manual on mfi interest rate statistics - european central bank

EN

MANUALON MFI INTEREST RATE

STATISTICS

REGULATION ECB/2001/18

October 2003

EC

B

EZ

B

EK

T

BC

E

EK

P

MA

NU

AL

ON

MFI

INT

ER

EST

RA

TE S

TA

TIS

TIC

S R

EG

ULA

TIO

N E

CB

/20

01

/18

Oct

obe

r 20

03

MANUALON MFI INTEREST RATE

STATISTICS

REGULATION ECB/2001/18

October 2003

© European Central Bank, October 2003

Address Kaiserstrasse 29

60311 Frankfurt am Main

Germany

Postal address Postfach 16 03 19

60066 Frankfurt am Main

Germany

Telephone +49 69 1344 0

Internet http://www.ecb.int

Fax +49 69 1344 6000

Telex 411 144 ecb d

All rights reserved.

Reproduction for educational and non-commercial purposes is permitted provided that the source is acknowledged.

ISBN 92-9181-371-0 (print)

ISBN 92-9181-372-9 (online)

3ECB • Manua l on MF I i n t e re s t r a t e s t a t i s t i c s • Oc tobe r 2003

Table of contents

1. Aim of this manual 7

2. Scope and uses of MFI interest rate statistics 8

3. Basic definitions 10

4. Types of interest rate 124.1 Nominal versus effective interest rates 124.2 Annualised agreed rate 13

4.2.1 Definition and annualised agreed rate formula 134.2.2 Clarification of variable n in the annualised agreed rate formula 144.2.3 Treatment of exceptional repayments of principal 154.2.4 The annualised agreed rate formula for indefinite loans 164.2.5 The annualised agreed rate formula applied to bank overdrafts 164.2.6 The annualised agreed rate formula for one-off deposits 164.2.7 Treatment of disagio 174.2.8 Treatment of agio 17

4.3 Narrowly defined effective rate 184.4 Annualising variable interest rates 214.5 Treatment of taxes, subsidies and regulatory arrangements 21

4.5.1 Taxes, subsidies and favourable rates 214.5.2 Special national practices including regulatory arrangements 234.5.3 The annualised agreed rate formula for subsidised loans 23

4.6 Annual percentage rate of charge 244.6.1 Definition and link to the Consumer Credit Directive 244.6.2 Indicator for other loan charges 264.6.3 Charges to be taken into account at national level 274.6.4 Period of fixation in the calculation of the APRC 274.6.5 Treatment of subsidies in the APRC 284.6.6 Treatment of non-profit institutions serving households in the APRC 29

5. Business coverage 305.1 Interest rates on outstanding amounts 30

5.1.1 Definition of outstanding amounts 305.1.2 Bad loans and loans for debt restructuring below market conditions 30

5.2 Interest rates on overnight deposits, deposits redeemable at notice and bankoverdrafts 315.2.1 The balance at the time reference point as indicator for new business 315.2.2 Determining the interest rate on an overnight deposit 325.2.3 Combined deposit and loan facilities 335.2.4 Regular savings on a deposit redeemable at notice 34

5.3 Interest rates on new business 355.3.1 Definition of new business 355.3.2 New business on deposits with agreed maturity 355.3.3 Matured deposit with agreed maturity 365.3.4 Regular savings on a deposit with agreed maturity 375.3.5 New lending with fixed interest rate and with initial rate fixation 385.3.6 Top-up loans 405.3.7 Conversion of a bank overdraft into another type of loan 405.3.8 Loan taken out in tranches 415.3.9 The definition of new business for variable interest rates 42

4 ECB • Manua l on MF I i n t e re s t r a t e s t a t i s t i c s • Oc tobe r 2003

5.3.10 Choice of money market index 435.3.11 Money market index with floor and ceiling 435.3.12 Change in the value of a currency as external index 445.3.13 Timing differences 445.3.14 ‘Cooling off’ period 455.3.15 Loan offer and preliminary offer 455.3.16 Loans for debt restructuring in the context of new business 465.3.17 Moratorium on a loan 46

6. Time reference point 476.1 Time reference point for interest rates on outstanding amounts, overnight

deposits, deposits redeemable at notice and bank overdrafts 476.2 Time reference point for interest rates on new business 48

7. Instrument categories 497.1 Summary tables of indicators 497.2 General provisions 517.3 Foreign currency deposits and loans 537.4 Breakdown by sector 537.5 Breakdown by type of instrument 55

7.5.1 Types of deposits 557.5.2 Types of loans 56

7.6 Breakdown by amount category 577.7 Breakdown by original maturity, period of notice period and initial rate fixation 58

7.7.1 Time bands 587.7.2 Original maturity and period of notice 587.7.3 Period of maturity for a loan taken out in tranches 597.7.4 Initial period of fixation of the interest rate 59

8. Treatment of credit card credit 628.1 Credit card linked to an overnight deposit account 628.2 Credit card linked to a loan account 638.3 Credit card not linked to any account 63

9. Specific national products 659.1 Step-up (step-down) deposits and loans 659.2 Savings plan for housing loans 669.3 Interest rate on zero-coupon-bond-like savings bond 669.4 Split of loan into two parts 679.5 Option of converting the deposit into equity shares 689.6 Interest rate linked to share price 699.7 Treatment of deposit comprising two components 699.8 Purchase of mortgage loans by a credit institution 729.9 Securitisation of mortgage loans by a credit institution 73

10. Aggregation of the data and reporting obligations 7410.1 Overview 7410.2 Statistical information at the level of the reporting agents 7410.3 National weighted average interest rates 7510.4 Aggregated results for the euro area 76


11. Selection of the reporting agents 7711.1 Selection of the actual reporting population 7711.2 Census versus sampling approach 7811.3 Stratification of the potential reporting population 7911.4 Minimum national sample size 82

11.4.1 Definition 8211.4.2 Theoretical background 8511.4.3 Assumptions and data availability 8711.4.4 Sample size and maximum random error 88

11.5 Special provisions in the case of group reporting 9011.6 Allocation of the sample across strata 9111.7 Estimation of total new business volume 9211.8 Maintenance of the sample 9511.9 Further sampling issues 9711.10Description of the euro area sample 98

Index of terms 101


1 Aim of this manual

On 20 December 2001, the GoverningCouncil of the European Central Bank (ECB)adopted Regulation ECB/2001/18 concerningstatistics on interest rates applied bymonetary financial institutions to deposits andloans vis-à-vis households and non-financialcorporations (hereinafter ‘the Regulation’).The Regulation was published in the OfficialJournal of the European Communities on 12January 20021 and came into force on 31January 2002. It defines the statisticalstandards according to which MFI interest ratesare collected and compiled in the EuropeanUnion (EU). The Regulation is binding on theMember States participating in monetaryunion.

This manual contains no additionalrequirements and has no binding legal status.Instead it aims to further clarify and illustratethe requirements laid down in the Regulation,mainly through the use of extendeddefinitions, explanations of the underlyingconcepts and examples. It also brings togetherdetailed transcriptions from other legal actsreferred to in the Regulation. A clear andconsistent understanding of the statisticalrequirements contained in the Regulation bythe statisticians in the national central banks(NCBs) of the European System of CentralBanks (ESCB) and also in the accessioncountries is essential for the production ofharmonised monetary financial institution

(MFI) interest rate statistics. The informationin the manual may also interest reportingagents and users of the statistics.

The manual is composed of 11 chapters.Chapter 2 sets out the scope of MFI interestrate statistics with special emphasis on theuse of these statistics for monetary policypurposes. Chapter 3 defines the main termscontained in the Regulation. Chapter 4discusses the types of interest rate compiledunder the Regulation. Chapter 5 concerns MFIinterest rates on new business and onoutstanding amounts and Chapter 6 explainsthe time reference point for these two mainkinds of statistics. Chapter 7 provides anoverview of the indicators available at theeuro area and at the national level. Chapters 8and 9 give guidance on the treatment ofspecific deposit and loan products. Chapter10 summarises the steps needed to aggregatethe individual data to obtain euro area results.Chapter 11 sets out the method of selectingthe reporting agents for MFI interest ratestatistics. It tackles a full range of samplingissues, including the stratification procedure,the definition of the minimum sample size,the way of allocating the sample across strataand the maintenance of the sample over time.

1 OJ L 10, 12.1.2002, p.24; also available at www.ecb.int.

ECB • Manua l on MF I i n t e re s t r a t e s t a t i s t i c s • Oc tobe r 20038

2 Scope and uses of MFI interest rate statistics

Since January 1999, statistics comprising a setof 10 euro area retail interest rates2 havebeen published on a monthly basis in the ECBMonthly Bulletin. These statistics arecompiled according to a short-term approach,using existing national interest rate statistics.While this approach has ensured that someretail interest rate statistics were available asfrom the start of monetary union, it hasserious limitations. The underlying data arenot harmonised and the results have had tobe used with caution, with the focus on thedevelopment of the interest rates over timerather than their level. Hence, the short-termapproach has provided only the minimuminitial data that the ECB required formonetary policy purposes.3

Work on the development of a steady-stateapproach started in summer 1999 with theaim of compiling a set of euro area interestrates on retail deposit and lending businessthat is harmonised, complete, detailed, andable to cope with financial innovation. Toavoid invoking a potentially misleadingcontrast between ‘retail’ and ‘wholesale’interest rates, expressions that can carrydifferent meanings, the statistics developedunder the steady-state approach are referredto as ‘MFI interest rate statistics’.

The scope of euro area MFI interest ratestatistics4 is all interest rates that MFIs residentin the euro area apply to euro-denominateddeposits and loans vis-à-vis non-financialsectors (other than government) resident inthe euro area, i.e. vis-à-vis households andnon-financial corporations of any size. Inpractice, mainly credit institutions need toreport MFI interest rate statistics.5 Thestatistics are compiled for the euro area as awhole and individually for each Member Statein order to give information about the leveland development of interest rates both ateuro area and at national level.

MFI interest rate statistics are collected atmonthly frequency. The first data is collected

for the reference month January 2003. Thestatistics have three main uses:

• MFI interest rate statistics are needed toanalyse the monetary transmissionmechanism, as monetary policy istransmitted through the economy via thechange in interest rates. First, the statisticsgive the possibility of studying the pass-through of changes in official rates andmarket interest rates to lending anddeposit interest rates faced by householdsand non-financial corporations. Informationabout the speed and extent of the pass-through is essential to understand theeffect of monetary policy on the economy.Second, changes in MFI interest rates affectthe cost of capital and so influenceinvestment decisions and substitutionbetween current and future consumption.MFI interest rate statistics are thereforevital for any economic analysis over time.Third, the statistics allow income effectsto be analysed, as changes in MFI interestrates affect the interest paid or receivedby households and non-financialcorporations and hence the disposableincome of these sectors. Finally, MFIinterest rate statistics enable users toanalyse the credit channel of monetarypolicy, in particular the cost spreadbetween self-financing and credit, the so-called external finance premium.

2 Euro area statistics are provided for interest rates on overnightdeposits, deposits with agreed maturity up to one year, depositswith agreed maturity up to two years, deposits with agreedmaturity over two years, deposits redeemable at notice up tothree months, deposits redeemable at notice over three months,loans to enterprises up to one year, loans to enterprises over oneyear, consumer loans, and housing loans to households. Moreinformation on these series and the actual data are available atwww.ecb.int.

3 In accordance with European Monetary Institution, Statisticalrequirements for Monetary Union, 1996.

4 The scope of MFI interest rate statistics in a non-participatingMember State is all interest rates that MFIs resident in thatMember State apply to deposits and loans in national currencyvis-à-vis non-financial sectors (other than government) residentin the Member State, i.e. vis-à-vis households and non-financialcorporations of any size.

5 More detailed definitions and further explanations are given inChapter 3.


• MFI interest rate statistics enhance themonetary analysis in the euro area. Ideally,prices and quantities are analysed together.Information on the remuneration of M3and its components is one essential factorto explain monetary growth and to assessits effects on price stability. For example,the own rate of M3 could be compared tothe rate of return of alternative assets toexplain portfolio shifts between monetaryand non-monetary assets. Similarly, detailedMFI interest rates allow the developmentof loans to the private sector to beanalysed.

• MFI interest rate statistics allow themonitoring of structural developments in thebanking and financial system and theanalysis of financial stability issues. Usersmay study the development of banks’interest rate margins and changes in theirprofitability, and potentially adversedevelopments that may damage financialstability, such as how quickly banks’interest rate margins react to externaldevelopments or how the interest burdenchanges for households and non-financialcorporations.

6 Where outstanding amounts consist of a significant part ofvariable rate business, the related rates may also provideinformation on the pass-through of interest rates.

Figure 1Use of the reported indicators for monetary policy purposes

SCOPE OF ANALYSIS

MONETARY TRANSMISSION

Interest rate channel Credit

channel

MONETARY

ANALYSIS

FINANCIAL

STABILITY

Pass-

through of

interest rates

Cost of capital

substitution

effect

Income

effect

External

finance

premium

Money

demand

Credit

demand

Bank compe-

tition, bank

profitability

Interest rates

on outstanding

amounts6

X X X X

Interest rates

on new

business

X X X X X X

In conclusion, MFI interest rate statistics areessential to well-informed monetary policydecision making. Monetary authorities needto be frequently and rapidly informed of thechanges in these interest rates, so as to assessthe reach, scope or delayed effect ofmonetary decisions and their change overtime. MFI interest rate statistics are designedto meet these needs.


3 Basic definitions7

In order to ensure the comparability of MFIinterest rate statistics with other macro-economic statistics produced at Europeanlevel, the Regulation relies to the extentpossible on existing frameworks such as theEuropean System of Accounts (ESA) 19958

and the ECB’s MFI balance sheet statistics.9

The Regulation therefore uses a number ofexpressions that are common to Europeanand in particular to euro area statistics, butalso terminology that is specific to MFIinterest rates. The main terms are defined inArticle 1 of the Regulation. In the followingthese definitions are explained in more detail.

A participating Member State10 (Article 1(1)) isa country that has adopted the single currencyin accordance with the Treaty establishingthe European Community. The Regulationonly refers to ‘participating Member State(s)’.For simplicity this manual uses ‘euro area’ tosignify ‘participating Member States’ eventhough, strictly, the euro area is broader thanthe participating Member States. It alsocomprises some territories or countriesassociated with participating Member Statesthat have been authorised to adopt the euroas their legal currency and in which the singlemonetary policy of the ECB is conducted (asin the case of Monaco and the Frenchoverseas territories of Saint-Pierre-et-Miquelon and Mayotte).

An entity is regarded as a resident of orresiding11 in a Member State (Article 1(1))when it has a centre of economic interest inthe territory of that Member State, i.e. whenit has engaged for a year or more in economicactivity in that territory, or when it hasregistered or indicated an intention tooperate permanently in that territory. ForMFI interest rate statistics, the interest ratesand weights refer to deposits by and loans tocustomers resident in the euro area.12 Nodistinction is made for customers betweendomestic residents and residents of the othereuro area Member States. For example, theBank of Greece collects interest rates applyingto the banking business of credit institutions

and other institutions resident in Greece forcustomers resident in Greece and in the otherparticipating Member States, but not forcustomers resident outside the euro area.13

Customers resident outside the euro areaare in principle faced with the same interestrates for their deposits and loans ascustomers resident in the euro area.However, customers resident outside theeuro area might prefer different types ofinstrument than customers resident in theeuro area. Hence, if customers residentoutside the euro area were covered by MFIinterest rate statistics, MFIs would need toapply different weighting schemes to calculateaverage interest rates on business for all(resident and non-resident) customers ofcredit institutions and other institutions. Thismight lead to different results.

The reporting scheme defined in theRegulation applies only to MFIs other thancentral banks and money market funds, i.e. tocredit institutions and other institutions (Article1(3)) included in the ‘list of MFIs’14. E-moneyinstitutions are credit institutions, and so theyare covered by MFI interest rate statistics. Inprinciple, money market funds should becovered by MFI interest rate statistics.

7 This chapter refers mainly to Article 1 of the Regulation.8 See Annex A to Council Regulation (EC) No 2223/96 of 25 June

1996 on the European system of national and regional accountsin the Community, OJ L 310, 30.11.1996, p. 1, as last amendedby Regulation (EC) No 359/2002 of the European Parliamentand of the Council, OJ L 58, 28.2.2002, p. 1.

9 Regulation ECB/2001/13 of 22 November 2001 concerning theconsolidated balance sheet of the monetary financial institutionssector, OJ L 333, 17.12.2001, p. 1, as amended by RegulationECB/2002/8 of 21 November 2002, OJ L 330, 6.12.2002, p.29.

10 Defined in Article 1 of Council Regulation (EC) No 2533/98 of23 November 1998 concerning the collection of statisticalinformation by the European Central Bank, OJ L 318,27.11.1998, p. 8.

11 Defined in Article 1 of Regulation (EC) No 2533/98.12 For non-participating Member States, this should read here and

in the rest of the document ‘resident in the same Member Stateas the reporting MFI’.

13 For non-participating Member States, the principle is the‘residency in the same Member State as the MFI’. For example,Danmarks Nationalbank collects interest rates referring to thebanking business of credit institutions and other institutionsresident in Denmark vis-à-vis customers resident in Denmark.

14 A comprehensive list of all MFIs in the EU is produced andpublished by the ECB. Further information and the list areavailable at www.ecb.int.


However, as their business is not to receivedeposits or grant loans, money market fundsdo not pay interest on their liabilities orreceive interest on their assets in the sameway as credit institutions and otherinstitutions. Therefore, a different method ofdata collection will have to be developed.The ECB will also consider in due course theextension of the reporting population toother financial intermediaries, except pensionfunds and insurance companies, and to centralgovernment and insurance corporations.

MFI interest rate statistics (Article 1(4)) coverinterest rates that resident credit institutionsand other institutions apply to euro-denominated15 deposits and loans vis-à-visnon-financial sectors other than governmentresident in the euro area. Non-financialsectors other than government are householdsand non-financial corporations (Article 1(2)) ofany size. Non-profit institutions servinghouseholds (NPISHs) are indistinguishablyincluded with households as they are not veryimportant for the purposes of MFI interestrate statistics at euro area level and in nearlyall Member States.16 The sectoral classificationfollows the principles established inRegulation ECB/2001/13 and in Chapter 2 ofthe ESA 95.17

The definition of the potential reportingpopulation (Article 1(5)) follows from thedefinition of the scope of MFI interest ratestatistics. In each Member State, the potentialreporting population comprises all residentcredit institutions and other institutions whichtake euro-denominated deposits from and/orgrant euro-denominated loans to householdsand/or non-financial corporations resident inthe participating Member States. This meansthat the customers may be resident anywherein the euro area, not necessarily in the same

Member State as the reporting creditinstitution or other institution. As explainedabove, the Regulation does not require adistinction between domestic residents andresidents of the other euro area MemberStates and does not apply to central banksand money market funds.

NCBs select the reporting agents (Article 1(1))for MFI interest rate statistics from thepotential reporting population, which for eachNCB comprises only resident entities. Allreporting agents together constitute theactual reporting population.18 Reporting agentsare the legal and natural persons that aresubject to the ECB’s statistical reportingrequirements. They include the entities that,according to the national law of their MemberState of residence, are neither a legal personnor a collection of natural persons, but canbe subject to rights and obligations. Thepersons legally representing these entitiesmust fulfil their reporting obligations.

15 For non-participating Member States, this should read here andin the rest of the document ‘national currency’.

16 In a few Member States, loans to NPISHs are a not negligiblepart of ‘other loans to households’.

17 See also ECB ‘Money and Banking Statistics – Sector Manual– Guidance for the statistical classification of customers’, secondedition, November 1999. The sectoral breakdown is furtherdiscussed in Chapter 7.4. For the treatment of NPISHs see alsoChapter 4.6.6.

18 Further discussed in Chapter 11.1.


4 Types of interest rate19

4.1 Nominal versus effective interestrates

The terms nominal and effective interest ratehave a range of different meanings dependingon the Member State and the context. This isillustrated in the following graphic.

Figure 2Type of rate

Nominalrates

Effectiverates

Two definitions on the extremes of this rangeof meaning are advertised nominal rate on theleft and annual percentage rate of charge onthe right:

• Advertised nominal rates are interest ratesthat are shown as the headline rates20 bythe credit institutions or other institutionsin the windows of their branches, theirleaflets, advertisements, newspapers, otherjournals, etc. These rates give customersan indication of the current interest ratelevel for different banking products. Theadvertised nominal rates might be appliedto highly standardised deposits and loanproducts, but are not necessarily the ratesthe credit institution or other institutionactually pays to or charges its customers.Households and non-financial corporationsmight be able to negotiate better termsthan those advertised. The advertisednominal rates might also be prime ratesthat the credit institution offers to its mostcreditworthy customers. In this case theactually applied rate on a deposit or loanmight be less favourable than theadvertised nominal rate.

• The annual percentage rate of charge (APRC)is an effective lending rate that covers thetotal costs of the credit to the consumer,i.e. the interest payments as well as allother related charges. The concept of ‘totalcosts for the consumer’ was designed for

the purposes of consumer protection. Thecompilation of the APRC is defined inCouncil Directive 87/102/EEC21 and furtherexplained in Chapter 4.6.1.

In view of the range of different meanings ofthe terms nominal and effective, it is notpossible to say in general which type of ratesMFI interest rates in fact are. What is possibleinstead is to describe briefly the main featuresof these rates and to leave it to the individualuser of the statistics to then call MFI interestrates nominal or effective:

• MFI interest rates are agreed rates: Thedata collected refers to the interest ratethat is individually agreed between a creditinstitution or other institution and itscustomer. MFI interest rates are hencedistinct from advertised nominal rates,because households and non-financialcorporations might be able to negotiatewith the credit institution or otherinstitution better terms and conditionsthan those advertised.

• MFI interest rates are annualised: They areconverted to an annual basis and quoted inpercentages per annum. This means thatMFI interest rates take into account thefrequency of interest payments. Ceterisparibus, the more frequent the interestpayments, the higher the MFI interest raterecorded in the statistics.22 Twopossibilities exist for annualising interestrates: either an algebraic formula leadingto the annualised agreed rate discussed inChapter 4.2.1, or successive approximationresulting in the narrowly defined effectiverate discussed in Chapter 4.3.

19 This chapter refers mainly to Part 1 of Annex II to the Regulation.20 Together with the advertised nominal rate, the corresponding

effective interest rate might be published but usually in smallprint rather than as the headline rate.

21 Council Directive 87/102/EEC of 22 December 1986 for theapproximation of the laws, regulations and administrativeprovisions of the Member States concerning consumer credit, OJL 42, 12.2.1987, p. 48, as last amended by Directive 98/7/ECof the European Parliament and of the Council, OJ L 101,1.4.1998, p. 17.

22 See also Equations 2 and 3 below.


• 43 out of the 45 MFI interest rates excludecharges: For deposits MFIs pay interest tothe customer but might also charge fees.Analogously, for loans the customer has topay an amount comprising an interest ratecomponent and a component made up ofother related charges. All but two MFIinterest rates refer only to the interestrate component and exclude all charges.The exceptions are the two additionalseries collected for consumer credit andloans to households for house purchases,where in addition to a rate without chargesthe APRC is also required that covers theinterest rate component and thecomponent of other charges.

4.2 Annualised agreed rate23

4.2.1 Definition and annualised agreedrate formula

The annualised agreed rate is defined inparagraph 1 of Annex II to the Regulation as‘the interest rate that is individually agreedbetween the reporting agent and thehousehold or non-financial corporation for adeposit or loan, converted to an annual basisand quoted in percentages per annum. Theannualised agreed rate shall cover all interestpayments on deposits and loans, but no othercharges that may apply. Disagio, defined asthe difference between the nominal amountof the loan and the amount received by thecustomer, shall be considered as an interestpayment at the start of the contract (time t0)and shall therefore be reflected in theannualised agreed rate.’

An annualised agreed rate is distinct from anadvertised nominal rate, as the customermight be able to negotiate with the creditinstitution or other institution better termsand conditions than those advertised. Anannualised agreed rate reflects thecreditworthiness and other qualities of thecustomer (in respect of loans) and thesolvency and other qualities of the creditinstitution as determined by the customer (inrespect of deposits). The annualised agreed

rate is influenced by the budget, capital orother constraints faced by the creditinstitution in granting loans and takingdeposits, including competition with othertypes of financial institution and product. It isa result of the demand and supply conditionsin the deposit and loan markets.

Paragraph 2 of Annex II to the Regulationprovides the formula for annualising theagreed interest rate, i.e. for convertinginterest payments that are due at regularintervals within a year to a yearly basis. It isapplied in cases where the interest paymentsthat are agreed between the credit institutionor other institution and the customer arecapitalised at regular intervals within a year,for example per month or quarter, ratherthan per annum:

1nr

1xn

ag −

+= [Equation 1]

Meaning of letters and symbols:

x Annualised agreed rate,

rag Interest rate per annum that is agreedbetween the reporting agents and thehousehold or non-financial corporation fora deposit or loan where the dates of theinterest capitalisation of the deposit andall the payments and repayments of theloan are at regular intervals in the year,and

n Number of interest capitalisation periodsfor the deposit and (re)payment periodsfor24 the loan per year, i.e. 1 for yearlypayments, 2 for semi-annual payments, 4for quarterly payments, and 12 for monthlypayments.

For example, a customer and a creditinstitution agree on a five-year loan at 10%per annum (p.a.) for the entire maturity,where the interest is paid at the end of eachquarter and the principal repaid at the end of

23 See also paragraphs 1 and 2 of Annex II to the Regulation.24 The text in strikethrough is included in the Regulation. However,

the application of the formula is clearer without it. See alsoChapter 4.2.2.


the fifth year. The annualised agreed rate forthis loan is then 10.3813% p.a. and calculatedas follows:

( ) 10381289.01410.011n

r1x

4nag =−+=−

+=

[Equation 2]

If, in the same example, the interest paymentswere at monthly frequency, then theannualised agreed rate would be slightlyhigher at 10.4713% p.a. and calculated asfollows:

( ) 10471307.011210.011n

r1x

12nag =−+=−

+=

[Equation 3]

In the case of daily interest capitalisation, n =365 should be used in Equation 1 followingthe convention of a standard year of 365 daysas specified in paragraph 12 of Annex II tothe Regulation.25 In the above loan example,the annualised agreed rate for monthlyinterest payments would be 10.5156% p.a.and calculated as follows:

( ) 1051558.0136510.011n

r1x

365nag =−+=−

+=

[Equation 4]

Equation 1 may also be used to derive theannualised agreed rate, for example in thecase of a deposit of EUR 10,000 that is placedfor two years where EUR 11,000 is paid outto the customer at maturity. During the twoyears, the customer earns 10%. Theannualised agreed rate is 4.8809% andcalculated as follows:

0488084.012

12

10.011n

r1x

21

nag =−

+=−

+=

[Equation 5]

4.2.2 Clarification of variable n in theannualised agreed rate formula

Question:

Which value should variable n take in theannualised agreed rate formula (Equation 1)when the interest payments and therepayment of a loan occur at different butregular intervals?

Case A: A customer and a credit institutionagree on a two-year loan at 10% p.a.with monthly interest payments. Theprincipal is repaid at the end of thesecond year. Should variable n inEquation 1 be:a) equal to 12 (based on the

frequency of interest payments),or

b) equal to ½ (based on therepayment frequency)?

Case B: A customer and a credit institutionagree on a five-year loan at 10% p.a.for the entire maturity, where theinterest should be paid at the end ofeach quarter and the principal shouldbe paid back in tranches on amonthly basis. Should n in Equation1 be:a) equal to 4 to only reflect

quarterly interest payments, orb) equal to 12 to also include the

repayments?

Answer:

As a general rule, for MFI interest ratestatistics the value of variable n in Equation 1is determined by the frequency of the interestpayments and not by the repayment periodsof the principal. If this rule is followed, theannualised agreed rate coincides with thenarrowly defined effective rate (NDER)26

whenever the interest payments are morefrequent or equally frequent than therepayments of the principal. This includes allcases where the principal is repaid at the endof the contract. An overview of possiblecombinations of interest payment andrepayment frequencies is given in thefollowing matrix.

25 If 360 days were used in the formula as a standard year insteadof the convention of 365 days, a different result would beachieved. The size of the difference depends on the level of theagreed interest rate rag.



It is expected that in general the frequency ofthe interest payments is higher or equal tothe repayment frequency of the principal (lightgrey boxes in Figure 3) and that hence theannualised agreed rate and the NDERcoincide. An example is Case A, wherevariable n in Equation 1 is equal to 12reflecting the number of interest paymentsper year. As the interest payments are atregular monthly intervals and the principal isrepaid in full at the end of the contract, i.e.situation E.M in Figure 3, Equation 1 providesfor n = 12 the same result as the NDER, i.e.an interest rate of 10.4713%.

In the remaining cases, where the frequencyof the interest payments is lower than therepayments of the principal (dark grey boxesin Figure 3), the annualised agreed ratedeviates from the NDER. The NDER thenrepresents the mathematically correctcalculation and the annualised agreed rate aclose approximation, which is in line with theRegulation. An example is Case B, wherevariable n in Equation 1 is equal to 4 reflectingthe quarterly interest payments. Since therepayments of the principal are at monthlyfrequency, the annualised agreed rate differs

Figure 3

from the NDER, i.e. situation M.Q in Figure3. The annualised agreed rate is 10.3813%and the NDER is 10.3795%.

The annualised agreed rates and the NDERsfor a 5-year loan at 10% p.a., with thecombinations of interest payment andrepayment frequencies as given in Figure 3,are summarised in Figure 4:

4.2.3 Treatment of exceptionalrepayments of principal

Question:

Is it in line with the Regulation to apply theannualised agreed rate formula (Equation 1)to financial contracts with regular interestcapitalisation, for example quarterly interestpayments, which provide the option to thecustomer to have exceptional repayments ?

Answer:

Repayments of the principal whether regularor irregular only influence the level of theannualised agreed rate and NDER if the

Figure 4

Frequency of interest payments

Monthly (.M) Quarterly (.Q) Yearly (.Y)

Monthly (M.) M.M M.Q M.Y

Quarterly (Q.) Q.M Q.Q Q.Y

Yearly (Y.) Y.M Y.Q Y.Y

Frequency of

repayment of

principal

End of contract (E.) E.M E.Q E.Y

Frequency of interest payments

Monthly (.M) Quarterly (.Q) Yearly (.Y)

Annualised

agreed rate

NDER Annualised

agreed rate

NDER Annualised

agreed rate

NDER

Monthly (M.) 10.4713 10.3813 10.3795 10.0000 9.0022

Quarterly (Q.) 10.4713 10.3813 10.0000 9.0331

Yearly (Y.) 10.4713 10.3813 10.0000

Frequency

of

repayment

of

principal End of contract

(E.)

10.4713 10.3813 10.0000


repayments are more frequent than theinterest payments. This is analogous to theissue discussed in Chapter 4.2.2. For example,in the case of monthly interest payments andexceptional repayments of the principal thatare not more frequent than monthly, theannualised agreed rate and the NDER coincidewith n = 12 in Equation 1 reflecting thefrequency of interest payments per year.Likewise, if the interest is paid quarterly andthe possibility for exceptional repayments ofthe principal occurs not more frequently thanquarterly, the annualised agreed rate and theNDER coincide with n = 4 in Equation 1.

It is assumed that in general the interest ispaid more frequently than exceptionalrepayments take place. Hence, the variable nin Equation 1 is determined by the frequencyof the interest payments and the possibilityof exceptional repayments can be ignored. Ifexceptional repayments occur more frequentthan the interest payments, then theannualised agreed rate deviates from theNDER. In that case the NDER represents themathematically correct calculation and theannualised agreed rate a close approximation,which is in line with the Regulation.

4.2.4 The annualised agreed rate formulafor indefinite loans

Question:

Can the annualised agreed rate formula(Equation 1) also be applied to loans thathave been granted to customers indefinitelybut still have regular interest payments?

Answer:

The variable n in Equation 1 refers not to thematurity of the loan but to the frequency ofthe interest payments. Hence, Equation 1 canbe applied.

4.2.5 The annualised agreed rate formulaapplied to bank overdrafts

Question:

What should n be in the annualised agreedrate formula (Equation 1) for bankoverdrafts27 that are characterised by havingirregular, rather than pre-defined, periods ofutilisation and/or repayment, and interestpayments that are based on the dailyoutstanding amount? Should n be equal to365?

Answer:

The use of n = 365 in Equation 1 assumesthat the interest is paid on a daily basis as inEquation 4. In this example, however, theinterest is computed (ex post) by using thedaily outstanding amount, but the interest ispaid at the end of the month. In the case ofmonthly interest payments variable n is equalto 12, in the case of quarterly interestpayments it is equal to 4.

4.2.6 The annualised agreed rate formulafor one-off deposits

Question:

Should the annualised agreed rate formula(Equation 1) be applied in the case of a one-off deposit with an agreed maturity of threemonths?

Answer:

For a one-off deposit with an agreed maturityof three months, two cases can bedistinguished. If the interest is paid at the endof each month, variable n in Equation 1 isequal to 12. If the interest is paid at the endof the three months, then n is equal to 4. Forexample if 10% is agreed for the 3-monthdeposit, then the annualised agreed rate forn = 12 is 10.47% and for n = 4 it yields10.38%.

27 Defined in Chapter 7.5.2.


4.2.7 Treatment of disagio

Question:

The last sentence in paragraph 1 of Annex IIto the Regulation states: ‘Disagio, defined asthe difference between the nominal amountof the loan and the amount received by thecustomer, shall be considered as an interestpayment at the start of the contract (time t0)and shall therefore be reflected in theannualised agreed rate.’ How should thefollowing cases of disagio be treated in theannualised agreed rate?

Case A: Disagio in the sense that the agreedmonthly interest payments are madeat the end of the previous period.

Case B: Disagio in the sense that the agreedmonthly interest payments are madeat the end of the previous period,with a different interest rate for thefirst period.

Case C: Disagio in the sense of a payment atthe beginning of the contract thathas no link with the subsequentinterest payments, which are basedon the outstanding amount of theloan in the previous period.

Answer:

In Cases A, B and C, the correct interest ratecan be given by the NDER. In Cases A and B,it might be possible to reflect the advancepayment for the first period as disagio inEquation 1. However, this formula ignoresfor all other periods that the interest is paidin advance and not at the end of the period.For such a complex product, only the NDERwill yield the correct result.

4.2.8 Treatment of agio

Question:

How should the annualised agreed rate bedetermined when the financial contract

includes an agio defined as the ‘inverse’ of adisagio,28 i.e. the price being higher than thenominal amount of the loan or non-negotiabledebt security? Should the agio beincorporated at all? An example is thefollowing purchase of a (borrowers’) noteloan (Schuldscheindarlehen), where the agio isincluded in the purchase price of the noteloan:

Purchase of note loan on 6 March 2003,nominal: EUR 5,112,918.81

Value date: 10 March 2003Interest rate: 7.60 %Original maturity: 25 March 2001 –

25 March 2004Purchase price: EUR 5,403,332.60

Agio (included in purchase price):EUR 5,403,332.60 – EUR 5,112,918.81 =EUR 290,413.79

Answer:

The (secondary) purchase of this note loanon 6 March 2003 is not considered to be newbusiness29 because no new funds are beinggranted to the borrower and no new termsand conditions are negotiated for theborrower. The (secondary) purchase is simplya change in ownership of the note loan. Newbusiness would only occur in the case of a(primary) issue of the note loan by a creditinstitution or other institution, where at thetime of issue the borrower receives moneythrough the issuance of a financial instrument,which according to the ECB’s MFI balancesheet statistics is classified as a loan ratherthan as debt security. If this primary issueinvolves an agio, it would be reflected in theMFI interest rate.

The agio is reflected in the MFI interest ratestatistics on outstanding amounts,30 irrespectiveof whether the note loan or non-negotiabledebt security is initially issued or subsequentlyacquired by the credit institution or otherinstitution. The agio results in a reduction of

28 See also Chapter 4.2.7.29 See also Chapter 5.3.1.30 See also Chapter 5.1.


the interest rate. The rate was 7.6% on EUR5,112,918.81, which results in an interestpayment of EUR 388,581.83. This amountapplied to the now higher purchase price ofEUR 5,403,332.60 is equivalent to an interestrate of 7.1915%. The latter is, from 6 March2003 onwards, reflected in the MFI interestrate statistics on outstanding amounts.

The same holds true for the (primary) issueand (secondary) purchase of any other non-negotiable debt security, which according toMFI balance sheet statistics is to be classifiedas a loan.

4.3 Narrowly defined effective rate31

Instead of the annualised agreed rate, NCBsmay require their reporting agents toimplement the NDER for all or some depositor loan instruments referring to new businessand outstanding amounts. The NDER refersto an annual basis and is defined as theinterest rate that equalises the present valueof all commitments other than charges(deposits or loans, payments or repayments,interest payments), future or existing, agreedby the credit institution or other institutionsand the household or non-financialcorporation. The NDER is equivalent to theinterest rate component of the APRC32 i.e. itdoes not take into account the component ofother charges. Hence, the basic formula forthe APRC in Annex II of Directive 98/7/EC33

applies, which is given as Equation 7 inChapter 4.6.1, but without the references toother charges. This Equation 7 is equivalentto the formula proposed by the InternationalSecurities Markets Association for theexponential interest rate calculation for allmaturities. Hence, in the case when years arepresumed to have 365 days and the amountof the deposit or loan is placed or paid out inone amount, the following applies:

∑∑=

−

=

+=

+=

N

1n

365D

n

N

1n 365D

nn

n)i1(*CF

)i1(

CFA

[Equation 6]


i Interest rateCFn Cash flow n, from the perspective of the

investor in the case of deposits and fromthe point of view of the credit institutionin the case of loans

N Number of cash flows associated withthe financial instrument

A Amount of the deposit (loan) initiallyplaced (paid out)

Dn Timing of the cash flow n, expressed indays after the first cash flow (in general,the date of investment of the deposit orvaluation of the loan).

The only difference between the NDER andthe annualised agreed rate is the underlyingmethod for annualising interest payments. TheNDER uses successive approximation and canbe applied to any type of deposit or loan. Theannualised agreed rate uses the algebraicformula in Equation 1 and is only applicableto deposits and loans where the dates ofinterest capitalisation are at regular intervals.

Both types of rates, the NDER and theannualised agreed rate, may be reported forthe purposes of MFI interest rate statistics.The reason is that for products with regularcapitalisation periods, where interestpayments occur more frequently or equallyfrequently than the repayments of theprincipal, including all cases where theprincipal is repaid in full at the end of thecontract, the annualised agreed rate and theNDER coincide.34 One formula can be derivedfrom the other. This applies also to productswith irregular or exceptional repayments ofthe principal as long as these do not occur

31 See also paragraph 3 of Annex II to the Regulation.32 Further discussed in Chapter 4.6.1.33 Directive 98/7/EC of the European Parliament and of the Council

of 16 February 1998 amending Directive 87/102/EEC for theapproximation of the laws, regulations and administrativeprovisions of the Member States concerning consumer credit, OJL 101, 1.4.1998, p. 17.

34 See also Chapter 4.2.2.


more frequently than the interest payments.35

Hence, for the majority of retail products theNDER and the annualised agreed rate lead tothe same result. For products with complexcash flow, however, only the NDER gives themathematically correct result. The annualisedagreed rate provides a close approximation,which is in line with the Regulation. It isassumed that the difference in results is smallespecially since MFI interest rate statisticsuse the convention of a standard year of 365days, as specified in paragraph 12 of Annex IIto the Regulation, i.e. ignoring the effect ofan additional day in leap years. The choicebetween the annualised agreed rate and theNDER does not therefore compromise thequality of the euro area statistics or thecomparability across countries.

Equation 2 gives an annualised agreed rate of10.3813% p.a. for a five-year loan at 10%where the interest is paid at the end of eachquarter and the principal repaid at maturity.The same result is achieved with Equation 6for the NDER, if the calculation refers to astandard year of 365 days ignoring that 2004is a leap year. This is shown in Figure 5,where t = 365 days per year, 365/12 days permonth are assumed and 365 is used in thediscount factor (1+NDER)^(-t/365).36

35 See also Chapter 4.2.3.36 The same result may be achieved by using 360 days consistently

in the calculation of the NDER. Consistently means that t = 360days per year and 30 days per month are assumed and also360 is used in the discount factor (1+NDER)^(-t/360).

5-year loan, quarterly interest rate payments, repayment of principal in 2005, standard year of 365 days:

tOutstand-

ing loan

Interest

rate p.a.

Interest

payments

Repayments

of principalCash flow

Discount factor =

(1+NDER)^(-t/365)

Present value

of cash flowNDER

15/11/2000 1.00 10,000 -10,000

14/02/2001 91.25 10,000 10% 250 0 250 0.98 243.90 0.10381316/05/2001 182.50 10,000 10% 250 0 250 0.95 237.95 0.10381315/08/2001 273.75 10,000 10% 250 0 250 0.93 232.15 0.10381315/11/2001 365.00 10,000 10% 250 0 250 0.91 226.49 0.10381314/02/2002 456.25 10,000 10% 250 0 250 0.88 220.96 0.10381316/05/2002 547.50 10,000 10% 250 0 250 0.86 215.57 0.10381315/08/2002 638.75 10,000 10% 250 0 250 0.84 210.32 0.10381315/11/2002 730.00 10,000 10% 250 0 250 0.82 205.19 0.10381314/02/2003 821.25 10,000 10% 250 0 250 0.80 200.18 0.10381316/05/2003 912.50 10,000 10% 250 0 250 0.78 195.30 0.10381315/08/2003 1003.75 10,000 10% 250 0 250 0.76 190.54 0.10381315/11/2003 1095.00 10,000 10% 250 0 250 0.74 185.89 0.10381314/02/2004 1186.25 10,000 10% 250 0 250 0.73 181.36 0.10381315/05/2004 1277.50 10,000 10% 250 0 250 0.71 176.93 0.10381314/08/2004 1368.75 10,000 10% 250 0 250 0.69 172.62 0.10381314/11/2004 1460.00 10,000 10% 250 0 250 0.67 168.41 0.10381313/02/2005 1551.25 10,000 10% 250 0 250 0.66 164.30 0.10381315/05/2005 1642.50 10,000 10% 250 0 250 0.64 160.29 0.10381314/08/2005 1733.75 10,000 10% 250 0 250 0.63 156.38 0.10381314/11/2005 1825.00 10,000 10% 250 10,000 10,250 0.61 6,255.28 0.103813

5,000 10,000 5,000 10,000.00

Figure 5


Deposit with agreed maturity of two years, interest rate payment at end of second year:

tOutstand-

ing deposit

Interest

rate

Reinvested

interest

Interest

payments

Payments

of

principal

Cash

flow

Discount factor =

(1+NDER)^(-t/365)

Present

value of

cash flow

NDER

15/11/2001 1 10,000 -10,000

15/11/2002 365 10,000 0 0 0 0 0.95 0 4.8809%15/11/2003 730 10,000 10% 0 1,000 10,000 11,000 0.91 10,000 4.8809%

1,000 10,000 1,000 10,000

The interest rate would be slightly lower, inthis example at 10.3759% p.a., if it was

recognised that 2004 is a leap year with 366days. This is shown in Figure 6.

Figure 6

5-year loan, quarterly interest rate payments, repayment of principal in 2005, 2004 is leap year:

tOutstand-

ing loan

Interest

rate p.a.

Interest

payments

Repayments

of principalCash flow

Discount factor =

(1+NDER)^(-t/365)

Present value

of cash flowNDER

15/11/2000 1.00 10,000 -10,000

14/02/2001 91.25 10,000 10% 250 0 250 0.98 243.91 0.10375916/05/2001 182.50 10,000 10% 250 0 250 0.95 237.96 0.10375915/08/2001 273.75 10,000 10% 250 0 250 0.93 232.16 0.10375915/11/2001 365.00 10,000 10% 250 0 250 0.91 226.50 0.10375914/02/2002 456.25 10,000 10% 250 0 250 0.88 220.98 0.10375916/05/2002 547.50 10,000 10% 250 0 250 0.86 215.59 0.10375915/08/2002 638.75 10,000 10% 250 0 250 0.84 210.33 0.10375915/11/2002 730.00 10,000 10% 250 0 250 0.82 205.21 0.10375914/02/2003 821.25 10,000 10% 250 0 250 0.80 200.20 0.10375916/05/2003 912.50 10,000 10% 250 0 250 0.78 195.32 0.10375915/08/2003 1003.75 10,000 10% 250 0 250 0.76 190.56 0.10375915/11/2003 1095.00 10,000 10% 250 0 250 0.74 185.92 0.10375914/02/2004 1186.25 10,000 10% 250 0 250 0.73 181.38 0.10375916/05/2004 1278.50 10,000 10% 250 0 250 0.71 176.91 0.10375915/08/2004 1369.75 10,000 10% 250 0 250 0.69 172.60 0.10375915/11/2004 1461.00 10,000 10% 250 0 250 0.67 168.39 0.10375914/02/2005 1552.25 10,000 10% 250 0 250 0.66 164.29 0.10375916/05/2005 1643.50 10,000 10% 250 0 250 0.64 160.28 0.10375915/08/2005 1734.75 10,000 10% 250 0 250 0.63 156.38 0.10375915/11/2005 1826.00 10,000 10% 250 10,000 10,250 0.61 6,255.12 0.103759

5,000 10,000 5,000 10,000.00

Equation 5 gives an annualised agreed rate of4.8809% p.a. for a deposit of EUR 10,000 thatis placed for two years, where the customerreceives EUR 11,000 from the credit

institution at maturity. The NDER for thisexample leads to the same result; thecalculation is shown in Figure 7.

Figure 7


4.4 Annualising variable interest rates

Question:

In March 2003, a customer and a creditinstitution agree on a two-year loan with avariable interest rate. The contract specifiesthe lending rate as a certain spread over anunderlying index. The value of the interestrate for the first month, i.e. March 2003, is10%. In March 2003 there is no informationregarding the future development of theinterest rate for the loan as it will rise and fallbased on the movement of the underlyingindex. In which way would the reporting agentcalculate the annualised agreed rate or NDERin March 2003? Would it assume that theinterest rate is 10% for the remaining periods?

Answer:

For the statistics on new business,37 theinterest rate that is taken into account in thecalculation of the annualised agreed rate orNDER is 10%. In general, for variable rates itis assumed that the interest rate remainsconstant at the level on the date of agreementon the contract, in this example at 10%. Thisis the same as assuming that the futureunknown development of the interest rate isnot taken into account. Hence, if the variablerate is agreed and paid per annum, theannualised agreed rate or NDER to berecorded as new business in March 2003 is10%. If the variable rate is to be paid permonth or quarter, the interest rate needs tobe annualised. The annualised agreed rate andNDER would then be 10.4713% and 10.3813%respectively.

The interest rate that is recorded for newbusiness is also reflected in the statistics onoutstanding amounts38 as the first reporting inMarch 2003. The subsequent reporting onoutstanding amounts reflects the interest rateapplied by the reporting institution at thetime of the calculation of MFI interest ratesand hence shows the variability of the interestrate over time.

4.5 Treatment of taxes, subsidies andregulatory arrangements39

4.5.1 Taxes, subsidies and favourable rates

The annualised agreed rate and the NDERreflect what the reporting agent pays ondeposits and receives for loans. If the amountpaid by one party and received by the otherdiffers, the point of view of the reporting agentdetermines the interest payment covered byMFI interest rate statistics. Following thisprinciple, MFI interest rates:

• are recorded on a gross basis before tax,since the pre-tax interest rates reflect whatreporting agents pay on deposits andreceive for loans,

• do not take into account subsidies grantedto households or non-financialcorporations by third parties such asgovernment, because these subsidies arenot paid or received by the reporting agent,

• reflect favourable rates that reporting agentsapply to their employees. Favourable ratesdo not include a subsidy granted by a thirdparty but are actually applied by thereporting agent.

Therefore, in the case of deposits, MFI interestrate statistics capture what the creditinstitution or other institution pays but notwhat the household or non-financialcorporation receives in terms of interestpayments. For example, if a customer receives5% p.a. on a deposit where 3% is actually paidby the reporting agent and the other 2% is asubsidy by a third party, which is transferredto the customer via the reporting agent, thenthe 3% p.a. is covered by MFI interest ratestatistics. Further examples are given in Figure8. The deposit interest rates covered by MFIinterest rate statistics are shaded, i.e. the 3%on the deposits of customers 1, 2 and 3 andthe 5% for the employee.

37 See also Chapter 5.3.1.38 See also Chapter 5.1.39 See also paragraphs 4 to 8 of Annex II to the Regulation.


Cus tomer 3

Cus tomer 2

Cus tomer 1

Emp loyee

Cre

dit i

nstit

utio

n Governm

ent

Pays 3% Transfers subs idy of 2% 5%

Pays 3%

Pays 3%

Pays 5%

Subsidy of 2%

Subs idy of 2% for cus tomer 3

Figure 8Subsidies and favourable rates on deposits

Analogously in the case of loans, MFI interestrate statistics capture what the creditinstitution or other institution charges interms of interest rates but not what thehousehold or non-financial corporation pays.For example, if a customer pays 6% p.a. for aloan where 10% is actually charged by thereporting agent but a third party deducts a

4% subsidy from this interest rate, and this istransferred to the customer via the reportingagent, the 10% p.a. is covered by MFI interestrate statistics. Further examples are given inFigure 9. The lending interest rates reflectedin MFI interest rate statistics are shaded, i.e.the 10% for the loans of customers 1, 2 and 3and the 6% for the loan of the employee.

Figure 9Subsidies and favourable rates on loans

Customer 3

Customer 2

Customer 1

Emp loyee

Cre

dit

in

sti

tuti

on

Go

vern

men

t

Charges 10%

Charges 10%

Charges 6%

Subsidy of 4%

Subsidy of 4% for customer 3

Charges 10%

Transfers subs idy of 4% 6%


4.5.2 Special national practices includingregulatory arrangements

Where national regulatory arrangements affectinterest payments, for example directsubsidies on certain types of instruments orinterest rate ceilings, these are reflected inMFI interest rate statistics. Any change to theregulatory arrangements, for example anincrease (decrease) in the level ofadministrated interest rates, is shown in thestatistics as an increase (decrease) of therespective MFI interest rate.

In addition to national regulatoryarrangements, the level and development ofMFI interest rates may also be influenced byspecial national practices that are not legallybinding. These can include nationalconventions, institutional arrangements andspecific deposit or lending products offeredat national level:

• National conventions cover usual bankingpractices which give rise to lower or higherthan usual interest rates, which are notnecessarily due to legal acts, for example ageneral but not legally enforced 0%remuneration on overnight deposits.

• Institutional arrangements are similar tonational conventions but affect only aspecific group of institutions. The legalstatus of these arrangements is irrelevantfor the purpose of analysing the MFIinterest rate statistics.

• Specific national products may, as a result oftheir special features, carry unusually highor low interest rates compared with otherproducts falling into the same instrumentcategory. The treatment of some specificproducts is defined in paragraphs 74 to 82of Annex II to the Regulation. Additionalspecific products are included in Chapter9 of this document.

All special national practices, comprisingregulatory arrangements, nationalconventions, institutional arrangements and

specific products, that are important becausethey have a significant influence on the leveland/or development of the deposit andlending rates set by MFIs, are documented inmethodological notes prepared by the ECB. Theaim of these is not to provide a completeoverview of existing regulatory arrangementsand specific national products but rather tocover those phenomena that are relevant fora correct analysis and interpretation of MFIinterest rate statistics. For example, thefollowing are covered: direct subsidies oncertain types of instruments, interest rateceilings, 0% remuneration on overnightdeposits, deposits placed at (close to) 0%interest, loans granted at (close to) 0%interest, and unusually high interest ratesoffered on specific national products.

4.5.3 The annualised agreed rate formulafor subsidised loans

Question:

How should the annualised agreed rate becalculated for a subsidised loan, where thecredit institution receives interest paymentsfrom its customer and the subsidy from thegovernment at different frequencies? Forexample, the credit institution receives 6%interest for a subsidised loan, 4% directly fromthe customer and 2% from the government.The interest payments by the customer takeplace monthly, whereas those by thegovernment are made only twice a year.Hence, no single value exists for the variablen in the annualised agreed rate formula(Equation 1): n would be 12 for the interestpaid by the customer and 2 for the interestpaid by the government.

Answer:

In this example of a subsidised loan at 6% p.a.where the customer pays 4% p.a. at monthlyfrequency and the government pays thesubsidy of 2 % p.a. every six months, the raterecorded in MFI interest rate statistics is6.1416% calculated as NDER.


Equation 1 cannot derive this result. Thissimple formula allows either the use of n =12, which gives 6.17%, assuming that bothcustomer and government pay monthly or,alternatively, it permits the division of theloan into two parts: one at 4% with n = 12and one at 2% with n = 2, leading to 6.08% intotal. Both results are incorrect. Equation 1can be applied as long as there is only onefrequency for all the payments but does notdeliver correct results for more complexproducts. NCBs are free to adjust Equation 1to make it fit to more complex nationalproducts. Alternatively, NCBs may ask thereporting agents to calculate the NDER forcomplex products.40

4.6 Annual percentage rate of charge41

4.6.1 Definition and link to the ConsumerCredit Directive

The APRC is defined in Article 1a(1)(a) ofDirective 87/102/EEC42 (hereinafter the‘Consumer Credit Directive’) as ‘that rate, on anannual basis which equalizes the present valueof all commitments (loans, repayment andcharges), future or existing, agreed by thecreditor and the borrower, [and] shall becalculated in accordance with the mathematicalformula set out in Annex II’. Annex II gives thefollowing basic equation expressing theequivalence of loans on the one hand andrepayments and charges on the other: “

[Equation 7]

Meaning of letters and symbols:K Number of a loanK’ Number of a repayment or a payment of

chargesAK Amount of loan number KA’K’ Amount of repayment number K’m Number of the last loanm’ Number of the last repayment or a

payment of charges

tK Interval, expressed in years and fractionsof a year, between the date of loan 1 andthose of subsequent loans 2 to m

tK’ Interval, expressed in years and fractionsof a year, between the date of loan 1 andthose of repayments or payments ofcharges 1 to m’

i Percentage rate that can be calculated(either by algebra, by successiveapproximations, or by a computerprogramme) where the other terms inthe equation are known from thecontract or otherwise.

Remarks:

(a) The amounts paid by both parties atdifferent times shall not necessarily beequal and shall not necessarily be paid atequal intervals.

(b) The starting date shall be that of the firstloan.

(c) Intervals between dates used in thecalculation shall be expressed in years orin fraction of a year. A year is presumedto have 365 days or 365.25 days or(for leap years) 366 days, 52 weeks or12 equal months. An equal monthis presumed to have 30.41666 days (i.e.365/12).

(d) The result of the calculation shall beexpressed with an accuracy of at leastone decimal place.43 When rounding to aparticular decimal place the following rule

40 This is analogous to countries that have to take into account thedisagio. The Regulation requires in paragraph 1 of Annex II thatthe disagio is reflected in MFI interest rates as if it were aninterest rate payment. This is not possible by means of theannualised agreed rate formula as given in Equation 1. NCBs ofMember States that use the annualised agreed rate and havedisagio need to adjust Equation 1 accordingly, which is furtherdiscussed in Chapter 4.2.7.

41 See also paragraphs 9 to 11 of Annex II to the Regulation.42 See footnote 21.43 This is a direct quote from Directive 87/102/EEC and only

included here to illustrate the requirements of the Directive.Paragraph 69 of Annex II to the Regulation specifies that ‘NCBsshall provide the MFI interest rates on outstanding amounts andon new business to the ECB with a detail of four decimal places.This shall be without prejudice to the decision taken by theNCBs on the level of detail at which they wish to collect the data.The published results shall not contain more than two decimalplaces.’ Hence it is up to the NCB to define the level of detail atwhich the reporting agents submit the data to the NCB.

��

′=′

=′

′=

=′+

′=

+

mK

1K

t

K

mK

1K

t

K

KK )i1(

A

)i1(

A


shall apply: if the figure at the decimalplace following this particular decimalplace is greater than or equal to 5, thefigure at this particular decimal place shallbe increased by one.

(e) Member States shall provide that themethods of resolution applicable give aresult equal to that of the examplespresented in Annex III of Directive 98/7/EC (where examples of calculation areprovided).”

The repayment amounts refer to the ‘totalcosts of the credit to the consumer’.44 Theseare ‘all the costs, including interest and othercharges, which the consumer has to pay forthe credit. The following charges areexcluded:(i) charges payable by the borrower for non-

compliance with any of his commitmentslaid down in the credit agreement;

(ii) charges other than the purchase pricewhich, in purchase of goods and services,the consumer is obliged to pay whetherthe transaction is paid in cash or by credit;

(iii) charges for the transfer of funds andcharges for keeping an account intendedto receive payments towards thereimbursement of the credit the paymentof interest and other charges exceptwhere the consumer does not havereasonable freedom of choice in thematter and where such charges areabnormally high; this proportion shall not,however, apply to charges for collectionof such reimbursements or payments,whether made in cash or otherwise;

(iv) membership subscriptions to associationsor groups and arising from agreementsseparate from the credit agreement, eventhough such subscriptions have an effecton the credit terms;

(v) charges for insurance or guarantees:included are, however, those designed toensure payment to the creditor, in theevent of the death, invalidity, illness orunemployment of the consumer, of a sumequal to or less than the total amount ofthe credit together with relevant interest

and other charges which have to beimposed by the creditor as a conditionfor credit being granted.”

The repayment amount comprises an interestrate component and a component made upof other (related) charges. The compositionof the component of other charges variesacross countries because the definitions inthe Consumer Credit Directive are applieddifferently and because national financialsystems and the procedure for securingcredits differ. No positive list of charges tobe taken into account in the calculation ofthe APRC exists. For the purposes of MFIinterest rate statistics, the definition of thecomponent of other charges has not beenfurther harmonised than what is already laiddown in the Consumer Credit Directive. Ifthis Directive is revised, the changes will alsoaffect the APRC collected for MFI interestrate statistics.

Although significant charges might be appliedin all lending categories, the Regulationrequires the compilation of an APRC only forconsumer credit and loans to households forhouse purchases (indicators 30 and 31 inFigure 21). For these two lending categories,EU legislation already obliges or recommendscreditors to provide consumers in the caseof new loans with the APRC:

• Under the Consumer Credit Directive,creditors are obliged to inform consumersin writing about the APRC for all newconsumer loans.45

• The Commission Recommendation of 1March 2001 on pre-contractual informationto be given to consumers by lendersoffering home loans46 gives guidance on

44 Article 1(d) of Directive 87/102/EEC.45 Further information available at http://europa.eu.int/comm/

consumers/policy/developments/cons_cred/index_en.html.46 OJ L 69, 10.3.2001, p. 23.


the information that should be provided toconsumers in the case of domestic andcross-border home loans. A home loan isdefined as ‘a credit to a consumer for thepurchase or transformation of the privateimmovable property he owns or aims toacquire, secured either by a mortgage onimmovable property or by a suretycommonly used in a Member State for thatpurpose’. Consumers should be given a‘European standard information sheet’ thatcovers the APRC. The guidelinesincorporated in the CommissionRecommendation were agreed in the formof a Voluntary Code of Conduct betweenthe mortgage-lending industry andconsumer groups. Member States and alllenders offering home loans in the EU,regardless of whether they are membersof the EU associations and federationswhich negotiated the Code of Conduct,are invited to comply with the CommissionRecommendation by 30 September 2002.If compliance is unsatisfactory, theCommission may consider proposing abinding legal instrument.47

The Consumer Credit Directive and theCommission Recommendation capture thelarge majority of loans to households andrequire information about the APRC. Theexceptions are consumer loans backed by amortgage guarantee and non-secured housingloans. These cases are expected to becovered by the forthcoming revision of theConsumer Credit Directive.

4.6.2 Indicator for other loan charges

In general, MFIs pay interest on deposits totheir customers but also charge fees.Analogously, MFIs usually receive paymentsfor the loans granted from their customerscomprising an interest rate component and acomponent of other related charges. Bothcomponents may have an impact on themonetary transmission mechanism.

The annualised agreed rates and NDERscollected for the purposes of MFI interest

rate statistics provide extensive informationabout the interest paid on deposits andcharged for loans by credit institutions andother institutions. The APRCs collected forconsumer credit and loans to households forhouse purchases incorporate indistinguishablyinformation about the interest rate andrelated fees charged by the same institutions.Hence, the APRC can change from one monthto another as a result of changes in theinterest rate component or the componentof other charges or both.

Since there are no methodological differencesbetween the annualised agreed rate and theNDER on the one hand and the APRC on theother hand, the APRC should not be lowerthan the annualised agreed rate (or NDER)because it includes charges in addition to theinterest rate. The mathematical differencebetween the APRC and the annualised agreedrate (or NDER) therefore represents thecomponent of other charges applied to loans.Indeed, the main aim of collecting the APRCfor the purposes of MFI interest rate statisticsis to construct an indicator of other loancharges applied to consumer credit and loansto households for house purchases, whichallows monitoring the development of loancharges over time.

Significant charges might be applied in alllending categories. However, MFI interest ratestatistics provide data only on charges onconsumer credit and loans to households forhouse purchases. In order to limit thereporting burden on credit institutions andother institutions, the Regulation requires noinformation related to charges for bankoverdrafts, loans to households for otherpurposes and loans to non-financialcorporations. For deposits, charges areassumed to be less significant than for loans,hence no data are collected.

47 Further information available at http://europa.eu.int/comm/internal_market/en/finances/consumer/homeloans-en.pdf.


4.6.3 Charges to be taken into account atnational level

Question:

The transposition of the Consumer CreditDirective into national law may require thecalculation of the APRC with respect to loansto households for house purchases based onassumptions that are not identical with thosein the Directive. In particular, the nationallegislation in one Member State requires theexclusion of mortgage protection insuranceas a relevant charge from the APRC. In thisMember State, mortgage protection insuranceis compulsory for the vast majority of housingloans. However, the credit institution orother institution does not always know theamount of the mortgage protection premiumapplicable to the individual borrower, as theinsurance policy may be arranged through aninsurance intermediary other than the lender.Accordingly, under national law the costs ofsuch insurance are not included in the APRCfor housing loans. Should the national definitionof the APRC be followed in this case for thepurposes of MFI interest rate statistics?

Answer:

Paragraph 11 of Annex II to the Regulationacknowledges that the composition of thefees to be taken into account in the APRCmay differ across countries, because theConsumer Credit Directive is differentlyapplied. Hence, if national legislation on theAPRC provides that mortgage protectioninsurance should not be considered as arelevant charge in the calculation of theAPRC, the reporting agents in that MemberState may exclude such mortgage protectionfrom the calculation of the interest rate.

4.6.4 Period of fixation in the calculationof the APRC

Question:

The transposition of the Consumer CreditDirective into national law may require the

calculation of the APRC with respect to loansto households for house purchases based onassumptions that are not identical with thosein the Directive. In particular, in one MemberState the legislator took the view that undueemphasis should not be placed on the initialinterest rate, which applies only for a shortperiod relative to the full term of the loan.Therefore, the national legislation requiresthe calculation of the APRC based on theassumption that the fixed interest rate at thebeginning of the contract applies only for theperiod specified and that the interest ratesapplicable to other periods of the contractare the current variable rates. Should thenational definition of the APRC be followedin this case for the purposes of MFI interestrate statistics?

Answer:

The initial period of fixation is discussed indetail in Chapter 7.7.4. In general, MFI interestrates on new business only reflect the interestrate that is agreed for the initial period offixation at the start of the contract or afterrenegotiations of the loan. If, after this initialperiod of fixation, the interest rateautomatically changes to a variable interestrate, which might be at a very different level,this is not reflected in new business statisticsbut in the statistics on outstanding amounts.For example, if a 10-year loan is grantedwhere it is agreed at time t0 that for the first12 months the interest is fixed at 10% andthen automatically adjusted to EURIBOR plusx basis points, then the MFI interest rate onnew business captures the 10%. The changesin the interest rate over time, i.e. the changefrom 10% to EURIBOR plus x basis pointsand then the movements of EURIBOR overtime, are only captured in the MFI interestrates on outstanding amounts.

In general, this approach should be followedfor all contracts with initial rate fixation.However, if according to the nationallegislation the calculation of the APRCwarrants a different treatment for certainproducts with initial rate fixation, then thesame treatment should also be applied for


these products to calculate the interest ratewithout charges. The reason for using thesame method of calculation for the interestrate with and without charges is that thedifference between them should serve as anindicator for loan charges as discussed inChapter 4.6.2.

In this example, the national legislationrequires that the APRC for loans tohouseholds for house purchases should bebased on the assumption that the fixed rateapplies for only the period specified and thatthe interest rates applicable to other periodsof the contract are the current variable rates.If the reporting agents follow this approachfor calculating the APRC for the purposes ofMFI interest rate statistics, this is reflected inthe new business indicator 31 in Figure 21referring to loans to households for housepurchases. The difference between indicator31 and the interest rate for loans tohouseholds for house purchases withoutcharges, i.e. the weighted average of newbusiness indicators 16 to 19 in Figure 20,should only be the charges. Hence, theassumption for the initial fixed rate that isapplied to indicator 31 also needs to beapplied for new business indicators 16 to 19referring to loans to households for housepurchases with variable rate and up to oneyear initial rate fixation, over one and up tofive years initial rate fixation, over five and upto ten years initial rate fixation, and over tenyears initial rate fixation. As a consequence,the NDER needs to be compiled for newbusiness indicators 16 to 19 in this MemberState, because the annualised agreed ratecould only take into account the fixed rate atthe beginning of the contract.

The alternative is that the credit institutionsignore the national legislation regarding theAPRC and report for the purposes of MFIinterest rate statistics the fixed rate includingcharges as the APRC and the fixed ratewithout charges as annualised agreed rate.

4.6.5 Treatment of subsidies in the APRC

Question:

The transposition of the Consumer CreditDirective into national law may require thatthe APRC reflects the total costs for a loanfrom the point of view of the customer. Forexample, a credit institution grants a loan at10% p.a., where the customer pays 6% andthe government transfers the remaining 4%as a subsidy directly to the credit institution.In this case, the APRC calculated accordingto national legislation would reflect the 6%p.a. paid by the customer plus any charges.Should the national definition of the APRC befollowed in this case for the purposes of MFIinterest rate statistics?

Answer:

The treatment of subsidies is discussed indetail in Chapter 4.5. In general, MFI interestrate statistics take the point of view of thereporting agent and not of the customer. Thismeans that in the above example, MFI interestrate statistics would reflect what the creditinstitution charges in terms of interest rate,i.e. the 10%, but not the 6% interest that thecustomer pays.

As explained in Chapter 4.6.2, there shouldbe no methodological differences between theAPRC and the annualised agreed rate (orNDER). The APRC should not be lower thanthe annualised agreed rate (or NDER) andthe difference between them solelyattributable to the existence of other chargesapplied to loans. It is therefore essential thatsubsidies are treated in the same way in theannualised agreed rate (or NDER) and theAPRC, as otherwise the indicator of otherloan charges cannot be interpreted.48 As aconsequence, when calculating the APRC forthe purposes of MFI interest rate statistics thepoint of view of the reporting agent needs tobe taken and the national legislation regardingthe treatment of subsidies ignored. Therefore,

48 This is analogous to the argumentation for the treatment of theinitial rate fixation in Chapter 4.6.4.


in the above example, both the annualisedagreed rate (or NDER) and the APRC needto reflect the 10% interest rate that is chargedby the credit institution. If national legislationwas followed for the calculation of the APRC,depending on the size of the related charges,the APRC could be lower than the annualisedagreed rate (or NDER) because the interestrate component would only reflect the 6%interest paid by the consumer.

In theory it would also be imaginable to applyin this Member State, for all types of rates,the treatment of subsidies as specified in thenational legislation, i.e. to take the point ofview of the consumer. However, this wouldnot be in line with paragraph 4 of Annex II tothe Regulation and would moreovercounteract the comparability of MFI interestrate statistics for the euro area.

4.6.6 Treatment of non-profit institutionsserving households in the APRC

As non-profit institutions serving households(NPISHs) are ‘not very important’ customersfor MFIs in terms of deposits placed and loanstaken as compared to households and non-financial corporations, following the ESA 95principle the household sector includesNPISHs for the purposes of MFI interest ratestatistics in general.49 However, for the APRCon loans to households for consumption andfor house purchases (indicators 30 and 31 inFigure 21), NCBs may grant a derogation toreporting agents regarding such loans toNPISHs.50 The reason is that loans to NPISHsmay not be covered by the national legislationreferring to the calculation of the APRC. It isexpected that loans to NPISHs forconsumption or house purchase are negligible.If they were not negligible, however, theexclusion of loans to NPISHs from the APRCfor consumer credit and loans to householdsfor house purchases and their inclusion in theannualised agreed rate or NDER would leadto a distortion of the indicator for other loancharges.51

Question:

If an NCB grants the derogation to itsreporting agents, this implies that all loans toNPISHs are excluded from the calculation ofthe APRC. Therefore, the APRC does notcover the interest rates on new consumercredit and new loans for house purchasesgranted to NPISHs. The question is whether,in the calculation of the APRC, NCBs shouldalso exclude the amount of new loans grantedto NPISHs?

Answer:

In order to calculate the APRC as a weightedaverage rate, the Regulation assumes that itis possible to use the new business volumesthat are provided for the calculation of theannualised agreed rate (or NDER).52

However, these new business volumesincludes the amount of new business inrespect of loans to households and NPISHs,whereas if the derogation is applied, theweighted interest rate to be calculated refersonly to households, i.e. excluding NPISHs.Hence, if the derogation is applied, reportingagents and NCBs require additional data tocalculate the APRC.

As the reporting agents apply the derogation,it can be assumed that they are in a positionto distinguish between households andNPISHs. Therefore, NCBs should askreporting agents to calculate the APRC withweights covering only households, i.e.excluding NPISHs, and to supply the NCBwith additional series for the amount of suchloans to households. The additional series onthe amount of loans to households forconsumption excluding NPISHs, i.e. indicator30, and loans to households for housepurchases excluding NPISHs, i.e. indicator 31,need to be transmitted to the ECB. Theamounts might not be significant but this isnot necessarily known for certain in advance.

49 See also Chapters 3 and 7.4.50 See footnote 1 to paragraph 9 in Annex II to the Regulation.51 See also Chapter 4.6.2.52 Further discussed in Chapters 10.3 and 10.4.


5 Business coverage53

5.1 Interest rates on outstandingamounts54

5.1.1 Definition of outstanding amounts

MFI interest rate statistics on outstandingamounts give information about the interestpaid and received by households and non-financial corporations, which allows theanalysis of any changes in the disposableincome of these sectors and their interestburden. From the point of view of the creditinstitution or other institution, the statisticsalso refer to the interest paid or received,which allows the exploration of changes ininterest rate margins and banks’ profitability.Interest rates on outstanding amounts arefurthermore needed to calculate the own rateof return on M3 and its components. A moreexhaustive list of uses is given in Chapter 2.

Outstanding amounts are defined as the stock ofall deposits placed by customers, i.e. householdsand non-financial corporations, with creditinstitutions or other institutions, and the stockof all loans granted by credit institutions orother institutions to their customers.

An interest rate on outstanding amountsreflects the weighted average55 interest ratelevel applied to the stock of deposits or loansin the relevant instrument category as at thetime reference point:56

• Interest rates on outstanding deposits coverall deposits placed and not yet withdrawnby customers in all the periods up to andincluding the reporting date.

• Interest rates on outstanding loans coverall loans withdrawn and not yet repaid bycustomers in all the periods up to andincluding the reporting date; this excludesbad loans and loans for debt restructuringat rates below market conditions.

MFI interest rates on outstanding amountsare therefore statistics on the interest ratesactually applied to all ‘open’ deposits and loans.

5.1.2 Bad loans and loans for debtrestructuring below marketconditions

Interest rates on bad loans and loans for debtrestructuring below market conditions are notcollected for the purposes of MFI interestrate statistics. As these loans in generalreceive little or no interest payment, theirinclusion would distort the results. Althoughthe interest rates on bad loans and loans fordebt restructuring below market conditionsare excluded from MFI interest rate statistics,their amounts are included in the weightinginformation used for aggregating MFI interestrates. The reason is that the amountsoutstanding are taken from MFI balance sheetstatistics and these cover bad loans and loansfor debt restructuring.

No harmonised definitions for bad loans andloans for debt restructuring apply. To theextent possible, NCBs should use existingnational definitions. It is recognised that thesedefinitions may differ across countries. TheECB does not recommend any (common)definitions. However, the ECB monitors thatthe applied national definitions exclude all badloans and loans for debt restructuring belowmarket conditions from MFI interest rates.Moreover, the ECB monitors that the nationaldefinitions do not exclude loans below marketconditions that are not bad loans or loans fordebt restructuring, as for example favourableinterest rates offered by credit institutions toemployees. As explained in Chapter 4.5.1,favourable rates are included in MFI interestrate statistics.

53 This chapter refers mainly to Part 2 of Annex II to the Regulation.54 See also paragraphs 14 to 16 of Annex II to the Regulation.55 Aggregations are discussed in Chapter 10.56 The time reference point is discussed in Chapter 6.1. Paragraph

15 of Annex II to the Regulation states that interest rates onoutstanding amounts cover all outstanding contracts that havebeen agreed in all the periods up to and including the reportingdate. This might be misleading. For example, if a loan is agreedin March 2003 but the first money only withdrawn in April2003, then the loan appears in the statistics on outstandingamounts for the first time in April 2003. See also the treatmentof a loan in tranches in Chapter 5.3.8.


5.2 Interest rates on overnight deposits,deposits redeemable at notice andbank overdrafts57

5.2.1 The balance at the time referencepoint as indicator for new business

The general concept of interest rates onoutstanding amounts is subject to Chapter5.1. Interest rates on new business areexplained in Chapter 5.3. Three instrumentcategories, i.e. overnight deposits, depositsredeemable at notice and bank overdrafts,58

form a separate group for which the interestrates on outstanding amounts and newbusiness coincide. The MFI interest rates forovernight deposits, deposits redeemable atnotice and bank overdrafts are designed tomeasure new business in each of the threeinstrument categories. They are thereforeincluded as new business indicators 1, 5, 6, 7,12 and 23 in Figure 20. The method forcompiling these indicators is, however, thesame as for interest rates on outstandingamounts. This is reflected by means of Figure19, which presents the new businessindicators 1, 5, 6, 7, 12 and 23 next to theMFI interest rates on outstanding amounts inFigure 18.

Overnight deposits, deposits redeemable atnotice (in particular non-transferable sightsaving deposits59) and bank overdrafts form aseparate group because they experience alarge number of inflows and outflowsthroughout the month. The increases anddecreases in the amount on these accountsarise from receipts and payments related tothe customer’s economic activity, and aretherefore related to transactions rather thanto the autonomous investment decisions ofthe customer. Also, the greater part of thedeposit or bank overdraft is usually turnedover during the period. The balance at thetime reference point60 is considered to be themost appropriate indicator for new business:

• For overnight deposit and depositredeemable deposits, the balance at thetime reference point reflects the amountthe customer has chosen to leave on this

type of deposit instead of placing themoney elsewhere.

• For bank overdrafts, the balance at thetime reference point reflects the amountthe customer has chosen to leave as adebit balances on a current or chequingaccount instead of borrowing the moneyelsewhere.

By leaving a ‘net’ (debit/credit) balance onthe overnight deposit61 or deposit redeemableat notice the customer has implicitly agreedto the terms and conditions of the account,which is a precondition for new business.The customer adjusts this balance as part ofhis or her portfolio management. The balanceat the time reference point is in fact theoutstanding amount at the time referencepoint, which means in other words that theconcept of new business is extended to thewhole stock in the case of overnight deposits,deposits redeemable at notice, and bankoverdrafts.

In addition to these conceptualconsiderations, there are also several practicalreasons for using the balance at the timereference point as an indicator for newbusiness on overnight deposits, depositsredeemable at notice and bank overdrafts. Ifinstead the definition of new business werebased on increases in the amount on existingdeposit or loan accounts, this would lead to aheavy reporting burden on credit institutionsand other institutions due to the large numberof inflows and outflows throughout themonth. Also, the weight based on all inflowsand outflows would overestimate the amountof new business on these accounts.

57 See also paragraphs 17 to 19 of Annex II to the Regulation.58 All defined in Chapter 7.5.59 Non-transferable sight savings deposits, which although legally

redeemable on demand are subject to significant penalties, havefeatures that are very close to overnight deposits and are offeredby credit institutions in some Member States. According toRegulation ECB/2001/13, they are classified as depositsredeemable at up to three months’ notice.

60 The time reference point is further discussed in Chapter 6.1.61 The overnight deposit becomes a bank overdraft in the case of a

debit balance; see also Chapter 7.5.2.


Analogous to the interest rates onoutstanding amounts discussed in Chapter5.1, the MFI interest rates for overnightdeposits, deposits redeemable at notice andbank overdrafts reflect the weighted averageinterest rate level applied to the stock ofdeposits or loans in the relevant instrumentcategory as at the time reference point:

• The interest rates for overnight depositsand deposits redeemable at notice coverall amounts placed and not yet withdrawnby customers in all the periods up to andincluding the reporting date.

• The interest rates on bank overdraftscover all amounts withdrawn and not yetrepaid by customers in all the periods upto and including the reporting date.

5.2.2 Determining the interest rate on anovernight deposit

Figure 10 shows an example of an overnightdeposit yielding 2% p.a. for an amount up toEUR 2,500 and 3% p.a. for any amount

Figure 10Overnight deposit

0

500

1000

1500

2000

2500

3000

3500

4000

4500

<-- 2

% 3

% -

->

Overnight deposit 0 1500 2500 3000 3500 1000 1000 0 2000 4000

Threshold 2500 2500 2500 2500 2500 2500 2500 2500 2500 2500

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9

MIR (OA) = MIR (NB) - 2.0% 2.0% 2.17% 2.29% 2.0% 2.0% - 2.0% 2.38%Weight (OA) =

Weight (NB)

- 1500 2500 3000 3500 1000 1000 - 2000 4000

- yielding 2% - 1500 2500 2500 2500 1000 1000 - 2000 2500

- yielding 3% - 0 0 500 1000 0 0 - 0 1500

exceeding EUR 2,500. The same figure couldalso represent a non-transferable sight savingdeposit or a bank overdraft. In the latter casethe account would not ‘yield’ but ‘cost’interest.

In this manual, MIR (OA) indicates the MFIinterest rate on outstanding amounts and MIR(NB) the MFI interest rate on new business.Weight (OA) and weight (NB) show theweighting information that would be appliedto the interest rates on outstanding amountsand new business respectively in aggregationswith values for other comparable accounts inthis or other reporting agents.62 As a resultof extending the definition of new businessfor overnight deposits to the whole stock,the interest rates and weights for newbusiness and outstanding amount coincide inFigure 10.

62 Aggregations are discussed in Chapter 10.


Each of the periods t0 to t9 represent onemonth. In this example, the MFI interest ratesare calculated as a snapshot of end-monthobservations.63 Hence, the amount andinterest rate observations in Figure 10 areend-month values. For example, at the end ofthe first month t1 an amount of EUR 1,500 isavailable on the overnight deposit yielding 2%interest and, therefore, the MFI interest rateon this account at time t1 is 2.0%. At time t3,EUR 3,000 is available on this deposit, whereEUR 2,500 yields 2% and EUR 500 yields 3%interest. Hence, the MFI interest rate for t3 iscalculated as (0.02 * 2500 + 0.03 * 500) /3000 = 2.17%. At time t9, the MFI interestrate is calculated as (0.02 *2500 + 0.03 *1500) / 4000 = 2.38%. These interest ratesreflect a snapshot of the deposit at the timeof data collection. They may of course differfrom the actual amount of interest earnedduring the month, the latter being covered byimplicit rates.

5.2.3 Combined deposit and loan facilities

Question:

How should the annualised agreed rate orNDER be calculated for financial products onaccounts that, depending on their balance,

Figure 11Combined deposit and loan facility

-3000

-2000

-1000

0

1000

2000

3000

4000

Lo

an

D

ep

osit

Amount 0 2000 3000 1000 -1000 -2000 2000 -300 0 -300

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9

can either be a deposit or a loan, such as forexample for overnight deposits and bankoverdrafts? Reporting agents do not know inadvance whether the account will be a depositor turn into a loan in the coming period.

Answer:

MFI interest rates are determined ex post, i.e.for the month prior to the reporting date. Tocalculate the MFI interest rate for a combineddeposit and loan facility, periods where theaccount was a (positive) overnight deposithave to be distinguished from periods wherethe account was a (negative) bank overdraft.For analytical reasons, it would not beappropriate to compute an average interestrate combining (low) overnight deposit ratesand (high) bank overdraft rates. An examplefor a combined deposit and loan facility isgiven in Figure 11.

For combined deposit and loan facilities, twocompilation methods need to bedistinguished:64

63 Further discussed in Chapter 6.1. An alternative calculationbased on implicit rates is possible but shall not be demonstratedhere.

64 The time reference point and hence the two compilationprocedures are further discussed in Chapter 6.1.


• If the interest rate is compiled as a snapshotof end-month observations, only one balanceduring the month is taken into account todecide whether the account in thereference month is an overnight depositor a bank overdraft. This balance is asnapshot at a certain point in time on thelast day of the month.

• If the interest rate is calculated as an implicitrate referring to the average of the month,for each daily balance the reporting agentneeds to assess whether the account is adeposit or a loan. The reporting agent thencalculates an average of the daily creditbalances and the daily debit balances toderive the average monthly stocks for thedenominator of the implicit rate. Also, forthe flows in the numerator the accruedinterest payable on deposits and receivableon loans need to be distinguished.

If in Figure 11 each of the periods t0 to t9

represents one month, then t1, t2, t3 and t6 arerecorded as overnight deposits and t4, t5, t7 andt9 as bank overdrafts. If each of the periods t0 tot9 represents one day and t9 the reporting day,then a bank overdraft is reported if the methodof end-period observations is chosen, becauseonly the situation at the time of data collectionis relevant.

5.2.4 Regular savings on a depositredeemable at notice65

Question:

A deposit redeemable at notice is subject toregular savings of EUR 200 per month for fiveyears. What should be reflected in thestatistics on new business and outstandingamounts?

Answer:

For deposits redeemable at notice theconcept of new business is extended to thewhole stock. Hence, the credit balance, i.e.the amount outstanding at the time referencepoint, is used as an indicator for new business.In this example, the initial savings of EUR2,000 and all subsequent savings of EUR 200are reflected in the amount outstanding onthe account and hence in MFI interest ratestatistics. An example for the first year isgiven in Figure 12.

Figure 12Deposit redeemable at notice

0

500

1000

1500

2000

2500

3000

Ou

tsta

nd

ing

am

ou

nt

2.0%

2.1%

2.2%

2.3%

2.4%

2.5%

2.6%

2.7%

2.8%

2.9%

3.0%

In

tere

st

ra

te

Deposit 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Interest rate 2.50% 2.50% 2.50% 2.70% 2.70% 2.65% 2.60% 2.55% 2.50% 2.45% 2.45% 2.45% 2.45%

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11 t12

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11 t12

MIR (OA) = MIR (NB)

- 2.5% 2.5% 2.7% 2.7% 2.65% 2.6% 2.55% 2.5% 2.45% 2.45% 2.45% 2.45%

Weight (OA) =

Weight (NB)

- 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

65 See also regular savings on deposits with agreed maturityexplained in Chapter 5.3.4.


5.3 Interest rates on new business66

5.3.1 Definition of new business

MFI interest rate statistics on new businessare statistics on the interest rates laid downin new agreements. They reflect the demandand supply conditions in the deposit and loanmarkets at the time of the agreement,including competition with other types offinancial institution and product. Thesestatistics are needed to analyse the pass-through of changes in official rates and marketinterest rates to lending and deposit interestrates faced by households and non-financialcorporations. They provide information aboutthe cost of capital and the cost spreadbetween self-financing and credit. MFI interestrates on new business allow the study ofprices and quantities together which helpsfor example to explain portfolio shifts. Theyalso show how quickly banks’ interest ratemargins react to external developments.

New business in these instrument categoriesis defined as any new agreement between thecustomer and the credit institution or otherinstitution. New agreements are:

• all financial contracts, terms and conditionsthat specify for the first time the interestrate of the deposit or loan, and

• all new negotiations of existing deposits andloans.

Prolongations of existing deposit and loancontracts that occur automatically, i.e. withoutany active involvement of the customer, anddo not involve any renegotiations of the termsand conditions of the contract, including theinterest rate, are not new business.

An interest rate on new business reflects theweighted average67 interest rate level that hasbeen agreed for all new deposits or loans inthe relevant instrument category during thereference month. Interest rates on newbusiness cover all new agreements during thewhole month68 prior to the reporting date.

MFI interest rate statistics on new businessare distinct from MFI interest rates onoutstanding amounts that, as explained inChapter 5.1, reflect the interest rates actuallyapplied to the stock of deposits and loans. Inthe extreme case, an interest rate laid downin a new agreement may never actually beapplied to any deposit or loan. For example,a customer and a credit institution mightagree on an interest rate for a certain amountof money but the customer might in the endchoose not to place any deposit with thisinstitution or analogously decide not towithdraw any of the money granted as a loan.As a consequence, both the agreed interestrate and amount would be reflected in theMFI interest rate statistics on new businessat the time of agreement, but never appear inthe MFI interest rate statistics on outstandingamounts.

5.3.2 New business on deposits with agreedmaturity

For most deposits with agreed maturity, i.e.classic time deposits where a fixed sum isplaced for a predefined period of time, newbusiness only arises when a new account isopened for the first time, at which point thedeposit amount and the interest rate areagreed. Normally, for deposits with agreedmaturity no further new business occurs untilmaturity. An exception to this rule isdiscussed in Chapter 5.3.4. The treatment ofmaturing deposits is subject to Chapter 5.3.3.

Figure 13 illustrates the calculation of MFIinterest rates on outstanding amounts andnew business in an example involving threeseparate deposits with agreed maturity:

a) at time t1 EUR 2,000 is placed at 4% p.a.with an agreed maturity of two years;

b) at time t2 EUR 2,000 is deposited at 5%p.a. also with an agreed maturity of twoyears; and

66 See also paragraphs 20 to 25 of Annex II to the Regulation.67 Aggregations are discussed in Chapter 10.68 Further discussed in Chapter 6.2.


c) at time t4 EUR 3,000 is paid in at 6% p.a.with an agreed maturity of three years.

In this example, at time t1, an amount of EUR2,000 is placed as a new deposit. At thattime, the MFI interest rate on new businessMIR (NB) and on outstanding amounts MIR(OA) is 4.0%, which is indicated in Figure 13.At time t2, EUR 2,000 is still on the first

deposit yielding 4% and EUR 2,000 is newlypaid into the second deposit yielding 5%.Hence, at time t2 the MFI interest rate onoutstanding amounts comprising bothdeposits is calculated as (0.04 * 2000 + 0.05 *2000) / 4000 = 0.045 = 4.5%, while the MFIinterest rate on new business at time t2 is 5%referring solely to the new second deposit.

Figure 13Deposit with agreed maturity

0

500

1000

1500

2000

2500

3000

3500

Ou

tsta

nd

ing

am

ou

nt

2 years maturity, 4% interest 2000 2000

2 years maturity, 5% interest 2000 2000

3 years maturity, 6% interest 3000 3000 3000

t0 t1 t2 t3 t4 t5 t6

t0 t1 t2 t3 t4 t5 t6

MIR (OA) - 4.0% 4.5% 5.0% 6.0% 6.0% 6.0% Weight (OA) - 2000 4000 2000 3000 3000 3000

MIR (NB) - 4.0% 5.0% - 6.0% - - Weight (NB) - 2000 2000 - 3000 - -

5.3.3 Matured deposit with agreedmaturity

Question:

A customer’s deposit with agreed maturitymatures. Following the standard procedure,the reporting agent informs the customer thatthe matured deposit can be withdrawn withina given period of time. In this example, thecustomer has three options:

Option 1: withdraw the funds,Option 2: inform the credit institution that

he or she wants to make a new

deposit on terms and conditionsidentical to the matured deposit,or

Option 3: do nothing, in which case the creditinstitution will automatically renewthe deposit, on identical terms.

Would options 2 and 3 constitute newbusiness?

Answer:

The task is to distinguish between newbusiness and an (automatic) prolongation ofan existing contract. The key issue is the


active involvement of the customer: an(automatic) prolongation without the activeinvolvement of the customer is not newbusiness, whereas a prolongation with activeinvolvement of the customer is new business.An additional aid for identifying new businessis given by the terms and conditions laid downin the existing contract, which in generaldetermine what happens when the depositwith agreed maturity reaches maturity.

Option 2 is new business because thecustomer is involved in new negotiations withthe credit institution. Whether the newdeposit is on terms and conditions identicalto the old deposit or on new terms andconditions is irrelevant for determiningwhether new business occurs.

Option 3 is not new business because nonew negotiations took place.

In general, the contract for a deposit withagreed maturity defines what happens withthe money after maturity, if it is notwithdrawn. For example, the contract mightdetermine that the deposit is originally placedfor one year at EURIBOR plus 50 basis points.If the funds are not withdrawn at maturity,then they are reinvested at the same terms,i.e. at EURIBOR plus 50 basis points. In thiscase, if the customer says nothing, it is notnew business but an (automatic) prolongationof the contract. If the customer rediscussesthe external index ‘EURIBOR’ or the spread‘plus 50 basis points’, then the customer isactively involved and the reinvestmentconstitutes new business.

It is also possible that the customer on placinga deposit with one-month maturity instructsthe credit institution to rollover the accounteach month provided the terms andconditions remain unchanged from one monthto the next. The customer stipulates that heor she wishes to be notified if there is achange in the terms and conditions attachedto the account. The automatic rolling over isnot new business. However, if the terms andconditions change, the customer is contactedand engages in new negotiations, this is

recorded as new business, independent ofwhether the outcome of the negotiations arethe old or new terms and conditions.

When the terms and conditions of a maturingdeposit with agreed maturity are renegotiatedand as a consequence the deposit is classifiedas new business, then the maturity of that(new) deposit is counted as commencing atthe point of the ‘new business’ classification.

5.3.4 Regular savings on a deposit withagreed maturity69

Question:

A company savings plan is linked to a depositwith agreed maturity. In the contract it isagreed that the company makes regularmonthly deposits of EUR 2,000 for a periodof two years. The first payment is made a fewweeks after the contract date.

a) What is to be considered as new business,the first deposit of EUR 2,000 or allpayments during the lifetime of thecontract?

b) What is to be considered as the startingdate, the contract date (in which case thereis not yet a payment and no amount ofnew business) or the first payment date?

Answer:

Ceteris paribus a credit institution or otherinstitution will offer a different interest rateon a deposit that is:

• progressively increasing over time fromEUR 2,000 in t0 to EUR 48,000 in t24,

• fixed at EUR 2,000 for two years, and

• fixed at EUR 48,000 for two years.

The company’s regular monthly savings in thisexample lead to a progressive increase in the

69 See also regular savings on deposits redeemable at notice inChapter 5.2.4.


outstanding amount on the deposit withagreed maturity. This increase is reflectedmonth after month in the MFI interest ratestatistics on outstanding amounts. In contrast,the MFI interest rate statistics on new businesscapture the deposit only once at the time ofagreement. Hence, the new business statisticscan either mirror the initially placed EUR2,000 or the maximum amount of EUR48,000, but cannot reflect the progressiveincrease of the amount on the deposit. Thetreatment of regular savings on deposits withagreed maturity in new business statisticscannot be defined in general, but depends onthe agreements laid down in the individualcontract:

Case 1: The customer and the creditinstitution agree that EUR 2,000 hasto be placed each month on theaccount until maturity. Thecustomer’s obligation can be enforcedby the credit institution in a way thatit is sure ex ante that at time t24 EUR48,000 will be accumulated on theaccount. In this case, the full amountof EUR 48,000 is reflected as newbusiness in MFI interest rate statisticsat the time of agreement on thecontract. Most likely the level of theinterest is linked to the commitmenton the part of the customer. Forvariable interest rates the value ofthe rate is determined at the time ofthe agreement.70

Case 2: The customer and the creditinstitution agree on a flexible savingsplan, which states that EUR 2,000should be placed each month on thedeposit until maturity. The paymentscan be lower or higher with (orwithout) a targeted maximum savingof EUR 48,000 at time t24. In this case,the amount accumulated on theaccount at maturity is certain only expost. Therefore, the credit institutionreports as new business the amountof EUR 2,000 when it is initiallyplaced. For variable interest rates the

value of the rate is determined at thetime the deposit is placed.71

In the absence of any knowledge to thecontrary, Case 2 should be assumed, i.e. thatthe amount accumulated on the account atmaturity is certain only ex post. Hence thedefault option is that credit institutions andother institutions report as new business theEUR 2,000 when it is initially placed. In thecase of variable interest rates the value of therate is determined at the time the deposit isplaced.

5.3.5 New lending with fixed interest rateand with initial rate fixation

Figure 14 shows the example of a loan that isgranted for ten years. At time t0 it is agreedbetween the customer and the creditinstitution that the interest is fixed at 10%p.a. for the first four years, (up to t3) and thatafter this initial period of fixation a newinterest rate level will be agreed for theremaining maturity of the loan.72 This newinterest rate might be fixed for anotherperiod or variable, but is in any case unknownat time t0. As an example, in Figure 14 theresult of the new negotiations at time t4 is afixed rate of 8% p.a. for the remaining sixyears of the loan. The MFI interest ratestatistics on new business capture at time t0

the interest rate of 10% agreed for the firstfour years and at time t4 the interest ratewhich is the result of the new negotiations.

70 This treatment is analogous to a loan in tranches discussed inChapter 5.3.8, where it is also the agreement that determinesthe amount and the time of recording as new business. However,the difference between the savings plan and the loan in tranchesis that for the savings plan the interest rate is linked to theincrease in the amount and for the loan to the full amountstated in the contract.

71 This treatment is analogous to the savings plans for housingloans described in paragraph 81 of Annex II to the Regulationand the plan d’épargne-logement defined in paragraph 82. Seealso Chapter 9.1.

72 The original maturity of this loan is 10 years with an initialperiod of fixation of four years. See also Chapter 7.7.4.


Figure 1410-year loan renegotiated after four years

Figure 15 provides another example of a 10-year loan. In this example, at time t0 thecustomer and the credit institutions agreethat the interest is fixed at 9% p.a. for thefirst 12 months, and that after this initialperiod of fixation the interest rateautomatically adjusts to EURIBOR plus x basispoints.73 This rate is then applied for the next12 months, after which it will againautomatically adjust to EURIBOR plus x basispoints. Only the interest rate of 9% for thefirst year is considered as new business attime t0. Neither the switch to variable ratesnor the associated automatic adjustments arereflected in the statistics on new business.They are not new agreements but part of theterms and conditions of the loan laid down attime t0. In contrast to the previous example,in this example the customer and the creditinstitution agreed on the variabilitymechanism for the entire maturity of the loan

0

2000

4000

6000

8000

10000

12000

Ou

tsta

nd

ing

am

ou

nt

5%

6%

7%

8%

9%

10%

11%

12%

In

terest

rate

Loan 10000 10000 10000 10000 10000 10000 10000 10000 10000 10000

Interest rate 10% 10% 10% 10% 8% 8% 8% 8% 8% 8%

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9

MIR (OA) 10% 10% 10% 10% 8% 8% 8% 8% 8% 8% Weight (OA) 10000 10000 10000 10000 10000 10000 10000 10000 10000 10000

MIR (NB) 10% - - - 8% - - - - - Weight (NB) 10000 - - - 10000 - - - - -

at t0, i.e. the day of the signature of the loancontract. A change from fixed to variableinterest rates or vice versa during the courseof the contract, which has been agreed at thestart of the contract (time t0), is not a newagreement but part of the terms andconditions of the loan laid down at time t0.These changes in the interest rate over timeare only captured in the MFI interest rate onoutstanding amounts.

73 The original maturity of the loan is ten years with an initialperiod of fixation of one year. This example is different to thecase where the initial period of fixation is very short as comparedto the whole maturity of the loan and the interest rate offeredduring this period is significantly below market conditions. Seealso Chapter 7.7.4.


Figure 1510-year loan with fixed interest rate for the first year

0

2000

4000

6000

8000

10000

12000O

uts

tan

din

g a

mo

un

t

3%

4%

5%

6%

7%

8%

9%

10%

In

terest

rate

Loan 10000 10000 10000 10000 10000 10000 10000 10000 10000 10000

Interest rate 9.00% 8.02% 8.31% 6.59% 5.88% 4.76% 4.36% 5.57% 7.16% 7.75%

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9

T0 t1 t2 t3 t4 t5 t6 t7 t8 t9

MIR (OA) 9.00% 8.02% 8.31% 6.59% 5.88% 4.76% 4.36% 5.57% 7.16% 7.75% Weight (OA) 10000 10000 10000 10000 10000 10000 10000 10000 10000 10000

MIR (NB) 9.00% - - - - - - - - - Weight (NB) 10000 - - - - - - - - -

5.3.6 Top-up loans

Question:

A customer has an outstanding consumercredit of EUR 10,000 at 9% and asks thecredit institution to lend a further EUR 5,000.How should this top-up loan be treated in MFIinterest rate statistics?

Answer:

If an agreement is reached for the incrementalamount, then only this agreement is newbusiness, i.e. the new loan of EUR 5,000. Theadditional loan could be at the same interestrate of 9% as the outstanding loan of EUR10,000, but also at a higher or lower rate.

If the whole loan is renegotiated, i.e. the EUR15,000, then the amount of EUR 15,000 isnew business. The interest rate recorded asnew business is then the one agreed in thenew negotiations between the customer and

the credit institution, which can be 9% as forthe old loan, or higher or lower.

5.3.7 Conversion of a bank overdraft intoanother type of loan

Question:

A customer incurred a bank overdraft of EUR10,000 on which the interest rate is 15% p.a.In agreement with the credit institution, thecustomer transforms this bank overdraft intoa consumer credit on which the interest rateis 10% p.a. Does the conversion representnew business?

Answer:

The conversion of a bank overdraft into aconsumer credit constitutes new business. Itrequires a new agreement between the creditinstitution and the customer. Both partiesare actively involved in new negotiations.


5.3.8 Loan taken out in tranches74

Figure 16Loan in tranches

0

2000

4000

6000

8000

10000

12000

Ou

tsta

nd

ing

am

ou

nt

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

In

terest

rate

Loan 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 10000 10000

Interest rate 8% 8% 8% 8% 8% 8% 8% 8% 8% 8% 8% 8% 8%

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11 t12

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11 t12

MIR (OA) - 8% 8% 8% 8% 8% 8% 8% 8% 8% 8% 8% 8% Weight (OA) - 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 10000 10000

MIR (NB) 8% - - - - - - - - - - - - Weight (NB) 10000 - - - - - - - - - - - -

A household or non-financial corporation isnormally expected to take out a loan – otherthan a bank overdraft – in full at the start ofthe contract (time t0). In some cases,however, it may be agreed that the customertakes out the loan in tranches at times t1, t2,t3, etc. instead of withdrawing the full amountat time t0. Figure 16 shows the first year of a10-year loan in tranches. In this example, aloan of EUR 10,000 is granted at time t0 for10 years fixed at 8% p.a. The customer takesout the first tranche of EUR 1,000 at time t1

and then further tranches of EUR 1,000 ineach of the following nine months.

For MFI interest rate statistics on newbusiness, the fact that the loan is taken out intranches is irrelevant. These statistics capturethe agreement between the customer and thecredit institution at time t0, i.e. the full amountof the loan granted at time t0 and the interestrate agreed in relation to this full amount.New business is therefore the interest rateof 8% for an amount of EUR 10,000 at timet0. For fixed and variable rates, the riskpremium included in the interest rate dependson the amount of the loan and the economicconditions. These facts can only be capturedconsistently by recording the interest rate

and the full amount of the loan at time t0 inMFI interest rates on new business. Asexplained in Chapter 5.3.1, MFI interest rateson new business are statistics on agreementsnot on the actually withdrawn credit.

In contrast, the MFI interest rates onoutstanding amounts are statistics on the actualwithdrawn credit at any point in time. Theyreflect the interest rates applied and theamount of the loan actually outstanding75 atthe time of data collection, i.e. the interestrate is 8% for an amount of EUR 1,000 in t1,8% for an amount of EUR 2,000 in t2, …, andfinally 8% for an amount of EUR 10,000 in t10

until maturity. No interest rate onoutstanding amounts is recorded at time t0 asno money has yet been withdrawn.

The credit institution or other institutiongranting the loan in tranches, for example forfinancing the building of a house, may chargeinterest on the amount granted but not yet

74 Loans in tranches that fall into the instrument categoriesconsumer credit, loans to households for house purchases andloans to non-financial corporations are covered. Bank overdrafts,which follow the treatment explained in Chapter 5.2.1, are notcovered. Umbrella contracts, which are dealt with in Part 6 ofAnnex II to the Regulation, are also excluded.

75 Derived from MFI balance sheet statistics.


withdrawn.76 In this example of an agreed lendingrate of 8% for the loan in tranches, the creditinstitution could ask for an additional 3% on theamount it has agreed to lend but the borrowerhas not yet taken out. The 3% is not covered byMFI interest rate statistics, neither as newbusiness nor as outstanding amounts. The 3% isnot considered part of the lending interest rate:it is a related charge that might be covered inthe APRC according to national conventions,but not in the annualised agreed rate or theNDER.

5.3.9 The definition of new business forvariable interest rates

A rise or fall of a variable interest rate in thesense of an automatic adjustment of theinterest rate performed by the creditinstitution or other institution is not a newagreement and therefore not new business.Such changes or movements in variableinterest rates over time are not captured byMFI interest rates on new business but byMFI interest rates on outstanding amounts.

A change from a fixed to a variable interestrate or vice versa during the course of the

contract is not a new agreement, if thepossibility of such a change is an agreed partof the terms and conditions of the deposit orloan. When the change occurs this is not newbusiness. However, a change from a fixed toa variable interest rate or vice versa is newbusiness, if the possibility of the change wasnot laid down in the initial contract. Thechange is therefore the result of negotiationsbetween the customer and the creditinstitution and constitutes new business.

MFI interest rate statistics on outstandingamounts and new business lead to differentresults not only for deposits and loans withinitially fixed or fully fixed interest rates, butalso for deposits and loans to which fully variableinterest rates apply. Since MFI interest ratestatistics reflect the interest rates that areactually agreed between the customer and thecredit institution or other institution, interestrates on outstanding amounts and new businessonly coincide for that part of the interest ratethat reflects the external index, i.e. in thisexample EURIBOR. The spread over thisexternal index might differ across customersbut also for the same customer over time.

Figure 17Variable interest rates

0%

2%

4%

6%

8%

10%

12%

14%

t0 t1 t2 t3 t4 t5 t6 t7 t8 t9

EURIBOR + x

EURIBOR + y

EURIBOR + z

EURIBOR

76 In Belgium this is referred to as ‘commission de non utilisation’,in Germany as ‘Bereitstellungszinsen’.


For example, in Figure 17 at time t0 a four-year loan at EURIBOR plus x basis points isagreed. The creditworthiness of thecustomer, his or her scope for negotiatingthe interest rate and market conditions varyover time. Hence, at time t2 the samecustomer might only reach agreement on afour-year loan at EURIBOR plus y basis points,but achieve EURIBOR plus z basis points attime t5, with z < x < y.

Contracts might be agreed betweencustomers and reporting agents that containvery flexible terms and conditions includingthe level that the variable interest rate maytake. In some cases, certain conventions areneeded in order to determine the level of thevariable interest rate that should be recordedas new business in MFI interest rate statistics.As not all possibilities can be covered in thismanual, the examples in Chapters 5.3.10 to5.3.13 are intended to provide generalguidance.

5.3.10 Choice of money market index

Question:

A loan of EUR 10 million is agreed, which istaken out in tranches.77 The contract specifiesthat at the time the withdrawals are beingmade the customer may choose which of thefollowing interest rates applies: one-month,three-month, six-month or 12-month LIBORplus 50 basis points. According to paragraph25 of Annex II to the Regulation, this loan intranches is new business at the time thecontract is agreed for the total amount ofEUR 10 million. At that time, however, it isunknown which interest rate the customerwill choose for the first and each of thefollowing withdrawals. Which interest rateshould be reported as new business?

Answer:

For a loan contract that provides thecustomer with a choice of money marketrates as an external index a convention isneeded for determining the value of the new

business rate. By convention, MFI interest ratestatistics on new business record as the valueof the variable interest rate either one-month,three-month, six-month or 12-month LIBORplus 50 basis points, whatever gives the lowestvalue at the time the contract is agreed. Therationale for this convention is that ceterisparibus the lowest rate is what the customerwould most likely have chosen, if he or shewithdrew the whole or a part of the amountat the day of agreement.

For determining the annualised value of thevariable rate it is assumed that the interestrate stays the same for the life of thecontract.78 For example, if at the time ofagreement on the contract 1-month LIBORplus 50 basis points is 4% p.a. and as such thelowest interest rate, then the annualisedagreed rate or NDER is determined based onthe 4% and taking into account the frequencyof interest rate payments.

The interest rates on outstanding amountsreflect the actual interest rate the customerchooses at each withdrawal, the developmentof the variable rate over time and also theactual amount of the loan which has beenwithdrawn.

5.3.11 Money market index with floor andceiling

Question:

A customer agrees to place a deposit with acredit institution that has an agreed maturityof one year receiving interest of 12-monthLIBOR plus 40 basis points with a floor of 2%and a ceiling of 6%. The interest rate floorand ceiling are not derivative contracts, whichmay or may not be exercised by the customeror the credit institution as described inparagraph 78 of Annex II to the Regulation.Instead, the floor and ceiling constitute acorridor for the possible interest rates. What

77 See also Chapter 5.3.8.78 This is in line with Chapter 4.4.


interest rate should be reported as newbusiness?

Answer:

The treatment is in principle the same as forany other variable interest rate but takinginto account the floor and ceiling for thepossible value. Hence, new business is the valueof 12-month LIBOR plus 40 basis points atthe time of agreement on the contract takinginto account that the interest rate cannot fallbelow 2% and cannot exceed 6%. Forexample, if at the time of agreement 12-month LIBOR is 1.5%, then 2% is recorded asnew business, because 1.5 plus 0.4 is belowthe agreed floor. If at the time of agreement12-month LIBOR is 4.5%, then 4.5 plus 0.4 =4.9% is the new business rate. If at the timeof agreement 12-month LIBOR is 5.9%, then6% is captured, because 5.9 plus 0.4 is abovethe agreed ceiling.

The annualised agreed rate or NDER onoutstanding amounts covers the interest rateapplied by the reporting agent at the time ofthe calculation of MFI interest rates, i.e. it isbased on the value of 12-month LIBOR at thetime of data collection. The floor and ceilingare taken into account in the same way as fornew business.

5.3.12 Change in the value of a currency asexternal index

Question:

A customer agrees with a credit institutionon a deposit with agreed maturity of oneyear where the interest rate is linked to thepercentage change in the value of a currencywith a floor of 2% and a ceiling of 6%, i.e. theinterest rate is always positive. Thepercentage change in the value of thecurrency is not known at the time ofagreement t0, but only at the time of maturity.What interest rate should be reported asnew business?

Answer:

In contrast to the example in Chapter 5.3.11,where a value can be given for 12-monthLIBOR at the time of the agreement on thecontract, the percentage change in the priceof a currency is only known ex post at thetime of maturity. Predictions for the value ofthe currency at maturity could be made todetermine a value for the (potential)percentage change of the currency at the timeof agreement. However, this is difficult andnot consistent with the method ofdetermining the value of a variable rate thatis linked to a money market or bond marketindex, where no predicted future value, butrather the current value of the index at thetime of agreement, is used. Hence, the onlyinterest rate that can be included in thestatistics on new business is the agreed floor,i.e. 2%, as this is known with certainty by thecustomer and the reporting agent at time t0.

79

The annualised agreed rate or NDER onoutstanding amounts covers the interest rateapplied by the reporting agent at the time ofthe calculation of MFI interest rates. Hence,it is based on the percentage change in theprice of the currency at the time of datacollection. The agreed floor and ceiling aretaken into account in the same way as inChapter 5.3.11.

5.3.13 Timing differences

Question:

A customer and a credit institution agree ona housing loan with a variable interest rate.The interest rate may be explicitly referencedto an external index, e.g. EURIBOR, or it maybe subject to change at the discretion of thelender. As the contractual rate is variable,the value of this rate may vary (within the

79 If no floor is agreed and no minimum return at all guaranteed,then a convention is needed for determining the level of theinterest rate on new business. In this case, by convention theannualised agreed rate or NDER on new business captures aninterest rate of 0% for the deposit. See also Chapters 9.6 and9.7.


terms of the contract) at any point from thereceipt of the letter of offer through toacceptance or withdrawal of the loan andthen during the life of the contract. Forcontracts with variable rates that have beenagreed, but where funds have not yet beenwithdrawn, some institutions record in theirIT system the prevailing rate for that producttype rather than the interest rate for thespecific contract. It is the view of theseinstitutions that reporting the interest ratefor the product type is both more accurateand more efficient as this obviates the needfor an additional significant data captureexercise. Furthermore, reporting the interestrate for the product type will allowinstitutions to provide data on the reportingdata for contracts that have been signed bythe borrower but for which the institutionhas not yet received the completed forms,which are in transit via post etc.

Answer:

The prevailing rate for the product type andthe rate quoted in the individual contractmight differ because the customers havedifferent creditworthiness. MFI interest ratestatistics on new business cover the actualrate in the individually agreed contract. Thevalue of the variable rate is determined basedon the value of the external index at the timeof agreement on the contract. If the customersigns first and then the credit institution,agreement is reached when the creditinstitution signs. If the credit institution signsfirst and then the customer, which is unusual,then the contract is accepted when thecustomer signs. There might be a transmissiontime for the contract back to the creditinstitution and the new business mighttherefore be recorded with a delay of somedays. As MFI interest rate statistics on newbusiness capture the average interest rateover the month, this delay will only beapparent at the end-month, in the sense thata new business on 30th March might only berecorded on 1st April because of thetransmission delay. This is accepted for MFIinterest rate statistics.

5.3.14 ‘Cooling off’ period

Question:

Laws for consumer protection might requirea “cooling off” period for consumer credit,thus necessitating for example a period of atleast 10 days between the conclusion of theagreement and the withdrawal of the funds.In practice, greater delays can occur. What isthe new business?

Answer:

The ‘cooling off’ period and hence thepossibility that a customer steps back fromthe contract has no influence on MFI interestrate statistics on new business. MFI interestrates on new business reflect the conditions,i.e. the interest rate and the amount of theloan, as laid down in the contract at the timeof agreement on the contract. The actual dayof the withdrawal of the funds is irrelevantfor interest rates on new business but shownonly in the statistics on outstandingamounts.80

5.3.15 Loan offer and preliminary offer

Question:

Before looking for a new house, a customermay have already acquired a loan offer from acredit institution or other institution. In theloan offer, the credit institution promises togrant a loan to the customer at specified,agreed terms in the event that the customerfinds a house. The advantage for the customeris that he or she can make a bid for a housewithout having to consult the creditinstitution first. The final loan contract issigned once the customer has found thehouse. In some Member States, lenders mustissue preliminary offers before granting loans.

80 This treatment is analogous to the loan in tranches, where it isalso the agreement that determines the amount and the time ofrecording as new business. The withdrawal of the first and allfollowing tranches is reflected only in the statistics on outstandingamounts. See also Chapter 5.3.8.


The preliminary offer specifies the terms andconditions of the loan, which cannot bechanged by the lender if the customer finallyagrees them. The offer stands for a givenperiod of time, the length of which isestablished by national law. During this time,the customer will generally look forcompeting offers from other lenders. Do loanoffers constitute new business?

Answer:

A loan offer is not new business. People mayattempt to buy houses that are worth lessthan the credit institution is willing to accept.In this case, a credit institution may eitherwithdraw the offer or lower the amount it iswilling to lend. New business therefore onlyarises when on the basis of the loan offer thefinal loan contract has formally been signed.Only when the customer signs the legallybinding contract does it constitute newbusiness.

5.3.16 Loans for debt restructuring in thecontext of new business

Question:

Bad loans and loans for debt restructuring atrates below market conditions are notincluded in the weighted average interest rateon outstanding amounts.81 Do new loans fordebt restructuring constitute new business?

Answer:

A loan for debt restructuring at rates belowmarket conditions could be seen as arenegotiation of an existing loan contract oreven as a new loan contract. Indeed,throughout the process of loan restructuringboth parties, the credit institution or otherinstitution and the household or non-financialcorporation, are actively involved in(re)negotiating the terms and conditions of

the contract. Nevertheless, loans for debtrestructuring at rates below marketconditions are not considered as newbusiness. The reason is that the interest rateagreed for a loan for debt restructuring isnot the result of the general demand andsupply conditions in the loan market at thetime of the agreement but rather what theindebted customer is able to pay. Hence,interest rates on loans for debt restructuringat rates below market conditions are, likeother bad loans, not the type of interest ratethat is supposed to be covered as an agreedrate by MFI interest rate statistics.82

5.3.17 Moratorium on a loan

Question:

If a customer temporarily stops therepayments on a loan and starts againsometime later, does the restart constitutenew business? Does the treatment dependon whether the bank had started to classifythis loan as a bad debt? Does it depend onwhether the customer signs a new agreementwith the bank?

Answer:

New business occurs only if the customersigns a new agreement. The restart of loanrepayments after a moratorium is per se notnew business. If the loan is considered as abad loan during the moratorium, it is takenout of the interest rates on outstandingamounts, but is reincluded when the customerrestarts to pay the interest.83

81 See also Chapter 5.1.2.82 See also Chapter 4.2.1.83 See also Chapter 5.1.2.


6 Time reference point84

6.1 Time reference point for interestrates on outstanding amounts,overnight deposits, depositsredeemable at notice and bankoverdrafts85

MFI interest rates on outstanding amounts,86

i.e. indicators 1 to 14 in Figure 18, may becompiled as a snapshot of end-monthobservations or as implicit rates referring tothe average over the month:

• As a snapshot of end-month observations,they are calculated as weighted averages87

of the interest rates applied to the stockof deposits and loans at a certain point intime on the last day of the month. At thatpoint in time, the reporting agents collectthe interest rates and the amounts involvedfor all outstanding contracts relevant forMFI interest rate statistics. As end-monthobservations, the MFI interest rates onoutstanding amounts have the same timereference point as MFI balance sheetstatistics. Hence, the amounts needed forweighting the interest rates may be takenfrom MFI balance sheet statistics. By meansof the collected interest rates and amounts,the reporting agents then compile aweighted average interest rate for eachinstrument category.

• As implicit rates referring to the average ofthe month, interest rates on outstandingamounts are calculated as quotients. Thenumerator is the accumulated flow ofinterest during the reference month, i.e.the accrued interest payable on depositsand receivable on loans, and thedenominator is the average monthly stock.At the end of the reference month, foreach instrument category the reportingagent needs to report the accrued interestpayable or receivable during the monthand the stock of deposits and loans onaverage during the same month. The NCBthen calculates from these data the monthlyimplicit interest rate per instrumentcategory. As averages of the month, these

implicit rates on outstanding amounts havethe same time reference point as the MFIinterest rates on new business, which issubject to Chapter 6.2.

As explained in Chapter 5.2.1, overnightdeposits, deposits redeemable at notice andbank overdrafts experience a large number ofinflows and outflows throughout the month.Therefore, the balance at the time referencepoint is taken as an indicator for the amountthe customer has chosen to leave on theovernight deposit or the deposit redeemableat notice, or has chosen to take out as a bankoverdraft instead of placing or borrowing themoney elsewhere. Hence, for indicators 1, 5,6, 7, 12 and 23 in Figure 19 and Figure 20 thecompilation procedure and the time referencepoint is exactly the same as for the interestrates on outstanding amounts. The choicebetween a snapshot of end-monthobservations and an implicit rate referring tothe average of the month has the followingimplications:

• In a snapshot of end-month observations, onlyone balance, i.e. the balance at a certainpoint in time on the last day of the month,is taken into account when computing theMFI interest rate. In the case of combineddeposit and loan facilities88 it is sufficientto look at this balance to decide whetherthe account in the reference month is anovernight deposit or a bank overdraft.

• The denominator in an implicit rate referringto the average of the month is based on theaverage of the daily balances on theovernight deposit, deposit redeemable atnotice or bank overdraft. Therefore, in thecase of combined deposit and loan facilitieseach day the reporting agent needs toassess whether the account is a deposit ora loan. The reporting agent then calculates

84 This chapter refers mainly to Part 3 of Annex II to the Regulation.85 See also paragraphs 26 to 31 of Annex II to the Regulation.86 Outstanding amounts are defined in Chapter 5.1.87 See also Chapter 10.88 Further discussed in Chapter 5.2.3.


an average of the daily credit balances andthe daily debit balances to derive theaverage monthly stocks for thedenominator of the implicit rate. Also, forthe flows in the numerator of the implicitrate accrued interest payable on depositsand receivable on loans need to bedistinguished. Reporting agents shall notreport weighted average interest ratescombining (low) overnight deposit ratesand (high) bank overdraft rates.

It is up to each NCB to decide which of thetwo methods, i.e. a snapshot of end-periodobservations or implicit rates referring toperiod averages, is more appropriate in thenational context. The ECB stronglyrecommends not mixing both approaches atnational level, as this makes the interpretationof the national data more difficult. However,if there are very good reasons to mix bothapproaches because the type of institutionsand their offered products require a differenttreatment, then this is accepted.

Basing the interest rates and the weights onlyon the last day of the month, as in the case ofa snapshot, is less accurate than referring tothe average stock during the month, asrequired for implicit rates. For example, day-of-the-month effects might distort the end-month results. However, these distortionsare expected to be small as a result of thenumber of accounts covered. Indeed, MFIinterest rates on outstanding amounts coverall deposits placed and not yet withdrawn orall loans withdrawn and not yet repaid bycustomers in all the periods up to andincluding the reporting date. A snapshot ofend-month observations only covers thosecontracts that are still outstanding at the timeof data collection. In contrast, an implicit ratereferring to the average of the month alsoincludes contracts that were outstanding atsome time during the month, but are notoutstanding at the end of the month.

The main difference between the twomethods is, however, the way the data iscollected. In the case of a snapshot, interestrates are directly collected, whereas in the case

of implicit rates, the interest rates are derivedas a quotient. Assuming that the profit andloss accounts provide enough detail, thenumerator in the quotient derived from theseaccounts should be of sufficiently high quality.The quality of the resulting implicit interestrates therefore depends on the denominator.Minimum standards for compiling the averagemonthly stock per instrument category arelaid down in the Regulation. The ideal is theaverage of daily stocks over the month, as itensures a close link between the flow in thenumerator and the reference stock in thedenominator. If instead the denominatorrefers to the end-month stock, then the flowin the numerator may include payments forcontracts that have already expired at theend of the month, resulting in inaccurateimplicit rates. Nevertheless, the ECB acceptsapproximations for the average of daily stocksif they fulfil the following minimum standards:

• For volatile instrument categories, i.e. atleast overnight deposits, depositsredeemable at notice, and bank overdrafts,the average monthly stock needs to bederived from daily balances.

• For all other instrument categories, theaverage monthly stock needs to be derivedfrom weekly or more frequent balances.

• For a transitional period of not more thantwo years from the entry into force of theRegulation, for loans with an agreedmaturity of over five years, the ECBaccepts end-month observations. However,at the latest from January 2004 and alsofor loans with an agreed maturity of overfive years, the average monthly stock mustbe derived from weekly or more frequentbalances.

6.2 Time reference point for interestrates on new business89

As defined in Chapter 5.3.1, MFI interest rateson new business reflect the average interest

89 See also paragraphs 32 and 33 of Annex II to the Regulation.


rate level applied to deposits and loans innew agreements that have been arrangedbetween customers and credit institutionsduring the reference month. The timereference point for new business rates istherefore the average of the month. Thisholds true for all indicators in Figure 20except 1, 5, 6, 7, 12 and 23, which refer toovernight deposits, deposits redeemable atnotice and bank overdrafts.

For each instrument category, reportingagents calculate the new business interest rateas a weighted average of all interest rates on

new business operations in the instrumentcategory during the reference month. Theythen transmit these new business interestrates referring to the average of the monthto the NCB of the participating Member Statein which they are resident. The transmissionincludes weighting information on the amountof new business conducted during thereporting month for each instrumentcategory. Reporting agents need to take intoaccount the new business operationsconducted during the entire month, ratherthan just a selected period during the month.

7 Instrument categories90

7.1 Summary tables of indicators

MFI interest rate statistics provide detailedmonthly information on 45 indicators, of whichtwo-thirds refer to new business and one-third to outstanding amounts. Covered areall essential instrument categories for euro-denominated deposits and loans faced byhouseholds and by non-financial corporations.The 45 indicators are in principle required atboth euro area level and national level, butnational deviations in the number of

indicators exist, which is further discussed inChapter 7.2.

Figure 18 shows the 14 instrument categoriesfor which MFI interest rates on outstandingamounts are collected at euro area level.Depending on the choice of NCBs, theinterest rates are either annualised agreedrates or NDERs and compiled either as asnapshot of end-month observations or asimplicit rates referring to the average of themonth.

Figure 18Indicators 1 to 14 on outstanding amounts91

Sector Type of instrument Original maturity Outstanding amount

indicator number:

Up to 2 years 1 To households With agreed maturity

Over 2 years 2

Up to 2 years 3 To non-financial

corporations

With agreed maturity

Over 2 years 4

Deposits

in EUR

Repos 5

Up to 1 year 6

Over 1 and up to 5 years 7

For house purchases

Over 5 years 8

Up to 1 year 9


To households

Consumer credit and

other loans

Over 5 years 11

Up to 1 year 12


Loans in

EUR

To non-financial

corporations

Over 5 years 14

90 This chapter refers mainly to Part 4, Appendix 1 and Appendix 2of Annex II to the Regulation.

91 In the following table ‘up to’ means ‘up to and including’ and‘households’ include NPISHs.


Figure 19 covers the six MFI interest ratesreferring to overnight deposits, depositsredeemable at notice and bank overdrafts.Depending on the choice of NCBs, they areeither annualised agreed rates or NDERs.Analogous to the MFI interest rates onoutstanding amounts in Figure 18, the sixindicators are compiled either as a snapshotof end-month observations or as implicit rates

referring to the average of the month. Sincethe MFI interest rates on overnight deposits,deposits redeemable at notice and bankoverdrafts are designed to measure newbusiness, they follow the numbering of thenew business indicators in Figure 20. Thebrackets around indicator numbers 1, 5, 6, 712 and 23 in the two following tables illustratetheir special status.

Figure 19Indicators 1, 5, 6, 7, 12 and 23 on new business92

Sector Type of instrument Period of notice New business

indicator number:

Overnight (1)

Up to 3 months notice (5)

To households

Redeemable at notice

Over 3 months notice (6)

Deposits

in EUR

To non-financial corporations Overnight (7)

To households Bank overdraft (12) Loans in

EUR To non-financial corporations Bank overdraft (23)

Figure 20 and Figure 21 list the 31 MFIinterest rates on new business that arecollected at euro area level. Indicators 1 to29 in Figure 20 are, depending on the NCBs’choice, either annualised agreed rates orNDERs and as such reflect only the interestrate paid or charged by credit institutionsand other institutions excluding any otherrelated fees. Indicators 30 and 31 in Figure 21are APRCs and therefore comprise an interestrate component and a component of other

charges. All new business rates other thanindicators 1, 5, 6, 7 12 and 23 are compiledas weighted average interest rates referringto the whole month.

92 In the following table ‘up to’ means ‘up to and including’ and‘households’ include NPISHs. For indicators 5 and 6, householdsand non-financial corporations are merged and allocated to thehousehold sector, since it owns about 98% of the outstandingamount of deposits redeemable at notice in all participatingMember States combined.


The total of 45 indicators in Figure 18, Figure20 and Figure 21 provide a balance betweenthe policy and analytical needs of the usersand the reporting burden on creditinstitutions and other institutions.

7.2 General provisions95

MFI interest rate statistics for the euro arearefer to instrument categories rather than toindividual products as is the case with thenational statistics on interest rates in many

Figure 20Indicators 1 to 29 on new business93

Sector Type of instrument Original maturity, period of notice, initial rate

fixation

New

business

indicator

number:

Overnight (1)

Up to 1 year maturity 2

Over 1 and up to 2 years maturity 3


Over 2 years maturity 4

Up to 3 months notice (5)

To

households

Redeemable at notice

Over 3 months notice (6)

Overnight (7)

Up to 1 year maturity 8

Over 1 and up to 2 years maturity 9

To non-

financial

corporations


Over 2 years maturity 10

Deposits

in EUR

Repos 11

Bank overdraft (12)

Variable rate and up to 1 year initial rate fixation 13

Over 1 and up to 5 years initial rate fixation 14

For consumption

Over 5 years initial rate fixation 15




For house purchases




To

households

For other purposes


Bank overdraft (23)



Other loans up to an

amount of EUR 1

million Over 5 years initial rate fixation 26



Loans in

EUR

To non-

financial

corporations

Other loans over an

amount of EUR 1

million Over 5 years initial rate fixation 29

Figure 21Indicators 30 and 31 referring to the APRC

Sector Type of instrument New business indicator number:

For consumption 30 Loans in EUR To households94

For house purchases 31

countries. For example, MFI interest ratesrefer to consumer credit with ‘up to x years’maturity or interest rate fixation, whereas inthe Unites States interest rates are collected

93 In the following table ‘up to’ means ‘up to and including’ and‘households’ include NPISHs. Furthermore, ‘variable rate’ is meantas a synonym for ‘floating rate’ the latter being the expressionused in the Regulation. For indicators 5 and 6, households andnon-financial corporations are merged and allocated to thehousehold sector, since it owns about 98% of the outstandingamount of deposits redeemable at notice in all participatingMember States combined.

94 In general including NPISHs, but NCBs may grant derogations inthis respect.

95 See also paragraphs 34 to 37 of Annex II to the Regulation.


for a 48-month new car loan as a typicalretail product. It is recognised that interestrates referring to typical retail products mightbe easier to interpret than interest ratesreferring to instrument categories and mightprovide a better comparison with capitalmarket interest rates. In the euro area,however, large national differences in thebanking business of credit institutions andother institutions exist. It is therefore notpossible to define a sufficient number oftypical retail products that are representative,or even available, in each Member State.

Also as a result of the national differences inproducts, deviations exist in the number ofindicators compiled per Member State. Insome Member States, resident creditinstitutions and other institutions do not offerdeposits or loans in some of the instrumentcategories listed in Figure 18 and Figure 20 tohouseholds and non-financial corporationsresident in the euro area. Where inapplicable,the NCBs in these Member States ignore theinstrument categories in the data collectionfrom reporting agents and collect less96 thanthe required 45 indicators at national level.An instrument category is inapplicable atnational level only if resident creditinstitutions and other institutions do not atall offer products belonging to this categoryto households and non-financial corporationsresident in the euro area. Data has to beprovided if some business exists, howeverlimited it is.97

NCBs need to cover each instrument categorylisted in Figure 18 and Figure 20 that exists inthe banking business of resident creditinstitutions and other institutions with euroarea households and non-financialcorporations, but not each product offered atnational level. Covering all products couldlead to a near-census situation in somecountries with a wide variety of products.However, NCBs cannot exclude a wholeinstrument category on the grounds that theamounts involved are very small. So if aninstrument category is only offered by oneinstitution, then this institution needs to be areporting institution. Furthermore, once a

credit institution or other institution isselected as a reporting agent, it has to coverfor each instrument category all interest ratesapplied to all the products that fit thiscategory. This principle implies that it is notup to NCBs to define a set of nationalproducts within each instrument category onwhich reporting agents collect data.

The exception to the rule of covering foreach instrument category all interest ratesapplied to all the products are interest rateson bad loans and loans for debt restructuring.As already explained in Chapter 5.1.2, badloans and loans for debt restructuring at ratesbelow market conditions are not covered byMFI interest rate statistics. As these loansreceive little or no interest payment, includingthem would distort the results for MFIinterest rate statistics.

If a new product belonging to an existinginstrument category is created, the reportinginstitutions cover it with the next reporting,as all reporting agents have to report on alltheir products. If an instrument category didnot exist in a Member State at the time thesample was first drawn, but one institutionsubsequently creates a new product belongingto this category, this institution needs to beselected into the sample at the time of thenext representativity check.98 The reason isthat with the creation of this new productthe instrument category is no longerinapplicable at national level.

96 Some NCBs collect more than 45 indicators at national level,because the statistical reporting requirements of the ECB arepart of a broader statistical reporting framework which the NCBestablishes under its own responsibility.

97 Council Regulation (EC) No 2533/98 concerning the collectionof statistical information by the European Central Bank specifiesin Article 8 that the ECB and NCBs may collect confidentialstatistical information for the tasks of the ESCB. The ECB andNCBs have to take all regulatory, administrative, technical andorganisational measures to protect confidential statisticalinformation, i.e. it should not be possible to identify reportingagents or any other legal person, natural person, entity orbranch, either directly from their name or address or from anofficially allocated identification code, or indirectly throughdeduction.

98 Further discussed in Chapters 11.8 and 11.9.


7.3 Foreign currency deposits andloans99

MFI interest rate statistics cover the interestrates applied by resident credit institutionsor other institutions to euro-denominateddeposits and loans vis-à-vis households andnon-financial corporations resident in theeuro area. Data on deposits and loans in othercurrencies is not required at euro area levelby the Regulation. However, in line with the4th recital and Article 3(2) of the Regulation,NCBs may collect data on foreign currencydeposits and loans at national level.

Dual currency loans exist where the currencyof the interest payments differs from thecurrency in which the loan is granted. Thetreatment of dual currency loans follows twoprinciples:

a) either both the amount lent and theinterest rate are covered by MFI interestrate statistics or both are excluded; and

b) one main use of MFI interest rate statisticsis to monitor the transmission of the ECB’smonetary policy.

Hence, if the lending rate agreed between thecustomer and the credit institution or otherinstitution is based on considerations relatingto the monetary policy as carried out by theESCB, then the interest rate and the amountof the loan should be covered by MFI interestrate statistics. If the interest rate agreed forthe loan is based on considerations relatingto the monetary policy of foreign centralbanks, then the interest rate and the amounton lent are beyond the scope of MFI interestrate statistics.

7.4 Breakdown by sector100

MFI interest rate statistics distinguish betweeninterest rates applied in the banking businessvis-à-vis households (including NPISHs) andvis-à-vis non-financial corporations. The ESA95 provides the standard for this sectoralclassification:

I Non-financial corporations are bodiesrecognised as independent legal entitieswhich are market producers and whoseprincipal activity is the production of goodsand non-financial services (ESA 95,paragraph 2.23). The institutional unitscovered are the following:(i) private and public corporations which

are market producers principallyengaged in the production of goods andnon-financial services;

(ii) cooperatives and partnershipsrecognised as independent legal entitieswhich are market producers principallyengaged in the production of goods andnon-financial services;

(iii)public producers which by virtue ofspecial legislation are recognised asindependent legal entities and which aremarket producers principally engagedin the production of goods and non-financial services;

(iv)non-profit institutions or associationsserving non-financial corporations,which are recognised as independentlegal entities and which are marketproducers principally engaged in theproduction of goods and non-financialservices;

(v) holding corporations controlling agroup of corporations which are marketproducers, if the preponderant type ofactivity of the group of corporations asa whole – measured on the basis ofvalue added – is the production ofgoods and non-financial services;

(vi)private and public quasi-corporationswhich are market producers principallyengaged in the production of goods andnon-financial services.

II Households are individuals or groups ofindividuals as consumers, and producers ofgoods and non-financial services exclusivelyfor their own final consumption, and asproducers of market goods and non-financial and financial services provided that

99 See also paragraph 38 of Annex II to the Regulation.100 See also paragraphs 39 to 41 of Annex II to the Regulation.


their activities are not those of quasi-corporations. It includes (ESA 95,paragraph 2.76):(i) individuals or groups of individuals

whose principal function isconsumption;

(ii) persons living permanently in institutionswho have little or no autonomy of actionor decision in economic matters (e.g.members of religious orders living inmonasteries, long-term patients inhospitals, prisoners serving longsentences, old persons living permanentlyin retirement homes);

(iii)individuals or groups of individualswhose principal function is consumptionand that produce goods and non-financial services for exclusively ownfinal use (services of owner-occupieddwellings and domestic servicesproduced by paid employees);

(iv)sole proprietorships and partnershipswithout independent legal status –other than those treated as quasi-corporations – which are marketproducers;

(v) non-profit institutions serving households(ESA 95, paragraph 2.87).

In particular, small businesses in the sense ofsole proprietorships and partnerships areallocated to the household sector or the non-financial corporation sector according to thefollowing principles (ESA 95, paragraph 2.99).Sole proprietorships and partnerships whichare market producers but not recognised asindependent legal entities:(i) are classified as households (see II above)

unless they are quasi-corporations.(ii) if they are quasi-corporations, i.e. if they

keep a complete set of accounts, have noindependent legal status, but an economicand financial behaviour that is differentfrom that of their owners and similar tothat of corporations (ESA 95, paragraph2.13 f):

(1) they are classified as non-financialcorporations (see I above), if they areprincipally engaged in the productionof goods and non-financial services;

(2) they are classified as financialcorporations and therefore notcovered by MFI interest rate statistics,if they are principally engaged infinancial intermediation and financialauxiliary activities.

Indicator 5 in Figure 18 and indicator 11 inFigure 20 refer to repos. At euro area levelno sectoral breakdown for repos is required,because their remuneration is often, althoughnot in all Member States, independent of theholding sector. Since currently at euro arealevel about 40% of repos placed by non-financial sectors are held by households and60% by non-financial corporations, the twosectors are merged for the purposes of MFIinterest rates on repos. Indicator 5 in Figure18 and indicator 11 in Figure 20 thereforeindistinguishably refer to both sectors insteadof being allocated to either households ornon-financial corporations. In line with the 4th

recital and Article 3(2) of the Regulation,NCBs may ask for a sectoral breakdown forrepos at national level.

Indicators 5 and 6 in Figure 20 refer todeposits redeemable at notice held byhouseholds. In the euro area, depositsredeemable at notice are overwhelminglyowned by households, i.e. currently about98% of all deposits redeemable at notice heldwith MFIs. Non-financial corporations holdabout 2% of these deposits. Since it was foundto be more cost-efficient to includehouseholds and non-financial corporationsindistinguishably in one sector, MFI interestrates are collected for deposits redeemableat notice vis-à-vis both sectors and thenentirely allocated to deposits redeemable atnotice held by households.


7.5 Breakdown by type of instrument101

7.5.1 Types of deposits

The instrument breakdown by type of depositfor MFI interest rate statistics and thedefinitions used follow MFI balance sheetstatistics and hence Regulation ECB/2001/13.It defines overnight deposits as deposits whichare ‘convertible into currency and/or whichare transferable on demand by cheque,banker’s order, debit entry or similar means,without significant delay, restriction orpenalty. Balances representing prepaidamounts in the context of electronic money,either in the form of ‘hardware based’ e-money (e.g. prepaid cards) or ‘software based’e-money, issued by MFIs, are included underthis item. This item excludes non-transferabledeposits which are technically withdrawableon demand but which are subject to significantpenalties:

• balances (interest-bearing or not) whichare transferable by cheque, banker’s order,debit entry or the like, without anysignificant penalty or restriction;

• balances (interest-bearing or not) whichare immediately convertible into currencyon demand or by close of business on theday following that on which the depositwas made, without any significant penaltyor restriction, but which are nottransferable;

• balances (interest-bearing or not)representing prepaid amounts in thecontext of ‘hardware-based’ or ‘software-based’ e-money (e.g. prepaid cards);

• loans to be repaid by close of business onthe day following that on which the loanwas granted.’102

MFI interest rates on overnight deposits, i.e.indicators 1 and 7 in Figure 20, cover allovernight deposits, whether or not they areinterest bearing. Zero-interest overnight

deposits influence the transmissionmechanism of monetary conditions, the ownrate of return on M3, the interest burden onhouseholds and non-financial corporations,etc., and are therefore captured by MFIinterest rate statistics.

According to Regulation ECB/2001/13 depositswith agreed maturity are ‘non-transferabledeposits which cannot be converted intocurrency before an agreed fixed term or thatcan only be converted into currency beforethat agreed term provided the holder ischarged some kind of penalty. This item alsoincludes administratively regulated savingsdeposits where the maturity related criterionis not relevant (classified in the maturity band‘over two years’). Financial products with roll-over provisions must be classified accordingto the earliest maturity. Although depositswith agreed maturity may feature thepossibility of earlier redemption after priornotification, or may be redeemable ondemand subject to certain penalties, thesefeatures are not considered to be relevantfor classification purposes.’ Indicators 1 to 4in Figure 18 and indicators 2 to 4 and 8 to 10in Figure 20 follow this definition.

The same Regulation defines depositsredeemable at notice as ‘non-transferabledeposits without any agreed maturity whichcannot be converted into currency without aperiod of prior notice, before the term ofwhich the conversion into cash is not possibleor possible only with a penalty. They includedeposits which, although perhaps legallywithdrawable on demand, would be subjectto penalties and restrictions according tonational practice (classified in the maturityband ‘up to and including three months’), andinvestment accounts without period of noticeor agreed maturity, but which containrestrictive drawing provisions (classified in the

101 See also paragraphs 42 to 50 of Annex II to the Regulation.102 This and the following definitions are found in Part 3 of Regulation

ECB/2001/13.


maturity band ‘over three months’).’ Thisdefinition applies to indicators 5 and 6 inFigure 20.

Also, Regulation ECB/2001/13 defines reposas ‘counterpart of cash received in exchangefor securities/gold sold by reporting agents ata given price under a firm commitment torepurchase the same (or similar) securities/gold at a fixed price on a specified futuredate. Amounts received by reporting agentsin exchange for securities/gold transferred toa third party (‘temporary acquirer’) are to beclassified under ‘repurchase agreements’where there is a firm commitment to reversethe operation and not merely an option todo so. This implies that reporting agentsretain effective (economic) ownership of theunderlying securities/gold during theoperation.’ Repos vis-à-vis households andnon-financial corporations, i.e. indicator 5 inFigure 18 and indicator 11 in Figure 20, arenon-advertised products, which implies thattheir prices cannot necessarily be identifieddirectly from market sources.

7.5.2 Types of loans

The instrument breakdown by type of loanfor MFI interest rate statistics and thedefinitions used follows as far as possible MFIbalance sheet statistics. Bank overdrafts areincluded in MFI interest rate statistics as anadditional separate type of instrument. Thedefinition is given in the following and thedefinition of consumer credit and loans tonon-financial corporations amendedaccordingly.

For the purposes of MFI interest ratestatistics, bank overdrafts (i.e. indicators 12and 23 in Figure 20 are defined as debitbalances on current or chequing accounts.The interest rate on bank overdrafts refersto the rate charged when an overnight deposit‘becomes negative’, i.e. the overnight depositand the bank overdraft are linked to the sameaccount. In contrast to loans to non-financialcorporations up to one year and consumercredit and other loans to households up to

one year, bank overdrafts are without adefined maturity and, in general, authorisedbut taken without giving prior notice to thebank. Usually, the credit institution or otherinstitution defines an upper limit for the sizeand the maximum period of the bankoverdraft the customer can accumulate. Allbank overdrafts are covered, whether theyare within or beyond the limit agreed betweenthe reporting agent and the household ornon-financial corporation. Typically, penaltiesapply if the overdraft is extended beyond theagreed limit. The penalties may be charged asan interest rate component, a component ofother charges, or a combination of both.Penalties on overdrafts applied as acomponent of other charges, i.e. in the formof special fees, are not covered by theannualised agreed rate or NDER, becausethese type of rates only cover the interestrate component of loans. If penalties onoverdrafts are applied as an interest ratecomponent, i.e. a higher interest rate, thishigher interest rate is reflected in MFI interestrate statistics.

Regulation ECB/2001/13 defines loans as‘funds lent by reporting agents to borrowers,which are not evidenced by documents orare represented by a single document (even ifit has become negotiable)’.

Loans to non-financial corporations comprise allloans to non-financial corporations regardlessof their size.103 In the case of outstandingamounts, i.e. for indicators 12 to 14 in Figure18, loans to non-financial corporations coverbank overdrafts. In contrast, new (other)loans to non-financial corporations, i.e.indicators 24 to 29 in Figure 20, exclude bankoverdrafts for the purposes of MFI interestrate statistics. Bank overdrafts to non-financialcorporations constitute as indicator 23 aseparate instrument category in Figure 20.

According to Regulation ECB/2001/13 loansgranted to households in the form of consumercredit are ‘loans granted for the purpose of



personal use in the consumption of goodsand services’. In the case of outstandingamounts, i.e. for indicators 9 to 11 in Figure18, consumer credit covers bank overdrafts.In contrast, new loans to households forconsumption, i.e. indicators 13 to 15 in Figure20 and indicator 30 in Figure 21, excludebank overdrafts for the purposes of MFIinterest rate statistics. Bank overdrafts tohouseholds are included in Figure 20 as partof a separate instrument category, i.e.indicator 12.

Regulation ECB/2001/13 defines loans tohouseholds for house purchases 104 as ‘creditextended for the purpose of investing inhousing, including building and homeimprovements. […] Lending for housepurchases comprises loans secured onresidential property that are used for thepurpose of house purchase and, whereidentifiable, other loans for house purchasesmade on a personal basis or secured againstother forms of assets.’ MFI interest ratestatistics covers indistinguishably secured andunsecured loans to households for housepurchases. In the case of outstanding amounts,i.e. for indicators 6 to 8 in Figure 18, loans tohouseholds for house purchases include bankoverdrafts, while in the case of new business,i.e. for indicators 16 to 19 in Figure 20 and31 in Figure 21, bank overdrafts are excluded.Instead, bank overdrafts to households arecovered in Figure 20 as part of a separateinstrument category, i.e. indicator 12.

Other loans to households are defined inRegulation ECB/2001/13 as ‘loans granted forpurposes such as business, debt consolidation,education, etc.’. In the case of outstandingamounts, i.e. for indicators 9 to 11 in Figure18, other loans to households include bankoverdrafts. In the case of new business, i.e.for indicators 20 to 22 in Figure 20, newloans to households for other purposesexclude bank overdrafts, because they arecovered in Figure 20 as part of a separateinstrument category, i.e. indicator 12.

For MFI interest rates on outstanding amounts,consumer credit, loans to households for

house purchases and other loans tohouseholds together cover all loans grantedto households by resident credit institutionsand other institutions. For MFI interest rateson new business, the possible loan types forhouseholds are bank overdrafts, consumercredit, loans to household for housepurchases and other loans to households.

7.6 Breakdown by amount category105

Large non-financial corporations haveconsiderable economic significance andaccount for a substantial share of bankingbusiness. The loans granted to large firmswill, due to the amounts involved, dominateany weighted average interest rate referringto the non-financial corporations sector as awhole. Small firms, however, may play aspecial role in the transmission of monetarypolicy, since they have limited access to capitalmarkets and are therefore more vulnerableto changes in the lending rates offered byMFIs. In general, larger firms are able tonegotiate the interest rate and conditionsmore than small firms so that larger non-financial corporations face less standardisedproducts with interest rates that are closerto (or identical with) market interest rates.

The size of non-financial corporations can bedefined on the basis of various measures suchas the number of employees or the turnover.Instead of the size of the non-financialcorporation, the size of the loan is considered tobe more relevant for distinguishing bankingconditions in broad terms for the purposes ofMFI interest rate statistics. Indeed, mostMember States confirmed a link between theamount of the loan granted and the interestrate. It should, however, be borne in mind thatthe amount of a loan is only one of many criteriathat are taken into account when negotiatingthe interest rate. The sum of all criteria maybest be summarised by ‘credit risk’.

104 ECB/2001/13 refers to ‘lending for house purchases’ which isused as a synonym.

105 See also paragraph 51 of Annex II to the Regulation.


For new (other)106 loans to non-financialcorporations, i.e. indicators 24 to 29 in Figure20, two size categories for the amount of theloan granted are distinguished, i.e. ‘up to andincluding EUR 1 million’ and ‘over EUR 1million’. The amount refers to the single loantransaction considered as new business. Itdoes not cover all business between the non-financial corporation and the reporting agent.

Amount categories apply only to new businessrates vis-à-vis non-financial corporations andonly to the instrument category of otherloans. No breakdown by amount categoriesis required for bank overdrafts or for depositsvis-à-vis non-financial corporations. Also nobreakdown by amount is requested fordeposits and loans vis-à-vis households or forinterest rates on outstanding amounts.

The breakdown by amount category for otherloans to non-financial corporations increasesthe comparability of the interest rates onsuch loans, since the size of the individualloans covered vary widely. As alreadymentioned at the beginning of this chapter,without amount categories a weightedaverage interest rate on loans to non-financialcorporations would be dominated by theinterest rates on the largest loans during thereporting month. This weighted averagelending rate might not be representative ofmost loans to non-financial corporations. Theindictors for (other) loans to non-financialcorporations of an amount up to and includingEUR 1 million are assumed to reflect theinterest rates faced by smaller corporations,which are ‘price takers’ rather than ‘pricemakers’.

7.7 Breakdown by original maturity,period of notice period and initialrate fixation107

7.7.1 Time bands

Depending on the type of instrument andwhether interest rates on outstandingamounts or new business are being lookedat, MFI interest rate statistics provide for a

breakdown by original maturity, period ofnotice, or initial period of fixation of theinterest rate. These breakdowns refer to timebands or ranges. For example, an interestrate on a deposit with an agreed maturity ofup to two years refers to an average interestrate across all deposits with an agreed originalmaturity between one day and a maximum oftwo years. Time bands or ranges are alsoused in the MFI balance sheet statistics.

The alternative would be to define exactpoints in time, i.e. interest rates oninstruments with a maturity, notice period orrate fixation of ‘x years/month’. Advantagesand drawbacks of basing statistics on exactproducts rather than instrument categoriesare discussed in Chapter 7.2.

7.7.2 Original maturity and period ofnotice

A breakdown by original maturity is applied inFigure 20 to new business on deposits withagreed maturity. In Figure 18, a breakdownby original maturity is required for all lendingrates on outstanding amounts and for alldeposit rates on outstanding amounts withthe exception of repos. For repos, i.e.indicator 5 in Figure 18 and indicator 11 inFigure 20, no maturity breakdown is requiredat euro area level, as repos are assumed tobe predominantly very short-term. At nationallevel, NCBs may ask for a maturity breakdownfor repos, which would be in line with the 4th

recital and Article 3(2) of the Regulation. Thedefinition of (and the breakdown by) originalmaturity follows Regulation ECB/2001/13,which states that ‘maturity at issue (originalmaturity) refers to the fixed period of life ofa financial instrument before which it cannotbe redeemed (e.g. debt securities) or beforewhich it can be redeemed only with somekind of penalty (e.g. some types of deposits)’.

The same Regulation defines further that the‘period of notice corresponds to the time

106 ‘Other’ refers to ‘other than bank overdrafts’.107 See also paragraphs 52 to 57 of Annex II to the Regulation.


between the moment the holder gives noticeof an intention to redeem the instrument andthe date on which the holder is allowed toconvert it into cash without incurring apenalty. Financial instruments are classifiedaccording to the period of notice only whenthere is no agreed maturity.’ For MFI interestrates a breakdown by period of notice isapplied in Figure 20 to new business ondeposits redeemable at notice.

7.7.3 Period of maturity for a loan takenout in tranches108

Question:

A loan is agreed between a credit institutionand a customer at t0, but the customer takesout the funds only two month later at t1.Should the original maturity of this loan bedefined as:

• the date of the contract (t0) until the endof the contract (tn), or

• the first value date (t1) until the end of thecontract (tn)?

Answer:

A household or non-financial corporation willnormally take out a loan (other than a bankoverdraft) in full at the start of the contract.Hence, in general, the contract date t0 andthe first value date t1 coincide. It may happenthat a household or non-financial corporationdecides to take out the loan in tranches attimes t1, t2, t3, etc. instead of withdrawing thefull amount at the start of the contract (timet0). Whatever the pattern of withdrawals, theoriginal maturity of the loan always refers tothe agreed original loan period according tothe loan contract, i.e. in this example to theperiod from t0 to tn. However, subject toclarification of the most appropriate definitionof ‘original maturity’ in MFI balance sheetstatistics, it is also acceptable in MFI interestrate statistics to define the original maturityas the value date until the end of the contract,i.e. from t1 to tn.

The loan actually appears in the MFI balancesheet statistics only at time t1. The sameapplies for MFI interest rate statistics onoutstanding amounts. In MFI interest ratestatistics on new business, the loan is recordedat the time of the agreement on the contract,i.e. at time t0.

7.7.4 Initial period of fixation of theinterest rate

The absence of any reference to fixed andvariable interest rates was often considered adrawback of the euro area retail interest ratestatistics compiled according to a short-termapproach.109 For the steady-state approach,users therefore expressed a strong need fora distinction between variable and fixedinterest rates on loans. In order to implementsuch a breakdown in a harmonised way acrossall Member States, variable and fixed needs tobe clearly defined. The definition needs to beunambiguous for all Member States and ableto cope with financial innovations, which tendto cause difficulties at the borderline.However, large national differences exist inthe retail banking business in the EU to theextent that there is no common view on theperiod of fixation that classifies a lendinginterest rate as fixed or variable. An interestrate which is fixed for one year and thenvariable might be considered as fixed in oneMember State and as variable in anotherdepending on the national lending practice.

To overcome this problem the initial period ofrate fixation was introduced as the basis for abreakdown for new lending business in MFIinterest rate statistics. It gives an indicationof the variability of interest rates at euroarea level, without prejudging whether a loanwith a specific fixation period is consideredas fixed or variable in the national context.The initial period of fixation is apredetermined period of time at the start ofa contract during which the value of theinterest rate cannot change. The value of the

108 Defined in Chapter 5.3.8.109 See description in Chapter 2.


interest rate is only considered to beunchangeable if it is defined as:

• an exact level, for example as 10%, or

• a spread over an external index at a certaintime, for example as six-month EURIBORplus two percentage points at a certain dayand time, which is equivalent to an exactinterest rate level.

If at the start of the contract the customerand the reporting agent agree on a procedurefor calculating the lending rate, this is notconsidered to be an initial rate fixation. Forexample, it might be agreed that six-monthEURIBOR plus two percentage points is theinterest rate for the first three years of theloan. As the value of this interest rate changesin line with EURIBOR during the three years,there is no initial rate fixation.

Three periods of initial rate fixation aredefined for new loans to households forconsumption and for other purposes, and fornew other loans to non-financialcorporations, i.e. indicators 13 to 15, 20 to22, and 24 to 29 in Figure 20; loans withoutany interest rate fixation are included asvariable or floating rates with loans that haveup to one year initial rate fixation:

• variable rate and up to (and including) oneyear initial rate fixation,

• over one and up to (and including) fiveyears initial rate fixation, and

• over five years initial rate fixation.

Considering that 95% of loans to householdsfor house purchases are granted with anoriginal maturity of over five years, and insome countries with interest rate fixation of10 years and more, for this type of loan fourperiods of initial fixation are distinguished;again, loans without any interest rate fixationare included as variable or floating rates withloans that have up to one year initial ratefixation:

• variable rate and up to (and including) oneyear initial rate fixation,

• over one and up to (and including) fiveyears initial rate fixation,

• over five and up to (and including) 10 yearsinitial rate fixation, and

• over 10 years initial rate fixation.

Normally, the initial period of fixation isshorter or equal to the original maturity ofthe loan.110 The initial period of fixation mightbe short and the interest rate agreed betweenthe customer and the reporting agent for thisinitial period of fixation might not berepresentative of the entire maturity of theloan. New business statistics only reflect theinterest rate that is agreed for the initialperiod of fixation at the start of a contract orafter renegotiation of the loan. If, after thisinitial period of fixation, the interest rateautomatically changes to a variable rate, whichmight be at a very different level, this is notreflected in the MFI interest rates on newbusiness but in those on outstanding amounts.Hence, both sets of statistics are needed tocapture the interest rate level anddevelopment in the euro area and at nationallevel. The two following examples furtherillustrate the interdependence of MFI interestrate statistics on new business and onoutstanding amounts.

A 10-year consumer credit is granted. At timet0, the customer and the reporting agent agreethat the interest rate is fixed at 10% p.a. forthe first four years and that a new interestrate level is negotiated at time t4.

111 The resultof the renegotiations at time t4 is a loan fixedat 8% p.a. for the remaining maturity. At timet0, the maturity of this loan is 10 years withan initial period of fixation of four years. Attime t4, the original maturity of the loan is still

110 In exceptional cases, the initial period of fixation may be longerthan the maturity of the loan. For example, if a loan is agreed towhich a fixed interest rate applies for the first year and this loanmay be repaid with one-month notice, then the shortest possiblematurity is one month and the period of initial rate fixation isone year.

111 See also Figure 14.


10 years and the classification of the loan inthe statistics on outstanding amounts istherefore independent of the newnegotiations.112 Hence, the MFI interest ratestatistics on outstanding amounts would reflectat time t4 the decrease from 10% to 8% p.a.in the category ‘maturity over five years’. Theresults of the renegotiation are also capturedas new agreements in MFI interest ratestatistics on new business; the 8% p.a. arerecorded for loans with an initial rate fixationof six years in the category ‘over five yearsinitial rate fixation’.

Another 10-year consumer credit is granted.This time the customer and the reportingagent agree at time t0 that the interest isfixed at 9% p.a. for the first 12 months andthat it then automatically adjusts to EURIBORplus x basis points.113 This rate is then appliedfor the next 12 months, after which it willagain automatically adjust to EURIBOR plus xbasis points. At time t0, the maturity of theloan is 10 years with an initial period offixation of one year. Only the interest rate of9% for the first year is considered as newbusiness at time t0. The movements in theinterest rate over time are captured in theMFI interest rate on outstanding amounts, butno further new business occurs.

The situation is different if the initial periodof fixation is very short as compared to thewhole maturity of the loan and the interestrate offered during this period is significantlybelow market conditions. For example, a 10-year loan is granted where the interest rateis fixed at 2% for the first six months andthen automatically adjusts to either a fixed toa variable interest rate at a level reflectingmarket conditions. To qualify for thetreatment described here, the 2% must be anintroductory offer and ‘eye-catcher’ to attractnew customers, in the sense that the interest

rate is significantly lower than current marketconditions, i.e. at least 200 basis points, andapplies for a very short period of the loan.The interest rate that applies after this initialperiod of fixation has to be already laid downin the contract agreed at time t0. It can befixed or variable but in any case at a levelreflecting market conditions. The treatmentof loans comprising such introductory offersto attract customers is the same as for step-up loans explained in Chapter 9.1. This meansthat the new business statistics cover the loanat time t0. The interest rate on new businesscan either be computed as a geometricaverage of the factors ‘1 + interest rate’ or asan NDER comprising the introductory offerand the interest rate agreed for after theinitial period of fixation. If a variable interestrate is agreed to apply after the initial periodof fixation, for example EURIBOR plus x basispoints, as a convention the value of thatvariable rate is for the purposes of MFIinterest rates on new business be determinedas at time t0. The statistics on outstandingamounts always reflect the interest rateactually applied by the reporting agent at thetime of calculation of the MFI interest rate,i.e. the 2% during the first months and thenthe agreed interest rate at a level reflectingmarket conditions.

For all step-up and step-down loans, the initialperiod of fixation is equal to the maturity ofthe loan, because a fixed interest rate isagreed for the whole maturity of the loan attime t0 when the contract is signed.

112 This treatment is in line with MFI balance sheet statistics. Thissituation of a loan that is renegotiated before it reaches maturityis different from the deposit with agreed maturity that isrenegotiated at maturity described in Chapter 5.3.3. In the lattercase, the maturing deposit with agreed maturity is renegotiated,therefore classified as new business, and the maturity of the(new) deposit counted as commencing at the point of the ‘newbusiness’ classification.

113 See also Figure 15.


8 Treatment of credit card credit

8.1 Credit card linked to an overnightdeposit account114

Question:

Holders of credit cards can draw cash ormake payments with a direct charge to anovernight deposit account. The creditinstitution sets the credit card limit, i.e. themaximum amount that can be drawn or paidwith the credit card. When the overnightdeposit account to which the credit cardwithdrawals are debited is not provisioned,the credit institution, if it authorises thewithdrawal, will consider an overdraft to havebeen incurred and will charge interest on thebalance. This transaction and interest rateshould be included in the MFI interest ratestatistics, but is the amount only the moneydrawn or the total limit available?

Furthermore, in general the cardholder makespurchases during the month paying with thecredit card. The receipts of the purchasesarrive successively at the credit institutionwhich issued the card. The credit cardcontract with the cardholder specifies thateverything that arrives during a month will bedebited to the cardholder’s overnight depositaccount on the 5th day of the following month.No interest is charged by the cardholder forthis deferral in payment which at mostamounts to one month. Should thesetransactions be included in the interest rateon overnight deposits (or bank overdrafts)given the enormous stock of outstandingtransactions at the end of a month that areto be settled only five days later?

Answer:

The agreement on a maximum amount thatmay be drawn with the credit card, which isoften referred to as a ‘credit limit’ or ‘creditfacility’, does not in itself constitute newbusiness. As long as no payments have beenmade with the credit card, no figure isrecorded in MFI interest rate statistics onnew business and on outstanding amounts.

If payments are made with the credit cardduring the month and the overnight depositaccount to which the card is linked issufficiently provisioned, again no new businessoccurs. The only (indirect) influence on MFIinterest rate statistics is the reducedoutstanding amount on the overnight deposit.

If payments are made with the credit cardand the overnight deposit account to whichthe card is linked is not provisioned, but thecredit institution authorises the withdrawal,then the account turns into a bank overdraft.The reporting agent will consider the amounton the account as a bank overdraft and chargeinterest. The MFI interest rate statistics forbank overdrafts reflect this interest rateapplied by the credit institution and the actualamount drawn, but not the total amountagreed as the credit limit.

The credit card credit in the form of a bankoverdraft may be reimbursed over severalmonths with the customer paying back a fixedamount each month. The loan repaymentsare reflected only as a decrease in theoutstanding amounts.

In this example, the credit card contractspecifies that all receipts that arrive at thecredit institution during a month will bedebited to the cardholder’s overnight depositaccount only on the 5th day of the followingmonth and no interest will be charged untilthe 5th day. Hence, during the referencemonth, in this example from the 6th day ofthe current month to the 5th day of thefollowing month, the credit institution isacting as a ‘letter box’ for the credit cardreceipts. The customer has not been granteda loan but rather a deferral in payment. Itwould hence seem more appropriate torecord the amounts covered by the interest-free period as receivables and not as loans inthe MFI balance sheet and hence to excludethem from MFI interest rate statistics.However, the exact treatment will depend

114 See also paragraph 75 of Annex II to the Regulation.


on the accounting rules applied. This needsto be taken up in the context of the MFIbalance sheet statistics. If the amountscovered by the interest-free period wereconsidered as receivables in MFI balance sheetstatistics, then only on the 5th day of thefollowing month the total amount of thecredit card receipts would become relevantfor MFI interest rate statistics and would haveto be reported as specified above. Chapter8.3 deals with the case where the interest-free period is considered to be a loan.

8.2 Credit card linked to a loanaccount115

Question:

Holders of credit cards can draw cash ormake payments with a direct charge to a loanaccount. As in Chapter 8.1, the creditinstitution defines the credit card limit, thereceipts arrive successively at the creditinstitution, and the contract specifies thateverything that arrives during a particularmonth will be debited to the cardholder’sloan account on the 5th day of the followingmonth. As above, no interest is charged forthis deferral in payment. Should only theportion drawn or the total limit be recorded?What is new business and what should beincluded in the statistics on outstandingamounts?

Answer:

Analogous to Chapter 8.1, so long as nopayments have been made with the creditcard no figure is recorded in MFI interestrate statistics. The agreement on a maximumamount that may be drawn with the creditcard does not constitute new business.

When a withdrawal is charged to the loanaccount, the interest rate statistics for newbusiness reflect the interest rate that thecredit institution applies to the loan. Theamount of new business is the money thathas actually been drawn, rather than the totalamount agreed as the credit limit.

The credit card credit in the form of a loanmay be reimbursed over several months withthe customer paying back a fixed amount eachmonth. The repayments of the loan arereflected only as decreases in the outstandingamounts.

As in Chapter 8.1, the credit card receiptswill be debited to the cardholder’s loanaccount only on the 5th day of the followingmonth without and no interest will be chargeduntil the 5th day. Again it would seem moreappropriate to record the amounts coveredby the interest-free period as receivablesrather than loans in the MFI balance sheet andtherefore to exclude them from MFI interestrate statistics. However, this should be takenup in the context of the MFI balance sheetstatistics. If the amounts covered by theinterest-free period were considered asreceivables then, in this case where the creditcard is linked to a loan account, on the 5th

day new business (amount and interest rate)should be recorded. On the same day theamount and interest rate should also enterthe statistics on outstanding amounts.Chapter 8.3 deals with the case where theinterest-free period is considered to be aloan.

8.3 Credit card not linked to anyaccount

Question:

In contrast to Chapters 8.1 and 8.2, acustomer’s credit card is not linked to adeposit or loan account. In fact, thecardholder does not have to have any accountwith the credit institution that issued thecredit card. The credit institution sets thecredit card limit, i.e. the maximum amountthat can be drawn or paid with the creditcard. During the month, the cardholder makespurchases, paying with the credit card. Thecredit card receipts arrive at the creditinstitution that issued the card. Each month-

115 This excludes loan accounts in the form of bank overdrafts,which are dealt with in Chapter 8.1.


end, the credit institution sends an invoice tothe value of the credit card receipts to thecardholder who is granted an interest-freeperiod of 56 days to pay the bill. A minimumamount of 3% of the balance must be paidafter this period has elapsed or additionalcharges will be imposed, in addition to aninterest payment. Interest is only chargedafter this interest-free period has passed forthe outstanding balance. A large number ofcardholders pay off their balance in full at thecounter or by bank transfer before the duedate and thereby incur no interest charges.Should the balances at the month-end thatare still covered by the interest free periodand so may incur no interest charges beincluded in the MFI interest rate statistics?Furthermore, should the credit card balanceswhich are attracting interest charges beclassified as overdraft, consumer credit orother loan?

Answer:

As in Chapters 8.1 and 8.2, the creditinstitution is in principle acting as a ‘letterbox’ for the credit card receipts. Thecustomer is not considered to have beengranted a loan but a deferral in payment andit would seem more appropriate to recordthe amounts covered by the interest-freeperiod as receivables rather than loans in theMFI balance sheet and to exclude them fromMFI interest rate statistics. However, in thecase where the credit card is not linked to anaccount, the interest-free balances incurredon credit cards are often included as loans inthe MFI balance sheet statistics. Again, thisshould be taken up in the context of the MFIbalance sheet statistics. If the amountscovered by the interest-free period wereconsidered as loans, then both the interest-free balances and the balance for whichinterest is charged should be treated as bankoverdrafts in MFI interest rate statistics. Themechanics of the credit card balances thatare not linked to an account are similar tobank overdrafts, i.e. without defined maturity,authorised but taken without giving priornotice, with a defined upper limit for theamount that can be drawn, and unsecured.

Furthermore, the interest rate that appliesafter the interest-free period is of a levelsimilar to overnight deposits becomingnegative.

As described in Chapter 8.1, the MFI interestrate statistics for bank overdrafts reflect theinterest rate applied by the credit institutionto the bank overdraft and the actual amountdrawn, but not the total amount agreed asthe credit limit. In this example, thecardholder is granted an interest-free periodof 56 days and hence 0% interest recordedfor 56 days for the newly incurred balances.For balances that are older than 56 days, theinterest rate that is charged by the cardissuing institution is reflected. If a minimumamount of 3% of the balance is not repaidwithin 56 days and a penalty is charged in theform of higher interest rates, this penalty isalso reflected in MFI interest rate statistics. Ifthe penalty in applied in form of fees or othernon-interest components, it is not covered.116

The repayments of the outstanding balancesduring the month are captured by decreasesin the outstanding amounts. If the cardholderpays off the balance in full at the counter orby bank transfer before the due date andthereby incurs no interest charges, only theinterest-free period is reflected for bankoverdrafts in the form of a 0% interest on theoutstanding amount.

The difference across countries in thetreatment of interest-free periods on creditcards in MFI balance sheet statistics hasimplications for MFI interest rate statistics. Insome countries the interest-free period isconsidered as a loan and included in thestatistics and in others it is off-balance sheetand hence excluded from the statistics. TheMFI interest rate statistics are therefore notfully harmonised in this respect, which needsto be accepted at this stage.

116 See also the treatment of penalties on bank overdrafts discussedin Chapter 7.5.2.


9 Specific national products

Not all specific national products can becovered by this manual. Hence, in the sameway as paragraphs 74 to 82 of Annex II to theRegulation, the treatment of the selectedproducts defined in this chapter should beused as a reference for other nationalproducts with similar characteristics.

9.1 Step-up (step-down) deposits andloans

Paragraph 74 of Annex II to the Regulationdefines a step-up (step-down) deposit or loanas ‘a deposit or loan with a fixed maturity towhich an interest rate is applied that increases(decreases) from year to year by a pre-fixednumber of percentage points’. As mentionedin Chapter 7.7.4, the initial period of fixationfor a step-up or step-down loan is equal tothe maturity of the loan, because a fixedinterest rate is agreed for the whole maturityat time t0 when the contract is signed.

An example of a step-up loan is given inFigure 22, where a four-year consumer creditis granted, for which the credit institutioncharges 5% interest in the first year, 7% inthe second, 9% in the third and 13% in thefourth. The new business statistics cover thestep-up loan at time t0 in the category ‘overone and up to five years initial rate fixation’.The interest rate on new business can eitherbe computed as a geometric average of thefactors ‘1 + interest rate’ or as an NDER.The NDER for this example is 8.2445% andgives the mathematically correct result. Thegeometric average of the factors ‘1 + interestrate’ provides an approximation. It is 8.46%for the example and calculated as follows:

MIR (NB) =

[Equation 8]

Unlike the NDER, the geometric average doesnot take into account the timing of the interest

Figure 22Step-up loan

0

500

1000

1500

Ou

tsta

nd

ing

am

ou

nt

2%

4%

6%

8%

10%

12%

14%

In

terest

rate

Loan 1000 1000 1000 1000

Interest rate 5% 7% 9% 13%

t0 t1 t2 t3

t0 t1 t2 t3

MIR (OA) 5% 7% 9% 13% Weight (OA) 1000 1000 1000 1000

MIR (NB) 8.24% (8.46) - - -

Weight (NB) 1000 - - -

0845998.01)13.01)(09.01)(07.01)(05.01(4 =−++++


rate payments. This means that the interestrates computed as a geometric average are thesame for a step-up loan with 5% in t0 and up to13% in t3 and for a step-down loan with 13% int0 and down to 5% in t3. The difference in resultbetween the NDER and the geometric averageis small and accepted for MFI interest ratestatistics on new business.

The statistics on outstanding amounts coverfrom time t0 to t3 the interest rates actuallyapplied by the reporting agent at the time ofcalculation of the MFI interest rate, i.e. 5% attime t0, and t1, 7% at time t1, 9% at time t2 and13% at time t3. These interest rates appear inthe instrument category ‘over 1 and up to 5years maturity’.

9.2 Savings plan for housing loans

Question:

According to paragraph 81 of Annex II to theRegulation, savings plans for housing loans arelong-term low return savings schemes that, aftera certain period of saving, give the saver theright to a housing loan at a discounted rate. InMFI balance sheet statistics, these savings plansare classified under deposits with agreedmaturity over two years. When the savings planis transformed into a loan, this loan is classifiedas loan to households for house purchases. Forthe purposes of MFI interest rate statistics,reporting agents report as new deposit businessthe interest rate that is agreed at the time theinitial deposit is placed. The correspondingamount of new business is the amount of moneythat has been placed. The increase of the amounton the deposit over time is only covered by theoutstanding amounts. At the time when thedeposit is transformed into a loan, this new loanis recorded as new lending business. The interestrate is the discounted rate that is being offeredby the reporting agent. The weight is the totalamount of the loan that is being granted to thehousehold.

a) Which amount should be covered as newdeposit business? Only the first tranche paidinto the account, or the total amount of

the deposit which has been agreed andwhich will be paid into the account intranches over time?

b) Which amount on the deposit should becovered by outstanding amounts? Thesingle tranche, which has been paid intothe account in the reference month, orthe cumulated amount of all tranches whichhave been paid into the account so far?

Answer:

Analogous to Chapter 5.3.4, the treatment innew business statistics depends on whether ornot it is known for certain ex ante which amountwill be held on the deposit account at maturity.In general, savings plans for house purchases arehighly flexible products and it is only known expost what amount has been accumulated on theaccount and can be transformed into a loan orpaid out to the customer. Hence, in general theamount of new business is the initial amount ofmoney that has been placed, i.e. the first tranchepaid into the account. The correspondinginterest rate for new business is the depositinterest rate that has been agreed for when theinitial deposit is placed.

If it is known ex ante for certain what exactamount will be held on the account atmaturity, then the total amount on thedeposit should be recorded as new business.

In both cases, the outstanding amount is thestock of all deposits placed by the customeron the account. At the time of reporting, theMFI interest rate statistics on outstandingamounts therefore reflect the accumulatedamount of all tranches that have been paidinto the account so far and the correspondinginterest rate.

9.3 Interest rate on zero-coupon-bond-like savings bond

Question:

In some countries certain non-marketablesavings bonds are classified as deposits in MFI


118034,015.0

225.0

1 15.0

280

80100

1 X

5.0

5.0

=−�

�

�

�

�

�

�

�

+=−

�

�

�

�

�

�

�

�

�

�

�

�

�

� −

+=

balance sheet statistics. These savings bondsare discounted ‘zero-coupon-bond-like’products involving only two cash flows: theinitial placement of the ‘deposit’ at adiscounted value and the final redemptionpayment at ‘nominal value’ which includes theinterest. Which amounts are to be consideredas new business and outstanding amounts forthe purpose of calculating interest rates onsavings bonds? Should the initial value of thedeposit placed in t0 be reflected both as newbusiness and outstanding amounts with anyaccrued interest over the lifetime to beexcluded for the savings bonds? Thistreatment would be in line with the MFIbalance sheet statistics, which requiresaccrued interest payable on deposits to beclassified separately as ‘remaining liabilities’.

Answer:

If, for example, the discounted value of thesavings bond is EUR 80 and the nominal valueat which the it is redeemed after two years isEUR 100, then at time t0 the amount of EUR80 is recorded as new business. The interestrate on new business is then 11.8034%. Itmay be calculated as an annualised agreedrate like in Equation 9 or as NDER shown inFigure 23.

[Equation 9]

The amount of EUR 80 is also the outstandingamount from time t0 until maturity, includingthe day of maturity where the bond isredeemed at EUR 100. The reason is that forMFI interest rate statistics, the EUR 20difference between the issuance andredemption value is interpreted as theinterest payment for the two years and notas principal. The interest of EUR 20 is knownex ante and interpreted as accrued over thematurity of the bond, in this example overtwo years. Hence, 11.8034% is reflected asthe interest rate on outstanding amounts forthe whole period. In this way, the savingsbond is treated just like a deposit with anagreed maturity of two years where theinterest is fixed at 12.5% p.a. and paid atmaturity.

9.4 Split of loan into two parts

Question:

A customer takes out a mortgage loan ofEUR 50,000 for 20 years at a variable interestrate. It is agreed with the credit institutionand laid down in the contract that at anypoint in time the customer may split themortgage. For example, after three years thecustomer may choose to split it into twoparts: EUR 20,000 at a fixed rate and EUR30,000 at a variable rate. How would thissplit be treated in MFI interest rate statistics?

Figure 23

Savings bond with agreed maturity of two years, interest rate payment at end of second year:

tOutstand-

ing deposit

Interest

rate

Reinvested

interest

Interest

payments

Payments

of

principal

Cash

flow

Discount

factor =

(1+NDER)^

(-t/365)

Present

value of

cash flow

NDER

15/11/2003 1 80 -80

14/11/2004 365 80 0 0 0 0 0.89 0.00 11.80340%14/11/2005 730 80 25% 0 20 80 100 0.80 80.00 11.80340%

20 80.00


Answer:

The possibility of the split is already agreed inthe contract for the mortgage loan of EUR50,000. The split does not constitute newbusiness if the same contract alreadydetermines the terms and conditions of thetwo loans after a possible split, i.e. the amountof money that would be lent at variable andat fixed interest rate as well as the interestrates applied to these amounts. In this case,the customer is free to choose the timing ofthe split without engaging in renegotiation ofthe contract. The change in the interest ratesfrom fully variable to partly fixed after thesplit is then only reflected in the statistics onoutstanding amounts.

If the split of the mortgage loan into twoparts implies new negotiation of the termsand conditions of the contract, for examplethe amount lent at variable and at fixed rateor the level of these interest rates, then eachof the loans after the split, i.e. the mortgageloan of EUR 20,000 and of EUR 30,000,constitutes new business.

9.5 Option of converting the depositinto equity shares

Question:

A deposit with agreed maturity of one year isplaced at a higher than market interest rate,currently for example at 14%, where atmaturity the credit institution (not thecustomer) has the option of converting thedeposit into the shares of a specific company.If at maturity the share price is high, thecredit institution will not exercise the option,but pay the 14% interest and return thedeposit. If at maturity the share price is low,then the credit institution will also pay the14% interest and exercise the option. In thiscase, at the time of the conversion intoshares, the customer will lose part of thedeposit because the value of the shares hasfallen. The interest rate is high because itincludes a risk premium for the possibility oflosing part of the capital. This instrument is

also referred to as reverse convertible. Howshould this instrument be treated in MFIinterest rate statistics?

Answer:

As there is no capital certainty for thedeposit, it is doubtful whether this instrumentshould be recorded on the MFI balance sheetas deposit liabilities. Its classification iscurrently under review. Assuming that theinstrument is recorded on the MFI balancesheet as deposit liabilities, then the treatmentin MFI interest rate statistics should be in linewith paragraph 78 of Annex II to theRegulation. It reads as follows: ‘Loans may beoffered to households or non-financialcorporations with associated derivativecontracts, i.e. an interest rate swap/cap/flooretc. As a convention, such associatedderivative contracts shall not be included inthe annualised agreed rate on new business.The annualised agreed rate on outstandingamounts shall always cover the rates appliedby the reporting agent at the time of thecalculation of MFI interest rates. Hence, inthe case that such a derivative contract isexercised and the reporting agent adjusts theinterest rate charged to the household ornon-financial corporation, this shall bereflected in the statistics on outstandingamounts.’

In this case of a deposit with one yearmaturity at an (higher than market) interestrate of 14%, where at maturity the creditinstitution has the option o converting thedeposit into shares, the annualised agreedrate or NDER on new business captures the14% for the deposit with agreed maturity.This reflects the agreement between thedepositor and the reporting agent and it isknown when the money is being placed.117

The gain or loss in capital at the time of theconversion into shares is only known ex post

117 Contrary to Chapter 9.6, in this case an interest rate is agreedex ante for the whole amount of money that is being placed.Although the inclusion of this agreed (higher than market)interest rate might overestimate the level of the deposit rate, thetreatment is in line with paragraph 78 of Annex II to theRegulation. Any adjustment of the interest rate of the kindproposed in Chapter 9.7 would be arbitrary.


when the product matures and thereforecannot be covered by the new business rate.

The annualised agreed rate or NDER onoutstanding amounts always covers the interestrate applied by the reporting agent at thetime of calculating the MFI interest rates.Therefore, until the day of maturity, the 14%on the deposit with agreed maturity iscaptured. At the time of the conversion intoshares, the amount converted is no longer adeposit and hence not included in MFI interestrate statistics, neither as new business nor asoutstanding amounts.

9.6 Interest rate linked to share price

Question:

A deposit is agreed where the creditinstitution does not guarantee any interestrate but the return on the deposit is entirelylinked to the stock market price of shares ina specific company. If, for example, the shareprice rises by 10%, the credit institution willpay 20% interest on the deposit to thecustomer. However, if the share price falls by30%, then the customer loses 30% of themoney placed as a deposit. How should thisinstrument be treated in MFI interest ratestatistics?

Answer:

Even more than in Chapter 9.5 it is doubtfulwhether this instrument should be recordedon the MFI balance sheet as deposit liabilities,because there is no capital certainty for thedeposit. Again, this needs to be taken up inthe context of the MFI balance sheet statistics.The alternative classification in the MFIbalance sheet would be as remaining liabilities,in which case this instrument would not becovered at all by MFI interest rate statistics.

If the instrument was recorded on the MFIbalance sheet as deposit liabilities, then aconvention is needed for the level of theinterest rate on new business. The return ofthe deposit is linked to the performance of

the stock market price of shares, which isonly known ex post and cannot be reflectedby the new business rate, and no minimumreturn at all is guaranteed.118 The conventionfor the purposes of MFI interest rate statisticsis that the annualised agreed rate or NDERon new business captures an interest rate of0% for the deposit.119

The annualised agreed rate or NDER onoutstanding amounts covers the interest rateapplied by the reporting agent at the time ofthe calculation of MFI interest rates, i.e. itshould be based on the stock market price ofthe shares of the specific company at thetime of calculation.

9.7 Treatment of deposit comprisingtwo components120

Question:

A customer and a credit institution agree ona deposit of EUR 10,000 comprising twoparts. These two parts are inextricably linkedand cannot be placed separately. The firstpart of EUR 6,000 is placed with an agreedmaturity of six months at a (higher thanmarket) fixed interest rate of 15% paid atmonthly frequency. The second part of EUR4,000 is placed with an agreed maturity ofthree years and the return is linked to astock exchange index with a guaranteedminimum return of 0%. How should thisproduct be treated in MFI interest rate

118 This is different to Chapter 5.3.12 where a floor of 2% is agreedas a minimum return, and to Chapter 9.7 where a minimumreturn of 0% is agreed for the second part of the deposit. It isalso different to Chapter 9.5 where a (higher than market)interest rate is agreed ex ante for the whole amount of moneythat is being placed.

119 This convention might underestimate the level of the depositrate. However, any adjustment of the interest rate of the kindproposed in Chapter 9.7 would be arbitrary.

120 This chapter does not cover financial products where the firstcomponent is a deposit and the second component an investmentin mutual fund shares/units or capitalisation products offered byinsurance companies. For these financial products only theamount and the interest rate relating to the first component, i.e.the deposit, is covered by MFI interest rate statistics. The amountand the interest rate placed in the second component areoutside the scope of MFI balance sheet statistics and not coveredby MFI interest rate statistics. In general, for these financialproducts the interest rate offered for the first component, i.e. thedeposit, is (close to) market interest rates.


%2151.336*40006*6000

6*4000*0

36*40006*6000

6*6000*1

12

15.01

12

=+

++�

�

�

�

�

�

�

�

−�

�

�

� +

statistics? Would the treatment change if the0% interest, i.e. the minimum return, on thesecond part of the deposit were notguaranteed, implying the possibility of a capitalloss for the investor?

Answer:

Paragraph 79 of Annex II to the Regulationreads as follows: ‘Deposits may be offeredcomprising two components: a deposit withagreed maturity to which a fixed interest rateis being applied and an embedded derivativewith a return that is linked to theperformance of a defined stock exchangeindex or a bilateral exchange rate, subject toa minimum guaranteed return of 0%. Thematurity of both components may be thesame or may differ. The annualised agreed rateon new business shall capture the interest ratefor the deposit with agreed maturity, as it reflectsthe agreement between the depositor andthe reporting agent and it is known when themoney is being placed. The return on theother component of the deposit linked to theperformance of a stock exchange index or abilateral exchange rate is only known ex postwhen the product matures and thereforecannot be covered by the new business rate.Hence, only the guaranteed minimum return of0% should be captured. The annualised agreedrate on outstanding amounts shall alwayscover the interest rate applied by thereporting agent at the time of the calculationof MFI interest rates. Until the day ofmaturity, the rate on the deposit with agreedmaturity shall be captured as well as theguaranteed minimum return on the depositcontaining the embedded derivative. Only atmaturity shall the MFI interest rates onoutstanding amounts reflect the annualised

interest rate that is paid by the reportingagent’ (emphasis added).

Hence, according to the Regulation, both partsof the deposit are covered by MFI interestrate statistics, i.e. the first part of EUR 6,000at 15% for a term of six months, and thesecond part of EUR 4,000 at the guaranteed0% for three years. However, the Regulationis silent about the precise treatment of thetwo parts, i.e. if they should be treated astwo separate deposits or as a single depositfor the purpose of calculating MFI interestrates. If both parts were at the same maturityand would hence fit into the same instrumentcategory for MFI interest rate statistics, noquestion would arise, as it would lead to thesame result whether both parts were treatedseparately or together.

If the two parts have different maturities, asin the example, the two parts of the depositare classified into two different instrumentcategories. The two parts should, however, belooked at together, because together theyform one contract, could not be placedindependently of each other, and hence theinterest rates on the two parts need to beseen as a ‘package’. For such products eitherthe NDER is calculated comprising the twocomponents or as an approximation aweighted average of the annualised agreedrates on both parts with the maturity and theamount placed as weighting information. Thedifference in the interest rate level resultingfrom the different compilation method needsto be accepted.

For the example, the calculation of theannualised agreed rate is given in Equation 10and of the NDER in Figure 24.

[Equation 10]


Figure 24

Deposit with two components: 6,000 at 15% for three months, 4,000 at 0% for 36 months, standard year of 365 days:

tOutstand-

ing loan

Interest

rate p.a.

Outstand-

ing loan

Interest

rate p.a.

Interest

payments

Repay-

ments of

principal

Cash flow

Discount

factor =

(1+NDER)^(-

t/365)

Present

value of

cash flow

NDER

01/01/2002 1.00 6,000 4000 -10,000.0

31/01/2002 30.42 6,000 15% 4000 0% 75.00 75.0 1.00 74.81 3.1371%28/02/2002 60.83 6,000 15% 4000 0% 75.00 75.0 0.99 74.61 0.0331/03/2002 91.25 6,000 15% 4000 0% 75.00 75.0 0.99 74.42 0.0330/04/2002 121.67 6,000 15% 4000 0% 75.00 75.0 0.99 74.23 0.0331/05/2002 152.08 6,000 15% 4000 0% 75.00 75.0 0.99 74.04 0.0330/06/2002 182.50 0 15% 4000 0% 75.00 6000 6,075.0 0.98 5,981.90 0.0331/07/2002 212.92 4000 0% 0.00 0.0 0.98 0.00 0.0331/08/2002 243.33 4000 0% 0.00 0.0 0.98 0.00 0.0330/09/2002 273.75 4000 0% 0.00 0.0 0.98 0.00 0.0331/10/2002 304.17 4000 0% 0.00 0.0 0.97 0.00 0.0330/11/2002 334.58 4000 0% 0.00 0.0 0.97 0.00 0.0331/12/2002 365.00 4000 0% 0.00 0.0 0.97 0.00 0.0331/01/2003 395.42 4000 0% 0.00 0.0 0.97 0.00 0.0328/02/2003 425.83 4000 0% 0.00 0.0 0.96 0.00 0.0331/03/2003 456.25 4000 0% 0.00 0.0 0.96 0.00 0.0330/04/2003 486.67 4000 0% 0.00 0.0 0.96 0.00 0.0331/05/2003 517.08 4000 0% 0.00 0.0 0.96 0.00 0.0330/06/2003 547.50 4000 0% 0.00 0.0 0.95 0.00 0.0331/07/2003 577.92 4000 0% 0.00 0.0 0.95 0.00 0.0331/08/2003 608.33 4000 0% 0.00 0.0 0.95 0.00 0.0330/09/2003 638.75 4000 0% 0.00 0.0 0.95 0.00 0.0331/10/2003 669.17 4000 0% 0.00 0.0 0.94 0.00 0.0330/11/2003 699.58 4000 0% 0.00 0.0 0.94 0.00 0.0331/12/2003 730.00 4000 0% 0.00 0.0 0.94 0.00 0.0331/01/2004 760.42 4000 0% 0.00 0.0 0.94 0.00 0.0329/02/2004 790.83 4000 0% 0.00 0.0 0.94 0.00 0.0331/03/2004 821.25 4000 0% 0.00 0.0 0.93 0.00 0.0330/04/2004 851.67 4000 0% 0.00 0.0 0.93 0.00 0.0331/05/2004 882.08 4000 0% 0.00 0.0 0.93 0.00 0.0330/06/2004 912.50 4000 0% 0.00 0.0 0.93 0.00 0.0331/07/2004 942.92 4000 0% 0.00 0.0 0.92 0.00 0.0331/08/2004 973.33 4000 0% 0.00 0.0 0.92 0.00 0.0330/09/2004 1003.75 4000 0% 0.00 0.0 0.92 0.00 0.0331/10/2004 1034.17 4000 0% 0.00 0.0 0.92 0.00 0.0330/11/2004 1064.58 4000 0% 0.00 0.0 0.91 0.00 0.0331/12/2004 1095.00 4000 0% 0.00 4000 4,000.0 0.91 3,645.99 0.03

450 10,000 450 10,000

At the time of agreement on the contract,the compiled average interest rate is reflectedboth as new business at six months maturityfor an amount of EUR 6,000 and as newbusiness at three years maturity for anamount of EUR 4,000. The MFI interest ratestatistics on outstanding amounts also reflectthe average interest rate based on theminimum guaranteed return on the twocomponents of the deposit at the timereference point. Hence, in this example theinterest rate on outstanding amounts both atsix months maturity for an amount of EUR

6,000 and at three years maturity for anamount of EUR 4,000 reflect the averageinterest rate as compiled above until therespective part of the deposit reachesmaturity.121 A higher than 0% interest ratethat is paid by the reporting agent at maturity

121 Products exist where the interest rate for the second part of thedeposit may have a return above 0% for a certain period of timebefore the deposit reaches maturity, even if at the date thecontract is agreed the minimum guaranteed return is 0%. In thiscase, the average interest rate on outstanding amounts isrecalculated at the time reference point in order to take intoaccount the change in the return for the second part of thedeposit.


of the second part of the deposit is onlyreflected in the interest rate on outstandingamounts for this instrument category atmaturity. As always, the interest rate needsto be annualised, i.e. it has to be taken intoaccount that the interest paid at maturity onthe EUR 4,000 is an one-off payment after 36months.

If there is no capital certainty for the secondcomponent of the deposit, the followingtreatment is applied in MFI interest ratestatistics. The amount placed in the first partof the deposit, in this example the EUR 6,000,is reflected in MFI interest rate statistics, butnot the EUR 4,000 placed in the second part,i.e. the remaining liability. Independent ofwhether or not there is capital certainty forthe second part of the deposit, the interestrate of 15% on the first part of the deposit isadjusted for the return on the second part,as both parts cannot be placedindependently.122 As the actual return on theremaining liability is only known ex post andno minimum return at all is guaranteed,analogous to Chapter 9.6, a convention isneeded. The convention for the purposes ofMFI interest rate statistics is that the returnon the remaining liability is assumed to be0%. Hence, the interest rate on the first partof the deposit is calculated as described aboveas NDER or weighted average of theannualised agreed rate taking into accountthe 15% as well as the assumed 0%. At thetime of agreement of the contract, 3.1371%or 3.2151% are reflected as new business withagreed maturity of six months and an amountof EUR 6,000. The MFI interest rate statisticson outstanding amounts reflect the same ratefor a deposit with agreed maturity of sixmonths and an amount of EUR 6,000 untilthis deposit matures after the six months.

9.8 Purchase of mortgage loans by acredit institution

Question:

A credit institution acquires the economicbut not legal ownership of mortgages that

are granted to households by a (related) lifeinsurance corporation. The mortgages inquestion have different starting dates and aretransferred normally in small, sometimes inlarge, portions from the life insurancecorporation to the credit institution.Originally, the life insurance corporationgrants the mortgage loans. It is in fact sellingmortgages through its network to householdsand these mortgages are funded or refinancedby the credit institution. The amount paid bythe credit institution to the life insurancecorporation is the going market rate at thetime of the purchase of the mortgage loansand includes among others a fee for handlingthe mortgage administration and collectionof payments by the life insurance corporation.The customer pays the interest for themortgage loan directly to the insurancecompany not to the credit institution. In MFIbalance sheet statistics, the credit institutionrecords these (purchased) mortgage loans asmortgage loans extended to households. Isthe transfer of mortgage loans new business?If it is which interest rate should be reportedfor new business and outstanding amounts?

122 This treatment is different to paragraph 80 of Annex II to theRegulation, which reads as follows: ‘Deposits with a maturity ofover two years as defined in Part 3 of Annex I to Regulation (EC)No 2423/2001 (ECB/2001/13) may contain pension savingsaccounts. The main part of pension savings accounts is placed insecurities and the interest rate on the accounts therefore dependson the yield of the underlying securities. The remaining part ofpension savings accounts is held in cash and the interest ratedetermined by the credit or other institution in the same way asfor other deposits. At the time when the deposit is placed, thetotal return to the household from the pension savings account isnot known and may also be negative. Also at the time thedeposit is placed, there is no interest rate agreed between thehousehold and the credit institution or other institution for thepart invested in securities, only for the remaining deposit part.Hence, only the deposit part that is not invested in securitiesshall be covered by MFI interest rate statistics. The annualisedagreed rate on new business that shall be reported is the rateagreed between the household and the reporting agent for thedeposit part at the time the deposit is placed. The annualisedagreed rate on outstanding amounts shall be the rate applied bythe reporting agent to the deposit part of the pension savingsaccounts at the time of calculation of the MFI interest rate.’ Thedifference in treatment is justified because the deposit with twocomponents, where one part is a remaining liability withoutcapital certainty and one part a short-term deposit, gains anunusually high interest rate on the short-term deposit whichcould distort the MFI interest rate statistics.


Answer:

The key for the treatment of such loans inMFI interest rate statistics is their recordingin MFI balance sheet statistics. The mortgageloans are originally granted by the insurancecompany and therefore initially appear on thebooks of the insurance company and not ofthe credit institution. There is in fact nocontractual relationship between thehousehold and the credit institution, onlybetween the household and the insurancecompany. The latter is beyond the scope ofMFI interest rate statistics on new business, asthe insurance company is obviously not acredit institution and therefore not includedin the potential reporting population123 of theRegulation.

Once the credit institution purchases themortgage loans, however, a loan tohouseholds for house purchases is recordedin the books of the credit institution. At thispoint in time, the interest rate actually paidby the household for the mortgage loan isrecorded in the MFI interest rate statistics onoutstanding amounts. The purchase of the loansby the credit institution does not constitutenew business. As these mortgage loans arecovered in MFI balance sheet statistics underloans to households for house purchases, theyare also included in the weighting informationwhich is used to calculate the euro areainterest rate for outstanding housing loans tohouseholds.

9.9 Securitisation of mortgage loans bya credit institution

Question:

A credit institution transfers the economicbut not legal ownership of mortgage loansthat it has granted to households to a financialvehicle corporation. Analogous to Chapter

9.8, the mortgages in question have differentstarting dates and are transferred normally insmall, sometimes in large portions from thecredit institution to the financial vehiclecorporation. Originally, the credit institutiongrants the mortgage loans and records themas loans to households for house purchasesin MFI balance sheet statistics. Once themortgage loans are sold to the financialvehicle corporation, the loans no longerappear in the MFI balance sheet of the creditinstitution. How should the securitisation betreated in MFI interest rate statistics?

Answer:

As in Chapter 9.8, the key for the treatmentof securitisation in MFI interest rate statisticsis the recording in MFI balance sheet statistics.The mortgage loans are originally granted bythe credit institution to the household andconstitute new business at the time ofagreement on the contracts. They are hencereflected as loans to households for housepurchases in the MFI interest rate statisticson new business and also appear at that timein MFI interest rate statistics on outstandingamounts.

However, once the credit institution sells themortgage loans to the financial vehiclecorporation, the loans disappear from thebooks of the credit institution. At that pointin time, they are removed from the MFIbalance sheet of this institution and also nolonger covered in MFI interest rate statisticson outstanding amounts. The transfer of theloans to the financial vehicle corporation doesnot constitute new business for MFI interestrate statistics.

123 Defined in Chapter 3 and further discussed in Chapter 11.1.


10 Aggregation of the data and reporting obligations124

10.1 Overview

To derive euro area aggregates for each ofthe instrument categories in Figure 18 andFigure 20, three levels of aggregation arenecessary: the first at the level of thereporting agents, the second at that of theNCBs and at the third at the ECB level. Thethree levels are illustrated in Figure 25.

The starting point for the compilation of MFIinterest rate statistics is the individual depositand loan products within the reporting agents.As a first step, each reporting agent collectsdata on all relevant products, aggregates themas appropriate and sends them to the NCB ofthe Member State in which it is resident. Ingeneral and as indicated in Figure 25,125 eachreporting agent provides for each instrumentcategory one average interest rate referringto the credit institution or other institutionas a whole. In the case of new business it alsoprovides the volume of the new deposits andloans.

As a second step in aggregation, each NCBcompiles for each instrument category aweighted average interest rate for eachinstrument category referring to the MemberState as a whole and reports it to the ECB. Inthe case of new business, the NCB alsoreports the volume of the new contracts atnational level.

The ECB carries out the third and last step ofaggregation by compiling for each instrument

Figure 25

Deposit and

loan

products

Instrument

categories

per country

Instrument

categories

at euro area

level

Interest rate

and volume per

instrument

category per

reporting agent

Interest rate

and volume per

instrument

category per

country

Interest rate

and volume per

product within

reporting agent

Instrument

categories per

reporting

agent

category a weighted average interest ratereferring to the euro area as a whole. Thereporting obligations are discussed in moredetail in the following chapters.

10.2 Statistical information at the level ofthe reporting agents126

Each reporting agent needs to submit thedata collected for MFI interest rate statisticsto the NCB of the Member State in which itis resident.

As explained in Chapter 6.1, MFI interestrates on outstanding amounts, i.e. indicators 1to 14 in Figure 18, may be compiled as asnapshot of end-month observations or asimplicit rates referring to the average overthe month. The same applies to overnightdeposits, deposits redeemable at notice and bankoverdrafts, i.e. indicators 1, 5, 6, 7, 12 and 23in Figure 19 or Figure 20. It is up to theNCBs to decide which of the two methods isbetter suited to the national context. Thedata that reporting agents need to provide tothe NCB depend on the chosen method:

• In the case of a snapshot of end-monthobservations, the reporting agents need to

124 This chapter refers mainly to Part 5 of Annex II to the Regulation.125 NCBs may also choose the compilation of implicit rates (see

Chapter 6.1) or require reporting agents to provide data on thelevel of individual deposit and loans. In both cases the reportingobligations differ from those illustrated in Figure 25.

126 See also paragraphs 59 to 65 of Annex II to the Regulation.


provide for each instrument category aweighted average interest rate referring tothe last day of the month. The weightedaverage interest rate covers all outstandingdeposits that have been placed and not yetwithdrawn by customers or all outstandingloans that have been withdrawn and notyet repaid by customers in all the periodsup to and including the reporting date.127

• In the case of implicit rates referring to theaverage of the month, reporting agentsprovide for each instrument category theaccrued interest payable or receivableduring the month and the stock of depositsand loans on average during the samemonth.

In both cases and hence independent of thechosen method, reporting agents need toprovide for bank overdrafts (i.e. indicators 12and 23 in Figure 20) the outstanding amountat the end of the month.

For each instrument category on new business,i.e. indicators 2 to 4, 8 to 11, 13 to 22, and24 to 31 in Figure 20 and Figure 21, allreporting agents need to provide a weightedaverage interest rate. This weighted averageinterest rate refers to all interest rates onnew business operations in the instrumentcategory during the entire reference month.128

In addition, they need to report for indicators2 to 4, 8 to 11, 13 to 22, and 24 to 29 inFigure 20 the amount of new businessconducted during the month. Reportingagents do not need to provide the amount ofnew business for loans to households forconsumption and for house purchasecompiled as APRC, i.e. indicators 30 and 31in Figure 21. These amounts can be derivedas aggregates from the more detailed datathat is provided for the other new businessrates.

Instead of asking reporting agents to provideweighted average interest rates on newbusiness and the corresponding volumereferring to the institution as a whole, NCBsmay also request data at the level of individualdeposits and loans.

NCBs may allow credit institutions and otherinstitutions which are resident in a singlenational territory and individually included inthe list of MFIs to report MFI interest ratestatistics together as a group.129 The groupthen becomes a notional reporting agent andhas the same reporting requirements as theother (individual) credit institutions and otherinstitutions that are reporting agents. So theyhave to provide the data on outstandingamounts and new business as defined in thischapter referring to the group as a whole. Inaddition they need to report every year foreach instrument category the number ofreporting institutions in the group and thevariance of interest rates across theseinstitutions. The number of reportinginstitutions and the variance should refer tothe month of October and be transmittedwith the October data.

10.3 National weighted average interestrates130

NCBs receive the data from the reportingagents and aggregate them to MFI interestrate statistics at the national level. These dataare then submitted to the ECB:

• For each instrument category onoutstanding amounts, i.e. indicators 1 to 14in Figure 18, and for each instrumentcategory on new business, i.e. indicators 1to 31 in Figure 20 and Figure 21, eachNCB provides a weighted average interestrate referring to the Member State as awhole.

• In addition, each NCB provides for each ofthe new business indicators 2 to 4 and 8 to29 in Figure 20 the amount of new businessconducted at national level in eachinstrument category during the referencemonth. These amounts of new businessrefer to the population total, i.e. to the

127 Bad loans and loans for debt restructuring at rates below marketconditions are excluded. See also Chapters 5.1 and 5.2.1.

128 See also Chapter 6.2.129 Further discussed in Chapter 11.5.130 See also paragraphs 66 to 71 of Annex II to the Regulation.


entire potential reporting population, notonly to the sample, i.e. the actual reportingpopulation.131 Therefore, where a samplingapproach is chosen at national level forselecting the reporting agents, the NCBmust apply expansion factors at nationallevel to derive the population total for theamount of new business.132 Thecomputation of selection probabilities andtheir use as expansion factors for derivingpopulation totals is explained in genericterms in Chapter 11.7.

The NCBs provide the interest rates onoutstanding amounts and on new business tothe ECB with a detail of four decimal places,i.e. 12.3456%. The level of detail at which theNCBs wish to collect the data from thereporting agents is up to them to define. Thepublished ECB results do not contain morethan two decimal places.

Together with the national data, NCBs needto provide methodological notes133 thatdocument any important special nationalpractices, comprising regulatoryarrangements, national conventions,institutional arrangements and specificproducts affecting MFI interest rate statistics.This information is essential in order tointerpret the level and the development ofthe deposit and lending rates set by MFIs.

NCBs carrying out a sampling approach forthe selection of the reporting agents alsoprovide an estimate of the sampling error forthe initial sample. A new estimate is requiredafter each maintenance134 of the sample. Thesampling error depends on the nationalsampling method used, i.e. the stratification,the sample size and its allocation, and theway the reporting agents are selected in eachstratum.

10.4 Aggregated results for the euroarea135

The ECB carries out the final level ofaggregation of the instrument categories foreach participating Member State to euro area

MFI interest rates statistics. It compilesweighted average interest rates for eachinstrument category referring to the euroarea as a whole:

• In the case of outstanding amounts, i.e.indicators 1 to 14 in Figure 18, andovernight deposits and deposits redeemableat notice, i.e. indicators 1, 5, 6 and 7 inFigure 20, the weighting information isderived from the MFI balance sheetstatistics. The interest rates on outstandingamounts compiled as a snapshot of end-month observations have the same timereference point as MFI balance sheetstatistics, whereas implicit rates refer tothe average over the month. Independentof the compilation procedure at nationallevel, the ECB uses the data on the size ofeach balance sheet item at the end of themonth in each Member State to calculatefrom the national interest rates weightedaverages for the euro area.

• In the case of new business and bankoverdrafts, i.e. indicators 2 to 4 and 8 to 31in Figure 20 and Figure 21, the weightinginformation is provided by the NCBs,which collect the data from the reportingagents.

The system of interest rates and weightinginformation is fully additive to allowcombinations of rates such as syntheticdeposit and lending rates.

131 Potential and actual reporting population is defined in Chapter 3and further discussed in Chapter 11.1.

132 No expansion factors are required for the weighted averageinterest rates where it is assumed that the estimate from thesample is the estimate for the entire potential reportingpopulation.

133 Their content is further discussed in Chapter 4.5.2.134 Further discussed in Chapter 11.8.135 See also paragraph 72 of Annex II to the Regulation. Further

discussed in Chapter 11.2.


11 Selection of the reporting agents136

11.1 Selection of the actual reportingpopulation137

MFI interest rate statistics provide for theeuro area as a whole and individually for eachMember State detailed information about theinterest rates that resident credit institutionsand other institutions apply to euro-denominated deposits and loans vis-à-vishouseholds and non-financial corporationsresident in the euro area. In each MemberState, the reporting agents for MFI interestrate statistics are selected by the respectiveNCB. Each NCB has the choice of declaringall or only a subset of resident creditinstitutions and other institutions as reportingagents.

The starting point for the selection of thereporting agents is the potential reportingpopulation, which provides the referencepopulation and sampling frame. For each NCB,the potential reporting population comprisesall resident credit institutions and otherinstitutions which take euro-denominated

Figure 26Decision tree

Minimum

sample size

Reporting

requirements

Potential reporting population (reference population or sampling frame)

Census Sample

Stratification of the potential

reporting population

Random sampling

within each

stratum

Selection of the

biggest institutions

within each stratum

Actual reporting population

Maintenance of the sample over time

deposits from and/or grant euro-denominatedloans to households and/or non-financialcorporations resident in the participatingMember States.138 In order to select from thepotential reporting population the actualreporting population, i.e. the reporting agentsfor MFI interest rate statistics, NCBs followthe procedure defined in Annex I to theRegulation, which is outlined as a ‘decisiontree’ in Figure 26.

An NCB first has the choice between a censusand a sample. In the case of a census, theNCB requires each resident credit institutionor other institution in the potential reportingpopulation to report MFI interest ratestatistics. In the case of a sample, the NCBstratifies the potential reporting population,which means that the entire referencepopulation is subdivided into sub-populations

136 This chapter refers mainly to Annex I to the Regulation.137 See also Article 2 of the Regulation.138 Defined in Article 1(5) of the Regulation. See also Chapter 3.


including institutions that as a result of theirlimited banking activity are of minorsignificance for MFI balance sheet statistics,i.e. so-called ‘tail institutions’. After thestratification, the NCB ideally selects thereporting agents at random from eachstratum. The alternative to drawing at randomis to select the largest institutions within eachstratum. In order to keep the samplerepresentative over time, the NCB needs torefresh the sample with joiners and adjust itfor leavers and other changes in thecharacteristics of the reporting agents. Toensure the quality of the output, theRegulation defines a minimum size for thenational sample.

The sampling approach aims to reduce thereporting burden on the whole bankingsector, as only some credit institutions andother institutions have to report, but mightrequire some small institutions to report MFIinterest rate statistics to ensure arepresentative range of results. Smallspecialised institutions might offer differentinterest rates than big universal banks.Following national procedures, the NCBneeds to inform the selected reporting agentsabout their reporting obligations.

11.2 Census versus sampling approach139

As stated in Chapter 11.1, NCBs first havethe choice between a census and a sample. Inthe case of a census, the NCB asks all creditinstitutions or other institutions in thatMember State to report MFI interest ratestatistics. In other words, the actual reportingpopulation is identical to the potentialreporting population. In the case of a sample,only a selection of the reference populationis asked to report, which implies that theactual reporting population is smaller thanthe potential reporting population. In general,the advantages of a sample survey ascompared to a census are as follows:

• Costs: A sample is less costly than a census,since data are collected only from a subsetof the potential reporting population.140

The total costs depend on the size of thesample, which is determined by the desireddegree of precision141 of the results.

• Timeliness: A sample gives more timelyresults than a census. Sampling limits thenumber of reporting agents and thereforethe number of reports that the NCBs needto collect, check and process.

• Accuracy:142 Sampling procedures may leadto higher quality data. The smaller volumeof work (fewer reports), make possiblemore careful guidance and monitoring ofthe reporting institutions, and more carefulchecking of the responses and dataprocessing. In general, sampling procedurestend to reduce measurement errors143 insurveys resulting from coding, editing,processing, non-response, incorrectanswers, etc.

The disadvantages of sampling procedures ascompared to a census are partly the costs forsetting up and maintaining the sample, butabove all the uncertainty resulting from thefact that only a part of the potential reportingpopulation is being observed. The resultsderived from a sample, i.e. from the actualreporting population, might therefore differfrom the true (unknown) values in thepotential reporting population. The errorsthat may occur because sampling proceduresare used are referred to as sampling errors.141

139 See also paragraphs 2 to 5 of Annex I to the Regulation.140 Council Regulation (EC) No 2533/98 of 23 November 1998

concerning the collection of statistical information by theEuropean Central Bank, OJ L 318, 27.11.1998, p. 8, specifies inArticle 3 that ‘without prejudice to the fulfilment of its statisticalreporting requirements, the ECB shall minimise the reportingburden involved’ and ‘may fully or partly exempt specific classesof reporting agents from its statistical reporting requirements’.The ESCB-internal ‘merits and costs exercise’ formalises theprocedures to be followed for justifying new statisticalrequirements. One step of this procedure is the assessment ofthe collection and compilation costs borne by reporting agents,NCBs and the ECB.

141 Variance and standard error, also referred to as sampling error,are measures for the precision of the sampling procedure:

( ) ( )[ ]2nnnÊÊˆVar θ−θ=θ

142 The mean square error aims to quantify the accuracy of theestimator:

( ) ( ) ( ) ( )[ ] BiasVarianceÊˆVarÊˆMSE2

nnn

2

nn +=θ−θ+θ=θ−θ=θ

143 Measurement errors are sometimes referred to as ‘non-samplingerrors’.


In contrast to measurement errors, whichmay occur both in a sample and in a census,sampling errors only occur in samples.

As explained in Chapter 10.4, the ECBperforms the final stage in aggregating theinstrument categories for each Member Stateto euro area results. To do so, the ECBrequires both interest rates and weights, i.e.the deposit and loan amounts per instrumentcategory and Member State. Whereas foroutstanding amounts the weights can bederived from the MFI balance sheet statistics,for new business the weights are based ondata collected from the actual reportingpopulation. Hence, in the case of a censusand a sample the interest rates on outstandingamounts and on new business, as well as theamount of new business, need to be collectedfrom the reporting agents. This implies that,in the case of a sample, not only the interestrates on outstanding amounts and newbusiness but also the amounts of new businessare sampling variables, i.e. variables measuredwith certainty after a random selection of thereporting agents. Sampling variables areestimates and therefore subject to samplingerrors.

To minimise the risk that the results of asample survey deviate from the true(unknown) value in the potential reportingpopulation, the aim is to compile arepresentative sample of credit institutionsand other institutions. A sample is consideredrepresentative of the potential reportingpopulation if it comprises all thecharacteristics of the institutions that arerelevant for MFI interest rate statistics andfound in the potential reporting population.In other words, the sample, i.e. the actualreporting population, should reflect therelevant characteristics of all creditinstitutions and other institutions in thepotential reporting population.

There are two main ways of creating a sample:

• In the case of non-random or purposivesampling, a number of typical units areselected from the potential reporting

population. A unit is typical if it conformsclosely, in the sampler’s view, to thecharacteristics of the potential reportingpopulation. The probability of selecting anyparticular unit cannot be quantified andtherefore the computation of the precisionof the sample is impossible.

• In the case of random sampling, each unit inthe population has a chance of beingselected in the sample and this selectionprobability can be quantified.

The advantages of random sampling overpurposive sampling are as follows:

• The reporting agents are selected bychance, i.e. objectively, and not subject tothe sampler’s view.

• The standard error and hence the precisionof the sampling procedure can bequantified.

• The degree of accuracy of the estimatesfrom the sample can be estimated.

• The optimal sample size can be defined,which is further discussed in 11.4.

• Selection probabilities and henceexpansions factors can be defined, which isfurther discussed in Chapter 11.7.

11.3 Stratification of the potentialreporting population144

To ensure that the sample of creditinstitutions and other institutions isrepresentative of the potential reportingpopulation, the Regulation requires that if aMember State decides on a sample, then thepotential reporting population needs to bestratified before any reporting agents areselected. Stratification implies that thepotential reporting population N is subdividedinto sub-populations or strata N1, N2, N3, …,

144 See also paragraphs 6 to 9 of Annex I to the Regulation.


NL. These are not overlapping and togethercomprise the entire reference population:

N1 + N2 + N3 + … + NL = N[Equation 11]

The advantage of stratification and hence ofstratified random sampling is that ‘extreme’samples, i.e. samples which are notrepresentative of the potential reportingpopulation, can be avoided, resulting in alower sampling error. For example, in thecase of MFI interest rate statistics, a samplecontaining only institutions specialising inhousing loans would be an extreme samplethat did not represent all institutions in thepotential reporting population. A furtheradvantage of stratified random sampling isthat, in addition to the estimated valuesreferring to the entire sample, informationcan also be gathered about each stratum.

To avoid extreme samples, first theheterogeneous potential reporting populationis divided into homogeneous strata. Next,from each stratum a certain number ofreporting agents are drawn, which togetherconstitute the sample. By selecting reportingagents from each of the homogeneous strata,the sample becomes heterogeneous and istherefore representative of theheterogeneous potential reporting population.The heterogeneity refers to the variance ofthe sampling variables.

A stratum is homogeneous if themeasurements for the sampling variables, i.e.the interest rates on outstanding amountsand the interest rates and amounts of newbusiness, vary little from one institution tothe other. For example, in the case of MFIinterest rate statistics, a stratum containingonly institutions specialising in housing loanscould be a homogeneous stratum, as interestrates might be similar. The result of suchhomogeneous strata is that the variance ofthe sampling variables within each stratum,i.e. the intra-stratum variance, is lower thanthe variance between strata, i.e. the extra-stratum variance. The decomposition of thetotal variance into intra-stratum and an extra-

stratum variance is also known as varianceanalysis or Huygens theorem. The univariateequation145 is as follows:

[Equation 12]

More generally:

Meaning of letters and symbols:Var(y) Total variance of y in the potential

reporting population, with y as thesampling variable

Varwithin(y) Variance of y within the strata(intra-stratum variance)

Varbetween(y)Variance of y between the strata(extra-stratum variance)

N Number of institutions in thepotential reporting population

Nh Number of institutions in stratumh, with h = 1, …, L

yh

Average y in stratum hy Average y in the potential

reporting populationwi Weight of the individual institution iwh Weight of the stratum h

145 The same kind of relationship exists for multiple variables. Invectorial notation it reads as follows:

Meaning of letters and symbols:wi Weight of institution iwh Weight of the stratum hyi Vector of observations for institution igh Vector which is the barycentre (‘centre of gravity’) of thestratum hg Vector which is the barycentre of the whole set of data

• A kind of distance: Generally, for numerical data, the

family of distances of Minkowski is applied, where the distancebetween two items, e.g. institutions or barycentres is computedwith the following formula, where i denotes an institution and j avariable:

.

The Euclidean distance is given with 1 and 2 j =α=λ .

( ) ( )

( ) ( ) ( )

( ) ( ) ( )

( ) ( ) ( )∑ ∑∑

∑ ∑∑

∑ ∑

= ==

= ==

= =

−+−=

−+−∗=

−+=

+=

L

1h

L

1h

2hh

N

1i

2hi

L

1h

L

1h

2hh

N

1i

2hi

hh

L

1h

L

1h

2hhhh

betweenwithin

yyNN1yy

N1yVar

yyNN1yy

N1N

N1yVar

yyNN1yVarN

N1yVar

yVaryVar)y(Var

h

h

∑∑∑ −+−=∈ h

2hh

h hi

2hii ggwgywInertia

λ

=

λ′′′

−α==− ∑1

p

1jjiijjiiii yy)y;d(yyy

( ) ( ) ( )∑ ∑∑= ==

−+−=L

1h

L

1h

2hh

N

1i

2hii yywyywyVar

h


Stratification requires that suitablestratification criteria are defined, which allowthe subdivision of all credit institutions andother institutions into homogeneous strata.Information on the stratification criteria mustbe available for each institution in thepotential reporting population. Thestratification criteria must relate to thepurpose of the survey, i.e. to the samplingvariables that are to be estimated from thesample. Sampling variables for MFI interestrate statistics are the interest rates onoutstanding amounts and on new business aswell as the amount of new business.146 Suitablestratification criteria can be derived, forexample, from the MFI balance sheetstatistics, national surveys conducted forsupervisory purposes, national retail interestrate statistics, or the list of MFIs.

If the institutions within each stratum showlittle variance in terms of the stratificationcriteria, and if there is a strong relationshipbetween the stratification criteria and thesampling variables, then the institutions withineach stratum are also likely to show littlevariance in interest rates and amounts of newbusiness. If a small number of creditinstitutions and other institutions is selectedfrom one stratum, the interest rates and theamounts of new business collected from theseinstitutions can be assumed to berepresentative of that stratum. Combining thedata from the different strata should giveresults that are representative of the potentialreporting population.

Within each Member State at least onestratification criterion is required as a minimumstandard by the Regulation. The aim is toensure that the sample of credit institutionsand other institutions is representative of thisMember State’s potential reporting populationand the standard error small. Ideally eachMember State defines a hierarchy ofstratification criteria. These should take intoaccount the national circumstances and hencebe specific to each Member State.

In many countries the banking business withhouseholds and non-financial corporations is

highly concentrated. The organisationalstructure of the national banking business mayamplify this effect. Some credit institutionsmight have a legal organisation where onlythe ‘head office’ is included in the list of MFIswhereas all the regional offices are treated as‘branches’. In this case the head office submitsone statistical report covering the wholeorganisation. In other banking groups, eachregional office might be organised as anindependent credit institution and as suchincluded in the list of MFIs. In this case, onestatistical report covers only the business ofthe reporting entity. Concentration in thebanking business and organisational differencesneed to be taken into account when designingthe sample. It might be necessary to workwith unequal probabilities for the selection ofthe reporting agents.147

Bigger countries might consider the region inwhich the credit institution or otherinstitution is located as a stratificationcriterion. Without aiming at the compilationof regional statistics, which goes beyond theuser requirements, regional (ex ante)stratification148 reduces the sampling errorwhere regional differences in interest ratesor in the type of customer exist. However,depending on the organisational structure,regional differences might be apparent atbranch level rather than at the level of creditinstitutions. Regional differences are capturedin interest rates and the new businessamounts, if the head office provides datacovering all branches. The alternative is todraw a sample at branch level.149 In the lattercase, the same minimum standards as for thesampling of credit institutions and otherinstitutions apply.

For the construction of the strata, the use ofquantitative data analysis techniques, such as

146 See also Chapter 11.2.147 Further discussed in Chapter 11.6.148 Stratification according to region which is decided by the sample

before any application of quantitative data analysis andregrouping techniques.

149 Sampling at branch level is allowed under certain conditions anddiscussed in Chapter 11.6. It is carried out without anyintermediate drawing, i.e. without sampling (in a first step) creditinstitutions and (in a second step) branches of the selectedinstitutions; this is further discussed at the end of this chapter.


principal component or factor analysis, andof regrouping techniques, such as clusteranalysis, is recommended. These statisticalmethods help to allocate the institutionsaccording to their statistical proximity, withsimilar ones allocated to the same stratumand dissimilar ones to different strata.

Since MFI interest rate statistics are aimed atproviding data on the level and developmentof interest rates both at euro area and atnational level, the country of residence of thecredit institutions and other institutions ischosen as the first ‘natural’ stratificationcriterion. Geographical ex ante stratificationgives NCBs flexibility to choose within theframework set by the Regulation the mostsuitable procedure for selection the reportingagents. Furthermore, it enables the nationalsamples or census procedures to be combinedinto a euro area sample ensuring reliablestatistics at euro area level and national level.In practice, the data are collected by NCBsfrom reporting agents at national level andthen aggregated to euro area results. Thestratification at euro area level can beillustrated in a schematic way150 as follows:

Figure 27

Potential stratification criteria

Member States Region Size of institution Type of business

Resulting strata

A1

A2

A3

Member State A

A4

B1

B2

B3

B4

B5

Member State B

B6

C1 Member State C

C2

D1

D2

D3

Member State D

D4

etc.

The random selection of the reporting agentstakes place after all strata are defined. Onlyat this stage, indicated as shaded cells inFigure 27, are credit institutions drawn fromthe sampling frame. Since there is nointermediate drawing, for example at the levelof regions or by the size of the institutions,the approach is referred to as single-stagesampling. If there were an intermediatedrawing, the sampling would occur in severalstages (multi-stage sampling) which wouldrequire more complex formulae and increasethe variance of the estimator.

11.4 Minimum national sample size151

11.4.1 Definition

The sample of reporting agents for MFIinterest rate statistics has to be of a size thatensures reliable euro area and nationalstatistics simultaneously. Trying to achieve

150 A brief description of the actual euro area sample for MFIinterest rate statistics is given in Chapter 11.10.

151 See also paragraphs 10 to 14 of Annex I to the Regulation.


reliable results for only the euro area couldlead to the situation that small countries inparticular have too few reporting agents tocompile a national set of statistics, whichwould conflict with the aim of providinginformation about the level and developmentof interest rates both at euro area and atnational level. So for the purposes of MFIinterest rate statistics the minimum nationalsample size is defined. The aggregate of thenational samples is then big enough to derivereliable data on the level and development ofinterest rates in the euro area.

The Regulation gives flexibility to NCBs whileensuring that the national results arecomparable and both the national and euroarea results are of high quality. It defines thatthe minimum national sample size shall be suchthat the maximum random error

( ) ( )θ∗≈θ∗= ααˆravzˆvarzD 2/2/

for interest rates on new business on averageover all instrument categories does notexceed 10 basis points152 at a confidence levelof 90%, where:

zα/2 Factor computed from the normaldistribution or any suitable distributionaccording to the structure of the data(e.g. t-distribution) assuming aconfidence level of 1-α,

( )θvar Variance of the estimator of parameterθ, and

( )θrav Estimated variance of the estimator ofparameter θ.

NCBs must prove compliance directly withappropriate data. Alternatively, in the absenceof such data, if they fulfil one or the other ofthe following criteria, the sample size may bepresumed large enough to meet the minimumrequirement.

(a) The minimum national sample size shallbe such that it covers at least 30% of theresident potential reporting population;where 30% of the resident potentialreporting population is greater than 100,

the minimum national sample size maynevertheless be limited to 100 reportingagents.

(b) The minimum national sample size shallbe such that the reporting agents in thenational sample cover at least 75% of thestock of euro-denominated depositsreceived from and at least 75% of thestock of euro-denominated loans grantedto households and non-financialcorporations resident in the participatingMember States.

As part of the preparation for defining theminimum national sample size, the ECBcarried out a series of simulations to derive atheoretical minimum sample size at nationallevel. In the simulations the sample size isbased on the acceptable maximum randomerror, the desired confidence level for theresults and the variance of the interest rates.The statistical details are given in Chapters11.4.2 to 11.4.4.

The data that was available to the ECB forthese simulations are of course not theharmonised MFI interest rate statistics thatare in fact the aim of the survey for which thesample size is being defined. Instead the ECBused data provided by NCBs from theircurrent national statistics on retail interestrates or from surveys conducted forsupervisory purposes. These national data arenot harmonised at euro area level. As a resultof the limited availability and comparability ofthese data, the definition of the minimumsample size could not be stringently based onthe simulation results. Deviations from thepure sampling theory were necessary. Hence,the definition takes into account, in ajudgmental way, any weaknesses in theassumptions and drawbacks of themathematical approach, tries to achieve abalance across countries and instrumentcategories and considers the reporting burdenon the banking sector imposed by the survey.

152 The absolute measure of 10 basis points at a confidence level of90% may be directly translated into a relative measure in termsof the acceptable maximum variation coefficient of the estimator.Further discussed in Chapter 11.4.2.


Paragraph 10 of Annex I to the Regulationgives priority to the definition of the minimumnational sample size that is closest to thetheoretical ideal, i.e. a size defined in termsof acceptable absolute or relative measure ofprecision. Compliance with this definitionrequires that appropriate data are available,i.e. data that are sufficiently detailed andderived from surveys that apply definitionsthat are sufficiently close to the concepts ofMFI interest rate statistics. These data areunlikely to be available to NCBs before thesurvey on MFI interest rate statistics isimplemented and reporting agents provide thefirst sets of data. Hence, the Regulation givestwo alternative and more pragmaticdefinitions of the minimum national samplesize, i.e. a coverage of 30% of the potentialreporting population or a coverage of 75% ofthe relevant business. The simulations by theECB have shown that in general thesepragmatic definitions lead to a sample sizethat ensures a maximum error for interestrates on new business on average over allinstrument categories that does not exceed10 basis points.

Once appropriate data are available, NCBsmay reassess their sample. They may reducethe sample size, if less than 30% of thepotential reporting population or less than75% of total euro-denominated deposits andloans lead to a maximum random error ofnot more than 10 basis points at a confidencelevel of 90%. Analogously, after a transitionalperiod153 they must increase the sample size,if 30% of the potential reporting populationor 75% of total euro-denominated depositsand loans would lead to a maximum randomerror of more than 10 basis points at aconfidence level of 90%. When appropriatedata are available at national level, the extentto which the national sample meets therequirement of a maximum random error of10 basis points at a confidence level of 90%may be shown by applying a formula which ismore targeted to the national samplingprocedure than Equation 18 in Chapter11.4.2. Therefore, at national level NCBsshould take into account in particular thegain in precision due to stratification, and

should also consider, where applicable, theselection of reporting agents with probabilityproportional to size.

In any case many NCBs select more reportingagents than defined as the minimum nationalsample size, i.e. a larger sample, which lowersthe national sampling error for that MemberState. In particular, an NCB might have toselect more reporting agents than the numberdefined as the minimum national sample size,because it may need to make the nationalsample more representative in the light ofthe structure of the national banking system.NCBs may also choose to apply a census. Theexhaustiveness of the census makes theestimator ‘perfect’ for the Member State inquestion. The variance of the estimator ineach stratum for that Member State, andhence the sampling error, is by definitionzero. This reduction in variance is attributableonly to the uncertainties connected with theselection and the sampling plan. Owing to theexhaustiveness of the statistics, themeasurement errors154 might increase sinceless time can be devoted to each report andreporting agent. Yet these measurementerrors are difficult to quantify.

If national samples comply with the definedminimum sample size or if NCBs select morereporting agents than the strict minimum,then the aggregate of these national samplesis large enough to derive reliable euro arearesults. Defining a minimum national samplesize thus ensures high quality for euro areadata and, in general, for the analysis byindividual Member State.

The minimum national sample size defined inthe Regulation refers to the initial sample andthe sample after maintenance. Therefore the

153 Annex IV to the Regulation defines for the sample a transitionalperiod up to the reference month of December 2006. Until thenNCBs that cover 30% of the potential reporting population or75% of total euro-denominated deposits and loans do not needto increase their sample even if that sample size leads to amaximum random error of more than 10 basis points at aconfidence level of 90%. The reason is that experience needs tobe gained with this completely new and complex survey andstability of the reporting population should be ensured for thefirst introductory years.



sample drawn at the beginning of the surveyneeds to be of the defined size. Due to theeffect of mergers and leavers the sample mightreduce in size over time, but needs to berefreshed at least at the next maintenanceperiod.155 When selecting new reportingagents, NCBs need to consider that they willrequire some time to implement the reportingsystem for MFI interest rate statistics. Henceit needs to be accepted that betweenmaintenance periods the sample might dropbelow the threshold given by the minimumnational sample size.

11.4.2 Theoretical background

The theoretical model for deriving theminimum sample size is based on simplerandom sampling without replacement,156 i.e.a sample where each unit can only be drawnonce. For reasons of simplification it isassumed that the average interest rate y iscalculated as a simple arithmetic mean insteadof a weighted average:157

, with[Equation 13]

The variance of this average interest rate yacross the institutions i in the simple randomsample without replacement is:

[Equation 14]

Since for MFI interest rate statistics theinterest rates Yi and Y in the potentialreporting population and hence the truevariance 2

yS are unknown, the variance of theaverage interest rate y across the institutionsi is estimated as follows:

[Equation 15]


yi

Interest rate offered by institution iin the sample

y Average interest rate across allinstitutions i in the sample

Yi

Interest rate offered by institution iin the potential reporting population

Y True (unknown) average interest ratein the potential reporting population

n Number of institutions in the sample

N Number of institutions in thepotential reporting population

E Expected value

True (unknown) variance of interestrates in the potential reportingpopulation

Observed variance of the interestrates across all institutions i in thesample

True (unknown) variance of theaverage interest rate in the sample

Estimated variance of the averageinterest rate in the sample

To derive the minimum national sample size n, aconfidence interval is defined. The sample size nshould be such that with a confidence level of1-α, the average interest rate y estimated bymeans of the sample lays within a certain rangearound the true (unknown) average interest ratein the potential reporting population Y. Thisrange should not be wider than plus or minusthe maximum random error D:

[Equation 16]

As explained above, the true variance of the

average interest rate in the sample )yvar( is

155 Further discussed in Chapter 11.8.156 In a simple random sample with replacement, the estimated

variance would be calculated without the finite populationcorrection (N-n)/n.

157 It is important to note that in the case of a weighted average,the variance of a ratio would have to be calculated.

( ) YyE =

( ){ }[ ]

( )∑=

−−

=

−=−=

N

1i

2i

2y

2y2

YY1N

1S thiw

NnN

nS

yEyE)yvar(

( )Nnf and yy

1n1s thiw

)f1(ns

NnN

ns

)yr(av

n

1i

2i

2y

2y

2y

=−−

=

−=

−=

∑=

∑=

=n

1iiy

n1y

)yvar(*zD with1)DyYDy(P

1)DyY(P

2/α=α−=+≤≤−

α−=≤−

)yvar(

)yr(av

2yS

2ys


unknown for MFI interest rate statistics. It isinstead estimated from national surveys forretail interest rate statistics, from datacollected for supervisory purposes, and othersources. Hence the maximum random errorD is:

[Equation 17]

Solving for n provides the formula for thesimulations to derive the minimum nationalsample size per instrument category used bythe ECB:

[Equation 18]

If the potential reporting population is verylarge and in the case of sampling withreplacement, which is equal to drawing from

an infinite population, then N

sz 2y2/

2 ∗α goes to

zero and the formula can be simplified to:

[Equation 19]

The Regulation requires that the maximumrandom error D does not exceed 10 basispoints at a confidence level of 90%. In otherwords, if all samples of size n were drawnfrom the potential reporting population, thenin 90% of the cases the estimated averageinterest rate in the sample should not deviatemore than 10 basis points from the trueaverage interest rate in the potentialreporting population.

The maximum random error D defined as anabsolute measure of error of 10 basis points isused in Equations 16 to 19. For example, areference population comprises N = 120institutions. The observed average interest

2y2/2/ s

nf1z)yr(avzD ∗−∗=∗= αα

2y2/

22

2y2/

22

2y2/

22

s)N1

n1(zD

sn

Nn1

zD

sn

f1zD

∗−∗=

∗−

∗=

∗−∗=

α

α

α

Nsz

D

szn 2

y2/2

2

2y2/

2

∗+

∗=⇒

α

α

2

2y2/

2

Dsz

n∗

=α

rate is 5% with a variance of 2ys = 0.4. At a

given confidence level, zα/2 can be derivedfrom statistical tables: given a confidence level1-α of 95% then zα/2 = 1.96 and given aconfidence level 1-α of 90% then zα/2 = 1.645.It follows from Equation 18 that the minimumnational sample size needs to comprise n =57 reporting agents. The problem is thathigher interest rates usually have a highervariance and would therefore require ceterisparibus a bigger sample size. If for the samepopulation the average interest rate was 10%with a variance of 2

ys = 1.6, then the minimumnational sample size would have to covern = 94 reporting agents.

Differences in the level and variance ofinterest rates across instrument categoriesexist in each Member State. Hence, in orderto define the minimum national sample sizefor MFI interest rate statistics, an ‘average’size over all instrument categories needs tobe taken. Furthermore, in order to overcomethe (potential) problem of increasing samplesize with increasing interest rates, theRegulation allows the translation of theabsolute measure of error D into a relativemeasure of error in terms of an acceptablemaximum variation coefficient CV of theestimator:158

[Equation 20]

The minimum national sample size is thencalculated as:

[Equation 21]

The task of the NCB is to translate theabsolute measure of error D = 10 basis pointsat a confidence level of 90% into a comparablerelative measure of error in term of anacceptable maximum variation coefficient CVfor all instrument categories. The ECB doesnot define this relative level of error.

y

sn

f1

yy of viationStandarddeCV

2y∗−

==

( )Ns

y*CV

sn 2

y2

2y

+=

158 See footnote 152.


11.4.3 Assumptions and data availability

The theoretical model explained in Chapter11.4.2 refers to a situation where a simplerandom sample is drawn without replacement.MFI interest rate statistics are indeed basedon a sampling procedure without replacementand hence refer to a finite referencepopulation. However, they are not based ona simple random sampling procedure butrather on stratified random sampling or acensus. To achieve the same level ofprecision,159 owing to the stratification of thepotential reporting population, fewerinstitutions need to be drawn with stratifiedrandom sampling than with simple randomsampling. The model hence overestimates theminimum sample size needed for MFI interestrate statistics.

Furthermore, the model assumes a normaldistribution of the sampling variables, i.e. theinterest rates and the amount of new businessin the potential reporting population. Anormal distribution of the sampling variablesmay be assumed if the potential reportingpopulation is very large and/or the samplelarge enough, which is the case for MFIinterest rate statistics. It may, however, beargued that the distribution of interest ratesin the potential reporting population is usuallyskewed, which means that the results of thesimulations need to be interpreted withcaution.

The higher the variance, the larger the sampleneeded to achieve the same level of precisionin the results. The true variance of theinterest rates in the potential reportingpopulation is unknown. For the purpose ofthe simulations, the variance is mainlyestimated from national surveys for retailinterest rate statistics and from data collectedfor supervisory purposes. As mentioned inChapter 11.4.1, these data are notharmonised and do not comply with the levelof detail and the definitions agreed for MFIinterest rate statistics. The (unharmonised)data provided by NCBs refer to weighted orsimple average interest rates, but also to themost commonly applied rates. Some of the

data are already derived from samples andothers based on a (near) census. Some NCBsreport data collected from ‘branches’,whereas others refer to ‘head offices’ or‘banking groups’. Some interest rates refer totypical products within an instrumentcategory and others to a range of verydifferent products in the same category.Finally, some NCBs have only limitedinformation on MFI/retail interest rates. As aconsequence, the estimates for the varianceof interest rates differ widely across MemberStates. These differences are a result of thediverse national banking systems and also ofthe dissimilar collection systems. For example,the variance of interest rates for typicalproducts is expected to be lower than if thefull range of interest rates in an instrumentcategory were taken into account.Furthermore, the variance over a smallnumber of institutions is expected to belower than if the estimate was based on alarger number. So the variances used in thesimulations are only rough approximations ofthe real variances that occur in MFI interestrate statistics.

Since a finite potential reporting populationand sampling without replacement areassumed for MFI interest rate statistics, thenational sample size also depends on the totalnumber of institutions in the potential reportingpopulation. This reference population maydiffer for each instrument category as not allinstitutions offer the full range of deposit andlending business vis-à-vis households and non-financial corporations. The ECB used in itssimulations in most cases the number ofcredit and other institutions as defined in thelist of MFIs. In most Member States, thisnumber overestimates the potential nationalreporting population as it includes institutionsthat are not active in deposit and lendingbusiness vis-à-vis households and non-financialcorporation. Therefore, the theoretical modeloverestimates the minimum sample size perinstrument category.

159 Defined in footnote 141.


11.4.4 Sample size and maximum randomerror

Figure 28 to Figure 31 illustrate the linkbetween the theoretical minimum sample sizederived by means of Equation 18 (verticalaxis) and the maximum random error D(horizontal axis) for a confidence level of 90%.Since the minimum sample size is alsodependent on the variance of the interestrates across MFIs and the size of the potentialreporting population in total or perinstrument category, this information is givenfor each of the series in the legend above thechart. To illustrate the influence of thevariance on the sample size, an instrumentcategory with a low variance, i.e. depositswith agreed maturity up to two years, andone with a high variance, i.e. consumer loansor loans to non-financial corporations up toone year, have been chosen for each of thefour countries A to D.

Figure 28 shows that when accepting amaximum random error D of 10 basis points,Country A would have to collect data from30 reporting agents on deposits with agreedmaturity up two years out of a potentialreporting population of 92 institutions forthis instrument category. Furthermore, it

would have to cover data from 58 reportingagents on consumer loans out of a potentialreporting population of 61 institutions forthe instrument category. The difference insample size is due to the difference invariance, which is with 0.17 low for thedeposit and with 4.28 high for the loancategory. A weighted average160 of the resultsof all instrument categories would lead to atheoretical overall minimum sample size of45 reporting agents for Country A out of anoverall potential reporting population of 116credit institutions and other institutions.

The same maximum error of 10 basis pointswould for Country B in Figure 29 lead to asample comprising 24 reporting agents fordeposits with agreed maturity up two yearsand 371 reporting agents for loans to non-financial corporations up to one year. Aweighted average of the results of allinstrument categories would lead to atheoretical minimum sample size of 71reporting agents for Country B out of apotential reporting population of 2,705 creditinstitutions and other institutions.

Figure 28Country A with a potential reporting population of N = 116 institutions

0

20

40

60

80

100

120

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Maximum random error D

Sam

ple

siz

e n

Minimum sample size - overall (N = 116)

Deposits with agreed maturity up to two years (variance = 0.17; N = 92)

Consumer loans (variance = 4.28; N = 61)

160 Computed from the minimum national sample size perinstrument category and the outstanding amounts per instrumentcategory, the latter being used as weighting information.


Applying the same model for Country C leadsin Figure 30 to a sample size of 44 reportingagents for deposits with agreed maturity uptwo years and 192 reporting agents for loansto non-financial corporations up to one year.

Figure 29Country B with a potential reporting population of N = 2,705 institutions

0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15


Sam

ple

siz

e n



Loans to enterprises up to one year (variance = 1.59; N = 2705)

Figure 30Country C with a potential reporting population of N = 848 institutions

The overall sample in Country C would needto comprise 150 reporting agents out of apotential reporting population of 848 creditinstitutions and other institutions.

0

100

200

300

400

500

600

700

800

900

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15


Sam

ple

siz

e n



Loans to enterprises up to one year (variance = 0.92; N = 848)

Finally, Figure 31 gives for Country D anoverall sample size of 63 reporting agents outof a potential reporting population of 214credit institutions and other institutions. Data

would need to be collected from 30 reportingagents on deposits with agreed maturity uptwo years and from 170 reporting agents onconsumer loans.


11.5 Special provisions in the case ofgroup reporting

The potential reporting population for MFIinterest rate statistics are all credit and otherinstitutions that take deposits from and grantloans to households and non-financialcorporations identified in the list of MFIs.However, as mentioned in Chapter 10.2,NCBs may allow credit institutions and otherinstitutions which are resident in a singlenational territory and individually included inthe list of MFIs to report MFI interest ratestatistics together as a group. Such groupscould, for example, be the Rabobanks in theNetherlands or the Caixas de Crédito AgrícolaMútuo in Portugal. The group becomes anotional reporting agent and has the samereporting requirements as the other(individual) credit institutions and otherinstitutions that are reporting agents.

The counting of the number of institutions inthe potential reporting population must beconsistent with the counting of them in theminimum sample size. For example, if the 400or so Rabobanks in the Netherlands arereporting together as a group they shouldeither be counted as one entity in both thepotential reporting population and the sample

Figure 31Country D with a potential reporting population of N = 214 institutions

0

50

100

150

200

250

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15


Sam

ple

siz

e n



Consumer loans (variance = 3.05; N = 214)

or they should each be counted individually inboth the potential reporting population andthe sample.

Groups that are reporting together need totell the NCB each year, for each instrumentcategory, the number of reporting institutionsand the variance of interest rates across theseinstitutions in the group. The number ofreporting agents and the variance must referto the month of October and be transmittedwith the October data.161 The reporting ofvariances in the group is intended tocompensate for the loss of information thatresults from reporting as a group, and isrequired by NCBs so they can estimate thetotal variance of interest rates in the nationalpotential reporting population. The varianceof interest rates across individual reportingagents is likely to be higher than the varianceacross groups. The reason is that each grouponly reports an average interest rate for thewhole group and these are supposed to bemore homogeneous than the rates offered bythe individual MFIs in the group. The averageof a group could, for example, even out

161 October was chosen as a ‘normal’ month, because it is not amonth of quarterly production as are March, June, Septemberand December, and not a ‘holiday’ month.


regional differences if those exist. If groupreporting led to underestimating the varianceof interest rates in the national potentialreporting population, then the requiredsample size would also be underestimated, asceteris paribus the sample needs to be smaller(larger) if the variance is lower (higher).Likewise, the sampling error would beunderestimated. So it is necessary to reportvariances within the group to achievecomparable results for all Member Statesbased on the same legal definition of areporting agent.

11.6 Allocation of the sample acrossstrata162

After defining the national strata and thenational sample size n, the sample is drawnby selecting the reporting agents from eachstratum. In this way the actual reportingpopulation is being defined. The total samplesize n is the sum of the sample sizes n1, n2, n3,…, nL for each of the strata:

n1 + n2 + n3 + … + nL = n [Equation 22]

Two issues need to be defined prior todrawing the reporting agents:

• the allocation of the sample size n amongthe strata, and

• the method for selecting the reportingagents.

Each NCB may choose the most appropriateallocation of the national sample size n amongthe strata. In other words, it is up to theNCBs to define how many reporting agentsnh have to be drawn from the total of creditinstitutions and other institutions Nh in eachstratum, as long as the sampling rate nh/Nh

for each stratum h fulfils the followingcondition:

0 < nh/Nh ≤ 1 [Equation 23]

The minimum standard is to select at leastone reporting agent from each stratum, i.e.

the sampling rate needs to be above zero (0< nh/Nh), which implies that it is not possibleto exclude one entire stratum from the actualreporting population. The number ofreporting agents nh may be the same for eachstratum (constant allocation), proportional tothe size of the stratum (proportional allocation),or dependent on the variance of the samplingvariables or another closely linked variable ineach stratum (optimal allocation163). Thesampling rate may also be one (nh/Nh = 1),which means that all credit institutions andother institutions in a stratum are selected asreporting agents. The allocation of thenational sample size across strata has aninfluence on the variance of the estimator.The best results and hence the lowest totalsampling error is in general achieved withoptimal allocation.

Regarding the method of selecting the reportingagents within each stratum, the Regulationgives NCBs the choice between randomsampling and the selection of the largestinstitutions per stratum,164 where one methodmay be used for one part of the strata andthe other method for the rest:

• The statistically ideal case is the randomselection of reporting agents in eachstratum. In a random sample each creditinstitution and other institution in thestratum has a known probability abovezero of being selected as a reporting agent.The random drawing of the institutions ineach stratum can then be carried out withequal probability for all institutions or withprobability proportional to the size of theinstitution. In the latter case, biginstitutions are more likely to be drawn.However, the small institutions also have aprobability of selection above zero andcould hence in theory also be selected.Random selection with probabilityproportional to size is highly recommendedwith extremely skewed populations.

162 See also paragraphs 15 to 20 of Annex I to the Regulation.163 Also known as Neyman optimum.164 Advantages and drawbacks are discussed in Chapter 11.2.


• The alternative to random sampling is theselection of the largest institutions in eachstratum. The aim of this procedure is toprotect small credit institutions and otherinstitutions from reporting as they mightbear very high costs in relation to biggerinstitutions. The selection of the largestinstitutions per stratum is not randomsampling, as small institutions have aprobability of selection equal to zero, butcan be justified with highly skewedpopulations. However, the procedure ofselecting the largest institutions in eachstratum can for the purposes of MFIinterest rate statistics be treated in thesame way as sampling with probabilityproportional to size.

The precondition for sampling with probabilityproportional to size, i.e. for random samplingand the selection of the largest institutions, isa strong statistical relation between thesampling variables and the size of the creditinstitutions and other institutions in thepotential reporting population. Samplingvariables are the interest rates and amount ofnew business. The size of the credit institutionor other institution is approximated by thesize of the relevant balance sheet items foreach institution. In some countries thecurrent surveys on retail interest ratestatistics might provide data on new businessamounts which could be compared to thesize of the corresponding balance sheet itemsfor these institutions. In other countries, datacollected for supervisory purposes might beused to test whether such a strongrelationship exists. Where such data are notavailable, NCBs should make assumptionsabout the relationship between the samplingvariables and the size of the institution. Theprecision of the results of the survey dependson the strength of the statistical relationship.

11.7 Estimation of total new businessvolume165

MFI interest rates for the euro area arecompiled as weighted averages of the interestrates applied in the Member States. For

interest rates on outstanding amounts theweighting information is derived from the MFIbalance sheet statistics. For interest rates onnew business, the new business amount perinstrument category and Member State isneeded for computing weighted average newbusiness rates for the euro area. The amountof new business is collected from thereporting agents together with the interestrates. Assuming that a sample is applied, thetotal amount of new business per MemberState and instrument category needs to beestimated from the results in the sample byapplying expansion, raising or inflation factors.The estimation of the population total is alsoreferred to as grossing-up.166

This chapter explains in generic terms thegrossing-up procedure for the amount of newbusiness, including the computation ofselection probabilities and their use asexpansion factors. Generic terms are usedbecause the precise formulae for a MemberState depend on the NCBs’ choice of strata,the allocation of the sample across the strata,the method of drawing the reporting agentsper stratum and also, in the case of the initialsample, on the data available as auxiliaryinformation. First, the estimation of thepopulation total from amounts derived bysimple random sampling is explained. Thenthe more complex procedure for samplingwith probability proportional to size isillustrated, which also applies to the selectionof the largest institutions within each stratum.As explained in Chapter 11.6, both therandom selection and the selection of thebiggest institutions in a stratum are treatedas random procedures for the purposes ofMFI interest rate statistics. Therefore, thesame general method may be used to estimatethe total amount of new business in thepotential reporting population from theresults of the sample.

165 Already mentioned in Chapter 10.3. See also paragraph 68 ofAnnex II to the Regulation.

166 No grossing-up is required for simple averages and ratios, becauseit is assumed that the estimate from the sample is also theestimate for the population.


In the case of simple random sampling within astratum, each credit institution and otherinstitution has the same chance of beingselected. At each draw the procedure givesan equal chance of selection to everyinstitution in the potential reportingpopulation, i.e. to each institution that hasnot already been drawn. Hence, eachinstitution has a known probability ofselection. Before the first draw, theprobability of selecting institution i is:

[Equation 24]

with n as the size of the sample and N as thesize of the potential reporting population. Theinverse of this selection probability is thenused as the expansion factor to estimate thetotal amount of new business in thepopulation from the sample:

[Equation 25]

The Horvitz-Thompson estimator for thepopulation total derived from a sample is:

[Equation 26]

with Y as the estimated total amount of newbusiness in the potential reporting population,yi as the amount of new business of institutioni and iπ as the probability of selectinginstitution i. In the case of simple randomsampling, the estimator from Equation 26 forthe total amount of new business in thepopulation becomes:

[Equation 27]

The starting point for sampling with probabilityproportional to size is also the calculation ofthe selection probability iπ for each institutioni. The selection probabilities are computedseparately for each stratum from the fixedsample size nh per stratum and the variable Ui

indicating the size of the institution i:

[Equation 28]

Nn

i =π

nN1

i

=π

∑∈ π

=si i

iyY

ssi

isi N

ni yNy

n1NyY ∗=∗== ∑∑

∈∈

∑=

∗=∗=πhN

1ii

ihihi

U

Unpn

Variable U signifies auxiliary information,which can differ between Member States. It isup to NCBs to define the most suitablenational variable U, which must be stronglycorrelated with the amount of new business,i.e. variable Y, in that Member State. Forexample, Ui could be the outstanding amountfor each instrument category for institution ifrom the MFI balance sheet statistics. Animportant restriction for the calculation of theselection probabilities is that the product ofthe fixed sample size nh and the size Ui ofinstitution i is smaller than the total of allinstitutions in the same instrument category:

[Equation 29]

This restriction guarantees the coherence ofthe selection probabilities. If the restrictionof Equation 29 is not fulfilled for institution i,this institution is selected automatically. Theprobability of its selection is hence set as 1.The other selection probabilities arerecalculated based on the exclusion of i withthe new restriction:

[Equation 30]

If this restriction is again not fulfilled,institution k is selected automatically and itsselection probability set to 1. The otherprobabilities are then calculated as follows:

[Equation 31]

etc.

Figure 32 gives an example for the calculationof the selection probabilities in the case ofsampling with probabilities proportional tosize.

∑=

<∗hN

1iiih UUn

( ) ∑−

=≠≠ <∗−

1N

1kikikh

h

UU1n

( ) ∑−

=≠≠ <∗−

2N

1ji,kji,kjh

h

UU2n


As in the case of simple random sampling, theHorvitz-Thompson estimator in Equation 26is used to estimate the total amount of newbusiness Y in the potential reportingpopulation from the result of the sample.Hence, inverses of the selection probabilitiescalculated by means of Equations 28 to 31,

i.e. i

1π , are used as expansion factors.

Assuming that there is a strong statisticalrelationship between the size of theinstitutions U and the amount of new businessY, the Horvitz-Thompson formula providesan unbiased estimate of the population total,i.e. the total amount of new business in thepotential reporting population. It is importantto note that pi, i.e. the size of institution i asshare of all institutions, in Equation 28,changes over time. In order to derive anunbiased estimate of the total amount of newbusiness from the sample, the pi and hencethe iπ need to be recalculated each month.

MFI interest rate statistics are based on aselection without replacement, i.e. each creditinstitution and other institution can only beselected once. In random sampling withprobability proportional to size, care has tobe taken that the selection probabilities areproportional to the size of the institutionsafter each draw. Hence, after the first

Figure 32

Institution Size Xi Xi/(sum Xi) Xi/(sum Xi-A) 4*Xi 3*Xi Selection probabilityA 16271700 0.3414 65086800 1B 9293500 0.1950 0.2960 27880500 0.8881 =3*0.2960C 6627700 0.1391 0.2111 19883100 0.6334 =3*0.2111D 4822600 0.1012 0.1536 14467800 0.4609E 3420100 0.0718 0.1089 10260300 0.3268F 2975100 0.0624 0.0948 8925300 0.2843G 2418600 0.0507 0.0770 7255800 0.2311H 669732 0.0141 0.0213 2009196 0.0640I 470700 0.0099 0.0150 1412100 0.0450J 260775 0.0055 0.0083 782325 0.0249K 203760 0.0043 0.0065 611280 0.0195L 126591 0.0027 0.0040 379773 0.0121M 47600 0.0010 0.0015 142800 0.0045N 34987 0.0007 0.0011 104961 0.0033O 20000 0.0004 0.0006 60000 0.0019

Sum Xi 47663445Sum Xi-A 31391745

institution is drawn, the selection probabilitiesof the remaining institutions need to beadjusted, the same after the second institutionis drawn, etc. For stratified random sampling,practical methods have been developed. Ifthe largest institutions within each stratum arechosen, a selection with replacement can beassumed. In this case the selectionprobabilities only need to be calculated oncebefore the first draw.

Each Member State either carries out a censusor has one sample for MFI interest ratestatistics covering all instrument categories.167

It is hence possible that a reporting agent hasno business in some instrument categories.For example, a sample comprises 50 reportingagents but for one instrument category only30 of them carry out business and haveoutstanding amounts. The outstandingamount, however, is used as auxiliary variableU indicating the size of the institutions. Forestimating the total amount of new businessfor this instrument category, the selectionprobabilities and hence expansion factors onlytake into account the 30 reporting agentsthat carry out business and have outstandingamounts. Whether the reporting agents alsohave new business or not in that month is

167 See also Chapter 11.9

N=15n=4


irrelevant in calculating the selectionprobabilities and expansion factors.

11.8 Maintenance of the sample168

NCBs that choose the sampling approachmust ensure that the sample remainsrepresentative over time. Maintenance of thesample over time is inherent to panel surveysthat aim to collect two or more measuresfrom the same sample units over time. Thefact that the same units report repeatedlyover time ensures the consistency of theanswers, also known as the panel effect.169

t1

Leavers Joiners t2

Panel of reporting agents n1

nd

(nb) n

2n

c = n

1 - n

d

Potential reporting population N1

Nd

Nb

N2 = N

c + N

bN

c = N

1 – N

d

Selection probability π1i

π’2i

π2i

The selection procedure for a panel does notdiffer from the selection of any other sample.The same methods are applied, but adaptedto the panel situation as necessary. What isdifficult to capture in a panel, but of noimportance in a one-off sample, is thevariability of the structure of the potentialreporting population compared to the actualreporting population, i.e. the reporting agents.The sample must therefore be adjusted forjoiners to the potential reporting population,leavers from the potential and actualreporting population, as well as for changesin the characteristics of the reporting agents.The following table describes the proceduresthat needs to be followed:

At time t1 of the drawing of the initial sample,n1 institutions of the total N1 have beenselected as reporting agents. At time t2, thepotential reporting population has changedto N2 institutions consisting of the commoninstitutions Nc, which are included in thepotential reporting population at t1 and at t2,as well as the joiners of the potentialreporting population Nb during the periodfrom t1 to t2. The common institutions Nc

consist of all institutions N1 in the potentialreporting population in t1 less the leavers Nd

during the period from t1 to t2.

If the initial sample n1 was drawn as a simplerandom sample and the sample n2 not adjustedfor joiners and leavers, then the sample n2

includes only the common institutions nc, i.e.the n1 institutions of the initial sample lessthe leavers nd. Hence, although the structureof the potential reporting population changesfrom t1 to t2 this is not reflected in the sample.The situation is as follows:

[Equation 32]

The sample n2 needs to be adjusted for thejoiners to the potential reporting population toremain representative of the potential reportingpopulation over time. In order to do so, it isnecessary to draw a sample nb from thepopulation of all joiners Nb. The sample n2 thenincludes the common institutions nc, i.e. the n1

institutions of the initial sample less the leaversnd, as well as the joiners nb in the period from t1

to t2. In this way the changes in the structure ofthe potential reporting population resulting fromjoiners from t1 to t2 are also reflected in thesample. The complementary selection of joininginstitutions nb among the total number of joinersNb is referred to as incremental sampling overtime. The situation is then as follows:

[Equation 33]

The sample n2 also needs to be adjusted forthe leavers from the potential and the actual

i2i1

1

1

2

c

d1c2

Nn

Nn

nnnn

π≠π

≠

−==

∑∑∈∈ π

+π

=≅

π≅π

≅=++

+−=+=

bc ni i2ni i122

i2i1

1

1

2

2

bc

bc

bd1bc2

11NN

Nn

Nn

NNnn

nnnnnn

168 See also paragraphs 21 to 26 of Annex I to the Regulation.169 See also Chapter 11.9.


reporting population. No adjustment isnecessary if there is proportionality betweenthe leavers in the potential reportingpopulation Nd and the leavers in the samplend (Case 1). If the institutions are leaving thepotential reporting population and theseinstitutions are not in the sample, the samplebecomes relatively too large for the size ofthe potential reporting population (Case 2).If relatively more institutions leave the samplethan the potential reporting population, thesample becomes too small over time andmight cease to be representative (Case 3).These three situations are illustrated in Figure33, where the circle signifies the potentialreporting population N, the uncolouredsquare the sample n and the dark square theleavers in the period from t1 to t2:

Figure 33

Case 1: Proportionality

N

leavers

n

N

leavers

n

N

leavers

n

Case 3:Leavers overrepresented in the sample

Case 2:Leavers underrepresented in the sample

Whereas in Case 1 the number of leaversfrom the potential reporting population can

be estimated using the formula dni

1 Nd

i1≅∑

∈π ,

this is not possible in Cases 2 and 3. In Case2 the number of leavers in the population isunderestimated and in Case 3 overestimated.If, in Cases 2 and 3, the inverses of theselection probabilities were used as expansionfactors, this would lead to biased results forthe population total. Therefore, in Cases 2and 3 the weights attached to each reportingagent in the sample must be adjusted, forexample by means of poststratification, i.e.stratification after the selection of the sample.The weight attached to each reporting agentis the inverse of its selection probability andhence the expansion factor for estimating thepopulation total. With poststratification, thesample becomes restratified. New selectionprobabilities and hence weights are allocated.

Even in Case 2, where the sample is relativelytoo big for the potential reporting population,none of the common institutions nc is takenout of the sample. This means that aninstitution that once implemented the MFIinterest rate reporting scheme will only berelieved of the reporting burden if it leavesthe list of MFIs or if the NCBs in agreementwith the ECB decides to reselect all reportingagents.

Finally, the sample needs to be adjusted forchanges in the characteristics of the reportingagents. These changes can occur because ofmergers, divisions, growth of the institution,etc. Some reporting agents might change thestratum. As in Cases 2 and 3 for leavers, thesample needs to be adjusted, for example bymeans of poststratification. In this case, thesample is restratified and new selectionprobabilities and hence weights are allocated.

According to the Regulation, NCBs thatchoose the sampling approach must check therepresentativity of their sample at least everyyear. If there are significant changes in thepotential reporting population, these arereflected in the sample after the annual check.In intervals of at least two years, the sample


must be refreshed to take account of joiners,leavers and other changes of thecharacteristics of the reporting agents. NCBsmay check and refresh their sample moreoften. Such adjustments of the sample overtime are statistically necessary to ensure thequality of panel surveys; in general they willnot lead to breaks in the time series.

The ECB leaves it to the discretion of theNCB when in the year it reviews the nationalsample. The ECB also leaves it to thediscretion of the NCB how much time itgrants to new reporting agents forimplementing the reporting requirements, butexpects that there are not more than 12months between the identification of a newreporting agents and the first reporting ofdata.

11.9 Further sampling issues170

To achieve consistency between MFI interestrate statistics on:

• outstanding amounts referring to deposits,

• outstanding amounts referring to loans,

• new business referring to deposits, and

• new business referring to loans,

NCBs should use the same sample of creditinstitutions and other institutions forcollecting these sets of statistics. It is,however, possible to use a sample for a subsetof MFI interest rate statistics and a census forthe rest, for example a sample for newbusiness and a census for outstandingamounts or a sample for new lending businessand a census for new deposit business andoutstanding amounts. The Regulation doesnot allow the use of two or more differentsamples.

By definition, the interest rates onoutstanding amounts comprise all depositsplaced and not yet withdrawn and all loanswithdrawn and not yet repaid by customers

170 See also paragraphs 27 and 28 of Annex I to the Regulation.171 See also Chapter 11.8.172 See also Chapter 7.2.

in all the periods up to and including thereporting date. This includes all new businessduring the month before and including thereporting date unless these new contractshave already been closed before the reportingdate. Moreover, as the MFI interest rates onnew business and on outstanding amountsare both weighted averages, not only theinterest rates but also the attached quantitiesare interlinked. The fact that the same creditinstitutions and other institutions report thestatistics both on outstanding amounts andnew business and do so repeatedly periodafter period leads to a panel effect.171 Thiseffect ensures consistency of answers for thetwo sets of statistics at one particular time(cross-sectional analysis) and for each set ofstatistics over time (time series analysis). If thesample differed for the collection of MFIinterest rates on new business andoutstanding amounts, a ‘time series paneleffect’ would still exist because the sameinstitutions were surveyed in each period. Iftwo samples were drawn, one for thereporting on new business and the other foroutstanding amounts, sampling effects mightlead to a different composition of the twoactual reporting populations, which mightthen lead to discrepancies between the twosets of statistics.

NCBs need to cover each instrument categorythat exists in the banking business of residentcredit institutions and other institutions witheuro area households and non-financialcorporations, but not each product offered atnational level.172 An instrument category isinapplicable at national level only if creditinstitutions and other institutions do not offerany such products to resident non-financialcorporations and households. NCBs mustprovide data if some business exists, howeverlimited this business is. Hence, if aninstrument category is only offered by oneinstitution, then this institution needs to berepresented in the sample. If an instrumentcategory did not exist in a Member State at


the time of the initial drawing of the sample,but afterwards one institution introduces anew product belonging to this instrumentcategory, then this institution needs to beselected into the sample at the time of thenext representativity check. If a new productis created that belongs to an existinginstrument category, the institutions in thesample need to cover it with the nextreporting, as all reporting agents are requiredto report for each instrument category allinterest rates applied to all the products thatfit this category.

Under certain conditions sampling is allowedat the level of branches rather than at thelevel of credit institutions or otherinstitutions. Sampling at the level of branchescan be undertaken for one or more of thestrata in the sample. The first precondition isthat the NCB decides on a census for thatstratum, i.e. that all credit institutions andother institutions in the stratum are subjectto reporting. The second precondition is thatthe NCB has a full list of branches that coversthe entire business of the credit institutionsand other institutions in the stratum. Thethird precondition is that the NCB hasappropriate data to assess the variance ofinterest rates on new business vis-à-vishouseholds and non-financial corporationsacross branches, and based on this variancethe number of branches that should report.The selected branches become notionalreporting agents and hence have the samereporting obligations as credit institutions andother institutions that are selected asreporting agents. Reporting at branch leveldoes not affect the liability as reporting agentof the credit institution or other institutionto which branches belong, i.e. if a branchreports incorrectly or fails to report, theinstitution included in the list of MFIs isliable, rather than the branch.

11.10 Description of the euro areasample

All euro area Member States except Greececurrently apply a sampling approach for

selecting the reporting agents. Hence only asubset of the potential reporting populationhad to implement the requirements of theRegulation and provides data for MFI interestrate statistics. Greece carries out a census.Apart from legal reasons in favour of a census,the Greek reporting agents and the interestrates they offer are heterogeneous to theextent that a sampling approach would haveresulted in a near-census. France combinescensus and sampling approach: a census isimplemented for collecting the interest rateson outstanding amounts, including bankoverdrafts, overnight deposits, and depositsredeemable at notice, and sampling is appliedfor (the rest of) MFI interest rate statisticson new business.

NCBs draw on a variety of stratification criteriain order to divide the potential reportingpopulation into homogeneous strata prior todrawing the sample. The variety ofstratification criteria reflects the diversity ofthe banking business in the euro area. SomeMember States stratify with respect to bankcategories. Others apply regional componentsto group the MFIs in the potential reportingpopulation. Furthermore, the type of productand customer, the degree of specialisation,the size of the institution and the number ofbranches are used to build up homogenousstrata. Some countries use principalcomponent or factor analysis to determinethe relevant stratification criteria and applycluster analysis to establish the strata. Theresulting number of strata per Member Statevaries between two and 15.

Most NCBs selected as reporting agents thelargest institutions within each stratum. Theprocedure is often combined with, or leadsto, a census in at least one stratum. TwoNCBs selected reporting agents randomly withprobability proportional to size. One MemberState applies sampling at branch level forinterest rates on new business. Their stratumof large universal banks covers about 25,000branches, from which the NCBs selectedrandomly with probability proportional to sizeabout 2,000 branches as reporting agents.


Two Member States allow group reporting.Therefore some of their credit institutionsand other institutions, which are individuallyincluded in the list of MFIs, report MFIinterest rate statistics together as a group,i.e. as if they were a single MFI. The groupbecomes a notional reporting agent.

The selection procedure leads to a euro areasample of more than 1,800 reporting agents

out of a total of more than 7,000 institutionsin the potential reporting population. Theeuro area sample covers about 80% of thestock of euro-denominated deposits receivedfrom and at least 80% of the stock of euro-denominated loans granted to households andnon-financial corporations resident in theparticipating Member States.


Index of terms

advertised interest rates, 12agio, 17annual percentage rate of charge (APRC),

12, 24, 50national legislation, 27, 28

annualised agreed rate, 13, 18, 50annualised agreed rate formula, 13annualising interest rates, 13, 18, 21

bad loans, 30, 52bank overdrafts, 16, 31, 40, 47, 50, 56, 74business coverage.

See MFI interest rates on new business oron outstanding amounts

capital certainty, lack of, 68, 69, 72ceiling, 43, 44census, 78, 98charges

in MFI interest rate statistics, 13in the APRC, 25indicator of, 26

consumer credit, 40, 45, 56, 60Consumer Credit Directive, 12, 24, 25, 27converting a deposit into shares, 68cooling off period, 45credit cards, 62, 63credit facility, 62credit institutions, 10credit limit, 62

deposits comprising two parts, 69deposits redeemable at notice,

31, 34, 47, 50, 54, 55, 74deposits with agreed maturity,

35, 36, 37, 43, 44, 55, 68derivative contract, 68, 70disagio, 13, 17

effective interest rate, 12euro area, 10euro area sample, 98expansion factors, 76, 96external index, 43, 44

favourable rates, 21fees. See chargesfixed interest rates, 38, 59floor, 43, 44

frequencyof interest payments, 14of interest rate statistics, 8of repayments of principal, 14

grossing-up.See expansion factors

group reporting, 75, 90, 99

Horvitz-Thompson estimator, 93households, 53Huygens theorem, 80

implicit rates, 47, 48, 75incremental sampling, 96indefinite loan, 16initial period of fixation, 27, 38, 59institutional arrangements, 23instrument categories, 51interest rates.

See MFI interest rates

loan in tranches, 41, 43, 59loan offer, 45loans for debt restructuring, 30, 46, 52loans to households for house purchases,

44, 46, 57, 66, 67, 72, 73loans to non-financial corporations, 56

matured deposit, 36maturity, 38, 39, 58, 59, 60maximum random error, 83, 86, 88measurement error, 78methodological notes, 23, 76MFI interest rate statistics

coverage of, 11scope of, 8uses of, 8

MFI interest rateslinked to share price, 69on new business, 35, 48, 50, 75on outstanding amounts, 30, 47, 49, 74on the amount granted but not yetwithdrawn, 42type of rate, 12

monetary financial institutions (MFIs), list of,10, 75, 81, 87, 90, 96, 98

moratorium on a loan, 46mortgage loans.

See loans to households for housepurchases


narrowly defined effective rate (NDER),18, 49, 50

national conventions, 23new business.

See MFI interest rates on new businessnew products, 52, 98nominal interest rate, 12non-financial corporations, 53, 57non-negotiable debt security, 17non-profit institutions serving households

(NPISHs), 11, 29, 54notional reporting agent, 75, 90

one-off deposit, 16original maturity. See maturityother institutions, 10other loans to households, 57outstanding amounts.

See MFI interest rates on outstandingamounts

overnight deposits, 31, 32, 47, 50, 55, 62, 74

panel survey, 95, 97participating Member State, 10period of notice, 58population total.

See expansion factorspreliminary offers, 45prime rates, 12prolongation of existing deposit or loan,

35, 36purposive sampling, 79

random sampling, 79, 91regular savings, 34, 37Regulation ECB/2001/18, 7regulatory arrangements, 23renegotiation of contract, 27, 35, 46, 60, 68repayments

exceptional repayments, 15frequency of, 14stop of, 46

reporting agents, 11reporting population

potential, 95reporting population

actual, 11, 78, 91, 95extension of, 11potential, 11, 73, 76, 78, 90representative of, 79stratification of potential, 79

repos, 54, 56resident, 10retail interest rates, 8reverse convertible, 68

sample, 78sample maintenance, 95sample size, 83, 85, 88, 91sampling at branch level, 81, 98sampling error, 76, 78sampling variables, 79sampling with probability proportional to

size, 91, 93, 98savings bonds, 67savings plan, 66.

See regular savingssecuritisation, 73selection of the largest institutions,

92, 94, 98selection probabilities.

See expansion factorsselection without replacement, 94short-term approach, 8single-stage sampling, 82snapshot, 47, 48, 74sole proprietorships, 54special national practices, 23specific national products, 23split of loan, 67standard year, 19steady-state approach, 8step-up and step-down deposit, 65step-up and step-down loan, 61, 65stratification, 79stratification criteria, 81, 98subsidies, 21, 28subsidised loan, 23

tail institutions, 78taxes, 21top-up loan, 40total costs of the credit to the consumer, 25

variable interest rates, 21, 42, 59.See also external index

variance, 80, 83, 88, 90

weighted average interest rate, 74

manual on mfi interest rate statistics - european central bank

Documents