University of Groningen Faculty of Economics and Business
“Bank bailouts and moral hazard in the financial crisis: recent evidence
from the European banking sector “
Master Thesis
Submitted by Josefine Rottmann (s2040271) On 16 July 2012
Word count: 6,683
2 Table of contents Abstract ... 3 1 Introduction ... 3 2 Literature review... 5 3 Methodology ... 10 3.1 Econometric model ... 10 3.2 Identification ... 13
4 Data and descriptive statistics ... 14
4.1 Sample and events ... 14
4.2 Identifying and explanatory covariates ... 16
5 Results ... 18
5.1 Main results ... 18
5.2 Larger banks ... 21
5.3 PIGS interaction with bailout probability ... 23
6 Discussion and conclusion ... 25
Appendix ... 27
3 Abstract
In order to test whether government bailouts induce moral hazard effects in terms of excessive risk taking by banks, this study uses a two-step model to estimate whether a higher expected bailout probability increases bank risk taking. This research uses recent data from the European banking sector from 2001-2011, including bank bailouts and failures during the recent financial crisis. The results show that an increase in expected bailout probability of 1% on average increases the likelihood of being in distress by 1.8 basis points. Also, it is shown that the larger a bank, the more likely it is to receive a government bailout, corroborating the too-big-to-fail argument.
Key words: banks, bailouts, government intervention, crisis, moral hazard, risk taking
1 Introduction
This paper analyses the impact of government bailouts and banks’ bailout expectations on
bank risk taking, in order to examine whether these bailouts induce moral hazard. I use 2001-2011 data from banks in 15 European countries, including bailouts and bank failures during the 2008-2009 financial crisis. In a two-step model, bailout probabilities and their impact on bank risk, measured as distress, are estimated. The results suggest that bank bailouts create moral hazard.
4
into a global economic crisis. The banking industry has been at the center of this crisis since the beginning and is now in the focus of public and political debate and scrutiny.
Particularly in Europe, the latest developments are revealing the depth of the crisis. The European debt crisis is still unfolding and the recent announcement of the large bailout sums Spain will need to rescue its banks cause even more uncertainty. In many European countries, such as the United Kingdom and Germany, the government has stepped in and rescued banking institutions with capital injections and large-scale rescue packages. Such government interventions may have helped to save some banks and avoid their exit from the market in the short term. However, the impact of such bailouts on bank risk taking is disputed, especially in the long term.
There is an ongoing debate among scholars regarding the impact of large-scale government bailouts on bank risk taking and the stability of the financial sector. In particular, it is debated whether such bailouts create moral hazard in banks. While there is a lot of empirical evidence from times prior to the recent crisis, there is still a lack of new evidence including the recent wave of bailouts. This study attempts to close this gap. Based on recent data from the financial crisis, the results show that an increase in bailout expectations by 1% on average increases the likelihood of being in distress by 1.8 basis points, which shows that bailouts do induce moral hazard. This study thus contributes new evidence to the moral hazard discussion.
5
previously presented ones and thus helps to get a better overview of the situation of the banking sector in Europe during the financial crisis.
This paper further contributes evidence supporting the “too-big-to-fail” argument, by
showing that moral hazard is stronger for larger banks. If such banks are in distress, governments step in to rescue them and avoid their exit from the market as such a failure would cause high costs. Bank size is positively significant and robust in predicting bailout probabilities and distress likelihoods. This shows that larger banks are both prone to take more risk and more likely to be bailed out by the government.
In the remainder of this paper, I will discuss the theoretical background concerning bank bailouts and moral hazard in Section 2, then present the methodology and the data used in Sections 3 and 4 to then analyze and discuss the results in Section 5 and conclude in 6.
2 Literature review
In the past five years, the financial sector has witnessed the “largest scope of government interventions” since the Great Depression (Hryckiewicz, 2012:2). The total amount of
government bailouts is estimated to $7.77 trillion in the United States, as of March 2009, and around $1 trillion in Europe, as of May 2010 (Wall Street Journal, 2010; Bloomberg, 2011). The recent developments in Spain and other Southern European countries give reason to anticipate even more bank bailouts in the near future.
6
Kirkpatrick, 2009; Mülbert, 2009). Banks have become notorious for incentive schemes rewarding high levels of risk taking by their managers (Bebchuk and Fried, 2003; Bebchuk and Spamann, 2009; Kirkpatrick, 2009). Bank incentive schemes allowed bank managers to enjoy the “full upside benefits of their risk-taking” while being shielded from the “downside exposure” (Dowd, 2009:153). These “perverse incentives” and the excessive
risk taking of banks are at the heart of the discussion surrounding the financial crisis (Bebchuk and Fried, 2003:21). Even though the United States, Germany and other Western countries are considered to have “some of the best corporate governance systems in the world”, these systems could apparently not avert excessive risk taking, and the corporate
governance of banks seems to have failed (Shleifer and Vishny, 1997:737). Recent regulations, such as Basel III, aim at preventing such excessive risk taking in the future by imposing stricter capital and liquidity requirements as well as improved transparency and disclosure (Bank for International Settlements, 2011).
7
Keynesianism, as many governments have intervened in the banking sector through bailouts (Dowd, 2009).
However, the actual impact of such bailouts on moral hazard and bank risk taking is controversial and there is an ongoing debate among scholars in the field as to whether or not government bailouts actually help to stabilize the financial sector and decrease bank risk taking, especially in the long run (Hryckiewicz, 2012). The issues surrounding moral hazard in the banking industry are at the heart of “the controversy over the causes of the present crisis” (Dowd, 2009:142). The moral hazard concept is “central to every policy discussion” regarding a financial crisis which justifies the importance of empirical research
testing the potential moral hazard effects of government interventions in the financial sector in times of crisis (Summers, 2007).
Bailout policies naturally have two “mutually offsetting effects”, i.e. the moral hazard
effect, which incentivizes banks to take more risk as they expect to be bailed out, and could hence cause distress, and the “risk-reducing value effect”, which in turn increases individual banks’ chance of survival (Cordella and Yeyati, 2003:301). Also, it is disputed how liberal governments and central banks, also called “lender of last resort”, should be in giving “access to last resort support” to distressed banks (Cordella and Yeyati, 2003:300).
8
hence at risk” (Mayes, 2004:516). Especially the small depositors are exposed to the risks
and do not have the power to monitor and protect themselves against excessive risk taking by the bank managers (Mülbert, 2009).
Furthermore, all banks, also the sound and solvent ones, can be subject to bank runs, which stem from a lack of trust in the bank’s solvency and stability (Kaufman, 1988; Mayes,
2004; Mülbert 2009). Such bank runs are considered as contagious and thus a run on one individual bank could not only cause the failure of other banks, but potentially destabilize the entire financial system (Kaufman, 1988). Therefore, in order to decrease the risk of such runs, government interventions can help to restore and safeguard trust and stability in the banking system.
Moreover, individual bank failures can result in spillover effects which could destabilize the banking sector further and result in uncertainty and more failures (Mayes, 2004). Due to its high levels of interconnectedness, the banking system is characterised by “powerful propagation mechanisms” so that what starts out as a small shock can evolve into a large
crisis through spillovers (Heremans, 2007:5). Thus, concerns regarding bank contagion can outweigh the moral hazard concerns and justify bailouts (Gotthard and Huang, 1999). If the potential contagion effects are substantial the risk of such spillover effects can be reduced by government interventions, such as bank bailouts (Summers, 2007).
However, other scholars argue that bailouts create moral hazard and thereby lead to increased risk taking (Gale and Vives, 2002; Freixas et al., 2004; Dowd, 2009; Gropp, Gruendl and Guettler, 2010; Dam and Koetter, 2012). Dowd (2009:142) even argues that moral hazard is a “pervasive and inevitable feature of the financial system” and played a
9
when it is in distress, it has no incentive to reduce the likelihood of distress, i.e. its level of risk taking, and is therefore incentivized to take more risk in order to make higher potential profits. Government interventions take “pressure off the bad banker” and can thus be
destabilizing (Dowd, 2009:158). Likewise, always bailing out banks in distress would cause “severe moral hazard problems” (Dam and Koetter, 2011:5).
The very existence of bailout schemes and legal frameworks for supporting distressed banks “will tend to change people’s actions” and the behavior of banks (Mayes, 2004: 518). The possibility of a bailout thus “indirectly encourages managerial shirking and risk taking”
(Gale and Vives, 2002:468). If this is the case, government bailouts have an adverse effect and they do not, as intended, stabilize the banking industry by reducing risk levels, but rather increase the risk appetite of banks. If a bank cannot expect to be bailed out, however, the threat of being closed can serve as an incentive for banks to be more prudent and not take too much risk (Suarez, 1994). It is important that banks “have enough at stake that they
fear the consequences” and thus try to avoid excessive risk levels (Mayes, 2004:544).
It has also been argued that some banks will be rescued mainly due to their sheer size (Financial Crisis Inquiry Commission, 2010; Moosa, 2010; Morrison, 2011). Particularly large banks can expect to receive a bailout when in distress as “policy makers had little alternative” but to rescue banks that were considered as ‘too big to fail’ (Morrison, 2011:
498). Moreover, in the case of widespread banking crises, the regulator may also face a “too-many-to-fail problem” if the number of banks in distress is so large that the regulator
10
chances of being bailed out increase (Acharya and Yorulmazer, 2007:2). In both cases, ‘too big to fail’ and ‘too many to fail’, there is moral hazard, as the banks can expect to be bailed out by the government when they are distressed.
The contradicting arguments against and in favor of government bailouts show that there seems to be a trade-off between the social cost of bank failures and the moral hazard effects of government bailouts (Cordella and Yeyati, 2003:301). These opposite effects have to be weighed against each other when discussing the appropriate policy responses to a banking crisis. If it can be shown that bailouts do indeed increase moral hazard and thus lead to more risk taking, the currently widespread policy approach of bailing out banks in distress on a large scale could be counterproductive in the long run. Large-scale government interventions are thus only justified if they lead to stability “without increasing significant risk afterwards” (Hryckiewicz, 2012:1). Understanding the impact of bailout policies on the
banking industry is thus essential in order to find the appropriate policies to handle banks in distress. This is even more so important as “the frequency of severe banking crises has increased significantly” in the past decades (Hellmann, Murdock and Stiglitz, 2000:147).
3 Methodology
3.1 Econometric model
11
bailed out by the government were in distress, thus these two variables are naturally endogenous. As I am trying to find out whether the expected bailout probability of a bank influences the bank’s risk taking and these two variables are endogenous, I use covariates that are orthogonal to bank risk taking in order to alleviate endogeneity problems. Thus, first, bailout probabilities are estimated using bank-specific and macroeconomic variables, but also country-level political variables which explain bailout decisions of the government, but do not explain bank distress. Then, these predicted bailout probabilities are used to estimate whether they increase the banks’ risk taking, measured as distress. If it can be
shown than an increased bailout probability increases risk taking by a bank, moral hazard exists.
The methodology is based on the approach and econometric model used by Dam and Koetter (2011). In particular, an estimation in two steps was conducted. In the first step, bailout probabilities for all banks, distressed and sound ones, are predicted using a set of covariates and bank characteristics. As the regulator will try “to be unpredictable” and is
very unlikely to announce ex-ante which sort of banks it will bail out and which not, bailout probabilities cannot be predicted as such (Dam and Koetter, 2011:5). They can, however, be inferred from observing past bailouts. Past bailouts, combined with the characteristics of the individual banks that were rescued, can help to infer a certain bailout probability for all banks. Thus, the first equation estimates bailout probabilities based on bank characteristics and other covariates. These bailout probabilities, , are predicted for all banks, sound and
distressed ones, according to the following bailout equation (Dam and Koetter, 2011).
12
In the second step, the predicted bailout probabilities from the bailout equation are then
related to a bank’s risk-taking behavior, i.e. the bank’s likelihood of running into distress.
The equation for estimating the distress likelihood is thus the following (Dam and Koetter, 2011).
' 1
zit Dit it X it (distress) (2)
With these two equations, it can be analysed whether the predicted bailout probabilities from the first step affect the distress likelihoods in the second step. If it can be shown that the expected bailout probability estimated with in the first equation affects the level of risk taking of banks, estimated in the second equation, this reveals that bailouts create moral hazard.
The number of observations used for both equations is different, as the bailout equation calculates bailout probabilities based on those banks that were bailed out and their characteristics. Thus, in the bailout equation, the number of observations used equals the number of bailouts in the dataset. In the distress equation, the distress likelihood is predicted using all banks in the sample.
13 3.2 Identification
Country-level political covariates are used as identifying covariates, as they explain government bailouts but do not directly affect the risk taking of banks. They thus are orthogonal to bank risk taking and serve to alleviate the endogeneity problem. Brown and Dinç (2011) argue that political variables are important as they have an effect on government decisions whether or not to rescue banks. Such variables can thus help to predict bailout probabilities. Therefore, I specify two identifying covariates that are related to the probability of a bailout, but not to the likelihood of bank distress. These two covariates are the event of an upcoming election, i.e. election year, and the political orientation of the current government.
The “political stance” of the government is included as it is likely to impact upon the incumbent government’s willingness to rescue distressed banks (Dam and Koetter, 2011:
13). More conservative, right-wing governments tend to be more in favor of business in general, and of injecting capital into large-scale bank rescue packages in particular, while more socialist, left-wing governments tend to be more reluctant to support banks with taxpayers’ money. The political orientation of the governments in the different European
countries has thus been included for each year, labeled as ‘-1’ for right/conservative/center-right governments and ‘+1’ for left/socialist/center-left.
Similar to Brown and Dinç (2011), I also include an election year variable, as the event of an upcoming election can be expected to impact upon the probability of a bailout. The prospect of a change in government could affect the banks‘ expectations regarding bailouts
14
political orientation of the government might change, which can cause uncertainty and impact upon banks‘ bailout expectations and hence their risk-taking behavior. An election
year dummy has been included to account for this.
Combined, these two identifying variables map the political situation in the different countries and can help to explain bailout probabilities without directly influencing bank distress. They are thus included in the first equation in order to predict bailout probabilities.
4 Data and descriptive statistics
4.1 Sample and events
The sample consists of all banks in 15 European countries1. The bank dataset was retrieved from Bankscope and includes 7105 banks and their financial figures from 2001-2011. This dataset builds the basis for the analysis. In the 15 countries, a total of 45 banks were bailed out by the government and 18 bank failures were recorded during the recent financial crisis2. Table 1 gives a detailed overview of bank bailouts and failures (see appendix).
When a bank is in distress, two events can occur: it can be bailed out or it can fail and exit the market. For the purpose of this research, bank bailouts and failures are defined as follows. A bailout is defined as a government intervention in order to rescue a bank in distress, such as capital injections, recapitalisations, large-scale rescue packages, purchases
1 Countries in which bank bailouts or failures could be observed during the 2008-2009 crisis: Austria,
Belgium, Denmark, Finland, France, Germany, Great Britain, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, and Sweden.
2 The information regarding bank bailouts and failures were gathered from various sources and does not claim
15
of bank shares or the nationalisation of a bank. These events show that the government is willing to help the bank and avoid its exit from the market. It is assumed that a distressed bank would cease to exist and exit the market without the government’s intervention. A
bank failure is then defined as an exit from the market, either in form of an acquisition by another bank or in form of a liquidation of the bank. In this case, the government does not step in and the bank disappears from the market. Table 2 presents the number of bailouts and failures during the crisis years. Note that no bailouts or failures from years prior to the recent financial crisis were included, as the focus of this study in on the most recent events. As can be seen in Table 2, around half of all bailouts occurred in 2008, and around one third in 2009. Two thirds of all failures were recorded in 2008 and 2009. This illustrates that the years 2008 and 2009 were the most severe years of the crisis.
Table 2
Events: bailouts and failures during the crisis years
Year Distress Bailout Failure
16
The event “distress” thus includes the events “bailout” and “failure” and means that the bank fails to exist as a going concern without any intervention by the government. Although various other measures of bank risk exist in the literature, they all capture different types of risk, such as liquidity risk, credit risk or market risk. But ultimately, the only concern for the regulator is whether the existence of a bank is endangered. As such, the event of being in distress measures the ultimate risk faced by a bank, namely that it ceases to exist.
4.2 Identifying and explanatory covariates
Several covariates have been selected in order to estimate bailout probabilities and distress likelihoods in the model. In order to alleviate endogeneity problems, country-level political variables have been selected as identifying covariates. These do not affect the risk levels of banks directly, but help to explain the probability of bailouts in these countries. These identifying covariates are the event of an upcoming election (election year dummy) and the political orientation of the incumbent government (political orientation government), as explained in section 3.2. Moreover, several explanatory covariates are used, which can be seen in Table 3.
On the bank level, two covariates indicating the banks‘ financial health and risk level are
17
position, more precisely its capacity to meet time liabilities and other risks (Bankers Almanac). Total assets is used to measure bank size.
Macroeconomic country-level covariates are used in order to explain bailout probabilities and distress likelihoods. The general state of the economic macro-environment in which banks operate can be expected to influence their behavior and their likelihoods of running into distress. Similar to Dam and Koetter (2011), the GDP growth rate and the unemployment rate of the 15 countries are used as macroeconomic covariates, as they may “influence both bailouts and risk-taking” (Dam and Koetter, 2011:15). As can be seen in
Table 3, both the GDP growth rates and the unemployment rates varied significantly. In 2008 and 2009, the GDP growth rates dropped considerably across Europe with, for example, growth rates of -3.5% in the Netherlands and -4.4% in Great Britain in 2009. Likewise, unemployment rates increased reaching the highest levels in Ireland and Spain with 18.4% and 29.2% in 2011, respectively. Combined, the bank-level and macroeconomic variables are used in the model to explain bailouts and predict distress likelihoods.
Table 3
Descriptive statistics explanatory variables
Variables Mean SD Min Max N
Bank-level variables
Total assets (th EUR) 13067812 92204163 72 2164103000 78155
Total capital ratio (%) 18,04 20,99 -5 869 15120
Equity/Total assets (%) 11,26 17,69 -775 138,02 53412
Macroeconomic variables
GDP growth rate (%) 1,25 2,35 -8,4 6,6 78161
Unemployment rate (%) 9,24 3,41 3,6 29,2 48874
18 5 Results
5.1 Main results
Table 4 shows the results of the two-step estimation of the bailout and distress equations. Specifications [1] and [2] show the probit estimates and specifications [3] and [4] show the OLS regression results. Probit and OLS have been conducted using both bank-level covariates, i.e. total capital ratio (Probit 1 and OLS 1) and equity/total assets (Probit 2 and OLS 2). The first column of each estimation shows the results of the bailout equation and the second column the results of the distress equation. The coefficient of determination, (pseudo) R², is around 32% on average, indicating a good fit of the equations. In line with Petersen (2009), standard errors are clustered on country level.
19 Table 4
Two-stage estimations of bailout and distress probabilities
Specification [1] Probit 1 [2] Probit 2 [3] OLS 1 [4] OLS 2
Equation Bailout Distress Bailout Distress Bailout Distress Bailout Distress
Explanatory covariates
Bailout probability 0.447 0.647*** 0.018 0.003
(1.370) (3.398) (1.099) (1.220)
Total Assets (th EUR) (log) 0.559*** 0.207*** 0.557*** 0.225*** 0.140*** 0.000 0.136** 0.001**
(3.152) (2.852) (2.740) (3.862) (3.809) (0.072) (3.014) (2.245)
Total capital ratio (%) -0.099 0.002*** 0.001 0.000
(-1.199) (2.710) (0.068) (0.272) Equity/total assets (%) -0.291*** 0.004** -0.003 0.000** (-3.196) (1.971) (-0.228) (2.614) GDP growth rate (%) -0.011 -0.097*** -0.046 -0.094*** -0.010 -0.001** -0.008 -0.000*** (-0.070) (-3.364) (-0.244) (-3.830) (-0.263) (-2.719) (-0.237) (-3.222) Unemployment rate (%) 0.072 -0.026* 0.136** -0.020** 0.022 -0.001* 0.027** -0.000** (1.498) (-1.940) (2.460) (-2.127) (1.645) (-1.798) (2.423) (-2.368)
Election year (dummy) -0.120 -0.526 -0.053 -0.072
(-0.137) (-0.449) (-0.224) (-0.291)
Political orientation government -0.325** -0.658** -0.040 -0.055
(-2.225) (-2.401) (-0.714) (-0.900)
Consolidated -9.180** -6.105*** -9.852** -6.732*** -2.106** 0.008 -2.051* -0.011*
(-2.428) (-4.450) (-2.308) (-5.779) (-2.287) (0.253) (-2.053) (-1.813)
(Pseudo) R² 0.459 0.278 0.505 0.348 0.493 0.014 0.472 0.008
Number of observations 39 12,455 45 40,040 39 12,455 45 40,040
Note: */**/*** denote the significance at the 10%/5%/1% levels respectively.
20
[4] help to further understand the economic significance of the results. As can be seen in specification [3], a 1% increase in the expected bailout probability increases risk taking, measured as the likelihood of distress, by 1.8 basis points. This is an economically significant increase in risk taking and is result of moral hazard.
Furthermore, it can be seen in specifications [1] and [2] that bank size, measured as total assets, is positive and statistically significant in predicting bailout probabilities and distress likelihoods, which indicates that the size of a bank has a significant impact on its expected bailout probability. This finding is robust across the different estimations. These results confirm that bigger banks are more likely to receive a bailout, supporting the too-big-to-fail argument. This shows that larger banks are prone to take more risk as they are more likely to be bailed out by the government. This is evidence in favor of moral hazard.
The effect of the election year dummy and the political orientation of the government, the political variables used in the first equation to predict bailout probabilities, on bailout probabilities is negative. Although the individual significances of the election year dummy may not seem high, it is the joint significance of both variables that is of interest, as these two covariates combined indicate the political situation in the different countries.
21
increases the probability of being bailed out when in distress. Likewise, having a higher equity-total-assets ratio decreases the bailout probability as banks with better capital adequacy are less likely to run into distress in the first place.
GDP growth rate, one of the economic country-level indicators, has a significant and negative effect on distress likelihoods in all four estimations, which shows that the overall economic situation in the country is an important factor in predicting bank distress. If the economy is shrinking, the likelihood of banks to run into distress increases. The unemployment rate plays a less significant role in predicting distress likelihoods, which could be explained by the fact that unemployment is a lagging indicator and takes some time to react to an economic shock, such as the recent crisis. Both GDP growth and unemployment rate seem to have little or no effect on predicting bailout probabilities, which suggests that the government’s decision whether or not to rescue distressed banks
depends more on other factors.
5.2 Larger banks
Furthermore, in order to examine whether the size of a bank, measured as total assets, does impact upon its likelihood of being bailed out, I estimated bailout and distress probabilities for subsets of larger banks ranked by total assets from the original sample, i.e. the top 1000, top 500 and top 200 largest banks. If it can be shown that for these banks, the impact of the expected bailout probability on distress, i.e. bank risk taking, is more significant, this would support the too-big-to-fail argument.
Table 5
Two-stage probit estimations of bailout and distresss probabilities for larger banks
Specification [1] All banks [2] Top 1000 banks [3] Top 500 banks [4] Top 200 banks
Equation Bailout Distress Bailout Distress Bailout Distress Bailout Distress
Explanatory covariates
Bailout probability 0.447 0.532 0.707* 2.444***
(1.370) (1.233) (1.892) (2.746)
Total Assets (th EUR) (log) 0.559*** 0.207*** 0.559*** 0.296*** 0.559*** 0.289*** 0.559*** 0.129
(3.152) (2.852) (3.152) (3.051) (3.152) (2.955) (3.152) (1.370)
Total capital ratio (%) -0.099 0.002*** -0.099 0.001 -0.099 0.008 -0.099 0.058
(-1.199) (2.710) (-1.199) (0.224) (-1.199) (0.698) (-1.199) (1.388)
GDP growth rate (%) -0.011 -0.097*** -0.011 -0.137*** -0.011 -0.145*** -0.011 -0.163***
(-0.070) (-3.364) (-0.070) (-6.044) (-0.070) (-6.787) (-0.070) (-5.493)
Unemployment rate (%) 0.072 -0.026* 0.072 -0.021 0.072 -0.026* 0.072 -0.061**
(1.498) (-1.940) (1.498) (-1.260) (1.498) (-1.873) (1.498) (-2.048)
Election year (dummy) -0.120 -0.120 -0.120 -0.120
(-0.137) (-0.137) (-0.137) (-0.137)
Political orientation government -0.325** -0.325** -0.325** -0.325**
(-2.225) (-2.225) (-2.225) (-2.225)
Consolidated -9.180** -6.105*** -9.180** -7.975*** -9.180** -8.035*** -9.180** -6.715***
(-2.428) (-4.450) (-2.428) (-4.646) (-2.428) (-4.264) (-2.428) (-3.023)
Pseudo R² 0.459 0.278 0.459 0.238 0.459 0.212 0.459 0.191
Number of observations 39 12,455 39 2,856 39 1,740 39 787
specifications [2], [3] and [4] show the results for the three subsets of larger banks. It can be observed that the significance of the bailout probability in predicting distress becomes more significant for the larger banks. It is significant at the 10% level in specification [3], i.e. for the largest 500 banks, and significant at the 1% level in specification [4], i.e. for the largest 200 banks. Thus, the larger the bank, the higher the significance of the expected bailout probability in predicting bank distress. This result supports the too-big-to-fail argument and is due to moral hazard in larger banks.
5.3 PIGS interaction with bailout probability
Moreover, in order to examine whether there are differences in the moral hazard effects between Northern and Southern European countries, the interaction between bailout probability and the PIGS countries, i.e. Portugal, Italy, Greece and Spain, is tested. For this purpose, I created a PIGS dummy variable as well as an interaction term using the PIGS dummy and the bailout probability. For interactions, OLS regressions are more useful, as probit interaction terms are very difficult to interpret (Ai and Norton, 2003). Table 6 summarizes the OLS regression results of the interaction of the PIGS countries with the bailout probability. For completeness, the results of the probit estimation of the interaction are reported in Table 7 (see appendix).
24
effect in the PIGS countries is less strong. This indicates that where a bank is located has an impact on bailout probabilities and it is thus important to include country-level variables into analyses of bank bailouts.
Table 6
Two-stage OLS estimation of interaction between PIGS and bailout probability
Specification [1] ALL [2] PIGS
Equation Bailout Distress Bailout Distress
Explanatory covariates
Bailout probability 0.003 0.001
(1.220) (0.212)
Total Assets (th EUR) (log) 0.136** 0.001** 0.136** 0.001**
(3.014) (2.245) (3.014) (2.458) Equity/total assets (%) -0.003 0.000** -0.003 0.000* (-0.228) (2.614) (-0.228) (2.082) GDP growth rate (%) -0.008 -0.000*** -0.008 -0.000*** (-0.237) (-3.222) (-0.237) (-3.170) Unemployment rate (%) 0.027** -0.000** 0.027** 0.000 (2.423) (-2.368) (2.423) (0.449)
Election year (dummy) -0.072 -0.072
(-0.291) (-0.291)
Political orientation government -0.055 -0.055
(-0.900) (-0.900)
PIGS -0.002**
(-2.468)
Interaction PIGS and bailout probability -0.006**
(-2.360)
Consolidated -2.051* -0.011* -2.051* -0.019*
(-2.053) (-1.813) (-2.053) (-2.138)
R² 0.472 0.008 0.472 0.009
Number of observations 45 40,040 45 40,040
Note: */**/*** denote the significance at the 10%/5%/1% levels respectively.
25 6 Discussion and conclusion
This study analyses whether government bailouts induce moral hazard effects in terms of excessive risk taking by banks. Based on the methodological approach used by Dam and Koetter (2011), I use a two-step model to predict bailout probabilities and their effect on bank risk taking, measured as the likelihood of bank distress. In the first step, the bailout equation, bailout probabilities are predicted based on bank characteristics and past bailouts. These predicted bailout probabilities are then used in the second step, the distress equation, to predict distress likelihoods for all banks. Together, these two equations test how expected bailout probabilities affect the risk taking behavior of banks and whether bailouts induce moral hazard in banks.
26
Furthermore, the results support the “too-big-to-fail” argument as bank size, measured as
total assets, is positively significant in predicting both bailout probabilities and bank distress. Also, the results for the subsets of larger banks show that bank size has a positive impact on the expected bailout probability. This shows that larger banks are more likely to take more risk as they are more likely to be bailed out by the government when they run into distress. Hence, moral hazard is stronger for larger banks. Moreover, there are differences in the moral hazard effect between Southern European countries, the PIGS countries, and the other European countries, suggesting that country-level variables do have an impact on banks’ expectation of being bailed out when in distress. This indicates that
including such variables in the analysis of bank bailouts and moral hazard is viable and important.
27
To conclude, this study contributes significant results to the moral hazard discussion in the context of the recent financial crisis. Yet, there is still a need for further research and more empirical evidence in order to fully understand the effect of government bailouts on risk taking in the banking industry and the appropriateness of such policy responses to widespread bank distress. This is particularly important as the frequency of banking crises has increased significantly in the past decades.
Word count: 6,683
Table 1: Overview bailouts and failures during the crisis in Europe
GOVERNMENT BAILOUTS*
Bank name and country Date of (first) bailout SIFI Details
Aareal Bank AG DE Feb 09 0 Equity recapitalization and government guarantees by the SoFFin
ABN Amro NL Oct 08 0 Nationalised by the Dutch government after a failed takeover by a consortium consisting
of RBS Group, Santander and Fortis
Allied Irish Bank IE Sep 08 0 Recapitalisation announced by the government in December 08; rescue package for the Allied Irish Bank and Bank of Ireland arranged by the government in February 2009; in Nov 10 the Irish government had to seek financial support from the European Central Bank as well as the IMF to sustain the bailout and guarantees for BoI and other Irish banks
Anglo-Irish Bank IE Jan 09 0 First nationalised under the Anglo Irish Bank Corporation Act, then merged with Irish Nationwide Building Society (new company name: Irish Bank Resolution Corporation), to be liquidated over a period of ten years, will thus exit from the market
Banca Monte dei Paschi di Siena
IT Dec 09 0 Oldest bank in Europe received two bailout packages, one in late 2009 and a second, even bigger one in June 2012, of about €3.9 billion
Bank of Ireland IE Sep 08 0 Recapitalisation announced by the government in December 08; rescue package for the Allied Irish Bank and Bank of Ireland arranged by the government in February 2009; in Nov 10 the Irish government had to seek financial support from the European Central Bank as well as the IMF to sustain the bailout and guarantees for BoI and other Irish banks
Bank of Scotland GB Oct 08 0 Was a part of HBOS, which was bailed out in the 2008 UK bank rescue package and eventually taken over by Lloyds TSB (now Lloyds Banking Group)
Bankia ES May 12 0 Nationalisation announced on 9 May 2012, 4th largest banking group in Spain
Banque Populaire
FR May 09 0 Capital injection through the government’s recapitalization scheme. The bank later merged with Caisse d'Epargne in 2009 (not a forced merger).
Bayerische Landesbank
DE Jan 09 0 Government guarantees by the SoFFin
BNP Paribas FR Mar 09 1 Government purchased preference shares worth €5.1 billion
Bingley deposits and branch network sold to Abbey National (owned by Grupo Santander), B&B branches rebranded Banco Santander in 2010
Caja de Ahorros Castilla La Mancha
ES Mar 09 0 Banco de Espana took over management of the bank in March 2009, board dismissed,
guarantee fund of €9 billion provided by the government; this is the first such bailout in Spain during the crisis
CajaSur ES May 10 0 Banco de Espana took over the bank's management, fired the board, saved with the
Spanish bank rescue fund (FROB), second such bailout in Spain since the crisis
Commerzbank DE Jan 09 1 Equity recapitalisation and guaranteed issues through the SoFFin
Crédit Agricole FR Oct 08 1 Capital injection by the government
Crédit Mutuel FR Oct 08 0 Repayment of bailout announced in October 2009
Danske Bank DK May 09 0 Received bailout through the Danish government’s bank rescue package
Dexia BE Oct 08 1 Received bailout from the Belgian, French and Luxembourg governments in Oct 08, split
up in late 2011; Belgian part (Dexia Banque Belgique) acquired by the Belgian
government and renamed (Belfius Banque & Assurances); government guarantees from the French, Belgium and Luxembourg governments worth € 90 billion; and € 95 billion worth into a bad bank
EBS IE Apr 10 0 The Irish government injected €875 million to recapitalize the bank
Fortis BE Oct 08 0 Government bailout first, then sold in parts Groupe Caisse
d’Epargne
FR May 09 0 Capital injection through the government’s recapitalization scheme. The bank later merged with Banque Populaire in 2009 (not a forced merger)
HBOS GB Oct 08 0 2008 UK bank rescue package; government bailout first, then taken over by Lloyds in January 2009
HSH Nordbank DE Jan 09 0 Government guarantees (SoFFin)
Hypo Real Estate DE Mar 09 0 Equity recapitalisation and guaranteed issues through the SoFFin; bailed out in 2008, nationalised in 2009
IKB DE Jan 09 0 Government guarantees (SoFFin); taken over by Lone Star in 2008 (KfW sold its 90%
stake to Lone Star)
ING Groep NL Oct 08 1 Capital injection plan by the Dutch Government to increase Tier 1 capital, repayment to be completed by 2013
KBC Group BE May 09 0 Loan guarantee by the Belgian government
Krajbanka LV Nov 11 0 Nationalised first, will be liquidated due to bankruptcy of Snoras (majority owner of Krajbanka)
Lloyds TSB GB Oct 08 1 2008 UK bank rescue package; UK Government acquired 43.4% stake in the Group's ordinary voting shares and 4 billion GBP of preference shares
National Westminster
GB Oct 08 0 Part of RBS Group, in which the UK government took a large stake in Oct 08 in order to recapitalise the Group
Natixis FR Aug 09 0 Guarantee on toxic assets from partially state-owned parent
Nordbanken SE Oct 08 0 Swedish government banking rescue (government announced it would guarantee all
banks and assumed bad bank debts)
Nordea Banken SE Oct 08 1 Swedish government banking rescue (government announced it would guarantee all
banks and assumed bad bank debts)
Northern Rock GB Feb 08 0 Suffered a bank run in 2008, taken into state ownership in February 2008, then split into two parts (assets and banking) in January 2010, the bank was then bought by Virgin Money in early 2012
RBS Group GB Oct 08 1 2008 UK bank rescue package; UK Government holds 84% stake
Roskilde Bank DK Aug 08 0 Acquired by Danmarks Nationalbank, 21 branches sold to other banks
Snoras LT Nov 11 0 Nationalised first, but could not be sustained, then declared bankrupt
Société Générale FR Oct 08 1 Government capital injection
WestLB DE Nov 09 0 Equity recapitalisation through the SoFFin
FAILURES
Bank name and country Date of failure SIFI Details Alliance &
Leicester
GB Oct 08 0 Taken over by Grupo Santander, renamed under parent brand Santander UK
Anglo-Irish Bank IE Jul 11 0 First nationalised under the Anglo Irish Bank Corporation Act, then merged with Irish Nationwide Building Society (new company name: Irish Bank Resolution Corporation), to be liquidated over a period of ten years, will thus exit from the market
Amagerbanken DK Feb 11 0 Assets transferred to Finansiel Stabilitet A/S (FSA), closed by administrators
Bank of Scotland GB Jan 09 0 Was a part of HBOS, which was bailed out in the 2008 UK bank rescue package and eventually taken over by Lloyds TSB (now Lloyds Banking Group)
Bank Trelleborg DK Jan 08 0 Acquired by Sydbank
Bingley deposits and branch network sold to Abbey National (owned by Grupo Santander), B&B branches rebranded Banco Santander in 2010
Capinordic Bank DK Feb 10 0 Declared insolvent, bankruptcy proceedings started in Feb 2010, assets transferred to FSA
Dresdner Bank DE May 09 0 Acquired by Commerzbank
DSB NL Oct 09 0 First Dutch bank to have collapsed, its bankruptcy was declared in October 2009
EiK Bank DK Sep 10 0 The bank became insolvent in 2010
Fionia Holding DK Feb 09 0 The banking activities in the current Fionia Bank are to be transferred to a new company which has been established and owned by Fionia Bank, but which is controlled by Financial Stability.
Fjordbank Mors DK Jun 11 0 Forced to close as it no longer met the FSA's increased solvency requirements; state-managed closure
Forstaedernes Bank
DK Oct 08 0 Acquired by Nykredit Bank
Fortis BE Oct 08 0 Government (from the Benelux governments) bailout first in Sep 08, then sold in parts in
Oct 08 with only the insurance business remaining
HBOS GB Jan 09 0 Taken over by Lloyds TSB (now Lloyds Banking Group) after having received a
government bailout in October 2008, now a wholly owned subsidiary of Lloyds Banking Group
IKB DE Oct 08 0 Government guarantees (SoFFin); then taken over by Lone Star in 2008 (KfW sold its
90% stake to Lone Star)
Krajbanka LV Nov 11 0 Nationalised first, will be liquidated due to bankruptcy
Northern Rock GB Nov 11 0 Bank run in 2008, nationalized in 2008 (had failed to find a commercial buyer), bought by Virgin Money in 2012, the NR brand is to be phased out and replaced by Virgin Money in 2012
Roskilde Bank DK Aug 08 0 Acquired by Danmarks Nationalbank, 21 branches sold to other banks, then liquidated
Snoras LV Nov 11 0 Nationalised first, but could not be sustained, then declared bankrupt
* banks in italics were first bailed out and eventually failed/exited the market
Table 7
Two-stage Probit estimation of interaction between PIGS and bailout probability
Specification [1] ALL [2] PIGS
Equation Bailout Distress Bailout Distress
Explanatory covariates
Bailout probability 0.647*** 0.472**
(3.398) (2.463)
Total Assets (th EUR) (log) 0.557*** 0.225*** 0.557*** 0.244***
(2.740) (3.862) (2.740) (4.490) Equity/total assets (%) -0.291*** 0.004** -0.291*** 0.004* (-3.196) (1.971) (-3.196) (1.920) GDP growth rate (%) -0.046 -0.094*** -0.046 -0.098*** (-0.244) (-3.830) (-0.244) (-4.326) Unemployment rate (%) 0.136** -0.020** 0.136** -0.006 (2.460) (-2.127) (2.460) (-0.424)
Election year (dummy) -0.526 -0.526
(-0.449) (-0.449)
Political orientation government -0.658** -0.658**
(-2.401) (-2.401)
PIGS -1.733***
(-7.634)
Interaction PIGS and bailout probability 1.655***
(4.788)
Consolidated -9.852** -6.732*** -9.852** -7.034***
(-2.308) (-5.779) (-2.308) (-6.387)
R² 0.505 0.348 0.505 0.358
Number of observations 45 40,040 45 40,040
Note: */**/*** denote the significance at the 10%/5%/1% levels respectively.
References
Acharya, V.V. & Yorulmazer, T. 2007. Too many to fail – An analysis of time-consistency in bank closure policies. Journal of Financial Intermediation, 16: 1-31.
Ai, C. & Norton, E.C. 2003. Interaction terms in logit and probit models. Economic letters, 80: 123-129.
Bankers Almanac. Total capital ratios. Available at
http://www.bankersaccuity.com/credit-risk/financial-data/financial-spreads/performance-ratios/total-capital/ (accessed 24 June
2012).
Bebchuck, L.A. & Fried, J.M. 2003. Executive compensation as an agency problem. The
Harvard John M. Olin Discussion Paper Series, Discussion Paper No. 421.
Bebchuk, L.A. & Spamann, H. 2009. Regulating bankers’ pay. The Harvard John M. Olin
Discussion Paper Series, Discussion Paper No. 641.
Bloomberg Markets Magazine. 2011. Secred Fed loans gave banks $13 billion undisclosed to Congress. Available at
http://www.bloomberg.com/news/2011-11-28/secret-fed-loans-undisclosed-to-congress-gave-banks-13-billion-in-income.html (accessed 01 July 2012).
Brown, C.O. & Dinç, I.S. 2011. Too many to fail? Evidence of regulatory forbearance when the banking sector is weak. Rewiev of Financial Studies, 24(4): 1378-1405.
Cordella, T., & Yeyati, E. L. 2003. Bank bailouts: Moral hazard vs. value effect. Journal of
Financial Intermediation, 12: 300-330.
Dam, L., & Koetter, M. 2011. Bank bailouts, interventions, and moral hazard. Bundesbank
Working Paper Series.
Dam, L., & Koetter, M. 2012. Bank bailouts and moral hazard: Empirical evidence from Germany. Review of Financial Studies, Advance Access published April 5, 2012. Demirgüc-Kunt, A. & Detragiache, E. 2002. Does deposit insurance increase banking system
stability? An empirical investigation. Journal of Monetary Economics, 49: 1373-1406. Dowd, K. 2009. Moral hazard and the financial crisis. Cato Journal, 29(1): 141-166.
Erkens, D., Hung, M., & Matos, P. 2009. Corporate governance in the recent financial crisis: Evidence from financial institutions worldwide. Working paper. UCLA.
Financial Stability Board. 2011. Policy measures to address systemically important financial institutions. Available at
http://www.financialstabilityboard.org/publications/r_111104bb.pdf (accessed 05 May
2012).
Freixas, X., Parigi, B. M., & Rochet, J. 2004. The lender of last resort: A twenty-first century approach. Journal of the European Economic Association, 2(6):1085–1115.
Gale, D., & Vives, X. 2002. Dollarization, bailouts, and the stability of the banking system. The
Quarterly Journal of Economics, 117(2): 467-502.
Goodhart, C.A.E. & Huang, H. 1999. A model of the lender of last resort. IMF Working Paper, WP/99/39.
Gropp, R., Gruendl, C., & Guettler, A. 2010. The impact of public guarantees on bank risk taking – evidence from a natural experiment. European Central Bank Working Paper Series, No. 1272.
Hellmann, T.F., Murdock, K.C. & Stiglitz, J.E. 2000. Liberalization, moral hazard in banking, and prudential regulation: Are capital requirement enough? American Economic Review, 90(1): 147-165.
Heremans, D. 2007. Corporate governance issues for banks. A financial stability perspective.
Working paper.Leuven.
Hryckiewicz, A. 2012. Government interventions – restoring or destructing financial stability in the long run? Wharton Financial Institutions Center Working Paper 12-02.
International Monetary Fund. 2009. Global Financial Stability Report April 2009. Responding to the financial crisis and measuring systemic risk. Washington, DC: International Monetary Fund.
Kaufman, G.G. 1988. Bank runs: Causes, benefits, and costs. Cato Journal, 7(3): 559-594. Kirkpatrick, G. 2009. The corporate governance lessons from the financial crisis. Financial
Market Trends. OECD.
Mayes, D.G. 2004. Who pays for bank insolvency? Journal of International Money and
Finance, 23: 515-551.
Morrison, A.D. 2011. Systemic risks and the ‘too-big-to-fail’ problem. Oxford Review of
Economic Policy, 27(3): 498-516.
Petersen, M. A. 2009. Estimating standard errors in finance panel data sets: Comparing approaches. Review of Financial Studies, 22 (1): 435-480.
Shleifer, A., & Vishny, R. W. 1997. A survey of corporate governance. The Journal of Finance, 52(2): 737-783.
Suarez, J. 1994. Closure rules, market power and risk-taking in a dynamic model of bank behaviour. LSE Financial Markets Group Discussion Paper, No. 196.
Summers, L. 2007. Beware moral hazard fundamentalists. Financial Times. Available at
http://www.ft.com/cms/s/0/5ffd2606-69e8-11dc-a571-0000779fd2ac.html#axzz20gE8pwPi (accessed 28 June 2012).
Wall Street Journal. 2012. Euro bailout: What is another trillion if you’ve already spent $1 trillion? Available at
