The characteristics of banks involved in M&A activiteis in the European banking sector : a comparison between the pre-crisis and crisis period

(1)

The characteristics of banks involved

in M&A activities in the European

banking sector

A comparison between the pre-‐crisis and crisis

period

Nynke Osinga

(2)

Table of C

ONTENTS

1 Introduction ... 2 2 Literature review ... 3 Diversification ... 4 Profitability ... 4 Efficiency ... 4

Loan portfolio and lending strategy ... 5

Liquidity and shock sensitivity ... 6

Size ... 7

Crisis ... 8

3 Data ... 9

Context, setting, background ... 9

European banking sector ... 9

Crisis ... 10

Regulation ... 12

Data sample ... 15

4 Empirical model ... 16

Multinomial logistic model ... 16

Variables ... 17 Regression ... 19 5 Goodness of fit ... 20 6 Results ... 21 Acquisitions ... 21 Mergers ... 24 7 Robustness tests ... 26 Acquisitions ... 26 Mergers ... 28 8 Discussion ... 29 9 Conclusions ... 31 10 Reference list ... 32 11 Appendix ... 35

(3)

1 Introduction

Mergers and acquisitions between banks have been a topic of interest for a long time now. However research on the topic has not been conclusive. On January 20th_{of 2013} the European Central Bank stated that the number of banks is decreasing rapidly. According to the ECB this is not only the result of insolvencies, but also; “quite often the consequence of banking sector mergers with a view to enhancing profitability” (ECB, 2013).

During the last decade the number of banking mergers and acquisitions has accelerated (Bankscope). The motives or reasons announced by the parties involved are often lower costs and growth opportunities. However, research on the benefits of M&A fails to find empirical evidence of these advantages (Focarelli et al., 2002, Berger et al., 1999). This raises questions about why banks engage in mergers and acquisitions. Motivated by that, this paper will research the motives for mergers and acquisitions, by analysing the characteristics of the banks involved.

Ex post research analyses the effects after a merger or acquisition; in ex ante research the presumed effects are analysed before a merger or acquisition. Most studies are an ex post analyses of the performance of banks evolved in an M&A deal. However Focarelli et al. (2002) focus on the ex ante performance. The paper ‘ Why do banks merge?’ (Focarelli et al., 2002) analyses the ex ante characteristics of Italian banks as a determinant for an M&A deal. The ex ante characteristics are then related to the subsequent performance of the deal. The results show that the main objective for mergers is to expand revenues from financial services by selling more services. For acquisitions the main objective is to improve the quality of the loan portfolio of the target bank, by transferring the superior lending competence of the acquiring bank to the target bank. The research performed in this paper will be similar to the ex ante research of Focarelli et al. (2002).

M&A is not only motivated by internal factors, it is also influenced by external factors. During periods of financial crisis M&As can be an important alternative to help resolve problems of financial distress. When banks are experiencing financial distress problems, M&As are often more efficient than bankruptcy or other means of exit due to the pre-existing franchise values of the merging firms (Berger et.al, 1999). In addition, regulators may also stimulate M&As during a financial crisis. The government may provide financial assistance or

(4)

bankruptcy.

For those reasons the motivations and characteristics of banks involved in a merger or acquisition might differ during times of crisis. This research will look at the differences of the likelihood of being a target or acquirer in an M&A activity before the crisis and during the crisis. The research question of this paper will be; what are the characteristics of banks involved in M&A activities in the European banking sector? A comparison between the pre-crisis and crisis period. To be able to answer that question the bank characteristics are analysed with the use of a multinomial logistic model. This paper adds to the existing literature because of the comparison between the periods before the crisis and during the crisis. Thereby, the existing literature focuses on a single country, while this research will focus on all European countries.

Given the difficult circumstances during the crisis and the on-going regulation the hypothesis is that banks will engage in M&As because they strive for risk

reduction through income diversification and loan portfolio improvement. This is because diversification lowers the volatility of profits and reduces risk, which might reduce capital requirements. Thereby the costs of bank supervision are tied to the perceived riskiness of the institution, so banks have additional incentives to reduce risk.

This paper has the following structure; first an overview of the existing literature is given in chapter 2, second the context, setting and background are illustrate in chapter 3 together with the dataset, chapter 4 demonstrates the empirical model, in chapter 5 the goodness of fit of the model is discussed, in chapter 6 the result are presented, chapter 7 provides a robustness check, chapter 9 briefly discusses some validity threats and chapter 8 contains the conclusions. The literature list and the appendix can be found at the end of this paper.

2 Literature review reasons for M&A in the banking sector

This chapter provides a background on the reasons and characteristics found in earlier research and literature.

(5)

2.1 Diversification

Diversification is one of the reasons for M&A. Earlier research has pointed out that M&A can result in considerable diversification gains (Lepetit et al, 2004). The presumed benefits of geographic or product diversification is that it will decrease the risk of the bank, by spreading the source of income (Liang and Rhoades, 1991). Especially in periods of crisis, it can be uncertain which services or skills will be requisite in the coming years as pointed out by Milbourn, Boot and Thakor (1999).

For the reasons mentioned above, banks with low diversification, might want to reduce their risk, by a merger or acquisition. Banks seeking for a diversification motivated M&A, might look for target banks with either a great number of services or a different product mix than the acquiring bank.

2.2 Profitability

One of the reasons to acquire or merge with a target bank might be because of its profitability. Banks that acquire banks with low profitability are presuming that they will be able to improve the performance of the target bank with better asset

management. Pasiouras, Tanna and Zopounidis (2007) formulate this as the

inefficient management hypothesis; indicating that the target banks have low profit due to inefficient management. Other earlier literature that is also consistent with the statement that profitability can be a motivation for an M&A, are the papers of

Focarelli et al. (1999), Wheelock and Wilson (2000) and Pasiouras and Zopounidis (2006); they find that banks with low profitability are more likely to be acquired. However, literature has not been conclusive, Hannan and Rhoades (1987) and

Pasiouras and Zopounidis (2006) did not find any empirical evidence for a correlation between profitability and the likelihood to be acquired.

2.3 Efficiency

Prior literature has shown that M&As can lead to efficiency increases. The efficiency is gained with economies of scale associated with a merger and acquisition, which leads to cost savings (Erel, 2011).

Overall, smaller and less efficient banks are acquired by or merged with larger, more efficient banks. This is consistent with the presumption that more efficient banks tend to transfer its superior knowledge and efficient management policies on to less efficient banks (Berger et al, 1999). In 2002 the European Central

(6)

Bank announced that small banks are mostly carried out to obtain a higher level of efficiency and in order to be big enough to survive (ECB 2002).

However, many cases of M&As did not bring any decrease in non-interest expenses nor provided evidence of efficiency gains. A possible explanation for this could be an increase in personnel expenses. When banks merge, the salary earned by the personnel working at one of the two banks often becomes the same, for that reason when the salaries of the target bank are lower, the acquiring bank increases those lower salaries. However, when the salaries of the personnel at the target bank are higher than the salaries at the buying bank, the higher salaries are seldom

decreased (Focarelli et al., 2002). That way an M&A is resulting in higher personnel costs, unless personnel becomes redundant and is fired.

2.4 Loan portfolio and lending strategy

For the European banking sector as a whole the loans-to-assets ratio in 2012 was 53.4% (EBF, ECB, 2013), which gives an indication of the importance of the loan portfolio and lending strategy of the banks.

The share of loans and quality of the portfolio of a bank could also affect the probability to be acquired or merged. Literature suggests that bank with a high

lending-to-asset ratios signal an assertive lending strategy, with a large customer base and an integrated market position; which could make the bank an attractive target. However, banks with a low lending-to-asset ratio signal a self-satisfied, conservative strategy, which could be made more profitable when implementing a more efficient strategy (Hannan and Rhoades, 1987).

Research demonstrates that target banks of acquisitions as well as mergers tend to have loan portfolios of poorer quality, witnessing a large number of bad loans compared to acquiring banks (Berger et al., 1999). A motivation for an M&A can be to implement a better lending strategy and improve the quality of the loan portfolio, resulting in a decrease of the share of bad loans and a reduction in lending to small-businesses. Due to the quality improvement of the loan portfolio consolidated banks might be permitted to lend more, without higher capital requirements (Hughes et al., 1999).

(7)

2.5 Liquidity & interbank market position

The liquidity position of banks could be part of the motivation leading to an

acquisition or merger. A bank with liquidity problems could be seeking for a merger of acquisition, in order to prevent a bank-run or bankruptcy (Pasiouras et al., 2011). However, Pasiouras et al. (2007) argue that a bank with excess liquidity may also be a good target, because the bank may experience a lack of investment opportunities. The motivation would be to increase the investment opportunities of the bank, while contributing to the overall liquidity of the two banks.

For the liquidity of a bank Pasiouras et al. (2007) are looking at the share of liquid assets divided by the customer & short term funding. A high percentage of liquid assets indicates a more liquid bank, which is better resistant against a bank-run. The study finds that banks with low liquidity are more likely to be a target, than banks with high liquidity.

The position of a bank on the interbank market could also influence the decision of a merger of acquisition. When a bank has lent more to other banks than it has borrowed from other banks; then the bank is a net creditor. Those banks are expected to be less sensitive to liquidity shocks and will thus be able to gain funding at low cost. This could make the bank an attractive target. Banks with a negative interbank balance are net debtors and are more exposed to liquidity shocks. They might want to diversify through a merger or acquisition, in order to reduce risk (Focarelli et al., 2002).

Focarelli et al. (2002) find that acquiring banks are relatively more sensitive to liquidity shocks, meaning that they are either a small net creditor or a net debtor. In case of a small creditor, this suggests that the bank is dynamic and has good

investment and lending opportunities and is therefore more likely to be part of an M&A. When the bank has a negative interbank balance, the bank has a poor liquidity position and for that reason wants diversify through an M&A.

Focarelli et al. (2002) also find that banks with low funding costs are indeed more likely to be a target in a merger or acquisition.

Ex-Post research on the performance of the bank after an M&A demonstrates that in the long-run the interbank balance increases as well as the lending-to-assets ratio (Focarelli et al., 2002). This suggests that the due to the improved liquidity, the bank was able to increase its lending.

(8)

2.6 Size

Large banks enjoy the benefits of economies of scale and economies of scope; they generally have better lending opportunities and can access funding against lower costs. As mentioned before, small banks might seek for a merger or acquisition, in order to grow in size such that they achieve the efficiency scale that is needed to be able to compete in the market (ECB, 2000).

Especially when the level of competition in market is high, the need to grow in size through M&A is high, because this might increase the market share (Focarelli et al., 2002). Another motivation for M&A could be the access to the financial safety net. The larger banks are the more important and have a bigger its impact on the financial stability; therefore in case of financial problems the government and

European Central Bank are more willing to safe the bank with their safety net (Berger et al., 1999). Thereby, when a large bank is perceived to be too big too fail, the bank might be considered to be less risky by debt- an shareholders, because they may presume that the government will protect the debt- and shareholders of the bank by providing guarantees (Berger et al., 1999).

Prior research on the effect of size on the likelihood to be part of an M&A has been inconclusive. Early literature found that the effect of size on the likelihood of M&A was not significant (Hannan and Rhoades, 1987 and Moore, 1996).

Other researches did find the effect of size to be significant, but were not consistent on whether it had negative or positive effect on the likelihood of M&A. Focarelli (2011) find an negative effect of size on the probability to be part of an merger or acquisition. Indicating that large banks are less likely to be a target or acquirer in a M&A than small banks. Pasiouras and Zopounidis (2006) as well find that smaller banks have increased probability to be part of an M&A as either a target or an acquirer. The findings of Wheelock and Wilson (2000) are partially consistent with the findings of Focarelli and Pasiouras and Zopounidis, they did not find a significant effect of size on the likelihood to be an acquirer, but did find that smaller banks have a significantly increased likelihood to be an acquisition target. An explanation for this could be that small banks are easier to acquire than large banks, because the small banks usually do not have resources to fight a hostile take over. It is also less expensive to buy or acquire a small bank than to buy or acquire a large bank (Pasiouras et al., 2007). Thereby a smaller bank is easier to incorporate into the

(9)

existing institution than a large bank, which generally has a more complicated financial structure (Berger et al., 1999). Finally restructuring after the merger or acquisition is easier to manage when the target banks is small than when the target bank is large (Focarelli et al., 2002).

Wheelock and Wilson (2004) however find that buying banks in a merger are on average larger, reporting a positive effect of size on the probability of being a buying bank in a merger. This could be explained by the fact large banks generally have greater resources and thereby an increased ability to buy or acquire compared to smaller banks.

Pasiouras et al. (2007) find a positive effect for acquirers as well as targets. Suggesting that banks engaging in an M&A are on average larger than banks that are not part of an M&A. They specify that the effect of size on the probability to be an acquirer is larger than the effect of size on the probability to be a target. They also considered the effect of size under different market environments and economic conditions and concluded that bank size provides part of the motivation for M&A no matter how the market and economic conditions are.

2.7 Crisis

In periods of crisis banks can experience difficult economic circumstances, endangering levels of exposure and high uncertainty. For that reason, a merger or acquisition is times of crisis, can be motivated by different reasons than in periods when there is no crisis.

Buch and DeLong (2004) find that when the GDP is lower, the volume of bank mergers increases. The costs in situations of financial distress or bankruptcy can be unlimited high and at the expense of the shareholders; therefore, an M&A would be in the best interest of the share and debt holders. Due to the pre-existing franchise values of the two banks, a merger or acquisition is also proved to be more effective then when the bank leaves the market (Berger et al., 1999).

Banks who are not jet in the situation of financial distress and who want to prevent this from happing, might want to restructure trough an M&A during times of crisis (Rossi and Volpin, 2004). Another way to avoid financial distress is by

(10)

provided and the diversification in product mix indicates a diversified risk portfolio and therefore a reduced probability of financial distress (Berger, 1999).

Similar to diversification, the improvement of the loan portfolio can reduce the risk of financial distress. Especially in times of crisis, the banks with a high number of bad loans are weak (Pasiouras et al., 2007). Therefore, a high number of bad loans might be an important reason for a merger or acquisition during the crisis. However, a high number of bad loans could also discourage other banks to acquire of merge with that bank, because the bad loans increase their risk and exposure.

Likewise, the interbank position and the liquidity of a bank are also of increased importance during a crisis. Due to the uncertain environment in periods of crisis, the low liquidity of a bank could lead to a bank-run if customers are afraid that they will not get their money back. As a result the bank could run into insolvency problems. The high exposure of banks with a negative interbank balance could as well be a bigger concern during periods of crisis.

The incentive to grow through an M&A might be more present during a crisis, since research has shown that large banks are more likely to survive a crisis. Focarelli et al., (2002) also found that the banks that went bankrupt where significantly smaller than banks involved in M&A. Also the access to the government-safety net is of bigger importance during times of crisis.

3 Data

3.1 Context, setting, background

The following chapter illustrates the context, setting and background of this research. Starting with the context of this paper by a circumscription of the European banking sector. The setting of this research will be explained by enlightening the crisis. For the background of this research the two periods (pre-crisis and crisis) will be explained and illustrated by the emphasis of the specific economic circumstances and the regulation during those years.

3.1.1 European banking sector

Europe consists of countries with open economies. This means that in the European financial market, countries can import or export capital from each other, which

(11)

prevents a bubble burst in countries with a capital surplus and a credit crunch in countries with a capital deficit.

The European financial market is bank based. The banks are financing about 70%-75% of the debt hold by customers and businesses in Europa (ECB, 2013). As a comparison, in the US this is only about 25%. European market also has a very high number of banks with different structures, creating a diversified banking system. However, the European banking sector is declining.

The aim to improve the quality of their capital, their liquidity ratios and their leverage ratios together with the upcoming regulation, pressures banks to decrease size and precede the movement of M&A activities and further specialisation. From 8,915 credit institutions in 2004, the European banking sector has only 7,861 credit institutions left in 2012, a decline of 11.8%. This is partly due to bankruptcies, but mostly due to mergers and acquisitions, where banks are merged away (ECB, 2013).

3.1.2 The crisis

The crisis started with the subprime mortgage crisis is the US. From the year 2001 until 2005 banks provided an increasing amount of mortgage loans. Due to high housing prices and the good interest terms, people found themselves in a good position. From 2002 banks took on more risk by providing subprime mortgages, which are mortgage loans provided to customers with a history of payment arrears. With those people the number of people buying a house increased. Therefore, the housing prices increased even further, resulting in an upward spiral for the demand for mortgages and increase in housing prices.

From 2006 the housing bubble bursted, the housing prices declined and the interest rate started to increase. People where no longer able to meet the payment obligations and loans started defaulting, resulting in a high number of bad loans. Due to the decrease in housing prices the collateral lowered in value, leaving banks with high losses.

From 2007 until 2009 the worsening financial state caused even more loans to default, which forced banks to make huge write-downs. The lack of trust in the market drove capital away from the banks, leaving banks short of capital and on the edge of bankruptcy.

One of the reasons, that caused this subprime mortgage crisis to lead to the global financial crisis, is the highly integrated banking market, which caused the

(12)

domino effect. Another reason is the use of new financial instruments, which were not correctly taken into account. When the crisis broke out, it exposed the high amount of obligations that were of high risk and low quality. Thereby banks were also more exposed to risk then they appeared to be due to the shadow banking system. Banks would invest in asset-backed securities (CDOs, covered bonds, CDAs) via conduits and SPVs and therefore the investments were not held in their own books. Due to that the capital requirements were much lower. However, when the vehicle got

downgraded and the refinancing of the losses was not possible, the bank had to take the assets on to their own balance sheet. Resulting the bank to be short of equity or even insolvent. Therefore, the banks had to deleverage. However if many banks are deleveraging this decreases the asset price, which leads to write-downs. This reduced the equity ratio even further, while the need for deleveraging increased even more.

With the low equity ratios of the banks, the concerns about insolvency rose, which induced bank-runs. Because nobody knew for sure which banks were in trouble, banks did no longer trust each other and were no longer borrowing to each other. Which ultimately caused the interbank market to dry up. In order to prevent bankruptcies troubled banks tried to merger with other banks. However, trust issues and the difficult market circumstances sometimes made it impossible to find a buyer.

In an attempt to maintain a stable market, central banks would provide liquidity support to the bank to make sure that a bankruptcy would not bring down the whole market. However, when the crisis kept on going the central banks was no longer able to provide enough support to prevent banks from going bankrupt. In some cases the government could then save the banks with state aid or our take them over and make the bank state-owned. Banks that were not saved faced bankruptcy.

Later in 2010 there was uncertainty about the financial position of some European governments, which led to the European sovereign debt crisis.

(13)

Source: European Central Bank (ECB)

3.1.3.1 Explanation on pre-crisis period 2004-2006

The chosen period for the pre-crisis period contains the years 2004, 2005 and 2006. This period has the same length as the crisis period and is most comparable to the crisis period in terms of technology improvements. Therefore, circumstances beside the crisis will have a miner influence on the difference in motivations for M&A between the two periods.

3.1.3.2 Regulation 2004-2006

The regulation applicable in the pre-crisis period is the Basel I accord. This is an international cooperation accord containing requirements concerning the capital ratios of the banks. Internationally active banks were required to maintain at least 8% (Tier 1 and Tier 2) capital as a percentage of risk weighted assets. The Basel I accord distinguishes between only three different risk weightings; business loans (100%), mortgage loans (50%), state loans (0%). This means that for mortgage loans there was only a capital requirement of 4%.

The biggest shortcoming of the Basel I accord is that is only distinguishes between three kinds of risk weights and does not take into account any operational risk. Therefore, in the 2004 the Basel II accord was announced and in 2006 the final version of Basel II was announced. However in Europe the Basel II was not

implemented until 2007. -0,08 -0,07 -0,06 -0,05 -0,04 -0,03 -0,02 -0,01 0 2005 2006 2007 2008 2009 % o f G D P Years

Graph 1: Government deficit EU-27

(14)

3.1.3.3 Economic circumstances 2004-2006

Before the crisis economic circumstances were good. The housing prices where high and people took on more mortgages. The banks lending activity was high. The GDP growth rate was relatively stable. Government deficits were declining and the unemployment was slightly decreasing. At the end of 2006 the US housing prices started to fall, also affecting the European housing market. However the real impact on the European banking sector was not until 2007.

Source: European Banking Federation (EBF), Eurostat 3.1.4.1 Explanation on crisis period 2007-2009

The crisis period considered in this research is the period from 2007 until 2009. The reason for this is that this is the period in which banks were mostly affected. The banks played a key role in the financial crisis, as well as mergers and acquisitions because they were important instruments used to prevent bankruptcy. Also the increased importance to improve their ratios in order to survive and the lack of trust between banks, make the financial crisis period of interest.

3.1.4.2 Regulation 2007-2009

The regulation during 2007-2009 was the Basel II accord. For the EU the Basel II accord was implicated through the CRD (capital requirement directive) as from 2007. Basel II, the follow-up of the Basel I accord, contains of three pillars. The first being capital requirements, the second being supervisory review and the third being market discipline. 0% 1% 2% 3% 4% 5% 6% 10 10,2 10,4 10,6 10,8 11 11,2 11,4 11,6 11,8 2004 2005 2006 G D P, EU R m ln Years

Graph 2: GDP pre-crisis EU-27

GDP

(15)

The first pillar allows banks to chose between two different risk weight approaches, the first required a minimum capital requirement of 15%, the second approach does not only consider the credit risk, but also the market and operational risk. That way each bank has a different capital requirement percentage based on its own portfolio. In 2008 an even more detailed alternative approach was available to calculate the required capital ratio.

The second pillar gives central banks the right to adjust the requirement according to circumstances. That way the supervisor can take into account the internal processes for dealing with residual risk such as systematic risk and liquidity risk.

The third pillar requires openness of information by the banks, which should allow for market discipline. Better-informed participants can react to the risk

assessment of the bank, by rewarding those that adept a sufficient risk management. In the light of the crisis several changes to Basel II accord were suggested in 2008 and 2009. Since the end of 2009 the Basel Committee is working on the Basel III accord, which includes higher capital requirements then in the Basel II accord.

3.1.4.3 Economic circumstances 2007-2009

When the financial crisis broke out the economic circumstance worsened. The GDP growth rate declined with a negative growth of -6% in 2009 (graph 3).

Source: European Banking Federation (EBF), Eurostat

-8% -6% -4% -2% 0% 2% 4% 6% 8% 11,4 11,6 11,8 12 12,2 12,4 12,6 2007 2008 2009 G D P, EU R m ln Years

Graph 3: GDP crisis period EU-27

GDP

(16)

The incentive to become or to stay healthy combined with the trust issues in the market, translates this into low or even negative returns on equity and high loan loss provisions (graph 4, graph 5).

Source (both graphs): European Banking Federation (EBF), Bankscope

3.2 Data sample

This research uses two datasets, the first providing all M&A activity and the second containing bank characteristics. This chapter will give a description of both datasets, followed by the use of the to dataset combined.

3.2.1 M&A activity

The first dataset contains all the mergers and acquisition between banks in Europe during the period from 2004 until 2009. This dataset was obtained from Thomson One. The total dataset contain 9907 observations. When the dataset is split up between the two periods, the dataset for the period of 2004 until 2006 contains 7261 observation and the dataset for 2007-2009 7435 observations. Meaning that according to the dataset there was more M&A activity during the crisis period.

The dataset contains all M&A activity between banks in Europe and with every M&A activity it specifies whether it is a merger or acquisition and who the acquirer or buyer was and who the acquired or target bank was.

-2% 0% 2% 4% 6% 8% 10% 12% 14% 2007 2008 2009 R o E (% )

Graph 4: Return on Equity

of banks top 335 capital holders

RoE (%) 0 50 100 150 200 250 300 350 400 2007 2008 2009 L o an lo ss p ro vi si o n s (EU R b ln )

Graph 5: Loan loss provision

of banks top 335 capital holders

(17)

3.2.2 Bank characteristics

The second dataset that is used for this research is the dataset from Bankscope. This dataset contains panel data on the characteristics of all banks over the world for the years 2004 until 2009. Because this research focuses on Europe, only the

characteristic of European banks will be used. Therefore, the dataset in matched with a list of all European countries; all banks that did not match were deleted. The dataset now exists of 14036 observations in total, from which 6907 for 2004 until 2006 and 7129 for 2007 until 2009. The observations are collapsed by name, such that for every variable the mean during that period is calculated per individual bank. By taking the average of the variables, banks with missing observation on one or two years do not have to be deleted.

4 Empirical model

4.1 Multinomial logistic model

This research looks at the effect of several bank characteristics on the probability of being an acquisition target, merger target, and acquirer or merger buyer. The

dependent variable is the M&A activity and the banks characteristics are the independent explanatory variables.

One of the models able to determine the probability is the linear probability model. This is a binary model in the sense that the outcome Y can only take on the value 0 or 1. It takes on the value 0 when the event does not occur and the value 1 when the event does occur. The dependent variable is therefore given by a Bernoulli distribution (1).

1 𝑌 = 1 𝑤𝑖𝑡ℎ 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑝_{0 𝑤𝑖𝑡ℎ 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 1 − 𝑝}

2 𝑌 = 𝛽!+ 𝛽!𝑥! + 𝛽!𝑥!+ 𝛽!𝑥!… + 𝛽!𝑥!+ 𝜀

This linear probability model can be given by the equation (2). The coefficients of the explanatory variables 𝑥_!, 𝑥_!, 𝑥_! are given by 𝛽_!, 𝛽_!, 𝛽_!. Where the intercept is given by 𝛽! and the error term is given by 𝜀. The calculated betas give the effect on the corresponding explanatory variables on the probability of Y being 1.

(18)

However, the model has some drawbacks. The betas calculated by the linear probability model can take on values beyond the [0.1] interval and therefore cannot be interpreted as real probabilities. Furthermore, the data sample on bank characteristics is heteroskedastic, which is the case when subcategories of the sample have different variances. This is a problem when using the OLS regression, since the OLS assumes the error terms to be uncorrelated and normally distributed.

Another seemingly suitable mode is the linear discriminant analysis. This model predicts the dependent group variable based on several predictor variables. However, this model also assumes the data to be homoscedastic.

To avoid this problem, the multinomial logistic model is used. The multinomial logistic model predicts the probability of an event to occur based on multiple independent variables. The dependent variable is nominal, meaning that it can take on different integers representing different events or categorical outcomes. The multinomial logistic model estimates its parameters with the maximum likelihood method. A further explanation of model and the estimation of its probabilities can be found in de Appendix.

In addition the Chow test is uses to test whether the outcomes before the crisis and during the crisis are significantly different from each other.

4.2 Variables

4.2.1 Dependent variable

The dependent variable used in this model is the variable Event. The variable is nominal and therefore can take on several discrete outcomes. When Event=0 it corresponds with no M&A occurring, 1=acquisition target, 2= acquisition, 3=merger target, 4=merger acquirer.

Table 1 M&A activity pre-crisis period

Event | Freq. Percent Cum. ---+--- 0 | 7,115 97.99 97.99 1 | 37 0.51 98.50 2 | 21 0.29 98.79 3 | 66 0.91 99.70 4 | 22 0.30 100.00 ---+--- Total | 7,261 100.00

(19)

Table 2 M&A activity crisis period

Event | Freq. Percent Cum. ---+--- 0 | 7,235 97.31 97.31 1 | 58 0.78 98.09 2 | 34 0.46 98.55 3 | 70 0.94 99.49 4 | 38 0.51 100.00 ---+--- Total | 7,435 100.00 4.2.2 Independent variable The independent variables are:

• avroaa

= average return on assets; this represents the profitability of the banks • avimp_loan_gross_loan

= average (impaired loan/gross loans); this represents the quality of the loan portfolio of the bank

• avliquid_asset_dep_ST

= average (liquid assets/deposits & short term debt); this is an indicator of the liquidity of the bank

• avpers_exp_assets

= average (personnel expenses/total assets); this indicates the efficiency of the bank

• avnon_int_gross_rev

= average (non interest income/gross revenues); this is a measure of the diversification of the bank

• avinterbank

= average interbank ratio; this shows the interbank market position of the bank

Table 3 Bank characteristics pre-crisis period

(20)

Table 4 Banks characteristics crisis period

Variable | Obs Mean Std. Dev. Min Max ---+--- avimp_loan~n | 3425 4.942647 7.727158 0 100 avroaa | 7423 .7575892 6.133704 -203.0233 78.90333 avinterbank | 5709 176.1523 193.2386 0 998.63 avliquid_a~T | 7131 46.81879 79.48913 0 983.33 avnon_int_~v | 7356 38.10658 36.71798 -600 780 ---+--- avpers_exp~s | 7129 .0240424 .0654178 0 2.668891 4.3 Regression

Given the variables the model in given by the following equation (13); (13) Pr 𝐸𝑣𝑒𝑛𝑡! = 𝐹 𝛽_! + 𝛽_!avroaa + 𝛽_!avimp_loan_gross_loan + 𝛽_!avliquid_asset_dep_ST + 𝛽_!avpers_exp_assets + 𝛽!avnon_int_gross_rev + 𝛽!avinterbank Where i=[0,1,2,3,4]

Since mergers and acquisitions involve a different commitment, they also might not happen for the same reasons. Therefore, the model is also estimated for mergers (14) and acquisitions (15) separately.

14 Pr 𝐸𝑣𝑒𝑛𝑡_! = 𝐹 𝛽_! + 𝛽_!avroaa + 𝛽_!avimp_loan_gross_loan + 𝛽!avliquid_asset_dep_ST + 𝛽!avpers_exp_assets + 𝛽_!avnon_int_gross_rev + 𝛽_!avinterbank Where i=[0,2,4] 15 Pr 𝐸𝑣𝑒𝑛𝑡! = 𝐹 𝛽_! + 𝛽_!avroaa + 𝛽_!avimp_loan_gross_loan + 𝛽_!avliquid_asset_dep_ST + 𝛽_!avpers_exp_assets + 𝛽!avnon_int_gross_rev + 𝛽!avinterbank Where i=[0,1,3]

(21)

The coefficient 𝛽 is the difference in the log odds as X changes by one. Usually the coefficients are not used for the interpretation of the model, because better

interpretation of the model can be gained by looking at the relative risk ratio (RRR). When the independent variable increases with one then it is RRR times more likely for the event to occur.

All three of the above models are performed with the 2004-2006 dataset and the 2007-2009 dataset, after that the outcomes are compared. The Chow test estimates whether the outcomes are statistically different from each other.

5 Goodness of fit

To check the validity of the model several tests are performed on all three of the equations, using both datasets. The outcomes of the test can be found in the appendix.

At first the sample size is checked. For a multinomial logistic model to be valid, there must be a large dataset and at least 10 observations per Event. In our model all categories have at least 10 observations, so that assumption is met.

Another typical test for the multinomial logistic model is the Hausman test. The Hausman test checks the independence of irrelevant alternatives, which is an assumption of the multinomial logistic model. The IIA assumption states that the odds of one event occurring over another event are independent of other alternative events outside the model. When performing the Hausman test the outcome supports H0 for all alternatives (Events) except for the base category. Normally this could be a problem, but since the events in this research cover all possible alternatives this wont be a problem.

After testing for IIA there must be tested whether a pair of events can be combined, which can be done with the Wald test or the LR test. In this research the Wald test is used. When performing the Wald test for the five categories it indicates that some categories can be combined. Hence, only the results for the two separated regressions for acquisitions (14) and mergers (15) are taken into account.

To check the overall fit of the model a goodness of fit test is performed, which is quite similar to the one that can be performed for linear regressions. Here the Pseudo R2 represents the difference in log-likelihood between the model without

(22)

explanatory variables (intercept only) and the full model. The bigger the R2, the better the fit of the model.

The prob>chi2 or p-value tests whether all the coefficients are different than zero. The outcomes show that all model are better than a empty model.

6 Results

In this chapter the key findings will be presented. The first output table will be shown in the text, others can be found in the Appendix.

6.1 Acquisitions

6.1.1 Acquisitions during pre-crisis period

The first regression that is done is the regression for acquisitions before the crisis. The regression gave the following outcome in table 6.

Table 6 Outcome multinomial logistic model for acquisitions 2004-2006

Multinomial logistic regression Number of obs = 2143 LR chi2(12) = 45.66 Prob > chi2 = 0.0000 Log likelihood = -330.40397 Pseudo R2 = 0.0646 Event | RRR Std. Err. z P>|z| [95% Conf. Interval] --- 1 |

avroaa | .8746158 .1830817 -0.64 0.522 .580279 1.31825 avimp_loan | 1.00242 .0314097 0.08 0.939 .9427103 1.065911 avliquid_a | .9975635 .0076139 -0.32 0.749 .9827516 1.012599 avpers_exp | .0482632 .7379065 -0.20 0.843 4.67e-15 4.99e+11 avnon_int_ | 1.017976 .0112353 1.61 0.106 .996192 1.040237 avinterban | 1.000962 .0010619 0.91 0.365 .9988829 1.003045 _cons | .0040512 .0023895 -9.34 0.000 .0012751 .0128718 --- 3 | avroaa | .8521515 .1141516 -1.19 0.232 .6553789 1.108003 avimp_loan | .983347 .0240909 -0.69 0.493 .9372454 1.031716 avliquid_a | 1.006079 .0028496 2.14 0.032 1.000509 1.011679 avpers_exp | 1.05e-19 1.87e-18 -2.46 0.014 7.93e-35 .0001391 avnon_int_ | 1.013059 .0056871 2.31 0.021 1.001973 1.024266 avinterban | .9952718 .0014093 -3.35 0.001 .9925134 .9980379 _cons | .0568829 .017463 -9.34 0.000 .0311648 .1038245 --- 5 | (base outcome)

(23)

Table 7 shows the direction of the effects and the significance levels of the independent variables on the likelihood of being a target and being an acquirer.

Table 7 Effect on the likelihood to be an acquisition target or acquirer 2004-2006

Variable Target Acquirer

Profitability - ns - ns

Quality loan portfolio - ns + ns

Liquidity - ns + **

Efficiency + ns + **

Diversification + ns + ***

Interbank market exposure - ns + ***

ns= Not significant; *Significance level of 10% ** Significance level of 5%; *** Significance level of 1%

None of the variables had a significant effect on the likelihood to be a target. The variables liquidity, diversification, efficiency and interbank market exposure had a significant positive effect on the probability to be an acquirer. Meaning that acquiring banks before the crisis are on average more efficient, more liquid, more diversified and more dependent on the interbank market.

6.1.2 Acquisitions during crisis period

The table (8) below shows the effects and significance levels for the explanatory variables during the crisis.

Table 8 Effect on the likelihood to be an acquisition target or acquirer 2007-2009

Variable Target Acquirer

Quality loan portfolio + ns + ns

Liquidity + ns + ***

Efficiency + ns + ***

Diversification + ns + ns

Interbank market exposure + ** + ***

For acquisitions during the crisis the interbank market exposure of the bank had a significant positive effect on the probability of being a target. With respect to the

(24)

probability of being an acquirer, the variable liquidity, efficiency and interbank market exposure had a positive effect. Which indicates that acquisition targets during the crisis were more interbank dependent and acquirers were more liquid, more efficient and also more dependent on interbank market.

6.1.3 Comparing pre-crisis and crisis outcomes

In order to compare the effects of variables for the before the crisis and during the crisis, the two model results are shown simultaneously.

Table 9 Simultaneous outcome pre-crisis and crisis period

Simultaneous results for precrisis, crisis

Number of obs = 4739 --- Robust

Coef. Std. Err. z P>|z| [95% Conf. Interval] ---+--- precrisis_1 avroaa | -.1339706 .1896228 -0.71 0.480 -.5056244 .2376832 avimp_loan | .0024169 .0145233 0.17 0.868 -.0260482 .0308821 avliquid_a | -.0024395 .0046585 -0.52 0.601 -.0115699 .006691 avpers_exp | -3.031086 12.07325 -0.25 0.802 -26.69421 20.63204 avnon_int_ | .0178166 .0065826 2.71 0.007 .004915 .0307182 avinterban | .0009615 .0009785 0.98 0.326 -.0009563 .0028793 _cons | -5.50874 .5457272 -10.09 0.000 -6.578346 -4.439134 ---+--- precrisis_3 avroaa | -.159991 .0857827 -1.87 0.062 -.328122 .0081401 avimp_loan | -.0167932 .0113687 -1.48 0.140 -.0390755 .0054891 avliquid_a | .0060605 .001846 3.28 0.001 .0024423 .0096787 avpers_exp | -43.70037 10.29621 -4.24 0.000 -63.88057 -23.52017 avnon_int_ | .012974 .0040091 3.24 0.001 .0051163 .0208317 avinterban | -.0047394 .0011596 -4.09 0.000 -.0070121 -.0024667 _cons | -2.866761 .2216709 -12.93 0.000 -3.301228 -2.432294 ---+--- precrisis_5 0 (omitted) ---+--- crisis_1 avroaa | -.0594182 .0355959 -1.67 0.095 -.129185 .0103486 avimp_loan | -.0201406 .0263775 -0.76 0.445 -.0718395 .0315583 avliquid_a | .0013991 .0035242 0.40 0.691 -.0055081 .0083063 avpers_exp | -8.403794 11.43436 -0.73 0.462 -30.81474 14.00715 avnon_int_ | .0075813 .0044815 1.69 0.091 -.0012023 .0163649 avinterban | -.0031452 .0011293 -2.79 0.005 -.0053587 -.0009318 _cons | -3.834047 .2864031 -13.39 0.000 -4.395387 -3.272707 ---+--- crisis_3 avroaa | -.0074168 .0430506 -0.17 0.863 -.0917944 .0769609 avimp_loan | -.0233436 .0152332 -1.53 0.125 -.0532001 .0065129 avliquid_a | .0059172 .0012678 4.67 0.000 .0034325 .008402 avpers_exp | -44.55763 14.45581 -3.08 0.002 -72.89049 -16.22476 avnon_int_ | .0016321 .003786 0.43 0.666 -.0057884 .0090526 avinterban | -.0036204 .0009904 -3.66 0.000 -.0055616 -.0016792 _cons | -2.752129 .2073788 -13.27 0.000 -3.158584 -2.345674 ---+--- crisis_5 0 (omitted) ---

(25)

To see whether the results are significantly different from each other a Chow test is performed. For targets as well as for acquirers the result of the Chow test for the whole model is that the null hypothesis cannot be rejected with a significance level of 5% (same with 10%). When testing per variable, the Chow test shows that the effect of the interbank market exposure on the probability of being a target is significantly different, where it has a significant positive effect during the crisis and non-significant effect before the crisis. The variable diversification has significantly different effect on the probability of being a acquirers, where it had a positive effect before to crisis and no significant affect during the crisis. The all outputs per individual variable can be found in Table 9.

Table 9 Chow test on significant difference between pre-crisis and crisis period

6.2 Mergers

6.2.1 Mergers during pre-crisis period

In the table below the directions and significance levels of the variables are presented.

Test Prob > chi2

[m1_1]avroaa=[m2_1]avroaa 0.6992 [m1_1]avimp_loan_gross_loan=[m2_1]avimp_loan_gross_loan 0.4538 [m1_1]avinterbank =[m2_1]avinterbank 0.0060 *** [m1_1]avliquid_asset_dep_ST =[m2_1]avliquid_asset_dep_ST 0.5111 [m1_1]avnon_int_gross_rev =[m2_1]avnon_int_gross_rev 0.1987 [m1_1]avpers_exp_assets=[m2_1]avpers_exp_assets 0.7466 [m1_3]avimp_loan_gross_loan =[m2_3]avimp_loan_gross_loan 0.7304 [m1_3]avroaa=[m2_3]avroaa 0.1119 [m1_3]avinterbank =[m2_3]avinterbank 0.4631 [m1_3]avliquid_asset_dep_ST =[m2_3]avliquid_asset_dep_ST 0.9490 [m1_3]avnon_int_gross_rev =[m2_3]avnon_int_gross_rev 0.0397 ** [m1_3]avpers_exp_assets=[m2_3]avpers_exp_assets 0.9615

(26)

Table 10 Effect on the likelihood to be a merger target or buyer 2004-2006

Variable Target Buyer

Quality loan portfolio + ns + ns

Liquidity - ns - *

Efficiency + ns + ns

Diversification + ns + ns

Interbank market exposure + ns + **

For mergers before the crisis there were no variables with a significant effect on the probability of being a target. For acquirers the only significant variables were liquidity, which had a negative effect and interbank market exposure, which had a positive effect. It indicates that before the crisis the buying banks in a merger were relatively liquid and dependent on the interbank market. Banks that were taken over during the pre-crisis period were not statistically different from banks that were not taken over.

6.2.1 Mergers during crisis period

The table below shows the directions and significance levels of the variables.

Variable Target Buyer

Profitability + ns + ns

Quality loan portfolio + ns + *

Liquidity - ns + ns

Efficiency + * + ***

Diversification + ns + ***

Interbank market exposure + ns + *

The only variable that is significant for merger targets during the crisis is the efficiency indicator, which has a significant positive effect on the likelihood to be acquired. The variables significant for buyer banks were the loan portfolio quality, efficiency, diversification and the interbank market exposure, all with a positive

(27)

effect. The findings indicate that both the targets and buyers during the crisis had a relatively high efficiency level, were buyers were also more diversified, more dependent on the interbank market and had better loan portfolio quality than non-involved banks.

6.2.3 Comparing pre-crisis and crisis outcomes

To in see whether the results from the pre-crisis and the crisis period are significantly different form each other, a Chow test is performed.

For merger targets the result of the Chow is that the null hypothesis cannot be rejected. When testing for significant differences between buyers before the crisis and during the crisis, the Chow test finds that for all variables together the outcomes are significantly different. When testing for significant differences per individual

variable, the Chow test finds that the effect of the quality of the loan portfolio and the diversification of the bank are significantly different between the periods. The results indicate that merger buyers during the crisis have significantly different

characteristics than buyers before the crisis. During the crisis buying banks are more diversified and have better-quality loan portfolios.

7 Robustness tests

As mentioned before size and the advantages related to it can be an important reason for M&A. Large banks generally have better resources, a more complicated financial structure and lower required capital ratios (Berger et al., 1999, Wheelock and Wilson 2004). Hence, it is reasonable to think that motives for M&A can vary according to bank size.

As a robustness check the regressions are re-estimated while distinguishing between small and large banks. The data samples for both acquisitions and mergers are split into two based on total asset size.

7.1 Small and large acquisitions

When re-performing the regressions, while distinguishing between small and large banks, the estimates during the pre-crisis period for target banks are again

(28)

more diversified and more exposed on the interbank market, as can be seen in table 12. The efficiency is no longer significant for both small and large acquirers.

Table 12 Effect on the likelihood to be an acquisition target or buyer 2004-2006

Variable Target(S) Acquirer(S) Target(L) Acquirer(L)

Profitability + ns - ns - ns - ns

Quality loan portfolio - ns + ns + ns + ns

Liquidity - ns + ns - ns + *

Efficiency + ns + ns - ns + ns

Diversification + ns + ns + ns + **

Interbank market exp + ns + ns - ns + **

During the crisis acquisition targets are not significantly different from non-involved, when distinguishing between small and large banks. The positive effect of the

interbank market exposure is no longer significant. Small acquirers during the crisis were significantly less diversified. The significantly positive effect of efficiency and interbank market exposure persists only for large acquirers. The effect of liquidity is no longer significant when distinguishing between small and large acquirers.

Table 13 Effect on the likelihood to be an acquisition target or buyer 2007-2009

Variable Target(S) Acquirer(S) Target(L) Acquirer(L)

Profitability - ns - ns - ns - ns

Quality loan portfolio + ns + ns + ns + ns

Liquidity - ns + ns + ns + ns

Efficiency + ns + ns - ns + *

Diversification + ns - *** + ns + ns

Interbank market exp + ns + ns + ns + **

When testing for significant differences between the pre-crisis and crisis period, the Chow test finds that for large banks there is no significant difference. For small banks, acquirers before the crisis are more diversified than during the crisis.

(29)

7.2 Small and large mergers

When re-performing the regressions, while distinguishing between small and large banks, the positive effects of liquidity and interbank market exposure for buyers are no longer significant. For small targets before the crisis there is still no variable significant, for large targets diversification has a significant positive effect.

Variable Target(S) Buyer(S) Target(L) Buyer(L)

Profitability - ns - ns - ns - ns

Quality loan portfolio + ns + ns + ns + ns

Liquidity - ns - ns - ns - ns

Efficiency + ns + ns + ns + ns

Diversification + ns + ns + * + ns

Interbank market exp - ns - ns + ns + ns

When distinguishing between small and large targets during the crisis, the efficiency variable is no longer significant. For merger buyers during the crisis, the effect of the interbank market exposure in no longer significant. The positive effect of the loan portfolio quality, the efficiency and the diversification due persist but only for large buyers during the crisis.

Variable Target(S) Buyer(S) Target(L) Buyer(L)

Profitability + ns - ns + ns + ns

Quality loan portfolio - ns + ns + ns + *

Liquidity + ns + ns - ns + ns

Efficiency -+ ns + ns + ns + *

Diversification - ns + ns + ns + **

Interbank market exp + ns - ns + ns + ns

When testing for significant differences between pre-crisis and during the crisis, the Chow test finds that targets as well as buyers are significantly different for both small and large banks. When performing the Chow test per variable, it finds that large

(30)

buyers during the crisis were significantly more efficient, more diversified and have better loan portfolio quality than before the crisis. Large targets were significantly more diversified before the crisis than during the crisis.

8 Discussion

The results suggest that profitability, as a motive for M&A, was no significant factor both before and during the crisis. The quality of the loan portfolio has shown to be a significant positive factor for merger acquirers during the crisis. Which is partly in line with the findings of Berger et al. (1999), who find that acquiring banks have less bad loans than target banks.

Acquisition acquirers before and during are motived to increase the

investment opportunities of the bank, while contributing to the overall liquidity of the two banks. Which is in line with the findings of Pasiouras et al. (2007). Liquidity also appears to be significant motivation for merger acquirers before the crisis as results suggest them to be less liquid than non-involved banks, which is an unexpected finding and not in line with any other literature. However, when distinguishing between small and large banks, the effect is not robust.

It is found that acquisition acquirers are motivated to transfer their superior knowledge and efficient management policies on to the less efficient banks, in line with the findings of Berger et al (1999). In mergers both acquirers and targets are more efficient during the crisis, which is in contrast with findings from earlier

research. A possible explanation for this could be that the crisis has caused bigger and more efficient banks to become a target of mergers. When performing a robustness check, the increased efficiency only persists for large merger acquirers, which can be explained by the economies of scale associated with bigger size.

Diversification is a significant factor in the motivation to reduce risk through an M&A, in the sense that acquisition acquirers before the crisis and merger acquirers during the crisis were more diversified. More diversified acquirers tend to take over less diversified targets.

During both periods the results show a trend of high exposure and high funding costs due to the interbank market position for acquirers of both mergers and acquisitions, which is in line with the findings of Focarelli et al. (2002). It is also

(31)

found that acquisition targets during the crisis are motivated to improve their

interbank market position in order to reduce risk. Earlier literature already mentioned this as a possible motivation, but never found empirical evidence for it (Focarelli et al., 2002).

When comparing the crisis period with the period before, it shows that during the crisis acquisition targets were motivated by reducing the cost of funding

(interbank market position), while before the crisis they were not. The difference could be explained by the interbank market drying-up during the crisis. Also acquisition acquirers were no longer motivated by diversification during the crisis, this is unexpected since diversification could lower capital requirements. Merger buyers during the crisis had increased diversification, compared to the period before the crisis. This could be due to the increased importance of diversification during the crisis, since it is associated with lower risk. During the crisis the quality of the loan portfolio was of increased importance for merger acquirers. The high quality of the portfolio might have been necessary to overcome trust-issues during the crisis. When comparing the crisis period with the period before, while examining small and large banks separately, the results over all sub-categories are consistent in that profitability was no significant factor in the probability of being involved in an M&A. The interbank market exposure was of positive influence for large acquisition acquirers both before and during the crisis. Small acquisition acquirers appear to be less diversified during the crisis than before the crisis. For large banks, merger acquirers are more diversified during the crisis, while merger targets are more diversified before the crisis. This difference among small and large banks is in line with earlier findings that large banks are usually more diversified while small banks operate in more specialized, niche markets (ECB, 2002). The quality of the loan portfolio of large merger acquirers was significantly better during the crisis and before the crisis. This difference between small and large merger acquirers can be explained by the fact that large banks usually lend to large institutions, which is less risky than when lending to small-businesses (Focarelli et al., 2002, Hughes et al., 1999). M&A during the crisis seems to be more motivated by the high efficiency level of large M&A acquirers than before the crisis. The difference can be explained by the increased need for efficiency in order to be able to survive during the crisis (ECB, 2002).

(32)

9 Conclusions

This research is conducted to the reasons for mergers and acquisitions between European banks, with a comparison between the pre-crisis and crisis period. With the use of a multinomial logistic model the likelihood of being involved in an M&A activity is estimated based on bank characteristics.

When testing the hypothesis it was found that the improvement of the loan portfolio is indeed a more imported motive for M&A during the crisis than before the crisis. However, the motive of diversification only accounts for mergers between large banks. Thereby, it was found that the interbank position was more important both before and during the crisis. Also, the M&A during the crisis seems to be more motivated by the high efficiency level of large M&A acquirers than before the crisis.

This paper adds to the existing literature by highlighting the differences between M&A characteristics before and during the crisis. The results show that there are significant differences, which might possibly have implications to the current regulation on M&A banks and the presumed risk profiles.

This research is subject to some limitations, which need to be considered. The first limitation is that the regressions are performed based on average data; therefore, the outcomes might be biased. For example when a bank is acquired in 2008 because it has a poor liquidity that year, then the average liquidity level over 2007-2009 might still be above average and thereby not showing in the odds ratio. Using only the liquidity of the bank in 2008 (year of M&A) would also give a biased outcome, since it is compared to the average liquidity of non-involved banks, which will have a better liquidity since 2008 was the worst year of crisis period. A possible way to solve this is to perform the regression for each year separately, however their will still be

fluctuations of the characteristics during the year. Thereby, the fraction of banks involved in M&A will be much smaller per year.

Another limitation is due to the dataset. The dataset only contains mergers and acquisitions between two banks in Europe. Mergers or acquisitions involving a bank in Europe and a bank in the US are not taken into account. Also banks taken over by the government are not taken into account, since the government is not a bank.

The latter might be subject for possible future research. Other possible future research is the likely wave of mergers and acquisitions resulting from banks that

(33)

failed the on-going Asset Quality Review (AQR) performed on behalf of the ECB and the upcoming stress tests performed by the European Banking Authority.

Reference list

Anderson, Ch.W., Becher, D.A. and Campbell, T.L., 2004, Bank mergers, the market for bank CEOs, and managerial incentives, Journal of Financial Intermediation, 13(1), 6- 27.

Bankscope, 2014, http://www.bvdinfo.com/en-gb/productscontainer/bankscope

Barnes, P., 2000, The identification of U.K. takeover targets using published

historical cost accounting data. Some empirical evidence comparing logit with linear discriminant analysis and raw financial ratios with industry-relative ratios,

International Review of Financial Analysis, 9(2), 147-162.

Berger, A. N., Demsetz, R. S., & Strahan, P. E., 1999, The consolidation of the financial services industry: Causes, consequences, and implications for the future.

Journal of Banking & Finance, 23(2), 135-194.

Boot, A. W. A. (1999). European Lessons on Consolidation in Banking, Journal of

Banking and Finance, 23, 609-613.

Buch, C. M., & DeLong, G., 2004, Cross-border bank mergers: What lures the rare animal? Journal of Banking & Finance, 28(9), 2077-2102.

Centraal Bureau Statistiek (CBS), 2014. Databases Bankscope. Retrieved February 13th 2014, http://www.cbs.dk/en/library/databases/bankscope

Cheng, D.C., B.E. Gup and L.D. Wall, 1989, Financial Determinants of Bank Takeovers, Journal of Money, Credit and Banking, 21 (4), 524-536.

Cheng, S., & Long, J. S., 2007, Testing for IIA in the multinomial logit model.

(34)

Demyanyk, Y., & Van Hemert, O., 2011, Understanding the subprime mortgage crisis. Review of Financial Studies, 24(6), 1848-1880.

Erel, I., 2011, The effect of bank mergers on loan prices: Evidence from the United States. Review of Financial Studies, 24(4), 1068-1101.

European Commision, 2014, Capital requirements regulation and directive – CRR/CRD IV. Retrieved on January 20th 2014,

http://ec.europa.eu/internal_market/bank/regcapital/legislation-in-force/index_en.htm

European Central Bank (ECB), 2014. Consolidated banking data. Retrieved: may 17th 2014, https://www.ecb.europa.eu/stats/money/consolidated/html/index.en.html

European Banking Federation (EFB), 2014. Factsheet: facts & figures 2011. Retrieved: February 13th 2014,

http://www.ebf-fbe.eu/uploads/Facts%20&%20Figures%202011.pdf

Focarelli, D., Panetta, F., & Salleo, C., 2002, Why do banks merge?. Journal of

Money, Credit, and Banking, 34(4), 1047-1066.

Gönül, F., & Srinivasan, K., 1993, Modeling multiple sources of heterogeneity in multinomial logit models: Methodological and managerial issues. Marketing Science,

12(3), 213-229.

Hannan, T., and S. Rhoades, 1987, Acquisition targets and motives: The case of the banking industry, The Review of Economics and Statistics, 69 (1), 67-74.

Hausman, J., & McFadden, D., 1984, Specification tests for the multinomial logit model. Econometrica: Journal of the Econometric Society, 1219-1240.

Hughes, J. P., Lang, W. W., Mester, L. J., & Moon, C. G., 1999, The dollars and sense of bank consolidation. Journal of banking & finance, 23(2), 291-324.

(35)

Johnsen, T., & Melicher, R. W., 1994, Predicting corporate bankruptcy and financial distress: information value added by multinomial logit models. Journal of Economics

and Business, 46(4), 269-286.

Lepetit, L., Patry, S., & Rous, P., 2004, Diversification versus specialization: an event study of M&As in the European banking industry. Applied Financial Economics, 14(9), 663-669.

Liang, J. N., & Rhoades, S. A., 1991, Asset diversification, firm risk, and risk-based capital requirements in banking. Review of Industrial Organization, 6(1), 49-59. Milbourn, T. T., Boot, A. W., & Thakor, A. V., 1999, Megamergers and expanded scope: Theories of bank size and activity diversity. Journal of Banking & Finance, 23(2), 195-214.

Pasiouras, F., Tanna, S., & Zopounidis, C., 2007, The identification of acquisition targets in the EU banking industry: an application of multicriteria approaches.

International Review of Financial Analysis, 16(3), 262-281.

Pasiouras, F., Tanna, S., & Gaganis, C., 2011, What drives acquisitions in the EU banking industry? The role of bank regulation and supervision framework, bank specific and market specific factors. Financial Markets, Institutions & Instruments,

20(2), 29-77.

Rossi, S., & Volpin, P. F., 2004, Cross-country determinants of mergers and acquisitions. Journal of Financial Economics, 74(2), 277-304.

Wheelock, D.C., and P.W Wilson, 2000, Why do banks disappear? The determinants of U.S. bank failures and acquisitions. The Review of Economics and Statistics, 82(1), 127-138.

Wheelock, D. and P. Wilson, 2004, ‘Consolidation in US banking: Which banks engage in mergers?’, Review of Financial Economics, 13(1), 7-39.

The characteristics of banks involved in M&A activiteis in the European banking sector : a comparison between the pre-crisis and crisis period