MONITORING IN SMALL BUSINESS LENDING: HOW TO OBSERVE THE UNOBSERVABLE

• DOI: http://doi.org/10.1111/jfir.12082
• Published date: 01 December 2015
• Author: Gabriele Sampagnaro,Antonio Meles,Vincenzo Verdoliva
• Date: 01 December 2015

MONITORING IN SMALL BUSINESS LENDING: HOW TO OBSERVE

THE UNOBSERVABLE

Gabriele Sampagnaro

University of Naples Parthenope

Antonio Meles

Second University of Naples

Vincenzo Verdoliva

Kingston Business School

Abstract

In this article we investigate the determining factors of bank monitoring in small

business lending. Unlike previous studies that rely on proxies of a bank’s monitoring

effort, we use a large and unique data set that includes the number of monitoring contacts

per year between a European bank and a large number of small firms from 2009 to 2012.

Our main results highlight that the frequency of a bank’s monitoring effort is negatively

related to the firm’s reputation and the strength of the bank–firm relationship, and

positively related to the bank–borrower proximity and the borrower’s credit risk level.

JEL Classification: D82, G21, G32

I. Introduction

Monitoring has valuable implications for both banks (interested in reducing monitoring

costs for competitive strategies) and firms (which seek to lower their borrowing costs),

but it has received little attention in the empirical literature because of the difficulty in

obtaining data and establishing an effective measure of a bank monitoring efforts (see,

among others, Coleman, Esho, and Sharpe 2006; Akhigbe and McNulty 2011). In fact,

monitoring is a combination of collection and inspection of credit information that may

pertain to different organizational levels and to different bank internal processes. Banks

are reluctant to provide such data because of confidentiality problems and costs of

extraction.

To address these issues, several studies have proposed proxy measures of a

bank’s monitoring efforts, such as credit ratings (e.g., Billett, Flannery, and Garfinkel

1995), loan loss provisions (e.g., Johnson 1997), bank size (e.g., Cook, Schellhorn, and

Spellman 2003), and labor input in the monitoring process (see, among others, Coleman,

We thank Drew Winters, the other editors of the Journal of Financial Research, and an anonymous referee for

constructive comments on a previous version of this paper.

The Journal of Financial Research Vol. XXXVIII, No. 4 Pages 495–510 Winter 2015

495

© 2015 The Southern Finance Association and the Southwestern Finance Association

Esho, and Sharpe 2006; Lee and Sharpe 2009). Most studies treat all borrower firms as a

homogenous group and do not recognize that a bank’s efforts spent on monitoring may

also depend on borrower reputation, potential information asymmetries, or the

borrower–lender relationship. One exception is Blackwell and Winters (1997), who

use a supposed measure of monitoring frequency (extracted from policy and procedure

handbooks from a sample of six banks) to find that banks monitor smaller firms more

frequently than larger firms, highly leveraged firms more frequently than less leveraged

firms, and firms with which they have a shorter banking relationship more frequently than

those with a longer banking relationship.

We build on this research by investigat ing the correlation between actua l

measures of bank monitoring effor ts and a set of variables that include both tra ditional

and nontraditional monit oring determinants. Speci fically, we use a large and uniqu e

data set that includes the numbe r of monitoring contacts pe r year between a large

Italian bank and approxim ately 30,000 clients (smal l firms) over a four-year period

(2009–2012).

Our contribution to the literature is based on four major findings. First, we find

strong empirical evidence that banks allocate more review and monitoring time to higher

risk borrower firms than lower risk firms. Second, we find that the larger the geographical

distance between a firm and a bank, the lower the level of bank monitoring because the

increase in distance causes an increase in monitoring costs. However, the effect of

distance also depends on credit quality: as distance increases, the bank may raise its

quality standards and accept only better quality loans that need less monitoring. Third,

we find that relationship banking works as substitute for monitoring effort. In fact, we

observe that borrower firms that have a close relationship with the bank and a higher

reputation (measured by both the length of credit relationship and the number of products

sold by a bank over time) are, ceteris paribus, less monitored than other firms. Finally, we

find strong evidence for the free-rider hypothesis suggested by Diamond (1984), which

predicts that when monitoring is costly, multiple lending relationships may lead to a

duplication of cost efforts and imply a free-rider problem among lenders, which reduces

the amount of monitoring.

Together, these findings reveal whether monitoring activities are correlated to

both traditional proxies for monitoring costs as predicted by theory (e.g., number of

lenders, credit risk classifications) and other indirect measures of monitoring not

considered in previous research (i.e., distance and banking products). By doing this, our

article may help scholars conducting applied empirical research select the most

appropriate identification variable for monitoring when direct observations are

unavailable.

II. Monitoring Metrics: A Brief Review

Bank monitoring represents one of the most important issues in the finance literature. For at

least three decades researchers have examined: (1) why banks dedicate considerable

resources to monitoring activities (e.g., Fama 1985; Chan, Greenbaum, and Thakor 1986;

Diamond 1984, 1991; Qi 1998; Boot and Thakor 2000), (2) what impact bank monitoring

496 The Journal of Financial Research