Do Lending Relationships Affect Corporate Financial Policies?

• Date: 01 March 2016
• DOI: http://doi.org/10.1111/fima.12097
• Published date: 01 March 2016
• Author: Hadiye Aslan

Do Lending Relationships Affect

Corporate Financial Policies?

Hadiye Aslan∗

This paper provides new evidence on how lending relationships impact firms’ financing and

investment decisions. I find that lending relationships have a significant impact on leverage

ratios, issuance choices, and the investment structures of relationship borrowers. The influence

of relationships is heightened for financially constrained firms. I find a significant decrease in

leverage, net debt issuing, and investment activity in the aftermath of lender-specific shocks to

lending relationships, including announcements of bank write-downs and downgrades in banks’

credit ratings. My findings are robust to controlling for confounding effects that might arise due

to unobserved demand and relationship changes.

Academics have recently focused considerable attention on the role of banks as relationship

lenders. The literature argues that relationship lending facilitates monitoring and screening and

can overcome problems of asymmetric information between the bank and the firm. There is a

large body of evidence regarding the costs and benefits of lending relationships on the terms of

loan contracts (Hoshi, Kashyap, and Scharfstein, 1991; Petersen and Rajan, 1994; Berger and

Udell, 1995; Bharath et al., 2011). The literature also provides evidence on whether initiating or

renewing a specific banking relationship creates value for the borrower (James, 1987; Lummer

and McConnell, 1989; Dahiya, Saunders, and Srinivasan, 2003).

However, the dynamic impact of lending relationships on the investment and financing policies

of firms is often ignored even though bank loans are the major source of external capital for cor-

porations in many economies. While it is interesting to recognize howa given lending relationship

creates value, we should understand howrelationship lending affects the dynamic investment and

financing decisions of borrowers through its influence on the borrowers’ access to funds from

the financial markets. The purpose of this study is to f ill these gaps in the literature. Using a

comprehensive loan sample covering a broad panel of US firms from 1990 to 2011, I provide a

broad-based analysis of the effect of lending relationships on corporate investment and financing

policies.1My main findings can be summarized as follows. Investment and financial policies are

systematically related to the presence of lending relationships. Depending on the definition of

lending relationships used in the empirical specifications, the estimated increase in debt ratios due

I thank Raghavendra Rau (Editor) and an anonymous referee for very helpful comments. I also thank Sandeep Dahiya,

YanivGrinstein, Maureen O’Hara,and participants at seminars at the University of Houston, Koc University, the American

Economic Association 2013 Annual Meetings, European Finance Association 2013 Annual Meetings, and the Lone Star

Finance Conference for useful comments or discussions on the issues addressed in this paper. I thank the European

Finance Association for awardingthis paper the 2013 S.A.C. Capital Advisors, L.P. Prize forthe Best Conference Paper.

∗Hadiye Aslan is an Assistant Professor in the J. Mack Robinson College of Business at Georgia State University in

Atlanta, GA.

1While earlier studies (Berger and Udell, 1995; Bharath et al., 2011) demonstrate that relationship lending results in

better loan terms, these papers do not directly estimate the real effects of relationships on subsequent firm behavior.

Financial Management •Spring 2016 •pages 141 – 173

142 Financial Management rSpring 2016

to relationships ranges from 0.029 to 0.040 (i.e., between 10.8% and 14.9% of mean leverage).

The evidence is also consistent with higher total investments for stronger lending relationship

firms. For instance, firms with lending relationships have 2.1% greater total investments relative

to non-relationship borrowers. This result represents an increase in investmentsof approximately

16.3% relative to the average investments of the firms in the sample.

However, there is substantial heterogeneity in these real effects with respect to the empirical

proxies of credit frictions. For example, while unrated borrowers have lower leverage and invest-

ment ratios overall, the benefits of borrowing from a relationship lender are especially strong for

these borrowers. More specifically, I find that a relationship borrower that is credit constrained

when compared to an unconstrained borrower has, on average, a 16.7% (an increase of 0.045

from a mean of 0.268) higher book leverage ratio and a 20.1% (an increase of 0.026 from a mean

of 0.129) higher total investment.

Analyzing the real effects of lending relationships poses challenging identification problems.

Since the matching between borrowers and lenders is not random, failing to control for this

endogeneity can confound relationship lending effects with clientele effects and could lead to

incorrect conclusions. To dispel these concerns, I use an endogenous switching regression model,

a generalization of the classic Heckman (1979) two-stage procedure, which also allows me to

pose a “what-if” question. For a relationship borrower, what would the alternative investment

and financing outcome be, had there been no such commitment? The resultant analysis provides

estimates of the (unobserved) counterfactual outcomes that are useful for the basic issue at

hand. I find that the counterfactual book leverage (total investments) for a relationship borrower

would have been 4.7% (2.2%) lower if the same borrower had instead opted for non-relationship

borrowing. Conversely, an average non-relationship borrower hypothetically would have had

higher leverage (total investments) by 4.1% (5.3%) if relationship lending had been employed.

Another concern is that a change in financing and investment activity would have occurred re-

gardless as to the existence of lending relationships. Toalleviate this worry, I consider disruptions

in relationships due to an adverse shock to the financial health of a lead relationship lender and

conduct an event study.I rely on the occurrence of bank-specific events, such as announcements

of bank write-downs or downgrades, to net out any unobservable demand effects. Finally, to

isolate the magnitude of the lender’s distress, I use an instrument that captures banks’ exposure

to toxic mortgage-backed securities (ABX exposure) during the recent financial crisis of 2007 as

a source of exogenous variation in the availability of credit to borrowers.

Overall, the results of this study complement the existing banking literature by quantifying

the real effects of lending relationships. The current research on banking relationships argues

that establishing a close lending relationship with a bank can mitigate information asymmetries

and create value for the borrower. This value creation could be in the form of lower financing

costs, fewer collateral requirements, contract flexibility, re-negotiations of credit contracts, or the

extension of additional loans when a corporation is in financial distress (Boot and Thakor, 1994;

Petersen and Rajan, 1994).

While relationship banking has benefits for borrowers, offsetting costs exist that prevent firms

from borrowing exclusively from relationship lenders. For example, Sharpe (1990) and Rajan

(1992) argue that while a relationship lender can reduce agency problems, the firm-specific

information about borrowers that banks obtain as a part of their relationships may create a hold-

up problem. In other words, the proprietary information produced and used by the relationship

lender increases its ex post bargaining power. This may be exploited to the detriment of the

firm (Greenbaum, Kanatas, and Venezia, 1989; Sharpe, 1990; Rajan, 1992). Thus, these firms

may choose non-noptimal investment and financing policies, and potentially valuableinvestment

opportunities may be lost. At the extreme, when a relationship bank is also a shareholder of the

Aslan rDo Lending Relationships Affect Corporate Financial Policies? 143

firm, the bank may influence corporate decisions in its favor, making other creditors less willing

to provide additional credit due to potential conflicts of interest.

Similarly,relationship banking can also affect the f irms’ choice of debt and investmentthrough

its effect on the non-pecuniary benefits received by the manager. In general, bank borrowing is

associated with strict monitoring, thereby mitigating agency problems. This monitoring behavior,

however, can reduce the manager’s non-pecuniary benefits created by discretion over resource

allocation. This reduction of non-pecuniary benefits by bank borrowing is likely to be stronger

in relationship banking than in an arm’s length relationship. In an arm’s length relationship,

monitoring by a bank is normally limited to ex ante and ex post monitoring, while interim

monitoring, which is often associated with dispatching bank members to client firms, plays an

important role in the relationship banking. Thus, when the costs of relationship banking exceed the

benefits perceived by the manager, the firm’s financing and investment policies can be distorted.

In the context of corporate policy, the net effect of lending relationships is ambiguous. This

study adds to the literature by building on a panel database of loans to analyze the dynamic

effects of relationship lending on the borrowers’ investment and financing policies.2Unlike

cross-sectional data (sampled at the loan level), a firm-year panel dataset can be used to analyze

the economic effects of lending relationships as they evolve over time. This formation allows

me to control for any unobservable borrower and time-series characteristics that are outside the

purview of an empirical framework that maypotentially lead to biased estimates. The time-series

dimension of the panel data also solves the problem of multiple loan observations in a given

year which, in cross-sectional studies, are treated as independent observations. However, these

observations are obviously not independent, as the accounting information is only updated on an

annual basis.

The remainder of the paper is organized as follows. Section I describes the data. In Section II,

I specify the empirical test design. I discuss the results in Sections III and IV, and provide my

conclusions in Section V.

I. Data and Empirical Specification

A. Sources of Data

I obtain details of loan transactions and the nature of the relationship between firms and their

banks from the Dealscan database distributed by the Loan Pricing Corporation.3The sources

of firm characteristics are Standard and Poor’s Compustat database and Center for Research in

Security Prices (CRSP) tapes. CRSP and Compustat data are merged using the historical header

file from CRSP, and the link file from Chava and Roberts (2008) is used to merge Dealscan

to Compustat/CRSP data. I narrow the sample by removing firms with total assets less than

2Ongena and Smith (2001) and Houston and James (1996) use a panel of firms to address issues associated with

relationship banks’ information monopolies, but do not tackle the capital structure issue. Berger and Udell (1995) and

Petersen and Rajan (1994) use data from the National Survey of Small Business Finance, so neither of these papers

have a panel of firms (inferences are made comparing firms at different points of their lending relationship). Degryse

and Cayseele (2000) use of a panel of firm-loan data on Belgium firms, but do not exploit the panel structure in their

empirical work. The data sample in Bharath et al. (2011) consists of a cross-sectional set of loans, where the deal-based

lending relationships are calculated based on the date of the loans.

3Dealscan has information on 50% to 75% of all US commercial loan volume into the early 1990s, with coverage

increasing to 80% to 90% later in the 2000s (Carey and Nini, 2007). It reports detailed information about the structure

of loan contracts including the identity of the borrowers and lenders, origination and maturity dates, the purpose of the

loan, the pricing, and the size of the deal.