The role of demand and supply in cyclical fluctuations of

household debt in Croatia

Ivana Herceg**Views expressed in this paper are solely those of the author’s

The 17th Dubrovnik Economic Conference, June, 2011

Motivation

"credit lies at the heart of the crisis" (Aikman, 2010)

theory of banks pro-cyclical behaviour typically associated with Minsky (1992): "from time to time, capitalist economies exhibit inflations and debt deflations which seem to have the potential to spin out of control. In such processes the economic system's reactions to a movement of the economy amplify the movement-inflation feeds upon inflation and debt deflation feeds upon debt deflation“

Motivation

recent evidence shows that fluctuations in the demand for loans rather than supply of loans, may be more important driver of aggregate household borrowing (Calem et al, 2011)

AIM: to identify whether recent fluctuations of household borrowing in Croatia were primarily caused by the more restrictive banks' lending policies or changes in household demand for loans

Methodology

two segments of lending policies are separately modelled in two steps:

criteria that households should satisfy in order to qualify for the loan (loan accessibility)

the highest loan amounts banks are willing to approve to the eligible households (loan amount availability)

Methodology – 1. step AIM: estimation of the probabilities for existence of

loan demand and loan supply for each household in the sample

methodological framework: full information availability - binary dependant variable

model probability of loan existence intersection of observable

demand and supply

non-existence of full information matrix - partial observability model

1 0

1 1 0

0 0 0

Demand

Supply

1itL

Methodology – 1. step

partial observability model (Poirier, 1980) only dummy variable of the loan existence is observable the probability distribution of the loan existence - normally

distributed bivariate process representing binary choice of credit demander and supplier

)0,0Pr()1,1Pr()1Pr( ** ititititit SDSDL

itL

),,()1Pr( sit

ditit XXFL

itditit eXD *

10* itit DD

itsitit XS * 10* itit SS

bivariate normal distribution correlation between

ite and

it

Methodology – 1. step

partial observability model (Poirier, 1980) and need to differ in at least one variable (the

exclusion restriction)

ditX

sitX

Demand Supply

Number of children Rural area of residence

Number of household members

Home ownership

Child birth Marital status

Amount of remittances Fixed-term working contract

Rural area of residence Life insurance investment

Financial difficulties in servicing living costs

Previously taken loans

Methodology – 2. step

AIM: estimation of the maximum loan amount that could be approved to the household that satisfies loan approving conditions

methodological framework stochastic frontier analysis (SFA)

distributional assumptions doesn’t allow correction for sample selectivity bias difficult to disentangle the impact of the banks' credit

policies changes from the improvement of the households' creditworthiness

quantile regression (QR) alleviates all mentioned disadvantages imposed by the SFA

Methodology - conjunction of two steps

sample selection correction, i.e. probability of loan existence, is incorporated into credit limit estimation (QR)

Heckman two-step correction procedure

Machado-Mata (MM) decomposition technique separate the effect of improved households'

creditworthiness from the effect of changed banks' lending policies

Data

Household Budget Survey (2008 and 2009)

sample of newly indebted households in 2009 percentage of newly indebted households

decreased in 2009 average amount of newly granted loans

decreased

characteristics of the newly indebted households in both years –similar

change in type of newly granted loans

Results-only loan amount availability segment

0

30

60

90

120

150

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96

credit limit percentiles

%

both the SFA and the QR analysis indicate that the credit limits generally increased between 2008 and 2009, with decreasing rise of credit supply in relation to the loan size

-40

-20

0

20

40

60

80

100

120

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96

credit limit percentile

%

credit limit change lending policies creditw orthiness

SFA approach QR approach

Results – only loan accessibility segment

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

1 23 45 67 89 111 133 155 177 199 221 243 265 287 309 331

pro

bability

2008_standards 2009_standards

trend_2009 standards trend_2008 standards

probability for loan existence decreased in 2009 due to combination of both supply, i.e. more rigorous banks' loan approving standards in household selection process, and demand, i.e. impaired households' propensity to borrow

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1 23 45 67 89 111 133 155 177 199 221 243 265 287 309 331

prob

abilit

y

2008_standards 2009_standards

trend_2009 standards trend_2008 standards

Demand Supply

Methodology – conjunction of two segments

in 2009 banks have tightened their lending standards and made access to credit more difficult - pro-cyclical behaviour

on average higher maximum loan amounts were made available to creditworthy households - counter-cyclical behaviour

increase in creditworthiness of household that were able to obtain loan - fled to quality

-20

0

20

40

60

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96

credit limit percentile

%

credi limit change lending policies creditw orthiness

Results-available credit limit usage

6

7

8

9

10

11

12

13

14

15

1 21 41 61 81 101 121 141 161 181 201 221 241 261 281 301 321

household

ln(n

ew lo

an)

credit limit new ly granted loan amount

in 2009 credit limits were used less compared to 2008 as a reaction to recession, households changed their consumption

patterns that led to reduced reliance on bank lending

6

7

8

9

10

11

12

13

14

1 21 41 61 81 101 121 141 161 181 201 221 241 261

household

ln(n

ew lo

an)

credit limit new ly granted loan amount

2008 2009

Conclusions

as a reaction to adverse economic developments banks only partially tightened their lending policies

pro-cyclical behaviour was principally conducted via process of household selection, while at the same time the maximum loan amounts made available to credit-worthy households even increased

relatively smaller exploitation of the available credit limits in 2009

household demand had the prominent role in cyclical fluctuations of household debt during 2009