You are on page 1of 28

Accounting Research Journal

Determinants of non-performing loans in banking sector in small developing island states: a study of Fiji
Ronald Ravinesh Kumar, Peter Josef Stauvermann, Arvind Patel, Selvin Sanil Prasad,
Article information:
To cite this document:
Ronald Ravinesh Kumar, Peter Josef Stauvermann, Arvind Patel, Selvin Sanil Prasad, "Determinants of non-performing
loans in banking sector in small developing island states: a study of Fiji", Accounting Research Journal, https://
doi.org/10.1108/ARJ-06-2015-0077
Permanent link to this document:
https://doi.org/10.1108/ARJ-06-2015-0077
Downloaded on: 20 May 2018, At: 05:53 (PT)
References: this document contains references to 0 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 2 times since 2018*
Access to this document was granted through an Emerald subscription provided by emerald-srm:178665 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

visit www.emeraldinsight.com/authors for more information.


About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication
Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.

*Related content and download information correct at time of download.


Determinants of non-performing loans in small developing economies: a case
of Fiji’s banking sector

Ronald Ravinesh Kumar


QUT Business School, School of Economics & Finance
Queensland University of Technology
Brisbane
Australia.
ronaldravinesh.kumar@hdr.qut.edu.au, ronaldkmr15@gmail.com;
+61 410114928; +61 07 3138 6608

School of Accounting & Finance


Faculty of Business & Economics
Laucala Campus, The University of the South Pacific
Suva
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

Fiji
kumar_RN@usp.ac.fj;

Business School, University of Bolton


Bolton
United Kingdom
E-mail: rrk1mpo@bolton.ac.uk.

Peter Josef Stauvermann


School of Global Business & Economics
Changwon National University
Changwon
Republic of Korea

Arvind Patel
School of Accounting & Finance, Faculty of Business & Economics
University of the South Pacific
Suva
Fiji

Selvin Prasad
School of Accounting & Finance, Faculty of Business & Economics
University of the South Pacific
Suva
Fiji.

Abstract

The banking sector stability depends in large part on the size of non-performing loans (NPLs).
Hence, the factors which explain the problem loans are very useful information for banks.

1
Notably, studies in this regard with respect to the small developing countries’ banking sector
have received less attention. Therefore, in this study, we examine the determinants of NPLs with
a case of Fiji’s banking sector, over the period 2000-2013. Our balanced sample consists of the
entire banking sector (5 commercial banks and 2 non-bank financial institutions). First, we
estimate a base model which comprise of bank-specific indicators that are related to bank
management and then we extend the estimations to include macroeconomic/structural factors
such as economic growth, inflation, real effective exchange rate, unemployment, remittances,
political instability and external events like the global financial crisis. The estimations are done
using pooled OLS, the random effects and the fixed effects regression methods, respectively. The
results show the following indicators have negative association with NPL and are statistically
significant with the conventional levels: return on equity, capital adequacy requirement, market
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

share based on assets, unemployment and time. On the other hand, the net interest margin has a
positive and statistically significant association with NPL. Subsequently, the stability of the
banking sector in small developing countries such as Fiji is largely dependent on banks’
profitability, solvency, size in terms of market share, and the presence of a learning curve, and
keeping a close tab on the interest rate spread between loans and deposits.

Keywords: non-performing loans; banking sector; pooled; random; fixed; regression; small
developing countries; Fiji.

2
1. Introduction
An effective financial and banking system is identified by its ability to match savings and
investments efficiently to support growth and development. However, the effectiveness of the
banking system can be hampered if banks expose themselves to risky lending and financial
investments which if not managed properly can result in an accumulation of non-performing
loans that can give rise to a de-stabilized financial system (Turner, 2010) and severe banking
crises (Reinhart and Rogoff, 2011).

Of course, it does not only take a global financial crisis (GFC) to jeopardize the banking
system in a country. Especially, the extent to which domestic banking system is affected by the
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

international financial market depends mainly on the current exchange rate system applied in the
country. For our purpose, we examine the factors that determine the non-performing loans in a
small economy, with Fiji as a case study. The reason for focusing on small developing countries
is because they have received lesser attention in the literature and more importantly, the banking
sector in these countries operates in an environment where economic activities and economies of
scope and scale are constrained.

The income of banks can be split into two parts, the interest income and the non-interest
income, the latter is mainly generated through off-balance sheet activities and fees. However,
because of the small market size and less opportunities to diversify risk in small developing
countries, it is not surprising to note a relatively huge interest rate spread. Further, since the
systemic risk of banks are mainly dependent on the economic performance of the domestic
economy, banks’ can be reluctant to provide large amounts of loans unless there is sufficient
collateral from the borrower that can be claimed by banks in case of default. Given these market
conditions, it is extremely important that banks in small developing countries are cognizant of
the plausible drivers of non-performing loans (NPLs) in order to effectively manage the loan
portfolio and to reduce the odds of default. A number of indicators can be linked to NPLs. On the
one hand, there are bank-specific indicators such as return on assets, solvency, interest rates on
deposit or loans, among others, which are directly under the control of the bank management,
and on the other hand, there are macro-economic and structural (systemic) indicators such as the

3
growth rate, inflation rate, unemployment rate, political climate, instances of global financial
depression, banks level of maturity, and economic policy, all of which can influence the NPLs.
We choose Fiji as a case country within the small developing economies because Fiji has
a relatively more developed financial sector in comparison to other small (developing)
economies in the Pacific and thus banks provide regular reports to the Reserve Bank of Fiji
(RBF) which makes data compilation less of a concern. The RBF has two objectives in principal
– to maintain a low inflation rate and an adequate level of foreign reserves. To realize these
objectives, the Fiji dollar is pegged to a basket of five currencies (US dollar, Australian dollar,
New Zealand dollar, Japanese yen and the Euro). As deemed necessary, discretionary
revaluations and devaluations (like in 2009) of the Fiji dollar are undertaken. Therefore, given
the fixed exchange rate system, capital controls become necessary and are subject to permanent
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

adjustments depending of the amount of available currency reserves. However, these restrictions
discourage foreign investments. Further, this characteristic of the exchange rate system
establishes the fact that the domestic banking system is only peripherally connected to the major
financial markets and consequently experience only indirect, second or third order effects, of the
GFC. Given this institutional and legal framework, which is typical for small developing
countries, the respective banking system differs from the ones we observe in developed countries
where the latter are fully integrated in the global financial market. Moreover, because of low
economic scale and progress in terms of economic development, the number of competing banks
is low and thus the sector is highly concentrated with the business operation primarily focused on
deposits, loans and mortgages (Sharma et al., 2014). Not surprisingly, the off-balance sheet
activities are mainly restricted to payment services like offering credit cards or online banking.

Another important fact is that the banking sector in Fiji has undergone a number of bank
transitions (takeover and acquisition) and in the past, experienced a major collapse of a
government controlled bank – the National Bank of Fiji (NBF). The failure of NBF was, in large
part, a result of poor management of funds and sustained accumulation of non-performing loans
(Grynberg, Munro and White, 2002; Lodhia and Burritt, 2004; Overton, 2003; Goundar, 2004
and 2005). In 1992, the collapse of the NBF resulted in an estimated loss of tax payer funds of
Fijian $280 million, representing 25% of Fiji’s national budget and triggered a large increase in
Fiji’s public debt (Grynberg, Munro and White 2002). Although the instability resulting in the

4
collapse of NBF and the subsequent bailout by the Fiji government are events of the past, the
loss incurred at a national level remains a terrifying example of a bank failure in small
developing countries and if nothing else, the collapse reverberate the importance of managing
and avoiding problem loans.

Against this background, the present study aims to contribute to the banking literature
and more specifically on the banking sector of small developing countries by examining the
bank-specific (idiosyncratic), macroeconomic (systematic) and structural determinants of NPL,
and using Fiji as a case study. The rest of the paper proceeds as follows. In Section 2, we provide
a brief literature survey on related studies, and a discussion on the determinants of NPL to gain
insights on the plausible and relevant factors to formulate hypothesis for testing. In Section 3, we
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

discuss the data, model and approach. In section 4, we present the results, and finally, in Section
5, we conclude.

2. Literature Survey
2.1 Some related studies
The macro and micro role of financial institutions are widely discussed (Beck, Levine
and Loayza, 2000; Levine 2005; Demirgüç-Kunt and Levine 2009), especially with a focus on
developed and emerging markets (Lee and Hsieh, 2013; Mirzaei, Moore and Liu, 2013) with a
common feature that these markets have highly developed financial institutions which are
inextricably linked. Nevertheless, banking sector studies with regards to developing (Mewga,
2011; Poshakwale and Qian 2011) and small economies are growing (Sharma and Nguyen, 2010;
Gounder and Sharma, 2012; Sharma and Gounder, 2013; Sharma et al. 2013, 2014 and 2015;
Kumar and Patel, 2014; Sharma and Gounder 2015), albeit slowly. For instance, Sharma and
Nguyen (2010) examine the banking development in Fiji over the period 1970-2006 and
compare it with 53 other countries. The authors conclude that in case of Fiji, although the legal
rules were weak, a combination of average law enforcement quality and high accounting
standards was critical in ensuring a reasonable performance in terms of banking development
over the periods. In another study, Sharma et al. (2013) show that larger banks relative to smaller
ones tend to be less efficient in providing loans or deposits. Further, Sharma et al. (2015) use the
nonparametric data envelopment analysis (DEA) to measure efficiency, where deposits are

5
treated as an input and output, and conclude from this that the overall efficiency levels of banks
in Fiji are lower than the corresponding ones in Australia. With the dynamic GMM and balanced
panel data they show that credit risks are positively correlated with efficiency and personnel
expenses are negatively correlated. Additionally, they revised the result from Sharma et al (2013)
that larger banks are less efficient than smaller ones. Kumar and Patel (2014) examine the degree
of competitiveness in the banking sector in Fiji using the Panzar and Ross (1987) method and
conclude among other things that: Fiji’s banking sector is concentrated and exhibit monopolistic
behavior; banks’ revenue is positively associated with interest expenses, capital adequacy
requirement and the number of branches, and negatively associated with personnel expenses,
other operating expenses and firm size measured by total assets. Further, Sharma et al. (2014)
find that previous period (lag-one) profitability and size are positively associated with
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

profitability and capital adequacy requirement whereas liquid assets to total assets have a
negative relationship. In a recent study, Sharma and Gounder (2015) examine the determinants
of bank profitability in Fiji over the periods 2000-2010 using dynamic generalized method of
moments (GMM) technique and find non-interest income as a single factor that influences
positively the return on asset and return on equity of banks. While these studies provide relevant
and important insights into small developing countries’ banking sector, we note that except for
Grynberg et al. (2002) which examine the NBF’s stability and the subsequent collapse, there are
no studies specifically focusing on bad or non-performing loans on small island economies.

2.2 Bank-specific factors


The non-performing loans (NPLs) are sometimes referred to as bad loans or problem loans.
Bank-specific factors that can influence NPLs include return on equity (ROE), solvency (SOL),
net interest margin (NIM), ratio of operating expenses to operating income as a measure of
inefficiency (INEF), and assets or market share (as a measure of SIZE).

In a study, Salas and Saurina (2002) state that banks with a lower amount of capital are more
inherent to risk which causes higher levels of non-performing loans. Moreover, the NPL and
ROE are negatively correlated because NPL can lead to setting up provisions and this lowers
directly the gross profits and thus lowers the ROE (Podpiera and Weill 2008; Abid et al., 2014;
Louzis et al., 2012; Makri et al., 2014). A high negative correlation can indicate a lack of

6
appropriate control of borrowers and/or a weak debt management and missing recovery
strategies. Alternatively, it can mean that targeted profit goals were exaggerated and the bank
had bad luck. On the other hand, high ROE and low NPL ratios indicate the bank had good luck
(Berger and DeYoung 1997). Makri et al. (2014) examine banks of 14 Euro zone countries and
show a negative relationship between ROE and NPL.

The SOL, defined as ratio between net income and sum of liabilities, is negatively
associated with NPL. Abid et al. (2014) use dynamic panel data methods and samples from 2003
to 2012 on 16 Tunisian banks to examine the impact of macroeconomic and bank specific factors
on the performance of bank loans. In principle, the argument is similar to that of ROE, i.e. if the
profits (income over expenditure) increase, the SOL ratio is high, and this implies lower NPLs.1
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

An alternative measure of SOL is the capital adequacy requirement ratio (CAR). The CAR
measures a bank’s capital position and is expressed as a ratio of capital to the risk weighted
assets. In other words, the ratio measures a bank’s capacity to meet the liabilities, credit risk and
operational risk.2 Some studies show that CAR ratio is negatively related to NPL (Berger and
DeYoung, 1997; Makri, Tsagkanos and Bellas, 2014; Dhar and Bakshi, 2015). This is because an
increase in CAR will enable banks to keep more funds aside to meet liabilities and other
obligations which will mean a decrease in the number of initial loans and hence the consequent
NPLs.

According to some studies (Demirgüç-Kunt and Detragiache, 1998; Louzis et al., 2012;
Abid at al., 2014), net interest margin (NIM),3 which is measured by the the difference between
lending and deposit rate, or credit to deposit ratio, has a positive association with the amount of
NPLs because higher NIM increases the interest burden (Sinkey and Greenawalt, 1991; Bofondi
and Ropele, 2011; Adebola et al., 2011). For instance, in a recent study, Messai and Jouini
(2013) examine 85 banks from 3 countries (Italy, Greece and Spain) over the period 2004 to

1
Solvency ratio is a key measure for an entity’s ability to meet its debts and other obligations. It indicates whether
the cash flow of a financial institution is sufficient to meet its short and long term obligations. Moreover, return on
equity (net profit as a ratio of shareholder capital) measures how much profit is earned compared to the total amount
of shareholders equity invested in the firm.
    
2
The solvency ratio can also be defined as: SOL =
      
3
It is a regulatory requirement of the RBF that banks in Fiji announce publicly if the interest rate spread exceeds 4%
(RBF, 1995), thus indirectly forcing the banks to keep the spread in the 4% range to avoid sending a negative signal
to the customers.

7
2008 in a panel regression and find a positive relationship between interest rate and NPL. A
rapid expansion of credits can also be an indication of poor screening and lending to borrowers
of lower quality which may increase NPL. Others (Angbazo 1997; Demirgüç-Kunt and Huizinga
1999; Valverde and Fernández, 2007) argue that a positive relationship is expected because a
high proportion of bad or problem loans may require banks to increase their interest margin to
compensate for their possible default risks. On the contrary, some scholars (Dewartpoint and
Tirole, 1994; Dhar and Bakshi, 2015) argue that NPL and NIM are negatively related because a
decrease in the latter can prompt banks to adopt more risky strategies for survival, thus resulting
in loan portfolios with higher probability of default. For instance, by taking 27 Indian banks into
account, Dhar and Bakshi (2015) show that the credit to deposit ratio is negatively related to
NPL.
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

The ROA is a measure of profitability of a bank and is defined as the ratio of income to
total assets. The ROA indicates the efficiency of the management of a company in generating net
income from all the resources owned by the institution. The effect of ROA on NPL can be
similar to ROE except that in case of the latter, the profits are related to equity and all the
remaining assets. The negative association is because a bank with higher profitability would have
less incentive to generate more profit thus less constrained to engage in risky loans which reduce
the amount of NPL (Godlewski, 2004; Louzis, et al. 2012; Messai and Jouini, 2013). On the
contrary, some studies (Makri et al. 2014; Dhar and Bakshi, 2015) find a positive association of
ROA, however, it is not statistically significant, and therefore fail to explain the variability in
NPL.

The INEF is defined as the ratio of operating expenses to operating income and is a
measure of inefficiency. A positive relationship between NPL and INEF is a sign of bad
management and wastage of resources, and over investing in loan management. However, a
negative relationship implies efficient management of problem loans (Berger and DeYoung
1997). The age or years of existence of a bank (AGE) can be an indicator of banks’ ability to
tolerate risk and hence the NPL. Studies (Wilson, 1967 and 1977; Broecker, 1990; Bofondi and
Gobbi, 2006) find that the AGE of a bank is negatively related to NPL because during startup, a
bank is expected to face an information asymmetry and the fact that it has to attract new

8
customers without being able to distinguish between good and bad borrowers. This is likely to
increase the odds of default. On the other hand, as a bank matures, a sufficient size of customer
base is established and hence banks can focus on streamlining their credit procedures and
ensuring better controls to minimize losses through NPL. However, we have some doubt if these
arguments are valid for banks which are following the Basel rules and where all banks have
access to credit scores about private and institutional borrowers. Therefore, it is more
informative to examine the size (SIZE) effect of a bank viz. the problem loans.

The proxy used for SIZE includes the number of branches, the market share, or the
Herfindahl-Hirschman Index (HHI), where the latter two can be based on total assets, loans or
revenues. The SIZE measured by the number of branches of a bank can have a positive
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

relationship with the NPL because an increase in the number of branches as a strategy to be more
competitive to avoid losing market shares and profits, can in turn expose banks to greater risks
(Calcagnini et al., 1999; Chortareas et al., 2013). However, more branches can provide easy
access to payment of loans and avenues to discuss any potential defaults thus enable banks to
take pro-active actions to minimize NPLs. In some studies, the market share or HHI shows a
negative association with NPL (Salas and Saurina, 2002; Rajan and Dhal, 2003; Hu et. al., 2004),
because SIZE in this sense is considered an approximate measure of diversification opportunities
for loans. However, it is also argued that as SIZE increases, the loan portfolios also increase and
hence the possibility of defaults due to a greater exposure to default risk (Abid et al., 2014).
Moreover, a larger bank can benefit from ‘too big to fail’ as the government will intervene to bail
out in the case of default, thus encouraging larger banks to expose themselves to greater default
risk.

2.3 Non-bank (macro/structural) specific factors


Some possible non-bank specific factors are macroeconomic or structural variables which
can influence the NPL. These include the real effective exchange rate (REER) which is the
exchange rate of the domestic currency against the basket currencies adjusted for inflation rates
between the domestic country and trading partners; the gross domestic product (GDP);
unemployment rate (UNEMP); inflation rate (INFL); amount of remittances (REMIT); political

9
crises (POLIT); and the GFC (FINCRS).4 For instance, during economic expansion (growth in
GDP per capita), households have the necessary streams of income to pay for the loans and
hence the probability of defaults is low (Bangia et al., 2002; Carey, 2002; Salas and Saurina,
2002; Fofack, 2005; Jimenez and Saurina, 2006; Quagliarello, 2007; Louzis et al., 2012; Messai
and Jouini, 2013; Abid at al., 2014). A decline in INFL is likely to improve financial conditions
of households thus enabling them to make prompt and regular repayments of loans (Demirgüç-
Kunt and Detragiache, 1998; Abid et al., 2014). On the other hand, when banks are less
responsive to changes in interest rates and keep lending rates sufficiently high or revise them
according to inflation projections (based on better information or projections), an increase of the
inflation rate may have no or even negative impact on the NPL. A higher REER implies a
depreciation of the national currency and a rise of the international competitiveness of export
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

oriented firms. This will decrease the NPL because as the profits of firms increase, their ability to
pay-up loans increases holding other things constant. It can be argued that UNEMP creates
income constraints and hence it can have a positive association with NPL. However, it is also
plausible that banks have appropriate credit screening processes and provide loans to customers
with smooth income flows and collaterals. This not only minimizes the possibility of loan losses,
but in times of high unemployment, banks can claim assets (fixed savings or similar assets) of
defaulting customers to recover the loan losses efficiently. In this regard, UNEMP can have a
negative impact on NPL as banks reduce or restrict lending. An economy receiving high
remittances (REMIT) can provide alternative income streams to meet financial obligations and
hence have negative impact on NPL. Other factors such as POLIT and FINCRS can have
positive impact on NPL as these events jeopardize economic activity. A political crisis can
severely damage the normal business operations and contribute to loan losses; and FINCRS can
have a positive impact on the loan losses, provided the financial sectors are closely linked to the
international financial system. However, FINCRS can have either no or negative association with
NPLs, if, as discussed, banks are peripherally linked. Another important factor is the banks
business history over time (TIME) that can reflect the overall improvement (or lack of it) of
banks in managing NPLs. A positive association of TIME with NPL indicates a lack of

4
Sami (2013) empirically shows that remittances are important for financial development. The authors’ thank a
reviewer for suggesting to include remittances, unemployment and financial crisis in the estimation.

10
improvement where as a negative impact indicates banks have become better in managing their
losses over time – an indication of banks operating over a learning curve.

3. Data, variable definition, and method


3.1 Data
The data on the bank specific variables is compiled from the annual key disclosure
statements reported by banks to the Reserve Bank of Fiji (RBF) (http://www.rbf.gov.fj/Left-
Menu/Regulatory-Framework/Published-Disclosure-Statements.aspx). The data on
macroeconomic variables is extracted from the World Development Indicators and Global
Development Finance (World Bank, 2015) database. Our sample of 7 banks comprises 5
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

commercial banks and 2 non-bank financial institutions (NBFIs). The 2 NBFIs are the Credit
Co-operation Fiji Limited (CCFL) and Merchant Finance Limited (MFL) both of which have
lending and deposit functions and are required by the RBF to provide the key disclosure
statements. The 5 commercial banks are: Australia and New Zealand Banking Group (ANZ),
WestPac Banking Corporation (WBC), Bank of the South Pacific (BSP), Bank of Baroda (BOB),
and Home Finance Corporation (HFC).5 We use a balanced panel sample from 2000-2013.

3.2 Variable definition and plausible signs


In Table 1, we define the variables and the respective expected signs in relation to
the NPL. Based on the expected signs, we formulate 14 hypotheses (see Table 1). For
instances, the first hypothesis is, H1: NPL and ROE are negatively associated, the second
hypothesis is, H2: NPL and SOL are negatively associated, the third hypothesis, H3: NPL
and NIM are positively associated, and so on.

<<INCLUDE TABLE 1 HERE>>

The bank-specific factors are ROE, INEF, SOL, NIM, and SIZE and the macroeconomic
and structural factors are GRWTH, INFL, REER, UNEMP, REMIT, POLIT, FINCRS and

5
Bred Bank is a French bank which joined Fiji’s banking sector in early 2013. Due to data unavailability, it is
excluded from the sample. Moreover, we exclude Habib Bank from the sample because it seized to operate after
2005 and was acquired by BSP. BSP also acquired Colonial Group in 2009.

11
TIME. The descriptive statistics of the variables are presented in Table 2. For the purpose of
regression estimation and to minimize standard errors, the variables are taken as percent and
transformed into natural logarithm.

<<INCLUDE TABLE 2 HERE>>

3.3 Method
The model set-up follows from Demirgüç-Kunt and Huizinga (1999). We examine the
significance of bank-specific and macroeconomic/structural variables explaining NPL in Fiji.
The general equation is expressed as:
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

"
NPL = α + ∑
%&  &(
&'&
#$ X + ∑,)#$
%& &'&*)&(
Z)  + ε (1)

")  ,   / . .

where 01234 represents the volume of non-performing loans (bad debts) as a percent of total
volume of loans of bank 5 at time 6; 78 represents the bank specific factors of bank 5 at time 6
(see Table 1). Further, we extend the base model to examine the statistical significance of
macroeconomic/structural variables denoted by 9; the ε is a stochastic disturbance term.
Subsequently, a more specific model for estimation is expressed as:

NPL = α + β$ ROE + β= SOL + β> NIM + βA INEF + βC SIZE +


%&&&&&&&&&&&&&&&&'&&&&&&&&&&&&&&&&( θG
 Z4 + ε (2)
)  (  )    / .. 

where Z include: INFL, GRWTH, 6


UNEMP, REER, REMIT, POLIT, FINCRS, and TIME,
respectively (Table 1). The subscript i denotes the specific bank (i = 1, … , 7); and the subscript
t denotes the year (t = 2000, … , 2013). In case of the fixed-effect estimation, we assume a one-
way error component: ε = µ + λ + ϑ where µ denotes the unobservable bank-specific
effect (certain characteristics of a bank’s staff such as skills, knowledge and experience in the
operational activities, and organizational and hierarchical structure may influence NPL); the λ is
a time-specific component and may include other macroeconomic factors like government debt,

^ _`a ^ _`a


6
More specifically, for estimation purpose, GRWTH = YZln ] b − ln ] b d + 1e × 100 to avoid
a .   a .  $
negative growth rates.

12
among other things not explicitly captured in the model; and ϑ denotes a random term which is
assumed to be independently and identically distributed. We examine equation (2) using three
methods – pooled OLS (PE)-, random effect (RE)-, and fixed effect (FE) -method, respectively.
The reason for using three methods is to identify the similarities and/or differences. Further, by
including the macroeconomic/structural factors and hence extending the base model, we can
examine the variations in the coefficients of the bank-specific variables. Subsequently, the
robustness of the results is supported if the results derived from application of the three methods
are consistent and on average, quantitatively similar.

4. Results
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

To examine the association between bank-specific and macroeconomic/structural factors


viz. NPL, first, we estimate the base model (model I), which consists only of the bank-specific
factors and then we extend the model to other factors (models II-X) as reported in Tables 3-5.
We estimate the respective models using the PE (Table 3), RE (Table 4) and FE (Table 5)
estimation, respectively. The discussion of the results is based on the average results from the
three estimations and ten models per estimation (Tables 3-5).

The results from all the estimations and models show that the coefficient of ROE is
negative and statistically significant at 1% level and hence confirm H1. On average, an increase
in ROE by 10% is associated with a decrease of the NPL by 0.37%. Moreover, based on the
estimations done using PE, we note that SOL is positive and significant at 5% level. However,
the coefficient of SOL is negative in the RE (Table 4), FE (Table 5), and model VII of PE (Table
3) estimations, and statistically significant at 10%, 1%, and 5%, respectively. Hence, based on
the overall and average results, we confirm H2. On average, a 10% increase in SOL will result in
a decrease of NPL by 0.02%. This is usually the case in small economies like Fiji, where banks
keep a large amount of capital aside because of the difficulty to diversify risk in a small economy
and the legal restrictions on banks to invest abroad. The coefficient of NIM is positive and
statistically significant at a 1% level in all the three estimations (Tables 3-5, models I-X), thus
confirming H3. Hence a 10% increase in NIM will increase the NPL by 0.08%.

13
We note the coefficient of INEF is negative but generally not statistically significant in
all three estimations (Tables 3-5, models I-X). Hence, the negative relationship weakly confirms
H4. On average, a 10% increase in INEF will result in a decrease of NPL by an average of
0.02%, which weakly indicates an efficient management of bad/problem loans. While this
argument is plausible, this factor plays a minimal role in explaining the variations in NPL.

The variable SIZE has on average, a negative coefficient and is statistically significant in
two – PE and RE (Tables 3 and 4) – out of the three estimations. Hence in general, the results
confirm H5. A large dominant bank can have low amount of loan losses since they have
sufficient resources to recover debts. An obvious reason is that a large bank has a bigger
selection regarding which loan requests they are willing to accept than small banks; naturally it
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

will prefer less risky investments. Further, a large bank can afford a good monitoring system and
incur the necessary fixed (overhead) costs compared to small banks. On the other hand, smaller
banks may want to attract small and unsecured loans to increase their loan portfolio and in doing
so, expose themselves to greater odds of defaults. Therefore, overall and on average, a 10%
increase in SIZE will reduce the NPL by 0.03%. The coefficient of HHI, a measure of
concentration, is negative but not statistically significant within the conventional levels, and
hence marginally explains the variations in NPLs. A higher HHI indicates that few large banks
dominate the market and hence have similar effect as SIZE.

In terms of macroeconomic factors (Tables 3-5, models III-X), we note that the
coefficients of GRWTH, INFL, REMIT are negative in all three but not statistically significant
within the conventional levels. Hence the results weakly confirm H7, H8 and H11. On average, a
10% increase in GRWTH, INFL and REMIT will reduce NPL by 0.005%, 0.015%, and 0.009%,
respectively. Further, the coefficient of UNEMP is negative and statistically significant, and
hence we fail to accept H9. Therefore, on average a 10% increase in UNEMP is associated with a
0.05% decline of NPL. The coefficient of FINCRS is negative however not statistically
significant. In this case, we weakly fail to accept H13. A reason for the negative association is
that the banking sector in Fiji and similar small developing economies as alluded earlier is due to
the banks’ weak integration in the international financial market. The banking sector in Fiji is
only subject to delayed and indirect effects ensuing from the international financial market.

14
Thus, banks which are remotely linked to the global financial markets have the necessary space
to adjustment and review the lending strategies to minimize problem loans. We also note the
coefficient of TIME (a measure of bank credit policy evolution and learning curve) is negative
and statistically significant in all the three estimations (Tables 3-5) and hence confirms H14.
Thus, banks are able to improve their NPL position over TIME.

Based on the three estimates, we note that the coefficient of REER is negative, however
not significant, and therefore weakly supports H10. An increase in REER implies a currency
depreciation which benefits domestic firms engaged in exports and households receiving
remittances, which can enhance the regularity in loans repayments. The coefficient of POLIT is
positive, however not statistically significant in FE and RE estimations. Nevertheless, the
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

positive association weakly confirms H12. Of course, political crisis have an adverse and direct
impact on investments, employment and income which creates serious cash flow constraints thus
resulting in higher odds of defaults. The weak association indicates the economic situation was
relatively stable in the last 20 years in Fiji and that banks have become relatively resilient to such
shocks.

<<INCLUDE TABLE 3 HERE>>

<<INCLUDE TABLE 4 HERE>>

<<INCLUDE TABLE 5 HERE>>

5. Conclusion
In this paper, we examined the determinants of NPLs in a banking sector in small
developing economies where we considered Fiji as a case study. We looked at 7 banks (5
commercial banks and 2 non-bank financial institutions). While we acknowledge that a
number of studies related to banking sector in small developing economies are emerging,
there is currently no study in regards to NPLs. Such studies are important because for

15
banks to remain stable, especially in small economies which are constrained by scale and
scope, problem loans need to be managed to avoid bank collapses. The bank-specific
factors that have a strong (based on statistical significance) negative association with NPL
are ROE, SOL and SIZE; and the factor that has a strong positive association with NPL is
NIM. Further, we note that INEF is negative and weakly associated with NPL. This
information is important for banks in order to understand the dynamics of NPL and to
effectively manage problem loans. The macroeconomic/structural factors that associate
negatively and strongly with NPL are UNEMP and TIME. This implies that with an
increase in UNEMP, banks become reluctant to provide loans or do so under strict scrutiny
and conditions to avoid any current or future NPLs. However, we note that banks are
responsive to structural changes and hence over TIME reorient their credit policy to ensure
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

quality loans and lower bad loans, which is also an indication of the presence of learning
curve for banks. All other factors are weakly associated with NPL.

Since most of the statistically significant indicators of NPL identified are bank-
specific, banks therefore can monitor them effectively to manage problem loans. For
example, to ensure bank stability in a small economy like Fiji, bankers and bank regulators
should focus on improving profitability (returns), solvency (capital), and increasing their
market share – as indicated by ROE, SOL and SIZE, respectively. The results seem to
suggest the strategy of increasing the size of banks which in most cases is easier for banks
that are already of a reasonably large size. A bigger sized bank is in better position to
increase the solvency and with an increasing market power, enhance the ROE and reduce
NPL. However, this strategy will marginalize small banks and in extreme case result in
small banks vanishing from the market. From the view of the bigger banks, this strategy
may seem to be promising, however from the perspective of the whole economy, this is a
questionable strategy because a higher ROE can deprive financial resources from other and
maybe more productive sectors of the economy. Additionally, few huge banks in the
economy create the problem of becoming ‘too big to fail’. At minimum, what we can learn
from the global financial crisis is that large banks recognize they are ‘too big to fail’ and
are willing to take irresponsible risks because in the worst case, they will be bailed out at
the expense of taxpayers. We also noted that a high interest margin has a high susceptibility

16
of odds of default and therefore while on one hand, banks may want to maximize returns
from lending, it is prudent that appropriate lending contracts are set up to avoid losses from
defaults. From borrower’s perspective, an increase in lending rates creates further
constraints to borrowing and puts additional repayment burden. However, banks can
increase the lending rates given the economic climate and within the caveats set by the
central bank of the country. It is also plausible that banks may keep the margin high to
discourage borrowings in the environment of low scope for investment, high
unemployment, few opportunities for diversification given the small economy, and off-
shore investment caveats set by the financial regulatory bodies such as the Reserve Bank of
Fiji. Another reason for banks preferring to keep a high margin is when households have
few investment options and hence choose to deposit or keep money safely in a bank even
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

when the latter offers low rates. Our results show that, in any case, a high net interest
margin can be counterproductive for banks (Kishore, 2012). Importantly, banks need to
monitor the macroeconomic/structural changes and review its lending policy regularly.
Nevertheless, banks in most cases have expert projections and adjust credit policy to
minimize problem loans. Our results show that unemployment and loan losses are
negatively associated, and a possible reason for this is that banks put in tighter measures to
recover loans during economic contractions combined with a strategy of rejecting to
provide loans to households which are more exposed to the risk of joblessness.

The results presented in the paper provide further insights on the strong and weak
determinants of NPL in a banking sector of a small developing economy such as Fiji.
However, while our models and estimations have tested for factors explaining NPL, we
understand that the causality may go in other directions. In this regard, the study can be
extended in multiple dimensions. For example, future research may consider the
determinants of the capital adequacy requirements, net interest margins, etc. It would be
interesting to consider other small economies as a comparative study to identify if there are
any differences in the way banks operate with respect to NPLs. Moreover, with larger
sample size, a slice and dice approach can be used to examine the pre- and post- effects of
structural breaks.7

7
The authors’ thank an anonymous reviewer for suggesting this point.

17
References
Abid, L., Ouertani, M.N. and Zouari-Ghorbel, S. (2014), “Macroeconomic and bank-specific
determinants of household's non-performing loans in Tunisia: A dynamic panel data”, Procedia
Economics and Finance, Vol. 13, pp. 58-68.
Abedola, S.S., Wan Yusoff, W.S. and Dalahan, J. (2011), “An ARDL approach to the determinants of
nonperforming loans in Islamic banking system in Malaysia”, Kuwait Chapter of Arabian Journal of
Business and Management Review, Vol. 1, No. 1, pp.20-30.
Admati, A. R., Demarzo, P.R., Hellwig, M.F. and Pfleiderer, P. C. (2011), “Fallacies, irrelevant facts, and
myths in the discussion of capital regulation: Why bank equity is not expensive”, Working Paper 86.
Rock Center for Corporate Governance at Stanford University, and Research Paper 2065. Stanford
Graduate School of Business, Stanford, CA.
Angbazo, L. (1997), “Commercial bank net interest margins, default risk, interest-rate risk, and off-
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

balance sheet banking”, Journal of Banking & Finance, Vol. 21, No. 1, pp. 55-87.
Bangia, A., Diebold, F.X., Kronimus, A., Schagen, C. and Schuermann, T. (2002), “Ratings migration
and the business cycle, with application to credit portfolio stress testing”, Journal of Banking & Finance,
Vol. 26, No. 2-3, pp. 445-474.
Beck, T., Levine, R., and Loayza, N. (2000), “Finance and the sources of growth”, Journal of Financial
Economics, Vol. 58, No. 1-2, pp. 261-300.
Berger, A.N. and DeYoung, R. (1997), “Problem loans and cost efficiency in commercial banks”,
Journal of Banking & Finance, Vol. 21, No. 6, pp. 849-870.
Bofondi, M. and Gobbi, G. (2006), “Informational barriers to entry into credit market”, Review of
Finance, Vol. 10, No. 1, pp. 39-67.
Bofondi, M. and Ropele, T. (2011). “Macroeconomic determinants of bad loans: Evidence from Italian
Banks”, Bank of Italy Occasional Papers 89. http://ssrn.com/abstract=1849872 or
http://dx.doi.org/10.2139/ssrn.1849872
Broecker, T. (1990), “Creditworthiness tests and interbank competition”, Econometrica, Vol. 58, No. 2,
pp. 429-452.
Calcagnini, G., Bonis, R. D. and Hester, D.D. (1999), “Determinants of bank branch expansion in Italy”,
Working Paper 9932, Social Systems Research Institute, The University of Wisconsin – Madison.
www.ssc.wisc.edu/econ/archive/wp9932.pdf
Carey, M. (2002), “A guide to choosing absolute bank capital requirements”, Journal of Banking &
Finance, Vol. 26, No. 5, pp. 929-951.
Chortareas, G.G., Girardone, C. and Ventouri, A. (2013), “Financial freedom and bank efficiency:
evidence from the European Union”, Journal of Banking & Finance, Vol. 37, No. 4, 1223-1231.
Demirgüç-Kunt, A. and Detragiache, E. (1998), “The determinants of banking crisis in developing and
developed countries”, IMF Staff Papers, Vol. 45, No. 1, pp. 81-109.
Demirgüç-Kunt, A. and Huizinga, H. (1999), “Determinants of commercial bank interest margins and
profitability: some international evidence”, World Bank Economic Review, Vol. 13, No. 2, pp. 379-408.
Demirgüç-Kunt, A. and Levine, R. (2009), “Finance and inequality: theory and evidence”, Annual Review
of Financial Economics, Vol. 1, No. 1, pp. 287-318.

18
Dewartpoint, M. and Tirole, J. (1994), The Prudential Regulation of Banks. MIT Press, MA.
Dhar, S., & Bakshi, A. (2015). Determinants of loan losses of Indian banks: A panel study. Journal of
Asia Business Studies, 9 (1), 17-32.
Fofack, H. (2005), “Non-performing loans in sub-Saharan Africa: causal analysis and macroeconomic
implications”, World Bank Policy Research Working Paper 3769.
http://elibrary.worldbank.org/doi/pdf/10.1596/1813-9450-3769
Godlewski, C.J. (2004), “Capital regulation and credit risk taking: empirical evidence from banks in
emerging market economies”, LaRGE, Institut d’Etudes Politiques, Université Robert Schuman,
Strasbourg 3. http://econwpa.repec.org/eps/fin/papers/0409/0409030.pdf
Gounder, N. and Sharma, P. (2012). “Determinants of bank net interest margins in Fiji - a small island
developing state”, Applied Financial Economics, Vol. 22, No. 19, pp. 1647-1654.
Goundar, R. (2004), “Fiji's economic growth impediments”, Journal of the Asia Pacific Economy, Vol. 9,
No. 3, pp. 301-324.
Goundar, R. (2005), “Neglected dimension of development: inequality, conflict and aid”, International
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

Journal of Social Economics, Vol. 32, No. 1-2, pp. 60-76.


Grynberg, R., Munro, D. and White, M. (2002), Crisis: the collapse of the National Bank of Fiji,
Crowford House Publishing.
Lodhia, S.K., and Burritt, R.L. (2004), “Public sector accountability failure in an emerging economy”,
International Journal of Public Sector Management, Vol. 17, No. 4, pp. 345-359.
Haldane, A., Brennan, S. and Madouros. V. (2010), “What is the contribution of the financial sector:
miracle or mirage?”, in: A. Turner, A. Haldane, P. Woolley, S. Wadhwani, C. Goodhart, A. Smithers, A.
Large, J. Kay, M. Wolf, P. Boone, S. Johnson, & R. Layard, R. (Eds.). The future of finance: The LSE
report, London School of Economics and Political Science, 87-121.
https://harr123et.files.wordpress.com/2010/07/futureoffinance5.pdf
Hausman, J.A. (1978), “Specification tests in econometrics”, Econometrica, Vol. 46, No. 6, pp.1251-71.
Hsiao, C. (2003), Analysis of panel data. 2nd edition, Cambridge: Cambridge University.
Hu, J-L., Li, Y. and Chiu, Y-H. (2004), “Ownership and nonperforming loans: evidence from Taiwan’s
banks”, The Developing Economies, Vol. 42. No. 3, pp. 405-420.
Jimenez, G. and Saurina, J. (2006), “Credit cycles, credit risk, and prudential regulation”, International
Journal of Central Banking, Vol. 2, No. 2, pp. 65-98.
Khemraj, T. and Pasha, S. (2009), “The determinants of non-performing loans: an econometric case study
of Guyana”, MPRA Paper No. 53128, MPRA. http://mpra.ub.uni-
muenchen.de/53128/1/MPRA_paper_53128.pdf
Kishore, R. (2012), “A comparison of bank margin and profitability”, International Journal of Education
Economics and Development, Vol. 3, No. 2, pp. 113-126.
Kumar, R.R. and Patel, A. (2014), “Exploring competitiveness in banking sector of a small island
economy: a study of Fiji”, Quality & Quantity, Vol. 48, No. 6, pp. 3169-3183.
Lee, C.C. and Hsieh, M.F. (2013), “The impact of bank capital on profitability and risk in Asian
banking”, Journal of International Money and Finance, Vol. 32, pp. 251-281.
Levine, R. (2005), “Finance and growth: theory and evidence”, in: P. Aghion and S. Durlauf (EDs.).
Handbook of Economic Growth, edition 1, Vol. 1, chapter 12, pages 865-934, Elsevier.

19
Louzis, D.P., Vouldis, A.T. and Metaxas, V.L. (2012), “Macroeconomic and bank-specific determinants
of non-performing loans in Greece: A comparative study of mortgage, business and consumer loan
portfolios”, Journal of Banking & Finance, Vol. 36, No. 4, pp. 1012-1027.
Overton, J.D. (2003), “Understanding coups”, Asia Pacific Viewpoint, Vol. 44, No. 3, pp. 351-355.
Makri, V., Tsagkanos, A. and Bellas, A. (2014), “Determinants of non-performing loans: The case of
Eurozone”, Panoeconomicus, Vol. 61, No. 2, pp. 193-206.
Messai, A.S. and Jouini, F. (2013), “Micro and macro determinants of non-performing loans”,
International Journal of Economics and Financial Issues, Vol. 3, No. 4, pp. 852-860.
Mirzaei, A., Moore, T. and Liu, G. (2013), “Does market structure matter on banks’ profitability and
stability? Emerging vs. advanced economies”, Journal of Banking & Finance, Vol. 37, No. 8, pp. 2920-
2937.
Mwega, F. (2011), “The competitiveness and efficiency of the financial services sector in Africa: A case
study of Kenya”, African Development Review, Vol. 23, No. 1, pp. 44-59.
Panzar, J.C. and Rosse, J.N. (1987), “Testing for “monopoly” equilibrium”, The Journal of Industrial
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

Economics, Vol. 35(4), pp. 443-456.


Podpiera, J. and Weil, L. (2008), “Bad luck or bad management? Emerging banking market experience”,
Journal of Financial Stability, Vol. 4, No. 2, pp. 135-148.
Poshakwale, S.S. and Qian, B. (2011), “Competitiveness and efficiency of the banking sector and
economic growth in Egypt”, African Development Review, Vol. 23, No. 1, pp. 99-120.
Quagliarello, M. (2007), “Banks’ riskiness over the business cycle: a panel analysis on Italian
intermediaries”, Applied Financial Economics, Vol. 17, No. 2, pp. 119-138.
Rajan, R. and Dhal, S.C. (2003), “Non-performing loans and terms of credit of public sector banks in
India: an empirical assessment”, Reserve Bank of India Occasional Papers, Vol. 24, No. 3, pp. 81-121.
https://www.rbi.org.in/scripts/PublicationsView.aspx?id=7206
Reinhart, C.M. and Rogoff, K.S. (2011), “From financial crash to debt crisis”, American Economic
Review, Vol. 101, No. 5, pp. 1676-1706.
Reserve Bank of Fiji (RBF) (1995), “Banking Supervision Policy Statement No: 17, Interest Spread
Disclosure Requirements for Banks”, Banking Act 1995, RBF.
Salas, V. and Saurina J. (2002), “Credit risk in two institutional regimes: Spanish commercial and savings
banks”, Journal of Financial Services Research, Vol. 22, No. 3, pp. 203-224.
Sami, J. (2013), “Remmittances, banking sector development and economic growth in Fiji”, International
Journal of Economics and Financial Issues, Vol. 3, No. 2, pp. 503-511.
Sharma, P. and Gounder, N. (2015), “Resilient through the GFC and beyond: what drives bank
profitability in small open Pacific economies?”, Journal of Asia-Pacific Business, Vol. 16, No. 3, pp. 191-
209.
Sharma, P. and Nguyen, D-T. (2010), “Law and banking development in a South Pacific island economy:
the case of Fiji, 1970-2006”, Journal of the Asia Pacific Economy, Vol. 15, No. 2, pp. 192-216.
Sharma, P., Gounder, N. and Xiang, D. (2015), “Level and determinants of foreign bank efficiency in a
Pacific island country”, Review of Pacific Basin Financial Markets and Policies, Vol. 18, No. 01, pp. 1-
26.
Sharma, P., Doca, E., Dakai, V. and Manoa, S. (2014), “An assessment of Fiji’s banking sector on a
global scale: 2000-2011”, Policy Research Working Paper #1, Reserve Bank of Fiji.

20
http://www.rbf.gov.fj/getattachment/Publications-%281%29/Working-Papers/Reserve-Bank-of-Fiji-and-
Griffith-University-Worki/Assessment-of-Fiji-s-BS-RBF-GU.pdf.aspx
Sharma, P. and Gounder, N.N. (2013), “Does finance matter for growth in the small, open Pacific Island
countries?”, The Journal of Pacific Studies, Vol. 33, No. 2, pp. 22-29.
Sharma, P., Gounder, N. and Xiang, D. (2013), “Foreign banks, profits, market power and efficiency in
PICs: some evidence from Fiji”, Applied Financial Economics, Vol. 23, No. 22, pp. 1733-1744.
Sinkey, J.F. and Greenawalt, M.B. (1991), “Loan-loss experience and risk taking behaviour at large
commercial banks”, Journal of Financial Services Research, Vol. 5, No. 1, pp. 43-59.
Turner, A. (2010), “What do banks do? Why do credit booms and busts occur and what can we do about
it?”, in A. Turner, A. Haldane, P. Woolley, S. Wadhwani, C. Goodhart, A. Smithers, A. Large, J. Kay,
M. Wolf, P. Boone, S. Johnson, & R. Layard, R. (Eds.). The future of finance: The LSE report, London
School of Economics and Political Science, 5-86.
https://harr123et.files.wordpress.com/2010/07/futureoffinance5.pdf
Valverde, S.C. and Fernández, F.R. (2007), “The determinants of bank margins in European banking”,
Journal of Banking & Finance, Vol. 31, No. 7, pp. 2043-2063.
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

Wilson, R. (1977), “A bidding model of perfect competition”, Review of Economic Studies, Vol. 44, No.
3, pp. 511-518.
Wilson, R.B. (1967), “Competitive bidding with asymmetric information”, Managements Science, Vol.
13, No. 1, pp. 816-820.
World Bank (2015), World Development Indicators and Global Development Finance, World Bank,
Washington D.C.

21
Table 1: Indicators
Hypothesis Hypothesis
Variable Name Definition
(H#) tested
Bank Specific
 
N.A. Non-performing loans (NPL)  = N.A.


H1 Return on Equity (ROE)  = (-)
  ! " #$%&
(  ) *+ ,   - *+.
H2 Solvency (SOL) ' = (-)
"/ 0 1 ! "" "

Net Interest Margin (NIM) 2 4  256  − 2 4  89 


H3 23 = (+)
:; < 

Inefficiency (INEF) 9 4>? 89  


H4 2= = (-)
9 4>?  @ A 
:;  ; < 
H5 Bank Size (SIZE) '2B = D (-)
∑E) :;  ; < 
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

-
 HI"
H6 Concentration (HHI) FF2 = ∑DE) G∑K N (-)
LM  J  HI"
Macroeconomic/Structural
J  TU J  TU
OP:F = ln S V − ln S V , (World
H7 Economic Growth (GRWTH) +%I  +%I W) (-)
Bank 2015)

H8 Inflation (INFL) 2= = ln X2 − ln X2W) , (World Bank 2015) (+)

H9 Unemployment (UNEMP) Y3 = 1 – 69;\6  4 , (World Bank 2015) (+)


Real effective exchange rate. Higher REER implies
H10 Real Effective Exchange rate (REER) (+)
depreciation of Fiji dollar (World Bank, 2015)
J ]I* "
H11 Workers’ Remittances (REMIT) 32: = , World Bank (2015) (-)
TU
1, 2000  2006
Political Crisis (POLIT) 2: = ^ (+)
0, ; cℎ 4
H12
1, 2007  2008
Financial Crisis (FINCRS) 2: = ^
0, ; cℎ 4
H13
H14 Bank policy/regulatory changes (TIME) :23 = 4  @4> ; . (-)
Notes: Hypotheses (1)-(7) are bank specific variables and (8)–(14) are macroeconomic and structural factors. N.A. – not applicable
Table 2: Descriptive statistics and correlation matrix
Descriptive Statistics
NPL ROE INEF NIM SOL SIZE HHI GRWTH INFL REER UNEMP REMIT
Mean 1.07 31.25 38.12 4.22 12.54 14.29 2711.29 0.92 3.91 99.79 48.31 5.27
Median 0.68 29.04 33.33 4.75 9.20 9.17 2678.20 1.10 3.55 99.47 48.80 5.37
Maximum 7.63 72.20 94.85 5.76 36.18 42.02 2930.10 4.95 8.67 108.00 49.90 6.77
Minimum -0.95 -56.85 15.47 1.79 0.84 1.54 2528.90 -2.41 0.76 91.65 46.10 2.61
Std. Dev. 1.38 17.28 16.54 1.43 7.97 13.14 120.73 2.01 2.18 4.52 1.50 1.04
Skewness 2.10 -0.95 1.32 -0.58 1.28 0.80 0.59 -0.08 0.75 0.36 -0.30 -0.87
Kurtosis 9.47 8.74 4.69 1.81 3.55 2.21 2.28 2.59 2.95 2.60 1.34 3.85
Jarque-Bera 242.90 149.47 40.21 11.41 27.89 12.90 7.82 0.80 9.15 2.81 12.69 15.29
Probability 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.67 0.01 0.25 0.00 0.00
Correlation
NPL 1.00
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

ROE -0.52 1.00


INEF -0.17 -0.06 1.00
NIM 0.26 0.08 0.12 1.00
SOL 0.43 -0.15 -0.31 -0.14 1.00
SIZE -0.42 0.43 0.37 0.00 -0.48 1.00
HHIL -0.11 -0.13 -0.27 -0.46 0.14 0.00 1.00
GRWTH -0.06 0.07 0.21 0.27 0.00 0.00 -0.41 1.00
INFL -0.10 -0.13 -0.11 -0.22 0.15 0.00 0.56 -0.03 1.00
REER -0.07 0.07 -0.11 -0.13 0.00 0.00 -0.24 -0.05 0.25 1.00
UNEMP -0.19 -0.18 0.08 -0.57 0.21 0.00 0.58 -0.33 0.63 0.10 1.00
REMIT -0.21 0.23 -0.03 -0.30 0.07 0.00 -0.15 0.35 -0.20 0.03 -0.20 1.00
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

Table 3: Pooled OLS: Dependent variable (NPL)


Independent
I II III IV V VI VII VIII IX X
Variable
A A A A A A A A A
Constant 4.793308 5.127477 4.784184 4.797575 4.761210 5.084818 4.791703 4.781735 4.794944 6.272860A
(0.033710) (0.203821 (0.034418) (0.033657) (0.106065) (0.145475) (0.034051) (0.033769) (0.034204) (0.558455)
-0.038865A -0.044162A -0.037812A -0.039060A -0.039034A -0.040306A -0.037893A -0.037738A -0.038956A -0.038506A
ROE
(0.006600) (0.006139) (0.006636) (0.006565) (0.006653) (0.006524) (0.006982) (0.006532) (0.006637) (0.006394)
0.004942B 0.007053A 0.005213B 0.005337B 0.005011B 0.005566A -0.005003B 0.005558A 0.004918B 0.005978A
SOL
(0.002109) (0.001770) (0.002114) (0.002116) (0.002130) (0.002095) (0.002123) (0.002103) (0.002120) (0.002080)
0.009848A 0.008924A 0.010526A 0.008974A 0.009971A 0.006993B 0.009456A 0.010578A 0.009714A 0.006396B
NIM
(0.002366) (0.002559) (0.002422) (0.002431) (0.002409) (0.002708) (0.002535) (0.002363) (0.002409) (0.002635)
INEF -0.002335 -0.005483B -0.001410 -0.002562 -0.002225 -0.001248 -0.002118 -0.001508 -0.002589 0.001154
(0.002937) (0.002766) (0.003023) (0.002925) (0.002971) (0.002934) (0.002990) (0.002927) (0.003042) (0.003134)
0.033303C -0.002566C -0.002269C -0.002429C -0.002400C -0.002507C -0.002468C -0.002394C -0.002811B
SIZE -
(0.001328) (0.001328) (0.001325) (0.001334) (0.001305) (0.001342) (0.001309) (0.001340) (0.001294)
-0.038726
HHI - - - - - - - -
(0.024751)
- -0.000629 -
GRWTH - - - - - - -
(0.000510)
-0.002195
INFL - - - - - - - - -
(0.001542)
0.006993
REER - - - - - - - - -
(0.021897)
B
-0.073770
UNEMP - - - - - - - - -
(0.035847)
-0.002050
REMIT - - - - - - - - -
(0.004626)
C
0.005334
POLIT - - - - - - - - -
(0.002777)
-0.000985
FINCRS - - - - - - - -
(0.002863)
-0.000743A
TIME - - - - - - - - -
(0.000280)
Adj − R- 0.512168 0.507402 0.514927 0.517546 0.507360 0.528740 0.507870 0.526022 0.507449 0.542238
F-stat 21.36782 20.98301 18.16162 18.34255 17.64970 19.13851 17.68370 18.94178 17.65562 20.15005
DW-stat 1.395730 1.508044 1.418383 1.479326 1.410810 1.458526 1.400222 1.494530 1.382088 1.480041
Notes: A, B and C indicates 1%, 5% and 10% level of statistical significance
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

Table 4: Random Effects: Dependent variable (NPL)


Independent
I II III IV V VI VII VIII IX X
Variable
A A A A A A A A A
Constant 4.811572 5.024476 4.805284 4.813450 4.820459 5.027369 4.810905 4.803551 4.813260 5.966869A
(0.028595) (0.166585) (0.029242) (0.028731) (0.084190) (0.118744) (0.029045) (0.028966) (0.028882) (0.464905)
-0.036147A -0.039120A -0.035605A -0.036227A -0.036077A -0.037907A -0.035867A -0.035542A -0.036123A -0.037290A
ROE
(0.005686) (0.005610) (0.005712) (0.005703) (0.005755) (0.005713) (0.006039) (0.005680) (0.005710) (0.005578)
-0.004506C -0.000289 -0.004201C -0.004008C -0.004510C -0.003811 -0.004440C -0.003763 -0.004498C -0.003432
SOL
(0.002370) (0.002037) (0.002388) (0.002422) (0.002397) (0.002384) (0.002393) (0.002413) (0.002378) (0.002372)
0.008407A 0.008247A 0.008858A 0.007962A 0.008376A 0.006307A 0.008313A 0.008934A 0.008260A 0.005637A
NIM
(0.001894) (0.002010) (0.001944) (0.001949) (0.001929) (0.002184) (0.002020) (0.001917) (0.001921) (0.002154)
INEF -0.002618 -0.005547B -0.001872 -0.002865 -0.002666 -0.001462 -0.002577 -0.002058 -0.003064 0.001048
(0.002723) (0.002743) (0.002818) (0.002741) (0.002765) (0.002768) (0.002754) (0.002737) (0.002850) (0.003041)
-0.005370B -0.005351B -0.005217B -0.005364B -0.005481B -0.005363B -0.005317B -0.005306B -0.005806A
SIZE -
(0.002198) (0.002199) (0.002205) (0.002201) (0.002194) (0.002202) (0.002197) (0.002203) (0.002192)
-0.026467
HHI - - - - - - - -
(0.020057)
- -0.000419 -
GRWTH - - - - - - -
(0.000406)
-0.001267
INFL - - - - - - - - -
(0.001238)
-0.001961
REER - - - - - - - - -
(0.017357)
C
-0.054133
UNEMP - - - - - - - - -
(0.028939)
-0.000542
REMIT - - - - - - - - -
(0.003669)
0.003471
POLIT - - - - - - - - -
(0.002230)
-0.001251
FINCRS - - - - - - - -
(0.002274)
-0.000578B
TIME - - - - - - - - -
(0.000232)
Adj − R- 0.408439 0.379783 0.408394 0.408295 0.402004 0.422754 0.402059 0.416494 0.403771 0.437911
F-stat 14.39459 12.87938 12.16007 12.15548 11.86809 12.83989 11.87055 12.53940 11.94821 13.59511
DW-stat 1.840409 1.752978 1.863448 1.901195 1.834477 1.895135 1.840100 1.911918 1.819357 1.940675
Notes: A, B and C indicates 1%, 5% and 10% level of statistical significance
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

Table 5: Fixed Effects: Dependent variable (NPL)


Independent
I II III IV V VI VII VIII IX X
Variable
A A A A A A A A A
Constant 4.809094 4.872838 4.799544 4.811824 4.831199 5.010080 4.808571 4.803782 4.811305 5.887093A
(0.036618) (0.172372) (0.037914) (0.036921) (0.085905) (0.128231) (0.037658) (0.036723) (0.037035) (0.475853)
-0.035200A -0.034951A -0.034528A -0.035326A -0.034963A -0.037324A -0.035054A -0.034804A -0.035177A -0.036875A
ROE
(0.005915) (0.005873) (0.005956) (0.005934) (0.006004) (0.006001) (0.006340) (0.005902) (0.005941) (0.005823)
-0.007270A -0.006361B -0.006819B -0.006924B -0.007342A -0.006805B -0.007243A -0.006602B -0.007300A -0.006285B
SOL
(0.002692) (0.002488) (0.002732) (0.002743) (0.002718) (0.002681) (0.002739) (0.002733) (0.002704) (0.002664)
0.007837A 0.007554A 0.008261A 0.007541A 0.007747A 0.005988A 0.007792A 0.008294A 0.007706A 0.005291B
NIM
(0.001906) (0.002018) (0.001956) (0.001956) (0.001943) (0.002201) (0.002034) (0.001933) (0.001932) (0.002173)
INEF -0.002104 -0.002606 -0.001332 -0.002320 -0.002220 -0.001079 -0.002083 -0.001670 -0.002548 0.001418
(0.002827) (0.003039) (0.002936) (0.002851) (0.002871) (0.002869) (0.002860) (0.002837) (0.002972) (0.003166)
-0.003819 -0.002760 -0.004147 -0.003669 -0.006087 -0.003752 -0.004174 -0.003969 -0.005805
SIZE -
(0.006934) (0.007020) (0.006969) (0.006991) (0.007006) (0.007046) (0.006914) (0.006970) (0.006828)
-0.009246
HHI - - - - - - - -
(0.020668)
- -0.000401 -
GRWTH - - - - - - -
(0.000411)
-0.000892
INFL - - - - - - - - -
(0.001249)
-0.004969
REER - - - - - - - - -
(0.017448)
C
-0.048493
UNEMP - - - - - - - - -
(0.029676)
-0.000248
REMIT - - - - - - - - -
(0.003717)
0.002854
POLIT - - - - - - - - -
(0.002241)
-0.001158
FINCRS - - - - - - - -
(0.002285)
-0.000537B
TIME - - - - - - - - -
(0.000236)
Adj − R- 0.698836 0.698476 0.698665 0.697109 0.695584 0.704574 0.695309 0.700999 0.696211 0.712736
F-stat 21.46219 21.42718 19.74176 19.60397 19.47022 20.27829 19.44630 19.95118 19.52508 21.05571
DW-stat 2.196995 2.183929 2.218734 2.238149 2.190450 2.242863 2.197627 2.244211 2.178358 2.289720
Notes: A, B and C indicates 1%, 5% and 10% level of statistical significance
Downloaded by INSEAD At 05:53 20 May 2018 (PT)

You might also like