Master Thesis
Business Economics – Finance Track
Student: Darryl Yorks
Student Number: 10205063 Supervisor: Tolga Caskurlu
Date: 15-08-16
FACULTY OF BUSINESS ECONOMICS
The effect of bank mergers on bank lending behavior
Abstract
This paper investigates the effect of bank consolidation on bank lending behavior. Using data for all indexed U.S. banks for the period 1987 to 2015. I find that on average the spread and interest rate tend to increase for loans with a minimum value of $75 million after a merger. For the small loans with a maximum value of $75 million the spread and interest rate remain unchanged. This increase is only present in the period 1990 to 2000. Furthermore, in the period 2001 to 2012 a merger also leads to a decrease in small firm lending but an increase in total lending over total assets.
2
Verklaring eigen werk
Hierbij verklaar ik, Darryl Yorks, dat ik deze scriptie zelf geschreven heb en dat ik de volledige verantwoordelijkheid op me neem voor de inhoud ervan.
Ik bevestig dat de tekst en het werk dat in deze scriptie gepresenteerd wordt origineel is en dat ik geen gebruik heb gemaakt van andere bronnen dan die welke in de tekst en in de referenties worden genoemd.
De Faculteit Economie en Bedrijfskunde is alleen verantwoordelijk voor de begeleiding tot het inleveren van de scriptie, niet voor de inhoud.
3 Table of contents
1. Introduction 4
2. Literature review 5
2.1 Rationale behind a merger 5
2.1.1 Value maximizing motives 6
2.1.2 Non-value maximizing motives 6
2.2 Small firm lending 7
2.3 Market power 9
3. Empirical strategy 10
3.1 Hypothesis 11
3.2 Methodology 11
4 Data and descriptive statistics 12
4.1 Data collection 13
4.2 Sample selection and statistics 13
5. Results 14
5.1 Interest spread 14
5.2 Interest rate 15
5.3 Small firm lending 16
5.4 Bank lending 17
6. Robustness checks 18
6.1 Subsamples based on loan size 18
6.1.1 Interest spread 18
6.1.2 Interest rate 19
6.1.3 Small firm lending 19
6.2 Subsamples based on time periods 20
6.2.1 Interest spread 20
6.2.2 Interest rate 21
6.2.3 Small firm lending 21
6.2.4 Lending 22
7. Conclusion 22
References 25
4 1. Introduction
The number of banks has steadily declined from the late 80’s until the beginning of the 90’s by almost 30% (Peek and Rosengren, 1998). This trend continued towards the financial crisis of 2007 to 2008. Over the period of 2007 until 2013, the number of banks shrank by 14% in the US (McCord, Prescott and Sablik, 2015). This decrease in the number of banks was not only caused by failures, but also by increased acquisition activity and by a drop in new bank entry (Berger, Saunders, Scalise and Udel, 1998). The focus of this thesis is on the acquisitions that partially cause the drop in the number of banks within the US.
With this decrease in the number of banks and the increase in acquisition activity, the banking landscape has changed. The share of total assets held by the eight largest banks has increased from 22,3% to 35,5% in the 90’s (Berger, Demsetz, and Strahan, 1999). Previous research is done in this field, but most of this date from the late 90’s or covers data from that period. Further, the changes are partially caused by changes in regulations and their effects cause banks to behave differently than before (Heid, 2007; Grosse and Schumann, 2014). Removing the intra- and interstate branching restrictions on banks during the 1980s and beginning of the 1990s caused an increase in the number of bank mergers (Berger, Kashyap, Scalise, Gertler and Friedman, 1995).
A merger or acquisition is often done to maximize shareholder value (Prompitak, 2009). After a bank merges with another bank, the total assets have increased. Berger et al. (1998) state that a large bank can change its’ lending behavior after a merger, which can influence the average interest rate charged or the focus of their. If a large bank tends to fund only large firms, or becomes less specialized in funding small firms, this can influence the whole economy as small firms have a significant impact on economic growth (Acs, 1993).
This study aims to provide evidence on the effect of bank consolidation on bank lending behavior. To do this I answer the following question: What is the effect of bank mergers on bank
lending behavior? To answer this question, the effect is split into four factors. I look at the effect
of mergers on the interest spread, interest rate charged, total lending and small firm lending. Each hypothesis uses the same base model with a changing dependent variable. This method is mainly based on a paper by Erel (2009) who finds a decrease in interest rates due to efficiency gains. Existing literature is divided about the effect of a merger. Papers by Peek and Rosengren (1998) suggest that a merger increases small firm lending, while Han, Zhang and Greene (2015) find the
5
opposite. Furthermore, Berger and Hannan (1989) argue that banks are no longer minimizing costs after a merger, assuming a merger leads to an increase in market power. This leads to an increase in the interest rate charged, but no increase in the spread.
To expand knowledge from existing research, total lending is added as there is limited literature on this effect. Additionally, the timeframe is expanded as this paper covers data from 1987 to 2015 to examine changes over time. Data from all indexed banks are used and the data are extracted from Wharton Research Data Services and Thomson ONE.
The remainder of this research consists of the following sections. Section 2 presents an analysis of the existing literature. Section 3 consists of the empirical method where I discuss the hypotheses and methodology. Section 4 presents the data and shows the summary statistics. Section 5 covers the results found from the regressions. Section 6 presents the results of the regressions on several subsamples. Section 7 summarizes and concludes the findings of this research.
2. Literature review
The existing literature reports different findings on the effect of mergers between banks on lending behavior and loan conditions. Some researchers find an increase of the interest spread (Berger and Hannan, 1989), a decrease in small firm lending (Han et al., 2015), or the opposite (Erel, 2009). This section first explains the rationale behind a merger. After that the effect of bank mergers on small firm lending is discussed. Finally, findings on the effect of market power are examined.
2.1 Rationale behind a merger
The rationale behind a merger can be explained by two types of motives: value maximizing and non-value maximizing. Value maximizing motives include increasing market power or efficiency. An example of a non-value maximizing motive is empire building. In this section both motives are discussed.
6 2.1.1 Value maximizing motives
The main rationale behind mergers and acquisitions (M&A) in the financial sector should be to maximize value, for example by increasing market power or to become more efficient (Prompitak, 2009).
Previous research is ambiguous about the efficiency gains in bank consolidation. On the one hand, Pilloff and Santomero (1998) find that acquiring banks tend on average to be more cost efficient than their targets. Peristiani (1993) argues that acquiring banks are more profitable because of smaller nonperforming loans ratios. These results indicate that acquiring firms can implement their strategy to the target to obtain similar profits as the acquiring firm (Savage, 1991). Cornett, McNutt and Tehranian (2006) find results in line with this theory. They look at performance changes around bank mergers and find that the operating performances of merged banks increase significantly after the merger. The increased performances originate not only from cost reduction, but also from revenue enhancement. Furthermore, larger mergers tend to produce greater performance gains than small mergers. On the other hand, Berger and Hannan (1989) and Boyd and Nicoló (2005) find negative efficiency gains, due to the effect of market power, which is further explained in section 2.3.
The “too big to fail” status might also be a main driver for taking over other banks. Banks can try to increase value by getting access to the safety net offered by the government. This includes discount window access, deposit insurance and payment system guarantees (Berger et al., 1999). O’Hara and Shaw (1990) are looking more in-depth to this matter and investigate what happens to bank equity values when some banks were announced to be “too big to fail” and what the consequences were. Results indicate positive welfare effects for the “too big to fail” banks due to the extra security provided by the government, whereas non-included banks suffered from negative welfare effects.
2.1.2 Non-value maximizing motives
A non-value maximizing merger motive for a firm is empire building (Berger et al., 1999). Managers tend to earn more money if they are in charge of a larger company. Taking over another bank is therefore likely to lead to an increase in salary. Furthermore, Berger et al. (1999) state that leaders want to protect their position or certain firm-specific sources of human capital. To achieve this they reduce the risk of bankruptcy to a level below one that will maximize
7
shareholders’ interest. This is backed by papers of Subrahmanyam, Rangan and Rosenstein (1997) and Cornett, Hoyakimian, Palia and Tehranian (2003). They show that there exists a positive relation between abnormal returns at acquiring banks and sensitivity of the CEO’s pay to performance of the bank. A possible solution to this problem could be to increase the CEO’s share in the acquiring bank. In this way a negative outcome will affect the CEO personally as well (Subrahmanyam et al, 1997).
Giving the CEO a larger share in the company does not come without risks. Once the manager has a substantial share in the firm, he or she might prevent the firm from being a target of a merger, which could be against the best interest of the shareholders (Hadlock, Houston and Ryngaert, 1999). These findings are in line with the concept that teams with large ownership or even a majority can cut off a possible takeover.
Not only managers and shareholders can influence merger decisions, also governments can play a direct role. The government looks into M&A’s and approves or disapproves them to make sure their safety nets, such as deposit insurance, will not be exploited and that they will not be over exposed to some firms. Next to this they can stimulate banks to merge, for example in times of financial crisis to increase financial stability. In addition banks can be offered subsidies to complete a specific merger.
2.2 Small firm lending
From the late 80’s until mid-90’s, the amount of commercial and savings banks declined by nearly 30% (Berger et al., 1999). In their paper, Peek and Rosengren (1998) mention a public policy concern from this shrinkage which is partially due to consolidation and the effect it can have on small firm lending. They state that small business borrowers rely more on banks to fulfill their credit needs, whereas larger firms can also access national credit markets and are thus less sensitive to changes in the banking industry. This sensitivity can cause that the small business credit needs are not fulfilled and possibly lead to missing investment opportunities or bankruptcy. Peek and Rosengren (1998) look for a possible effect of bank mergers on small firm lending by examining changes in business loans of $1 million or less for nonfarm, nonresidential, commercial and industrial purposes. Their results suggest that the effect of a merger on small firm lending depends on the share of the portfolio that is dedicated to small firm lending by the acquiring bank. They also find that acquiring banks tend to be more specialized in small business
8
lending than non-acquirers of the same size. These findings indicate that after a merger, banks are more likely to increase their small firm lending rather than to decrease it. Additionally, Erel (2009) finds a similar result in his study. His findings suggest that, on average, the amount of small-business lending by the acquiring firm increases after the merger.
On the contrary, findings by Han et al. (2015) suggest the opposite. In their research, they use the US Survey of Small Business Finances (SSBF) which only includes small businesses with fewer than 500 employees. Han et al. (2015) have access to a total of 7801 firms over the years 1998 to 2003. They use total lines of credit over total assets to reflect bank credit availability, and their results show a decrease in a firms’ liquidity position once the concentration of the bank market increases. Assuming that a merger increases market power, this shows that after a merger the consolidated firm is less willing to lend to small firms. These results are similar to the effect Berger et al. (1998) find. The research of Han et al. (2015) extends on the work by Berger et al. (1998) as it also reports that a decrease in small firm lending is partially offset by other banks in the local area. Strahan and Weston (1998) find results along the same line. They argue that cross-section small business lending increases for every dollar of total assets for banks with an asset size smaller than $300 million. For larger companies small business lending slowly increases, but the importance in the portfolio decreases. Featherstone (1996) confirms this result when looking at the effect of bank consolidation on bank lending behavior in rural areas; specifically in the agricultural sector. He finds that, in general, consolidated banks are lending larger and more profitable.
Contrary, Berger and Black (2011) discuss the usage of “hard” and “soft” information by banks, where “hard” information denotes quantitative information which is easily transferred to others and “soft” information is qualitative information that is hard to quantify and communicated to others. In their paper they analyze bank size and the use of different technologies in small business lending and want to examine if bank size influences the focus and comparative advantage of a bank. They use the same data as Han et al. (2015) and they link it with the December 1997 Call Report data for banks, as they use the size of lending institutions for their organizational form. Their findings suggest that large banks have a comparative advantage when it comes to “hard” lending technologies, such as fixed-asset lending, but this advantage only accounts for small and large firms, not the average sized firms. Furthermore, small banks have an advantage in relationship lending. This advantage is the strongest for lending
9
to the largest firms. This means that bank consolidation may not reduce credit availability to small firms. Petersen and Rajan (2002) report findings along the same lines. They conclude that small business lending increases due to improvements in communicating “hard” information, which leads to fewer poor decisions in lending without personal contact.
2.3 Market power
Existing literature shows ambiguous results on the effect of an increase in market power. Where Berger and Hannan (1989) find a positive effect of market power on profits, Keely (1990) finds a negative effect. Berger and Hannan (1989) find that firms with significant market power pay 25 to 100 basis points lower deposit rates to their customers compared to firms with low market power. In their research, they look at 470 banks in 195 local banking markets over a 2.5 year period and try to determine if deposit rates move in favor of customers or not. Their findings suggest that, if a bank manages to increase their market power, for example with a merger, it is able to charge prices that increase profit margins.
Berger and Hannan (1989) extend their research in 1998 by examining whether firms with high market power charge higher prices, while minimizing costs. They look at a sample of over 5000 U.S. banks within a period of 10 years. They measure efficiency by looking at the predicted costs of an efficient bank compared to that of actual banks in the sample. Results suggest that firms with higher market power do not obtain higher profits since costs are not minimized. A possible explanation for this is that managers tend to avoid cost minimization and instead shirk or spend resources to maintain or increase market power (Tullock, 1967; Posner, 1975). However, it is hard to measure whether using these resources has a direct effect on profits.
Boyd and Nicoló (2005) find similar results to those of Berger and Hannan in 1989 and 1998. Boyd and Nicoló (2005) examine the effect of competition on risk taking by banks. Their findings imply an increase of risk taking by firms with more market power, which leads them to increase interest rates and increase bankruptcy risk for borrowing firms. Contrary, Keely (1990) finds the opposite. Banks with more market power, measured as a larger market-to-book ratio, tend to hold more capital than others and keep interest rates on a lower point. According to Keely (1990) this is due to the anticompetitive measures taken by the government. Banks with large market power can become more valuable by becoming “too big to fail” and granting them deposit insurance.
10
Moreover, Sapienza (2002) finds a positive relationship between market concentration and prices in banking. In his research, Sapienza (2002) investigates the Italian market to see how loan contracts for manufacturing firms change over time. He finds that small mergers are actually improving loan terms. However, the bigger the merge becomes, the greater is the monopoly effect and the increase in the cost of loans.
Further, Erel (2009) finds a decrease in spreads and interest rates after a merger, due to the efficiency gains being passed on to the borrowing company. In his paper he analyzes nearly all large banks and a sample of small and medium-sized banks in the United States over the period 1990 to 2000. He measures the effect of a merger up to three years after it occurs, thereby following the research of Focarelli and Panetta (2003) who find that efficiency gains are fully realized after three years.
In contrast to previous findings, Scott and Dunkelberg (2003) find that bank mergers that affected about 25% of the firms in their data had no significant change on the ability to obtain a loan. Also, this did not change their existing contracts. However, it did increase the fees for the banking services.
Previous studies on small firm lending and loan prices around mergers mainly use data from the 80’s until the early 2000’s. In this paper I expand this period to examine if there is a change over time. Also the banking landscape has changed since the 90’s, especially due to the crisis from 2007/2008 (Grosse and Schumann, 2014). Further, regulations have increased and banks are less prone to risks (Heid, 2007). Therefore I expect to find results along the line of Han et al. (2015) and Berger and Hannan (1989) who find that conducting a merger causes the bank to reduce small firm lending and an increase in spreads due to extended market power, assuming a merger leads to an increase in market power.
3. Empirical strategy
In this section I first explain the objective of this research, along with the different hypotheses and the expected findings. After that, I discuss the required data and the construction of the regressions. Finally, the construction of the various variables is discussed.
11 3.1 Hypothesis
The objective of this research is to empirically examine the effects of a bank merger on the banks’ lending behavior. To investigate this effect, I test the following hypotheses:
Bank consolidation increases the interest spread.
Bank consolidation increases the interest rates charged on loans. Bank consolidation increases bank lending.
Bank consolidation decreases small firm lending.
I expect to find that bank consolidation will increase interest rates charged and spreads, as a result of an increase in market power. This increase is suggested by the papers of Berger and Hannan (1989) and Sapienza (2002). Bank lending is also expected to increase as a consolidated firm is expected to aim on lending to larger firms, which usually require larger loans (Featherstone, 1996). I expect findings in line with results of Boyd and Nicoló (2005) who find a decrease in small firm lending. Additionally, I expect the merger to cause banks to take more risk and exploit their market power.
3.2 Methodology
The methodology from papers by Erel (2009) and Sapienza (2002) are used as a basis. With these studies as building blocks, the different hypotheses are tested by the following equations:
𝑆𝑝𝑟𝑒𝑎𝑑𝑖,𝑘,𝑡 = 𝛼 + 𝛽1𝐴𝑓𝑡𝑒𝑟𝑀𝑒𝑟𝑔𝑒𝑟𝑘,𝑡 + 𝛽2𝐴𝑓𝑡𝑒𝑟𝑀𝑒𝑟𝑔𝑒𝑟𝑘,𝑡∗ 𝑆𝑖𝑧𝑒𝑅𝑎𝑡𝑖𝑜𝑘,𝑡 + 𝜋1 𝑋𝑖,𝑘,𝑡 + 𝜋2 𝑌𝑘,𝑡−1+ 𝑦𝑡+ 𝑓𝑘 + 𝑧𝑖+ 𝜀𝑖,𝑘,𝑡 (1) 𝑅𝑎𝑡𝑒𝑖,𝑘,𝑡 = 𝛼 + 𝛽1𝐴𝑓𝑡𝑒𝑟𝑀𝑒𝑟𝑔𝑒𝑟𝑘,𝑡+ 𝛽2𝐴𝑓𝑡𝑒𝑟𝑀𝑒𝑟𝑔𝑒𝑟𝑘,𝑡∗ 𝑆𝑖𝑧𝑒𝑅𝑎𝑡𝑖𝑜𝑘,𝑡 + 𝜋1 𝑋𝑖,𝑘,𝑡 + 𝜋2 𝑌𝑘,𝑡−1+ 𝑦𝑡+ 𝑓𝑘+ 𝑧𝑖 + 𝜀𝑖,𝑘,𝑡 (2) 𝑆𝐹𝑙𝑒𝑛𝑑𝑖𝑛𝑔𝑖,𝑘,𝑡 = 𝛼 + 𝛽1𝐴𝑓𝑡𝑒𝑟𝑀𝑒𝑟𝑔𝑒𝑟𝑘,𝑡+ 𝛽2𝐴𝑓𝑡𝑒𝑟𝑀𝑒𝑟𝑔𝑒𝑟𝑘,𝑡∗ 𝑆𝑖𝑧𝑒𝑅𝑎𝑡𝑖𝑜𝑘,𝑡+ 𝜋1 𝑋𝑖,𝑘,𝑡 + 𝜋2 𝑌𝑘,𝑡−1 + 𝑦𝑡+ 𝑓𝑘 + 𝑧𝑖 + 𝜀𝑖,𝑘,𝑡 (3) 𝐿𝑒𝑛𝑑𝑖𝑛𝑔𝑖,𝑘,𝑡 = 𝛼 + 𝛽1𝐴𝑓𝑡𝑒𝑟𝑀𝑒𝑟𝑔𝑒𝑟𝑘,𝑡 + 𝛽2𝐴𝑓𝑡𝑒𝑟𝑀𝑒𝑟𝑔𝑒𝑟𝑘,𝑡 ∗ 𝑆𝑖𝑧𝑒𝑅𝑎𝑡𝑖𝑜𝑘,𝑡+ 𝜋1 𝑋𝑖,𝑘,𝑡 + 𝜋2 𝑌𝑘,𝑡−1+ 𝑦𝑡+ 𝑓𝑘+ 𝜀𝑖,𝑘,𝑡 (4)
12 Spread is the difference between the effective interest rate on loan i of bank k in year t and the
Treasury rate of the same duration as the loan. Treasury bills with a one-, three- and six month maturity are used, along with one-, two-, three-, five-, seven-, ten-, twenty- and thirty-year Treasury notes. Since the maturity of the loan does not always perfectly match those of the notes or bills, the nearest match is used. Rate, is the effective annual interest rate on loan i of bank k in year t. SFlending is measured as total assets of the borrowing company over total assets of the lending company and Lending is measured by total loans over total assets.
I use a dummy variable along with an interaction term. AfterMerger equals 1 for the first three years after a merger. I use the time period of three years, as Focarelli and Panetta (2003) argue that the gains from a merger are fully internalized after three years. I refer to a “merger” if the acquiring firm owns more than 50% of the shares after the consolidation. The interaction consists of AfterMerger multiplied by SizeRatio. SizeRatio denotes the ratio of total assets of the targeted firm over that of the acquiring firm. SizeRatio is excluded from the regression due to multicollinearity with the interaction between AfterMerger and SizeRatio.
Every equation includes several control variables. On bank-level (𝑌𝑘,𝑡−1) I control for
Ln(Total Assets), which is the logarithm of total assets. Also, Non-Performing Loans Ratio,
which is the non-performing loans over total loans from previous year, is used as a control variable. Using the lagged variable is necessary, because the ratio from the previous period will influence loan pricing in the coming period and not simultaneously. ROA and Capitalization, measured as net income over total assets from last year and equity over total assets respectively are also used as control variables. On a loan-level (𝑋𝑖,𝑘,𝑡) I control for Loan Size, computed as the logarithm of the size of the loan. In addition the Average Spread is included, which is measured as the average spread from that year. Firm fixed (𝑓𝑘) and yearly fixed (𝑦𝑡) effects are both added in every regression, along with a control for loan type (𝑧i).
4. Data and descriptive statistics
In this section I first discuss the different databases I use and how they are combined. After that, the sample focus of this paper is explained. Finally the summary statistics are presented and discussed for all firms and loans used in the sample.
13 4.1 Data collection
The data sources I use in this research are the Thomson ONE database and Wharton Research Data Services (WRDS). Thomson ONE provides data on over one million deals since the 1970’s and covers more than 92.000 syndicated loans since 1982. For the merger section this database reports the target firm, acquirer, date announced and effective, shares acquired and value of the transaction. For the syndicated loans, it reports the borrowing company, along with announcement-, closing- and maturity date, lending firm(s), and proceeds per lending firm. WRDS is used to collect the firm fundamentals of banks granting the loans.
A limitation of the databases is the lack of information. In Thomson ONE the target firms’ total assets is not always reported and there is limited information about the characteristics of borrowing firms in the syndicated loans part. Furthermore, WRDS does not always report firms’ data in consecutive years and therefore some firms are excluded partially or completely. Both databases are matched to get mergers and loans granted in line with the banks’ income statement and balance sheet reports. I match them manually, since the CUSIP identifier possibly refers to different firms within both databases. The only option to merge loans with the balance sheet reports is firm name. Furthermore, both databases report firm names differently and this leads to false negatives during the merge. By doing this manually, I avoid these false negatives.
4.2 Sample selection and statistics
This research includes all US banks from the WRDS database that report data in at least 7 consecutive years over the period between 1987 and 2015. In the beginning of the 1990’s there was an increase in the number of bank mergers due to the removal of intra- and interstate branching restrictions on banks during the 1980’s and beginning of the 1990’s. These restrictions were officially removed by the Riegle-Neal Act in 1994 (Berger et al., 1995). Due to the increase in mergers during this period, Erel (2009) chooses to look at the period between 1990 and 2000. In this paper the time span is extended with data up to 2015.
Table 1 provides statistics of all banks from 1987 to 2015. The mean bank in the sample has total assets of $12,2 billion. The median bank has total assets of $1,0 billion. The total loans have a mean and median of $6,3 billion and $656,1 million respectively. Capitalization is measured by the ratio of equity over total assets. The mean and median of Capitalization are
14
9,5% and 8,8% respectively. Return on Assets is the ratio of net income over total assets from preceding year. The mean and median of this ratio are 0,4% and 1%. Non-Performing Loans
Ratio is equal to loans that are not collected added by loans that are ninety days late, divided by
total loans. The non-performing loans ratio is on average 2,2% and has a median of 1%. Panels B and C describe the summary statistics of the acquiring and non-acquiring banks respectively. Compared to the acquiring banks, the non-acquiring banks have a lower average ($1,9 billion to $17,4 billion) and median ($525 million to $1,6 billion) in total assets. Moreover, they have a lower average return on assets (-0,04% to 0,9%) and a higher non-performing loans ratio (2,5% to 2%).
Table 2 reports summary statistics for the loans included in the sample. The average loan size is $145,0 million. The average length of the loans granted is 50,5 months. The average annual interest rate is 4,34% and tends to decrease as loan size increases. The average spread, calculated as the annual interest rate minus the Treasury rate of the same duration as the loan, is 2%. The average spread also tends to decrease as loan size increases.
5. Results
In this section I present and discuss the results from the different regressions. First I examine the dependent variable Spread. Then I discuss the results for Rate. Afterwards I research the findings for SFlending. Finally I discuss the results from Lending.
5.1 Interest spread
Table 3 shows the results from Equation (1), where Spread is the dependent variable. The sample includes all firms over the period 1987-2015. Note that the number of observations differs across the columns due to missing data of the Total Assets Borrowing Firm and multicollinearity of the loan-fixed effects. All regressions control for year effects and firm-fixed effects. Only Column (6) controls for loan type.
In column (1) AfterMerger is included along with the lagged firm size, lagged non-performing loans ratio and average spread. Column (2) includes the interaction between
AfterMerger and SizeRatio. Here the coefficient AfterMerger is not significantly different from
15
zero, meaning that a merger, regardless of its’ size, on average does not significantly affect the spread of new loans given in the period three years after the merger.
However, the interaction becomes significant at a 10% significance level and a coefficient of 0,121 once the controls Ln(Loan Size), Maturity, ROA and Capitalization are added in column (4). This means the Spread increases on average after a merger depending on the size of the target firm. Furthermore, loan size and the length of the loan show to be significant at a 1% significance level. They both report a negative relation with spread, meaning an increase in loan size or the duration of the loan lead to a decrease in spread. This is in line with the theory of economies of scale (Benston, 1972). The added bank effects are all statistically indifferent from zero.
After adding Ln(Total Assets Borrowing Firm) in column (5) the merger alone again has no significant effect on the spread. The same accounts for column (6) where I add a control for loan type. Note that in the last two columns the amount of observations drops from 26.616 to 8.450 due to the missing data on total assets of the borrowing firm.
The findings in column (1), (2) and (3) are partially in line with those of Scott and Dunkelberg (2003), who report no significant changes in profit margins after a merger. However, the results in column (4), (5) and (6) suggest the opposite of findings by Erel (2009) who finds a decrease in loan spreads, since banks pass their efficiency gains on to the borrowing company. 5.2 Interest rate
Table 4 represents the results for Equation (2) where interest rate is the dependent variable. The sample includes data for all firms over the period 1987-2015. Note that the number of observations differs across the columns due to missing data of the Total Assets Borrowing Firm. All regressions control for year effects and firm-fixed effects, but Column (6) also controls for loan-fixed effects. The setup is the same as in Table 3 and adds new variables to the regressions over the different columns.
Contrary to the findings in Table 3, AfterMerger and AfterMerger*SizeRatio do not report any significant values over the different columns, which indicates that a merged bank does not significantly alter the interest rate charged on loans three years after the merger. This is in line with findings by Scott and Dunkelberg (2003) who find no significant changes in interest rates after a merger.
16
In the first two columns Ln(Total Assets) shows a significant negative value at a 1% significance level. This indicates that larger banks tend to charge a lower interest rate on loan on average. These findings go against the results from Berger and Hannan (1989), assuming an increase in total assets leads to an increase in market power. They argue that a larger firm tends to increase interest rates due to the increase in market power. After I add different variables over the other four columns this value is no longer significant. The same accounts for the Non-Performing
Loans Ratio.
Further, in column (3) to (6) the bank effects are all statistically indifferent from zero like in Table 3. The loan effects do show a significant value at a 1% significance level. Like in Table 3 Loan Size shows a negative value. This is again in line with the theory of economies of scale (Benston, 1972). However, Maturity has a positive relationship with Rate. This indicates that banks on average charge higher interest rates over loans with a longer duration.
Combining the results in Table 4 with those from Table 3 indicates that merged banks keep their interest rates constant in the period three years after the merger stable on average, but increase their spreads. This conforms to findings by Berger and Hannan (1989) who report that an increase in market power leads to an increase in profit margins. This is under the assumption that an increase in total assets leads to an increase in market power.
5.3 Small firm lending
Table 5 represents the results for Equation (3), where SFlending is the dependent variable. It includes all firms in the sample. Here the regressions are different from those in the previous two tables. Ln(Total Assets Borrowing Firm) is excluded as an independent variable, since adding it leads to collinearity. This is because SFlending is equal to the total assets of the borrowing firm divided by the total assets of the lending firm.
In column (1) Ln(Total Assets) shows a coefficient of -0,733, which is significant at a 1% significance level. This suggests that larger banks tend to lend to smaller firms. A 1% increase in total assets leads to a decrease of 73,3 basis points in the small firm ratio. This significance remains over the other columns. The results suggest the opposite of findings by Han et al. (2015) and Berget et al. (1998) who argue that an increase in size leads to a decrease in small firm lending. Non-Performing Loans Ratio also shows a significant relationship with SFlending at a 1% significance level over all the columns, but has a positive coefficient. This means that banks with a higher ratio of non-performing loans tend to lend to larger firms.
17
In column (3) to (5) I add the other independent variables. Ln(Loan Size) shows a significant coefficient at a 1% significance level for all three regressions. This implies that larger loans are on average granted to larger firms. Also capitalization is significant, but at a 10% significance level.
Over all the five columns AfterMerger shows no significant value, but the interaction variable shows a significant relationship with SFlending at a 1% significance level in column (2), (4) and (5). This indicates that in the three years after a merger, the merged bank tends to lend to larger firms on average. This conforms to findings by Han et al. (2015) and Berget et al. (1998) who state that an increase in size leads to a decrease in small firm lending.
5.4 Bank lending
Table 6 shows the findings for Equation (4). The number of observations decreases, because it includes one observation per firm per year. This is to avoid duplicates, as a bank can grant multiple loans in a single year. Furthermore, the dependent variables are only balance sheet items and the average spread charged. The results cover all firms in the sample over the whole period. In column (1) and (2), the after merger dummy shows a statistically significant value at a 1% significance level. The result implies that in the three years after a merger the combined firm tends to increase the total lending over total asset ratio. As the interaction shows no significant coefficient, the size of the merger has no significant effect. These results are against the findings by Keely (1990), who argues that larger firms tend to hold more capital as they are less risk loving. This is under that assumption that an increase in size leads to an increase in market power.
In all columns Non-Performing Loans Ratio shows a significant negative coefficient at a 1% significance level. This indicates that an increase in the share of non-performing loans leads to a decrease in future lending. Furthermore, columns (3) and (4) suggest there is a positive relationship between ROA and Lending. The coefficients are both significant at a 1% significance level and indicate that an increase in returns over total assets lead to an increase in lending. Finally, in the first two columns, Average Spread is statistically significant at a 1% significance level and shows a positive relationship between the two variables. This means that an increase in the average spread leads to an increase of lending over total assets. The significance level becomes 5% after I add both ROA and Capitalization.
18 6. Robustness checks
In this section I divide the sample in several subsamples. First, the loans are separated into a small and large loan group, to investigate if there is a difference between loan sizes. After that, the data are split into two periods.
6.1 Subsamples based on loan size
In this section the sample is divided into small and large loans. Small loans are loans with a maximum value of $75 million per bank. Large loans are defined as loans larger than $75 million. There are several reasons on why the results could differ between loan sizes. Erel (2009) argues that loan size is related to borrower size. Additionally, he claims that acquirers tend to overtake smaller targets. This results in a new, and most likely, smaller loan portfolio added to the acquirers’ loan portfolio.
6.1.1 Interest spread
Table 7 displays the results for the regressions on Equation (1), with Spread as the dependent variable. The regressions are constructed similar to the last three columns of Table 3, where I only exclude Ln(Total Assets Borrowing Firm) and the control for loan type in column (4) and add them in column (5) and (6) respectively. Note the decrease in observations when both are added. This potentially changes the estimates in magnitude and statistical significance, but this does not seem to be the case in my results. Column (1) to (3) show results for small loans and (4) to (6) show the results for large loans.
The results in column (1) to (3) are similar to those in Table 3. The merger dummy is not statistically different from zero in all the three columns. However, unlike findings in Table 3, the interaction variable shows no statistically significant value at a 10% significance level. This suggests that a merger has no significant effect on the spread of small loans in the first three years after a merger, regardless of the size of the target firm. Additionally, Non-Performing Loans
Ratio shows a negative coefficient that is significant at a 1% significance level. The relationships
between Spread and the remaining variables remain the same as in Table 3.
In column (4) to (6) the results tend to slightly deviate from Table 3. In column (4)
Ln(Total Assets) shows a significant value at a 5% significance level which is not the case in
19
Dissimilar to the small loans subsample, the interaction variable shows a significant value at a 5% significance level in column (4) and (6) indicating that for large loans the merger and its’ size do affect the spread on loans in the period three years after the merger. Furthermore, the
Non-Performing Loans Ratio alters from a negative coefficient in column (1) to (3), to a positive
coefficient in column (4) to (6). This indicates that an increase in the Non-Performing Loans
Ratio leads to a decrease in Spread for small loans, but an increase for large loans. 6.1.2 Interest rate
Table 8 shows the results for Equation (2), where Rate is the dependent variable. It follows the same setup as Table 7.
The results in column (1) to (3) are similar to findings in Table 4. Both the dummy and the interaction are statistically indifferent from zero. However, like in Table 7, where Spread is the dependent variable, Non-Performing Loans Ratio shows a negative relationship with the dependent variable at a 1% significance level. This implies that banks are lowering their interest rates, once a larger portion of their portfolio becomes non-performing.
In column (4) the interaction shows a positive coefficient which is significant at a 10% significance level, indicating that for large loans the interest rates charged increases after the merger on average. For this increase accounts the larger the target bank, the larger the increase in interest rates charged. The significance is no longer present in column (5) and (6). Further, like in Table 7 the Non-Performing Loans Ratio shifts from a negative coefficient for the small loans to a positive coefficient for the large loans.
6.1.3 Small firm lending
Table 9 reports the results for Equation (3), with SFlending as the dependent variable. The setup is similar to that of Table 7, but here only the last two columns of Table 5 are included. This is because Ln(Total Assets Borrowing Firm) is not included.
Column (1) and (2) show the findings for small loans. The other two columns report findings for large loans.
Column (1) and (2) show no significant relationship between AfterMerger and SFlending, but do show a significant one between the interaction and SFlending at a 10% significance level. These findings are in line with Table 5, but the relationship is statistically less strong. Moreover the first two columns indicate that there is no significant relationship between firm size and small
20
firm lending when it comes to small loans, unlike in Table 5. This indicates that firm size does not influence the firm size small loans are granted to. In column (1) Maturity is significant at a 10% significance level, indicating that small loans with a longer maturity are more often granted to larger firms.
The results in the last two columns report the same relationships in Table 5. Compared to the first two columns the independent variables tend to have a more significant impact on small firm lending for large loans than for small loans. Ln(Total Assets), Non-Performing Loans Ratio,
Ln(Loan Size) and Capitalization are all statistically significant for large loans and not for small
loans. This indicates that firm and loan effects hardly influence small firm lending. 6.2 Subsamples based on time periods
In this section the sample is divided into two samples based on different time periods, to examine if the effect of a merger has changed after the period that Erel (2009) covers. Period A is the period that Erel (2009) covers, which is from 1990 to 2000. Period B is from 2001 to 2012, which covers the added data.
6.2.1 Spread
Table 10 displays the results for the regressions on Equation (1), with Spread as the dependent variable. The regressions are constructed in a similar way to those in section 6.1. Column (1) to (3) show results for the period A and (4) to (6) show those for period B.
In period A I find different results than Erel (2009) does. In his paper the results indicate a significant value for the merger dummy and interaction. Table 10 only reports significance for the interaction in column (2), (3), (4) and (6). Contrary to the findings by Erel (2009), the coefficient is positive. This indicates an increase in spread after a merger and the larger the size of the merger compared to the total assets of the acquirer, the larger the effect. Compared to the results in Table 3, firm size and the non-performance ratio indicate a significant coefficient in column (1) which is not the case in Table 3. This statistical significance is no longer present in column (2) and (3).
The results in the last three tables are comparable to those in Table 3. This is possibly caused by the large amount of observations in period B. Compared to period A, the results over
21
period B are similar. Only the significance of the coefficients from the interaction decreases over time.
6.2.2 Interest rate
Table 11 displays the results for the regressions on Equation (2), with Rate as the dependent variable. The regressions are constructed identically to those in section 6.2.1.
In the second column, the interaction is significant at a 5% significance level and the merger dummy is not statistically different from zero. In Table 4 the merger dummy and the interaction both show no value that is significantly different from zero. The interaction shows a positive value and indicates that after a merger the interest rate increases, depending on the size of the acquired firm. Again, this can be caused by the small share of observations in period A. In column (3) the interaction is only significant at a 10% significance level.
In column (4) to (6) the results are similar to the findings in Table 4. Compared to period A the effect of a merger has decreased as both the dummy and the interaction are statistically indifferent from zero. Other than that the effects remain similar.
6.2.3 Small firm lending
Table 12 shows the findings for the regressions on equation (3), where SFlending is the dependent variable. The regressions are constructed similarly to those in section 6.1.3. However, the sample is now divided into period A and B. The first two columns report findings on period A and the last two report them on period B.
The findings in period A differ from those in Table 5. The interaction variable shows no statistically significant value, where this is the case in Table 5. The same accounts for Ln(Total
Assets), Non-Performing Loans Ratio, Average Spread and Capitalization.
However, findings in column (3) and (4) are almost the same as in Table 5 when it comes to statistical significance of the coefficients. The results indicate that over time firm characteristics have become more impactful when it comes to the size of firms that banks are focusing their business on. Do note the difference in the amount of observations in both periods as this can possibly bias the results.
22 6.2.4 Lending
Table 13 represents the findings of the regressions on Equation (4), where Lending is the dependent variable. The regressions are constructed as in the last two columns in Table 6, only now the sample is divided into period A and B. Column (1) and (2) show the results for period A, where column (3) and (4) show the results for period B.
In the first two columns the results are similar to those in Table 6. However,
Capitalization shows a negative relationship with Lending in period A unlike Table 6. This
shows that in period A firms that are more capitalized tend to have a lower ratio of total loans over total assets. In period B this relationship becomes positive. Furthermore, in column (3) the merger dummy shows a statistically significant value at a 10% significance level, which indicates that after merging the combined firm tends to increase Lending. Other than that the findings remain similar.
7. Conclusion
The aim of this paper is to provide evidence on the effect of bank consolidation on bank lending behavior. This is done by dividing lending behavior into four different measures: Spread, Rate,
SFlending and Lending. Spread is the difference between the effective interest rate charged on a
loan, minus the Treasury rate of the same duration as the loan. Rate equals the effective annual interest rate. SFlending denotes the total assets of the borrowing company over total assets of the lending company. Lending equals total loans over total assets. These four variables are used as dependent variables in the constructed regression to examine what the effects of a merger are. Findings indicate an increase in the spread on loans in several regressions after I control for the size of the borrowing firm and loan type. This effect only seems to be present for large loans, as it is no longer statistically significant for loans smaller than $75 million. The size of the acquired firm tends to be of significant importance, as the interaction between the merger dummy and the size ratio shows a statistically significant coefficient. However, I find a change of the merger effect over time. In the period 1990 to 2000 the interaction reports a significant value at a 1% significance level, but becomes statistically indifferent from zero before I control for Loan
Type for the period 2001 to 2012. The results from the different subsamples connect with papers
23
market power leads to an increase in profit margins.
In the case of interest rates, the results suggest different findings than for the spread. Only for large loans the coefficient is significantly different from zero before I control for Ln(Total
Assets Borrowing Firm) and Loan Type. After I add these two variables to the regression in
period A the coefficient is also significantly different from zero. This positive effect comes from the fact that a firm conducts a merger in both cases and SizeRatio shows significant importance as only the interaction is statistically significant. This suggest that the size of the merger is important for large loans, but not for small loans. For the different periods the results are similar to the findings over the different periods on the spread. In the first period the interaction is significant and positive, but in the second period this is no longer true for the interest rate. The outcomes from the whole sample suggest that a merger does not tend to affect interest rates at all. The results from SFlending indicate there is a decrease small firm lending as the interaction shows a positive and significant relationship with SFlending. Once I divide the sample into small and large loans, the interaction variable still reports a positive and significant coefficient. This suggests that after a merger small firm lending decreases and loans are granted to larger firms on average compared to the bank that grants the loan. This effect increases as the size ratio increases. However, AfterMerger * SizeRatio is no longer significantly different from zero in the period 1990-2000. These findings indicate that only the mergers in recent periods have caused a decrease in small firm lending within three years after a merger.
Findings suggest that after conducting a merger, firms tend to increase their lending before I control for ROA and Capitalization. This does not depend on how large the acquired total assets are compared to that of the acquirer. After dividing the sample into the two periods, the results are only significant for the merger dummy in the most recent period.
Linking the findings to the existing literature, they are not completely in line with previous papers. Papers by Erel (2009) and Keely (1990) find a decrease in spread and interest rate respectively. The decrease in spread is caused by the efficiency gains being passed on to the customers. The lower interest rates are caused by an increase in market power. Banks with more market power tend to be more valuable due to anticompetitive measures taken by the government. These measures cause an increase in market power to be harder to achieve. Also, a bank can increase its value by becoming “too big to fail”. This causes them to become less risk loving. The results by Scott and Dunkelberg (2003) are similar to the findings on small firm
24
lending in period A. They show no significant difference after a merger. However, the findings in the most recent period are in line with the results of Han et al. (2015) and Strahan and Weston (1998). They argue that small firm lending tends to decrease after a merger. The increase in lending can be lead back to the results from Boyd and Nicoló (2005), who argue that larger firms are more willing to take risks.
All in all, this research suggests that a merger leads to an increase in the spread, interest rate and lending, but a decrease in small firm lending and lending in general within three years after the merger. This shows that the government should be careful with allowing banks to merge. It can cause small firms to go bankrupt, due to limited credit availability. Furthermore, it can lead to banks becoming “too big to fail” and exploit the government safety net. However, it does not lead to efficiency gains being passed on to firms that are granted a loan.
Regarding this research, there are several limitations. The selected sample is biased towards large firms due to limited data access. The syndicated loans from Thomson ONE only cover a limited share of the total loans that banks are granting, which can lead to a sample selection bias. Furthermore, this research relies on the assumption that an increase in total assets leads to an increase in market power. This does not have to be the case as a merger can also serve to penetrate a new market. For further research the Herfindahl-index can be used as an alternative for this assumption to measure the market concentration. Also, a more complete loan database could be used to expand the loan specific controls, such as fixed or floating rates and collateral. Along with this database, a larger number of banks can then be included in the sample which makes it possible to examine if there is a difference between large and small banks.
25 References
Acs, Z. (1999). Are small firms important?. Boston: Kluwer Academic.
Benston, G. (1972). Economies of Scale of Financial Institutions. Journal of Money, Credit and Banking, 4(2), 312.
Berger, A. and Black, L. (2011). Bank size, lending technologies, and small business finance. Journal of Banking & Finance, 35(3), 724-735.
Berger, A., Demsetz, R. and Strahan, P. (1999). The consolidation of the financial services industry: Causes, consequences, and implications for the future. Journal of Banking &
Finance, 23(2-4),135-194.
Berger, A. and Hannan, T. (1989). The Price-Concentration Relationship in Banking. The Review
of Economics and Statistics, 71(2), 291.
Berger, A. and Hannan, T. (1998). The Efficiency Cost of Market Power in the Banking Industry: A Test of the “Quiet Life” and Related Hypotheses. Review of Economics and Statistics, 80(3), 454-465.
Berger, A., Kashyap, A., Scalise, J., Gertler, M. and Friedman, B. (1995). The Transformation of the U.S. Banking Industry: What a Long, Strange Trip It's Been. Brookings Papers on
Economic Activity, 1995(2), 55.
Berger, A., Saunders, A., Scalise, J., & Udell, G. (1998). The Effects of Bank Mergers and Acquisitions on Small Business Lending. Journal of Financial Economics, 50(2),
187-229.
Boyd, J. H., & De Nicolo, G. (2005). The theory of bank risk taking and competition revisited. The Journal of finance, 60(3), 1329-1343.
Cornett, M., Hovakimian, G., Palia, D. and Tehranian, H. (2003). The impact of the manager– shareholder conflict on acquiring bank returns. Journal of Banking & Finance, 27(1),
26
Cornett, M., McNutt, J. and Tehranian, H. (2006). Performance Changes Around Bank Mergers: Revenue Enhancements versus Cost Reductions. Journal of Money, Credit, and Banking, 38(4), 1013-1050.
Featherstone, A. (1996). Post-Acquisition Performance of Rural Banks. American Journal of
Agricultural Economics, 78(3)728-733.
Focarelli, D. and Panetta, F. (2003). Are Mergers Beneficial to Consumers? Evidence from the Market for Bank Deposits. American Economic Review, 93(4), 1152-1172.
Grosse, S. and Schumann, E. (2014). Cyclical behavior of German banks' capital resources and the countercyclical buffer of Basel III. European Journal of Political Economy,
34, S40-S44.
Han, L., Zhang, S. and Greene, F. (2015). Bank market concentration, relationship banking, and small business liquidity. International Small Business Journal.
Hadlock, C., Houston, J. and Ryngaert, M. (1999). The role of managerial incentives in bank acquisitions. Journal of Banking & Finance, 23(2-4), 221-249.
Heid, F. (2007). The cyclical effects of the Basel II capital requirements. Journal of Banking
& Finance, 31(12), 3885-3900.
Keeley, M. (1990). Deposit Insurance, Risk, and Market Power in Banking. The American
Economic Review, 80(5), 1183-1200.
McCord, R., Prescott, S., & Sablik, T. (2015). Explaining the Decline in the Number of Banks
since the Great Recession (EB15-03). Retrieved from Community Banking website:
https://www.communitybanking.org/documents/Session3_Paper2_Prescott-
Summary.pdf
O'Hara, M., & Shaw, W. (1990). Deposit insurance and wealth effects: the value of being “too big to fail”. The Journal of Finance, 45(5), 1587-1600.
27
Peek, J. and Rosengren, E. (1998). Bank consolidation and small business lending: It's not just bank size that matters. Journal of Banking & Finance, 22(6-8), 799-819.
Peristiani, S. (1993). The Effects of Mergers on Bank Performance. Federal Reserve of New
York Studies on Excess Capacity in the Financial Sector. 1993.
Petersen, M. and Rajan, R. (2002). Does Distance Still Matter? The Information Revolution in Small Business Lending. The Journal of Finance, 57(6), 2533-2570.
Piloff, Steven J., and Anthony M. Santomero. The value effects of bank mergers and acquisitions. Bank Mergers & Acquisitions. Springer US, 1998. 59-78.
Posner, R. (1975). The Social Costs of Monopoly and Regulation. Journal of Political Economy, 83(4), pp.807-827.
Prompitak, D. (2009). The impacts of bank mergers and acquisitions (M&A’s) on bank
behaviour (Doctoral dissertation, University of Birmingham, Birmingham, United
Kingdom). Retrieved from http://etheses.bham.ac.uk/779/1/Prompitak10PhD.pdf
Sapienza, P. (2002). The Effects of Banking Mergers on Loan Contracts. The Journal of Finance, 57(1), 329-367.
Savage, D. T. (1991). Mergers, branch closings, and cost savings. Board of Governors of the Federal Reserve System.
Scott, J. and Dunkelberg, W. (2003). Bank Mergers and Small Firm Financing. Journal of
Money, Credit, and Banking, 35(6a), 999-1017.
Strahan, P. and Weston, J. (1998). Small business lending and the changing structure of the banking industry. Journal of Banking & Finance, 22(6-8), 821-845.
Subrahmanyam, V., Rangan, N. and Rosenstein, S. (1997). The Role of outside Directors in Bank Acquisitions. Financial Management, 26(3), 23.
28
Tullock, G. (1967). The welfare costs of tariffs, monopolies, and theft. Economic Inquiry, 5(3),
29 Appendix A: Tables
Table 1: Statistics of banks. The statistics shown in panel A describe the complete sample of banks used in this research from 1987 to 2015. Panel B describes the statistics of acquiring banks. These are banks that have acquired at least one bank during this time period. Panel C lists the same statistics for banks that have not merged with another bank in these years. Total Assets are gross total assets of the bank taken from the balance sheet. Ln(Total Assets) is the logarithm of Total Assets. Total Loans displays the total loans outstanding and Total Deposits show the total deposits a bank has. Net Interest Income represents total interest and dividends received from earning assets minus total interest paid for use of debt and deposits. Capitalization is measured as equity over total assets. Return on Assets is the ratio of net income over total assets form preceding year. Non-Performing Loans Ratio is measured as loans that are not collected added by loans that are ninety days late, over total loans.
Mean Median Std Dev. Minimum Maximum
No. of observations
Panel A:All Banks
Total Assets ($ million) 12.190 1.035 93.124 0,003 2.573.126 15.067
Ln(Total Assets) 7,22 6,94 1,66 -5,81 14,76 15.067
Total Loans ($ million) 6.320 656,10 39.601 0,000 892.647 14.751
Total Deposits ($ million) 7.581 790,29 50.542 0,000 1.363.427 15.032
Net Interest Income ($ million) 329,99 31,12 2.260 -518,77 51.523 13.745
Capitalization 0,095 0,088 0,045 0,00003 1,00 14.989
Return on Assets 0,004 0,010 0,195 -19,34 3,47 15.057
Non-Performing Loans Ratio 0,022 0,010 0,111 0,000 10,97 13.537
Panel B: Acquiring Banks
Total Assets ($ million) 17.438 1.599 113.895 0,580 2.573.126 9.994
Ln(Total Assets) 7,65 7,38 1,65 -0,545 14,76 9.994
Total Loans ($ million) 8.957 1.033 48.375 0,000 892.647 9.763
Total Deposits ($ million) 10.759 1.211 61.721 0,000 1.363.427 9.978
Net Interest Income ($ million) 480,08 47,06 2.788 -518,77 51.523 8.929
Capitalization 0,092 0,087 0,042 0,0003 1,00 9.958
Return on Assets 0,009 0,010 0,023 -1,93 0,316 9.986
Non-Performing Loans Ratio 0,020 0,010 0,068 0,000 2,68 8.809
Panel C:Non-Acquiring Banks
Total Assets ($ million) 1.850 525 6.394 0,003 127.633 5.073
Ln(Total Assets) 6,38 6.263 1,33 -5,81 11,76 5.073
Total Loans ($ million) 1.160 338 4.150 0,000 76.279 4.988
Total Deposits ($ million) 1.305 400 4.205 0,000 81.821 5.054
Net Interest Income ($ million) 51,72 16,43 223 -13,70 6.773 4.816
Capitalization 0,100 0,090 0,051 0,00003 0,989 5.031
Return on Assets -0,004 0,008 0,334 -19,34 3,47 5.071
Non-Performing Loans Ratio 0,025 0,011 0,163 0,000 10,97 4.728
30
Table 2 Summary statistics loans. The statistics describe 32.713 loans which were successfully linked to a bank in the sample from 1987 to 2015. “All Loans” denotes all loans covered in the sample, where the following rows cover different subsamples based on loan size.
Average loan size (Million $) Average months until maturity Average interest rate (%) Average spread (%) No. of observations All Loans 144,97 50,45 4,34 1,98 32.713
Loan Size ≤ $25 million 12,95 48,51 4,94 2,74 8.043
$25 Million < Loan Size ≤ $75
Million 49,90 51,06 4,49 2,24 9.336
$75 Million < Loan Size ≤ $250
Million 144,69 51,77 3,97 1,64 11.075
Loan Size >$250 Million 603,41 49,33 4,07 1,19 4.259
31 Table 3: The effect of bank mergers on loan spread. The dependent variable is Spread, the effective annual interest rate on the loan deducted by the Treasury rate of the same duration. AfterMerger is a dummy variable equal to 1 for the first three years after a merger. Ln(Total Assets) is the logarithm of total assets. Non-Performing Loans Ratio is non-performing loans over total loans from the previous period. Average Spread is the average spread per year for all banks except the observed bank. AfterMerger*SizeRatio is an interaction between the merger dummy and size ratio. SizeRatio is total assets of the acquired firm over total assets of the acquiring firm. Ln(Loan Size) is the logarithm of the loan size. Maturity is the amount of months to maturity of the loan. ROA is net income over total assets at the end of previous year. Capitalization is equity over total assets. Ln(Total Assets Borrowing Firm) is total assets of the borrowing firm.
Dependent variable: Spread
(1) (2) (3) (4) (5) (6) AfterMerger -0,024 -0,024 -0,042 -0,040 -0,000 -0,028 (-0,71) (-0,70) (-1,32) (-1,25) (-0,01) (-0,68) Ln(Total Assets) -0,109** -0,109** 0,071 0,062 0,040 0,046 (-2,02) (-2,02) (1,36) (1,19) (0,54) -(0,68) Non-Performing Loans Ratio -4,713** -4,684** -2,569 -2,161 3,363 1.497 (-2,13) (-2,09) (-1,20) (-1,00) (1,08) (0,53) Average Spread 0,775*** 0,775*** 0,681*** 0,683*** 0,526*** 0,461*** (26,01) (25,98) (24,28) (24,33) (12,21) (11,77) AfterMerger * SizeRatio 0,007 0,121* 0,179* 0,188* (0,09) (1,67) (1,68) (1,96) Ln(Loan Size) -0,338*** -0,338*** -0,151*** -0,140*** (-56,16) (-56,19) (-14,27) (-14,25) Maturity -0,083*** -0,083*** -0,156*** -0,199*** (-18,93) (-18,92) (-20,76) (-25,07) ROA -0,207 -0,855 1,145 -0,488 (-0,08) (-0,31) (0,30) (-0,14) Capitalization 0,716 0,712 1,128 0,213 (0,53) (0,52) (0,60) (0,12) Ln(Total Assets Borrowing Firm) -0,130*** -0,139*** (-16,06) (-18,60)
Year effects Yes Yes Yes Yes Yes Yes
Firm-Fixed effects Yes Yes Yes Yes Yes Yes
Loan Type No No No No No Yes
Adjusted R² 0,260 0,260 0,347 0,347 0,386 0,494
No. of observations 26.616 26.616 26.616 26.616 8.450 8.450
32 Table 4: The effect of bank mergers on loan interest rate. The dependent variable is Rate, the effective annual interest rate on the loan deducted by the Treasury rate of the same duration. AfterMerger is a dummy variable equal to 1 for the first three years after a merger. Ln(Total Assets) is the logarithm of total assets. Non-Performing Loans Ratio is non-performing loans over total loans from the previous period. Average Spread is the average spread per year for all banks except the observed bank. AfterMerger*SizeRatio is an interaction between the merger dummy and size ratio. SizeRatio is total assets of the acquired firm over total assets of the acquiring firm. Ln(Loan Size) is the logarithm of the loan size. Maturity is the amount of months to maturity of the loan. ROA is net income over total assets at the end of previous year. Capitalization is equity over total assets. Ln(Total Assets Borrowing Firm) is total assets of the borrowing firm.
Dependent variable: Rate
(1) (2) (3) (4) (5) (6) AfterMerger -0,026 -0,027 -0,045 -0,043 -0,017 -0,045 (-0,79) (-0,81) (-1,43) (-1,37) (-0,38) (-1,13) Ln(Total Assets) -0,183*** -0,181*** 0,005 -0,002 0,023 0,028 (-3,46) (-3,41) (0,09) (-0,04) (0,32) (0,44) Non-Performing Loans Ratio -5,361** -5,453** -3,283 -2,939 3,341 1,683 (-2,47) (-2,48) (-1,57) (-1,40) (1,11) (0,62) Average Spread 0,695*** 0,695*** 0,603*** 0,605*** 0,451*** 0,388*** (23,76) (23,72) (22,05) (22,09) (10,88) (10,36) AfterMerger * SizeRatio -0,022 0,102 0,112 0,120 (-0,29) (1,44) (1,09) (1,30) Ln(Loan Size) -0,334*** -0,335*** -0,148*** -0,137*** (-57,07) (-57,09) (-14,55) (-14,62) Maturity 0,117*** 0,117*** 0,090*** 0,040*** (27,54) (27,54) (12,48) (5,31) ROA -2,083 -2,627 1,814 0,109 (-0,79) (-0,99) (0,50) (0,03) Capitalization 0,455 0,452 1,359 0,693 (0,34) (0,34) (0,75) (0,42) Ln(Total Assets Borrowing Firm) -0,136*** -0,142*** (-17,38) (-19,92)
Year effects Yes Yes Yes Yes Yes Yes
Firm-Fixed effects Yes Yes Yes Yes Yes Yes
Loans-Fixed effects No No No No No Yes
Adjusted R² 0,662 0,662 0,706 0,706 0,778 0,820
No. of observations 26.617 26.617 26.617 26.617 8.450 8.450
