See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/4789290

Syndicated Loans

Article  in  Journal of Financial Intermediation · February 2000

DOI: 10.2139/ssrn.152689 · Source: RePEc

CITATIONS READS
172 9,506

2 authors:

Steven A. Dennis Donald J. Mullineaux

University of North Dakota University of Kentucky
20 PUBLICATIONS   611 CITATIONS    60 PUBLICATIONS   2,136 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Donald J. Mullineaux on 08 September 2020.

The user has requested enhancement of the downloaded file.

 SYNDICATED LOANS

Steven A. Dennis
California State University at Fullerton

Donald J. Mullineaux
University of Kentucky

Abstract

This paper analyzes the market for syndicated loans, a hybrid of private and
public debt, which has grown at well over a 20 percent rate annually over the last
decade and totaled over $1 trillion in 1997. While loan sales have been heavily
researched, there has been little work on syndications, which differ in character
from sales. We present empirical evidence that the extent to which a loan can be
syndicated increases as information about the borrower becomes more
transparent, as the syndicate’s managing agent becomes more “reputable,” as
the loan’s maturity increases, and as the loan lacks collateral. The lead manager
holds larger proportions of information-problematic loans in its own portfolio.
Loan syndications, like loan sales, appear to be motivated, in part, by capital
regulations, but the liquidity position of the agent bank does not influence its
syndication behavior. Activity in the syndication market is broadly consistent with
Diamond’s (1991) “life-cycle” model of borrower choice, but our results also
support the view that contract characteristics and reputation can serve as
“substitutes” for information in the debt market.

KEY WORDS: LOAN SYNDICATIONS, BANKING, AGENCY COSTS,

REPUTATION

Contact:
Donald J. Mullineaux
Gatton College of Business and Economics
University of Kentucky
Lexington, KY 40506-0034
Phone: (606) 257-2890
Fax: (606) 257-9688
E-mail: mullinea@pop.uky.edu
 Syndicated Loans

A significant literature has emerged in recent years on the topic of

borrower choice between private and public debt. A consensus view suggests

that a critical factor driving the choice is the character and quality of the

information available about the borrower. Diamond (1991) develops a formal

model which involves borrowers shifting from private sources (financial

intermediaries such as banks and insurance companies) to the public markets

(commercial paper and bonds) as the quality of the information about the firm

improves and the borrower develops a “reputation” in the form of a history of

successful debt repayments. Carey, Prowse, Rea and Udell (CPRU, 1993)

propose an extended continuum, with firms gravitating from insider finance,

through venture capital, bank loan finance, private placements, and the public

debt markets as information and collateral become increasingly available and the

borrower’s repayment “track record” improves. As borrowers become less

“information problematic,” the characteristics of the lenders and the underlying

debt contracts vary systematically. Bank loans tend to be relatively short term,

involve extensive covenants, and are frequently re-negotiated. The majority of

public-debt contracts are longer term, involve relatively loose covenants, and are

almost never restructured. These contractual characteristics have been

extensively examined and rationalized in the literature in papers such as Berlin

and Loeys (1988), Berlin and Mester (1992), and Rajan and Winton (1995).

As CPRU note, some borrowing techniques and their associated contracts

involve overlap between the public and private markets. They focus primarily on

1
private placements, which involve lenders who are intermediaries (typically

insurance companies) and contracts that have relatively strict covenants and

reasonably frequent restructuring, which are all characteristics of private debt.

Like public debt, however, private placements are issued in large amounts by

sizable firms at fixed interest rates, and sales of these debt claims to investors

normally are facilitated by investment banks or commercial bank holding

company affiliates.1 In addition, James (1987) finds that announcing private

placement financing has no effect on the equity returns to the borrowing firm,

whereas loan announcements have a significantly positive impact.

Still another type of financing that involves characteristics of both public

and private debt is syndicated loans. Syndication involves the sale of a relatively

large commercial loan in “parcels” to a group of institutional buyers, whereas a

private placement typically is the sale of a “whole” debt contract to a single lender

(although some private placements involve a relatively small group of lenders).

In principle, any loan could be syndicated by any financial institution that acts as

a loan originator. In practice, only certain kinds of loans and certain types of

institutions engage in syndications. This paper identifies the factors that

influence a loan’s syndication potential. Our maintained hypothesis is that the

characteristics of the borrower, of the lender, and the loan contract itself can play

some role. By sorting out these influences empirically, we hope to: (1) provide

further evidence on the significance of information problems and mechanisms for

resolving them in financial contracting and (2) specify where syndicated loans “fit”

on the private/public debt continuum.

2
 An analysis of syndicated loans also may provide indirect evidence

concerning the role of relationships in financial arrangements. Recent papers by

Berger and Udell (1995), Petersen and Rajan (1994), and Cole (1998) provide

evidence that an ongoing relationship between a borrower and financing agent

can serve as a mechanism for attenuating agency and information problems.

When borrowers seek multiple loans from the same bank over time, a repayment

history accumulates and the lender forms a more extensive and dynamic

information set based on multiple assessments of financial statements,

discussions with managements, and possible renegotiations. When lending is

complemented with the provision of deposit, cash-management and operations-

based (e.g., payroll) services, the information set becomes still broader and

deeper. Berger and Udell (1995) find that interest rates and collateral

requirements on lines of credit decline with the length of a bank-borrower

relationship, while Petersen and Rajan (1994) provide evidence that dependence

on trade credit decreases with the length of a relationship. Cole (1998) finds that

the probability a small business will receive credit increases in the presence of a

relationship, especially if the borrower obtained multiple services in that context.

When lenders provide funds to borrowers in the context of a syndicated

loan, the elements that facilitate establishing and deepening a relationship are

less likely to be present. While the lead bank may have some form of

relationship with the borrower, this is less likely to be the case for participating

members. Since the buyer of the syndicated loan cannot rely on relationships

with the borrower as a substitute for other mechanisms that resolve agency

3
problems, evidence that certain loan contract characteristics play a different role

in a syndication context relative to a relationship setting would confirm the

relevance of relationships as a factor for resolving information problems.

Section I of this paper provides an overview of the loan syndication

market. Section II specifies a model that identifies the various factors which

affect the syndication potential of individual commercial loans. Section III

provides estimates of the model and interprets the results. Section IV provides

some conclusions.

I. Overview of the Loan Syndication Market

Syndicating loans is a centuries old process that has shown significant

growth in the 1990’s. Gold Sheets Annual, a publication of Loan Pricing

Corporation, reports that loan syndication volume hit a record high $888 billion in

1996 compared to $137 billion in 1987, a compound annual growth rate of

roughly 23 percent.2 In 1997, loan syndications exceeded $1 trillion for the year

as a whole, compared to roughly $300 million of private placements. Syndicated

financings in 1996 were employed largely for general corporate purposes (49.5

percent) and for debt repayment (33.5 percent), which represents a considerable

shift from the late 1980’s when syndicated loans were used primarily to finance

mergers and acquisitions and leveraged buyout activities. The rapid growth in

volume has been accompanied by declining spreads and fees. In 1996, the

average rate spread over LIBOR on BB credits averaged 71 basis points,

4
compared to 130 basis points in 1992. Average fees were lower by about 10

basis points over the same period. Buyers of syndicated credits included

commercial and investment banks, insurance companies, mutual funds, and

other institutional funds managers. The American Banker (November 18, 1997)

reports that over $14 billion of loans were syndicated in 1997 through the

Internet, including a $4 billion loan to Compaq Computer, Inc., using a secure

private communications system called Intralinks.

The syndication market has grown significantly more than the private-

placement market in recent years and, according to some practitioners, has

begun to converge with the junk bond component of the public debt market in

terms of borrowers, investors, and underwriters. The convergence has been

facilitated by innovation in contractual structures in the syndication market,

including the establishment of loan “packages” containing tranches with longer-

than-average maturities (10-12 years), bullet repayments of principal, and call

protection. These various innovations, in turn, are fueled by competition between

commercial and investment banks for syndication share and the increasing

availability of information about such deals. The American Banker reports

(August 1, 1997) that three information services—Loan Pricing Corporation,

Securities Data Company and Portfolio Management Data LLC—compete

aggressively in this market, updating their databases on-line on a daily basis.

Securities Data Company reports that among the top 25 managing agents in the

first ten months of 1997, investment banks had 18 percent market share,

5
compared to 5 percent over the same period in 1996 (American Banker,

December 12, 1997).

In a syndicated loan, two or more banks agree jointly to make a loan to a

borrower. As emphasized by Gorton and Pennachi (1995), loan syndications

differ from loan sales. A loan sale typically involves a “participation contract”

which grants the buyer a claim on all or part of a loan’s cash flows. The buyer of

a participation is an “indirect lender” with no relationship to the borrower. From a

relationship viewpoint, the purchaser of a participation is in a position similar to

the buyer of a public debt contract. In a syndication setting, each bank is a direct

lender to the borrower, with every member’s claim evidenced by a separate note,

although there is only a single loan agreement contract. One lender will typically

act as managing agent for the group, negotiating the loan agreement, then

coordinating the documentation process, the loan closing, the funding of loan

advances, and the administration of repayments. The agent collects a fee for

these services.

The syndicate members typically will have less interaction with the

borrower than the lead bank over the life of the loan. Consequently, the benefits

of an on-going relationship as a means of resolving agency problems are less

operative for the participating members, save for relationships based on prior

transactions with the borrower in question.

Agent banks have several potential motivations for syndicating loans.

Regulators limit the maximum size of any single loan to a portion of the bank’s

equity capital, so syndication can be a method to avoid “overlining.” Syndication

6
also may reflect a voluntary diversification motive, a mechanism for managing

interest rate risk, or a strategy for enhancing fee income. Participating banks

may be motivated by a lack of origination capabilities in certain geographic

regions or in certain types of transactions or a desire to economize on origination

costs. Pennachi (1988) suggests that loan purchasing banks may have funding

advantages relative to originators.

The agent bank commonly issues a commitment letter to a borrower in

which it may commit to fund an entire loan facility, or alternatively some portion

thereof, with a promise to use “good faith efforts” to arrange commitments from

other lenders for the remainder. If the agent commits for the entire amount, the

loan can be syndicated after it is closed, to the advantage of the borrower in the

sense that the funds are received more promptly.3 Otherwise, the loan must be

syndicated prior to closing. The agent bank prepares an “information

memorandum” that contains descriptive and financial information concerning the

borrower (including projections of cash flows). Recipients of the memorandum

sign a confidentiality agreement. The agent typically will meet with prospective

members to explain the terms of the credit, describe the borrower’s business and

prospects (often with presentations by the borrower’s management), and answer

questions.

The agent bank negotiates and drafts all the loan documents, but

syndicate participants can provide comments and suggestions when the

syndication occurs prior to closing. The participants are not generally involved in

the negotiations with the borrower, however. Acting as an intermediary, the

7
agent bank attempts to satisfy the potentially competing objectives of the

borrower and syndicate members.

The agent bank also facilitates the administration of the loan, typically

acting as a middleman for draws upon and repayment of the loan. The agent

calculates required interest payments, obtains waivers and amendments to the

loan documents, and in the case of a secured loan, holds all pledged collateral

(or is the grantee of relevant security interests) on behalf of the syndicate

members. The agent bank collects a fee for its services, which typically falls in

range of 10 to 40 basis points as a percentage of the facility. In some

transactions, the roles of the agent are divided among several institutions. A

“lead bank” negotiates the documents, puts together the syndicate, and closes

the loan and an “administrative agent” handles post-closing loan administration.

On occasion, a “collateral agent” is designated to monitor and administer the

collateral backing the loan. Fees are split in the case of multiple agents.

The “agency section” of the syndicated loan agreement formally

designates the agent bank and will provide for its removal under specified

conditions. The language typically exculpates the agent from any potential

liability to the syndicate members except where it results from “gross negligence

or willful misconduct.” While the language is crafted to temper or negate the

presence of a fiduciary duty on the part of the agent, attorneys typically counsel

the agent to administer the credit in good faith and exercise prudence and

reasonableness.4 Although standard provisions permit the agent to declare an

event of default, typically the agent will seek the prior advice of the member

8
banks. Indeed, the loan agreement will identify which decisions require the

consent of a designated proportion of the member banks. Unanimity is normally

required for any reduction in principal, interest or fees or for extensions of any

terms of the credit. In brief, the syndicate participants delegate some monitoring

responsibilities to the managing agent both at the loan origination or due

diligence stage and at the post-closing loan review stage.5

The loan syndication market invites potential agency problems involving

both adverse selection and moral hazard. The agent bank may possess

information unavailable to the syndicate participants. If the borrower is a long-

time customer, the originating agent bank may have obtained idiosyncratic

information regarding prospective performance that is not reflected in financial

statement data. Examples of such might involve judgments concerning

management expertise, the nature of customer-supplier relationships, or the

borrower’s capacity to adapt successfully to changing market conditions. The

originating bank has an incentive to syndicate those loans on which its “inside

information” is less favorable, to the potential economic detriment of the

participant banks.

As Gorton and Pennachi (1995) and others have noted, sales of loans

also generate potential moral hazard problems, since the seller has less

incentive to monitor once the loan is removed from the balance sheet.

Monitoring is a costly activity, but after the sale of a loan the benefits accrue to

the buyer rather than the loan originator. The moral hazard problem is potentially

less severe in the case of a loan syndication than a loan sale, since the

9
purchasing bank in a syndication holds a note against the borrower and has the

right to setoff against deposits. Nonetheless, the participating members have

delegated some monitoring responsibilities to the agent bank, in the sense that

members rely on the agent’s loan documentation, its enforcement of covenants,

and its administration of collateral. As the agent syndicates a larger proportion of

an individual loan, its incentive to monitor ex post declines monotonically. In

some instances, the agent bank will syndicate the entire amount of a loan facility.

These “information asymmetries” between the agent bank and syndicate

members are quite similar in nature to those which have been used to motivate

the existence of financial intermediaries. Intermediaries have been shown to

have a comparative advantage in solving these agency-related problems. (See

Bhattacharya and Thakor (1993) for a survey of this “existence” literature). How

are these similar problems overcome in the context of loan syndications? And

what do the “solutions” to these problems imply about where syndications “fit” in

the information spectrum? We hypothesize that the factors which determine

when a loan can be syndicated include the characteristics of (1) the borrower, (2)

the agent bank, and (3) the loan contract itself.

Research on loan syndications is relatively limited. In the only paper we

could uncover in a literature search, Simons (1993) examines empirically the

motives for syndications and addresses the issue of whether managing agents

are likely to “exploit” the syndicate member banks. She reports that the capital

position of the agent bank is a major factor affecting syndication activity and

suggests that diversification is the primary motive for syndication. Using bank

10
examiner ratings for the syndicated loans in her sample, Simons finds that

managing agents syndicate larger percentages of individual loans as the

examiner ratings improve.6 These ratings represent ex post evaluations of loan

quality, but Simons suggests “these loans may look less attractive to participants

even before they are criticized by examiners “ and that “the lead banks concern

with maintaining their reputations in the marketplace may lead them not only to

avoid abuses, but to promote risky loans even less aggressively than safe loans”

(p. 49). We investigate these and other issues more broadly and systematically

in our study, with a sample that includes non-syndicated as well as syndicated

loans. Simons’ sample consists only of syndicated loans.

II. Factors Affecting Loan Syndications

We specify a model that identifies the factors that determine whether a

loan is likely to be syndicated and, if so, to what extent. Our dependent variable,

the percentage of a particular loan that is syndicated, is somewhat similar to that

of Gorton and Pennachi (1995), who examine the proportion of a loan that is

sold. Their sample consisted strictly of loans sold, however. Our sample

includes non-syndicated loans, which assume a value of zero. The proportion of

a loan that will be syndicated depends generally on the agent’s underlying

motives for selling and on the scope of the agency problems associated with

syndication. Accordingly, variables that reflect potential information asymmetries

are a critical part of the model, as are variables that represent potential solutions

11
to these problems. Some of the latter variables may be captured in certain

characteristics of the individual loan, although some loan attributes—such as

loan size—may be related to the underlying motivations for syndication.

Similarly, certain characteristics of the agent bank may affect the syndication

potential of a given loan, either as a mechanism for resolving agency problems or

as a reflection of the underlying motives for syndications.

Pennachi (1988) demonstrates how agency problems limit loan sales in

the sense that a seller’s ability to market loans depends on the buyer’s

perception of the seller’s incentive to monitor. Pennachi argues that when the

benefits to monitoring are negligible, a loan can be sold in its entirety.

Greenbaum and Thakor (1987) demonstrate formally that, under certain

conditions, banks will sell or securitize higher-quality assets and retain lower-

quality loans on their balance sheets. Mester (1992) presents evidence

suggesting that it is less costly for a bank to monitor a loan that it has originated

compared to a loan that it has purchased. The implication of this research is that

loans that involve information which is “transparent” (easy to access, process,

and interpret) have higher syndication potential than loans involving “opaque”

(fuzzy, incomplete, difficult to observe and interpret) information. We employ

several different measures of the quality of the information available in a specific

loan transaction, including the existence of a public credit rating (either a bond

rating or a commercial paper rating), whether the borrower is a publicly-listed

firm, and the annual sales of the borrower in the year the loan closed. We argue

that information is likely to be more transparent when the borrower has a credit

12
rating or is a listed firm or when the borrower is large (as reflected in annual

sales). Increased transparency in turn raises the likelihood that a larger

proportion of a particular loan can be syndicated.

Certain characteristics of the loan itself may affect the agent bank’s

capacity to syndicate either because the characteristic serves to attenuate

agency costs or because it influences the perceived value to the buyer for non-

agency related reasons. The maturity and the status of the loan with respect to

collateral are two such characteristics. A number of potential channels exist that

might affect a loan’s syndication potential and the likely impacts are not uni-

directional.

If there is significant potential for the lead/agent bank to exploit the

syndicate members, then keeping loan maturity short could serve to minimize

such a prospect. Short-term loans involve less opportunity for the agent bank to

shirk, for example, and short maturities are likely to involve frequent requests for

renewals, which triggers more frequent monitoring of the borrower and the agent

by the syndicate members. Gorton and Pennachi (1989) argue that “banks are

less likely to shirk in information production or covenant monitoring” when selling

loan “strips,” which are short-term segments of longer-term loans (p. 130). The

reason is that the selling bank intends to resell the strip on the date it matures to

avoid having to fund it. These arguments suggest that lengthening a loan’s

maturity would reduce its potential for successful syndication. On the other hand,

frequent renewals also increase the overall (and duplicative) monitoring costs for

the set of syndicate banks. Diamond (1984) demonstrates how the avoidance of

13
duplicative monitoring costs helps provide a rationale for the existence of

financial intermediaries. Syndication results in duplicative monitoring by its very

nature. Since the majority of syndicated loans involve variable-rate pricing,

which minimizes interest rate risk, syndicate members might prefer longer-term

claims on the borrower’s cash flows to avoid “excessive” monitoring costs. In

addition, Rajan (1992) has emphasized that short-maturity loans create an

opportunity for the originating bank to extract rents from borrowers on the

renewal date whenever ex post information reveals a favorable state. If this is

the case, managing agents would prefer to hold larger proportions of such loans

in their own portfolios to avoid sharing such rents with syndicate members. If

avoiding duplicative monitoring costs or potential rent extractions are relevant

considerations, then lengthening a loan’s maturity would enhance its syndication

potential. We include the loan’s maturity as a variable in our model, but the likely

sign of this variable is ambiguous.

Similarly, the presence of collateral could, in principle, increase or reduce

a loan’s syndication potential. When a loan is fully secured, the quality of the

lenders’ monitoring effort assumes less importance. Collateral accordingly

reduces the sensitivity of the loan’s cash flows to any information differences

between the agent and syndicate member banks, suggesting that the presence

of collateral would raise the likelihood that a loan could be syndicated. On the

other hand, Berger and Udell (1990) demonstrate that collateral typically is

associated with riskier loans. This suggests that collateral could serve as a

signal that a particular loan involves a high degree of opaque information. Rajan

14
and Winton (1995) demonstrate formally that collateral is more likely to be

observed in loans to firms that need monitoring and that “collateralization of

private debt will be correlated with financial distress at the firm level and poor

business conditions at the aggregate level, both of which have empirical support”

(p. 115-16). These arguments suggest that the presence of collateral should

reduce the prospects of syndicating a loan. We include a dummy variable for the

presence of collateral in our model, but again we are agnostic about its sign.

If the agency problems between the agent and syndicate members are

potentially significant, another factor that could attenuate these problems is the

formation of a “reputation” for non-exploitative behavior by the lead bank.

Reputation has been proposed as a general solution to agency problems in

contracting in numerous settings, including audit quality [DeAngelo (1981)], bond

ratings [Wakeman (1981)], dividends [ Easterbrook (1984)], underwriting [Booth

and Smith (1986)], and the abnormal stock price response to loan agreement

announcements [Billett, Flannery, and Garfinkel (1995)]. James (1992)

emphasizes the role of relationship-specific assets in the pricing of underwriting

services. He notes that underwriters obtain information while evaluating a

potential issue that can be useful in underwriting subsequent offerings of the

same firm. In addition, the “investment banker may also identify an informed

client base for the firm’s stock which is also a durable transaction-specific asset”

(p. 1687). These arguments carry over directly to loan underwriting in a

syndicate context. A bank that has established transaction-specific assets

(reputation) should have lower costs in syndicating loans than banks that have

15
eschewed such investments. Gorton and Haubrich (1990) argue that the

reputation of the selling back may replace the selling bank’s equity as the

mechanism to ensure performance in a loan sale with no recourse.

In our model, we develop a proxy for “reputational capital” based on the

assumption that a large volume of repeat business between an agent originator

and a syndicate member reflects a relationship containing significant transaction-

specific investments. Using data from a period prior to that used in our

regressions, we determine the amount of repeat business generated by each

syndicate manager by creating “client lists.” We proceed through the data

chronologically by manager, summing the volume of loans sold to all members

who enter into multiple syndicates with this lead bank. We use this variable as

one measure of the reputation of the managing agent.

This technique for measuring reputation could erroneously reflect serial

correlation in syndications, in that these leading managers simply may have an

established strategy to syndicate a large proportion of their loans. As an

alternative measure of reputation, we employ the senior, unsecured debt rating of

the agent bank. We hypothesize that a loan can be more readily syndicated

when the lead bank has a higher credit rating. In addition, we include a dummy

variable equal to one if the loan originator is a bank and zero for non-banks, as

another reputational factor. Non-banks are relatively recent entrants into the

syndication market as originators, and we hypothesize that banks have more

“reputational capital” in the market than non-banks. This hypothesis would be

confirmed by a positive coefficient on this dummy. Preece and Mullineaux (1994)

16
and Billett, Flannery, and Garfinkel (1995) provide evidence that non-banks

recently have “mimicked” some of the unique characteristics of banks in

commercial lending, but banks may well maintain a comparative advantage in

syndications on reputational grounds.

We also include two different measures of the scale of the loan facility in

our model. The first is simply the size of the loan in dollars and the second is the

loan-facility size divided by the equity capital of the agent bank. As either of

these variables increases, a larger proportion of a loan is likely to be syndicated,

reflecting either discretionary or regulatory-driven motives for diversification.

In an expanded version of this “base-case” model, we employ several

other variables relating to the managing agent bank which have been utilized in

empirical studies of loan sales, such as Pavel and Phillis (1987). Among these

are the rate of growth in the loan portfolio over the most recent 12 months, the

charge-off ratio for total loans, and the ratio of non-current loans to total loans.7

The loan-growth variable is a rough measure of the extent to which the agent

bank is liquidity constrained. A higher value for this variable should be

associated with an increased prospect for syndication, other things equal. The

other two variables are measures of overall loan quality in the agent bank

portfolio. Increases in these ratios could reduce the likelihood that the originator

can syndicate large portions of individual commercial loans.

Many studies, including Berger and Udell (1993) and Pennachi (1988),

have suggested that regulatory requirements for bank capital may induce banks

to sell loans. Selling loans without recourse or syndicating loans results in their

17
removal from the originating bank’s balance sheet. Commercial loans are

required to be “backed” by equity capital equal to 4 percent of the loan amount.

We include two separate measures of “capital adequacy” in our expanded model:

the ratio of equity capital to total assets and the ratio of equity capital to risk-

adjusted assets (the Tier I capital ratio) in the year the loan is syndicated. If loan

syndications are motivated by capital requirements, increases in these capital

ratios should reduce the prospect that a given loan will be syndicated.

To re-capitulate, we propose a base-case model which takes the following

form:

SYNPERC = f (BONDRATE, MATURITY, SECURED, REPEAT, BANK, FACSIZE)

where SYNPERC is the percentage of a particular loan that is syndicated,

BONDRATE is a dummy variable which proxies for the scope of information

asymmetries and is equal to one if the borrower has a public bond rating and

zero otherwise, MATURITY is the maturity of the loan facility, SECURED is a

dummy variable equal to one if the facility is collaterallized and zero otherwise,

REPEAT is a continuous variable reflecting the dollar volume of repeat

transactions for each originating agent bank based on activity in the pre-sample

period, BANK is a dummy variable equal to one if the managing agent is a bank

and zero otherwise, and FACSIZE is the dollar amount of the loan facility. The

coefficients of BONDRATE, REPEAT, BANK, and FACSIZE are hypothesized to

be positive, while the signs of the MATURITY and SECURED coefficients are

ambiguous. As alternative measures of information asymmetry to BONDRATE,

we employ CPRATE (one if the borrower has a commercial paper rating and zero

otherwise), TICKER, which is a dummy variable equal to one if the borrower is

18
listed on the NYSE, AMEX, or NASDAQ stock exchanges and zero otherwise,

and SALES which reflects the annual sales of the borrowing firm in the year of

the syndication. The coefficients of the alternative dummies should be positive,

since the presence of a commercial paper rating or listing of a borrower’s stock is

suggestive of increased information transparency. A similar argument can be

made for SALES; larger firms may be characterized by fewer information

asymmetries. As an alternative measure of managing agent reputation to

REPEAT, we employ BANKRATE, which is the senior, unsecured debt rating of

the agent bank. The coefficient of BANKRATE should be positive; the higher the

credit rating of the agent, the larger will be the proportion of its loans that can be

syndicated. We estimate this model using a sample of 3,410 loans originated by

banks and non-banks over the period 1987-95.

We also estimate an expanded version of the model for loans originated

only by banks. The additional variables each represent characteristics of the

managing agent bank in the year the loan is syndicated. These include:

LNGROWTH, the rate of growth in loans; CHARGE-OFFS, the ratio of loans

charged off to total loans; NCURRENT, the ratio of non-current loans to total

loans; EQUITY, the ratio of equity capital to total assets; and TIER I, the ratio of

core capital to risk-adjusted assets. The proportion of loans that can be

syndicated should increase with LNGROWTH, which proxies for the degree the

agent is liquidity constrained, and should decline with the remaining three

variables, which measure the quality of the overall loan portfolio and the degree

to which the bank is capital constrained.

19
III. Estimates of the Model

Since our study requires detailed information on the characteristics of

individual loans, we employ data from a private database compiled by the Loan

Pricing Corporation (LPC). The data set contains information on approximately

30,000 loan facilities involving some 2,500 lenders. Much of the data gathered

by LPC comes from commitment letters or credit agreements contained in public

filings with the Securities and Exchange Commission, but LPC also reports deals

obtained from direct research at the lending banks. LPC attempts to confirm the

data from SEC filings with senior management at the lender and reports when a

transaction is “fully confirmed,” “partially confirmed,” or “unconfirmed.” We

selected all non-private placement, fully-confirmed loan transactions where we

could identify either the managing agent’s share or the percentage of the loan

syndicated. This resulted in a sample of 3,410 loan transactions of which 1,526

were syndicated. Non-syndicated loans are those with a percent syndicated

equal to zero. Table I provides some descriptive statistics on the sample.

Of the 3,410 transactions, 2,769 (81.2%) were originated with borrowers

who lacked a senior, unsecured S&P credit rating. About 93 percent of the non-

syndicated loans went to non-rated borrowers, while roughly 67 percent of the

syndicated loans went to unrated companies. Even larger proportions of the loan

facilities were with firms who lacked commercial-paper ratings. Roughly half of

the loan transactions are with publicly-listed firms. A larger proportion of the non-

20
syndicated loans are secured, although the majority of loans involve collateral in

both classes. Banks originated a larger proportion of the syndicated loans (97%)

than the non-syndicated facilities (86%). Not surprisingly, syndicated loans are

much larger than non-syndicated ones and are more likely to be originated with

larger firms. The magnitudes of the differences are roughly the same for means

and medians. The mean maturity of the syndicated loans is almost 50 percent

larger than the average for non-syndicated loans and the median is twice as

large.8 Table 1 also shows the average share held by a managing agent in a

syndication setting is 32 percent, with a range from 0 to 90.7 percent. A value of

zero means the entire amount of the loan was syndicated. Roughly 50 percent of

the syndications involve an assignment minimum with an average value of $7.5

million. Prior approval from the borrower for assignments is required in roughly

43 percent of the syndications and managing agent approval for subsequent

assignments is required 45 percent of the time in our sample. The majority of the

loans in our sample are for general corporate purchases, working capital, or debt

repayment and no significant differences exist between syndicated and non-

syndicated loans with respect to loan purpose or loan type. A strong majority of

both types of loans are either revolvers or term loans.

We estimate the model(s) with maximum likelihood techniques, using a

doubly censored Tobit model (since the dependent variable is censored at 0 and

100 percent.) The results of the base case model are presented in Table II.

Three of the four measures of information asymmetry are significant at the

.01 level (BONDRATE, CPRATE, AND TICKER), while the coefficient of

21
borrower sales is significant only at the .10 level and only when BANKRATE is

used as a measure of reputation. Having a public credit rating or being publicly

listed allows a managing agent to syndicate larger proportions of particular loans,

presumably because the information available about such borrowers is relatively

transparent. When the information set is opaque, the agent holds a larger

proportion of the loan in its portfolio, perhaps in part as a signal regarding the

“quality” of the credit. Sales appears to be a weak proxy for the information

characteristics of the borrower.

The coefficient of the loan’s maturity is positive and significant at the .01

level. Longer maturity enhances a loan’s syndication potential, presumably

because longer-term loans economize on duplicative monitoring costs for the

syndicate banks. Another interpretation is that managing agents prefer to hold

larger proportions of short-term loans in their portfolios because they offer

opportunities to extract rents from borrowers in the renewal stage when favorable

information is revealed. Our findings are not consistent with the notion that short

maturities, and consequent frequent re-contracting, are a solution to potential

agency problems within the syndicate.

The coefficient of the dummy variable for collateral is negative and

significant at the .01 level, suggesting that agents are able to syndicate smaller

proportions of secured relative to unsecured loans. While collateral serves to

reduce the sensitivity of a loan’s value to the character of the information that

supports it, collateral also can serve as a signal of a higher level of risk. In the

syndication market, the latter effect appears to dominate.

22
 The variables which serve as proxies for the reputation of the managing

agent are all positive and significant at either the .01 or .05 level, suggesting that

reputation can serve as a method for attenuating potential agency problems, as

many researchers have claimed. The proportion of a loan that can be syndicated

increases as the managing agent is more heavily involved in repeat business, as

the agent’s credit rating improves, and as the agent is a bank rather than a non-

banking institution. These results suggest that certain managing agents have

successfully developed transactions-specific assets that facilitate their capacities

to syndicate loans. Our results are also similar to those of Billett, Flannery and

Garfinkel (1995), who found that the identity of the lender affects the size of the

equity market’s response to a loan agreement announcement. The identity of the

managing agent bank also influences the amount of a particular loan that can be

syndicated.

Finally, the size of the loan facility is a significant factor in determining the

proportion of a loan that is syndicated. The larger the loan, the larger the

percentage the managing agent will place with participating members.

Table III reports the results of the estimation of the bank-only model,

which includes several agent-specific variables that have been employed in

various studies of loan sales rather than syndications. The coefficient of

LNGROWTH, the rate of increase in loans, is generally insignificant, suggesting

that agent banks are not generally motivated by liquidity constraints in their

syndication activity.9 When the LNGROWTH coefficient is significant (one of four

cases), it is incorrectly signed according to the liquidity hypothesis. The

23
coefficient of the charge-off ratio is likewise generally insignificant, suggesting

that a higher level of “bad loans” in the agent bank’s portfolio does not inhibit its

capacity to syndicate. When instead the ratio of non-current loans to total loans

is included in our regressions, its coefficient is significant (at the .01 level) in one

of two cases and has a positive sign, suggesting that agent banks syndicate

larger proportions of commercial loans when the overall quality of the loan

portfolio deteriorates. Again, evidence implies that the credit quality of the

managing agent bank’s overall portfolio does not inhibit its capacity to

syndicate.10 Both measures of capital adequacy at the agent bank have negative

and significant coefficients, suggesting that agents which are capital constrained

are likely to syndicate larger proportions of their loans. These results are similar

to the findings in Simons (1993) for syndications and Pavel and Phillis (1987) for

loan sales and are consistent with the “regulatory tax hypothesis” in the loan

sales literature [Berger and Udell (1993)]. All of the variables in the base-case

model remain significant when these additional variables are included and each

maintains its original sign. In broad terms, our evidence suggests there are

some similarities and some differences in the factors that drive loan syndications

and loan sales.

IV. Concluding Discussion

In this paper, we have analyzed one of the most rapidly growing sectors of

the financial system, the market for syndicated loans. The volume of syndicated

24
lending in 1997 was over 8 times the amount of new junk-bond issues. Our main

objective was to identify the factors that enhance the prospects that an individual

loan can be syndicated. We found evidence that the characteristics of the

borrower, the managing agent, and the loan contract itself all are of some

relevance. In particular, the better the quality of the information about the

borrower as reflected in either credit ratings or listing on a stock exchange, the

larger is the proportion of its debt that can be syndicated. This confirms the now

standard view that the scale and scope of information asymmetries is relevant to

where and how a firm finances. We also found evidence that the reputation of

the managing agent, as reflected in the volume of repeat business or in the

agent’s own credit rating, can attenuate some of the agency problems that

adhere in loan syndications. While many authors, such as Gorton and Haubrich

(1990) and Gorton and Pennachi (1989) have suggested that reputation might

resolve moral-hazard problems in the loan sales market, our evidence confirms a

reputational effect. This result confirms the relevance of relationships as a

mechanism for resolving agency problems in debt contracting. While the context

of syndication makes it difficult for members to develop and exploit relationships

with the borrower, participatory banks can and do rely on relationships with the

lead bank to mitigate problems associated with information asymmetrics.

We also found that an individual loan’s characteristics can affect the

degree to which it can be syndicated. Indeed, the syndication market serves as

a type of “laboratory” in which some important hypotheses and principles in

finance are tested in a real-world setting. Much research in finance and

25
economics has suggested, for example, that keeping contractual relations brief

can be a mechanism for mitigating potential agency problems. This logic

suggests that short-term loans should be more readily syndicated. Our evidence

finds an opposite result—that lengthening a loan’s maturity allows an agent to

syndicate larger proportions of a given loan. Two reasons suggest why this may

be a logical outcome. By its nature, syndication involves duplicative monitoring

costs in what Simons (1993) labels “secondary intermediation.” When loan

maturities are lengthened, the scale of these duplicative costs is reduced and the

value of participating in syndications as a lending strategy increases. A second

reason why longer-term maturity loans may be more heavily syndicated is that,

as Rajan (1992) has demonstrated, short-term loan contracts create an

opportunity for the lender to extract rents from the borrower at the contract

renewal stage, if favorable information emerges ex post. While syndicate

members might prefer to share in any such rent creation, they cannot

contractually commit the agent to re-syndicate such loans at the renewal stage.

Consequently, larger proportions of short-term loans involving the potential for

rent extraction will be held in the managing agent bank’s portfolio. The fact that

longer-maturity loans are more likely to be syndicated implies that the loan

syndication market tends to overlap somewhat with the private-placement market

with respect to maturity.

Another loan characteristic that could serve to attenuate agency problems

between the agent and members is collateral, since pledging an asset to secure

a loan reduces the impact of any information asymmetries between the

26
contracting parties. This logic implies that secured loans should be more readily

syndicated. Again, our evidence is to the contrary. Smaller portions of the

collaterallized loans tend to be syndicated, perhaps because collateral acts as a

signal that information about the borrower tends to be more opaque than

transparent or as a signal of prospective financial distress. The presence of

collateral might also signal the absence of an on-going relationship between the

borrower and the lead bank. Berger and Udell (1995) report evidence that

sufficiently extensive relationships can substitute for collateral as devices for

resolving agency problems. On balance, potential agency problems within the

syndicate appear to be primarily resolved by the managing agent’s behavior and

characteristics rather than the loan’s characteristics. Our results for the collateral

and maturity variables do suggest, however, that information asymmetries are

relevant to the issue of a loan’s salability in a syndication setting.

On the supply side, some of the agent bank’s balance sheet

characteristics also facilitate syndication. In particular, capital constrained banks

syndicate larger proportions of their loans, which complements findings in the

loan-sales literature. However, liquidity considerations do not drive syndication

behavior as evidence shows it does in the loan sales market. Nor does the

overall credit quality of the managing agent’s portfolio matter much for

syndications.

Our evidence can be interpreted as broadly consistent with Diamond’s

(1991) life-cycle model of borrower choice and with the notion that the locus of an

individual firm on an information continuum extending from highly opaque to fully

27
transparent will largely determine a borrower’s likely source of financing. As

CPRU note, Diamond’s model is highly theoretical and abstracts from differences

in the features of various financial contracts, such as maturity and the presence

of collateral and covenants. They argue that these features, and the dynamic

nature of information production at both the loan origination and the monitoring

(loan review) stages, are critical to understanding the multiplicity of debt

contracts and financing sources. CPRU focus primarily on private placements

which they characterize as appropriate for firms who are “less information

problematic” than commercial bank borrowers and “more information

problematic” than borrowers who can access the public credit markets. Where

do syndicated loans fit on this continuum? Loan syndication also involves the

sale of contracts which are, on average, more informationally opaque than

contracts sold in the public debt market and perhaps even somewhat more

opaque than private placements.11 For example, a larger proportion of the

syndicated loans in our sample involve collateral (75%) than in Kwan and

Carlton’s (1993) sample of 658 private placements (33%).

Perhaps the most fundamental contribution of our paper is additional

empirical support for the hypothesis that the character and quality of information

concerning borrowers affects the salability of debt claims. Where information is

less than fully transparent, debt contracts tend to be marketed to investors with

specialized monitoring skills who rely on contractual characteristics and seller

reputation to resolve information asymmetry and agency problems.

28
 References

Berger, Allen and Gregory Udell, 1990, Collateral, loan quality, and bank risk,
Journal of Monetary Economics 25, 21-42.

Berger, Allen and Gregory Udell, 1993, Securitization, risk, and the liquidity
problem in banking, in Michael Klausner and Lawrence White, Eds.:
Structural Change in Banking (Irwin Publishing).

Berger, Allen and Gregory Udell, 1995, Relationship lending and lines of credit in
small-firm finance, Journal of Business 68, 351-82.

Berlin, Mitchell and Jan Loeys, 1988, Bond covenants and delegated monitoring,
Journal of Finance 43, 397-412.

Berlin, Mitchell and Loretta Mester, 1992, Debt covenants and renegotiation,
Journal of Financial Intermediation 2, 19-133.

Bhattacharya, Sudipto and Anjan Thakor, 1993, Contemporary banking theory,

Journal of Financial Intermediation 3, 2-47.

Billett, Matthew, Mark Flannery and Jon Garfinkel, 1995, The effect of lender
identity on a borrowing firm’s equity return, Journal of Finance 50, 699-
718.

Booth, James and Richard Smith, 1986, Capital raising, underwriting, and the
certification hypothesis, Journal of Financial Economics 15, 261-81.

Carey, Mark, Stephen Prowse, John Rea, and Gregory Udell, 1993, The
economics of private placements: a new look, Financial Markets,
Institutions, and Instruments 2, 1-67.

Cole, Rebel, 1998, The importance of relationships to the availability of credit,

Journal of Banking and Finance 22, 959-77.

DeAngelo, Linda, 1981, Auditor size and audit quality, Journal of Accounting and
Economics 3, 183-99.

Diamond, Douglas, 1984, Financial intermediation and delegated monitoring,

Review of Economic Studies 51, 393-414.

29
Diamond, Douglas, 1991, Monitoring and reputation: the choice between bank
loans and privately-placed debt, Journal of Political Economy 99, 689-721.

Easterbrook, Frank, 1984, Two agency costs explanations of dividends,

American Economic Review 74, 650-59.

Fox, Charles, 1989, Structuring and documenting syndicated bank credits, in The
Banker’s Guide to Multi-Bank Credits (Executive Enterprises Publications).

Gorton, Gary and Joseph Haubrich, 1990, The loan sales market, in George
Kaufman, Ed.: Research in Financial Services: Private and Public Policy
(JAI Press).

Gorton, Gary, and George Pennachi, 1989, Are loan sales really off balance
sheet?, Journal of Accounting, Auditing, and Finance 4, 125-45.

Gorton, Gary and George Pennachi, 1995, Banks and loan sales: Marketing non-
marketable assets, Journal of Monetary Economics 35, 389-411.

Greenbaum, Stuart and Anjan Thakor, 1987, Bank funding modes: Securitization
versus deposits, Journal of Banking and Finance 11, 379-401.

James, Christopher, 1987, Some evidence on the uniqueness of bank loans,

Journal of Financial Economics 19, 217-35.

James, Christopher, 1988, The use of loan sales and commercial letters of credit
by commercial banks, Journal of Monetary Economics 47, 1865-86.

Klein, Benjamin and Keith Leffler, 1981, The role of market forces in assuring
contractual performance, Journal of Political Economy 89, 651-71.

Kwan, Simon and Willard Carlton, 1993, The structure and pricing of private
placement corporate loans, University of Arizona working paper.

Mester, Loretta, 1992, Traditional and nontraditional banking: An information

theoretic approach, Journal of Banking and Finance 16, 545-66.

Nassberg, Robert, 1981, Loan documentation: Basic, but critical, Business

Lawyer 36, 843-933.

Pavel, Christine and David Phillis, 1987, Why banks sell loans: An empirical
analysis, Economic Perspectives, Federal Reserve Bank of Chicago, 3-14.

Pennachi, George, 1988, Loan sales and the cost of bank capital, Journal of
Finance 43, 375-96.

30
Petersen, Mitchell and Raguraham Rajan, 1994, The benefits of lending
relationships: Evidence from small business data, Journal of Finance 49,
3-37.

Preece, Dianna and Donald Mullineaux, 1994, Monitoring by financial

intermediaries: Banks versus nonbanks, Journal of Financial Services
Research 8, 193-202.

Preece Dianna and Donald Mullineaux, 1996, Monitoring, loan renegotiability and
firm value, Journal of Banking and Finance 20, 577-91.

Rajan, Raguraham and Andrew Winton, 1995, Covenants and collateral as

incentives to monitor, Journal of Finance 50, 1113-1146.

Rajan, Raghuram, 1992, Insiders and outsiders: The choice between informed
and arm’s length debt, Journal of Finance 47, 1367-1400.

Simons, Katerina, 1993, Why do banks syndicate loans?, New England

Economic Review, Federal Reserve Bank of Boston, 45-52.

Wakeman, Lawrence, 1981, The real function of bond rating agencies, Chase
Financial Quarterly, 1, 3-12.

FOOTNOTES

*The authors wish to thank Marcia Cornett, Mark Flannery, Chris James, Tony Saunders, and Joe

Sinkey for their comments and suggestions. The usual disclaimer applies.
1
From a legal standpoint, private placements do not represent “underwritings,” with the exception

of certain private placements issued under SEC Rule 144A. CPRU (1993) argue that Rule 144A

securities have characteristics nearly identical to public bonds.

2
The recent volume of syndication activity is especially remarkable, given that the volume of

commercial and industrial loans on U.S. bank balance sheets as of December 31, 1996 was

$783.2 billion, according to the Federal Reserve Bulletin. The syndication activity reported in

Gold Sheets includes sales by non-U.S. banks, however.

3
In the event of a post-closing syndication, the managing agent and the participating member

sign an “assignment and assumption agreement” in which the agent assigns the agreed upon

amount of its commitment to the member. Legally, this process is described as “novation,” which

31
represents an amendment to the original loan agreement. The borrower in turn delivers a note to

the member reflecting the assignment amount. Managing agents can establish a minimum

required assignment amount, which results in a limit on the number of banks in the syndicate.

Borrowers sometimes demand the capacity to approve the participating banks and, on occasion,

the managing agent will require that syndicate banks seek their approval if the member wishes to

subsequently assign their own interest to some other bank. Fox (1989) reports that post-closing

syndications are “increasingly common.”

4
Nassberg (1981) notes that the agent bank is typically subject to a liability claim in the event of

“bad faith” or “gross neglect.” Such claims are infrequent, but in 1989 Security Pacific Bank was

sued by 10 plaintiff banks, which held $67.6 million of its syndicated loans, on the grounds that

Security Pacific withheld information. The judge ruled in favor of Security Pacific (Forbes, May

27, 1991).
5
CPRU (1993) note that private placements facilitated by agents, rather than placed directly by

the borrower, involve a similar form of delegated monitoring. They claim that potential agency

problems are more severe in the post-closing stage, since lenders will likely replicate fully the due

diligence efforts of the agent. Simons (1993) claims that, although syndicate members are

expected to perform their own credit analyses rather than rely on representations made by the

agent, buyers typically will rely strictly on the loan documentation provided by the lead bank in

conducting such evaluations. This suggests the agency problems may be equally severe in both

stages of the monitoring process in the case of loan syndications.

6
Simons reports that banks syndicated roughly 82 percent of loans rated “pass” or “special

mention” by examiners, but only 70 percent of loans rated “substandard” or “doubtful,” and 53

percent of loans rated “loss.”

7
Ideally, we would also examine the role of charged-off commercial loans and non-current

commercial loans as determinants of syndication behavior. Unfortunately, our data source, the

FDIC’s Institutional Directory Web Page, does not contain charge-offs or non-current loans by

loan type.

32
8
CPRU (1993) report that the average size of a private placement (in 1989) was $76 million with

a mean maturity of nine years. While our data cover a longer time horizon, the average facility

size for syndicated loans ($242 million) appears substantially larger than for private placements

and the average maturity is notably shorter (3.65 years).

9
Liquidity considerations have been shown to be a factor influencing loan sales by Pavel and

Phillis (1987) and Berger and Udell (1993), among others.

10
Simons reports that banks syndicate smaller percentages of individual loans as the credit

quality of the loan in question deteriorates as measured by subsequent examiner ratings.

11
Preece and Mullineaux (1996) report evidence that announcements of syndicated loan

agreements fail to produce the positive returns to borrower equity associated with non-syndicated

loan announcements. This result is similar to James’ (1987) findings for private placements.

33
 Table I

Descriptive Statistics for the Sample

Syndicated Nonsyndicated
Full Sample Loans Loans
Sample Size 3410 1526 1884
Loans to non-rated borrowers (senior, 2769 1016 1753
unsecured debt)
Loans to non-rated borrowers 3068 1251 1796
(commercial paper)
Loans to publicly-listed borrowers 1729 858 871
Number of secured loans 2766 1139 1627
Observations with bank/non-bank 3142/268 1478/48 1664/220
lenders
Borrower Sales: Mean/Median (Min/Max) $767m/$113m $1.5b/$337m $207m/$50m
(13,000/109b) (1m/109b) (13,000/14.6b)

Facility size: Mean/Median (Min/Max) $119m/$15m $242m/$69m $18.7m/$5m

(42,856/8.6b) (.6m/8.6b) (42,856/1.4b)
Maturity in years: Mean (Min/Max) 2.9/2.2 3.7/3.2 2.5/1.5

Syndication information:
Agent Share: Mean (Min/Max) 32%
(0/90.7)
Assignment Minimum: Mean $7.5m
(Min/Max) (10,000/25m)
Company Consent for Assignment 661/865
(Yes/No)
Agent Consent for Assignment 679/847
(Yes/No)
 Table II

Estimates of the Base-Case Models(s) for Percent of Loan Syndicated

Tobit regressions of the percent of an individual loan that is syndicated on the following independent variables: BONDRATE, a dummy equal to
one if the borrower has a senior, secured debt rating and 0 otherwise; CPRATE, a dummy equal to one if the borrower has a commercial paper
rating and 0 otherwise; TICKER, a dummy equal to one if the borrower is listed on the NYSE, AMEX or NASDAQ and 0 otherwise; SALES, the
annual sales of the borrower in the year the loan is syndicated; MATURITY, the maturity of the loan (in days); SECURED, a dummy equal to one if
the loan is collateralized and 0 otherwise; REPEAT, the number of repeat transactions in the pre-sample period conducted with the managing
agent bank syndicating the loan; BANKRATE, the senior, secured debt rating of the agent bank; BANK, a dummy equal to one if the managing
agent has a bank charter and 0 otherwise; and FACSIZE, the dollar value of the size of the loan facility. The numbers in parentheses are standard
errors. The sample size is 3,410. Three *Õs denotes significance at the .01 level, two at .05 and one at .10.

Model A Model B Model C Model D Model E Model F Model G Model H

INTERCEPT 4.05*** 4.04*** 4.05*** 4.03*** 4.03*** 4.02*** 4.06*** 4.04***
(0.04) (0.04) (0.04) (0.04) (0.04) (0.05) (0.04) (0.04)
BONDRATE 0.104*** 0.127***
(0.017) (0.017)
CPRATE 0.083*** 0.102***
(0.022) (0.022)
TICKER 0.045*** 0.035***
(0.014) (0.015)
SALES 2.222 E-12 5.532 E-12*
(2.45 E-12) (2.96 E-12)
MATURITY 3.92 E-5*** 4.39 E-5*** 4.53 E-5*** 5.173 E-5*** 4.11 E-5*** 4.61 E-5*** 4.22 E-5*** 4.74 E-5***
(8.07 E-6) (8.31 E-6) (8.32 E-6) (8.66 E-6) (8.21 E-6) (8.55 E-6) (8.45 E-6) (8.82 E-6)
SECURED -0.065*** -0.062*** -0.074*** -0.075*** -0.076*** -0.081*** -0.082*** -0.079***
(0.018) (0.018) (0.018) (0.019) (0.018) (0.018) (0.018) (0.020)
REPEAT 0.002*** 0.002*** 0.002*** 0.002***
(0.0002) (0.0002) (0.0002) (0.0002)
BANKRATE 0.004** 0.006*** 0.006*** 0.005**
(0.002) (0.002) (0.002) (0.002)
BANK 0.137*** 0.190*** 0.154*** 0.214*** 0.160*** 0.227*** 0.151*** 0.224***
(0.039) (0.040) (0.039) (.040) (0.039) (0.04) (0.04) (0.041)
FACSIZE 8.47 E-11*** 1.123 E-10*** 6.98 E-11*** 9.43 E-11*** 8.67 E-11*** 1.24 E-10*** 7.84 E-11*** 1.031 E-10***
(1.77 E-11) (1.89 E-11) (1.89 E-11) (2.05 E-11) (1.18 E-11) (2.04 E-11) (2.09 E-11) (2.34 E-11)
 TABLE III

Estimates of the Expanded Model for Percent of Loans Syndicated

Tobit regression estimates based on a sample of bank-only syndications that includes the variables in the base-case model with the exception of
BANK, CPRATE and SALES, but also includes as independent variables: LNGROWTH, the rate of growth in loans of the agent bank in the year
the loan is syndicated; CHARGE-OFFS, the ratio of loans charged off by the agent bank to total loans, NCURRENT, the ratio on non-current loans
to total loans, EQUITY, the ratio of equity capital to total assets, and TIER I, the ratio of core capital to risk adjusted assets. We also replace the
size of the facility in these regressions with RELSIZE, which is the facility size divided by the equity of the managing agent bank. The figures in
parentheses are standard errors. The sample size is 1,663. Three *Õs denotes significant at the .01 level, two at .05 and one at the .10 level.

Model A Model B Model C Model D

INTERCEPT 4.41*** 4.43*** 4.36*** 4.28***
(0.07) (0.08) (0.05) (0.05)
BONDRATE 0.114*** 0.151*** 0.117***
(0.02) (0.02) (0.02)
TICKER 0.037***
(0.02)
MATURITY 3.65 E-5*** 3.32 E-5*** 4.57 E-5*** 3.35 E-5***
(1.00 E-5) (1.1 E-5) (1.1 E-5) (1.0 E-5)
SECURED -0.108*** -0.114*** -0.139*** -0.112***
(0.02) (0.02) (0.02) (0.02)
REPEAT 0.002*** 0.003*** 0.003***
(3.35 E-5) (3.33 E-5) (3.34 E-4)
BANKRATE 0.011***
(0.003)
LNGROWTH -0.053 -0.054 -0.114*** -0.056
(0.17) (0.04) (0.04) (0.04)
CHARGE-OFFS -6.53 E-4 -0.008
(0.01) (0.12)
NCURRENT 0.029*** 0.005
(0.007) (0.007)
EQUITY -0.022*** -0.023***
(0.006) (0.006)
TIER 1 -0.004*** -0.003***
(0.001) (0.001)
RELSIZE 1.43 E-4*** 1.53 E-4*** 1.05 E-4*** 1.43 E-4***
(5.1 E-5) (5.5 E-5) (4.5 E-5) (5.2 E-5)

View publication stats