimated

Data release deviation news) standard error

US change in non-farm payrolls 3.0 0.4UK retail sales including fuel 2.1 0.7US ISM non-manufacturing 2.1 0.5UK manufacturing PMI 1.7 0.8Italy manufacturing PMI 1.6 0.6UK manufacturing production 1.3 0.3UK net lending secured on dwellings 1.3 0.5US ISM manufacturing 1.3 0.3Italy industrial production 1.3 0.4US existing home sales 1.2 0.4

Sources: Bloomberg and Bank calculations.

(a) UK releases in blue, US releases in magenta and euro-area releases in orange. Data releases are orderedaccording to their impact on UK ten-year spot interest rates. The impact is estimated from regressions ofdaily changes in ten-year rates on macro news over January 2003–May 2015, and standard errors are robustto heteroskedasticity. See appendix for more details.

(1) The only exception is French consumer confidence news, which was found to besignificant for both short and long-term interest rates. However this series is notincluded in the ESIs as it entered with the opposite sign to what would be expected,suggesting this was likely to be a spurious result rather than a genuine reflection ofthe impact of news about French consumer confidence on UK interest rates.

(2) In the case of the historically weighted ESI, a normalisation of the sensitivitycoefficient is needed for technical reasons, which are explained in the appendix. Inorder to allow for an easier comparison, the sensitivity coefficient for the equallyweighted ESI was also normalised, such that it averages one over 2003–08.

(3) At times the coefficient is close to zero, suggesting that interest rates areunresponsive to such data news. In other periods, the coefficient even becomesnegative. It is likely that this reflects other news on the day of the data releaseswhich is distorting the reaction to the data news themselves.

264 Quarterly Bulletin 2015 Q3

The broad pattern of moves in sensitivity is similar accordingto both estimates. Short-term interest rates were particularlysensitive to data news at the beginning of the financial crisis.Over the period since then, sensitivity has frequently beenlow. This may be consistent with the low level of interestrates and their proximity to their ‘lower bound’ over thisperiod. Since there is a limit to how low nominal interest ratescan be, then as they approach this level they will be unable tofall further in response to negative data news, which wouldreduce the sensitivity of interest rates to data news. Swansonand Williams (2014a, 2014b) put particular emphasis on thisexplanation for recent variation in sensitivity.

Even during this period, however, there have also been periodsof elevated sensitivity, such as the spike upwards during thesummer of 2013. This coincided with an increased focus onthe prospects for monetary policy in the United States, anepisode that is considered in more detail in the next section.The final section of the article considers more broadly some ofthe potential reasons why the sensitivity may vary over time.These include factors such as changes in marketmicrostructure, as well as changes in central bankcommunication.

Long-term interest ratesChart 4 and Chart 5 illustrate how the sensitivity of long-termUK interest rates to data news has varied over time, for theequally weighted and historically weighted ESIs respectively.

3

2

1

0

1

2

3

4

5

May2003

May05

Sensitivity

95% confidence interval Index

May07

May09

May11

May13

May15

6

7

4

+–

Sources: Bloomberg and Bank calculations.

(a) Based on 90-day rolling regressions of daily changes in three-year rates on an equallyweighted ESI. The estimated coefficient is normalised to average one over 2003–08 so as tofacilitate the comparison with Chart 3. Standard errors are robust to heteroskedasticity.See main text for more details.

Chart 2 Estimated sensitivity of short-term UK interestrates (three-year) to an equally weighted ESI(a)

4

3

2

1

0

1

2

3

4

5

6

7 Index

May2003

May05

Sensitivity

95% confidence interval

May07

May09

May11

May13

May15

+

–

Sources: Bloomberg and Bank calculations.

(a) Based on 90-day rolling regressions of daily changes in three-year rates on a historicallyweighted ESI. The estimated coefficient is normalised so intuitively it can broadly bethought of as averaging one over 2003–08. See appendix for more details. Standard errorsare bootstrapped to account for the fact that the ESI is a generated regressor. The y-axisscale is restricted to show variation in the coefficient estimate more clearly. As a result, the95% confidence interval is not fully captured in the chart when the standard errors spike in2003 and 2008.

Chart 3 Estimated sensitivity of short-term UK interestrates (three-year) to a historically weighted ESI(a)

5

4

3

2

1

0

1

2

3

4

5

6 Index

May2003

May05

Sensitivity

95% confidence interval

May07

May09

May11

May13

May15

+

–

Sources: Bloomberg and Bank calculations.

(a) Based on 90-day rolling regressions of daily changes in ten-year rates on an equally weightedESI. The estimated coefficient is normalised to average one over 2003–08 so as to facilitatethe comparison with Chart 5. Standard errors are robust to heteroskedasticity. See maintext for more details.

Chart 4 Estimated sensitivity of long-term UK interestrates (ten-year) to an equally weighted ESI(a)

4

3

2

1

0

1

2

3

4

5

Index

+

–

May2003

May05

Sensitivity 95% confidence interval

May07

May09

May11

May13

May15

6

5

Sources: Bloomberg and Bank calculations.

(a) Based on 90-day rolling regressions of daily changes in ten-year rates on a historicallyweighted ESI. The estimated coefficient is normalised so intuitively it can broadly bethought of as averaging one over 2003–08. See appendix for more details. Standard errorsare bootstrapped to account for the fact that the ESI is a generated regressor.

Chart 5 Estimated sensitivity of long-term UK interestrates (ten-year) to a historically weighted ESI(a)

Topical articles UK market interest rates and macroeconomic data news 265

As with short-term interest rates there is a high degree ofvariation over time. The standard errors around thecoefficient estimates are larger, so sensitivity of long-terminterest rates is less precisely estimated and, perhapsreflecting this, there is less similarity in the dynamics of thetwo measures of the sensitivity of long rates than in those forshort-term rates. Despite displaying different dynamics oversome periods, however, changes in the sensitivity of long-terminterest rates tend to be positively correlated with those ofshort-term interest rates, maybe as a result of having commondrivers.

These results are broadly similar to those for theUnited Kingdom of Swanson and Williams (2014b), and thosefor the United States of Goldberg and Grisse (2013),Faust et al (2007) and Swanson and Williams (2014a), whoalso find evidence that the sensitivity of short and long-terminterest rates to data news varies significantly over time.

How much of the observed variation in interest ratescan be attributed to data news?The estimates of sensitivity of interest rates to data newsshow the extent to which interest rates respond to a givenamount of news. The approach can also be used to investigatehow much of the variation in interest rates can be explainedby economic data news, using the ‘R-squared’ statistic of theestimated regressions. The R-squared captures the proportionof the variation in the dependent variable (daily changes ininterest rates) that can be explained by the explanatoryvariable (the ESI).

Given that macro data releases receive considerable attentionfrom market participants, it is perhaps surprising to see thatthey can only account for a relatively small proportion of thevariation in short-term interest rates (Chart 6). The impact ofdata news has, however, varied over time. During certainsubperiods, data news explains up to 40% of the totalvariation in short-term interest rates; but these periods areinfrequent. The findings are similar for long-term interestrates (Chart 7).(1) Swanson and Williams (2014b) find similarresults for the United Kingdom using a slightly differentsample period.

These findings suggest that while macroeconomic data newsdoes affect market interest rates systematically, there are anumber of other important factors. These may include policycommunications, geopolitical developments, movements inother asset prices, and technical factors in the bond market.Although this may appear surprising given the attention datanews receive, this result is actually broadly consistent with thefindings of a range of research on financial markets, whichsuggests that the volatility of asset prices cannot be explainedby changes in underlying economic fundamentals alone. Muchof this research gives prominence to changes in the risk premiarequired by investors to explain variation in asset prices.

Although economic fundamentals can affect risk premia, theserisk premia are also likely to depend on a broad range of otherpotential drivers.

It is also possible that the low R-squared values can also partlybe explained by the drawbacks of the approach used. The factthat the impact of data releases is identified using daily datameans that the measure of interest rate changes could also becapturing the effects of other events on the same day as thereleases, which results in the impact of data news beingmismeasured.(2) Perhaps consistent with that, there is

0

5

10

15

20

25

30

35

40

45

Equally weighted

Historically weighted

Per cent

May2003

May05

May07

May09

May11

May13

May15

Sources: Bloomberg and Bank calculations.

(a) The plot depicts the R-squared statistic from 90-day rolling regressions of daily changes inthree-year interest rates on two alternative ESIs. See main text for more details.

Chart 6 Estimated proportion of the observed variationin short-term UK interest rates (three-year) explained bythe ESIs(a)

0

5

10

15

20

25

30

35

40

45

Equally weighted

Historically weighted

May2003

May05

May07

May09

May11

May13

May15

Per cent

Sources: Bloomberg and Bank calculations.

(a) The plot depicts the R-squared statistic from 90-day rolling regressions of daily changes inten-year interest rates on two alternative ESIs. See main text for more details.

Chart 7 Estimated proportion of the observed variationin long-term UK interest rates (ten-year) explained bythe ESIs(a)

(1) These estimates include days in which there are no data releases. If the estimation isrepeated only considering days in which there are non-zero ESI values, the share ofexplained variation of short and long-term interest rate increases, but it still typicallyremains low.

(2) A number of papers, such as Faust et al (2007) and Goldberg and Grisse (2013) havetackled this issue by relying on high-frequency intraday data to study time variationin the sensitivity of interest rates to macroeconomic data news.

266 Quarterly Bulletin 2015 Q3

The sensitivity of other asset prices to datanews

The impact of macroeconomic data on financial markets is notlimited to interest rates, and a similar approach to that used inthis article can be applied to a range of other asset prices. Thisbox illustrates this by examining the sensitivity to macro datanews of the sterling-US dollar exchange rate.(1)

A historically weighted ESI is constructed using thesame technique as that applied to interest rates in the maintext (see the appendix for more details). This involvescombining the data releases found to significantly affect thesterling-US dollar exchange rate,(2) weighting each releaseaccording to its past impact on the exchange rate.

The exchange rate is expected to respond to data releases tothe extent that these contain news about the outlook of onecountry relative to the other. Consistent with this, both UKand US data releases are found to be significant (Table 1).Typically, positive data news on UK activity causes sterling toappreciate (resulting in a positive coefficient in Table 1), whilepositive data news on US activity causes the US dollar toappreciate (and sterling to depreciate, resulting in a negativecoefficient in Table 1).(3) This could be consistent with theimpact of these releases on UK/US interest rates, to the extentthat changes in exchange rates reflect changes in interest ratedifferentials across countries.(4)

Following the same process as for interest rates, rollingregressions can then be estimated of changes in the spotexchange rate on this historically weighted ESI. Chart Ashows that the sensitivity of the sterling-US dollar exchangerate to data news also varies considerably over time. In

(1) For an examination of the reaction to data news in equity markets see Andersen et al(2007).

(2) In this case, only UK and US releases were considered.(3) In Table 1 the coefficient on the US ISM non-manufacturing composite index is

positive suggesting that positive news in this data release caused the dollar todepreciate. It is possible this reflects measurement error (due to other news takingplace on the same day) rather than a well-identified impact of the release on theUS dollar.

(4) This can be explained according to the uncovered interest parity (UIP) framework.Positive data news in the United Kingdom, which results in an increase in UK interestrates, will increase the expected return for an investor in the United Kingdom relativeto the United States. This will encourage investors to shift their investments from theUnited States to the United Kingdom, increasing demand for sterling, causing it toappreciate. Technically, this assumes that there is no change in either the exchangerate risk premium or the expected long-run level of the exchange rate. If data newswere to affect either of these components, the direction of the impact of data newson the exchange rate could be positive or negative.

4

3

2

1

0

1

2

3

4

5

6 Index

May2003

May05

Sensitivity

95% confidence interval

May07

May09

May11

May13

May15

+

–

Sources: Bloomberg and Bank calculations.

(a) Based on 90-day rolling regressions of daily changes in US$ per £ exchange rate on ahistorically weighted ESI. The estimated coefficient is normalised so intuitively it can broadlybe thought of as averaging one over 2003–08. See appendix for more details. Standard errorsare bootstrapped to account for the fact that the ESI is a generated regressor. The y-axis scaleis restricted to show variation in the coefficient estimate more clearly. As a result, the 95%confidence interval is not fully captured in the chart when the standard errors spike in 2008.

Chart A Estimated sensitivity of the US dollar-sterlingbilateral exchange rate to a historically weighted ESI(a)

0

5

10

15

20

25

30

Per cent

May2003

May05

May07

May09

May11

May13

May15

35

40

45

Sources: Bloomberg and Bank calculations.

(a) The plot depicts the R-squared statistic from 90-day rolling regressions of daily changes inUS$ per £ exchange rate on a historically weighted ESI. See main text for more details.

Chart B Estimated proportion of the observed variationin the US dollar-sterling bilateral exchange rateexplained by a historically weighted ESI(a)

Table 1 Data releases included in the historically weighted ESI forthe US dollar-sterling bilateral exchange rate (US$ per £)(a)

Estimated impact onUS$ per £ (percentage

change per standard EstimatedData release deviation news) standard error

UK GDP — first estimate 0.32 0.08UK manufacturing PMI 0.28 0.07UK net lending secured on dwellings 0.25 0.10UK GDP — third estimate 0.25 0.09UK manufacturing production 0.19 0.03UK PPI input prices 0.17 0.06US ISM non-manufacturing composite 0.14 0.06US change in non-farm payrolls -0.13 0.06US factory orders -0.11 0.06US advance retail sales -0.11 0.04US GDP -0.11 0.05

Sources: Bloomberg and Bank calculations.

(a) UK releases in blue, US releases in magenta. Data releases are ordered according to their impact on the US$per £ exchange rate. The impact is estimated from regressions of daily changes in the US$ per £ exchangerate on macro news over January 2003–May 2015, and standard errors are robust to heteroskedasticity. Seeappendix for more details.

Topical articles UK market interest rates and macroeconomic data news 267

evidence that data news can explain a greater proportion ofthe variation in interest rates at lower frequencies, when theshort-lived ‘noise’ affecting interest rates on a day-to-daybasis is more likely to have washed out. Altavilla, Giannoneand Modugno (2014) find that the power of ESIs for explainingchanges in bond and equity prices increases significantly whenlooking at monthly or quarterly changes in the case of theUnited States. A final shortcoming of the approach could bethat the survey measure of expectations used might not fullyreflect the expectations of market participants — for instancebecause it is not representative of all the active investorswithin the government bond market.

Case study: heightened sensitivity of interestrates to data news during the summer of 2013

To illustrate how this approach can be used to helpunderstand the drivers of changes in market interest rates, thissection examines the changes in the sensitivity of interestrates to data news during the summer of 2013. During thisperiod there was a sustained increase in short and long-terminterest rates, in both the United Kingdom and theUnited States. From their low point in May 2013, short-termrates increased by around 75 basis points by mid-September,in both the United Kingdom and the United States. Over asimilar period long-term rates increased by around145 basis points in the United States and by 130 basis pointsin the United Kingdom.

A relevant question for conjunctural analysis within centralbanks at the time was to understand what was driving theseincreases in UK and US interest rates. One aspect of that wasthe extent to which it reflected the reaction of interest ratesto data news.

Chart 1 shows that over this period there was an accumulationof upside data news in both the United Kingdom and theUnited States, which would be consistent with an increase ininterest rates in both countries.

Focusing on the reaction of UK interest rates, Chart 8 showsthat the sensitivity of short-term UK interest rates to data

news rose sharply around that time, becoming statisticallyhigher than average at times during the second half of 2013.The sensitivity of long-term UK interest rates also increasedmarkedly. Hence, it appears that interest rates wereresponding by more than usual to the positive data news overthis period, contributing to the increase in interest ratesobserved.

Using the approach discussed in the previous section, theestimated R-squared values from this period can be used toquantify the extent to which the combination of this positivedata news and increased sensitivity of interest rates to datanews can explain the changes in interest rates during 2013.Chart 6 and Chart 7 show that, for the historically weightedESI, data news can account for around a quarter of the movesin short and long-term interest rates over certain periods of2013. This suggests an important role for data news.Although, this also means that, even during this period ofheightened sensitivity, it still only accounted for a relativelysmall proportion of the variation in interest rates.

particular, the sensitivity has been at very low levels sinceUK and US short-term interest rates moved close to theireffective lower bounds following the financial crisis. Thesensitivity estimate has increased back towards average levelsover the past year.

For exchange rates, data news generally explains even less ofthe observed variation in the asset price than was the case forinterest rates (Chart B). Again, this is consistent with theresults of Swanson and Williams (2014b) and, more generally,

with other research which finds that it is particularly difficultto explain exchange rate dynamics solely on the basis of thereaction to changes in macroeconomic fundamentals.(1)

(1) For a brief survey of this literature see Evans and Lyons (2008).

1

0

1

2

3

4

Mar. May July Sep. Nov. Jan. Mar.

Index

Sensitivity

95% confidence interval

2013 14

+

–

Sources: Bloomberg and Bank calculations.

(a) Based on 90-day rolling regressions of daily changes in three-year interest rates on ahistorically weighted ESI. The estimated coefficient is normalised so intuitively it canbroadly be thought of as averaging one over 2003–08. See appendix for more details.Standard errors are bootstrapped to account for the fact that the ESI is a generatedregressor.

Chart 8 Estimated sensitivity of short-term interestrates (three-year) to a historically weighted ESI over2013 and early 2014(a)

268 Quarterly Bulletin 2015 Q3

There are a number of possible explanations for this increasein the sensitivity of interest rates to data news. To betterunderstand the drivers of this in the case of UK interest rates,the sensitivity can be estimated separately for the UK andUS data releases included in the historically weighted ESI.Chart 9 shows that for short-term UK interest rates there wasan increase in the sensitivity to both UK and particularlyUS data news. The sensitivity to UK data news picked up toaround its past average level, while the sensitivity to US datanews increased more markedly to above its past average. Theseparate estimates for UK and US data news are based onfewer observations, and so are less precisely estimated.Nonetheless, combined with the fact that interest ratesincreased in both the United Kingdom and the United States,this suggests an important role for internationaldevelopments.

In the United States, beginning in May 2013 there wasincreased speculation about the future prospects for monetarypolicy. In particular, communication by the Federal ReserveBoard Open Market Committee (FOMC, responsible forsetting monetary policy in the United States) suggested it wasconsidering reducing the pace at which it was purchasingassets from the private sector as part of its QE policy (thiscame to be known as ‘tapering’ the pace of purchases).Changes in the stock of assets expected to be purchased bythe Federal Reserve could have had a direct impact onUS interest rates. In addition, as this communication by theFOMC was interpreted as signalling a turning point in thestance of US monetary policy, it could also have resulted inmarket participants paying closer attention to macroeconomicdata news, as they formed their expectations for the futurepath of monetary policy.

That could explain an increase in the sensitivity of US interestrates to US data news. And given the close correlationbetween market interest rates internationally, this could alsohave increased the sensitivity of UK interest rates to US datanews, as investors assessed the implications of changes in theUnited States for UK interest rates.

One potential reason for the less marked increase in thesensitivity of UK interest rates to UK data news, relative toUS data news, is the introduction of forward guidance by theMPC in August 2013.(1) It is difficult, however, to draw strongconclusions about the effect of forward guidance on thesensitivity of UK interest rates to data news using thetechniques outlined in this article. This is because it is hard toknow what might have happened to the sensitivity of interestrates to UK data news in the absence of the MPC’s forwardguidance, particularly given the other developments in globalmarkets around this time. Moreover, the MPC specificallyreferred to an unemployment threshold when it firstintroduced forward guidance.(2) But since UK labour marketdata news has not on average had a significant effect onUK interest rates over the full sample period, they are notincluded in the historically weighted ESI. Therefore anyimpact on the sensitivity of interest rates to these labourmarket releases will not be captured by this method.

The sensitivity of interest rates to data news could also havebeen affected by a number of other factors. For instance,changes in government bond market conditions following thefinancial crisis could have made periods of sharp increases inmarket volatility around key events more likely. The nextsection considers some of the potential drivers of the changingsensitivity of interest rates to macroeconomic data news overtime, drawing on existing research where possible.

Potential drivers of the time variation insensitivity

While it appears that the sensitivity of interest rates to datanews varies considerably over time, the technique used in thisarticle cannot be used to identify the reasons for this variation.

One factor frequently cited in the literature as potentiallyaffecting the sensitivity of interest rates to data news is thedegree of uncertainty and risk aversion in financial markets.Goldberg and Grisse (2013) find that the sensitivity of

1

0

1

2

3

4

UK data

US data

Index

–

+

Mar. May July Sep. Nov. Jan. Mar.2013 14

Sources: Bloomberg and Bank calculations.

(a) Based on 90-day rolling regressions of daily changes in three-year interest rates on ahistorically weighted ESI. The estimated coefficient is normalised so intuitively it canbroadly be thought of as averaging one over 2003–08. See appendix for more details. Eachline corresponds to a separate regression which considers UK and US data releasesseparately.

Chart 9 Estimated sensitivity of short-term interestrates (three-year) to two historically weighted ESIs over2013 and early 2014, separating UK and US data news(a)

(1) The Chancellor had asked the MPC to consider the case for adopting some form offorward guidance when setting the MPC’s remit in March 2013. For more detail onthe forward guidance policy adopted by the MPC in August 2013, seewww.bankofengland.co.uk/publications/Documents/inflationreport/2013/ir13augforwardguidance.pdf.

(2) The MPC stated that it intended, at a minimum, to maintain the exceptionallyaccommodative stance of monetary policy until economic slack had beensubstantially reduced, provided that this did not put at risk either price stability orfinancial stability. In particular, and subject to those conditions, the MPC stated itintended not to raise Bank Rate from 0.5% at least until the Labour Force Surveyheadline measure of the unemployment rate had fallen to a threshold of 7%.

Topical articles UK market interest rates and macroeconomic data news 269

US interest rates to key US data releases (such as non-farmpayrolls, GDP and CPI) is lower when uncertainty and riskconditions are elevated. They suggest that marketparticipants are less likely to take strong signals from datanews when the relationship between these and the economicoutlook is more uncertain.(1)

Other papers have found that asset price sensitivity dependson the magnitude and direction of data news. Looking atexchange rates, Andersen et al (2003) and Ehrmann andFratzscher (2005) find that negative data news has a largereffect than positive news of the same magnitude, and thatlarger news (in absolute terms) seems to have a more thanproportional impact.

As discussed earlier, in recent years the sensitivity of interestrates could also have been affected by the fact that interestrates have been closer to their lower bound. Given that thereis a limit to how low nominal interest rates can fall, at verylow levels it is likely interest rates will respond less to(particularly negative) data news. Swanson and Williams(2014a, 2014b) find this to be an important factor for thesensitivity of interest rates to data releases in theUnited States, United Kingdom and Germany at differentpoints in time since the beginning of the financial crisis.

It is possible that changes in central bank communicationscould also have an effect on interest rate sensitivity, asdescribed in the previous section.

Finally, it is possible that structural changes in financialmarkets following the financial crisis may have affected thesensitivity of interest rates to data news. For example, prior tothe financial crisis, intermediaries in government bondmarkets tended to hold government bonds themselves afterpurchasing them from clients, waiting until they could find anappropriate buyer. More recently, the Bank’s market contactssuggest that these intermediaries often try to directly matchbuyers and sellers rather than hold the bonds themselves. Tothe extent that intermediaries holding these bonds reducedsome of the volatility in market prices, these changes in theway the market operates could have led to an increase in thesensitivity of market prices in the face of developments suchas macroeconomic data news. Set against that, there is alsosome indication that these intermediaries are now less activein speculatively trading in these markets themselves. Thatreduced activity could act in the opposite direction, reducingthe sensitivity of market prices to news more generally,including in response to data news. These hypotheses havenot been widely explored in previous research, most likely dueto the absence of data on these developments.

Conclusion

This article has explored the link between macroeconomicdata releases and changes in market interest rates. It findsthat both short and long-term interest rates in theUnited Kingdom respond to data releases that differ frommarket participants’ expectations prior to the release,although their sensitivity to data news varies over time.

In part reflecting these changes in sensitivity, the importanceof data news in explaining the variation in interest rates alsovaries considerably over time. During certain periods it canexplain up to 40% of the variation in short and long-terminterest rates. But on average data news can only account fora relatively small proportion of the total variation inUK interest rates. This is consistent with previous researchwhich suggests that the volatility of asset prices cannot beexplained by changes in underlying economic fundamentalsalone.

That said, there appears to be an important role for bothdomestic and foreign data news. Indeed, for short andlong-term interest rates, foreign news is as important asdomestic news. This is consistent with the United Kingdombeing a small open economy and with an important role forspillovers from international financial markets to UK assetprices.

The technique used in this article cannot be used to identifythe drivers of variation over time in the sensitivity of interestrates to data news. But other research suggests it may berelated to factors such as the degree of uncertainty and riskaversion, central bank communication, and the absolute sizeand direction of data news. In recent years, it could also havebeen affected by the low level of interest rates and theirproximity to their ‘lower bound’, and changes in the structureof financial markets.

(1) Consistent with this, Pericoli and Veronese (2015) find that interest rates respond lessto data news during periods when there is more disagreement among forecastersabout upcoming macroeconomic releases.

270 Quarterly Bulletin 2015 Q3

Appendix

This appendix describes the data used throughout the articleand explains in detail the econometric method underlying theanalysis.

DataThe sample considered comprises daily data on marketinterest rates, exchange rates and macroeconomic data news.We consider 110 different types of macroeconomic datareleases (24 UK series, 31 US series and 55 euro-area series).These include activity indicators (such as GDP or retail sales),inflation estimates (such as CPI inflation), labour marketindicators (such as the unemployment rate) and businesssurveys (such as the purchasing managers’ index (PMI)). Thesample period is from January 2003 to May 2015.

Market interest rates are measured using the Bank ofEngland’s zero-coupon government bond yields. To capturethe effect on both short and long-term interest rates,estimations are performed using both the three-year andten-year spot government bond yields. Spot exchange ratesare used as provided by Bloomberg.

The daily change in market interest rates and exchange rates ismeasured as the change from close of business on the day ofthe release relative to the rate at close of business on the dayprior to the release.(1) A one-day window is used to try toisolate the effect of the data release. But it is possible thatthis window will also include a range of other events whichcan affect interest rates.(2)

Macroeconomic data news is constructed by comparingheadline releases with market expectations, as measured bythe Bloomberg survey of market participants.(3) This involvesa survey of a panel of analysts from private institutions such asinvestment banks, which is conducted in the days prior to eachdata release. The number of respondents varies across datareleases and over time. The news component is calculated asthe difference between the headline data release and themedian expectation from the survey. While this survey allowsus to estimate the news component of data releases it is notperfect, not least because the pool of survey respondentsmight not be representative of the entire universe of investorsparticipating in the market.

MethodThe relatively low frequency of releases of individualmacroeconomic data series is an obstacle to measuring thetime variation in the sensitivity of market interest rates toparticular series over short horizons. To overcome this issuean economic surprise index (ESI) is constructed. It aggregatesdifferent types of macroeconomic releases into a(higher-frequency) single indicator. In order to account for therelative importance of different types of releases, one of the

approaches used to construct the ESI follows Swanson andWilliams (2014a, 2014b) in weighting news in each data seriesaccording to its average importance in explaining dailychanges in yields over the whole sample. Daily changes in thisESI are then compared to daily changes in the asset pricesconsidered to measure the sensitivity of these asset prices tomacroeconomic data news.

The weight placed on different data series and the sensitivityof the asset price to the ESI are estimated simultaneouslyfrom the following equation:

(1)

where ∆it represents daily changes in the asset priceconsidered, ατ is an annual dummy variable used to pick uptrends in the variation of asset prices, wi represent the weightsattached to different macroeconomic data news (xt

i), and δτ isthe aggregate sensitivity of the asset price in question to theESI (Σi=1

n wi xti ) which is allowed to vary in each calendar year.

εt is a residual.

This non-linear specification is estimated using maximumlikelihood. This allows joint estimation of the weights (wi) andthe aggregate sensitivity (δτ). However, to separately identifythese two sets of coefficients, it is necessary to normalise theaggregate sensitivity in a benchmark period.(4) The benchmarkperiod is chosen to be between the start of 2003 and the endof 2008, the part of our sample where Bank Rate was not nearits lower bound.

In order to obtain estimates of how the sensitivity of assetprices to data news changes over shorter periods of time thancaptured by the annual change in δτ, 90-day rolling OLSregressions are run using the daily ESI constructed above (W� = Σi=1

n wixti ).(5) The specification used is:

(2)

A variable selection process is conducted in order to obtain amore precise estimation when using the methodology thatweights different macroeconomic series according to theirhistorical impact on interest rates. This narrows down thesample of over 100 types of macroeconomic data news fromthe United Kingdom, United States and the euro area by only

(1) Special consideration is given to UK bank holidays, as US and euro-area data releaseson those days are allowed to affect UK interest rates on the following day.

(2) As discussed, a potential extension to the work discussed in this article which mighthelp to alleviate this problem would be to use intraday data to more accuratelycapture the change in interest rates around the precise time of the data release.

(3) For some of the data releases, the forecasts provided by Bloomberg are based onunderlying data from Markit.

(4) In practice this means making δτ average 1 over 2003–08.(5) Standard errors are obtained using a bootstrap to account for the sampling

uncertainty derived from using a constructed regressor.

Topical articles UK market interest rates and macroeconomic data news 271

considering those that significantly affect the asset price inquestion. In the first stage, all data news from a given origin(United Kingdom, United States or euro area) are consideredin turn in three separate regressions. Those variables found tobe significant at the 5% level in this first stage are taken intoaccount for a second stage, in which variables from differenteconomies are considered together. This interaction renders

some of these variables non-significant, and hence they aresubsequently dropped before confirming a final set ofsignificant releases.(1) This process is done separately for eachasset price considered, as the set of significant data newsmight change depending on asset characteristics. Once theset of asset-specific significant data news is formed, themethodology described above can be applied.

(1) Data releases with less than 25 non-zero news observations throughout the sampleare also dropped, as are those which have a negligible impact (an impact of less than1 basis point in bonds and 0.1% in exchange rates per standard deviation of news).

272 Quarterly Bulletin 2015 Q3

References

Altavilla, C, Giannone, D and Modugno, M (2014), ‘The low frequency effects of macroeconomic news on government bond yields’, Board ofGovernors of the Federal Reserve System Finance and Economics Discussion Series No. 2014-52.

Andersen, T G, Bollerslev, T, Diebold, F X and Vega, C (2003), ‘Micro effects of macro announcements: real-time price discovery in foreignexchange’, American Economic Review, Vol. 93, No. 1, pages 38–62.

Andersen, T G, Bollerslev, T, Diebold, F X and Vega, C (2007), ‘Real-time price discovery in global stock, bond and foreign exchange markets’,Journal of International Economics, Vol. 73, No. 2, pages 251–77.

ap Gwilym, O, Buckle, M, Clare, A D and Thomas, S H (1998), ‘The transaction-by-transaction adjustment of interest rate and equity indexfutures markets to macroeconomic announcements’, Journal of Derivatives, Vol. 6, No. 2, pages 7–17.

Balduzzi, P, Elton, E J and Clifton Green, T (2001), ‘Economic news and bond prices: evidence from the US Treasury market’, Journal ofFinancial and Quantitative Analysis, Vol. 36, No. 4, pages 523–43, Cambridge University Press.

Brooke, M, Danton, G and Moessner, R (1999), ‘News and the sterling markets’, Bank of England Quarterly Bulletin, Vol. 39, No. 4,pages 374–83, available at www.bankofengland.co.uk/archive/Documents/historicpubs/qb/1999/qb990402.pdf.

Clare, A and Courtenay, R (2001), ‘Assessing the impact of macroeconomic news announcements on securities prices under different monetarypolicy regimes’, Bank of England Working Paper No. 125, available atwww.bankofengland.co.uk/archive/Documents/historicpubs/workingpapers/2001/wp125.pdf.

Ederington, L H and Lee, J H (1993), ‘How markets process information: news releases and volatility’, Journal of Finance, Vol. 48, No. 4,pages 1,161–91.

Ehrmann, M and Fratzscher, M (2005), ‘Exchange rates and fundamentals: new evidence from real-time data’, Journal of International Moneyand Finance, Vol. 24, No. 2, pages 317–41.

Evans, M D D and Lyons, R K (2008), ‘How is macro news transmitted to exchange rates?’, Journal of Financial Economics, Vol. 88, pages 26–50.

Faust, J, Rogers, J H, Wang, S-Y B and Wright, J H (2007), ‘The high frequency response of exchange rates and interest rates to macroeconomicannouncements’, Journal of Monetary Economics, Vol. 54, No. 4, pages 1,051–68.

Goldberg, L S and Grisse, C (2013), ‘Time variation in asset price responses to macro announcements’, Federal Reserve Bank of New York StaffReports No. 626.

Gürkaynak, R S, Sack, B and Swanson, E (2005), ‘The sensitivity of long-term interest rates to economic news: evidence and implications formacroeconomic models’, American Economic Review, Vol. 95, No. 1, pages 425–36.

Joyce, M and Read, V (1999), ‘Asset price reactions to RPI announcements’, Bank of England Working Paper No. 94, available atwww.bankofengland.co.uk/archive/Documents/historicpubs/workingpapers/1999/wp94.pdf.

Joyce, M, Tong, M and Woods, R (2011), ‘The United Kingdom’s quantitative easing policy: design, operation and impact’, Bank of EnglandQuarterly Bulletin, Vol. 51, No. 3, pages 200–12, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/qb110301.pdf.

Malik, S and Meldrum, A (2014), ‘Evaluating the robustness of UK term structure decompositions using linear regression methods’, Bank ofEngland Working Paper No. 518, available at www.bankofengland.co.uk/research/Documents/workingpapers/2014/wp518.pdf.

Pericoli, M and Veronese, G (2015), ‘Forecaster heterogeneity, surprises and financial markets’, Banca d’Italia Working Papers No. 1020.

Swanson, E T and Williams, J C (2014a), ‘Measuring the effect of the zero lower bound on medium- and longer-term interest rates’, AmericanEconomic Review, Vol. 104, No. 10, pages 3,154–85.

Swanson, E T and Williams, J C (2014b), ‘Measuring the effect of the zero lower bound on yields and exchange rates in the UK and Germany’,Journal of International Economics, Vol. 92, pages S2–S21.

Topical articles Estimating market expectations of changes in Bank Rate 273

• The Bank uses a variety of methods to extract information about market participants’expectations of the future path of Bank Rate.

• This article examines some techniques for estimating, using market prices, market expectationsof the timing of future changes in Bank Rate and the probability of Bank Rate being changedwithin a given period of time.

• These techniques are useful because the expected timing of changes in Bank Rate cannot bedirectly inferred from the mean expected path of the level of Bank Rate.

Estimating market expectations of changes in Bank RateBy David Elliott of the Bank’s Sterling Markets Division and Joseph Noss of the Bank’s Capital Markets Division.(1)

(1) The authors would like to thank Nicola Anderson, Ben Morley, David Murphy andMatthew Osborne for their help in producing this article.

Overview

The Bank of England’s Monetary Policy Committee (MPC)sets its policy rate, Bank Rate, in order to influence marketinterest rates, the level of activity in the economy, andinflation. People’s expectations of future Bank Rate can havean important influence on economic activity. The MPC is,therefore, naturally interested in understanding how marketexpectations of its future policy are evolving.

The Bank uses a combination of surveys and financial marketprices — interest rates and options prices — to obtaininformation on market participants’ expectations of thefuture level of Bank Rate. But it is also interested inexpectations of the timing of future changes in Bank Rate.There is an important distinction between these twoexpectations. While market interest rates and optionsprices provide direct information on the expected level ofBank Rate at a given point in time, they do not, bythemselves, provide direct information on the time atwhich Bank Rate is expected to change. To estimate thelatter requires some assumptions to be made aboutBank Rate’s possible future paths.

One means of doing so, discussed in this article, is theLibor Market Model (LMM), a framework originally developedto price interest rate derivatives. The LMM can also be usedto estimate the probability that market participants attachto a change in the level of Bank Rate occurring within a givenperiod of time.

The techniques described here are not exhaustive, and noneoffers a definitive view of market participants’ expectationsof future Bank Rate or the timing of its changes. Allestimates involve a significant degree of uncertainty and relyon assumptions. In particular, those derived from marketprices are affected by credit, liquidity and term premia.The Bank therefore continues to monitor and develop arange of indicators to assess market expectations ofmonetary policy.

274 Quarterly Bulletin 2015 Q3

The Bank of England’s Monetary Policy Committee (MPC) setsBank Rate, the interest rate paid on commercial banks’reserves at the Bank of England. Expectations of the futurelevel of Bank Rate — including the timing and pace of itschanges — affect a range of market interest rates thatinfluence financial asset prices and the cost of credit in thewider economy. The MPC is, therefore, naturally interested inunderstanding how these expectations are evolving. Suchexpectations may also contain useful information aboutinvestors’ perceptions of economic developments.

The Bank has for some time used surveys to gatherinformation about expectations of the future level ofBank Rate. Such expectations can also be estimated from theinterest rates on instruments traded in financial markets. Andthe prices of options contracts written on these instrumentscan give a guide to the weight that investors attach todifferent possible levels of interest rates around these centralexpectations.

Surveys and financial market prices can also be informativeabout market participants’ expectations of the timing offuture changes in Bank Rate. But there is an importantdistinction between deriving market expectations of futurelevels and future changes in Bank Rate. Market interest rates,in themselves, can provide information on the time at whichthe expected level of Bank Rate reaches a certain point; butderiving expectations about the time at which Bank Ratechanges requires further assumptions about its possible paths.

This article begins by reviewing the means by which the Bankextracts information about expectations of the future level ofBank Rate. It then explains some techniques used to extractinformation about the timing of future changes in Bank Rate— including the Libor Market Model (LMM), a frameworkoriginally developed to price interest rate derivatives.

Estimating market expectations of the futurelevel of Bank Rate

The Bank uses a combination of market-based andsurvey-based measures to estimate expectations of the futurelevel of Bank Rate. This section discusses these measures.(1)

Market-based measuresThe Bank uses forward interest rates — that is, the rates atwhich investors can agree today to borrow or lend over someperiod beginning in the future(2) — to assess marketexpectations of the future level of Bank Rate. These forwardrates can be derived from a range of financial instruments suchas bonds, futures and swaps.

Few financial instruments reference Bank Rate directly. Butforward rates can be derived from overnight index swaps (OIS)

and instruments referencing Libor (the London interbankoffered rate). OIS are contracts involving payments based onthe average overnight interest rate that prevails over theirlifetime. For sterling contracts, the relevant overnight interestrate is the sterling overnight index average (SONIA).(3) AndLibor is a quoted measure of the interest rates at which banksjudge they can borrow from other banks.(4)

Forward rates derived from these market interest rates do notcorrespond perfectly to market participants’ expectations ofBank Rate. This is because they include varying levels ofcredit, liquidity and term premia, which compensate investorsfor, respectively, the risk of counterparty default; the cost oftaking or closing a position in the contract in the future; andthe risk associated with borrowing or lending over a longperiod, compared to undertaking a series of (otherwiseidentical) shorter-term transactions.(5)

Since SONIA is an overnight rate, it contains a smaller creditpremium than the rates underlying forward Libor rates, whichare generally of a longer maturity. For this reason, forwardOIS rates are the Bank’s preferred means of inferring marketexpectations of the level of Bank Rate. Market intelligencesuggests that participants in interest rate markets also use theforward OIS curve as a summary measure of marketexpectations of future monetary policy. Over the pastthree years, SONIA has tended to trade a few basis pointsbelow Bank Rate. This ‘wedge’ has to be considered wheninterpreting the forward OIS curve as a measure of Bank Rateexpectations, and is discussed in more detail in the box onpage 275.

Survey-based measuresThe Bank also uses surveys to gather information on Bank Rateexpectations. One such survey is the Reuters survey ofeconomists, which has been running since the late 1990s.While the precise format has varied, in recent years the surveyhas asked economists for their expectations of futureBank Rate both after the next MPC meeting, and at the end ofevery quarter thereafter, out to a horizon of around18 months. Other surveys of monetary policy expectationsinclude the Bank’s quarterly survey of external forecasters(SEF) — which is carried out in advance of the Inflation Report— and surveys carried out by Bloomberg and HM Treasury.

Survey measures have the advantage of not being affected bycredit, liquidity and term premia, or the SONIA-Bank Rate

(1) For more detail, see Joyce and Meldrum (2008).(2) For example, the one-year interest rate, five years forward, is the rate at which

investors can currently agree to borrow or lend for a one-year period starting infive years’ time.

(3) SONIA is compiled by the Wholesale Markets Brokers’ Association. For furtherdetails, see www.wmba.org.uk.

(4) Libor is compiled by the ICE Benchmark Administrator. For further details, seewww.theice.com/iba/libor.

(5) A number of recent studies estimate the level of term premia incorporated in marketinterest rates; for a summary, see Guimaraes (2014).

Topical articles Estimating market expectations of changes in Bank Rate 275

The wedge between SONIA and Bank Rate

Alongside the MPC decision on 5 March 2009, the Bankimplemented a ‘floor’ system for implementing monetarypolicy. Under this framework, all reserves balances areremunerated at Bank Rate. This should, in theory, keepovernight interest rates close to Bank Rate: were marketinterest rates to fall below Bank Rate, banks could earn arisk-free profit by borrowing reserves in the market anddepositing them with the Bank, where they earn Bank Rate.All else being equal, this should drive overnight interest ratesback towards Bank Rate.

Since the introduction of the floor system, overnight rateshave typically traded close to Bank Rate, with volatility athistorically low levels. But since mid-2012, overnight interestrates, as measured by SONIA, have traded at an average ofaround 7 basis points below Bank Rate, although this ‘wedge’has narrowed to around 4 basis points in recent months(Chart A). Market intelligence contacts report that bankshave been relatively unwilling to borrow cash offered bynon-banks (without reserves accounts) even at rates belowBank Rate, for two reasons:(1)

• First, there has been a fall in banks’ demand for short-termborrowing. This reflects several factors, including banks’ongoing efforts to reduce their reliance on short-termwholesale funding.

• Second, banks have become less willing to borrow toarbitrage overnight rates against their reserves accounts.Such borrowing increases their balance sheet size andleverage. As a result, banks may have increased the returnsthey require to justify a given quantity of borrowing.

This wedge complicates analysis of market expectations ofBank Rate and its future changes. Any estimate of theseexpectations must make an assumption about expectations ofthe level of the wedge in the future.

One way to gain an insight into the degree to which marketparticipants expect the wedge to persist is from the prices ofSONIA-Bank Rate basis swaps — contracts that involvepayments linked to the average value of the differencebetween SONIA and Bank Rate that has prevailed over theirlife. The prices of these securities allow for the construction ofa SONIA-Bank Rate forward curve, which, abstracting fromcredit, liquidity and term premia, gives an indication of theSONIA-Bank Rate wedge that market participants expect toapply at a given future time (Chart B). The SONIA-Bank Rateforward curve currently suggests that markets expect thewedge between the two rates to decrease over the next year.

There are, however, reasons to be cautious of drawing a firmconclusion on investor expectations of the future value of theSONIA-Bank Rate wedge from swaps. Market intelligencesuggests that the market for these swaps is relatively illiquid,with the volume of traded contracts far lower than that inother interest rate markets, including that for OIS. As such,their prices may embody liquidity premia to compensateinvestors for the ease with which they can take or closepositions. This may drive the prices of these instruments awayfrom a level commensurate with a pure read on investorexpectations of the wedge.

+

–

1.5

1.0

0.5

0.0

0.5

1.0

2006 08 10 12 14

Percentage points

Floor system introduced

Chart A The SONIA-Bank Rate wedge(a)(b)

Sources: Bloomberg and Bank calculations.

(a) SONIA minus Bank Rate.(b) Data to 30 July 2015.

+

–

5

4

3

2

1

0

1Basis points

July Oct. Jan. Apr. July Oct.2015 16

Chart B MPC-dated forward SONIA-Bank Rate basisswap rates(a)

Source: Bloomberg.

(a) Data to 30 July 2015.

(1) For further details see Jackson and Sim (2013).

276 Quarterly Bulletin 2015 Q3

wedge. But they are only carried out periodically, meaningthat they are not available for day-to-day monitoring. Theyare also typically carried out over a period of several days,which potentially complicates their interpretation if, forexample, the sample period coincides with economic news.

There is also the possibility that market prices and surveyscapture different types of information. Survey respondentsmight report their modal expectations (the outcome forBank Rate that they believe to be most likely), whereas marketprices are indicative of market participants’ mean expectations(the probability-weighted average outcome for Bank Rate). Ifthe distribution of possible outcomes for Bank Rate is skewed,this may cause expectations obtained from surveys to differfrom those derived from financial market prices, even if themodal expectation is the same in both cases.

Together, these considerations mean that the Bank usessurveys and market-based measures as complements, ratherthan relying on one or the other. The Bank also gathersmarket intelligence on monetary policy expectations throughits regular dialogue with market participants.

Deriving the market’s view of the distributionof possible future levels of Bank Rate

The measures of expectations discussed in the previoussection provide the Bank with a useful summary of marketviews about the future level of Bank Rate. But the Bank is alsointerested in the perceived balance of risks around thesecentral expectations. It therefore also monitors a range ofsurvey-based and market-based measures of the distributionof possible future levels of Bank Rate.

Since 2014, the Bank’s SEF and the Reuters survey ofeconomists have included questions about the probabilityrespondents attach to different future levels of Bank Rate, inaddition to the questions about their central projections.

It is also possible to derive market perceptions of thedistribution of possible future levels of Bank Rate from interestrate options. Options are contracts giving the holder the right(but not the obligation) to buy or sell an asset on (or before) aspecified future date at a specified price (the ‘strike’ price). Anoption’s price therefore reflects the weight that investorsattach to the possibility of the price of the underlying assetreaching the strike price. And if option prices for a range ofstrike prices are available, it is possible to infer informationabout the weight that investors attach to the underlying assetprice reaching these different levels.

As explained in the previous section, forward OIS rates are theBank’s preferred means of inferring market expectations of thelevel of Bank Rate. But options on OIS rates are not widely

traded. Instead, a guide to the market-implied distribution offuture levels of Bank Rate can be obtained from optionsreferencing Libor, which are more frequently traded.

One way to display this information is in the form of arisk-neutral probability density function (PDF).(1)(2) Chart 1shows a recent example of a risk-neutral PDF of three-monthLibor. Possible levels of the interest rate are measuredhorizontally and probability is measured vertically, so that thearea under the line corresponds to the probability arisk-neutral investor might attach to three-month Libor lyingwithin the corresponding range. The risk-neutral PDF has beenpositively skewed recently, which may reflect market viewsthat interest rates are close to their effective lower bound.

Estimating market expectations of futurechanges in Bank Rate

The previous sections discussed some ways of extractinginformation about market participants’ expectations of thelevel of Bank Rate at a given point in time, along with theweight they attach to outcomes around that centralexpectation. But the Bank is also interested in expectations ofthe timing of changes in Bank Rate. This section begins bydescribing one simple proxy for measuring these expectations,and explains its limitations. It then introduces the LMM, atechnique that can be used to estimate market expectations ofthe future path of Bank Rate, including the expected timing ofits changes.

(1) See Breeden and Litzenberger (1978) and Clews, Panigirtzoglou and Proudman(2000). Estimated PDFs can be obtained fromwww.bankofengland.co.uk/statistics/Pages/impliedpdfs/.

(2) The probabilities shown in Chart 1 are those that would be perceived by a‘risk-neutral’ investor who was indifferent between a pay-off with certainty and agamble with the same expected pay-off. In the likely case that investors are in factrisk-averse, the actual probabilities they attach to different levels of Libor may differto those suggested by the risk-neutral PDF.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5Probability density

0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00Three-month Libor in December 2015

Chart 1 Option-implied risk-neutral probability densityfunction of Libor, as of 30 July 2015

Sources: Bloomberg and Bank calculations.

Topical articles Estimating market expectations of changes in Bank Rate 277

The techniques considered here can be applied whenBank Rate is at any level, and can be used to estimateexpectations of both increases and decreases in Bank Rate.The remainder of the article applies the techniques to thecurrent conjuncture, in which Bank Rate is at 0.5% and theupward-sloping forward OIS curve indicates that marketsgenerally expect the next change in Bank Rate to be anincrease.

A simple proxy based on OIS ratesUnder the assumption that market participants expect theMPC to change Bank Rate in increments of 25 basis points,one simple proxy for when markets expect Bank Rate tochange is the date at which forward OIS rates reach a level25 basis points above or below the current level of Bank Rate.For example, Chart 2 shows a recent forward OIS curve. Thissimple proxy would suggest that markets expected Bank Rateto change at the date when the curve reaches 0.75%. Marketintelligence suggests that participants in interest rate marketstypically use variants of this proxy to estimate marketexpectations of the timing of changes in Bank Rate.

This proxy can be adjusted to control for any differencebetween SONIA and Bank Rate. If market participants expecta difference to persist, then it might be appropriate tocompare forward OIS rates to SONIA, rather than Bank Rate.In practice, it is difficult to determine market expectations ofthe future difference with confidence (see the box onpage 275). For simplicity, therefore, the remainder of thearticle assumes that market participants do not expect asignificant difference to persist. The remainder of the articlealso abstracts from credit, liquidity and term premia.

The distinction between the expected level ofBank Rate and the expected timing of its changesThe date at which forward OIS rates reach 25 basis pointsabove or below the current level of Bank Rate may be areasonable measure of the date at which the expected levelof Bank Rate reaches that level. But that date may differ tothe date at which Bank Rate is expected to change to thatlevel.

This distinction is illustrated in Figure 1, which shows twopurely hypothetical future paths of Bank Rate, illustrated bythe red and blue lines. Market participants are assumed toplace a 50% probability on each. In the scenario shown by thered line, Bank Rate increases first in period 1 and quicklyincreases again in period 2. In the scenario shown by theblue line, Bank Rate remains unchanged for longer, beforeincreasing once in period 5. The green line shows the meanexpected level of Bank Rate at each point in time: since a 50%probability is attached to each of the two paths, this is thesimple average of the two. This line can be thought of as ahypothetical forward OIS curve.

There are two points to take from this example:

• The mean expected level of Bank Rate — as measured bythe green line — increases to 0.75% in period 2, because inthat period, one of the possible paths is at 0.5% and theother is at 1%.

• But the mean expected time of Bank Rate’s first increase islater, in period 3. This is because one of the possible pathsfirst increases in period 1 and the other does so in period 5;the mean time is halfway between these dates.

0.0

0.5

1.0

1.5

2.0

2.5

2015 16 17 18 19 20

Per cent

Chart 2 Forward OIS curve, as of 30 July 2015(a)

Sources: Bloomberg and Bank calculations.

(a) Instantaneous forward interest rates derived from the Bank’s OIS curves.

0.00

0.25

0.50

0.75

1.00

1.25

0 1 2 3 4 5 6

Blue Bank Rate path

Red Bank Rate path

Mean expected level

Time

Per cent

Blue path reaches 0.75%

Time of mean expected first change

Red path reaches 0.75%

Mean expected level reaches 0.75%

Figure 1 Two hypothetical future paths of Bank Rate(a)

(a) These paths of Bank Rate are entirely hypothetical and bear no relation to MPC policy.

278 Quarterly Bulletin 2015 Q3

This simple example shows that differences in the expectedlevel of Bank Rate do not necessarily correspond to theexpected timing of its changes. This means that the simpleproxy based on OIS rates may not give a clear read on marketexpectations of the timing of changes in Bank Rate.(1)

In order to estimate expectations of the timing of changes inBank Rate, it is therefore necessary to make some assumptionsabout its possible future paths. One means of doing so is touse the LMM,(2) a framework originally developed to priceinterest rate derivatives.(3)

The Libor Market ModelThe LMM uses information from current forward interest ratesand options prices to identify a statistical distributiongoverning the future path of interest rates. The analysis in thisarticle is based on an LMM calibrated so that the distributionof the forward rate at each maturity has a mean consistentwith the current forward OIS curve, and volatility consistentwith the prices of options on Libor.(4) More details areprovided in the annex at the end of this article.

While the option-implied PDFs discussed above provide anindication of market participants’ view of the distribution offuture Bank Rate at a given point in time, they do not provideany information on how markets expect Bank Rate to movebetween these points in time. The advantage of the LMM isthat it models changes in interest rates as well as their level,which means it can be used to estimate a wider range ofdistributions and statistics, under the assumptions set out inthe annex.

Once calibrated, the LMM can be used to simulate possiblepaths of interest rates, the distribution of which is consistentwith current forward rates and options prices. Two examplesof the simulated paths of forward rates are shown by thegreen and purple lines in Figure 2. These simulated paths canthen be used to make inferences about market expectations ofBank Rate.

For example, the mean expected time at which Bank Ratechanges to a given level corresponds to the average time atwhich the simulated paths of interest rates reach that level.The two paths in Figure 2 reach 0.75% at t1 and t2. In thissimple example, placing equal weight on each path, theexpected time at which Bank Rate changes to 0.75% wouldtherefore be the average of t1 and t2.

By simulating many such paths of forward interest rates, theLMM can be used to estimate a PDF showing the weight arisk-neutral market participant might place on Bank Ratechanging at different future times. A recent example of such aPDF is illustrated in Chart 3. Whereas the horizontal axis inChart 1 shows possible levels of interest rates at a given time,this axis in Chart 3 shows possible dates of changes in

Bank Rate. It is important to remember that these resultsabstract from all risk premia, as well as any expecteddifference between SONIA and Bank Rate. In practice, credit,liquidity and term premia, as well as any expected wedgebetween SONIA and Bank Rate, may lead market participants’true expectations to differ from the estimates presented inthis article.

(1) The forward OIS curve plots the means of a set of probability distributions overpossible interest rates at given points in time. On the other hand, the expected dateof the next change in Bank Rate is the mean of a probability distribution over dates atwhich a given rate is reached. The relationship between these distributions willgenerally not be simple.

(2) Other models that have been used for similar purposes include those in Andreasenand Meldrum (2015) and Bauer and Rudebusch (2014).

(3) See Brace, Gatarek and Musiela (1997), Miltersen, Sandmann and Sondermann (1997)and Jamshidian (1997).

(4) When using the LMM to make inferences about Bank Rate, it would be preferable touse the prices of options on Bank Rate or OIS. Since these are not widely traded,options on Libor are used here. This should be a reasonable approximation to theextent that the volatility of OIS rates is similar to the volatility of Libor.

0.0

0.5

1.0

1.5

2.0

2.5

3.0Per cent

Date

t1 t2

Figure 2 Two examples of simulated paths of forwardinterest rates

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08Probability density

2015 16 17 18 19 20 21

Chart 3 Estimated risk-neutral probability densityfunction of the time of the next increase in Bank Rate,as of 30 July 2015(a)

Sources: Bloomberg and Bank calculations.

(a) Estimated using the LMM.

Topical articles Estimating market expectations of changes in Bank Rate 279

Chart 4 compares the mean and mode of the estimateddistribution of the expected timing of the first rise inBank Rate. This provides an indication of how the location andshape of the distribution shown in Chart 3 have changed overtime. For most of the sample, the mean date has been laterthan the modal (most likely) date. This is because theestimated distribution has tended to be positively skewed;that is, market participants appear to place more weight onthe first increase in Bank Rate occurring after the estimatedmodal date than before it. The gap between the mean andmode has also varied over time. In particular, this gapwidened in late 2014 and early 2015, as market interest ratesfell and the estimated mean expected time of the first increasein Bank Rate moved later.

The LMM can also be used to estimate the weight that arisk-neutral market participant might attach to Bank Ratechanging within a given period of time. Chart 5 shows howsuch estimates have varied over time.

Conclusion

This article has discussed a range of market-based andsurvey-based tools that can be used to estimate marketexpectations of Bank Rate. Some of these methodologies areused to estimate market expectations of the level of Bank Rateat a given point in time, and some for measuring marketexpectations of the timing of its changes.

None of the methods discussed offers a definitive view ofexpectations of the future path of Bank Rate. The estimatesderived from all of them are subject to uncertainty and rely onassumptions. In particular, those derived from market pricesare affected by credit, liquidity and term premia, as well as anydifference between SONIA and Bank Rate. The Bank thereforecontinues to monitor and develop a range of indicators toassess market expectations of future monetary policy.

2012

2013

2014

2015

2016

2017

2018

2019

2012 13 14 15

Estimated mean date

Estimated modal date

Chart 4 Estimated mean and modal expected dates ofthe next increase in Bank Rate(a)

Sources: Bloomberg and Bank calculations.

(a) Estimated using the LMM. Data to 30 July 2015.

0

20

40

60

80

100

2012 13 14 15

One year

Two years

Three years

Per cent

Chart 5 Estimated risk-neutral probability of aBank Rate increase within one, two and three years(a)

Sources: Bloomberg and Bank calculations.

(a) Estimated using the LMM. Data to 30 July 2015.

280 Quarterly Bulletin 2015 Q3

AnnexOutline of the Libor Market Model

Full technical details of the Libor Market Model (LMM) aredescribed in Brace, Gatarek and Musiela (1997). This annexprovides an outline based on the implementation and notationin Brigo and Mercurio (2006).

The LMM specifies a set of equations governing changes ineach of a set of forward interest rates. The parameters ofthese equations are then chosen to minimise the sum ofsquared differences between observed implied interest ratevolatilities and those given by the model. The remainder ofthis annex discusses this procedure in more detail.

Evolution of forward interest ratesThe LMM models a set of forward interest rates of maturities,T1, T2, …, TM.

Each forward interest rate is assumed to follow a ‘diffusionprocess’, whereby the rate of increase in the interest rate inany small space of time dt is the sum of a deterministic rate ofincrease (or drift) and a normally distributed randomfluctuation:

where Fi (t) denotes the forward interest rate applyingbetween times Ti-1 and Ti viewed at time t;µi denotes the deterministic rate of drift;σi denotes the volatility of the return on the forwardrate Fi (t); anddWi denotes normally distributed random fluctuationswith mean zero and variance dt.

The random fluctuations have correlation given by:

Given this specification, it can be shown that in order toensure that the modelled forward interest rates do not permitinvestors the opportunity of arbitrage (that is, the opportunityto generate riskless profit by agreeing to borrow and lend atdifferent forward interest rates simultaneously), the drifts µi

must satisfy the relationship:

where the function q(t) indexes a set of times t where Tq (t)-1 < t < Tq (t). This implies that the drift of forward rate Fi

is determined by its volatility σi and by past forward rates Fj

and their volatilities σj.

Functions for volatility and correlationAbsent any other assumptions, calibrating the LMM to a largenumber of forward rates becomes very computationallyburdensome. Fitting the model to n forward rates necessitatesthe calibration of n × n variance and covariance parameters.

In order to improve the model’s tractability, functional formsare imposed that govern how the volatilities associated witheach forward rate, and the correlations between them, varyover time. In the specification used here, these functions takethe form

and

where a, b, c, d and β are constant parameters to beestimated.

This specification is used elsewhere in the recent literature.(1)

It reduces the complexity of the model, while permittingsufficient flexibility for the modelled volatilities to matchthose observed in the market. Intuitively, it implies that thevolatilities of forward rates of maturities close together aremore highly correlated than those of forward rates ofmaturities far apart.

CalibrationCalibrating the model amounts to choosing the parametersa, b, c, d and β in order to minimise the difference between theimplied volatilities at different horizons given by the model,and those observed in the market.

To do so, it is useful to express the implied volatilities given bythe model as an analytical function of the parameters to beestimated. Here the analytical approximation developed byRebonato (1999) is used:

where is the (squared) implied volatility of theinterest rate applying between times p and q, Sp, q (0) is theforward swap rate between times p and q, and the weightswi (t) are given by the formula:

dF tF

dt t dW( )

( ) i , ,Mi

ii i i for 1µ σ= + =

E dW t dW t( ) ( ) i, j , ,M.i j i j, for 1ρ( )= =

t tt F t

F t

( ) ( )

( )i ij i j j j

j jj q t

i ,

( ) 1∑µ στ ρ σ

τ( ) ( )=+=

σ ( )( ) ( )= − + +( )− −t a T t b e di i

c T ti

ei j

i j

,ρ = β− −

w F w F

St t dt( ) ( )p q

LFM i i j j

p qi j p

q

i j i

T

j,,

, ,

p(0) (0) (0) (0)

(0)

2

21 0∑ ∫υ ρ σ σ( ) =

= +

p qLFM,

2υ( )

(1) See summary in Brigo and Mercurio (2006).

Topical articles Estimating market expectations of changes in Bank Rate 281

where P (t, u) is the price of a bond at time t with tenor u – t,and τi is the fraction of time between Ti–1 and Ti.

Numerical optimisation is then used to choose parametersthat minimise the sum of squared differences between thepredicted implied volatilities and observed implied volatilities;that is:

The distinction between expected levels and expectedchangesOnce calibrated, the LMM can be used to estimate the time atwhich Bank Rate is first expected to increase to a given levelx%, that is:

where Bt refers to Bank Rate in period t and E is theexpectations operator.

This time may differ to the time at which forward overnightindex swap rates first reach x%, which instead indicates thetime at which the expected level of Bank Rate first reaches x%(abstracting from the effects of term premia and theSONIA-Bank Rate wedge). That is:

w tP t T

P t t( )

( , )

( , )i

i i

k kk p

q

1∑τ

τ=

= +

a,b ,c ,d , argmin .p qLFM

p qa,b,c,d, , ,

2 ∑β υ σ( ) ( )= −β

E t B xmin : %t{ }≥

t E B xmin : % .t{ } ≥

282 Quarterly Bulletin 2015 Q3

References

Andreasen, M and Meldrum, A (2015), ‘Market beliefs about the UK monetary policy lift-off horizon: a no-arbitrage shadow rate termstructure model approach’, Bank of England Staff Working Paper No. 541, available atwww.bankofengland.co.uk/research/Documents/workingpapers/2015/swp541.pdf.

Bauer, M and Rudebusch, G (2014), ‘Monetary policy expectations at the zero lower bound’, Federal Reserve Bank of San Francisco WorkingPaper No. 18.

Brace, A, Gatarek, D and Musiela, M (1997), ‘The market model of interest rate dynamics’, Mathematical Finance, Vol. 7, Issue 2, pages 127–55.

Breeden, D and Litzenberger, R (1978), ‘Prices of state-contingent claims implicit in options prices’, Journal of Business, Vol. 51, No. 4,pages 621–51.

Brigo, D and Mercurio, F (2006), Interest rate models — theory and practice: with smile, inflation and credit, Springer Finance.

Clews, R, Panigirtzoglou, N and Proudman, J (2000), ‘Recent developments in extracting information from options markets’, Bank of EnglandQuarterly Bulletin, February, pages 50–60, available at www.bankofengland.co.uk/archive/Documents/historicpubs/qb/2000/qb000101.pdf.

Guimaraes, R (2014), ‘Expectations, risk premia and information spanning in dynamic term structure model estimation’, Bank of EnglandWorking Paper No. 489, available at www.bankofengland.co.uk/research/Documents/workingpapers/2014/wp489.pdf.

Jackson, C and Sim, M (2013), ‘Recent developments in the sterling overnight money market’, Bank of England Quarterly Bulletin, Vol. 53,No. 3, pages 223–32, available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2013/qb130304.pdf.

Jamshidian, F (1997), ‘Libor and swap market models and measures’, Finance and Stochastics, Vol. 1, No. 4, pages 293–330.

Joyce, M and Meldrum, A (2008), ‘Market expectations of future Bank Rate’, Bank of England Quarterly Bulletin, Vol. 48, No. 3, pages 274–82,available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/qb080301.pdf.

Miltersen, K, Sandmann, K and Sondermann, D (1997), ‘Closed form solutions for term structure derivatives with log-normal interest rates’,The Journal of Finance, Vol. 52, No. 1, pages 409–30.

Rebonato, R (1999), ‘On the simultaneous calibration of multi-factor log-normal interest-rate models to Black volatilities and to the correlationmatrix’, The Journal of Computational Finance, Vol. 2, No. 4, pages 5–27.

Topical articles Over-the-counter (OTC) derivatives and central clearing 283

• Over-the-counter (OTC) derivatives markets have grown significantly over recent decades, andthe United Kingdom is an important international centre for them. These markets facilitate thehedging of risk, but they can also give rise to complex exposures within the financial system.

• Following the financial crisis, policymakers have promoted reforms to these markets. Theseinclude the greater use of central counterparties (CCPs) to ‘centrally clear’ transactions, managingrisk within the system.

• The concentration of risk within CCPs does however highlight other challenges, including the needfor supervisory co-operation internationally. Authorities are working to address these issues.

Over-the-counter (OTC) derivatives,central clearing and financial stabilityBy Arshadur Rahman of the Bank’s Financial Market Infrastructure Directorate.(1)

Overview

Derivatives are contracts that derive their value from anunderlying asset (equities for example) or reference price(such as interest rates). They can be used to mitigate avariety of financial risks. OTC trades are those transactedbilaterally between parties, as opposed to being executed onan exchange. The market for OTC derivatives has grown overthe past two decades, and as of December 2014 stood atapproximately US$630 trillion in terms of outstandingnotional value. When OTC derivatives are traded bilaterally,they involve the risk that a counterparty fails to meet theirobligations under the contract. This risk can be mitigated byusing a CCP to centrally clear the transaction. The CCP actsas buyer to every seller, and seller to every buyer, simplifyingthe network of exposures within the system.

Central clearing is therefore recognised as a key way tomanage systemic risk. Following the financial crisis of 2007–09, G20 leaders agreed to reform the structure of OTCderivatives markets, requiring that contracts which aresufficiently standardised be centrally cleared. TheUnited States and Japan have already implemented this‘clearing obligation’ for certain interest rate and creditderivative contracts. The European Union is scheduled to doso in 2016. Approximately 50% of interest rate contractsand 20% of credit derivative contracts outstanding globallyare now centrally cleared. The proportion of the flow of newcontracts which is centrally cleared is higher still: since theintroduction of the clearing obligation in the United States in2013, for example, 80% of new interest rate contracts and70% of new credit derivative contracts have been centrally

cleared. In managing risk in the financial system, CCPs dohowever concentrate risk within themselves. This willbecome an increasingly important consideration, since thereis likely to be further migration towards central clearing, andmore concentration of activity among a small number ofCCPs and their users. While some of these CCPs are locatedin the United Kingdom, at the end of 2014 60% of themargin required to support transactions cleared by UK-basedCCPs was provided by non-UK participants, underscoring theglobal nature of this activity.

Regulators around the world are addressing the increasedconcentration and cross-border nature of central clearingactivity, by working together to ensure consistency ofapproach across jurisdictions based on robust regulatorystandards. This is in terms both of the scope of thederivatives contracts which are captured by the clearingobligation, and also in the application of prudentialrequirements for CCPs. Market fragmentation may result ifrules are not applied consistently in the different jurisdictionsin which CCPs and their participants operate. Authoritiesalso need to ensure that mechanisms are in place to mitigatethe risk arising from extreme circumstances in which CCPscould experience financial or operational difficulties.

Overall, while there has been significant progress inimproving the robustness of OTC derivatives markets over the past five years, there is work still to do. The Bank of England, along with other stakeholders, willcontinue contributing to this work.

(1) The author would like to thank Paul Alexander for his help in producing this article.

284 Quarterly Bulletin 2015 Q3

Introduction

Over-the-counter (OTC) derivatives markets have grownsignificantly over the past two decades, and constitute asystemically important component of financial servicesactivity.(1) The use of central counterparties (CCPs) in OTCderivatives markets has also increased over this period. This isimportant because ‘central clearing’, as it is known, is a keyway of managing risk within the system.

Previous Bulletin articles have explained the function of CCPs,and identified increased banking sector exposures to them.(2)

This article builds on these by exploring the increasing use ofCCPs in the OTC derivatives market. It begins with anoverview of OTC derivatives markets and how these havegrown over time, before looking at the increased use of centralclearing. It then considers factors affecting the regulatorydecision to require certain contracts to be cleared via CCPs(the ‘clearing obligation’) and the outlook for a furtherincrease in central clearing. The article concludes bydiscussing some of the policy implications of thesedevelopments.

Setting the scene: OTC derivatives and theCCPs that clear them

The financial crisis of 2007–2009 highlighted shortcomings inthe identification and management of risk in OTC derivativesmarkets. There was a lack of transparency about the size ofbilateral positions in OTC derivatives contracts. Thecombination of opacity and concerns over the adequacy ofcollateral, and counterparty risk management arrangementsmore generally, created an environment in which confidencecould be lost rapidly.(3) This ultimately contributed tosignificant market disruption in the aftermath of the collapseof Lehman Brothers and the near-collapse of AIG in September2008, both of whom were major participants in OTCderivatives markets.

In response to the crisis, G20 leaders committed to reform thestructure of OTC derivatives markets and to improve theirtransparency. It was agreed in Pittsburgh in September 2009that all standardised OTC derivative contracts should betraded on exchanges or electronic trading platforms, whereappropriate, and cleared through CCPs; and that OTCderivatives contracts should be centrally reported to bodiescalled ‘trade repositories’.(4) The Financial Stability Board(FSB) has been tasked with monitoring progress with theimplementation of these reforms.(5)

What are OTC derivatives and how are they used?(6)

Derivatives are contracts that derive their value from anunderlying asset (for example, equities or commodities) orreference price (for example, interest rates, foreign exchange

rates or credit indices). Derivatives can be split into two broadgroups, based on how they are traded. Exchange-tradedderivatives (ETDs) are highly standardised contracts traded onregulated exchanges. OTC derivatives, by contrast, are tradedbilaterally between counterparties, and may have morebespoke terms. Central clearing of ETDs is a long-establishedpractice, but central clearing of OTC derivative contracts onlybecame widely available from the late 1990s.(7) In terms ofoutstanding notional value (the nominal amount referenced ina contract to calculate the cash flows arising) Chart 1 showsthat the OTC derivatives segment is much larger, constitutingaround 90% of the overall derivatives market.

Chart 2a shows that the notional value of outstanding OTCderivatives grew rapidly in the early 2000s, particularly from2006 to 2008. The notional value is a measure of activity butnot necessarily of economic exposure or of risk. Theoutstanding market value of contracts, shown in Chart 2b, issignificantly lower. Following the onset of the financial crisis,the growth of the market slowed. As of end-2014, the marketstood at approximately US$608 trillion by gross notionalvalue,(8) or US$20 trillion by market value.

By far the largest proportion of activity is in interest ratederivatives. These are contracts which are used to hedgeagainst the risk of changes in interest rates. For example, amanufacturer with a variable-rate bank loan may seek to swap

(1) See Murphy (2009).(2) See Nixon and Rehlon (2013) and Liu, Quiet and Roth (2015).(3) See Murphy (2013).(4) See G20 Communiqué, September 2009;

www.treasury.gov/resource-center/international/g7-g20/Documents/pittsburgh_summit_leaders_statement_250909.pdf.

(5) For further detail on the FSB’s work, see www.financialstabilityboard.org/what-we-do/policy-development/otc-derivatives/ and for the latest progress report on thereforms to derivatives markets, see www.financialstabilityboard.org/wp-content/uploads/OTC-Derivatives-Ninth-July-2015-Progress-Report.pdf.

(6) For further detail on how derivatives can be used, see the International Swaps andDerivatives Association’s (ISDA’s) information brochure;www2.isda.org/attachment/NjQ3Mw==/ISDA%20FINAL%202014.pdf.

(7) See Norman (2011).(8) This total excludes contracts which have not been allocated to one of the five main

asset classes, which increases the total to around US$630 trillion.

Over-the-counter US$630 trillion

Exchange traded US$65 trillion

Source: Bank for International Settlements.

(a) By outstanding gross notional value.

Chart 1 Composition of global derivatives contracts bytrading arrangement as at the end of 2014(a)

Topical articles Over-the-counter (OTC) derivatives and central clearing 285

the variable (and therefore uncertain) interest payments underthe loan arrangement for fixed-rate interest payments, toallow it to better plan its future payment obligations. Thenext largest proportion of activity is in foreign exchange (FX)derivatives, followed by credit derivatives. These arecontracts which are used to protect against the default of aparticular entity or group of entities to which the contractbuyer may be financially exposed. For example, a bank mayseek to protect itself against the default of a firm or group offirms to which it has extended loans. The annex to this articleprovides more detail on the different types of derivativecontracts.

What are CCPs and what is central clearing?OTC derivatives transactions necessarily involve counterpartycredit risk. This is the risk that one counterparty fails to meetits obligations to the other. In that case, the non-defaulting

party is exposed to losses due to adverse price movements inthe value of the portfolio until it is able to replace thedefaulter with a new counterparty.(1) This risk is particularlyimportant in OTC derivatives contracts because they mayhave a term of many years.

CCPs mitigate and manage counterparty risk within thesystem by ‘clearing’ transactions: in effect, they stand as thebuyer to every seller and the seller to every buyer within acentrally cleared market, thereby simplifying the network ofexposures within the system and reducing their size via‘multilateral netting’.(2) Reflecting these arrangements, eachmarket participant has only an aggregate counterpartyexposure to the CCP. This can be preferable to multipleexposures across a range of other, possibly less creditworthy,counterparties. However, this does concentrate risk within theCCP itself.

A CCP manages the counterparty credit risk that it faces in anumber of ways. First, it applies strict membership criteria towould-be direct participants, known as ‘clearing members’. Itthen requires clearing members to provide ‘margin’ (collateral)in the form of cash or other liquid assets to offset the risksrelated to the exposures to each member. These exposuresarise from transactions undertaken by the clearing memberboth on its own behalf and from transactions it undertakes onbehalf of its clients. Clearing members in turn collect marginfrom clients to manage their own counterparty exposure.CCPs calculate margin very conservatively to cover potentiallosses in the event of a counterparty default.(3)

Margin is provided in two forms: ‘initial margin’ is posted atthe beginning of a transaction to cover potential futureadverse changes in the value of the contract, and isrecalculated on a regular basis. Additional ‘variation margin’ isposted to cover actual adverse changes in the market value ofthe contract during its life. CCPs generally also requireclearing members to contribute to a mutualised default fund,which protects the CCP in the event that the margin it holds isinsufficient to cover losses on the positions of a defaultingmember.

The United Kingdom as a major international centrefor OTC derivatives activity and central clearingAccording to data collected by the Bank for InternationalSettlements (BIS), the United Kingdom is the single largestglobal venue for OTC derivatives activity: it accounts for

(1) For further detail, see Hull (2009) and Norman (2011).(2) When Party A and Party B trade a set of contracts, they can agree to net the

exposures which arise (‘bilateral netting’), but the net exposure cannot be furthernetted against Party A’s exposures to Party C. When all contracts in a market arecleared through a CCP, the CCP can go further and net the exposures which arisefrom Party A’s transactions with all its original counterparties, to a single netexposure (‘multilateral netting’). For further detail on this process, see Nixon andRehlon (2013).

(3) The calculation is conservative partly because the same methodology is applied to allparticipants — that is, it is not reduced for some participants by factoring in theestimated likelihood of default.

0

100

200

300

400

500

600

700

1998 2000 02 04 06 08 10 12 14

Interest rate

Credit

Foreign exchange

Equity

Commodity

US$ trillions

Chart 2 Size of global OTC derivatives markets(a) By outstanding gross notional value

0

5

10

15

20

25

30

35

1998 2000 02 04 06 08 10 12 14

Interest rate

Credit

Foreign exchange

Equity

Commodity

US$ trillions

Source: Bank for International Settlements.

(b) By outstanding market value of contracts

286 Quarterly Bulletin 2015 Q3

almost half of all global activity in interest rate derivatives,and over a third of global activity in foreign exchangederivatives contracts (Chart 3).(1) The United Kingdom is alsoa major centre for the central clearing of OTC derivativescontracts: it is home to four CCPs, which between themaccount for most of the cleared activity in OTC interest ratederivatives globally, and a substantial proportion of thecleared activity in the other asset classes.(2)

The size of the UK market and the systemic importance of some of the CCPs within it make it essential for theBank of England to understand and monitor the risks arisingfrom this sector. The Bank is the supervisor of UK CCPs, and infulfilling this function it works closely with other authorities inregulatory colleges (see Supervisory co-operation sectionlater). The Bank’s Financial Policy Committee (FPC) considersrisks to the UK financial system more broadly, including fromoutside the core banking system. It intends to conduct anin-depth analysis of the derivatives market during 2016.(3)

How has central clearing of OTC derivativescontracts changed over time?

The clearing obligationThe largest jurisdictions have implemented the G20commitment that standardised OTC derivatives contractsshould be centrally cleared, by establishing legal requirementsfor central clearing of specified types of transaction. InEurope, the framework for this ‘clearing obligation’ isestablished by the European Market Infrastructure Regulation,commonly known as EMIR, which was enacted inAugust 2012.(4)

It is the responsibility of the European Securities and MarketsAuthority (ESMA) to propose which products should besubject to mandatory clearing. This must then be approved bythe European Commission, Council and Parliament.(5) In theUnited States, the relevant provisions are detailed within theDodd-Frank Act, and enforced by the Commodity FuturesTrading Commission (CFTC) and the Securities and ExchangeCommission (SEC).(6)

The OTC derivatives contracts currently offered for centralclearing may come from any of the five asset classes shown inChart 2, but by far the largest in terms of notional valuescleared are the interest rate and credit derivative asset classes.Chart 4 shows that the proportion of the outstanding stock ofderivatives transactions in these two asset classes that arecentrally cleared has increased over recent years. As ofDecember 2014, it is estimated that around 50% of the overallmarket for interest rate and 20% of the overall market forcredit derivatives were centrally cleared. The FX asset class,while larger than the credit asset class and the second largestoverall, contains very few contracts that are currently offeredfor central clearing by CCPs: as a proportion of overall FXactivity, the amount centrally cleared is estimated to be lessthan 1%.(7)

Chart 4 shows the proportion of the stock of historicaltransactions that have been centrally cleared. When lookingonly at the flow of new transactions, the proportion ofcentrally cleared activity is notably higher. Chart 5 shows theproportion of the flow of new contracts that are centrallycleared in the United States (where the clearing obligation has

(1) Location was determined as being the location of the sales desk of the reportingentity; where no sales desk was involved in a deal, the trading desk was used todetermine the location of deals.

(2) The four CCPs located in the United Kingdom are: CME Clearing Europe Limited,ICE Clear Europe Limited, LCH.Clearnet Limited and LME Clear Limited.

(3) See Bank of England Financial Stability Report, July 2015;www.bankofengland.co.uk/publications/Documents/fsr/2015/fsrfull1507.pdf.

(4) For further detail on the EMIR implementation timetable, seewww.fca.org.uk/firms/markets/international-markets/emir.

(5) See ESMA webpages on the clearing obligation; www.esma.europa.eu/page/OTC-derivatives-and-clearing-obligation.

(6) See Dodd-Frank Act, available at www.sec.gov/about/laws/wallstreetreform-cpa.pdf.(7) See BIS (2013), ‘Triennial survey’, available at www.bis.org/publ/rpfx13.htm and

ForexClear.

US$1.35 trillion

US$0.63 trillion

US$0.20 trillion

US$0.10 trillion

US$0.48 trillion

United Kingdom

United States

France

Germany

Rest of the world

Chart 3 Average daily turnover by notional value ofglobal OTC derivatives in April 2013(a) Interest rate derivatives

US$1.69 trillion

US$0.64 trillion

US$0.28 trillion

US$0.22 trillion

US$1.38 trillion

United Kingdom

United States

Singapore

Japan

Rest of the world

Source: Bank for International Settlements.

(b) FX derivatives

Topical articles Over-the-counter (OTC) derivatives and central clearing 287

been in place since 2013) as a percentage of weekly aggregatetransaction volumes. Between 2013 Q4 and 2015 Q2, onaverage, 80% of new interest rate derivatives and 70% of newcredit derivative index contracts were centrally cleared.

Which contracts are (or will be) subject to the clearingobligation?To assess which contracts should fall under the clearingobligation, regulators may start by considering contracts thatare already offered for central clearing by CCPs. Alternatively,they may first assess the market as a whole to see whichcontracts are sufficiently liquid and standardised to justify theimposition of a clearing obligation.(1) In Europe, ESMA’s broadobjectives include the reduction of systemic risk. It willconsider various criteria when selecting classes of OTC

derivatives for mandatory clearing, including their degree ofstandardisation, their volume and liquidity, and the availabilityof fair, reliable and generally accepted pricing information.These are important factors in the CCP’s ability to valuecontracts, and sell the positions into a liquid market in theevent of a clearing member default. It is of critical importancethat CCPs are able to effectively risk-manage the classes ofderivatives that are subject to the clearing obligation. TheBank has contributed to ESMA’s determination process, byanalysing large volumes of confidential transaction-level dataon various interest rate, credit and FX derivatives, to assess theliquidity of these contracts and their consequent suitability forthe clearing obligation.

To date, regulators around the world have focused on fourmain interest rate derivatives contracts and one main type ofcredit default swap (CDS) contract for the clearing obligation.

• ‘Plain vanilla’ interest rate swaps (IRS) — contracts to swap afixed interest rate cash flow for a variable (or ‘floating’) rateone on some notional principal amount;

• Basis swaps — contracts to swap two cash flows, both ofwhich are based on floating rates;

• Forward rate agreements (FRAs) — short-term contractsthat specify the interest rate that will apply to theborrowing/lending of a notional principal amount starting atsome point in the future;

• Overnight index swaps (OIS) — short-term IRS contracts inwhich the floating payment is based on an index rate fordaily overnight unsecured lending; and

• CDS index contracts — these provide protection against thedefault of any member of a list (or index) of entities, knownas ‘names’. These differ from CDS covering only individualentities, known as ‘single name’ contracts.

The annex to this article provides further details on thedifferent types of contracts.

Table A summarises the scope of the clearing obligations ofthe United States, Japan and the European Union (EU). Theformer two have already been implemented, and the latter iscurrently in the process of being ratified.(2) Japan was the firstjurisdiction to implement a clearing obligation, focusing onyen-denominated contracts. The US regime began in early2013 and covers a much wider range of contracts,denominated in any of the four most globally liquid currencies— US dollar, euro, sterling and yen. The EU, which is due to

(1) Factors which regulators should consider are set out in the International Organizationof Securities Commissions’ ‘Requirements for mandatory clearing’, 2012;www.financialstabilityboard.org/2012/02/cos_120202/.

(2) http://europa.eu/rapid/press-release_IP-15-5459_en.htm?locale=en.

0

10

20

30

40

50

60

Apr. Aug. Dec. Apr. Aug. Dec. Apr. Aug. Dec.

OTC interest rate derivatives

OTC credit derivatives

2012 1413 15

Per cent

Sources: DTCC trade information warehouse reports and Bank calculations.

(a) When calculating the proportion of centrally cleared activity, the volume in respect of thetwo novated (CCP-intermediated) contracts (that is, buyer-CCP and CCP-seller) is halved, torepresent the original buyer-seller transaction and avoid inflating the numerator.

Chart 4 Proportion of the outstanding stock of OTCinterest rate and credit derivatives that is centrallycleared(a)

20

30

40

50

60

70

80

90

100

Oct. Feb. June Oct. Feb. June0

Single-currency interest rate derivatives

Credit derivative indices

Per cent

2013 14 15

Source: Financial Stability Board.

(a) Dashed lines represent the average percentage cleared over the observation period.

Chart 5 Percentage of new OTC derivatives transactionscentrally cleared in the United States(a)

288 Quarterly Bulletin 2015 Q3

begin implementation of the clearing obligation from 2016,could capture an even wider range of contracts, subject to theoutcome of a consultation on clearing contracts denominatedin a broader set of European Economic Area (EEA) currencies.(1)

Under EMIR, once the clearing obligation takes effect, it will bephased in. CCP clearing members will be required to centrallyclear all transactions involving designated contracts from sixmonths after the rules take effect, large financial companieswhich are not clearing members within twelve months,smaller financial firms within 18 months and any other firmssubject to the clearing obligation within three years. Thisrepresents a significant expansion in the range of firms whocentrally clear their transactions, with some, such asnon-financial companies engaging in non-hedging activityabove certain thresholds, being required to centrally clear forthe first time.

Looking ahead: what factors will affectfurther migration towards central clearing?

The outlook for further migration to central clearingCentral clearing is likely to increase over the next few years,driven by general market practice (that is, firms progressivelystreamlining their arrangements for operational convenience),the introduction of the clearing obligation in the EU, and, asexplained below, increased margin and capital costs in respectof non-centrally cleared transactions. However, it willprobably not increase to cover 100% of the overall market. Asmentioned earlier, there will always be a certain proportion ofcontracts that are inherently unsuited to central clearingbecause they are not sufficiently standardised and liquid.

Contracts that are not centrally cleared (because in theabsence of an applicable clearing obligation, marketparticipants have either chosen not to clear through a CCP, orcannot do so because no CCPs offer to clear the contract) willbe subject to new bilateral margin requirements. These aredue to be phased in from September 2016.(2) They will be setto mitigate bilateral counterparty risk on these contracts andto incentivise (or at the least not undermine) central clearing.In addition, most FSB member jurisdictions have madechanges to their prudential frameworks to require highercapital requirements for non-centrally cleared derivatives —this was also part of the original G20 commitment of 2009.

The FSB has analysed the extent to which outstanding OTCderivatives may be capable of being centrally cleared. Chart 6shows that of the interest rate derivatives transacted by16 large dealers, plain vanilla IRS contracts are the most‘clearable’, and the most cleared, followed by FRAs. Currently,just over US$90 trillion of IRS contracts (or roughly half themarket) is centrally cleared. The analysis suggests scope for afurther increase in central clearing.(3) Notional values tradedin CDS contracts are much lower than in interest ratecontracts, and the proportions currently centrally cleared arealso lower: approximately US$0.9 trillion of Europeancorporate index contracts and US$0.5 trillion of Americancorporate index contracts are centrally cleared (or 37% and41% respectively). More complex contracts such as

(1) www.esma.europa.eu/system/files/esma-2015-807_-_consultation_paper_no_4_on_the_clearing_obligation_irs_2.pdf.

(2) See Bank for International Settlements and the International Organization ofSecurities Commissions’ ‘Margin requirements for non-centrally cleared derivatives’,March 2015; www.bis.org/bcbs/publ/d317.htm.

(3) It should be noted that the FSB has drawn this data from a range of sources.Estimates of how many additional contracts can be cleared assume there are noconfounding contractual, legal or other issues.

Table A Clearing obligations for the United States, Japan and the EU(a)

Jurisdiction Asset class Effective from Currencies(b) Maturities

United States IRS 11 March 2013 USD, EUR, GBP, JPY 28 days–50 years (30 years for JPY)BasisFRA 3 days–3 years OIS USD, EUR, GBP 7 days–2 years CDS indices USD, EUR Mainly 5 years, some 3, 7, 10 years

Japan IRS 1 November 2012 JPY Up to 30 years BasisIRS 1 July 2014 JPY/EUR Up to 10 years BasisCDS indices 1 November 2012 JPY 5 years(c)

EU IRS 2016 USD, EUR, GBP, JPY 28 days–50 years (30 years for JPY)BasisFRA USD, EUR, GBP 3 days–3 yearsOIS 7 days–3 yearsCDS indices EUR 5 yearsIRS In consultation SEK 28 days–15 years

CZK, DKK, HUF, NOK, PLN 28 days–5 yearsFRA SEK 3 days–2 years

NOK, PLN 3 days–1 year

(a) The complete list of clearing obligations already in place around the world also includes: China (IRS), Korea (IRS) and India (FX Forwards).(b) Currency abbreviations: USD = US dollar, EUR = Euro, GBP = Pound sterling, JPY = Japanese yen, SEK = Swedish krona, CZK = Czech koruna, DKK = Danish krone, HUF = Hungarian forint, NOK = Norwegian krone and

PLN = Polish zloty.(c) For Japanese CDS index contracts, maturity is not specified in primary legislation, but currently only five-year contracts are centrally cleared.

Topical articles Over-the-counter (OTC) derivatives and central clearing 289

cross-currency swaps and cap/floor contracts within theinterest rate asset class (see annex) are generally lessstandardised and more challenging to risk manage for CCPs, sothey are not presently offered for central clearing.

Overall, this analysis implies that there is potential for themajority of plain vanilla IRS contracts, OIS, basis swaps andFRAs to be centrally cleared, along with a sizable proportion ofCDS index contracts on European and American corporateentities. Single-name corporate CDS contracts appear to alsohave potential for increased central clearing, although inpractice this is challenging because the aggregate outstandingnotional values comprise a large number of individualcontracts on different single names, each of which may havelimited liquidity.

Indicators of liquidity include the size and depth of the marketin a specific contract. This can be measured in a number ofways, including the average daily number of trades and tradevolumes (measured in notional values traded, or outstandingat a given point in time). Additional measures include marketconcentration, which is concerned with whether there are asufficient number of active market participants to enable aCCP to exit a derivatives position inherited from a defaultingclearing member.

Contracts have specific features that will affect their liquidity.Among the most important are the currency of denominationand contract maturity. Chart 7 shows that centrally clearedactivity for IRS contracts as measured by notional value tradedis concentrated at shorter maturities of up to five years. Themost liquid contracts by currency are those denominated inUS dollars, followed by euros. CDS markets are even moreconcentrated, both by maturity and by currency: five-yearcontracts are the most actively traded, and most contracts aredenominated either in euros or US dollars, with less trading inother currencies.(1) Therefore, when considering a givencontract for the clearing obligation, regulators also need todetermine which specific currencies and maturities aresufficiently liquid and therefore appropriate to capture underthe obligation.

Other likely market trends in central clearing over themedium termFurther migration towards use of CCPsWhile central clearing is likely to increase, this may not beentirely reflected in the outstanding notional values at CCPs

0 50 100 150 200

Interest rate swaps

Overnight indexed swaps

Basis swaps

Forward rate agreements

Cross-currency swaps

Swaptions

Caps and floors

Other

Centrally cleared

Offered for central clearing but not cleared

Not currently offered for central clearing

US$ trillions

Chart 6 Actual and potential central clearing of OTC derivativesby product typeOutstanding notional amounts at end-June 2015

(a) Interest rate derivatives — large dealers

0.0 0.5 1.0 1.5 2.0 2.5

European corporate indices

Americas corporate indices

Americas mortgage-backed indices

Asia-Pacific corporate indices

Sovereign indices

Americas single-name corporates

European single-name corporates

European single-name sovereigns

Asia-Pacific single-name corporates

Americas single-name sovereigns

Other

Centrally cleared

Offered for central clearing but not cleared

Not currently offered for central clearing

US$ trillions

Source: Financial Stability Board.

(a) Contracts specifying the Americas refer to both North and South American markets.

(b) Credit derivatives — all counterparties(a)

0

20

40

60

80

100

120

0–2 years

2–5 years

5–10 years

10–30 years

Euro

Japa

nese

yen

Poun

d st

erlin

g

Cana

dian

dol

lar

Aust

ralia

n do

llar

New

Zea

land

dol

lar

Swis

s fra

nc

Swed

ish

kron

a

US

dolla

r

US$ billlions

Sources: LCH.Clearnet and Bank calculations.

Chart 7 Average daily notional value traded for IRScontracts cleared during 2014 in LCH SwapClear

(1) See Benos, Wetherilt and Zikes (2013) for an examination of the CDS market.

290 Quarterly Bulletin 2015 Q3

because of the impact of trade compression. This is a tool toreduce outstanding exposures through cancelling trades thatresult in offsetting positions. It is therefore useful to look atdaily trading activity alongside outstanding notional values fora more accurate sense of centrally cleared activity.

Market concentration among CCPs and their clearingmembersA small number of CCPs clear most of the available OTCderivatives activity due to economies of scale in the provisionof clearing services — for example, the greater scope fornetting. This makes those CCPs systemically important.Furthermore, there are a relatively small number of clearingmembers for these CCPs, and fewer still that offer clientclearing.(1) Those clearing members that do offer clientclearing become more important within the system becausenon-clearing member firms would otherwise be unable toaccess central clearing, hindering their ability to undertakeOTC derivatives transactions (especially if these contractsbecome subject to the clearing obligation).

Chart 8 shows the significant derivative exposures betweenCCPs and other institutions within the UK financial system.From this, the central role played by individual CCPs is evident.

High degree of internationalisationA large proportion of OTC derivatives activity takes placeoutside the home jurisdiction of the CCP clearing the relevantmarket, involving at least one overseas participant: Chart 9shows that the majority of initial margin requirements at

UK CCPs was accounted for by clearing members not locatedin the United Kingdom themselves, with 39% provided byclearing members based outside the European Economic Area.

Given that so much OTC derivatives activity is cross-border bynature, the timing and scope of implementation of theclearing obligation in different jurisdictions is particularlyimportant.

Policy implications: the systemic importanceof CCPs and the internationalisation ofcentral clearing

The growth of central clearing, the concentration of activitywithin a few CCPs and their clearing members, and thecross-border nature of much OTC derivatives activity allcreate challenges. Policymakers have worked to address thesechallenges both at national levels and internationally, via theFSB and other bodies such as the Committee on Payments andMarkets Infrastructures (CPMI) and the InternationalOrganization of Securities Commissions (IOSCO). Thechallenges, and the responses of policymakers, are detailed inthis section.(2)

Enhancing CCP resilienceThe concentration of business in CCPs is increasing thesystemic importance of many CCPs and creates the risk that

(1) For example, there are currently 97 clearing members of LCH SwapClear and 21members of ICE Clear Europe’s CDS clearing service.

(2) The developments identified in this section are closely related to the four safeguardsfor a resilient and efficient global framework for central clearing identified by the FSBin 2012: (i) fair and open access by market participants to CCPs, based on transparentand objective criteria; (ii) co-operative oversight arrangements between all relevantauthorities, both domestically and internationally, that result in robust andconsistently applied regulation and oversight of global CCPs; (iii) resolution andrecovery regimes that ensure the core functions of CCPs are maintained during timesof crises and that consider the interests of all jurisdictions where the CCP issystemically important; and (iv) appropriate liquidity arrangements for CCPs in thecurrencies they clear. See FSB’s ‘OTC derivatives markets reforms: third progressreport on implementation’, June 2012; www.financialstabilityboard.org/wp-content/uploads/r_120615.pdf.

CCP

Other financial institutions

Sources: Prudential Regulation Authority and Bank calculations.

(a) Each node represents a financial institution and the size of the node is scaled by the totalamount of exposures to that institution.

(b) Exposures are measured by current market values, net of collateral. For any two institutionsin the network, the greater the exposures between them, the more closely they arepositioned.

Chart 8 Significant derivatives exposures within theUK financial system as at end-June 2014(a)(b)

40%

21%

39%

United Kingdom

European Economic Area (excluding United Kingdom)

Other jurisdictions

Sources: CME Clearing Europe, ICE Clear Europe, LCH.Clearnet Ltd, LME Clear and Bankcalculations.

Chart 9 Distribution of initial margin requirements atUK CCPs by location of clearing member (as at end-2014)

Topical articles Over-the-counter (OTC) derivatives and central clearing 291

they are perceived to be ‘too big to fail’. Internationalinitiatives to address this risk have focused on establishingstrong regulatory standards for CCPs, ensuring that they haverecovery plans, and ensuring that authorities have resolutionpowers in the extreme event that a CCP neverthelessexperiences distress.

Prudential standards It is clearly critical that CCPs are subject to strong riskmanagement standards. In 2012, regulators agreed newinternational standards for financial market infrastructures,including CCPs: the CPMI-IOSCO ‘Principles for financialmarket infrastructures’.(1) These set strong requirements inareas such as credit and liquidity risk management and theeligibility of different asset types to serve as collateral. InEurope, the Principles were effected through the relevantrequirements of EMIR, and in the United Kingdom they havebeen embedded in the Bank’s supervisory approach to CCPs.(2)

Internationally, the work of authorities to promote CCPresilience is ongoing.(3) For example, a review has beeninitiated of stress-testing arrangements across CCPs.(4)

Stress-testing methodologies are very important for theresilience of CCPs because they determine CCPs’ estimates ofthe losses they could face in extreme but plausiblecircumstances, which in turn drive the size of CCPs’ defaultfunds.

CPMI and IOSCO have also issued standards to encouragegreater transparency in the approach of CCPs in managingtheir risks. This highlights the important risk governance roleplayed by clearing members, their clients, and other relevantparties. Increased transparency should allow thesestakeholders to monitor the quality of CCP riskmanagement.(5)

Recovery and resolutionWhile every effort should be made to minimise the risk of CCPfailure, it is impossible to remove this risk entirely. And CCPsmay fail for a variety of reasons unrelated to the default of aclearing member (for example, through failures in their ITsystems or other operational infrastructure).

Regulators therefore also focus on recovery arrangements,which are intended to help the CCP remain viable even duringperiods of stress, so as to ensure continuity of critical clearingfunctions.(6) To this purpose, CPMI and IOSCO have issuedguidance on recovery arrangements for financial marketinfrastructures.(7) Tools identified in the report (and alreadyimplemented in the United Kingdom) include CCPs’ use of theright to call for additional funds from clearing members;haircutting (that is, applying a proportionate reduction ofvalue to) a CCP’s obligation to pay its clearing membersvariation margin; and, if necessary, terminating outstandingcentrally cleared contracts to cap the CCP’s exposure.

It is important that authorities also consider what action maybe necessary in the event recovery measures prove to beinsufficient. In such a scenario, however unlikely it is to arise,resolution might be necessary. The goal of a resolution regimeis to ensure that critical clearing functions are maintained tosupport financial stability, without exposing public funds toloss.

The FSB has published guidance on the types of resolutiontools that should be available to authorities.(8) These arebroadly similar to those which may be applied to banks — forexample the power to transfer all or part of a failing CCP’sactivities to an acquirer, or temporarily to a ‘bridge’ entity. Itis important that resolution authorities have a flexible toolkitin order to be able to respond to different circumstances. TheUK resolution regime already covers CCPs and the EuropeanCommission is expected to bring forward proposals for anEU-wide resolution regime for CCPs in the near future.

Central bank liquidity accessLiquidity management is first and foremost the responsibilityof CCPs themselves. EMIR, following the CPMI-IOSCOPrinciples, establishes strong requirements for CCPs in thisarea. But central banks can also play an important role as theultimate liquidity provider to CCPs, reflecting the systemicimportance of the services they provide. In 2012 a group ofmajor central banks adopted the policy that there should be‘no technical obstacles’ to the provision of liquidity to a CCPthat is fundamentally viable but experiencing a temporaryliquidity need.(9)

In November 2014, the Bank widened access to its sterlingfacilities to include CCPs.(10) This provides access to anaccount at the Bank, allowing CCPs to deposit sterling funds.It also means that CCPs experiencing a clearing memberdefault can (if they fail to source funding from the market) usethe margin posted with them by that member — or other

(1) See the Committee on Payment and Settlement Systems (CPSS — the former nameof CPMI) and the Technical Committee of the International Organization ofSecurities Commissions’ ‘Principles for financial market infrastructures’, April 2012;www.bis.org/cpmi/publ/d101a.pdf.

(2) See The Bank of England’s approach to the supervision of financial marketinfrastructures, April 2013;www.bankofengland.co.uk/financialstability/Documents/fmi/fmisupervision.pdf.

(3) Details on this can be found in the FSB’s work programme;www.financialstabilityboard.org/wp-content/uploads/FSB-Chairs-letter-to-G20-April-2015.pdf.

(4) See the Bank for International Settlements press release;www.bis.org/press/p150311.htm.

(5) See CPMI-IOSCO’s ‘Public quantitative disclosure standards for centralcounterparties’, February 2015; www.bis.org/cpmi/publ/d125.htm.

(6) Extensive analysis has been conducted on how best to distribute losses betweenclearing members in the event that these exceed the pre-funded prudentialresources of the CCP. For more detail, see Elliott (2013).

(7) See CPMI-IOSCO’s ‘Recovery of financial market infrastructures’, October 2014;www.bis.org/cpmi/publ/d121.pdf.

(8) See the Annex to the FSB’s ‘Key attributes of effective resolution regimes forfinancial institutions’, October 2014; www.financialstabilityboard.org/wp-content/uploads/r_141015.pdf.

(9) See the FSB’s ‘OTC derivatives market reforms: third progress report onimplementation’, June 2012; www.financialstabilityboard.org/wp-content/uploads/r_120615.pdf.

(10) See Bank of England press release;www.bankofengland.co.uk/publications/Pages/news/2014/144.aspx.

292 Quarterly Bulletin 2015 Q3

assets they may hold — as collateral to obtain sterling liquidityfrom the Bank. Using assets as collateral to obtain temporaryfunding may be preferable to the possible alternative of sellingthem off at a loss under volatile or illiquid marketconditions.(1)

Supervisory co-operationCCPs clearing OTC derivatives may have clearing membersfrom many different countries and may clear productsdenominated in a number of different currencies. The distressof such a CCP could therefore impact financial stability inother jurisdictions via losses incurred by institutions thatprovide critical services in that jurisdiction, or by disruptingfinancial market activity there. The authorities responsible forthe supervision of clearing members and markets, and forfinancial stability more broadly in those jurisdictions, thereforehave a legitimate interest in the soundness of such CCPs.

As home to global CCPs, the UK authorities have taken theinitiative to establish regulatory ‘colleges’ consisting of a widerange of other international authorities.(2) These collegesprovide a vehicle for the sharing of information about thoseCCPs and the approach that the Bank is taking to theirsupervision, and for those authorities to input into the Bank’sapproach. In addition, within the EU, EMIR establishedregulatory colleges of relevant EU authorities for each EU CCP.Co-operation between authorities will also be important in thecontext of the preparation of resolution plans for CCPs.

Broader international co-ordination and timelyimplementation of consistent clearing rulesOTC derivatives transactions frequently involve counterpartiesfrom more than one jurisdiction. For these parties to be ableto clear their transactions at the same CCP, the use of thatCCP must be permitted in both jurisdictions, especially whencentral clearing of a transaction is mandated. If CCPs wereunable to operate in certain jurisdictions, there would be a riskthat clearing and trading activity more generally wouldbecome fragmented along geographical lines. That wouldhave undesirable implications for access to markets and formarket liquidity.

One approach has been to subject CCPs to dual regulation,whereby authorities in one jurisdiction require a CCPestablished in another jurisdiction to be subject to their ownregulatory regime(s) as well as to the regulatory regime thatapplies in the CCP’s home jurisdiction. However this approachcarries costs: it is likely to involve some duplication of effortby regulators and it can also be burdensome for CCPs as, forexample, they may be required to put in place differentarrangements to meet regulatory requirements that havesimilar objectives but which are expressed differently. There isalso a risk that regulatory requirements in differentjurisdictions actually conflict, so that a CCP cannotsimultaneously comply with both regimes. Recognising these

issues, G20 leaders have encouraged jurisdictions to ‘defer’where appropriate to the rules of other jurisdictions.(3)

In the EU, EMIR establishes a regime based on the ‘recognition’of other jurisdictions, provided certain conditions are met,primarily that they apply requirements for CCPs which areequivalent to those which are set in the EU. Where theseconditions are met, ESMA will defer to the relevant non-EUauthority in the supervision of CCPs from that jurisdiction.(4)

It is also desirable that authorities in different jurisdictionsco-operate in considering which OTC derivatives productsshould be subject to a clearing mandate. EU and USauthorities have committed to do this.(5)

Potential adverse impacts of CCP policiesAs CCPs become more central to OTC derivatives markets, therisk increases that they could take actions which have theeffect of imposing stress on other parts of the financialsystem. For example, under the CPMI-IOSCO Principles, CCPsare expected to consider the potentially adverse effects of themodels they use to calculate clearing member marginrequirements. This includes in particular the impact ofprocyclicality, whereby margin requirements may increaserapidly during periods of market stress. In such situations,CCPs may require their clearing members to post more margindue to worsening market conditions, but these conditionsthemselves may make it more difficult to source thatadditional margin, and also further reduce the value of themargin already posted, thus driving up margin requirementseven more. Procyclicality can cause liquidity stress, becauseclearing members posting margin might have to findadditional liquid assets precisely at times when they are leastable to do so.(6) EMIR carries requirements specificallydesigned to mitigate these risks, and the potential formacroprudential authorities to have a role in influencing CCPmargin policies is also likely to be a subject of debate amongpolicymakers in the coming years.(7)

(1) In addition, the Bank and the European Central Bank agreed to extend the standingswapline between them to facilitate the provision of multi-currency liquidity by bothcentral banks to CCPs established in the United Kingdom and the euro area inMarch 2015 as part of a wider agreement on co-operative arrangements for CCPs. See Bank of England press release;www.bankofengland.co.uk/publications/Pages/news/2015/044.aspx.

(2) This puts into effect Responsibility E of the Principles, which requires authorities toco-operate with each other to promote the safety and efficiency of financial marketinfrastructures.

(3) Paragraph 71 of the September 2013 G20 Leaders’ St. Petersburg Declaration stated:‘We agree that jurisdictions and regulators should be able to defer to each otherwhen it is justified by the quality of their respective regulatory and enforcementregimes, based on similar outcomes, in a non-discriminatory way, paying due respectto home country regulation regimes.’ For the full statement, see https://g20.org/wp-content/uploads/2014/12/Saint_Petersburg_Declaration_ENG_0.pdf.

(4) Further information on this approach can be found on ESMA’s website;www.esma.europa.eu/page/Third-non-EU-countries.

(5) See the July 2013 ‘Path Forward’ press release;www.cftc.gov/PressRoom/PressReleases/pr6640-13.

(6) For a more detailed examination of procyclicality, see Murphy, Vasios and Vause(2014).

(7) See for example the European Systemic Risk Board (ESRB) report on the efficiency ofmargin requirements under EMIR to limit procyclicality;www.esrb.europa.eu/pub/pdf/other/150729_report_pro-cyclicality.en.pdf?3326abd623e59d361b84c385a69b6d04.

Topical articles Over-the-counter (OTC) derivatives and central clearing 293

Conclusion

In the EU, the clearing obligation is expected to take effectfrom 2016, first covering interest rate derivatives. This willfurther increase the centrally cleared share of the OTCderivatives market, bringing about a further simplification ofthe network of exposures within the system.

Managing the growing systemic importance of CCPs is critical.Ultimately, it will determine the effectiveness of the clearingobligation in mitigating systemic risk. Regulators around theworld will therefore continue to promote strengthened

prudential and risk management standards for both CCPs andtheir users, in areas such as stress testing and marginrequirements. Authorities will also continue their work toensure that there are mechanisms in place in extremecircumstances where a CCP’s viability is threatened.

The global nature of OTC derivatives markets also highlightsthe ongoing importance of co-operation between regulators.The Bank will continue to collaborate with other regulatorsand stakeholders, to ensure a timely and consistent approachto strengthening the global regulatory framework for OTCderivatives and central clearing.

294 Quarterly Bulletin 2015 Q3

AnnexWhat are OTC derivatives and which contracts can becentrally cleared?

OTC derivatives contracts fall into three broad types:

Swaps involve an agreement between two parties to exchangeone series of cash flows for another. For example, the twoparties might agree to exchange a set of fixed Japanese yencash flows for a set of US dollar ones, or a set based on a fixedinterest rate for those based on a floating rate.

Forwards involve an agreement between two parties topurchase a defined asset at a fixed price in the future. Forexample, the two parties might agree that one will buy a setamount of a commodity in a year’s time at a price agreedtoday. Futures are standardised forwards, typically traded ona regulated exchange rather than over-the-counter (theytherefore fall outside the scope of this paper).

Options are contracts that give the holder the right but notthe obligation to buy or sell something at a fixed price in thefuture. Swaptions are options on the exchange of future cashflows.

There are many variants of these basic types, and they can becombined into a huge variety of structures. ‘Exotic’derivatives are contracts that may encompass any of thepreviously mentioned structures, but which are more complexin terms of their payment calculation method and/orreference price. For example, a barrier option is a derivativescontract whose pay-off depends on whether or not theunderlying price has reached or exceeded a predeterminedlevel. Each of the five main asset classes has its own standardset of products taken from these types. These are detailed inthe table below.

Table A1 Summary of main OTC derivative contract types(a)

Asset class Further detail on selected products

Interest rate • ‘Plain vanilla’ fixed-float interest rate swaps (IRS) are the most commonly traded contracts. These involve a market participant paying cash flows at a predetermined fixed interest rate on a notional principal amount for a fixed period (normally between one and 30 years), receiving in return a floating rate on the same notional amount over the same period. These contracts can be useful for firms seeking to replace a variable-rate debt obligation with a set of fixed interest payments.• Basis swaps are the same as plain vanilla IRS, but use a floating rate for both payment legs of the contract.• Cross-currency swaps are IRS contracts that involve exchanging cash flows in different currencies.• Forward rate agreements (FRAs) are the second most common type of OTC interest rate derivative. An FRA is an agreement that a certain interest rate will apply to borrowing/lending a certain principal amount during a specified future period. FRAs tend to be shorter in duration (typically less than two years), with effective dates which may follow the contract trade date by several months.• Overnight indexed swaps (OIS) are a type of IRS contract in which the floating payment leg is based on an index rate for daily overnight unsecured lending between commercial banks. These contracts are also typically shorter in duration, lasting up to 18 months.• Caps and Floors allow a buyer to receive payments when the reference rate exceeds (Cap) or falls below (Floor) a certain threshold — these can be used to hedge (protect) against excessive fluctuations in interest rates.

Credit • Single-name credit default swaps (CDS) afford the contract buyer protection against the debt default of a reference entity, typically a corporate institution or a country (sovereign). Single-name corporates are the most numerous CDS contracts available by number of different contracts, but also among the least frequently traded.• Index CDS contracts provide protection against any constituent members of the index; sovereign index contracts will often be grouped by geographical region (for example, Western Europe). Corporate index contracts may additionally be grouped by industry sector (for example, financial services). Another underlying reference source would be portfolios of residential mortgage-backed debt, which may be grouped into an index as is, or as is more often the case, split into ‘tranches’, or layers of debt based on the probability of repayment. Index CDS contracts are the most frequently traded CDS contracts.• Total return swaps (and their indices) allow the contract buyer to receive any income generated by the underlying asset (typically an index of corporate names), rather than solely providing protection against the default of the asset.

Foreign exchange (FX) • Spot FX contracts allow counterparties to agree an exchange of currencies shortly after the trade date. In FX forward contracts, the exchange of currencies occurs at a deferred date. These are the most common FX contracts.• Non-deliverable forwards (NDFs) are used to hedge the currency risk arising from less liquid or restricted currencies, with one leg usually denominated in US dollars. They are similar to FX forwards, except that rather than exchanging payments in the two different currencies, the net dollar value of both legs (including that of the less liquid currency) is paid. NDFs are the only OTC FX derivatives contracts that are currently offered for central clearing by CCPs — and the trade volumes are very small compared with FX forward contracts.

Equity These are divided into various subtypes, based on how the return on the underlying equity is calculated (for instance, on its price, dividend or volatility), and whether the underlying equity is a single name, single index, or basket of equities. Certain equity index options and contracts for difference are centrally cleared, though volumes are limited.

Commodity These are divided into various subtypes, based on the underlying commodity (such as metals, energy or agricultural produce), and whether they are a single commodity or an index of commodities. Certain commodity swaps and options are centrally cleared by some CCPs, though volumes are fairly small.

Sources: CME Clearing Europe, European Commodity Clearing, ICE Clear Europe, ISDA, LCH.Clearnet Ltd and LME Clear.

(a) Products in blue are currently offered for central clearing by CCPs.

Topical articles Over-the-counter (OTC) derivatives and central clearing 295

References

Benos, E, Wetherilt, A and Zikes, F (2013), ‘The structure and dynamics of the UK credit default swap market’, Bank of England FinancialStability Paper No. 25, available at www.bankofengland.co.uk/financialstability/Documents/fpc/fspapers/fs_paper25.pdf.

Elliott, D (2013), ‘Central counterparty loss-allocation rules’, Bank of England Financial Stability Paper No. 20, available atwww.bankofengland.co.uk/financialstability/Documents/fpc/fspapers/fs_paper20.pdf.

Hull, J (2009), Options, futures and other derivatives, 7th edition, Prentice-Hall.

Liu, Z, Quiet, S and Roth, B (2015), ‘Banking sector interconnectedness: what is it, how can we measure it and why does it matter?’, Bank of England Quarterly Bulletin, Vol. 55, No. 2, pages 130–38, available atwww.bankofengland.co.uk/publications/Documents/quarterlybulletin/2015/q202.pdf.

Murphy, D (2009), Unravelling the credit crunch, 1st edition, Chapman and Hall.

Murphy, D (2013), OTC derivatives, bilateral trading and central clearing: an introduction to regulatory policy, market impact and systemic risk,1st edition, Palgrave Macmillan.

Murphy, D, Vasios, M and Vause, N (2014), ‘An investigation into the procyclicality of risk-based initial margin models’, Bank of EnglandFinancial Stability Paper No. 29, available at www.bankofengland.co.uk/financialstability/Documents/fpc/fspapers/fs_paper29.pdf.

Nixon, D and Rehlon, A (2013), ‘Central counterparties: what are they, why do they matter, and how does the Bank supervise them?’, Bank of England Quarterly Bulletin, Vol. 53, No. 2, pages 147–56, available atwww.bankofengland.co.uk/publications/Documents/quarterlybulletin/2013/qb130206.pdf.

Norman, P (2011), The risk controllers, John Wiley and Sons.

Recent economic andfinancial developments

Quarterly Bulletin Recent economic and financial developments 297

298 Quarterly Bulletin 2015 Q3

• A change to the Monetary Policy Committee’s (MPC’s) monetary policy communication scheduletook effect in August, with the simultaneous release of the August MPC policy decision, minutesand the Inflation Report.

• There were recurring bouts of financial market volatility, particularly in Chinese equities, withsome spillover to developed-economy equity prices and exchange rates.

• Corporate credit spreads continued to rise, particularly US high-yield spreads, reflecting theimpact of a further fall in the price of oil on the energy sector.

• Both short-term and long-term market interest rates in the United Kingdom and United Statesdeclined over the review period, reflecting concerns about the outlook for global growth.

Overview

A change to the Bank’s monetary policy communicationschedule took effect in August, with the simultaneousrelease of the August MPC policy decision, minutes andInflation Report. There was considerable interest ahead ofthe change, but in the event, the market reaction to thepolicy decision and accompanying communications wasmuted.

The review period as a whole was coloured by a number ofepisodes of heightened financial market volatility. Thevolatility was initially confined largely to Chinese equityprices, with the authorities there enacting a range ofmeasures to halt declines in the domestic stock market. Butfurther falls in Chinese equity prices in August led to a rise inrisk aversion and declines in share prices in developedmarkets. Both the FTSE All-Share and the S&P 500 endedthe review period materially lower than at the start.

The sterling exchange rate index ended the period over aper cent higher than at the start. But there were somesignificant moves in a number of exchange rate pairs duringthat time, particularly around the period of peak volatility inequity markets, which contacts suggested was at least partlydue to the rapid reversal of cross-border trades that hadbeen funded in euros and yen.

UK and US short-term market interest rates fell over thereview period as a whole, with the market-implied timing of

the first rate rise being pushed out by several months in bothcountries. Contacts attributed this primarily to internationaldevelopments that had increased fears of a slowdown inglobal growth, with particular concerns about the prospectsfor China. These worries were evident in commodity prices,with the oil price falling to new post-crisis lows.Nevertheless, some market contacts continued to think thatthe Federal Open Market Committee (FOMC) might raiserates in 2015. At the time of going to print, the attention ofmarket participants was focused keenly on the outcome ofthe 17 September policy meeting of the FOMC.

Concerns about the outlook for international growth andinflation also pushed down on long-term interest rates. Andthere was a decline in medium-term market-implied inflationexpectations in the United Kingdom, United States andeuro area. There was also a material drop in long-termmeasures of inflation expectations in the United States, butthese remained resilient in the United Kingdom. That wasthought to partly reflect relatively price-insensitive demandfor inflation compensation from UK pension funds andinsurers.

In euro-area fixed-income markets, sentiment improvedfollowing the successful resolution of negotiations over therelease of the final tranche of funds under the Greek bailoutprogramme. As a result, there was a further decline inperiphery government bond spreads over bunds.

Markets and operations

Recent economic and financial developments Markets and operations 299

In discharging its responsibilities to ensure monetary andfinancial stability, the Bank gathers information from contactsacross a range of financial markets. Regular dialogue withmarket contacts provides valuable insights into how marketsfunction, and provides context for the formulation of policy,including the design and evaluation of the Bank’s own marketoperations. The first section of this article reviewsdevelopments in financial markets between 3 June and3 September 2015. The second section describes the Bank’sown operations within the Sterling Monetary Framework.

Monetary policy and interest ratesA change in the Monetary Policy Committee’s (MPC’s)communication strategy took effect in August, as theCommittee moved to a schedule in which it released theminutes of its policy meeting and the policy decision on thesame day.(1) The Bank also published the AugustInflation Report to coincide with the policy decision. Contactslooked ahead to the first policy decision under the newschedule with anticipation, with some suggesting that thechange might lead to higher volatility on such days relative tothe past, but lower volatility on other days. In the event, themarket reaction was muted compared with the response tomany past Inflation Reports (Chart 1).

UK and US short-term market interest rates declined over thereview period as a whole (Chart 2). Much of that fall occurredin August, following considerable global financial marketturbulence, which some contacts thought was likely to delaythe timing of rate increases. Subsequent commentary fromFederal Open Market Committee (FOMC) members atJackson Hole, anticipating a pickup in inflation, was thought bycontacts to have left open the possibility that theFederal Reserve might still raise rates in 2015. At the time ofgoing to print, the attention of market participants was

focused keenly on the outcome of the 17 September policymeeting of the FOMC.

Heightened concern about the prospects for global growthalso pushed down on long-term government bond yields inthe United States and United Kingdom (Chart 3) and therewere sharp falls in a range of commodity prices. A number ofcommodity prices fell to multiyear lows, with the Brentfront-month oil contract declining to US$43 per barrel inAugust — the lowest since 2009. Contacts suggested that thelatest drop in commodity prices was a reflection of slowingglobal demand. That was in contrast to the decline observedin the price of oil towards the end of 2014 and into the firsthalf of this year, which was typically attributed to supply-sidefactors, including the decision of members of the

(1) These changes were introduced following Kevin Warsh’s review of the MPC’stransparency practices and procedures, available atwww.bankofengland.co.uk/publications/Pages/news/2014/168.aspx.

+–10 5 0 5 10

Feb. 2012

May 2012

Aug. 2012

Nov. 2012

Feb. 2013

May 2013

Aug. 2013

Nov. 2013

Feb. 2014

May 2014

Aug. 2014

Nov. 2014

Feb. 2015

May 2015

Aug. 2015

Basis points

Sources: Bloomberg and Bank calculations.

(a) Reaction between 10:25 and 11:45, apart from August 2015 (11:55–14:00). Green dashedlines show the average intraday reaction to all Inflation Reports published betweenMay 2009 and May 2015.

Chart 1 One-year, one-year forward overnight indexswap (OIS) rate intraday reaction to Inflation Reportpublication(a)

US dollar

Sterling

Euro

+

–0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

2015 16 17 18 19 20

Per cent

Dashed lines: 3 June 2015Solid lines: 3 September 2015

Sources: Bloomberg and Bank calculations.

(a) Instantaneous forward rates derived from the Bank’s OIS curves.

Chart 2 Instantaneous forward interest rates derivedfrom OIS contracts(a)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

Jan. Apr. July Oct. Jan. Apr. July

United Kingdom

United States

Germany

Previous Bulletin

Per cent

2014 15

Sources: Bloomberg and Bank calculations.

(a) Yields to maturity derived from the Bank’s government liability curves.

Chart 3 Selected ten-year government bond yields(a)

300 Quarterly Bulletin 2015 Q3

Organization of Petroleum Exporting Countries not to reducesupply in order to keep prices stable.

In the euro area, the European Central Bank (ECB) continuedits programme of large-scale asset purchases. Following themarket volatility in August, comments from ECB policymakerswere thought to have increased expectations that the ECBwould loosen policy further, for example by extending its assetpurchase programme. At the end of the review period, theECB Governing Council announced that the limit for purchaseswould be increased from 25% to 33% of each individualsecurity. Separately, following a period of heighteneduncertainty in late June and early July concerning Greece’sfiscal position, a deal was reached between Greece and itscreditors, enabling disbursement of the next tranche of fundsunder the Economic Adjustment Programme. This helped tolift sentiment in the euro area and allay fears of Greek exitfrom the eurozone. After rising sharply, Greek sovereignspreads to bunds ended the review period lower and there waslimited contagion to other European periphery bond spreads(Chart 4).

There was a decline in implied medium-term inflationexpectations across advanced economies, which contactsattributed to falling commodity prices and fears of slowingglobal growth. Spot five-year inflation swap rates finished thereview period markedly lower in the United Kingdom,United States and euro area. But while US five-year, five-yearforward inflation swap rates also declined, those of theUnited Kingdom and euro area remained largely unchanged(Chart 5).

Contacts continued to attribute the relative resilience ofUK long-term inflation expectations to the hedging activities

of liability-driven investors such as insurers and pension funds.A large proportion of net issuance of inflation-linked bonds inthe United Kingdom is held by such investors, while there isless structural demand for long-term inflation protection inthe United States and euro area.

Foreign exchangeThe sterling exchange rate index (ERI) rose by 1.4% over theperiod as a whole. There was a broad-based appreciation ofsterling over the first part of the review period, with thesterling ERI reaching its highest level since March 2009(Chart 6). The appreciation was in large part due to a riseagainst the euro (Chart 7), which fell following the resolutionof negotiations between Greece and its creditors. Contactsreported that reduced uncertainty about the position ofGreece within the single currency, coupled with very lowinterest rates in the euro area relative to some other countries,had encouraged a ‘carry trade’ whereby investors borrow ineuro to finance investments in other countries (with yieldssufficiently great to offset the borrowing costs, plus perceivedforeign currency risks associated with this strategy). Much ofthat carry trade later reversed after a very sharp pickup involatility in equity prices in late August, which lowered therisk-adjusted returns from the strategy (see the box onpages 302–03 for further discussion of volatility in equity andforeign exchange markets). Sterling also declined against thedollar toward the end of the review period, following strongerUS activity GDP data.

Elsewhere, the People’s Bank of China (PBoC) announced achange in the methodology for setting the daily renminbiexchange rate, linking the currency explicitly to the previousday’s closing price, with allowances to take account of supplyand demand factors. The change was seen by contacts as afurther step towards liberalisation of the renminbi exchange

0

2

4

6

8

10

12

14

16

18

20

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Jan. May Sep. Jan. May Sep.

Portugal (right-hand scale) Ireland (right-hand scale)

Italy (right-hand scale)

Spain (right-hand scale)

Greece (left-hand scale)

Per centPer cent

Previous Bulletin

2014 15

Sources: Bloomberg and Bank calculations.

Chart 4 Selected ten-year European peripherygovernment bond spreads to bunds

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Jan. Apr. July Oct. Jan. Apr. July

Sterling

US dollar

Euro

Previous Bulletin

Per cent

2014 15

Dashed lines: Spot five-year swap

Solid lines: Five-year, five-year forward swap

Sources: Bloomberg and Bank calculations.

(a) Swap rates derived from the Bank’s inflation swap curves.

Chart 5 Selected inflation swap rates(a)

Recent economic and financial developments Markets and operations 301

rate. Some speculated that the move was also a means ofslightly weakening the exchange rate to improvecompetitiveness. But the depreciation since the shift inmethodology has been relatively modest, the renminbi fallingby approximately 3.8% versus the dollar (Chart 8).Commentators suggested that data on Chinese foreignexchange reserves pointed to continuing large-scaleintervention by PBoC to support the currency in the periodthat followed the announcement.

There were depreciations of a number of emerging marketcurrencies against the dollar and the pound, which intensifiedfollowing the PBoC’s change in the renminbi exchange ratefixing methodology. Measures of emerging market exchangerate volatility implied by option prices rose to levels last seenduring the ‘taper tantrum’ in 2013 (Chart 9). Currencies ofcommodity exporters came under particular pressure due tofalling commodity prices.

Equity marketsThere was considerable volatility in Chinese stock marketsduring the summer, with the Shanghai Stock ExchangeComposite Index falling by 36% by the end of the reviewperiod (Chart 10). The decline occurred over two distinctepisodes. The first, which began in June, saw equities fall by29%. Sharp moves in prices caused a large number of sharesto be suspended, with trading in over half of the shares on theShanghai exchange halted at one point. Contacts attributedthe drop in Chinese equity prices to a range of factors,including short-term tightness in money markets, and effortsby the Chinese authorities to limit the extent to which retailinvestors could take on debt to leverage their positions instocks. The Chinese authorities subsequently introduced anumber of measures to support the market.

70

75

80

85

90

95

100

105

110

2005 07 09 11 13 15

Index: January 2005 = 100

Previous Bulletin

Source: Bank calculations.

Chart 6 Sterling ERI

+

–

4.0

3.0

2.0

1.0

0.0

1.0

2.0

3.0

4.0

5.0

Other

Emerging markets(a) Japanese yen

US dollarEuro

£ ERIPer cent

3 June to18 August

18 August to3 September

Overall change

Sources: Bloomberg, Thomson Reuters Datastream and Bank calculations.

(a) The emerging market currencies in the narrow sterling ERI are: Chinese renminbi,Czech koruna, Indian rupee, Polish zloty, Russian rouble, South African rand and Turkish lira.

Chart 7 Contributions to changes in the sterling ERI

5.90

6.00

6.10

6.20

6.30

6.40

6.50

6.60Jan. Apr. July

Reference band Onshore renminbi

Offshore renminbi Reference rateRenminbi per dollar

(inverted scale)

Renminbi depreciation

2015

Sources: Bloomberg, People’s Bank of China and Bank calculations.

Chart 8 Onshore and offshore dollar-renminbi exchangerates, the official reference rate and target trading band(rates against US dollar, inverted scale)

0

50

100

150

200

250

300

350

400

450

2008 09 10 11 12 13 14 15

Index: 1 January 2008 = 100

Source: JPMorgan Chase & Co.

Chart 9 Implied volatility of emerging market currencies

302 Quarterly Bulletin 2015 Q3

Equity and foreign exchange market volatility

Volatility in the Chinese equity market spilled over to equityprices in developed markets, in light of growing concernsabout the prospects for the economic outlook in China, andthe possible impact on the rest of the world. There were alsosizable moves in various currency pairs. But while there wasthought to be some news in developments in China, theextent of the volatility in developed markets on 24 August wasgreater than might have been expected. Contacts suggestedthat the extent of the disruption was likely to have been due,at least in part, to market structural factors.

Equity market volatilityThe peak period of equity market volatility occurred onMonday 24 August. The S&P 500 fell 3.7% on the day(Chart 2), while the FTSE All-Share suffered its worst one-daydecline since March 2009, dropping by 4.5%. The Euro Stoxxalso fell a considerable distance. There were also large movesin equity derivatives markets. The VIX index — a measure of30-day implied volatility on the S&P 500 — recorded thelargest two-day increase in its history and reached an intradayhigh of 53%, a level not seen since March 2009. The impliedvolatility of the VIX (the VVIX) rose to its highest ever level(Chart A).

While the precise trigger for the large drop in the S&P remainsunclear, contacts thought that it might have stemmed from alarge drop in the price of the S&P futures contract — whichtrades nearly continuously around the clock — shortly beforethe US market open. That then caused trading in the futurescontract to be halted. According to contacts, the suspension

of trading in the future created uncertainty about whereUS stocks would open, and led a number of firms to reducethe amount of capital committed to market-making. Theextent of the move also released market makers from theobligation to post indicative prices, adding to uncertainty. Atthe same time, the fall in the S&P future caused some‘stop-loss’ orders to be activated automatically. Those ordersthen had to be executed at the prevailing market price, puttingimmediate downward pressure on equities at the open. Thecombination of sell orders, reduced market-making capacity,uncertainty, and, perhaps, the additional influence ofelectronic trading systems, resulted in rapid price falls andheightened volatility.

In fact, price moves were sufficiently large that they led to theactivation of circuit breakers — introduced after the 2010‘flash crash’ — which suspended trading in a number of shares.The New York Stock Exchange saw over a thousand pauses inthe trading of individual securities and exchange-traded fundson the day, compared with around ten on a typical tradingday.

Halts in trading of stocks and the S&P future also meant thatoptions prices could not be priced accurately, as there was nounderlying price for the options to reference. Many marketmakers stopped quoting prices on certain options during theperiod of greatest instability, and those that continued toprovide quotes widened the bid-offer spreads significantly.

Spillovers to the foreign exchange marketContacts reported that the intraday volatility in equitymarkets on 24 August prompted a wave of risk reduction,particularly affecting carry trades that had been financed inyen and euro (which contacts note has grown in use as afunding currency over the past year). Consistent with amaterial decline in euro and yen-funded carry trades, net shortspeculative positioning in both the euro and the yen versus theUS dollar fell sharply in the week to 25 August. And data fromEmerging Portfolio Fund Research show sizable outflows fromUS equity funds in the week ending 26 August.

It is also likely that many speculative investors would havepurchased carry funding currencies very quickly once equityvolatility picked up, in anticipation of the reversal of carrytrades, adding to the upward momentum in those currencies.And contacts added that relatively thin summer markets,along with nervousness related to the timing of lift-off in theUnited States, were likely to have contributed to the ensuingvolatility. The strengthening of the Japanese yen versus theUS dollar on 24 August was especially marked, with the valueof the dollar against the yen falling from ¥122 to ¥116 — anearly 5% intraday decline. There was considerable volatilityin a number of other currency pairs as well.

0

10

5

15

20

25

30

35

40

45

0

20

40

60

80

100

120

140

160

180

Jan. Apr. July Oct. Jan. Apr. July

Per cent Per cent

VVIX(a) (right-hand scale)

VIX(b) (left-hand scale)

2014 15

Source: Chicago Board Options Exchange.

(a) The VIX of VIX (VVIX) index is a measure of the market-implied volatility of the VIX index,estimated from a weighted average of VIX index option prices.

(b) The Chicago Board Options Exchange Market Volatility Index (VIX) is a measure of themarket-implied volatility of the S&P 500, estimated from a weighted average of S&P 500index option prices.

Chart A Option-implied equity volatility of the S&P 500(VIX) and option-implied volatility of the VIX

Recent economic and financial developments Markets and operations 303

Explaining the size of the moves, contacts have often noted inthe past that market makers have become less willing toprovide liquidity. And that would be even more the caseduring a period of volatility, in which the market wasbecoming increasingly ‘one-way’. Contacts also thought thatliquidity on electronic matching platforms had disappearedvery rapidly, as participants withdrew. However, there werenot widespread reports that platforms had closed, in contrastwith what occurred during the period of volatility thatfollowed the removal of the floor on the Swiss franc versus theeuro earlier in the year. Instead, platforms quickly shifted toindicative pricing or much wider bid-offer spreads.

Consistent with a rapid reduction in liquidity, contacts notedthat there was ‘gapping’ in various currency pairs, with marketprices jumping to new levels with no trades in between. Thatwas evident even in euro-dollar and dollar-yen exchange rates,two of the most heavily traded pairs. Contacts suggested thatsuch pricing dynamics were likely to be a more frequentoccurrence in foreign exchange markets compared with thepast, in light of ongoing structural changes.

A second period of volatility occurred in August. Althoughthere was no obvious immediate driver, there had been agrowing expectation that the PBoC would lower the reserverequirement ratio for banks around that time. And contactsthought that those expectations reached a peak in the daysimmediately prior to the crash. Contacts suggested that itwas the absence of such a change that led to a reassessmentof the willingness of the authorities to loosen policy tosupport growth. In turn, that resulted in a material drop in theShanghai Composite Share Index on 24 August and furtherdeclines in the following days. Against this backdrop, thePBoC subsequently reduced both the reserve requirementratio and its main policy rate on 25 August.

The August volatility in Chinese equity markets wasassociated with falls in advanced-economy equity pricesalthough some of these moves subsequently reversed (seethe box on pages 302–03 for a discussion of equity marketvolatility). The FTSE All-Share and S&P 500 ended the reviewperiod down 11% and 8% respectively (Chart 11).

Corporate bonds and bank funding marketsIn contrast to the volatility in equity markets,advanced-economy corporate bond spreads rosecomparatively modestly over the review period. In theUS market, contacts attributed rising spreads to particularlystrong debt issuance to finance mergers and acquisitions andshare buybacks. It was also suggested that some US dollarissuance had been brought forward in anticipation ofprospective policy tightening by the Federal Reserve later inthe year.

US high-yield bond spreads also widened materially(124 basis points) over the review period (Chart 12). Contactsnoted that US high-yield energy spreads continued to beclosely correlated with oil price movements, rising as energyprices fell. But contacts thought that the risk of spillover tothe wider US high-yield bond market was lower than it hadbeen a year ago, as many fund managers had reduced theirexposure to the energy sector since then.

0

1000

2000

3000

4000

5000

6000

Jan. Apr. July Oct. Jan. Apr. July

Index: 19 December 1990 = 100

2014 15

Sources: Bloomberg and Bank calculations.

(a) The index is quoted in domestic currency terms.

Chart 10 Shanghai Stock Exchange Composite Index(a)

60

70

80

90

100

110

120

130

140

Jan. Mar. May July Sep. Nov. Jan. Mar. May July Sep.

DJ Euro Stoxx

FTSE All-Share

S&P 500

Topix

2014

Previous Bulletin

Indices: 1 January 2014 = 100

15

MSCI Emerging Markets index

Sources: Bloomberg and Bank calculations.

(a) Indices are quoted in domestic currency terms, except for the MSCI Emerging Markets index,which is quoted in US dollar terms. The MSCI Emerging Markets index is a free-floatweighted index that monitors the performance of stocks in global emerging markets.

Chart 11 International equity indices(a)

Against the backdrop of Greece-related uncertainty early inthe review period, primary corporate bond issuance byEuropean firms was weak and there was a marked pickup innew issue premia. Contacts also reported that uncertaintyaround Greece had weighed on primary issuance by banks,with a number of deals having been postponed. But contactswere generally fairly sanguine about the pause, noting thatEuropean banks did not have a pressing need to raise funds,due to readily available medium-term funding facilities fromthe ECB. And UK banks remained well-funded, given proactiveissuance before and after the general election. Thesubsequent resolution of negotiations over the Greek bailoutprogramme resulted in a flurry of issuance in European andUK bank funding markets. Senior unsecured spreadscontinued to edge higher, although there was no obvious

spillover during the period of extreme volatility in some othermarkets (Chart 13).

Operations

Operations within the Sterling Monetary Frameworkand other market operationsThis section provides an update of the Bank’s operationswithin the Sterling Monetary Framework (SMF) over thereview period, as well as its other market operations.Collectively, these operations help implement the Bank’smonetary policy stance and provide liquidity insurance toinstitutions when deemed necessary.

The aggregate level of central bank reserves is closelymonitored by the Bank, as it affects monetary conditions inthe UK economy. The level of central bank reserves is affectedby (i) the stock of assets purchased via the Asset PurchaseFacility (APF); (ii) the level of reserves supplied by operationsunder the SMF; and (iii) the net impact of other sterling flowsacross the Bank’s balance sheet. Over the review period,aggregate reserves rose to a high of £319 billion, largelyreflecting the larger-than-usual injections of reserves throughthe Bank’s Indexed Long-Term Repo (ILTR) operations(discussed below).

Operational Standing FacilitiesSince 5 March 2009, the rate paid on the OperationalStanding Deposit Facility has been zero, while all reservesaccount balances have been remunerated at Bank Rate. As aconsequence, there is little incentive for reserves accountholders to use the deposit facility. Reflecting this, the averageuse of the deposit facility was £0 million in the three monthsto 9 July 2015.(1)

The rate charged on the Operational Standing Lending Facilityremained at 25 basis points above Bank Rate. However, giventhe large aggregate supply of reserves, there was no demandfrom market participants to use the lending facility. Theaverage use of the lending facility was also £0 million over thequarter to 9 July 2015.

Indexed Long-Term Repo operationsThe Bank conducts regular ILTR operations as part of itsprovision of liquidity insurance to banks, building societies andbroker-dealers. During the review period, the Bank offered aminimum of £5 billion via six-month repos in each of its ILTRoperations on 9 June, 7 July and 11 August 2015 (Table A).

Participation in, and usage of, ILTR operations has continuedto remain higher than last year, although the total amountallocated in each operation remained below the minimum

304 Quarterly Bulletin 2015 Q3

Previous Bulletin

0

50

100

150

200

250

300

350

400

450

2011 12 13 14 15

United Kingdom(d)

United States(b)

Europe(c)

Basis points above mid-swaps

Asset quality review results announced

Sources: Bloomberg, Markit Group Limited and Bank calculations.

(a) Constant-maturity unweighted average of secondary market spreads to mid-swaps of banks’five-year senior unsecured bonds, where available. Where a five-year bond is unavailable, aproxy has been constructed based on the nearest maturity of bond available for a giveninstitution and the historical relationship of that bond with the corresponding five-yearbond.

(b) Average of Bank of America, Citi, Goldman Sachs, JPMorgan Chase & Co., Morgan Stanleyand Wells Fargo.

(c) Average of Banco Santander, BBVA, BNP Paribas, Crédit Agricole, Credit Suisse,Deutsche Bank, ING, Intesa, Société Générale, UBS and UniCredit.

(d) Average of Barclays, HSBC, Lloyds Banking Group, Nationwide, Royal Bank of Scotland andSantander UK.

Chart 13 Indicative senior unsecured bank bondspreads(a)

Previous Bulletin

0

200

400

600

800

1,000

1,200

Jan. Apr. July Oct. Jan. Apr. July

Basis points

High-yield energy corporates

High-yield non-energy corporates

High-yield corporates

2014 15

Sources: BofA Merrill Lynch Global Research and Bank calculations.

Chart 12 US corporate bond option-adjusted spreads

(1) Operational Standing Facility usage data are released with a lag.

Recent economic and financial developments Markets and operations 305

£5 billion on offer (Chart 14) and is somewhat lower thanusage in the previous quarter. This continued to reflect usageof the ILTR by some participants as a source of term repoliquidity. Over the review period, with £2.1 billion of ILTRsmaturing, the increased participation in ILTR operationsresulted in a net addition of £4 billion of central bank reserves.This was in addition to the existing supply of aggregatereserves which were provided largely through the Bank’s APFoperations.

Contingent Term Repo FacilityThe Contingent Term Repo Facility (CTRF) is a contingentliquidity facility that the Bank can activate in response toactual or prospective market-wide stress of an exceptionalnature. The Bank reserves the right to activate the facility as itdeems appropriate. In light of market conditions throughoutthe review period, the Bank judged that CTRF auctions werenot required.

Discount Window FacilityThe Discount Window Facility (DWF) is a bilateral on-demandfacility provided to institutions experiencing a firm-specific ormarket-wide liquidity shock. It allows participants to borrowhighly liquid assets in return for less liquid collateral inpotentially large size and for a variable term. The Bankpublishes quarterly data of DWF usage with a lag. The averagedaily amount outstanding in the DWF in the three months to31 March 2014 was £0 million.

Other operationsFunding for Lending SchemeThe Funding for Lending Scheme (FLS) was launched by theBank and HM Treasury on 13 July 2012. The initial drawdownperiod for the FLS ran from 1 August 2012 until 31 January2014. The drawdown period for the FLS extension opened on3 February 2014 and will run until 29 January 2016. Thequantity each participant can borrow in the FLS is linked totheir lending to the UK real economy, with the incentivescurrently skewed towards supporting lending to smallbusinesses.

The Bank publishes quarterly data showing, for each groupparticipating in the FLS extension, the amount borrowed fromthe Bank and the net quarterly flows of lending. During thesecond quarter of 2015, the number of groups participating inthe FLS extension was 34. Of these, eleven participants madedrawdowns of £5.1 billion in total. Participants also repaid£0.9 billion, taking total outstanding drawings to £61.4 billion.

US dollar repo operationsOn 23 April 2014, in co-ordination with other central banksand in view of the improvement in US dollar fundingconditions, the Bank ceased the monthly 84-day US dollarliquidity-providing operations. The seven-day US dollaroperations will continue until further notice. The network ofbilateral central bank liquidity swap arrangements provides aframework for the reintroduction of further US liquidityoperations if warranted by market conditions. There was nouse of the Bank’s US dollar facilities throughout the reviewperiod.

Bank of England balance sheet: capital portfolioThe Bank holds an investment portfolio that is approximatelythe same size as its capital and reserves (net of equityholdings, for example in the Bank for International

Table A Indexed Long-Term Repo operations(a)

Total Collateral set summary

Level A Level B Level C

9 June 2015 (six-month maturity)

Minimum on offer (£ millions) 5,000

Total bids received (£ millions) 1,637 1,207 0 430

Amount allocated (£ millions) 1,637 1,207 0 430

Clearing spread (basis points) 0 n.a. 15

7 July 2015 (six-month maturity)

Minimum on offer (£ millions) 5,000

Total bids received (£ millions) 3,090 275 40 2,775

Amount allocated (£ millions) 2,315 275 40 2,000

Clearing spread (basis points) 0 5 15

11 August 2015 (six-month maturity)

Minimum on offer (£ millions) 5,000

Total bids received (£ millions) 2,135 1,340 0 795

Amount allocated (£ millions) 2,135 1,340 0 795

Clearing spread (basis points) 0 n.a. 15

(a) The minimum amount on offer is the size of the operation that the Bank is willing to allocate, in aggregate,across all collateral sets at the minimum clearing spreads.

Three-month Level A allocated (left-hand scale)

Three-month Level B allocated (left-hand scale)

Six-month Level A allocated (left-hand scale)

Six-month Level B allocated (left-hand scale)

Six-month Level C allocated (left-hand scale)

Level A clearing spread (right-hand scale)

Level B clearing spread (right-hand scale)

Level C clearing spread (right-hand scale)

Amount allocated (£ millions) Clearing spread (basis points)

0

1,000

2,000

3,000

4,000

5,000

6,000

0

10

20

30

40

50

60

70

Jan.Apr.

JulyOct.

Jan.Apr.

JulyOct.

Jan.Apr.

2011 12 13 14 15

JulyOct.

Jan.Apr.

July JulyOct.

Jan.Apr.

(a) Where there has not been any allocation to a collateral set, no clearing spread is marked.

Chart 14 ILTR reserves allocation and clearing spreads(a)

Settlements, and the Bank’s physical assets) and aggregatecash ratio deposits. The portfolio consists ofsterling-denominated securities. Securities purchased by theBank for this portfolio are normally held to maturity, thoughsales may be made from time to time, reflecting, for example,risk or liquidity management needs or changes in investmentpolicy. The portfolio currently includes around £5.5 billion ofgilts and £0.2 billion of other debt securities.

Asset purchasesAlongside the publication of the Inflation Report on12 February 2014, the MPC announced that it intends tomaintain the stock of purchased assets, including reinvestingthe cash flows associated with all maturing gilts held in theAPF, at least until Bank Rate has been raised from its currentlevel of 0.5%. There were no gilts held in the APF that hadmatured during the review period, and as a result there wereno reinvestment operations.

The total stock of gilts outstanding in the APF, measured asproceeds paid to sellers, remained unchanged at £375 billion.The stock of gilts comprised of £77.9 billion of purchases inthe 3–7 years residual maturity range, £139.5 billion in the7–15 years residual maturity range and £157.5 billion with aresidual maturity of greater than 15 years (Chart 15).

Gilt lending facilityThe Bank continued to offer to lend gilts held in the APF viathe Debt Management Office in return for otherUK government collateral. In the three months to 30 June2015, the daily average value of gilts lent, as part of the giltlending facility, was £330 million. The average daily lending inthe previous quarter was somewhat higher at £632 million.

Corporate bondsThere were no purchases of corporate bonds during the reviewperiod. Future purchase or sale operations through thescheme will be dependent on market demand, which the Bankwill keep under review in consultation with its counterparties.Reflecting the recent lack of activity, the scheme currentlyholds no bonds.

Secured commercial paper facilityThe Bank continued to offer to purchase secured commercialpaper backed by underlying assets that are short term andprovide credit to companies or consumers that supporteconomic activity in the United Kingdom. No purchases weremade during the review period.

306 Quarterly Bulletin 2015 Q3

0255075

100125150175200225250275300325350375400

Feb. Feb. Feb. Feb. Feb. Feb. Feb.

15+ years

7–15 years

0–7 years£ billions

14131211102009 15

(a) Proceeds paid to counterparties on a settled basis.(b) Residual maturity as at the date of purchase.

Chart 15 Cumulative gilt purchases by maturity(a)(b)

Report

Quarterly Bulletin Report 307

308 Quarterly Bulletin 2015 Q3

On 25 June 2015, the Bank of England and the Centre forEconomic Policy Research (CEPR) hosted their thirteenthMonetary Policy Roundtable. These events provide a forumfor economists to discuss key issues relevant to monetarypolicy in the United Kingdom.(1) As with previous Roundtablediscussions, participants included a range of economists fromprivate sector financial institutions, academia, public sectorbodies and industry associations. There were two topics ofdiscussion:

• what are the global drivers of inflation? and

• why are real interest rates so low?

This note summarises the main issues raised byparticipants.(2) The Roundtables are conducted under‘Chatham House Rule’ and so opinions expressed at themeeting are not attributed to individuals. This summarydoes not represent the views of the Bank of England, theMonetary Policy Committee or the CEPR.

What are the global drivers of inflation?

Inflation had been low internationally in the period precedingthe Roundtable, particularly in advanced countries. Forexample, in the United Kingdom the latest CPI inflationoutturn at the time of the Roundtable was just 0.1% (for thetwelve months to May 2015). In contrast, UK CPI inflation hadaveraged 1.8% during 2000–08 and 3.1% in the period2009–13.

From a monetary policy perspective, identifying the drivers ofinflation is important because of the lags between a change inmonetary policy and its effects on inflation. When inflation ispushed away from target due to factors that are likely topersist, a monetary policy response may be warranted to bringinflation back towards the target. In contrast, monetarypolicy makers may ‘look through’ a period where inflation istemporarily away from target and not adjust policy on thebasis of factors that are judged likely to be transitory.

In that context, the first session of the Roundtable discussedparticipants’ views of the likely drivers of recent and futureinflation. The three speakers highlighted negative output gapsand falls in commodity prices as having contributed to lowinflation in recent years. They also discussed additional

drivers, including the possibility of more persistent factorssuch as an extended period of stagnant demand growth acrosscountries.

It was generally agreed that negative output gaps (theestimated gap between economic output and potentialeconomic output) had contributed to low inflation inadvanced economies since the dramatic falls in economicactivity in 2008–09. Central banks had responded byproviding unprecedented monetary stimulus, which was likelyto have stimulated economic activity. But in general it waslikely that output remained below potential output in manyeconomies, which was likely to be a drag on inflation. Someparticipants cautioned against viewing the relationshipbetween inflation and output gaps in a deterministic way,however. Historical relationships between output gaps andinflation were not stable; in the past, similar rates of inflationhad been observed alongside very different estimated levels ofthe output gap. This suggested that other factors were oftenmore important drivers of fluctuations in inflation.

Speakers also pointed to commodity prices as an obviousdriver of lower inflation in the recent period. Oil prices hadfallen sharply in 2014 Q4 following OPEC’s decision tomaintain production rather than cutting output in the face ofadditional energy supply from other sources and falling prices.The key question was how oil prices were likely to evolve inthe future and hence the outlook for supply and demand inthat market.

One speaker noted that global demand for oil had remainedstrong, partly in response to the fall in prices. Global oildemand had increased by more than would have beensuggested by changes in global GDP growth alone. And theincrease in demand had been broad-based: from consumers aswell as firms, and from across many regions of the world. Thespeaker argued that there was scope for further increases indemand, particularly from some developing countries.

On the other hand, strong oil supply was likely to keep priceslow in the near term. This partly reflected record output fromSaudi Arabia and also a reluctance of some OPEC countries to

Monetary Policy Roundtable

(1) This report was prepared by David Elliott of the Markets Directorate of the Bank,together with Joanna Konings and Jon Relleen of the Monetary Analysis Directorate.

(2) For both this and previous summaries, see www.bankofengland.co.uk/publications/Pages/other/monetary/roundtable/default.aspx.

Report Monetary Policy Roundtable 309

reduce supply given the importance of the income stream fortheir public finances. Time lags between falls in oil prices andreductions in supply by non-OPEC producers would also act tokeep prices low for a period. However, reductions in supplycapacity were eventually likely to affect oil production, helpingto support prices in the medium term. Rising production costscould also support prices and over time it was suggested thatthe oil price may return to around US$100 per barrel.

In light of the different factors influencing the price of oil,participants discussed what would be the most appropriateforecasting method to use. The Bank’s forecasting conventionassumes that oil prices evolve in line with the latest futurescurve, although this approach has well-known limitations,particularly for forecasting at longer horizons. Someparticipants argued for a structural approach based on oilproduction costs and other supply considerations. It wasagreed that these approaches could be informative, althoughsome argued that the futures curve was probably at least asgood as alternative approaches for forecasting over the shortto medium-term horizons most relevant for monetary policy.

Participants also highlighted some longer-term, persistentdrivers of low inflation. One speaker related low inflation toconcerns about a prolonged period of stagnant nominaldemand growth, perhaps related to wealth being concentratedin the hands of those with a high propensity to save. Theprocess of economic rebalancing within the euro area was alsomentioned. Some European countries running large andpersistent current account deficits might become morecompetitive by having persistently lower inflation than theirmajor trading partners. If inflation was relatively low incountries running surpluses it might have to be even lower indeficit countries. This rebalancing was likely to be a long-termprocess, so any effect on inflation might be persistent.

Another speaker argued that movements in inflationexpectations might indicate market concerns about outputremaining below its potential level, and the associateddisinflationary pressure. Each of the different measures ofinflation expectations had strengths and weaknesses, but thefinancial market-based five-year, five-year forward inflationrate should give an indication of central bank credibility as itembodies inflation expectations at a time horizon whentemporary influences on inflation, such as a fall in the oil price,should have disappeared. Implied five-year inflation ratesfive years forward were lower than historical averages in theeuro area and to a lesser extent in the United States, althoughnot in the United Kingdom.

The speaker argued that falls in five-year, five-year forwardinflation rates during 2014 and the early part of 2015 hadpartly reflected market concerns about the willingness and/orability of central banks to bring about conditions that wouldreturn inflation to target. Against a backdrop of low inflation,

concerns may arise if central banks are not perceived to beeasing policy sufficiently, or there are questions about theeffectiveness of monetary policy close to the lower bound forpolicy rates.

In summary, the drivers of inflation highlighted in the firstsession included a range of factors, some of which were likelyto be transitory, such as previous falls in oil prices, along withfactors that may be more persistent. On balance, the mix ofdrivers suggested that inflation was likely to rise as thetemporary factors dissipated, as long as monetarypolicy makers could demonstrate their willingness and abilityto return inflation towards targeted levels.

Why are real interest rates so low?

In the second session of the Roundtable, participants discussedwhat factors had driven falls in the level of real interest rates.Real interest rates in advanced economies had fallensignificantly since the 1980s, and in recent years the real yieldsavailable on index-linked government bonds had often beennegative.(1)

The key question for monetary policy makers was what signalthey should take from this. For example, did it reflectstructural, or at least very persistent, changes inmacroeconomic conditions? And did the low level offorward-looking real yields indicate that economic growth wasexpected to be weak for a protracted period, or were realyields being influenced by other factors such as risk premia?In that context, participants discussed the potential role offundamental macroeconomic factors as well as technicalfinancial market factors, and whether these factors were likelyto be temporary or permanent.

Speakers suggested several interrelated macroeconomicfactors that could have led to lower real interest rates. Theseincluded a slowdown in productivity growth, demographicchanges and deleveraging following the financial crisis.

Several participants argued that low real interest rates couldreflect reduced productivity growth, which had been slow byhistorical standards in several advanced economies overrecent years. However, productivity had been reasonablystrong before the financial crisis so this could not explain thefalls in real yields prior to the crisis. And there were mixedviews on whether productivity growth would continue to beweak going forward. For example, one speaker suggested thata more optimistic view was that technological innovationswould lead to large productivity gains in the future. Someparticipants also suggested that current productivity mightnot be as low as official estimates suggested, as technological

(1) See, for example, King, M and Low, D (2014), ‘Measuring the ‘world’ real interest rate’,NBER Working Paper No. 19887.

310 Quarterly Bulletin 2015 Q3

innovation could have made productivity growth moredifficult to observe and measure.

A number of participants argued that demographic factorscould be contributing to low real interest rates. For example,longer retirement periods might incentivise individuals toincrease their level of saving, which would tend to push downreal interest rates. Absent reforms such as significantincreases in the age of retirement, this factor couldpersistently reduce interest rates. One speaker argued thatthe increasing number of retirees could be related to changesin corporate behaviour that might also slow productivitygrowth. Specifically, retirees may be more interested inincome than capital gains, and so favour buying shares incompanies that pay large dividends rather than firms thatdirect a large proportion of their earnings towards capitalexpenditure. This could help to explain why ratios of dividendsto earnings have increased in recent years, while investmenthas been relatively weak. And continued low investmentmight signal weak productivity growth in the future.

Some speakers suggested that deleveraging following thefinancial crisis, where borrowers sought to repay large debtsaccumulated before the crisis, had also reduced real interestrates. This argument suggested that some of the reduction inreal interest rates might be temporary — reversing once thedeleveraging cycle and recovery from the financial crisis werecomplete. That said, there was evidence that it takes aparticularly long time for economies to recover from financialcrises, so the effects could last a reasonably long time.

In addition, participants discussed several financial marketfactors that could be contributing to low real interest rates.These factors generally concerned the effect of investors’decisions on the demand for bonds, either due to riskpreferences or regulations that encouraged holding bonds.

For example, one participant suggested that Solvency II, a newregulatory regime for European insurance companies, hadincreased the extent to which these firms bought bonds tohedge their liabilities, supporting the prices of those bonds andso reducing their yields. Others discussed how regulation andconcerns about counterparty credit risk had led to more tradesbeing collateralised, generating greater demand forhigh-quality bonds to be used as collateral. And someparticipants argued that investors had increased theirallocation to bonds in order to reduce the riskiness of theirbalance sheets, while others noted that some central bankshad contributed to the high global demand for bonds throughasset purchase programmes.

These types of factors may have reduced the risk premiumbuilt into the price paid for a bond, also known as the ‘termpremium’. One speaker presented survey and model-basedevidence suggesting that the majority of the reduction in real

rates in recent years reflected falls in term premia, rather thanfalls in expectations for future interest rates. This speaker alsopointed out that long-term interest rates can be volatile,meaning that they may not give a consistently clean read onstructural macroeconomic developments.

Several participants suggested that although there was a largequantity of government bonds outstanding, fewer sovereignswere considered to be risk-free than in the past. This meantthat purchases of bonds for collateral or to hedge risks wereconcentrated among a smaller number of the highest-qualitysovereign issuers, exacerbating the falls in market interestrates for those countries. One contact pointed out that ifthese sovereigns were to reduce debt issuance, there could befurther falls in the yields on their bonds.

Participants also discussed the fact that while interest rateswere low, returns on other asset classes such as equities hadnot fallen to the same extent. This was in contrast to previousepisodes of low interest rates, which had tended to involvelow returns across most assets. Broadly speaking, assetreturns should be linked to the marginal product of capital, sothe apparent disparity between returns on bonds and on otherassets might suggest that factors specific to the bond marketwere at play. For example, some participants thought thatinvestors now attached a higher value to the hedgingcharacteristics of government bonds, whose returns might beexpected to negatively covary with returns availableelsewhere. Alternatively, changes in regulation or central bankpolicies such as quantitative easing could have pushedgovernment bond yields below where they would otherwisehave been.

The very high degree of international comovement betweenadvanced-economy government bond yields was alsodiscussed. This suggested that moves in yields were largelydriven by global rather than country-specific factors. Thatseemed consistent with increases in trade and financiallinkages between countries over recent decades, which werelikely to have amplified spillovers from shocks in one countryto another, and the increasingly global focus of investors.

It was also pointed out that the current low level of yields wasnot dramatically out of line with experience over a muchlonger period, such as the past 100 years or more. It may bethe case that yields in the 1970s and 1980s were in factunusually high. And so the recent falls in yields could reflect amove back towards very long-run averages.

Overall, most participants agreed that there was a great dealof uncertainty about what signal to take from the low levels ofreal interest rates. To the extent that they reflectedfundamental macroeconomic factors, it was possible that theygave a negative signal about the long-term growth potentialof the global economy. On the other hand, some of the

Report Monetary Policy Roundtable 311

financial market factors suggested that yields could movesharply upwards if investor behaviour changed or there weresignificant adjustments to central bank policy. It was generallyagreed that global factors were key determinants of realinterest rates, given the high level of correlation between

yields across countries, and that the apparent differentialbetween yields on government bonds and those available onother assets was a puzzle that warranted furtherconsideration.

Quarterly Bulletin Appendices 313

Appendices

314 Quarterly Bulletin 2015 Q3

The articles that have been published recently in theQuarterly Bulletin are listed below. Articles fromDecember 1960 to Winter 2004 are available on theBank’s website at:

www.bankofengland.co.uk/archive/Pages/digitalcontent/historicpubs/quarterlybulletins.aspx.

Articles from Spring 2005 onwards are available at:

www.bankofengland.co.uk/publications/Pages/quarterlybulletin/default.aspx.

Articles

2011 Q4– Understanding recent developments in UK external trade– The financial position of British households: evidence from the 2011 NMG Consulting survey– Going public: UK companies’ use of capital markets– Trading models and liquidity provision in OTC derivatives markets

2012 Q1– What might be driving the need to rebalance in the United Kingdom?– Agents’ Special Surveys since the start of the financial crisis– What can the oil futures curve tell us about the outlook for oil prices?– Quantitative easing and other unconventional monetary policies: Bank of England conference summary– The Bank of England’s Special Liquidity Scheme– Monetary Policy Roundtable

2012 Q2– How has the risk to inflation from inflation expectations evolved?– Public attitudes to monetary policy and satisfaction with the Bank– Using changes in auction maturity sectors to help identify the impact of QE on gilt yields– UK labour productivity since the onset of the crisis — an international and historical perspective– Considering the continuity of payments for customers in a bank’s recovery or resolution– A review of the work of the London Foreign Exchange Joint Standing Committee in 2011

2012 Q3– RAMSI: a top-down stress-testing model developed at the Bank of England

– What accounts for the fall in UK ten-year government bond yields?– Option-implied probability distributions for future inflation– The Bank of England’s Real-Time Gross Settlement infrastructure– The distributional effects of asset purchases– Monetary Policy Roundtable

2012 Q4– The Funding for Lending Scheme– What can the money data tell us about the impact of QE?– Influences on household spending: evidence from the 2012 NMG Consulting survey– The role of designated market makers in the new trading landscape– The Prudential Regulation Authority

2013 Q1– Changes to the Bank of England– The profile of cash transfers between the Asset Purchase Facility and Her Majesty’s Treasury– Private equity and financial stability– Commercial property and financial stability– The Agents’ company visit scores– The Bank of England Bank Liabilities Survey– Monetary Policy Roundtable

2013 Q2– Macroeconomic uncertainty: what is it, how can we measure it and why does it matter?– Do inflation expectations currently pose a risk to the economy? – Public attitudes to monetary policy– Cross-border bank credit and global financial stability– The Old Lady of Threadneedle Street– Central counterparties: what are they, why do they matter and how does the Bank supervise them?– A review of the work of the London Foreign Exchange Joint Standing Committee in 2012

2013 Q3– Macroprudential policy at the Bank of England– Bank capital and liquidity– The rationale for the prudential regulation and supervision of insurers– Recent developments in the sterling overnight money market– Nowcasting world GDP and trade using global indicators– The Natural Rate Hypothesis: an idea past its sell-by date– Monetary Policy Roundtable

Contents of recent Quarterly Bulletins

Quarterly Bulletin Appendices 315

2013 Q4– SME forbearance and its implications for monetary and financial stability– Bringing down the Great Wall? Global implications of capital account liberalisation in China– Banknotes, local currencies and central bank objectives– Banks’ disclosure and financial stability– Understanding the MPC’s forecast performance since mid-2010– The financial position of British households: evidence from the 2013 NMG Consulting survey– What can company data tell us about financing and investment decisions?– Tiering in CHAPS– The foreign exchange and over-the-counter interest rate derivatives market in the United Kingdom– Qualitative easing: a new tool for the stabilisation of financial markets

2014 Q1– Money in the modern economy: an introduction– Money creation in the modern economy– The Court of the Bank of England– Dealing with a banking crisis: what lessons can be learned from Japan’s experience?– The role of business model analysis in the supervision of insurers– Nowcasting UK GDP growth– Curiosities from the vaults: a Bank miscellany– Monetary Policy Roundtable

2014 Q2– The UK productivity puzzle– The Bank of England as a bank– Credit spreads: capturing credit conditions facing households and firms– Assessing the risk to inflation from inflation expectations– Public attitudes to monetary policy– How have world shocks affected the UK economy?– How has the Liquidity Saving Mechanism reduced banks’ intraday liquidity costs in CHAPS?– Risk managing loan collateral at the Bank of England– Sterling Monetary Framework Annual Report 2013–14– A review of the work of the London Foreign Exchange Joint Standing Committee in 2013

2014 Q3– Innovations in payment technologies and the emergence of digital currencies– The economics of digital currencies– How might macroprudential capital policy affect credit conditions?– Household debt and spending– Enhancing the resilience of the Bank of England’s Real-Time Gross Settlement infrastructure

– Conference on Monetary and Financial Law– Monetary Policy Roundtable– Changes to the Bank’s weekly reporting regime

2014 Q4– Bank funding costs: what are they, what determines them and why do they matter?– Why is the UK banking system so big and is that a problem?– The interaction of the FPC and the MPC– The Bank of England’s approach to resolving failed institutions– The potential impact of higher interest rates on the household sector: evidence from the 2014 NMG Consulting survey

2015 Q1– Investment banking: linkages to the real economy and the financial system– Desperate adventurers and men of straw: the failure of City of Glasgow Bank and its enduring impact on the UK banking system– Capital in the 21st century– The Agencies and ‘One Bank’– Self-employment: what can we learn from recent developments?– Flora and fauna at the Bank of England– Big data and central banks

2015 Q2– Mapping the UK financial system– Banking sector interconnectedness: what is it, how can we measure it and why does it matter?– The prudential regulation of insurers under Solvency II– A bank within a bank: how a commercial bank’s treasury function affects the interest rates set for loans and deposits– Do inflation expectations currently pose a risk to inflation?– Innovations in the Bank’s provision of liquidity insurance via Indexed Long-Term Repo (ILTR) operations– A review of the work of the London Foreign Exchange Joint Standing Committee in 2014

2015 Q3– How has cash usage evolved in recent decades? What might drive demand in the future?– Bank failure and bail-in: an introduction– Insurance and financial stability– How much do UK market interest rates respond to macroeconomic data news?– Estimating market expectations of changes in Bank Rate– Over-the-counter (OTC) derivatives, central clearing and financial stability– Monetary Policy Roundtable

316 Quarterly Bulletin 2015 Q3

The Bank of England publishes information on all aspects of its work in many formats. Listed below are some of themain Bank of England publications. For a full list, please referto our website:

www.bankofengland.co.uk/publications/Pages/default.aspx.

Staff working papers

An up-to-date list of staff working papers is maintained on the Bank of England’s website at:

www.bankofengland.co.uk/research/Pages/workingpapers/default.aspx

where abstracts of all papers may be found. Papers publishedsince January 1997 are available in full, in portable documentformat (PDF).

No. 539 Bank leverage, credit traps and credit policies(July 2015)Angus Foulis, Benjamin Nelson and Misa Tanaka

No. 540 The rate elasticity of retail deposits in theUnited Kingdom: a macroeconomic investigation(August 2015)Ching-Wai (Jeremy) Chiu and John Hill

No. 541 Market beliefs about the UK monetary policy lift-offhorizon: a no-arbitrage shadow rate term structure modelapproach (August 2015)Martin M Andreasen and Andrew Meldrum

No. 542 Unconventional monetary policies and themacroeconomy: the impact of the United Kingdom’s QE2 andFunding for Lending Scheme (August 2015)Rohan Churm, Michael Joyce, George Kapetanios andKonstantinos Theodoridis

No. 543 Interest rates, debt and intertemporal allocation:evidence from notched mortgage contracts in theUnited Kingdom (August 2015)Michael Carlos Best, James Cloyne, Ethan Ilzetzki andHenrik Jacobsen Kleven

No. 544 Exchange rate regimes and current accountadjustment: an empirical investigation (August 2015)Fernando Eguren-Martín

No. 545 Into the light: dark pool trading and intraday marketquality on the primary exchange (September 2015)James Brugler

No. 546 Regulatory arbitrage in action: evidence frombanking flows and macroprudential policy (September 2015)Dennis Reinhardt and Rhiannon Sowerbutts

No. 547 Extreme downside risk and financial crises(September 2015)Richard D F Harris, Linh H Nguyen and Evarist Stoja

No. 548 A heterogeneous agent model for assessing theeffects of capital regulation on the interbank money marketunder a corridor system (September 2015)Christopher Jackson and Joseph Noss

External MPC Unit discussion papers

The MPC Unit discussion paper series reports on researchcarried out by, or under supervision of, the external membersof the Monetary Policy Committee. Papers are available fromthe Bank’s website at:

www.bankofengland.co.uk/research/Pages/externalmpcpapers/default.aspx.

The following papers have been published recently:

No. 41 The relevance or otherwise of the central bank’sbalance sheet (January 2014)David Miles and Jochen Schanz

No. 42 What are the macroeconomic effects of assetpurchases? (April 2014)Martin Weale and Tomasz Wieladek

Monetary and Financial Statistics

Monetary and Financial Statistics (Bankstats) contains detailed information on money and lending, monetary andfinancial institutions’ balance sheets, banks’ income andexpenditure, analyses of bank deposits and lending, externalbusiness of banks, public sector debt, money markets, issues of securities, financial derivatives, interest and exchange rates, explanatory notes to tables and occasional relatedarticles.

Bankstats is published on a monthly basis, free of charge, onthe Bank’s website at:

www.bankofengland.co.uk/statistics/Pages/bankstats/default.aspx.

Further details are available from the Statistics and RegulatoryData Division, Bank of England: telephone 020 7601 5432;email srdd_editor@bankofengland.co.uk.

Bank of England publications

Quarterly Bulletin Appendices 317

Articles that have been published in recent issues of Monetary and Financial Statistics can also be found on theBank’s website at:

www.bankofengland.co.uk/statistics/Pages/ms/articles.aspx.

Financial Stability Report

The Financial Stability Report is published twice a year underthe guidance of the Financial Policy Committee (FPC). Itcovers the Committee’s assessment of the outlook for thestability and resilience of the financial sector at the time ofpreparation of the Report, and the policy actions it advises toreduce and mitigate risks to stability. The Bank of Englandintends this publication to be read by those who areresponsible for, or have interest in, maintaining and promotingfinancial stability at a national or international level. It is ofespecial interest to policymakers in the United Kingdom andabroad; international financial institutions; academics;journalists; market infrastructure providers; and financialmarket participants. The Financial Stability Report is availableat:

www.bankofengland.co.uk/publications/Pages/fsr/default.aspx.

Handbooks in central banking

The series of Handbooks in central banking provide concise,balanced and accessible overviews of key central bankingtopics. The Handbooks have been developed from studymaterials, research and training carried out by the Bank’sCentre for Central Banking Studies (CCBS). The Handbooksare therefore targeted primarily at central bankers, but arelikely to be of interest to all those interested in the varioustechnical and analytical aspects of central banking. TheHandbook series also includes ‘Technical Handbooks’ which areaimed more at specialist readers and often contain moremethodological material than the Handbooks, incorporatingthe experiences and expertise of the author(s) on topics thataddress the problems encountered by central bankers in theirday-to-day work. All the Handbooks are available via theBank’s website at:

www.bankofengland.co.uk/education/Pages/ccbs/handbooks/default.aspx.

The Bank of England’s Sterling MonetaryFramework (the ‘Red Book’)

The ‘Red Book’ describes the Bank of England’s framework forits operations in the sterling money markets, which is designedto implement the interest rate decisions of the Monetary

Policy Committee while meeting the liquidity needs, and socontributing to the stability of, the banking system as a whole.It also sets out the Bank’s specific objectives for theframework, and how it delivers those objectives. Theframework was introduced in May 2006. The ‘Red Book’ isavailable at:

www.bankofengland.co.uk/markets/Documents/money/publications/redbook.pdf.

Cost-benefit analysis of monetary andfinancial statistics

The handbook describes a cost-benefit analysis (CBA)framework that has been developed within the Bank to ensurea fair balance between the benefits derived from good-qualitystatistics and the costs that are borne by reporting banks.Although CBA is a well-established approach in othercontexts, it has not often been applied to statistical provision,so techniques have had to be adapted for application to the Bank’s monetary and financial statistics. The handbook alsodiscusses how the application of CBA has enabled cuts in boththe amount and the complexity of information that is requiredfrom reporting banks.

www.bankofengland.co.uk/statistics/Pages/about/cba.aspx.

Credit Conditions Survey

Developments in credit conditions are of key interest to theBank of England in its assessment of economic conditions.This quarterly survey of bank and building society lenders is aninput to this assessment. The survey covers secured andunsecured lending to households and small businesses; andlending to non-financial corporations, and to non-bankfinancial firms. Copies are available on the Bank’s website at:

www.bankofengland.co.uk/publications/Pages/other/monetary/creditconditions.aspx.

Bank Liabilities Survey

Developments in lenders’ balance sheets are of key interest tothe Bank of England in its assessment of economic conditions.The aim of this quarterly survey of banks and building societylenders is to improve understanding of the role of lenders’liabilities and capital in driving credit and monetary conditions.The survey covers developments in the volume and price ofbank funding; developments in the loss-absorbing capacity ofbanks as determined by their capital positions; anddevelopments in the internal price charged to business unitswithin individual banks to fund the flow of new loans,

318 Quarterly Bulletin 2015 Q3

sometimes referred to as the ‘transfer price’. Copies areavailable on the Bank’s website at:

www.bankofengland.co.uk/publications/Pages/other/monetary/bls/default.aspx.

Credit Conditions Review

This quarterly publication presents the Bank of England’sassessment of the latest developments in bank fundingconditions and household and corporate credit. It drawsmainly on long-established official data sources, such as theexisting monetary and other financial statistics collected bythe Bank, and other data sources such as surveys of businessesand data from other organisations. The analysis also draws onthe results of the Bank of England’s Bank Liabilities Survey andCredit Conditions Survey. Copies are available on the Bank’swebsite at:

www.bankofengland.co.uk/publications/Pages/creditconditionsreview/default.aspx.

Quarterly Bulletin

The Quarterly Bulletin explores topical issues relating to theBank’s core purposes of monetary and financial stability.Some articles present analysis on current economic andfinancial issues, and policy implications. Other articlesenhance the Bank’s public accountability by explaining theinstitutional structure of the Bank and the various policyinstruments that are used to meet its objectives. TheQuarterly Bulletin is available at:

www.bankofengland.co.uk/publications/Pages/quarterlybulletin/default.aspx.

Inflation Report

The Bank’s quarterly Inflation Report was first published in1993. The Report sets out the detailed economic analysis and inflation projections on which the Bank’s Monetary Policy Committee bases its interest rate decisions, andpresents an assessment of the prospects for UK inflation. TheInflation Report is available at:

www.bankofengland.co.uk/publications/Pages/inflationreport/default.aspx.

Publication dates

Publication dates for 2015 are as follows:

Quarterly Bulletin Inflation ReportQ1 12 March February 12 FebruaryQ2 18 June May 13 MayQ3 18 September August 6 AugustQ4 15 December November 5 November

© Bank of England 2015ISSN 0005-5166Printed by Park Communications Limited