What is the impact of problem loans on Japanese bank productivity growth?

• Date: 01 May 2019
• Published date: 01 May 2019
• Author: Emmanuel Mamatzakis,Anh Nguyet Vu,Roman Matousek
• DOI: http://doi.org/10.1111/fmii.12112

DOI: 10.1111/fmii.12112

ORIGINAL ARTICLE

What is the impact of problem loans on Japanese

bank productivity growth?

Emmanuel Mamatzakis1Roman Matousek2Anh Nguyet Vu1

1University of Sussex,School of Business,

Management and Economics

2Nottingham University Business School

Correspondence

EmmanuelMamatzakis, University of Sussex,

Schoolof Business, Management and Economics,

Falmer,Brighton, BN1 9SL, UK.

Email:e.mamatzakis@sussex.ac.uk

Abstract

This paper examines for the first time the impact of problem loans

on Japanese productivity growth. We exploit a new data set of

Japanese problem loans classified into two categories: bankrupt

and restructured loans. We opt for a novel and flexible productiv-

ity growth decomposition that allows to measure the direct impact

of these problem loans on productivity growth. The results reveal

that Japanese bank productivity growth was severely constrained

by bankrupt and restructured loans early in 2000s, whilst some per-

sistence of the negative impact of problem loans on productivity

growth is observed in the late 2000s. Thereafter,there is only some

partial recovery in the productivity growth from 2012 to 2015. Fur-

ther, we also perform cluster analysis to examine convergence or

divergenceacross regions and over time. We observe limited conver-

gence, though Regional Banks seem to form clusters in some regions.

KEYWORDS

bank productivity, bankrupt loans, cluster analysis, Japan, restruc-

tured loans

JEL CLASSIFICATION

D24, G21, C33

1INTRODUCTION

Therehas been extensive theoretical and empirical research into the field of firm efficiency and productivity (Heshmati,

Kumbhakar, & Sun, 2014; Subal C. Kumbhakar & Tsionas, 2016; Sun, Kumbhakar, & Tveterås, 2015). In terms of bank

efficiency and productivity,the outbreak of the global financial crisis has driven a surge of banking studies (Matousek,

Rughoo, Sarantis, & Assaf,2015; Tsionas, Assaf, & Matousek, 2015), unfolding the paramount importance of financial

intermediarieswithin the economic system. In terms of productivity growth, evidence is rather limited with studies that

apply parametric methods to evaluate bank productivity (Boucinha, Ribeiro, & Weyman-Jones, 2013;Feng & Serletis,

2010; Feng & Zhang, 2012, 2014). As indicated in a reviewof non-parametric productivity applied in banking by Fethi

c

2019 New YorkUniversity Salomon Center and Wiley Periodicals, Inc.

Financial Markets,Inst. & Inst. 2019;28:213–240. wileyonlinelibrary.com/journal/fmii 213

214 MAMATZAKISET AL.

and Pasiouras (2010), the majority of studies adopt a non-parametric approach(Fukuyama & Weber, 2005, 2010; Liu

& Tone,2008).

Our study extends the literatureon bank productivity by opting for a parametric estimation technique. We decom-

pose bank productivity growth into different components, namely the effects of problem loans, which are essentially

uncontrollable inputs, quasi-fixed input, returns to scale, and technological change. The Japanese banking system is of

interest as its performance has been undermined by an unprecedented volume of bankrupt and restructured loans.

These loans are referred as risk-monitored loans disclosed in accordance with the Japanese Banking Law.Moreover, in

Japanese banking literature,bank efficiency studies have dominated the research field of bank performance, for exam-

ple Barros, Managi, and Matousek (2012), Drakeand Hall (2003), Fukuyama and Weber (2005), Fukuyama and Weber

(2010), Yang and Morita (2013). Japanese bank productivity has been rather neglected (Assaf, Barros, & Matousek,

2011; Fukuyama, 1995; Fukuyama,Guerra, & Weber, 1999).

We, thus, fill a gap in the literature and apply for the first time a productivity growth decomposition to Japanese

banks, where problem loans’ impact would be revealed.1Previous literature has considered nonperforming loans as

uncontrollable inputs or undesirable outputs in the banking production process (Assaf, Matousek, & Tsionas, 2013;

Barros et al., 2012; Drake & Hall, 2003; Fukuyama& Weber, 2008; Glass, McKillop, Quinn, & Wilson, 2014; Hughes &

Mester,2010; Mamatzakis, Matousek, & Vu, 2016). We follow Drake and Hall (2003) and Hughes and Mester (2010) to

treat bankrupt and restructured loans as uncontrollable inputs in our productivity decomposition.2Given the exten-

sive volume of bankrupt and restructured loans in Japan, we expect that they have an impact on bank productivity.

Arguably, banks may receive payments of the principal and interest on these loans subject to borrowers’ financial

health. These overdue loans in turn would raisebank's operating costs in the short-run. Hence, one would expect these

loans to deteriorate bank productivity.

Alongside bankrupt and restructured loans, we also employ equity as a quasi-fixedinput (Berger & DeYoung, 1997;

Hughes, Mester, & Moon, 2001; Ray& Das, 2010). Within a short period, it would be unfeasible to adjust the level of

equity considerablyand quickly (Lozano-Vivas & Pasiouras, 2014). In the event of unexpectedlosses, the level of equity

is of utmost importance to ensure bank safety and soundness, preventing banks from temporary illiquidity and insol-

vency (Diamond & Rajan, 2000). Equity would also serve as a cost-reducing factor due to less interest paid for debt

financing (Hughes & Mester,2013). Finally, the inclusion of equity in our productivity decomposition is of importance

for Japan. The reason is that during the banking crisis in the late 1990s, there was a prolonged period of undercapitali-

sation until the early 2000s. The Japanese authorities responded by injecting public capital four times between March

1998 and June 2003 (Hoshi & Kashyap, 2010), hoping to stabilise the financial marketand revive the banking industry.

The contribution of this study can be summarised in the following ways. First, we expand the parametric method-

ological literature of bank productivity growth (Boucinha et al., 2013; Casu, Ferrari, & Zhao, 2013; Lozano-Vivas&

Pasiouras, 2014) as opposed to the broadly applied nonparametric one (Alam, 2001; Berg, Forsund,& Jansen, 1992;

Delis, Molyneux, & Pasiouras, 2011; Fiordelisi & Molyneux, 2010; Grifell-Tatjé& Lovell, 1997; Kao & Liu, 2014; Whee-

lock & Wilson, 1999). Our paper refers to Japan, which servesas an excellent case study given the trouble of its banking

industry (Barros, Managi, & Matousek, 2009; Fukuyama, 1995; Fukuyama & Weber,2002). Second, we exploit a new

data set of bankrupt and restructured loans, which are disaggregated from “risk-monitored loans” disclosed subject

to the Japanese Banking Law. The adopted approach enables a comprehensive analysis by allowing for the impact of

these loans on total factor productivity growth. Finally,we test for convergence – catching up effect – among Japanese

banks and geographic regions by using club convergence analysis proposed by Phillips and Sul (2007). This method-

ology is superior compared to others such as the 𝛽-convergence and 𝜎-convergence. For a paneldata variable, it can

detect convergence in sub-groups even though convergence is rejected at the whole panel level. After more than a

decade of the banking crisis and structural reforms, converging on productivity growth would indicate that Japanese

banks havebeen brought together to facilitate further necessary policy changes, especially on the ground of continued

deflation and low economic growth.

Our results show that productivity growth in Japanese commercial banks is impaired by the impact of bankrupt

loans. The destructive effect of these loans varies over time, appearing to capture events such as government inter-

ventions, the global financial crisis, and the Tohokutsunami/earthquake. Interestingly, restructured loans are among