Description
In finance, a loan is a debt evidenced by a note which specifies that, among other things, the principal amount, interest rate, and date of repayment. A loan entails the reallocation of the subject asset(s) for a period of time, between the lender and the borrower.
Bank loans and borrower value during the recent financial crisis Empirical evidence from France
Abstract We investigate the impact of bank loan announcements on borrower value during the recent boom and bust cycle of the 2000's using a sample of 253 large loans to French borrowers. We find no si gnificant stock market reaction to bank loan announcements during the boom period but a significa nt and negative one during the financial crisis. Hence, although we document significant cha nges in bank behavior during the crisis with conservative contractual and organizational m odifications, we cannot provide empirical support for the certification value of bank loans during a
period of increased informational asymmetries. However, bank loan announcements for larger firms receiving large loans funded by international pools of lenders contribute to borrower value even during the crisis.
Keywords: bank loans, borrower value, financial crisis, event study, syndicated lending, Europe. JEL classification: G14, G20.
A previous version of this paper circulated under the title: "Are bank loans still special (especial ly during a crisis)? Empirical evidence from a European country". I thank the participants of the AFFI 2012 Inte rnational Conference (Strasbourg), FEBS 2012 Conference (London), INFINITI 2012 Conference ( Dublin), IFABS 2012 Conference (Valencia), MFS 2012 Conference (Krakow) and 10th Corporate Finance Wo rkshop (Gent) in particular Pramuan Bunkanwanicha, Taufiq Choudhry, Issam Hallak, Iftekhar Hasan, Dorota Skala and Jonas Standaert for insightful discussions. The usual disclaimer applies.
1
Bank loans and borrower value during the recent financial crisis Empirical evidence from France
This version: September 2012
Abstract We investigate the impact of bank loan announcements on borrower value during the recent boom and bust cycle of the 2000's using a sample of 253 large loans to French borrowers. We find no si gnificant stock market reaction to bank loan announcements during the boom period but a significa nt and negative one during the financial crisis. Hence, although we document significant cha nges in bank behavior during the crisis with conservative contractual and organizational m odifications, we cannot provide empirical support for the certification value of bank loans during a
period of increased informational asymmetries. However, bank loan announcements for larger firms receiving large loans funded by international pools of lenders contribute to borrower value even during the crisis.
Keywords: bank loans, borrower value, financial crisis, event study, syndicated lending, Europe. JEL classification: G14, G20.
2
1. Introduction
The ongoing economic and financial turmoil that started in 2007 has (again) put financial institutions in the center of harsh debate and massive critics, in particular with respect to their role in fuelling and propagating the crisis as well as in provoking a credit crunch. Indeed, according to Dell'Ariccia et al. (2008) and Purnanandam (2011), banks had gradually relaxed their screening and monitoring standards before the crisis, especially in the US sub?prime mortgage market. Then, they sharply curtailed new credit and forced firms to reduce investments hence propagating the financial crisis to the real economy (Duchin et al. 2010; Ivashina and Scharfstein 2010; Santos 2011). These findings are somehow disturbing because according to the seminal contributions by Diamond (1984, 1991) and Fama (1985), financial intermediaries are considered as efficient in evaluating, screening and monitoring borrowers and play a specific role in managing the problems resulting from imperfect information on firms. As banks are believed to produce valuable private information regarding borrower's risk profile and quality, bank loan announcement should convey valuable information to the market about the borrower's financial situation. Empirical evidence tends to support the view that bank loans are thus "special" according to several authors, who find positive and significant abnormal returns for borrower's stocks around the date of a bank loan announcement (James 1987; Lummer and McConnell 1989; Preece and Mullineaux 1996; Focarelli et al. 2008). Bank loan signaling and certification role should be even more crucial during episodes of boom and bust such as the most recent one starting in the aftermath of the Internet bubble followed by the financial turmoil of 2007?2008. Indeed, de Haas and van Horen (2010) show that banks tighten screening and monitoring during a financial crisis when information asymmetries are exacerbated. Thus, the value of bank loan signaling and certification should be even more important during periods of financial turmoil, leading eventually to larger positive stock market reactions following a bank loan announcement. However, empirical
3
evidence from different episodes of crisis around the world (South?East Asia, Russia or Norway) show that the adverse shocks to banks also affect their borrowers' performance (Bae et al. 2002; Ongena et al. 2003; Chava and Purnanandam 2011). Hence, it is also possible to observe negative stock market reaction following a bank loan announcement during periods of financial crisis when lending banks experience episodes of financial distress, which can adversely impact their borrowers. Indeed, more recent empirical evidence seems to question the "specialness" of bank loans. Billett et al. (2006) find that bank loans are not "special" at all when abnormal returns are estimated over a longer period while Fields et al. (2006) suggest the diminishing market reaction to bank loan announcement is consistent with the dramatic change in the financial market. The results of event studies performed on samples from emerging markets' borrowers even show negative abnormal returns for bank loan announcements (Bailey et al. 2012 and Huang et al. 2012 for China and Godlewski et al. 2011 for Russia). < Insert Figure 1 > These issues are even more important regarding the largest market for external corporate financing in terms of bank debt: the syndicated lending market1. Its development provides a representative proxy for the boom and bust cycle (see Figure 1) with 2 trillion USD and 3000 issues in 2002, then 4.5 trillion USD and 9000 issues in 2007 and 4 trillion and 6500 issues in 2011. If we establish a parallel between loan syndication and securitization2, we can wonder if such techniques have reduced the incentives of lenders to properly perform their screening duties, as shown by Mian and Sufi (2009) and Keys et al. (2010) in the case of loan securitization. Also, due to the particular structure of syndicated loans, issues related to
1
A syndicated loan is granted by a pool of banks composed of lead (arrangers) and participant ba nks that provide funding to a borrower under a single agreement. 2 A securitization does not change the contract between the borrower and the original lender. Instead a new contract is created by the lender and a third party to sell the cash flow from the underlying loan. In a syndicated loan, all lenders are and remain part of one loan contract with the borrower.
4
informational frictions are more complicated and sever in such a setting. Private information available to some lenders may create an adverse selection problem while moral hazard problem may arise when the participant banks delegate some monitoring tasks to the lead bank. This market provides an excellent laboratory to investigate our main research question: are bank loans (still) "special", especially during a crisis? We aim here at revisiting the issue of bank loan "specialness", i.e. the certification value of bank lending, with a particular focus on the recent boom and bust cycle. To do so we perform an empirical investigation of stock market reactions to bank loan announcements during the 2000?2009 period using event study methodology. We perform empirical test of loan, bank syndicate and borrower characteristics influencing stock market reaction. We investigate if the stock market perception is different over the boom and bust period and to which loan, syndicate, and borrower characteristics this perception is the most sensitive. We also check if the recent crisis induced a shift in banks and borrowers behavior, in particular in terms of loan and syndicate characteristics during the boom and bust periods. We focus on the French syndicated lending market for several reasons. First, next to deals for US companies, syndicated loans to French companies are important sources of external financing. Indeed, such loans are constantly listed in the top global deals. For instance, among the 5 top deals ranging from 15 to 25 billion USD in 2011, French company CADES raised 16.6 billion USD through a syndicated deal. In the first quarter of 2012, among the top 10 syndicated loans, Eiffarie raised 4.6 billion USD. Second, our focus on the French syndicated lending market is motivated by its specific features, as bank syndicates lending to French companies are larger and less concentrated when compared to syndicates in the US or the UK (Godlewski et al. 2012). This particular structure may have important consequences on screening and monitoring of borrowers, thus influencing bank loan's "specialness" in France, especially during a financial crisis. Third, recent concerns regarding French banks liquidity and
5
solvency with respect to the Eurozone sovereign debt crisis appeal for a better understanding of stock market perception of bank lending decisions in this area3. The rest of the article is organized as follows. We present the relevant literature and testable hypotheses in section 2. Section 3 is devoted to the description of the data and methodology. Results are displayed and discussed in section 4. Finally, section 5 concludes the article.
2. Related literature and hypotheses
In this section we survey the relevant literature dealing with the "specialness" of bank loans and the syndicated lending market. We also discuss the impact of boom and bust cycles on bank behavior. 2.1. The "specialness" of bank loans and the syndicated lending market
There is consensus in the literature that bank loans are significantly different from other forms of corporate external finance. Indeed, financial intermediation theory argues that banks are unique institutions because they gain insider information and knowledge on firms through lending and deposit relationships (Fama 1985; Diamond 1991). Hence, the traditional informational view of bank loans argues that banks, as large creditors, can produce valuable private information about borrowing firms through initial screening and monitoring. Therefore, lending decisions reveal positive private information about the firms because banks would lend to high?quality borrowers, rather than to those of low?quality, to maximize the value of the loans. A large body of empirical research shows that announcements of bank loan agreements are associated with positive abnormal returns for borrowers on average. In other words, stock markets treat bank loan financing as good news and bank loan announcements therefore convey positive information regarding borrower's conditions. Indeed, bank loans, or debt more generally, can create value by reducing overinvestment by non?congruent
3
"What's the Matter With the French Banks?", The Wall Street Journal, 13/9/2011; "Moody's Downgrade: SocGen, Credit Agricole's Liquidity Problems Larger Than Greece", Forbes, 14/9/2011
6
managers (Jensen and Meckling 1976) or by giving a manager the opportunity to signal the quality of the firm and his willingness to be monitored by lenders (Diamond 1991)4. Thus, bank loans are considered as "special", starting with the seminal work of James (1987) who finds a sizeable average excess return following announcements that firms have signed a bank loan agreement. Many further studies confirm and refine this result. Lummer and McConnell (1989) report significant average excess returns for favorable loan revision announcements while Slovin et al. (1992) show that bank loan announcements are particularly good news for firms with severe information asymmetry, such as small firms. According to Best and Zhang (1993), firms that face greater earnings uncertainty and lack sufficient evaluation and monitoring by other stakeholders benefit most from bank loan announcements. Higher positive excess returns following loan announcements are also associated with more reputable lenders (Billett et al. 1995). The global syndicated lending market represents a significant portion of external financing for companies, as almost 4 trillion USD of debt had been raised on this market in 2011 (Thomson Reuters 2011). The benefits of loan syndication both for lenders (portfolio risk and sources of revenues diversification) and borrowers (mostly lower costs as compared to bond issues or a series of bilateral loans) largely explain the success of syndicated lending. A syndicated loan embeds both features of bank lending: transactional and relationship (Altunbas et al. 2006). It is therefore also "special" as any bank loan and most of empirical research tends to show that it is true. Indeed, loans generate positive abnormal returns and consequently are special when they are made by syndicates with fewer lenders (Preece and Mullineaux 1996) or with larger portions of the loan retained by arrangers (Focarelli et al. 2008).
4
In contrast, announcements of SEO (seasoned equity offerings) generate an average negativ e abnormal return, whereas announcements of public bond issues generate zero or slightly negativ e equity returns, according to previous research.
7
Overall, an empirical consensus seems to emerge from previous research regarding bank loans' "specialness" as certification and signaling device regarding borrowers' quality. Hence we can expect to observe a positive reaction of investors to a bank loan announcement, materialized by a significant and positive abnormal return for the borrowing firm's stock around the announcement date. However, there also exists empirical evidence showing that bank loan announcements can be considered as bad news with negative abnormal returns (Billett et al. 2006). Such findings are particularly frequent in the case of emerging market economies (Bailey et al. 2012; Godlewski et al. 2011; Huang et al. 2012). These recent findings may question the empirical consensus in favor of bank "specialness". Furthermore, the specific features of syndicated lending may have potential adverse effects on the stock market reaction to bank loan announcements. Indeed, syndicated loans have their drawbacks because the nature of a syndicated loan may expose the banking pool's members to the adverse consequences of informational frictions and potential agency costs. First, private information about the borrower can create adverse selection problems, as the arranger may be inclined to syndicate loans for unreliable borrowers. Second, participating banks may delegate monitoring to the arranger, but the banks are not in the loop as to what the arranger is doing, which might result in situations of moral hazard. Thus, for all these reasons we might also observe an insignificant or even negative abnormal return for the borrower around the bank loan announcement date. 2.2. Bank lending during boom and bust periods
Much of the research on bank lending behavior, qualified as procyclicality in a boom and bust framework, has focused on credit crunches during business cycle downturns. Several hypotheses for these crunches were tested and partially validate. Hence, it appears that credit crunches can be explained by reduced risk taking by banks (Wagster 1999; Furfine 2001), implementation of tougher regulatory capital standards (Berger and Udell 1994; Hancock et al.
8
1995) or increasing supervisory toughness (Peek and Rosengren 1995; Gambacorta and Mistrulli 2004), as well as reduced loan demand (Bernanke et al. 1991). More recently, Dell'Ariccia and Marquez (2006) argue that banks may loosen their lending standards and thus lead to deteriorated loan portfolios, lower profits, and expanded aggregate credit because information asymmetries decrease during economic growth periods. With respect to the most recent episode of boom and bust, Demyanyk and Van Hemert (2011) find that the quality of loans deteriorated for six consecutive years before the crisis and that securitizers were aware of it. De Haas and van Horen (2010) provide additional evidence regarding bank lending behavior during the global financial crisis by analyzing changes in the structure of syndicated loans. They find an increase in retention rates among syndicate arrangers during the crisis, especially in case of important information asymmetries between the borrower and the syndicate or within the syndicate. They interpret their findings as a "wake?up call" with increased screening and monitoring by banks during the bust period starting in 2007. Following these results, we can expect that such reaction in bank lending behavior should translate in a greater certification and signaling role of bank loans and hence their "specialness" during a crisis. We could observe a positive stock market reaction to bank loan announcements during the bust cycle if investors believe in a "wake?up call" of banks. Indeed, with increased information asymmetries during a crisis period, banks should react mainly through two channels: contractual and organizational. Regarding the former, banks should become more conservative in order to better mitigate adverse selection and moral hazard problems between the borrower and the lenders by adjusting main loan terms (more collateral, more covenants, and longer maturities) in order to screen and monitor borrowers more tightly. Regarding the latter, banks can also adapt the structure of the syndicating pool with fewer lenders and with a greater percentage of local banks in order to reduce informational frictions within the syndicate and enhance borrower's monitoring. These
9
contractual and organizational adjustments should translate into a greater certification value of bank loans during a crisis and hence significant and positive abnormal returns for borrower's stock around a bank loan announcement date. On the contrary, during the boom or pre?crisis period, we shall observe insignificant abnormal returns as less information asymmetries lead to relaxed lending standards which in turn diminish the certification value of a bank loan. However, we may also obtain an opposite result with stock markets sanctioning bank loan announcements perceived as signals of borrower weakness during economic and financial turmoil. Indeed, and in particular on the syndicated lending market, troubled borrowers could be the first to ask for bank debt funding, especially in the form of credit lines. Indeed, Ivashina and Scharfstein (2010) show that after Lehman Brothers collapse, borrowers, especially those financially constrained, heavily drawn down their lines of credit, inducing banks to limit new loans. Hence, it could also be the case that investors would sanction external financing from distress lenders, especially during a severe financial crisis. For all these reasons, we could also find significant and negative abnormal returns for borrower's stock because of investors' concerns regarding firm's and/or lender's conditions.
3. Data and methodology
In this section, we provide a description of the data and relevant descriptive statistics, followed by an explanation of the methodology. 3.1. Data
Data on equity prices, loan and syndicate characteristics and borrower balance sheet for French companies over the 2000?2009 timespan are extracted via the Bloomberg Professional Terminal Server. Bloomberg provides detailed information on the terms of loan agreements, the composition and structure of the lending syndicate and accounting data for the borrowing companies. The main filter we apply concerns stock price availability over at least 150 trading days before the date of bank loan announcement. Additional filters concern syndicate characteristics and balance sheet data availability. The final full sample contains 253 bank loan 10
announcements, each for a unique company, a figure which is within the range of events in previous studies (from 117 to 728 events) as reported by Maskara and Mullineaux (2011). The number of bank loan announcements increase over time, with 6 events in 2000, a peak of 50 events in 2007 and 23 events in 2009. Table 1 provides summary definitions and descriptive statistics for main loan, syndicate and balance sheet variables for the full sample. Overall, syndicated loans for French borrowers are large (almost 700 MLN USD) with a maturity of almost three years and a spread close to 130 bps over Libor or Euribor. A typical loan facility is composed of more than two tranches5. Half of the loans are term loans and 40% are revolving loans. One out of five loans is secured and has covenants. Bank syndicates are composed of almost nine lenders of which an important part bear arranger titles (such as lead arranger, mandated arranger or arranger). More than 2/3 of lenders are French banks and we observe a similar figure for the arranging banks. We remark that figures for league table lenders6 are very similar to those for French banks (actually, French banks in the sample are often listed on Bloomberg League tables). The sample contains large firms with respect to their balance sheet or sales (40 and 8 BLN USD respectively). Common equity and debt ratios represent each roughly 1/3 of total assets while Ebitda amounts for more than 10% of interest expenses. Firms are relatively liquid according to their current ratios, with a good profitability with respect to profitability margins
5
Syndicated loans can be "tranched" into heterogeneous components that can then be distributed across lenders differentiated by their risk aversion. This technique is somehow close to tranching in a securitization process. 6 We consider a lender to be part of the league table if it is listed as one of the first 25 fin ancial institutions in the Bloomberg Underwriter Rankings Table, computed according to lender's marke t share, amount issued and number of issues between 2000 and 2009 for the European Market. We choose the 25th rank as a cutoff because below this rank the market share of a lender is lower than 1%.
11
as well as return on assets. Overall, these figures suggest a good level of the firms' quality in our sample7. 3.2. Methodology
We consider a bank loan announcement as an event and identify bank loan announcement dates in Bloomberg and consider this date as day 08. We adopt a classic approach based on the market model which relates the return of a given stock to the return of the market index: (1) where is the return on the share price of borrower on day , the stock market return is 1
on day and and are the parameters to be estimated over an estimation window. an error term with 100, where 0 and . Returns are defined as /
is the daily closing stock market price at time and we proxy for the market
return using the SBF 250 stock index return9. We consider an estimation window of 100 to 10 days prior to the event date10 and we use OLS regressions to estimate the market model. Daily abnormal returns are obtained following: | where is the actual return on the share price of borrower on day while | (2) is the
normal return where
is the conditioning information for the market model, i.e. the market
return. Following previous studies (see Maskara and Mullineaux 2011 for a summary), we examine seven different event windows: three symmetric ones (one?day [0,0], three?days [?1,+1], five?days [?2,+2]) and four asymmetric ones (two?days [?1,0]; [0,1] and three?days [? 2,0]; [0,2]). The latter, especially [?1,0] and [?2,0], serve also the purpose of verifying the
7 8
The most important industrial sectors in the sample are Consumer (37,93%) and Industrial (20,69%). It is necessary to make sure that there is no other corporate news that could influence stock returns within an event window. We check carefully and find no contamination caused by other events around our event dates. 9 Our results do not change when using CAC 40 or SBF 120 stock index but provide lower statisti cal quality of the regressions (R² lower than 10%). 10 Using longer estimation windows ranging from 150 days to 30 days prior to the event date gi ves similar results.
12
existence of potential information leakage. For each event window, we compute the cumulative abnormal returns (CAR): , where and ? (3)
are respectively the lower and upper bounds of an event window.
Standardized cumulative abnormal returns (SCAR) are obtained by dividing CAR by
?
, where is the number of days within a given event window and
is
the variance of the abnormal return estimated from equation (1). Cumulative average abnormal returns (CAAR) are defined as: , ? , (4)
where N is the number of borrowers in the sample on day . Cumulative average standardized abnormal ? returns
?
(CASAR) .
are
obtained
by
dividing
CAAR
by
We proceed in two steps. First, we perform a univariate analysis using t?tests to investigate the statistical significance of CAAR and CASAR with the null hypothesis being that the CAAR or CASAR equals 0. We also perform similar tests (t?test or chi²?test depending on the nature of the variable under consideration) to investigate the statistical significance of differences in various loan, syndicate and borrower variables with respect to positive and negative CAAR. Then, we repeat the tests with respect to two different periods of our sample: before and after the crisis. We define the period between January 2000 and August 2007 as the Pre?crisis period while the period from September 2007 to December 2009 is considered as the Crisis period. Again we test the statistical significance of CAAR and CASAR as well as of various variables with respect to positive and negative CAAR for the Crisis and Pre?crisis sub? periods. Second, we perform a multivariate analysis by regressing the borrowers' CAR on a Crisis dummy (equal to 1 for bank loan announcements during the Crisis period, i.e. between
13
September 2007 and December 2009, 0 otherwise) and interaction terms between the latter and most important variables related to loan, syndicate and borrower characteristics. By doing so we aim at testing the robustness of the univariate results in a multivariate setting, in particular regarding the effect of the crisis on stock market reaction to bank loan announcements. We also want to investigate more in details the influence of the interaction between the crisis effect and other loan, syndicate and borrower characteristics on the abnormal returns. We focus on the main loan terms such as size, maturity, type, the presence of collateral or covenants. We also consider the size of the syndicate as well as main balance sheet characteristics of the borrowers such as size, solvency and profitability. The equation of interest to be estimated by OLS regression with robust standard errors can be formally defined as: (5) is the cumulative abnormal return for borrower and is defined in equation (3). and are the coefficients for the Crisis dummy and the interaction variables between the Crisis dummy and several Variables of interest related to loan, syndicate and borrower characteristics. We test several specifications depending on the vector of control variables (Controls), which can be loan and syndicate related only, or also including borrower characteristics. The latter implies a reduction in the number of usable observations due to balance sheet data availability.
4. Results
In this section, we first discuss the univariate results regarding stock market reaction to bank loan announcements for the full sample period and for the boom and bust periods. We also investigate loan, syndicate and borrower characteristics related to positive and negative stock market reaction. Then we provide the multivariate results relating the CARs to loan, syndicate
14
and borrower variables. Finally, we investigate the financial constraints issues of borrowers and their impact on stock market reaction. 4.1. Univariate results
We first present our main univariate results regarding stock market reaction to bank loan announcements over the full time span of the sample as well as by boom and bust periods in Table 2. We first discuss results provided in columns 2 to 3 in table 2. We remark that the CAAR are positive for 40% to 50% of bank loan announcements. Nevertheless, we observe systematically negative stock market reaction but only significant for three event windows: [? 2,0], [?1,0] and [0,0], with approximately ?0.30 for CAAR and ranging from ?0.07 to ?0.09 for CASAR11. We conclude that bank debt financing through a syndicated loan by French companies is considered as a negative signal by the stock market. Furthermore, we can also claim that some form of information leakage seems to be at work as significant reaction is observed for windows before the loan announcement event. This first series of results do not confirm previous findings that bank loans are special (James 1987; Lummer and McConnell 1989; Preece and Mullineaux 1996; Focarelli et al. 2008). We rather provide empirical support for conclusions reached by Billett et al. (2006), Fields et al. (2006), Bailey et al. (2012), Godlewski et al. (2011) and Huang et al. (2012). In the French case, bank loan announcements are considered as bad news by the stock market refuting bank's specialness arguments as well as certification and signaling role of bank debt financing. We now turn to the results provided in columns 4 to 9 in table 2. First of all we remark that most of stock market reactions are negative, confirming previous results. Thus bank loan announcements are considered as a negative signal by investors. However, these reactions appear to be significant only during the crisis period as CAARs and CASARs are statistically
11
We reach similar conclusions when using alternative t statistics such as Patell (1976) or Boehmer et al. (1991).
15
different from 0 mainly for the [?2,0] and [?1,0] event windows, while there are no significant market reactions during the pre?crisis period. Furthermore, for these particular event windows, both CAAR and CASAR are statistically different regarding the sub?periods under investigation according to the results of t?tests in column 9. Finally, in absolute value, CAAR and CASAR are overall larger during the crisis. For instance, the CAAR for the [?1,0] event window is more than 20 times larger during crisis than before. It is also twice the CAAR for the full time period under investigation (2000?20009). Overall, we can claim that a bank loan announcement is perceived differently with respect to the economic environment (Crisis vs. Pre?crisis) and that it is considered as a negative signal by market participants during the crisis period, while it is not considered as a signal at all before the turmoil. Hence, bank loan announcements appear to be considered as bad news during a period of economic and financial turmoil, while they are perceived as insignificant during a boom period. Although contrary to some of previous empirical findings, this result receives support from recent research on the 2007?2008 crisis. Dell'Ariccia and Marquez (2006) and Demyanyk and Van Hemert (2011) have shown that banks have relaxed their lending standards during the boom period leading to a deterioration of their loan portfolio's quality and of the certification value of bank loan announcements. Our results for the pre?crisis period support the hypothesis that relaxed lending standards during a period of reduced informational asymmetries diminish the certification value of bank lending. Even if De Haas and van Horen (2010) provide evidence on a "wake?up call" with increased screening and monitoring by banks during the bust period starting in 2007, our results tend to show that providing a loan to a borrower during the crisis is perceived negatively by the stock market. This can be related to several issues. First, even with more conservative lending standards, investors can still doubt in the capacity of banks to identify valuable borrowers on the credit market, especially if banks are perceived as more vulnerable due to an adverse economic and financial environment. Second, we can also expect that on
16
average, lower quality borrowers need to apply for bank loans during a crisis, especially through credit lines (Ivashina and Scharfstein 2010). For illustrative purposes, we also provide in figure 2 the evolution of CARs over a [? 10,10] time window for the whole sample (red line), the Crisis (blue line) and the Pre?crisis (green line) periods. We remark that the full sample CAR experience a sharp decline one day before the event date and drops to around ?0.3. Then it remains at this level for around two days. We observe similar patterns for both Crisis and Pre?crisis CARs but the drop is larger for the Crisis CAR which almost reaches ?0.6. However, further investigation is needed to better understand these results and verify which features of the loan contract, the syndicate and the borrower play a significant role in shaping stock market reaction. In what follows we focus on the most significant CAAR using the [?1,0] window12. We aim now at investigating those characteristics that are associated with a positive stock market reaction. To do so we perform t?tests or chi²?tests (depending on the nature of the variable under consideration) on the difference of various variables with respect to a dummy equal to 1 if the CAAR [?1,0] is positive (122 events), and equal to 0 if the CAAR [? 1,0] is negative (131 events). The results are displayed in Table 3. Regarding loan characteristics, we observe that the only significant feature is the facility amount. The stock market reaction is positive for larger loans (twice as large as loans with negative CAAR). This result can be linked to our findings regarding bank syndicate characteristics, as we remark that larger syndicates with fewer local lenders are associated with positive CAAR. Regarding firm characteristics, we remark that significant differences in stock market reaction are essentially related to firm size measured with total assets and sales.
12
All results are similar when using other less significant windows as well as CASAR.
17
According to these results, the French stock market considers that large loans, funded by large syndicates of which a smaller proportion is composed of local banks, are a positive signal. Indeed, a larger loan funded by a more diffuse syndicate can be considered as a good signal regarding borrower's quality. The size of the loan can be interpreted as reinforcing the certification and signaling role of the bank lending decision (Mosebach 1999) while a larger syndicate is usually associated with less informational frictions and their subsequent consequences in terms of adverse selection and moral hazard in the relationship between the borrower and the lenders (Lee and Mullineaux 2004; Sufi 2007; Bosch and Steffen 2011). This can also be related to our findings regarding borrower characteristics. Indeed, the market values positively loan announcement by large firms with important sales, thus more visible and less opaque companies with sustained economic activity. Meanwhile, the presence of numerous lenders can also serve as a device to mitigate eventual liquidity risk in funding the loan to the borrower as well as a risk diversification device, in particular when funding a large loan (Gatev and Strahan 2009). However, the result regarding syndicate size does not confirm previous results by Preece and Mullineaux (1996) who show, using a sample of bank loans provided to US borrowers, a positive reaction to loans funded by smaller syndicates. A positive market response to bank loan announcement involving less local lenders is more puzzling. Indeed, one could expect the opposite as local lenders presence help to mitigate the adverse consequences of informational asymmetries both between the borrower and the syndicate as well as within the syndicate (Berger et al. 2001). However, this effect is not systematically true as shown recently by Fungá?ová et al. (2011). Hence, we can argue here that the larger presence of foreign lenders can be considered by the stock market as a better and/or more objective signal regarding deal and borrower quality. This argument is even more appealing with respect to the recent fragility of French banks following the 2007?2008 crisis.
18
Next, we investigate more in details stock market reaction to bank loan announcements during the recent boom and bust cycle. The results are displayed in Table 4. First of all, we remark that there are significant differences regarding loan maturity and contractual features such as loan collateralization or covenants. Indeed, maturity is more than three times larger during crisis, and one out of three loan contracts are secured and have covenants, while these features are only present for less than 20% of loans before the crisis. These loan characteristics tend to show a change in bank behavior during the crisis due to increase borrower default risk, uncertainty and informational frictions. In particular, loan characteristics aiming at reducing adverse selection (security) and moral hazard (covenants) problems are reinforced during the crisis period. Larger maturities imply also that banks provide longer term funding to dilute the cost of bank debt for borrowers even at the expense of larger spreads. These results are in line with the "wake?up call" argument provided by de Haas and van Horen (2010). Meanwhile, we also remark that the only significant feature of the bank syndicate that changes significantly is the number of lenders, which is reduced by three banks during the crisis. This again confirms a change in bank behavior and is consistent with changes in loan characteristics as a smaller syndicate is better suited to cope with borrower monitoring and mitigating agency costs within the syndicate (Lee and Mullineaux 2004; Sufi 2007; Bosch and Steffen 2011). It can also be explained by the difficulties of financial institutions during that period and thus their weaker willingness to fund syndicated loans. Finally, the only borrower characteristic exhibiting a significant (although statistically weak) change during the crisis is profitability which is twice larger than before. Overall we find that bank behavior has changed during the crisis and has become more conservative as a reaction to increased informational asymmetries. Finally, we investigate differences in loan, syndicate and borrower characteristics for positive and negative stock market reactions during and before the crisis (Table 5). Regarding
19
loan characteristics, apart from loan size which exhibits similar features for positive CAAR as for the full sample (larger loans are associated with positive stock reaction), we remark that during the crisis, loans with larger spreads (70 bps larger on average) and more tranches (1 tranche more on average) were associated with a positive stock market reaction. It is also worth noticing that a positive reaction is associated with an average loan size of 1 billion USD during the crisis while the same is true for a 700 million USD loan before the crisis. The evidence is completely inverted for spread: before the crisis, positive reactions are related to lower spreads while they are associated with larger spreads during the crisis. The spread result can be analyzed within the Spence costly signal framework. In an environment plagued with greater uncertainty and thus informational asymmetry, the capacity to pay a higher spread can be interpreted as a signal regarding the expected performance of the borrower. But we can also consider that the stock market perceives higher spreads as a signal of reinforced lending standards of the banks, i.e. more risk adjusted loan pricing, and greater certification value. This can be related to the result regarding the tranching of syndicated loans. Following recent evidence by Maskara (2010), multiple tranches actually create economic value and provide benefits for riskier borrower even if on average, the credit spread for a multi?tranches loan is larger. This is because without tranching, such spread would be even larger, eventually leading to adverse selection effects. We also observe differences regarding bank syndicate features, as the size of the syndicate and the number of arrangers are significantly different for positive and negative stock market reaction but only before the crisis. Larger syndicates with more arrangers are associated with positive CAAR according to the argument relating such syndicate structure with less informationally problematic deals and borrowers. Other syndicate features such as the percentage of local lenders or arrangers exhibit similar level of significance as for the full period (cf. Table 3).
20
Finally, we also remark differences regarding borrower characteristics such as such as size (measured by total assets or sales) which exhibit similar significant levels by CAAR during crisis or no?crisis periods as for the full period (cf. Table 3). In other words, size matters as larger borrowers experience positive abnormal returns around the bank loan announcement date, in particular during the financial crisis. 4.2. Multivariate results
We present now the results of a multivariate analysis of the relationship between borrower's market value, measured by the most significant CAR [?1,0], and various loan, syndicate and borrower characteristics, with a particular focus on the effect of the recent financial crisis. The latter is captured with a dummy variable equal to one if the bank loan announcement occurs between September 2007 and December 2009. We also interact this dummy with other variables, for which we restrained our analysis to the most important and significant characteristics (regarding the univariate analysis results). These are the loan facility amount, maturity or the size of the lending syndicate, as well as loan terms variables such as the type of the loan, the number of tranches as well as the presence of collateral and covenants13. Due to limited data availability, we restrict the borrower characteristics to three main variables: sales, equity ratio and operating margin. Results are displayed in table 6. First we discuss the results obtained with loan and syndicate variables only, i.e. regressions (1) - (8). We remark a significant and negative coefficient for the Crisis dummy, confirming univariate results. Issuing a loan during the recent bust is not considered by investors as a positive signal regarding the borrower's profile. In other words, the recent financial crisis reduces (even destroys) borrower's market value steaming from a bank loan announcements.
13
We are unable to provide viable regressions results when including the Spread variable as it oft en unavailable in our sample.
21
Next we observe that among the seven variables capturing different characteristics of the loan or the syndicate, only two bear significant coefficients when interacted with the Crisis dummy. These are the Facility and Lenders (in log) variables. The result for the loan amount confirms our univariate conclusions as larger loans are associated with positive CAAR even during the crisis. Indeed, the combined coefficient for log(Facility) equals (?0.033+0.315) which is larger than the Crisis dummy coefficient. Hence even during a bust period where bank loan announcements are perceived as negative news and can destroy firm's market value, announcements of large loans can counteract this adverse effect and end up in a positive effect on the market value of the borrower. An additional confirmation for that result is the fact that the individual coefficient for log(Facility) is always significant and positive across all regressions (1) to (8). The result for the size of the syndicate is statistically weaker, as the coefficient for the interaction term is significant at the 10% confidence level only and the log(Lenders) coefficient itself is not significant. In other words, the crisis effect dominates the syndicate size effects with respect to abnormal returns. This conclusion somehow confirms the univariate result where we found that this syndicate characteristic doesn't matter for the stock market reaction during crisis. When turning to the specifications (9) to (11) with borrower's balance sheet variables, we confirm again the negative relationship between the crisis and stock market reaction to bank loan announcements. We also remark that two of the three main borrower characteristics, size measured by log(Sales) and solvency measured by the common equity to total assets ratio, bear significant and negative coefficients. The borrower size effect confirms univariate findings as the combined coefficient for the Crisis dummy remains positive (? 0.1114+0.3266) although the interaction term is significant and negative. Overall, size matters for a bank loan announcement to be perceived as a positive signal for investors during the recent financial crisis, as larger borrowers receiving large loans experience positive abnormal returns around the bank loan announcement date.
22
The result for equity ratio can appear as counterintuitive because less capitalized firms can be considered as more fragile, especially during a crisis. However, we can also remind that the corollary of equity is debt which has been found to work as a signaling and disciplining device (Leland and Pyle 1977; Ross 1977), helping to solve adverse selection that results from information asymmetries between firm insiders and outsiders. Indeed, debt can reduce agency costs resulting from conflicts of interest between shareholders and managers as it increases the pressure on managers to perform and stop wasting company resources and increase their effort by restricting the 'free cash?flow' at the disposal of managers (Jensen 1986). Moreover, a high?quality firm can issue more debt than a low?quality firm, because the issuance of debt leads to a higher probability of default due to debt?servicing costs. Thus, receiving a bank loan during a financial crisis can be viewed by investors as a strong signal regarding borrower's quality certified by the lenders. Finally, we can also argue that equity investors value the monitoring role of bank debt holders in a period of adverse economic conditions and increased uncertainties regarding borrower's prospects and behavior. 4.3. Financial constraints In a nutshell, our findings do not support the certification value of bank loan announcements during the crisis although we uncover a "wake?up call" effect of banks' behavior, with more conservative loan terms as well as syndicate structures. Although some of our results are consistent with previous findings regarding loan and borrower characteristics related to a positive abnormal return, such as loan or firm size, we cannot fully explain the drivers behind a negative stock market reaction during the crisis relying solely on the certification and signaling arguments. Indeed, lenders' reputation do not seem to matter for stock market perception as we do not find any statistically significant difference between positive and negative abnormal returns across the boom and bust periods for syndicates with large or small portions of league table members. Thus we turn to alternative explanations related to financial constraints of borrowing firms following notably Ivashina and Scharfstein
23
(2010), by examining several characteristics of borrowers related to their retained earnings, free cash flow, (short and long term) borrowings and (total and available) lines of credit14. We replicate the previous analysis on a battery of six variables related to borrower's financial constraints in Table 7. < Insert Table 7> We first remark that retained earnings and free cash flow (FCF) are systematically larger for positive stock market except for FCF during the boom period. The difference is particularly dramatic for retained earnings which become heavily negative for negative abnormal returns during the crisis. We observe a similar pattern for short and long term borrowings, as investors positively react to bank loan announcements to firms with large borrowings, especially during the crisis. Hence, it seems that investors value bank loan announcements to less financially constrained borrowers. However, none of these variables exhibit a statistically significant difference in means according to the t?tests. Finally, we also uncover that the stock market perception of bank loans is also sensitive to borrower's line of credits as this variable is again much larger as compared to firms with negative abnormal returns. Furthermore, this characteristic appears to be statistically significant in terms of difference between the positive and negative stock market reaction. Thus we have some empirical support for the argument that the negative stock market reaction to bank loan announcements is related to borrower's financial constraints. This effect is persistent whenever the period under investigation (i.e. crisis vs. pre?crisis.). Hence, investors seem to be sensitive to firms' financial conditions when they apply for a bank loan, and when these conditions are weak, the stock market perception of bank loan announcements is significantly negative. Moreover, following our results, this effect might be stronger than the "classical" certification value effect which should be even stronger during a crisis when lenders adjust
14
Unfortunately, investigating such detailed variables comes at the cost of losing a substantial portion of the sample due to data unavailability. The number of available observations for the six variables under con sideration in the full sample is 200, 188, 196, 200, 61, and 58 respectively. This is why we do no t include them in our main analysis in previous sections.
24
loan terms and syndicate structures to greater informational frictions in the economy following a "wake?up call". Overall, it seems that investors are more sensitive to borrowers' financial conditions during a bust period even if lenders' behaviors change to reflect a tougher economic environment.
5. Conclusion
We have empirically revisited the issue of bank loans "specialness" with a particular focus on the recent boom and bust cycle to provide a better understanding of stock market perception of bank loan announcements in the case of a major European country. Using a sample of 253 loan announcements to French borrowers from January 2000 until December 2009 we have computed CAAR and CASAR for the whole period as well as for the boom and bust sub?periods. We have then investigated various loan, syndicate and borrower characteristics that could influence stock market reaction. Regarding the full sample results over the 2000?2009 timespan we found significant and negative stock market reaction to bank loan announcements. This first finding does not support the consensus of (positive) bank loan specialness first provided by James (1987) but rather more recent conclusions by Billett et al. (2006). In our case, bank loan announcements are actually perceived as bad news. However, we also document which loan, syndicate and borrower characteristics are associated with a positive reaction. We find that larger loans funded by numerous lenders of which a smaller proportion is local banks to large borrowers are related to a positive abnormal return. This series of results is more in line with previous literature (Mosebach 1999; Lee and Mullineaux 2004; Sufi 2007; Bosch and Steffen 2011). We then investigate the effect of the recent boom and bust cycle on stock market perception of bank loan announcements using both univariate and multivariate analysis. First of all we find that the average negative stock market reaction to bank loan announcements is essentially due to the loans provided during the financial crisis from 2007 to 2009, while the reaction is not significant for loans before the crisis. The latter result confirms that certification 25
value of bank loan announcements is reduced when lending standards are relaxed due to less informational asymmetries during a boom period. On the contrary, the former result doesn't confirm that the certification value of bank lending increases during bust periods when bank behavior is more conservative as a reaction to greater informational asymmetries. This last result is rather surprising as we uncover a significant change in bank lending behavior over the cycle, following notably recent evidence by de Haas and van Horen (2010). During the crisis period, loans have larger maturities, are more often secured and have covenants, and are funded by much smaller syndicates. These results clearly indicate a "wake? up call" effect of the crisis on bank screening and monitoring activities, with the reinforcement of contractual (loan) and organizational (syndicate) features aiming at mitigating adverse selection and moral hazard problems during a period of greater uncertainty and risk. Second, we look into the characteristics of loans, syndicates and borrowers that are related to positive and negative stock market reaction over the boom and bust cycle. Although empirical evidence tends to support a reinforcement of contractual and organizational means of screening and monitoring by banks during the crisis, we do not find support for these characteristics to be related to positive abnormal returns. However, we find that larger loan spreads and multi?tranches deals are associated with a positive market reaction during the crisis. Moreover, larger loans to larger borrowers funded by syndicates with fewer local lenders are also positively perceived by investors. We also uncover that a positive reaction during the crisis is associated with a borrower's lower common equity ratio. Finally, we also uncover investors' sensitivity to borrower's financial conditions, in particular their financial constraints during a bust period. Indeed, we find that the stock market perception is systematically positive for bank loan announcements to firms which are less financially constrained, especially with respect to the availability of lines of credit. This result suggests that even if lenders adjust loan terms and syndicate structure to reflect tougher
26
economic environment with greater information asymmetries, investors put more value on the certification effect of loans to financially unconstrained borrowers. Overall, our findings can be considered as questioning bank loans "specialness", especially in a period of crisis. However, several results also confirm established conclusions regarding the effects of such characteristics as loan and firm size or the structure of syndicates on stock market perception of bank loan announcements. We document a significant change in bank behavior over the economic cycle, with reactions in terms of loan and syndicate features to a crisis environment. We also uncover the signaling role of loan spreads and borrower financial structure as well as the economic advantages of loan tranching. However, more research needs to be done to better understand bank specialness in the current economic and financial environment.
27
References
Altunbas, Yener, Blaise Gadanecz, and Alper Kara, 2006. Syndicated Loans: A Hybrid of Relationship Lending and Publicly Traded Debt. Palgrave Macmillan. Bae, Kee?Hong, Jun?Koo Kang, and Chan?Woo Lim, 2002. The value of durable ba nk relationships: evidence from Korean banking shocks. Journal of Financial Economics 64(2), 181-214. Bailey, Warren, Wei Huang, and Zhishu Yang, 2012. Bank Loans with Chinese Characteristic s: Some Evidence on Inside Debt in a State?Controlled Banking System. Journal of Quantitative Fin ancial Analysis 46(6), 1795?1830. Berger, Allen N., Leora F. Klapper, and Gregory F. Udell, 2001. The ability of banks to lend to informationally opaque small businesses. Journal of Banking & Finance 25(12), 2127-67. Berger, Allen N., and Gregory F. Udell, 1994. Did Risk?Based Capital Allocate Bank Credit a nd Cause a 'Credit Crunch' in the United States? Journal of Money, Credit and Banking 26(3), 585-628. Bernanke, Ben S., Cara S. Lown, and Benjamin M. Friedman, 1991. The Credit Crunch. Brookings Papers on Economic Activity 1991(2), 205-47. Best, Ronald, and Hang Zhang, 1993. Alternative Information Sources and the Informatio n Content of Bank Loans. Journal of Finance 48(4), 1507-22. Billett, Matthew T., Mark J. Flannery, and Jon A. Garfinkel, 1995. The Effect of Lender Identity on a Borrowing Firm's Equity Return. Journal of Finance 50(2), 699-718. ——— 2006. Are Bank Loans Special? Evidence on the Post?Announcement Performance of Bank Borrowers. Journal of Financial and Quantitative Analysis 41(4), 733-51. Boehmer, Ekkehart, Jim Masumeci, and Annette B. Poulsen, 1991. Event?study methodolog y under conditions of event?induced variance. Journal of Financial Economics 30(2), 253-72. Bosch, Oliver, and Sascha Steffen, 2011. On syndicate composition, corporate structure and the certification effect of credit ratings. Journal of Banking & Finance 35(2), 290-99. Chava, Sudheer, and Amiyatosh Purnanandam, 2011. The effect of banking crisis on ba nk? dependent borrowers. Journal of Financial Economics 99(1), 116-35. De Haas, Ralph, and Neeltje van Horen, 2010. The crisis as a wake?up call. Do banks tigh ten screening and monitoring during a financial crisis? Netherlands Central Bank, Research Department. Dell'Ariccia, Giovanni, Deniz Igan, and Luc Laeven, 2008. Credit Booms and Lending Standards: Evidence From The Subprime Mortgage Market. CEPR Discussion Papers. Dell'Ariccia, Giovanni, and Robert Marquez, 2006. Lending Booms and Lending Standards. Journal of Finance 61(5), 2511-46. Demyanyk, Yuliya, and Otto Van Hemert, 2011. Understanding the Subprime Mortgage Crisis. Review of Financial Studies 24(6), 1848 -1880. Diamond, Douglas W., 1984. Financial Intermediation and Delegated Monitoring. Review of Economic Studies 51(3), 393-414. ——— 1991. Monitoring and Reputation: The Choice between Bank Loans and Directly Placed Debt. Journal of Political Economy 99(4), 689-721. Duchin, Ran, Oguzhan Ozbas, and Berk A. Sensoy, 2010. Costly external finance, corpora te investment, and the subprime mortgage credit crisis. Journal of Financial Economics 97(3), 418-35. Fama, Eugene F. 1985. What's different about banks? Journal of Monetary Economics 15(1), 29-39. Fields, L. Paige, Donald R. Fraser, Tammy L. Berry, and Steven Byers, 2006. Do Bank Lo an Relationships Still Matter? Journal of Money, Credit, and Banking 38, 1195-1209. Focarelli, Dario, Alberto Franco Pozzolo, and Luca Casolaro, 2008. The pricing effect of certification on syndicated loans. Journal of Monetary Economics 55(2), 335-49. Fungá?ová, Zuzana, Christophe J. Godlewski, and Laurent Weill, 2011. Asymmetric Information
and Loan Spreads in Russia. Eastern European Economics 49(1), 13-29. Furfine, Craig, 2001. Bank Portfolio Allocation?: The Impact of Capital Requirements , Regulatory Monitoring , and Economic Conditions. Journal of Financial Services Research 20(1), 33-56. Gambacorta, Leonardo, and Paolo Emilio Mistrulli, 2004. Does bank capital affect lendin g behavior? Journal of Financial Intermediation 13(4), 436-57. Gatev, Evan, and Philip E. Strahan, 2009. Liquidity risk and syndicate structure. Journal of Financial Economics 93(3), 490-504.
28
Godlewski, Christophe J., Bulat Sanditov, and Thierry Burger?Helmchen, 2012. Bank Lendin g Networks, Experience, Reputation, and Borrowing Costs: Empirical Evidence from the French Syndicated Lending Market. Journal of Business Finance & Accounting 39(1?2), 113-140. Godlewski, Christophe J., Zuzana Fungá?ová, and Laurent Weill, 2011. Stock Market Reaction to Debt Financing Arrangements in Russia. Comparative Economic Studies 53, 679?693. Hancock, Diana, Andrew J. Laing, and James A. Wilcox, 1995. Bank capital shocks: Dynami c effects on securities, loans, and capital. Journal of Banking & Finance 19(3?4), 661-77. Huang, Weihua, Armin Schwienbacher, and Shan Zhao, 2012. When Bank Loans Are Bad News: Evi dence From Market Reactions to Loan Announcements Under the Risk of Expropriation. Journal of International Financial Markets, Institutions & Money 22(2), 233?252. Ivashina, Victoria, and David Scharfstein, 2010. Bank lending during the financial crisis of 2008. Journal of Financial Economics 97(3), 319-38. James, Christopher, 1987. Some evidence on the uniqueness of bank loans. Journal of Financial Economics 19(2), 217-35. Jensen, Michael C., 1986. Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers. American Economic Review 76(2), 323-29. Jensen, Michael C., and William H. Meckling, 1976. Theory of the firm: Managerial behavi or, agency costs and ownership structure. Journal of Financial Economics 3(4), 305-60. Keys, Benjamin J., Tanmoy Mukherjee, Amit Seru, and Vikrant Vig, 2010. Did Securitization Lead to Lax Screening? Evidence from Subprime Loans. Quarterly Journal of Economics 125(1), 307 -362. Lee, S.W., and D.J. Mullineaux, 2004. Monitoring, Financial Distress, and the Structure of Commercial Lending Syndicates. Financial Management 33, 107-30. Leland, Hayne E., and David H. Pyle, 1977. Informational Asymmetries, Financial Structure, and Financial Intermediation. Journal of Finance 32(2), 371-87. Lummer, Scott L., and John J. McConnell, 1989. Further evidence on the bank lending process and the capital?market response to bank loan agreements. Journal of Financial Economics 25(1), 99-122. Maskara, Pankaj K., 2010. Economic value in tranching of syndicated loans. Journal of Banking & Finance 34(5), 946-55. Maskara, Pankaj K., and Donald J. Mullineaux, 2011. Information asymmetry and self?selection bias in bank loan announcement studies. Journal of Financial Economics 101(3), 684-94. Mian, Atif, and Amir Sufi, 2009. The Consequences of Mortgage Credit Expansion: Eviden ce from the U.S. Mortgage Default Crisis. Quarterly Journal of Economics 124(4), 1449 -1496. Mosebach, Micheal, 1999. Market response to banks granting lines of credit. Journal of Banking & Finance 23(11), 1707-23. Ongena, Steven, David C Smith, and Dag Michalsen, 2003. Firms and their distressed bank s: lessons from the Norwegian banking crisis. Journal of Financial Economics 67(1), 81-112. Patell, James M., 1976. Corporate Forecasts of Earnings Per Share and Stock Price Behavi or: Empirical Test. Journal of Accounting Research 14(2), 246-76. Peek, Joe, and Eric Rosengren, 1995. Bank regulation and the credit crunch. Journal of Banking & Finance 19(3?4), 679-92. Preece, Dianna, and Donald J. Mullineaux, 1996. Monitoring, loan renegotiability, and fir m value: The role of lending syndicates. Journal of Banking & Finance 20(3), 577-93. Purnanandam, Amiyatosh, 2011. Originate?to?distribute Model and the Subprime Mortgag e Crisis. Review of Financial Studies 24(6), 1881 -1915. Ross, Stephen A., 1977. The Determination of Financial Structure: The Incentive?Signallin g Approach. Bell Journal of Economics 8(1), 23-40. Santos, João A. C., 2011. Bank Corporate Loan Pricing Following the Subprime Crisis. Review of Financial Studies 24(6), 1916 -1943. Slovin, Myron B., Shane A. Johnson, and John L. Glascock, 1992. Firm size and the information content of bank loan announcements. Journal of Banking & Finance 16(6), 1057-71. Sufi, A., 2007. Information Asymmetry and Financing Arrangements: Evidence from Syndicated Loans. Journal of Finance 62, 629-68.
Thomson Reuters, 2011. Global Syndicated Loans Review. Thomson Reuters. Wagster, John D., 1999. The Basle Accord of 1988 and the international credit crunch of 1989? 1992. Journal of Financial Services Research 15(2), 123-43.
29
Figure 1 Worldwide syndicated loans amounts and issues
This figure displays the evolution of the yearly loan amounts (left scale) and number of issues (ri ght scale) on the global syndicated lending market (source: Thomson Reuters).
30
0,8 0,6 0,4 0,2 CA R 0 ?10 ?9 ?8 ?7 ?6 ?5 ?4 ? 3 ?2 ? 1 ?0,2 ?0,4 ?0,6 0 1 2 3 4 5 6 7 8 9 10
Event time CAR CAR Crisis CAR Pre crisis
Figure 2 CAR around the event date
This figure displays the CARs over an event window of 21 days around the event date. The red li ne represents the CARs computed over the full sample period. The blue line represents the CARs compute d over the Crisis period. The green line represents the CARs computed over the Pre crisis period.
31
Table 1. Definitions and descriptive statistics for loan, bank syndicate and borrower characteristics
This table displays means and standard deviations for main loan, bank syndicate and borrower characteristics. Sample pe riod is 2000 until 2009. The number of observations varies because of data availability for particular variables. Data source: Bloomberg Professional Terminal Server. Variable Definition N Mean Std dev. Facility Loan facility amount in MLN USD 253 674.121 1 349.495 Spread Loan spread in bps 135 128.801 103.780 Maturity Loan maturity in years 253 2.921 4.703 Tranches Number of tranches in a loan facility 253 2.419 2.315 Term loan Dummy equal to 1 if the loan is a term loan, 0 otherwise 253 0.509 0.500 dummy Revolving loan Dummy equal to 1 if the loan is a revolving loan, 0 otherwise 253 0.387 0.488 dummy Secured dummy Dummy equal to 1 if the loan is secured, 0 otherwise 253 0.217 0.413 Covenants Dummy equal to 1 if the loan has covenants, 0 otherwise 253 0.233 0.423 dummy Lenders Number of lenders in the syndicate 253 8.565 7.945 Arrangers Number of arrangers in the syndicate 233 6.733 6.260 Local lenders Percentage of local (French) lenders in the syndicate 220 66.235 26.646 Local arrangers Percentage of local (French) arrangers in the syndicate 209 64.844 27.970 League table Percentage of lenders in the syndicate that are listed on the 230 66.190 23.090 lenders Bloomberg League table League table Percentage of arrangers in the syndicate that are listed on 214 66.801 23.934 arrangers the Bloomberg League table Assets Total assets in MLN USD 200 40 045.630 198 298.210 Sales Total Sales in MLN USD 206 7 990.400 17 459.120 Debt ratio Total debt / Total assets 200 34.082 21.854 Equity ratio Common equity / Total assets 200 28.322 24.433 Interest Ebitda / Total interest expenses 176 12.113 21.530 coverage ratio Current ratio (Cash + Accounts receivable + Short term investments + 176 134.936 89.671 Inventories) / Current liabilities Profit margin Net income / Total sales 206 9.461 25.981 Return on assets Net income / Total assets 199 3.750 5.825
32
Table 2. Stock market reactions to bank loan announcements
This table displays CAAR (cumulative average abnormal return) and CASAR (cumulative average abnormal standardized return) for the selected seven event windows over the entire sample time span (January 2000 until December 2009) and by boom (Pre?crisis) and bust (Crisis) period. Crisis starts from September 2007. The number of ban k loan announcement events is 253 (147 during the Pre?crisis period and 106 during the Crisis period). ***, ** and * indic ate CAAR and CASAR statistically different from 0 at the 1%, 5% and 10% confidence level according to Student tests. In th e last two columns, ** and * indicate a statistically significant difference in means at the 5% and 10% confidence level for the CAAR and CASAR between the Crisis and Pre?crisis periods. Data source: Bloomberg Professional Terminal Server. Crisis Pre?crisis Event CAAR CASAR Percent t?test t?test window of for for CAAR CASAR CAAR CASAR positive CAAR CASAR CAAR [0,0] ?0.4291* ?0.1558 ?0.2066 0.75 0.83 ?0.3001** ?0.0961 0.3872 [?1,1] [?2,2] [?2,0] [?1,0] [0,1] [0,2] ?0.3068* ?0.3062 ?0.2570 ?0.0938** ?0.0767 ?0.0397 0.4812 ?0.6458*** 0.4098 0.4549 ?0.3676 ?0.0033 ?0.2162*** ?0.0860 0.0319 0.0310 ?0.3092 ?0.5626 0.0032 ?0.0733 ?0.1042 2.29** 0.11 ?0.89 2.49** 0.09 ?0.91 ?0.3121 ?0.2622 ?0.2996 ?0.0828 ?0.0456 ?0.0726* 0.4587 0.4812 0.4662 ?0.5909 ?0.2228 ?0.6025** ?0.1589* ?0.0563 ?0.1814*** ?0.0561 ?0.2906 0.0118 ?0.0500 ?0.0237 ?0.0397 0.0127 1.07 ?0.11 1.74* 1.19 0.15 2.22**
33
Table 3. Loan, bank syndicate and borrower characteristics by stock market reaction
This table displays means and standard deviations for main loan, bank syndicate and borro wer characteristics by positive and negative CAAR for the [?1,0] event window (122 and 131 events respectively) and the results of t?tests or chi?2 tests for the means. The latter is used for binomial t est of proportion (dummy variables) while the former is used for Student tests of means (c ontinuous variables). All explanatory variables are defined as in previous tables. ***, ** and * indicate a statistically significant difference in means at the 1%, 5% and 10% confidence level for the relevant variables. Sample period is 2000 until 2009. Data source: Bloomberg Professional Terminal Server. Positive CAAR [?1,0] Negative CAAR [?1,0] Variable Mean Std dev. Mean Std dev. t?test or chi?2 test Facility 900.521 1 804.383 459.911 619.135 2.57** Spread 129.278 109.380 128.345 98.929 0.05 Maturity 2.861 0.2 Tranches 0.501 0.369 0.235 0.401 Lenders Arrangers 9.764 7.383 26.202 70.842 65.030 22.672 Assets Sales 70 860.070 11 849.140 0.95 Equity ratio 16.644 1.150 33.962 4.188 4.096 2.552 2.978 2.615 0.487 5.229 2.292 0.501 1.994 0.530 0.493
0.89 Term loan dummy 0.484 0.426 1.64 9.090 7.316
0.46 Revolving loan dummy 0.406 0.37 Secured dummy 0.200 0.268 7.430 6.132 60.708 0.401 0.444 6.520 5.049 26.104 58.306 23.518 25.334 29 199.330 9 796.820 35.493 29.849 26.075 0.502 11.533 3.290
0.47 Covenants dummy
0.200 2.33** 71.469 27.724
1.51 Local lenders 26.948 22.670 0.05 280 611.410 22 575.270 26.700
3.05*** Local arrangers 67.271 66.716
3.31*** League table lenders 66.880 2.09** 25.924 30.627 10.922 1.392 12.474 6.091
0.73 League table arrangers
11 026.210 4 488.950 16.479 15.353 13.479
2.98*** Debt ratio 32.585 0.93 Interest coverage ratio
0.76 Current ratio 1.303 0.67 Profit margin 6.140 1.83* Return on assets 5.555 1.09
34
Table 4. Loan, bank syndicate and borrower characteristics (crisis vs. pre?crisis period)
This table displays means and standard deviations for main loan, bank syndicate and borrow er characteristics by Crisis and Pre?crisis period (106 and 147 events respectively) and the results of t?tests or chi?2 tests for the means. The latter is used for binomial test of proportion (dummy variables) while the former is used for Student tests of means (continuous variables). All explanatory variables are defined as i n previous tables. ***, ** and * indicate a statistically significant difference in means at the 1%, 5% and 10% confidence level for the relevant variables. Data source: Bloomberg Professional Terminal Server. Crisis Pre?crisis Variable Mean Std dev. Mean Std dev. t?test or chi?2 test Facility 731.750 1 717.186 632.565 1 009.220 ?0.53 Spread 139.690 114.370 123.884 98.890 ?0.82 Maturity 4.844 6.486 1.535 1.820 ?5.11*** Tranches 2.566 1.809 ?0.80 Term loan dummy 0.476 0.473 0.501 0.429 2.879 0.557 2.313 0.499
1.59 Revolving loan dummy 0.330 0.497 2.51 Secured dummy 0.163 0.311 9.830 6.880 69.475 0.371 0.465 9.148 6.903 26.307 67.008 23.657 23.273 65.101 64.035 27.320 0.177 3.27***
0.292 0.457 6.04** Covenants dummy 0.383 Lenders Arrangers 6.811 6.505 26.750 63.418 68.121 24.289 Assets 6.22** 5.460 5.128
0.47 Local lenders 28.408 22.187 ?1.31
?1.49 Local arrangers 64.926
?0.91 League table lenders
?1.03 League table arrangers 69.435 59 408.280 275 540.020 ?1.03 Sales 7 209.810 35.768 17.665 ?0.97 Equity ratio Interest coverage ratio Current ratio Profit margin 29.300 11.233 1.415 13.153 5.870 ?1.46 20.106 1.249 31.168 ?0.89
Debt ratio
26 309.740 115 456.080 9 016.570 21 949.030 13 107.110 ?0.69 32.887 24.399 29.262 12.738 1.305 6.653 4.184 15.260 22.562 0.547 20.918 5.770 3.441 27.656 0.46 ?0.69
?1.70* Return on assets
35
Table 5. Loan, bank syndicate and borrower characteristics by stock market reaction (crisis vs. pre?crisis period)
This table displays means and standard deviations for main loan, bank syndicate and borrower characteristics by positive and negative CAAR for the [?1,0] event window and the results of t?t ests or chi?2 tests for the means over Crisis and Pre?crisis periods. The chi?2 test is used for binomial test of proportion (dummy variables) while the t?test is used for Student tests of means (continuous variables). All explanatory variables are defined as in previous tables. ***, ** and * indicate a statistically significant difference in means at the 1%, 5% and 10% confidence level for the relevant variables. Data source: Bloomberg Professional Terminal Server. Crisis Pre?crisis Positive CAAR [?1,0] Variable Facility Spread Mean Std dev. Negative CAAR [?1,0] Mean Std dev. t?test or chi?2 test ?1.80* ?2.09** 7.142 2.081 0.509 0.466 Positive CAAR [?1,0] Mean Std dev. Negative CAAR [?1,0] Mean 478.550 138.900 1.636 1.698 0.414 1.39 0.191 0.157 7.958 5.814 59.127 2.03** 22.538 68.102 13 227.300 4 904.900 33.988 28.405 20.765 1.537 12.197 3.202 Std dev. 716.470 99.322 1.538 2.486 0.497 0.500 0.395 0.366 7.173 4.935 27.380 57.364 0.47 23.588 37 714.200 11 462.800 30.447 15.284 0.07 1.612 19.403 5.573 t?test or chi?2 test ?1.85* 2.006 1.914 0.517 0.504 0.11 1.39 ?2.49** 69.496 28.572 64.043 ?0.61 ?1.22 ?1.88* 26.868 14.888 1.04 38.812 4.974
Secured dummy Covenants dummy Lenders Arrangers
Assets Sales Debt ratio
1 088.200 2 464.800 436.780 47.660 780.420 1 213.200 176.900 123.100 105.900 96.461 108.600 97.132 1.49 Maturity 4.937 5.668 4.768 ?0.14 1.534 0.01 Tranches 3.188 3.541 2.052 ?1.96* 2.147 1.14 Term loan dummy 0.612 0.492 0.504 0.80 0.503 2.42 Revolving loan dummy 0.306 0.351 0.482 0.382 0.489 2.62 0.245 0.435 0.333 0.476 0.70 0.121 0.329 0.286 0.456 0.333 0.476 0.75 0.207 0.409 6.854 5.348 6.776 5.598 ?0.07 11.627 9.224 6.425 5.073 6.569 5.220 0.13 7.917 8.292 ?1.84* Local lenders 63.579 23.693 73.868 27.505 1.85* 25.135 2.25** Local arrangers 60.056 26.395 72.078 27.129 69.870 26.991 2.52** League table lenders 66.853 21.937 69.072 23.150 65.875 24.328 0.45 League table arrangers 71.211 23.056 64.302 26.327 65.950 22.096 0.39 122 950.000 406 523.000 8 298.700 12 197.300 ?1.71* 38 738.000 156 639.000 15 717.000 31 420.000 4 024.100 7 603.400 ?2.25** 9 399.500 14 249.400 33.790 15.792 37.359 19.061 0.91 31.842 16.978 0.47 Equity ratio 23.938 15.269 33.545 13.991 2.99*** 39.128 ?0.28 Interest coverage ratio 11.039 19.472 11.369 29.305 10.546 12.361 ?0.98 Current ratio 1.257 0.464 1.330 0.525 1.279 0.572 ?0.47 Profit margin 4.766 1.53 7.011 11.109 6.276 27.859 ?0.19 Return on assets 5.864 1.4 3.345 6.437 3.543 5.253 0.18
36
Table 6. Regressions of CAR on main loan, bank syndicate and borrower characteristics
This table displays the results of OLS regressions with robust standard errors (in parentheses) where the dependent variable is the CAR [?1,0]. Crisis is a dummy variable equal to 1 if the bank loan announ cement occurs between September 2008 and December 2009, 0 otherwise. All explanatory variables are defined as in previous tables. Dummies for industrial sectors included but not shown. ***, ** and * indicate a statistically significant coefficient at the 1%, 5% and 10% confidence level. Data source: Bloomberg Professional Terminal Server. Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Crisis Crisis x log(Facility) Crisis x log(Maturity) Crisis x log(Lenders) Crisis x log(Tranches) Crisis x Term loan dummy Crisis x Secured dummy Crisis x Covenants dummy Crisis x log(Sales) Crisis x Equity ratio Crisis x Operating margin ?0.6759** (0.3073) ?0.0330** (0.0154) ?0.3624 (0.2795) ?0.2760* (0.1450) ?0.0215 (0.2795) ?0.2902 (0.4454) ?0.2038 (0.5999) ?0.5165 (0.4991) ?0.1114*** (0.0420) ?3.1248*** (1.0638) ?1.9462 (1.2738) ?1.0108*** (0.3456)
37
Table 6. continued
Variables log(Facility) log(Maturity) log(Lenders) log(Tranches) Term loan dummy Secured dummy Covenants dummy log(Sales) Equity ratio Operating margin Intercept N R²
(1) 0.2979** (0.1282) ?0.0310 (0.2148) ?0.1200 (0.2033) 0.0996 (0.2310) 0.2199 (0.3947) ?0.0130 (0.4679) 0.0962 (0.3294)
(2) 0.3150** (0.1293) ?0.0408 (0.2160) ?0.1205 (0.2035) 0.1137 (0.2309) 0.2150 (0.3952) ?0.0198 (0.4679) 0.0994 (0.3299)
(3) 0.2880** (0.1292) 0.0139 (0.2021) ?0.1255 (0.2012) 0.1077 (0.2312) 0.2656 (0.4058) ?0.0450 (0.4624) 0.0333 (0.3352)
(4) 0.2859** (0.1279) ?0.0886 (0.2161) 0.0380 (0.1989) 0.1559 (0.2313) 0.1866 (0.3977) ?0.0102 (0.4699) 0.0941 (0.3324)
(5) 0.2975** (0.1304) ?0.1517 (0.2174) ?0.0671 (0.2056) 0.1541 (0.2597) 0.2467 (0.4089) ?0.0329 (0.4663) 0.0559 (0.3396)
(6) 0.2994** (0.1296) ?0.1218 (0.2200) ?0.0793 (0.2051) 0.1164 (0.2300) 0.4059 (0.4517) ?0.0259 (0.4651) 0.0677 (0.3355)
(7) 0.2951** (0.1291) ?0.1518 (0.2127) ?0.0672 (0.2059) 0.1378 (0.2267) 0.2464 (0.4101) 0.0897 (0.5180) 0.0526 (0.3357)
(8) 0.3029** (0.1292) ?0.1409 (0.2133) ?0.0774 (0.2052) 0.1501 (0.2291) 0.2604 (0.4047) ?0.0492 (0.4718) 0.3606 (0.3851)
(9) 0.1195 (0.1534) ?0.0221 (0.2402) ?0.1247 (0.2324) 0.0402 (0.2657) 0.1829 (0.4237) 0.1451 (0.5018) 0.2013 (0.3781) 0.2742*** (0.0922) ?0.0502 (0.4466) ?0.3693 (0.7971) ?4.1374 (2.7696) 152 0.1372 1.4954
(10) 0.1195 (0.1530) ?0.0616 (0.2456) ?0.1137 (0.2344) 0.0502 (0.2656) 0.1967 (0.4288) 0.1272 (0.5025) 0.2062 (0.3790) 0.3266*** (0.0921) ?0.0936 (0.4430) ?0.4476 (0.7892) ?4.6000 (2.8178) 152 0.1257 1.7212
(11) 0.1180 (0.1507) ?0.0237 (0.2392) ?0.1011 (0.2229) 0.0613 (0.2654) 0.0675 (0.4152) 0.0895 (0.5149) 0.2413 (0.3689) 0.2267** (0.0933) 0.4249 (0.3057) ?0.2385 (0.8022) ?3.9515 (2.7226) 152 0.1464 2.4899
(12) 0.1298 (0.1474) ?0.1723 (0.2484) 0.0123 (0.2311) ?0.0075 (0.2532) 0.2630 (0.4410) 0.2073 (0.4811) 0.1917 (0.3880) 0.2394*** (0.0877) ?0.0869 (0.4252) 0.8072 (0.6532) ?4.7261* (2.7604) 152 2.4585
?5.1849** (2.4616) 195 0.0959 2.8214
?5.5395** (2.4860) 195 0.0943 2.0527
?5.1431** (2.4543) 195 0.0869 1.8463
?5.3442** (2.4784) 195 0.0899 1.8491
?5.5182** (2.5405) 195 0.0765 1.5041
?5.5458** (2.5129) 195 0.0785 1.8876
?5.4655** (2.5016) 195 0.0770 1.4944
?5.6365** (2.5046) 195 0.0802 1.5009
0.1086 F?stat.
38
Table 7. Financial constraints variables by stock market reaction (crisis vs. pre?crisis period)
This table displays means and standard deviations for six variables related to borrower's financial constraints by positive and negative CAAR for the [?1,0] event window and the results of t?tests or W ilcoxon / Kruskall?Wallis tests for the means over Crisis and Pre?crisis periods. The non?parametric Wilcoxon / Kruskall?Wallis tests are used for the Total lines of credit and Total available lines of cre dit variables as they have the least observations. The t?test is used for Student tests of means for all other variables. ***, ** and * indicate a statistically significant difference in means at the 1%, 5% and 10% confidence level for the relevant variables. Data source: Bloomberg Professional Terminal Server. Crisis Pre?crisis Positive CAAR (?1,0) Variable Retained earnings Free cash flow Short term borrowings Mean 1 416,513 Std dev. 6 246,156 776,106 21 239,490 16 057,010 66 122,470 ?1,43 Long term borrowings 1 749,968 Total lines of credit Total available lines of credit 3 084,038 5 276,217 4 331,588 14 069,640 Negative CAAR (?1,0) Mean ?14 098,780 127,891 533,339 4 617,426 997,369 911,257 Std dev. 101 592,700 741,922 844,621 ?1,68* 1 961,061 1 859,941 ?2.10** / 4.42** ?0.72 / 0.52 2 994,070 1 555,670 3 108,824 1 592,442 253,688 469,572 239,762 508,919 ?3.38*** / 11.47*** ?2.09** / 4.38** t?test or W. / K?W test ?0,97 ?1,21 ?1,39 2 851,609 Positive CAAR (?1,0) Mean 1 693,006 216,056 8 576,254 4 264,132 Std dev. 4 558,637 3 291,180 39 552,460 ?1,18 Negative CAAR (?1,0) Mean 1 267,558 358,286 1 118,950 5 901,872 Std dev. 6 060,784 890,314 4 443,141 18 536,490 t?test or W. / K?W test ?0,43 0,32
28 733,180 133 740,200
39
doc_189949881.docx
In finance, a loan is a debt evidenced by a note which specifies that, among other things, the principal amount, interest rate, and date of repayment. A loan entails the reallocation of the subject asset(s) for a period of time, between the lender and the borrower.
Bank loans and borrower value during the recent financial crisis Empirical evidence from France
Abstract We investigate the impact of bank loan announcements on borrower value during the recent boom and bust cycle of the 2000's using a sample of 253 large loans to French borrowers. We find no si gnificant stock market reaction to bank loan announcements during the boom period but a significa nt and negative one during the financial crisis. Hence, although we document significant cha nges in bank behavior during the crisis with conservative contractual and organizational m odifications, we cannot provide empirical support for the certification value of bank loans during a
period of increased informational asymmetries. However, bank loan announcements for larger firms receiving large loans funded by international pools of lenders contribute to borrower value even during the crisis.
Keywords: bank loans, borrower value, financial crisis, event study, syndicated lending, Europe. JEL classification: G14, G20.
A previous version of this paper circulated under the title: "Are bank loans still special (especial ly during a crisis)? Empirical evidence from a European country". I thank the participants of the AFFI 2012 Inte rnational Conference (Strasbourg), FEBS 2012 Conference (London), INFINITI 2012 Conference ( Dublin), IFABS 2012 Conference (Valencia), MFS 2012 Conference (Krakow) and 10th Corporate Finance Wo rkshop (Gent) in particular Pramuan Bunkanwanicha, Taufiq Choudhry, Issam Hallak, Iftekhar Hasan, Dorota Skala and Jonas Standaert for insightful discussions. The usual disclaimer applies.
1
Bank loans and borrower value during the recent financial crisis Empirical evidence from France
This version: September 2012
Abstract We investigate the impact of bank loan announcements on borrower value during the recent boom and bust cycle of the 2000's using a sample of 253 large loans to French borrowers. We find no si gnificant stock market reaction to bank loan announcements during the boom period but a significa nt and negative one during the financial crisis. Hence, although we document significant cha nges in bank behavior during the crisis with conservative contractual and organizational m odifications, we cannot provide empirical support for the certification value of bank loans during a
period of increased informational asymmetries. However, bank loan announcements for larger firms receiving large loans funded by international pools of lenders contribute to borrower value even during the crisis.
Keywords: bank loans, borrower value, financial crisis, event study, syndicated lending, Europe. JEL classification: G14, G20.
2
1. Introduction
The ongoing economic and financial turmoil that started in 2007 has (again) put financial institutions in the center of harsh debate and massive critics, in particular with respect to their role in fuelling and propagating the crisis as well as in provoking a credit crunch. Indeed, according to Dell'Ariccia et al. (2008) and Purnanandam (2011), banks had gradually relaxed their screening and monitoring standards before the crisis, especially in the US sub?prime mortgage market. Then, they sharply curtailed new credit and forced firms to reduce investments hence propagating the financial crisis to the real economy (Duchin et al. 2010; Ivashina and Scharfstein 2010; Santos 2011). These findings are somehow disturbing because according to the seminal contributions by Diamond (1984, 1991) and Fama (1985), financial intermediaries are considered as efficient in evaluating, screening and monitoring borrowers and play a specific role in managing the problems resulting from imperfect information on firms. As banks are believed to produce valuable private information regarding borrower's risk profile and quality, bank loan announcement should convey valuable information to the market about the borrower's financial situation. Empirical evidence tends to support the view that bank loans are thus "special" according to several authors, who find positive and significant abnormal returns for borrower's stocks around the date of a bank loan announcement (James 1987; Lummer and McConnell 1989; Preece and Mullineaux 1996; Focarelli et al. 2008). Bank loan signaling and certification role should be even more crucial during episodes of boom and bust such as the most recent one starting in the aftermath of the Internet bubble followed by the financial turmoil of 2007?2008. Indeed, de Haas and van Horen (2010) show that banks tighten screening and monitoring during a financial crisis when information asymmetries are exacerbated. Thus, the value of bank loan signaling and certification should be even more important during periods of financial turmoil, leading eventually to larger positive stock market reactions following a bank loan announcement. However, empirical
3
evidence from different episodes of crisis around the world (South?East Asia, Russia or Norway) show that the adverse shocks to banks also affect their borrowers' performance (Bae et al. 2002; Ongena et al. 2003; Chava and Purnanandam 2011). Hence, it is also possible to observe negative stock market reaction following a bank loan announcement during periods of financial crisis when lending banks experience episodes of financial distress, which can adversely impact their borrowers. Indeed, more recent empirical evidence seems to question the "specialness" of bank loans. Billett et al. (2006) find that bank loans are not "special" at all when abnormal returns are estimated over a longer period while Fields et al. (2006) suggest the diminishing market reaction to bank loan announcement is consistent with the dramatic change in the financial market. The results of event studies performed on samples from emerging markets' borrowers even show negative abnormal returns for bank loan announcements (Bailey et al. 2012 and Huang et al. 2012 for China and Godlewski et al. 2011 for Russia). < Insert Figure 1 > These issues are even more important regarding the largest market for external corporate financing in terms of bank debt: the syndicated lending market1. Its development provides a representative proxy for the boom and bust cycle (see Figure 1) with 2 trillion USD and 3000 issues in 2002, then 4.5 trillion USD and 9000 issues in 2007 and 4 trillion and 6500 issues in 2011. If we establish a parallel between loan syndication and securitization2, we can wonder if such techniques have reduced the incentives of lenders to properly perform their screening duties, as shown by Mian and Sufi (2009) and Keys et al. (2010) in the case of loan securitization. Also, due to the particular structure of syndicated loans, issues related to
1
A syndicated loan is granted by a pool of banks composed of lead (arrangers) and participant ba nks that provide funding to a borrower under a single agreement. 2 A securitization does not change the contract between the borrower and the original lender. Instead a new contract is created by the lender and a third party to sell the cash flow from the underlying loan. In a syndicated loan, all lenders are and remain part of one loan contract with the borrower.
4
informational frictions are more complicated and sever in such a setting. Private information available to some lenders may create an adverse selection problem while moral hazard problem may arise when the participant banks delegate some monitoring tasks to the lead bank. This market provides an excellent laboratory to investigate our main research question: are bank loans (still) "special", especially during a crisis? We aim here at revisiting the issue of bank loan "specialness", i.e. the certification value of bank lending, with a particular focus on the recent boom and bust cycle. To do so we perform an empirical investigation of stock market reactions to bank loan announcements during the 2000?2009 period using event study methodology. We perform empirical test of loan, bank syndicate and borrower characteristics influencing stock market reaction. We investigate if the stock market perception is different over the boom and bust period and to which loan, syndicate, and borrower characteristics this perception is the most sensitive. We also check if the recent crisis induced a shift in banks and borrowers behavior, in particular in terms of loan and syndicate characteristics during the boom and bust periods. We focus on the French syndicated lending market for several reasons. First, next to deals for US companies, syndicated loans to French companies are important sources of external financing. Indeed, such loans are constantly listed in the top global deals. For instance, among the 5 top deals ranging from 15 to 25 billion USD in 2011, French company CADES raised 16.6 billion USD through a syndicated deal. In the first quarter of 2012, among the top 10 syndicated loans, Eiffarie raised 4.6 billion USD. Second, our focus on the French syndicated lending market is motivated by its specific features, as bank syndicates lending to French companies are larger and less concentrated when compared to syndicates in the US or the UK (Godlewski et al. 2012). This particular structure may have important consequences on screening and monitoring of borrowers, thus influencing bank loan's "specialness" in France, especially during a financial crisis. Third, recent concerns regarding French banks liquidity and
5
solvency with respect to the Eurozone sovereign debt crisis appeal for a better understanding of stock market perception of bank lending decisions in this area3. The rest of the article is organized as follows. We present the relevant literature and testable hypotheses in section 2. Section 3 is devoted to the description of the data and methodology. Results are displayed and discussed in section 4. Finally, section 5 concludes the article.
2. Related literature and hypotheses
In this section we survey the relevant literature dealing with the "specialness" of bank loans and the syndicated lending market. We also discuss the impact of boom and bust cycles on bank behavior. 2.1. The "specialness" of bank loans and the syndicated lending market
There is consensus in the literature that bank loans are significantly different from other forms of corporate external finance. Indeed, financial intermediation theory argues that banks are unique institutions because they gain insider information and knowledge on firms through lending and deposit relationships (Fama 1985; Diamond 1991). Hence, the traditional informational view of bank loans argues that banks, as large creditors, can produce valuable private information about borrowing firms through initial screening and monitoring. Therefore, lending decisions reveal positive private information about the firms because banks would lend to high?quality borrowers, rather than to those of low?quality, to maximize the value of the loans. A large body of empirical research shows that announcements of bank loan agreements are associated with positive abnormal returns for borrowers on average. In other words, stock markets treat bank loan financing as good news and bank loan announcements therefore convey positive information regarding borrower's conditions. Indeed, bank loans, or debt more generally, can create value by reducing overinvestment by non?congruent
3
"What's the Matter With the French Banks?", The Wall Street Journal, 13/9/2011; "Moody's Downgrade: SocGen, Credit Agricole's Liquidity Problems Larger Than Greece", Forbes, 14/9/2011
6
managers (Jensen and Meckling 1976) or by giving a manager the opportunity to signal the quality of the firm and his willingness to be monitored by lenders (Diamond 1991)4. Thus, bank loans are considered as "special", starting with the seminal work of James (1987) who finds a sizeable average excess return following announcements that firms have signed a bank loan agreement. Many further studies confirm and refine this result. Lummer and McConnell (1989) report significant average excess returns for favorable loan revision announcements while Slovin et al. (1992) show that bank loan announcements are particularly good news for firms with severe information asymmetry, such as small firms. According to Best and Zhang (1993), firms that face greater earnings uncertainty and lack sufficient evaluation and monitoring by other stakeholders benefit most from bank loan announcements. Higher positive excess returns following loan announcements are also associated with more reputable lenders (Billett et al. 1995). The global syndicated lending market represents a significant portion of external financing for companies, as almost 4 trillion USD of debt had been raised on this market in 2011 (Thomson Reuters 2011). The benefits of loan syndication both for lenders (portfolio risk and sources of revenues diversification) and borrowers (mostly lower costs as compared to bond issues or a series of bilateral loans) largely explain the success of syndicated lending. A syndicated loan embeds both features of bank lending: transactional and relationship (Altunbas et al. 2006). It is therefore also "special" as any bank loan and most of empirical research tends to show that it is true. Indeed, loans generate positive abnormal returns and consequently are special when they are made by syndicates with fewer lenders (Preece and Mullineaux 1996) or with larger portions of the loan retained by arrangers (Focarelli et al. 2008).
4
In contrast, announcements of SEO (seasoned equity offerings) generate an average negativ e abnormal return, whereas announcements of public bond issues generate zero or slightly negativ e equity returns, according to previous research.
7
Overall, an empirical consensus seems to emerge from previous research regarding bank loans' "specialness" as certification and signaling device regarding borrowers' quality. Hence we can expect to observe a positive reaction of investors to a bank loan announcement, materialized by a significant and positive abnormal return for the borrowing firm's stock around the announcement date. However, there also exists empirical evidence showing that bank loan announcements can be considered as bad news with negative abnormal returns (Billett et al. 2006). Such findings are particularly frequent in the case of emerging market economies (Bailey et al. 2012; Godlewski et al. 2011; Huang et al. 2012). These recent findings may question the empirical consensus in favor of bank "specialness". Furthermore, the specific features of syndicated lending may have potential adverse effects on the stock market reaction to bank loan announcements. Indeed, syndicated loans have their drawbacks because the nature of a syndicated loan may expose the banking pool's members to the adverse consequences of informational frictions and potential agency costs. First, private information about the borrower can create adverse selection problems, as the arranger may be inclined to syndicate loans for unreliable borrowers. Second, participating banks may delegate monitoring to the arranger, but the banks are not in the loop as to what the arranger is doing, which might result in situations of moral hazard. Thus, for all these reasons we might also observe an insignificant or even negative abnormal return for the borrower around the bank loan announcement date. 2.2. Bank lending during boom and bust periods
Much of the research on bank lending behavior, qualified as procyclicality in a boom and bust framework, has focused on credit crunches during business cycle downturns. Several hypotheses for these crunches were tested and partially validate. Hence, it appears that credit crunches can be explained by reduced risk taking by banks (Wagster 1999; Furfine 2001), implementation of tougher regulatory capital standards (Berger and Udell 1994; Hancock et al.
8
1995) or increasing supervisory toughness (Peek and Rosengren 1995; Gambacorta and Mistrulli 2004), as well as reduced loan demand (Bernanke et al. 1991). More recently, Dell'Ariccia and Marquez (2006) argue that banks may loosen their lending standards and thus lead to deteriorated loan portfolios, lower profits, and expanded aggregate credit because information asymmetries decrease during economic growth periods. With respect to the most recent episode of boom and bust, Demyanyk and Van Hemert (2011) find that the quality of loans deteriorated for six consecutive years before the crisis and that securitizers were aware of it. De Haas and van Horen (2010) provide additional evidence regarding bank lending behavior during the global financial crisis by analyzing changes in the structure of syndicated loans. They find an increase in retention rates among syndicate arrangers during the crisis, especially in case of important information asymmetries between the borrower and the syndicate or within the syndicate. They interpret their findings as a "wake?up call" with increased screening and monitoring by banks during the bust period starting in 2007. Following these results, we can expect that such reaction in bank lending behavior should translate in a greater certification and signaling role of bank loans and hence their "specialness" during a crisis. We could observe a positive stock market reaction to bank loan announcements during the bust cycle if investors believe in a "wake?up call" of banks. Indeed, with increased information asymmetries during a crisis period, banks should react mainly through two channels: contractual and organizational. Regarding the former, banks should become more conservative in order to better mitigate adverse selection and moral hazard problems between the borrower and the lenders by adjusting main loan terms (more collateral, more covenants, and longer maturities) in order to screen and monitor borrowers more tightly. Regarding the latter, banks can also adapt the structure of the syndicating pool with fewer lenders and with a greater percentage of local banks in order to reduce informational frictions within the syndicate and enhance borrower's monitoring. These
9
contractual and organizational adjustments should translate into a greater certification value of bank loans during a crisis and hence significant and positive abnormal returns for borrower's stock around a bank loan announcement date. On the contrary, during the boom or pre?crisis period, we shall observe insignificant abnormal returns as less information asymmetries lead to relaxed lending standards which in turn diminish the certification value of a bank loan. However, we may also obtain an opposite result with stock markets sanctioning bank loan announcements perceived as signals of borrower weakness during economic and financial turmoil. Indeed, and in particular on the syndicated lending market, troubled borrowers could be the first to ask for bank debt funding, especially in the form of credit lines. Indeed, Ivashina and Scharfstein (2010) show that after Lehman Brothers collapse, borrowers, especially those financially constrained, heavily drawn down their lines of credit, inducing banks to limit new loans. Hence, it could also be the case that investors would sanction external financing from distress lenders, especially during a severe financial crisis. For all these reasons, we could also find significant and negative abnormal returns for borrower's stock because of investors' concerns regarding firm's and/or lender's conditions.
3. Data and methodology
In this section, we provide a description of the data and relevant descriptive statistics, followed by an explanation of the methodology. 3.1. Data
Data on equity prices, loan and syndicate characteristics and borrower balance sheet for French companies over the 2000?2009 timespan are extracted via the Bloomberg Professional Terminal Server. Bloomberg provides detailed information on the terms of loan agreements, the composition and structure of the lending syndicate and accounting data for the borrowing companies. The main filter we apply concerns stock price availability over at least 150 trading days before the date of bank loan announcement. Additional filters concern syndicate characteristics and balance sheet data availability. The final full sample contains 253 bank loan 10
announcements, each for a unique company, a figure which is within the range of events in previous studies (from 117 to 728 events) as reported by Maskara and Mullineaux (2011). The number of bank loan announcements increase over time, with 6 events in 2000, a peak of 50 events in 2007 and 23 events in 2009. Table 1 provides summary definitions and descriptive statistics for main loan, syndicate and balance sheet variables for the full sample. Overall, syndicated loans for French borrowers are large (almost 700 MLN USD) with a maturity of almost three years and a spread close to 130 bps over Libor or Euribor. A typical loan facility is composed of more than two tranches5. Half of the loans are term loans and 40% are revolving loans. One out of five loans is secured and has covenants. Bank syndicates are composed of almost nine lenders of which an important part bear arranger titles (such as lead arranger, mandated arranger or arranger). More than 2/3 of lenders are French banks and we observe a similar figure for the arranging banks. We remark that figures for league table lenders6 are very similar to those for French banks (actually, French banks in the sample are often listed on Bloomberg League tables). The sample contains large firms with respect to their balance sheet or sales (40 and 8 BLN USD respectively). Common equity and debt ratios represent each roughly 1/3 of total assets while Ebitda amounts for more than 10% of interest expenses. Firms are relatively liquid according to their current ratios, with a good profitability with respect to profitability margins
5
Syndicated loans can be "tranched" into heterogeneous components that can then be distributed across lenders differentiated by their risk aversion. This technique is somehow close to tranching in a securitization process. 6 We consider a lender to be part of the league table if it is listed as one of the first 25 fin ancial institutions in the Bloomberg Underwriter Rankings Table, computed according to lender's marke t share, amount issued and number of issues between 2000 and 2009 for the European Market. We choose the 25th rank as a cutoff because below this rank the market share of a lender is lower than 1%.
11
as well as return on assets. Overall, these figures suggest a good level of the firms' quality in our sample7. 3.2. Methodology
We consider a bank loan announcement as an event and identify bank loan announcement dates in Bloomberg and consider this date as day 08. We adopt a classic approach based on the market model which relates the return of a given stock to the return of the market index: (1) where is the return on the share price of borrower on day , the stock market return is 1
on day and and are the parameters to be estimated over an estimation window. an error term with 100, where 0 and . Returns are defined as /
is the daily closing stock market price at time and we proxy for the market
return using the SBF 250 stock index return9. We consider an estimation window of 100 to 10 days prior to the event date10 and we use OLS regressions to estimate the market model. Daily abnormal returns are obtained following: | where is the actual return on the share price of borrower on day while | (2) is the
normal return where
is the conditioning information for the market model, i.e. the market
return. Following previous studies (see Maskara and Mullineaux 2011 for a summary), we examine seven different event windows: three symmetric ones (one?day [0,0], three?days [?1,+1], five?days [?2,+2]) and four asymmetric ones (two?days [?1,0]; [0,1] and three?days [? 2,0]; [0,2]). The latter, especially [?1,0] and [?2,0], serve also the purpose of verifying the
7 8
The most important industrial sectors in the sample are Consumer (37,93%) and Industrial (20,69%). It is necessary to make sure that there is no other corporate news that could influence stock returns within an event window. We check carefully and find no contamination caused by other events around our event dates. 9 Our results do not change when using CAC 40 or SBF 120 stock index but provide lower statisti cal quality of the regressions (R² lower than 10%). 10 Using longer estimation windows ranging from 150 days to 30 days prior to the event date gi ves similar results.
12
existence of potential information leakage. For each event window, we compute the cumulative abnormal returns (CAR): , where and ? (3)
are respectively the lower and upper bounds of an event window.
Standardized cumulative abnormal returns (SCAR) are obtained by dividing CAR by
?
, where is the number of days within a given event window and
is
the variance of the abnormal return estimated from equation (1). Cumulative average abnormal returns (CAAR) are defined as: , ? , (4)
where N is the number of borrowers in the sample on day . Cumulative average standardized abnormal ? returns
?
(CASAR) .
are
obtained
by
dividing
CAAR
by
We proceed in two steps. First, we perform a univariate analysis using t?tests to investigate the statistical significance of CAAR and CASAR with the null hypothesis being that the CAAR or CASAR equals 0. We also perform similar tests (t?test or chi²?test depending on the nature of the variable under consideration) to investigate the statistical significance of differences in various loan, syndicate and borrower variables with respect to positive and negative CAAR. Then, we repeat the tests with respect to two different periods of our sample: before and after the crisis. We define the period between January 2000 and August 2007 as the Pre?crisis period while the period from September 2007 to December 2009 is considered as the Crisis period. Again we test the statistical significance of CAAR and CASAR as well as of various variables with respect to positive and negative CAAR for the Crisis and Pre?crisis sub? periods. Second, we perform a multivariate analysis by regressing the borrowers' CAR on a Crisis dummy (equal to 1 for bank loan announcements during the Crisis period, i.e. between
13
September 2007 and December 2009, 0 otherwise) and interaction terms between the latter and most important variables related to loan, syndicate and borrower characteristics. By doing so we aim at testing the robustness of the univariate results in a multivariate setting, in particular regarding the effect of the crisis on stock market reaction to bank loan announcements. We also want to investigate more in details the influence of the interaction between the crisis effect and other loan, syndicate and borrower characteristics on the abnormal returns. We focus on the main loan terms such as size, maturity, type, the presence of collateral or covenants. We also consider the size of the syndicate as well as main balance sheet characteristics of the borrowers such as size, solvency and profitability. The equation of interest to be estimated by OLS regression with robust standard errors can be formally defined as: (5) is the cumulative abnormal return for borrower and is defined in equation (3). and are the coefficients for the Crisis dummy and the interaction variables between the Crisis dummy and several Variables of interest related to loan, syndicate and borrower characteristics. We test several specifications depending on the vector of control variables (Controls), which can be loan and syndicate related only, or also including borrower characteristics. The latter implies a reduction in the number of usable observations due to balance sheet data availability.
4. Results
In this section, we first discuss the univariate results regarding stock market reaction to bank loan announcements for the full sample period and for the boom and bust periods. We also investigate loan, syndicate and borrower characteristics related to positive and negative stock market reaction. Then we provide the multivariate results relating the CARs to loan, syndicate
14
and borrower variables. Finally, we investigate the financial constraints issues of borrowers and their impact on stock market reaction. 4.1. Univariate results
We first present our main univariate results regarding stock market reaction to bank loan announcements over the full time span of the sample as well as by boom and bust periods in Table 2. We first discuss results provided in columns 2 to 3 in table 2. We remark that the CAAR are positive for 40% to 50% of bank loan announcements. Nevertheless, we observe systematically negative stock market reaction but only significant for three event windows: [? 2,0], [?1,0] and [0,0], with approximately ?0.30 for CAAR and ranging from ?0.07 to ?0.09 for CASAR11. We conclude that bank debt financing through a syndicated loan by French companies is considered as a negative signal by the stock market. Furthermore, we can also claim that some form of information leakage seems to be at work as significant reaction is observed for windows before the loan announcement event. This first series of results do not confirm previous findings that bank loans are special (James 1987; Lummer and McConnell 1989; Preece and Mullineaux 1996; Focarelli et al. 2008). We rather provide empirical support for conclusions reached by Billett et al. (2006), Fields et al. (2006), Bailey et al. (2012), Godlewski et al. (2011) and Huang et al. (2012). In the French case, bank loan announcements are considered as bad news by the stock market refuting bank's specialness arguments as well as certification and signaling role of bank debt financing. We now turn to the results provided in columns 4 to 9 in table 2. First of all we remark that most of stock market reactions are negative, confirming previous results. Thus bank loan announcements are considered as a negative signal by investors. However, these reactions appear to be significant only during the crisis period as CAARs and CASARs are statistically
11
We reach similar conclusions when using alternative t statistics such as Patell (1976) or Boehmer et al. (1991).
15
different from 0 mainly for the [?2,0] and [?1,0] event windows, while there are no significant market reactions during the pre?crisis period. Furthermore, for these particular event windows, both CAAR and CASAR are statistically different regarding the sub?periods under investigation according to the results of t?tests in column 9. Finally, in absolute value, CAAR and CASAR are overall larger during the crisis. For instance, the CAAR for the [?1,0] event window is more than 20 times larger during crisis than before. It is also twice the CAAR for the full time period under investigation (2000?20009). Overall, we can claim that a bank loan announcement is perceived differently with respect to the economic environment (Crisis vs. Pre?crisis) and that it is considered as a negative signal by market participants during the crisis period, while it is not considered as a signal at all before the turmoil. Hence, bank loan announcements appear to be considered as bad news during a period of economic and financial turmoil, while they are perceived as insignificant during a boom period. Although contrary to some of previous empirical findings, this result receives support from recent research on the 2007?2008 crisis. Dell'Ariccia and Marquez (2006) and Demyanyk and Van Hemert (2011) have shown that banks have relaxed their lending standards during the boom period leading to a deterioration of their loan portfolio's quality and of the certification value of bank loan announcements. Our results for the pre?crisis period support the hypothesis that relaxed lending standards during a period of reduced informational asymmetries diminish the certification value of bank lending. Even if De Haas and van Horen (2010) provide evidence on a "wake?up call" with increased screening and monitoring by banks during the bust period starting in 2007, our results tend to show that providing a loan to a borrower during the crisis is perceived negatively by the stock market. This can be related to several issues. First, even with more conservative lending standards, investors can still doubt in the capacity of banks to identify valuable borrowers on the credit market, especially if banks are perceived as more vulnerable due to an adverse economic and financial environment. Second, we can also expect that on
16
average, lower quality borrowers need to apply for bank loans during a crisis, especially through credit lines (Ivashina and Scharfstein 2010). For illustrative purposes, we also provide in figure 2 the evolution of CARs over a [? 10,10] time window for the whole sample (red line), the Crisis (blue line) and the Pre?crisis (green line) periods. We remark that the full sample CAR experience a sharp decline one day before the event date and drops to around ?0.3. Then it remains at this level for around two days. We observe similar patterns for both Crisis and Pre?crisis CARs but the drop is larger for the Crisis CAR which almost reaches ?0.6. However, further investigation is needed to better understand these results and verify which features of the loan contract, the syndicate and the borrower play a significant role in shaping stock market reaction. In what follows we focus on the most significant CAAR using the [?1,0] window12. We aim now at investigating those characteristics that are associated with a positive stock market reaction. To do so we perform t?tests or chi²?tests (depending on the nature of the variable under consideration) on the difference of various variables with respect to a dummy equal to 1 if the CAAR [?1,0] is positive (122 events), and equal to 0 if the CAAR [? 1,0] is negative (131 events). The results are displayed in Table 3. Regarding loan characteristics, we observe that the only significant feature is the facility amount. The stock market reaction is positive for larger loans (twice as large as loans with negative CAAR). This result can be linked to our findings regarding bank syndicate characteristics, as we remark that larger syndicates with fewer local lenders are associated with positive CAAR. Regarding firm characteristics, we remark that significant differences in stock market reaction are essentially related to firm size measured with total assets and sales.
12
All results are similar when using other less significant windows as well as CASAR.
17
According to these results, the French stock market considers that large loans, funded by large syndicates of which a smaller proportion is composed of local banks, are a positive signal. Indeed, a larger loan funded by a more diffuse syndicate can be considered as a good signal regarding borrower's quality. The size of the loan can be interpreted as reinforcing the certification and signaling role of the bank lending decision (Mosebach 1999) while a larger syndicate is usually associated with less informational frictions and their subsequent consequences in terms of adverse selection and moral hazard in the relationship between the borrower and the lenders (Lee and Mullineaux 2004; Sufi 2007; Bosch and Steffen 2011). This can also be related to our findings regarding borrower characteristics. Indeed, the market values positively loan announcement by large firms with important sales, thus more visible and less opaque companies with sustained economic activity. Meanwhile, the presence of numerous lenders can also serve as a device to mitigate eventual liquidity risk in funding the loan to the borrower as well as a risk diversification device, in particular when funding a large loan (Gatev and Strahan 2009). However, the result regarding syndicate size does not confirm previous results by Preece and Mullineaux (1996) who show, using a sample of bank loans provided to US borrowers, a positive reaction to loans funded by smaller syndicates. A positive market response to bank loan announcement involving less local lenders is more puzzling. Indeed, one could expect the opposite as local lenders presence help to mitigate the adverse consequences of informational asymmetries both between the borrower and the syndicate as well as within the syndicate (Berger et al. 2001). However, this effect is not systematically true as shown recently by Fungá?ová et al. (2011). Hence, we can argue here that the larger presence of foreign lenders can be considered by the stock market as a better and/or more objective signal regarding deal and borrower quality. This argument is even more appealing with respect to the recent fragility of French banks following the 2007?2008 crisis.
18
Next, we investigate more in details stock market reaction to bank loan announcements during the recent boom and bust cycle. The results are displayed in Table 4. First of all, we remark that there are significant differences regarding loan maturity and contractual features such as loan collateralization or covenants. Indeed, maturity is more than three times larger during crisis, and one out of three loan contracts are secured and have covenants, while these features are only present for less than 20% of loans before the crisis. These loan characteristics tend to show a change in bank behavior during the crisis due to increase borrower default risk, uncertainty and informational frictions. In particular, loan characteristics aiming at reducing adverse selection (security) and moral hazard (covenants) problems are reinforced during the crisis period. Larger maturities imply also that banks provide longer term funding to dilute the cost of bank debt for borrowers even at the expense of larger spreads. These results are in line with the "wake?up call" argument provided by de Haas and van Horen (2010). Meanwhile, we also remark that the only significant feature of the bank syndicate that changes significantly is the number of lenders, which is reduced by three banks during the crisis. This again confirms a change in bank behavior and is consistent with changes in loan characteristics as a smaller syndicate is better suited to cope with borrower monitoring and mitigating agency costs within the syndicate (Lee and Mullineaux 2004; Sufi 2007; Bosch and Steffen 2011). It can also be explained by the difficulties of financial institutions during that period and thus their weaker willingness to fund syndicated loans. Finally, the only borrower characteristic exhibiting a significant (although statistically weak) change during the crisis is profitability which is twice larger than before. Overall we find that bank behavior has changed during the crisis and has become more conservative as a reaction to increased informational asymmetries. Finally, we investigate differences in loan, syndicate and borrower characteristics for positive and negative stock market reactions during and before the crisis (Table 5). Regarding
19
loan characteristics, apart from loan size which exhibits similar features for positive CAAR as for the full sample (larger loans are associated with positive stock reaction), we remark that during the crisis, loans with larger spreads (70 bps larger on average) and more tranches (1 tranche more on average) were associated with a positive stock market reaction. It is also worth noticing that a positive reaction is associated with an average loan size of 1 billion USD during the crisis while the same is true for a 700 million USD loan before the crisis. The evidence is completely inverted for spread: before the crisis, positive reactions are related to lower spreads while they are associated with larger spreads during the crisis. The spread result can be analyzed within the Spence costly signal framework. In an environment plagued with greater uncertainty and thus informational asymmetry, the capacity to pay a higher spread can be interpreted as a signal regarding the expected performance of the borrower. But we can also consider that the stock market perceives higher spreads as a signal of reinforced lending standards of the banks, i.e. more risk adjusted loan pricing, and greater certification value. This can be related to the result regarding the tranching of syndicated loans. Following recent evidence by Maskara (2010), multiple tranches actually create economic value and provide benefits for riskier borrower even if on average, the credit spread for a multi?tranches loan is larger. This is because without tranching, such spread would be even larger, eventually leading to adverse selection effects. We also observe differences regarding bank syndicate features, as the size of the syndicate and the number of arrangers are significantly different for positive and negative stock market reaction but only before the crisis. Larger syndicates with more arrangers are associated with positive CAAR according to the argument relating such syndicate structure with less informationally problematic deals and borrowers. Other syndicate features such as the percentage of local lenders or arrangers exhibit similar level of significance as for the full period (cf. Table 3).
20
Finally, we also remark differences regarding borrower characteristics such as such as size (measured by total assets or sales) which exhibit similar significant levels by CAAR during crisis or no?crisis periods as for the full period (cf. Table 3). In other words, size matters as larger borrowers experience positive abnormal returns around the bank loan announcement date, in particular during the financial crisis. 4.2. Multivariate results
We present now the results of a multivariate analysis of the relationship between borrower's market value, measured by the most significant CAR [?1,0], and various loan, syndicate and borrower characteristics, with a particular focus on the effect of the recent financial crisis. The latter is captured with a dummy variable equal to one if the bank loan announcement occurs between September 2007 and December 2009. We also interact this dummy with other variables, for which we restrained our analysis to the most important and significant characteristics (regarding the univariate analysis results). These are the loan facility amount, maturity or the size of the lending syndicate, as well as loan terms variables such as the type of the loan, the number of tranches as well as the presence of collateral and covenants13. Due to limited data availability, we restrict the borrower characteristics to three main variables: sales, equity ratio and operating margin. Results are displayed in table 6. First we discuss the results obtained with loan and syndicate variables only, i.e. regressions (1) - (8). We remark a significant and negative coefficient for the Crisis dummy, confirming univariate results. Issuing a loan during the recent bust is not considered by investors as a positive signal regarding the borrower's profile. In other words, the recent financial crisis reduces (even destroys) borrower's market value steaming from a bank loan announcements.
13
We are unable to provide viable regressions results when including the Spread variable as it oft en unavailable in our sample.
21
Next we observe that among the seven variables capturing different characteristics of the loan or the syndicate, only two bear significant coefficients when interacted with the Crisis dummy. These are the Facility and Lenders (in log) variables. The result for the loan amount confirms our univariate conclusions as larger loans are associated with positive CAAR even during the crisis. Indeed, the combined coefficient for log(Facility) equals (?0.033+0.315) which is larger than the Crisis dummy coefficient. Hence even during a bust period where bank loan announcements are perceived as negative news and can destroy firm's market value, announcements of large loans can counteract this adverse effect and end up in a positive effect on the market value of the borrower. An additional confirmation for that result is the fact that the individual coefficient for log(Facility) is always significant and positive across all regressions (1) to (8). The result for the size of the syndicate is statistically weaker, as the coefficient for the interaction term is significant at the 10% confidence level only and the log(Lenders) coefficient itself is not significant. In other words, the crisis effect dominates the syndicate size effects with respect to abnormal returns. This conclusion somehow confirms the univariate result where we found that this syndicate characteristic doesn't matter for the stock market reaction during crisis. When turning to the specifications (9) to (11) with borrower's balance sheet variables, we confirm again the negative relationship between the crisis and stock market reaction to bank loan announcements. We also remark that two of the three main borrower characteristics, size measured by log(Sales) and solvency measured by the common equity to total assets ratio, bear significant and negative coefficients. The borrower size effect confirms univariate findings as the combined coefficient for the Crisis dummy remains positive (? 0.1114+0.3266) although the interaction term is significant and negative. Overall, size matters for a bank loan announcement to be perceived as a positive signal for investors during the recent financial crisis, as larger borrowers receiving large loans experience positive abnormal returns around the bank loan announcement date.
22
The result for equity ratio can appear as counterintuitive because less capitalized firms can be considered as more fragile, especially during a crisis. However, we can also remind that the corollary of equity is debt which has been found to work as a signaling and disciplining device (Leland and Pyle 1977; Ross 1977), helping to solve adverse selection that results from information asymmetries between firm insiders and outsiders. Indeed, debt can reduce agency costs resulting from conflicts of interest between shareholders and managers as it increases the pressure on managers to perform and stop wasting company resources and increase their effort by restricting the 'free cash?flow' at the disposal of managers (Jensen 1986). Moreover, a high?quality firm can issue more debt than a low?quality firm, because the issuance of debt leads to a higher probability of default due to debt?servicing costs. Thus, receiving a bank loan during a financial crisis can be viewed by investors as a strong signal regarding borrower's quality certified by the lenders. Finally, we can also argue that equity investors value the monitoring role of bank debt holders in a period of adverse economic conditions and increased uncertainties regarding borrower's prospects and behavior. 4.3. Financial constraints In a nutshell, our findings do not support the certification value of bank loan announcements during the crisis although we uncover a "wake?up call" effect of banks' behavior, with more conservative loan terms as well as syndicate structures. Although some of our results are consistent with previous findings regarding loan and borrower characteristics related to a positive abnormal return, such as loan or firm size, we cannot fully explain the drivers behind a negative stock market reaction during the crisis relying solely on the certification and signaling arguments. Indeed, lenders' reputation do not seem to matter for stock market perception as we do not find any statistically significant difference between positive and negative abnormal returns across the boom and bust periods for syndicates with large or small portions of league table members. Thus we turn to alternative explanations related to financial constraints of borrowing firms following notably Ivashina and Scharfstein
23
(2010), by examining several characteristics of borrowers related to their retained earnings, free cash flow, (short and long term) borrowings and (total and available) lines of credit14. We replicate the previous analysis on a battery of six variables related to borrower's financial constraints in Table 7. < Insert Table 7> We first remark that retained earnings and free cash flow (FCF) are systematically larger for positive stock market except for FCF during the boom period. The difference is particularly dramatic for retained earnings which become heavily negative for negative abnormal returns during the crisis. We observe a similar pattern for short and long term borrowings, as investors positively react to bank loan announcements to firms with large borrowings, especially during the crisis. Hence, it seems that investors value bank loan announcements to less financially constrained borrowers. However, none of these variables exhibit a statistically significant difference in means according to the t?tests. Finally, we also uncover that the stock market perception of bank loans is also sensitive to borrower's line of credits as this variable is again much larger as compared to firms with negative abnormal returns. Furthermore, this characteristic appears to be statistically significant in terms of difference between the positive and negative stock market reaction. Thus we have some empirical support for the argument that the negative stock market reaction to bank loan announcements is related to borrower's financial constraints. This effect is persistent whenever the period under investigation (i.e. crisis vs. pre?crisis.). Hence, investors seem to be sensitive to firms' financial conditions when they apply for a bank loan, and when these conditions are weak, the stock market perception of bank loan announcements is significantly negative. Moreover, following our results, this effect might be stronger than the "classical" certification value effect which should be even stronger during a crisis when lenders adjust
14
Unfortunately, investigating such detailed variables comes at the cost of losing a substantial portion of the sample due to data unavailability. The number of available observations for the six variables under con sideration in the full sample is 200, 188, 196, 200, 61, and 58 respectively. This is why we do no t include them in our main analysis in previous sections.
24
loan terms and syndicate structures to greater informational frictions in the economy following a "wake?up call". Overall, it seems that investors are more sensitive to borrowers' financial conditions during a bust period even if lenders' behaviors change to reflect a tougher economic environment.
5. Conclusion
We have empirically revisited the issue of bank loans "specialness" with a particular focus on the recent boom and bust cycle to provide a better understanding of stock market perception of bank loan announcements in the case of a major European country. Using a sample of 253 loan announcements to French borrowers from January 2000 until December 2009 we have computed CAAR and CASAR for the whole period as well as for the boom and bust sub?periods. We have then investigated various loan, syndicate and borrower characteristics that could influence stock market reaction. Regarding the full sample results over the 2000?2009 timespan we found significant and negative stock market reaction to bank loan announcements. This first finding does not support the consensus of (positive) bank loan specialness first provided by James (1987) but rather more recent conclusions by Billett et al. (2006). In our case, bank loan announcements are actually perceived as bad news. However, we also document which loan, syndicate and borrower characteristics are associated with a positive reaction. We find that larger loans funded by numerous lenders of which a smaller proportion is local banks to large borrowers are related to a positive abnormal return. This series of results is more in line with previous literature (Mosebach 1999; Lee and Mullineaux 2004; Sufi 2007; Bosch and Steffen 2011). We then investigate the effect of the recent boom and bust cycle on stock market perception of bank loan announcements using both univariate and multivariate analysis. First of all we find that the average negative stock market reaction to bank loan announcements is essentially due to the loans provided during the financial crisis from 2007 to 2009, while the reaction is not significant for loans before the crisis. The latter result confirms that certification 25
value of bank loan announcements is reduced when lending standards are relaxed due to less informational asymmetries during a boom period. On the contrary, the former result doesn't confirm that the certification value of bank lending increases during bust periods when bank behavior is more conservative as a reaction to greater informational asymmetries. This last result is rather surprising as we uncover a significant change in bank lending behavior over the cycle, following notably recent evidence by de Haas and van Horen (2010). During the crisis period, loans have larger maturities, are more often secured and have covenants, and are funded by much smaller syndicates. These results clearly indicate a "wake? up call" effect of the crisis on bank screening and monitoring activities, with the reinforcement of contractual (loan) and organizational (syndicate) features aiming at mitigating adverse selection and moral hazard problems during a period of greater uncertainty and risk. Second, we look into the characteristics of loans, syndicates and borrowers that are related to positive and negative stock market reaction over the boom and bust cycle. Although empirical evidence tends to support a reinforcement of contractual and organizational means of screening and monitoring by banks during the crisis, we do not find support for these characteristics to be related to positive abnormal returns. However, we find that larger loan spreads and multi?tranches deals are associated with a positive market reaction during the crisis. Moreover, larger loans to larger borrowers funded by syndicates with fewer local lenders are also positively perceived by investors. We also uncover that a positive reaction during the crisis is associated with a borrower's lower common equity ratio. Finally, we also uncover investors' sensitivity to borrower's financial conditions, in particular their financial constraints during a bust period. Indeed, we find that the stock market perception is systematically positive for bank loan announcements to firms which are less financially constrained, especially with respect to the availability of lines of credit. This result suggests that even if lenders adjust loan terms and syndicate structure to reflect tougher
26
economic environment with greater information asymmetries, investors put more value on the certification effect of loans to financially unconstrained borrowers. Overall, our findings can be considered as questioning bank loans "specialness", especially in a period of crisis. However, several results also confirm established conclusions regarding the effects of such characteristics as loan and firm size or the structure of syndicates on stock market perception of bank loan announcements. We document a significant change in bank behavior over the economic cycle, with reactions in terms of loan and syndicate features to a crisis environment. We also uncover the signaling role of loan spreads and borrower financial structure as well as the economic advantages of loan tranching. However, more research needs to be done to better understand bank specialness in the current economic and financial environment.
27
References
Altunbas, Yener, Blaise Gadanecz, and Alper Kara, 2006. Syndicated Loans: A Hybrid of Relationship Lending and Publicly Traded Debt. Palgrave Macmillan. Bae, Kee?Hong, Jun?Koo Kang, and Chan?Woo Lim, 2002. The value of durable ba nk relationships: evidence from Korean banking shocks. Journal of Financial Economics 64(2), 181-214. Bailey, Warren, Wei Huang, and Zhishu Yang, 2012. Bank Loans with Chinese Characteristic s: Some Evidence on Inside Debt in a State?Controlled Banking System. Journal of Quantitative Fin ancial Analysis 46(6), 1795?1830. Berger, Allen N., Leora F. Klapper, and Gregory F. Udell, 2001. The ability of banks to lend to informationally opaque small businesses. Journal of Banking & Finance 25(12), 2127-67. Berger, Allen N., and Gregory F. Udell, 1994. Did Risk?Based Capital Allocate Bank Credit a nd Cause a 'Credit Crunch' in the United States? Journal of Money, Credit and Banking 26(3), 585-628. Bernanke, Ben S., Cara S. Lown, and Benjamin M. Friedman, 1991. The Credit Crunch. Brookings Papers on Economic Activity 1991(2), 205-47. Best, Ronald, and Hang Zhang, 1993. Alternative Information Sources and the Informatio n Content of Bank Loans. Journal of Finance 48(4), 1507-22. Billett, Matthew T., Mark J. Flannery, and Jon A. Garfinkel, 1995. The Effect of Lender Identity on a Borrowing Firm's Equity Return. Journal of Finance 50(2), 699-718. ——— 2006. Are Bank Loans Special? Evidence on the Post?Announcement Performance of Bank Borrowers. Journal of Financial and Quantitative Analysis 41(4), 733-51. Boehmer, Ekkehart, Jim Masumeci, and Annette B. Poulsen, 1991. Event?study methodolog y under conditions of event?induced variance. Journal of Financial Economics 30(2), 253-72. Bosch, Oliver, and Sascha Steffen, 2011. On syndicate composition, corporate structure and the certification effect of credit ratings. Journal of Banking & Finance 35(2), 290-99. Chava, Sudheer, and Amiyatosh Purnanandam, 2011. The effect of banking crisis on ba nk? dependent borrowers. Journal of Financial Economics 99(1), 116-35. De Haas, Ralph, and Neeltje van Horen, 2010. The crisis as a wake?up call. Do banks tigh ten screening and monitoring during a financial crisis? Netherlands Central Bank, Research Department. Dell'Ariccia, Giovanni, Deniz Igan, and Luc Laeven, 2008. Credit Booms and Lending Standards: Evidence From The Subprime Mortgage Market. CEPR Discussion Papers. Dell'Ariccia, Giovanni, and Robert Marquez, 2006. Lending Booms and Lending Standards. Journal of Finance 61(5), 2511-46. Demyanyk, Yuliya, and Otto Van Hemert, 2011. Understanding the Subprime Mortgage Crisis. Review of Financial Studies 24(6), 1848 -1880. Diamond, Douglas W., 1984. Financial Intermediation and Delegated Monitoring. Review of Economic Studies 51(3), 393-414. ——— 1991. Monitoring and Reputation: The Choice between Bank Loans and Directly Placed Debt. Journal of Political Economy 99(4), 689-721. Duchin, Ran, Oguzhan Ozbas, and Berk A. Sensoy, 2010. Costly external finance, corpora te investment, and the subprime mortgage credit crisis. Journal of Financial Economics 97(3), 418-35. Fama, Eugene F. 1985. What's different about banks? Journal of Monetary Economics 15(1), 29-39. Fields, L. Paige, Donald R. Fraser, Tammy L. Berry, and Steven Byers, 2006. Do Bank Lo an Relationships Still Matter? Journal of Money, Credit, and Banking 38, 1195-1209. Focarelli, Dario, Alberto Franco Pozzolo, and Luca Casolaro, 2008. The pricing effect of certification on syndicated loans. Journal of Monetary Economics 55(2), 335-49. Fungá?ová, Zuzana, Christophe J. Godlewski, and Laurent Weill, 2011. Asymmetric Information
and Loan Spreads in Russia. Eastern European Economics 49(1), 13-29. Furfine, Craig, 2001. Bank Portfolio Allocation?: The Impact of Capital Requirements , Regulatory Monitoring , and Economic Conditions. Journal of Financial Services Research 20(1), 33-56. Gambacorta, Leonardo, and Paolo Emilio Mistrulli, 2004. Does bank capital affect lendin g behavior? Journal of Financial Intermediation 13(4), 436-57. Gatev, Evan, and Philip E. Strahan, 2009. Liquidity risk and syndicate structure. Journal of Financial Economics 93(3), 490-504.
28
Godlewski, Christophe J., Bulat Sanditov, and Thierry Burger?Helmchen, 2012. Bank Lendin g Networks, Experience, Reputation, and Borrowing Costs: Empirical Evidence from the French Syndicated Lending Market. Journal of Business Finance & Accounting 39(1?2), 113-140. Godlewski, Christophe J., Zuzana Fungá?ová, and Laurent Weill, 2011. Stock Market Reaction to Debt Financing Arrangements in Russia. Comparative Economic Studies 53, 679?693. Hancock, Diana, Andrew J. Laing, and James A. Wilcox, 1995. Bank capital shocks: Dynami c effects on securities, loans, and capital. Journal of Banking & Finance 19(3?4), 661-77. Huang, Weihua, Armin Schwienbacher, and Shan Zhao, 2012. When Bank Loans Are Bad News: Evi dence From Market Reactions to Loan Announcements Under the Risk of Expropriation. Journal of International Financial Markets, Institutions & Money 22(2), 233?252. Ivashina, Victoria, and David Scharfstein, 2010. Bank lending during the financial crisis of 2008. Journal of Financial Economics 97(3), 319-38. James, Christopher, 1987. Some evidence on the uniqueness of bank loans. Journal of Financial Economics 19(2), 217-35. Jensen, Michael C., 1986. Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers. American Economic Review 76(2), 323-29. Jensen, Michael C., and William H. Meckling, 1976. Theory of the firm: Managerial behavi or, agency costs and ownership structure. Journal of Financial Economics 3(4), 305-60. Keys, Benjamin J., Tanmoy Mukherjee, Amit Seru, and Vikrant Vig, 2010. Did Securitization Lead to Lax Screening? Evidence from Subprime Loans. Quarterly Journal of Economics 125(1), 307 -362. Lee, S.W., and D.J. Mullineaux, 2004. Monitoring, Financial Distress, and the Structure of Commercial Lending Syndicates. Financial Management 33, 107-30. Leland, Hayne E., and David H. Pyle, 1977. Informational Asymmetries, Financial Structure, and Financial Intermediation. Journal of Finance 32(2), 371-87. Lummer, Scott L., and John J. McConnell, 1989. Further evidence on the bank lending process and the capital?market response to bank loan agreements. Journal of Financial Economics 25(1), 99-122. Maskara, Pankaj K., 2010. Economic value in tranching of syndicated loans. Journal of Banking & Finance 34(5), 946-55. Maskara, Pankaj K., and Donald J. Mullineaux, 2011. Information asymmetry and self?selection bias in bank loan announcement studies. Journal of Financial Economics 101(3), 684-94. Mian, Atif, and Amir Sufi, 2009. The Consequences of Mortgage Credit Expansion: Eviden ce from the U.S. Mortgage Default Crisis. Quarterly Journal of Economics 124(4), 1449 -1496. Mosebach, Micheal, 1999. Market response to banks granting lines of credit. Journal of Banking & Finance 23(11), 1707-23. Ongena, Steven, David C Smith, and Dag Michalsen, 2003. Firms and their distressed bank s: lessons from the Norwegian banking crisis. Journal of Financial Economics 67(1), 81-112. Patell, James M., 1976. Corporate Forecasts of Earnings Per Share and Stock Price Behavi or: Empirical Test. Journal of Accounting Research 14(2), 246-76. Peek, Joe, and Eric Rosengren, 1995. Bank regulation and the credit crunch. Journal of Banking & Finance 19(3?4), 679-92. Preece, Dianna, and Donald J. Mullineaux, 1996. Monitoring, loan renegotiability, and fir m value: The role of lending syndicates. Journal of Banking & Finance 20(3), 577-93. Purnanandam, Amiyatosh, 2011. Originate?to?distribute Model and the Subprime Mortgag e Crisis. Review of Financial Studies 24(6), 1881 -1915. Ross, Stephen A., 1977. The Determination of Financial Structure: The Incentive?Signallin g Approach. Bell Journal of Economics 8(1), 23-40. Santos, João A. C., 2011. Bank Corporate Loan Pricing Following the Subprime Crisis. Review of Financial Studies 24(6), 1916 -1943. Slovin, Myron B., Shane A. Johnson, and John L. Glascock, 1992. Firm size and the information content of bank loan announcements. Journal of Banking & Finance 16(6), 1057-71. Sufi, A., 2007. Information Asymmetry and Financing Arrangements: Evidence from Syndicated Loans. Journal of Finance 62, 629-68.
Thomson Reuters, 2011. Global Syndicated Loans Review. Thomson Reuters. Wagster, John D., 1999. The Basle Accord of 1988 and the international credit crunch of 1989? 1992. Journal of Financial Services Research 15(2), 123-43.
29
Figure 1 Worldwide syndicated loans amounts and issues
This figure displays the evolution of the yearly loan amounts (left scale) and number of issues (ri ght scale) on the global syndicated lending market (source: Thomson Reuters).
30
0,8 0,6 0,4 0,2 CA R 0 ?10 ?9 ?8 ?7 ?6 ?5 ?4 ? 3 ?2 ? 1 ?0,2 ?0,4 ?0,6 0 1 2 3 4 5 6 7 8 9 10
Event time CAR CAR Crisis CAR Pre crisis
Figure 2 CAR around the event date
This figure displays the CARs over an event window of 21 days around the event date. The red li ne represents the CARs computed over the full sample period. The blue line represents the CARs compute d over the Crisis period. The green line represents the CARs computed over the Pre crisis period.
31
Table 1. Definitions and descriptive statistics for loan, bank syndicate and borrower characteristics
This table displays means and standard deviations for main loan, bank syndicate and borrower characteristics. Sample pe riod is 2000 until 2009. The number of observations varies because of data availability for particular variables. Data source: Bloomberg Professional Terminal Server. Variable Definition N Mean Std dev. Facility Loan facility amount in MLN USD 253 674.121 1 349.495 Spread Loan spread in bps 135 128.801 103.780 Maturity Loan maturity in years 253 2.921 4.703 Tranches Number of tranches in a loan facility 253 2.419 2.315 Term loan Dummy equal to 1 if the loan is a term loan, 0 otherwise 253 0.509 0.500 dummy Revolving loan Dummy equal to 1 if the loan is a revolving loan, 0 otherwise 253 0.387 0.488 dummy Secured dummy Dummy equal to 1 if the loan is secured, 0 otherwise 253 0.217 0.413 Covenants Dummy equal to 1 if the loan has covenants, 0 otherwise 253 0.233 0.423 dummy Lenders Number of lenders in the syndicate 253 8.565 7.945 Arrangers Number of arrangers in the syndicate 233 6.733 6.260 Local lenders Percentage of local (French) lenders in the syndicate 220 66.235 26.646 Local arrangers Percentage of local (French) arrangers in the syndicate 209 64.844 27.970 League table Percentage of lenders in the syndicate that are listed on the 230 66.190 23.090 lenders Bloomberg League table League table Percentage of arrangers in the syndicate that are listed on 214 66.801 23.934 arrangers the Bloomberg League table Assets Total assets in MLN USD 200 40 045.630 198 298.210 Sales Total Sales in MLN USD 206 7 990.400 17 459.120 Debt ratio Total debt / Total assets 200 34.082 21.854 Equity ratio Common equity / Total assets 200 28.322 24.433 Interest Ebitda / Total interest expenses 176 12.113 21.530 coverage ratio Current ratio (Cash + Accounts receivable + Short term investments + 176 134.936 89.671 Inventories) / Current liabilities Profit margin Net income / Total sales 206 9.461 25.981 Return on assets Net income / Total assets 199 3.750 5.825
32
Table 2. Stock market reactions to bank loan announcements
This table displays CAAR (cumulative average abnormal return) and CASAR (cumulative average abnormal standardized return) for the selected seven event windows over the entire sample time span (January 2000 until December 2009) and by boom (Pre?crisis) and bust (Crisis) period. Crisis starts from September 2007. The number of ban k loan announcement events is 253 (147 during the Pre?crisis period and 106 during the Crisis period). ***, ** and * indic ate CAAR and CASAR statistically different from 0 at the 1%, 5% and 10% confidence level according to Student tests. In th e last two columns, ** and * indicate a statistically significant difference in means at the 5% and 10% confidence level for the CAAR and CASAR between the Crisis and Pre?crisis periods. Data source: Bloomberg Professional Terminal Server. Crisis Pre?crisis Event CAAR CASAR Percent t?test t?test window of for for CAAR CASAR CAAR CASAR positive CAAR CASAR CAAR [0,0] ?0.4291* ?0.1558 ?0.2066 0.75 0.83 ?0.3001** ?0.0961 0.3872 [?1,1] [?2,2] [?2,0] [?1,0] [0,1] [0,2] ?0.3068* ?0.3062 ?0.2570 ?0.0938** ?0.0767 ?0.0397 0.4812 ?0.6458*** 0.4098 0.4549 ?0.3676 ?0.0033 ?0.2162*** ?0.0860 0.0319 0.0310 ?0.3092 ?0.5626 0.0032 ?0.0733 ?0.1042 2.29** 0.11 ?0.89 2.49** 0.09 ?0.91 ?0.3121 ?0.2622 ?0.2996 ?0.0828 ?0.0456 ?0.0726* 0.4587 0.4812 0.4662 ?0.5909 ?0.2228 ?0.6025** ?0.1589* ?0.0563 ?0.1814*** ?0.0561 ?0.2906 0.0118 ?0.0500 ?0.0237 ?0.0397 0.0127 1.07 ?0.11 1.74* 1.19 0.15 2.22**
33
Table 3. Loan, bank syndicate and borrower characteristics by stock market reaction
This table displays means and standard deviations for main loan, bank syndicate and borro wer characteristics by positive and negative CAAR for the [?1,0] event window (122 and 131 events respectively) and the results of t?tests or chi?2 tests for the means. The latter is used for binomial t est of proportion (dummy variables) while the former is used for Student tests of means (c ontinuous variables). All explanatory variables are defined as in previous tables. ***, ** and * indicate a statistically significant difference in means at the 1%, 5% and 10% confidence level for the relevant variables. Sample period is 2000 until 2009. Data source: Bloomberg Professional Terminal Server. Positive CAAR [?1,0] Negative CAAR [?1,0] Variable Mean Std dev. Mean Std dev. t?test or chi?2 test Facility 900.521 1 804.383 459.911 619.135 2.57** Spread 129.278 109.380 128.345 98.929 0.05 Maturity 2.861 0.2 Tranches 0.501 0.369 0.235 0.401 Lenders Arrangers 9.764 7.383 26.202 70.842 65.030 22.672 Assets Sales 70 860.070 11 849.140 0.95 Equity ratio 16.644 1.150 33.962 4.188 4.096 2.552 2.978 2.615 0.487 5.229 2.292 0.501 1.994 0.530 0.493
0.89 Term loan dummy 0.484 0.426 1.64 9.090 7.316
0.46 Revolving loan dummy 0.406 0.37 Secured dummy 0.200 0.268 7.430 6.132 60.708 0.401 0.444 6.520 5.049 26.104 58.306 23.518 25.334 29 199.330 9 796.820 35.493 29.849 26.075 0.502 11.533 3.290
0.47 Covenants dummy
0.200 2.33** 71.469 27.724
1.51 Local lenders 26.948 22.670 0.05 280 611.410 22 575.270 26.700
3.05*** Local arrangers 67.271 66.716
3.31*** League table lenders 66.880 2.09** 25.924 30.627 10.922 1.392 12.474 6.091
0.73 League table arrangers
11 026.210 4 488.950 16.479 15.353 13.479
2.98*** Debt ratio 32.585 0.93 Interest coverage ratio
0.76 Current ratio 1.303 0.67 Profit margin 6.140 1.83* Return on assets 5.555 1.09
34
Table 4. Loan, bank syndicate and borrower characteristics (crisis vs. pre?crisis period)
This table displays means and standard deviations for main loan, bank syndicate and borrow er characteristics by Crisis and Pre?crisis period (106 and 147 events respectively) and the results of t?tests or chi?2 tests for the means. The latter is used for binomial test of proportion (dummy variables) while the former is used for Student tests of means (continuous variables). All explanatory variables are defined as i n previous tables. ***, ** and * indicate a statistically significant difference in means at the 1%, 5% and 10% confidence level for the relevant variables. Data source: Bloomberg Professional Terminal Server. Crisis Pre?crisis Variable Mean Std dev. Mean Std dev. t?test or chi?2 test Facility 731.750 1 717.186 632.565 1 009.220 ?0.53 Spread 139.690 114.370 123.884 98.890 ?0.82 Maturity 4.844 6.486 1.535 1.820 ?5.11*** Tranches 2.566 1.809 ?0.80 Term loan dummy 0.476 0.473 0.501 0.429 2.879 0.557 2.313 0.499
1.59 Revolving loan dummy 0.330 0.497 2.51 Secured dummy 0.163 0.311 9.830 6.880 69.475 0.371 0.465 9.148 6.903 26.307 67.008 23.657 23.273 65.101 64.035 27.320 0.177 3.27***
0.292 0.457 6.04** Covenants dummy 0.383 Lenders Arrangers 6.811 6.505 26.750 63.418 68.121 24.289 Assets 6.22** 5.460 5.128
0.47 Local lenders 28.408 22.187 ?1.31
?1.49 Local arrangers 64.926
?0.91 League table lenders
?1.03 League table arrangers 69.435 59 408.280 275 540.020 ?1.03 Sales 7 209.810 35.768 17.665 ?0.97 Equity ratio Interest coverage ratio Current ratio Profit margin 29.300 11.233 1.415 13.153 5.870 ?1.46 20.106 1.249 31.168 ?0.89
Debt ratio
26 309.740 115 456.080 9 016.570 21 949.030 13 107.110 ?0.69 32.887 24.399 29.262 12.738 1.305 6.653 4.184 15.260 22.562 0.547 20.918 5.770 3.441 27.656 0.46 ?0.69
?1.70* Return on assets
35
Table 5. Loan, bank syndicate and borrower characteristics by stock market reaction (crisis vs. pre?crisis period)
This table displays means and standard deviations for main loan, bank syndicate and borrower characteristics by positive and negative CAAR for the [?1,0] event window and the results of t?t ests or chi?2 tests for the means over Crisis and Pre?crisis periods. The chi?2 test is used for binomial test of proportion (dummy variables) while the t?test is used for Student tests of means (continuous variables). All explanatory variables are defined as in previous tables. ***, ** and * indicate a statistically significant difference in means at the 1%, 5% and 10% confidence level for the relevant variables. Data source: Bloomberg Professional Terminal Server. Crisis Pre?crisis Positive CAAR [?1,0] Variable Facility Spread Mean Std dev. Negative CAAR [?1,0] Mean Std dev. t?test or chi?2 test ?1.80* ?2.09** 7.142 2.081 0.509 0.466 Positive CAAR [?1,0] Mean Std dev. Negative CAAR [?1,0] Mean 478.550 138.900 1.636 1.698 0.414 1.39 0.191 0.157 7.958 5.814 59.127 2.03** 22.538 68.102 13 227.300 4 904.900 33.988 28.405 20.765 1.537 12.197 3.202 Std dev. 716.470 99.322 1.538 2.486 0.497 0.500 0.395 0.366 7.173 4.935 27.380 57.364 0.47 23.588 37 714.200 11 462.800 30.447 15.284 0.07 1.612 19.403 5.573 t?test or chi?2 test ?1.85* 2.006 1.914 0.517 0.504 0.11 1.39 ?2.49** 69.496 28.572 64.043 ?0.61 ?1.22 ?1.88* 26.868 14.888 1.04 38.812 4.974
Secured dummy Covenants dummy Lenders Arrangers
Assets Sales Debt ratio
1 088.200 2 464.800 436.780 47.660 780.420 1 213.200 176.900 123.100 105.900 96.461 108.600 97.132 1.49 Maturity 4.937 5.668 4.768 ?0.14 1.534 0.01 Tranches 3.188 3.541 2.052 ?1.96* 2.147 1.14 Term loan dummy 0.612 0.492 0.504 0.80 0.503 2.42 Revolving loan dummy 0.306 0.351 0.482 0.382 0.489 2.62 0.245 0.435 0.333 0.476 0.70 0.121 0.329 0.286 0.456 0.333 0.476 0.75 0.207 0.409 6.854 5.348 6.776 5.598 ?0.07 11.627 9.224 6.425 5.073 6.569 5.220 0.13 7.917 8.292 ?1.84* Local lenders 63.579 23.693 73.868 27.505 1.85* 25.135 2.25** Local arrangers 60.056 26.395 72.078 27.129 69.870 26.991 2.52** League table lenders 66.853 21.937 69.072 23.150 65.875 24.328 0.45 League table arrangers 71.211 23.056 64.302 26.327 65.950 22.096 0.39 122 950.000 406 523.000 8 298.700 12 197.300 ?1.71* 38 738.000 156 639.000 15 717.000 31 420.000 4 024.100 7 603.400 ?2.25** 9 399.500 14 249.400 33.790 15.792 37.359 19.061 0.91 31.842 16.978 0.47 Equity ratio 23.938 15.269 33.545 13.991 2.99*** 39.128 ?0.28 Interest coverage ratio 11.039 19.472 11.369 29.305 10.546 12.361 ?0.98 Current ratio 1.257 0.464 1.330 0.525 1.279 0.572 ?0.47 Profit margin 4.766 1.53 7.011 11.109 6.276 27.859 ?0.19 Return on assets 5.864 1.4 3.345 6.437 3.543 5.253 0.18
36
Table 6. Regressions of CAR on main loan, bank syndicate and borrower characteristics
This table displays the results of OLS regressions with robust standard errors (in parentheses) where the dependent variable is the CAR [?1,0]. Crisis is a dummy variable equal to 1 if the bank loan announ cement occurs between September 2008 and December 2009, 0 otherwise. All explanatory variables are defined as in previous tables. Dummies for industrial sectors included but not shown. ***, ** and * indicate a statistically significant coefficient at the 1%, 5% and 10% confidence level. Data source: Bloomberg Professional Terminal Server. Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Crisis Crisis x log(Facility) Crisis x log(Maturity) Crisis x log(Lenders) Crisis x log(Tranches) Crisis x Term loan dummy Crisis x Secured dummy Crisis x Covenants dummy Crisis x log(Sales) Crisis x Equity ratio Crisis x Operating margin ?0.6759** (0.3073) ?0.0330** (0.0154) ?0.3624 (0.2795) ?0.2760* (0.1450) ?0.0215 (0.2795) ?0.2902 (0.4454) ?0.2038 (0.5999) ?0.5165 (0.4991) ?0.1114*** (0.0420) ?3.1248*** (1.0638) ?1.9462 (1.2738) ?1.0108*** (0.3456)
37
Table 6. continued
Variables log(Facility) log(Maturity) log(Lenders) log(Tranches) Term loan dummy Secured dummy Covenants dummy log(Sales) Equity ratio Operating margin Intercept N R²
(1) 0.2979** (0.1282) ?0.0310 (0.2148) ?0.1200 (0.2033) 0.0996 (0.2310) 0.2199 (0.3947) ?0.0130 (0.4679) 0.0962 (0.3294)
(2) 0.3150** (0.1293) ?0.0408 (0.2160) ?0.1205 (0.2035) 0.1137 (0.2309) 0.2150 (0.3952) ?0.0198 (0.4679) 0.0994 (0.3299)
(3) 0.2880** (0.1292) 0.0139 (0.2021) ?0.1255 (0.2012) 0.1077 (0.2312) 0.2656 (0.4058) ?0.0450 (0.4624) 0.0333 (0.3352)
(4) 0.2859** (0.1279) ?0.0886 (0.2161) 0.0380 (0.1989) 0.1559 (0.2313) 0.1866 (0.3977) ?0.0102 (0.4699) 0.0941 (0.3324)
(5) 0.2975** (0.1304) ?0.1517 (0.2174) ?0.0671 (0.2056) 0.1541 (0.2597) 0.2467 (0.4089) ?0.0329 (0.4663) 0.0559 (0.3396)
(6) 0.2994** (0.1296) ?0.1218 (0.2200) ?0.0793 (0.2051) 0.1164 (0.2300) 0.4059 (0.4517) ?0.0259 (0.4651) 0.0677 (0.3355)
(7) 0.2951** (0.1291) ?0.1518 (0.2127) ?0.0672 (0.2059) 0.1378 (0.2267) 0.2464 (0.4101) 0.0897 (0.5180) 0.0526 (0.3357)
(8) 0.3029** (0.1292) ?0.1409 (0.2133) ?0.0774 (0.2052) 0.1501 (0.2291) 0.2604 (0.4047) ?0.0492 (0.4718) 0.3606 (0.3851)
(9) 0.1195 (0.1534) ?0.0221 (0.2402) ?0.1247 (0.2324) 0.0402 (0.2657) 0.1829 (0.4237) 0.1451 (0.5018) 0.2013 (0.3781) 0.2742*** (0.0922) ?0.0502 (0.4466) ?0.3693 (0.7971) ?4.1374 (2.7696) 152 0.1372 1.4954
(10) 0.1195 (0.1530) ?0.0616 (0.2456) ?0.1137 (0.2344) 0.0502 (0.2656) 0.1967 (0.4288) 0.1272 (0.5025) 0.2062 (0.3790) 0.3266*** (0.0921) ?0.0936 (0.4430) ?0.4476 (0.7892) ?4.6000 (2.8178) 152 0.1257 1.7212
(11) 0.1180 (0.1507) ?0.0237 (0.2392) ?0.1011 (0.2229) 0.0613 (0.2654) 0.0675 (0.4152) 0.0895 (0.5149) 0.2413 (0.3689) 0.2267** (0.0933) 0.4249 (0.3057) ?0.2385 (0.8022) ?3.9515 (2.7226) 152 0.1464 2.4899
(12) 0.1298 (0.1474) ?0.1723 (0.2484) 0.0123 (0.2311) ?0.0075 (0.2532) 0.2630 (0.4410) 0.2073 (0.4811) 0.1917 (0.3880) 0.2394*** (0.0877) ?0.0869 (0.4252) 0.8072 (0.6532) ?4.7261* (2.7604) 152 2.4585
?5.1849** (2.4616) 195 0.0959 2.8214
?5.5395** (2.4860) 195 0.0943 2.0527
?5.1431** (2.4543) 195 0.0869 1.8463
?5.3442** (2.4784) 195 0.0899 1.8491
?5.5182** (2.5405) 195 0.0765 1.5041
?5.5458** (2.5129) 195 0.0785 1.8876
?5.4655** (2.5016) 195 0.0770 1.4944
?5.6365** (2.5046) 195 0.0802 1.5009
0.1086 F?stat.
38
Table 7. Financial constraints variables by stock market reaction (crisis vs. pre?crisis period)
This table displays means and standard deviations for six variables related to borrower's financial constraints by positive and negative CAAR for the [?1,0] event window and the results of t?tests or W ilcoxon / Kruskall?Wallis tests for the means over Crisis and Pre?crisis periods. The non?parametric Wilcoxon / Kruskall?Wallis tests are used for the Total lines of credit and Total available lines of cre dit variables as they have the least observations. The t?test is used for Student tests of means for all other variables. ***, ** and * indicate a statistically significant difference in means at the 1%, 5% and 10% confidence level for the relevant variables. Data source: Bloomberg Professional Terminal Server. Crisis Pre?crisis Positive CAAR (?1,0) Variable Retained earnings Free cash flow Short term borrowings Mean 1 416,513 Std dev. 6 246,156 776,106 21 239,490 16 057,010 66 122,470 ?1,43 Long term borrowings 1 749,968 Total lines of credit Total available lines of credit 3 084,038 5 276,217 4 331,588 14 069,640 Negative CAAR (?1,0) Mean ?14 098,780 127,891 533,339 4 617,426 997,369 911,257 Std dev. 101 592,700 741,922 844,621 ?1,68* 1 961,061 1 859,941 ?2.10** / 4.42** ?0.72 / 0.52 2 994,070 1 555,670 3 108,824 1 592,442 253,688 469,572 239,762 508,919 ?3.38*** / 11.47*** ?2.09** / 4.38** t?test or W. / K?W test ?0,97 ?1,21 ?1,39 2 851,609 Positive CAAR (?1,0) Mean 1 693,006 216,056 8 576,254 4 264,132 Std dev. 4 558,637 3 291,180 39 552,460 ?1,18 Negative CAAR (?1,0) Mean 1 267,558 358,286 1 118,950 5 901,872 Std dev. 6 060,784 890,314 4 443,141 18 536,490 t?test or W. / K?W test ?0,43 0,32
28 733,180 133 740,200
39
doc_189949881.docx