Bank loan refinancing: the effect on stock prices in Europe

Bank loan refinancing: the effect on stock prices in Europe

Caspar A. Slagboom (S1613359)

MSc Thesis Business Administration – Finance

University of Groningen

Supervisor: dr. H. Gonenc

June, 2013

Abstract

This study investigates the influence of bank loan refinancings on stock

prices of 210 refinancing deals in Europe. In addition, the effect of the

recent credit crunch (2008-2010) and the liquidity position of a firm are

tested for their influence on the stock price reactions. Using the event

study methodology, abnormal returns for refinancing deals in the period

2005 – 2013 are calculated. This paper finds that the effect of bank loan

refinancings on stock prices is small. Also, the liquidity position and the

credit crunch have no significant influence on abnormal returns. Yet, this

research finds that the loan type has an impact on stock prices; when the

refinancing is a term loan, instead of a revolving credit facility, abnormal

returns are positive. Finally, this paper finds that firm size has a negative

effect on stock prices reactions.

Key words:

2

1. Introduction

How do shareholders of a company react to an announcement of a bank loan refinancing? Does the liquidity position of a firm effect the stock price reactions to such announcements? This paper examines the effects of bank loan refinancing announcements on shareholder’s value. A bank loan refinancing is a replacement of an old bank loan with a new bank loan; the new loan repays the old loan. Bank loan refinancings occur on a frequent basis.

Numerous research has explored the effect of bank loans and their effect on stock prices, but research lacks evidence on bank loan refinancings. Since a bank loan refinancing is legally a new bank loan, and refinancings show equal characteristics to bank loan renewals, literature on those two forms of financing is employed. Research on new bank loans demonstrates that announcements of new bank loans send positive signals to the market; stock prices increase (James (1987), Diamond (1984), Mikkelson and Partch (1986)). These stock price reactions are explained by the unique relationship between a bank and a company (James, 1987). Banks invest in fundamental credit analysis and information-gathering technology (Benston and Smith (1976), Campbell and Kracaw (1980)). Further, they gain access to private information about their customers as result of close relationships. This results in positive stock price reactions for bank loans. Other forms of external corporate financing are found to produce negative or neutral effects. Fama (1985) explains that announcements of bank loan renewals send clear positive signals to the market because banks are already familiar with the customer. Lummer and McConnell explain that abnormal returns following the announcement of a renewal, depend on the contract differences between the old and new loan. So, if new bank loans and bank loan renewals send clear, positive or negative, signals to the market, how will the announcement of bank loan refinancings have effect on shareholder’s value creation?

3 of the loan has an effect on the stock price reactions. When a bank loan refinancing is a term loan its abnormal returns are significantly higher than when it is a RCF or a mix of both loan types. Furthermore, when a refinancing is announced by an Italian company, the abnormal returns are also found to be higher. Finally, this paper shows that firm size, measured in total assets, has a negative influence on the stock price reactions.

This research contributes to the existing literature by investigating refinancings separately from other new loans. As explained, much research has been conducted about bank loans and bank loan announcements, but the literature investigating bank loan refinancings is scarce. Because a refinancing is in legal terms a new bank, the existing literature includes bank loan refinancings in new loans. Bank loan refinancings are not investigated individually. In addition, previous literature has investigated bank loan renewals. A bank loan renewal is an extension to the maturity of the existing bank loan. Hence, a loan renewals has resemblance to a bank loan refinancing. Although, bank loan refinancings have been investigated as part of new bank loans, and they show similarities to bank loan renewals, it is hard to establish the contribution of this paper to the existing literature. Given that existing literature on bank loan refinancings is scarce, the results from this paper cannot directly be compared with previous works.

4

2. Literature review

In this section, an overview of the ideas and results of previous literature is given. Using this prior research, the general theory behind the information content of the announcements of bank loans and refinancings is explained. This includes the theories that support the reasoning behind the hypotheses. First of all, the concept of refinancing and the reasons why a company refinances its outstanding bank loans will be shortly explained.

2.1 Definition

Bank loan refinancing is the replacement of an old bank loan with a new bank loan. In essence, the new loan repays the old loan. In the case of a refinancing, an entire new debt contract is composed and the old contract is terminated. There are several reasons why a company is prepared to refinance its debt obligations, beyond the predetermined or natural contract limits. Firstly, a company could take advantage of better interest rates. For example, if a current bank loan was contracted in a time interest rates were higher than they are today, a company has an incentive to renegotiate the current loan in order to take advantage of better interest rates. Secondly, a company might want to improve the covenant agreements in the debt contract. This can also be a consequence of the increased risk that the company breaks through its covenants in the future. Another reason for debt refinancing is a company’s improved credit rating. An improved credit rating gives the company a better position in renegotiations. A refinancing can also materialize if there is a mismatch between the loan contract and the credit standing of a company. Further, in many cases debt refinancing occurs to extend the loan tenor. In that case, the loan approaches the date of maturity while the company is still in need of cash; or the company is not financially ready to repay the debt.

5 price reactions between the announcement of new loans and refinancings. Finally, note that refinancing should not be confused with debt restructuring. Debt restructuring is the replacement of debt when a company is subject to financial distress.

2.2 Uniqueness of bank loans

6 announcements of bank credit agreements, the effect should be observed around announcements of renewals, not initiations. Although loan renewals do not exactly equal loan refinancing, their features concerning the natural outgrowth of business interactions could potentially be equal. Slovin and Young (1990) illustrate the value of company-bank relationships from another perspective. Using data from initial public offerings (IPO), they explain that the length of the relationship between firm and bank decreases information asymmetry. Although IPO’s are dissimilar to bank loans, this study underlines the importance of the relationship between banks and companies. Slovin and Young show that the relationship between bank and firm has a positive effect on the pricing of IPOs. Companies that issue an IPO possess substantial asymmetric information because they are not thoroughly known to investors. This asymmetric information is likely to lead to under-pricing of the IPO. Slovin and Young argue that an existing relationship between a bank and the firm reduces the asymmetric information because of the confidential nature of the bank and its access to inside information. Their research finds that there is a negative relationship between the presence of bank-firm relationship and the degree of under-pricing of an IPO. This implies that, as a result of a bank’s confidence in a company, the relationship between bank and firm indeed send a positive signal to outside investors.

2.3 Bank loans and abnormal returns

Historically, much research has been done on the information content of different types of external financing by companies. Most of these studies conclude that equity financing, public debt or private placements show zero or negative stock returns. However, as described above, researchers find evidence for the uniqueness of bank loans as a result of the underlying confidential relationship between company and bank.

James (1987) wrote the first, and a widely cited, paper that investigates the abnormal returns of an event study around the announcement of bank credit agreements. This event study reports a significant positive two-day announcement return for new bank loan agreements and a negative response for private placements. His results provide evidence supporting the uniqueness of bank loans.

7 basis of this information and the banks’ decisions, which become publicly available, provide signals about borrowers’ creditworthiness. Lummer and McConnell report larger positive announcement returns for larger loans, presumably because investors believe that larger loans have bigger economic effect. Additionally, Lummer and McConnell distinguish between new bank loans and renewals. For new loans, the borrower’s abnormal return around announcement is not significant. The loan renewals are divided into three different groups: favourable, unfavourable and mixed renewals. The difference between the three groups of renewals is determined by the favourability of these revised dimensions in the contract (maturity, interest rate, absolute value, security and debt covenants). For favourable loan renewals, the excess return is significantly positive; for unfavourable loan renewals, it is significantly negative. The theory behind the favourable and unfavourable renewals and their effect on stock prices is clear. The mixed renewals have some favourable changed dimensions, and some unfavourable changed dimensions. For this category, high positive announcement effects are found. This deserves some theoretical explanation. Lummer and McConnell explain that, if the borrower cannot meet certain covenants of the loan agreement, those terms must be relaxed if default is to be avoided. Other terms are then made more restrictive to insure that the bank’s credit risk position does not depress. Indeed, many of the mixed loan renewals follow a violation of a debt covenant or a missed interest payment. If such financial problems are not already known to outsiders, a mixed renewal contains negative information. However, as the bank has confidence in the firm and re-enters in a new agreement, this also sends a positive signal to the market. The stock price reaction is neutral. Now, if the market already possesses the negative information on the financial problems, the announcement effect is solely positive.

Lummer and McConnell interpret these results as proof that banks enter new agreements with no information advantage compared to other investors. However, as banks continue the credit relationships with their customers, from a monitoring, control and cross-sell perspective, they are granted access to inside information which gives them a relative information advantage.

Different loan types

8 agreements and straight lines of credit. A line of credit is an less formal understanding between a bank and a firm about the maximum amount the firm may borrow in a given period. A revolving credit agreement is a formal arrangement between a bank and a firm for a renewable loan commitment. Straight line of credit, being less formal, appear not to have a stock price reaction, whereas positive abnormal returns are associated with revolving credit facilities. McDonald suggests that revolving credit agreements are viewed as positive implied audits. They send clear signals to the market and convey positive information about the firm, given the bank’s willingness to enter into a loan arrangement. In contrast, the straight line credit is more obscure and is considered a less reliable signal to the market.

Mikkelson and Partch (1986) also find that stock returns around announcements are positive for bank credit. In their paper, they analyse the stock price effects of various types of financing events. On average they find a negative, statistically significant, stock price response to the announcement of common stock and convertible debt offerings. The average price reaction to the announcement of preferred stock, straight debt, private placements of debt and term loans is small and not significant. Yet, the average price response to the announcement of revolving credit agreements is positive and significant.

Dennis and Lu (2006) find that private debt placement is more similar to public bond issues than to bank loans in terms of the price reaction at the announcement . They denote that indeed only the announcement of bank loans has a positive announcement effect.

9 relationship between the number of syndicates in the loan agreement and the abnormal return triggered by the announcement.

The effect of firm size

Slovin, Johnson and Glascock (1992) investigate whether share price responses to loan announcements differ between large and small firms. Since large firms (based on market capitalization) are well monitored and have an acquired reputation, banks have less comparative advantage in the external financing process relative to the capital markets. On the other hand, for small firms, moral hazard and adverse selection problems are more severe; they have shorter corporate histories, lesser reputation, and less public information is available for investors. Hence, small firms receive greater benefit from a bank’s screening and monitoring services. They explain that, for large companies, broader and richer public information is available and fewer screening and monitoring services are entailed than for small firms. The authors find positive share price effects only for small firms. Both renewals and initiations of loan announcements are significantly positive. For large firms, there is little evidence that bank credit decisions transfer information to capital markets. Their results are consistent with the views of Fama (1985) and Diamond (1984), that small, less prestigious firms gain greater value from the screening and monitoring entailed in bank lending relations.

Abnormal returns in non-US markets

10 Huang, Schwienbacher and Zhao (2012) argue that positive announcement effects only occur in the case that lending is efficient. Their paper investigates the announcement effects of bank loans in China. The Chinese banking sector is characterized by banks controlled by government entities and banks themselves can have considerable interest in non-financial firms causing conflicts of interest and corruption. They argue that lenders must have the incentive and ability to screen and monitor borrowers. Also, lending decisions must be based on the quality of the projects financed; not on private benefits to bank managers or political considerations. The authors find that loan announcements lead to negative abnormal returns in China. Additionally, they find that a firm’s vulnerability to expropriation creates negative abnormal returns. However, this effect only occurs when the loan is borrowed from an inefficient banks.

Third party influences

Best and Zhang (1993) re-examine the information content of the bank loans by addressing two issues. First, they recognize that parties other than banks (such as financial analysts), also perform evaluation and monitoring roles. Consequently, if these other information sources can sufficiently resolve informational asymmetries, bank loan announcements should transfer little relevant information to the market. However, if these other sources create noisy signals, or have a disadvantage in information production compared to banks, bank loan announcements should carry useful information to the market. The second issue Best and Zhang address, is whether banks spend equal efforts in evaluating all borrowers. It is feasible that banks first use indicators from other sources to screen the borrower, and then decide how to allocate investigation resources. Their results suggest that banks rely upon other indicators as initial screening devices to determine where to increase their evaluation and monitoring efforts most efficiently. If the indicators are ‘good’, banks do little further research. If the indicators are negative or unchanged, banks have incentives to expand investigation. Best and Zhang conclude that for new credit agreements there is no information advantage.

11 because there exist shared benefits from longstanding customer relationships. Second, underwriting or monitoring technologies may become more effective when they are more specialized. If the negotiation and managing of high-risk loans requires different skills than the negotiation and managing of low-risk loans, individual lenders may choose to specialize. These specialized banks are rated more reputable.

Bank loans and financial crises

Boscaljon and Ho (2003) investigate the effect of banking relationships and information content on stock prices during the Asian crisis in 1997. They also provide additional insights into the unique position of commercial banks. More specifically, this paper examines the role commercial banks play in conveying information to the market, before and after the Asian financial crisis. Their findings suggest that loan announcement effects are largest after the beginning of this crisis. Boscaljon and Ho present two explanations for these larger effects. For a start, in consistency with Billett, Flannery and Garfinkel (1995), the reformation of the banking industry should lead to loans made by higher quality lenders after the Asian crisis. A credit crunch often comes together with a flight to quality by banks and investors, as they seek less risky investments. This increases the quality of the banks and will give higher abnormal returns. Secondly, during times of greater economic uncertainty loan announcements are expected to convey more information than in normal economic times (Best and Zhang (1993), Morck and Nakamura (1999)). Their results explain that bank loan announcements convey more information in a banking system with higher quality lenders. Also, commercial banks play an increased role in providing information to the markets in times of economic uncertainty. The authors also find that contract renewals results in positive abnormal returns, but loan contract initiations do not.

12 firms that have not issued debt for two years, experience less abnormal returns than low-leveraged firms or firms that have issued equity recently.

What if loans stay unannounced or unpublished?

Fery et al. (2003) investigate differences in announcement effects between published and non-published corporate loan announcements. They do this because prior research only use one source of information for credit announcements. They find that borrowers’ share prices react positively to announcement that are published in both the financial press and by a dedicated information provider. No statistically significant reaction is observed for the non-published credit agreements. Maskara and Mullineaux (2011) criticize the uniqueness of bank loans. They argue that previous research implicitly assumes that all loans are equally likely to be announced. In practice, only one in the four loans is announced. Their argument is that announcements are selective. They examine the factors that affect the decision to announce a loan and how sample selection problems can bias the outcomes of loan announcement research. They find that the majority of the announced loans are announced by firms that receive large loans (large relative to the borrowers assets) or by firms that experience serious cash flow problems. They argue that companies only announce loans that are expected to give a positive market return, and refuse to announce when a loan announcement will have a negative effect. Further, they find that loans that stay unannounced are granted to large, solvent firms where company information is publicly available and where investors are not surprised by new loans. Investors perceive these loans as normal business deals necessary to keep the business running. Maskara and Mullineaux find that there is no abnormal return for loan agreements, and that only for the smallest announced loans is a positive abnormal return discovered. Lummer and McConnell (1989) also suggest as a limitation to their research that there may be a reporting bias because non-renewals are systematically not announced.

Differences over time

13 produce more positive returns for smaller firms and for those with poor stock price performance prior to the announcement. They explain this change is caused by the reduction of the value of monitoring provided by commercial bank loans. This is divided into two developments. In the first place, the costs and effort of acquiring financial information on borrowers has reduced strongly in the last decades with the rise of Internet and other global communication platforms. This has an impact on the importance of banks as an intercessor and monitor of companies. Secondly, increasing evidence exists of the important change in the intermediation process. Federal Reserve’s Flow of Funds accounts report that the use of market sources for financing (commercial paper and corporate bonds) are equal to the use of bank credit in 1980. Yet, by the year of 2003, the credit markets provided about ten times more financing than banks did. This decreased role of the commercial banks in the credit flow process indicates that bank loan relationships matter less. For those reasons, stock price reactions to loan announcements has disappeared. A note to this research has to be made; the research by Fields et al. is based on US data. The US market for funding is identified as a market-centred financial system, while the European market for funding is a bank-centred financial system (Allen and Gale, 1995)). Berger (1999) and Boot (2000) characterize a bank-centre financial system by three conditions. They explain the following characteristics:

“ 1) lenders gather non-public information, 2) the information gathering process occurs over an extended time period and includes numerous interactions with borrowers, and 3) the resulting information remains private.”

Considering these conditions, it might not be likely that the diminishing abnormal returns perceived in the US, are to be observed for European bank loan announcements.

14

2.4 Refinancing Risk

Liquidity risk from debt arises from the borrower’s loss of control rents in the event that lenders are unwilling to refinance. Liquidity risk is the risk that a solvent but illiquid borrower is unable to obtain refinancing (Diamond, 1991). In this case it is usually referred to as ‘refinancing risk’. A borrower that cannot refinance their existing debt and has no sufficient funds on hand to repay their lenders, may have a liquidity problem. The borrower may be considered technically insolvent; even though their assets are greater than their liabilities, they cannot raise liquid funds to pay their creditors. Insolvency may lead to bankruptcy, even when the borrower has a positive net worth (He and Xiong, 2012). A company that is unable to repay its old debt in full, gives considerable control to its lenders because the company can only pay off its debt by issuing a new loan. Lenders have the right to liquidate or take control in another way. So, in the case that a company is not liquid enough to repay its bank loan, shareholders may become concerned about the company being liquidated by its lenders. On the other hand, a company which is successful in issuing new debt to repay the bank loan, could signal a positive signal to its shareholders. In that case, shareholders are relieved that the liquidity risk is taken away. Despite the refinancing, the company may still not be in a safe position, as the moment of repayment has just been pushed forward. Bearing that in mind, the signal send to shareholders could be neutral, as shareholders are still skeptical about the company’s future. The likelihood of refinancing also depends on the future cash flows that can be assigned to the lenders. If a company has small expected future cash flows, a lender is less likely to refinance the loan (Diamond, 1991). When a company’s investment is financed by bank loans, it is possible that the duration before the investment is generating cash flows is longer than the maturity of the loan. In this particular case the loan is likely to be refinanced because borrower has the ability to assign future cash flows to the lenders.

15

2.5 Hypotheses

The main topic that is addressed in this paper is how bank loan refinancing announcements instantly influence stock prices. In the literature, both new loans and loan renewals have been thoroughly discussed. The majority of the previous research agrees that abnormal returns found for new bank loan announcement is positive. Literature states that bank loans are unique because banks have the ability to monitor companies and gain information that is not publicly available. Bank loan refinancings in particular have not been investigated extensively. Remember that in legal terms a refinancing is new loan agreement. James (1987), Mikkelson and Partch (1986) and Diamond (1984) have found positive abnormal returns for new bank loans. On the contrary, Lummer and McConnell find no significant results for new bank loans, whereas stock price reactions to loan renewals can either be positive or negative. Fama (1985) explains positive abnormal returns for renewals because banks are already familiar with their clients. The market reacts positive because banks have confidence in re-entering in a new agreement. So, new bank loans and bank renewals can have positive as well as negative effects on abnormal returns. Therefore, the following two hypotheses have been established;

H1a: Bank loan refinancing announcements produce significant positive abnormal returns. H1b: Bank loan refinancing announcements produce significant negative abnormal returns

16

H2a: The level of liquidity has a positive influence on the stock price reactions to bank loan refinancing announcements.

H2b: The level of liquidity has a negative influence on the stock price reactions to bank loan refinancing announcements.

According to Boscaljon and Ho (2004), abnormal returns in economically tight situations are higher than in economically safe situations. If a bank is willing to refinance a bank loan when credit is market-wide scarcely available, the signal sent to the investors is strong. The credit crunch (2008-2010) was an economically tight situation on a worldwide scale. It has led to a credit crunch. Banks were persistent to supply companies with much credit. According with the literature of Boscaljon and Ho, the next hypothesis is established;

17

3. Methodology

When investigating the impact of refinancing announcements on a firm’s value, an event study is conducted. The central idea of an event study is that the change in company value, assuming rationality and efficiency in the stock market, will be translated instantly in stock prices, showing abnormal returns (Serra, 2002). Hence, principal to event studies is the measurement of abnormal returns. There are several methods abnormal returns can be calculated. In this paper, the ‘market model’ method is used (MacKinlay, 1997). Subsequently, a parametric event study test, as demonstrated by Brown and Warner (1985), and a non-parametric event study test, the Wilcoxon signed-rank test, are applied on the abnormal returns.

In the rest of this section, the calculation of abnormal returns are explained and a further explication on the parametric and non-parametric test statistics are provided. Additionally, an outline of the multivariate regression analysis, used to examine the effect of several independent variables on the abnormal returns, is given.

Abnormal returns

Abnormal returns show the difference between the returns realized when an event happens and the average of an historical estimation window. According to Brown and Warner (1980, 1985), abnormal returns can be achieved using three different methods. The easiest method is the mean adjusted returns, where the stock returns are adjusted with a calculated historical average return of the identical stock. The second method that Brown and Warner explain, is the market-adjusted return. This method assumes that the market portfolio of risky assets is a linear combination of all securities. Therefore, the market adjusted abnormal return of security 𝑖 is the difference between its own return and the that on the market portfolio. The third method is the market model. This model relates the return of any given security to the return of the market portfolio. The market model represents a potential improvement over the other two methods. By removing the portion of the return that is related to variation in the market’s return, the variance of the abnormal return is reduced. This in turn, can lead to increased ability to detect event effects (MacKinlay, 1997). Accordingly, the market model abnormal returns are used in this paper.

18 window is negative. In this paper, the event window is business day [-2] through [+2], where [0] is the event day. The estimation window is the period from business day [-200] through [-3]. From this estimation period, the expected stock- and market returns are calculated.

The stock returns are computed as the daily natural logarithmic returns of a company’s stock prices. For the market return, the daily logarithmic stock market returns are calculated. The stock market index used for this research is the Morgan Stanley Capital International European index (MSCI Europe). The MSCI Europe captures large and mid-cap representations across 16 developed European markets. The MSCI is a broad based stock index, which is essential to reflect a well-diversified market portfolio. As this paper focuses on Western European firms, this index is assessed to be more appropriate than the MSCI World index, for example.

According to MacKinlay, the market model’s linear specification follows from the assumed joint normality of asset returns. Under general conditions Ordinary Least Square (OLS) is a consistent estimation procedure for the market model parameters. For any security 𝑖 the market model is

(1)

where

𝑖,𝑡 is the return on the shares for company 𝑖 at time 𝑡 𝑚,𝑡 is the return on the market portfolio 𝑚 at time 𝑡 𝑖, 𝑖 are the parameters of the model

is the statistical margin of error for which the expected value E( )=0 and the

variation VAR( =

For the 𝑖 company in the event time, the OLS estimators of the market model parameters for an estimation window of observations are:

-2

-200 0 +2

19

̂

̂

̂

∑

∑

Given the market model parameter estimates, one can measure and analyse abnormal returns. When the estimation window outcomes of formulae 2 and 3 are inserted in formula 1, the expected returns can be calculated. Subsequently, the abnormal returns can be calculated by subtracting the expected return estimated in the estimation window from the actual returns in the event window. Using the market model to measure the normal return, the sample abnormal return is:

̂

̂

̂

where

̂ is the abnormal return of company 𝑖 at time

is the expected return estimated in the estimation window for company 𝑖 at time

20 Subsequently, the average of the abnormal returns of all the companies for each day in the event window is calculated by:

∑

where AAR is the average abnormal return of event day 𝑡. According to MacKinlay (1997), the average abnormal returns need to be accumulated to draw any conclusions. Therefore, the cumulative average abnormal returns are computed by:

∑

where is the cumulative abnormal return of the event window. The CAAR is calculated for four different event windows, namely [-2,2], [-1,1], [0,2] and [0,1]. For both, the AAR and the CAAR, the student’s t-test is conducted to test for significance in the event window. The significance of the event window is calculated using the standard deviation of the average abnormal return measured over all the companies per event day.

MacKinlay explains that event studies can be biased if a company has two events and the estimation- or event window of the two events overlap. To prevent this biasness, the dataset is checked for overlap of events in the event- and estimation window by the same company. If an overlap is diagnosed, one of the two events is excluded from the dataset.

Diagnostic tests

21

Non-parametric test

The student’s t-test statistic relies on the assumption that the abnormal returns follow a normal distribution. Previous studies have shown that abnormal returns distributions are not normally distributed but show fat tails and are right skewed. When testing for positive abnormal performance, parametric tests reject too often. When testing for negative abnormal performance, they reject too seldom. When the assumption of normality of abnormal returns is violated, parametric test are not well specified. Non- parametric tests are well-specified and more powerful at detecting a false null hypothesis of abnormal returns (Serra, 2002). In this paper, the Wilcoxon signed-rank test is used as a non-parametric test. This test considers that both, the sign and the magnitude of the abnormal returns, are important. The test assumes that none of the absolute values are equal, and that each is different from zero.

Multivariate regressions

In order to use the abnormal returns in a regression analysis, a cross-section analysis is used. Therefore, the cumulative abnormal returns (CAR) of each company are calculated for the event window. For the regression analysis, the cumulative abnormal return of each company for event window [0,2] is used. This CAR is also tested for significance with the student’s t-test. The cumulative abnormal returns are now used for a cross-sectional regression analysis. The regression will be run multiple times to test for different sets of dummy variables. The regression formula is exhibited below;

𝑡

∑ _𝑖 𝑚𝑚 _𝑖

𝑖 (10)

where

22 is the liquidity of company 𝑖 measured as the cash and cash equivalents level

divided by the total assets of the firm

𝑡 is another liquidity variable measured as the long-term debt of company 𝑖 divided by its EBITDA

is the logarithmic size of the bank loan

is a variable of the size of company 𝑖 measured as the logarithm of the company’s total assets.

is the tenor of the company’s loan measured as the years until the maturity

of the loan

∑ 𝑚𝑚

is the accumulation of various dummy variables

is the error term of the regression equation

23 this ratio, and it is likely that this measure has influence on abnormal returns following a refinancing announcement.

The loan size variable describes the size of new loan. The logarithmic function of the loan size is calculated because the variable is strongly skewed to the right. The firm size variable represents the size of each firm measured in the firm’s total assets. As the total assets of the firms in the sample are skewed to the right, the logarithmic values of the total assets are calculated. The tenor variable is the time to maturity of the newly issued loan. The summation of the dummy variables expresses different sets of dummies. Different dummy effects are to be tested in this paper. Hence, the regression is performed five times. At first, a crisis dummy is employed to test whether the financial crisis effects the announcement effect of bank loan refinancings. In the second regression, a dummy for each year (2005-2012) is inserted in the formula to investigate differences between individual years. In the third regression, dummies for country differences are entered (France, Germany, Italy, Netherlands, Norway, Spain, Sweden, Switzerland and United Kingdom). Fourthly, dummies are used to show the difference between different loan types (RCF, Term Loan and a mix between RCF and Term Loan). The final regression includes dummies to investigate industry differences. The different industries are: Indusrials, Consumer Goods and Services, Real Estate, Basic Materials, Media and Telecoms, Oil/Gas and Petrochemical, Technology and Transport.

4. Data

24 All refinancing deals are issued by European companies. The timeframe of the refinancing deals is between January 1st, 2005 and March 1st, 2013. In the case that the estimation windows of two deals by one firm overlap, one of the deals is taken out of the sample. In graph 1, the deal frequency per year is presented.

Graph 1: Frequency of refinancing announcements per year.

All deals have a deal value of 1 billion dollars or above. Further, most loans in the sample are syndicated loans. Syndication means that multiple banks are lending to the company to aim at risk sharing. This is reasonable, because all loans are above one billion dollars, and most financial institutions are not willing to attain exposure larger than one billion dollars to a single company. The companies operate in 8 different industries. Financial- and utility companies are left out. These types of companies are usually highly levered, and can disturb the statistical outcomes. In table 1 below, the distribution among the different industries, as well as the distribution among countries, are presented.

The refinancings deals are spread over a period that is highly influenced by the credit crunch. Therefore, the deals are separated between ‘crisis’ and ‘no crisis’ deals . The date that is used as the launch of the credit crunch is September 15th, 2008. On this date, former American investment bank Lehman Brothers declared bankruptcy. According to Ivashina and Scharfstein (2010), the collapse of Lehman Brothers intensified the credit crunch and made it sensible in Europe. After the collapse, the unwillingness of banks to lend out money, blew over from the United States to Europe. A sudden

25 decrease in the availability of bank credit was the result. December 31st, 2010 is used as the end of the credit crunch, as the international availability of credit increased again after that date.

In addition, the deals are spread out over different loan types. The deals are subdivided between four different categories. The first is a revolving credit facility (RCF). In a RCF, the bank gives commitment to a borrower for a certain loan amount, but the borrower is not obligated to take out the full loan. The company is allowed to withdraw the amount needed at any moment in time during the period of the tenor. The second category is a term loan, which implies that the borrower takes out the entire committed loan. A term loan is generally issued with a pre-determined financing goal, whereas a RCF is often issued to supply financial flexibility for the company. The third category consists of a mix of a RCF and a term loan. In the last category, all the other loans are accumulated. Further, note that all deals that have M&A or M&A related purposes are filtered out. M&A can have an effect on stock prices, disturbing abnormal returns generated by bank loan refinancings.

Table 1: Distribution of refinancing announcements over four categories. The four categories are country, industry, loan type and crisis.

Country Industry Loan type Crisis

Austria 2 Basic Materials 22 RCF 143 Crisis 60

Belgium 4 Consumer Goods & Services 70 Term 23 No crisis 150

Denmark 4 Industrials 49 Mix 38

Finland 5 Media/Telecoms 23 Other 6

France 45 Oil/Gas/Petrochemical 14 Total 210

Germany 27 Real Estate 19 Total 210

26 For the company’s share prices and MSCI prices, Thomson Reuters Datastream is used. From this database, all the prices in the estimation- and event window are assembled to facilitate the calculations of the abnormal returns. Finally, the financial information used for the OLS regression comes from each company’s most recent annual report before the announcement day, based on its latest fiscal year . For example, if a company announces the refinancing deal at April 27th, 2010 and the company’s fiscal year ends on March 31st, the financial data is obtained from the annual report of the fiscal year ending March 31st, 2010. This information is also available on ThomsonOne. Note that all financial information is expressed in US dollars.

Table 2: Summary statistics of independent variables, or variables that contribute to the independent variables. Cash, LT Debt, Assets, EBITDA and loan size are all denoted in millions of dollars. The tenor is denoted in number of years. The cash ratio is a ratio of cash divided by total assets. The Debt-to-EBITDA is a ratio of LT Debt divided by EBITDA.

Mean Median Max Min St. Dev. Skew. Kurt. Jar.-Bera

Cash 3092.7 1549.9 22149.1 16.5 3871.7 2.2 8.4 430.6 LT Debt 7457.8 4150.4 59548.2 14.2 9289.7 3.0 14.4 1446.3 Assets 33245.2 24076.0 135001.2 1120.6 32557.9 1.4 4.1 78.0 EBITDA 4088.6 2151.6 30533.8 63.7 5087.7 2.5 10.6 733.6 Loan size 2721.6 2055.0 9520.0 1000.0 1778.9 1.9 6.5 235.9 Tenor 4.7 5.0 7.0 1.0 1.4 -0.9 4.3 40.6 Cash ratio 0.0888 0.0741 0.4271 0.0033 0.0638 7.8 7.8 305.7 Debt-to-EBITDA 2.5252 1.9657 14.6737 0.0284 2.3355 2.9 13.5 1249.6

In table 2, the summary statistics of the borrower characteristics and the independent variables are displayed. The cash, LT Debt, Assets, EBITDA and Loan size are measured in millions of dollars. The tenor is measured in years. The cash ratio and debt-to-EBITDA ratio indicate the relative liquidity of the firm. For example, the maximum amount of the cash ratio is 0.4271; this means that the cash and cash equivalent of a firm is 42.7% of its total assets. The debt-to-EBITDA ratio means, as explained in the methodology, the long-term debt of a company divided by its EBITDA. To clarify, again look at the maximum amount of 14.7. This means that the long term debt is 14.7 times as large as its operating income (EBITDA). Thus, the company would have to operate on the same rate for 14.7 years in order to naturally repay its debt.

27 Bera should not be rejected at a probability level of 0.05, because 1.7527 does not exceed the critical value of 5.99. Thisimplies that the abnormal returns in the estimation window are normally distributed. The mean and median of AARs are both zero. The maximum and minimum average abnormal return are 0.33% and -0.46%, respectively.

5. Results

In this section, the empirical results of the abnormal returns are presented and discussed. As described in the methodology, the market model, developed by Brown and Warner (1980, 1985) and MacKinlay (1997), is used to calculate the abnormal returns. This is complemented by the student’s t-test and the Wilcoxon signed-rank test. Subsequently, the results of the multivariate regression analysis are presented and discussed.

Abnormal returns

As presented in section 3, the average abnormal returns and the cumulative average abnormal returns are calculated. In table 4, the results of these calculations are displayed. In the upper part of the table, the average abnormal returns (AAR) for an event window of five days, starting two days prior to the announcement, ending two days after the announcement. In the lower part of the table, the cumulative average abnormal returns (CAAR) are shown. The CAARs are calculated for four different event windows, namely [-2,2], [-1,1], [0,1] and [0,2]. The AAR and CAAR are accompanied by the student’s t-test, the p-value of the t-test and the p-value of the Wilcoxon signed-rank test.

Table 3: Descriptive statistics average abnormal returns in estimation window.

28

Table 4: The upper half of the table displays the average abnormal returns in the event window [-2,2]. The lower half of the table displays the cumulative average abnormal return in several event windows. Also are the significance levels of the student’s t-test and the Wilcoxon signed-rank test shown. * significant at level 10% ** significant at level 5% and *** significant at level 1%.

Event

day AAR Student's t-test P-value Wilcoxon p-value

-2 -0.0020 -1.6739 0.0955* 0.1797

-1 -0.0008 -0.5866 0.5580 0.2271

0 0.0006 0.4792 0.6322 0.9102

1 0.0001 0.1171 0.9069 0.4029

2 0.0023 2.0365 0.0428** 0.0416**

CAAR Student's t-test P-value Wilcoxon p-value

[-2,2] 0.0003 0.1389 0.9173 0.8160

[-1,1] 0.0000 -0.0122 0.9902 0.9967

[0,2] 0.0032 1.4306 0.1538 0.3798

[0,1] 0.0008 0.4259 0.6706 0.8896

As the table reveals, the average abnormal return for the two days prior to the announcement of a refinancing is negative (-0.2% and -0.08% respectively). The announcement day itself and the two days following the announcement day show positive AARs (0.06%, 0.01% and 0.23%, respectively). Yet, the t-test and Wilcoxon test for day [-1], [0] and [1] are not significant. The AAR at day [-2] is significant at a level of 10%, and the AAR at day [2] is significant at a level of 5%. Nevertheless, the AARs, 0.2% and 0.23% respectively, do not have strong economic value. Therefore, no conclusions from these results can be drawn regarding the effect of refinancing announcement on shareholder’s value.

For event windows [-2,2], [0,2] and [0,1] the cumulative average abnormal returns show positive results of 0.03%, 0.32% and 0.08%, respectively. For event window [-1,1] has a CAAR of 0.00%. Nonetheless, the CAARs for these event windows are not significant, according to the student’s t-test and the Wilcoxon signed-rank t-test.

Summarizing, event days [-2] and [2] produce significant AARs. However, these results even each other out in CAAR [-2,2] and their economic value is small. All the other AARs and CAARs are not significant. As a conclusion of these results, hypotheses 1a and 1b are reject.

29 Germany, Norway, Sweden, and Switzerland, the cumulative average abnormal returns are positive for both event windows, but the values are not considerably high. For Italy, the CAARs are also positive for both event windows, with a value of 1.58% and 1.77%. Compared to the other countries, these values are relatively high. Also, if Italy’s CAARs are compared to the CAARs of the whole sample, the values are high. The CAARs for the UK are negative in both event windows. Bearing in mind that the major financial center of Europe is centered in London, this is a notable result. All the other outcomes are ambiguous, having positive and negative CAARs for event window [0,2] and [-2,2].

Table 5: Subdivision of average abnormal returns between different countries. Each day within the event window [-2,2] is displayed individually. The cumulative average abnormal returns for event windows [-2,2] and [0,2] are presented in the two left columns. Countries that have less than 8 refinancing events are accumulated in category ‘others’.

Country Obs. -2 -1 0 1 2 [-2,2] [0,2] France 45 -0.0022 -0.0007 -0.0028 0.0005 0.0027 -0.0024 0.0005 Germany 27 0.0007 0.0018 0.0026 -0.0041 0.0060 0.0069 0.0045 Italy 12 0.0003 -0.0023 0.0081 0.0013 0.0084 0.0158 0.0177 Netherlands 14 -0.0002 -0.0031 -0.0038 -0.0015 0.0057 -0.0029 0.0004 Norway 9 -0.0002 -0.0054 0.0125 0.0001 0.0024 0.0095 0.0105 Spain 8 0.0057 0.0085 -0.0058 0.0030 -0.0114 0.0001 -0.0142 Sweden 23 -0.0011 0.0000 0.0051 0.0042 0.0005 0.0087 0.0098 Switzerland 19 -0.0035 0.0114 0.0054 -0.0013 0.0005 0.0125 0.0046 UK 30 -0.0034 -0.0074 -0.0028 0.0012 0.0007 -0.0116 -0.0008 Others 23 -0.0063 -0.0009 -0.0055 -0.0036 0.0024 -0.0139 -0.0066

30 Considering that the majority of the refinancings in the sample is an RCF, it has a large impact on the overall result. Observing the abnormal returns of term loans, we find a positive 1.31% cumulative increase for [0,2]. This is relatively high compared to results from the entire sample. An explanation for that result could be, that a term loan is usually issued with a predetermined purpose.

A RCF is often issued to supply liquidity and monetary flexibility to a company. When a mix between term loans and RCF’s is issued CAARs are ambiguous; the CAAR for event window [-2,2] is negative, whereas the CAAR for [0,2] is positive.

Table 7: Subdivision of average abnormal returns between different industries. Each day within the event window [-2,2] is displayed individually. The cumulative average abnormal returns for event windows [-2,2] and [0,2] are presented in the two left columns. Industry Obs. -2 -1 0 1 2 [-2,2] [-2,2] Industrials 60 0.0026 0.0030 0.0016 0.0020 0.0024 0.0116 0.0060 Consumer G&S 76 -0.0009 -0.0025 0.0011 0.0012 0.0012 0.0001 0.0035 Real Estate 26 -0.0018 0.0027 0.0005 -0.0008 0.0087 0.0093 0.0084 Transport 6 0.0046 -0.0014 -0.0026 -0.0017 0.0015 0.0004 -0.0028 Basic Materials 23 -0.0126 0.0027 -0.0069 -0.0059 0.0002 -0.0226 -0.0127 Oil/Gas/Petrochemical 15 -0.0040 -0.0082 0.0083 0.0012 -0.0031 -0.0058 0.0065 Technology 8 -0.0014 0.0003 -0.0008 -0.0083 0.0080 -0.0022 -0.0011 Media/Telecoms 26 -0.0036 -0.0011 -0.0018 -0.0010 0.0043 -0.0033 0.0014

In table 7, the abnormal returns are subdivided between the different industries in the sample. Industrials and Real Estate firms have positive CAARs for both event windows. The most conspicuous result is the negative CAARs for the ‘Basic Materials’ industry of -2.26% and -1.27%, for event

Table 6: Subdivision of average abnormal returns between different loan types. Each day within the event window [-2,2] is displayed individually. The cumulative average abnormal returns for event windows [-2,2] and [0,2] are presented in the two left columns.

Loan type Obs. -2 -1 0 1 2 [-2,2] [0,2]

RCF 143 -0.0003 0.0007 -0.0008 -0.0011 0.0024 0.0008 0.0000

Term Loan 23 -0.0079 0.0044 0.0099 -0.0004 0.0036 0.0096 0.0131

Other 6 0.0009 0.0028 -0.0090 -0.0059 0.0014 -0.0097 -0.0135

31 windows [-2,2] and [0,2], respectively. Considering the other industries, the results only show small positive or negative returns.

In table 8, the results are split between different years of refinancing announcements. The years that mainly cover the credit crunch (2008, 2009 and 2010) have positive CAARs. The year before the credit crunch, 2007, also has positive CAARs. In this year the credit crunch had already begun in the US, so this could be an explanation for that result. The other years have negative or uncertain CAARs.

Multivariate regression

In tables 9-13, the results of the multivariate regression analysis are demonstrated. Each table shows the results of a different regression formula. As described in the methodology, in each regression formula other dummy variables have been inserted. In the left column of each table, the variable coefficients are shown, and in the right column the coefficient’s probability. The correlation matrixes in the appendix demonstrate that no multicollinearity is found . The White tests, also included in the appendix, prove that no heteroskedasticity is found in the regressions.

Table 8: Subdivision of average abnormal returns between different years. Each day within the event window [-2,2] is displayed individually. The cumulative average abnormal returns for event windows [-2,2] and [0,2] are presented in the two left columns.

32 In all tables, the cash ratio has a positive effect on the cumulative abnormal return (CAR). However, the p-values appear to be insignificant. The other variable that is used in this paper as a measure of liquidity, the debt-to-EBITDA ratio, also shows a positive effect on CAR (except for the last regression analysis). The values of this variable are lower than the results from cash ratio. Yet, the p-values also appear not significant. Consequently, hypotheses 2a and 2b are rejected.

The size of the new loan has a positive effect on the CAR. Again, this variable is found to be insignificant. The size of the firm shows a negative effect on cumulative abnormal returns. For each individual regression analysis, the results are significant at a level of 1%. This result is in accordance with the literature by Slovin, Johnson and Glascock (1992), Fama (1985) and Diamond (1984). Slovin, Johnson and Glascock found abnormal returns for small firms, whereas little abnormal returns are observed for large firms. Fama and Diamond argued that smaller firms gain greater value from screening and monitoring by banks than large firms.

The tenor of the loan has a small positive effect on the CAR (expect for the result in table 12). Yet, the p-values are not significant. All R-squared probabilities fall between 0.05 and 0.1; this means that between 5 and 10% of the variability among the variables is explained.

In table 9, the dummy variable for the financial crisis is inserted in the regression analysis. The effect of the crisis dummy is positive but has no significant p-value. This outcome results in a rejection of hypothesis 3.

33

Table 10: Results from the multivariate regression analysis with dummies for the loan type included. The regression coefficients are displayed in the middle column and its p-value in the right column. * significant at level 10% ** significant at level 5% and *** significant at level 1%.

Coefficient Probability Constant 0.0111 0.7929 Cash/Total Assets 0.0266 0.4887 Debt-to-EBITDA 0.0001 0.8649 Ln Loan Size 0.0083 0.1545 Ln Firm Size -0.0090 0.0033*** Tenor -0.0015 0.4352 RCF dummy 0.0022 0.1449

Term Loan dummy 0.0272 0.0857*

Mix dummy 0.0164 0.2805

R-squared 0.0635

In table 11, results of country dummy variables are included. All the country dummies appear insignificant, except for Italy. When a refinancing is announced in Italy, the CAR is expected to be higher by 2.39%. This results is significant at a level of 10%.

Table 12 contains dummy variables for the year a refinancing is announced in. All years appear not significant. Finally, table 13 includes dummy variables for each industry. Again, none of the industry dummies has a significant p-value.

Table 9: Results from the multivariate regression analysis with a financial crisis dummy included. The regression coefficients are displayed in the middle column and its p-value in the right column. * significant at level 10% ** significant at level 5% and *** significant at level 1%.

34

Table 11: Results from the multivariate regression analysis with country dummies included. The regression coefficients are displayed in the middle column and its p-value in the right column. * significant at level 10% ** significant at level 5% and *** significant at level 1%.

Coefficient Probability Constant 0.0088 0.8359 Cash/Total Assets 0.0131 0.7380 Debt-to-EBITDA 0.0011 0.3145 Ln Loan Size 0.0085 0.1359 Ln Firm Size -0.0081 0.0027*** Tenor -0.0011 0.5414 France dummy 0.0085 0.3405 Germany dummy 0.0112 0.2478 Italy dummy 0.0239 0.0516* Netherlands dummy 0.0090 0.2478 Norway dummy 0.0221 0.1043 Spain dummy -0.0133 0.3620 Sweden dummy 0.0147 0.1623 Switzerland dummy 0.0132 0.2215 UK dummy 0.0031 0.7436 R-squared 0.0930

Table 12: Results from the multivariate regression analysis with dummies for each year included. The regression coefficients are displayed in the middle column and its p-value in the right column. * significant at level 10% ** significant at level 5% and *** significant at level 1%.

35

Table 13: Results from the multivariate regression analysis with industry dummies included. The regression coefficients are displayed in the middle column and its p-value in the right column. * significant at level 10% ** significant at level 5% and *** significant at level 1%.

Coefficient Probability Constant 0.0020 0.9650 Cash/Total Assets 0.0150 0.7207 Debt-to-EBITDA -0.0001 0.9288 Ln Loan Size 0.0087 0.0132 Ln Firm Size -0.0076 0.0062*** Tenor -0.0005 0.7662 Industrials dummy 0.0130 0.4276

Consumer G&S dummy 0.0087 0.587

Real Estate dummy 0.0154 0.3877

Basic Material dummy -0.0023 0.8965

Oil/Gas/Petr. dummy 0.0109 0.5519

Technology dummy 0.0093 0.6422

Media& Telco. dummy 0.0083 0.6291

R-squared 0.0646

6. Conclusions

This paper examined the effects of bank loan refinancings on shareholder’s value of the borrowing company. An event study analysis has been performed to examine stock price reactions. The market model approach is used to execute the event study. In order to measure shareholder’s value creation, abnormal returns are calculated. The abnormal returns are analyzed in an event window starting two days prior to the refinancing announcement, and ending two days after the announcement. Additionally, an OLS regression analysis was executed to test what factors that influence these announcement effects. For this paper, a new dataset containing 210 European bank loan refinancings has been collected. Each bank loan refinancing has a deal value larger than one billion dollars.

36 Most previous research find positive stock price reactions to the announcement of new bank loans and renewals, as opposed to most other forms of external financing. According to the literature, these positive abnormal returns are observed because of the creation of extensive company-bank relationships. Banks invest in information-gathering technology and fundamental credit analysis that gives them a competitive advantage in evaluating risky lending opportunities. Secondly, banks gain access to private information about their customers over time as a result of intimate continuing business relationship with them. As Fama (1985) explained, due to the low priority of among other debt claims, signals from a credit renewal process, which can be compared with a refinancing process, are credible and consequently reduce monitoring costs. For that reason, bank loans are pronounced to be unique as opposed to other methods of external corporate financing.

The results from this research are not in consensus with the literature. The cumulative abnormal returns prove to be not significant. The average abnormal returns do appear significant on two individual event days. On event day [-2], the average abnormal return is -0.2%. On event day [2], the AAR is 0.23%. Although these results are found to be significant, their economic values are low. Also, the two significant outcomes balance each other out, leaving negligible cumulative average abnormal returns. Following the results based on this sample, no strong evidence is found concluding that bank loan refinancing announcements contribute to an increase or a decrease in shareholder value. An explanation for this result could be the number of syndicate lenders. As explained by Preece and Mullineaux (1996), there is a negative relationship between the number of syndicate lenders and abnormal returns triggered by the a loan announcement.

The regression analysis has some significant results. The size of the firm has a negative effect on cumulative abnormal returns. This is in accordance with the results found by Slovin, Johnson and Glascock (1992). The variables used to indicate the liquidity of a company do not show significant results.

37 a predetermined purpose. An RCF is often issued to supply liquidity and monetary flexibility to a company.

The results also show that a refinancing announcement originated by an Italian firm, produces positive abnormal returns. Further, based on the multivariate regression results refinancing risk (measured in a firm’s short- and long term liquidity), does not affect shareholder’s value. Additionally, regression analysis indicates that the credit crunch does not induce stock price reaction.

7. Limitations

This paper knows some limitations. First of all, an important limitation to this paper is the absence of specific refinancing literature. Therefore, it is hard to establish the contribution of this paper to existing literature, and this paper cannot be directly compared to outcomes of other researches. Secondly, Lummer and McConnell suggest there may be a reporting bias because refinancings that are not completed are systematically not announced. A company does not want to present negative news.

Next, to find what factors drive stock price reactions it is a logical step to look at the difference between the terms of the existing loan contract and the new loan contract. An equal method as Lummer and McConnell should then be applied. This method is conceivable, but it requires examination of every individual debt contract. This would demand a lot of time, and given the time constraint of this Master thesis it is not a realistic option. Further, as the independent variables measuring the liquidity (Debt/EBITDA and Cash/Total Assets) do not give significant results, other measures of liquidity could be used (for example, the cash conversion cycle). Additionally, the liquidity risk is derived from the total outstanding debt. It might be more effective to use the amount of outstanding debt of the terminating loan, because this is the amount that has to be repaid immediately and has a direct influence on a company’s liquidity. However, to find out the exact amounts of outstanding debt of the old loan at the announcement of the new loan would also demand a lot of time.

38

8. Future Research

Since literature on bank loan refinancings is scarce, future research can be incredibly interesting. The refinancing of bank loans occurs on a highly regular basis; more extensive research is recommended. In addition, as this paper claims that loan renewals and bank loan refinancings have comparable features, comparable research is supported. Hence, the bank loan refinancing should be investigated pursuing the method used by Lummer and McConnell (1989). This implies that all individual loan contracts should be analysed carefully, so that the terminated and new loan contracts can be compared. In this way, the refinancings can be denominated as ‘favourable’ or ‘unfavourable’ refinancings. Along these lines, refinancings can be examined more accurately because the favourability of contract changes may influence stock returns.

39

REFERENCES

Aintablian, S. & Roberts, G.S., 1999. A note on market response to corporate loan announcements in Canada. Journal of Banking and Finance. 24, 381-393.

Allen, F. & Gale, D., 1995. A welfare comparison of intermediaries and financial markets in Germany and the US. European Economic Review. 39, 179-209.

Asquith, P., Beatty, A. & Weber, J. 2005. Performance pricing in bank debt contracts. Journal of

Accounting & Economics. 40, 101-208.

Benston, G. & Smith, C., 1976. A transaction cost approach to the theory of financial intermediation.

Journal of Finance. 31, 215-231.

Berger, A., 1999. The big picture of relationship finance. Business Access to Capital and Credit. Federal Reserve System Research Conference, 390-400.

Best, B. & Zhang, H., 1993. Alternative information sources and the information content of bank loans. Journal of Finance. 48, 1507-1522.

Billett, M.T., Flannery, M.J. & Garfinkel, J.A., 1995. The effect of lender identity on a borrowing firm’s equity return. Journal of Finance. 50, 699-718.

Billett, M.T., Flannery, M.J. & Garfinkel, J.A., 2006. Are bank loans special? Evidence on the post-announcement performance of bank borrowers. Journal of Financial and Quantitative

Analysis. 41(4), 733-751.

Black, F., 1975. Bank funds management in an efficient market. Journal of Finance Economics. 2, 323- 339.

Boot, A., 2000. Relationship banking: what do we know? Journal of Financial Intermediation. 9, 7-25. Boscaljon, B. & Ho, C.C., 2004. Information content of bank loan announcements to Asian

corporations during periods of economic uncertainty. Journal of Banking & Finance. 29, 369-389.

Brown, S.J., & Warner, J. B., 1980. Measuring security price performance. Journal of Financial

Economics, 8, 205-258.

Brown, S.J., & Warner, J. B., 1985. Using daily stock returns: the case of event studies. Journal of

Financial Economics, 14, 3-31.

Campbell, T. & Kracaw, W., 1980. Information production, market signalling and the theory of intermediation. Journal of Finance. 35, 863-882.

Chava, S. & Roberts, M.R., 2008. How does financing impact investment? The role of debt covenants.

Journal of Finance. 6, 2085-2121.