The effects of state-ownership and state support on European significant banks

(1)

The effects of state-ownership and state support on

European significant banks

Rijksuniversiteit Groningen

Thesis Msc. Finance Supervisor: Mario Hernandez

September 2014 – January 2015

(2)

2

Table of content

1. Introduction ...3

2. Research topic ...5

3. Relevance ...6

4. Review of literature and hypotheses formulation ...7

4.1 Hypothesis 1: customer deposits ...7

4.2 Hypothesis 2: lending ...7

4.3 Hypothesis 3: loan loss provisions ...8

4.4 Hypothesis 4: robustness...9

5. Methodology ...11

5.1 Definitions of state involvement ...11

5.2 Data ...13

5.3 Statistical analysis and used regressions ...16

6. Results and discussion ...21

6.1 Customer deposits ...21

6.2 Growth in outstanding loans ...26

6.3 Loan loss provisions ...28

6.4 Robustness ...31

7. Conclusion and recommendation ...38

(3)

3

1. Introduction

The financial crisis that started in the summer of 2007 had a considerable impact on society and the lives of many around the world. The imploding housing market in the United States exposed the major risks banks took all over the world in the years before and resulted in billions of losses for banks, insurance companies and other investors. Funding of many banks dried out, the interbank money market came to a standstill and a credit crunch was born.

All these events had an impact on the real economy. For firms in all kind of industries, the possibilities to take credit vanished, major layoffs were announced, the gross domestic product (GDP) declined rapidly and depositors feared for bank runs. This resulted in significant government interventions and made political connections with the financial industry more alive than before.

With the massive failure of the financial sector, major banks especially, the general public demanded change. With more strict supervision and less bonuses, banks should be safer in the future and fulfil their utility function as financial intermediaries between deposit holders and loan takers.

Politicians all over the world responded by imposing new legislation and guidelines, and used all other kinds of political power in an attempt to fix the financial system and fulfil the demands of their voters. Supervisors, as the national central banks, responded by implementing new rules and guidelines as well such as Basel III.

Another major development in this field is the centralization of banking supervision within the Eurozone (the countries in the European Union, which use the Euro as currency). In October 2013, the European Central Bank (ECB) announced that they would take over the main supervisory tasks from the National Central Banks of the member states, starting November 2014.

(4)

4

Now the worst part of the financial crisis is over, governments are making plans on how to manage the direct influence they gained by these injections and the way forward. Therefore, it is more important than ever before to investigate the effects of state ownership and state support on banks. If there are any real effects on the capability to absorb new financial shocks, lending to the general public or the collection of deposits, these effects should be incorporated in the strategy governments will follow.

The path this thesis follows to investigate these possible effects start with the defining of the research topic in section two. Thereafter the relevance will be explained in the next section, while in section four relevant literature is reviewed and the four hypothesis of this thesis are formulated. In the fifth section, the used methodology is discussed. Results are presented and discussed in section six. A conclusion is drawn and recommendations for future research are made in the last section.

(5)

5

2. Research topic

This thesis will investigate the influence of state-ownership and state support on European significant banks. Given the severe consequences of the financial crisis of 2007-2009, the aftermath on financial institutions and the subsequent government interventions, a new financial landscape is defined which offers interesting opportunities for research.

This thesis will take a further look into the effects of state involvement in European significant banks on deposit holders who want a safe place to store their savings, on companies and individuals who need loans to fund investments or buy homes, on governments who are trying to kick-start the economy and on regulators trying to prevent new crises. This main research topic about the effects of state-ownership and state support on European significant banks in the aftermath of the financial crisis will therefore be divided in four research questions.

The first research question focuses on the effect on customer deposits, since this is seen as a stable source of funding in Basel III (Basel Committee on Banking Supervision, 2014) and therefore is important in times when other, less stable, funding sources becomes more difficult to obtain for financial institutions. The second research question will be driven by the effects on lending to the general public like mortgages to individuals and loans to business. Thirdly, a closer look will be taken at the loan loss provisions of banks. Loan loss provisions could be an indication of the risk appetite in the pass and the ability of banks to manage their loan portfolio. The fourth and last research question will focus on the robustness of European significant banks. This thesis selected two of the main risk indicators used by regulators, the leverage and Tier 1 capital ratio, and the outcomes of the comprehensive assessment performed by the ECB to reflect the robustness of a bank.

(6)

6

3. Relevance

The amount of funds governments spent to help major banks survive during the most recent financial crisis is significant. As mentioned earlier, in the Eurozone the total amounts to 275 billion euro (ECB, 2013). This money is guaranteed by the taxes paid by the citizens of these countries. Since the real economy is recovering slowly, the time has come for governments and parliaments of these countries to decide about the future of these banks. Need they be privatized, by selling the shares to the public, recover (party) the injected money and make them ‘private’ again or should they stay state-owned and serve the economy as ‘public bank’?

The ECB defined 130 European banks in the Eurozone as ‘significant’. These banks will be subject to central supervision by this institution. Some of these banks are state-owned for many years now, some have been nationalized in recent years, others have received state support in another form and a large group remained private. Together they cover approximately 85% of Eurozone bank assets and the ECB considers them of significant importance. By investigating their assets during an asset quality review and testing their robustness during a stress test, together called the comprehensive assessment, the real quality of the assets is tested. According to the ECB this is needed to create more transparency, repair any problems that they find and build confidence with all stakeholders (ECB, 2013).

(7)

7

4. Review of literature and hypotheses formulation

To investigate the relationship between state-ownership, state support and the effects of this ownership or support in the aftermath of the financial crisis, one main research question and four research questions were defined in section two, in order to structure this thesis. To answer these research questions, four hypotheses are formulated below, based on recent literature. These hypotheses will be empirically tested, using quantitative data from sources described in section five.

4.1 Hypothesis 1: customer deposits

Theoretical evidence suggests public banks will be less vulnerable to deposit withdrawals or bank runs during a crisis (Brei & Schclarek, 2014). The authors suggested that the state has better access to additional funds during a crisis, which could be used for recapitalization of public banks, compared to their private banks and that public banks are more credible in promising a future recapitalization, if needed. This suggests depositors will have a preference for banks in which the government is a major shareholder during or just after a crisis, to protect their deposits, knowing the probability of a bank run is lower in such banks compared to their privately held rivals.

Bertay et al. (2014) found in their empirical research along 1633 banks from 111 countries in the period 1999-2010, that state banks experience relatively low growth rates of their total liabilities during economic booms, especially of their non-deposit liabilities, compared to their private rivals. Consequently, based on this result, it is logically to assume state-owned banks are more familiar with deposit as core funding of their activities and could exploit this experience during times banks are fighting for scare and stable funding, as during a financial crisis.

The findings mentioned result in the following hypothesis:

European significant banks that have state involvement attract more deposits compared to their privately held rivals in the aftermath of the financial crisis.

4.2 Hypothesis 2: lending

(8)

8

banks will reduce their lending more strongly, compared to their public rivals (Brei & Schclarek, 2014).

Empirical research performed on a sample of US banks, suggest banks that hold more illiquid assets, increased their liquid asset holdings and decreased lending during the financial crises of 2007-2009 (Cornett, McNutt, Strahan, & Tehranian, 2011). These researchers also found that banks that relied more on deposits and other stable funding, lent relatively more. Based on this finding, the researchers concluded that liquidity risk exposure was an important driver in the decline of credit production during the crisis. Research of Beltratti & Stulz (2012) confirm this finding, using a smaller but more globally focused sample.

Bertay et al. (2014) found that lending by state banks in high-income countries is countercyclical and that these state banks expand their lending relatively more during banking crises. These researchers also found that foreign-owned banks’ lending is procyclical. This suggest these foreign-owned banks have direct access to funding from their international parents to take advantage of opportunities, but are also scaling down activities when their parents need the funding for other activities in other markets.

The evidence obtained from prior research on this topic, combined with the prior literature researched in the previous paragraph, follows to the following hypothesis:

European significant banks that have state involvement will lend more compared to their privately held rivals in the aftermath of the financial crisis.

4.3 Hypothesis 3: loan loss provisions

Researchers found that state banks report relatively high additional non-performing loans during economic upswings (Bertay, Demirgüç-Kunt, & Huizinga, 2014). According to these researchers this finding may be the result of deteriorating loan quality of state banks during economic upswings or that these banks report non-performing loans more evenly over the business cycle.

(9)

9

Chinese commercial banks have much better financial positions in terms of non-performing loan ratios. Also Sapienza (2004) found that state-owned banks have significantly more non-performing loans compared to privately held banks, based on a sample of Italian banks in the 1990s.

The findings suggest that there is a relationship between state involvement and the ratio of non-performing loans and loan loss provisions. Bertay et al. suggest this could be due to more prudency in both lending to customers as in reporting about their asset quality. Wen also found that non-stated owned Chinese banks have a higher asset growth rate. This could indicate more risk taking and could support the suggestion of Bertay et al. that state owned banks are more prudent in lending. Sapienza suggest the government owned banks do the opposite. They charge less interest and lend more to businesses in depressed areas, possibly based on their political connections.

Although the researched literature is not unanimous in the explanation why, all four papers indicate state-owned banks will report more non-performing loans and therefore loan loss provisions, compared to their privately held rivals. This leads to the following hypothesis.

European significant banks that have state involvement will take more loan loss provisions compared to their privately held rivals in the aftermath of the financial crisis.

4.4 Hypothesis 4: robustness

Researchers found that large banks with less leverage in the year 2006 performed better during the crisis (Beltratti & Stulz, 2012). These researchers performed empirical research using a sample of 164 “large banks” from 32 countries around the world, with total assets in excess of $50 billion. They also found that banks that used more deposits as funding performed better during the financial crisis of 2007-2009.

(10)

10

owned banks are better able to recapitalize. Brei & Schclarek (2014) support this suggestion. The last hypotheses is therefore:

(11)

11

5. Methodology

This section will discuss the main focus points and the design of the methodology that will be used to test the formulated hypotheses. First of all the used definitions for state involvement will be defined, secondly the data sources and sample are discussed and this section will end with the used statistical analyses and formulated regressions.

5.1 Definitions of state involvement

This thesis focusses on the effects of government involvement. To investigate this possible effect, distinction will be made between a wider definition of state support and a more narrow definition of state-ownership. State support, as interpreted in this thesis, includes all kinds of direct support of the state since the start of the financial crisis in 2007, like capital injections or significant explicit state guarantees. State ownership is straighter forward: direct government influence by the holding of shares in the bank.

This distinction provides the opportunity to take a closer look into state involvement, compared to the single variable of state-ownership previous researchers used. State-ownership bypasses the potential influence the government has on banks they supported or the interpretation given by the general public or investors to these supported banks. The state-ownership variable makes it possible to test the more direct influence of the state, by holding direct shares and the related voting power. The state support variable tests the more indirect influence of the state, via the guarantees and the dependency of banks of this support.

(12)

12

banks that have received state support is increased, and therefore increases the state involvement as a significant factor.

Research of Panetta et al. (2009) showed that the announcement of recapitalizations, debt guarantees and assets purchases or guarantees by the government had no positive impact on bank stock prices. However, the researchers tried to investigate the effect of these government interventions on lending, but concluded it may be too early to investigate this topic because of the short time span between those interventions and the time of their research. Also the working paper of Laeven and Valencia (Laeven & Valencia, 2010) supports the importance of government interventions, such as guarantees and asset purchases.

Given the fact that governments used billions to support European banks and the suggestion from previous researchers that this support could be important, it is interesting to include this into the analysis. Additionally, all four of the previous mentioned banks that received state support during the financial crisis did not repay this support in full at the end of 2013. Therefore, the dependency on the government is still there and has there been for multiple years. Hence, if there is a significant effect of this state support, it is logical to assume this would emerge the characteristics of these banks. In addition, when there is a significant effect, also this effect should be incorporated in the strategy going forward in the coming years.

To make this wider definition of state support measurable, this thesis will interpretate recapitalizations, debt guarantees and asset purchases or asset guarantees initiated by the government since the start of the financial crisis and still active during 2013 as criteria for state support.

For the interpretation of state-ownership, a link will be made with previous research. Beltratti & Stulz (2012) defined state ownership as the state owning at least 10% of the share of the specific bank. In their sample, 8% of the banks fulfilled these criteria. This sample was collected of data from the end of 2006.

(13)

13

For the definition of state ownership, this thesis will follow the 50% threshold as defined by La Porta et al., Bertay et al. and Cornett et al. For this threshold is chosen, since this can be seen as direct ownership because the voting power of the majority of the shares could be used to directly influence the strategy of the bank.

5.2 Data

As mentioned before, the sample chosen are the 130 European banks that are marked as significant by the ECB. With the 85% of total assets of all Eurozone banks (ECB, 2013), results found in this sample form a good proxy for all Eurozone banks and could therefore be used for policy making decisions in the Eurozone.

Table 1 shows the distribution of the significant banks across the 19 members of the Eurozone. Germany is the country with the most significant banks, not surprisingly since this is the largest country in the Eurozone by population as well as GDP (Eurostat, 2014).

Table 1: Significant banks per country

Country Number of significant banks

Country Number of significant banks Austria 6 Latvia 3 Belgium 6 Lithuania 3 Cyprus 4 Luxembourg 6 Estonia 3 Malta 3 Finland 3 Portugal 3 France 13 Slovakia 3 Germany 25 Slovenia 3 Greece 4 Spain 15

Ireland 5 The Netherlands 7

Italy 15

The criteria used by the ECB to classify a bank as significant are in principal based on the total assets of the bank (total assets above 30 billion euro or more than 20% of GDP of the country where the headquarter is located), the importance for the (local) economy or the bank is among the three largest credit institutions in the country where the headquarter is located (ECB, 2013).

(14)

14

validated with other sources, as the published annual reports of the significant banks themselves. Any missing information will be added via the same published annual reports and other public sources, like the websites, investor presentations, analyst reports and press releases of European significant banks.

Table 2 shows the variables obtained to test our hypotheses. The variables customer deposits, deposit ratio, loans outstanding, non-performing loans, Tier 1 ratio, leverage (calculated as equity divided by the total assets) and the result of the ECB comprehensive assessment will be used as dependent variables, in the corresponding regression analysis. The deposit ratio is calculated from the obtained customer deposit divided by the loans outstanding.

Table 2:Obtained variables and source

Variable Source Variable Source

Customer deposits Bankscope Total assets Bankscope GDP per capita Bankscope Non-performing loans Bankscope Equity Bankscope Inflation Bankscope Result ECB

comprehensive assessment

ECB Return on average assets

Bankscope

Loans outstanding Bankscope Cost to income Bankscope Listed on stock exchange Bankscope State support Pulbic sources Ownership structure Bankscope and public

sources

Tier 1 ratio Bankscope and public sources

To represent business cycles, in all regressions the growth in GDP per capita and inflation are included. To control for overall country characteristic the GDP per capita is added. For the determination of the location of the bank, the location provided by the ECB is followed.

Also firm specific variables are included in all regressions to control for bank specific characteristics. These variables include the log of the total assets, the growth of the total assets, the return on average assets, the listing of a bank on a stock exchange and the cost-to-income ratio.

In addition the lagged of the dependent variable is included and the dependent variables of the other regressions. In some regressions, specific variables are excluded since these variables did not add any explanatory power.

(15)

15

In table 3 summary statistics are given of the variables used to measure the state involvement in European significant banks, the outcomes of the ECB comprehensive assessment and the listing on a stock exchange. In total, 36 banks are for more than 50% owned by the state and 52 were supported by the state since the start of the financial crisis and did not repay this support in full before 2013. In total 59 banks are either owned by the stated or are still receiving support of the state. Interesting to see, is that 24 banks in the sample failed for the ECB test and did not have enough core capital to prevent shortfalls as a result of the performed asset quality review and stress test. 68 banks are listed on a stock exchange.

Table 3: Descriptive statistics of dummy variables

Variable N Yes No

State ownership 122 36 86

State support 122 52 70

State involvement 122 59 63

Failed ECB comprehensive assessment 122 24 98 Listed on stock exchange 122 68 54

Summary statistics of the most important other variables are given in table 4. From this table, country specific variables are excluded since these will be influenced by the over- and underweight of some countries in the sample. In total, 8 banks are taken out of scope since these banks are already consolidated in the results of other banks in the sample or due to the fact these banks have no loans outstanding and are not collection deposits (i.e. clearing houses).

Interesting is that the mean Tier 1 ratio improved in 2013 and the leverage declined. This could indicate the robustness of the European significant banks improved. Another indicator the robustness of banks have improved is the fact that the mean deposit ratio declined in 2013, indicating more customer deposits were used to fund outstanding loans. This is a results of the declining amount of outstanding loans combined with the increasing amount of customer deposits.

(16)

16

Table 4: Descriptive statistics of most important variables

Variable N Minimum Maximum Mean Std. Dev.

Tier 1 Ratio (2013) 109 5.8000 44.0200 14.2080 5.2861 Tier 1 Ratio (2012) 109 0.6000 39.9300 12.6757 5.1414 Return On Avg Assets (2013) 117 -13.5190 19,7000 -0.0756 3.0397 Growth in loans outstanding (2013) 117 -0.3432 0.4734 -0.0362 0.1196 Growth in loans outstanding (2012) 115 -0.8482 2.4398 -0.0050 0.2867 Cost to Income Ratio (2013) 114 11.5940 656.7570 71.3345 63.4872 Deposit ratio (2013) 115 0.0600 35.8700 1.1657 3.3182 Deposit ratio (2012) 115 0.0200 48.5500 1.2271 4.4907 Non-performing loans (2013) 98 0.0880 44.8630 11.5948 10.7126 Non-performing loans (2012) 103 0.0420 40.3560 9.5893 8.6785 log Total Assets (2013) 117 5.7539 9.2553 7.846750 0.6494 Leverage (2013) 116 0.0170 0.2118 0.0676 0.0374 Leverage (2012) 116 -0.0330 0.2044 0.0581 0.0424 Growth total assets (2013) 117 -0.4484 0.4093 -0.0507 0.1126 Growth customer deposits (2013) 115 -0.8898 0.7249 0.0152 0.1958 Growth customer deposits (2012) 113 -0.9502 1.2549 0.0381 0.2268

Performed correlation tests showed strong correlation between several dependent variables and their lagged dependent variables. These correlations reported to be strong for Tier 1 ratio, leverage, growth in loans outstanding, deposit ratio and non-performing loans. To correct for this correlation, the lagged dependent variables will be included in the regression analysis. This is conform earlier mentioned methodology.

5.3 Statistical analysis and used regressions

To test the formulated hypothesis in section four, multiple OLS regressions will be performed. This approach is similar to approaches used by Beltratti and Stulz (2012), Bertay et al. (2014) and Cornett et al. (2010), all studies that investigated the impact of state-ownership on banks.

(17)

17

time with the state support variable, will show if state-ownership and state support, independently from each other, have a significant effect on European significant banks. This is necessary since some state support resulted in state-ownership according to the defined definition in this thesis, but not all state-ownership resulted in state support during the financial crisis. To control for the fact that outsiders could perceive banks that did receive state support, but are not state owned the regression will be run a third time. This time including the state involvement variable, to which a bank classifies if it is either state-owned, received state support or both.

Hypothesis 1

To answer the first hypothesis, two regressions will be performed. The first regression will focus on the growth in total customer deposits in 2013, while the second hypothesis takes deposit ratio as the dependent variable. Since customer deposits are seen as stable sources of funding, recently emphasized again by the implementation of the new Basel III guidelines, this is a relevant topic.

The first regression will investigate the growth of deposits and the variables explaining that growth. Included in the regression are the lagged dependent variable and the controlling variables for firm characteristics, the business cycle and for country specifics.

Regression 1

Growth deposits (2013) = β1* growth deposits (2012) + β2 * log total assets (2013) + β3 * GDP growth per capita (2013) + β4 * GDP per capita (2013) + β5 * non-performing loans ratio (2013) + β6 * leverage (2013) + β7 * inflation rate (2013) + β8 * listed + β9 * return on average assets (2013) + β10 * deposit ratio (2013) + β11 * cost to income ratio (2013) + β12 * ECB result + β13 * growth loans (2013) + β14 * tier 1 ratio (2013) + β15 * state variable.

(18)

18

same controlling variable are included as in the previous regression, however some variables like the listing on stock exchanges, return on average assets and the result of the ECB comprehensive assessment are excluded. These variables added no explanatory power.

Regression 2

Deposit ratio (2013) = β1* deposit ratio (2012)+ β2 * log total assets (2013) + β3 * GDP growth per

capita (2013) + β4 * GDP per capita (2013) + β5 * non-performing loans ratio (2013)+ β6 * leverage (2013)+ β7 * inflation rate (2013) + β8 * growth deposits (2013) + β9 * cost to income ratio (2013) +

β10 * growth total assets (2013) + β11 * growth loans (2013) + β12 * tier 1 ratio (2013) + β13 * state

variable.

Hypothesis 2

For the second hypothesis, one regression is formulated. This regression looks into the variables explaining the growth in loans outstanding. Again, the lagged dependent variable and the control variables for firm characteristics, country specific variables and firm variables are included. Variables that added not explanatory power are excluded.

Regression 3

Growth loans (2013) = β1* growth loans (2012) + β2 * log total assets (2013) + β3 * GDP growth per

capita (2013) + β4 * GDP per capita (2013) + β5 * non-performing loans ratio (2013)+ β6 * leverage (2013)+ β7 * listed + β8 * return on average assets (2013) + β9 * deposit ratio (2013) + β10 * cost to

income ratio (2013) + β11 * growth total assets (2013) + β12 * tier 1 ratio (2013) + β13 * state

variable.

Hypothesis 3

(19)

19 Regression 4

Non-performing loans ratio (2013) = β1* non-performing loans ratio (2012) + β2 * log total assets (2013) + β3 * GDP growth per capita (2013) + β4 * GDP per capita (2013) + β5 * leverage (2013)+ β6 *

listed + β7 * return on average assets (2013) + β8 * deposit ratio (2013) + β9 * cost to income ratio (2013) + β10 * tier 1 ratio (2013) + β11 * state variable.

Hypothesis 4

For the last hypothesis, that investigated the factors that explain the robustness of banks, three regressions are formulated. The first regression, with Tier 1 ratio as dependent variable, is directly linked to the method Basel III uses to measure the reserves of a bank. The Tier 1 ratio measures the risk weighted assets, compared to the risk weighted capital. The lagged Tier 1 ratio variable is included and control variables for firm characteristics, country specific and business cycles are also taken into account. Again, variables added not any explanatory power have been excluded.

Regression 5

Tier 1 ratio (2013) = β1* Tier 1 ratio (2012) + β2 * log total assets (2013) + β3 * GDP growth per

capita (2013) + β4 * GDP per capita (2013) + β5 * non-performing loans ratio (2013)+ β6 * leverage (2013)+ β7 * inflation rate (2013) + β8 * listed + β9 * return on average assets (2013) + β10 * deposit

ratio (2013) + β11 * cost to income ratio (2013) + β12 * state variable.

(20)

20

Likewise the previous regressions, the lagged leverage ratio is included, together with the controlling variables for the business cycle, firm characteristics and country specifics. Non-explanatory variables are excluded.

Regression 6

Leverage (2013) = β1* Leverage (2012) + β2 * log total assets (2013) + β3 * GDP growth per capita (2013) + β4 * GDP per capita (2013) + β5 * non-performing loans ratio (2013) + β6 * Tier 1 ratio (2013) + β7 * inflation rate (2013) + β8 * listed + β9 * return on average assets (2013) + β10 * deposit

ratio (2013) + β11 * cost to income ratio (2013) + β12 * state variable.

Compared to the first two regressions used to test the fourth hypothesis, the last regression is focussed on a more qualitative and forward looking variable that could explain the robustness of a significant bank: the outcome of the ECB comprehensive assessment. Since the ECB did look into the loan books and assessed the quality of this loans and thereafter performed a stress test, this outcome can be seen as a predictor of the robustness of a European significant bank (as per year end 2013). The last regression investigates the variables that explain the outcome of the ECB test. Included are again the controlling variables and non-explanatory variables are excluded. Since this is the first time this test was performed in this way, no previous results are available and can be included in the regression.

Regression 7

ECB result = β1* Leverage (2013) + β2 * log total assets (2013) + β3 * GDP growth per capita (2013)

+ β4 * GDP per capita (2013) + β5 * non-performing loans ratio (2013) + β6 * Tier 1 ratio (2013) +

β7 * inflation rate (2013) + β8 * listed + β9 * return on average assets (2013) + β10 * deposit ratio

(21)

21

6. Results and discussion

In this section, the results of the regressions performed to test the previous formulated hypothesis will be discussed.

The involvement of the state is tested in three separate analysis: state ownership of the significant bank, state support of the significant bank and state involvement of the significant bank (state ownership of the European significant bank and/or state support of the European significant bank).

6.1 Customer deposits

The first hypothesis states that European significant banks in which the government is involved directly, will attract more customer deposits. This hypothesis is tested by running the in the previous chapter defined first two regression: growth in customer deposits and deposit ratio. Both regressions were run three times: the first time with the state ownership variable, the second time with the narrow definition of state support (state support) and the third time with the wide definition of state support (state involvement).

Regression 1: Growth in customer deposits

Table 5: Model summary regression 1

Model R R Square Adjusted R Square Std. Error of the Estimate State ownership 0.708 0.502 0.404 0.143 State support 0.714 0.509 0.412 0.142 State involvement 0.708 0.501 0.402 0.143

(22)

22

In the second run, with state support as state variable, the model has a R2 of 0.509 and an adjusted R2 of 0.412. The state support is not significant as an explaining variable at the 10% significance level in this model.

Table 6: Coefficients regression 1.

*** significance at 10%, ** significance at 5%, * significance at 1%

Variable Regression 1 Regression 2 Regression 3

(Constant) 0,135 (0.650) 0.130 (0.661) 0.127 (0.670) Lagged growth customer deposits 0.303*** (0.001) 0.330*** (0.001) 0.309*** (0.001) Log total assets -0,016

(0.653) -0,011 (0.752) -0.015 (0.671) Growth GDP per capita 0.451 (0.638) 0.621 (0.519) 0.477 (0.621) GDP per capita 0.000 (0.174) 0.000 (0.409) 0.000 (0.233) Non-performing loan ratio 0.000 (0.971) -0.001 (0.700) 0.000 (0.900) Leverage 0.557 (0.441) 0.517 (0.442) 0.583 (0.389) Inflation rate -0.020 (0.528) -0.017 (0.582) -0.017 (0.577) Listed 0.021 (0.625) 0.030 (0.424) 0.030 (0.433) Return on average assets 0.029*** (0.010) 0.027** (0.017) 0.029** (0.011) Deposit ratio -0.058 (0.142) -0.058 (0.139) -0.055 (0.164) Cost-to-income ratio 0.001** (0.011) 0.001** (0.017) 0.001** (0.013) Result ECB test -0.074

(0.146) -0.075 (0.136) -0.070 (0.164) Growth loans outstanding 0.532*** (0.009) 0.537*** (0.008) 0.523*** (0.010) Tier 1 Ratio -0.008** (0.029) -0.009** (0.013) -0.009** (0.018) State ownership -0.021 (0.639) n/a n/a

State support n/a 0.043 (0.247)

n/a State involvement n/a n/a 0.005

(23)

23

Again the growth in customer deposits of the previous year (at 1% level), return on average assets in the current year (at 5% level), the cost-to-income ratio in the current year (at 5% level), the growth in outstanding loans in the current year (at 1% level) and the tier 1 ratio in the current year (at 5% level) are explaining the growth in customer deposits.

The last run, including the state involvement variable, does make no difference. The R2 _{(0.501) and the adjusted R}2_{(0.402) are comparable, the state} involvement variable is not significant at the 10% significance level and the same variables are explaining the growth in deposits: growth in customer deposits of the previous year (at 1% level), return on average assets in the current year (at 5% level), the cost-to-income ratio in the current year (at 5% level), the growth in outstanding loans in the current year (at 1% level) and the tier 1 ratio in the current year (at 5% level).

Regression 2: Deposit ratio

For the second regression, the first run resulted in a model with a better R2_of 0.987 and adjusted R2_{of 0.985. Again, the state ownership variable is not significant} at the 10% significance level. Variables that do explain the deposit ratio are the deposit ratio of the previous year (at 1% level), GDP growth per capita in the current year (at 5% level), GDP per capita in the current year (at 1% level), non-performing loans ratio in the current year (at 5% level), leverage ratio in the current year (at 1% level), growth in total customer deposits in the current year (at 1% level), cost to income ratio of the current year (at 5% level), growth in total assets in the current year (at 5% level), growth in outstanding loans in the current year (at 1% level) and Tier 1 ratio (at 1% level).

(24)

24

10% significance level. The deposit ratio of the previous year (at 1% level), the log of total assets (at 10% level), GDP growth per capita in the current year (at 5% level), GDP per capita in the current year (at 1% level), non-performing loans ratio in the current year (at 5% level), leverage ratio in the current year (at 1% level), growth in total customer deposits in the current year (at 1% level), cost to income ratio in the current year (at 5% level), growth in total assets in the current year (at 1% level), growth in outstanding loans in the current year (at 1% level) and the tier 1 ratio in the current year (at 5% level) do explain the deposit ratio in this model.

Table 8: Coefficients regression 2

(Constant) -0.066 (0.545) -0.064 (0.560) -0,058 (0.598) Lagged deposit ratio 0.945***

(0.000)

0.946*** (0.000)

0.945*** (0.000) Log total assets 0.023*

(0.081) 0.022* (0.100) 0.021 (0.106) Growth GDP per capita 0.796*** (0.030) 0.735** (0.047) 0.758** (0.039) GDP per capita 0.000*** (0.003) 0.000*** (0.007) 0.000*** (0.007) Non-performing loan ratio 0.003** (0.014) 0.003** (0.011) 0.003** (0.011) Leverage -0.782*** (0.002) -0.756*** (0.002) -0.776*** (0.002) Inflation rate 0.015 (0.215) 0.016 (0.186) 0.016 (0.194) Growth customer deposits 0.380*** (0.000) 0.385*** (0.000) 0.380*** (0.000) Cost-to-income ratio 0.000** (0.023) 0.000** (0.024) 0.000** (0.023) Growth total assets 0.288**

(0.012) 0.293*** (0.009) 0.290*** (0.010) Growth loans -0.522*** (0.000) -0.538*** (0.000) -0.529*** (0.000) Tier 1 Ratio 0.003* (0.055) 0.003** (0.047) 0.003** (0.047) State ownership -0.008 (0.608) n/a n/a

State support n/a -0.012 (0.385)

n/a State involvement n/a n/a -0.012

(25)

25

The last run, with the state involvement variable, resulted again in the same R2 (0.987) and adjusted R2 (0.985). Also the state involvement is not a significant variable at the 10% significance level. Again the same variable do explain the deposit ratio: The deposit ratio of the previous year (at 1% level), GDP growth per capita in the current year (at 5% level), GDP per capita in the current year (at 1% level), non-performing loans ratio in the current year (at 5% level), leverage in the current year (at 1% level), growth in total customer deposits in the current year (at 1% level), cost to income ratio in the current year (at 5% level), cost-to-income ratio (at 5% level), growth in total assets in the current year (at 1% level), growth in outstanding loans in the current year (at 1% level) and the tier 1 ratio in the current year (at 5% level).

Based on the results of the first two regressions, there is not enough evidence to support the hypothesis that European significant banks where the government is actively involved, attract more deposits than European significant banks were the government is not involved. None of the runs resulted in significant results for the variables interested in for this hypotheses: the first hypothesis has to be rejected.

The results are not in line with the outcomes of the Bertay et al. (2014) and Brei & Schclarek (2014). A possible reason could be that they really looked at period of crisis, while the timeframe of this thesis is focussed on the aftermath after the financial crisis 2007. Another explanation could be the contamination of the dataset by the chosen definition of state-ownership. Since banks that received an equity injection during the financial crisis which resulting in a state-ownership are included, the effects of required restructuring of these banks are included in this thesis and can be the opposite of the behaviour of banks that have been state-owned for more years.

(26)

26

6.2 Growth in outstanding loans

Second, the lending of European significant banks is topic of interest. The hypothesis is that European significant banks that have active government involvement, will lend more to the general public. This is tested by running the defined third regression three times: the first time with the state ownership variable, the second time with the state support variable and the third time with the state involvement variable.

Regression 3: the growth in outstanding loans

The model in the first run, in which the state ownership variable is included, has a R2 of 0.807 and an adjusted R2 of 0.775. The state ownership has variable is significant at the 5% significance level and has a positive influence on the growth of outstanding loans. Other variables that explain the growth in outstanding loans are the growth of outstanding loans in the previous year (at 1% level), non-performing loans ratio in the current year (at 5% level), return on average assets in the current year (at 1% level), deposit ratio in the current year (at 5% level) and growth in total assets in the current year (at 1% level).

The second run, including the state support variable, resulted in a model with a R2 of 0.791 and an adjusted R2 of 0.757. The state support variable is not significant at the 10% significance level. Variables that do explain the growth in outstanding loans are the growth of outstanding loans in the previous year (at 1% level), return on average assets in the current year (at 1% level), growth in total assets in the current year (at 1% level) and the deposit ratio (at 10% level).

(27)

27

1% level), return on average assets in the current year (at 1% level), deposit ratio in the current year (at 5% level) and growth in total assets in the current year (at 1% level).

Variable Regression 1 Regression 2 Regression 3 (Constant) -0.087 (0.352) -0.106 (0.272) -0.113 (0.240) Lagged growth outstanding loans 0.184*** (0.000) 0.189*** (0.001) 0.195*** (0.000) Log total assets 0.005

(0.644) 0.008 (0.484) 0.010 (0.421) Growth GDP per capita -0.397 (0.184) -0.400 (0.206) -0.381 (0.221) GDP per capita 0.000 (0.135) 0.000 (0.206) 0.000 (0.272) Non-performing loan ratio -0.002** (0,035) -0.001 (0.143) -0.001 (0.101) Leverage 0.124 (0.580) 0.105 (0.652) 0.127 (0.585) Listed 0.012 (0.412) -0.005 (0.717) -0.004 (0.765) Return on average assets 0.012*** (0.001) 0.012*** (0.002) 0.012*** (0.002) Deposit ratio 0.029** (0.025) 0.026* (0.052) 0.027** (0.040) Cost-to-income ratio 0.000 (0.187) 0.000 (0.344) 0.000 (0.313) Growth total assets 0.637***

(0.000) 0.633*** (0.000) 0.634*** (0.000) Tier 1 Ratio -0.001 (0.359) 0.000 (0.799) 0.000 (0.690) State ownership 0.039** (0.012) n/a n/a

(0.256)

(28)

28

are owned by the government are better able to create new loans. One possible explanation is that banks that are supported during the crisis and still have to repay the state support, and are busy with restructuring the bank, while European significant banks that are owned by the government are more robust - refer also to the results of regressions five, six and seven - and therefore able to supply loans to the general public, since they have less pressure to restructure and can focus on their core business. Given this results, it is only possible to partly accept the current hypothesis. State-ownership has a significant effect on lending, while state support is not associated to the growth of loans outstanding.

The results of state-ownership are consistent with the results that Cornett et al. (2011), Beltratti & Stulz (2012) and Brei & Schclarek (2014) found, although the explanation for this results could not be verified by this thesis. As noted in the previous paragraph, this thesis noted no significant association between state-ownership and state support with the deposit ratio, one of the main arguments used by Cornett et al. and Beltratti and Stulz for the explaination of this outcome. Again, it could be that the timeframe differs too much and the results are limited to a certain point in the cycle or that the dataset is polluted by the chosen definition.

Another interesting outcome of the third regression is the positive influence of the return on average assets on the growth in loans outstanding. This indicates that banks that are able to make higher returns on their current assets, are able to reinvest that in loans. This could indicate that successful European banks are able to gain market share and invest in new profitable investment, and therefore strengthen their position even further. This is consistent with the outcome from paragraph 6.1, indicating that banks with better operational results are able to attract more deposits.

6.3 Loan loss provisions

(29)

29

Regression 4: non-performing loans

Model R R Square Adjusted R Square Std. Error of the Estimate State ownership .972 ,946 ,938 2.605 State support .972 ,946 ,938 2.600 State involvement .972 ,946 ,938 2.601

(Constant) 9.764** (0.045) 9.872** (0.042) 10.009** (0.040) Lagged non-performing ratio 1.043*** (0.000) 1.051*** (0.000) 1.050*** (0.000) Log total assets -0.910

(0.127) -0.957 (0.111) -0.960 (0.110) Growth GDP per capita -99.841*** (0.000) -101.510*** (0.000) -101.109*** (0.000) GDP per capita 0.000 (0.684) 0.000 (0.842) 0.000 (0.826) Leverage -3.587 (0.760) -3.621 (0.757) -4.189 (0.721) Listed 0.231 (0.760) 0.251 (0.707) 0.227 (0.735) Return on average assets -0.061 (0.741) -0.049 (0.790) -0.053 (0.776) Deposit ratio -0.187 (0.779) -0.179 (0.787) -0.219 (0.743) Cost-to-income ratio 0.001 (0.919) 0.001 (0.879) 0.001 (0.888) Tier 1 ratio 0.008 (0.906) 0.011 (0.862) 0.012 (0.843) State ownership -0.071 (0.919) n/a n/a

State support n/a -0.363 (0.579)

n/a State involvement n/a n/a -0.355

(0.592)

(30)

non-30

performing loans are the level of non-performing loans in the previous year (at 1% level) and growth of GDP per capita in the current year (at 1% level).

The model of the second, with the state support variable, resulted in a model with a R2 of 0.972 and an adjusted R2 of 0.938. The state support variable is not significant at the 10% significance level. Variables that do explain the level of non-performing loans in this model are again the level non-non-performing loans in the previous year (at 1% level) and growth of GDP per capita in the current year (at 1% level).

The last model, with the state involvement variable, had a model with a R2 of 0.946 and an adjusted R2 of 0.938. In this model is the state involvement is not significant at the 10% significance level. And also for this model the same two variables explain the level of non-performing loans: the level non-performing loans in the previous year (at 1% level) and growth of GDP per capita in the current year (at 1% level).

Given the results of the fourth regression, the third hypothesis is rejected. All the runs show no significant results for the variables that would explain active involvement of the government in the European significant banks.

The results show two clear outcomes, resulting from all three runs. First, first the level of non-performing loans depends on the level of non-performing loans of the previous year. The second is that the level of non-performing loans depends on the growth of the GDP per capita. When the GDP per capita in the main country of the European significant bank grows, the level of non-performing will be lower compared to a declining GDP per capita. This is conform the general view that banks in the countries that where hit the hardest have the worst loan portfolio.

(31)

31

6.4 Robustness

The fourth hypothesis concerns the robustness of the significant banks. To test the robustness, we formulated three proxies: the tier 1 ratio, the leverage ratio and the results of the ECB comprehensive assessment. These three proxies for the robustness of the European significant banks will be tested in three runs: the first time with the state ownership variable, the second time with the narrow definition of state support and the third time with the wide definition of state support.

Regression 5: Tier 1 ratio

Model R R Square

Adjusted R Square

Std. Error of the Estimate

State ownership 0.934a 0.873 0.853 2.1057 State support 0.937a 0.878 0.859 2.0653 State

involvement

0.936a 0.877 0.858 2.0706

The first run, including the state ownership variable, resulted in a model with a R2 of 0.873 and an adjusted R2 of 0.853. The state ownership variable is significant at the 5% significance level and has a positive influence on the tier 1 ratio. Three other variables explain the tier 1 ratio: the tier 1 ratio of the previous year (at 1% level), GDP growth per capita (at 10% level), GDP per capita in the current year (at 5% level) and listed on the stock exchange (at 5% level).

The model of the second run, including the state support variable, gave a R2_of 0.878 and an adjusted R2_{of .859. The state support variable is significant at the 1%} significance level and has also a positive influence on the tier 1 ratio. Again the tier 1 ratio of the previous year (at 1% level), GDP per capita in the current year (at 1% level) and listing on a stock exchange (at 1% level) are influencing the tier 1 ratio.

(32)

32

(Constant) 5.005 (0.213) 4.372 (0.266) 3.697 (0.348) Lagged Tier 1 ratio 0.935***

(0.000)

0.959*** (0.000)

0.951*** (0.000) Log total assets -0.157

(0.746)

0.096 (0.840)

0.112 (0.815) Growth GDP per capita -22.882*

(0.093) -17.264 (0.198) -18.846 (0.160) GDP per capita 0.000** (0.019) 0.000*** (0.003) 0.000*** (0.003) Non-performing loan ratio -0.026

(0.465) -0.036 (0.311) -0.033 (0.348) Leverage 14.565 (0.119) 13.104 (0.153) 15.570* (0.091) Inflation rate 0.600 (0.181) 0.401 (0.354) 0.434 (0.317) Listed -1.448** (0.021) -2.089*** (0.000) -1.970*** (0.000) Return on average assets -0.093

(0.537) -0.134 (0.368) -0.119 (0.424) Desposit ratio 0.234 (0.676) 0.028 (0.958) 0.206 (0.707) Cost-to-income ratio 0.006 (0.236) 0.007 (0.170) 0.007 (0.158) State ownership 1.474** (0.026) n/a n/a

State support n/a 1.491*** (0.005)

n/a State involvement n/a n/a 1.456***

(0.006)

Regression 6: Leverage

involvement

(33)

33

(Constant) 0.060** (0.013) 0.055** (0.020) 0.052** (0.029) Lagged leverage 0.786*** (0.000) 0.790*** (0.000) 0.803*** (0.000) Log total assets -0.004

(0.219)

-0.002 (0.467)

-0.002 (0.488) Growth GDP per capita -0.327***

(0.001) -0.297*** (0.001) -0.314*** (0.001) GDP per capita 0.000*** (0.000) 0.000*** (0.000) 0.000*** (0.000) Non-performing loan ratio 0.000

(0.676) 0.000 (0.912) 0.000 (0.872) Tier 1 ratio 0.001*** (0.000) 0.001*** (0.000) 0.001*** (0.000) Inflation rate -0.004 (0.102) -0.006** (0.035) -0.006** (0.039) Listed 0.004 (0.278) 0.001 (0.849) 0.001 (0.734) Return on average assets 0.004***

(0.000) 0.004*** (0.000) 0.004*** (0.000) Deposit ratio -0.005 (0.142) -0.006* (0.072) -0.005 (0.131) Cost-to-income ratio 0.000*** (0.000) 0.000*** (0.000) 0.000*** (0.000) State ownership 0.008* (0.062) n/a n/a

State support n/a 0.008** (0.015)

n/a

State involvement n/a 0.008**

(0.017)

(34)

34

The second run, including the state support variable, modelled a R2 of 0.897 and an adjusted R2 of 0.881. The state support variable is significant at the 5% significance level and has a positive influence on the leverage in the current. Also the leverage of the previous year (at 1% level), growth of GDP per capita in the current year (at 1% level), GDP per capita in the current year (at 1% level), tier 1 ratio in the current year (at 1% level), the inflation rate in the current year (at 5% level), return on average assets in the current year (at 1% level) and the cost-to-income ratio in the current year (at 1% level) have a positive influence on the leverage.

The model created for the third run, including the state involvement variable, showed a R2 of 0.896 and an adjusted R2 of 0.881. The state support variable was significant at the 5% significance level and had a positive influence on the leverage. Other relevant variables are the leverage ratio of the previous year (at 1% level), growth in GDP per capita in the current year (at 1% level), GDP per capita in the current year (at 1% level), tier 1 ratio in the current year (at 1% level), inflation rate in the current year (at 5% level), return on average assets in the current year (at 1% level) and the cost-to-income ratio in the current year (at 1% level).

Regression 7: ECB comprehensive assessment

The first run, including the state ownership variable, modelled a R2 of 0.488 and an adjusted R2 of 0.411. The state ownership variable is significant at the 10% significance level and has a negative influence on the chance that a European significant bank failed for the ECB comprehensive assessment. Other variables that explained the outcome of the ECB comprehensive assessment are the leverage in the current year (at 5% level), log of total assets in the current year (at 1% level), growth of GDP per capita in the current year (at 1% level) and the cost-to-income ratio in the current year (at 5% level).

involvement

(35)

35

(Constant) 2.367*** (0.000) 2.431*** (0.000) 2.409*** (0.000) Leverage -3.068** (0.034) -2.984** (0.042) -2.896** (0.049) Log total assets -0.252***

(0.001)

-0.256*** (0.001)

-0.257*** (0.001) Growth GDP per capita -5.833***

(0.004) -5.658*** (0.007) -5.764*** (0.006) GDP per capita 0.000 (0.876) 0.000 (0.804) 0.000 (0.844) Non performing loan ratio 0.007

(0.177) 0.004 (0.529) 0.004 (0.494) Tier 1 ratio 0.002 (0.773) -0.003 (0.740) -0.003 (0.731) Inflation rate -0.031 (0.643) -0.013 (0.841) -0.011 (0.865) Listed 0.073 (0.435) 0.160* (0.060) 0.163* (0.056) Return on average assets 0.007

(0.744) 0.004 (0.860) 0.005 (0.834) Deposit ratio -0.133 (0.119) -0.114 (0.183) -0.107 (0.213) Cost-to-income 0.001** (0.050) 0.001 (0.129) 0.001 (0.124) State ownership -0.187* (0.061) n/a n/a

(0.506)

During the second run, including the state support variable, the model showed a R2 of 0.469 and an adjusted R2 of 0.389. The state support variable is not significant at the 10% significance level. Variables that explained the outcome of the ECB comprehensive assessment in the model are: leverage in the current year (at 5% level), log of total assets in the current year (at 1% level), the growth of GDP per capita in the current year (at 1% level) and the listing on a stock exchange (at 10% level).

(36)

36

10% significance level. Only the leverage of the current year (at 5% level), log of total assets in the current year (at 1% level), the growth of GDP per capita in the current year (at 1% level) and the listing on a stock exchange (at 10% level) are significant in this model.

The last three regressions show clear results. European significant banks owned by the government have higher tier 1 ratios, a lower leverage and have better performed during the ECB comprehensive assessment. Hence, these European significant banks are more robust compared to their privately owned rivals and have a lower chance to run into difficulties. These results are confirm the expectations stated in the thesis based on the studies of Beltratti & Stultz (2012), Bertay et al. (2014) and Brei & Schclarek (2014).

Also the European significant banks that still actively receiving state support have higher tier 1 ratios and lower leverage. This could indicate that the support these banks have received has helped them to become more robust and give them a buffer for potential future economic downturns. However, the state support showed not to be a significant variable for the results of the ECB comprehensive assessment. Since this was more a qualitative exercise, the assessment did take a more in-depth investigation into the assets quality compared to the tier 1 ratio which is more based on models and risk-classification given to specific asset categories. This could be an indication that these banks are still not strong enough to withstand a possible economic downturn and the state support that already has been given was not enough. Further research could focus on this topic.

Concluding, there is there is enough evidence to accept the hypothesis and state that state involvement have a positive effect on the robustness of European significant banks.

(37)

37

Other interesting results are the significant relation between the leverage and the return on assets and with the cost-to-income ratio, but not with the Tier 1 ratio and the ECB results. Apparently European significant banks with better operational results have relatively more equity, but these operational results are not explaining the Tier 1 ratios and the outcomes from the ECB results.

Another outcome that is interesting, is that the GDP per capita has a significant positive influence on the Tier 1 ratio and is associated with a lower leverage, but that this influence is limited. The growth of the GDP per capita is negatively associated with the chance that an European significant bank failed for the ECB comprehensive assessment and passively associated with the leverage, but no association with Tier 1 ratio. A possible explanation could be that countries that have faced growing GDP per capita have less bad assets (and therefore less hidden losses) and more confidence in the future, and could therefore handle higher leverages. However, this results are not very clear and deserve addition future research.

(38)

38

7. Conclusion and recommendation

This thesis analysed the effected of state involvement, in the form of state-ownership and state support, on the European banks that were marked as significant by the ECB. The results are interesting. State-owned banks are associated with better Tier 1 ratios, lower leverage, better chance to pass for the ECB comprehensive assessment and more growth in loans outstanding to the general public. State supported banks are also associated with a better Tier 1 ratio and lower leverage, but this state support does not have a significant influence on the chance of passing for the ECB comprehensive assessment and on the growth in loans outstanding.

The results of this theses are relevant for policy decisions taken by governments. First of all, state involvement is associated with more robustness of European significant banks. Better Tier 1 ratios and lower leverage could lead to a lower possibility of state aid in the future. This could be a reason for governments to keep the shares they gained and use the influence they have to increase the Tier 1 ratio and lower the leverage even further. Also the associating between state-ownership and growth in loan outstanding is relevant for policy makers. Since the economy is recovering only slowly from the financial crisis and small- and medium enterprises are indicating that the willingness of banks to lend is declined, state-owned banks could help to kick-start the economy by generating new loans and fund new investments.

This thesis showed different results for state supported and state-owned European significant banks on the growth of loans outstanding and on the forward looking robustness of banks. Future research should investigate this indication further, because this could be of interest for future state aid in the case of an emergency. If state-ownership has more positives on the future of the bank (in terms of loans outstanding to kick start the economy and on forward looking robustness), this should be incorporated in the decision making process during an emergency

(39)

39

(40)

40

References

Basel Committee on Banking Supervision. (2014). Basel III: The Net Stable Funding

Ratio.

BBC. (2008, 10 13). UK banks receive £37bn bail-out. Retrieved from BBC.co.uk: http://news.bbc.co.uk/2/hi/business/7666570.stm

Beltratti, A., & Stulz, R. (2012). The credit crisis around the globe: Why did some banks perform better? Journal of Financial Economics, 1-17.

Bertay, A., Demirgüç-Kunt, A., & Huizinga, H. (2014). Bank ownership and credit over the business cycle: Is lending by state banks less procyclical? Journal of

Banking & Finance.

Brei, M., & Schclarek, A. (2014). A theoretical model of bank lending: Does ownership matter in times of crisis? Journal of Banking & Finance.

Cornett, M., Guo, L., Khaksari, S., & Tehranian, H. (2010). The impact of state ownership on performance differences in privately-owned versus state-owned banks: An international comparison. Journal of Financial Intermediation. Cornett, M., McNutt, J., Strahan, P., & Tehranian, H. (2011). Liquidity risk

management and credit supply in the financial crisis. Journal of Financial

Economics, 297-312.

De Nederlandsche Bank. (2008, 10 19). Government reinforces ING’s core capital by

EUR 10 billion. Retrieved from DNB.nl:

http://www.dnb.nl/en/news/news-and-archive/persberichten-2008/dnb189474.jsp

Dutch Ministry of Finance. (2013, 02 01). State of the Netherlands nationalises SNS

REAAL. Retrieved from Government.nl:

http://www.government.nl/news/2013/02/01/state-of-the-netherlands-nationalises-sns-reaal.html

ECB. (2013). Note comprehensive assessment.

Eurostat. (2014). Key figures on Europe, 2014 edition.

La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2002). Government Ownership of Banks. Journal of Finance, 265-301.

Laeven, L., & Valencia, F. (2010). Resolution of Banking Crises: The Good, the Bad, and the Ugly. IMF Working Paper.

(41)

41

NRC. (2008, 03 10). Dutch government nationalises Fortis and ABN Amro. Opgehaald van NRC.nl: http://vorige.nrc.nl/international/article2008873.ece Panetta, F., Faeh, T., Grande, G., Ho, C., King, M., Levy, A., . . . Zaghini, A. (2009).

An assessment of financial sector rescue programmes. BIS Papers.

Sapienza, P. (2004). The effects of government ownership on bank lending. Journal

of Financial Economics, 357-384 .