Motivations for Loan Herding by Chinese Banks and Its Impact on Bank Performance

• Published date: 01 July 2019
• DOI: http://doi.org/10.1111/cwe.12285
• Author: Chung‐Hua Shen,Yen‐Hsien Lee,Hao Fang,Chien‐Ping Chung
• Date: 01 July 2019

©2019 Institute of World Economics and Politics, Chinese Academy of Social Sciences

China & World Economy / 29–52, Vol. 27, No. 4, 2019

29

Motivations for Loan Herding by Chinese Banks and

Its Impact on Bank Performance

Hao Fang, Yen-Hsien Lee, Chung-Hua Shen, Chien-Ping Chung*

Abstract

This study uses a dynamic herding model that considers intertemporal and cross-

sectional correlation to conrm that loan herding occurs among joint-stock commercial

banks (JSCBs) and city commercial banks (CCBs). We clarify the motivations for bank

loan herding. We find that loan herding by both JSCBs and CCBs results more from

following the behavior of other same-type banks than different-type banks because of

characteristic herding or reputational concerns. Loan herding by JSCBs is motivated

by investigative herding, whereas loan herding by CCBs results from informational

cascades. Moreover, loan herding has a signicantly harmful impact on the operating

performance of CCBs but not JSCBs, which may be explained by the irrational behavior

of CCBs. Our results will help Chinese bank supervisors develop appropriate policies

for handling loan herding.

Key words: nancial crisis, irrational herding, loan herding

JEL codes: G2, G4

I. Introduction

Studies of herding behavior have attracted considerable attention. In psychology,

because of informational asymmetry and the agency problem, a group of investors may

follow one another into or out of the same securities over a certain period; this is known

as herding behavior (Wermers, 1999). Previous studies of herding behavior focused

on the irrational trading behavior of institutional investors in capital markets. The

Lakonishok, Shleifer and Vishny (LSV) (1992) herding measure is a static indicator that

*Hao Fang, Associate Professor, School of Economics, Qu Fu Normal University, China. Email: hihaoa@gmail.

com; Yen-Hsien Lee, Associate Professor, Department of Finance, Chung Yuan Christian University, Taiwan,

China. Email: yh@cycu.edu.tw; Chung-Hua Shen (corresponding author), Professor, Institute of Banking and

Monetary Studies, Nanjing Audit University, China. Email: chshen01@ntu.edu.tw; Chien-Ping Chung, Associate

Professor, College of Management, National Taiwan University of Science and Technology, Taiwan, China.

Email: thomas6311@g2.usc.edu.tw. This study was supported by the Planning Research Project of Social

Science in Shandong Province of China (No. 19CJJJ09).

Hao Fang et al. / 29–52, Vol. 27, No. 4, 2019

©2019 Institute of World Economics and Politics, Chinese Academy of Social Sciences

30

examines the cross-sectional dependence of trades within a specic period. Extending

the LSV measure to a dynamic model, Sias (2004) measured cross-sectional temporal

dependence directly. Namely, he measured the extent to which institutional investors

follow each other’s trades over adjacent periods. Recently, the concept of herd behavior

has been extended from institutional investors in the capital market to banks in the loan

market.

The term “loan herding” describes the phenomenon of banks following other banks’

loan decisions in lending to industrial rms. Uchida and Nakagawa (2007) demonstrated

that Japanese banks exhibited herd behavior in the loan market and proposed that

cyclical herding behavior during the late 1980s contributed to the decline of the Japanese

economy. Nakagawa et al. (2012) showed that herding by Japanese banks in business

lending had a negative impact on both GDP and land prices for the following two years.

de Juan (2003) found similar evidence regarding the branching decisions of Spanish

banks. Buch and Lipponer (2006) reported that foreign direct investment made by

German banks in Organisation for Economic Cooperation and Development (OECD)

countries represented bank herding. Past research has also provided evidence of herding

by banks in China. Using a vector autoregressive model, Chen et al. (2012) found that

state-owned banks (SOBs) and joint-stock banks incidentally exhibit the same credit

behavior and that city commercial banks (CCBs) exhibit irrational herd behavior based on

reputation. Using data on bank lending to colleges and universities, Zou and Deng (2008)

applied Bayes’ theorem on incomplete information to show that a bank loan decision-

maker tends to follow other decision-maker’s lending decisions when information on the

borrower is lacking. Deng and Li (2012) constructed a herd model with multi-subject

communication and found that herd effects arise from information communication and

imitation behavior among agents, which can generate bank runs. However, these studies

did not use the dynamic herding model that considers intertemporal and cross-sectional

correlation or examine the following behavior among banks in lending. Therefore, it is

worth combining these approaches to conduct further analysis.

In this paper, we extend the dynamic herding model in capital markets to explore the

existence of intertemporal loan herding among banks. Although Uchida and Nakagawa

(2007) examined the cross-sectional dependence of lending positions among banks

within a single period, the herding model of a single period cannot reveal the following

behavior of banks in lending; only an intertemporal herding model can clarify whether

banks follow the loan behavior of other banks. Sias (2004) proposed an intertemporal

herding model and decomposed the cross-sectional correlation of institutional trading

into institutions following their own trades (own cascades), institutions following

the trades of other institutions of the same type (same-type herding) and institutions