Macroeconomic stress test- Comparative review of the Serbia and

Macroeconomic stress testComparative review of the Serbia and Czech Republic
Viktorija Misic1
Abstract:
After the collapse of investment bank Lehman Brothers back in 2008 and the US subrime crisis,
the main question became the quantification of financial sector vulnerabilities. Since the collapse
of commercial banks can lead to financial instability, one of adequate measures would be to
examine the effect of shocks to various risk factors on the financial condition of banks. Credit
risk, being a crucial risk in banks, is the risk that a borrower will default on his financial
obligations. It can lead to big losses in banking books.
Within the framework of macroeconomic stress test we investigate two countries Czech Republic
and Republic of Serbia. The Czech Republic became the member of European Union in 2004
and published first Financial Stability report in 2005. The Czech financial system proved
resilient to the effects of the global financial crisis. During the last three years, banks further
strengthened capitalization levels, with total capitalization increasing to 15.9 percent by June
2011. Hence, the Czech banking sector was one of the few in Central and Eastern Europe (CEE)
which, so far, did not require public support (International Monetary Fund, 2012). Although
Serbia became a candidate for European Union membership in 2012, starting from 1. July 2010,
the National Bank of Serbia made affords in strengthening stability of the financial system. As a
part of the Financial Sector Assessment program (FSAP) stress test were conducted to assess the
resilience of the Serbian banking sector to a set of extreme but plausible shocks.
This paper describes the methodology of credit stress test, implementation and practically apply
macroeconomic stress test as well as the results of Serbian and Czech banking sector. Our focus
is on calculation of default rate for corporate and household sector under the scenario 1, scenario
2 and scenario 3 in case of Czech Republic and Republic of Serbia.
Key words: bank, financial stability, stress tests, credit risk, macro stress test, non-performing
loans.
JEL classification: C22, E24, E31, E27, G21, G28.
Student doktorského studia, Katedra hospodarske a socialni politiky, Národohospodářská fakulta VŠE v Praze,
Nám. W. Churchilla 4, 130 67 Praha 3; e-mail: [email protected].
1
1. Introduction
The stress tests are carried out by EBA (European Banking Authority) and the national
supervisory authorities in EU member states, in close cooperation with the European Systemic
Risk Board, the European Central Bank and the European commission. It has become a standard
element of Financial Sector Assessment Programs (FSAPs), implemented jointly by the IMF and
the World Bank (IMF (2003)). The ESRB was established at 2011 with aim to identify systemic
risks and issue recommendations to mitigate those risks.
Stress testing is also important part of the New Basel Accord (BIS, 2004). Basel II (Pillar 1minimum capital requirements and Pillar 2-supervisory review process) requires banks to
perform stress tests. The bank which applies internal rating based approach has to target those
requirements. Basel III proposals are designed to lead to greater strength of commercial banks.
The capital requirements are getting raised and new regulatory requirements regarding bank
liquidity and leverage are introduced, as well as additional capital buffers. (BCBS 2010) 2.
Many empirical studies have employed these macro credit models. Boss (2002) uses the
macroeconomic credit model to analyze the stress situation for bank default probability in
Austria and finds that industrial production, inflation rate, stock index, nominal short-term
interest rates, and oil prices are the determinant factors of default probability.
Baboucek & Jancar (2005) employ the vector autoregression model (VAR) using the
NPLs and the macroeconomic factors for the Czech Republic. Through macro prudential
analysis they use an unrestricted VAR model to empirically investigate transmission involving a
set of macroeconomic variables who characterized Czech economy.
Drehmann, Sorensen & Stringa (2010) estimate the integrated impact of the credit and the
interest rate risks on the banks' portfolios, assessing the banks' economic value, the future
earnings and the capital adequacy.
Peura & Jokivuolle (2003) measure the capital adequacy by simulating the difference
between the bank's actual capital and the minimum capital requirements and they determine
whether the estimated bank's capital buffer is sufficient over the business cycles.
Pain (2003) found an empirical relationship between banks’ loan loss provisions and
macroeconomic indicators such as GDP growth, real interest rates, credit growth and the
concentration of the loan portfolio.
In their work Kalirai and Scheicher (2002) model the impact of key macroeconomic
variables, such as indicators of general economic activity, price stability, households’ and
corporate sectors’ situation, financial market and external events, on aggregated loan loss
provisions (LLP) using a linear regression model and a sensitivity analysis for macro stress
2
Basel III will require banks to hold 4.5% of common equity (up from 2% in Basel II) and 6% of Tier I capital (up
from 4% in Basel II) of risk-weighted assets (RWA).
testing. Short-term interest rates, GDP growth rates, the stock index and industrial production are
found to influence LLP.
Pesola (2001, 2007) confirmed that macroeconomic shocks jointly with financial fragility
generate banks loan losses. He employs an econometric model based on panel data to assess the
relationship between the ratio of banks’ loan losses and enterprise bankruptcies per capita and
macroeconomic variables. His findings suggest that high corporate and household indebtedness,
combined with negative macroeconomic shocks contributed to the banking crisis in Sweden,
Norway and Finland.
All of these studies confirmed that macroeconomic variables could affect the bank’s
portfolio and increase credit risk measured by LLP3 and NPL.
2. Credit risk and macroeconomic stress test - Serbia and Czech Republic
The main credit risk parameters are PD- probability of default, LGD- loss given default,
EAD- exposure at default. PD is used to predict the volume of the gross inflow of nonperforming loans NPL’s. Changes in the risk factors can lead to upgrades as well as downgrades
of risk parameters (The PD is by far the most popular risk parameter which is followed in stress
tests). For example, an increase in price of resources such as oil or energy can have a negative
impact on PDs in the automobile or any other industry consuming lots of energy, but it could
have a positive impact on the PDs in the country trading these resources.4
Depending on the availability of data, credit risk factors and their correlations with
macrovariables can be estimated using data on loan performance (historical NPLs, default rates,
recovery rates, loan-loss provisions (LLPs) or cost of credit) or using microdata on corporate
sector from credit registries and eventually household sector data (Čihak, 2007).
Non-performing loan (NPL) is one of important indicators to evaluate status of portfolio in
commercial banks in Serbia and Czech Republic. As NPL rate gets higher, bank need more
provision to cover losses on these non-performing loan. Therefore, NPL ratio can reasonably
represent the default risk of commercial bank. The share of NPLs in Serbia’s total banking sector
loans has been increasing since 2008 and achieve the peak in Q1 2012 where total loans of the
banking sector past due for more than 90 days making up 20,4 percent of gross loans (National
Bank of Serbia, 2012).
According to Czech National bank, in contrast to 2009–2010, there was no further strong growth
in NPLs, and their ratio to total loans declined slightly to 6% at the end of 2011 (compared to
3
Loan loss provisioning is a non-cash expense for banks to account for future losses on loan defaults. Banks assume
that a certain percentage of loans will default or become slow-paying. Banks enter a percentage as an expense when
calculating their pre-tax incomes. This guarantees a bank's solvency and capitalization if and when the defaults
occur. They appear on the income statement as an operating expense.
4
Engelmann B., Rauhmeier, R, (2006), The Basel II Risk Parameters: Estimation, Validation, and Stress Testing,
Dresdner Bank, Germany.
6.3% at the end of 2010). An international comparison between selected EU countries shows that
the NPL ratio in the Czech Republic is similar to that in Slovakia (5.6%), higher than in Austria
(2.7%) and Belgium (2.8%), and lower than in Poland (8.2%), Slovenia (11.8%). A moderate
decline in credit risk is also indicated by the evolution of risk costs, defined as net provisioning
relative to total loans and by the evolution of loan restructuring in both the household segment
and the non-financial corporations segment. As we can see from the figure below, corporate NPL
in 31.10 reached 7.51 of total loans. In 2002 at the end of January the NPL reached peak of 17.7
and after that never was at the same level. NPL house was in 2008 in Q2 in the lowest level at
2.9 of total loans, but after that constantly growing. (Czech National Bank, 2011/2012).
Figure 1. Total NPL ratio, NPL corporate and NPL household ratios- Czech Republic
Source: Author’s calculation.
The increase in the unemployment rate in 2012 Q3 could make problem for household in
payment loan obligations. The increasing unemployment causes the default rate to grow as more
people lose jobs and their creditworthiness decreases. According to Ministry of labour and social
affairs of Czech Republic the lowest unemployment rate was in 2008 Q3 and reached 5.0% and
in Q32012 reached 8.4 percent. The peak was in Q4 2010 was 9.6%.
The Czech financial sector is dominated by a few large banks. Banks account for 84 percent of
the financial sector assets. The banking system’s assets grew rapidly from 2000 to 2008,
especially household loans, but balance sheet growth had almost stopped in 2009 as a result of
the crisis. The 5 largest banks control more than 70 percent of total bank assets, and the 3 largest
ones about 60 percent. (International Monetary Fund, 2012)
The Serbian banking sector at end-Q3 2012 comprised of 33 banks- 21 in foreign and 12 in
domestic ownership. Among domestically owned banks, 9 banks were state-owned (either by
holding a majority share or being the largest individual shareholder) and 3 were in the ownership
of private individuals. Foreign-owned banks dominated the market – they accounted for 74% of
total assets, 74% of total capital and 71% of employment of the banking sector, and posted profit
of RSD 17.5 bln. (National Bank of Serbia, 2012)
Serbia’s banking sector is well capitalized. According to National Bank of Serbia Capital
adequacy ratio in Q3 2012 is 16.4 percent (in Q3 2011 was 19.7%). The share of non-performing
loans in total loans is rising, mostly as a result of foreign exchange-induced credit risk. In Q1
2012 the capital adequacy ratio decreased as a result of the combined effect of a fall in regulatory
capital and a rise in risk-weighted assets triggered, among other things, by dinar depreciation.
Banking sector assets reached 83,5 of GDP in 2011. Bank’s loan portfolio is still dominant,
accounting for close to 60% of total banking sector assets in 2011. (National Bank of Serbia,
2011).
Table 1.Selected parameters of the Serbian banking sector
Number
of banks
Total domestic 12
banks
Total foreign 21
banks
Profit
bln.
21.7
22.9
in Assets
bln
685
1.965
in Assets in % Capital
bln
26
135
74
411
in Capital
%
25
in
75
Source: www.nbs.rs
3. Macroeconomic credit risk model
Sorge & Virolainen (2006) highlight two approaches that explicitly link the default
probabilities and macroeconomic variables- Wilson (1997a, 1997b) and Merton (1974) models.
These two authors adopt the Wilson framework to perform a macro stress test on credit default
probability in Finland and find that default probability distribution by Monte Carlo simulation is
significantly different from its normal distribution in stress situations. In comparison, the Merton
model integrates asset price changes into default probability evaluation.
Merton's model (1974) was originally developed for the firm`s level but extended for the
purposes of the macro stress testing. Merton model integrates asset price changes into default
probability evaluation. Merton’s type model for the Czech economy was used in Jakubik (2007).
Jakubik & Schmieder (2008) apply the model on the household and the corporate sectors for the
Czech Republic and Germany. They test the effects of macroeconomic variables on NPL as a
measure of the default rate. They conclude that key macroeconomic determinants, such as
interest rates, exchange rates, inflation, GDP growth and the level of indebtedness, can
meaningfully simulate corporate default rates for both countries. The results show the greater
macroeconomic shocks in Czech Republic than in Germany.
Hamerle, Liebig & Scheule (2004) use factors models to forecast the default probabilities
of the individual borrowers in Germany. Merton's model was used also in Drehmann (2005) for
the stress testing the corporate exposures of the banks in the UK. He concluded that is quite
reassuring as even in the worst conditions expected losses of banks corporate exposures are not
high enough to cause a bank failure.
One of the few credit risk models that explicitly links macroeconomic factors and
corporate sector default rates was developed by Wilson (1997a, 1997b). Wilson's logistic model
was used in studies of Boss (2002) and Virolainen (2004). Boss (2002) and Boss et al. (2006)
estimate the relationship between the macroeconomic variables and the credit risk for the
corporate default rate in the Austrian banking sector. Virolainen (2004) and Virolainen,
Jokivuolle & Vähämaa (2008) develop the macroeconomic credit risk model that estimates the
probability of default in Finnish industries. The idea is to model the relationship between default
rates and macroeconomic factors and to simulate the evolution of default rates over time by
generating macroeconomic shocks to the system. These simulated future default rates and
estimates expected and unexpected credit portfolio losses including also the current
macroeconomic situation.
For purpose calculation default rates we use Wilson model (1997a,b) in line with
Virolainen (2004). First, the average default rate for industry j is modeled by the logistic
functional form as
ds,t= 1/1+exp(ys,t)
where ds,t is the default rate in industry s at time t, and ys,t is the industry-specific macroeconomic
index, whose parameters must be estimated.
According with Boss (2002) we adopt formulation that a higher value for ys,t implies a better
state of the economy with a lower default rate ds,t and vice versa. The logistic functional form is
given by:
L(ds,t ) ln (1-ds,t/dj,t)=ys,t
The logit transformed default rate (the industry-specific macroeconomic index) is assumed to be
determined by a number of exogenous macroeconomic factors.
Ys,t=ß s,0 + ß s,1 x1,t + ß s,2 x2,t+….+ ß s,n xn,t+ ε s,t ,
where βs =(βs,1 βs,2, …., βs,n ) is a set of regression coefficients to be estimated for the sth industry,
xn,t (x1,t,x2,t…xn,t) is the set of explanatory macroeconomic factors (e.g. GDP, exchange rate,
unemployment rate, etc.), and ε s,t is a random error assumed to be independent and identically
normally distributed.
3.1.Macroeconomic credit risk model for corporate sector-Serbia and Czech Republic
Typically, the credit risk models include a measure of credit risk as dependent variable and
macroeconomic variables (i.e., output measures, interest rates, inflation, and the exchange rate)
as explanatory variables. The macroeconomic data for the Czech Republic have been taken from
the time series archives (ARAD) of the Czech National Bank. For corporate sector as
independent variables we used GDP growth rate, unemployment rate, exchange rate CZK/ EUR
and exchange rate CZK/USD.
Table 2. Summary Statistics, Czech Republic, Corporate sector, observations from 2002:1 to
2012:3
Variable
Mean
Median
Minimum Maximum Std. Dev.
gdp
0.704
0.8
-3.3
2.4
1.05
un
8.323
8.6
5.0
10.3
1.34
czk_usd
22.43
21.23
15.89
36.23
4.88
czk_eur
27.90
27.78
24.29
32.98
2.75
Source: Author’s calculation. Summary Statistics, Gdp is gross domestic product, un- unemployment
rate, exchange rate czk_usd, exchange rate czk_eur.
The macroeconomic credit risk model for corporate sector for Czech Republic is:
ln (1- pd corp,t/ pd corp,t) = α + ß1gdp,t+ ß2un,t+ ß3czk_eur,t+ ß4czk_usd,t
Table 3. Czech Republic, corporate sector, using observations 2002:1-2012:3 (T = 43)
Dependent variable: ln_nplc
Variable
Denoted
Coefficient
Std. Error
t-ratio
p-value
Constant
const
-3.01529
0.460675
-6.5454
<0.00001
***
GDP growth rate
gdp
-0.0710079
0.032693
-2.1720
0.03617
**
Unemployment
rate
Exchange rate
CZK/EUR
Exchange rate
CZK/USD
un
0.249528
0.0279239
8.9360
<0.00001
***
czk_eur
-0.130102
0.0248833
-5.2285
<0.00001
***
czk_usd
0.089606
0.0137262
6.5281
<0.00001
***
R-squared
0.841888
Adjusted R-squared
F(4, 38)
50.58383
Hannan-Quinn
-7.776307
P-value(F)
1.02e-14
Akaike criterion
-11.02369
rho
0.603986
Durbin-Watson
0.797412
0.825244
Source: Author’s calculation. Significant at 1% level. Dependent variable: ln_corp.
As shown in the table, the outcome of our analysis demonstrates a important influence of the
exchange rate czk_usd. According to t- test we might say that unemployment rate and exchange
rate CZK/USD are the substantional explanatory variable. An appreciated exchange rate raises
the prices of domestic goods in foreign currency. Appreciated exchange rates results in a higher
default rates in the corporate sector for both countries. Positive impact of the CZK/EUR
exchange rate on the default rate might be the results of the preference for loans denominated in
the euro. GDP have negative signs means that increasing GDP affects positively demand for
company’s goods. Such increases may lead in better creditworthiness of the firms.
The negative impact of depreciation of the domestic currency on the default rate is given by the
fact that the currency depreciation favours domestic exporters and increases their profits. The
increasing GDP stimulates the demand for goods that corporations produce and that increases
their profits and ability to repay the debt. The probability of default decreases.
Figure 2. Independent and Dependent variables for corporate sector- Czech Republic
Source: Author’s calculaton. Un- unemployment rate, nplc- is nonperforming loan for corporate sector,
exch is exchange rate CZK/EUR and exchczk_usd is exchange rate CZK/USD.
The figures shows that nonperforming loan achieve a peak in 2010Q1. Increased unemployment,
negative industrial production and local currency depreciation caused problem in repaying debt.
Figure 3. Actual and estimates value for Czech corporate sector ln_npl
Source: Author’s calculation.
Figure 3.shows performance of the estimated model. In 2008 default rate was at the lowest level
at about 3% default rate. Furthermore, leading the collapse in worldwide market and distress
situation during the crisis the default rate constantly increased. This situation reflects on
corporate sector ability to pay debt. After 2008 the default rate constantly decrease and peak
9.1% in 2010. The explanatory variables of corporate sector in case of Serbia we use GDP,
producer price index, GDP Euro area and industrial production.
Table 4. Summary Statistics, Serbia, Corporate sector, observations from 2002:1 to 2012:3
Variable
Mean
Median
Minimum
Maximum
Std. Dev.
gdp
2.97
3.1
-4.1
13.7
3.64
ppi
8.63
9.6
-0.2
17.1
5.41
gdp_euro
0.6
1.1
-4.7
2.7
1.57
ind
2.99
3.8
-9.5
13.7
4.68
Source: author’s calculation. Summary Statistics, using the observations for the years from 2002:1 to
2012:3 for the variables: gdp represent Serbia’s gdp growth rates ind- industrial production growth rate,
rsd_eur- RSD/EUR exchange rate growth rate, gdp_eu- is GDP for Euro area (17 countries), ppi- Producer
price index.
The euro area has fallen into a mild recession in Q3 2012 (-0.1%), which adversely affected the
growth outlook of countries in the region which are also important export markets for Serbia.
The exchange rate CZK/USD for Czech Republic and Serbia’s producer price index experiences
the highest volatility with the standard deviation of 4,9 and 5,4. The mean of the rate of the
exchange rate suggests the decreasing path over the period on average with relatively high
standard deviation of 4.8%.
The macroeconomic credit risk model for corporate sector for Serbia is:
ln (1- pd corp,t/ pd corp,t)= α + ß1gdp,t-3 + ß3ind,t-5+ ß4 gdp_eu,t+ ß5ppi-4
Table 5. Serbia corporate sector, observations from 2003:2-2012:3 (T = 38)
Dependent variable: ln_corp
Variable
Constant
Denoted
const
Coefficient
-1.90265
Std. Error
0.0795383
t-ratio
-23.9212
p-value
<0.00001
***
Gross domestic
product
Producer price
index
GDP euro area- 17
countries
Industrial
production
gdp_3
-0.040939
0.0136293
-3.0037
0.00506
***
ppi_4
0.0442509
0.00687925
6.4325
<0.00001
***
gdp_euro
-0.0718627
0.0266462
-2.6969
0.01093
**
ind _5
-0.0402331
0.00962748
-4.1790
0.00020
***
R-squared
0.772705
Adjusted R-squared
0.745154
F(4, 33)
28.04644
Hannan-Quinn
2.308343
P-value(F)
3.33e-10
Akaike criterion
-0.604860
rho
0.388063
Durbin-Watson
1.198908
Source: Author’s calculation. All variables are calculated as growth rates. Significant at 1% level.
Dependent variable: ln_corp.
The negative impact of depreciation of the domestic currency on the default rate is given by the
fact that the currency depreciation favours domestic exporters and increases their profits, which
in turn helps to decrease their default rates. Apart from the PPI growth rate all coefficients of the
explanatory variables have positive signs.
The results show the negative impact of GDP growth on the default rate in the small exportoriented country. The increasing GDP stimulates the demand for goods that corporations produce
and that increases their profits and ability to repay the debt. In that situation the probability of
default decreases.
Figure 4. Frequency distribution, corporate sector
Source: Author’s calculation. Gaussian distribution. Based on data from www.nbs.rs.
3.2.Macroeconomic credit risk model for household sector for Serbia and Czech Republic
The macroeconomic credit risk model for the household sector for Serbia is:
ln (1- pd house,t/ pd house,t) = α + ß1gdp,t-3 + ß2rsd_eur,t-1+ ß3cpi,t-5
where pd house,t is the default rate defined as the portion of households non-performing loans to
total households loan in time t.
Table 6. Summary Statistics, Serbia, Household sector, using the observations 2002:1 - 2012:3
Variable
Mean
Median
gdp
2.97
3.1
-4.1
13.7
3.64
cpi
11.17
10.3
3.2
29.1
5.02
rsd_eur
-1.58
-1.13
-15.6
4.6
3.51
Source: Author’s calculation.
Minimum Maximum Std. Dev.
Depreciations of dinar against the euro favour exporting companies but negatively affect the
households whose loans are denominated in euro. Depreciations of dinar causes that the exported
goods are much more competitive abroad. That can increase the profit of export oriented
companies. Regarding the households who took the loans denominated in euro, the depreciation
of local currency can increase their default rate.
The household sector data for Czech Republic we used: unemployment rate, household
consumption and interest rate on new business. The macroeconomic credit risk model for
household sector for Czech Republic is:
ln (1- pd house,t/ pd house,t) = α + ß1un,t-2 + ß2inter_house, t-6,+ ß3consump,t-1
Table 7. Summary Statistics, Czech Republic, Household sector, data from 2002:1-2012:3
Variable
Mean
Median
Minimum Maximum
Std. Dev.
un
8.32
8.6
5.0
10.3
1.34
inter_house
11.86
11.54
8.52
14.82
2.01
consump
0.46
0.5
-1.8
2.3
0.9
Source: Author’s calculation. Summary statistics, un – unemployment rate for Czech Republic,
inter_house5 is loans to household- interest rate on new business (% p.a), consump – GDP by type of
expenditure-quarterly percentage changes, seasonally adjusted, final consumption expenditure of
household.
.
Table 8. Czech household sector, observations from 2003:3-2012:3 (T = 37)
Dependent variable: ln_nplh
Variable
Denoted
Coefficient
Std. Error
t-ratio
p-value
Constant
const
-4.50524
0.167484
-26.8995
<0.00001
***
Unemployment
rate
un_2
0.117082
0.0126175
9.2793
<0.00001
***
inter_house_6 0.0360472
0.0104673
3.4438
0.00158
***
0.0229645
-2.2401
0.03194
**
Interest rate for
new business
Final consumption
expenditurehousehold
5
consump_1
-0.0514416
According to explanation of methodology interest rate on new business (ARAD)- the average rate is the rate
applied by banks on CZK-denominated loans to clients. New business includes all new agreements between the
banks and their clients in the course of the reference period.
R-squared
0.781989
Adjusted R-squared
0.762170
F(3, 33)
39.45616
Hannan-Quinn
-54.55342
P-value(F)
5.08e-11
Akaike criterion
-56.82512
rho
0.555256
Durbin-Watson
0.868940
Source: Author’s calculation. All variables are calculated as growth rates. Significant at 1% level.
Dependent variable: ln_house.
The positive sign in unemployment as explanatory variable means that with rising
unemployment rate the people will be unable to meet their obligations. Unemployment was
found to be relevant with a lag of two quarters. In the case of the interest rate on new business
the statistically best results were achieved for a long lag of six quarters. The adjusted coefficient
determination (indicates a good determination of the dependent variables by independent
variables), R2 is relatively high in our calculation, according with that higher coefficient is an
indicator of a better goodness of fit for the observations. The Durbin-Watson statistics indicate
low autocorrelation in the data because its values are below two.
Figure 5. Frequency distribution, Czech household sector.
Source: Author’s calculation.
Table 9. Serbia household sector, using observations 2003:2-2012:3 (T = 38)
Dependent variable: ln_house
Variable
Denoted
Coefficient
Std. Error
t-ratio
p-value
Constant
const
-2.28419
0.101453
-22.5147
<0.00001
***
Gross domestic
product growth
rate
gdp_3
-0.0407908
0.0109713
-3.7180
0.00072
***
Consumer price
index
cpi_5
-0.0260206
0.00793974
-3.2773
0.00242
***
Exchange rate
RSD/EUR
rsd_eur_1
-0.0212304
0.0105022
-2.0215
0.05116
*
R-squared
0.508784
Adjusted R-squared
0.465441
F(3, 34)
11.73866
Hannan-Quinn
4.548319
P-value(F)
0.000020
Akaike criterion
2.217757
rho
0.652836
Durbin-Watson
0.722714
Source: Author’s calculation. All variables are calculated as growth rates. Significant at 1% level.
Dependent variable: ln_house.
The negative effect of the inflation on the default rate is demonstrated in the deterioration of the
real value of a credit obligation. The outcome of the regression analysis shows that the corporate
default rate for Serbia’s household depends negatively on the GDP.
The recession and
deterioration in euro area makes Serbian economy vulnerable. In figure 6 it can be seen that
default rates and ability of households to repay debt are constantly decreasing after 2007. The
peaks were at the end of 2009 and 2010. In this period the default rate reached almost 8.3
percent.
Figure 6. Actual and estimates value of the ln_house for Serbia
Source: Author’s calculation.
Figure 7. Density of normal distribution
Source: Author’s calculation.
4. Macro stress test outcomes for corporate and household sectors
We choose three scenarios for Czech Republic and Serbia: Scenario 1 represents the actual data
(taking account a lags), Scenario 2 are the worst historical data6 and predictions and Scenario 37
(+ 2 pp deviation from scenario 2).
Table 10. Macroeconomic scenarios-Czech Republic
Variable
(% q-o-q)
Denoted by
GDP real growth
gdp
-0.3
1.9
3.9
Unemployment
rate_2
un
8.4
8.8
10.8
Exchange rate
CZK/EUR growth
rate (q-o-q)
Exchange rate
CZK/USD growth
rate (q-o-q)
Consumption
expenditure of
households_1
Interest rate on new
business_6
Czk_eur
3.55
-6.58
-8.58
Czk_usd
-1.7
-19.9
-21.9
consump
-1.1
1.2
3.2
14.33
15.25
17.25
Inter_house
Scenario 1%
Scenario 2%
Scenario 3%
Source: Author’s calculation. Corporate sector don’t have a lag so we for Scenario 1 we used data from Q32012.
Int_h je interest rate for household – interest rate on new business house. For this variable -scenario 2 we use
historical data. The worst case was in 28.2. 2011. Interest rate was 15.25. Un rate scenario 1- Ministry Finance of
Czech Republic, 2014 (registred unemployment).
Depreciation of CZK ag. USD and appreciation of CZK ag. EUR, when compared with previous
quarter same year was -1.7 and 3.55. In Q3 2011 CZK is appreciated for 0.69 percent. Exchange
rate CZK/EUR for Scenario 1 is data from 2014 forecast according to Ministry Finance of Czech
Republic. Depreciation CZK ag. USD is for Scenario 2 using the values from 2008Q4 (19.9). In
that period the biggest drop in value of local currency is recorded. Moreover, neither 3M
PRIBOR, wage growth rate, unemployment rate, nor household debt/ GDP were among the most
important explanatory variables.
6
7
Boss (2002) used the historically observed maximum movements of the macro variables for scenario.
Virolainen (2004) sets shocks by increasing or decreasing the values of the variables by certain percentage points.
Table 11. Different macroeconomic scenarios for Serbia.
Variable
(% q-o-q)
Denoted
by
Scenario 1%
Scenario 2 %
Scenario 3
Real GDP growth
rates_3
gdp
0.5
-4
-6.0
Inflation growth ratesCPI_5
cpi
12.1
14
16
Exchange RSD/EUR
growth rates_1
rsd_eur
-3.3
-15
-17
Industrial production
growth rate_5
ind
-0.6
-7.3
-9.3
GDP growth rates Euro
area
gdp_eu
-0.1
-4.7
-6.7
Inflation growth ratesPPI_4
ppi
12.3
17
19
Source: Hystorical and Hypothetical data, author’s calculation.
Table 12. Default rates Serbia
Scenario 1, in %
Scenario 2, in %
Scenario 3,in %
Household default rate
7.3
10.3
11.02
Corporate default rate
19.3
39.03
48.7
Source: Author’s calculation
Table 13. Default rates Czech Republic
Scenario 1, in %
Scenario 2, in %
Scenario 3, in %
Household default rate
5
4.8
5.8
Corporate default rate
4.9
9.5
12.1
Source: Author’s calculation.
5.Conclusion
In this study, macroeconomic stress test was applied to Czech and Serbian banking system.
These two countries are interesting for comparation of default rates for household and corporate
sector. We use time frame of ten years and quarterly data.
Further, Wilson model has been used to estimate default rates. These default rates are calculated
based on non-performing loans for corporate and household sectors. The both countries has also
own local currency which was applied in stress tests. At the moment, Serbian domestic currency
is quoted in range of 113-114 per one euro.
The results show significantly high rate of default in corporate sector. The most important
problems faced by corporate as well as household sector are high rate of unemployment,
depreciation of local currency and high inflation rate. The model show meaningful for the Czech
corporate sector, which this was not the case for household sector the same country.
We should be concerned with high default rate for Serbia’s corporate sector. Without
FDI, with official unemployment rate of 26%, high inflation of 11.9 Adversely macroeconomic
environment affected possibility of repaying debt for the both sectors which resulted in higher
bank losses and increased credit supply. Current macroeconomic situation in Serbia can only
adversely effect on default rates and Capital adequacy ratio.
Increasing public spending which results in high deficit, has for results has deteriorating
investor’s confidence. According to National Bank of Serbia, the current credit rating of country
is BB- and assigned by agencies Standard&Poor’s and Fitch.
Appendix A
Additional specification of macro stress testing model
Table A.1. Tests for OLS model for Czech Republic and Serbia-Corporate Sector
HeteroscedasticityWhite test
Normality of residualsnormally distributed
errors
AutocorelationLMF test
Serbia – Null hypothesis
heteroscedasticity not present,
p-value = 0.155828
error is normally distributed
p-value = 0.473424
Czech –Null hypothesis
heteroscedasticity not present,
p-value =0.0387867
error is normally distributed
p-value =0.148209
no autocorrelation
p-value = 0.0410029
no autocorrelation
p-value = 0.000588037
Source: Author’s calculation.
Table A.2.
Tests for OLS model for Czech Republic and Serbia-Household Sector
Serbia – Null hypothesis
Czech –Null hypothesis
Heteroscedasticity- heteroskedasticity not
heteroscedasticity not present,
White test
present,
p-value 0.458231
p-value= 0.363371
Normality of
error is normally distributed, error is normally distributed
residualsp-value = 0.465866
p-value = 0.0197159
normally
distributed errors
Autocorrelationno autocorrelation,
no autocorrelation,
LMF test
p-value = 0.000502393
p-value = 5.10684e-005
Table A.3. Correlation coefficients, Corporate sector, Czech Republic, using the
observations 2002:1 - 2012:3, 5% critical value (two-tailed) = 0.3008 for n = 43
gdp
un
ln_nplc
czk_eur
czk_usd
1.0000
0.2006
-0.0783
0.3845
0.2457
gdp
1.0000
0.7791
0.5244
0.5544
un
1.0000
0.3548
0.6066
ln_nplc
1.0000
0.8733
czk_eur
1.0000
czk_usd
Source: Author’s calculation. Gdp is gross domestic product, un- unemployment rate, ln_nplc- nonperforming
loan of corporate sector, czk_eur - exchange rate CZK/EUR, czk_usd - exchange rate CZK/USD.
The all variables show correlation between -1 and 1 and so linear dependence between (two)
variables.
Table A.4. Correlation coefficients, Corporate sector, Serbia, using the observations 2002:1 2012:3, 5% critical value (two-tailed) = 0.3008 for n = 43
gdp
ind
ln_nplcorp
ppi
gdp_euro
1.0000
0.8484
-0.6643
0.1789
0.6283
gdp
1.0000
-0.6188
0.1229
0.6106
ind
1.0000
0.2603
-0.3615
ln_nplcorp
1.0000
0.2787
ppi
1.0000
gdp_euro
Source: Author’s calculation. Gdp is gross domestic product, ind – industrial production, ln_nplcorp- nonperforming
loan for corporate sector, ppi- producer price index, gdp_euro- gross domestic product of euro area 17 countries.
Table A.5. Correlation coefficients, Household sector, Czech Republic, observation
from 2002:1 - 2012:3, 5% critical value (two-tailed) = 0.3008 for n = 43
un
inter_house
consum
ln_nplh
1.0000
-0.1988
0.0138
0.7151
un
1.0000
-0.5999
-0.2929
inter_house
1.0000
0.0091
consum
1.0000
ln_nplh
Source: Author’s calculation. Un is unemployment rate, inter_house- interest rate on new business, consum- final
consumption expenditure and nonperforming loan for household sector (ln_nplh).
Table A.6. Correlation coefficients, Serbia- household sector, using the observations 2002:1 2012:3
5% critical value (two-tailed) = 0.3008 for n = 43
gdp
cpi
Rsd_eur
npl_ratioh
1.0000
0.3727
0.0940
-0.6186
gdp
1.0000
0.1767
-0.4919
cpi
1.0000
-0.2070
Rsd_eur
1.0000
npl_ratioh
Source: Author’s calculation. Gdp - gross domestic product, growth rate, cpi- consumer price index, rsd_eur is
exchange rate RSD/EUR and npl_ratioh- non performing loan for household sector
Literature
BCBS (2004), “International Convergence of Capital Measurement and Capital Standards”,
Revised Framework, Bank for International Settlements, June 2004.
BCBS (2010), “A global regulatory framework for more resilient banks and banking systems”,
Bank for International Settlements, Dec. 2010 (rev June 2011).
Boss, M (2002), “A macroeconomic credit risk model for stress testing the Austrian credit
portfolio”, Financial Stability Report 4, pp. 64-82, Oesterreichische Nationalbank.
Cihak, M. (2007), “Introduction to Applied Stress Testing”, IMF Working paper WP/07/59,
Monetary and Capital Markets Department, International Monetary Fund.
Czech National bank, Financial stability, on-line:<http://www.cnb.cz/en/financial_stability/>
Czech Statistical office, Index spotrebitelskych cen-zivotnich nakladu (ctvrtletne), 09.10.2012,
on-line:http://www.czso.cz/csu/redakce.nsf/i/isc_cr
Czech National Bank, Time Series database- ARAD, on-line:<
http://www.cnb.cz/cnb/STAT.ARADY_PKG.STROM_DRILL?p_strid=0&p_lang=EN>
Czech National Bank, Financial Stability Report 2011/2012, on-line:<
http://www.cnb.cz/miranda2/export/sites/www.cnb.cz/en/financial_stability/fs_reports/fsr_20112012/fsr_2011-2012_financial_sector.pdf>
Drehmann, Stringa M., Sorensen S., (2010), “The integrated impact of credit and interest risk on
banks: A dynamic framework and stress testing application”, Journal of Banking and Finance,
vol 43 (4), pp 713-729.
Drehmann M, (2005),” A Market Based Macro Stress Test for the Corporate Credit Exposures of
UK Banks”, Bank of England, On-line: <http://www.bis.org/bcbs/events/rtf05Drehmann.pdf>
Hamerle, A., T. Liebig, & H. Scheule (2004): “Forecasting Credit Portfolio Risk." Discussion
Paper Series 2: Banking and Financial Supervision 1, Deutsche Bundesbank.
Engelmann B., Rauhmeier, R, (2006), “The Basel II Risk Parameters: Estimation, Validation,
and Stress Testing”, Dresdner Bank, Germany.
Eurostat, Real GDP growth rate-volume, on-line:
<http://epp.eurostat.ec.europa.eu/tgm/table.do?tab=table&init=1&plugin=1&language=en&pcod
e=tec00115>, updated 2012-04-02.
Eurostat, News release, Euro indicators, 174/2012, online:<http://epp.eurostat.ec.europa.eu/cache/ITY_PUBLIC/2-06122012-AP/EN/2-06122012-APEN.PDF>, 6.December 2012.
International Monetary Fund, 2003, “European Financial Integration, Stability and Supervision,”
in: “Euro Area Policies: Selected Issues,” IMF Country Report No. 05/266, August, pp. 113–46,
available at http://www.imf.org/external/np/sta/fsi/eng/guide/index.htm
International Monetary Fund (2012), Czech Republic: Tehnical Notel on Stress Testing the
Banking sector , IMF Country Report No. 12/74, July 2012. Washington, D.C.,
on-line:< http://www.imf.org/external/pubs/ft/scr/2012/cr12174.pdf>
Jakubík, P. and C Schmieder (2008): “Stress testing credit risk: comparison of the Czech
Republic and Germany”, Financial Stability Institute Award 2008 Winning Paper.
Jakubík, P., (2007), "Credit Risk in the Czech Economy," Working Papers IES 2007/11, Charles
University Prague, Faculty of Social Sciences, Institute of Economic Studies, revised March
2007.
Jancar M., Baboucek, I. (2005), “A VaR Analysis of the Effects of Macroeconomic Shocks to
the Quality of the Aggregate Loan Portfolio of the Czech Banking Sector”, Czech National Bank
Working Paper Series 1/2005.
Kalirai, H., and M. Scheicher (2002).,“Macroeconomic Stress Testing: Preliminary Evidence for
Austria”, Financial Stability Report No. 3: 58–74. Oesterreichische Nationalbank.
Merton R. (1974): On the Pricing of Corporate Debt: The Risk Structure of Interest Rates,
Journal of Finance, 29, 449–470.
National Bank of Serbia, Republic of Serbia’s long-term credit rating, on-line:<
http://www.nbs.rs/internet/english/17/17_2/rejting_srbija.html>
National Bank of Serbia (2011), “Annual Financial Stability Report 2011”, National Bank of
Serbia, on-line: <http://www.nbs.rs/export/sites/default/internet/english/90/90_4/fsr_2011.pdf>
National Bank of Serbia (2012) , “Banking Sector in Serbia- Second quarter Report, September
2012” , Banking Supervision Department , National Bank of Serbia, on-line:
<http://www.nbs.rs/export/sites/default/internet/english/55/55_4/quarter_report_II_12.pdf>
National Bank of Serbia (2011) , “Banking Supervision - Third quarter report, 2011”, Banking
Supervision Department, National Bank of Serbia, on-line:
<http://www.nbs.rs/internet/english/55/55_4/quarter_report_IV_11.pdf>
National Bank of Serbia (2012), Statistical Bulletin, on-line
<http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_01_12.pdf>, January 2012.
National Bank of Serbia (2010), Statistical Bulletin, on-line:
< http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_12_11.pdf>, December 2010.
National Bank of Serbia, Republic of Serbia’s long-term credit rating,
on-line:< http://www.nbs.rs/internet/english/17/17_2/rejting_srbija.html>
National Bank of Serbia (2009), Statistical Bulletin, on-line:
<http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_12_09.pdf>, December 2009.
National Bank of Serbia (2008), Statistical Bulletin, on-line:
< http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_12_08.pdf>, December 2008January 2009.
National Bank of Serbia (2007), Statistical Bulletin, on-line:
<http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_12_07.pdf>, December 2007.
National Bank of Serbia (2006), Statistical Bulletin, on-line:
<http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_12_06.pdf>, December 2006.
National Bank of Serbia (2005), Statistical Bulletin, on-line:
<http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_12_05.pdf>, December 2005.
National Bank of Serbia (2004), Statistical Bulletin, on-line:
<http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_12_04.pdf>, December 2004.
National Bank of Serbia (2003), Statistical Bulletin, on-line:
http://www.nbs.rs/static/nbs_site/gen/english/90/statisticki/sb_12_03.pdf, December 2003.
Pain, D. (2003),” The Provisioning Experience of the Major UK Banks? A Small Panel
Investigation”, Bank of England Working Paper, January.
Pesola, J. (2001),” The Role of Macroeconomic Shocks in Banking Crises”, Bank of
Finland, Discussion Paper, No. 6.
Pesola, J. (2007), “Financial fragility, macroeconomic shocks and banks loan losses: evidence
from Europe”, Bank of Finland Research, Discussion papers 15.
Peura and Jokivuolle (2003), ‘Simulation-based stress testing of banks’ regulatory capital
adequacy’, Bank of Finland Discussion Paper Series 4.
Sorge M. (2004), “Stress-testing financial systems: an overview of current methodologies“,
Monetary and Economic Department, Basel Committee on Banking Supervision, Basel.
Sorge M., Virolainen K. (2006), “A comparative analysis of macro stress-testing methodologies
with application to Finland”, Journal of Financial Stability, Volume 2, issue 2, June 2006, pages
113-151.
Virolainen, K. (2004), “Macro Stress Testing with a Macroeconomic Credit Risk Model for
Finland”, Bank of Finland Discussion paper No.18/2004.
Wilson, T C (1997a): “Portfolio credit risk (I)”, Risk, vol 10, issue 9, pp 111-117.
Wilson, T C (1997b): “Portfolio credit risk (II)”, Risk, vol 10, issue 10, pp 56-61.