Transcript Information Sharing, Lending and Default: International

CENTRO STUDI
C ENTRE
FOR
DI
S T U D IE S
ECONOM IA
IN
E
E C O N O M IC S
FINANZA
AND
F IN A N C E
D IPARTIMENTO DI SCIENZE ECONOMICHE
U NIVERSITÀ DEGLI STUDI DI SALERNO
Information sharing in credit markets: a survey
Tullio Jappelli - Marco Pagano
Conference on The Economics of Consumer Credit:
European Experience and Lessons from the US
European University Institute, 13/14 May 2003
1
Outline
A. The role of credit information systems
• Reducing adverse selection
• Disciplinary effect of default disclosure
• Preventing over-borrowing by multiple lenders
B. Empirical evidence
• Historical
• Macro
• Micro
• When is public intervention useful?
C. Issues in the design of credit information systems (transition economies)
2
Asymmetric information in credit markets
• The data needed to screen credit applications and to monitor borrowers
are not freely available to banks. In their absence, banks face adverse
selection and moral hazard problems.
• Adverse selection arises when the lender does not observe some
characteristics of the borrower. Moral hazard when he does not
observe certain actions by the borrower, which affect the probability of
repayment: for instance, the borrower’s effort to manage his project
and avoid default
• In either case, asymmetric information between borrowers and lenders
can prevent the efficient allocation of credit.
3
How can banks mitigate adverse selection
and moral hazard problems?
•
•
•
•
Collateral
Large equity stakes
Reputation
Acquisition of information … but how?
Alternatives:
1. Acquire information directly by spending resources, or
2. Acquire it by exchanging it with other lenders.
Exchange what?
• Credit history
• Amount of debt
• Other information
4
Information sharing arrangements
• Private Credit Bureaus (especially in Anglo-Saxon countries).
Principle of reciprocity. They collect, file and distribute data supplied
voluntarily by members
• Public Credit Registers (PCR) managed by central banks
(Continental Europe and Latin America). Their data are compulsorily
reported by lenders. Universal coverage of banking institutions.
• Credit bureaus focus on consumer and small business loans, which are
generally ignored by PCR.
• Conversely, PCR often provide a more complete picture of large
corporate loans.
5
Type of information shared
• Black or negative information: defaults and arrears.
•
White or positive information:
- debtor’s current overall loan exposure and guarantees
- past credit history other than defaults and arrears
- other characteristics, such as employment, income or line of business.
• Often credit bureaus process these data, assigning a credit score to
borrowers based on statistical risk analysis.
• There is great international variability in the type of data reported and
distributed by private credit bureaus and PCR.
6
The World Bank survey of credit
information sharing
• In 1999-2001 the World Bank conducted a survey of credit bureaus
and PCR.
• PCR: 41 countries out of 77 who replied to the survey operate a PCR.
Most Latin American countries, 7 European, many former socialist
countries. Mostly in the 1990s. Oldest in Germany (1934), France
(1946), Italy and Spain (1962).
• Prevalence of minimum reporting threshold in PCR, with wide
international variability
• Private credit bureaus: Anglo-Saxon countries, Europe, East-Asia.
• Market segmentation in the US: Dun and Bradstreet for commercial
loans, specialized credit bureaus for consumers. Not elsewhere.
7
The four effects of information sharing
1.
Credit bureaus improve banks’ knowledge of applicants’
characteristics and permit more accurate prediction of repayment
probability.
2.
Credit bureaus reduce the informational rents that banks could
otherwise extract from their customers.
3.
Credit bureaus work as a borrower discipline device: every
borrower knows that a default damages his reputation with other
potential lenders, cutting him off from future credit or making it
more expensive.
4.
Credit bureaus eliminate the incentive to over-borrow when
borrowers draw credit simultaneously from many banks without any
of them realizing.
8
1. Reducing adverse selection
•
Information sharing allows banks to improve statistical risk management,
and make more accurate predictions about repayment probabilities of credit
applicants, partly overcoming adverse selection problems.
Better knowledge of borrowers’ characteristics leads to safer lending and
lower defaults.
•
The effect on lending is ambiguous. The increase in lending to safe
borrowers may not compensate for the reduction in lending to risky borrowers.
• Kalberg and Udell (2003) capture the economies of scale associated
with information exchange from multiple sources on each borrower.
The use of more than one source increases the precision of the signal
about the borrower.
9
2. Reducing borrowers’ holdup
•
In segmented credit markets banks have private information on borrowers.
This informational advantage confers to banks market power over their
customers and generates a hold-up problem: borrowers anticipate that banks
will charge predatory rates in the future and exert low effort to perform.
If they commit to exchange information about borrowers’ types, banks restrain
their own future ability to extract informational rents. A larger portion of the
total surplus is left to entrepreneurs.
•
Borrowers have a greater incentive to invest effort in their project to ensure
their success. This reduces defaults and increases lending.
•
Gehrig and Stenbacka (2003) consider a situation in which banks are already
competing. In the presence of switching costs, future informational rents are a
stimulus to competition. Information sharing avoids these rents, thus reducing
competition. In this sense, information sharing is an anti-competitive
device.
10
3. Disciplinary effect of default disclosure
Sharing black information disciplines borrowers. Borrowers try to avoid
the implied loss of reputation and the penalty of higher interest rates. They
exert more effort to avoid default, interest rates fall and lending increases.
However:
1. Exchanging other data may reduce this disciplinary effect. A high-quality
borrower will not be concerned about his default being reported to outside
banks if these are also told that he is a high-quality client.
2. Over time, creditors learn more about their borrowers, reducing the value of
each additional piece of negative information, so incentives to comply are
also reduced (Vercammen, 1995).
11
4. Eliminating incentives to over-borrow from multiple lenders
• The literature on information sharing is mostly based on single-lender,
bilateral information exchange. Multiple bank relationships are
commonplace in most countries. In the absence of information sharing,
multiple lending creates an incentive to over-borrow.
• Anticipating this moral hazard, lenders might ration the amount of
credit supplied, require a higher interest rate, or deny all credit unless
assisted by collateral.
• Exchanging positive information on borrowers’ debt level eliminates
this problem, and improves the terms offered by lenders. The supply
of lending and/or the interest rates offered to credit seekers improves.
12
Summary of theoretical predictions
•
The incentive to share information depends on borrowers’ mobility,
heterogeneity of the population, degree of competition of the banking sector.
•
Exchanging different types of information address different information
problems in the credit market:
•
Information about borrower characteristics relieves adverse selection
problems.
•
Sharing default information tends to correct moral hazard problems.
•
Information about borrowers’ debt exposure removes the particular form of
moral hazard deriving from borrowers’ ability to borrow from multiple
lenders.
•
All models predict that information sharing reduces default rates, whereas the
prediction concerning its effect on lending is less clear-cut. Some models
predict that a more competitive banking structure leads to less information
sharing, others that it leads to more sharing.
13
Empirical questions
• What are lenders’ incentives to promote the exchange of information? Why
do credit bureaus operate in some countries but not in others?
• Can we expect credit bureaus to increase lending activity? Will better risk
management reduce the default rate? Are credit bureaus ultimately
beneficial to economic activity and welfare?
• If credit bureaus do not emerge spontaneously in the market, should
governments promote the exchange of information between lenders? And
even granted that government should have a role, what is the best way to
intervene and to enforce information sharing?
• Historical evidence
• Macroeconomic evidence
• Microeconomic evidence
• Causality issue: one the one hand, information sharing can increase lending.
On the other, higher lending increases the incentive to exchange
14
information.
The appearance of credit bureaus
•
•
•
•
•
•
•
The incentive to share credit data should correlate with:
Mobility of borrowers between various lenders.
Heterogeneity of borrowers.
The size of the loan market.
Technical innovation.
Regulation (Privacy Laws)
The group of countries where credit bureaus are more active also
exhibits high mobility of borrowers and deep consumer credit
markets.
15
Borrowers’ mobility and information
sharing
250
us
C redit reports per cap ita
200
150
ja p a n
b e lg iu m
uk
100
fin la n d
n e th e rla
g e rm a n y
50
a u stra li
sw e d e n
n o rw a y
ita ly
fr a n ce
0
0
5
10
Residential mobility
15
20
Figure 3
Figure 3. Credit Reports and Residential M obility
16
The effect of information sharing: descriptive statistics
Variable
Total Sample
No Information
Sharing
Black Information
Only
Black and White
Information
Bank Lending / GDP (%)
60.53
31.10
67.57
66.42
Credit Risk
7.77
15.20
5.11
7.14
Log GDP
7.19
5.96
6.77
7.79
GDP Growth Rate (%)
3.45
4.53
2.87
3.38
Rule of Law
7.24
4.80
8.14
7.59
Creditor Rights
2.15
3.14
2.20
1.83
French Origin
0.40
0.43
0.20
0.48
German Origin
0.12
0.00
0.00
0.22
Scandinavian Origin
0.10
0.00
0.30
0.04
English Origin
0.37
0.57
0.50
0.26
40
7
10
23
Number of observations
17
Effect of information sharing on Bank Lending / GDP
Variable
GDP Growth Rate
Log GDP
Rule of Law
Creditor Rights
Ordinary Least Squares
(1)
(2)
(3)
Trimmed
Regression
(4)
2.61
(0.85)
5.30
(1.73)
7.47
(3.14)
6.58
(2.12)
2.93
(0.89)
4.96
(1.51)
6.25
(2.46)
8.32
(2.76)
-54.86
(-2.71)
24.77
(1.52)
23.18
(2.38)
-67.93
(-3.03)
2.17
(0.62)
2.23
(0.61)
7.72
(3.64)
5.27
(1.07)
-7.01
(-0.65)
26.67
(1.24)
-44.46
(-3.18)
29.38
(1.82)
15.65
(1.43)
-42.65
(-1.22)
0.30
(0.15)
5.08
(2.17)
4.90
(3.94)
8.57
(2.55)
0.80
(0.12)
19.83
(1.82)
-29.98
(-2.52)
36.42
(4.90)
26.95
(5.00)
-60.02
(-3.32)
-1.19
(-0.68)
5.34
(2.00)
4.87
(2.89)
9.96
(3.23)
2.46
(0.31)
14.66
(1.42)
-29.22
(-2.59)
36.46
(3.50)
27.23
(2.92)
-60.64
(-2.96)
0.46
0.50
0.67
0.81
-.-
40
40
40
36
40
French Origin
German Origin
Scandinavian Origin
Black Information Only
Black and White Information
Constant
Adjusted R square
Number of observations
Robust
Regression
(5)
18
Effect of information sharing on Credit Risk
Variable
GDP Growth Rate (%)
Log GDP
Rule of Law
Creditor Rights
French Origin
Ordinary Least Squares
(1)
(2)
Trimmed
Regression
(3)
-0.63
(-2.05)
-0.57
(-1.21)
-1.65
(-4.31)
-0.45
(-1.07)
-4.26
(-1.91)
-2.99
(-1.76)
30.59
(9.67)
-0.56
(-1.97)
-0.34
(-0.74)
-1.67
(-4.74)
-0.09
(-0.17)
0.90
(0.73)
-2.76
(-2.32)
2.19
(1.42)
-4.54
(-2.15)
-2.40
(-1.37)
27.51
(8.90)
-0.63
(-2.59)
-0.04
(-0.14)
-1.88
(-8.34)
-0.17
(-0.47)
1.33
(1.41)
-1.96
(-1.81)
2.62
(1.78)
-3.08
(-1.75)
-2.15
(-1.65)
26.27
(9.74)
-0.61
(-2.06)
-0.21
(-0.43)
-1.71
(-5.45)
-0.09
(-0.17)
1.04
(0.70)
-2.46
(-1.41)
2.23
(1.18)
-3.78
(-1.89)
-2.22
(-1.23)
26.49
(7.09)
0.78
0.84
0.90
-.-
35
35
30
35
German Origin
Scandinavian Origin
Black Information Only
Black and White
Information
Constant
R square
Number of observations
Robust Regression
(4)
19
Microeconomic evidence
• Some papers analyze the predictive power of credit scoring tools,
generally finding that credit report are an important tool for assessing
consumer credit risk (Chandler and Parker, 1992; Barron and Staten,
2003).
• Kalberg and Udell (2003) show that trade credit history in D&B
reports improves default predictions relative to financial statements
alone.
• Galindo and Miller (2001) find a positive relation between access to
finance (debt) and an index of information sharing in the Worldscope
database: well performing credit registries reduce the sensitivity of
investment to cash flows.countries with better information sharing
•
20
When is public intervention useful?
• Information sharing arrangements do not guarantee an efficiency gain
under all circumstances. Their ability to increase efficiency depends
on the way they are designed.
• In some situations, sharing information about defaults may elicit a
level of effort by borrowers above the socially efficient level. In
others, it falls short of it.
• The social efficiency of information sharing also depends on the type
of information shared and, more generally, on the design of the sharing
mechanism.
• Compulsory information sharing may discourage banks from
searching new information, reduce screening and monitoring, and kill
relationship lending.
21
Reasons for creating a Public Credit Register
1.
•
•
Antitrust policy
If information sharing depends on the banks’ initiative, they
may not implement it even if it is socially efficient  for fear
of increased competition.
Policy intervention in these cases could be justified as any
antitrust policy.
2. Enhance banking stability and supervision.
•
Bank managers may take too much risk. A public register
helps supervision and prudential intervention.
3. Help conglomerate banks
•
Information sharing can help conglomerate banks to figure out
their own consolidated liability position.
22
Determinants of the Presence of Public Credit Registers
Descriptive Statistics
Variable
Total Sample
PCR Present
PCR Absent
Creditor Rights
2.14
1.59
2.50
Rule of Law
7.08
6.67
7.34
Pre-existence of a Private Credit Bureau
0.51
0.29
0.65
English Origin
0.38
0.12
0.54
French Origin
0.39
0.71
0.19
German Origin
0.14
0.11
0.15
Scandinavian Origin
0.09
0.06
0.12
43
17
26
Number of observations
23
Explaining the presence of PCR
Variable
Creditor Rights
Rule of Law
Pre-existence of a Private Credit
Bureau
French Origin
Probit Regressions
(1)
(2)
(3)
(4)
-0.16
(-2.37)
-0.01
(-0.11)
-0.39
(-2.24)
-0.07
(-0.81)
-0.01
(-0.09)
-0.41
(-2.04)
0.49
(3.35)
0.566
(1.77)
0.476
(1.16)
4,312.77
(2.16)
-118.17
(-0.12)
12,437.21
(2.30)
2,199.30
(0.95)
366.06
(0.36)
13,143.05
(2.13)
-11,988.05
(-1.65)
-15,801.16
(-1.72)
-10,216.27
(-0.96)
43
43
41
41
German Origin
Scandinavian Origin
Number of observations
Tobit Regressions
24
Summary of empirical evidence
• Credit is positively correlated with the presence of information
sharing arrangements (also controlling for country size, growth,
and variables capturing respect for the law and protection of
creditor rights)
• Public and private information sharing mitigates defaults.
• PCR are introduced to compensate, at least partly, for the weak
protection that the state offered to creditors’ interests.
• The impact of private information sharing arrangements is
similar to that of public credit registers.
25
C. Issues in the architecture of credit
information systems
•
•
•
•
•
•
•
Relationship between private and public systems
Dosage between black and white information
Memory of the system
Identifying company groups
Cross-border lending
Privacy protection
Informal markets and poor protection of creditor rights
26
I. Private and public information sharing:
substitutes or complements?
• The case for the introduction of a PCR is stronger in countries where
private information sharing arrangements do not exist or are primitive
(transition economies).
• Public arrangements can put out of business existing credit bureaus or
discourage the creation of new ones.
• The higher the minimum reporting threshold of a PCR, the larger
the scope for private initiative.
• Credit bureaus complement PCR. They can provide greater detail than
PCR, merge various types of data and provide credit-scoring services.
27
II. Dosage between black and white information
• Default information is most effective in correcting moral hazard problems in
the credit market.
• Loan information helps in estimating the total indebtedness of credit seekers.
This helps to correct incentive problems connected with multiple lenders.
Is more information always better?
• A system that provides much information about borrowers’ characteristics
may lead banks to identify more easily high-quality borrowers.
• But such borrowers will be less worried to be reported as defaulters, trusting
that their reputation will not be stained by such an event. As a result, they
may exert less effort to avoid default.
28
III. Memory of the system
1. How long should default records kept?
2. Should they be removed after (late) repayment?
• A system with infinite memory creates a high incentive to repay, but
ex ante deters borrowing. The risk of being eternally black- listed may
deter even borrowers with relatively solid prospects.
• A system where records are kept for a very short time exerts little
discipline.
• Neither extremes are generally desirable. Memory must be set
trading off borrowers’ discipline and the need to offer them a second
chance (an interesting case: the Belgian Central Office for Credit to
Private Individuals).
29
IV. Identifying company groups
•
A credit information system must identify debtors and their liabilities
unambiguously. While for individuals this is a relatively simple task,
this may not the case for companies, which are often parts of
complex group structures.
•
This is a very difficult practical task for two reasons.
1.
The loans to the various subsidiaries may go undetected to a PCR
because each of them does not exceed the PCR minimum
reporting threshold.
2.
Often corporate groups borrow heavily via foreign subsidiaries. But
the latter debt is typically undetected by the group’s domestic PCR
 banks will not have a clear picture of the group’s total
indebtedness.
30
V. Cross-border lending
• As companies grow integrated into world capital markets, national
credit information systems become unable to identify their total
indebtedness.
• The credit bureaus’ response: branching abroad, buying foreign
bureaus or linking up with them.
• The PCR response: coordinate national registers (difficult).
• European PCR feature deeply ingrained differences, and cannot agree
on common rules, increasing danger of their displacement by private
multinational bureaus.
New PCR have the opportunity of designing the PCR so as to ensure
compatibility with the systems of their main commercial partners.
31
VI. Privacy protection
• Information sharing finds an obvious limit in privacy protection. This
differs widely both within Europe and between the US and Europe,
and these differences appear to have affected the development of credit
information systems.
• One should not necessarily take a negative view of the effect of
privacy laws on credit information systems. Divulging certain types of
information may lead people to become “too cautious”, and reduce risk
taking. A moderate concern for privacy may indirectly serve also
economic efficiency.
•
There is also one privacy-protection rule that improves the accuracy of
the data stored by credit information systems: entitling individuals with
the right to inspect and correct mistaken information.
32
VII. Informal markets and protection of
creditor rights
• In developing countries the role of informal lending is larger than in
developed economies. Since typically both credit bureaus and PCR
base their information on data reported by formal lenders, their utility
is much reduced in these countries.
• This limitation of information sharing systems could be overcome by
opening access of PCR also to informal lenders.
• PCR and credit bureaus are more frequent in countries where creditor
rights receive a relatively poor protection and the law is less effectively
enforced. In this sense, PCR appear to act as a partial substitute for the
lack of good judicial enforcement.
33
Future research
• The type of information exchanged and design of information sharing
mechanism matter at least as much as the decision to set up such
mechanism.
• The market may not provide spontaneously the socially desirable
information sharing mechanism.
• Key design issues: public vs. private mechanisms; dosage between black
and white information sharing; “memory” of the system; enforcement of
creditor rights and judicial efficiency.
•
•
•
•
Future research:
Models with multiple lenders.
Effect of privacy laws on economic efficiency.
Microeconomic evidence on the value of information sharing.
Effects of information sharing in other areas: labor markets / insurance
markets.
34