Transcript Folie 1 - University of Cagliari

Small business banking and financing:
a global perspective
Cagliari, 25 May 2007
Default Rates in the Loan Market for SMEs:
Evidence from Slovakia
Christa Hainz
University of Munich, CESifo, and WDI
Jarko Fidrmuc
Anton Malesich
University of Munich, CESifo, and
Comenius University Bratislava
Comenius University Bratislava
Motivation
• SMEs in emerging markets face barriers in access to finance:
-
SMEs contribute significantly to growth and employment in the new EU
member states (EBRD, 2005).
-
SMEs crucially depend on external financing provided by locally
operating banks.
• The lending boom and concerns about future stability:
-
Markets are attractive for foreign banks (Claeys and Hainz, 2006).
-
Financial vulnerability increases during a lending boom (Coricelli et al.,
2006, Duenwald et al., 2005, Honohan and Klingebiel, 2000).
Loans to Households and Domestic Firms
This paper
• Research questions:
-
What are the typical default rates of loans to SMEs in Slovakia?
-
Which factors determine default?
-
What are the implications for financial vulnerability?
• Results:
-
On average, 6 per cent of the SMEs default on their loan.
-
We find large sectoral differences.
-
Indebtedness increases the probability of default only for firms
with above average indebtedness.
-
Natural persons are less likely to default. This suggests effect of
liability on incentives.
What determines defaults?
• Hypothesis 1: More highly indebted firms are more likely
to default.
-
If firms are highly indebted, when successful they have to pay a
higher proportion of their payoff to the bank.
-
Incentive to exert effort suffers.
• Hypothesis 2: Firms are more likely to default if they are
less profitable and less liquid.
-
Probability that firm becomes bankrupt depends on profitability
and liquidity (Altman, 1968).
What determines defaults?
• Hypothesis 3: The higher is the debtor’s liability, the less
likely the firm is to default.
-
If the debtor is fully liable, he internalizes the effect of his
investment decision on payoffs.
-
Debtor’s incentives are distorted if he is not (fully) liable (Bester,
1987, Holmström, 1996).
Data Description
• Unique data of loans to 667 SMEs in Slovakia
-
provided by one of the major banks (foreign investor),
-
between 2000 and 2005,
-
1496 observations.
• Data on whether firms have become defaulted
-
Default: delay of repayment > 90 days.
• Financial data from the firm’s annual balance sheets
-
reported as shares on total assets or liabilities for the previous year,
-
total sales indicate the size of the SMEs (€ 1 to 10 million).
Defaults
• 90 SMEs (6 per cent) defaulted on their loan.
International Comparison:
• Syndicated loans – five year period (Altman and Suggit, 2000)
-
4.6 per cent for companies with an original S&P rating B,
23.5 per cent for companies with an original S&P rating Caa.
• SMEs in the US
-
2.7 per cent (Agarwal and Hauswald, 2007).
• SMEs in Sweden
-
0.9 – 2.3 per cent (Jacobsen, Lindé, Roszbach, 2005).
Observations by Years
350
defaults
non-defaults
300
250
200
150
100
50
0
Jan. 2000-June 2001
Jan. 2001-June 2002
Jan. 2002-June 2003
Jan. 2003-June 2004
Jan. 2004-June 2005
Data Structure by Legal Forms
Non-Default Companies
Default Companies
cooperative
24%
state
enterpris
0.0%
joint stock
company
29%
natural
person
1%
cooperative
20%
limited
liability
company
46%
natural
person
5%
state
enterprise
0%
joint stock
company
24%
limited
liability
company
51%
Descriptive Statistics
Bank Cash & bank
Total sales
accounts
loans
SKK mill.
A: Non-default companies
0.298
0.152
100319
Mean
27.727
0.853
298431
Max
-0.237
-0.190
30115
Min.
0.832
0.125
65584
Std. Dev.
B: Default companies
0.100
0.177
114200
Mean
0.715
0.666
291358
Max
-0.120
0.006
30142
Min.
0.138
0.147
71465
Std. Dev.
C: F-Test of equal mean and variance between the sub-samples
5.082**
3.258*
3.747*
Mean
0.024
0.071
0.053
p-value
*
36.439***
1.381
1.187
Variance
0.000
0.052
0.300
p-value
Earnings
before tax.
0.033
0.488
-0.321
0.078
-0.038
0.171
-0.617
0.119
66.804***
0.000
2.343***
0.000
Determinants of Defaults
• We estimate following probit (and marginal probability) models


P qti  1 | t 1  1   2 Cti1  Zit 1   ti
-
Debt channel: Bank loans as a share of total liabilities (Ct-1),
-
Liquidity channel: Cash and bank accounts as a share of
total assets (Z1,t-1),
-
Profitability channel: Earnings before taxation as a share
of total assets (Z2,t-1),
-
Further control variables: industry, time and legal form dummies.
Basic Estimation
P1
P2
0.797*
1.195**
(0.409)
(0.484)
-1.675***
-1.804***
(0.392)
(0.416)
-4.612***
-5.214***
(0.728)
(0.815)
-1.381***
-1.073***
(0.100)
(0.237)
No
Yes
Number of observations
1496
1496
Pseudo-R2
0.134
0.183
Bank loans
Cash and bank accounts
Earnings before taxation
Constant
Industry, time, leg. form dummies
Sensitivity Analysis
• Control for possible selection bias by including industry,
time, and legal form dummies.
• Highly indebted SMEs may have higher default probabilities. We split up the sample into companies with debt
levels below/ above median level of credits (12% of total
liabilities).
• Panel probit estimations reflect the possible effects of
unobservable firm characteristics and the selection bias.
Loan Size
Bank loans
Cash and bank accounts
Earnings before taxation
Constant
Industry, time, leg. form dummies
Credit size
Number of observations
P3a
P3b
-0.587
1.733***
(2.834)
(0.659)
-2.551***
-1.555***
(0.683)
(0.554)
-7.691***
-3.561***
(1.421)
(1.127)
-1.761***
-1.720***
(0.533)
(0.399)
Yes
Yes
Small
Large
748
748
 For firms with a high level of credits, indebtedness increases
the probability of default significantly.
Panel Estimations
REP1
REP2
REP3
0.791
1.474*
2.948**
(0.619)
(0.809)
(1.325)
-2.205***
-2.571***
-2.392***
(0.566)
(0.673)
(1.003)
-5.294***
-6.621***
-4.865***
(1.042)
(1.351)
(2.063)
-1.848***
-1.475***
-2.762***
(0.217)
(0.431)
(0.906)
Industry, time, leg. form dummies
No
Yes
Yes
Credit size
All
All
Large
1496
1496
748
Bank loans
Cash and bank accounts
Earnings before taxation
Constant
Number of observations
 Effects are robust to inclusion of firm fixed effects.
Industry-Specific Effects
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
Probit
Large Credits
Panel-Probit
Panel Probit for Large Credits
-0.4
Agriculture
Construction
Retail trade
Other services
 Default probabilities differ largely across industries.
Legal-Form Effects
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
Probit
Large Credits
Panel-Probit
Panel Probit for Large Credits
-1.0
Natural persons
Limited liability company (s.r.o.)
Joint stock company (a.s.)
Natural persons are much less likely to default than other legal forms.
Conclusions
• Default rates of loans to SMEs in Slovakia were higher than in
mature markets (already before lending boom started).
• Evidence that defaults depends on
- Indebtedness (for those with indebtedness above average),
- Legal form
- Thus, incentives matter.
• Should we worry about the “lending boom”?
- Possibly yes, if leverage of SMEs increases.
- No, if new loans are made to SMEs.
- The banks perform comparably well in this market.