Lending activity and credit supply to firms
during the crisis
Evidence from the Croatian micro level data *
Tomislav Ridzak, Financial Stability Department
Croatian National Bank
*The views expressed in this article are those of the author and do not necessarily represent the views of, and should not be
attributed to the CNB
Motivation

Crisis brought the need to rebalance the Croatian
economy


Cutting the losses and transferring resources to more productive uses
(depends on capitalization, legal framework, own interest of bank
management).
Anecdotal evidence point to a lot of inertia in loan
dynamics:

Loan prolongation / restructuring instead of cutting the losses.
Introduction


This research aims to analyse the credit supply to
individual enterprises in Croatia during the crisis and to
detect possible credit misallocation
Steps:

get an overview on dynamics of the credit activity in
Croatia using credit creation and destruction;
 explore differences in lending patterns before and during
the crisis and find possible evidence of loan misplacement
(evergreening, zombie lending) using firm level data.
International evidence on loan reallocation

1st approach: reducing exposures to enterprises in
distress and turning to new and promising projects:


US in each recession from 1979 to 2008 (Contessi and Francis,
2009).
2nd approach: extending time limits for loan
repayments (evergreening, zombie lending):

Mostly associated with the case of Japan - Peek and Rosengreen
(2003) find that the banks increased loans to severely impaired firms in
Japan because of corporate affiliations and government pressure.
 Recent financial crisis exposed all the vulnerabilities of the banking
systems in the US and Europe - Albetrazzi and Marchetti (2010) find
evidence of evergreening in Italy.
Credit creation and destruction
Credit creation and destruction

The creation was calculated as the sum of all increases in total loan
amounts to individual clients (relative to the balance at the end of the
previous period)
 loanb,i ,t if loanb,i,t  0
Creationb ,t 
i
 loan
b ,i ,t 1
i

The destruction was calculated as the sum of all decreases in total
loan amounts to individual clients (relative to the balance at the end of
the previous period)
 loan if loan

 loan
b ,i ,t
Destructio nb ,t
b ,i , t
0
i
b ,i ,t 1
i

The excess credit growth which measures the reallocation in excess
of net credit change is defined as follows
Excess credit growthb ,t  Creationb ,t  Destructio nb ,t  | Creationb ,t  Destructio nb ,t |
Credit creation and destruction (median of
all banks)
23%
21%
19%
17%
15%
13%
11%
9%
7%
Median of destruction (all sectors)
Median of creation (all sectors)
Dec-09
Sep-09
Jun-09
Mar-09
Dec-08
Sep-08
Jun-08
Mar-08
Dec-07
Sep-07
Jun-07
Mar-07
5%
Credit reallocation: Excess credit growth
30.0%
25.0%
20.0%
15.0%
10.0%
5.0%
Reallocation (all sectors)
Reallocation (real estate and business services)
Reallocation (construction)
Dec-09
Sep-09
Jun-09
Mar-09
Dec-08
Sep-08
Jun-08
Mar-08
Dec-07
Sep-07
Jun-07
Mar-07
0.0%
Summary: crisis and pre-crisis
Average from
Average from
30/6/2007 to 30/9/2008 30/9/2008 to 31/12/2009
(2)
(1)
(2) - (1)
Median of destruction (all sectors)
10.18%
8.69%
-1.50%
Median of creation (all sectors)
12.80%
9.64%
-3.17%
Median of destruction (construction)
9.90%
9.29%
-0.61%
Median of creation (construction)
12.16%
9.16%
-3.00%
Median of destruction (real estate
and business services)
8.68%
6.73%
-1.95%
Median of creation (real estate and
business services)
11.78%
8.66%
-3.11%
Reallocation (all sectors)
18.46%
16.29%
-2.17%
Reallocation (construction)
19.78%
15.83%
-3.95%
Reallocation (real estate and business
services)
16.10%
11.77%
-4.34%
Lending patterns
Data

Bank × company panel dataset in two periods:




CRISIS: begins 30.9.2008, ends 31.12.2009
PRE-CRISIS: begins 30.6.2007, ends 30.9.2008
Dependent variable, corporate lending from bank b to
company i, is introduced in a regression as the change
in the loans deflated by firm’s average assets
Explanatory variables are:



banks’ performance and financial strength
firms’ financial indicators and other characteristics
interaction terms
Explanatory variables

Bank variables:




Firm variables:




Liquidity, profitability, capitalization and share of bad loans to
firms in total loans
Bank size dummy
Biggest creditor dummy
Z-Score dummy for companies riskier than median
Dummy for low productive companies (TFP below the median for
the sector)
Small firm dummy
Interactions between bank and firm variables
Estimation strategy, Lending patterns and
evergreening

I use multiple lending and fixed effects to single out the
effect of bank variables, i.e. credit supply on change in
loans
yb ,i  x b,i β  u i  eb ,i


Fixed effects will pick-up the unobservable part of the
error term (ui) that does not vary among banks
As a result the obtained coefficients on the bank specific
variables can be interpreted as drivers of supply
Estimation results: credit supply

Two bank specific variables are significant determinant of
credit supply to firms in the crisis period:



Can this be interpreted as evidence of capitalization related
credit crunch?


Capitalization
Non-performing loan ratio
Yes, as only about 8 per cent of the decrease in loans from low
capitalized banks is substituted by loans from high capitalized banks (5
per cent in the pre-crisis period)
Results are robust across specifications
Estimation results: evergreening
Eq Name:
Dep. Var:
SPEC_1
SPEC_2
SPEC_3
SPEC_4
SPEC_5
SPEC_6
Change in loans from bank to firm over average assets in crisis period
Bank specific control variables
-2,5766
[0.8894]***
Big bank dummy × Z_SCORE_R
(Capital adequacy ratio <= median) × Z_SCORE_R
-0,6169
[0.9448]
(Capital adequacy ratio <= median) × Z_SCORE_R ×
Bigest creditor dummy
0,7319
[0.7218]
(Capital adequcy ratio <= median) × Z_SCORE_R ×
TFP_LOW
-0,5622
[0.8402]
(Capital adequacy ratio <= median) × Z_SCORE_R ×
TFP_LOW × Biggest creditor dummy
1,9187
1,9059
[0.9164]* [0.9242]**
0,1204
[0.8211]
-1,8853
[0.8448]**
Big bank dummy × Small company dummy
Observations:
R-squared:
7243
0,5896
7243
0,5904
7243
0,5896
6787
0,5718
6787
0,5720
7243
0,5900
Conclusion

Bank capitalization was significant factor of loan supply
in the crisis period

Less capitalized banks extended the loans to
problematic enterprises (evergreening)

Small and medium sized banks are forced to deal with
the clients that are financially less stable, because of the
competition from the big banks

Bank-firm relationship in Croatia seems to be strong