Nonperforming loans in the euro area: Are core–periphery banking markets fragmented?

• Author: Dimitrios Anastasiou,Mike Tsionas,Helen Louri
• DOI: http://doi.org/10.1002/ijfe.1651
• Published date: 01 January 2019
• Date: 01 January 2019

RESEARCH ARTICLE

Nonperforming loans in the euro area: Are core–periphery

banking markets fragmented?

Dimitrios Anastasiou

1

| Helen Louri

2,4

| Mike Tsionas

2,3

1

Department of Accounting and Finance,

Athens University of Economics and

Business, Athens, Greece

2

Department of Economics, Athens

University of Economics and Business

3

Department of Economics, Lancaster

University Management School, UK

4

London School of Economics, European

Institute, Hellenic Observatory, UK

Correspondence

Helen Louri, Professor, Department of

Economics, Athens University of

Economics and Business and Research

Associate LSE EI/HO, 76 Patission Street,

Athens 10434, Greece.

Email: elouri@aueb.gr

Funding information

Research Center of AUEB

JEL Classification: C23; C51; G21; G2

Abstract

The objectives of this study are, first, to examine the causes of nonperforming

loans (NPLs) in the euro area for the period 2003Q1 to 2016Q1 and, second, to

investigate if there is fragmentation between core and periphery banking mar-

kets. By employing both fully modified ordinary least squares (FMOLS) and

Bayesian panel‐cointegration vector autoregression techniques, we estimate

the long‐run effects of both bank‐specific and macroeconomic factors on NPLs.

We find that NPLs in the euro area have performed an upward (much higher

in the periphery) shift after 2008 and are mostly related to worsening

macroeconomic conditions. A chi‐square test comparing the estimated

coefficients for the core and periphery NPLs rejects the hypothesis of equality

revealing another aspect of financial fragmentation in the euro area that leaves

the periphery more vulnerable. Such findings can be helpful when designing

macroprudential as well as NPL resolution policies.

KEYWORDS

bank‐specific determinants, financial fragmentation, FMOLS estimation, macroeconomic

determinants, nonperforming loans, panel cointegrated VAR

EBA updated Risk Dashboard confirms that

elevated NPLs and low profitability are the

main challenges for the EU banking

sector.

The European Banking Authority, periodical update

of Risk Dashboard, April 3, 2017

1|INTRODUCTION

The recent financial crisis, which started in the United

States in 2007, was the catalyst that revealed the weak-

nesses in the international banking system. One such

important weakness was the high credit risk undertaken

that resulted in increases in nonperforming loans (NPLs).

According to the European Central Bank (ECB), a loan is

considered to be nonperforming (or impaired) when

payments of interest are past due by 90 days or more.

Losses due to NPLs reduce banks'profits, and when these

are not enough (as is often the case in recent years),

losses reduce banks'capital, creating a pressing need for

recapitalization and constraining credit provision to the

economy. A high rate of NPLs may also cause expecta-

tions about the stability of the banking system to deterio-

rate, creating systemic risk that may in turn lead to a run

on deposits, significantly reducing the intermediation

power of banks.

Therefore, it is of urgent importance to understand

and identify the factors that affect NPLs and provide

some guidance both to the banks so that they improve

their credit policies and to the governments and banking

supervisors so that appropriate preventive measures and

stress‐testing models are adopted. It should be noted that

NPLs, along with the liquidity risk, are the most signifi-

cant types of risk for commercial banks, especially after

Received: 30 May 2017 Revised: 29 January 2018 Accepted: 26 June 2018

DOI: 10.1002/ijfe.1651

Int J Fin Econ. 2019;24:97–112. © 2018 John Wiley & Sons, Ltd.wileyonlinelibrary.com/journal/ijfe 97

the 2007 financial crisis. A typical example of an NPL‐

caused banking crisis is the Asian crisis of 1997, the cost

of which has been high not only in fiscal terms but also

concerning losses to the real economy.

The problem of the NPLs has expanded in the euro

area, but countries located in the euro area periphery

and especially those involved in International Monetary

Fund‐European Union support programmes such as Por-

tugal, Ireland, and Greece seem to face the biggest threat.

Thus, in 2016, whereas NPLs in the euro area reached

12% and were less than 2% in Germany, they exceeded

35% in Greece and reached 20% in Italy and Ireland.

NPLs present the most significant challenge that euro

area periphery banks face. Managing this problem will

be a defining factor, which will permit (often called zom-

bie) banks to start contributing to the growth of their

economies again (ECB, 2015).

But are NPLs really different between core and

periphery countries? The euro area (one of the largest

currency areas globally) is a unique example of a mone-

tary union of 19 members with some fiscal autonomy,

where convergence has been obstructed by the crisis

and financial fragmentation is often accused of causing

further divergence. Do NPLs react similarly to the same

explanatory variables in all countries? Or do they have

different responses in the periphery thus increasing the

vulnerability of its banks? Are there signs of financial

fragmentation (especially visible in money and capital

markets after the sovereign debt crisis) meaning that

NPLs are to a significant extent determined by the coun-

try of origin rather than its underlying fundamentals? Or

is it the case that NPLs are higher in the periphery just

because macroeconomic conditions have diverged and

growth has stagnated?

The contribution of this paper, which looks for signs

of fragmentation in the determinants of euro area NPLs

in the period 2003 Q1 to 2016 Q1 is threefold: First, both

fully modified OLS (FMOLS) and Bayesian panel‐

cointegrated vector autoregression (VAR) are used as

econometric methodologies, improving our estimates

with respect to endogeneity problems. Second, a new

interest rate margin (MIR), which is an indirect measure

of the cost of borrowing adjusted for risk, is examined as

a potential determinant and found to exert a significantly

positive effect. Finally, the application of a chi‐square test

for measuring fragmentation of the estimated coefficients

is also worth noting, because its results support the exis-

tence of fragmentation between core and periphery banks

regarding their impaired loans.

The rest of the paper is organized as follows. In

Section 2, we offer an extended literature review of the

factors that affect NPLs. Section 3 describes the data,

the variables, and the models employed. In Sections 4

and 5, we present the econometric methodologies and

the empirical results, respectively. Section 6 concludes.

2|LITERATURE REVIEW

2.1 |Initial hypotheses behind NPL

creation

One of the earliest studies that tried to understand the

reasons behind NPLs is Berger and DeYoung (1997). They

applied Granger‐causality techniques in order to test four

hypotheses about the relationship between loan quality,

cost efficiency, and bank capital, taking a sample of U.S.

commercial banks for the period 1985–1994. These four

hypotheses were referred to as “bad luck,”“bad manage-

ment,”“skimping,”and “moral hazard.”Based on their

results, they came to the conclusion that the bad manage-

ment hypothesis was superior to the others for the whole

sample. They also found that another reason for increas-

ing NPLs is the fact that bank capital ratios were gener-

ally low, hinting at moral hazard incentives driving

inadequately capitalized banks towards taking a high

portfolio risk.

Along the same line, Podpiera and Weill (2008) exam-

ined the existence of a causal relationship between NPLs

and cost efficiency. They expanded on the Granger cau-

sality framework used by Berger and DeYoung (1997)

by applying the generalized method of moments (GMM)

and using dynamic panel estimators. They found support

only for the bad management hypothesis.

2.2 |NPLs and macroeconomic conditions

A series of studies on NPLs focused exclusively on the

role of macroeconomic or country‐specific determinants

and found that they exerted a most significant effect. In

particular, Espinoza and Prasad (2010) found that the

NPL ratio rises when economic growth slows and risk

aversion decreases as well as when interest rates increase.

In particular, NPLs and real GDP growth were found to

have a notable inverse relationship. Nkusu (2011) follow-

ing a similar methodology to Espinoza and Prasad (2010)

found that an aggravation in the macroeconomic envi-

ronment, as proxied by sluggish growth, decreasing asset

prices, and higher unemployment, is interrelated with

debt service problems. On the contrary, an improving

macroeconomic environment implies a decrease in NPLs.

The importance of the current account balance was

stressed by Kauko (2012) who examined its interrelation

with the development of NPLs, especially in the recent

financial crisis. His main result was that the rapid credit

expansion in the period 2000–2005 could be considered

as an important risk factor only if combined with the

98 ANASTASIOU ET AL.