the structure of investment bank syndicates and the ... · the structure of investment bank...
Post on 22-Mar-2020
6 Views
Preview:
TRANSCRIPT
The structure of investment bank syndicates
and the quality of bond underwriting
Arthur Krebbers
a, Andrew Marshall
b, Patrick McColgan
c
Abstract
We analyze how the structure of bookrunner syndicates affects the quality of their service to
issuers as reflected in the pricing of euro-denominated bond tranches. We find that domestic
banks obtain lower at-issue credit spreads for issuing firms, highlighting the importance of
ongoing banking relationships in achieving the highest quality of intermediary services. Larger
bookrunner syndicates are associated with higher credit spreads and the inclusion of a passive
bookrunner to the syndicate reduces credit spreads. Bookrunner reputation is of secondary
importance in the pricing of our sample bond tranches. We find significant changes in the
structure of bookrunner syndicates following the global financial crisis, and our findings on the
impact of bookrunner syndicates for pricing are driven by tranches issued during the post-crisis
period. This highlights that investors place greater emphasis on the information provided by
financial intermediaries when there is greater heterogeneity in syndicate structure during the
post-crisis period.
JEL Classification: G11, G12, G24
Keywords: At-issue credit spread; Bookrunner geography; Passive bookrunners.
a Department of Accounting and Finance, University of Strathclyde, Glasgow, G4 0QU, UK,
Email: Arthur.krebbers@strath.ac.uk.
b Department of Accounting and Finance, University of Strathclyde, Glasgow, G4 OQU, UK,
Tel: 44-141-548-3894, Email: a.marshall@strath.ac.uk
c Department of Accounting and Finance, University of Strathclyde, Glasgow, G4 0QU, UK,
Tel: 44-141-548-3690, Email: patrick.mccolgan@strath.ac.uk
We thank Robert Faff, Andrey Golubov, Peter Limbach, Dimitris Petmezas, seminar participants
at the 2015 BAFA Scottish Area Group Conference (Edinburgh), and the University of
Strathclyde for helpful comments on earlier drafts of this paper. All errors remain our own.
1
1. Introduction
The role of financial intermediaries as bookrunners1 on corporate bond tranches should be
relatively uncontentious. Bond bookrunners perform two main functions: they have a
certification role, evaluating the issuer and its prospects on behalf of investors (Chemmanur and
Fulghieri, 1994), and a placement role, which involves marketing and distributing the securities
(Kessel, 1971). The corporate bond market is dominated by large well-rated firms (Denis and
Mihov, 2003), the documentation and terms are standardized, typically devoid of complex
structures or covenants and the investors tend to be institutional. Yet research on the reputation
of financial intermediary syndicates in the bond market concludes that there is substantial
heterogeneity in the quality of service provided. For example Fang (2005) finds that the
reputation of an issuer’s lead bookrunner has a positive impact on the quality of service
provided. Shivdasani and Song (2011) find increased competition from commercial banks led to
an increase in the popularity of co-led bookrunner syndicates during the 1996 to 2000 economic
boom. However, these issuers were associated with a greater incidence of subsequent earnings
restatements and fraud lawsuits, suggesting that competition from commercial banks led to a
reduction in the quality of the financial intermediary services.
This paper seeks to provide new and further evidence on how the structure of financial
intermediary syndicates affects the quality of service they provide. We focus on the European
bond market to study bookrunner syndicates and the quality of their service due to changes in
funding patterns for European issuing firms and changes in regulations facing banks in Europe
after the financial crisis. Large European firms have been moving towards a more capital
1 The term bookrunner is more commonly used in European markets to describe investment banking activity in the
placement of corporate bonds. In European bond markets investment banks typically manage bond issues on a best
efforts basis and place capital at risk only for a very short period prior to placement of the bond with investors.
Banks’ primary function is to build the order book for bond tranches. As such, we use the term bookrunner to
describe the role of investment banks in placement and underwriting in this paper.
2
markets based model of funding, and external pressures on bookrunner syndicate structures have
grown since the global financial crisis (Eurofi, 2014). Stricter regulations since 2008 have forced
banks to maintain higher capital ratios and many have sought to develop more ancillary business
from their clients, including increased presence in bond issues (Chivukula et al., 2014). As a
result, banks have begun to place increasing pressure on issuing firms to offer more bond
bookrunner mandates, leading to an increase in bookrunner syndicate sizes (Stothard, 2013).
However bond issuers have recognized possible inefficiencies in large syndicates and have
adopted bookrunner distinctions (active and passive) that leave the placement role in the hands of
active bookrunners.2 Banks that the issuer feels are less capable of performing the placement
roles for the envisaged tranche can be relegated to a passive status.3 It is now increasingly
common in European bond markets for a subset of banks to be relegated to a passive certification
role (Moore, 2009; Stothard, 2013). Therefore we suggest that the decision on how best for
issuers to structure their bookrunner syndicates has become increasingly important in the
European bond market.
With the exception of Fang (2005) who examines the decision to appoint a high
reputation investment bank, and Shivdasani and Song (2011) who examine the decision to
appoint co-led underwriting syndicates, there is little evidence on the relation between
bookrunner syndicate structure and the quality of intermediary service provided in the bond
market. Therefore this paper examines the impact of reputation and three additional related
characteristics of the overall bookrunner syndicates on the quality of services provided, as
2 Bookrunners who perform both roles are classified as active bookrunners; if only certification they are classified as
passive bookrunners. 3 Note that this distinction is different from the practice of selecting lead bookrunners. Lead bookrunners purely
coordinate the placement role, but they do not perform it exclusively. In the euro-denominated bond market there is
no designated lead underwriter or advisor, which is the focus of much of this prior literature on US investment
banking syndicates.
3
measured by the at issue spread. We also examine whether the relation between bookrunner
syndicate structure and quality of service changed surrounding the period of the global financial
crisis.
First, we consider the size of the bookrunner syndicate. Increasing the number of
intermediaries running the order book could provide benefits to the issuer through ability to sell
bonds to a large group of potential investors (Kessel, 1971), but can increase coordination and
free-rider difficulties within the syndicate (Diamond, 1996; Shivdasani and Song, 2011).
Second, we consider the assigned roles of the bookrunners in the syndicate as either
passive or active. As the active placement role involves higher possible coordination
inefficiencies and free-rider risks, limiting the number of active bookrunners involved in
placement is expected to reduce free-rider costs. On the other hand including a passive
bookrunner in the syndicate rather than appointing an additional active bookrunner can limit
coordination costs and a possible improvement in quality of intermediary service.
Third, we examine the role of bookrunner reputation, based on their ranking in corporate
bond league tables, on the quality of intermediary service. Bond issuers appoint reputable
bookrunners; on the basis that these intermediaries will be incentivized to perform better quality
services so as to maintain a high league ranking and therefore strong reputation (Chemmanur and
Fulghieri, 1994). Alternatively this high reputation group of bookrunners can be oligopolistic and
distrusted by investors if they are seen to exploit their reputation to sell low quality tranches to
generate higher fee income (Chemmanur and Krishnan, 2012). We also extend this literature to
examine how intermediary reputation interacts with bookrunner’s assigned role in the syndicate
and the geographic location of bookrunners to affect the quality of service provided.
4
Fourth, issuers in European bond markets can choose between domestic and non-
domestic bookrunners. This is important as European capital markets are characterized by a
relatively high degree of information, cultural and linguistic barriers. We argue that domestic
banks are better placed to carry out a certification role and understand an issuer’s preferences in
regards to the placement role through ongoing relationships with the issuing firm. Alternatively,
a domestic bank can be argued to be weaker in the placement role as they are less able to attract
price-competitive international investors to the offering (Massa and Zalkodas, 2014).
Finally, as our sample data spans the 2008 global financial crisis, an event which had a
direct impact on regulations facing banks and corporate bond credit spreads we consider if there
has been changes in the relation between bookrunner syndicate structure and quality of service
over this period. The post-crisis time period is characterized by a significant increase in
syndicate size and an increased propensity to appoint passive bookrunners. This allows issuing
firms to differentiate tranche quality through the structure of the financial intermediary group.
Our sample comprises 1,224 investment grade euro-denominated public bond tranches
issued by 324 Western European firms between 2001 and 2012. The bond tranches are
significantly more homogeneous than earlier bond market studies and the issuers are generally
large and well-established firms in the European capital markets. Theoretically, differences
between the at-issue credit spread for these tranches should be small and fully explainable by
tranche characteristics such as the rating, tenor and issue size (Longstaff et al., 2005). Therefore
the influence of financial intermediaries should be limited for these issuers and therefore our
sample provides a rigorous test of the residual influence of bookrunner characteristics on the
quality of their service.
5
After controlling for endogenous matching of issuers and bookrunners we find that the
trend of increasing bookrunner syndicate size is detrimental to the quality of service. We
interpret this as evidence of coordination and free-rider problems in the placement of bond
tranches. However, bond issuers are able to lower the at-issue credit spreads and reduce the free-
rider costs of larger syndicates through introducing passive bookrunners to the syndicate.
The effect of bookrunners geography is to our knowledge untested in the literature and
we find that the fraction of non-domestic, active non-domestic bookrunners and high reputation
active non-domestic bookrunners is negatively related to the quality of intermediary service. This
suggests that domestic bookrunners are able to achieve a higher price for bonds issued by virtue
of having a stronger banking relationship and becoming more informed about the issuer. We also
find that bookrunner reputation is of secondary importance to the pricing of bond tranches. When
we exclude geographical effects, we find that higher reputation bookrunners are able to offer
better quality services, which is consistent with the reputation theory of Fang (2005). However,
the benefits of using domestic bookrunners outweigh reputation effects.
The significance of our results is driven by tranches issued after the 2008 global financial
crisis. Prior to the financial crisis we find no relation between bookrunner syndicates and quality
of service in the euro-denominated investment grade bond market. The post crisis period is
characterized by larger bookrunner syndicates and increased segregation in bookrunner roles
between active and passive. This makes within syndicate coordination and free-rider costs more
of a concern and strengthens the benefits from introducing passive bookrunners to the syndicate.
Our findings contribute to several strands of finance literature. First, they add to literature
on the growing importance of intermediary relations in increasingly competitive markets for
bond underwriting (Shivdasani and Song, 2011). Our new findings on bookrunner geography
6
suggest that bookrunners are likely to increasingly focus on a particular set of clients they are
able to develop a strong relationship with, such as a geographical focus.
Second, our study adds to the debt market agency literature which typically focuses on
debtholder-shareholder conflicts (Jensen and Meckling, 1976). We find that the size and role
split of a bookrunner syndicate have a significant impact on the at-issue credit spread and
interpret this as evidence of intra-bookrunner agency costs. As funding markets become more
reliant on intermediated sources of capital, such agency costs become increasingly important
(Shivdasani and Song, 2011). Moreover, the growth in bookrunner syndicate sizes will require a
clearer role differentiation between syndicate members.
Third, our findings contribute to the literature on the impact of the global financial crisis
on the role of financial intermediary syndicates. The relation between syndicate structure and
credit spreads in the post crisis period highlights that the relative importance that investors and
issuers place on the certification and placement functions of financial intermediaries varies over
time with financial market conditions.
Finally, our results add to the literature on the determinants of bond credit spreads
(Longstaff et al., 2005). Post-financial crisis, bank capital legislation has made the secondary
market for corporate bonds increasingly illiquid (International Capital Market Association, 2014)
and therefore the determinants of the primary market credit spread increasingly relevant for
investors. We suggest that models seeking to evaluate the at-issue credit spread should
incorporate the various aspects of the bookrunner syndicate composition.
The remainder of this paper is organized as follows. Section 2 develops our testable
hypotheses. Section 3 describes the sample selection process, our dependent variable of the
quality of bookrunner services, and the empirical determinants. Section 4 describes our
7
regression design. Section 5 reports the results of our empirical analysis. Finally, Section 6
concludes.
2. Hypotheses development
In this section we develop our empirical predictions relating to the four key dimensions of the
bookrunner syndicate structure that we suggest can influence the quality of their service. We also
explain the possible impact of the financial crisis on the syndicate structure and quality
relationship.
2.1. Bookrunner syndicate size
The impact of the number of financial intermediaries on the quality of service to capital markets
issuers has been debated for a number of years. Kessel (1971) finds that the search benefits from
a larger number of bookrunners results in lower at-issue credit spreads. Corwin and Schultz
(2005) and Andres et al. (2014) find similar results for equity IPOs and high yield bonds
respectively. Alternatively, as bookrunner syndicates grow the degree of coordination required
increases and the resulting free-rider problems within the bookrunner syndicate also increase
(Diamond, 1996; Shivdasani and Song, 2011). We therefore propose the following competing
hypotheses:
H1a: There is a positive relation between bookrunner syndicate size and quality of service.
H1b: There is a negative relation between bookrunner syndicate size and quality of service.
2.2. Allocation of bookrunner responsibilities
8
The certification bookrunner role entails both legal and reputational risks for intermediaries. The
certification role includes the negotiation of the bond tranche terms and conditions in the bond
documentation and the due diligence performed on the issuer. A bond’s legal documentation
does not distinguish between active and passive bookrunners and any losses incurred by
investors through insufficient due diligence is the responsibility of the entire syndicate.
Therefore a similar degree of commitment to the certification role is required for both active and
passive bookrunners. Given that this work is performed independently by each bookrunner,
coordination risks are largely absent from certification. The more bookrunners that perform this
role the greater the certification benefits offered to the issuer.
The placement role is managed jointly and exclusively by the active bookrunners. It is
therefore common for active bookrunners to receive a relatively larger share of the overall fees
of a bond tranche.4 Placement incorporates pre-announcement strategic discussions, marketing,
managing of the order book, and price setting. As the number of bookrunners increases
placement becomes increasingly difficult to coordinate. Free-rider incentives increase as each
active bookrunners’ contribution becomes more difficult to assess for the issuer.
Following these arguments, we expect that bookrunner certification is unrelated to the
allocation of roles between active and passive bookrunners. Issuers receive stronger certification
as they add more bookrunners to the syndicate, irrespective of whether they perform an active or
passive role. Focusing on the placement role, we expect that increasing the number of active
bookrunners in the syndicate leads to the same trade-off between search benefits from larger
syndicates against coordination difficulties and free-rider incentives amongst a larger number of
active bookrunners. Therefore, we propose the following competing hypotheses:
4 For Verizon Wireless’ USD 49bn multi-tranche bond issue priced in September 2013, the firm paid four active
bookrunners combined fees of $166.57m and seven passive bookrunners total fees of $89.1m (O’Malley, 2015).
9
H2a: There is a positive relation between active bookrunner use and quality of service.
H2b: There is a negative relation between active bookrunner use and quality of service.
Holding the size of the bookrunner syndicate constant, we also expect that the inclusion
of passive bookrunners in the syndicate limits the scope for coordination costs associated with a
higher number of active bookrunners. Issuing firms retain the certification benefits from a larger
bookrunner group without incurring the free-rider and coordination problems inherent in larger
syndicates comprised by active bookrunners (Shivdasani and Song, 2011). Therefore we expect
that the appointment of a passive bookrunner leads to an improvement in quality of placement
services. We propose that:
H3: There is a positive relation between passive bookrunner use and quality of service.
2.3. Bookrunner reputation
Classic product market models suggest that a producer’s reputation is positively related to the
quality of its goods (Allen, 1984). Chemmanur and Fulghieri (1994) adapt this model to the
reputation of bookrunners for equity offerings, arguing that higher reputation bookrunners offer
higher quality services and allow the issuing firm to sell shares at a higher price. However
empirical support for this prediction has been mixed in the equity markets (e.g. Logue et al.,
2002).
Fang (2005) argues that the bookrunner reputation theory is more likely to hold in bond
markets due to the frequency of bond market issuance and the greater relative bargaining
10
position of bond issuing firms as these are larger and more established firms. We also suggest
that in contrast to the equity market (Slovin et al., 2000) the underwriting risks taken by bond
bookrunners is negligible.5 Underwriting risk is restricted to an investor cancelling the order in
the period between the public bookbuilding and the actual delivery of the bonds to investors, a
period typically lasting five business days (Standard and Poor’s, 2013). As such, the main
reputational risk for bookrunners is through insufficient investor interest in the bond tranche. In
practice this is a rare occurrence given the sophistication of the typical investor and the relative
stability of most issuers. Reputational damage can occur if the order book built up before the
bond is priced is less than the target amount communicated to the market. This could result in an
issue being downsized or withdrawn, which would have a negative impact for both active and
passive bookrunners since both are highlighted in external communication on the tranche and
both would be associated with the failed offering.
Chemmanur and Krishnan’s (2012) market power hypothesis suggests a negative relation
between bookrunner reputation and quality of service. In this theory, the impact of reputation on
quality is weak amongst financial intermediaries as their reputation is established largely through
their long-term relationships with investors, and not simply the quality of their service. High
reputation bookrunners are incentivized to reduce the quality of their services and exploit their
existing network of bond investors to ensure successful placement.
Empirical studies have produced mixed findings on the relation between bookrunner
reputation and quality of service in the bond market. Fang (2005) finds support for the reputation
hypothesis as tranches led by high reputation bookrunners have lower at-issue credit spreads.
Andres et al. (2014) find support for the market power hypothesis in a sample of high yield bond
5 In practice issues are normally sold on a best efforts basis requiring that bookrunners only commit to purchasing
the securities after they have engaged in public bookbuilding. During this process investor orders are taken, a
clearing price is determined and the bonds are allocated to investors.
11
issues as those led by high reputation banks have higher at-issue credit spreads, which they
suggest reflects changes to high reputation bookrunners’ incentives following the repeal of the
Glass-Steagall Act. This resulted in growth in bookrunner competition leading to reduced
bookrunner fees resulting in the premium fee for offering higher quality services reducing
(Shivdasani and Song, 2011).
Given the conflicting reputation and market power arguments we propose the following
alternative hypothesis for the relation between bookrunner reputation and the quality of their
service:
H4a: There is a positive relation between bookrunner reputation and quality of service.
H4b: There is a negative relation between bookrunner reputation and quality of service.
2.4. Bookrunner geography
There is little research on how the geographic location of bookrunners relative to the issuing firm
impacts on the quality of the intermediary service. However loan market studies do show that
banks with stronger client relationships offer higher quality services. Fama (1985) emphasizes
the role that relationship banks can play in generating information on borrowing firms. Puri
(1996) finds that investors paid higher prices for securities underwritten by commercial banks
rather than investment houses, which is attributed to this certification role. Also Datta et al.
(1999) find that credit spreads on bond initial public offerings are negatively related to the
strength of the issuing firm’s banking relationships.
As we examine bond issues across a number of European Union countries we can
differentiate between domestic or non-domestic bookrunners based on the country where the
12
issuer and bookrunner are headquartered. We expect that domestic banks will enjoy stronger
relationships with issuing firms through previous provision of retail and investment banking
services and therefore offer a higher quality certification service. Differences in culture and
language within the European Union can lead to higher information asymmetries for non-
domestic bookrunners, reducing their ability to carry out the screening and due diligence
required for this role.
On the other hand, given a degree of information immobility across European borders,
hiring non-domestic bookrunners can improve the quality of the placement service and the
resulting pricing of bond tranches. Massa and Zalkodas (2014) find that US firms with access to
both the domestic and international bond markets tend to issue in international markets resulting
in lower at-issue credit spreads. They argue that international bond investors are able to offer
competitive pricing due to the portfolio benefits they obtain through diversifying away from
domestic firms. These search benefits could overcome the known home bias in financial analyst
forecasts and investor asset allocations (Bae et al., 2008). However we expect that the search
benefit from using non-domestic bookrunners is limited for our sample bond tranches given the
size and international stature of the issuing firms.
Therefore we suggest that the decision to use domestic or non-domestic bookrunners is
determined by a trade-off between the expected higher quality of certification offered by
domestic bookrunners against potentially greater search and resulting placement benefits through
using non-domestic bookrunners. This leads us to the following competing hypotheses on the
geographic location of intermediaries in the bookrunner syndicate:
13
H5a: There is a negative relation between non-domestic bookrunner use and quality of
service.
H5b: There is a positive relation between non-domestic bookrunner use and quality of
service.
2.5. Impact of the global financial crisis
In this section we develop hypotheses relating to the impact of the global financial crisis on the
relation between bookrunner syndicate structure and the quality of financial intermediary service.
Practitioner reports and the financial press suggest that the financial crisis led to significant
changes in the structure of bookrunner syndicates in the bond market. At the same time, credit
spreads increased during the financial crisis and subsequent European sovereign debt crisis. One
of our motivations for this study was that following the crisis, banks have sought to increase fee
income through ancillary business generated from banking clients including in the bond market
(Stothard, 2013). This has led to a growth in bookrunner syndicate sizes and issuing firms have
responded by adopting an active and passive bookrunner distinction with greater frequency
(Eurofi, 2014; Chivukula et al., 2014).
Shivdasani and Song (2011) argue that investor’s incentives to collect additional
information to evaluate securities are stronger during a recession, which reduces the importance
of bookrunner certification. As the euro-denominated bond market matures over time from its
formation in 1999 and issuing firms and their in-house advisors become more established in the
market this can further reduce the importance of the certification function of bookrunner
syndicates in the post-crisis time period. If this is the case, we expect that our previous
14
hypotheses for the relation between bookrunner syndicates and quality of intermediary service
are stronger in the pre-crisis period.
Alternatively, given the low risk securities in our sample investors can be expected to
place little importance on underwriter certification in the pre-crisis period. We argue that the
post-crisis period is characterized by greater information asymmetries between issuers and
investors and a reduced trust in the credit ratings assigned to issuing firms. We expect that
investors will place greater importance on the screening and due diligence carried out by
financial intermediaries in the certification process for sample securities during this later time
period. If this is the case, we expect that our previous hypotheses on the relation between
bookrunner syndicates and quality of service are stronger in the post-crisis period.
Given these alternative explanations for the relation between bookrunner syndicate
structure and at-issue credit spreads, we propose the following competing hypotheses:
H6a: The relation between bookrunner syndicate structure and quality of service is driven
by the pre-crisis period.
H6b: The relation between bookrunner syndicate structure and quality of service is driven
by the post-crisis period.
3. Sample construction, dependent variable and empirical determinants
In this section we set out the sample selection process, our dependent variable of the quality of
bookrunner services, and the proxies for the bookrunner syndicate characteristics (our
explanatory variables). An overview of the sources and calculations for each of the empirical
determinants can be found in Table 1.
15
[Insert Table 1 about here]
3.1. Sample construction
Our sample is based on a Dealogic Debt Capital Markets Analytics search of all euro-
denominated senior unsecured bond tranches by Western European corporates from January 1,
2001 up to December 31, 2012.6 We exclude tranches issued by financial institutions
7 and
secured tranches due to their distinct credit risk profile.8 We include only tranches that are fully
distributed. We also exclude tranches that Dealogic identifies as non-investment grade from the
sample since the price of such tranches is best measured by an at-issue yield rather than a spread
given their equity-like features (Blume et al., 1991).9
In order to focus on tranches that have been marketed to a wide syndicate of European
investors we filter out tranches that are domestically placed, privately placed, single bookrunner-
led, retail-targeted, smaller than EUR 200m, fungible10
, and with a maturity of less than 1 year.
Such tranches will have been sold to a small subset of European investors and are therefore
expected to have cleared at different and less competitive at-issue credit spreads (Blackwell and
Kidwell, 1988; Massa and Zalkodas, 2014). Moreover lower transaction costs incurred through
private placements would imply lower credit spread sensitivity for issuers using this format
6 Dealogic defines Western Europe as including Austria, Belgium, France, Germany, Greece, Ireland, Luxembourg,
Netherlands, Portugal, Spain, Switzerland and UK. 7 The sample does include captive finance firms whose main line of business is to provide lending services to
customers of their industrial parent. Examples include Volkswagen Financial Services and Renault Credit
International Banque. 8 Stulz and Johnson (1985) note that the value of secured debt is largely linked to the value of the collateral assigned
to the bond as opposed to the overall credit worthiness of the firm. 9 In practice a small number of unrated tranches remains in our sample. Dealogic classifies these tranches as
investment grade where the issuer is unrated and no covenants are identified for the tranche. These tranches have the
characteristics of investment grade securities even though the issuer and security are unrated. 10
These are typically known as taps. They have the same terms and conditions as one of an issuer’s existing bonds
and effectively result in an increase in the outstanding amount of this bond. Taps tend to be sold to a small number
of existing holders in the bond.
16
(Blackwell and Kidwell, 1988). These filters produce a final sample of 1,224 bond tranches
issued by 324 firms and we report summary statistics for these tranches in Table 2.
[Insert Table 2 about here]
3.2. Dependent variable - measuring the quality of bookrunner services
The literature on the quality of financial intermediary services typically uses the price of the
asset intermediated as a proxy for quality. The bond literature shows that issuers consider ex ante
observable firm and tranche characteristics related to the risk of default when selecting the
optimal structure of their bookrunner syndicate (Longstaff et al., 2005).11
We measure the quality
of bookrunner services through the at-issue credit spread on bond tranches relative to the
benchmark midswap rate, which is an inverse measure of the price obtained. The at-issue credit
spread is widely quoted in key trade publications including Thomson Reuters’ International
Financing Review (IFR) and Euromoney Institutional Investors’ GlobalCapital as a measure of
issue quality in the bond market.12
In the European bond market the appropriate at-issue credit spread is the spread over the
euro mid-swap rate, which is equivalent to the market’s expectations of the return generated
through the Euribor interbank market over the same tenor. The euro mid-swap rate is an actively
11
Alternative proxies for bookrunner quality can include fees and tranche characteristics including size and tenor.
However, tranche size and tenor are difficult to interpret as measures of issue quality without an understanding of
issuer’s optimal preferences. Moreover, fee information is not required to be reported for tranches sold into Europe
and is unavailable for our sample. Fee details are also likely to be uninformative for our sample tranches because
bond market fees are typically negotiated in advance and are more closely linked to borrower and tranche
parameters such as credit quality and tenor (Melnik and Nissim, 2003). From our discussion with practitioners they
suggest that for plain vanilla bond tranches, bookrunner fees are typically offered on a take it or leave it basis,
meaning no differentiation according to quality of bookrunner service provided. 12
For example, in an article “Blue chips thrive as EDF takes EUR 2bn, Daimler goes to 10 years,” GlobalCapital,
September 4, 2012 is quoted as describing the French utility as having “...sold a EUR 2bn 10 year bond in January
[2012] that had a 3.875% coupon and was priced at 168.6bp over mid-swaps.”
17
traded contract where the purchaser receives 6 month Euribor for a specified period of time. The
euro mid-swap is used for pricing primary euro-denominated corporate bond offerings, as there
is no uniform government bond across the Eurozone.
For our sample of bond tranches we collect the at-issue credit spread in the first instance
from Dealogic, which has records of the pricing details for most of its bonds obtained directly
from the final term sheets of each tranche. For the fixed rate tranches where Dealogic did not
record an at-issue credit spread we calculate this manually through retrieving the at-issue yield to
maturity from the bond prospectus and deducting the benchmark midswap rate as of the date of
issuance. The spread over midswap is not available for the 101 floating rate note tranches in our
sample as they are typically priced over 3 month Euribor, whereas the euro midswap is based on
market expectations of the 6 month Euribor. For these tranches we approximate the spread over
midswap by manually replacing their 3 month Euribor based credit spread with a 6 month
spread. We convert by adding the applicable 6v3 basis swap spread to their at-issue spread over
3 month Euribor. As historic data on this swap is only available from January 2004 we can apply
this calculation to 71 out of the 101 floating rate tranches. Following these additional exclusions,
at-issue credit spread is available for 1,194 tranches.
For these tranches, the mean (median) at-issue credit spread is 1.370% (0.960%). Despite
our sample period covering the global financial crisis, the mean figure is comparable to the
1.35% treasury spread reported by Fang (2005) for her subsample of issues underwritten by high
reputation bookrunners. As would be expected it is noticeably lower than the 4.97% spread for
high yield bonds reported by Andres et al. (2014) between 2000 and 2008.
3.3. Empirical determinants of quality - bookrunner syndicate structure
18
3.3.1. Bookrunner syndicate size
The number of bookrunners is a count of each individual bank appointed to any bookrunner role
in a bond tranche offering. Dealogic extract this information from the final term sheets on each
tranche. In cases where the Dealogic data is incomplete, we collect these details from the
relevant bond prospectus. As reported in Table 2 the sample mean (median) number of
bookrunners is 4.074 (4.000), which is considerably higher than reported in earlier studies.
Andres et al. (2014) record an average of 3.13 bookrunners in their sample of high yield
tranches, and Shivdasani and Song (2011) find that only 12.3% of their sample of corporate bond
issues is intermediated by 4 or more bookrunners. The larger syndicate size reflects the criteria of
tranches using at least 2 bookrunners, our focus on non-domestically placed bonds, the trend of
increasing bookrunner syndicates over time, and larger tranches require more bookrunners for
effective distribution.
3.3.2. Bookrunner active-passive role split
We split the total bookrunner syndicate between active and passive bookrunners. Information on
the involvement of each bookrunner is obtained through searching press coverage of each
tranche published in IFR and GlobalCapital. News sources make clear distinctions between those
banks with active and passive roles on each tranche. This information is sourced from the
communication distributed by active bookrunners during the marketing phase of the syndication
process.13
Tranches in our sample have an average of 3.680 active bookrunners and 8.40% of the
sample tranches have at least one passive bookrunner. In our empirical testing we focus on the
13
It is important for active bookrunners to make this distinction. As only the active bookrunners take responsibility
for the placement role, only their bond sales force is required to actively market and source orders for the tranche.
19
number of active bookrunners in the syndicate and a dummy variable for those tranches that
appoint at least one passive bookrunner to the syndicate.
3.3.3. Bookrunner reputation
Following prior bookrunner studies we use league table rankings to proxy for bookrunner
reputation (Fang, 2005; Golubov et al., 2012). In Table 3 we firstly construct a sample-specific
league table for all bookrunners and, secondly solely for active bookrunners. To create these
league tables we mimic Bloomberg and Dealscan league table methods in assigning league table
credits equally across the bookrunner syndicate for each trade; either the full syndicate or only
the active bookrunners.14
[Insert Table 3 about here]
The main publishers of European bond league tables, namely Dealogic and IFR, publish
rankings focused almost exclusively on the Top 10 and therefore we focus on this benchmark to
classify our sample bookrunners as top tier. We define high reputation bookrunners as those in
the Top 10 of our league tables. The Top 10 banks in both the overall and active role league
tables are dominated by French, German, UK, and US banks. We measure bookrunner syndicate
reputation as the percentage of the total bookrunner syndicate and active bookrunner group on a
14
One difference in our method and Bloomberg and Dealscan is in the treatment of mergers and acquisitions
amongst competing bookrunners. To avoid artificially inflating the reputation of smaller banks we do not
retrospectively assign bookrunner credits from the target bank to the acquiring bank. For instance, in the case of
RBS’ take-over of ABN AMRO’s investment banking division in 2007, RBS does not obtain league table credits for
ABN AMRO’s earlier led transactions.
20
tranche that is part of the Top 10. On average, Top 10 bookrunners constitute 62.0% of the total
bookrunners and 63.0% of the active bookrunners on a tranche.15
3.3.4. Bookrunner geography
To measure the role of bookrunner geography in the quality of bookrunner service we examine
the proportion of non-domestic bookrunners in the syndicate, (domicile is based on the country
of incorporation of the issuer and the bookrunner). Following our method for league table
rankings we focus on the bookrunner syndicate at the time of issue and do not retrospectively
account for the effects of mergers amongst bookrunners.16
Non-domestic bookrunners account
for 65.9% of total bookrunner syndicate for our sample and this ratio is almost identical to the
fraction of non-domestic active bookrunners. To consider the relative influence of geography and
reputational effects we construct a proxy for the number of non-domestic Top 10 active
bookrunners on a tranche. 43.7% of active bookrunners are classified as both non-domestic and
high reputation.
4. Construction of empirical tests
4.1. Model specification
Previous studies have found that the propensity to appoint top tier bookrunners is determined by
issuer and tranche parameters that also influence credit spreads (Gande et al., 1999; Puri, 1996).
Therefore, we use two-stage instrumental variable regressions that account for endogeneity in the
matching between issuers and bookrunners. Our first stage regressions examine the determinants
15
At least one of the Top 10 bookrunners in our sample is involved in 96.1% of tranches. The near-uniform presence
of a Top 10 bookrunner means an indicator variable for using a top tier bank, as used most commonly in prior
literature (Fang, 2005; Golubov et al., 2012), would be a meaningless proxy for reputation in our sample. 16
Again in the case of the RBS take-over of ABN AMRO, for the purpose of earlier ABN AMRO-led trades the
bank is seen as based in the Netherlands, i.e. it is not retrospectively seen as a UK bank.
21
of bookrunner syndicate structure, and second stage regressions examine the determinants of at-
issue credit spreads after-controlling for bookrunner-issuer matching in the first stage
specification. All regressions are estimated with standard errors clustered at the bond level.
4.2. Instrumental variables
To estimate these models we require at least one instrumental variable that appears only in the
first stage regressions. Such factors that are expected to influence the parameters of a bookrunner
syndicate but not the at-issue credit spread. Given our focus on a wide variety of bookrunner
syndicate parameters we use three instrumental variables that we expect are related to at least
one aspect of the structure of the investment banking syndicate. We use a Southern Europe
dummy, issue frequency, and a debut dummy.17
The Southern Europe dummy takes the value one if the issuer’s principal headquarters are
in Greece, Italy, Portugal or Spain, and zero otherwise. Southern European banks tend to be
smaller, reflecting these countries’ more fragmented banking systems and lower GDP per capita
levels (Cavalier, 2014). Holding constant the size of a borrowing firm’s funding and liquidity
requirements, Southern European issuers are expected to require a larger syndicate of
relationship banks to provide their required range of banking services. These banks can lobby for
bond bookrunner business, resulting in larger average bookrunner syndicates and a greater
likelihood of relegating some banks to a passive role. Table 3 also shows few high reputation
bookrunners are based in Southern Europe and so we expect a negative relation between
reputation and the Southern Europe dummy if these issuers are more likely to use domestic
17
Fang (2005) and Golubov et al. (2012) use a scope instrumental variable that ranks lead underwriters’ prior
experience advising or underwriting transactions for the issuing firm in the M&A, equity, and bond markets. A
higher score implies a strong prior relationship between the investment bank and the issuing firm and increases the
likelihood of appointing a top tier investment bank. Given the absence of a lead bookrunner for our sample and the
resulting focus on the overall bookrunner syndicate, this instrument is less informative in our analysis.
22
bookrunners. To compensate for this, Southern European issuers can use more non-domestic
banks to increase investor search benefits.
Issue frequency is a count of the number of euro-denominated tranches issued by a firm
during the sample period. Frequent bond issuers are expected to have dedicated in-house bond
issuance expertise, are more likely to build relationships with reputable banks, and hence require
smaller bookrunner syndicates (Fang, 2005). Moreover, frequent issuers are expected to be better
known by the major bond investors, which reduce the potential placement benefits from hiring
non-domestic bookrunners.
The debut dummy takes the value of one if the tranche is the company’s first appearance
in the euro-denominated bond market, and zero otherwise. We construct this variable for an
extended sample starting in 1999; the first year of bond issues in the euro-denominated market.18
If the first transaction is multi-tranche, all tranches are labelled as debut. Prior research shows
that debut bond issuers are likely to be smaller firms and are more likely to appoint their
relationship banks to bookrunner roles (Yasuda, 2005). This suggests that bookrunner syndicates
on debut tranches will be smaller and have a lower proportion of non-domestic banks.
4.3. Control variables
We control for a range of other firm- and tranche-specific variables that prior studies have found
to influence the at-issue credit spread and the choice of bookrunner parameters. The firm
characteristics we use are size, profitability, the level of intangible assets, leverage, growth
opportunities and public ownership.19
For our sample of investment grade firms the average
issuer has EUR 57.89bn of assets, a book leverage ratio of 34.1% and generates operating profit
18
Issuance and data availability in the euro-denominated bond market is limited and incomplete prior to 2002. 19
See Fang (2005) and Shivdasani and Song (2011) for a comprehensive summary of these variables.
23
of 14.8% of total assets. 13.8% of our sample firms have majority government ownership. Credit
spreads for such firms are expected to be lower due to government regulation and expected
government support in cases of financial distress (Standard and Poor’s, 2010). Collectively, these
summary statistics show that our sample firms are very large and low credit risk issuers.
The tranche-specific parameters we use are the credit rating, maturity, tranche size and
whether the tranche is part of a multi-tranche offering.20
The credit rating is expected to have a
substantial influence on the credit spread, being a key proxy of default risk. It is measured as a
numeric scale of the S&P tranche rating, ascending from 1 for AAA to 10 for BBB- and 11 for
unrated tranches. The sample mean of 7.239 lies between an A- and a BBB+. The mean tranche
tenor is 7.3 years and size is EUR 0.816bn. Both are in line with benchmark index standards.21
29.6% of our sample tranches are part of multi-tranches offers.
5. Empirical analysis
5.1. Determinants of bookrunner syndicate structure
In this section we report the findings of the first stage regressions of bookrunner syndicate
structure. The determinants of syndicate size characteristics are presented in Table 4. Model 1
presents the determinants of the number of bookrunners in the syndicate. All three instrumental
bookrunner characteristics are statistically significant. Issuers from Southern Europe tend to
appoint more bookrunners, and debut and frequent issuers appoint fewer bookrunners. We expect
Southern European issuers to appoint larger bookrunner syndicates given the smaller size of their
domestic banking system and the limited ability of domestic banks to distribute large-scale
20
See Asquith et al. (2013), Collin-Dufresne et al. (2001), Elton et al. (2001), and Longstaff et al. (2005) for a
comprehensive summary of these variables. 21
The iBoxx index requires a EUR 0.5bn minimum issue size for inclusion and includes a 5-7 year index as well as
a 7-10 year index.
24
offerings. Frequent issuers are expected to have dedicated in-house advisory groups and the
issuer’s reputation in the market reduces the need for larger bookrunner syndicates. Our finding
that debut issuers use smaller bookrunner syndicates is consistent with Yasuda (2005), who finds
that debut issuers are more likely to use a smaller pool of existing relationship banks to
underwrite their debut issue. For our control variables, the number of total bookrunners is also
positively related to our credit rating scale, where we assign low values to tranches with the
strongest credit ratings, suggesting that weaker rated firms appoint more bookrunners to ensure
distribution of their bonds. Syndicate size is negatively related to the dummy for majority
government ownership, suggesting that public ownership reduces the need for a large
bookrunner syndicate to ensure distribution of tranches. Unsurprisingly, larger and more
complex multi-tranche offerings require larger bookrunner syndicate groups.
[Insert Table 4 about here]
The determinants of the number of active bookrunners shown in Model 2 are largely
similar to the determinants of the number of total bookrunners in Model 1. The main differences
are that the debut dummy is not statistically significant. The lack of statistical significance for the
debut dummy suggests that although debut firms tend to have smaller overall bookrunner
syndicates, they appoint a comparable number of active bookrunners given the greater degree of
support in placement activities they could require. Again, our credit rating measure, tranche size,
and the multi-tranche dummy are positively related to the number of active bookrunners in the
syndicate. We also now find that firm size and leverage are positively related to the number of
25
active bookrunners, which again is likely to reflect the number of banking relationships, issue
size and complexity, and perceived riskiness of these issuing firms.
Model 3 presents the results for a first stage Heckman (1979) probit model where the
dependent variable is a dummy set equal to one for those bookrunner syndicates using a passive
bookrunner, and zero otherwise. We find that Southern European firms are more likely to
appoint passive bookrunners. We expect that such issuers face the greatest pressure to appoint
more bookrunners because they engage with larger numbers of domestic relationship banks in
the fragmented Southern European banking market. To maintain these banking relationships,
issuing firms are required to reward this greater number of banks with bond market business.
Firms issuing in larger amounts and multi-tranche offerings are also more likely to appoint
passive bookrunners, suggesting these issuers require greater certification from a larger number
bookrunners, but use passive bookrunners to limit the coordination and free rider problems
inherent in larger active bookrunner syndicates.
We extend our analysis of syndicate structure in Table 5 to consider the determinants of
bookrunner syndicate reputation and the geographical split between domestic and non-domestic
bookrunners. Models 1 and 2 examine the determinants of the proportion of total and active
bookrunners that are classified as high reputation banks respectively. We find that Southern
European firms appoint significantly fewer Top 10 bookrunners, likely reflecting the relatively
lower league table ranking of their domestic relationship banks. We also find that debut firms are
likely to have a lower proportion of Top 10 bookrunners. This can reflect a preference for
domestic banks regardless of league table position, which is consistent with Yasuda’s (2005)
finding that debut issuers are more likely to use a small pool of existing relationship banks to
underwrite their debut issue. It is also consistent with debut issues being smaller and less
26
complex, hence placing less emphasis on the reputation benefits of Top 10 bookrunners. For our
control variables, we find that larger and longer maturity tranches are more likely to involve the
appointment of high reputation bookrunners, likely reflecting the need to use high reputation
banks with a larger distribution network to raise larger amounts of capital at longer maturities.
Fang (2005) finds similar results for her sample of US bond issues. Also profitable firms use
higher reputation bookrunners, suggesting that high reputation intermediaries are less likely to be
associated with relatively poorer performing firms.
[Insert Table 5 about here]
Models 3 to 5 examine the proportion of non-domestic bookrunners appointed to a
tranche. Model 3 examines the proportion of non-domestic bookrunners and models 4 and 5
examine the proportion of non-domestic active bookrunners and the proportion of non-domestic
active Top 10 bookrunners in the syndicate respectively. We find that non-domestic bookrunners
are more likely to be appointed by Southern European firms and less likely to be mandated by
frequent and debut issuers. The results for Southern European issuers reflect the smaller size of
their local relationship banks, and hence greater need to appoint non-domestic banks to distribute
their bond offerings. The results for frequency are somewhat surprising as one would expect
frequent borrowers to have more sizeable funding needs and hence more likely to develop
relationships with non-domestic banks. The opposite relationship likely reflects that many
frequent issuers in the euro-denominated corporate bond markets have partial government stakes
or that frequent issuers issue off standardized terms and documentation, over a short arrangement
period, and place less reliance on the search benefits of non-domestic bookrunners. The result for
27
debut is consistent with our view that such firms are smaller and more likely to have smaller
relationship banking syndicates and be dominated by domestic banks. For our control variables,
we find a lower propensity for majority government-owned firms to appoint non-domestic active
bookrunners, suggesting these issuers can face political pressure around the domicile of their
financial intermediaries. Larger and unprofitable firms and multi-tranche issues are more likely
to appoint non-domestic bookrunners, suggesting a greater need by these issuers to attract a
wider investor base for their offerings. Highly rated tranches are also more likely to appointment
of non-domestic bookrunners, consistent with reputation concerns where non-domestic banks are
unwilling to manage the order book for higher risk bonds for issuing firms when information
asymmetries are higher.
5.2. The impact of bookrunner syndicate structure on quality of service
In this section we present the results for our second stage regressions of the determinants of at-
issue credit spreads. The models presented are derived from the first stage models examining the
determinants of bookrunner syndicate structure presented in Tables 4 and 5.
After controlling for issuer-bookrunner syndicate matching, we find in Model 1 that
bookrunner syndicate size is positively related to at-issue credit spread (p=0.018). This confirms
Hypothesis 1b that larger bookrunner syndicates perform lower quality services. This finding is
consistent with agency theory predictions on larger syndicates of financial intermediaries
(Diamond, 1996). It also suggests that the growth in bookrunner syndicates in Europe over the
2000s has resulted in similar coordination and free-rider conflicts following the repeal of the
Glass-Steagall Act in the US a decade earlier (Shivdasani and Song 2011).
28
[Insert Table 6 about here]
Model 2 examines the impact of the number of active bookrunners on the quality of
service. Consistent with Hypothesis 2b, the number of active bookrunners is positively related to
the at-issue credit spread (p=0.022) (lower quality). This supports our findings for overall
syndicate size and shows that larger active bookrunner syndicates are associated with lower
quality intermediary services given the coordination costs and free-rider incentives within the
placement function.
Model 3 examines the effect of passive bookrunners on the at-issue credit spread.
Consistent with Hypothesis 3, the passive bookrunner Mills ratio is negatively related to at-issue
credit spreads (p=0.046). This confirms that the decision to use a passive bookrunner reduces the
at-issue credit spread, suggesting an improvement in the quality of underwriting service. We
argue that passive bookrunners can reduce the coordination and free-riding problems inherent in
larger bookrunner groups by using passive bookrunners. These banks perform a due diligence
role independent of other bookrunners in the syndicate that allows the issuing firm to benefit
from certification. They do not contribute to the distribution of the bond tranche where free-rider
costs within the group structure are associated with a reduction in the quality of underwriting
service (Shivdasani and Song, 2011).
For our control variables we find that across models 1 to 3, low profitability and high
leverage firms have higher at-issue credit spreads, suggesting that bond investors consider these
are important firm-level measures of tranche risk. As expected, we find a positive relation
between our credit rating and at-issue spreads. Multi-tranche offerings and issues by firms with
higher market-to-book ratios have lower at-issue credit spreads. Multi-tranches offers are
29
expected to be used by larger and less risky issuers and we expect that market-to-book reflects
the bond market’s perception of investment opportunities for the tranche proceeds. We find weak
evidence in model 2 that larger issuers have lower at-issue spreads, but the result is not robust in
the remaining models.
Collectively, our results on the active-passive split provide new and robust support for the
agency perspective on syndicates of financial intermediaries (Diamond, 1996). Larger syndicates
of bookrunners actively involved with the selling of a tranche are associated with a lower quality
of service. However issuers who elect to relegate at least one of these intermediaries to passive
roles receive higher quality of service from their bookrunner syndicate in the form of improved
pricing for their bond tranche. We attribute this effect to a reduction in coordination problems
and the resulting scope for free-riding amongst the active bookrunners.
In Table 7 we examine the effects of bookrunner reputation and geographic proximity on
the quality of service. Model 1 studies the impact of the proportion of total bookrunners that are
classified as Top 10 on the at-issue credit spread. Consistent with Hypothesis 4a, the proportion
of Top 10 bookrunners is negatively related to at-issue credit spreads (p=0.025). This is
consistent with reputational benefits incentivizing top tier bookrunners to perform higher quality
services (Chemmanur and Fulghieri, 1994; Fang, 2005). Although potentially appealing for high
yield bond issues we expect the market power hypothesis is less likely to explain the behavior for
our sample of European investment grade issuers. The market is dominated by large and well
established borrowers and higher reputation bookrunners are therefore incentivized to perform
high quality services for these issuers as they envisage repeat business in the future.
Model 2 examines the proportion of Top 10 active bookrunners. We find bookrunner
reputation is also negatively related to credit spreads (p=0.048), confirming reputational
30
incentives hold for both the certification and the placement workstreams. Both models therefore
support Hypothesis 4a and the theory that high reputation bookrunners provide higher quality of
intermediary services. This complements and extends the earlier findings for lead investment
bank reputation in the bond market by Fang (2005) and in M&A advisory roles by Golubov et al.
(2012).22
[Insert Table 7 about here]
In models 3 to 5 of Table 7 we test Hypotheses 5a and 5b and examine the impact of
bookrunner geographic proximity on at-issue credit spread. In model 3 we find that the
proportion of non-domestic bookrunners is positive and weakly significant to at-issue credit
spreads (p=0.083). This provides some support for the previously untested theory that non-
domestic banks offer lower quality bookrunner services. More generally, it is consistent with the
prediction that strong banking relationships can provide valuable certification for issuing firms in
debt markets (Datta et al. 1999; Drucker and Puri, 2005).23
In model 4 we examine the proportion of non-domestic active bookrunners. This is
positively related to at-issue credit spreads (p=0.039) and suggests that the relationship benefits
22 We repeat this analysis of bookrunner reputation replacing the proportion of Top 10 bookrunners with the average
bookrunner ranking for the overall syndicate and for the group of active bookrunners. A higher value for average
bookrunner ranking implies a great fraction of lower reputation bookrunners in the syndicate. We find that this
variable is positively related to the at-issue credit spread and the coefficient is significant at the 5% level in both
regression models. This confirms the robustness of our earlier findings and suggests that the lower the average
league table position of the bookrunner syndicate, the weaker their ability to deliver the best pricing for the issuer. 23
In further testing we also examine the impact of non-domestic bookrunners located in the largest three European
economies; namely France, Germany and the UK. Analyzing the make-up of the non-domestic active bookrunner
syndicate, we find that around half, are from Germany, France or the United Kingdom. Bookrunners domiciled in
these countries should offer the most meaningful search benefits through providing access to investors in the main
domiciles for European capital markets. We again find a positive relation between the proportion of non-domestic
bookrunners headquartered in these economies and the at-issue credit spread, suggesting that relationship
mechanisms implied by geographic proximity between issuers firms and bookrunner syndicates outweigh search
benefits through hiring bookrunner from Europe’s largest capital markets.
31
of using domestic bookrunners continue to apply when only considering active bookrunners who
are able to offer search benefits to the issuer (Kessel, 1971).
The final model considers the influence of bookrunner geographic proximity when
measured against bookrunner reputation. Model 5 examines the proportion of non-domestic
active bookrunners that are part of the Top 10 league table, and so are expected to perform high
quality services due to reputational incentives. We find a positive and significant relation
between this measure of bookrunner geography and at-issue credit spreads (p=0.022). This
suggests that irrespective of whether a non-domestic bank is a high reputation bookrunner, an
issuer receives higher quality service from a domestic bookrunner.24
Collectively, these findings
support Hypothesis 5a that using a higher proportion of non-domestic bookrunners is associated
with a lower quality of intermediary services.
Consistent with our earlier findings reported in Table 6, for the control variables we find
that lower rated bond tranches are priced at higher at-issue credit spreads and in most cases that
profitability and growth opportunities are negatively related to spreads. The positive relation
from leverage and spreads and the negative relation between multi-tranche issues and credit
spreads is weaker in Table 7. We also find a negative relation between issuer size and spreads,
suggesting that larger and therefore lower risk firms achieve lower credit spreads on their bond
tranches. Larger tranches command higher at-issue spreads suggesting that investors require
compensation in the form of higher returns for purchasing larger issues.
24
In addition, we re-estimate our findings for the subsample of issuers who do not have a domestic Top 10
bookrunner in their bookrunner syndicate. If reputational effects are the primary drivers of bookrunner syndicate
selection, the benefit of a non-domestic top tier bookrunner are most acute for these issuers. For the overall sample,
629 tranches do not use a domestic Top 10 bookrunner. For these issuers, 65.9% of their active bookrunners are high
reputation non-domestic investment banks. Compared to the overall rate of 43.7%, this emphasizes that for firms
without a high reputation domestic investment bank to run their order book, there is a greater emphasis on
appointing non-domestic high reputation banks. We again find that the proportion of non-domestic active
bookrunners is positive for this subsample, although insignificant. This suggests that only those issuers whose best
domestic bank is not regarded as a high reputation bookrunner in reported league tables should be indifferent
between appointing a non-domestic high reputation bookrunner and a lower reputation domestic bookrunner.
32
Overall, Table 7 shows that high reputation bookrunners do offer high quality services,
on average, for the tranches where they form the majority of the bookrunner syndicate. However,
we also find that domestic bookrunners offer higher quality services than non-domestic
bookrunners, even when these non-domestic intermediaries are classified as high reputation
bookrunners. We speculate that the difference between top tier and non-top tier bookrunners
largely reflects top tier bookrunners being skilled at serving clients they have a strong business
relationship with, which will predominantly be their domestic client base.
5.3. Impact of the global financial crisis on bookrunner syndicates and at-issue credit spreads
In this section we examine the impact of the global financial crisis on the relation between
bookrunner syndicate structure and quality of service. To highlight the structural changes in the
euro-denominated bond market over the sample period, Figure 1 presents quarterly data on the
number of bond tranches issued and the average at issue-credit spread over our sample period.
Spreads decline initially and reach a minimum between 2003 and 2007 as banks compete with
bond investors to offer lower spreads for borrowing firms. As expected, we observe a sharp spike
in credit spreads in Q3 2008. Although spreads decline after this stage the average value remains
above the average credit-spread observed at any stage in the pre-crisis period. This reflects the
continued European sovereign debt crisis during the latter part of our sample period. This clear
demarcation point in the data supports a focus on two distinct pre- and post-crisis periods.
[Insert Figure 1 about here]
33
The first post-crisis year, 2009, is the year of record issuance in the euro-denominated
corporate debt market. This reflects companies diversifying away from bank and short-term
sources of finance following the financial crisis. We therefore conduct subsample tests for the
period before and after the recent global financial crisis. We take September 1, 2008 as the
starting date for the crisis period being the month of Lehman Brothers’ bankruptcy filing.
Panel A of Table 8 reports mean and median comparisons for bookrunner syndicates over
the pre- and post-crisis subsamples. Panel B reports this same data for instrumental and tranche
variables. It is clear that European corporate bond markets remained highly accessible for our
sample of well-established investment grade firms throughout the post-crisis period.
[Insert Table 8 about here]
Post-crisis, we find a significant increase in the size of bookrunner syndicates. The mean
(median) syndicate size increases from 3.211 (3.000) to 4.865 (4.000). The increase in syndicate
size results from both an increase in the number of active bookrunners and increased use of
passive bookrunners. We find limited evidence of an increase in the proportion of active Top 10
bookrunners, but otherwise no evidence of issuers using higher reputation bookrunners in the
post-crisis period. Focusing on geographical characteristics of the sample, we find no obvious
increase in the propensity to use non-domestic bookrunners, for the overall and active
bookrunner groups. The median proportion of high reputation bookrunners increases from 40%
to 50% in the post-crisis period and the difference is weakly significant at the 10% level.
For tranche characteristics, the mean (median) at-issue credit spread is 0.711% (0.600%)
in the pre-crisis period, which increases significantly in the post-crisis period to 1.946%
34
(1.600%). We find that firms issuing post-financial crisis issued less frequently over the entire
sample period. We expect that such firms switched out of bank financing and into the bond
market following the financial crisis. As expected given the variable definition, debut issuers are
less frequent in the latter time period. We find a reduction of one grade in the mean and median
credit rating from A- to BBB+ for tranches issued post-crisis, suggesting a reduction in the credit
quality of firms during the financial crisis period. Finally, we find a decline in the average, but
not the median, tranche size post-crisis and a reduction in the frequency of multi-tranche
offerings. With the exception of increased credit spread, these summary statistics for tranche
characteristics in Panel B offer no evidence of increased tranche size or complexity that would
necessitate the larger and more heterogeneous bookrunners syndicates found in Panel A.
We examine the role of the financial crisis on the relation between bookrunner syndicates
and at-issue credit spreads by estimating separate regressions for the pre- and post-crisis periods.
Table 9 repeats the tests in Table 6 for bookrunner syndicate size and the active-passive split
surrounding the financial crisis.25
Panel A reports results for the pre-financial period and Panel B
for the post-crisis period. The post-crisis coefficient for bookrunner syndicate size in model 1 is
positive and statistically significant (p=0.017). The post-crisis sample is characterized by larger
bookrunner syndicate sizes that are more likely to be affected by free-rider incentives in larger
groups. The bookrunner syndicate size coefficient in the pre-crisis period is insignificant. This
reflects that smaller and more homogeneous syndicates of intermediaries are less prone to
agency costs, meaning that firms are better able to trade off the marginal search benefits against
the free-rider agency costs of adding an additional bookrunner to the syndicate.
25
We report only the coefficients for bookrunner characteristics in the second stage regressions of at-issue credit
spread determinants. All first stage regressions and firm and tranche characteristics have been included in our
analysis but are omitted for brevity.
35
[Insert Table 9 about here]
The post-crisis results for bookrunner active-passive split, shown in models 2 and 3 of
Table 9, are statistically significant and in line with our main regressions. The number of active
bookrunners is positively related to at-issue credit spreads (p=0.020) and the passive bookrunner
Mills ratio is weakly significant and negatively related to at-issue spreads (p=0.093). This
finding is unsurprising since passive bookrunners are significantly less common in the period
prior to the financial crisis given the more manageable bookrunner syndicate sizes.26
In Table 10 we consider the impact of bookrunner reputation and geography on credit
spreads in the pre- and post-crisis periods. We find that the post-crisis results are weaker than for
the earlier tests in Table 7. For reputation in model 1 the proportion of Top 10 bookrunners is
weakly and negatively related to issue spreads (p=0.072) However, our measure of active
bookrunner reputation in model 2 is insignificant in the pre- and post-crisis period. This provides
limited evidence in support of the reputation hypothesis of Fang (2005) and generally supports
our finding for the full sample that bookrunner reputation is of secondary importance in the
pricing of bond tranches for our sample of investment grade issuers.
[Insert Table 10 about here]
Finally, the pre- and post-crisis results for bookrunner geography are also shown in Table
10 in models 3 to 5. The results are qualitatively in line with our earlier findings in Table 7. The
coefficients on the post-crisis results are all positive and statistically significant at the 10% level
26
We report the coefficient on the passive bookrunner Mills ratio for the pre-crisis period in Table 9 for
completeness, but note that meaningful interpretation of the variable is limited given the small number of our sample
bond tranches used a passive bookrunner prior to the financial crisis.
36
or better. The fact that these results are significant only for the post-crisis period suggests that the
growth in bookrunner syndicate sizes and resulting higher costs of communication and
information dissemination for the issuer increases the value of hiring a greater proportion of
domestic banks with lower information frictions. Alternative explanations based on post-
financial crisis changes in the frequency of using non-domestic bookrunner syndicate appear
unconvincing. Table 8 highlights that the overall proportion of non-domestic bookrunners, active
or otherwise, has not changed significantly from pre- to post-crisis.
Collectively, we find that post-financial crisis tranches drive our earlier findings on the
relation between bookrunner syndicate structure and quality of service for our sample of
investment grade bond tranches issued by large European firms. This provides support for
Hypothesis 6b. Our findings reflect the post-crisis structural trends in bookrunner syndicates
highlighted in Table 8, most notably larger bookrunner syndicate sizes and increased use of
passive bookrunner roles. Prior to the financial crisis bookrunner syndicate structure is unrelated
to the pricing of our sample of low risk bond tranches. However, these structural trends have and
continue to impact the corporate bond markets and are hence relevant for bond issuers.
Moreover, our results suggest that the relative importance of the certification and placement role
of financial intermediaries in the bond market varies with financial market conditions. The
heterogeneity of the at-issue spread primarily began in the years following the financial crisis,
indicating more volatile markets, and strengthened the importance of bookrunner certification to
investors.
37
6. Conclusions
In this paper we examine the impact of a range of bond bookrunner syndicate characteristics on
the quality of bookrunner services pre- and post-financial crisis. We use two-stage regression
models on the at-issue credit spread as our measure of quality. Our sample incorporates 1,224
euro-denominated investment grade public bond tranches made by 324 Western European firms
from 2001 to 2012.
This research extends prior evidence studying the influence of bookrunner syndicate
structure (Fang, 2005; Shivdasani and Song, 2011). For our sample we can jointly analyze highly
homogeneous issuers and heterogeneous bookrunner syndicate structures. Our sample of
investment grade rated tranches includes only large and well-established borrowers, firms who
should in theory enjoy limited benefits from the certification and placement services offered by
their bookrunners. At the same time the euro-denominated bond market is characterized by a
high degree of geographic dispersion amongst bookrunners, as well as recent structural trends
that have resulted in rapid changes in bookrunner syndicate sizes and roles.
We find that a broad range of bookrunner parameters are of importance to the pricing of
corporate bond tranches. We find that the highest quality bookrunner services are offered by
domestic banks. This result continues to hold when comparing to non-domestic active and high
reputation bookrunners. Non-domestic bookrunners should offer greater search benefits, but this
is outweighed in our sample by the greater certification benefits provided by domestic
bookrunners.
Syndicate structure becomes especially important amongst the larger post-financial crisis
bookrunner syndicates, being associated with greater coordination and free-rider costs. Assuming
that geographic proximity reflects strength of relationship, these results suggest that there is an
38
important distinction between insider and outsider bookrunners with insider bookrunners better
able to perform their certification and placement responsibilities.
We find that reputation is of lesser importance. Top 10 bookrunners are only able to offer
higher quality services when introduced as a domestic bookrunner. We conjecture that in a
culturally diverse market such as the European Union higher reputation banks have risen to the
top of the league tables through winning a large share of domestic capital markets business.
Our results also show that the larger bookrunner syndicates that have emerged since the
financial crisis tend to perform lower quality services, reflecting coordination and free-rider
problems (Diamond, 1996). Issuers are able to reduce such agency costs by relegating some of
their bookrunners to a passive status, only performing the certification role, while a subset of
trusted banks maintain an active status, responsible for both placement and certification
workstreams. This role split is a fairly recent innovation in the capital markets, to our knowledge
not yet studied in prior literature. It limits the degree of coordination required for the placement
related workstream without impacting the benefits received through the non-coordinated
certification-related workstreams. Overall, our findings suggest that both the identity and
responsibility of appointed bookrunners is of importance to the quality of service provided by
these financial intermediaries.
39
References: Allen, F., 1984. Reputation and product quality. RAND Journal of Economics 15, 311-327.
Andres, A., Betzer, A., Limbach, P., 2014. Underwriter reputation and the quality of
certification: Evidence from high-yield bonds. Journal of Banking and Finance 40, 97-115.
Asquith, P., Au, A.S., Covert, T., Pathak, P.A., 2013. The market for borrowing corporate bonds.
Journal of Financial Economics 107, 155-182.
Bae, K.H., Stulz, R.M., Tan, H., 2008. Do local analysts know more? A cross-country study of
the performance of local analysts and foreign analysts. Journal of Financial Economics 88,
581-606.
Blackwell, D.W., Kidwell, D.S., 1988. An investigation of cost differences between public sales
and private placements of debt. Journal of Financial Economics 22, 253-278.
Blume, M.E., Keim, D.B., Patel, S.A., 1991. Returns and volatility of low-grade bonds 1977-
1989. Journal of Finance 46, 49-74.
Cavalier, D., 2014. Southern European banks (Italy, Spain, Portugal), mirrors of the crisis. BNP
Paribas Economics Research. Available at: http://economic-
research.bnpparibas.com/Views/DisplayPublication.aspx?type=document&IdPdf=24226.
Chemmanur, T.J., Fulghieri, P., 1994. Investment bank reputation, information production and
financial intermediation. Journal of Finance 49, 57-79.
Chemmanur, T.J., Krishnan, K., 2012. Hetrogeneous beliefs, IPO valuation, and the economic
role of the underwriter in IPOs. Financial Management 41, 769-811.
Chivukula, R., Variankaval, R., Zenner, M., 2014. Corporate finance with a sprig of Basel: Basel
III implications for non-banks. JPMorgan. Available at
https://www.jpmorgan.com/jpmpdf/1320637007385.pdf.
Collin-Dufresne, P., Goldstein, R.S., Martin, J.S., 2001, The determinants of credit spread
changes. Journal of Finance 56, 2177-2207.
Corwin, S.A., Schultz, P., 2005. The role of IPO underwriting syndicates: Pricing, information
production, and underwriter competition. Journal of Finance 60, 443-486.
Denis, D.J., Mihov, V.T., 2003. The choice among bank debt, non-bank private debt, and public
debt: Evidence form new corporate borrowings. Journal of Financial Economics 70, 3-28.
Diamond, D.W., 1996. Financial intermediation as delegated monitoring: A simple example.
Federal Research Bank of Richmond Economic Quarterly 82, 51-66.
Datta, S., Iskandar-Datta, M., Patel, A., 1999. Bank monitoring and the pricing of corporate
public debt. Journal of Financial Economics 51, 435-449.
Drucker, S., Puri, M., 2005. On the benefits of concurrent lending and underwriting. Journal of
Finance 60, 2763-2799.
Elton, E.J., Gruber, M.J., Agrawal, D., Mann, C., 2001. Explaining the rate spread on corporate
bonds. Journal of Finance 56, 247-277.
Eurofi, 2014. Stimulating EU corporate bond and equity markets. Available at:
http://www.eurofi.net/wp-content/uploads/2014/09/Stimulating-EU-corporate-bond-Web.pdf
Fama, E.F., 1985. What's different about banks? Journal of Monetary Economics 15, 29-39.
Fang, L.H., 2005. Investment bank reputation and the price and quality of underwriting services.
Journal of Finance 60, 2729-2761.
Gande, A., Puri, M., Saunders, A., 1999. Bank entry, competition, and the market for corporate
securities underwriting. Journal of Financial Economics 54, 165-195.
Golubov, A., Petmezas, D., Travlos, N.G., 2012. When it pays to pay your investment banker:
New evidence on the role of financial advisors in M&As. Journal of Finance 67, 271-312.
40
Heckman, J.T., 1979. Sample selection bias as a specification error. Econometrica 47, 153-161.
International Capital Market Association, 2014. The current state and future evolution of the
European investment grade corporate bond secondary market: perspectives from the market.
ICMA Secondary Market Practices Committee report. Available at
http://www.icmasyndicate.org/Regulatory-Policy-and-Market-Practice/Secondary-
Markets/survey-report-liquidity-in-the-european-secondary-bond-market-perspectives-from-
the-market/.
Jensen, M.C., Meckling, W., 1976. Theory of the firm: Managerial behavior, agency costs and
capital structure. Journal of Financial Economics 3, 305-360.
Kessel, R., 1971. A study of the effects of competition in the tax-exempt bond market. Journal of
Political Economy 79, 706-738.
Logue, D.E., Foster-Johnson, L., Rogalski, R.J., Seward, J.K., 2002. What is special about the
roles of underwriter reputation and market activities in initial public offerings? Journal of
Business 75, 213-243.
Longstaff, F.A., Mithal, S., Neis, E., 2005. Corporate yield spreads: Default risk or liquidity?
New evidence from the credit default swap market. Journal of Finance 60, 2213-2253.
Massa, M., Zalkodas, A., 2014. Investor base and corporate borrowing: Evidence from
international bonds. Journal of Financial Economics 92, 95-110.
Melnik, A., Nissim, D., 2003. Debt issue costs and issue characteristics in the market for U.S.
dollar denominated international bonds. European Finance Review 7, 277-296.
Moore, H.N., 2009. Seeking credit: In bond sales, some banks more equal than others. Wall
Street Journal, March 9.
O’Malley, C., 2015. Bonds without borders: a history of the Eurobond market. First Edition.
Wiley.
Puri, M., 1996. Commercial banks in investment banking: Conflict of interest or certification
role? Journal of Financial Economics 40, 373-401.
Shivdasani, A., Song, W.L., 2011. Breaking down the barriers: Competition, syndicate structure,
and underwriting incentives. Journal of Financial Economics 99, 581-600.
Slovin, M.B., Sushka, M.E., Lai, K.W.L., 2000. Alternative flotation methods, adverse selection,
and ownership structure: Evidence from seasoned equity issuance in the U.K. Journal of
Financial Economics 57, 157-190.
Standard and Poor’s, 2010. Rating government-related entities: Methodology and assumptions.
Global credit portal. Available at:
http://www.standardandpoors.com/spf/upload/Ratings_EMEA/2010-12-
09_CBEvent_CriteriaGenRatingGov-RelatedEntities.pdf.
Standard and Poor’s, 2013. HY bond market primer. McGraw Hill Financial. Available at:
https://www.lcdcomps.com/d/pdf/hyprimer.pdf.
Stothard, M., 2013. Big banks’ share of corporate debt at new low. Financial Times, February
21.
Stulz, R.M., Johnson, H., 1985. An analysis of secured debt. Journal of Financial Economics 14,
501-521.
Yasuda, A., 2005. Do bank relationships affect the firm’s underwriter choice in the corporate-
bond underwriting market? Journal of Finance 60, 1259-1292.
41
Table 1
Variable definitions and data sources
The table presents variable definitions for bookrunner, bond, and firm characteristics for a sample of 1,224 euro-
denominated public bond tranches issued by 324 Western European firms during 2001-2012.
Variable Name Calculation Source
Panel A: Bookrunner (BR) characteristics
Bookrunner quantity
Number of BRs
A count of the total number of bookrunners on a tranche. Dealogic, Bond
prospectus
Bookrunner active-passive split
Number of Active BRs A count of the total number of active bookrunners on a
tranche.
Dealogic, Bond
prospectus, Financial
press
Passive BR An indicator variable taking the value of one if a tranche
includes a passive bookrunner, and zero otherwise.
Dealogic, Bond
prospectus, Financial
press
Bookrunner reputation
% of Top 10 BRs The percentage of bookrunners on a tranche that are a Top
10 bank by deal value during the sample period.
Dealogic, Bond
prospectus
% of Active Top 10
BRs
The percentage of active bookrunners on a tranche that are a
Top 10 bank by deal value during the sample period.
Dealogic, Bond
prospectus, Financial
press
Bookrunner geography
% of Non-domestic
BRs
The percentage of bookrunners headquartered in a different
country to the issuer.
Dealogic, Bond
prospectus
% of Non-domestic
Active BRs
The percentage of active bookrunners headquartered in a
different country to the issuer.
Dealogic, Bond
prospectus, Financial
press
% of Non-domestic
Active Top 10 BRs
The percentage of active bookrunners headquartered in a
different country to the issuer and who are ranked as a Top
10 bank by deal value during the sample period.
Dealogic, Bond
prospectus, Financial
press
42
Panel B: Tranche and firm characteristics
Dependent variable
At-issue credit
spread
At-issue yield to maturity minus the benchmark euro midswap
rate for the equivalent tenor.
Dealogic, Bond prospectus
Instrumental variables
Southern Europe An indicator variable set equal to one if the issuer is domiciled
in Greece, Italy, Portugal or Spain, and zero otherwise.
Company reports
Frequency Total number of tranches issued by the borrowing firm during
the sample period.
Dealogic
Debut An indicator variable set equal to one if the bond represents the
firm’s first syndicated public bond in the euro-denominated
market, and zero otherwise.
Dealogic
Control variables
Firm size The natural logarithm of the issuer’s book value of total assets
in EUR billions.
Worldscope
Profitability Earnings before interest, taxes, depreciation and amortization
(EBITDA) divided by book value of total assets.
Worldscope
Intangible assets One minus the ratio of net property, plant, and equipment
divided by the book value of total assets.
Worldscope
Leverage Book value of total debt divided by total assets. Worldscope
Growth
opportunities
Book value of total assets plus market value of equity minus
book value of equity, divided by the book value of total assets.
Worldscope
Publicly owned An indicator variable equal to one if 50% or more of the firm’s
shares are owned by the national government, and zero
otherwise.
Company reports
Credit rating The numeric value for the S&P rating assigned to the bond
tranche on the issue date, ascending from 1 for AAA to 10 for
BBB- and 11 for unrated tranches.
S&P
Maturity The natural logarithm of the tenor of the tranche in years. Dealogic
Tranche size The natural logarithm of the amount issued in EUR billions. Dealogic
Multi-tranche An indicator variable equal to one if the issuer sells 2 or more
tranches in the same currency on the same day, and zero
otherwise.
Dealogic
43
Table 2
Summary statistics of bookrunner characteristics and control factors
The table presents summary statistics of a sample of 1,224 euro-denominated public bond tranches issued by 324
Western European firms during 2001-2012. All variables are defined in Table 1.
Number of
Observations Mean Median St. Dev
Panel A: Bookrunner (BR) characteristics
Number of BRs 1224 4.074 4.000 2.134
Number of Active BRs 1224 3.680 4.000 1.329
Passive BR 1224 0.084 - -
% of Top 10 BRs 1224 0.620 0.625 0.259
% of Active Top 10 BRs 1224 0.630 0.667 0.264
% of Non-domestic BRs 1224 0.659 0.667 0.248
% of Non-domestic Active BRs 1224 0.658 0.667 0.255
% of Non-domestic Active Top 10 BRs 1224 0.437 0.500 0.258
Panel B : Tranche and firm characteristics
At-issue credit spread 1194 1.370 0.960 1.188
Southern Europe 1224 0.171 - -
Frequency 1224 8.821 7.000 6.946
Debut 1224 0.242 - -
Firm size 1095 57.888 36.431 56.329
Profitability 1095 0.148 0.112 0.905
Intangible assets 1095 0.650 0.675 0.204
Leverage 1095 0.341 0.350 0.154
Growth opportunities 1043 1.342 1.197 0.495
Publicly owned 1224 0.138 - -
Credit rating 1224 7.239 7.000 2.336
Maturity 1224 7.302 7.000 4.895
Tranche size 1224 0.816 0.750 0.500
Multi-tranche 1224 0.296 - -
44
Table 3
Bookrunner rankings
The table ranks investment banks according to their activity as bookrunners for our sample of 1,224 euro-denominated public bond tranches issued
by 324 Western European firms during 2001-2012.
Any Bookrunner role Active Bookrunner role
Apportioned issuance
Apportioned issuance
Rank Bookrunner EURbn % cumulative
%
Number
of
tranches
Rank Bookrunner EURbn % cumulative
%
Number
of
tranches
1 Deutsche Bank 91.5 9.2% 9.2% 387 1 Deutsche Bank 99.3 10.0% 10.0% 376
2 BNP Paribas 89.9 9.0% 18.2% 427 2 BNP Paribas 92.2 9.2% 19.2% 407
3 SocGen 74.7 7.5% 25.7% 361 3 SocGen 71.4 7.2% 26.3% 341
4 HSBC 70.8 7.1% 32.7% 312 4 Barclays 64.3 6.4% 32.8% 269
5 Barclays 64.2 6.4% 39.2% 287 5 HSBC 63.6 6.4% 39.2% 291
6 JPMorgan 60.6 6.1% 45.3% 259 6 JPMorgan 63.2 6.3% 45.5% 238
7 Citi 60.1 6.0% 51.3% 274 7 Citi 56.5 5.7% 51.1% 239
8 RBS 57.8 5.8% 57.1% 288 8 RBS 55.5 5.6% 56.7% 264
9 Credit Agricole
CIB 52.8 5.3% 62.4% 262 9
Credit Agricole
CIB 49.5 5.0% 61.7% 243
10 UniCredit 34.7 3.5% 65.8% 166 10 UniCredit 29.4 2.9% 64.6% 143
11 Commerzbank 27.0 2.7% 68.5% 115 11 ABN AMRO* 26.6 2.7% 67.3% 94
12 ABN AMRO* 25.8 2.6% 71.1% 96 12 Commerzbank 25.1 2.5% 69.8% 110
13 ING 24.1 2.4% 73.5% 134 13 Credit Suisse 23.9 2.4% 72.2% 95
14 Natixis 24.1 2.4% 76.0% 139 14 Morgan Stanley 22.5 2.3% 74.4% 84
15 Credit Suisse 23.2 2.3% 78.3% 105 15 Santander 21.2 2.1% 76.6% 106
16 Santander 23.0 2.3% 80.6% 136 16 Natixis 20.7 2.1% 78.6% 128
17 Morgan Stanley 20.1 2.0% 82.6% 94 17 ING 20.0 2.0% 80.6% 117
18 BBVA 18.7 1.9% 84.5% 96 18 UBS 14.8 1.5% 82.1% 73
19 UBS 16.6 1.7% 86.1% 92 19 Dresdner
Kleinwort 14.6 1.5% 83.6% 57
20 Goldman Sachs 16.1 1.6% 87.7% 74 20 Goldman Sachs 14.2 1.4% 85.0% 62
21 BoAML 15.5 1.6% 89.3% 92 21 BBVA 14.1 1.4% 86.4% 81
22 Dresdner
Kleinwort 14.6 1.5% 90.8% 57 22 BoAML 13.8 1.4% 87.8% 73
23 Intesa Sanpaolo 12.5 1.3% 92.0% 78 23 Merrill Lynch* 12.2 1.2% 89.0% 41
24 Merrill Lynch* 10.8 1.1% 93.1% 41 24 Intesa Sanpaolo 10.5 1.0% 90.1% 65
25 Mitsubishi 10.6 1.1% 94.2% 63 25 Lehman
Brothers 9.4 0.9% 91.0% 37
* marks pre-acquisition/merger entities, i.e. the league table credits of ABN AMRO before the takeover of its investment banking activities by RBS
45
Table 4
Determinants of bookrunner group size and active-passive split
The table reports first stage regressions predicting the determinants of bookrunner syndicate size and role allocations
for a sample of 1,224 euro-denominated public bond tranches issued by 324 Western European firms during 2001-
2012. Models 1 and 2 are ordinary least squares regressions and model 3 presents the results of a probit regression of
the decision to use a passive bookrunner. All variables are defined in Table 1. P-values are calculated from bond
level-clustered standard errors. ***, **, and * denote significance at the 1%, 5%, and 10% levels respectively.
Model 1 Model 2 Model 3
Number of BRs Number of Active BRs Passive BR Probit
Constant -12.835
*** -7.735
*** -10.611
***
(0.000) (0.000) (0.006)
Southern Europe 0.950
*** 0.783
*** 0.484
**
(0.000) (0.000) (0.011)
Frequency -0.046
*** -0.045
*** -0.018
(0.000) (0.000) (0.320)
Debut -0.337
* -0.127 -0.2905
(0.053) (0.270) (0.212)
Firm size 0.059 0.188
*** -0.091
(0.614) (0.001) (0.431)
Profitability 0.014 0.007 -0.033
(0.104) (0.257) (0.270)
Intangible assets -0.364 -0.131 -0.131
(0.206) (0.589) (0.769)
Leverage -0.556 1.010
*** -0.899
(0.393) (0.003) (0.152)
Growth opportunities 0.028 -0.010 0.121
(0.877) (0.912) (0.477)
Publicly owned -0.544
* -0.106 -0.076
(0.071) (0.603) (0.801)
Credit rating 0.092
** 0.115
*** 0.064
(0.026) (0.000) (0.172)
Maturity -0.026 0.109 -0.035
(0.837) (0.117) (0.792)
Tranche size 0.816
*** 0.473
*** 0.454
**
(0.000) (0.000) (0.018)
Multi-tranche 1.057
*** 0.462
*** 0.608
***
(0.000) (0.000) (0.001)
Year and Sector controls yes yes yes
Number of Observations 1017 1017 1017
R2 / Pseudo R
2 0.386 0.404 0.261
Wald χ2
104.14***
(0.000)
F-statistic 16.66
*** 23.18
***
(0.000) (0.000)
46
Table 5
Determinants of bookrunner reputation and geography
The table reports first stage ordinary least squares regressions predicting the determinants of bookrunner syndicate
reputation and the geographic location of syndicate members for a sample of 1,224 euro-denominated public bond
tranches issued by 324 Western European firms during 2001-2012. All variables are defined in Table 1. P-values are
calculated from bond level-clustered standard errors. ***, **, and * denote significance at the 1%, 5%, and 10%
levels respectively.
Model 1 Model 2 Model 3 Model 4 Model 5
% of Top 10
BRs
% of Active Top
10 BRs
% of Non-
domestic BRs
% of Non-
domestic Active
BRs
% of Non-
domestic Active
Top 10 BRs
Constant -0.524 -0.781 0.745
* 0.783
* -0.125
(0.276) (0.108) (0.080) (0.078) (0.786)
Southern
Europe
-0.100***
-0.082***
0.066***
0.088***
0.110***
(0.000) (0.004) (0.004) (0.000) (0.000)
Frequency -0.000 0.0002 -0.005
** -0.005
** -0.002
(0.885) (0.917) (0.022) (0.022) (0.262)
Debut -0.049
* -0.047
* -0.062
** -0.055
** -0.062
**
(0.072) (0.089) (0.017) (0.038) (0.024)
Firm size -0.019 -0.020 0.023
* 0.024
* 0.004
(0.178) (0.168) (0.071) (0.061) (0.777)
Profitability 0.009
*** 0.009
*** -0.009
*** -0.009
*** -0.005
***
(0.000) (0.000) (0.000) (0.000) (0.000)
Intangible
assets
-0.004 0.008 -0.049 -0.029 -0.018
(0.945) (0.893) (0.401) (0.636) (0.751)
Leverage -0.082 -0.093 0.057 0.075 -0.035
(0.261) (0.217) (0.445) (0.326) (0.642)
Growth
opportunities
0.017 0.020 0.001 0.003 -0.007
(0.490) (0.415) (0.967) (0.913) (0.773)
Publicly owned 0.060 0.070
* -0.139
*** -0.148
*** -0.002
(0.151) (0.092) (0.001) (0.001) (0.966)
Credit rating 0.006 0.006 -0.017
*** -0.020
*** -0.007
(0.302) (0.304) (0.006) (0.002) (0.288)
Maturity 0.032
* 0.032
* -0.019 -0.024 0.002
(0.051) (0.064) (0.231) (0.138) (0.886)
Tranche size 0.058
** 0.071
*** 0.001 -0.001 0.035
(0.015) (0.003) (0.945) (0.981) (0.135)
Multi-tranche 0.012 0.010 0.047
** 0.050
** 0.048
**
(0.575) (0.674) (0.018) (0.016) (0.039)
Year and
Sector controls yes yes yes yes yes
Number of
Observations 1017 1017 1017 1017 1017
R2 / Pseudo R
2 0.109 0.096 0.127 0.138 0.107
F-statistic 16.66
*** 23.18
*** 20.93
*** 24.49
*** 6.94
***
(0.000) (0.000) (0.000) (0.000) (0.000)
47
Table 6
Regression analysis of impact of bookrunner quantity and active-passive split on at-issue credit spread
The table reports second-stage regressions for two-stage regression models predicting the at-issue credit spread of
1,224 euro-denominated public bond tranches issued by 324 Western European firms during 2001-2012. First stage
selection models are presented in Table 4 and examine the determinants of bookrunner syndicate structure. Second
stage outcome regressions examine the determinants of at-issue credit spreads against predicted bookrunner
characteristics from the first stage selection model. Models 1 and 2 report results for selection models based on
instrumental variable two-stage least squares (IV-2SLS) regressions. Model 3 reports results for a selection model
using a Heckman two-stage regression. All variables are defined in Table 1. P-values are calculated from bond level-
clustered standard errors. ***, **, and * denote significance at the 1%, 5%, and 10% levels respectively.
Model 1 Model 2 Model 3
Number of BRs Number of Active BRs Passive BR
Constant 0.535 -0.185 1.357
(0.767) (0.909) (0.554)
Bookrunner syndicate size 0.180
** 0.206
**
(0.018) (0.022)
Passive BR mills ratio -0.326
**
(0.046)
Firm size -0.038 -0.058
* -0.020
(0.324) (0.090) (0.633)
Profitability -0.072
*** -0.071
*** -0.059
***
(0.000) (0.000) (0.000)
Intangible assets 0.098 0.056 0.064
(0.517) (0.706) (0.678)
Leverage 0.733
*** 0.454
** 0.856
***
(0.002) (0.041) (0.001)
Growth opportunities -0.260
*** -0.254
*** -0.289
***
(0.002) (0.002) (0.001)
Publicly owned -0.007 -0.086 -0.102
(0.967) (0.545) (0.469)
Credit rating 0.174
*** 0.166
*** 0.171
***
(0.000) (0.000) (0.000)
Maturity -0.021 -0.048 -0.014
(0.729) (0.418) (0.813)
Tranche size 0.011 0.059 0.036
(0.915) (0.498) (0.727)
Multi-tranche -0.306
*** -0.212
*** -0.281
**
(0.007) (0.009) (0.017)
Year and Sector controls yes yes yes
Number of Observations 1017 1017 1017
R2 / Pseudo R
2 0.474 0.498 0.496
48
Table 7
Regression analysis on impact of bookrunner reputation on at-issue credit spread
The table reports second-stage regressions for two-stage regression models predicting the at-issue credit spread of
1,224 euro-denominated public bond tranches issued by 324 Western European firms during 2001-2012. First stage
selection models are presented in Table 4 and examine the determinants of bookrunner syndicate structure. Second
stage outcome regressions examine the determinants of at-issue credit spreads against predicted bookrunner
characteristics from the first stage selection model. All models are based on instrumental variable two-stage least
squares (IV-2SLS) regressions. All variables are defined in Table 1. P-values are calculated from bond level-
clustered standard errors. ***, **, and * denote significance at the 1%, 5%, and 10% levels respectively.
Model 1 Model 2 Model 3 Model 4 Model 5
% of Top 10
BRs
% of Active Top
10 BRs
% of Non-
domestic BRs
% of Non-
domestic Active
BRs
% of Non-
domestic Active
Top 10 BRs
Constant -2.887
* -3.629
* -3.033
* -1.784 2.472
(0.083) (0.053) (0.059) (0.242) (0.484)
Bookrunner
reputation
-2.108**
-2.346**
(0.025) (0.048)
Bookrunner
geography
1.560* 1.644
** 1.676
**
(0.083) (0.039) (0.022)
Firm size -0.076
* -0.079
* -0.078
** -0.0767
** -0.027
(0.065) (0.070) (0.044) (0.041) (0.711)
Profitability -0.053
*** -0.052
*** -0.055
*** -0.065
*** -0.549
(0.000) (0.000) (0.000) (0.000) (0.612)
Intangible
assets
0.028 0.050 0.105 0.01858 0.231
(0.883) (0.801) (0.560) (0.906) (0.581)
Leverage 0.421 0.399 0.545
** 0.550
** 0.530
(0.109) (0.155) (0.036) (0.025) (0.318)
Growth
opportunities
-0.207**
-0.198**
-0.263***
-0.271***
-0.135
(0.017) (0.032) (0.005) (0.004) (0.578)
Publicly owned 0.006 0.040 0.096 -0.119 -0.466
(0.972) (0.839) (0.648) (0.410) (0.154)
Credit rating 0.200
*** 0.201
*** 0.216*** 0.192
*** 0.179
***
(0.000) (0.000) (0.000) (0.000) (0.000)
Maturity 0.054 0.059 0.006 -0.007 -0.007
(0.448) (0.436) (0.928) (0.919) (0.951)
Tranche size 0.277
*** 0.321
*** 0.163** 0.138
* -0.172
(0.002) (0.003) (0.029) (0.077) (0.414)
Multi-tranche -0.091 -0.093 -0.187** -0.125 -0.056
(0.310) (0.342) (0.041) (0.114) (0.703)
Year and Sector
controls yes yes yes yes yes
Number of
Observations 1017 1017 1017 1017 1017
R2 / Pseudo R
2 0.312 0.236 0.407 0.401 0.385
49
Table 8
Univariate comparison of tranche and issuer characteristics for pre and post financial crisis subsamples
The table reports summary statistics for the pre- and post-financial crisis subsamples of 1,224 euro-denominated public bond issues made by 324
Western European firms during 2001-2012. Pre-financial crisis includes all tranches priced before September 2008 while post-financial crisis
includes all tranches priced after this date. P-values are reported for ANOVA tests of difference in means and Kruskal-Wallis tests of difference in
medians.
Pre-financial crisis
Post-financial crisis
Independent variables Number of.
Observations Mean Median
Number of
Observations Mean Median
T-test
of
means
Wilcoxon-
Rank-Sum
Panel A: Bookrunner characteristics
Number of BRs 586 3.211 3.000 638 4.865 4.000 0.000 0.000
Number of Active BRs 586 3.160 3.000 638 4.157 4.000 0.000 0.000
Passive BR 586 0.017 0.000 638 0.146 0.000 0.000 0.000
% of Top 10 BRs 586 0.611 0.667 638 0.627 0.600 0.304 0.391
% of Active Top 10 BRs 586 0.613 0.667 638 0.645 0.667 0.040 0.059
% of Non-domestic BRs 586 0.670 0.667 638 0.649 0.667 0.141 0.298
% of Non-domestic Active BRs 586 0.671 0.667 638 0.647 0.667 0.105 0.167
% of Non-domestic Active Top 10 BRs 586 0.424 0.400 638 0.449 0.500 0.083 0.092
Panel B: Tranche characteristics
At-issue credit spread 556 0.711 0.600 637 1.946 1.600 0.000 0.000
Southern Europe 586 0.162 0.000 638 0.179 0.000 0.441 0.442
Frequency 586 9.420 7.000 638 8.271 7.000 0.004 0.094
Debut 586 0.321 0.000 638 0.169 0.000 0.000 0.000
Credit rating 586 6.792 7.000 638 7.650 8.000 0.000 0.000
Maturity 586 7.507 7.000 638 7.113 7.000 0.159 0.681
Tranche size 586 0.850 0.750 638 0.784 0.750 0.023 0.488
Multi-tranche 586 0.341 0.000 638 0.254 0.000 0.001 0.001
50
Table 9
Impact of bookrunner quantity and active-passive split on at-issue credit spread surrounding financial crisis
The table reports two-stage regression analysis predicting the at-issue credit spread of 1,224 euro-denominated
public bond tranches issued by 324 Western European firms during 2001-2012. First stage selection models examine
the determinants of bookrunner syndicate structure. Second stage outcome regressions examine the determinants of
at-issue credit spreads against predicted bookrunner characteristics from the first stage selection model. Pre-financial
crisis includes all tranches priced before September 2008 while post-financial crisis includes all tranches priced after
this date. Models 1 and 2 report results for selection models based on instrumental variable two-stage least squares
(IV-2SLS) regressions. Model 3 reports results for a selection model using a Heckman two-stage regression. All
variables are defined in Table 1. P-values are calculated from bond level-clustered standard errors. ***, **, and *
denote significance at the 1%, 5%, and 10% levels respectively.
Model 1 Model 2 Model 3
Number of BRs Number of Active BRs Passive BR
Panel A: Pre-financial crisis
Bookrunner syndicate size -0.004 0.010
(0.972) (0.933)
Passive BR mills ratio 0.008
(0.914)
Firm and Tranche controls yes yes yes
Year and Sector controls yes yes yes
Number of Observations 474 474 474
R2 / Pseudo R2 0.554 0.556 0.555
Panel B: Post-financial crisis
Bookrunner syndicate size 0.232
** 0.264
**
(0.017) (0.020)
Passive BR mills ratio -0.530
*
(0.093)
Firm and Tranche controls yes yes yes
Year and Sector controls yes yes yes
Number of Observations 543 543 543
R2 / Pseudo R
2 0.370 0.412 0.415
51
Table 10
Impact of bookrunner reputation on at-issue credit spread surrounding financial crisis
The table reports two-stage regression analysis predicting the at-issue credit spread of 1,224 euro-denominated
public bond tranches issued by 324 Western European firms during 2001-2012. First stage selection models examine
the determinants of bookrunner syndicate structure. Second stage outcome regressions examine the determinants of
at-issue credit spreads against predicted bookrunner characteristics from the first stage selection model. Pre-financial
crisis includes all tranches priced before September 2008 while post-financial crisis includes all tranches priced after
this date. All models are based on instrumental variable two-stage least squares (IV-2SLS) regressions. All variables
are defined in Table 1. P-values are based on standard errors calculated from bond level-clustered standard errors.
***, **, and * denote significance at the 1%, 5%, and 10% levels respectively.
Model 1 Model 2 Model 3 Model 4 Model 5
% of Top 10
BRs
% of Active
Top 10 BRs
% of Non-
domestic BRs
% of Non-
domestic Active
BRs
% of Non-
domestic Active
Top 10 BRs
Panel A: Pre-financial crisis
Bookrunner
reputation
0.353 0.404
(0.381) (0.362)
Bookrunner
geography
1.175 1.004 0.942
(0.228) (0.251) (0.144)
Firm and Tranche
controls yes yes yes yes yes
Year and Sector
controls yes yes yes yes yes
Number of
Observations 474 474 474 474 474
R2 / Pseudo R2 0.527 0.521 0.2928 n.m.f. n.m.f.
Panel B: Post-financial crisis
Bookrunner
reputation
-4.246* -3.404
(0.072) (0.170)
Bookrunner
geography
1.893* 2.020
** 2.417
***
(0.069) (0.023) (0.009)
Firm and Tranche
controls yes yes yes yes yes
Year and Sector
controls yes yes yes yes yes
Number of
Observations 543 543 543 543 543
R2 / Pseudo R
2 n.m.f. 0.055 0.331 0.328 0.287
52
Figure 1
Pattern of tranche issue volume and at-issue credit spreads over time
The chart presents average statistics for number of tranches issued and at-issue credit spread for 1,224 euro-denominated public bond tranches
issued by 324 Western European firms during 2001-2012.
0
0.5
1
1.5
2
2.5
3
3.5
0
10
20
30
40
50
60
70
80
Q1
-01
Q3
-01
Q1
-02
Q3
-02
Q1
-03
Q3
-03
Q1
-04
Q3
-04
Q1
-05
Q3
-05
Q1
-06
Q3
-06
Q1
-07
Q3
-07
Q1
-08
Q3
-08
Q1
-09
Q3
-09
Q1
-10
Q3
-10
Q1
-11
Q3
-11
Q1
-12
Q3
-12
At-issue credit spread
No. tranches issued
No. tranches Av. Credit Spread
top related