THE DETERMINANTS OF UNSECURED BORROWING: EVIDENCE FROM THE BRITISH HOUSEHOLD PANEL SURVEY. Documentos de Trabajo N.º 0511. Ana del Río and Garry Young



Similar documents
Why Did the Demand for Cash Decrease Recently in Korea?

Morningstar Investor Return

Working Paper No Net Intergenerational Transfers from an Increase in Social Security Benefits

Chapter 8: Regression with Lagged Explanatory Variables

Risk Modelling of Collateralised Lending

Appendix D Flexibility Factor/Margin of Choice Desktop Research

Measuring macroeconomic volatility Applications to export revenue data,

PROFIT TEST MODELLING IN LIFE ASSURANCE USING SPREADSHEETS PART ONE

How To Calculate Price Elasiciy Per Capia Per Capi

USE OF EDUCATION TECHNOLOGY IN ENGLISH CLASSES

Finance and Economics Discussion Series Divisions of Research & Statistics and Monetary Affairs Federal Reserve Board, Washington, D.C.

Supplementary Appendix for Depression Babies: Do Macroeconomic Experiences Affect Risk-Taking?

Economics Honors Exam 2008 Solutions Question 5

BALANCE OF PAYMENTS. First quarter Balance of payments

Vector Autoregressions (VARs): Operational Perspectives

Duration and Convexity ( ) 20 = Bond B has a maturity of 5 years and also has a required rate of return of 10%. Its price is $613.

The Real Business Cycle paradigm. The RBC model emphasizes supply (technology) disturbances as the main source of

Edinburgh Research Explorer

DOES TRADING VOLUME INFLUENCE GARCH EFFECTS? SOME EVIDENCE FROM THE GREEK MARKET WITH SPECIAL REFERENCE TO BANKING SECTOR

4. International Parity Conditions

II.1. Debt reduction and fiscal multipliers. dbt da dpbal da dg. bal

The Greek financial crisis: growing imbalances and sovereign spreads. Heather D. Gibson, Stephan G. Hall and George S. Tavlas

Individual Health Insurance April 30, 2008 Pages

A Re-examination of the Joint Mortality Functions

The impact of self-employment on labour-productivity growth: A Canada and United States comparison

The Economic Value of Medical Research

Principal components of stock market dynamics. Methodology and applications in brief (to be updated ) Andrei Bouzaev, bouzaev@ya.

Journal Of Business & Economics Research September 2005 Volume 3, Number 9

Rationales of Mortgage Insurance Premium Structures

MACROECONOMIC FORECASTS AT THE MOF A LOOK INTO THE REAR VIEW MIRROR

Interest Rates, Taxes and Corporate Financial Policies

Market Liquidity and the Impacts of the Computerized Trading System: Evidence from the Stock Exchange of Thailand

The Influence of Positive Feedback Trading on Return Autocorrelation: Evidence for the German Stock Market

Optimal Investment and Consumption Decision of Family with Life Insurance

The retirement-consumption puzzle and involuntary early retirement: Evidence from the British Household Panel Survey. Sarah Smith

Chapter 1.6 Financial Management

A Note on Using the Svensson procedure to estimate the risk free rate in corporate valuation

GoRA. For more information on genetics and on Rheumatoid Arthritis: Genetics of Rheumatoid Arthritis. Published work referred to in the results:

The impact of dark matter on sustainability of current account imbalance

Hedging with Forwards and Futures

When Is Growth Pro-Poor? Evidence from a Panel of Countries

Adversity or Strategy?: The Effects of Credit Constraint and Expectation on Mortgage Default and Personal Bankruptcy Decisions

Debt Accumulation, Debt Reduction, and Debt Spillovers in Canada, *

The Impact of Surplus Distribution on the Risk Exposure of With Profit Life Insurance Policies Including Interest Rate Guarantees.

Present Value Methodology

Estimating Time-Varying Equity Risk Premium The Japanese Stock Market

DYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS

SUBJECT SA0 OF THE INSTITUTE AND FACULTY OF ACTUARIES

The Impact of Surplus Distribution on the Risk Exposure of With Profit Life Insurance Policies Including Interest Rate Guarantees

Bid-ask Spread and Order Size in the Foreign Exchange Market: An Empirical Investigation

11/6/2013. Chapter 14: Dynamic AD-AS. Introduction. Introduction. Keeping track of time. The model s elements

The Transport Equation

CHARGE AND DISCHARGE OF A CAPACITOR

LEASING VERSUSBUYING

Does Option Trading Have a Pervasive Impact on Underlying Stock Prices? *

Default Risk in Equity Returns

The Grantor Retained Annuity Trust (GRAT)

Contrarian insider trading and earnings management around seasoned equity offerings; SEOs

Loans, Interest Rates and Guarantees: Is There a Link? 1

Measuring the Effects of Exchange Rate Changes on Investment. in Australian Manufacturing Industry

The naive method discussed in Lecture 1 uses the most recent observations to forecast future values. That is, Y ˆ t + 1

Working paper No.3 Cyclically adjusting the public finances

Documenting Uncertain Times: Post-graduate Transitions of the. Academically Adrift. Cohort. Josipa Roksa

Segmentation, Probability of Default and Basel II Capital Measures. for Credit Card Portfolios

CURRENT ACCOUNTS IN THE EURO AREA: AN INTERTEMPORAL APPROACH. José Manuel Campa and Ángel Gavilán an. Documentos de Trabajo N.

Relationships between Stock Prices and Accounting Information: A Review of the Residual Income and Ohlson Models. Scott Pirie* and Malcolm Smith**

INTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES

How To Calculate A Person'S Income From A Life Insurance

WORKING P A P E R. Does Malpractice Liability Reform Attract High Risk Doctors? SETH A. SEABURY WR-674-ICJ. December 2009

The Interaction of Guarantees, Surplus Distribution, and Asset Allocation in With Profit Life Insurance Policies

Determinants of Capital Structure: Comparison of Empirical Evidence from the Use of Different Estimators

Distributing Human Resources among Software Development Projects 1

Small and Large Trades Around Earnings Announcements: Does Trading Behavior Explain Post-Earnings-Announcement Drift?

UNDERSTANDING THE DEATH BENEFIT SWITCH OPTION IN UNIVERSAL LIFE POLICIES. Nadine Gatzert

Idealistic characteristics of Islamic Azad University masters - Islamshahr Branch from Students Perspective

Factors Affecting Initial Enrollment Intensity: Part-Time versus Full-Time Enrollment

ARCH Proceedings

The Sensitivity of Corporate Bond Volatility to Macroeconomic Announcements. by Nikolay Kosturov* and Duane Stock**

Preliminary. Comments welcome. Equity Valuation Using Multiples

Table of contents Chapter 1 Interest rates and factors Chapter 2 Level annuities Chapter 3 Varying annuities

Research. Michigan. Center. Retirement. Behavioral Effects of Social Security Policies on Benefit Claiming, Retirement and Saving.

The Interaction of Public and Private Insurance: Medicaid and the Long-Term Care Insurance Market

Day Trading Index Research - He Ingeria and Sock Marke

Does Capital Punishment Have a Deterrence Effect on the Murder Rate? Issues and Evidence

DEBT REVOLVERS FOR SELF CONTROL. Carol C. Bertaut and Michael Haliassos DEPARTMENT OF ECONOMICS UNIVERSITY OF CYPRUS. Discussion Paper

Stock Returns with Consumption and Illiquidity Risks

The Behavior of China s Stock Prices in Response to the Proposal and Approval of Bonus Issues

William E. Simon Graduate School of Business Administration. IPO Market Cycles: Bubbles or Sequential Learning?

MULTI-PERIOD OPTIMIZATION MODEL FOR A HOUSEHOLD, AND OPTIMAL INSURANCE DESIGN

Does informed trading occur in the options market? Some revealing clues

The Information Content of Implied Skewness and Kurtosis Changes Prior to Earnings Announcements for Stock and Option Returns

Dynamic Hybrid Products in Life Insurance: Assessing the Policyholders Viewpoint

GOOD NEWS, BAD NEWS AND GARCH EFFECTS IN STOCK RETURN DATA

Monetary Policy & Real Estate Investment Trusts *

The Identification of the Response of Interest Rates to Monetary Policy Actions Using Market-Based Measures of Monetary Policy Shocks

Market Efficiency or Not? The Behaviour of China s Stock Prices in Response to the Announcement of Bonus Issues

Internal and External Factors for Credit Growth in Macao

Migration, Spillovers, and Trade Diversion: The Impact of Internationalization on Stock Market Liquidity

ANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS

Cointegration: The Engle and Granger approach

Transcription:

THE DETERMINANTS OF UNSECURED BORROWING: EVIDENCE FROM THE BRITISH HOUSEHOLD PANEL SURVEY 2005 Ana del Río and Garry Young Documenos de Trabajo N.º 0511

THE DETERMINANTS OF UNSECURED BORROWING: EVIDENCE FROM THE BRITISH HOUSEHOLD PANEL SURVEY

THE DETERMINANTS OF UNSECURED BORROWING: EVIDENCE FROM THE BRITISH HOUSEHOLD PANEL SURVEY (*) Ana del Río (**) BANCO DE ESPAÑA Garry Young (***) BANK OF ENGLAND (*) This paper was largely wren when Ana Del-Río was on secondmen a he Bank of England. The views expressed in his paper are hose of he auhors, and no necessarily hose of he Bank of England or Banco de España. We are graeful o Andrew Beno, Olympia Bover, Peer Brierley, John Ermisch, Sephen Jenkins, Merxe Tudela, wo anonymous referees and paricipans a seminars held a he Bank of England, Banco de España and he 2003 Brish Household Panel Survey research conference for helpful commens and suggesions; any remaining errors are our own. (**) Banco de España. C/ Alcalá, 48. 28014 Madrid, Spain. email: adelrio@bde.es. (***) Macro Prudenial Risk Division, Financial Sabily, Bank of England, Threadneedle Sree, London, EC2R 8AH. email: garry.young@bankofengland.co.uk. Documenos de Trabajo. N.º 0511 2005

The Working Paper Series seeks o disseminae original research in economics and finance. All papers have been anonymously refereed. By publishing hese papers, he Banco de España aims o conribue o economic analysis and, in paricular, o knowledge of he Spanish economy and s inernaional environmen. The opinions and analyses in he Working Paper Series are he responsibily of he auhors and, herefore, do no necessarily coincide wh hose of he Banco de España or he Eurosysem. The Banco de España disseminaes s main repors and mos of s publicaions via he INTERNET a he following webse: hp://www.bde.es. Reproducion for educaional and non-commercial purposes is permed provided ha he source is acknowledged. BANCO DE ESPAÑA, Madrid, 2005 ISSN: 0213-2710 (prin) ISSN: 1579-8666 (on line) Depóso legal: Imprena del Banco de España

Absrac Household indebedness has risen sharply in recen years, wh large increases in boh secured and unsecured borrowing. In his paper, waves 5 and 10 of he Brish Household Panel Survey (BHPS) for 1995 and are used o examine he deerminans of paricipaion in he unsecured deb marke and he amoun borrowed. Prob models for paricipaion are esimaed and age, income, posive financial prospecs and housing enure are found o be very significan and have he expeced sign according o a life-cycle model for consumpion. Regressions o explain he level of borrowing by individuals sugges ha income is he main variable explaining cross-secional differences in unsecured debs. The increase in aggregae unsecured deb beween 1995 and does no seem o be closely linked o changes in he deerminans of deb marke paricipaion and has been mainly associaed wh he larger amouns borrowed by hose wh debs. Increases in income, beer educaional qualificaions and improved prospecs regarding he financial suaion conribued o his resul. The major par of he overall increase in unsecured deb is no explained by variables a he individual level, bu is accouned for by common, unmodelled macroeconomic facors. Key words: Unsecured deb, Brish Household Panel Survey JEL classificaion: C21, D14, E21

Summary Unsecured borrowing by households, mainly in he form of personal loans, overdrafs and cred cards, has grown rapidly over he pas en years or so. This has raised concerns ha could cause widespread financial difficulies and defaul among households who migh sruggle o keep up wh heir deb repaymens. The validy of such concerns will depend o a large exen on he ype of people who have increased heir indebedness and wheher hey are borrowing more because heir economic circumsances have changed and hey feel more confiden abou aking on addional financial commmens. Borrowing for hese reasons is unlikely o be as risky as increased borrowing whou a change in underlying economic condions. This paper examines survey evidence on he deerminans and disribuion of unsecured deb using waves 5 and 10 for 1995 and of he Brish Household Panel Survey (BHPS). Previous work in he Bank has used he BHPS o analyse he overall financial posion of households, including he disribuion of unsecured deb across differen income and age groups. This paper looks in more deail a he deerminans of he cross-secional disribuion of unsecured deb and wheher his disribuion has changed over ime. Tha makes possible o assess wheher unsecured deb has increased because he facors deermining s use have changed or wheher more deb is held for given circumsances. One of he key risks associaed wh unsecured deb is ha is increasingly used by high risk borrowers. Despe he increased prevalence of cred cards, here is no evidence from he BHPS ha paricipaion in he unsecured deb marke rose beween 1995 and. In boh years, around 39% of people claimed o have some deb in his form. These may no be he same people, as he BHPS suggess ha 35% of he mos indebed quarile in 1995 had no unsecured deb in. Bu he evidence suggess ha here has been no subsanial change in he facors ha deermine wheher an individual is likely o have unsecured deb or no. In line wh sandard life-cycle consideraions, economeric analysis indicaes ha he main deerminan of he paricipaion decision is he age of he borrower, wh 20 o 30-year olds mos likely o borrow unsecured. Oher saisically significan facors are income, economic prospecs, qualificaions, job saus, housing saus and he exen of morgage borrowing. While here is no clear saisical evidence of a change in he deerminans of paricipaion in he unsecured cred marke beween 1995 and, here was, hough, a sriking increase in he amoun of deb held by borrowers beween hese wo years. According o economeric esimaes, he main deerminan of he level of unsecured borrowing of borrowers is he level of individual income. Age seems o be less imporan in deermining he amoun of unsecured borrowing han he decision o paricipae in he unsecured marke. The oher saisically significan deerminans of he amoun of borrowing are economic prospecs, qualificaions, job saus, housing saus and he exen of morgage borrowing. Bu, as wh he paricipaion decision, here is lle evidence of a major change in he imporance of hese deerminans beween 1995 and, alhough here does appear o have been a sligh increase in he relaive borrowing of hose wh high incomes. Insead, he main change beween hese years has been an increase in he amoun borrowed hroughou he disribuion. This suggess ha facors affecing all curren and poenial borrowers, regardless of heir personal characerisics, were mos imporan in explaining he rise in unsecured deb beween 1995 and.

Thus he rise in unsecured borrowing appears no o have been concenraed whin poor risk groups, bu o have been a general phenomenon affecing hose likely o be borrowers o a similar exen. While is no possible, on he basis of he informaion available, o explain he cause of his shif, is consisen wh lower raes of ineres on unsecured deb. According o he heory oulined in his paper, lower raes on unsecured deb would raise boh he unsecured and secured borrowing of hose unable o borrow as much as hey would like a secured ineres raes, whou encouraging furher borrowing by hose who are unlikely o paricipae in he unsecured marke. This would improve he welfare of hose who had been consrained by enabling hem o spread heir spending more smoohly over ime. Of course, more unsecured deb involves greaer risks even if deb is no concenraed among high risk groups. Some individuals do have very high levels of deb in relaion o heir income and ha exposes hem o he risk ha hey will no be able o repay. Bu here is no evidence ha his suaion worsened beween 1995 and.

1 Inroducion Borrowing by UK households has risen rapidly in recen years and by he end of 2004 he overall amoun owed was worh over 140% of annual pos-ax income. The majory of his deb is accouned for by morgages, secured on he borrower s home, bu he proporion ha is unsecured had reached around 20% of household income, almos double wha was in 1994. Unsecured deb, mainly in he form of cred cards, overdrafs and personal loans, differs from secured deb in erms of s purposes, cos, flexibily and risk. Tradionally, s main purpose has been o finance durable consumpion while secured deb has financed house purchase, bu he purposes for which deb is used are changing. During he 1990s, more use was made of unsecured deb o finance holidays, clohing or special occasions 1 while secured deb increasingly financed consumpion hrough morgage equy whdrawal [see Davey (2001)]. These developmens sugges ha deb is much less closely relaed o paricular purchases han in he pas. One of he facors behind shifs in he composion of borrowing is likely o be changes in he price a which people are able o borrow. Usually, unsecured borrowing is more expensive because of he greaer risk o lenders in he absence of collaeral. The greaer availabily of secured borrowing on beer erms would herefore encourage morgage equy whdrawal and he widespread subsuion of secured for unsecured deb. However, here is also an increasing amoun of aggregae unsecured deb ha does no bear any explic ineres, arising from purchases offering ineres-free cred or from he use of cred cards ha do no bear ineres if seled a he end of each monh. This would end o encourage he subsuion of unsecured for secured deb. Tha boh ypes of borrowing have risen rapidly a he same ime suggess ha hese relaive price effecs may have offse each oher. Bu also carries wh he possibily ha he risk of unsecured lending has changed as he characerisics of borrowers have alered. Recen research a he Bank of England [May, Tudela and Young (2004)] has described he disribuion of unsecured and secured deb across households in Brain as of Sepember 2004 using a specially commissioned survey. This paper aemps o go furher and assess wha lies behind he greaer use of unsecured deb by Brish households, since his poenially has implicaions for boh macroeconomic and financial sabily. I does his by means of a deailed invesigaion of he deerminans of borrowing a he individual level using informaion from he 1995 and waves of he BHPS. This aemps o clarify he ype of facors ha influence borrowing and wheher he imporance of hese facors has changed over ime. Is ha people are borrowing more because hey feel more confiden abou he fuure, or is simply more convenien for hem o finance spending in his way? Wha are he characerisics of borrowers and have hese changed recenly? Overall levels of borrowing can be analysed using life-cycle permanen income hypohesis models, where deb allows individuals o smooh consumpion over he life cycle and o finance he purchase of asses such as houses and consumer durables. Changing levels of borrowing can hen be explained whin his framework as a response o facors affecing spending, aking ino accoun cred consrains and oher supply-side facors ha migh influence he way in which spending is financed. In his paper we exend his model o ake accoun of differences beween secured and unsecured deb. Because secured deb is 1. Informaion on he purposes of unsecured borrowing is available from surveys of borrowers. The weigh of durable consumpion as he end-use of unsecured deb decreased from near 68% in 1995 o around 54% in 2002. Meanwhile, he weigh of he financing of holidays, clohes and special occasions increased from below 3% of he sock of unsecured deb in 1995 o near 10% in 2002. While canno accoun for he change in overall borrowing, he share of new loans used for loan consolidaion rose from around 6% o more han 12%. Source: NOP Financial Research Survey. BANCO DE ESPAÑA 11 DOCUMENTO DE TRABAJO N.º 0511

ypically cheaper han unsecured deb, will end o be used in preference o unsecured deb by individuals wh access o boh ypes of deb. This poins o he imporance in he empirical analysis of aking ino accoun he circumsances of he individual borrower, including heir posion in he housing marke. Our empirical analysis builds on previous household-level sudies, mainly concerned wh US households. For insance, Cox and Jappelli (1993) use he 1983 Survey of Consumer Finances (SCF) o esimae a cross-secional demand for deb equaion for US households. They find a posive relaionship wh permanen income and ne worh and a negaive relaionship wh curren income and age. Duca and Rosenhal (1993) also use he 1983 SCF and find ha he demand for deb of young households is posively relaed o wealh, income and household size and negaively relaed o unemploymen. Crook (2001) focuses on a more recen period and finds ha US households demand for deb is relaed posively o home ownership, family size, and job saus, while negaively relaed o ne worh, age and risk aversion. For Brish households, Bridges and Disney (2002) find ha he access o unsecured deb of low-income households is posively associaed wh income-relaed and income generaing characerisics. Banks e al. (2002) describe he disribuion of Brish household deb according o he BHPS as a par of a very comprehensive analysis of he disribuion of financial wealh in he UK in. Cox e al. (2002) also use he BHPS o analyse he changes in he disribuion of household deb-income raios, income and asses across borrowers and conclude ha he increase in deb-income raios of Brish households during he second half of he 1990s was larger among he younges and lowes-income households 2. The paper is organised as follows. Secion 2 exends a sandard life-cycle model of consumpion o ake accoun of he relaionship beween secured and unsecured borrowing. Secion 3 oulines he empirical mehod used. Secion 4 describes unsecured deb in he BHPS and examines he deerminans of deb in a cross-secional approach. Secion 5 focuses on deb changes using he panel dimension in he BHPS. Secion 6 concludes. 2. We invesigae he impac of unsecured deb on financial disress among Brish households in a separae paper [Del-Río and Young (2005)]. BANCO DE ESPAÑA 12 DOCUMENTO DE TRABAJO N.º 0511

2 The heoreical deerminans of unsecured deb One of he mos imporan characerisics of unsecured deb is ha is usually expensive relaive o oher possible mehods of finance, such as secured borrowing or running down asse holdings. This suggess ha s use migh be concenraed among hose who do no have access o cheaper finance. In his secion, we ouline a calibraed version of he life-cycle model of consumpion, where households are able o borrow a relaively low raes agains he secury of heir house, bu have o pay higher raes for unsecured borrowing 3. In conras o oher models, here are no quanaive cred consrains in he unsecured marke in his model, insead households are limed in he amoun hey can borrow a lower secured ineres raes. This model capures many of he key characerisics of he UK deb marke, in ha a borrower s financial decisions appear o be srongly ied o heir posion in he housing marke. The model provides a framework for undersanding he effec of facors which vary across households, hereby helping o explain cross-secional differences in indebedness, and facors which change over ime. Households are assumed o be economically acive for hree periods, reflecing differen phases of he life cycle. During his ime hey earn an exogenous income sream and consume non-durable goods and housing. They aim o maximise ineremporal uily and o die solven. Their ineremporal uily funcion a he beginning of heir lives is given by: V 0 = 3 = 1 α 1 α 1 ( h c ) ( 1 γ ) γ ( 1 + δ ) where h and c represen heir ownership of housing and consumpion of goods respecively, α is a parameer indicaing heir preference for housing relaive o goods, γ is he coefficien of relaive risk aversion and δ indicaes heir rae of ime preference. Households face he following flow budge consrain: (2.1) s + u = pc + qh + (1+r-1)s-1 + (1 + r-1 + η-1)u-1 - qh-1 - y, =1,2,3,4. (2.2) where s and u are socks of secured and unsecured deb respecively, y is exogenous nominal non-propery income, p and q are he prices of goods and housing respecively, r is he rae of ineres on secured deb and η is he premium on unsecured borrowing. I is assumed ha all households aim o die wh zero ne worh, so ha a he beginning of he period afer heir deah (a dae 4) he proceeds from he sale of he house is sufficien o pay off all remaining deb. Households can use secured and unsecured deb o smooh heir spending over ime. The use of secured and unsecured deb is assumed o be consrained such ha: s φ q h (2.3) 0 u (2.4) The firs expression saes ha secured deb canno exceed a proporion, φ, of he value of he household s house qh. The second expression saes ha unsecured deb canno be negaive (so households canno lend a high unsecured ineres raes). The choice of how much secured and unsecured deb o borrow is hen deermined joinly wh ha of how much o consume and how much o spend on housing. 3. No disincion is made a his sage beween a household and he individuals whin. Our empirical analysis considers he borrowing decisions of individuals raher han households. BANCO DE ESPAÑA 13 DOCUMENTO DE TRABAJO N.º 0511

Opimal housing and non-housing consumpion are derived by maximising (2.1) subjec o (2.2), (2.3) and (2.4). There are hree possible soluions a any dae, depending on which of he borrowing consrains are binding. These are refleced in he firs order condions for ineremporal consumpion over ime (2.5) and he choice beween housing and goods (2.6), wren for he case when he secured deb consrain is binding: v c v c v c v h + 1 + 1 ( 1 + r + η ) p = ( 1 + δ ) p i + 1 ( 1 + r + η ) p = ( 1 + r + η ) q q η q φ + 1 (2.5) (2.6) In his case, boh iner-emporal and inra-emporal consumpion decisions are affeced by he rae of ineres on unsecured borrowing. When he secured deb consrain is no binding, he premium on unsecured deb drops ou of hese condions. This has he effec of raising curren relaive o fuure consumpion and changing he effecive relaive price of housing. The hird possible oucome is a corner soluion where household borrowing violaes he secured borrowing consrain if consumpion choices can be made a secured borrowing raes, bu is whin he consrain when choices are made a higher unsecured raes. Depending on he exac specificaion of preferences, he model can be solved for he opimal ime profile of consumpion of housing and goods ha saisfy he budge consrain and he erminal condion. Ieraive mehods need o be used, because behaviour a any dae depends on wheher secured borrowing consrains are expeced o be binding in he fuure. The seady-sae soluion is illusraed here by means of a model calibraed roughly o he UK suaion. The hree periods of he model can be hough of as represening 15 years each. The resuls are shown in Table A for differen scenarios and for wo differen levels of he premium on unsecured borrowing. In he main case, income in he (15-year) firs period is 300,000, consisen wh annual income of 20,000. This rises o he equivalen of 40,000 per annum in he second sage of life, before falling back o 15,000 per annum in he las sage of life (which includes reiremen). The rae of ime preference has been se equal o he rae of ineres on secured deb (0.3, equivalen o 2% per annum) so ha in he absence of consrains individuals would smooh consumpion over heir life-cycle by borrowing when young, saving when in middle age and running down heir asses in old age and a deah. Bu he imposion of a lim on secured borrowing of up o 90% of he value of owned housing prevens individuals from reaching his opimum. In he case where he premium on unsecured deb is 0.1, consumpion of goods in he firs period is virually equal o income and he sock of housing is 54,400, jus over 2.5 imes annual income. This is financed by secured deb of 48,900, he maximum possible given he secured borrowing consrain. Unsecured borrowing in he firs period is relaively small a 500. Afer he firs period, he borrowing consrain in he model no longer binds, so ha individuals choose he same level of consumpion in he second and hird sages of life. The paern of income over ime means ha individuals build up financial asses during he second sage of life and run hese down in he hird sage, so ha when hey die he value of heir house is sufficien o pay off heir secured deb. Only in he firs period of life is any unsecured deb borrowed. In he same circumsances, bu where he premium on unsecured deb is lower a 0.05, individuals are beer able o smooh consumpion over ime. Consumpion is sill lower in he firs sage of life and higher hereafer because of he limaion on secured borrowing, bu is higher in he firs period han would have been wh a higher unsecured BANCO DE ESPAÑA 14 DOCUMENTO DE TRABAJO N.º 0511

borrowing premium. This is financed boh by higher secured and unsecured borrowing; secured borrowing is higher since individuals choose o buy a larger sock of housing, wh he addional amoun of consumer spending effecively financed by unsecured borrowing, which eases he secured borrowing consrain somewha. Noe ha despe he lower rae of ineres on unsecured borrowing, hose in he second and hird sage of life do no use eher because hey do no need o borrow or because cheaper secured borrowing is available. Table A: Comparaive saisics of calibraed model,housands Unsecured premium = 0.1 Unsecured premium = 0.05 Period c h s u y c H s u y Main case: 1 295.1 54.4 48.9 0.5 300 325.8 66.1 59.5 32.4 300 2 409.8 93.5-86.8 0 600 385.6 87.9-71.5 0 600 3 409.8 93.5 71.9 0 225 385.6 87.9 67.6 0 225 Higher secured borrowing lim ( φ = 0.95) 1 297.3 54.8 52 0 300 325.8 66.1 62.8 29.2 300 2 408.2 93.1-85.8 0 600 385.6 87.9-71.5 0 600 3 408.2 93.1 71.6 0 225 385.6 87.9 67.6 0 225 Low ownership of housing ( α =0.0005) 1 315.4 0.55 0.5 15.5 300 345.5 0.66 0.6 45.6 300 2 424.4 0.92-152.9 0 600 401.9 0.87-135.6 0 600 3 424.4 0.92 0.7 0 225 401.9 0.87 0.67 0 225 Low income expecaions 1 267.1 60.9 28.1 0 300 2 267.1 60.9 3.6 0 300 3 267.1 60.9 46.9 0 225 Less paience ( δ =0.4) 1 376.9 69.4 62.5 83.8 300 410.8 83.4 75.1 119.1 300 2 389.2 88.8 7.1 0 600 361.4 82.4 18.9 0 600 3 289.3 66 50.8 0 225 268.7 61.3 47.1 0 225 Main case: δ = 0.3, α = 0.05, γ =0.25, φ = 0.90. The second case we consider is of a higher secured borrowing lim of 95% of he value of he individual s house, raher han 90% as in he base case. The main effec is in causing individuals o subsue secured deb for unsecured deb wh lle or no noiceable effec on firs period spending in eher case. The reason for his negligible impac is ha he change in he borrowing lim does lle o alleviae he consrain. In he absence of any resricion on secured borrowing, individuals in he same circumsances would choose o inves 80,000 in housing and borrow 140,000, a loan o value raio of 175%, using he addional resources o finance consumpion of goods. Hence, he relaxaion of he consrain does lle o move individuals o heir opimum posion. The hird case shows he behaviour of hose who are no owner-occupiers because hey have no ase for home ownership (oher han o a rivial exen). Consumpion smoohing is prevened by he higher effecive rae of ineres on borrowing, which resuls in less consumpion han opimal being chosen in he firs sage of life. This is clearly less of a problem when he premium on unsecured borrowing is lower. In he cross-secion, in comparison wh hose who have a sronger preference for housing, unsecured borrowing is higher for hose who do no have access o he secured deb marke, reflecing a higher level BANCO DE ESPAÑA 15 DOCUMENTO DE TRABAJO N.º 0511

of non-housing consumpion, alhough heir overall level of borrowing is lower since hey do no have o finance he purchase of a house. Their ne worh is also lower since hey have no housing wealh. The fourh case illusraes he imporance of income expecaions on borrowing. Wh second sage income expeced o be he same as in he firs sage of life, is possible for he individual o smooh consumpion whou recourse o unsecured deb. Noe ha he sock of housing purchased in he firs sage of life is higher han for hose who have higher lifeime incomes bu are consrained from borrowing as much as hey would like, indicaing he effec of he borrowing consrain on he inraemporal consumpion decision. The fifh case shows he effec of less paience. In he high-unsecured premium case, his leads o a hump-shaped pah of consumpion, wh he premium on borrowing causing impaien individuals o resrain heir desire for presen consumpion. Despe his, heir unsecured borrowing exceeds heir secured borrowing. In he low unsecured premium case, here is no hump shape in consumpion as individuals are more able o achieve heir preferred consumpion pah. The implicaions of he model are ha unsecured deb is likely o be used more by hose who are young, impaien, wh srong income expecaions and no access o cheaper secured deb. I is likely o be used mos when unsecured borrowing coss leas. The model shows ha here are suaions where unsecured and secured deb end o move in oppose direcions in response o shocks, when he consrain on secured borrowing is relaxed for example, and when hey move ogeher, as when unsecured borrowing raes are reduced. BANCO DE ESPAÑA 16 DOCUMENTO DE TRABAJO N.º 0511

3 Esimaing he empirical deerminans of unsecured deb The previous discussion is inended o provide a framework for undersanding he deerminans of unsecured deb raher han a model o be esimaed. I suggess ha unsecured borrowing is likely o be relaed o he household s posion in he life-cycle, rae of ime preference, access o cheaper secured finance, income and income expecaions and he cos of unsecured borrowing. In his secion, we describe how he deerminans of he demand for unsecured deb may be esimaed empirically. Suppose ha he demand funcion for unsecured deb by individual i a dae, D, is of he following general form: D = a + f ( Y ) + b Z c r + ε (3.1) i u where ai is an individual-specific fixed effec, Y represens he income and oher economic circumsances of he individual, he funcion f(.) denoes ha he (ime-varying) relaionship beween D and Y could be non-linear, Z incorporaes he individual s demographic and oher personal characerisics, including age-relaed effecs, r u he individual-specific ineres rae levied on unsecured deb and ε he unobserved deerminans of unsecured deb. The coefficiens, oher han he fixed effec, ai, can poenially vary over ime. Supply condions in he unsecured cred marke are refleced in he effecive rae of ineres charged on unsecured deb. Where individuals are cred-consrained, he effecive rae can be hough of as being high enough o equae heir demand for unsecured loans o he supply. For example, he effecive ineres rae migh be given by: r u = r + ϕ g Y ) (3.2) ( where r is he base rae, ϕ is he premium ha financial insuions charge o he riskies individuals and he negaive (possibly non-linear) relaionship beween he effecive rae and individual income reflecs lower perceived lending risks a high levels of individual income. Subsuing (3.2) ino (3.1) hen gives a reduced form deb funcion: D = a + ( f ( Y ) + c g ( Y )) + b Z c ( r + ϕ ) + ε (3.3) i This expression helps o clarify a number of poins relevan o how is esimaed. Firs, in cross-secions, is impossible o esimae he individual specific fixed effecs, αi, and he inercep erm has o be imposed a he same value across all individuals. This means ha any genuine fixed effecs become par of he error erm. If hese are correlaed wh any of he explanaory variables, he relevan coefficiens will be biased. So if individuals wh a paricularly high rae of ime preference also choose o work more and so earn higher incomes, he esimaed coefficien on income will oversae he response of unsecured borrowing o a change in income. Second, also in cross-secions, here is no variaion across individuals in macroeconomic variables such as he base rae of ineres, r, so heir impac becomes par of he overall inercep erm. Third, we allow he relaionship beween unsecured deb and income o be non-linear. In cross-secion esimaion below we do his by esimaing separae inercep coefficiens according o he individual s posion in he income disribuion. We use a similar approach o esimae age effecs. Fourh, wh panel daa wh a sufficienly long ime dimension, is possible o avoid he biases due o individual-specific fixed effecs and o idenify he impac of macroeconomic facors, bu is also necessary o assume ha coefficiens are eher consan over ime, or have a relaively simple srucure. BANCO DE ESPAÑA 17 DOCUMENTO DE TRABAJO N.º 0511

In fac, mos of our esimaion effor is in esimaing cross-secion regressions for 1995 and, bu we do make some use of he limed panel informaion available as a check ha he cross-secion resuls are no srongly affeced by correlaion beween individual fixed effecs and explanaory variables. A furher economeric issue is how o deal wh he fac ha mos people in our sample do no paricipae in he unsecured deb marke. This could arise, as in he heoreical model, because hey do no wan any unsecured deb a prevailing ineres raes; could also reflec quanaive cred consrains or high enry coss ha preven hem reaching heir desired deb posion. These wo possibilies presen differen economeric problems [see Wooldridge (2002)]. In he former case, he economeric issue is ha many observaions are a a corner soluion, in his case wh no unsecured deb, and simply esimaing he parameers in (3.3) over all households, including hose whou deb, using say a Tob approach would be highly influenced by paricipaion decisions. In he laer case, he economeric issue arises because of possible non-random sample selecion ha prevens some individuals paricipaing in he marke. In principle, such biases can be avoided by following he approach of Heckman (1979), which involves esimaing a model of cred marke paricipaion, where his is condioned on facors addional o hose ha deermine he amoun of deb borrowed. Sudies similar o ours, such as Duca and Rosenhal (1993), Cox and Jappelli (1993) and Crook (2001), use a wo-sep Heckman procedure ha involves including wo addional erms in he deb equaion o capure, firsly, he probabily of an individual paricipaing in he cred marke, and secondly, heir no being cred consrained. In he firs sep, prob models are used o esimae he probabily of paricipaing in he marke and he probabily of being unconsrained 4. Then, he esimaed effecs are included as addional regressors in (3.3), so ha he parameers can be inerpreed as hose of a rue demand funcion. We do no follow his approach for wo reasons. Firs, unlike he US surveys, he BHPS does no provide any direc measure ha would make possible o discriminae beween consrained and unconsrained individuals, alhough is unlikely ha many individuals in he UK are unable o borrow a all 5. Second, he implemenaion of he Heckman procedure is que problemaic whou a srong heoreical case for supplemenary variablesha affec he paricipaion decision bu do no influence he amoun borrowed 6. Parly for his reason, our resuls from using a Heckman approach are no differen o hose from simple OLS cross-secion regressions 7. This also suggess ha any corner-soluion biases are small. Given his and he resuls in previous sudies we will focus separaely on he paricipaion equaion (using a prob model) and on deb equaions, using simple OLS cross-secion regressions for hose wh deb, excluding non-paricipans. 4. According o Jappelli (1990), cred consrains affec 19% of households and as many as 30% of young households [Duca and Rosenhal (1993)]. Oher relaed works are, Zeldes (1989), Cox and Jappelli (1993), Crook (1996), Gross and Souleles (2002). For he Uned Kingdom, Davies and Weber (1991), using household-level daa and idenifying unconsrained households as hose wh savings, found evidence of declining liquidy consrains bu no of loosening cred conrains in he 1980s. Bayoumi (1993) found ha sofer liquidy consrains due o financial deregulaion during he 1980s had a significan effec on UK consumpion. More recenly, Fernández-Corugedo and Muellbauer (2002) esimaed an index of non-price cred condions, and found evidence of looser cred resricions during he 1980s and second half of he 1990s. 5. One possible proxy is given in wave 5, when individuals sae wheher hey hink ha was a righ ime o use cred in he hypoheical case ha hey waned o buy somehing big. One of he possible answers o his quesion was Can ge cred and ha was only seleced by 2.4% of 8,774 respondens. 6. Cox and Jappelli (1993) used years of educaion, occupaion, area income, employmen saus, and rural/urban saus as supplemenary variables for he probabily of having posive deb. Duca and Rosenhal (1993) and Crook (1996), by conras, assumed ha he same variables deermined he probabily of having deb and he amoun borrowed (allowing for differen parameers in he paricipaion and deb equaions). 7. Resuls using he Heckman procedure are available upon reques. When conducing he wo-sep Heckman procedure, we have added dummies for region, race and employmen saus o he paricipaion equaion, as facors ha migh influence paricipaion in he unsecured deb marke whou having much effec on he overall amoun borrowed. BANCO DE ESPAÑA 18 DOCUMENTO DE TRABAJO N.º 0511

4 A cross-secional analysis of he deerminans of unsecured deb 4.1 The daa The BHPS 8 is an annual naional survey of he economic and demographic characerisics of Brish individuals and households. The firs wave covered a represenaive sample of he populaion of Grea Brain in 1991. This sample has remained broadly represenaive 9 given ha he same individuals are re-inerviewed each year and, if hey spl-off from original households, all adul members of heir new households are also inerviewed. In 1991 he survey included around 5,500 households and 10,300 individuals (aged over 16 years). Informaion on unsecured deb and financial asses is available only in waves 5 and 10 of he BHPS covering 1995 and. For unsecured deb, individuals are asked abou he overall amoun hey owe, excluding cred card and oher bills being paid off in he monh of he inerview 10. They are shown a card o promp hem abou he forms in which hey may have borrowed. In 1995, he promp card conained he following lis of deb insrumens: hire purchase agreemens, personal loans (from bank, building sociey or oher financial insuion), cred cards, caalogue or mail order purchase agreemens, DSS Social Fund loan, any oher loan from a privae individual, or anyhing else. In, wo addional insrumens, overdrafs and suden loans, were added o his lis. This change in he lis of unsecured deb insrumens affecs any analysis aemping o compare responses across he wo waves of he survey. As boh ypes of insrumen were available in 1995, is no clear how respondens wh overdrafs or suden loans would have included his ype of borrowing in heir answers o he survey a ha ime whou being promped. For example, hey could have considered borrowing on overdrafs as a form of personal loan. Bu he change in quesion mus leave room for doub ha his was he case. As shown in Table B, overdrafs represened nearly 7% of he oal number of deb insrumens menioned in. Suden loans were a less significan 1% of oal deb insrumens. If borrowing using hese insrumens were enirely omed in 1995, bu no, hen a comparison would oversae he increase in unsecured household deb. There is some evidence agains his in ha Redwood and Tudela (2004) find ha unsecured deb is more underrepored relaive o aggregae figures in han in 1995. This migh sugges ha he new lised insrumens in were included in oher caegories in 1995. Throughou his analysis we assess he sensivy of esimaes o his poenial problem by changing he sample in. 8. The Brish Household Panel Survey (BHPS) is managed by he ESRC UK Longudinal Sudies Cenre wh he Insue for Social and Economic Research a he Universy of Essex. Deailed informaion can be found in Brice e al. (2002), available a hp://www.iser.essex.ac.uk/bhps/. 9. The sample excludes households locaed norh of he Caledonian Canal in Scoland. Since 1997, new samples have been added o he BHPS aimed a exending he coverage of some paricular regions and groups of populaion. We exclude hem o keep he sample represenaive of he Brish populaion as a whole. 10. If individuals do no know he exac amoun hey owe, hey are asked o indicae wheher is more han 100, more han 500, more han 1,500, or more han 5,000. Depending on he case we assign a deb of 50, 300, 1,000, 3,250 or 7,000. This affecs 310 borrowers (ou of 6,889). If individuals repor ha he deb is a join commmen we assign half of he value. In we can discover which par of he deb is a sole commmen bu we do no use his informaion since is no available for 1995. Join commmens affec 984 and 709 individuals ou of 3,481 and 3,458 debors in 1995 and respecively. For each year, all unsecured deb values above he 99 h percenile are recorded o he value of he 99 h percenile. This is also done for unsecured deb-income raios. BANCO DE ESPAÑA 19 DOCUMENTO DE TRABAJO N.º 0511

Table B: Number of unsecured deb insrumens by age group 1995 age hire personal purchase loan cred cards mail order purchase DSS Social Fund loan loans from individuals somehing else overdraf suden loan Toal Toal (%) 16-20 18 74 30 74 11 15 5 - - 227 4.7% 20-25 110 258 144 136 19 33 8 - - 708 14.7% 25-30 146 239 189 149 17 28 7 - - 775 16.1% 30-35 192 210 215 179 12 24 3 - - 835 17.3% 35-40 130 166 157 128 7 14 7 - - 609 12.6% 40-45 115 138 150 89 10 6 5 - - 513 10.6% 45-50 101 120 139 95 2 9 6 - - 472 9.8% 50-55 59 73 104 63 3 5 4 - - 311 6.5% 55-60 49 31 48 25 1 1 - - 155 3.2% 60+ 58 26 76 47 1 3 4 - - 215 4.5% Toal 978 1335 1252 985 83 138 49 4820 100% Toal (%) 20.3% 27.7% 26.0% 20.4% 1.7% 2.9% 1.0% 100 age hire personal purchase loan cred cards mail order purchase DSS Social Fund loan loans from individuals somehing else overdraf suden loan Toal Toal (%) 16-20 13 44 51 51 10 11 56 94 6 336 5.8% 20-25 65 175 170 87 15 28 152 173 11 876 15.1% 25-30 142 233 238 114 20 25 113 77 9 971 16.7% 30-35 136 239 257 134 15 17 106 28 3 935 16.1% 35-40 136 212 278 125 14 19 76 12 6 878 15.1% 40-45 91 157 170 86 8 4 63 3 6 588 10.1% 45-50 72 110 124 55 3 4 33 3 8 412 7.1% 50-55 57 85 121 58 2 3 29 2 8 365 6.3% 55-60 28 61 69 44 3 17 1 3 226 3.9% 60+ 46 39 76 53 11 6 231 4.0% Toal 786 1355 1554 807 90 111 656 393 66 5818 100% Toal (%) 13.5% 23.3% 26.7% 13.9% 1.5% 1.9% 11.3% 6.8% 1.1% 100 A likely reason for he addion of he suden loan caegory in he survey is ha loans had by hen become he main form of financial suppor for sudens. Up o and including academic year 1997/98 sudens were funded under a differen se of arrangemens, inroduced in 1990/91, when non income-assessed suden loans were inroduced o provide exra resources owards living expenses and parially o replace grans. The main gran raes were frozen a heir 1990/91 values unil 1994/95 when he shif from gran o loan was acceleraed by reducing he level of gran raes and increasing loan raes. Furher deails on he exen of suden loan finance are provided by Callender and Wilkinson (2003). This shif in he suden finance regime owards loans is also likely o disor unsecured deb marke paricipaion and borrowing, especially among individuals and heir families who have been sudens during he new regime. Again, we ry o avoid he disorion by varying he sample o exclude hose affeced. 4.2 Preliminary daa descripion Table C shows ha he proporion of individuals reporing ha hey had any unsecured deb did no change beween 1995 and, wh around 39% of individuals who answered his quesion claiming o have a leas one form of unsecured deb in boh years 11. Significanly, among hose wh some unsecured deb, he mean amoun almos doubled from 1,489 in 1995 o 2,793 in. Indeed, unsecured deb approximaely 11. Abou 5% of individuals did no answer his quesion in boh 1995 and. BANCO DE ESPAÑA 20 DOCUMENTO DE TRABAJO N.º 0511

doubled a mos poins of he disribuion wh he median rising from 700 per debor in 1995 o 1,500 in and he 90 h percenile rising from 4,000 o 8,000. The raio of unsecured deb o income also rose a mos poins of he disribuion. For individuals wh some deb, he median of he raio increased from 8% in 1995 o 12% in while he 90 h percenile rose from around 42% in 1995 o 70% in, alhough he increase is more modes when we exclude full-ime sudens from he sample 12. Table C: Individual deb levels and deb-income raios of deb-holders 1995 (a) 1995 (a) Excluding full ime sudens Individuals wh no deb 5,353 5,182 5,182 4,899 4,890 Individuals wh deb 3,431 3,458 3,004 3,276 3,275 Proporaion of debors 0.39 0.40 0.37 0.40 0.40 Deb (Levels ) sample size 3,265 3,227 2,827 3,133 3,089 mean 1,489 2,973 2,708 1,492 2,936 10 h percenile 60 100 100 60 100 30 h percenile 250 500 500 250 500 50 h percenile 700 1500 1500 700 1500 70 h percenile 1,600 3,500 3,000 1625 3,500 90 h percenile 4,000 8,000 7,500 4,000 8,000 Deb-income raio (%) (b) sample size 3265 3257 2827 3133 3089 mean 21 39 24 18 31 10 h percenile 1 1 1 1 1 30 h percenile 3 5 4 3 5 50 h percenile 8 12 10 7 11 70 h percenile 17 26 22 16 24 90 h percenile 42 70 54 39 61 (a) The number of households wh unsecured deb excludes families whose deb is only in he form of suden loans or overdrafs. Deb and deb-income raios are calculaed excluding households wh overdrafs or suden loans, no maer if hey have oher ype of insrumens. (b) Excluding individuals wh income below 100 As suggesed by he life-cycle model and he simple model of Secion 2, here are clear differences in unsecured deb marke paricipaion by age. Table D shows ha in boh 1995 and, around 60% of individuals aged 20 o 35 years old had a leas one form of unsecured deb. This fracion decreases wh age o 10% for individuals older han 60. There is also a clear increasing relaionship beween unsecured deb marke paricipaion and income which is similar in boh 1995 and 13. 12. Noe ha individuals provide he oal amoun of deb hey owe and he differen classes of insrumens hey use. There is no informaion on deb by insrumen. Therefore, in he hird column of Table C, when excluding households wh overdrafs we are also excluding he deb ha hese borrowers may hold in oher insrumens. Since individuals wh overdrafs are usually high debors hese figures migh be biasing downward he rue figure. 13. Income groups are deciles of he income disribuion of he oal sample in 1995. In decile values are updaed wh he Reail Price Index. BANCO DE ESPAÑA 21 DOCUMENTO DE TRABAJO N.º 0511

Table D: Proporion of borrowers and sample weighs by age and income. Proporion of borrowers Sample weighs of Excluding f each age and income % Toal sample sudens group 1995 (a) 1995 1995 age groups 16-20 24 33 26 34 34 9 8 20-25 58 60 57 59 59 9 9 25-30 58 62 61 58 62 10 10 30-35 57 57 56 57 57 11 10 35-40 54 55 54 54 55 9 10 40-45 48 49 49 49 49 8 9 45-50 43 41 41 43 41 9 8 50-55 40 33 33 40 33 7 9 55-60 28 31 31 28 31 6 6 60+ 10 11 11 10 11 21 21 Toal 39 40 39 39 40 100 100 income groups (by deciles) 1 24 30 25 30 31 10 9 2 29 29 27 29 27 10 8 3 29 30 28 28 28 10 7 4 32 29 29 31 29 10 8 5 36 31 30 36 30 10 10 6 40 39 39 39 39 10 11 7 48 44 43 48 44 10 11 8 49 49 49 49 49 10 11 9 55 55 55 55 55 10 12 10 50 50 49 50 50 10 12 Toal 39 40 39 39 40 100 100 (a) Households wh overdrafs or suden loans and wh no oher ype of unsecured deb are excluded There is no significan change in he overall paricipaion rae beween 1995 and. Whin individual age groups, here is an increase in paricipaion among hose who are under 30 ha is paricularly marked for 16 o 20 year olds. Much of his is likely o reflec he shif in suden finance beween 1995 and as here is no change in paricipaion for 16 o 25 year olds once full-ime sudens are excluded (fourh and fifh columns of Table D). Similarly, he apparen increase in paricipaion among he lowes income group appears o be due o greaer paricipaion by full-ime sudens. Once hey are excluded, here is no change in paricipaion in he lowes income group. In conras o paricipaion, here is no clear evidence of any sysemaic effec of age on he amoun borrowed (see op panels of Char 1 and Char 1b), apar from for he oldes and younges groups who end o borrow less. The amoun borrowed ends o rise in line wh he level of income, wh he unsecured deb o income raio being fairly similar across all bu he lowes income groups, who have by far he highes levels of unsecured deb in relaion o income 14. 14. In Cox e al. (2002) he unsecured deb-income raio seems o be negaively correlaed wh age and income. Discrepancies can arise since heir sudy focuses on households, no on individuals, and income variables and groups can differ. In addion our analysis excludes all new samples in he BHPS since 1997. BANCO DE ESPAÑA 22 DOCUMENTO DE TRABAJO N.º 0511

Deb levels and deb-income raios are significanly higher for all groups in han in 1995. The increase seems o be more imporan for he lowes income decile and hose beween 20 and 25, alhough his appears o be affeced by he change in BHPS quesionnaire and mehods of suden finance. When hose wh only overdrafs or suden loans are excluded he increase in deb levels is much more modes for hese groups (see dashed lines in Char 1) 15. However, excluding hese groups may leave ou people wh high deb levels for oher reasons. Some evidence in favour of his view is ha when full-ime sudens are excluded (Char 1b), here remains a large increase in deb for hose wh low incomes and hose beween 20 and 25. Char 1: Unsecured deb levels and deb-income raios by age and income 4000 Mean unsecured deb of debors (consan prices 1995) 25% Unsecured deb-income raio (Median) 3500 3000 1995 (a) 20% 1995 (a) 2500 15% 1500 10% 1000 500 5% 0 16-20 20-25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60+ 0% 16-20 20-25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60+ age age 4000 3500 3000 2500 Mean unsecured deb of debors (consan prices 1995) 1995 (a) 30% 25% 20% (0.61) Unsecured deb-income raio (Median) 1995 (a) 15% 1500 1000 10% 500 5% 0 1 2 3 4 5 6 7 8 9 10 income deciles 0% 1 2 3 4 5 6 7 8 9 10 income deciles (*) 0.60 All figures are calculaed excluding individuals wh income below 100 pounds (a) Excluding households wh overdrafs or suden loans 15. The increase in deb for all deciles is saisically significan. BANCO DE ESPAÑA 23 DOCUMENTO DE TRABAJO N.º 0511

Char 1b: Unsecured deb levels and deb-income raios by age and income Sample excluding full ime sudens 4000 Mean unsecured deb of debors (consan prices 1995) excluding full ime sudens 25% Unsecured deb-income raio (Median) excluding full ime sudens 3500 1995 20% 1995 3000 2500 15% 1500 10% 1000 5% 500 0 16-20 20-25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60+ age 0% 16-20 20-25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60+ age 4000 3500 Mean unsecured deb of debors (consan prices 1995) excluding full ime sudens 1995 50% 45% 40% Unsecured deb-income raio (Median) excluding full ime sudens 1995 3000 35% 2500 30% 1500 25% 20% 15% 1000 10% 500 5% 0 1 2 3 4 5 6 7 8 9 10 income deciles All figures are calculaed excluding individuals wh income below 100 pounds 0% 1 2 3 4 5 6 7 8 9 10 (*) 0.95 income deciles Table E and Char 2 show he relaionship beween unsecured deb and various measures of housing wealh and herefore secured deb capacy. There is a clear negaive relaionship beween paricipaion in he unsecured deb marke and ne housing wealh 16, bu he relaionship beween he amoun borrowed and ne housing wealh is less clear (see Char 2). In 1995, he unsecured deb-income raio appears o be independen of ne housing wealh. The former increased mos beween 1995 and for hose wh low housing wealh, such ha here is a sligh decreasing relaionship in. 16. Ne housing wealh is he value of he residenial house ne of morgages. Since hese are household variables in BHPS we assign half he value of he house and morgage o he firs and second person owning he accommodaion. BANCO DE ESPAÑA 24 DOCUMENTO DE TRABAJO N.º 0511

Table E: Proporion of borrowers and sample weighs by age and income. Proporion of borrowers Excluding f % Toal sample sudens 1995 (a) 1995 Ne housing wealh no housing wealh 37 41 38 38 39 < percenile10 67 66 66 66 67 10-30 55 55 55 55 54 30-50 39 39 39 39 39 50-70 32 32 31 31 31 70-90 26 24 23 23 25 more han 90 21 21 21 21 21 Toal 38 39 38 38 39 Housing saus Oher 40 43 41 41 40 Owner occupier, no morgage 16 16 15 15 16 Living wh owner-occupiers 23 29 26 26 25 Owner occupier wh morgage 53 53 52 52 53 Living wh morgagers 34 40 37 37 42 Toal 39 40 39 39 40 Morgage-debors < percenile 20 42 41 41 41 43 20-40 51 54 53 53 52 40-60 60 60 60 60 60 60-80 60 59 58 58 60 more han 80 53 56 56 56 53 Toal 53 54 54 54 53 Financial wealh no financial wealh 41 41 39 39 42 quarile 1 51 55 54 54 55 quarile 2 45 50 48 48 48 quarile 3 34 38 36 36 35 quarile 4 25 26 25 25 25 Toal 40 42 41 41 41 (a) Households wh overdrafs or suden loans and wh no oher ype of unsecured deb are excluded BANCO DE ESPAÑA 25 DOCUMENTO DE TRABAJO N.º 0511

Char 2: Unsecured deb levels and deb-income raios by housing-wealh and secured deb 3500 3000 2500 1500 1000 500 Mean unsecured deb of debors (consan prices 1995) 1995 (a) 0 no housing < 10-30 30-50 50-70 70-90 more han wealh percenile10 90 16% 14% 12% 10% 8% 6% Unsecured deb-income raio (Median) 4% 1995 2% (a) 0% no housing < 10-30 30-50 50-70 70-90 more han wealh percenile10 90 3000 Mean unsecured deb of debors (consan prices 1995) 18% 16% Unsecured deb-income raio (Median) 2500 14% 12% 10% 1500 1000 500 0 Oher 1995 (a) Owner occupier, no morgage Living wh owneroccupiers Owner occupier wh morgage Living wh morgagers 8% 6% 4% 2% 0% Oher Owner occupier, no morgage Living wh owneroccupiers Owner occupier wh morgage 1995 (a) Living wh morgagers 3500 3000 2500 Mean unsecured deb of debors (consan prices 1995) 16% 14% 12% 10% Unsecured deb-income raio (Median) 8% 1500 6% 1000 500 0 < percenile 20 1995 (a) 20-40 40-60 60-80 more han 80 4% 2% 0% < percenile 20 1995 (a) 20-40 40-60 60-80 more han 80 All figures are calculaed excluding individuals wh income below 100 pounds (a) Excluding households wh overdrafs or suden loans BANCO DE ESPAÑA 26 DOCUMENTO DE TRABAJO N.º 0511

Turning o housing saus, owner-occupiers wh a morgage have a higher propensy o hold unsecured deb han oher groups in boh 1995 and. They also have higher amouns of unsecured deb han owners wh no secured deb. Furher, among households wh morgages, here seems o be a weak posive correlaion beween he level of unsecured deb and he secured deb-income raio in and a sronger posive relaionship beween he unsecured deb-income raio and he secured deb-income raio in. These general relaionships are broadly consisen wh he heoreical model of Secion 2, alhough is no clear why hose wh relaively low secured deb-income raios choose o have any unsecured deb raher han seeking o increase heir lower-cos secured deb 17. As regards financial wealh 18, here seems o be a negaive relaionship beween he size of financial asses and paricipaion in he unsecured deb marke (see Table E and Char 3) 19. In, here is a relaively clear decreasing relaionship beween unsecured deb-income raios and financial wealh 20. Those wh a low level of financial asses are more likely o hold unsecured deb o finance consumpion. All hese figures poin o a que generalised increase in he average unsecured deb of borrowers beween 1995 and, while paricipaion raes were broadly unchanged. Char 3: Unsecured deb levels and deb-income raios by financial wealh 3000 Mean unsecured deb of debors (consan prices 1995) 14% 12% Unsecured deb-income raio (Median) 2500 10% 8% 1500 1000 500 1995 (a) 6% 4% 2% 1995 (a) 0 Zero fin.wealh 1 2 3 4 Quariles 0% Zero fin.wealh 1 2 3 4 Quariles 17. The increase in he unsecured deb of hose wh relaively high morgage deb-income raios could be relaed o a change in he morgage marke in 1998. From ha ime, borrowers wh a secured loan o housing value raio of less han 0.9 were exemped from paying for morgage indemny insurance. This migh have caused some borrowers o subsue unsecured for secured borrowing. Fernández-Corugedo and Muellbauer (2002) esimae ha his raised he long-run sock of aggregae unsecured cred by 8%. 18. This variable does no include asses in he form of pension funds or insurance producs. 19. Financial wealh groups are perceniles for hose wh posive financial asses. We consider separaely hose wh no financial wealh. We assign equal shares if savings are held joinly. 20. The same paern is observed when considering only liquid financial asses. BANCO DE ESPAÑA 27 DOCUMENTO DE TRABAJO N.º 0511