Journal of Banking & Finance 36 (2012) 1437–1451

Contents lists available at SciVerse ScienceDirect

Journal of Banking & Finance

journal homepage: www.elsevier.com/locate/jbf

Common information asymmetry factors in syndicated loan structures

Claudia Champagne ⇑, Frank Coggins
Department of Finance, Université de Sherbrooke, 2500 Blvd. de l’Université, Sherbrooke, Canada J1K 2R1

a r t i c l e i n f o a b s t r a c t

Article history: This paper provides a comprehensive study of the syndicate structure and its relationship to information
Received 21 April 2011 asymmetry and loan spread by using principal component analysis on a large set of 40 structure-related
Accepted 10 December 2011 variables. A total of six structure components are identified and related to syndicate quality, syndicate
Available online 16 December 2011
members’ heterogeneity or share concentration, lead arranger’s characteristics, lead lender’s or syndi-
cate’s location, lender–borrower relationships, and lead institution type. In multivariate settings, all six
JEL classification: components are significant determinants of loan spread, either directly or indirectly through their impact
C31
on other components. Lead share retention, previous lender–borrower relationships and syndicate qual-
C38
F34
ity are shown to be bilaterally related to loan spread. Structure components differ regionally, which can
G21 provide an explanation for the European pricing discount observed in the literature. An Asian discount is
L14 observed and cannot be explained by structure differences.
Ó 2011 Elsevier B.V. All rights reserved.
Keywords:
Syndicated loan market
Principal component analysis
Syndicate structure
Information asymmetry
European pricing discount

1. Introduction
Extensive research conducted in the past 15 years on syndi-
cated loans concludes that one of the key differences between syn-
dicated loans and bilateral loans (or sole lender loans) is the
addition of within-syndicate information asymmetries. The way
the syndicate is structured serves as a mechanism to address po-
tential agency problems that can arise in such a context. However,
the syndicate structure is complex and involves many characteris-
tics that may mitigate or enhance within-syndicate information
asymmetries and their associated premium. Therefore, despite
the fact that syndicated loans have, by definition, more informa-
tion asymmetries (between syndicate members) than bilateral
loans, the overall level of this information asymmetry is hard to
determine and can still vary from one syndicate to another.1
Although recent papers have examined the impact of the struc-
ture on loan terms or the link between structure and information
asymmetry, most have focused on one or two structure measures
⇑ Corresponding author. Tel.: +1 819 821 8000x62976; fax: +1 819 821 7934.
E-mail addresses: claudia.champagne@usherbrooke.ca (C. Champagne), frank.
coggins@usherbrooke.ca (F. Coggins).
1
As evidenced, for instance, by the high standard deviations of the structure
variables and components in the sample, available in Tables 1 and 3 and discussed
later.
at a time, typically retention by the lead arranger and the number
of lenders. Although very important, these variables do not capture
the multidimensionality of the syndicate structure that includes
many different characteristics that combine and interact to in-
crease or decrease the diversification and information asymmetry
premiums of a loan.2 Judging the information asymmetries within
a syndicate solely on its size or arranger retention overlooks a
variety of factors that can influence them. These factors range from
previous lender–lender or lender–borrower relationships to the dif-
ference in experience between the lead and syndicate participants.
The first purpose of the paper is therefore to identify the princi-
pal components of the syndicate structure. This will not only allow
us to capture all the major characteristics of a syndicate structure
without generating unnecessary multicollinearity or overidentifi-
cation problems in multivariate settings, but also to combine
structure metrics into a small number of significant, easily inter-
pretable, components. These components are not only useful to
better understand the multidimensionality and complexity of a
syndicate structure through parsimonious models but also to com-
pare the structure of different syndicates along more than one
factor. The second objective of the paper is to use these structure
components in a multivariate setting to better understand how
2
See Ivashina (2009) for a complete discussion of these two premiums.

0378-4266/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.jbankfin.2011.12.009
1438 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451
they are related to one another and to information asymmetries
and to test their impact on loan spreads.
The contributions of this paper to the syndicated loans literature
are threefold. First, by identifying common syndicate structure fac-
tors related to information asymmetries across syndicated loans,
this paper provides a new approach to characterize and quantify
the multidimensional structure of a syndicate. While most papers
focus on one or two variables, usually related to syndicate size
and lead share retention, to proxy for syndicate structure, we use
a large set of variables to capture different aspects of a syndicate
that may increase or reduce information asymmetries. Starting
with this set of 40 structure variables, we show that six components
account for more than 60% of the variability in international syndi-
cate structures and that the components can be interpreted as the
quality of the syndicate, the heterogeneity of its members and syn-
dicate concentration, the characteristics of the lead arranger, the
geography of the syndication and its lead arranger, the average rela-
tions between the borrower and the syndicate members, and the
lender institution type. Second, this paper re-examines the impact
of syndicate structure on loan spreads using principal components
that capture the multidimensionality of syndicate structure, and
finds that all six components are significant determinants of loan
spread, either directly or indirectly though their impact on other
structure components. This simultaneous estimation of different
structure variables has never been done, and results support a num-
ber of different theories purported to explain loan spreads, such as
the agency, certification premium and home bias theories, and offer
new insights into their interaction.3 Further, we show that lead ar-
ranger share retention is not the only structure-related endogenous
determinant of spread, and that syndicate quality and previous len-
der–borrower relationships are also bilaterally related to loan spread.
Finally, we use structure components to compare three major
syndicated loan markets: US, Europe and Asia. We find that struc-
ture components, as well as their impact on loan spread, differ
regionally. US syndicates are, on average, of higher quality and
with more reputable and more-experienced lead arrangers than
European or Asian syndicates. They are also associated with the
strongest and longest lender–borrower relationships. Further, loan
spreads are less sensitive to most structure components in Europe
than in the US, while they are less sensitive to lead arranger char-
acteristics and syndicate concentration in Asia, everything else
held equal. Results also provide an explanation for the European
pricing discount observed in the literature. Specifically, the ob-
served discount can be attributable to different syndicate struc-
tures for European borrowers, which lead to lower diversification
and certification premiums, respectively. Finally, a new Asian pric-
ing discount is observed, which cannot be explained by different
syndicate structures.
The remainder of the paper is organized as follows. Section 2
reviews the literature on determinants of syndicate structure and
loan spreads. Section 3 presents the methodology and the results
obtained from the principal component analysis. Section 4 tests
the components in univariate and multivariate models of loan
spread and syndicate structure. Section 5 concludes the paper.
2. The determinants of syndicate structure and loan spread
To our knowledge, no previous study has addressed the issue of
the multidimensionality of the syndicate structure and the identi-
fication of structure components. However, there is a large body of
literature on the relationship between syndicate structure and
3
See, for example, Dennis and Mullineaux (2000), Sufi (2007), Ivashina (2009),
Cook et al. (2003), and Carey and Nini (2007). The theories are discussed in more
detail in Section 2.
information asymmetries and on the relationship between the
structure and loan terms.
2.1. Syndicate structure and information asymmetries
The structure of a loan syndicate has been extensively studied
in the past 15 years. This research has generally concluded that
the structure is conditioned by information asymmetries. Specifi-
cally, lender–borrower information asymmetries create informa-
tion asymmetries or frictions between syndicate members that
may lead to agency problems. There are two types of agency prob-
lems observed in this context: moral hazard and adverse selection
problems. The first problem, moral hazard, occurs when the lead
arranger reduces its incentive to monitor the loan optimally once
it is not responsible for the totality of it (Jensen and Meckling,
1976). The second problem, adverse selection, arises when the lead
arranger has private information about the borrower acquired
through due diligence or prior relationships with the borrower. If
the other members of the syndicate do not have access to this
information, a lemons problem can occur if the lead retains a larger
portion of the best-quality loans and a smaller portion of the low-
er-quality loans. The two types of agency problems have been
studied in the context of syndicated loans, as detailed in the fol-
lowing paragraphs.
While the structure of the syndicate can theoretically be seen
both as a consequence of or a solution to agency problems, studies
generally conclude that the syndicate is structured to reduce infor-
mation asymmetries between the participants involved. In accor-
dance with adverse selection predictions, the characteristics of
the lead arrangers have been shown to be significant determinants
of syndicate structure. For example, the proportion of the loan that
is retained by the lead arranger is shown to be negatively related to
the reputation of the lead (Panyagometh and Roberts, 2010). The
quantity and quality of information about the borrower also have
an impact on the syndicate structure. They are negatively related
to the share retained by the lead lender (Simons, 1993) and posi-
tively related to the number of lenders in the syndicate (Dennis
and Mullineaux, 2000). Panyagometh and Roberts (2002) find that
lead lenders syndicate a larger portion of loans that are subse-
quently upgraded, a sign that lead banks do not engage in exploit-
ative behavior, while Jones et al. (2001) observe a negative relation
between loan rating and lead share. However, they highlight that
arrangers may still exploit their informational advantage and
syndicate more of the low quality loans than the syndicate
members would have accepted under a symmetric-information
environment.
Consistent with moral hazard, loan syndicates can also imply a
free riding problem that reduces each lender’s incentive to monitor
and renegotiate if necessary. For instance, Preece and Mullineaux
(1996) find that the syndicate size (i.e. the number of lenders) is
negatively related to abnormal returns following loan announce-
ments because of the higher renegotiation costs. Further, Esty
and Megginson (2003) conclude that fewer lenders represent best
practices to promote monitoring efficiency and flexibility in
restructuring and that, in countries with strong creditor rights
and reliable legal enforcement, lenders create smaller and more
concentrated syndicates to facilitate monitoring and low cost con-
tracting. Lee and Mullineaux (2004) observe that smaller and more
concentrated syndicates can avert hold-out behavior by syndicate
members or coordination problems in renegotiations. Sufi (2007)
observes that lead arrangers retain a larger share and form more
concentrated syndicates when borrowers require more intense
due diligence and monitoring. Lead arranger reputation or past
lead–borrower relationships can help mitigate the information
asymmetries within the syndicate. Nonetheless, because low cost
restructuring can encourage borrowers to default strategically,
 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451
creditors may have an incentive to increase the size of the syndi-
cate to make default more costly or to impose a future penalty
on defaulting firms (Bolton and Scharfstein, 1996; Chowdhry,
1991).
2.2. Syndicate structure and loan terms
The impact of syndicate structure on loan terms has also been
studied, mostly using the syndicate size or lead arranger share as
a proxy for structure. Following agency theory, the way a syndicate
is structured affects the loan spread either through a diversification
or information asymmetry premium. Ivashina (2009) argues that
in equilibrium the information asymmetry premium required by
the participants is offset by the diversification premium required
by the lead arranger, which increases with the lead share. An in-
crease in the lead share therefore increases loan spreads. Syndicate
structure can also be related to non-price loan terms. For instance,
Coleman et al. (2006) find that larger banking syndicates lend for
longer maturities, but due to a decline in contractual flexibility
and monitoring, lend at lower yield spreads. Consistent with the
presence of lender–borrower information asymmetries, Cook
et al. (2003) conclude that lenders can extract a certification pre-
mium that is related to their reputation. Further studies find that
the location of the lead arranger or the syndication market can
have an impact on the price of the loan and often observe a home
bias (e.g. Carey and Nini, 2007; Houston et al., 2007). Finally, the
institution type of the lender can influence loan spreads. For in-
stance, Harjoto et al. (2006) and Nandy and Shao (2010) observe
that bank-led syndicates are associated with lower spreads and
lower-risk borrowers.
Overall, although the significance of syndicate structure on loan
terms has been studied, it has been on one proxy at a time, with
little or no consideration of interactions between structure compo-
nents or the bilateral relationship between loan spreads and syndi-
cate structure.
3. Syndicate structure components
3.1. Sample of syndicated loans
We generate an international sample that consists of 20,336
loan syndicates between 1998 and 2009 from Dealscan by Reuters’
Loan Pricing Corporation (LPC). Corporate information about the
lenders and borrowers is taken from the Compustat Global
database.
3.2. Syndicate structure components
One of the difficulties when trying to study the structure of syn-
dicates or account for it in empirical analyses is the number and
variety of variables that capture different aspects of the structure.
Such variables can be related to, among other things, the quality or
reputation of the lead arrangers or participants, past relationships
between syndicate members or the concentration of shares. These
variables can all be directly or indirectly related to the information
asymmetry and/or diversification premium discussed earlier.
Further, they combine and interact in ways that are not yet well
understood.
We use principal component analysis to better understand the
multidimensionality of syndicate structure and to develop a smal-
ler number of artificial variables or components that will account
for most of the variance in the observed structure variables. We
consider a total of 40 structure variables to build the components.
These structure variables are partly based on previous research
that focuses mostly on syndicate size and loan retention (e.g. Esty
1439
and Megginson, 2003; Ivashina 2009) and partly based on our own
measures. The variables are defined in Appendix A. They represent
characteristics of the syndicate size (e.g. number of leads and par-
ticipants), share concentration (e.g. lead retention and syndicate
concentration index), relationship between the lenders and the
borrower (e.g. average intensity and duration of the relationships),
quality of the entire syndicate (e.g. average reputation and experi-
ence), heterogeneity between the arrangers and the participants
(e.g. reputation and experience range), lead type (e.g. bank or
investment bank) and geography (e.g. lead or syndication region).
Table 1 shows the descriptive statistics for the 40 structure vari-
ables studied, and their correlation coefficient with loan spread.4
The structure variables that are most correlated with spread,
which includes a syndicate information asymmetry premium, nota-
bly comprise concentration measures (HH-INDEX: 0.33 and TOP3:
0.34), the number of lenders (LENDERS: 0.195), the institution type
of the lead arranger (LEAD-BANK: 0.22 and LEAD-INVEST: 0.15) and
the number of countries or industries involved in the syndicate
(COUNTRIES: 0.247 and TYPES: 0.19). Because of data availability,
principal component analysis is conducted on two sets of structure
variables: (i) the subset of 36 items (on the total sample of 20,336
observations) that excludes the four concentration-related
variables (i.e. LEAD-EXPOSURE, HH-INDEX, LEAD-SHARE AND TOP3),
and (ii) the entire set of 40 structure variables (on a restricted sample
of 7718 observations). We perform the following orthogonal linear
transformation of the data matrix X containing the structure
variables:
Y T ¼ XT W
P
T
¼V
where the matrices W, R and V are given by a singular value decom-
position of X. Because the structure variables are measured on dif-
ferent scales or units, we use the correlation matrix instead of the
covariance matrix. To determine the number of components that
will be retained for subsequent analysis, we use a combination of
criteria. Based on the Kaiser criterion, or the eigenvalue-one crite-
rion, only components with an eigenvalue greater than 1 are con-
sidered.5 According to the Kaiser criterion, 11 components would
be retained. However, only the first five components account for at
least 5% each of the total variance, which is often considered as
another criterion for inclusion. Further, a look at the cumulative
percentage of variance (not shown) indicates that the first six com-
ponents collectively account for 61.75% of the variance. An examina-
tion of the scree plot of the eigenvalues (not shown) indicates that
the number of components is more likely between 5 and 7. Finally,
the rotated factor patterns, obtained using orthogonal varimax rota-
tion, indicate that components 7–11 are not significantly loaded by
enough items that are not also loading on other components.6 Upon
reviewing the above criteria, and to ensure the interpretability of all
the components, we retain six components for subsequent analyses.
The rotated factor patterns obtained using six components are avail-
able in Panel A of Table 2.
Based on the highest loading variables for each component,
which are arbitrarily defined as variables that load with an optimal
weight greater than |0.4| for a particular component while not
4
The entire correlation coefficient matrix between the 40 structure variables is
untabulated but is available upon request. Correlation coefficients vary greatly,
ranging from 1.3% to 96.5%, and can be positive or negative according to the structure
variable.
5
Components’ eigenvalues, as well as the proportion of variance explained, are
untabulated but available upon request.
6
The rotated factor patterns for the 11 components with eigenvalues greater than
1 are not shown but are available upon request. Further, we arbitrarily set the weight
at 0.4 to determine ‘‘significant’’ variables in a component.
1440 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451

Table 1
Descriptive statistics for the syndicate structure variables. This table presents summary statistics for the 40 different variables described in Appendix A and used in the principal
component analysis in Section 3 of the paper, as well as their correlation with loan spreads. N is the sample size.

Variable N Mean Std. dev. Min. Max. Correlation with SPREAD

LENDERS 20336 10.8973 9.18 2.00 141.00 0.195
LEADS 20336 1.9820 2.41 0.00 37.00 0.101
PARTICIPANTS 20336 8.9154 8.99 0.00 140.00 0.172
TYPES 20336 1.6527 0.78 1.00 5.00 0.191
COUNTRIES 20336 3.9597 3.17 1.00 30.00 0.247
LEAD EXPOSURE 11994 0.0268 0.48 0.00 20.54 0.045
HH INDEX 7718 0.1893 0.16 0.01 1.00 0.331
LEAD SHARE 7718 36.1096 26.99 0.00 100.00 0.121
TOP3 7718 53.9565 27.38 6.36 103.65 0.336
REL-PARTICIPANTS 20336 1.0340 1.98 0.00 60.00 0.090
REL-LEAD 20336 1.7833 2.56 0.00 102.00 0.094
DURATION-PARTICIPANTS 20336 14.7618 30.04 0.00 766.00 0.093
DURATION-LEAD 20336 23.5283 35.48 0.00 774.00 0.125
LEAD-BANK 20336 0.8969 0.30 0.00 1.00 0.223
LEAD-INVEST 20336 0.0479 0.21 0.00 1.00 0.151
REPUTATION-LEAD 20336 0.2074 0.33 0.00 1.00 0.046
MARKETSHARE-LEAD 20336 0.0113 0.01 0.00 0.04 0.062
IMPORTANCE-LEAD 20336 0.1600 0.26 0.00 1.00 0.043
EXPERIENCE-LEAD 20336 478.5633 466.92 1.00 1811.00 0.042
INTENSITY-SYND 20336 132.9382 150.85 0.00 1874.00 0.148
DURATION-SYND 20336 70.2041 38.83 0.00 297.00 0.153
REPUTATION-SYND 20336 0.0845 0.09 0.00 1.00 0.063
MARKETSHARE-SYND 20336 0.0072 0.00 0.00 0.03 0.143
IMPORTANCE-SYND 20336 0.0788 0.09 0.00 0.75 0.099
EXPERIENCE-SYND 20336 301.3252 191.73 1.00 1431.00 0.091
ASYMMETRY-INTENSITY 20336 643.3279 530.41 0.00 1988.00 0.198
ASYMMETRY-DURATION 20336 159.3518 84.11 0.00 323.00 0.151
ASYMMETRY-REPUTATION 20336 0.4312 0.41 0.00 1.00 0.140
ASYMMETRY-MARKETSHARE 20336 0.0181 0.01 0.00 0.04 0.136
ASYMMETRY-IMPORTANCE 20336 0.3914 0.38 0.00 1.00 0.172
ASYMMETRY-EXPERIENCE 20336 803.8002 507.53 0.00 1810.00 0.121
INTERNATIONAL 20336 0.2432 0.43 0.00 1.00 0.030
LEAD-US 20336 0.6573 0.47 0.00 1.00 0.098
LEAD-JAPAN 20336 0.0278 0.16 0.00 1.00 0.071
LEAD-UK 20336 0.0498 0.22 0.00 1.00 0.013
SYND-US-CA 20336 0.6851 0.46 0.00 1.00 0.157
SYND-LATAMERICA 20336 0.0097 0.10 0.00 1.00 0.026
SYND-EUROPE 20336 0.1202 0.33 0.00 1.00 0.044
SYND-AFRICA-MIDDLE 20336 0.0088 0.09 0.00 1.00 0.039
SYND-ASIA 20336 0.1663 0.37 0.00 1.00 0.157

significantly loading on any other component, we can interpret the
six components as follows7:
1. Syndicate quality component (QUALITY): measures the quality
of syndicate members and their relationships with one another
as indicated by the length and the intensity (measured by the
average number of common deals between lenders) of their
past connections, as well as the average reputation (based on
the lenders’ league table ranking), market share, importance
in the loan market (based on volume ranking) and experience
(based on the number of previous transactions) of the syndicate
members. The stronger the past relationships between lenders,
the higher the trust and cohesion within the syndicate, which
can reduce coordination problems and lower information fric-
tions between members.
2. Syndicate heterogeneity component (HETEROGENEITY): mea-
sures the heterogeneity within the syndicate in terms of the
asymmetry of the intensity and duration of past alliances
between its members. It also captures the heterogeneity that
is due to the number of lenders and participants. The higher
the heterogeneity of the syndicate, the lower the cohesion
and concentration of the syndicate and the higher the informa-
tion asymmetries.
3. Lead arranger component (LEAD): measures the quality of the
syndicate’s lead arranger in terms of reputation, market share,
importance and experience. Given that the syndicated loan
market is one of repeat transactions, the higher the quality of
the lead arranger, the lower the information asymmetries
within the syndicate.
4. Syndicate geography component (GEOGRAPHY): measures the
geographical location of the lead arranger and the region of syn-
dication. Everything else held equal, a foreign lead arranger
increases information asymmetries.
5. Relationship component (RELATIONSHIP): measures the inten-
sity and duration of the relationship between the syndicate
members (leads and participants) and the borrower. Everything
else held equal, the stronger the relationship, the lower the
information asymmetry.
6. Institution type component (TYPE): indicates the type of lead
arranger in terms of its financial sector (bank or investment
firm). A bank typically has more experience in selecting and
monitoring borrowers, which will reduce information frictions.

7
Analysis of the principal components on the entire set of 40
As a robustness test, a principal component analysis is also conducted on the
quantiles of the structure variables instead of their actual measures (in Table 1) to
structure variables, including the four concentration variables,
obtain uniform units across the variables. Untabulated resulting components and shows that the second component is now related also to concen-
rotated factor optimal weights are relatively similar. tration variables (HH-INDEX and TOP3), while the GEOGRAPHY
Table 2
Rotated factor patterns for the six structure components. This table summarizes the rotated factor patterns for the six structure components retained in Section 3 of the paper. Rotation is performed with varimax. Panel A presents the
rotated factor patterns on the subset of 36 structure variables using the variables as defined in Appendix A and measured in Table 1. Panel B presents the rotated factor patterns for the entire set of 40 structure variables, including
concentration variables, using the variables as they are defined in Appendix A and measured in Table 1. In Panels A and B, variables with a loading greater than |0.40| for a specific component are shaded in gray. In Panel A, variables that
meaningfully load on a particular component and not on any other are bolded. Number of observations is 20,336 for Panel A and 7718 for Panel B.

C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451

1441
1442 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451
and the LEAD components simply change position (in terms of the
percentage of variance explained), as shown in Panel B of Table 2.
The remaining components are relatively similar. Interestingly, the
proportion of the variance explained by the second component is
similar in both cases (with or without concentration variables),
which may indicate that the two components, CONCENTRATION
and HETEROGENEITY, can act as substitutes.
3.3. Weighted factor-based scores
To use the components in subsequent analysis, we calculate
weighted factor-based scores. A factor-based score is a linear com-
posite of the variables that demonstrate meaningful loadings for
the component in question. Because they are not true principal
components, they can demonstrate nonzero correlation with one
another.8 However, two of the advantages of factor-based scores
are their tractability and interpretability given that they are based
on fewer distinct variables. Based on the meaningful loading vari-
ables for each component, which are arbitrarily defined as variables
that load with an optimal weight greater than |0.4| for a particular
component (identified in gray shading in Panel A of Table 2) while
not significantly loading on any other component (identified in bold
in Panel A of Table 2), variables are weighted by their loading factor
and added together to obtain factor-based scores.9 The following
artificial variables are thus created:
QUALITY W 1  INTENSITY  SYND þ W 2  DURATION  SYND þ W 3  MARKETSHARE  SYND
þW 4  IMPORTANCE  SYND þ W 5  EXPERIENCE  SYND
HETEROGENEITY W 1  LENDERS þ W 2  PARTICIPANTS þ W 3  ASYMMETRY  INTENSITY þ W 4  ASYMMETRY  DURATION
LEAD W 1  REPUTATION  LEAD þ W 2  MARKETSHARE  LEAD þ W 3  IMPORTANCE  LEAD þ W 4  EXPERIENCE  LEAD
GEOGRAPHY W 1  LEADS þ W 2  INTERNATIONAL þ W 3  LEAD  US þ W 4  LEAD  UK þ W 5  SYND
US  CA þ W 6  SYND  EUROPE þ W 7  SYND  ASIA
RELATIONS W 1  REL  PARTICIPANTS þ W 2  REL  LEAD þ W 3  DURATION  PARTICIPANTS þ W 4  DURATION  LEAD
TYPE W 1  LEAD  BANK þ W 2  LEAD  INVEST
where Wi is the weight of the variable in the component, as given in
Panel A of Table 2, with two exceptions. For the GEOGRAPHY and
TYPE components, positive and negative loadings make the interpre-
tation of the factor-based score difficult. To address this issue, we re-
code LEAD-US, SYND-US-CA and LEAD-INVEST to vary in the opposite
direction, while the remaining loadings are multiplied by 1 to ob-
tain positive weights everywhere.10 A high score for TYPE therefore
implies that the lead arranger is a bank and not an investment firm.
Similarly, the highest score for GEOGRAPHY would be for a structure
in which the lead is not from the US but from the UK and that is not
syndicated in the US or Canada but in Europe or Asia. For share-re-
lated variables, the HETEROGENEITY component is changed to a con-
centration component11:
CONCENTRATION ¼ W 1  LENDERS þ W 2  PARTICIPANTS þ W 3
 HH  INDEX þ W 4  TOP3 þ W 5
 ASYMMETRY  DURATION
8
A component score, by contrast, is a linear composite of the optimally-weighted
observed variables and includes all the variables. As an untabulated robustness check,
all the estimations and analyses are also conducted using component scores with
qualitatively similar results.
9
Weighted factor-based scores based solely on the first condition (i.e. weight
greater than |0.4|) yield very similar results.
10
Because rotated factor weights are based on the correlation matrix, the recoding
simply changes the signs of the weights and not their value.
11
Results are qualitatively similar if the two meaningful concentration variables are
added to the five original HETEROGENEITY components.
We recalculate the scores for the six components, including CON-
CENTRATION, using weights from Panel B of Table 2. Again, to sim-
plify the interpretation of the resulting factor-based score easier,
we recode HH-INDEX and TOP3 such that all the loadings are posi-
tive. A higher factor-based score therefore implies a lower concen-
tration index or top3 share, everything else held equal.
4. Analysis of structure components
4.1. Univariate analysis of the structure components
As discussed in Section 2, the loan syndicate is structured to re-
duce within-syndicate information asymmetry that can be related
to (lead-) lender–borrower information asymmetries. Panels A and
B of Table 3 show weighted factor-based scores for the six struc-
ture components (including the CONCENTRATION component or
not) identified in Section 3, conditional on two borrower character-
istics commonly associated with the level of borrower information
asymmetries: borrower’s opaqueness (measured by the absence of
a credit rating on the loan) and borrower’s home country’s state of
economic development (based on the IMF’s classification of emerg-
ing and developed countries). Opaque or emerging-country bor-
rowers are associated with lower quality syndicates and leads,
based on their QUALITY factor-based scores, but with syndicates
that are more homogeneous. Opaque and emerging-country
borrowers have weaker RELATIONS components and are associated
with more concentrated syndicates (i.e. lower CONCENTRATION
components), which is consistent with Sufi (2007). These results
suggest that the structure components are indeed related to infor-
mation asymmetries.
Panel C of Table 3 shows the values for the 6 structure compo-
nents, conditional on borrowers’ regions.12 US borrowers are asso-
ciated with syndicates of higher quality (456.33) than European
(396.43) or Asian (218.93) borrowers. They are also associated with
higher-quality leads. Syndicates for European borrowers are based
on weaker relationships between the borrower and lenders, and
are the most heterogeneous and diffuse. Asian borrowers are associ-
ated with the most homogeneous and concentrated syndicates and
with the lowest-quality syndicates and leads.
4.2. Multivariate analysis of loan spread
Studies of the impact of the syndicate structure on loan spreads
typically consider one or two structure measures at a time, which
does not capture the multidimensionality of syndicate structure.
Because structure variables are correlated, adding all or most rele-
vant variables induces important multicollinearity or overidentifi-
cation problems in any multivariate analysis. To address these
12
Samples are divided according to borrower region and not lead arranger region
because the latter is included in the structure components. However, all estimations
and analyses are also conducted using samples based on lead arranger region, with
very similar results (in the majority of cases, the region of the borrower is the same as
the lead arranger’s). Results are available upon request.
Table 3
Descriptive statistics for structure components, borrower-specific and loan-specific variables, conditional on borrower characteristics. This table summarizes the summary statistics for the structure component scores and factor-based
scores, borrower-specific and loan-specific variables used in the analyses throughout the paper. Component scores and weighted factor-based scores are defined in Section 3 of the paper. The remaining variables are defined in
Appendix B. N is the number of observations. Borrower industry, loan purpose, loan type, and year dummy variables are not reported to save valuable journal space.

Panel A: Structure components conditional on borrower transparence Panel B: Structure components conditional on borrower country ’s economic development
Equality of Developed-country borrowers Emerging-country borrowers Equality of
Rated borrowers Opaque borrower Means Variances Means Variances
N Mean Std. dev. N Mean Std. dev. t-Value F-value N Mean Std. dev. N Mean Std. dev. t-Value F-value
Components based on 36 structure variables
QUALITY 7089 477.575 281.11 8817 356.423 272.62 27.38 1.06 14,818 423.170 283.63 1088 236.746 205.22 28.06 1.91
HETEROGENEITY 7089 705.890 425.83 8817 468.279 393.92 36.16 1.17 14,818 587.805 426.50 1088 388.582 358.44 17.45 1.42
LEAD 7089 475.183 407.71 8817 325.268 352.48 22.47 1.34 14,818 404.421 390.00 1088 224.041 262.97 20.99 2.20
GEOGRAPHY 7089 1.189 1.31 8817 2.580 2.35 47.18 3.25 14,818 1.770 1.94 1088 4.538 2.18 40.76 1.26
RELATIONS 7089 37.882 42.90 8817 18.601 29.43 32.23 2.13 14,818 28.270 37.71 1088 12.540 27.50 17.69 1.88
TYPE 7089 1.535 0.45 8817 1.566 0.37 4.67 1.45 14,818 1.553 0.41 1088 1.547 0.35 0.52 1.33

C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451

Components based on 40 structure variables
QUALITY 2176 426.350 222.14 3726 299.712 242.98 20.4 1.20 5290 363.961 243.74 612 194.621 177.95 21.34 1.88
CONCENTRATION 2176 206.409 82.25 3726 145.772 83.99 26.96 1.04 5290 171.371 89.23 612 140.096 74.55 9.61 1.43
GEOGRAPHY 2176 1.288 1.66 3726 2.712 2.57 25.85 2.39 5290 1.891 2.16 612 4.746 2.59 26.27 1.43
LEAD 2176 425.511 371.66 3726 264.790 324.11 16.79 1.31 5290 340.405 357.34 612 182.648 250.11 14.03 2.04
RELATIONS 2176 48.017 52.99 3726 23.664 40.02 18.57 1.75 5290 34.545 47.19 612 16.201 38.88 10.79 1.47
TYPE 2176 0.775 0.17 3726 0.749 0.22 5.03 1.63 5290 0.768 0.19 612 0.679 0.30 7.08 2.59
All borrowers US Borrowers European borrowers Asian borrowers Non-US borrowers
N Mean Std. dev. N Mean Std. dev. N Mean Std. dev. N Mean Std. dev. N Mean Std. dev.
Panel C: Structure components conditional on borrower’s region
QUALITY 15,906 410.418 282.91 10,781 456.328 291.40 2084 396.432 224.54 2421 218.932 204.10 5125 313.842 236.72
HETEROGENEITY 15,906 574.178 425.17 10,781 614.113 439.85 2084 656.967 368.77 2421 330.707 320.95 5125 490.170 379.05
LEAD 15,906 392.082 385.35 10,781 456.869 413.06 2084 353.752 293.12 2421 151.548 207.89 5125 255.797 273.10
GEOGRAPHY 15,906 1.960 2.08 10,781 0.887 0.58 2084 4.529 2.44 2421 4.310 1.99 5125 4.217 2.27
RELATIONS 15,906 27.194 37.31 10,781 32.482 39.94 2084 13.299 18.16 2421 18.148 34.17 5125 16.069 27.96
TYPE 15,906 1.552 0.41 10,781 1.543 0.42 2084 1.557 0.44 2421 1.585 0.31 5125 1.573 0.37
Factor-scores with concentration variables:
QUALITY 5902 346.402 243.29 3774 405.904 250.04 493 359.126 179.79 1519 193.361 170.73 2128 240.875 189.44
CONCENTRATION 5902 168.128 88.33 3774 169.563 93.36 493 213.532 86.08 1519 147.603 67.84 2128 165.585 78.58
GEOGRAPHY 5902 2.187 2.38 3774 0.960 0.60 493 5.096 3.40 1519 4.207 2.45 2128 4.365 2.76
LEAD 5902 324.046 351.05 3774 395.116 379.83 493 333.592 280.74 1519 143.066 210.60 2128 198.004 247.25
RELATIONS 5902 32.643 46.73 3774 39.512 48.95 493 14.766 20.27 1519 21.747 43.51 2128 20.460 39.71
TYPE 5902 0.759 0.21 3774 0.763 0.20 493 0.811 0.05 1519 0.731 0.25 2128 0.752 0.22
Borrower and loan variables
SIZE 15,906 6.932 1.63 10,781 6.847 1.55 2084 7.483 1.77 2421 6.726 1.75 5125 7.112 1.78
RELAMT 15,906 0.434 1.56 10,781 0.451 1.67 2084 0.768 1.91 2421 0.111 0.32 5125 0.398 1.29
LEVERAGE 15,906 1.933 27.40 10,781 1.716 32.57 2084 2.621 7.87 2421 2.208 11.89 5125 2.389 9.89
PROFIT 15,906 36.987 117.08 10,781 33.837 129.30 2084 58.327 33.80 2421 30.738 118.20 5125 43.612 85.51
DEBT A 15,906 0.628 0.26 10,781 0.631 0.28 2084 0.663 0.27 2421 0.603 0.18 5125 0.621 0.22
OPAQUE 15,906 0.554 0.50 10,781 0.397 0.49 2084 0.799 0.40 2421 0.997 0.06 5125 0.885 0.32
ECONDEV 15,906 9.988 0.69 10,781 10.242 0.06 2084 10.046 0.25 2421 9.044 1.18 5125 9.455 1.02
EMERGING 15,906 0.068 0.25 10,781 0.000 0.00 2084 0.001 0.03 2421 0.314 0.46 5125 0.212 0.41
LEGAL 15,906 0.790 0.41 10,781 1.000 0.00 2084 0.368 0.48 2421 0.283 0.45 5125 0.349 0.48
MTY 15,906 3.719 0.70 10,781 3.680 0.70 2084 3.880 0.72 2421 3.773 0.67 5125 3.802 0.70
AMT 15,906 18.943 1.92 10,781 19.305 1.22 2084 20.179 1.52 2421 16.225 2.45 5125 18.183 2.73
TRANCHES 15,906 1.970 1.29 10,781 1.830 1.05 2084 2.798 2.08 2421 1.924 1.13 5125 2.264 1.65
MULT-TRANCHES 15,906 0.542 0.50 10,781 0.518 0.50 2084 0.677 0.47 2421 0.560 0.50 5125 0.594 0.49
SECURED 15,906 0.400 0.49 10,781 0.482 0.50 2084 0.180 0.38 2421 0.259 0.44 5125 0.228 0.42
COVENANT 15,906 0.485 0.50 10,781 0.681 0.47 2084 0.065 0.25 2421 0.054 0.23 5125 0.073 0.26

1443
(continued on next page)
1444 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451

issues, we use a different methodology in which each loan’s

weighted factor-based score for each of the six components is
Std. dev.

added as an explanatory variable in a multivariate model of loan

31.21
10.10

3.28
3.84
0.12

0.47
0.46
spread. As mentioned above, syndicate structure can affect loan
spread either through a diversification or information asymmetry
premium. With the exception of HETEROGENEITY and CONCENTRA-
Non-US borrowers

21.700

TION, which can be related to both types of syndicate-related

Mean

5.554

2.439
0.987

0.319
0.294
2.076

premium, the remaining five components should only be related di-

rectly to loan spread through the information asymmetry premium.
Therefore, if borrower and loan characteristics are controlled for,
5125
5125
5125
5125
5125
5125
5125

the sign, magnitude, and significance of the components’ coeffi-

N

cients should reflect their impact on the information asymmetry

premium.
Std. dev.

The appropriate model to consider both the syndicate structure

36.24
12.75

4.18
4.94
0.47
0.45

and loan spread must be determined, because both variables have

0.08

been modeled in numerous ways in the literature. The relationship

between loan spreads and the syndicate structure is still unclear.
While the majority of studies examine unilateral relations (see,
24.895
Mean

2.589
3.237
0.993

7.507
0.319
0.284
Asian borrowers

for example, Angbazo et al., 1998), Ivashina (2009) provides evi-

dence that endogenous relationships capture the simultaneous
determination of the spread and the structure more appropriately,
at least as measured by the lead arranger share. Thus, the research
2421
2421
2421
2421
2421
2421
2421

strategy adopted herein is to study loan spread and syndicate

N

structure both separately and simultaneously. The general form

of the spread model examined is the following:
Std. dev.

25.01

X X
6.27

1.64
1.73
0.15

0.47
0.47

6 N
SPREADl ¼ b0 þ bi  STRUCTUREi;l þ bi  X i;l þ el ð1Þ
i¼1 i¼7
European borrowers

In model (1), SPREADl is the all-in loan spread over LIBOR for loan l,
18.832
Mean

1.512
0.976

3.870
0.332
0.317
1.430

STRUCTUREi,l is the weighted factor-score for one of the six syndi-

cate structure components identified previously for loan l, and Xj,l
is one of the loan-specific, borrower-specific or calendar control
2084
2084
2084
2084
2084
2084
2084

variables for loan l. To control for the most potential risk factors,
including loan type and purpose, observations are taken at the
N

facility level.13 Based on existing theories and the availability of

variables, we use the following set of exogenous variables, where
Std. dev.

the variable definitions are provided in Appendix B

30.68
2.65

2.22
0.34
0.29
2.03
0.04

Xð1Þ ¼ ½SIZE; RELAMT; LEVERAGE; PROFIT; DEBTA; OPAQUE; ECON

 DEV; EMERGING; LEGAL; MTY; AMT; TRANCHES; MULT
26.197
Mean

3.189

2.459
0.999

0.132

2.802
0.094

 TRANCHE; SECURED; COVENANT; SENIOR; BORROWER

US Borrowers

 COUNTRY; BORROWER
 INDUSTRY; TYPE; PURPOSE; YEAR
10,781
10,781
10,781
10,781
10,781
10,781
10,781

Descriptive statistics for the components and the control vari-

N

ables used in the regressions are available in Panel C of Table 3,

while Table 4 shows the results for the evaluation of model (1)
Std. dev.

using the structure components as exogenously-determined

30.92
6.23

2.51
2.85
0.39
0.37
0.07

right-hand side variables.14 Preliminary results in Section 4.1 show

that syndicate structures are intrinsically different across regions,
which may have an impact on the effect of the structure on loan
spread. To test this hypothesis, we divide the sample into the three
24.748
Mean

3.951

2.336
2.685
0.995

0.193
0.158

largest syndicated loan markets: US, Europe, and Asia.

All borrowers

Panels A and B of Table 4 show results for US borrowers. Panel A

shows results using the components based on 36 structure vari-
15,906
15,906
15,906
15,906
15,906
15,906
15,906

ables, excluding concentration variables, while Panel B includes

the concentration variables in the components. The QUALITY com-
N

ponent is significantly negatively related to the spread, with a coef-

ficient of 0.028 in Panel A and 0.038 in Panel B, indicating that
Table 3 (continued)

stronger cohesion and higher-quality syndicate members can

POOL-LENDERS
POOL-LEADS
FIRST-SYND

INFO-SYND
FIRST-ALL

INFO-ALL

13
Untabulated robustness tests show that results are similar when done on deal
SENIOR

level observations.
14
To formally detect multicollinearity in all the models used in the study, Variable
Inflation Factors (VIF) are calculated. None of the VIFs are problematic.
Table 4
Multivariate regressions for loan spreads for US and European borrowers. This table summarizes the results for regression model (1) when loan spreads are regressed against the six structure components measured with weighted factor-
based scores, controlling for loan-specific and borrower-specific variables. The structure components are defined on the subset of 36 structure variables in Panels A, C, D and F and on the subset of 40 structure variables in Panels B, E and
H. Multivariate regressions are estimated using OLS and t-values are corrected for heteroskedasticity.

US borrowers European borrowers US and European borrowers

Panel A – With Panel B – With Panel C – With Panel D – With Panel E – With Panel F – With Panel G – Without Panel H – With
components based components based components based components based components based components based components components based
on 36 structure on 40 structure on 36 structure on 36 structure on 40 structure on 36 structure on 40 structure
variables variables variables variables variables variables variables
Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value

C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451

INTERCEPT 872.437 22.81*** 714.395 10.86*** 857.704 22.42*** 424.246 3.73*** 1323.229 6.36*** 415.125 3.63*** 538.651 5.08*** 906.827 4.89***
QUALITY 0.028 6.48*** 0.038 5.55*** 0.016 2.98*** 0.013 1.27 0.036 1.79* 0.055 2.29** 0.033 5.06***
HETEROGENEITY 0.041 12.62*** 0.045 11.6*** 0.009 1.25 0.018 1.19
CONCENTRATION 0.079 3.29*** 0.049 0.94 0.069 3.17***
LEAD 0.014 4.82 *** 0.010 2.23** 0.014 3.66*** 0.021 2.80*** 0.005 0.40 0.043 2.88*** 0.009 2.19**
GEOGRAPHY 3.735 2.04** 2.445 1.01 1.955 0.90 1.117 1.05 1.405 1.20 2.658 1.61 1.430 1.34
RELATIONS 0.082 2.92*** 0.002 0.06 0.055 1.70* 0.047 0.38 0.219 1.30 0.150 0.69 0.004 0.12
TYPE 33.903 13.11*** 68.547 9.36*** 21.683 6.99*** 23.156 4.84*** 11.473 0.20 18.940 1.98** 68.491 9.45**
SIZE 8.252 6.02*** 10.232 4.48*** 9.184 6.70*** 17.367 8.05*** 19.239 5.78*** 17.281 7.98*** 12.915 11.07*** 11.981 6.70***
RELAMT 0.354 0.56 2.288 0.74 0.269 0.43 3.607 2.94*** 1.342 1.28 3.763 3.07*** 0.914 1.58 1.166 0.99
LEVERAGE 0.057 1.85* 0.009 0.17 0.055 1.81* 0.435 1.69* 2.069 2.71*** 0.413 1.59 0.059 1.91* 0.004 0.08
PROFIT 0.033 4.25*** 0.024 3.13*** 0.032 4.10*** 0.242 3.72*** 0.265 2.38** 0.239 3.68*** 0.033 4.29*** 0.024 3.15***
DEBT A 83.241 21.26*** 77.545 13.22*** 82.047 20.93*** 17.681 2.18** 19.922 1.26 17.123 2.10** 79.097 21.98*** 72.126 12.94***
OPAQUE 3.623 1.49 0.249 0.07 53.943 6.00*** 14.509 2.54** 10.985 1.27 6.129 0.27 5.762 2.53** 0.204 0.06
ECONDEV 30.355 2.95*** 13.144 0.75 29.867 2.90*** 34.829 3.42*** 16.198 0.92
LEGAL 7.756 1.56 4.301 0.52 8.217 1.65* 16.020 3.26*** 11.582 1.38
QUALITY  OPAQUE 0.027 3.30*** 0.048 1.87*
HETEROGENEITY  OPAQUE 0.009 1.65* 0.008 0.51
LEAD  OPAQUE 0.004 0.72 0.029 1.68*
GEOGRAPHY  OPAQUE 7.287 1.93* 2.436 1.27
RELATIONS  OPAQUE 0.108 1.73* 0.174 0.68
TYPE  OPAQUE 37.017 7.02*** 6.670 0.61
EUROPE 28.97 5.83*** 14.91 1.58
Loan term variables Yes Yes Yes Yes Yes Yes Yes Yes
Borrower country fixed effects No No No Yes Yes Yes Yes Yes
Borrower Industry fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Loan type fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Loan purpose fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes
N 10,781 3774 10,781 2084 493 2084 12,865 4267
Adj. R2 0.5058 0.5250 0.5081 0.6418 0.7652 0.6468 0.4969 0.5396
F-value 257.60*** 100.30*** 232.95*** 83.93*** 36.64*** 75.97*** 318.64*** 112.11***

N is the number of observations used in each model specification. Borrower industry, loan purpose, loan type, loan term and year dummy variables are not reported to save valuable journal space.
*
Indicates significance at the 10% levels.
**
Indicates significance at the 5% levels.
***
Indicates significance at the 1% levels.

1445
1446 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451
diminish the information asymmetry premium. A higher HETERO-
GENEITY component in the syndicate is also related to a lower
spread. Although higher heterogeneity implies lower concentra-
tion and cohesion, which are associated with adverse selection
and moral hazard problems, it can also imply that the loan is dif-
fuse among many lenders. The negative sign indicates that the re-
lated information asymmetry premium is more than offset by the
reduction in the diversification premium, which is consistent with
Ivashina (2009). Moreover, the CONCENTRATION coefficient in Pa-
nel B is negative, indicating that larger and more dispersed (less
concentrated) syndicates are related to lower spreads. The positive
LEAD component coefficients of 0.014 in Panel A and 0.01 in Panel
B show that more reputable and more-experienced lead arrangers
are associated with higher spreads, which is consistent with the
certification premium observed by Cook et al. (2003). US borrowers
pay higher spreads when their lead arranger is foreign or when
their loan is syndicated in Europe or Asia as opposed to the US
(captured by the GEOGRAPHY component), which is evidence of a
domestic bias. However, this component is not significant in Panel
B when concentration variables are added to the components.
Finally, the TYPE component indicates that bank-led syndicates
(investment firm-led) are associated with lower (higher) spreads,
which is consistent with Harjoto et al. (2006) and Nandy and Shao
(2010), discussed earlier. The remaining coefficients are in line
with the literature; larger, more profitable and less indebted
(measured with DEBTA) borrowers are related to lower spreads.
To further test whether the components are related to informa-
tion asymmetries, we include interactive variables combining each
of the components with OPAQUE, a measure of borrower opaque-
ness that is typically related to lender–borrower information
asymmetries, in model (1). Results are available in Panel C and
show that borrower information asymmetries are related to all
the components except LEAD. The quality of the syndicate has a
stronger impact on loan spreads when the borrower is opaque
(0.027) than for transparent borrowers (0.016). A similar phe-
nomenon occurs for RELATIONS and TYPE. The heterogeneity of
the syndicate is positively (negatively) related to the loan premium
when borrowers are opaque (transparent), which indicates that a
more homogeneous (concentrated) syndicate is beneficial when
the borrower needs more due diligence and monitoring. LEAD is
only a significant (and positive) determinant of the spread for
transparent borrowers, which may indicate that the certification
premium is offset by a reduction in syndicate information asym-
metries for opaque borrowers only. These results support the the-
ories and studies on syndicate structure discussed in Section 2.
They show that syndicate-related premiums are explained not only
by lead arranger concentration but also, simultaneously, by the
quality and cohesion of the entire syndicate, the characteristics
of the lead arranger, the location of the lead and the syndication
market, lender–borrower relationships and the type of lender. Fur-
ther, borrower information asymmetries are related not only to the
concentration of the syndicate, but also to the other structure
components.
4.3. European and Asian markets
Panels D–F present a similar analysis on a sub-sample of Euro-
pean borrowers. According to panel D (Panel E), only LEAD (QUAL-
ITY) is a significant determinant of the spread for European
borrowers. When conditioning on the opaqueness of the borrower
in panel F, the quality and cohesion of the syndicate (QUALITY) and
its lead arranger’s reputation (LEAD) are negatively related to the
spread for opaque borrowers, and positively related to spread for
transparent borrowers. These results show that that spread models
differ according to the region and that loan spreads on the Euro-
pean market are less sensitive to syndicate structure than on the
American market. Depending on the distribution of the sample,
one should therefore be careful when extending observations or
conclusions based on US samples to other countries.
These results are interesting in the context of the European
pricing discount observed on the syndicated loan market. Carey
and Nini (2007) observe a discount of 30 bps in Europe and cannot
explain it with differences in borrower, loan and lender character-
istics. The authors use syndicate size as a proxy for the syndicate
structure along with the fraction of lenders with each nationality.
Houston et al. (2007) conclude that North American borrowers
with European lead arrangers can get a 40 bps discount. The
authors include the number of leads and domestic lead indicators
as syndicate structure proxies. Although these syndicate character-
istics are certainly important factors in determining loan spreads,
they may be a poor proxy for the multidimensional syndicate
structure. Given the large differences in at least five of the six
structure components between European and US borrowers in pa-
nel C of Table 3, different syndicate structures could help explain
the European discount. This is formally verified in panels G and
H, in which we combine the two sub-samples of US and European
borrowers and include a dummy variable to capture a European
effect in the model.15 In Panel G, when no structure component is
included, European borrowers are associated with lower-spread
loans (28.97 bps lower). In Panel H, with the inclusion of structure
components, the European pricing effect disappears. Further, it is
not captured by the GEOGRAPHY component of the syndicate struc-
ture.16 The pricing discount observed in previous papers is therefore
potentially attributable to different syndicate structures in Europe,
such as more diffuse or lower quality syndicates and lead arrangers
(see Table 3). These differences can lead to lower concentration and
certification premiums.
The multivariate analysis of loan spreads for Asian borrowers is
presented in Table 5. The sensitivity of loan spreads to structure
components is also different from that of US borrowers. Loan
spreads in Asia are related to the heterogeneity or concentration
of the syndicate only weakly or not at all, and are not influenced
by the reputation of the lead arranger, which may be related to
the low LEADS component in Asia (see Table 3). Further, Asian
borrowers pay lower spreads when they deal with a non-US lead
or have their loan arranged in Europe or Asia. From the coefficients
in panel C, we see that opaque Asian borrowers benefit much more
from a higher quality syndicate or lead arranger than US or Euro-
pean borrowers, while transparent borrowers, who do not benefit
as much from the reduction in syndicate information asymmetries,
are charged a higher certification premium.17 Like their US counter-
parts, opaque Asian borrowers pay a higher premium with a more
heterogeneous syndicate. Nonetheless, contrary to US borrowers,
Asian opaque (transparent) borrowers pay a higher (lower) spread
when they have more and longer past relationships with syndicate
members.
Using a similar methodology as for the European discount, we
test for an Asian discount. Results are available in panels D and
E. Results in panel D shows that an Asian effect is observed and
is even larger than the European pricing effect observed previously,
with a lower spread of more than 107 bps for Asian borrowers as
opposed to US borrowers. The inclusion of structure components
in panel E reduces the effect to 88.86 bps but does not remove
its significance. Based on our results, it therefore appears that a
15
Because spread models are region-dependent, results using an OLS regression
may be biased. However, because the European discount was previously inferred with
such a model on international samples, we wish to keep the same methodology here.
16
Another specification in which the GEOGRAPHY component is removed from the
model generates similar results.
17
Interactive variables with GEOGRAPHY and LEADS are associated with high
multicollinearity and are removed from the model to ensure valid inference.
 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451 1447

Table 5
Multivariate regressions for loan spreads for Asian borrowers. This table summarizes the results for regression model (1) when loan spreads are regressed against the six structure
components measured with weighted factor-based scores, controlling for loan-specific and borrower-specific variables. The structure components are defined on the subset of 36
structure variables in Panels A and C and on the subset of 40 structure variables in Panels B and E. Multivariate regressions are estimated using OLS and t-values are corrected for
heteroskedasticity.

Asian borrowers US and Asian borrowers

Panel A – With Panel B – With Panel C – With Panel D – Without Panel E – With
components based on components based on components based on structure components components based on
36 structure variables 40 structure variables 36 structure variables 40 structure variables
Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value Coeff. t-Value
*** *** ***
INTERCEPT 303.577 7.26 202.355 3.38 115.864 0.80 506.345 13.39 410.686 9.60***
QUALITY 0.021 2.21** 0.037 2.65*** 2.243 3.37*** 0.045 7.35***
HETEROGENEITY 0.011 1.72* 1.855 4.15***
CONCENTRATION 0.044 1.35 0.122 6.43***
LEAD 0.005 0.68 0.012 1.18 1.394 3.61*** 0.007 1.68*
GEOGRAPHY 2.637 3.10*** 2.053 2.28** 2.301 2.70*** 0.537 0.60
RELATIONS 0.118 2.67*** 0.073 1.62 12.418 2.61*** 0.041 1.56
TYPE 21.160 4.47*** 17.494 2.27** 21.638 4.53*** 55.308 9.80***
SIZE 9.980 7.84*** 7.581 4.31*** 9.985 7.88*** 16.695 16.75*** 12.320 9.16***
RELAMT 29.371 5.16*** 43.491 3.88*** 29.640 5.23*** 0.134 0.22 5.071 1.87*
LEVERAGE 0.094 0.81 0.072 0.60 0.091 0.79 0.056 1.85* 0.005 0.09
PROFIT 0.001 0.09 0.064 0.71 0.001 0.06 0.026 3.58*** 0.029 3.77***
DEBT A 28.614 3.19*** 24.790 1.99** 28.741 3.22*** 93.109 25.03*** 81.453 15.27***
OPAQUE 103.773 4.27*** 6.453 0.16 322.239 2.27** 1.467 0.62 4.581 1.46
ECONDEV 1.740 0.89 5.010 1.83* 1.897 0.97 2.889 1.09 4.964 1.86*
EMERGING 26.806 4.54*** 34.978 4.58*** 26.827 4.57*** 41.088 5.76*** 27.835 3.77***
LEGAL 0.206 0.05 2.230 0.40 0.559 0.14 30.630 5.50*** 11.056 1.89*
QUALITY  OPAQUE 2.264 3.40***
HETEROGENEITY  OPAQUE 1.844 4.13***
LEAD  OPAQUE 1.389 3.59***
RELATIONS  OPAQUE 12.300 2.59***
ASIA 107.10 18.19*** 86.86 10.80***
Loan term variables Yes Yes Yes Yes Yes
Borrower country fixed effects No No No Yes Yes
Borrower industry fixed effects Yes Yes Yes Yes Yes
Loan type fixed effects Yes Yes Yes Yes Yes
Loan purpose fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
N 2421 1519 2421 13,202 5293
Adj. R2 0.2183 0.2000 0.2261 0.4437 0.4468
F-value 16.02*** 9.43*** 15.43*** 264.20*** 93.92***

N is the number of observations used in each model specification. Borrower industry, loan purpose, loan type, loan term and year dummy variables are not reported to save
valuable journal space.
*
Indicates significance at the 10% levels.
**
Indicates significance at the 5% levels.
***
Indicates significance at the 1% levels.

pricing discount, which cannot be explained by different syndicate
structures, persists for Asian borrowers.18
4.4. Complementary results
We re-examine the results by taking into account bilateral rela-
tions not only between structure and spread but also between
structure components. Untabulated results using endogenous
models show that the signs of the relationships between the
components and the spread are similar to those using an OLS esti-
mation of model (1), with the exception of the geography compo-
nent, which is no longer a direct significant determinant of spread,
but rather an indirect determinant through its influence on HETER-
OGENEITY, CONCENTRATION, or RELATIONS. Overall, the effect of the
structure components on loan spreads is generally stronger with
the endogenous model. Three components are endogenously
18
The discount appears to be present for most countries in Asia, including Japan,
although some countries have very few observations from which strong inferences
can be drawn. Dividing the sample into Asian crisis (1997) and post-crisis periods or
winsorizing the data (at the 1% and 5% percentiles) to remove outliers do not change
the results.
determined with loan spread: heterogeneity (or concentration),
relations and quality. Results also show that structure components
can also be related to other components. For instance, a higher
RELATIONS component is associated with higher-quality and more
heterogeneous syndicates, while the presence of a bank as lead
arranger increases the heterogeneity and lowers the concentration
of the syndicate. It is also associated with stronger relationships.19
5. Conclusion
This paper is the first comprehensive study of all dimensions of
a syndicate structure. We start with a very large set of 40 struc-
ture-related variables and use principal component analysis to
identify six principal components of the syndicate structure, there-
by providing a new approach to characterize and quantify the mul-
tidimensional structure of a syndicate. We can interpret the six
components as the quality of the syndicate, the heterogeneity of
its members or share concentration, the characteristics of the lead
arranger, the geography of the syndication and its lead arranger,
19
Untabulated results using endogenous models for European and Asian borrowers
are also consistent with previous results.
1448 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451

the average relations between the borrower and the syndicate
members, and the lenders’ institution types.
In univariate settings, we can associate opaque or emerging-
country borrowers with lower quality syndicates and leads, weaker
previous relationships with the lenders and more homogeneous or
concentrated syndicates. We re-examine the impact of syndicate
structure on loan spreads using multivariate regressions and find
that the six structure components are significant determinants of
loan spreads, either directly or indirectly through their impact on
other structure components. Consistent with agency theory, high-
er-quality syndicates with stronger cohesion can diminish the infor-
mation asymmetry premium. Heterogeneous or less concentrated
syndicates are related to lower spreads, which is consistent with a
reduction of the diversification premium. In keeping with certifica-
tion premium theory, reputable and more-experienced lead arrang-
ers are associated with higher spreads. However, results show that
lead quality is a significant (and positive) determinant of spread
only for transparent borrowers. For opaque borrowers, the benefits
of a higher-quality lead offset the certification premium. Results
from the endogenous estimation of structure components show
that three structure components have bilateral relationships. We
demonstrate that components interact with each other to ulti-
mately affect loan spreads. For instance, in accordance with home
bias theory, US borrowers pay higher spreads with a foreign lead
or syndication region. However, when structure components are
modeled endogenously with the spread, the geography component
is no longer a direct determinant of loan spreads but rather an
indirect determinant through its influence on other structure
components.
Structure components also appear to differ regionally. US bor-
rowers are associated with syndicates and leads of higher quality
than European or Asian borrowers. Syndicates for European bor-
rowers are based on weaker lender–borrower relationships and
are the most heterogeneous and diffuse. Asian borrowers are asso-
ciated with the most homogeneous and concentrated syndicates,
and with the lowest-quality syndicates and leads. Multivariate
spread models also differ according to borrower region. Loan
spreads are less sensitive to syndicate structure on the European
market than on the American market. However, opaque Asian bor-
rowers benefit much more from a higher quality syndicate or lead
arranger than their US or European counterparts. Finally, we can
explain the European pricing discount observed in the literature
by different syndicate structures for European borrowers than
for US borrowers, such as more diffuse syndicates and lower-
quality syndicates and lead arrangers. These differences can in
turn lead to lower concentration and certification premiums,
respectively.
Acknowledgements
Special thanks to an anonymous referee for many constructive
comments and suggestions, which helped us improve the paper.
We also thank Stefanie Kleimeier and other participants at the
2011 FMA European Conference and PBFEAM Conference. Finally,
we would like to acknowledge the Research Group in Applied Fi-
nance (GReFA), the Fonds de recherche sur la société et la culture
(FQRSC) and the Institut de Finance Mathématique de Montréal
(IFM2) for their financial support. The usual disclaimer applies.

Appendix A. Definitions of syndicate structure variables

This appendix describes the different syndicate structure variables used in the principal component analysis. Unless otherwise men-
tioned, variables are either from Dealscan (loan terms, syndicate structure, and lenders) or from Compustat Global (borrower
characteristics).

Variable Definition

(a) Variables related to the syndicate size

LENDERS Total number of distinct lenders in the syndicate.
LEADS Total number of lead arrangers in the syndicate. Lenders are considered in the lead
arranger category if they get lead arranger credit from Dealscan.
PARTICIPANTS Total number of participants (non lead) in the syndicate.
TYPES Total number of distinct financial institution types represented by members of the
syndicate (e.g. if the syndicate involves only commercial banks, then the variable is equal
to 1; if the syndicate involves commercial banks and insurance companies, then the
variable is equal to 2). Industries are grouped into five categories: banks, insurance
companies, investment banks, funds and other. The variable proxies for syndicate
heterogeneity.
COUNTRIES Total number of distinct countries represented by the members of the syndicate (e.g. if the
syndicate involves only US lenders, then the variable is equal to 1; if the syndicate involves
lenders from the US and UK, then the variable is equal to 2). The variable proxies for
syndicate heterogeneity.

(b) Variables related to share concentration

LEAD EXPOSURE Ratio of the total loan amount to the lead arranger’s assets.
HH-INDEX Herfindahl–Hirschman index as measured by the sum of the squares of the loan share of
each individual lender in the syndicate at loan origination.
LEAD-SHARE Share of the loan retained by the lead arranger at loan origination. If there is more than one
lead arranger, it is the total sum of shares they detain.
TOP3-SHARE Sum of share held by the lenders with the three largest shares at loan origination.
(continued on next page)
 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451 1449

Appendix A (continued)

Variable Definition

(c) Variables related to the relationship between the lenders and the borrower
REL-PARTICIPANTS Average number of past loans in the 5-year period prior to the deal active date with each
lender in the syndicate
REL-LEAD Average number of past loans in the 5-year period prior to the deal active date with the
lead arranger(s).
DURATION-PARTICIPANTS Average length of relationship between the borrower and each lender in the syndicate,
measured in number of months between the first deal and current deal active date,
DURATION-LEAD Length of relationship between borrower and lead arranger, measured in number of
months since first deal.

(d) Variables related to the institution type of the lead arranger

LEAD-BANK One if the main lead arranger is a bank, 0 otherwise. If there is more than one lead arranger
for the deal, the main lead arranger is identified as the one with the largest share. When
lender share is not available, the main lead bank is identified with the lender role within
the syndicate.
LEAD-INVEST One if the main lead arranger is an investment firm, 0 otherwise.

(e) Variables related to the quality of the entire syndicate

REPUTATION-LEAD Inverse of the lead arranger’s ranking in terms of league table credit during the previous
year.
MARKETSHARE-LEAD Ratio of the total amount of loans arranged by the lead arranger in the previous year to the
total volume of loans arranged in the previous year.
IMPORTANCE-LEAD Inverse of the lead arranger’s ranking in terms of volume. If there is more than 1 lead
arranger for the deal, the lead arranger with the best ranking is taken.
EXPERIENCE-LEAD Total number of loans arranged by the lead arranger in the previous year.
INTENSITY-SYND Average number of past common deals in the 5-year period prior to the deal active date
between each pair of lenders in the syndicate.
DURATION-SYND Average length of relationship between all the pairs of lenders in the syndicate, measured
in number of months between the first common deal and current deal active date.
REPUTATION-SYND Average of the inverse rankings of syndicate lenders in terms of league table credit during
the previous year.
MARKETSHARE-SYND Average market shares (measured by the ratio of total amount of loans by the lender in the
previous year to the total volume of loans arranged in the previous year) of syndicate
lenders.
IMPORTANCE-SYND Average of the inverse rankings of syndicate lenders in terms of volume during the
previous year.
EXPERIENCE-SYND Average number of deals by syndicate lenders in the previous year.
ASYMMETRY-INTENSITY Range in the number of past common deals in the 5-year period prior to the deal active
date between each pair of lender in the syndicate.
ASYMMETRY-DURATION Range in the length of relationship between all the pairs of lenders in the syndicate
(measured in number of months).
ASYMMETRY-REPUTATION Range in the inverse rankings of syndicate lenders in terms of league table credit during
the previous year.
ASYMMETRY-MARKETSHARE Range in the market shares of syndicate lenders during the previous year.
ASYMMETRY-IMPORTANCE Range in the inverse rankings of syndicate lenders in terms of volume during the previous
year.
ASYMMETRY-EXPERIENCE Range in the number of deals by syndicate lenders in the previous year.

(f) Variables related to the geography of the lead or the syndication

INTERNATIONAL One if the borrower is from the same country as the lead arranger, 0 otherwise.
LEAD- COUNTRY One if the main lead arranger is from a specific country, 0 otherwise. Three countries are
considered: US (LEAD-US), Japan (LEAD-JAPAN) and UK (LEAD-UK).
SYND-REGION Set of five dummy variables used to capture where the syndicate is arranged. The regional
dummies are for US & CANADA (SYND-US-CA), Latin America (SYND-LAT-AMERICA),
Western Europe (SYND-EUROPE), Africa and Middle East (SYND-AFRICA-EAST) and Asia/
Pacific (SYND-ASIA).
1450 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451

Appendix B. Definitions of control variables used in the multivariate models analysis

This appendix describes the different structure variables used in the multivariate models. The variables are divided into two categories:
(i) borrower-specific variables and (ii) loan-specific variables. Unless otherwise mentioned, variables are either from Dealscan (loan terms,
syndicate structure, and lenders) or from Compustat Global (borrower characteristics).

Variable Definition

(i) Borrower-specific variables:

SIZE Log of the inflation-adjusted US dollar book value of the assets of the borrower observed at the nearest date
before the loan active date, adjusted using the Consumer Price Index (CPI).
RELAMT Ratio of the loan amount to borrower size.
LEVERAGE Borrower’s debt-to-equity ratio observed at the nearest date before the loan active date.
PROFIT Borrower’s return on equity (ROE) observed at the nearest date before the loan active date.
OPAQUE One if the borrower is unrated, 0 otherwise.
ECONDEV Borrower’s home country’s level of economic development as measured by the per capita GNP obtained from
the International Monetary Fund.
EMERGING One if the borrower’s home country is considered to be emerging, 0 otherwise. Identification of emerging
countries is from the International Monetary Fund.
LEGAL One if the borrower’s home country’s legal system is civil law, 0 if common law. To facilitate the interpretation
of the coefficients and to limit the number of dummy variables, countries that are categorized as socialist are
removed from these estimations.
POOL-LENDERS Number of distinct lenders that were involved in loans with the borrower in the previous five years.
POOL-LEADS Number of distinct lead arrangers that were involved in loans with the borrower in the previous five years.
FIRST-SYND One if the borrower is tapping the syndicated loan market for the first time, 0 otherwise.
FIRST-ALL One if the borrower has never borrowed through any distribution method recorded in the LPC database, in the
five-year period prior to the deal active date, 0 otherwise
INFO-SYNDICATION Number of times that the borrower has borrowed on the syndicated loan market through syndications during
the five-year period prior to the active date of the deal (based only on the entries in the LPC database).
INFO-ALL Number of times that the borrower has borrowed through any distribution method recorded in the LPC
database, in the five-year period prior to the deal active date.
INDUSTRY Set of eight dummy variables based on the four-digit SIC code classification of the borrower’s industry:
agriculture, forestry and fishing, construction, finance, insurance and real estate, manufacturing, mining, retail
trade, services, and transportation, communications, etc. Robustness tests using a sub-sample of non-financial
borrowers yield similar results.
EUROPE One if the borrower is from Europe, 0 otherwise.
ASIA One if the borrower is from Asia, 0 otherwise.

(ii) Loan-specific variables:

SPREAD Total (fees and interest) annual spread paid over LIBOR for each dollar drawn down from the loan net of
upfront fees.
MTY Natural logarithm of the maturity of the loan as measured by the number of months until loan expiration.
AMT Natural logarithm of the deal amount in US dollars adjusted for inflation using the CPI between 1994 and
2009. In the case of multiple-facility deals, AMT, MTY, TYPE and PURPOSE take the values corresponding to the
facility with the largest amount.
TRANCHES Number of tranches in the deal.
MULT-TRANCHE One if the deal includes more than one tranche (or facility), 0 otherwise.
SECURED One if the loan is secured, 0 otherwise.
COVENANT One if the loan has special covenants, 0 otherwise.
SENIOR One if the loan is senior, 0 otherwise.
LOAN-TYPE Set of five distinct binary variables to account for the following loan types: 364-day facility, floating rate note,
letter of credit, term loan and revolver/line of credit. The remaining facilities are grouped into another class
and serve as the control variable.
PURPOSE Set of five dummy variables designed to capture the following loan purposes: recapitalization, acquisitions,
working capital, debt restructuring and other purposes. The general corporate purpose category serves as the
control group.
YEAR Set of indicator variables to control for general trends in the market over the 1994–2009 period.

References
Angbazo, L., Mei, J., Saunders, A., 1998. Credit spreads in the market for highly
leveraged transaction loans. Journal of Banking and Finance 22, 1249–1282.
Bolton, P., Scharfstein, D., 1996. Optimal debt structure and the number of creditors.
The Journal of Political Economy 104 (1), 1–25.
Carey, M., Nini, C., 2007. Is the corporate loan market globally integrated? A pricing
puzzle. Journal of Finance 62 (6), 2969–3007.
Chowdhry, B., 1991. What is different about international lending? Review of
Financial Studies 4 (1), 121–148.
Coleman, A., Esho, N., Sharpe, I., 2006. Does bank monitoring influence loan contract
terms? Journal of Financial Services Research 30 (2), 177–198.
 C. Champagne, F. Coggins / Journal of Banking & Finance 36 (2012) 1437–1451
Cook, D., Schellhorn, C., Spellman, L., 2003. Lender certification premiums. Journal of
Banking and Finance 27 (8), 1561–1579.
Dennis, S., Mullineaux, D., 2000. Syndicated loans. Journal of Financial
Intermediation 9, 404–426.
Esty, B., Megginson, W., 2003. Creditor rights, enforcement, and debt ownership
structure: evidence from the global syndicated loan market. Journal of Financial
and Quantitative Analysis 38 (1), 37–58.
Harjoto, M., Mullineaux, D., Yi, H., 2006. A comparison of syndicated loan pricing at
investment and commercial banks. Financial Management 35 (4), 49–70.
Houston, J., Itzkowitz, J., Naranjo, A., 2007. Borrower beyond borders: The
geography and pricing of syndicated bank loans. Unpublished manuscript.
Ivashina, V., 2009. Asymmetric information effects on loan spreads. Journal of
Financial Economics 92, 300–319.
Jensen, M., Meckling, W., 1976. Theory of the firm: Managerial behavior,
agency costs and ownership structure. Journal of Financial Economics 3 (4),
305–360.
1451
Jones, J., Lang, W., Nigro, P., 2001. Recent trends in bank loan syndications: Evidence
for 1995–1999. Federal Reserve Bank of Chicago Proceedings, May, pp. 328–352.
Lee, S., Mullineaux, D., 2004. Monitoring, financial distress, and the structure of
commercial lending syndicates. Financial Management 33 (3), 107–130.
Nandy, D., Shao, P., 2010. Institutional investment in syndicated loans. Unpublished
manuscript.
Panyagometh, K., Roberts, G., 2002. Private information, incentive conflicts, and
determinants of loan syndications. Unpublished manuscript.
Panyagometh, K., Roberts, G., 2010. Do lead banks exploit syndicate participants?
Evidence from ex post risk. Financial Management 39 (1), 273–299.
Preece, D., Mullineaux, D., 1996. Monitoring, loan renegotiability, and firm value:
the role of lending syndicates. Journal of Banking and Finance 20, 557–593.
Simons, K., 1993. Why do banks syndicate loans? New England Economic Review,
Federal Reserve Bank Boston, 45–52.
Sufi, A., 2007. Information asymmetry and financing arrangements: evidence from
syndicated loans. The Journal of Finance 62 (2), 629–668.