MEASURED IN GOVERNMENT DEBT AND LIQUIDITY RATIOS OF INDIVIDUAL BANKS

(1)

RIJKSUNIVERSITEIT GRONINGEN: FACULTY OF ECONOMICS AND BUSINESS

AN ANALYSIS OF

FINANCIAL STABILITY OF

GOVERNMENTS AND

BANKS

MEASURED IN GOVERNMENT DEBT AND LIQUIDITY

RATIOS OF INDIVIDUAL BANKS

Master thesis International Economics and Business

Ilse Arends

Student number: s1875973

Email: i.arends.1@student.rug.nl

Master: International Economics and Business Supervisor: Michiel Gerritse

Co-supervisor: Anna Samaryna

(2)

ABSTRACT

The purpose of this study is to describe and analyze whether the financial stability of governments have an effect on the financial stability of individual banks in the country. The financial stability of the government is valued by government debt and the stability of individual banks by their liquidity ratio. The empirical basis of the study is a difference-in-difference fixed effects model. The results show that government debt has a significant negative impact on the liquidity of individual banks. Robustness checks were performed and these confirmed the results.

(3)

1. INTRODUCTION

The financial crisis of 2008 had a major influence on the world economy, especially after the bankruptcy of the Lehman Brothers, the financial markets became truly volatile (ECB, 2010). Due to the major impact the crisis had, more attention was drawn to the causation of the crisis. Financial stability came as a topic to the foreground and became increasingly important for governments and the banking sector. To verify that the financials of the countries and banks in Europe remain stable, a new capital accord came into place, resulting in Basel III (Basel Committee on Banking Supervision, 2013). The need for new accords and ratios demonstrates the need for extra measures for financial stability, and the urgency of sound financials.

The financial crisis, that included a debt crisis and banking crisis, hit the world hard. Budgets of governments were affected, which resulted in government debt levels to increase (Olivares-Caminal, 2011). Governments could be placed in positions were they needed to default, which would especially reduce the short-term funding for domestic banks (IMF, 2012). The events of the last years illustrated what the effect of high government debts can be on the stability of banks (Gennaioli, Martin and Rossi, 2014). The new directives (Basel Committee on Banking Supervision, 2013) as well as the crisis implied the link between financial stability of governments and banks.

(6)

The research focuses on the financial stability of governments and the effect of this on the financial stability of individual banks. The link regarding financial stability will be explained by liquidity ratios of the banks in the different countries, related to the government debt of the country (Basel Committee on Banking Supervision, 2013). The added value of this research is the channel through which financial stability of a country can influence the financial stability of a bank that is located in that specific country. Therefore, the main research question is:

Looking at the debt ratios of domestic governments and the liquidity coverage ratios of banks, does the financial stability of a country affect the financial stability of an individual bank?

The question will be answered by measuring liquidity ratios of banks located in Europe, whereof the location will be matched to a country. Afterwards, the effect of the debt of a country on the liquidity of an individual bank will be checked. This will be done for domestic banks as well as banks that are located in several countries. To do this, a difference-in-difference model with fixed effects is used to measure the impact of government debt on liquidity of individual banks. The difference indicates whether domestic banks are more sensitive to domestic government debt or not.

(7)

2. LITERATURE REVIEW

2.1 FINANCIAL STABILITY FROM GOVERNMENTS TO BANKS

Especially after the last crisis of 2008, where two crises took place at the same time, namely the debt crisis and the banking crisis, the links between governments and banks came in the spotlight (Bofonfi, Carpinelli and Sette, 2013). These last crises certainly illustrated the link between government default and the financial stability of banks (Gennaioli, Martin and Rossi, 2013). Researchers statistically proved that the increase in risk for a sovereign default increases the probability of a responding banking crisis (Borensztein and Panizza, 2008). To illustrate what happened during the crisis, the government needed to support the banks with public funds, which results in a transfer from private debt to public debt (Diacon, Donici and Maha, 2013). Contagion fears started to spread, while governments were wrestling with their financial situation and banks were in liquidity problems. As this fear rose, worst case scenarios were considered. One of the major concerns was about the risk that the banking sector in Europe was facing (Black et al., 2013). If a default of a government would lead to the collapse of an European bank, the results occurring could be tremendous. This type of scenario highlights the need for further clarifications of the risks associated to government that lead to financial instability of banks. The growing research on this topic indicates that there is growing awareness of the link between banks and governments (Perez, 2014; Reinhart and Rogoff, 2011; Gennaioli et al., 2014). Last years, more research has been done to provide a more solid link between banks and governments (Perez, 2014) and to add to the current body of research, this paper will discuss the influence of government debt on bank liquidity.

(8)

withstand within the system. To address financial stability, policymakers and researchers have focused on clarifying and assessing different measures (Gadanecz and Jayaram, 2009). The set of soundness indicators of the IMF (2014) is such an example.

First the two indicators for financial stability that are used in the research will be described. There after their relation and the differences in banks that are exposed to several governments and the banks that are not, will be explained.

2.2 FINANCIAL STABILITY INDICATORS

Individual indicators are used to analyze the functioning and stability of the financial system (Gadanecz and Jayaram, 2009). Regarding this, the paper entails one indicator for government financial stability and one for measuring the financial stability of banks. It must be noted, that it would be best to investigate the financial stability by more than one indicator (IMF, 2014), however, this will be beyond the scope of this paper.

2.2.1 Government debt

(9)

institutional reforms and increases in taxes, which will hurt voters directly (Shirakawa, 2012). When looking at the three options presented, the options explain why governments default, although it damages the financial system.

Accordingly, the financial stability of a countries’ financial system is affected by government debt. High debt burdens of governments reduce the ability of the financial system to manage shocks in the system (Higgins and Merfeld, 1987). Besides this, sovereign debt is difficult to structure and therefore the effects of high debt levels will remain for a longer period than for private sector debt. This is, amongst other reasons, due to limits in enforceability of regulations regarding payments. Another reason is that government debt have a major role in the operations of financial markets. When a government can repay its debt, this is often due to the ability to grow in GDP, and whether the government is able to generate enough regarding their primary balance (Brooke et all., 2013) Also, the ECB indicates that the growing government debt rates state low growth possibilities and financial instability (ECB, 2014). Since government debt is one of the most well-known indicators of government performance, this indicator will be used for the indication of financial stability of governments.

2.2.2 Liquidity of individual banks

The declining quality of capital and equity, as well as insufficient liquidity during the financial crisis accounted for a near termination of the banking system (BIS, 2011:1). When declining and insufficient ratios came together with periods of downturn or crisis, confidence in solvency and liquidity of the banking system and banks in general was lost. The financial stability of the sector is outlined by indicators as real interest rates, risk measures, capital and liquidity ratios and quality of loans. All these indicators could reflect the complications in the banking sector (Gadanecz and Jayaram, 20091). Since the crisis, a lot of indicators are used to measure the stability of banks. The IMF (2014) introduced several financial soundness indicators, with ratios as nonperforming loans to total gross loans, return on assets, liquid asset ratio, and a few more. It also introduced an encouraged set of indicators to identify stability more in-depth (IMF, 2014). From the commonly used financial indicators are the soundness indicators most popular. In this research, the financial stability of banks is measured in only one indicator as the dependent variable. The BIS featured the fact that

(10)

strong capital and equity requirements are not sufficient when looking at the financial stability of a bank, a strong liquidity base is as essential (Basel Committee on Banking Supervision, 2013). This was especially underlined by the last crisis. During this crisis is became visible that, although banks had sufficient capital buffers, they still experienced distress due to liquidity problems (Basel Committee on Banking Supervision, 2013).

The liquidity ratio is chosen because of the relation with financial stability as well as the indication whether a bank is able to absorb shocks in the short term. The liquidity ratio includes whether banks are able to convert assets in short term money, which can be used to absorb shocks (Liang, 2013). This means that low liquidity ratios indicate that banks will not be able to absorb shocks in the short term, and when shocks do appear this can lead to severe damage of the bank (Tarulo, 2014). Growing awareness indicates the importance of the liquidity ratio of banks (Deléchat et al., 2012). Indicators as return on assets, and others, do indicate the effect on financial stability. When looking at the existing literature, especially liquidity ratios and loan ratios are commonly used indicators of financial stability of banks (Basel Committee on Banking Supervision, 2013; Holmstrom and Tirole, 1998; IMF, 2014). It is noted that banks, to some extent, have control over their liquidity position, while it does not have direct control over impairment of loans (IMF, 2014). For the purpose of this paper, the liquidity ratio is used as a dependent variable. To illustrate the difference aspects of financial stability for banks the loan ratio, consisting of impaired loans, is taken into account as a robustness check. Next to this, it must be noted that these ratios are short term liquidity ratios, while it is also of importance to check for financial stability on the long term (Pierret, 2014). This effect is captured in a robustness check with an indicator for bank solvency.

(11)

on Banking Supervision, 2013). Looking at the influences on liquidity of individual banks, four main influences can be seen: opportunity costs and shocks to funding, bank characteristics, macroeconomic fundamentals, and Moral hazard and safety nets (Deléchat et al., 2012: 5-7). As Deléchat et al. (2012: 5) indicated, opportunity costs and shocks to funding is what the banks perceive to happen, since they will not have too much liquidity as it is costly, while they would like to be prepared when volatility appears. Whether this is likely to occur will have influence on the amount of liquidity possessed by a bank. Regarding bank characteristics, this relates to whether banks are constrained by market imperfections to hold liquidity. Financial constrained banks are more likely to have higher liquidity buffers. While macroeconomic fundamentals entail the fact that banks are in need of more liquidity during recessions and crises and have less need for liquidity when advanced times appear (Deléchat et al., 2012: 6). Moral hazard and safety nets relate to whether a banks has the need to have liquidity buffers, since availability of arrangements could reduce this (Deléchat et al., 2012: 7). These influences have in common that liquidity is used to dampen the first shocks facing a bank.

To ensure against liquidity risk, it is for banks fundamental to have sufficient liquidity (Diamond and Dybvig, 1983; Diamond and Rajan, 2001). It is not only fundamental for banks to have sufficient liquidity, it is also a requirement set by central banks (ECB, 2014). Banks have several options to ensure sufficient liquidity. Loans that are not performing can lead to withdrawals and because of an imperfect capital market, it can become too expensive to raise liquidity again in the short run (Deléchat et al., 2012). In this case, to regain liquidity banks can lend from interbank markets, a central bank or other financial agencies. Even though these mechanisms are into force, illiquid banks could still fail because of the absent of a last resort lender such as the domestic government (Rochet and Vives, 2004). Liquidity is needed to be financial stable, however if liquidity becomes too high this could also have negative impacts on the stability of banks in the long run (Gray, 2011).

(12)

others what the loan is worth. This indicates that a banks can get cash, and therefore liquidity, but only for a higher price, and maybe not even in a sufficient amount. Even when there is a suspicion that the bank cannot meet the liquidity demand, more depositors may be wanting to withdraw and therefore this can lead to a bank run (Tarulo, 2014). This demonstrates that low liquidity ratios can lead to financial instability and even a bank run.

2.2.3 Government debt and the liquidity of individual banks

To relate banks’ and governments’ financial stability, the levels of outstanding government debt is, in many cases, making significant effects on the domestic financial market (Brooke et all., 2013). When banks buy bonds from their own government they expose themselves to sovereign risk. (Perez, 2014; Reinhart and Rogoff, 2011; Bolton and Jeanne, 2011). The default of a government will therefore have a significant impact on the financial system of that country (Perez, 2014). There are several manners in which riskiness of a government and an increase in the government debt influences the liquidity of banks and therefore their financial stability (Popov and Van Horen, 2013).

(13)

guarantees issued to the banks. This results in lower liquidity for banks as well, since value on the balance sheet in downgraded (Angeloni and Wolff, 2012; BIS 2011). To be exposed to the government is needed for public liquidity, since central banks in general demand a certain rate of government bonds when issuing liquidity to banks (Bolton and Jeanne, 2011). In the end a sovereign default will reduce the financial stability and wealth of a bank, which as well reduces the liquidity rate (Perez, 2014).

In the majority of the countries, banks are significantly exposed to government debt. This indicates that when a government defaults, this has an influence on the banks trough the four channels indicated earlier in this paper (BIS, 2011; Sosa Padilla, 2012). Decreasing of the assets of the bank directly relates to the liquidity of the bank, which will reduce in this case. As Popov and Van Horen (2014) noticed, European banks tend to have a significant amount of government debt securities on their balance sheets. This indicates that European banks are especially sensitive to government debt changes. Increased holdings of government debt even further increases the link between sovereign risks and banking risks (Angeloni and Wolff, 2012). Other studies found that the government bonds holdings by banks are less related, it is not the main determent of banks’ performance (Angeloni and Wolff, 2012). This indicates that there can be other ways in which government debt in transferred to the banking system. Especially regarding high debt levels, countries are now at high risk and less stable. Next to this, governments are especially responsible for the domestic real financial system (Angeloni and Wolf, 2012), and therefore should keep a close eye on their own financial statements.

When measuring the effect of government debt on the liquidity of banks, hypotheses need to be formulated. Looking at the literature discussed above, the following hypothesis can be formulated:

H1

Government debt has impact on the liquidity of individual banks

H2

The impact of government debt on the liquidity of individual banks is negative

(14)

taxes and government guarantees as well (Angeloni and Wolff, 2012; Bolton and Jeanne, 2011). These theories are less extensive, however, to control for this, an indicator for a difference-in-difference model is introduced to assure reliability and validity standards are met. This indicator is whether the bank is multinational or not and will be explained in the following paragraph.

2.2.4 Multinational banks

Multinational banks have been growing significantly, this indicates that banks are increasingly exposed to the different governments and that risk can be transferred at a faster pace (Navaretti et al., 2010). Multinational banks are exposed to other governments as well, where domestic banks are only exposed to their own domestic government. Banks that are exposed to other government can transfer liquidity to and from the other location. Especially during local crisis, multinational banks can transfer liquidity whereas domestic banks had to come up with other solutions (De Haas and Lelyveld, 2011; Navaretti et al., 2010). Whereas multinational banks are strong and can transfer liquidity during local crisis, they are exposed more heavily to global crisis, as the crisis of 2008 showed. (De Haas and Lelyveld, 2011). This indicates that multinational banks had more liquidity problems during the global crisis, where domestic banks have lower liquidity during local crisis. Moreover, multinational banks are conducive to react to changes of governments, which suggests that multinational banks are exposed to a different set of variables than their domestic counterparts (Aspach, Nier and Thiesset, 2005).

(15)

that banks that are exposed to multiple governments have less liquidity than the domestic banks, this seemingly reflects their admittance to diversified sources of funding.

To identify the difference between banks that are exposed to multiple governments and banks that are exposed to one government, it is taken into account whether a bank is multinational or not (Bofondi, Carpinelli and Sette, 2013). Since the exposure to the different governments has a difference influence on the liquidity of the banks, the difference between the exposures of governments is captured by this variable. To address the exposure of a bank to one government, the multinational bank would have a different liquidity ratio than its domestic counterpart. Therefore, the stylized fact indicates this difference in effect (Bofondi, Carpinelli and Sette, 2013; De Haas and Lelyveld, 2011):

S1

(16)

2.3 SUMMARY OF HYPOTHESES

As can be subtracted from the literature the financial stability of banks is affected by the financial stability of governments. The relationship between the stability of governments and banks is expected to be negative. This indicates that when government debt levels rise the liquidity of banks will decline (Perez, 2014). As could be seen from the figure, when government debt rises the liquidity of banks go down. The figure measures liquidity as the mean of the banks during the year, and government debt likewise the mean of the countries during that year. The data in the figure is distracted form the final dataset used during the research.

Figure 1: Liquidity means and government debt means per year

(17)

3. DATA AND

METHODOLOGY

3.1 DATA COLLECTION AND LIMITATIONS

To elaborate on the link between governments’ financial stability and the financial stability of banks, this research investigates whether government debt has influence on the risk of individual banks. To investigate the link between government debt and the liquidity of banks, 843 banks are examined over twenty-eight countries of the European Union over a time period of ten year, 1999 to 2009. The liquidity ratios of the banks originate from the Bankscope database (Bureau van Dijk, 2014). The ratio is measured in net loans divided by the total assets of the bank. The location of the banks are matched to the countries of the European Union. The government debt2 data is coming from Eurostat (2014), the government debt is measured in comparison to Gross Domestic Product (GDP). A factor that indicates whether the bank is multinational or only active domestically (Claessens and Van Horen, 2013) will account for the bank’s exposure to different governments. Below a table can be found with the sources of the data and how it is measured.

Table 1: Data collection

Data What does it measure Data source

Liquidity of individual banks

Liquidity ratio: net loans divided by total assets

Bankscope by Bureau van Dijk, 2014

Government debt Total debt hold by governments in

comparison to Gross Domestic Product

Eurostat, 2014

Multinational dummy

Whether the bank was exposed to other governments as well

Claessens and Van Horen, 2013

Impaired loans ratio of individual banks

Impaired loans ratio: impaired loans divided by total loans

Equity ratios of individual banks

Equity ratio: equity divided by total assets

(18)

The limitations of the data concern the different data sets used in the study. Starting with the data for liquidity ratios, which was available for a significant amount of banks in Europe. When introducing the variable Multinational, this variable is not available for all banks and only available till the year 2009. Next to this, the Multinational bank data set (Claessens and Van Horen, 2013) was partly available for other banks in Europe than the data set of Bankscope (Bureau van Dijk, 2014). Government debt is available for almost all countries, though it should be noted that the for Croatia, Greece, Estonia and Poland some years are missing (Eurostat, 2014). These data limitations are taken into account when performing the research.

3.2 DATA ANALYSIS

Regarding the liquidity ratio, the ratio includes the net loans divided by the total assets of the bank (Bankscope, 2014), where the variable government debt included the total debt of the government in a given year (Eurostat, 2014). To describe the data accurately summary statistics were computed. Next to this, a correlation matrix is computed to check for correlations within the data. The summary statistics and correlation matrix can be found in Appendices 1 and 2.

3.3 CONCEPTUAL MODEL

(19)

3.4 THE MODEL

As for panel data regressions, the analysis should hold for several assumptions. The first assumption that has to be checked is heteroskedasticity. To test this the Breach Pagan test is performed, for this test the null hypothesis assumes homoscedasticity, which implies that the deviations of the error term are constant and therefore are not dependent on the independent variable (Greene, 2000: 601). When testing for this, the test statistic is found to be significant, therefore the null hypothesis can be rejected (Appendix 3). A possible bias to the analysis is regarding an endogeneity problem. As this research investigates the influence of government financial stability on the financial stability of banks, other researchers have found that bank stability could also lead to government financial stability (Angeloni and Wolff, 2012; Bolton and Jeanne, 2011). To account for this possible bias a difference-in-difference model will be used, this method will be explained more in-depth later.

Since the dataset is constructed by panel data there are three possible options regarding regression models; pooled regression model, fixed effects or random effects model

(Torres-Financial Stability of a Bank Financial Stability of a Government Exposure of a Bank to more Governments Sensitivity difference to Government by Banks Measured in Liquidity

Measured in Government Debt

Measured in Multinational Dummy

(20)

Reyna, 2007). When pooling the data of the different banks in a country together, it is assumed that all banks are interacting with the other variables equally. In theory differences in banks are expected, since banks are behaving different from each other and are not all affected in the same manner by events (Gennaioli, Martin and Rossi, 2014). Since it is expected that the banks are different regarding their liquidity using the pooled regression method is estimated to be inappropriate. Not all banks have the same relation to governments, therefore the assumption that all banks have the same relationship can be dropped. The dropped assumption leads to a choice between a random or fixed effects model, since these two models allow for bank heterogeneity.

One of the most important advantages of the fixed effect model is that the model controls for differences when time is not changing between observations. The model is created to research the causes of changes with an entity, in this case within a bank (Kohler and Kreuter, 2009). When looking at the random effects model, this model assumes the variation across observations is random and uncorrelated with the independent variables. This model is recommended to use when differences across observations are assumed to have an effect on the relationship with the independent variables, a possibility is that time effects are also incorporated (Gertler et al., 2010).

To test econometrically which model will fit best, the Hausman test is performed (Hausman, 1978). When the Chi square statistic is low and the test is found to be significant a fixed effect model is preferred, whereas the value is high a random effects model is preferred. Regarding this research the null hypothesis that the random effects model can be used is rejected (Appendix 3), and therefore a fixed effects model will be used to test for the hypotheses.

(21)

The standard regression includes the influence of government debt on the liquidity of individual banks. This is done, because the research investigates the link between government debt and the liquidity of banks. First an estimation is made with only those two variables in order to capture the change in coefficients.

(1)

𝑙𝑖𝑞_{𝑖,𝑐,𝑡}= 𝛽₁𝑔𝑜𝑣𝐷_𝑐,𝑡 + 𝑓𝑒_{𝑖,𝑐,𝑡}+ 𝜀

Where liq is the liquidity of an individual bank i in country c in a year t. govD is the government debt in percentage of GDP in country c in year t. The fixed effects are captured by the fixed effects coefficient (fe).

The research focuses on the influence of government debt on the liquidity of individual banks. In the literature this is the main relationship, however, there is a possibility that the liquidity ratios of individual banks have an influence on the government debt. To deal with this endogeneity problem, a difference-in-difference model is used (Bertrand, Duflo and Mullainathan, 2004). A difference-in-difference model compares two groups, before and after the treatment. The difference in the banks is whether a bank is multinational or not, this difference indicates whether the bank is influenced by multiple government debts. The difference-in-difference model can be combined with fixed effects, since there are more than two time periods investigated (Menaldo, 2011). There needs to be caution with the difference-in-difference model, since one needs to be confident that there are no other factors that affect the difference in trends between domestic and multinational banks, in that case the estimation is biased (Gertler et al., 2010).

(22)

By making use of a difference-in-difference model any standard differences in the liquidity of individual banks between being multinational and not being multinational before exposure to the government debt will be eliminated. The consequence is that any difference in the mean level of the liquidity of individual banks between being multinational or not, after the exposure to government debt can be correctly attributed to the government debt (Menaldo, 2011). This method disregards the likelihood that an unobserved factor is correlating with the units in government debt, and is also correlating with a higher level of the liquidity of individual banks ex ante, and is advancing any difference in the average outcome between the banks being multinationals and the banks that are domestic are observed after the exposure to government debt. The difference-in-difference model used, disregards the problem that unobserved factors might be driving correlations between the government debt and the liquidity of individual banks, therefore it isolates the treatment effect (Menaldo, 2011).

(2)

𝑙𝑖𝑞𝑖,𝑐,𝑡 = 𝛽1𝑔𝑜𝑣𝐷𝑐,𝑡 + 𝛽2𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖,𝑡+ 𝛽3𝑔𝑜𝑣𝐷𝑐,𝑡× 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖,𝑡 + 𝑓𝑒𝑖,𝑐,𝑡+ 𝜀 Where Multinational is whether the individual bank i is multinational in year t. And the variable govD x Multinational is a sensitivity term for bank i in year t. The difference within government debt when bank i is multinational at time t is interpreted as the difference in sensitivity (Ashenfelter and Card, 1985) for a multinational bank versus a domestic bank to government debt regarding its’ liquidity. This difference in sensitivity of banks is measured as: (2.1) 𝑑 𝑙𝑖𝑞_{𝑖,𝑐,𝑡} 𝑑 𝑔𝑜𝑣𝐷_𝑐,𝑡 = 𝛽1 𝑔𝑜𝑣𝐷𝑐,𝑡+ 𝛽3 𝑔𝑜𝑣𝐷𝑐,𝑡 × 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖,𝑡 { 1 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙 0 𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐

(23)

(3)

𝑙𝑖𝑞_{𝑖,𝑐,𝑡}= 𝛽₁𝐺𝑜𝑣𝐷_𝑐,𝑡 + 𝛽₂𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑖,𝑡+ 𝛽₃𝑔𝑜𝑣𝐷_𝑐,𝑡× 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑖,𝑡 + 𝛽₄𝐶𝑐𝑜𝑑𝑒_𝑐 + 𝛽₅𝑇𝑐𝑜𝑑𝑒_𝑡+ 𝑓𝑒_{𝑖,𝑐,𝑡}+ 𝜀

This could be analyzed in a different manner as well. To control for country and time fixed effects, as well as the correlation between the two, dummies are tabulated and a dummy called Dummy Country Year (DCY) is created. This dummy is made for every country within every year, as a two way fixed effects. Therefore, it takes out all means of countries for every years. Next to this, the fixed effect is added to control for the other fixed effects (fe), including bank fixed effects as well as a correlation between bank and year effects. The final regression includes all fixed effects for country, year and banks.

(4)

𝑙𝑖𝑞𝑖,𝑐,𝑡= 𝛽1𝐺𝑜𝑣𝐷𝑐,𝑡 + 𝛽2𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖,𝑡+ 𝛽3𝑔𝑜𝑣𝐷𝑐,𝑡× 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖,𝑡 + 𝛽4𝐷𝐶𝑌𝑐,𝑡 + 𝑓𝑒𝑖,𝑐,𝑡+ 𝜀

(24)

4. EMPIRICAL RESULTS

4.1 RESULTS

A fixed effects model is used to test for the influence of government debt on liquidity of the individual banks. When using regression 1, de baseline model, the results suggest that government debt has a negative influence on the liquidity of individual banks, this result is significant on all levels. This indicates that for a given bank, as debt varies across time by one percentage point, the liquidity ratio of the bank will decrease by 0.12 percentage point.

When continuing with the difference-in-difference model, a multinational dummy (Multinational) is added to the regression as well as an interaction term. The regression indicates that government debt has a significant and negative impact on the liquidity of individual banks. So, for a given bank, as debt varies across time by one percentage point, the liquidity ratio of the bank will decrease by 0.09 percentage point in model 2, the difference-in-difference model. This illustrates that when a bank is multinational the bank has more liquidity than a domestic counterpart. Next to this, multinationals are less sensitive regarding their liquidity for domestic government debt.

When looking at the regression with the variables Ccode and Tcode, the signs and significance do not change significantly. The liquidity ratio of a bank will decrease by 0.10 percent point, when government rises with 1 percent point. When performing the regression the Ccode is omitted because of collinearity reasons. This indicates that the variables incorporate information that is almost the same as another variable, which means that the variable can be safely dropped. Even though the variable might be important for the research, the variable is comparable to another variable included in the regression. The country fixed effects emittance can be explained by the fixed effects of banks incorporating the influence of countries on the particular bank, as well as the variable government debt since this variable already accounts partly for the differences between countries.

(25)

dummies in the previous regression. The effect of government debt on a given bank is, as debt varies across time by one percentage point, that the liquidity ratio of the bank will decrease by 0.34 percentage point. This indicates that the fixed effects are controlling for other influences that are more positive and were first captured by the variable government debt. When controlling for other effects, the effect of government debt on liquidity of individual banks is still significant. An increase in the sensitivity of multinationals to government debt, indicates that the multinational is slightly less sensitive to government debt than their domestic counterparts, although the signs stay the same as in the last regression.

In the following graph the results can be seen, a more extensive table can be found in Appendix 4. First the baseline regression is shown, where after the difference in difference indicator is added. After this a country and time dummy is added, where the country dummy is omitted due to collinearity. Lastly, the interaction dummy of country and year (DCY) is replacing the country and time dummy. When looking at the different outcomes of the models all relationships between government debt and the liquidity ratios of individual banks are negative and significant.

Table 2: Results regressions

Liquidity baseline diff in diff Ccode+Tcode DCY

Government Debt (0.117) (0.093) (0.101) (0.336) 0.000* 0.000* 0.000* 0.000* Multinational 9.449 8.405 3.635 0.000* 0.000* 0.032* GovDMultinational (0.064) (0.093) (0.069) 0.015 0.000* 0.013

Notes Including bank

fixed effects

Inlcuding bank fixed effects

Ccodes omitted Including bank and time fixed effects

Including bank, time and country fixed effects

(26)

4.2 DISCUSSION

The discussion of the results will be linked to answering the main research question: Looking

at the debt ratios of domestic governments and the liquidity coverage ratios of banks, does the financial stability of a country affect the financial stability of an individual bank?

To measure financial stability of the government the variable government debt is used, and to measure the financial stability of an individual bank the variable liquidity is used. This can be seen in the conceptual model. To account for endogeneity problems the dummy variable

Multinational is added in the difference-in-difference model. When looking at the results and

keeping track of the stylized fact, the most important beta for the research is the one that captures the relation between of government debt on the liquidity ratio of the individual bank.

When looking at the four regressions performed the models have in common that they indicate that the relationship is significant. Which confirms the first hypothesis stated:

Government debt has impact on the liquidity of individual banks. This relates to the literature

in the way that government debt relates to the liquidity of individual banks. Looking at the stability of the countries and banks, it is stated that the government stability does have a significant influence regarding their debt policy on the stability of banks in the country. This is corresponding with the literature regarding the link between banks and the government (BIS, 2011; Perez, 2014). The significant link between government debt and liquidity ratios of individual banks indicate that when a government defaults, this has consequences for banks trough the four channels explained earlier in this paper (BIS, 2011; Sosa Padilla, 2012). Decreasing of the assets of the bank directly relates to the liquidity of the bank, which will reduce in that case. This leads to the second hypothesis: The impact of government debt on the

liquidity of individual banks is negative. This hypothesis can be confirmed, since the sign is in

(27)

states that it only influences the liquidity with 0.34 when government debt rises by 1 percentage point.

To answer the main research question: Looking at the debt ratios of domestic governments

and the liquidity coverage ratios of banks, does the financial stability of a country affect the financial stability of an individual bank? Yes, the financial stability of a government

influences the financial stability of an individual bank. As illustrated by the literature the last crisis highlighted the link between governments’ and banks’ financial stability (Gennaioli, Martin and Rossi, 2013). The research performed found new evidence for this link by measuring the government debt and liquidity ratios of banks. The findings add to the growing literature on the link between governments and banks in Europe. The findings also imply that the fears that existed concerning the risk that the banking sector was facing when governments were on the point to default (Black et al., 2013) were not without reason. To verify the results against policy actions taken by the Basel Committee on Banking Supervision, it is clear that the committee pays more attention to the liquidity setting of a bank (Basel Committee on Banking Supervision, 2013), since this is truly influenced by government debt.

4.3 ROBUSTNESS OF RESULTS

To test for robustness different methods are used. In short, the government debt variable will be lagged since this variable influences liquidity over time, the dependent variable liquidity will be replaced by other potential variables that indicate bank financial stability. Lastly, a few countries in Europe were more heavily affected by the crisis of 2008 than others, the most affected countries are taken out of the regression.

(28)

(5)

𝑙𝑖𝑞_{𝑖,𝑐,𝑡} = 𝛽₁𝐿𝐺𝑜𝑣𝐷_𝑐,𝑡 + 𝛽₂𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑖,𝑡+ 𝛽₃𝐿𝑔𝑜𝑣𝐷_𝑐,𝑡× 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑖,𝑡 + 𝛽₄𝐷𝐶𝑌_𝑐,𝑡 + 𝑓𝑒_{𝑖,𝑐,𝑡}+ 𝜀

When performing the model (5), the result of the regression does not change significantly. The signs and significance of the main indicator stay the same. Which means that lagging the government debt variable has almost the same results. This indicates that the government debt does also have impact in the long run on liquidity of individual banks.

The IMF (2014) stated the financial soundness indicators, which include next to liquidity ratios, also equity ratios, nonperforming loans and others. These other variables are also measuring the financial stability of an individual bank. When looking at the existing literature next to liquidity ratios, loan ratios are common used indicators of financial stability of banks (Basel Committee on Banking Supervision, 2013; Holmstrom and Tirole, 1998; IMF, 2014). As ratios and nonperforming loan ratios are both short term ratios measuring financial stability of a bank, this will be captured in the following robustness check. The difference between impaired loans and nonperforming loans is that impaired loans is an accounting technique that considers cases in which there is a chance that the creditor is not able to repay the loan. While nonperforming loans is a regulatory approach, which entails that the loans that are 90 days past due (Klein, 2013). Recognizing this difference, the impaired loans variable is seen as nonperforming loans in this robustness check.

(6)

𝐼𝑚𝑝𝑎𝑖𝑟𝑒𝑑𝐿𝑜𝑎𝑛𝑠_{𝑖,𝑐,𝑡}

= 𝛽₁𝐺𝑜𝑣𝐷_𝑐,𝑡 + 𝛽₂𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑖,𝑡+ 𝛽₃𝑔𝑜𝑣𝐷_𝑐,𝑡× 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑖,𝑡 + 𝛽₄𝐷𝐶𝑌_𝑐,𝑡 + 𝑓𝑒_{𝑖,𝑐,𝑡}+ 𝜀

(29)

Solvency and liquidity are two different factors that measure financial stability. While solvency is more long term focused, liquidity is concentrated on the short term (Pierret, 2014). Those two are both important, therefore the robustness check is also done by an making use of an indicator of solvency, in this case an equity to assets ratio.

(7)

𝐸𝑞𝑢𝑖𝑡𝑦_𝐴𝑠𝑠𝑒𝑡𝑠𝑖,𝑐,𝑡

= 𝛽₁𝐺𝑜𝑣𝐷_𝑐,𝑡 + 𝛽₂𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑖,𝑡+ 𝛽₃𝑔𝑜𝑣𝐷_𝑐,𝑡× 𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑖,𝑡 + 𝛽4𝐷𝐶𝑌𝑐,𝑡 + 𝑓𝑒𝑖,𝑐,𝑡+ 𝜀

When the liquidity ratio is alternated by a long term indicator of financial stability for a bank, as an equity to asset ratio is, the relationship between government debt and the dependent variable becomes positive and significant. This indicates that the effect of higher government debt on the long run financial stability of a bank is positive. The explanation for this can be found in the fact that banks may not be liquid but still can be solvent in the long run. This implies that banks might not be able to stand sudden shocks in the environment, but are able to survive in the long run (Pierret, 2014). Next to this reason, the positive effect can be explained by countercyclical behavior of a bank. When government debt rises, banks can feel the need to keep a close eye on their financial stability position (Padilla, 2012) and increase their equity ratios (Pierret, 2014).

The third kind of robustness check is done by omitting the PIIGS countries: Portugal, Ireland, Italy, Greece and Spain. A regression is performed without these five countries. The reason to omit the countries; Portugal, Ireland, Italy, Greece and Spain is that these countries have proven to be less financial stable during the last crisis. Especially those countries’ debt burdens were growing tremendously (Schmidt, 2014). The debt crisis started in Ireland, spreading to Spain, Portugal, Italy and especially affecting Greece (Teague, 2013). These five countries were most effected by the crises and therefore are omitted as a robustness check.

(30)

Table 3: Results regressions robustness Standard

model

Lag debt Dependent

variable: Equity to Assets Dependent variable: Impaired loans to total loans PIIGS Government Debt (0.336) (0.310) 0.198 (0.044) (0.338) 0.000* 0.000* 0.000* 0.850 0.000* Multinational 3.635 2.880 3.312 0.348 3.417 0.032 0.125 0.001* 0.874 0.053 GovDMult (0.069) (0.063) (0.026) (0.019) (0.059) 0.013 0.042 0.122 0.564 0.067

(31)

5. CONCLUSION

5.1 GENERAL CONCLUSION

To conclude, the research question will be answered and the road to answering the research question will be discussed. The crises of 2008 highlighted the importance of the link between governments and banks. To further research the financial stability link between governments and banks, a financial stability indicator is specified for both entities. To measure the financial stability of governments the government debt is used as variable. To measure the financial stability of banks the liquidity ratio is used. Form this relation the research question can be formulated: Looking at the debt ratios of domestic governments and the liquidity coverage

ratios of banks, does the financial stability of a country affect the financial stability of an individual bank?

To answer the research question a fixed effect model is used. To get rid of potential endogeneity problems a difference-in-difference model is added. The variable used for the difference-in-difference model is a dummy which indicates whether a bank is multinational or not. For this variable a stylized fact is important: Being multinational or domestic is a characteristic of a banks which makes the bank different sensitive to government debt. To the difference-in-difference fixed effect model other fixed effects are added, such as country and year fixed effects. The fixed effects are added by making use of matrices or one-way fixed effects. This is done to test the hypotheses: Government debt has impact on the liquidity of

individual banks and The impact of government debt on the liquidity of individual banks is negative. The results of the regressed models indicate that government debt has a significant

(32)

were most heavily exposed to the crises of 2008. Omitting these countries did not significantly change the results.

To sum up, as in line with the literature, financial stability of governments does have an effect on the financial stability of banks. Measured in the government debt and liquidity ratios of individual banks this relationship is significant and negative. An overall conclusion of the hypotheses tested can be found in the following table.

Table 4: Conclusion hypotheses

Hypothesis Confirmation Robustness

H1 Government debt has impact on the liquidity of individual banks

Significance Confirmed

All robustness checks confirm, except when dependent variable is the

impaired loans ratio

H2 The impact of government debt on the liquidity of individual banks is negative

Relationship Confirmed

All robustness checks confirm, except when dependent variable is the equity ratio

5.2 IMPLICATIONS

The research contributed to the growing literature on the link between governments’ financial stability and the financial stability of banks (Perez, 2014). New links between governments and banks still need to be investigated. This research implies that the link between government debt and liquidity of banks is significant related and when performing robustness checks with other variables the link was still existent. When drawing implications from the research it can be found that governments need to be aware of their debt levels since this has a significant impact on the financial stability of banks. New regulations are already in place for banks itself (Basel Committee on Banking Supervision, 2013), however, it should be noted that governments have an impact on these ratios as well. One of the main lessons learned from the research is that governments have a role in the stability of banks, and therefore, need to watch their own debt levels.

5.3 LIMITATIONS AND FUTURE RESEARCH

(33)

within Europe. Due to data limitations the crisis of 2008 could not be taken into account, although this would be the most interesting period to study, since this is the time when a lot of countries were facing financial stability issues (Bofonfi, Carpinelli and Sette, 2013). Therefore, future research should focus on the impact of financial instability of governments on the financial stability of banks taking into account the crisis of the last years.

Regarding the literature government bond holdings state whether a bank is more or less exposed to the developments in government debt (BIS, 2011; Popov and Van Horen, 2013). When focussing on the financial stability of banks this is less of an issue, however, it is important to notice to what extent which bank is affected on the balance sheet. An indirect effect could take place via the bond holdings, therefore, other research has to be done to find out whether there is an indirect effect taking place. This effect can also take place via other indicators, research has to show whether this is the case and to what extent.

Furthermore, other indicators can be used for measuring financial stability of governments as well as for banks. The variables used in the paper and regressions are not the only indicators for financial stability, therefore other variables could be used as well. Other variables to check for financial stability can include: bail out policies, credit risk measures, balance sheet indicators etc. (Israël, 2013; IMF, 2014). To further investigate other indicators for financial stability new research has to be performed.

(34)

SOURCES

Allen, W.A. and Wood, G. (2006). Defining and achieving financial stability. Journal of

Financial Stability, 2 (2006), p. 152–172.

Angeloni, C., and Wolff, G. (2012) Are Banks Affected by Their Holdings of Government Debt? Bruegel, Working Paper, 07.

Ashenfelter, O. and Card, D. (1985) Using the Longitudinal Structure of Earnings to Estimate the Effects of Training Programs. Review of Economics and Statistics, 67, p. 648–660.

Aspachs, O., Nier, E., and Tiesset, M. (2005) Liquidity, Banking Regulation and the Macroeconomy. Mimeo (London: London School of Economics).

Bank for International Settlement (BIS) (2011) The impact of sovereign credit risk on bank funding conditions. Committee of Global Financial System, Papers, 43.

Basel Committee on Banking Supervision (2013) Basel III: The Liquidity Coverage Ratio and liquidity risk monitoring tools. Bank for International Settlements. p. 1-75. Available from: http://www.bis.org/publ/bcbs238.pdf. [Accessed: 10/2014]

Black, L. et al. (2013) The Systemic Risk of European Banks during the Financial and Sovereign Debt Crises. Board of Governors of the Federal Reserve System, International Finance Discussion Papers, 1083. Available from:

http://www.federalreserve.gov/pubs/ifdp/2013/1083/ifdp1083.pdf. [Accessed: 12/2014]

Blundell-Wignall, A. and Slovik, P. (2010) The EU Stress Test and Sovereign Debt Exposures. OECD Financial Affairs Division, OECD Working Papers on Finance, 4. Available from: www.oecd.org/daf/fin. [Accessed: 11/2014]

Bofondi, M., Carpinelli, L. and Sette, E. (2013) Credit supply during a sovereign debt crisis.

(35)

http://www.bancaditalia.it/pubblicazioni/econo/temidi/td13/td909_13/en_td909/en_tema_909. pdf. [Accessed: 11/2014]

Bolton, P. and Jeanne, O. (2011) Sovereign Default Risk and Bank Fragility in Financially Integrated Economies. IMF, Economic Review, 59, p.162-194. Available from:

http://www.nber.org/papers/w16899.pdf. [Accessed: 12/2014]

Borensztein, E. and Panizza, U. (2008) The costs of sovereign default. IMF, Working Papers.

Brooke, M. et all. (2013) Sovereign default and state-contingent Debt. Bank of England, Financial Stability Paper, 27, p. 1-20. Available from:

http://www.bankofengland.co.uk/research/Documents/fspapers/fs_paper27.pdf. [Accessed: 12/2014]

Bureau van Dijk (2014) Liquidity ratio 34 – Net Loans/ Total Assets. Bankscope. Available from: Library of the University of Groningen. [Accessed: 12/2014]

Claessens, S. and Van Horen. N. (2013) Foreign banks: Trends and impact. Journal of Money,

Credit and Banking (forthcoming). Database available from:

http://www.dnb.nl/en/onderzoek-2/databases/bank.jsp. [Accessed: 12/2014]

Committee on the Global Financial System (CGFS) (2011) The impact of sovereign credit risk on bank funding conditions. CGFS, Papers, 43, p. 1-52. Available from:

http://www.bis.org/publ/cgfs43.pdf. [Accessed: 12/2014]

De Haas, R and Van Lelyveld, I. (2011) Multinational Banks and the Global Financial Crisis: Weathering the Perfect Storm. European Bank for Reconstruction and Development. Working Paper, 135.

Deléchat, C., et al., (2012) The Determinants of Banks' Liquidity Buffers in Central America.

IMF, Working paper, 12 (301), p. 1-43. Available from:

(36)

Diacon, P. E., Donici, G. A. and Maha, L. G. (2013) From the Global Crisis to the National Crises: The Case of the European Union Countries. Procedia Economics and Finance, 6, p 41-47.

Diamond, D. W. and Dybvig, P. H. (1983) Bank Runs, Deposit Insurance, and Liquidity.

Journal of Political Economy, 91 (3), p. 401–19.

Diamond, D. W. and Rajan, R. G. (2001) Liquidity Risk, Liquidity Creation, and Financial Fragility: A Theory of Banking. Journal of Political Economy, University of Chicago Press, 109 (2), p. 287–327.

European Central Bank (ECB) (2010) Lessons for financial stability and monetary policy.

ECB. Available from:

http://www.ecb.europa.eu/pub/pdf/other/greatfinancialcrisisecbcolloquiumpapademos201203 en.pdf. [Accessed: 12/2014]

European Central Bank (ECB) (2013) Financial Integration in Europe. ECB, p. 1-158.

European Central Bank (ECB) (2014) Financial Stability Review. ECB, p. 1-109. Available from:

http://www.ecb.europa.eu/pub/pdf/other/financialstabilityreview201405en.pdf??b9ad281efe7 751025dd1bb9b6673e07c. [Accessed: 12/2014]

Eurostat (2014) Government deficit/surplus, debt and associated data. Database European

Commission. Available from:

http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=gov_10dd_edpt1&lang=en. [Accessed: 12/2014]

Gadanecz, B. and Jayaram, K. (2009) Measures of financial stability – a review. IFC Bulletin, 31, p. 365-380. Available from: http://www.bis.org/ifc/publ/ifcb31ab.pdf. [Accessed:

(37)

Gennaioli, N., Martin, A. and Rossi, S. (2014) Banks, Government Bonds, and Default: What do the Data Say? IMF, Working Paper, 14 (120). Available from:

https://www.imf.org/external/pubs/ft/wp/2014/wp14120.pdf. [Accessed: 11/2014]

Gertler, P. J., et al., (2010) Impact Evaluation in Practice: Ancillary Material. The World

Bank. Available from:

http://webcache.googleusercontent.com/search?q=cache:Zhgd59JXozQJ:siteresources.worldb

ank.org/EXTHDOFFICE/Resources/5485726-1295455628620/21.Technical_Dif_in_Dif_Premand_Holla_ENG_PP.pptx+&cd=5&hl=en&ct =clnk&gl=nl. [Accessed: 01/2015]

Gray, S. (2011) Central Bank Balances and Reserve Requirements. IMF, Working Paper, 11 (36).

Greene, W. (2000) Econometric Analysis. Upper Saddle River: NJ, Prentice-Hall.

Hausman, J.A. (1978) Specification tests in econometrics. Econometrica, 46 (6), p. 1251-1271.

Higgins, B. and Merfeld, T. J. (1987) Debt, Financial Stability And Public Policy. Federal

Reserve Bank of Kansas City, Economic review, p. 1-12. Available from:

https://www.kansascityfed.org/publicat/econrev/EconRevArchive/1987/1q87higg.pdf. [Accessed: 12/2014]

Holmstrom, B. and Tirole, J. (1998) Private and Public Supply of Liquidity. Journal of

Political Economy, 106, p.1-40. Available from:

http://isites.harvard.edu/fs/docs/icb.topic1002789.files/Week%209%20-

%20October%2025%20and%2027/Holmstrom-Private%20and%20public%20supply%20of%20liquidity.pdf. [Accessed: 12/2014]

International Monetary Fund (IMF) (2012) Sovereigns, Banks, and emerging markets: Detailed analysis and policies. IMF, Chapter 2. Available from:

(38)

International Monetary Fund (IMF) (2014) Financial Soundness Indicators (FSIs)—Concepts and Definitions. IMF. Available from:

http://fsi.imf.org/misc/FSI%20Concepts%20and%20Definitions.pdf. [Accessed: 01/2015]

Israël, J. M. et al. (2013) Statistics and Indicators for Financial Stability Analysis and Policy.

ECB Occasional paper series, 145, p. 1-36. Available from:

http://www.ecb.europa.eu/pub/pdf/scpops/ecbocp145.pdf. [Accessed: 01/2015]

Klein, N. (2013) Non-Performing Loans in CESEE: Determinants and Impact on Macroeconomic Performance. IMF, Working Paper , 13 (72), p. 1-26. Available from: http://www.imf.org/external/pubs/ft/wp/2013/wp1372.pdf. [Accessed: 01/2015]

Kohler, U and Kreuter, F. (2009) Data Analysis Using Stata. State Press, 2.

Kumhof, M. and Tanner, E. (2005) Government Debt: A Key Role in Financial Intermediation. IMF, Working Paper, 05 (57), p. 1-30. Available from:

http://www.imf.org/external/pubs/ft/wp/2005/wp0557.pdf. [Accessed: 12/2014]

Liang, N. (2013) Systemic Risk Monitoring and Financial Stability. Journal of Money, Credit

and Banking, 45 (1), p. 1-8.

Marianne Bertrand, M., Duflo, E. and Mullainathan, S. (2003) How much should we trust difference-in-difference estimates? p.1-32. Available from:

http://economics.mit.edu/files/750. [Accessed: 01/2015]

Menaldo, V. (2011) What is endogeneity bias and how can we address it? University of

Washington, Political Science, p. 1-4. Available from:

http://faculty.washington.edu/vmenaldo/Endogeneity%20Bias.pdf. [Accessed: 12/2014]

(39)

Noyer, C. (2012) Central banking in a context of high public debt. Banque de France. Financial Stability Review, 16, p. 1-252. Available from:

https://www.banque-france.fr/uploads/tx_bdfgrandesdates/FSR16-20-04.pdf. [Accessed: 11/2014]

Olivares-Caminal, R. (2011) The EU architecture to avert a sovereign debt crisis. OECD. Available from: http://www.oecd.org/finance/financial-markets/49191980.pdf. [Accessed: 11/2014]

Perez (2014) Sovereign debt, domestic banks and the provision of public liquidity? Job

Market Paper, Stanford University, p. 1-74. Available from:

http://stanford.edu/~dperez1/Perez_JMP.pdf. [Accessed: 12/2014]

Pierret, D. (2014) Systemic risk and the solvency-liquidity nexus of banks. p. 1-51. Available from: https://www.stern.nyu.edu/sites/default/files/assets/documents/con_047616.pdf.

[Accessed: 01/2015]

Popov and Van Horen (2014) Exporting Sovereign Stress: Evidence from Syndicated Bank Lending during the Euro Area Sovereign Debt Crisis. Review of Finance (forthcoming), p. 1-54. Available from:

http://www.dnb.nl/en/binaries/Exposure_to_sovereign_debt_and_syndicated_lending_RoF_w _tcm47-313313.pdf . [Accessed: 11/2014]

Reinhart, C. M. and Rogoff, K. S. (2011) From financial crash to debt crisis. American

Association, American Economic Review, 101 (5), p. 1676-1706. Available from:

http://www.nber.org/papers/w15795. [Accessed: 11/2014]

Rochet, J. C. and Vives, X. (2004) Coordination Failures and the Lender of Last Resort: Was Bagehot Right After All?. Journal of the European Economic Association, 2 (6), p. 1116– 1147.

(40)

Shirakawa, M. (2012) Sustainability of government debt: preconditions for stability in the financial system and prices. Banque de France, Financial Stability Review, 16, p.169-181. Available from:

https://www.banque-france.fr/fileadmin/user_upload/banque_de_france/publications/Revue_de_la_stabilite_financ iere/2012/rsf-avril-2012/FSR16-article-16.pdf. [Accessed: 01/2015]

Solomos D. K., Koumparoulis D. N. (2012) The Recent Reform of European Governance: A Critical Review. Available from:

http://download.springer.com/static/pdf/335/art%253A10.1007%252Fs10272-012-0431-x.pdf?auth66=1411810990_b5dd3246414b5bb51638608e9ce111dc&ext=.pdf . [Accessed: 11/2014]

Sosa Padilla, C. (2012) Sovereign Defaults and Banking Crises. Manuscript, McMaster

University. Available from:

http://mpra.ub.uni-muenchen.de/41074/1/MPRA_paper_41074.pdf. [Accessed: 11/2014]

Tarullo, D. K. (2014) Liquidity Regulation. Financial Stability Board, p. 1-21. Available from: http://www.financialstabilityboard.org/wp-content/uploads/Tarullo-Liquidity-Regulation.pdf. [Accessed: 12/2014]

Teague, M. (2013) The Role of the "PIIGS" in the European Sovereign Debt Crisis. Equities. Available from: http://www.equities.com/editors-desk/international-investing/europe/the-role-of-the-piigs-in-the-european-sovereign-debt-crisis. [Accessed: 12/2014]

Torres-Reyna, O. (2007) Panel Data Analysis Fixed and Random Effects using Stata (v. 4.2).

Princeton University. Available from: http://www.princeton.edu/~otorres/Panel101.pdf.

(41)

APPENDICES

Appendix 1: Summary statistics

Variable Observations Mean Std Dev Min Max

Liquidity 32352 55.92462 23.20254 (1.861) 100 Government Debt 56564 62.63509 22.26116 6.1 126.8 Multinational 9904 .3178514 .4656649 0 1 GovDMultinational 9520 12.95209 24.32793 0 126.8 Year 57376 2004 3.162305 1999 2009 Country 57376 13.9326 7.496005 1 28 Bank 57376 2608.978 1505.976 1 5218 Equity to Assets 33847 11.55631 17.34991 (749.655) 138.017

Impaired loans to loans 5498 5.708081 7.549202 (.541) 247.997

Lagged GovD 51409 61.6533 22.00614 6.1 114.7

Appendix 2: Correlation matrix

Liquidity Government Debt Multinational GovD Multinational Lagged GovD Equity to Assets Impaired loans to loans Liquidity 1.0000 Government Debt 0.0735 1.0000 Multinational (0.1087) (0.2543) 1.0000 GovDMultinational (0.0944) 0.0598 0.8311 1.0000 Lagged GovD 0.0873 0.9800 (0.2780) 0.0314 1.0000 Equity to Assets 0.1132 (0.2264) (0.0040) (0.0555) (0.2040) 1.0000 Impaired loans to loans (0.0457) 0.0321 0.0437 0.0533 0.0145 0,1001 1.0000

Appendix 3: Breusch-Pagan and Hausman test

Test Breusch-Pagan Hausman

Nul hypothesis Constant variance Difference in coefficients not systematic

Chi2 60.16 146.37

Prob>Chi2 0.0000* 0.0000*

Result No constant variance Difference in coefficients is systematic: use

(42)

Appendix 4: Results regression standard

In Appendix 4 the baseline (4.1), difference-in-difference (4.2) regressions can be found. As well as the regressions with the Ccode and Tcode dummies (4.3) and the one with the matrix dummy DCY (4.4)

Appendix 4.1: Results regression standard baseline

Liquidity Coef. Std. Err. t P>|t| [95% Conf. Interval]

GovD -.1167149 .0076787 -15.20 0.000 -.1317656 -.1016642 _cons 6.316.833 .4822284 130.99 0.000 6.222.314 6.411.352 sigma_u 2.448.382 sigma_e 74.593.204

rho .91506403 (fraction of variance due to u_i)

F test that all

u_i=0 F(4456, 27614) = 61.64 Prob > F = 0.0000

Appendix 4.2: Results regression standard diff-in-diff

GovD -.0929574 .0174661 -5.32 0.000 -.1271985 -.0587164 Multinational 9.448.773 1.593.965 5.93 0.000 6.323918 12.57363 GovDMultinational -.0640259 .0262383 -2.44 0.015 -.1154641 -.0125876 _cons 5.672.548 1.018.515 55.69 0.000 54.72875 58.7222 sigma_u 25.330.638 sigma_e 82.376.798

F test that all u_i=0

F(842, 5106) =

(43)

Appendix 4.3: Results regression standard Ccode + Tcode

(44)

_cons 5.565.315 1.213.686 45.85 0.000 5.327.381 580.325

sigma_u 24.976.829

sigma_e 80.802.354

F test that all u_i=0 F(842, 5096) = 50.02 Prob > F = 0.0000

Appendix 4.4: Results regression standard DCY

(45)

(46)

(47)

(48)

(49)

(50)

(51)

_cons 7.265.202 3.468.827 20.94 0.000 6.585.155 7.945.249

sigma_u 28.561.246

sigma_e 76.620.564

F test that all

(52)

Appendix 5: Results regression with lagged government debt

Liquidity baseline diff in diff

Ccode + Tcode DCY Debt (.2047553) (.1826998) (.1159513) (.3097857) 0.000 0.000 0.000 0.000 Multinational 8.103741 7.995387 2.879055 0.000 0.000 0.125 DM (.0753697) (.1044962) (.0628039) 0.011 0.000 0.042

country codes omitted

Appendix 6: Results regression with dependent variable Impaired loans to total loans

Impaired Loans to Loans baseline diff in diff

Ccode + Tcode DCY Debt .1417306 .1359597 .0754276 (.0440629) 0.000 0.000 0.006 0.850 Multinational (1.910199) (2.292131) .3481302 0.286 0.219 0.874 DM .0249183 .0344536 (.0192412) 0.403 0.249 0.564

country codes omitted Appendix 7: Results regression with dependent variable Equity to Assets

Equity to Assets baseline diff in diff

Ccode + Tcode DCY Debt .0493186 .0635054 .0824457 .1977658 0.000 0.000 0.000 0.000 Multinational (.6560833) (.273751) 3.311726 0.480 0.769 0.001 DM .0135111 .0147002 (.0257981) 0.375 0.336 0.122

country codes omitted Appendix 8: Results regression without the PIIGS countries