Directory UMM :Data Elmu:jurnal:J-a:Journal Of Banking And Finance:Vol25.Issue2.2001:
Journal of Banking & Finance 25 (2001) 431±444
www.elsevier.com/locate/econbase
A note on: Capital adequacy and the
information content of term loans and lines of
credit
P. Andre
a,1
, R. Mathieu
b,*
, P. Zhang
c,2
a
Ecole
des Hautes Etudes
Commerciales, 3000 Chemin de la C^
ote-Sainte-Catherine, Montr
eal, Que.,
Canada H3T 2A7
b
School of Business and Economics, Wilfrid Laurier University, Waterloo, Ont., Canada N2L 3C5
c
School of Accountancy, University of Waterloo, Waterloo, Ont., Canada N2L 3G1
Received 2 October 1998; accepted 25 October 1999
Abstract
This study examines the information content conveyed by the disclosure of credit
agreements in a Canadian setting. We argue that the introduction of the 1988-capital
adequacy requirements lead banks to reduce their level of commitment at the issuance of
lines of credit to avoid their inclusion in the calculation of the capital ratio. As a result,
after 1988, the disclosure of lines of credit is expected to be less informative than the
disclosure of term loans since banks may exert less eort to screen and monitor ®rms. Our
results are consistent with the argument that the dierence between the market reactions
at the disclosure of term loans and lines of credit is signi®cant after 1988. We also provide
evidence that ®rm size and concentration of borrowing aect the market reaction at the
disclosure of bank credit agreements. Ó 2001 Elsevier Science B.V. All rights reserved.
JEL classi®cation: G21
Keywords: Information of bank credit agreements; Capital adequacy
*
Corresponding author. Tel.: +1-519-884-1970, ext. 3142; fax: +1-519-884-0201.
E-mail addresses: paul.andre@hec.ca (P. AndreÂ), rmathieu@wlu.ca (R.
pzhang@uwaterloo.ca (P. Zhang).
1
Tel.: +514-340-6528; fax: +514-340-5633.
2
Tel.: +519-888-4567, ext. 6548; fax: +519-888-7562.
0378-4266/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 4 2 6 6 ( 9 9 ) 0 0 1 3 0 - 2
Mathieu),
432
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
1. Introduction
Bank credit agreements (term loans and lines of credit) provide ®rms with
important access to capital. These agreements, when disclosed, can provide a
signal to the market of ®rmsÕ ability to raise capital and banksÕ assessment of
®rmsÕ value (Fama, 1985; James, 1987; James and Wier, 1990; Slovin and
Young, 1990; Lummer and McConnell, 1989). The empirical evidence indicates
that the characteristics of both banks and ®rms can aect the degree of informativeness of bank credit agreements (Johnson, 1996; Preece and Mullineaux, 1996; Petersen and Rajan, 1994). The empirical ®ndings also suggest
that the informativeness of bank credit agreements depends on how well ®rms
are monitored (Slovin et al., 1992; Best and Zhang, 1993).
In this paper, using a sample of Canadian ®rms, we examine the relationship
between the informativeness of credit agreements and banksÕ level of commitment. BanksÕ level of commitment at the issuance of a credit agreement is
aected by several factors such as time to maturity, the collateral oered on the
loan, and their ability to unconditionally cancel the credit agreement. If banks
reduce their level of commitment when providing ®nancing, their risk exposure
would be reduced which may lead them to exert lower eort in screening and
monitoring ®rms. In such a case, we expect to observe a lower market reaction
at the disclosure of bank credit agreements.
In 1988, Canadian authorities changed the capital adequacy requirements to
adopt the recommendations of the Bank of International Settlements. We argue that the introduction of the 1988-capital ratio results in a reduction of
banksÕ level of commitment at the issuance of lines of credit so that they can
avoid some costs associated with the issuance of o-balance sheet instruments.
Therefore, by comparing the information content conveyed by the disclosure of
term loans and lines of credit before and after 1988, we can test the impact of
banksÕ level of commitment on the market reaction at the disclosure of bank
credit agreements. 3
Our sample consists of 122 announcements of new and revised bank credit
agreements in the Canadian market during the period 1982±1995. Our results
indicate that the information content conveyed by the disclosure of lines of
credit diers before and after 1988 for small ®rms. The information content
conveyed by lines of credit and term loans for small ®rms diers after 1988, but
not before. Furthermore, the information content conveyed by term loans
3
The conditions on the credit agreements can be used as proxies for bankÕs level of commitment.
Those conditions include, among others, time to maturity, the collateral oered on the loan, and
their ability to unconditionally cancel the credit agreement. However, in our sample of
announcements, information with respect to the time to maturity and the collateral oered on
the loan are not always disclosed, and information on banksÕ ability to unconditionally cancel is not
available.
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
433
before and after 1988 is the same. We view these results as evidence that the
introduction of the 1988-capital adequacy requirements has reduced the informativeness of lines of credit.
We also document in this paper signi®cant average announcement excess
returns when bank credit agreements are new, when they are provided to small
®rms, when they are issued by single banks or when renewals are favorable.
The remainder of the paper is structured as follows. Section 2 describes the
1988-capital adequacy requirements and presents our predictions. Data collection and methodology are discussed in Section 3. Section 4 presents the
empirical results and Section 5 summarizes the paper.
2. Capital adequacy requirements and the information of bank credit agreements
To maintain public con®dence in the banking system, the Canadian government insures deposits up to $60,000. To limit the level of risk chosen by
banks, the government imposes capital adequacy measures that link the capital
issued by banks to their investing activities. Prior to 1988, the capital adequacy
guidelines suggested that banks maintain 1 dollar of capital for every 30 dollars
of assets. However, this basic ratio did not take into account banksÕ exposure
to risk and included only on-balance sheet activities.
In 1988, Canadian authorities changed the capital adequacy requirements to
comply with the recommendations of the Bank of International Settlement,
referred to as the Basle Accord. The Basle Accord introduces a common definition of capital, uses a weighting system to calculate the minimum capital
requirement, and takes into consideration o-balance sheet activities.
Before the introduction of the Basle Accord, the used portion of a line of
credit had no impact on the required minimum level of capital. However, after
1988, lines of credit may be included in the calculation of the capital ratio. Like
all o-balance sheet instruments, the inclusion of lines of credit follows a twostep procedure. First, a credit conversion factor (i.e., a weight) is applied to
obtain the asset equivalents. Second, the weight applied to similar assets is used
to convert these asset equivalents into risk-adjusted o-balance sheet activities.
The guidelines published by the Superintendent of ®nancial institutions
de®ne the weights used for the conversion of o-balance sheet instruments.
With respect to lines of credit, the following weights are used to obtain the
asset equivalents:
· 50%: commitments with an original maturity exceeding 1 year, including underwriting commitments and commercial credit lines.
· 0%: commitments with an original maturity of 1 year or less or that are unconditionally cancelable at any time without prior notice.
Banks can avoid the impact of lines of credit on the calculation of capital
adequacy requirements by either reducing the length of the credit agreements
434
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
or introducing a clause that allows them to unconditionally cancel the lines of
credit at any time without prior notice. Discussions with bankers revealed that
most Canadian banks have changed the basic conditions associated with lines
of credit to avoid their inclusion in the calculation of the capital ratio. The
consequence of these changes in the basic conditions of lines of credit is to
lower the commitment of banks to their clients. As a result, we expect to observe a weaker market reaction at their disclosure after 1988 and the following
predictions are developed accordingly:
Prediction 1: The information content conveyed by the disclosure of lines
of credit and term loans prior to 1988 does not dier.
Prediction 2: The information content conveyed by the disclosure of lines of
credit and term loans after 1988 diers.
Prediction 3: The information content conveyed by the disclosure of lines
of credit before and after 1988 diers.
Prediction 4: The information content conveyed by the disclosure of term
loans before and after 1988 does not dier.
3. Data and methodology
The Globe & Mail on CD-ROM and the database of Canadian Business and
Current Aairs (CBCA) are used to obtain the announcements of Canadian
bank credit agreements in the Globe & Mail over the period 1982±1995. 4; 5
Only ®rms with stock prices on the TSE Western daily ®le are included in the
sample, and we obtain a total of 150 announcements. From this total, 28
observations are discarded since there is other information in the announcements about the ®rms. The ®nal sample consists of 122 announcements. Descriptive statistics of the sample are provided in Table 1.
To analyze the information content of bank credit agreements, we examine
the relationship between changes in ®rmsÕ market value at the announcement
of bank credit agreements using an event study methodology. As in Brown and
Warner (1980), James (1987), and Lummer and McConnell (1989), the excess
returns are calculated using the market model, and the two-day event window
4
The Globe & Mail is a daily newspaper that specializes in economic issues and is the Canadian
equivalent of the Wall Street Journal. During our sample period, it was the only nation-wide
newspaper.
5
Given that some parent companies are not Canadian corporations, there is a possibility that the
announcements in Canada lag the announcements in foreign countries. We conduct a search of the
Wall Street Journal using our entire sample and ®nd no evidence that such a lag exists.
435
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
Table 1
Descriptive statistics
Full sample
Number of
observations 122
(A) Sample composition
New credit agreements
Revised credit agreements
New and revised credit agreements
Lines of credit
Term loans
Lines of credit and term loans
91
27
4
46
64
12
(B) Loan amount (in millions of dollars)
Number of
Mean
observationsa
Full sample
New agreements
Revised agreements
New and revised agreements
Lines of credit
Term loans
Lines of credit and term
loans
Minimum
Maximum
111
85
23
3
317
337
259
1978
140
123
211
246
0.5
0.5
5
22
2520
2520
1400
326
44
56
11
352
279
370
161
106
200
5
0.5
78
2250
2520
2200
(C) Single versus multiple banks
Single bank
Full sample
Lines of credit
Term loans
Median
38
13
24
Multiple
banks
72
30
32
Unknown
12
3
8
(D) Canadian versus non-Canadian banks
Canadian
Nonbanksb
Canadian
banks
Full sample
Lines of credit
Term loans
a
49
25
19
22
4
18
Represents the number of observations for which the information is provided.
Canadian banks include credit agreements provided by a group of Canadian banks (or a single
Canadian bank) or a group of banks that include Canadian banks. This information was not always available in the announcements.
b
436
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
is de®ned as the day of the announcement in the Globe & Mail t 0 and the
previous day t ÿ1. The parameters of the market model on daily returns
are estimated over the period ranging from t ÿ170 to t ÿ21 prior to the
announcement. 6 Tests of statistical signi®cance (z-statistics) of the average
abnormal returns are based on standardized prediction errors using the parameters of the market model (see James, 1987).
4. Empirical results
4.1. The information content of bank credit agreements
Before examining the impact of the introduction of the 1988-capital
adequacy requirements on the market reaction, we present the average excess returns for the two-day announcement period around the disclosure of
bank credit agreements controlling for various dimensions of the credit
agreements, ®rms, and banks. These results are presented in Panel A of
Table 2. Consistent with prior studies, we observe signi®cant positive market
reactions at the announcements of bank credit agreements. We also obtain
evidence that the market reaction is stronger when ®rms obtain loans
from single banks than from multiple banks. The market reaction is signi®cant when the credit agreements are provided to small ®rms but not to
large ®rms. The empirical evidence suggests that the market reaction is
signi®cant at the disclosure of new credit agreements and favorably revised
agreements.
The average announcement excess return is statistically dierent from zero
when the credit agreement involves a term loan (announcement excess return of
2.20% with a z-statistic of 2.45) and both a term loan and a line of credit
(announcement excess returns of 5.24% with a z-statistic of 3.27), but it is not
statistically dierent from zero when it involves a line of credit (announcement
excess returns of 1.59% with a z-statistic of )0.02). While the signi®cant positive market reaction at the disclosure of term loans is consistent with the results obtained by James (1987), the lack of signi®cance at the disclosure of lines
of credit is not. Our results suggest that the market may perceive the information content of a commitment to lend dierently from the actual lending.
However, the null hypothesis that the two market reactions are the same
cannot be rejected.
6
To correct for thin trading problems, missing returns are calculated using the bid and ask
prices. More precisely, the price used to calculate the return for a day is assumed to be the mean of
the bid and ask prices when no transaction is recorded for that day.
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
437
4.2. Investigation of causes for the dierence between the information content of
term loans and lines of credit
The predictions developed in Section 2 can potentially explain the dierence
between the market reactions at the disclosure of lines of credit and term loans.
Panel B of Table 2 presents the average announcement excess returns at the
disclosures of both lines of credit and term loans before and after 1988. The
market reactions at the disclosure of both types of agreement before 1988 are
not signi®cant. The null hypothesis that the market reactions are the same
cannot be rejected, which is consistent with prediction 1.
The market reaction at the disclosures of term loans after 1988 is signi®cant
while the reaction at the disclosure of lines of credit is not. The null hypothesis
that the market reactions are the same is rejected at 5% level (t test of 2.27).
This result is consistent with prediction 2. The null hypothesis that the market
reactions at the disclosure of lines of credit before and after 1988 are the same
cannot be rejected, which is contrary to prediction 3. The null hypothesis that
the market reactions at the disclosure of term loans after and before 1988 are
the same cannot be rejected, which is consistent with prediction 4.
Results in Panel A of Table 2 and previous studies (for example, Slovin et
al., 1992) provide evidence that ®rm size aects the information content of
bank credit agreements. We reexamine the predictions developed in Section 2
by dividing the sample according to ®rm size. Results are presented in Panels C
and D of Table 2.
For small ®rms before 1988, the average announcement excess returns are
6.63% for lines of credit (z-statistic of 0.53) and )0.04% for term loans (zstatistic of 1.37). 7 After 1988, the average announcement excess returns are
0.06% for lines of credit (z-statistic of 1.18) and 4.69% for term loans (z-statistic
of 2.97). Furthermore, all four predictions are supported. 8 Therefore, for
small ®rms, the introduction of the 1988-capital adequacy requirements has
signi®cantly reduced the information content of lines of credit, while the informativeness of term loans is not aected.
7
Note that the negative (but not signi®cant) market reaction at the disclosure of term loans is
surprising since we expected to have a positive and signi®cant market reaction. Further
investigation reveals that a positive and signi®cant market reaction is obtained when one speci®c
negative observation is excluded from the test (the impact of one observation can be important in
this test given the small sample size). Even with the exclusion of this variable, we still cannot reject
the null that the two market reactions are the same.
8
Prediction 3, which was not supported using the entire sample, is supported for small ®rms by
using the Mann±Whitney U test. The use of a non-parametric test is justi®ed by the small number
of lines of credit before 1988 in the sample. The results for the other predictions are consistent using
both parametric and non-parametric tests.
438
Table 2
Average announcement excess returns
Number of observations
122
91
27
4
23
61
61
38
72
49
22
46
64
12
2.27
2.27
1.97
4.24
2.86
2.97
1.57
4.67
0.68
3.25
1.06
1.59
2.20
5.24
(B) Average announcement excess returns and capital adequacy: Full samplec
Before 1988
Lines of credit
13
4.82
Term loans
22
1.14
After 1988
Lines of credit
33
0.32
Term loans
54
3.30
(C) Average announcement excess returns and capital adequacy: Small ®rmsd
Before 1988
Lines of credit
7
6.63
Term loans
11
)0.04
After 1988
Lines of credit
15
0.06
Term loans
28
4.69
z-statistic
Proportion of positive
excess returns (%)
2.68
2.22
1.03
1.51
1.89
2.19
1.59
2.59
0.73
1.38
0.97
)0.02
2.45
3.27
57
58
50
75
48
59
54
66
48
55
59
48
61
67
0.37
1.42
62
64
0.47
3.35
42
61
0.53
1.37
86
64
1.18
2.97
40
61
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
(A) Average announcement excess returns
Full sample
New agreements
Revised agreements
New and revised agreements
Favorable revisions
Small ®rmsa
Large ®rms
Single bank
Multiple banks
Canadian banksb
Foreign banks
Lines of credit
Term loans
Lines of credit and term loans
Announcement period
excess returns (%)
Table 2 (Continued)
Number of observations
Announcement period
excess returns (%)
a
Proportion of positive
excess returns (%)
0.54
0.02
33
64
0.45
1.73
44
62
We use the median of the total value of ®rmsÕ assets to distinguish between small and large ®rms.
Canadian banks include credit agreements provided by a group of Canadian banks (or a single Canadian bank) and a group of banks that include
Canadian banks.
c
The null hypothesis that the market reactions are the same at the disclosure of lines of credit and term loans before 1988 cannot be rejected. The null
hypothesis that the market reactions are the same at the disclosure of lines of credit and term loans after 1988 can be rejected at 5% level (t-test of 2.27).
The null hypothesis that the market reactions are the same at the disclosure of lines of credit before and after 1988 cannot be rejected. The null
hypothesis that the market reactions are the same at the disclosure of term loans before and after 1988 cannot be rejected.
d
The null hypothesis that the market reactions are the same at the disclosure of lines of credit and term loans before 1988 cannot be rejected. The null
hypothesis that the market reactions are the same at the disclosure of lines of credit and term loans after 1988 can be rejected at 5% level (t-test of 2.16).
The null hypothesis that the market reactions are the same at the disclosure of lines of credit before and after 1988 can be rejected at 10% level using the
Mann±Whitney U-test (see footnote 10). The null hypothesis that the market reactions are the same at the disclosure of term loans before and after 1988
cannot be rejected.
e
We cannot reject any null hypothesis that any two market reactions are the same.
*
Signi®cant at 0.10 level.
**
Signi®cant at 0.05 level.
***
Signi®cant at 0.01 level.
b
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
(D) Average announcement excess returns and capital adequacy: Large ®rmse
Before 1988
Lines of credit
6
2.70
Term loans
11
2.32
After 1988
Lines of credit
18
0.54
Term loans
26
1.70
z-statistic
439
440
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
For large ®rms (Table 2, Panel D), the market reaction is not signi®cant at
the disclosure of lines of credit before 1988 (the average announcement excess
returns is 2.70 and the z-statistic is 0.54) and at the disclosure of term loans (the
average announcement excess returns is 2.32 and the z-statistic is 0.02). After
1988, the market reaction is not signi®cant at the disclosure of lines of credit
(the average announcement excess returns is 0.54 and the z-statistic is 0.45),
while it is signi®cant at the disclosure of term loans (the average announcement
excess returns is 1.70 and the z-statistic is 1.73). However, predictions 2 and 3
are not supported.
We perform additional tests to explore other causes that could explain the
dierence between the market reactions at the disclosure of term loans and
lines of credit. First, we examine the possibility that there was initial borrowing
when ®rms received lines of credit. Second, we examine if there is a dierence
between the market reaction at the disclosure of revolving credit agreements
versus other types of lines of credit. Third, we then examine the possibility that
the relevant event-window for lines of credit diers from the one for term
loans. 9 The results of these tests failed to explain the dierence between the
two market reactions. Therefore, the 1988-capital adequacy requirements best
explains the dierence between the information content conveyed by the disclosure of lines of credit and term loans.
4.3. Multivariate analysis
In this section, we use multivariate regressions to test the relationship between announcement excess returns and the variables used in the univariate
analysis to examine the market reaction to the disclosure of bank credit
agreements. As in Lummer and McConnell (1989) and Johnson (1996), we
control for heteroscedasticity in cross-sectional stock returns by using a
weighted least squares regression using the inverse of the relevant standard
prediction errors as weight. We start with a simple model that consists of four
variables as follows:
PEi a b1 SMALL82±87i b2 SMALL88±95i b3 LARGE82±87i
b4 LARGE88±95i ei ;
where PE is the two-day excess return, SMALL82±87 a dummy variable that
takes the value of 1 when a small ®rm receives a term loan before 1988 and
takes the value of 0 when it receives a line of credit, SMALL88±95 a dummy
9
We examine the average announcement excess returns for windows ranging from 2 to 11 days
prior to the announcement. We ®nd that 62% of our sample ®rms report non-trivial corporate news
items during ÿ4 to ÿ11 days window. As a result of this noise, it is not appropriate to extend the
window beyond ÿ3 days.
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
441
variable that takes the value of 1 when a small ®rm receives a term loan after
1988 and takes the value of 0 when it receives a line of credit, LARGE82±87 a
dummy variable that takes the value of 1 when a large ®rm receives a term loan
before 1988 and takes the value of 0 when it receives a line of credit,
LARGE88±95 a dummy variable that takes the value of 1 when a large ®rm
receives a term loan after 1988 and takes the value of 0 when it receives a line of
credit; ei is a noise term.
Given the results obtained in the univariate analysis, we expect to have a
positive sign for the variable SMALL88±95 since banksÕ level of commitment
at the issuance of lines of credit after 1988 is expected to be lower. The coef®cients of the variables SMALL82±87, LARGE82±87, and LARGE88±95 are
expected to be non-signi®cant. 10
Results of this regression are presented in Columns B and C of Table 3. The
adjusted R2 of the model is 4.3%. The only signi®cant variable is SMALL88±95
(t-statistic of 2.92) and it has the expected sign. Therefore, the results indicate
that the introduction of the 1988-capital adequacy rules have changed the
market perception regarding the level of information conveyed by lines of
credit and term loans.
Based on the results of the univariate analysis, we add the following variables to the model:
LOGASSET
NUMBK
NATBK
NR
the logarithm of the total value of assets;
a dummy variable that takes the value of 1 when the loan is
provided by a single bank and takes the value of 0 when the
loan is provided by multiple banks or when the information
is not provided;
a dummy variable that takes the value of 1 when the loan is
provided by a Canadian bank and takes the value of 0 when
the loan is provided by a non-Canadian bank or when the
information is not provided;
a dummy variable that takes the value of 0 when it is a new
loan and takes the value of 1 when it is a revised loan.
The predicted signs of the above variables, presented in Column A of Table
3, are based on the results of the univariate analysis and the evidence provided
in the literature.
10
We do not test prediction 3 in the multivariate analysis. As discussed in the univariate analysis,
because of the small number of lines of credit before 1988 in our sample, we could not reject the null
hypothesis that the market reaction at the diclosure of lines of credit before and after 1988 are the
same using parametric tests.
442
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
Table 3
Multivariate analysisa
Independent
variables
(A)
Sign
(B)
Results
(C)
t-Statistic
(D)
Results
(E)
t-Statistic
(F)
Results
(G)
t-Statistic
Intercept
SMALL82±87
SMALL88±95
LARGE82±87
LARGE88±95
LOGASSET
NUMBK
NATBK
NR
NUMBKInt
NATBKInt
?
?
+
?
?
)
+
+
)
+
+
)0.005
0.019
0.028
0.006
0.003
)1.14
1.16
2.92
0.52
0.53
0.058
0.009
0.024
)0.005
0.004
)0.003
)0.002
)0.004
)0.008
1.52
0.47
2.02
)0.36
0.56
)1.58
)0.25
)0.59
)0.83
0.079
0.017
0.035
0.002
0.007
)0.003
)0.012
)0.006
)0.006
0.030
0.001
1.94
0.85
2.47
0.13
0.88
)1.89
)1.26
)0.74
)0.64
1.78
0.13
Number of
observationsb
Adjusted R2
122
0.043
115
0.069
115
0.079
a
PEi a b1 SMALL82±87i b2 SMALL88±95i b3 LARGE82±87i b4 LARGE88±95i
b5 LOGASSETi b6 NUMBKi b7 NATBKi b8 NRi b9 NUMBKInt
i
b10 NATBKInt
i ei ;
where PE is the two-day excess return, SMALL82±87 a dummy variable that takes the value of 1
when a small ®rm receives a term loan before 1988 and takes the value of 0 when it receives a line of
credit, SMALL88±95 a dummy variable that takes the value of 1 when a small ®rm receives a term
loan after 1988 and takes the value of 0 when it receives a line of credit, LARGE82±87 a dummy
variable that takes the value of 1 when a large ®rm receives a term loan before 1988 and takes the
value of 0 when it receives a line of credit, LARGE88±95 a dummy variable that takes the value of 1
when a large ®rm receives a term loan after 1988 and takes the value of 0 when it receives a line of
credit, LOGASSET the logarithm of the total value of assets, NUMBK a dummy variable that
takes the value of 1 when the loan is provided by a single bank and takes the value of 0 when the
loan is provided by multiple banks or when the information is not provided, NATBK a dummy
variable that takes the value of 1 when the loan is provided by a Canadian bank and takes the value
of 0 when the loan is provided by a non-Canadian bank or when the information is not provided,
NR a dummy variable that takes the value of 0 when it is a new loan and takes the value of 1 when
it is a revised loan, NUMBKInt an interactive dummy variable that takes the value of 1 if the ®rm
receives a term loan from a single bank and takes the value of 0 otherwise, and NATBKInt is an
interactive dummy variable that takes the value of 1 if the ®rm receives a term loan from a
Canadian bank and takes the value of 0 otherwise.
b
There are a total of 115 observations in the regression since some information is missing for few
®rms and we exclude ®rms that receive both a new and a revised credit agreement.
*
Signi®cant at 0.10 level.
**
Signi®cant at 0.05 level.
***
Signi®cant at 0.01 level.
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
443
Columns D and E of Table 3 present the results. The adjusted R2 of the
model is 6.9%. The variable SMALL88±95 is still signi®cant (t-statistic of 2.02)
with the predicted sign. This con®rms the results of the previous model as well
as the univariate analysis. However, it is surprising that none of the other
variables is signi®cant.
Given the interaction that seems to exist between some of the variables and
the type of credit agreements (term loan and line of credit), we further expand
our model by including the following two variables: 11
NUMBKInt
NATBKInt
an interactive dummy variable that takes the value of 1 if
the ®rm receives a term loan from a single bank and
takes the value of 0 otherwise;
an interactive dummy variable that takes the value of 1 if
the ®rm receives a term loan from a Canadian bank and
takes the value of 0 otherwise.
Columns F and G of Table 3 present the results of the regression with the
interaction variables. The inclusion of these additional variables increases the
adjusted R2 from 6.9% to 7.9%. Once again, the coecient of the variable
SMALL88±95 is signi®cant (t-statistic of 2.47) with the expected sign. With the
inclusion of the two interaction variables, the coecient of the variable
LOGASSET becomes signi®cant (t-statistic of )1.89) with the expected sign.
Finally, the variable NUMBKInt is signi®cant (t-statistic of 1.78) with the expected sign. The remaining variables are not signi®cant.
5. Conclusion
Our paper contributes to the banking literature by examining the impact of
the 1988-capital adequacy requirements on the informativeness of bank credit
agreements. We provide evidence that, for small ®rms, the information content
conveyed by the disclosure of lines of credit before and after 1988 signi®cantly
diers. We also provide evidence that the information content conveyed by the
disclosure of term loans and lines of credit signi®cantly diers after but not
before 1988. Finally, the null hypothesis that the market reactions at the disclosure of term loans before and after 1988 are the same cannot be rejected. We
view these results as evidence that the market perceives that banks reduced
their level of commitment when issuing lines of credit after the implementation
of the 1988-capital adequacy requirements.
11
We also included other variables such as the relative loan amount (loan amount over total
assets plus loan amount) and growth opportunities (market value over equity). The inclusion of
these variables does not aect the main results.
444
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
We also examine the eect of concentration of borrowing, ®rm size, nationality of the lending bank, and new versus renewed loans on the information
content conveyed by bank credit agreements. We obtain evidence that the
market reaction is stronger when credit agreements are issued by single banks
than by multiple banks. The market reaction is signi®cant when credit agreements are provided to small ®rms. Our empirical evidence also suggests that the
market reaction is signi®cant at the disclosure of new credit agreements.
Acknowledgements
The authors would like to thank Jean-Francßois Gosselin Labbe and John
John D'Argensio for their research assistance. Financial support from Canadian Academic Accounting Association is gratefully acknowledged. The authors would like to thank Glenn Feltham, Terry Levesques, Theresa Libby,
Sean Robb and seminar participants at the 1998 Northern Finance Association
Meetings, the 1999 Canadian Academic Accounting Association, European
Accounting Association Meetings and the Ecole
des HEC for their useful
comments.
References
Best, R., Zhang, H., 1993. Alternative information sources and the information content of bank
loans. Journal of Finance 48, 1507±1522.
Brown, S.J., Warner, J.B., 1980. Measuring security price performance. Journal of Financial
Economics 8, 205±258.
Fama, E.F., 1985. What's dierent about banks? Journal of Monetary Economics 15, 29±39.
James, C., 1987. Some evidence of the uniqueness of bank loans. Journal of Financial Economics
19, 217±235.
James, C., Wier, P., 1990. Borrowing relationships, intermediation, and the cost of issuing public
securities. Journal of Financial Economics 28, 149±173.
Johnson, S.A., 1996. The eect of bank reputation on the value of bank loans agreements. Journal
of Accounting Auditing and Finance 12 (1), 83±100.
Lummer, S.L., McConnell, J.J., 1989. Further evidence on the bank lending process and the
capital-market response to bank loan agreements. Journal of Financial Economics 25, 99±122.
Petersen, M.A., Rajan, R.J., 1994. The bene®ts of lending relationships: Evidence from small
business data. Journal of Finance 49, 3±37.
Preece, D., Mullineaux, D.J., 1996. Monitoring, loan renegotiability and ®rm value: The role of
lending syndicates. Journal of Banking and Finance 20, 577±593.
Slovin, M.B., Johnson, S.A., Glascock, J.L., 1992. Firm size and the information content of bank
loan announcements. Journal of Banking and Finance 16, 1057±1071.
Slovin, M.B., Young, J.E., 1990. Bank lending and initial public oerings. Journal of Banking and
Finance 14, 729±740.
www.elsevier.com/locate/econbase
A note on: Capital adequacy and the
information content of term loans and lines of
credit
P. Andre
a,1
, R. Mathieu
b,*
, P. Zhang
c,2
a
Ecole
des Hautes Etudes
Commerciales, 3000 Chemin de la C^
ote-Sainte-Catherine, Montr
eal, Que.,
Canada H3T 2A7
b
School of Business and Economics, Wilfrid Laurier University, Waterloo, Ont., Canada N2L 3C5
c
School of Accountancy, University of Waterloo, Waterloo, Ont., Canada N2L 3G1
Received 2 October 1998; accepted 25 October 1999
Abstract
This study examines the information content conveyed by the disclosure of credit
agreements in a Canadian setting. We argue that the introduction of the 1988-capital
adequacy requirements lead banks to reduce their level of commitment at the issuance of
lines of credit to avoid their inclusion in the calculation of the capital ratio. As a result,
after 1988, the disclosure of lines of credit is expected to be less informative than the
disclosure of term loans since banks may exert less eort to screen and monitor ®rms. Our
results are consistent with the argument that the dierence between the market reactions
at the disclosure of term loans and lines of credit is signi®cant after 1988. We also provide
evidence that ®rm size and concentration of borrowing aect the market reaction at the
disclosure of bank credit agreements. Ó 2001 Elsevier Science B.V. All rights reserved.
JEL classi®cation: G21
Keywords: Information of bank credit agreements; Capital adequacy
*
Corresponding author. Tel.: +1-519-884-1970, ext. 3142; fax: +1-519-884-0201.
E-mail addresses: paul.andre@hec.ca (P. AndreÂ), rmathieu@wlu.ca (R.
pzhang@uwaterloo.ca (P. Zhang).
1
Tel.: +514-340-6528; fax: +514-340-5633.
2
Tel.: +519-888-4567, ext. 6548; fax: +519-888-7562.
0378-4266/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 4 2 6 6 ( 9 9 ) 0 0 1 3 0 - 2
Mathieu),
432
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
1. Introduction
Bank credit agreements (term loans and lines of credit) provide ®rms with
important access to capital. These agreements, when disclosed, can provide a
signal to the market of ®rmsÕ ability to raise capital and banksÕ assessment of
®rmsÕ value (Fama, 1985; James, 1987; James and Wier, 1990; Slovin and
Young, 1990; Lummer and McConnell, 1989). The empirical evidence indicates
that the characteristics of both banks and ®rms can aect the degree of informativeness of bank credit agreements (Johnson, 1996; Preece and Mullineaux, 1996; Petersen and Rajan, 1994). The empirical ®ndings also suggest
that the informativeness of bank credit agreements depends on how well ®rms
are monitored (Slovin et al., 1992; Best and Zhang, 1993).
In this paper, using a sample of Canadian ®rms, we examine the relationship
between the informativeness of credit agreements and banksÕ level of commitment. BanksÕ level of commitment at the issuance of a credit agreement is
aected by several factors such as time to maturity, the collateral oered on the
loan, and their ability to unconditionally cancel the credit agreement. If banks
reduce their level of commitment when providing ®nancing, their risk exposure
would be reduced which may lead them to exert lower eort in screening and
monitoring ®rms. In such a case, we expect to observe a lower market reaction
at the disclosure of bank credit agreements.
In 1988, Canadian authorities changed the capital adequacy requirements to
adopt the recommendations of the Bank of International Settlements. We argue that the introduction of the 1988-capital ratio results in a reduction of
banksÕ level of commitment at the issuance of lines of credit so that they can
avoid some costs associated with the issuance of o-balance sheet instruments.
Therefore, by comparing the information content conveyed by the disclosure of
term loans and lines of credit before and after 1988, we can test the impact of
banksÕ level of commitment on the market reaction at the disclosure of bank
credit agreements. 3
Our sample consists of 122 announcements of new and revised bank credit
agreements in the Canadian market during the period 1982±1995. Our results
indicate that the information content conveyed by the disclosure of lines of
credit diers before and after 1988 for small ®rms. The information content
conveyed by lines of credit and term loans for small ®rms diers after 1988, but
not before. Furthermore, the information content conveyed by term loans
3
The conditions on the credit agreements can be used as proxies for bankÕs level of commitment.
Those conditions include, among others, time to maturity, the collateral oered on the loan, and
their ability to unconditionally cancel the credit agreement. However, in our sample of
announcements, information with respect to the time to maturity and the collateral oered on
the loan are not always disclosed, and information on banksÕ ability to unconditionally cancel is not
available.
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
433
before and after 1988 is the same. We view these results as evidence that the
introduction of the 1988-capital adequacy requirements has reduced the informativeness of lines of credit.
We also document in this paper signi®cant average announcement excess
returns when bank credit agreements are new, when they are provided to small
®rms, when they are issued by single banks or when renewals are favorable.
The remainder of the paper is structured as follows. Section 2 describes the
1988-capital adequacy requirements and presents our predictions. Data collection and methodology are discussed in Section 3. Section 4 presents the
empirical results and Section 5 summarizes the paper.
2. Capital adequacy requirements and the information of bank credit agreements
To maintain public con®dence in the banking system, the Canadian government insures deposits up to $60,000. To limit the level of risk chosen by
banks, the government imposes capital adequacy measures that link the capital
issued by banks to their investing activities. Prior to 1988, the capital adequacy
guidelines suggested that banks maintain 1 dollar of capital for every 30 dollars
of assets. However, this basic ratio did not take into account banksÕ exposure
to risk and included only on-balance sheet activities.
In 1988, Canadian authorities changed the capital adequacy requirements to
comply with the recommendations of the Bank of International Settlement,
referred to as the Basle Accord. The Basle Accord introduces a common definition of capital, uses a weighting system to calculate the minimum capital
requirement, and takes into consideration o-balance sheet activities.
Before the introduction of the Basle Accord, the used portion of a line of
credit had no impact on the required minimum level of capital. However, after
1988, lines of credit may be included in the calculation of the capital ratio. Like
all o-balance sheet instruments, the inclusion of lines of credit follows a twostep procedure. First, a credit conversion factor (i.e., a weight) is applied to
obtain the asset equivalents. Second, the weight applied to similar assets is used
to convert these asset equivalents into risk-adjusted o-balance sheet activities.
The guidelines published by the Superintendent of ®nancial institutions
de®ne the weights used for the conversion of o-balance sheet instruments.
With respect to lines of credit, the following weights are used to obtain the
asset equivalents:
· 50%: commitments with an original maturity exceeding 1 year, including underwriting commitments and commercial credit lines.
· 0%: commitments with an original maturity of 1 year or less or that are unconditionally cancelable at any time without prior notice.
Banks can avoid the impact of lines of credit on the calculation of capital
adequacy requirements by either reducing the length of the credit agreements
434
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
or introducing a clause that allows them to unconditionally cancel the lines of
credit at any time without prior notice. Discussions with bankers revealed that
most Canadian banks have changed the basic conditions associated with lines
of credit to avoid their inclusion in the calculation of the capital ratio. The
consequence of these changes in the basic conditions of lines of credit is to
lower the commitment of banks to their clients. As a result, we expect to observe a weaker market reaction at their disclosure after 1988 and the following
predictions are developed accordingly:
Prediction 1: The information content conveyed by the disclosure of lines
of credit and term loans prior to 1988 does not dier.
Prediction 2: The information content conveyed by the disclosure of lines of
credit and term loans after 1988 diers.
Prediction 3: The information content conveyed by the disclosure of lines
of credit before and after 1988 diers.
Prediction 4: The information content conveyed by the disclosure of term
loans before and after 1988 does not dier.
3. Data and methodology
The Globe & Mail on CD-ROM and the database of Canadian Business and
Current Aairs (CBCA) are used to obtain the announcements of Canadian
bank credit agreements in the Globe & Mail over the period 1982±1995. 4; 5
Only ®rms with stock prices on the TSE Western daily ®le are included in the
sample, and we obtain a total of 150 announcements. From this total, 28
observations are discarded since there is other information in the announcements about the ®rms. The ®nal sample consists of 122 announcements. Descriptive statistics of the sample are provided in Table 1.
To analyze the information content of bank credit agreements, we examine
the relationship between changes in ®rmsÕ market value at the announcement
of bank credit agreements using an event study methodology. As in Brown and
Warner (1980), James (1987), and Lummer and McConnell (1989), the excess
returns are calculated using the market model, and the two-day event window
4
The Globe & Mail is a daily newspaper that specializes in economic issues and is the Canadian
equivalent of the Wall Street Journal. During our sample period, it was the only nation-wide
newspaper.
5
Given that some parent companies are not Canadian corporations, there is a possibility that the
announcements in Canada lag the announcements in foreign countries. We conduct a search of the
Wall Street Journal using our entire sample and ®nd no evidence that such a lag exists.
435
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
Table 1
Descriptive statistics
Full sample
Number of
observations 122
(A) Sample composition
New credit agreements
Revised credit agreements
New and revised credit agreements
Lines of credit
Term loans
Lines of credit and term loans
91
27
4
46
64
12
(B) Loan amount (in millions of dollars)
Number of
Mean
observationsa
Full sample
New agreements
Revised agreements
New and revised agreements
Lines of credit
Term loans
Lines of credit and term
loans
Minimum
Maximum
111
85
23
3
317
337
259
1978
140
123
211
246
0.5
0.5
5
22
2520
2520
1400
326
44
56
11
352
279
370
161
106
200
5
0.5
78
2250
2520
2200
(C) Single versus multiple banks
Single bank
Full sample
Lines of credit
Term loans
Median
38
13
24
Multiple
banks
72
30
32
Unknown
12
3
8
(D) Canadian versus non-Canadian banks
Canadian
Nonbanksb
Canadian
banks
Full sample
Lines of credit
Term loans
a
49
25
19
22
4
18
Represents the number of observations for which the information is provided.
Canadian banks include credit agreements provided by a group of Canadian banks (or a single
Canadian bank) or a group of banks that include Canadian banks. This information was not always available in the announcements.
b
436
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
is de®ned as the day of the announcement in the Globe & Mail t 0 and the
previous day t ÿ1. The parameters of the market model on daily returns
are estimated over the period ranging from t ÿ170 to t ÿ21 prior to the
announcement. 6 Tests of statistical signi®cance (z-statistics) of the average
abnormal returns are based on standardized prediction errors using the parameters of the market model (see James, 1987).
4. Empirical results
4.1. The information content of bank credit agreements
Before examining the impact of the introduction of the 1988-capital
adequacy requirements on the market reaction, we present the average excess returns for the two-day announcement period around the disclosure of
bank credit agreements controlling for various dimensions of the credit
agreements, ®rms, and banks. These results are presented in Panel A of
Table 2. Consistent with prior studies, we observe signi®cant positive market
reactions at the announcements of bank credit agreements. We also obtain
evidence that the market reaction is stronger when ®rms obtain loans
from single banks than from multiple banks. The market reaction is signi®cant when the credit agreements are provided to small ®rms but not to
large ®rms. The empirical evidence suggests that the market reaction is
signi®cant at the disclosure of new credit agreements and favorably revised
agreements.
The average announcement excess return is statistically dierent from zero
when the credit agreement involves a term loan (announcement excess return of
2.20% with a z-statistic of 2.45) and both a term loan and a line of credit
(announcement excess returns of 5.24% with a z-statistic of 3.27), but it is not
statistically dierent from zero when it involves a line of credit (announcement
excess returns of 1.59% with a z-statistic of )0.02). While the signi®cant positive market reaction at the disclosure of term loans is consistent with the results obtained by James (1987), the lack of signi®cance at the disclosure of lines
of credit is not. Our results suggest that the market may perceive the information content of a commitment to lend dierently from the actual lending.
However, the null hypothesis that the two market reactions are the same
cannot be rejected.
6
To correct for thin trading problems, missing returns are calculated using the bid and ask
prices. More precisely, the price used to calculate the return for a day is assumed to be the mean of
the bid and ask prices when no transaction is recorded for that day.
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
437
4.2. Investigation of causes for the dierence between the information content of
term loans and lines of credit
The predictions developed in Section 2 can potentially explain the dierence
between the market reactions at the disclosure of lines of credit and term loans.
Panel B of Table 2 presents the average announcement excess returns at the
disclosures of both lines of credit and term loans before and after 1988. The
market reactions at the disclosure of both types of agreement before 1988 are
not signi®cant. The null hypothesis that the market reactions are the same
cannot be rejected, which is consistent with prediction 1.
The market reaction at the disclosures of term loans after 1988 is signi®cant
while the reaction at the disclosure of lines of credit is not. The null hypothesis
that the market reactions are the same is rejected at 5% level (t test of 2.27).
This result is consistent with prediction 2. The null hypothesis that the market
reactions at the disclosure of lines of credit before and after 1988 are the same
cannot be rejected, which is contrary to prediction 3. The null hypothesis that
the market reactions at the disclosure of term loans after and before 1988 are
the same cannot be rejected, which is consistent with prediction 4.
Results in Panel A of Table 2 and previous studies (for example, Slovin et
al., 1992) provide evidence that ®rm size aects the information content of
bank credit agreements. We reexamine the predictions developed in Section 2
by dividing the sample according to ®rm size. Results are presented in Panels C
and D of Table 2.
For small ®rms before 1988, the average announcement excess returns are
6.63% for lines of credit (z-statistic of 0.53) and )0.04% for term loans (zstatistic of 1.37). 7 After 1988, the average announcement excess returns are
0.06% for lines of credit (z-statistic of 1.18) and 4.69% for term loans (z-statistic
of 2.97). Furthermore, all four predictions are supported. 8 Therefore, for
small ®rms, the introduction of the 1988-capital adequacy requirements has
signi®cantly reduced the information content of lines of credit, while the informativeness of term loans is not aected.
7
Note that the negative (but not signi®cant) market reaction at the disclosure of term loans is
surprising since we expected to have a positive and signi®cant market reaction. Further
investigation reveals that a positive and signi®cant market reaction is obtained when one speci®c
negative observation is excluded from the test (the impact of one observation can be important in
this test given the small sample size). Even with the exclusion of this variable, we still cannot reject
the null that the two market reactions are the same.
8
Prediction 3, which was not supported using the entire sample, is supported for small ®rms by
using the Mann±Whitney U test. The use of a non-parametric test is justi®ed by the small number
of lines of credit before 1988 in the sample. The results for the other predictions are consistent using
both parametric and non-parametric tests.
438
Table 2
Average announcement excess returns
Number of observations
122
91
27
4
23
61
61
38
72
49
22
46
64
12
2.27
2.27
1.97
4.24
2.86
2.97
1.57
4.67
0.68
3.25
1.06
1.59
2.20
5.24
(B) Average announcement excess returns and capital adequacy: Full samplec
Before 1988
Lines of credit
13
4.82
Term loans
22
1.14
After 1988
Lines of credit
33
0.32
Term loans
54
3.30
(C) Average announcement excess returns and capital adequacy: Small ®rmsd
Before 1988
Lines of credit
7
6.63
Term loans
11
)0.04
After 1988
Lines of credit
15
0.06
Term loans
28
4.69
z-statistic
Proportion of positive
excess returns (%)
2.68
2.22
1.03
1.51
1.89
2.19
1.59
2.59
0.73
1.38
0.97
)0.02
2.45
3.27
57
58
50
75
48
59
54
66
48
55
59
48
61
67
0.37
1.42
62
64
0.47
3.35
42
61
0.53
1.37
86
64
1.18
2.97
40
61
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
(A) Average announcement excess returns
Full sample
New agreements
Revised agreements
New and revised agreements
Favorable revisions
Small ®rmsa
Large ®rms
Single bank
Multiple banks
Canadian banksb
Foreign banks
Lines of credit
Term loans
Lines of credit and term loans
Announcement period
excess returns (%)
Table 2 (Continued)
Number of observations
Announcement period
excess returns (%)
a
Proportion of positive
excess returns (%)
0.54
0.02
33
64
0.45
1.73
44
62
We use the median of the total value of ®rmsÕ assets to distinguish between small and large ®rms.
Canadian banks include credit agreements provided by a group of Canadian banks (or a single Canadian bank) and a group of banks that include
Canadian banks.
c
The null hypothesis that the market reactions are the same at the disclosure of lines of credit and term loans before 1988 cannot be rejected. The null
hypothesis that the market reactions are the same at the disclosure of lines of credit and term loans after 1988 can be rejected at 5% level (t-test of 2.27).
The null hypothesis that the market reactions are the same at the disclosure of lines of credit before and after 1988 cannot be rejected. The null
hypothesis that the market reactions are the same at the disclosure of term loans before and after 1988 cannot be rejected.
d
The null hypothesis that the market reactions are the same at the disclosure of lines of credit and term loans before 1988 cannot be rejected. The null
hypothesis that the market reactions are the same at the disclosure of lines of credit and term loans after 1988 can be rejected at 5% level (t-test of 2.16).
The null hypothesis that the market reactions are the same at the disclosure of lines of credit before and after 1988 can be rejected at 10% level using the
Mann±Whitney U-test (see footnote 10). The null hypothesis that the market reactions are the same at the disclosure of term loans before and after 1988
cannot be rejected.
e
We cannot reject any null hypothesis that any two market reactions are the same.
*
Signi®cant at 0.10 level.
**
Signi®cant at 0.05 level.
***
Signi®cant at 0.01 level.
b
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
(D) Average announcement excess returns and capital adequacy: Large ®rmse
Before 1988
Lines of credit
6
2.70
Term loans
11
2.32
After 1988
Lines of credit
18
0.54
Term loans
26
1.70
z-statistic
439
440
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
For large ®rms (Table 2, Panel D), the market reaction is not signi®cant at
the disclosure of lines of credit before 1988 (the average announcement excess
returns is 2.70 and the z-statistic is 0.54) and at the disclosure of term loans (the
average announcement excess returns is 2.32 and the z-statistic is 0.02). After
1988, the market reaction is not signi®cant at the disclosure of lines of credit
(the average announcement excess returns is 0.54 and the z-statistic is 0.45),
while it is signi®cant at the disclosure of term loans (the average announcement
excess returns is 1.70 and the z-statistic is 1.73). However, predictions 2 and 3
are not supported.
We perform additional tests to explore other causes that could explain the
dierence between the market reactions at the disclosure of term loans and
lines of credit. First, we examine the possibility that there was initial borrowing
when ®rms received lines of credit. Second, we examine if there is a dierence
between the market reaction at the disclosure of revolving credit agreements
versus other types of lines of credit. Third, we then examine the possibility that
the relevant event-window for lines of credit diers from the one for term
loans. 9 The results of these tests failed to explain the dierence between the
two market reactions. Therefore, the 1988-capital adequacy requirements best
explains the dierence between the information content conveyed by the disclosure of lines of credit and term loans.
4.3. Multivariate analysis
In this section, we use multivariate regressions to test the relationship between announcement excess returns and the variables used in the univariate
analysis to examine the market reaction to the disclosure of bank credit
agreements. As in Lummer and McConnell (1989) and Johnson (1996), we
control for heteroscedasticity in cross-sectional stock returns by using a
weighted least squares regression using the inverse of the relevant standard
prediction errors as weight. We start with a simple model that consists of four
variables as follows:
PEi a b1 SMALL82±87i b2 SMALL88±95i b3 LARGE82±87i
b4 LARGE88±95i ei ;
where PE is the two-day excess return, SMALL82±87 a dummy variable that
takes the value of 1 when a small ®rm receives a term loan before 1988 and
takes the value of 0 when it receives a line of credit, SMALL88±95 a dummy
9
We examine the average announcement excess returns for windows ranging from 2 to 11 days
prior to the announcement. We ®nd that 62% of our sample ®rms report non-trivial corporate news
items during ÿ4 to ÿ11 days window. As a result of this noise, it is not appropriate to extend the
window beyond ÿ3 days.
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
441
variable that takes the value of 1 when a small ®rm receives a term loan after
1988 and takes the value of 0 when it receives a line of credit, LARGE82±87 a
dummy variable that takes the value of 1 when a large ®rm receives a term loan
before 1988 and takes the value of 0 when it receives a line of credit,
LARGE88±95 a dummy variable that takes the value of 1 when a large ®rm
receives a term loan after 1988 and takes the value of 0 when it receives a line of
credit; ei is a noise term.
Given the results obtained in the univariate analysis, we expect to have a
positive sign for the variable SMALL88±95 since banksÕ level of commitment
at the issuance of lines of credit after 1988 is expected to be lower. The coef®cients of the variables SMALL82±87, LARGE82±87, and LARGE88±95 are
expected to be non-signi®cant. 10
Results of this regression are presented in Columns B and C of Table 3. The
adjusted R2 of the model is 4.3%. The only signi®cant variable is SMALL88±95
(t-statistic of 2.92) and it has the expected sign. Therefore, the results indicate
that the introduction of the 1988-capital adequacy rules have changed the
market perception regarding the level of information conveyed by lines of
credit and term loans.
Based on the results of the univariate analysis, we add the following variables to the model:
LOGASSET
NUMBK
NATBK
NR
the logarithm of the total value of assets;
a dummy variable that takes the value of 1 when the loan is
provided by a single bank and takes the value of 0 when the
loan is provided by multiple banks or when the information
is not provided;
a dummy variable that takes the value of 1 when the loan is
provided by a Canadian bank and takes the value of 0 when
the loan is provided by a non-Canadian bank or when the
information is not provided;
a dummy variable that takes the value of 0 when it is a new
loan and takes the value of 1 when it is a revised loan.
The predicted signs of the above variables, presented in Column A of Table
3, are based on the results of the univariate analysis and the evidence provided
in the literature.
10
We do not test prediction 3 in the multivariate analysis. As discussed in the univariate analysis,
because of the small number of lines of credit before 1988 in our sample, we could not reject the null
hypothesis that the market reaction at the diclosure of lines of credit before and after 1988 are the
same using parametric tests.
442
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
Table 3
Multivariate analysisa
Independent
variables
(A)
Sign
(B)
Results
(C)
t-Statistic
(D)
Results
(E)
t-Statistic
(F)
Results
(G)
t-Statistic
Intercept
SMALL82±87
SMALL88±95
LARGE82±87
LARGE88±95
LOGASSET
NUMBK
NATBK
NR
NUMBKInt
NATBKInt
?
?
+
?
?
)
+
+
)
+
+
)0.005
0.019
0.028
0.006
0.003
)1.14
1.16
2.92
0.52
0.53
0.058
0.009
0.024
)0.005
0.004
)0.003
)0.002
)0.004
)0.008
1.52
0.47
2.02
)0.36
0.56
)1.58
)0.25
)0.59
)0.83
0.079
0.017
0.035
0.002
0.007
)0.003
)0.012
)0.006
)0.006
0.030
0.001
1.94
0.85
2.47
0.13
0.88
)1.89
)1.26
)0.74
)0.64
1.78
0.13
Number of
observationsb
Adjusted R2
122
0.043
115
0.069
115
0.079
a
PEi a b1 SMALL82±87i b2 SMALL88±95i b3 LARGE82±87i b4 LARGE88±95i
b5 LOGASSETi b6 NUMBKi b7 NATBKi b8 NRi b9 NUMBKInt
i
b10 NATBKInt
i ei ;
where PE is the two-day excess return, SMALL82±87 a dummy variable that takes the value of 1
when a small ®rm receives a term loan before 1988 and takes the value of 0 when it receives a line of
credit, SMALL88±95 a dummy variable that takes the value of 1 when a small ®rm receives a term
loan after 1988 and takes the value of 0 when it receives a line of credit, LARGE82±87 a dummy
variable that takes the value of 1 when a large ®rm receives a term loan before 1988 and takes the
value of 0 when it receives a line of credit, LARGE88±95 a dummy variable that takes the value of 1
when a large ®rm receives a term loan after 1988 and takes the value of 0 when it receives a line of
credit, LOGASSET the logarithm of the total value of assets, NUMBK a dummy variable that
takes the value of 1 when the loan is provided by a single bank and takes the value of 0 when the
loan is provided by multiple banks or when the information is not provided, NATBK a dummy
variable that takes the value of 1 when the loan is provided by a Canadian bank and takes the value
of 0 when the loan is provided by a non-Canadian bank or when the information is not provided,
NR a dummy variable that takes the value of 0 when it is a new loan and takes the value of 1 when
it is a revised loan, NUMBKInt an interactive dummy variable that takes the value of 1 if the ®rm
receives a term loan from a single bank and takes the value of 0 otherwise, and NATBKInt is an
interactive dummy variable that takes the value of 1 if the ®rm receives a term loan from a
Canadian bank and takes the value of 0 otherwise.
b
There are a total of 115 observations in the regression since some information is missing for few
®rms and we exclude ®rms that receive both a new and a revised credit agreement.
*
Signi®cant at 0.10 level.
**
Signi®cant at 0.05 level.
***
Signi®cant at 0.01 level.
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
443
Columns D and E of Table 3 present the results. The adjusted R2 of the
model is 6.9%. The variable SMALL88±95 is still signi®cant (t-statistic of 2.02)
with the predicted sign. This con®rms the results of the previous model as well
as the univariate analysis. However, it is surprising that none of the other
variables is signi®cant.
Given the interaction that seems to exist between some of the variables and
the type of credit agreements (term loan and line of credit), we further expand
our model by including the following two variables: 11
NUMBKInt
NATBKInt
an interactive dummy variable that takes the value of 1 if
the ®rm receives a term loan from a single bank and
takes the value of 0 otherwise;
an interactive dummy variable that takes the value of 1 if
the ®rm receives a term loan from a Canadian bank and
takes the value of 0 otherwise.
Columns F and G of Table 3 present the results of the regression with the
interaction variables. The inclusion of these additional variables increases the
adjusted R2 from 6.9% to 7.9%. Once again, the coecient of the variable
SMALL88±95 is signi®cant (t-statistic of 2.47) with the expected sign. With the
inclusion of the two interaction variables, the coecient of the variable
LOGASSET becomes signi®cant (t-statistic of )1.89) with the expected sign.
Finally, the variable NUMBKInt is signi®cant (t-statistic of 1.78) with the expected sign. The remaining variables are not signi®cant.
5. Conclusion
Our paper contributes to the banking literature by examining the impact of
the 1988-capital adequacy requirements on the informativeness of bank credit
agreements. We provide evidence that, for small ®rms, the information content
conveyed by the disclosure of lines of credit before and after 1988 signi®cantly
diers. We also provide evidence that the information content conveyed by the
disclosure of term loans and lines of credit signi®cantly diers after but not
before 1988. Finally, the null hypothesis that the market reactions at the disclosure of term loans before and after 1988 are the same cannot be rejected. We
view these results as evidence that the market perceives that banks reduced
their level of commitment when issuing lines of credit after the implementation
of the 1988-capital adequacy requirements.
11
We also included other variables such as the relative loan amount (loan amount over total
assets plus loan amount) and growth opportunities (market value over equity). The inclusion of
these variables does not aect the main results.
444
P. Andre et al. / Journal of Banking & Finance 25 (2001) 431±444
We also examine the eect of concentration of borrowing, ®rm size, nationality of the lending bank, and new versus renewed loans on the information
content conveyed by bank credit agreements. We obtain evidence that the
market reaction is stronger when credit agreements are issued by single banks
than by multiple banks. The market reaction is signi®cant when credit agreements are provided to small ®rms. Our empirical evidence also suggests that the
market reaction is signi®cant at the disclosure of new credit agreements.
Acknowledgements
The authors would like to thank Jean-Francßois Gosselin Labbe and John
John D'Argensio for their research assistance. Financial support from Canadian Academic Accounting Association is gratefully acknowledged. The authors would like to thank Glenn Feltham, Terry Levesques, Theresa Libby,
Sean Robb and seminar participants at the 1998 Northern Finance Association
Meetings, the 1999 Canadian Academic Accounting Association, European
Accounting Association Meetings and the Ecole
des HEC for their useful
comments.
References
Best, R., Zhang, H., 1993. Alternative information sources and the information content of bank
loans. Journal of Finance 48, 1507±1522.
Brown, S.J., Warner, J.B., 1980. Measuring security price performance. Journal of Financial
Economics 8, 205±258.
Fama, E.F., 1985. What's dierent about banks? Journal of Monetary Economics 15, 29±39.
James, C., 1987. Some evidence of the uniqueness of bank loans. Journal of Financial Economics
19, 217±235.
James, C., Wier, P., 1990. Borrowing relationships, intermediation, and the cost of issuing public
securities. Journal of Financial Economics 28, 149±173.
Johnson, S.A., 1996. The eect of bank reputation on the value of bank loans agreements. Journal
of Accounting Auditing and Finance 12 (1), 83±100.
Lummer, S.L., McConnell, J.J., 1989. Further evidence on the bank lending process and the
capital-market response to bank loan agreements. Journal of Financial Economics 25, 99±122.
Petersen, M.A., Rajan, R.J., 1994. The bene®ts of lending relationships: Evidence from small
business data. Journal of Finance 49, 3±37.
Preece, D., Mullineaux, D.J., 1996. Monitoring, loan renegotiability and ®rm value: The role of
lending syndicates. Journal of Banking and Finance 20, 577±593.
Slovin, M.B., Johnson, S.A., Glascock, J.L., 1992. Firm size and the information content of bank
loan announcements. Journal of Banking and Finance 16, 1057±1071.
Slovin, M.B., Young, J.E., 1990. Bank lending and initial public oerings. Journal of Banking and
Finance 14, 729±740.