The good and the bad of public debt
The influence of sovereign credit risk on
the profitability of banks in the eurozone
The good and the bad of public debt
The influence of sovereign credit risk on
the profitability of banks in the eurozone
Sjoerd Peter den Daas
1Abstract
This paper analyzes the influence of public debt levels and domestic interest rates on the profitability of eurozone banks over the period 1999-2011. Results imply that bank profitability is positively and significantly affected by local public debt levels. Interest rates on 10-year bonds are significantly negatively affecting profitability. During the financial crisis, the effects for the two variables become stronger, indicating that sovereign credit risk is gaining importance, confirming a diverging pattern within the eurozone. Sovereign interest rates prove to have more explanatory power than domestic public debt levels. The results indicate that country of operation matters for bank profitability, and that banks in crisis years tend to benefit from local debt. This study finds that this particularly is the case for safe haven countries, such as Germany and Austria.
Keywords: public debt level, interest rate, yield, sovereign credit risk, profitability, bank performance, eurozone, financial crisis
Acknowledgements
“I hear and I forget. I see and I remember. I do, and I understand…” – Confucius (551 BC-479 BC)
In this preface, I would like to thank all the people from the University of Groningen and Uppsala University for the inspiring time I have had, in the broadest sense of the word. In particular, I would like to thank my first supervisor during my Master’s thesis, Dr. Henk von Eije. Without his support and feedback, I would never have been able to provide you with the last chapter of my career as a student at the University of Groningen, my Master’s thesis. Additionally, I want to thank Dr. Wim Westerman, for challenging me to take the best out of my period in the International Financial Management Program.
Also, I am truly thankful to my managers at the RTL Z news desk, Perry Feenstra and Jeroen Windt, for the opportunities they gave me. They facilitated their time and resources for the sometimes difficult combination of my internship, which from February on has become my job, and my Master’s thesis. And although this combination has sometimes led to some frictions, I am very happy to already have had the opportunity to exchange my views with many people from the working field. One of the highlights during the early stage writing this thesis was the interview that I had for RTL Z with the former president of the Dutch Central Bank, Dr. Arnout Wellink. This interview inspired and motivated me to continue to work out the ideas that I already had developed when building up my research framework.
2 Table of contents
1. Introduction 2
2. Literature review and hypotheses development 6
2.1 Thirteen years of the eurozone: a risk-free market? 6
2.2 The effects of public debt on economic growth 10
2.3 The influence of public debt levels on the profitability of banks 12
2.4 The influence of sovereign interest rates on the profitability of banks 14
2.5 The influence of public debt and sovereign interest rates during the financial crisis 15
3. Methodology 18
3.1 Equation 18
3.2 Dependent variable 18
3.3 Independent variables 19
3.4 Regression and panel data techniques 23
4. Data and descriptive statistics 24
4.1 Data 24
4.2 Descriptive statistics 25
5. Results and findings 27
5.1 The influence of public debt on the profitability of banks 27
5.2 The influence of sovereign interest rates on the profitability of banks 29
5.3 The influence of public debt and sovereign interest rates during the financial crisis 30
5.4 Control variables 32
5.5 Robustness check and clarification 34
6. Summary and conclusion 36
7. References 40
3 1. Introduction
4 government. Banks typically are the biggest holders of securities issued by governments within the eurozone (Angeloni and Wolff, 2012). That emphasizes the important role of banks within the system, and the intertwined connectedness of banks and governments. Particularly in the years since the outbreak of the financial crisis late 2007, the absolute level of domestic debt securities on banks’ balance sheets has increased (Merler and Pisani-Ferry, 2012 and Angeloni and Wolff, 2012). Banks are thus much more likely to be affected by sovereign tensions relative to non-financial firms. The debate on the effects of public debt and sovereign interest rates on economic growth is evolving (Reinhart and Rogoff, 2010). However, little is known as about the exact impact of public debt and sovereign interest rates on the profitability of banks. The research question designed is as follows:
‘What is the influence of sovereign credit risk on the profitability of banks in the eurozone?’
5 for the eurozone, but that country differences increasingly will occur. This paper therefore explores the impact for the eurozone as a whole, both for listed and unlisted banks. I create a panel dataset of 5,722 banks throughout the eurozone, which have reported bank profitability in one or more years in the period 1999-2011. For each eurozone country, macroeconomic variables are collected, including the explanatory variables for public debt and sovereign interest rates. This paper builds on the existing literature on bank profitability, by controlling for variables that have empirically shown to be impacting bank profitability. To explore the coefficients during the crisis period, dummy variables have been introduced for crisis years.
I find a significant and positive relation between local public debt and bank profitability. This impact of public debt becomes stronger during the financial crisis. However, when looking at the country differences, only for Austria, France Germany and Luxembourg public debt shows to have a positive impact on bank profitability. During the crisis, the coefficient for public debt only strengthens for banks that are originated in Germany, but remains also positive in Austria. In line with previous studies, market yields on sovereign bonds show to have a significantly negative impact on bank profitability. This effect becomes stronger during the crisis. These results imply that, since the outbreak of the financial crisis, markets have been reassessing eurozone sovereign credit risk. Furthermore, yield shows to have more explanatory power than public debt, particularly in explaining the effects of the financial crisis on bank profitability.
6 2. Literature review and hypotheses
In this section, I review the literature on the influence of sovereign risks on the operational performance of banks. Within this literature review, hypotheses will be formulated.
2.1 Thirteen years of the eurozone: a risk-free market?
8 missed the coercive power, although it had proper tools to back monetary policies as negotiated in this Stability and Growth Pact. Therefore, Buti et al. (2003) argue that the specification of sanctions as well as the enforceability is to be improved. In case of violation, sanctions should be sufficiently large, and misbehavior in good times should be tackled. This has become even more apparent in the last couple of years, showing systemic downside risks and declined, or even negative, economic growth. The eurozone leaders have therefore agreed upon stronger budgetary discipline and are working on new legislation that makes it possible to automatically put sanctions on excessive deficits, unless a qualified majority opposes this (European Council, 2011).
Figure 1 (Source: Eurostat, 2012)
9 Bolton and Jeanne (2011) argue, financial integration without fiscal integration will result in an inefficient equilibrium supply of government debt. What happens in the monetary union of the eurozone, is that the riskier members of the currency union will issue too much debt, thereby enjoying lower yields due to the fact that these countries are backed by the union that is safer as a whole. This again is one of the symptoms of the free-rider problem. The safer members on the other hand, inefficiently restrict the amount of debt that is issued on bond markets. Debt levels in riskier countries therefore have been rising throughout these years, as they were able to raise a relatively higher amount of debt rather than the amount that would have been possible when standing alone. Comparable with the debt level in figure 1, figure 2 shows that interest rates for individual member states remained broadly stable at comparable levels, from the introduction of the euro until late 2007. This changed with the outbreak of the financial crisis. In August 2007, the ECB started initiating its liquidity operations,
10 thereby supporting European banks that had severe exposure to assets-backed securities in the United States. The operation followed the troubles arising in the U.S. housing market. Back then, there were little worries about Europe. The crisis entered a more acute phase with the fall of Lehman Brothers in September 2008, severely hurting not only the U.S., but Europe as well (Lane, 2012). During the start of this global financial crisis, investors started to reassess their international exposure. As Milesi-Ferretti and Tille (2011) put it, investors were repatriating funds back to their home markets. During the gross of 2008 and 2009, European markets remained relatively calm, but at the end of 2009, a new phase was being entered by worse than expected macroeconomic statistics on Spain and Ireland, and the restatement of the Greek deficit (Lane, 2012). From early 2010 therefore, markets started to also reassess the different sovereigns within the eurozone. Not soon after, Greece was the first country that sought assistance of the European Union in May 2010. Investors had already started to doubt the ability of the eurozone to be a fully converging risk free single market with that one single currency, the euro. As a result, interest rate differentials in government bonds of the different nation states of the eurozone started to diverge again, as shown in figure 2.
2.2 The effects of public debt on economic growth
11 within the European Union, mainly spent on healthcare and education. As Faraglia et al. (2010) argue, governments have the ability to borrow long term and are so able to roll over their borrowing. However, it is a different story with fiscal adjustments. In most cases, tax policies cannot be abruptly altered, so policy changes will not lead to government balance surpluses in the short run.
Several authors find that when debt levels exceed certain limits, economic growth will slow down, or even turn negative. Reinhart and Rogoff (2010) find that, at the moment when public debt reaches a level of roughly 90% of GDP, median economic rates will go down with one percent on average. Other authors (among others: Cecchetti et al., 2011, Kumar and Woo, 2010, Checherita and Rother, 2010, Caner et al., 2010) find a similar correlation of high debt levels and economic growth at similar percentages. All of these authors find non-linear relationships, implying that there is an optimum level. For emerging markets there is evidence for a threshold of roughly 64 percent of debt to GDP. The most important contributions to the literature within this field are collected in appendix 1. It is a clear observation that debt levels particularly have risen in Europe. Although the debate as to whether these high debt levels are also causally related to economic growth, remains unsolved (Panizza and Presbitero, 2012), one can at least not ignore that government finances are the single most important indicator of a country’s fiscal sustainability. High public debt, as already suggested by Diamond (1965), leads to potentially higher taxes that need to be paid to fulfill the increased interest payments needs. This will likely reduce the lifetime consumption of taxpayers. Eichengreen (2010) moreover argues that long term interest rates are likely to rise when debt burdens increase, as has been the case in the years since 2007. This will, other things kept equal, lower the level of investments and thereby slow down economic growth rates, since relatively more public money is paid on interest expenses. Following Adam (2011), increased expenditures can only be growth enhancing when the government debt level is sufficiently low. So even though the stimulus packages released in the recent years might be successful in the short run, these will lead to an increased deficit and thereby to an increased debt to GDP level that could lead to the contrary in the long run. Lower economic growth leads to lower bank profitability. However, this does not yet incorporate the other effects that come along with increased public debt levels.
12 2.3 The influence of public debt levels on the profitability of banks
13 to obtain liquidity from the respective central bank. Furthermore, high debt levels could lead to rating downgrades of that specific sovereign, almost always flowing through to domestic banks, with lower credit ratings and thereby diminished market access of these banks (Panetta et al., 2011). Via the same mechanism, high debt levels could lead to increased yields2 on the market of credit default swaps. Thereby, a lesser ability to meet all debt obligations is signaled. This has particularly been visible during the recent financial crisis. Several authors have studied this link between sovereign and banking crises. The latter namely has historically often been followed by steeply rising public debt levels (Frangakis, 2011). This mechanism also works the other way around (Reinhart and Rogoff, 2010). The absolute level of public debt after a typical banking crisis, as they say, is on average 86% higher relative to the period prior to that crisis. Reinhart and Rogoff (2011b) argue, that this is not necessarily related to the bailing out banks or the recapitalization of the banking system. These costs, as the authors say, are difficult to measure, and studies diverge upon the fact as to whether or not these are directly visible in public debt levels. But it is clear, that European member states in general but eurozone countries in particular, spent significant amounts of money by means of capital injections into banks. It is mostly West- and Central European countries that are listed on top in terms of guarantees relative to the country’s GDP. Combined with all the guarantees outstanding, of which it is not yet fully clear whether and when these are going to be called upon, no less than 36.5% of the euro area and 43.6% of the EU27 GDP has been approved as a measure to support the banking system3. Denmark here appears to bias the total approved measures in the European Union upwards, by approving the largest interventions of the whole continent. For Belgium and The Netherlands respectively, a safety net of 74.6% and 42.2% of the country’s GDP is approved to guarantee domestic banks. Logically, a guarantee that is not being called upon does not bring any direct fiscal costs in the short run, since guarantees alone do not increase public debt levels. All these guarantees can be seen as explicit, since these are the ones that can either or not be called upon. Additionally, government debt-to-GDP levels are an important asset in determining implicit government guarantees. As Panetta et al. (2011) find, financial markets assess the willingness and the ability of governments to support banks during systemic downside events. Such an implicit guarantee could as a consequence lower funding costs for domestic banks, and thereby increase profitability. Although these government guarantees tempt to be large, the higher debt level rather is the result of declined tax income (Reinhart and Rogoff, 2009). But it remains the case that the economic costs of the
2 In the further of this paper, yield and interest rates are used interchangeably
14 2007-2009 crises are on average larger, relative to past crises (Frangakis, 2011). Though, also this author argues that the direct fiscal costs of dealing with the crises appear to be lower. Although the exact linkage between banks and a sovereign's debt level yet remains to be unclear, it seems obvious that banks and sovereigns have never been as intertwined as they have become in the recent years. The question is to what extent public debt levels are really affecting the profitability of these banks. The above reasoning suggests the following hypothesis:
• H1: Domestic public debt levels have a positive effect on the profitability of banks
2.4 The influence of sovereign interest rates on the profitability of banks
15 crisis. These banks have difficulties to find funding, since the collateral, mostly home-sovereign bonds thus, is perceived to be too risky.
This riskiness, when resulting in a sovereign's downgrading, in such a case often results in downgrades for domestic banks, as what has recently been shown in Spain. As a result, interest rates on Spanish government bonds has significantly increased. Thereby, a vicious circle is being entered in which governments face more downward pressure, since risks increase via the mechanism that implicit bank guarantees from the government will become explicit, thereby again increasing the domestic interest rates, increasing the funding costs for banks and decreasing profitability. This line of thought suggests the following research hypothesis:
• H2: Sovereign 10-year market interest rates have a negative effect on the profitability of banks
2.5 The influence of public debt and sovereign interest rates during the financial crisis
In times of cyclical downswings, banks are less willing to loan money on account of increased credit risk. This effect is even stronger when banks are forced to maintain higher capital buffers (Bikker and Hu, 2002). This procyclicality of banks is confirmed by Dietrich and Wanzenried (2011) in a recent study running from 1999-2009. Especially in the current European debt crisis, more stringent institutional regulation has made banks to move their tier-1 ratio up to higher levels. In the recent years, with the Basel III accord, banks need to hold more capital, as put in the core tier 1 ratio. The more capital these banks are able to attract, the less risky these banks are becoming. But, this could at the same hurt bank profitability (Eichengreen, 2011). However, the amount of domestic government bond holdings relative to the core tier 1 ratio decreases, whilst keeping the same level of absolute holdings. During the financial downturn, the difference in interest rates between short- and long-term loans made banks to use short term deposits in order to finance longer term loans has become larger, thereby increasing profitability. Furthermore, Bikker and Hu (2002) find that profits in GDP growth environments higher than 2%, mostly turn out to be 2.5 times as high relative to bank profits in GDP growth environments under 2%. This is explained by the fact that during a boom, capital reserves accumulate much faster and additional capital is far easier to obtain. But there appears to be more than just the economic growth effect, that affects bank profitability during times of crisis.
17 This seems to be the case for public debt, and particularly for domestic interest rates. As from the beginning of the financial crisis, but particularly during the European debt crisis, markets started to reassess sovereign credit risk. These eurozone-wide risks were, till the eruption of the financial crisis, perceived to be rather low (De Grauwe and Moesen, 2009). However, this safe-assets view has shifted to a view in which sovereign credit risks have gained importance, thereby increasing the effects on the profitability of banks. Particularly in times of crises, several eurozone countries have their interest payments seen to go up. In the period from April to July 2011, when the EBA stress test was released, Greek debt was a matter of concern for the market valuation of domestic banks. In the period afterwards, up till December, Italian debt holdings, as well as Portuguese and Irish debt holdings were an issue, again mainly hurting balances of the Italian, Portuguese and Irish banks, respectively. Greek debt holdings in this period remained to have a material effect on the banks' market values throughout the country of Greece. When a bank’s collateral shrinks, its credibility moves downwards almost immediately. Banks at the same time could obtain abnormal returns on bonds from highly indebted countries as Portugal and Spain on secondary markets, when news about prospects of a possible Greek bailout was circulating in the media back in 2010 (Mink and de Haan, 2012). This signals the willingness of the eurozone leaders that a bail-out would not be limited only to Greece. This could be seen as a 'wake-up call'; new information about Greece made investors to reassess the vulnerability of other countries. For most types of government debt securities, there are liquid markets in which these securities can be traded. The value of debt securities moves in the opposite direction of domestic interest rates. Following this line of reasoning, it is expected that the influence of interest rates on bank profitability is stronger after the crisis. Banks can, during the financial crisis, therefore face increased difficulties in financing their own balance sheets. The underlying securities provide less collateral to obtain liquidity from central banks, thereby decreasing wholesale funding. Also, in times of crisis, the impact of rating downgrades of a specific sovereign deems to be higher. Lower credit ratings will flow through to the creditworthiness of domestic banks, thereby diminishing market access for funding (Panetta et al., 2011). In line with this theory, the following hypotheses are suggested:
• H3a During the financial crisis (2008-2011), the coefficient of the influence of public debt levels on the profitability of banks becomes stronger
18 3. Methodology
In this section, the methodology used to test the hypotheses is presented. The characteristics of the collected data comprise both time series and cross-sectional elements. First, the characteristics of the model are presented after which I discuss both the dependent, and the independent variables. Second, a summary of the different variables and the expected coefficients are presented, in table 3.1. Lastly, the different panel data techniques used are discussed.
3.1 Equation
In the recent literature, several models have been proposed to link the different variables that impact the profitability of banks (ROAA). Hereby, two important types of determinants have gained much attention – the internal changes, or within effects, and the external changes, or dynamic reallocation effects. Following Staikouras and Wood (2011), this distinction is important, since failure to count for external changes can lead to misinterpretations about the underlying strengths of the financial industry. Therefore, both aspects have been taken into account when estimating the influence of the different variables on the profitability of banks. To test the hypotheses, an equation on profitability has been put together based on the empirical work of, among others, Demirgüç-Kunt and Huizinga (1999), Athanasoglou et al. (2008) and Hauner et al. (2008). The equation is specified as follows. First the regression with the domestic public debt variable will be tested. Secondly, the regression with the domestic interest rate variable will be used in the equation, and in the third regression, the two macroeconomic variables will be combined. Lastly, the crisis variable will be included for both the public debt and the yield variable.
3.2 Dependent variable
19 captured, by using this variable (Angeloni and Wolff, 2012). Since not all of the banks included in the sample are listed, I have chosen not to include the return on shares as an alternative measure of profitability. Another often used profitability measure, the net interest margin (NIM), measures the profit earned on interest activities but has not been included because of the fact that this variable is likely not to be immediately affected by the domestic sovereign’s debt-to-GDP level, since the NIM does for instance not account for losses in sovereign bond holdings.
3.3 Independent variables
In this section, the independent variables are discussed. First, the two main explanatory variables are put forward, namely the debt-to-GDP level and the domestic sovereign interest rate on 10 year bonds. Second, the control variables that have gained evidence in the literature to be affecting bank profitability are discussed, starting with external variables, followed by the internal variables.
3.3.1 Local public debt
20 3.3.2 Interest rates
Other than the debt-to-GDP levels, the sovereign interest rate can be regarded as being a forward looking variable that indicates the current credibility of a country with regard to all future expected events. The mechanisms through which this influences the profitability of banks are similar to the ones associated with high debt-to-GDP levels. Countries paying high interest rates to raise money, are more likely to face problems in refinancing themselves, and might be subject to the same channels that are influencing the profitability of banks, as seen with high debt-to-GDP levels. The loss in value of the government debt securities hereby will make the collateral that can be used to obtain liquidity from central banks lower. Furthermore, credit downgrade risks of the sovereign decreases the value of both explicit and implicit government guarantees, and most often leads to downgrades of domestic banks, thereby increasing funding costs. Therefore, higher interest rates are expected to give a negative sign, decreasing the profitability of banks.
3.3.3 Other macroeconomic variables
Economic growth: banks are being able to collect higher profits in GDP-growth environments. The demand for lending increases when the economy is showing cyclical upswings (Dietrich and Wanzenried, 2011). On the other side of the balance, especially during a boom, capital is readily available in economic growth environments, and positive NPV projects are easier to access (Bikker and Hu, 2002). Additionally, probabilities on individual and corporate defaults are lower when growth is higher (Staikouras and Wood, 2011). Although the growth variable in the literature appears not to always be statistically significant, a positive sign is expected.
21 3.3.4 Bank specific determinants
Bank size: bank size, is being measured in terms of total assets. In order to reduce scaling effects, the logarithm of the total assets for each bank is taken rather than the absolute level of assets. The relationship between the size of banks and profitability is rather ambiguous. Large banks appear to benefit from economies of scale, and have a higher degree of product and loan diversification, thereby reducing risk (Smirlock, 1985). Additionally, larger banks in terms of assets appear to be benefiting from being systemically large. Systematically large banks are judged by financial markets to have become too big to fail (Demirgüç-Kunt and Huizinga, 2010). Thereby, these banks implicitly enjoy the government's guarantees, and can therefore be more profitable. Larger banks namely face decreased funding costs. However, the same authors argue that many banks have grown far beyond this point, and rather have become too big to save. This could be the case for larger banks in terms of assets on the one hand, but also for relatively smaller banks in highly indebted countries. When banks have grown too large, agency costs tend to increase slightly (Stiroh and Rumble, 2006). Following the literature however, this does not outweigh the benefits of being large. Therefore, the expected sign of size is positive.
Loans to assets: The net loans to total assets ratio provides a measure of risk since it incorporates the riskier loans relative to other assets, such as government debt securities. The loans to assets ratio appears to be inversely related to the bank’s return on average assets (Staikouras and Wood, 2011). When the amount of riskier loans outstanding increases, bank funding costs will increase. That could reduce the profitability of the firm. At the same time, however, loans pay high returns. In the literature though, this does not outweigh the risk on losses. Banks that are more dependent on loan-earning assets namely, are less profitable than banks that have larger sources of other income (Dietrich and Wanzenried, 2011). Therefore, the expected sign is negative.
22 Capital ratio: banks that hold high levels of equity over their total assets appear to be the best performing banks (Demirgüc-Kunt and Huizinga, 1999). This can be explained by the fact that maintaining higher capital ratios implies lower bankruptcy costs, as a result of that funding costs will be lower4. The capital ratio, following Dietrich and Wanzenried (2011), is highly correlated with deposits, since customers are, particularly in the years since the financial crisis, looking for safer banks to deposit their savings.
Customer deposits: the level of customer deposits as a percentage of total funding gives an indication of the dependence of banks to their customers. In countries with higher debt-to-GDP levels, the possibility of higher future distortionary taxation exists. Therefore, people increase their savings to anticipate on this. This brings more deposits within the banks. Additionally, in times of crises, people seem to increase their savings (Frangakis, 2011). For banks operating in high debt-to-GDP level countries, this could indicate that when customer deposits rise, funding costs could become less. However, profitability not necessarily increases, since particularly in the years after the financial crisis, attractive investment opportunities have turned out to be more difficult to find. Therefore, the expected sign remains to be ambiguous.
Listing: Following Ianotta (2007), listed banks might face more pressure from shareholders, analysts and financial markets in general. Additionally, listed banks in general are larger and better capitalized relative to their non-listed counterparts. This has been found to be related to the funding costs, which appear to be lower for listed firms (Ianotta, 2007). Listed firms, however, also face higher costs, as a result of the fact that being listed brings about more reporting requirements. This creates a significant negative impact on listed bank (Staikouras and Wood, 2011). Listed banks nevertheless appear to be more profitable than banks that are not, particularly during times of financial crisis (Dietrich and Wanzenried, 2011). The influence of being listed is therefore expected to have a positive impact on bank profitability. A dummy variable (DLISTING) will be used to control for this.
IFRS: banks can report either using their local GAAP, or under IFRS. In case of the latter, one could observe that all assets and liabilities at each point in time are being valued at their true market value, and not at the value
23 at maturity. Implementation of IFRS has shown to be positively affecting the profitability of most sectors (Iatridis, 2010 and Iatridis and Dalla, 2011). This is a result of the fact that under IFRS, less income smoothing is documented (Doukakis, 2010). Therefore, a positive sign is expected. I control for the accounting standard of the different banks making use of a dummy variable (DIFRS).
Variables Variable description Expected sign
Dependent variable
ROAA
Return on average assets, measured as the yearly net profits over
total average assets (%)
Independent variables
Public debt
Gross government debt over GDP (%) at the end of the year for each
country +/-
Yield
Average yearly market interest rate on 10 year government bond for
each country (%) -
Economic growth Annual GDP growth for each country (%) +
Inflation Annual inflation rate for each country (%) +
Bank size Logarithm of the total assets for each financial institution (lnassets) + Loans to assets Net loans to total assets (%) at the end of each year - Loan loss provisions Loan loss provisions over total loans (%) -
Capital ratio Equity to total assets (%) +
Customer deposits The level of customer deposits over total funding (%) +/-
DListing
Dummy variable that takes 0 for unlisted banks, and 1 for banks
listed on the stock exchange +
DIFRS Dummy variable that takes 0 for local GAAP and 1 for IFRS +
Table 1: Proposed variables and expected signs
3.4 Regression and panel data techniques
debt-to-24 GDP level and domestic interest rates on 10-year bonds, are derived to control for the influence on bank profitability. The model can therefore be tested in Eviews by making use of multiple linear panel regression models. Since there are cross-sectional observations that have no data available for one or more variables at one or more different points in time, the dataset can be typified as being unbalanced. The largest advantage of panel data is, that it can address broader ranges of issues and tackle more complex problems (Brooks, 2008). Since making use of panel data combines cross-sectional with time-series data, the number of degrees of freedom will be increased. More importantly, variables that would otherwise have been omitted and could have led to biased results, are implicitly included by adding a dummy for each firm that captures such omissions. All regressions will be estimated using fixed effects, both for cross-section and for time effects. This will strengthen the power of the tests deducted. Since Eviews automatically accounts for missing observations, all of the following techniques can all be used. Corrections are applied for HAC and standard errors.
4. Data and descriptive statistics
The data used to test the hypotheses are derived from the BankScope database, as built up by Bureau van Dijk. The analysis has been conducted making use of data stemming from individual banks. I will present the data selection of the 6,262 banks that have (dis)continued operations in the period from 1999-2011, and that have reported their profit for at least one of the periods.
4.1 Data
25 interest income, ownership, loans relative to customer deposits, customer deposits to total funding, ownership and the reporting standard of the bank have been derived from the BankScope database.
Table 2: Banks’ countries of origin
4.2 Descriptive statistics
In this section, descriptive statistics are presented. Previous literature acknowledges the relationship of the different variables on bank profitability, these variables are presented in table 4.2. In order to improve the generalizability of this study, and enabling the dataset to better fit the normality requirements, extreme outliers are erased. For the dependent variable profit, the lower 1% as well as the upper 1% of the observations has been removed. The same yields for the dependent variables capital ratio, loans to total assets and loan loss provisions to total assets. Following Kothari et al. (2005), data trimming could lead to the omission of extreme observations. But since the dataset can be considered to be sufficiently large, this is unlikely to lead to biased results. To correct for the omission of the extremes, robustness checks and clarifications will be performed at the end of the results section.
Country No. Of banks % of total observations
26 4.2.1 Summary statistics
Table 4.2 presents the summary statistics. The descriptive statistics show, in several cases, lower medians relative to the means within the sample. This indicates skewness. However, by trimming the dataset, skewness has already been limited to a minimum level.
Mean Median Maximum Minimum Std. Dev.
Firm year observations Profit 0.52 0.3 10.01 -4.34 1.03 40,974 Public debt 84.86 94 165.3 3.7 25.75 74,386 Yield 4.21 4.22 16.24 2.41 0.9 74,021 Economic growth 1.54 1.7 10.5 -14.3 2.41 74,342 Inflation 1.98 1.9 12.2 -1.7 1.02 74,386 Size 13.48 13.27 21.51 3.61 1.91 41,933 Loans to assets 56.32 60.64 100 -0.04 22.6 40,852
Loan loss provisions 17.77 15.39 162.64 -58.68 20.44 37,376
Capital ratio 9.42 6.39 86.3 0.66 10.84 41,055
Customer deposits 67.07 74.59 99.99 0.16 25.23 38,413
Table 3: Descriptive statistics for the variables selected
4.2.2 Correlations
27 the domestic sovereign (Demirgüç-Kunt and Huizinga, 2010). Via the same mechanism, the correlation of size and capital ratio is significantly negatively correlated (-0.43). This is also the case for the correlation of size and the core tier 1 ratio, although to a lesser extent (-0.18). Again, as Demirgüç-Kunt and Huizinga (2010) find, this can be seen as an effect of the positive aspects of being too big to fail, implicitly leaning on the domestic sovereign.
5. Results and findings
In this section, I will test the hypotheses as proposed in the second section. Firstly, I report the results of the panel least squares regression on the influence of public debt on the profitability of banks. Second, I present the influence of interest rates on 10-year bonds in the domestic country on the profitability of banks.5 Thirdly, I will check whether the signs of public debt and yield have significantly changed during the financial crisis. Lastly, I analyze whether the influence of public debt and interest rates have had different implications for the different eurozone countries to clarify results and to check for robustness.
5.1 Evidence for the effects of domestic public debt on eurozone bank profitability
In the following models, H1 is being tested. The regression model uses panel OLS techniques. Several regressions were run, which are presented in table 4. The results of this study show that banks that are originated in environments in which domestic debt levels are running up, can benefit from using local debt when it comes down to the profitability. In other words, each additional unit of public debt leads to additional profits for banks, although, in the first model, this sign is not significant. Model 1 here, is the baseline model in which only the influence of public debt on the profitability of banks has been regressed. In this baseline model, public debt levels have been isolated from the other explanatory variable, yield. The different control variables however, are included in all models. In this first model the beta for public debt comes to a positive sign of 0.0002. However, this small sign is not significant and does not allow me to accept the null hypothesis, that debt has a positive influence on the profitability of banks. The sign of public debt is remains positive throughout the different models in table 4. When adding yield as a second explanatory variable next to the control variables, the coefficient is somewhat less strong (0.001), while the T-statistic slightly moves up. In the later models, when
28 Table 4: The influence of sovereign tensions on the profitability of banks6
6 Cross section and time fixed effects are used for all models. Robust t-statistics are provided in the parentheses. All results are corrected for heteroskedasticity and
autocorrelation. ***/**/* indicate significance levels at the 1%, 5% and 10%-levels, respectively (two-tailed)
Independent variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Intercept -1.227** -1.212 -1.235* -1.612*** -1.576** -1.844***
(-1.78) (-1.55) (-1.58) (-2.35) (-2.02) (-2.36)
Public debt 0.0002 0.001 0.002* 0.003***
(0.27) (0.61) (1.92) (3.50)
Public debt crisis 0.006*** 0.003***
(10.93) (4.92) Yield -0.033** -0.034** -0.047*** -0.045*** (-2.18) (-2.24) (-2.90) (-2.74) Yield crisis -0.339*** -0.285*** (-12.32) (-8.99) Economic growth 0.025*** 0.013** 0.014** -0.001 -0.007 0.010* (4.78) (2.38) (2.45) (-0.24) (-1.20) (1.70) Inflation 0.030*** 0.028*** 0.029*** 0.019** -0.012 0.008 (3.39) (2.86) (2.93) (2.19) (-1.08) (0.82) Size 0.090*** 0.107*** 0.106*** 0.104*** 0.124*** 0.120*** (5.64) (6.50) (6.36) (6.23) (7.53) (7.24) Loans to assets -0.003*** -0.003*** -0.003*** -0.003*** -0.002*** -0.003*** (-5.93) (-6.10) (-6.12) (-5.02) (-4.47) (-4.65) Loan loss provisions 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001***
29 a crisis dummy as well yield is included, the coefficient for public debt becomes significant. Coefficients for the influence of public debt on bank profitability then vary between 0.002 and 0.003. These consistent results are in line with the literature. Hauner (2008) find that higher levels of debt could raise the profitability of banks. This was however, since the sample period ran from 2001 to 2003, still in the time of the safe assets view. Government bonds were, up to the European sovereign debt crisis, still seen as safe assets within the eurozone. The significance in the later models is remarkable, to some extent. When public debt increases, the value of implicit government guarantees will be lower (Demirgüc-Kunt and Huizinga, 2010). In line with Angeloni and Wolff (2012) however, these significantly positive signs can be explained by the fact that banks are the biggest holders of domestic debt. As long as financial markets do not question the creditworthiness of single countries, the results should show that there is indeed no negative sign with regard to public debt expansion. With a level of 87.4% of debt-to-GDP at the end of the sample period within the eurozone, the results suggest that the public debt level on average still not touched upon the level where public debt starts to hurt the balance sheets of banks. The positive effects of public debt may be stronger in these results however, if one considers that the majority of the observations included are from Germany. German government debt securities are by definition regarded to be safer than Greek debt securities, thereby the results could have been biased upwards. It is therefore important to interpret these results with care. In Greece and Italy, public debt levels have by far exceeded the levels that are assumed to be critical for the gross of the sample period. For countries such as Germany, The Netherlands and Spain, this is not the case; although for these countries debt levels also have been observed to be growing, since the financial crisis. The influence of public debt on the profitability of banks might differ throughout the eurozone, and it is therefore considered to be worthwhile to analyze the effects also by country.
5.2 Evidence for the effects of domestic interest rates on eurozone bank profitability
30 interest rates contains all publicly available information about the creditworthiness and perceived risk of that specific sovereign, as judged by financial markets. Therefore, it is not surprising that the impact is stronger, relative to the effect of the public debt level, making interest rates to have a larger influence on bank profitability. When combining the variable yield with the coefficient of public debt in model 3, the coefficient for yield slightly moves downwards. With -0.034, yield remains negatively affecting bank profitability at the 5%-level. The results show that when the model is expanded with the different crisis dummies, that are analyzed later in this chapter, the explanatory power of the coefficient for yield increases. The coefficient for yield becomes increasingly negative, ranging from -0.045 to -0.047, and becomes significant at the 1%-level. These results are in line with the literature. When following Panetta et al. (2011), balance sheets are weakened when sovereign interest rates are going up as a result of high debt levels, as the market value of the debt security goes down. Second, the lower debt securities’ value makes that banks can obtain less liquidity from the central bank, since its collateral has gone down. Third, high debt levels, particularly when combined with high interest rates, can make credit rating agencies to reassess the creditworthiness of the sovereign negatively. A downgrade of a sovereign thereby almost always flows through to domestic banks, increasing funding costs. Lastly, banks operating in higher indebted governments are less likely to be able to benefit from implicit, or explicit guarantees. All of these four channels will negatively impact the profitability of banks, through the rise in sovereign credit risk by means of an increase of domestic interest rates.
5.3 Effects of domestic macroeconomic changes since the financial crisis
32 effects of the financial crisis, that can not be explained by the two variables public debt and yield, differ throughout the models. In the first four models, not including the crisis dummies for public debt and yield, the coefficients for the financial crisis are negative and statistically significant. As expected, bank profitability will suffer in times of financial crisis. The coefficients vary from -0.090 in model 1 to -0.351 in model 4. More importantly, the dummy variable for the financial crisis plays an important role when deriving conclusions from the crisis dummy coefficients of public debt and yield. As can be observed, when the crisis dummy for public debt is inserted in model 4, the coefficient of the crisis dummy is most negative, with -0.351. In model 5 however, when yield and the yield crisis dummy are applied, the coefficient of the crisis dummy becomes significantly positive (0.619). This suggests that domestic interest rates have played a significant role during the financial crisis in explaining bank profitability. Furthermore, the results suggest that there are positive aspects of the financial crisis that cannot be explained by model 5. In model 6 however, when public debt and the public debt crisis dummy are added to the regression, the coefficient of the crisis dummy decreases to 0.415 and is no longer statistically significant. This again confirms that during crisis times, local public debt has explanatory power with regard to bank profitability. Since the crisis dummy in model 6 is not statistically significant, based on the coefficient of the crisis dummy I cannot draw conclusions about the positive effects of the financial crisis. However, one could argue that the financial crisis within the eurozone could also have led to some positive effects on bank profitability, for banks located in countries that are still outperforming other eurozone members. As mentioned, the majority of the sample consists out of German banks, which are assumed to be safer. Therefore, a country analysis is worthwhile.
5.4 Control variables
34 times, for tax purposes. In bad times, a negative loan loss provision could be used to ‘borrow money’ from the bank (Kanagaretnam et al., 2003). This line of reasoning suggests, that higher loan loss provisions are taken when banks are more profitable. However, it would still be the case that after a loan loss provision, the return on assets will be lower. The coefficient for customer deposits, as a percentage of total funding, is found to have a positive and statistically significant effect on the profitability of banks. This is in line with the literature. Particularly when debt levels are running up, people tend to increase their savings (Frangakis, 2011). This is a result of the fact that when debt levels increase during times of crises, people will anticipate on higher future distortionary taxes. In my analysis though, no distinction has yet been made between normal times and times of financial turmoil. Possibly, this is one of the reasons that this effect is relatively small (0.001). Listed banks show to be more profitable than its non-listed counterparts, significant at the 1%-level with a sign of 0.132 to 0.167 in the different models. This is in line with the literature, since listed banks face more pressure from shareholders and financial markets (Ianotta, 2007). Also, funding costs appear to be lower for listed banks In general, banks that are listed appear to be better capitalized and since listed banks on average are larger, these banks could also derive implicit and explicit benefits from the domestic sovereign for being systemically large (Demirgüc-Kunt and Huizinga, 2010). Whether banks report their results by making use of the local GAAP or IFRS, also makes a difference. The coefficient for IFRS ranges from 0.101 and 0.123, whereby all of the coefficients are significant at the 1% confidence level. This is in line with the literature, which proves that firms that value their assets and liabilities making use of the true market value, or IFRS, significantly face higher profitability (Iatridis and Dalla, 2011). The majority of the banks that report using IFRS are listed. Although the correlation of the dummies for IFRS and listed banks was not exceeding the set limit of 0.8, results should be interpreted with caution.
5.5 Robustness check and clarification
35 analyzed, further again including a dummy variable that takes 1 during the financial crisis and 0 during non-crisis years. Apart from Germany (16,734), only the observations for Austria and France are larger than 1,000. The other countries vary between 58 and 551. Therefore, the results should be interpreted with caution.
36 measures the impact of the crisis, is significantly negative at the 10%-level (-14.63). This suggests that for Austria, crisis effects other than public debt and yield have a negative effect on the profitability of banks. Furthermore, a significantly negative coefficient can be seen for yields during the financial crisis for Ireland, Luxembourg and Slovenia. For all these three countries, this explains the negative effects of the financial crisis, since the crisis dummy in both cases is significantly positive at the 10%-level. This positive coefficient suggests that for these three countries, benefits are reaped from the financial crisis that cannot be explained by the model. The coefficients for public debt in Slovakia indicate that after the public debt crisis, the effects of public debt on bank profitability are less positive. In Slovenia, the coefficient for public debt becomes negative. This can be explained by the fact that automatic stabilizers in Eastern- European countries are much lower than in Central Europe. This implies a lesser ability of these governments to insure higher income levels, and therefore a higher enactment of fiscal stimulus programs, that could make the public debt levels in Slovakia and Slovenia to increase at a faster pace, particularly in times of a crisis, thereby increasingly hurting profitability.
It can be concluded, that the coefficients both for public debt levels, as well as for sovereign interest rates differ widely throughout the eurozone. For Germany, the coefficient for yield shows to be significantly changing its sign in times of financial turmoil. The analyzed domestic variables with regard to sovereign risk do play a role in explaining bank profitability within the different eurozone countries. However, the results of this section are limited, as a result of the limited observations that could be derived for the individual member states, and therefore shows many insignificant variables. Therefore, results do not allow me to draw country-specific conclusions for the majority of the countries.
6. Summary and conclusion
37 return on average assets (ROAA) over the whole sample period. This makes the sample an unbalanced panel. In order to empirically test the hypothesis, a regression analysis was performed. As a robustness check, country-level regressions are applied.
I find a significant and positive relation between public debt levels and bank profitability. The results for public are thereby broadly in line with previous studies that have been done is this field. Up to the financial crisis, domestic macroeconomic variables were not perceived to play a major role. When investors do not doubt the creditworthiness of the country, debt levels give room for expansion. Although the relation between public debt and bank profitability is positive, it only plays a minor role in explaining bank profitability. Since the financial crisis however, the impact of public debt on bank profitability has become stronger. For banks, it could therefore be beneficial to make use of local debt. That is a remarkable conclusion. Other than aimed at with the establishment of the Maastricht Treaty, group cohesion in the eurozone could not be maintained. Since the start of the financial crisis, investors have reassessed their assets, and concluded that the eurozone as a whole can no longer be regarded to be a fully financially integrated monetary union. Banks located in countries that have high debt levels such as Italy and Greece are therefore expected to rather see negative effects of public debt. The majority of the observations however is from Germany. This suggests, that the results particularly hold for German banks. As a robustness check, a country analysis is applied. Only for Austria, France, Germany and Luxembourg public debt had a positive impact on bank profitability, during normal times. During the crisis, the coefficient for public debt only strengthens for banks that are originated in Germany, and weakens, but remains positive for Austria. From an institutional perspective, the results thereby suggest that the coervice power of Germany has strengthened. These namely are all countries that somehow depend on Germany, and resemble country characteristics. Moreover, it is these countries that are seen as safe havens by investors during crisis times. As a result, banks that operate in countries such as Austria and Germany that can reap benefits from the financial crisis by using local debt. For other countries, including Greece and Italy, not enough evidence was found to draw country-specific conclusions on the influence of public debt.
38 coefficient strengthens during the financial crisis. Particularly in times of financial turmoil namely, the negative effects of sovereign yields on bank profitability increase. Contradicting the rather backward looking variable of public debt, yield contains forward looking information on a country’s creditworthiness. It is therefore not surprising, that the negative crisis effects are fully explained by yields. Banks are the biggest holders of sovereign debt securities. When interest rates increase, the real value of these securities will go down. Thereby, balance sheets are weakened, leaving less collateral for these banks to obtain liquidity from the central bank. The results of this study suggest that interest rates are well capturing the crisis effects on the profitability of banks. The crisis dummy, that explains the effects of the crisis on bank profitability, was negative when only public debt was included in the sample. This implies, that public debt did not fully capture the sovereign risk factor. With the inclusion of yields, the coefficient of the crisis dummy variable becomes positive, but insignificant. The yield coefficients both in normal times, and increasingly in crisis times, therefore show to be valid explanatory variables in explaining the decrease in bank profitability. When accounting for the negative effects of interest rates as well as for the effects of public debt, there appear to still be benefits for some countries, including Luxembourg, Germany and The Netherlands. The crisis in these countries thereby seems to lead to positive side-effects. These crisis effects have not been specifically included in the models, other than the dummy variable for the financial crisis. This suggests that there might be other aspects that have not been captured in the different models, from which safe-haven countries in a monetary union reap benefits during times of financial turmoil. Literature suggests that this positive sign could be a result of a country’s ability to bail out banks. Further research however is needed to clarify these findings.
40 7. References
Adam, K. (2011), Government debt and optimal monetary and fiscal policy, European Economic Review, vol. 55, iss. 1, pp. 57-74
Angeloni, C. and G.B. Wolff (2012), Are banks affected by their holdings of government debt? Bruegel Working Paper, iss. 7
Barro, R.J. (1974), Are government bonds net wealth? Journal of Political Economy, vol. 82, iss. 6, pp. 1095-1117
Bikker, J.A. and H. Hu (2002), Cyclical patterns in profits, provisioning and lending of banks, De Nederlandsche Bank Staff Reports, no. 86
Bolton, P. and O. Jeanne (2011), Sovereign default risk and bank fragility in financially integrated economies, Centre for Economic Policy Research, Discussion Paper, no. 8358
Boyd, J.H. and B. Champ (2006), Inflation, banking and economic growth, Federal Reserve Bank of Cleveland
Bruggink, A. (1989), Performance control in banking: theory and application, Nederlands Instituut voor het Bank- en Effectenbedrijf, pp. 60-74
Buiter, W.H., G. Corsetti and N. Roubini (1993), 'Excessive deficits': sense and nonsense in the Treaty of Maastricht, Economic Policy, vol. 8, iss. 16, pp. 58-100
Buti, M., S. Eijffinger and D. Franco (2003), Revisiting the stability and growth pact: grand design or internal adjustment? Centre for Economic Policy Research, Discussion Paper, no. 3692
41 Cecchetti, S.G., M.S. Mohanty and F. Zampolli (2011), The real effects of debt, Bank of International Settlements Working Papers, no. 352
Checherita, C. and P. Rother (2010), The impact of high and growing government debt on economic growth – an empirical investigation for the euro area, European Central Bank Working Paper Series no. 1237
Côté, D. and C. Graham (2004), Convergence of government bond yields in the eurozone: the role of policy harmonization, Bank of Canada, Working Paper, no. 23
European Central Bank, Money Market Study,
http://www.ecb.int/pub/pdf/other/euromoneymarketstudy201012en.pdf?55794d21c4d43bf20d676356f512a828
Davis, E., C. Gouzouli, M. Spence and J. Star (1993), Measuring the performance of banks, Business Strategy Review, vol. 4, iss. 3
De Bondt, G. (2002), Retail bank interest rate pass-through: new evidence at the euro area level, European Central Bank, Working Paper, no. 136
De Grauwe, P. (2006), What have we learnt about monetary integration since the Maastricht Treaty? Journal of Common Market Studies, vol. 44, iss. 4, pp. 711-730
De Grauwe, P. and X. Moesen (2009), Gains for all: a proposal for a common euro bond, Intereconomics/Review of the European Economic Policy, vol. 44, iss. 3, pp. 132-135
Demirgüç-Kunt, A. and H. Huizinga (1999), Determinants of commercial bank interest margins and profitability: some international evidence, World Bank Economic Review, vol. 13, iss. 2, pp. 379-408
Demirgüç-Kunt, A. and H. Huizinga (2010), Are banks too big to fail or too big to save? International evidence from equity prices and CDS spreads, European Banking Center Discussion Paper, no. 2010-15
42 financial intermediation, Journal of Money, Credit and Banking, vol. 36, iss. 11, pp. 593-623
Diamond, P.A. (1965), National debt in a neoclassical growth model, American Economic Review, vol. 55, iss. 5, pp. 1126-1150
Dietrich, A. and G. Wanzenried (2011), Determinants of bank profitability before and during the crisis: evidence from Switzerland, International Financial Markets Institutions and Money, vol. 21, iss. 3, pp. 307-327
Dolls, M., C. Fuest and A. Peichi (2012), Automatic stabilizers and economic crisis: US vs. Europe, Journal of Public Economics, vol. 96, iss. 4, pp. 279-294
Doukakis, L.C. (2010), The persistence of earnings and earnings components after the adoption of IFRS, Managerial Finance, vol. 36, iss. 11, pp. 969-980
Durbin, J. and G.S. Watson (1951), Testing for serial correlation in least squares regression II, Biometrika, vol. 38, iss. 1-2, pp. 159-177
Eichengreen, B. (2011), Crisis and growth in the advanced economies: what we know, what we don't, and what we can learn from the 1930s, Comparative Economic Studies, vol. 53, iss. 3, pp. 383-406
European Banking Authority (2011), EU-wide stress test, aggregate report.
Frangakis, M. (2011), The rising public debt in the EU: implications for policy, Journal of contemporary European Studies, vol. 19, iss. 1, pp. 7-20
Eurostat, Euro indicators news release, iss. 6 February 2012
43 Fama, E.F. (1991), Efficient capital markets: II, The Journal of Finance, vol. 46, iss. 5, pp. 1575 – 1617
Faraglia, E., A. Marcet and A. Scott (2010), In search of a theory of debt management, Journal of Monetary Economics, vol. 57, iss. 10, pp. 821-836
Hsing, Y. (2010, Government debt and the long-term interest rate: application of the loanable funds model to Greece, Journal of Economic Integration, vol. 25, iss. 4, pp. 722-733
Golin, J. (2001), The bank credit analysis handbook: guide for analysts, bankers and investors, John Wiley and Sons
Grahl, J. (2011), Crisis in the eurozone, Soundings, vol. 47, iss. 47, pp. 143-158
Ianotta, G., G. Nocera and A. Sironi (2007), Ownership structure, risk and performance in the European banking industry, Journal of Banking and Finance, vol. 31, iss. 7, pp. 2127-2149
Iatridis, G. (2010), IFRS adoption and financial statement effects: the UK case, International Research Journal of Finance and Economics, iss. 38, pp. 165-172
Iatridis, G. and K. Dalla (2011), The impact of IFRS implementation on Greek listed companies: an industrial sector and stock market index analysis, International Journal of Managerial Finance, vol. 7, iss. 3, pp. 284-303
Jensen, M.C. (1978), Some anomalous evidence regarding market efficiency, Journal of Financial Economics, vol. 6, iss. 2-3, pp. 95-101
Kanagatretnam, K., G.J. Lobo and R. Mathieu (2003), Managerial incentives for income smoothing through bank loan loss provisions, Review of Quantitative Finance and Accounting, vol. 20, iss. 1, pp. 63-80
44 Kostadinova, I., I. Stanislav Eminescu and A. Tokofai (2011), EU27 and euro area government expenditure-to-GDP ratios falling after 10 quarters of growth, Eurostat Statistics in Focus, iss. 13, pp. 1-12
Kothari, S.P., J.S. Sabino and T. Zach (2005), Implications of survival and data trimming for tests of market efficiency, Journal of Accounting and Economics, vol. 39, iss. 1, pp. 129-161
Kumar, M.S. and J. Woo (2010), Public debt and growth, International Monetary Fund Working Paper, vol. 10, iss. 174
Lane, P.R. (2012), The European sovereign debt crisis, Journal of Economic Perspectives, forthcoming
Lever, H. and C. Huhne (1985), Debt and Danger, Harmandsworth, The Athlantic Monthly Press, pp. 11-20
Maudos, J. and J.F. De Guevera (2004), Factors explaining the interest margin in the banking sectors of the European Union, Journal of Banking and Finance, vol. 28, iss. 9, pp. 2259-2281
Milesi-Ferretti, G.M. and Tille, C. (2011), The great retrenchment: international capital flows during the global financial crisis, vol. 26, iss. 66, pp. 285-330
Mink, M. and J. de Haan (2012), Contagion during the Greek sovereign debt crisis, DNB Working Paper, no. 335
Panetta, F., R. Correa, M. Davies, A. Di Cesare, J. Marques, F. Nadal de Simone, F. Signoreti, C. Vespro, S. Vildo, M. Wieland and A. Zaghini (2011), The impact of sovereign credit risk on bank funding conditions, Bank for International Settlements – Committee on the Global Financial System, iss. 43
45 Radaelli, C.M. (2000), Policy transfer in the European Union: institutional isomorphism as a source of legitimacy, International Journal of Policy and Administration, vol. 13, iss. 1, pp. 25-43
Reinhart, C.M. And K.S. Rogoff (2009), The aftermath of financial crises, American Economic Review, vol. 99, iss. 2, pp. 466-472
Reinhart, C.M. And K.S. Rogoff (2010), Growth in a time of debt, American Economic Review, vol. 100, iss. 2, pp. 573-578
Reinhart, C.M. And K.S. Rogoff (2011a), The forgotten history of domestic debt, The Economic Journal, vol. 121, iss. 552, pp. 319-350
Reinhart, C.M. And K.S. Rogoff (2011b), A decade of debt, National Bureau of Economic Research Working Paper, no. 16827
Sarrat de Tramezaigues, G.R. (2010), Assessing the 2005 growth and stability pact reform: a new fiscal policy rule for the eurozone is required, International Journal of Economics and Finance, vol. 2, iss. 5, pp. 140-150
Savage, J.D. (2001), Budgetary collective action problems: convergence and compliance under the Maastricht Treaty on European Union, Public Administration Review, vol. 61, iss. 1, pp. 43-53
Schlicht, E. (2006), Public debt as private wealth: some equilibrium considerations, Metroeconomica, vol. 57, iss.4, pp. 494-520
Staikouras, C.K. and G.E. Wood (2011), The determinants of European bank profitability, International Business and Economics Research Journal, vol. 3, iss.6, pp. 57-68
47 Appendix 1: the influence of public debt on economic growth
Authors Time span Dataset Critical debt to GDP-level Effect on growth
Reinhart & Rogoff (2010) 18xx - 2010 64 countries 90% 1% decline above critical level
Cecchetti et al. (2011) 1980-2010 18 OECD countries 85% Every 10%-pt increase of debt leads to 0,1%-pt decline
Checherita & Rother (2010) 1970-2011 12 eurozone countries
90-100%, lower confidence
interval at 70% Not researched
Kumar & Woo (2010) 1970-2007 38 advanced and
emerging economies 70-80%
Every 10 %-pt increase of debt leads to 0,2%-pt decline
Caner et al. (2010) 1980-2008 99 advanced and emerging countries
77% (64% in emerging countries)
Every %-pt increase of debt leads to 0,017%-pt decline, 0,02%-pt in emerging
49 Appendix 3: Correlation matrix
Public
debt Yield
Economic
growth Infl. Size
Loans assets Loss prv assets Capital ratio Customer
deposits DListed DIFRS DCRISIS
50 Appendix 4: Country-level analysis7
Independent
variable Austria Belgium Cyprus Finland France Germany Greece Ireland Italy Luxembourg Netherlands Portugal Slovakia Slovenia Spain
Public debt 0.043** 0.209 -0.018 -0.089 0.003** 0.010** -0.110 0.013 -0.186 0.579** -0.003** 0.032 0.065 0.043 -0.059 (-2.01) (1.29) (-0.40) (-0.84) (-0.67) (0.02) (-0.98) (0.63) (-1.03) (2.50) (-0.05) (0.44) (0.82) (0.55) (-0.62) Public debt crisis -0.017* -0.347 0.025 -0.17 -0.431 0.002** 0.187 -0.007 1.278 -0.832*** -0.028** -0.091 -0.012* -0.130* 0.086
(-1.69) (-1.27) (0.28) (-0.60) (-0.67) (0.02) (0.23) (-0.36) (0.63) (-3.29) (-0.04) (-0.45) (-0.11) (-1.60) (0.34) Yield -0.043* 0.818** -0.154 0.47 0.011* 0.002*** 0.550 0.055 -1.555 -0.014 0.041* 1.040 -0.131 0.345 0.003*** (-1.70) (2.34) (-0.77) (0.74) (0.10) (0.00) (0.79) (0.35) (-1.26) (-0.26) (0.10) (0.44) (-0.96) (1.31) (0.00) Yield crisis 0.098 -0.529 -1.599 -3.717 -0.068 -0.014*** -4.966 -0.281* 28.365 -0.937** -1.830 1.107 1.675 -0.399* -1.178 (2.02) (-1.22) (-1.20) (-1.24) (0.18) (-0.00) (-0.21) (-1.73) (0.74) (-2.13) (-0.55) (0.44) (1.36) (-1.65) (-0.38) Dumcrisis -14.630* 1.370 7.200 19.490 1.159 0.129** 0.757 1.650* -265.540 11.240*** 6.497* -0.018*** -8.449 4.183* -0.39*** (-1.94) (1.31) (1.60) (0.88) (0.48) (0.02) (0.02) (1.87) (-0.68) (2.83) (0.12) (-0.00) (-1.26) (1.72) (-0.08) Observations 1863 203 128 194 1439 16734 61 58 689 551 97 69 136 72 151 Adjusted R Squared 0.44 0.08 0.63 0.03 0.16 0.48 0.54 0.65 0.42 0.59 0.75 0.67 0.26 0.86 0.18 F-statistic 6.60 2.50 8.05 2.31 2.04 9.51 3.14 4.45 1.74 9.65 7.40 4.14 2.63 22.22 6.15 Durbin-Watson 2.17 1.25 2.01 1.71 23.35 1.22 2.71 3.26 41.26 1.39 2.08 3.80 2.23 1.36 0.71
7 Cross section fixed effects are used for all models. Robust t-statistics are provided in the parentheses. All results are corrected for heteroskedasticity and autocorrelation. ***/**/* indicate significance levels at the 1%, 5% and 10%-levels, respectively (two-tailed)
