Description
This study aims to investigate the impact of IT loans on the performance of banks. Using panel data covering the period from 1998 to 2002, the results of the empirical analysis indicate that such IT loans have a significant negative influence as banks increase their loan ratios to the IT industry.
Banks and Bank Systems, Volume 5, Issue 2, 2010
108
Hsien-Chang Kuo (Taiwan), Lie-Huey Wang (Taiwan), Yi-Hsun Lai (Taiwan), Shuhui Yu (Taiwan),
Chinfu Wu (Taiwan)
Loan policy and bank performance: evidence from Taiwan
Abstract
This study aims to investigate the impact of IT loans on the performance of banks. Using panel data covering the
period from 1998 to 2002, the results of the empirical analysis indicate that such IT loans have a significant negative
influence as banks increase their loan ratios to the IT industry. The implications are that, during a macroeconomic
downturn, banks will suffer losses if they abruptly offer loans to the IT industry and do not handle such business
appropriately. Besides, this study also finds that there is a positive correlation between the consumer loan ratio and
bank performance, suggesting that it is beneficial for banks to exploit the consumer loans market. Conversely, we find
that the changes in the political and economic environment have a negative effect on the performance of banks during
the year 2000, which lies in the middle of the sample period.
Keywords: loan policy, bank performance, IT industry, IT loan, consumer loan.
JEL Classification: C23, G21, G32.
Introduction
©
The role of a bank in a financial market is that of a
financial intermediary, which makes use of loan and
deposit services to effectively channel the idle funds of
the general public into valuable production and other
investment projects. As indirect finance is still the
major funding channel for enterprises, banks through
their intermediation bridge the gap between the supply
and demand for funds, thereby reducing the costs of
exchange, and efficiently supervising the capital
utilization of the enterprises. In addition, banks play a
key role in the development of the overall economy
of the society as it creates profit for itself.
In recent years, we have observed that the financial
institutions in Taiwan have faced the double
predicament of a decline in credit business and an
increase in nonperforming loans (NPLs). For example,
in February 2001, the total loans of the Bank of
Taiwan, Taiwan Cooperative Bank, First Commercial
Bank, Hua Nan Commercial Bank, and Chang Hwa
Commercial Bank slid to a new 22 month-low. As for
the loan-deposit ratios in recent years, we find that the
loan-deposit ratio for all banks was 87.90% in 1997,
but was only 72.66% in 2002, indicating that the banks
have not been doing well in the loan market.
In the 1990s, the lifting of restrictions on the
establishment of new banks in Taiwan, along with
existing financial institutions being allowed to engage
in banking business, resulted in an increasingly large
number of banks and especially fierce competition in
Taiwan’s financial markets. However, in recent
years, as the investment environment has deterio-
rated, traditional industries have hollowed out and the
incomes of the general populace have decreased,
there has been a significant decline in the ability of
industry to purchase or upgrade mechanical equip-
ment and people have been less willing to consume.
©Hsien-Chang Kuo, Lie-Huey Wang, Yi-Hsun Lai, Shuhui Yu, Chinfu
Wu, 2010.
This has resulted in a fall in the demand for loans. In
addition to the influence of the external environment
upon the demand for loans, the government’s request
for risk controls has also indirectly frustrated the loan
decisions of banks that have higher debt ratios (Lown
and Peristiani, 1996; Berger and Udell, 1994). In
terms of the high NPLs ratios, the statistical data
clearly show that the banks in Taiwan are facing high
pressure in this regard. For instance, the NPLs ratio
for Taiwan’s financial institutions greatly increased
from 3% in 1995 to 6.84% in 2002, reaching a peak
of 8.78% in the first quarter of 2002. Using the
domestic banks (see Table 1) as an illustration, there
is no doubt that the domestic banks are characterized
by low profitability and high NPLs ratios.
From the perspective of industries, we find that
Taiwan’s industries are undergoing a transformation
from traditional industries to high-technology indust-
ries (IT industries). In the 1990s, the market value of
listed IT firms was only 2.7% of total listed firms; in
1995, it had increased to 13.4%; and in 1999, it was
54.2%. From 1994 to 1999, the average annual growth
rate of the IT industry was 19.03%, and the ratio of its
market value among manufacturing industries increa-
sed from 15.83% to 28.99%; moreover, it ranked fifth
among the world’s leading IT industries. This transfor-
mation was a showcase for the rapid transformation
and the superiority of Taiwan’s development in the IT
industry. Furthermore, the government in Taiwan has
actively allocated resources and industrial strategies
into its “Two Trillion, Twin Stars” project. On the one
hand, it is enhancing the existing superior industries
(including the semiconductor and TFTLCD
industries), and expects that the output production
value of these industries will reach a total value of two
trillion NT dollars, thereby making it the third largest
IT industry in the world; this is the so-called “Two
Trillion” project. On the other hand, it aims to continue
to build upon the superior research, development talent
and social environment that Taiwan has as a founda-
tion, in order to develop global star industries, such as
Banks and Bank Systems, Volume 5, Issue 2, 2010
109
its digital content industry and bio-technology indust-
ry, which is referred to as the “Twin Stars” project. As
the government and the general public set these plans
into action, and inject resources into the “Two Trillion,
Twin Stars” project, it is expected that the process of
transforming Taiwan’s industry will be accelerated.
Table 1. Financial data of domestic banks in Taiwan
(1998-2002)
Financial data of domestic banks in Taiwan from 1998 to 2002.
The data are taken from Statistics Office of the Bureau of
Monetary Affairs, Ministry of Finance, Financial Statistics and
Financial Statistics Index.
Item 1998 1999 2000 2001 2002
Annual total assets
(million NT)
175,479 192,601 207,751 217,408 220,970
Return on total assets 0.7 0.54 0.46 0.26 -0.47
Net income rate 9.57 8.01 7.27 4.5 -10.57
Nonperforming
loans ratio
4.37 4.88 5.34 7.48 6.12
Since the transformation of Taiwan’s industries
started to take place, the IT industry has gradually
replaced traditional industry as the primary driving
force of economic development. Through the
extension of loans to the IT industry, banks can
generate profits, sustain themselves, and allocate
further funds to the industry, which will enable the
industry to gain enough funds for investment and
will help develop the economy as a whole.
Nevertheless, since the industry started to be
transformed, the profitability of banks has decreased
as a whole (see Figure 1).
From Figure 1 we can see that during the industrial
transformation and the development of the IT
industry, the profitability and loan quality of the
banks continuously worsened. It may have been
caused by the special characteristics of the IT
industry. While the IT industry has experienced a
very rapid pace of change, a big requirement for
investment and a long return period compared to the
traditional industry, it has encountered much greater
risk than traditional industries. Meanwhile, many
firms in the new technology industries do not have
enough collateral, and therefore, banks have to
worry about NPLs and bad debts when lending to
the IT industry. The loss may lead to potential
bankruptcy. Therefore, the dilemma between profi-
tability and potential loss causes banks to hesitate in
their lending policies.
-200
-100
0
100
200
300
1995 1996 1997 1998 1999 2000 2001 2002
Year
LINE 1
LINE 2
LINE 3
Notes: The data are taken from Financial Statistics Monthly, Republic of China (Taiwan), published by the Economic Research
Department, Central Bank of the Republic of China (Taiwan), and the Taiwan Economic J ournal Data Bank. Line 1 is the domestic
banks’ earning deflator; line 2 is the domestic banks’ NPL ratio deflator; and line 3 is the IT firms’ market value to all listed firms’
market value deflator (the deflator is 100 in 1995).
Fig. 1. Domestic banks and IT industry development
When banks make the maximization of profitability
their target, loan policy takes into consideration the
competitiveness and resources of a bank to determine
how many resources it will allocate to its credit
departments, or how much funding each credit
branch (e.g., in terms of business loans, consumer
loans, etc.) can have. Under limited bank resources,
the quality of the bank’s loan policy determines the
ability of the bank to achieve its goal. This study does
not evaluate the quality of bank loan policy by
observing the history of its strategy. It rather analyzes
the profitability of different loan combinations, and
the kinds of borrower that can help the bank achieve
maximum profit, as it seeks to further discuss
whether a bank should reallocate its business
resources and change its loan policy to achieve or
accelerate the achievement of its goals. This study
analyzes the effects of loan policy toward the IT
industry on the banks’ operating performance, and
hopes to answer the question as to whether banks in
Taiwan should increase their loans to the IT industry
in order to maximize their profits.
This study seeks to clarify whether loans to the IT
industry can become the primary source of profit for
banks from the perspective of protecting the financial
Banks and Bank Systems, Volume 5, Issue 2, 2010
110
markets and promoting economic development. If the
results show that the increase in the flows of capital
between financial institutions and the IT industry will
boost the profitability of financial institutions, it
becomes clear that this kind of loan policy should be
implemented. The reinforcement of the loan
relationships between banks and the IT industry will
be significant for the stability of the financial market
and the development of the economy. On the
contrary, if the results show that increasing loans to
the IT industry will result in a loss, banks should
either be more actively seeking other sources of profit
or reevaluating and readjusting the existing structure
of interactions with industries in order to increase
profitability and reduce their bad debts.
In addition, this study also discusses the loan ratio in
relation to a bank’s assets: even though loan
services are the main sources of income of banks,
banks can still allocate their resources. Besides
loans, banks also handle bonds and securities
investments. In particular, since the rise of the bond
market, the bonds that banks possess have increased
significantly (from 1998 to 2002 that growth has
almost doubled). It, thus, needs to be asked whether
allocating assets to loan services would result in
profitability. The inclusion of the loans to total
assets ratio in this study helps us to understand
whether banks’ resources in terms of loan services
are beneficial for creating the banks’ profitability.
The remainder of this article is organized as follows.
Section 1 discusses previous studies, highlighting
those studies related to loan policies and the banks’
performance. Section 2 explains the variables that
this study adopts and introduces the methodology
and the model specifications. Section 3 presents the
analysis and discusses the empirical results. The
final section provides the conclusions.
1. Literature review
Since the loan spread is the basic source of profit for
banks, the banks’ loan policies are closely tied to
their operating performance. Many scholars in
discussing the optimal loan policies make the
maximization of the banks’ profitability their
primary goals.
Pringle’s (1974) assumption was based on the
borrowers and interest rates remaining unified, so that
banks could earn profit by controlling the amount of
loans and investing in government bonds, to set the
target function as in the case when banks pursued the
goal of maximizing private wealth. Pringle deduced
that the optimal loan policies (the amount of the
loans), would be affected by the spread of deposits
(minus the risk-free rate of interest), the spread of
loans, the amount of the deposits in the next period,
and the capital of the bank. The study emphasized
that the risk-free rate of interest did not influence the
bank’s decision regarding the amount of the loans.
Graddy and Kyle (1979) also started with the
equation on the balance sheet, and concluded that the
amount of loans, deposits, bank capital and labor cost
were factors that were significant to a bank’s
profitability. In addition, Molyneux et al. (1998),
under the assumption that managers of banks pursued
to maximize profitability, minimize risk, and
maximize expected utility, deduced that the two
major factors influencing a bank’s amount of loans
were the profitability of the loans and the risk faced
by the bank itself.
Besides making theoretical deductions, scholars
have also empirically discussed the influence of the
amount of loans on a bank’s operating performance.
Taking Molyneux et al. (1998) as an example, in
that paper they discussed how loan behavior affects
bank performance, utilizing the data for foreign
banks in the United States from the first quarter of
1990 to the third quarter of 1992. With the 2SLS
method, they found that with the return on total
assets (ROA) as the proxy variable for operating
performance, when the amounts of the loans
increase, the bank’s performance is significantly
positively affected. Graddy and Kyle (1979) made
use of the sample data for 463 commercial banks in
1974, and adopted OLS and 3SLS estimations to
find that, when banks raised the amount of the loans
in their assets, at the 0.01 significance level, doing
so would have a significantly positive influence on
the labor costs ratio (labor cost to total assets).
Kwan (2002) examined the banks in 7 Asian
countries (Hong Kong, Indonesia, South Korea,
Malaysia, Philippines, Singapore, and Thailand)
from 1992 to 1999, using regression analysis, and
empirically found that when banks increased the
amount of their loans in earning assets, at the 0.01
significance level, they would significantly and
positively influence the labor cost ratio (the labor
cost to earning assets). This conclusion is in
agreement with Graddy and Kyle (1979).
In financial markets, the conditions of borrowers are
not identical; therefore, in addition to the banks’
decisions as to the amounts of the loans, the targets of
the loans and their condition are also important
factors related to bank performance. By under-
standing the borrower, the risk premium can be
ascertained, and the profit erosion from bad debts can
be decreased. Basically borrowing can be categorized
as consisting of business loans and consumer loans.
With differences in time periods, regions, and
subjects of study, scholars categorize the amounts of
the loans, and then discuss the influence of different
types of loans on bank performance. For example,
Fraser et al. (1974) used American banks in 1969 and
1970 as the study sample, and applied canonical
Banks and Bank Systems, Volume 5, Issue 2, 2010
111
correlation analysis to find that the composition of
loans (the ratio of different types of loans including
real estate, agriculture, consumer and business
loans) would significantly affect the bank at the 0.01
significance level.
In addition, Arshadi and Lawrence (1987), based on
Graddy and Kyle’s (1979) research, divided the
factors that influenced bank performance into seven
types, including cost, structure, loan composition,
deposit composition, regional factors, economic
conditions and scale. The loan composition type
categorized the total amount of loans into real-estate
loans, agricultural loans, consumer loans and
business loans. Through the categorization, their
study empirically analyzed new banks in America
from 1980 to 1984 (the purpose of the study was to
discuss the influence of the factors on bank
performance after the new banks had entered the
market), using canonical correlation analysis. It was
found that the above-mentioned factors did exhibit a
significant correlation with operating performance.
Furthermore, different types of loan and bank
performance had different degrees of correlation. For
example, the study empirically found that agricultural
loans and business loans were significantly
negatively correlated with operating performance,
and the authors reached the conclusion that loan
composition directly influenced bank performance.
The meaning of a functional policy under a loan
policy is that, when the bank makes a loan policy
decision to allocate resources to different
departments, each department can, based on the
direction of the policy, design functional policies to
assist in the execution of the loan policy and attain
profitability. For instance, by observing the
borrowers, it may be possible to decrease risk and
loss associated with the loan. Glover (1939) once
categorized the factors that should be taken into
consideration in lending into nine items, including
expecting a future surplus, the financial statement of
the borrower, additional conditions, such as
collateral, the business management of the borrower,
the business operations of the borrower, the purpose
of the loan, the relationship with the borrower, the
scale of the borrower’s business, and the term of the
loan, and stated that the banks should reinforce their
understanding of these factors to increase executive
capability. In addition, Kuo (2001) further introduced
a support system to the loan policy, as a result of
which the data of the borrower (such as the financial,
managerial, and economic aspects) were put into the
system to scientifically increase the quality of the
loans and the profitability of banks.
In addition to the above-mentioned discussions of
loan policy and bank performance in terms of the
amounts of the loans and the types of loans, there
are other studies that regard NPLs as an explicit
factor of the severity of the loan policy, the reason
being that the amount of the loans is sometimes
determined by the external factors of the
environment, and not necessarily by the willingness
of the manager to take risks. However, when a bank
chooses a high-risk loan policy, it should reasonably
reflect on the higher amount of bad loans, and create
a positive influence on the performance of the bank.
This is referred to as the policy hypothesis by
scholars. Jordan (1998) took the new banks in
England from 1989 to 1992 as his sample to test the
hypothesis. The study used cost efficiency and profit
efficiency as proxy variables for bank performance.
Through examination, he found that the amount of
bad loans did have a significant positive correlation
with profitability, proving that the effect of the policy
hypothesis did exist. In other words, assuming that
the hypothesis is valid, when banks have high NPLs,
it means that they have easier loan policies, or are
more willing to engage in loan services. The above-
mentioned hypothesis indicates the existence of this
effect, but it is worth noting that Berger and Robert
(1997) once created three different hypotheses as to
the causes of bad loans, namely, bad luck, bad
management, and skimping and moral-hazard.
Skimping meant that, in the pursuit of high
performance, the resources on monitoring were
reduced, which resulted in a high NPLs ratio.
Although the phenomenon of a high NPLs ratio and a
high return did occur, its direction of influence was
directly opposite to that of the policy hypothesis; the
two hypotheses did not contradict each other.
Meanwhile, interest rates on loans can reflect the
preferences of a bank’s lending policy. Liew (1970)
thought that when interest rates for loans increased,
profitability would increase, and would cause a bank
to increase its preference for risks, thus resulting in
an easier loan policy (when interest rates for loans
increase, the debtor’s risks also increase, resulting in
a higher probability of default). Conversely, Lown
and Peristiani (1996) thought that when banks tried
to reduce loans, they would also increase interest
rates to increase the difficulties for the borrowers.
Nonetheless, in an empirical analysis of the
relationship between loans and interest rates,
Hannan (1991) found that whatever the amount of
the loan was, the interest rate and the amount of the
loan exhibited a significant negative correlation.
The empirical finding of Lown and Peristiani (1996)
exhibited significant positive correlations in small
banks, and insignificant negative correlations in
large banks, indicating that the two loan policies had
a mixed influence on the interest rate.
From the previous literature we may clearly see that
there are many different factors related to the
differences in banks’ lending policies. Different
Banks and Bank Systems, Volume 5, Issue 2, 2010
112
lending policies (severeness, composition) have
different effects on bank performance. However, the
previous literature rarely studies the influence of loan
policies on bank performance in a country that is
transforming its industries from traditional industries
to IT industries. When we find that Taiwan faces the
problem of low performance for banks and the
transformation of industry, we consider the loan
policies of banks towards the IT industry and their
relationship with performance. On the one hand, it
discusses the aspects that the previous studies have
rarely touched upon; on the other hand, it provides
appropriate suggestions for banks in the face of
today’s dilemma. We trust that this study will be
valuable and will contribute to related studies.
2. The method
2.1. Variables. 2.1.1. Bank performance variables.
The purpose of this study is to discuss the
relationship between loan policies and bank
performance. There are many factors involved in
measuring bank performance. The performance
referred to in this study is operating performance,
i.e., banks are evaluated from the perspective of
profitability. The important consideration for loan
policies concerns whether banks in Taiwan will
increase the ratio of loans to the IT industry, which
has higher risk. In terms of the explanatory variable,
the return on total assets (ROA) and the return on
equity (ROE) serve as the proxy variables for bank
performance (Arshadi and Lawrence, 1987; Bourke,
1989; Webster, 1997; J ordan, 1998; Molyneux et
al., 1998; J oyce, 2001; Sigler and Porterfield, 2001).
2.1.2. Loan policy variables. The loan policy
variables include the loan policies of banks toward
the IT industry, the consumer loan ratio, as well as
the total loans to total assets ratio. Through the design
of the above-mentioned variables, the analysis is
expected to indicate whether increasing the loan ratio
and amount for the IT industry, increasing the
consumer loan ratio, and utilizing assets in loan
services can effectively increase the bank’s
profitability, i.e., the analysis of the adjustment of the
loan policy following the transformation of industry.
(1) IT loan ratio
While discussing the relationship between
borrowers and bank performance, the purpose is to
understand the bank’s allocation of funds for loans.
When the borrowers are different, there will be
differences in terms of bank performance. In terms
of the loans, many previous studies used “the loans
for specific borrowers to total loans” as a proxy
variable for this decision (Fraser and Rose, 1971;
Fraser et al., 1974; Arshadi and Lawrence, 1987).
Some scholars have also elected to use “the loans
for specific borrowers to total assets” as the proxy
variable (Graddy and Kyle, 1979). The difference is
that the latter variable takes into consideration the
bank’s asset allocation policy and loan policy at the
same time. This study uses “the loans to the IT
industry to total loans” (IT loan ratio) as the proxy
variable. Its function is to discuss whether the IT
industry has a more significant influence than other
borrowers when banks select borrowers. Of course,
when banks do not have other ideal borrowers,
whether or not lending to the IT industry can result
in profitability is also an important issue.
(2) Consumer loan ratio
With the decreases in the traditional industry’s
demand for funds, banks can try to increase
profitability not only through increasing loan
services for the IT industry, but also by extending
the consumer loan market. Increasing the consumer
loan ratio is also one of the banks’ main loan
policies in recent years. The consumer loans have
continued to reach new highs in recent years, and
the consumer loan ratio has increased from 32.81%
in 1997 to 37.61% in 2001. It is mainly caused by
not only traditional personal loan services, but also
the proliferation of credit cards and cash cards.
Since Makto Bank first started to provide this
service, 30 financial institutions have begun to offer
cash card services. Compared to the high risk and
uncertainty of loans in the IT industry, whether
increasing amount of consumer loans and the
consumer loan ratio (consumer loans to total loans)
should be included in banks’ lending policies to
increase profitability or not will be empirically
analyzed in this study. In this study, in order to
analyze the influence of the newly-launched
consumer loan services on bank performance, the
consumer loans in this study will exclude mortgages
as well as car, housing and welfare loans for the
sake of accuracy in this study.
(3) Total loans to total assets ratio (Loan ratio)
To examine the allocation of bank assets, this study
also applies “total loans to total assets” (the loan
ratio) as the proxy variable for the asset allocation
policy. The meaning of this variable involves
measuring the allocation of bank assets, and it also
reflects the risk faced by the bank. When this ratio is
higher, it means that banks will be more enthusiastic
in offering loans to the general public instead of
letting them remain unused or using them to
purchase securities, which have much higher risk.
Under the basis of high risk and high returns,
increasing the total loans to total assets ratio should
have a positive influence on bank performance;
however, if the risk management is ineffective,
which causes the NPLs ratio to be too high and
erodes profitability, the result will be negative.
Molyneux et al. (1998) empirically found that the
Banks and Bank Systems, Volume 5, Issue 2, 2010
113
foreign banks in America were in a situation where
the loan ratio was significantly negatively correlated
with the ROA.
2.1.3. Bank’s structure variables. The bank’s
structural factors in this study include the debt ratio,
the bank scale, and the number of bank branches.
Their purpose is to analyze whether the high debt
characteristic of the bank industry and bank scale
influence the profitability of banks. In addition, in the
financial environment of Taiwan, where there are
large numbers of banks and bank branches, it needs
to be asked whether increasing the number of bank
branches has a negative influence on profitability, or
impacts the operation of banks because of high cost.
(1) Debt ratio
On the one hand, the debt ratio represents a bank’s
consideration of cost for its sources of funds, mea-
ning that when the bank uses more external capital
its profitability will be more limited. On the other
hand, when the debt ratio is too high, the bank will
have to monitor its risk and cannot engage in high-
risk high-return services (Lown and Peristiani,
1996; Berger and Udell, 1994). It also compresses
the profitability of a bank. Under this theory, the
debt ratio should be negatively correlated with bank
performance.
(2) Bank scale
When a bank has a larger scale of operations, it
means that it has more resources and services for its
consumers, and it can also decrease its costs through
scale economies; its size is, thus, beneficial for
profitability. However, under poor management,
when the bank cannot cover its huge costs, its
oversized scale of operations will adversely affect
the bank’s performance. In terms of empirical
studies, Miller and Noulas (1996) found that larger-
scale banks had lower operating performance,
showing that bank scales were negatively correlated
with their performance. Meanwhile, Zhang (1996)
categorized the scales of banks by dividing total
assets into large, medium and small, and found that
when the assets scale was too big or too small, it
would become non-efficient, indicating that only the
medium scale would be beneficial for bank
performance. This study uses total assets as the
proxy variable for bank scale.
(3) Number of branches
The number of branches, which can increase the
competitiveness of a bank, represents the bank’s
ability to provide service and convenience. Under
this theory, the number of bank branches will be
positively correlated with the expected profitability
of the bank. Arshadi and Lawrence’s (1987) study
also shows that there is a positive correlation. On
the other hand, there are a total of 3,068 branches of
financial institutions in Taiwan. With a limited
market, the overabundance of branches will render
the operation inefficient, result in unused resources,
and affect the profitability of the banks adversely.
Anderson et al. (1982) found that having too many
branches would negatively influence bank
performance. Therefore, this study includes the
number of bank branches as an explanatory variable.
2.2. Major events during the study period. During
the period from 1998 to 2002, the financial
environment of Taiwan underwent major changes.
In the middle of the year 2000, for the first time in
Taiwan’s political history the ruling party changed.
The original ruling party, the Kuomintang (KMT,
the Chinese Nationalist Party), lost the election, and
the Democratic Progressive Party (DPP), which had
never had the experience of ruling before, became the
ruling party. On the one hand, the previous ruling
party had launched many financial reforms and, on
the other hand, the inexperience of the new
government was reflected on its indecisive policies,
such as those related to the nuclear power plants and
the reform of the farmers’ and fishermen’s associa-
tions. Meanwhile, the lack of communication or a
relationship between the new government and China
also increased the domestic political crisis. Using a
study by Business Environment Risk Intelligence
(BERI) as an example, the analysis shows that the
investment environment in Taiwan in 2003 was at the
lowest risk level 1A and was ranked 4
th
globally,
while Taiwan’s political risk was ranked 11
th
globally
(the lower the rank, the less the risk), indicating that
the political environment of Taiwan adversely
affected the investment environment in Taiwan. In
addition, in 2000, after the ruling party had changed,
Taiwan’s political risk score decreased by 6, and its
rank declined to 16
th
globally. The political risk
indicator did not return to its former level until 2002.
In addition to political factors, during this time,
Taiwan’s economic environment also underwent
drastic change. From the fourth quarter of 2000
onwards, Taiwan’s GDP growth rate began to slide
down. In the second, third and fourth quarters of
2001, there was even negative growth, which had
not occurred in 20 years. Furthermore, the private
investment growth rate and the investment amount
also began to decrease in 2001, showing that the
economic environment had undergone significant
changes since the middle of the year 2000.
In particular, there was the bursting of the Internet
bubble in the year 2000. With this unexpected turn
of events in the development and profitability of the
electronic business market, investors in this market
encountered huge losses. Amazon.com, for example,
Banks and Bank Systems, Volume 5, Issue 2, 2010
114
saw its stock price fall from 113 USD in 1999 to 40
USD in 2000. A number of well-known internet
companies also suffered huge losses, which
impacted both the domestic and foreign economic
environment.
To sum up, after the middle of the year 2000, the
following questions are raised. Do the increases in
political risk, the changes in the domestic economic
and investment environment, and the effects of the
Internet bubble on the international economic
environment negatively influence the domestic
economy? Do those influences result in a loss of
profitability, or a loss of performance indirectly
through the bank’s business relationship with
various corporations?
This study discusses the impact of such influences on
bank performance as a result of these drastic changes.
Thus, we set a dummy variable, referred to as TIME,
before the middle of the year 2000 as TIME =0, and
after the middle of the year 2000 as TIME =1, in
order to analyze the impact of the change in the
ruling party on Taiwan’s financial sector.
The operating definitions of the variables in this
study will be shown in Table 2.
2.3. Statistical method and model specifications.
The sample in this study consists of banks covering
the period from 1998 to 2002, with half a year as a
period, or a total of 10 periods, including cross-
sectional and time series panel data. Since banks
differ from each other, we should not directly
adopt the ordinary least squares (OLS) method, but
through testing should select the most appropriate
model for the data based on the OLS model,
namely, the fixed effects model (FE model) and the
random effects model (RE model), to avoid
estimation bias. The distinctive feature of the FE
model is that, by using panel data, the intercept
term will change because of the different
characteristics of banks, but the error term of the
sample does not change with time, indicating that
the error term remains constant at different times,
i.e., during the observation period, the individual
sample is reflected by the intercept term of the
regression equation. Therefore, the FE model is
also referred to as the dummy variable method; at
this time, the intercept changes only with the
differences among banks, but not with time. The
RE model, by contrast, mainly assumes that the
individual differences in the sample are random,
and are independent of the independent variable. It,
therefore, shows that the differences in terms of the
individual samples will be reflected in the error
term in the regression equation. Hence, the RE
model can also be referred to as the error compo-
nent model.
Table 2. Operating definitions of variables
In this table we present the operating definitions of return on assets
(ROA), return on equity (ROE), high-technology industries loan
ratio (IT loan ratio, ITRATIO), consumer loan ratio (CONRATIO),
loan ratio (LORATIO), debt ratio (DRATIO), bank scale (SCALE),
number of branches (BRANCH), and time dummy (TIME).
Variable Operating definition
Return on assets (ROA)
(Net gain after tax before interest ÷ Total
assets) ×100%.
Return on equity (ROE)
(Net gain after tax before interest ÷ Total
equity) ×100%.
IT loan ratio (ITRATIO) (The loans for IT industry ÷ Total loans) ×100%.
Consumer loan ratio
(CONRATIO)
[(Consumer loans minus housing, car,
maintenance, and welfare loans) ÷ Total loans]
×100%.
Loan ratio (LORATIO) (Total loans ÷ Total assets) ×100%.
Debt ratio (DRATIO) (Total debt ÷Total assets) ×100%.
Bank scale (SCALE) Logarithm of total assets.
Number of branches
(BRANCH)
Logarithm of the number of branches.
Time factor (TIME)
Dummy variable: before the middle of year 2000,
TIME = 0; after the middle of year 2000, TIME = 1.
To compare which of the models would be more
appropriate for the data, we have to test each of them.
The comparison between the OLS model and the FE
model can be achieved by means of the F-test, the F-
test being a distribution that corresponds to F (N-1,
NT-N-K-1). The comparison between the OLS model
and the RE model can be achieved by means of the
Lagrange Multiplier (LM), where the LM is a chi-
square distribution with one degree of freedom. When
the FE model and the RE model are both superior to
the OLS model, we should select from the FE model
and the RE model, and use the Hausman test to test
whether the intercept term and the independent
variable are statistically correlated. If the intercept term
and independent variable are not statistically correla-
ted, we should adopt the RE model; if, alternatively,
the intercept term and independent variable are
statistically correlated, we should adopt the FE model.
Through the above-mentioned hypothesis tests, we
select the most appropriate model for the data type
from the OLS model, FE model and RE model, in
order to decrease the estimation bias. This will be
helpful to the accuracy of this study.
This study mainly discusses the effect of loan
policy, the bank’s structural factors and changes of
the politico-economical environment in 2000 on
bank performance. Based on the above-mentioned
variable specifications, we select the appropriate
variables for the analysis. This study constructs
three models for analysis.
The purpose of the first model (Model 1) is to
discuss the relationship between borrowers, resource
allocation and bank operating performance. Thus, in
Model 1, we select three variables as independent
variables: the IT loan ratio (ITRATIO), the consumer
loan ratio (CONRATIO), and the loan ratio
Banks and Bank Systems, Volume 5, Issue 2, 2010
115
(LORATIO). The dependent variables are denoted
by operating performance (OP) in equation (1).
When we use ROA as the proxy variable for bank
operating performance, it is shown in Model 1-1. In
addition, this study attempts to again treat ROE as
the proxy variable for bank performance, and places
it in Model 1-2 as the dependent variable for the
model to analyze and compare it with Model 1-1.
Through the specification of Model 1, we expect
that after the empirical analysis is completed, the
result can help banks to select borrowers and design
loan policies, in order to increase their profitability.
it it i it
CONRATIO ITRATIO OP
2 1 0
it it
LORATIO
3
. (1)
In Model 2, in addition to the same independent va-
riables as in Model 1, i.e., ITRATIO, CONRATIO,
and LORATIO, we further incorporate the bank’s
structural variables, including the debt ratio
(DRATIO), bank scale (SCALE), and the number of
branches (BRANCH), into the model to discuss the
influences that these variables have on the bank’s
profitability, and thereby, increase the explanatory
power. In Model 2, we use ROA (in Model 2-1) and
ROE (in Model 2-2) separately to represent the
factors, and to measure, discuss and compare the
influence of loan policy and structural factors on the
bank’s performance:
it it t i it
CONRATIO ITRATIO OP
2 1 0
it it it
SCALE DRATIO LORATIO
5 4 3
it it
BRANCH
6
. (2)
The purpose of Model 3 is to discuss whether the
bank’s performance exhibits significant change
before and after the middle of the year 2000, and
whether the influence the borrowers have on the
bank’s operating performance has changed
significantly after this time period. Thus, in Model 3,
in specifying the independent variables, we include
the three variables, the same as in Model 1; we also
include the middle of the year 2000 as the time
factor. The model with ROA as the dependent
variable is Model 3-1, and the model with ROE as the
dependent variable is Model 3-2. Through the
specification of the model, we can analyze the
influence that the time factor has on the bank’s
operating performance.
it it i it
CONRATIO ITRATIO OP
2 1 0
it it it
TIME LORATIO
4 3
. (3)
3. Results
3.1. Data source and descriptive statistics. Our
data are collected from the “Financial Report
Database” in the Taiwan Economic Journal (TEJ ),
and a part of the undisclosed data is deduced from
the estimation
1
. The sample consists of all domestic
financial institutions and small to medium
commercial banks listed on the Taiwan Stock
Exchange from 1998 to 2002. Banks with missing
data are excluded from the sample. As a result, a total
of 27 listed banks (270 semiannual observations) are
included in this study. The descriptive statistics of the
variables are presented in Table 3.
Table 3. Descriptive statistics of variables
The sample consists of 270 observations on Taiwanese listed
banks from1998-2002. Data are taken fromthe “Financial Report
Database” in the Taiwan Economic Journal (TEJ). ROA is the
ratio of net gain after tax before interest to total assets. ROE is the
ratio of net gain after tax before interest to total equity. ITRATIO
is the ratio of the loans for IT industry to total loans. CONRATIO
is the ratio of consumer loans minus housing, car, maintenance,
and welfare loans to total loans. LORATIO is the ratio of total
loans to total assets. DRATIO is the ratio of total debt to total
assets. SCALE is the logarithmof total assets. BRANCH is the
logarithmof the number of branches.
Variables N Minimum Maximum Average Std. deviation
ROA 270 -2.44 1.45 0.19 0.68
ROE 270 -57.59 15.64 1.72 10.53
ITRATIO 270 0.00 52.59 3.80 11.06
CONRATIO 270 0.00 40.89 50.18 4.63
LORATIO 270 41.51 78.55 65.84 6.22
DRATIO 270 88.81 97.28 92.29 1.64
SCALE 270 4.84 7.18 5.81 0.70
BRANCH 270 3.18 5.16 3.98 0.54
From Table 3, we may find that, among the observa-
tions, banks have huge differences in profitability.
Taking ROA as an example, the maximum is 1.45%,
and the minimum is -2.44%. We find that banking is
not an “easy to operate” industry and slight mistakes
can result in huge losses.
Moreover, with regard to the IT loan ratio and the
consumer loan ratio, we may find that the loan
policies differ significantly among banks. From the
significant differences in the banks, we may find
that domestic banks do not have a general consensus
on the loan policy for optimal profitability; this is
also why this study is valuable.
In addition, from the loan ratio, we may learn that
all of the samples treat loaning as the main business,
and the differences are slight. In terms of the assets
scale, the maximum is ten times the minimum,
showing that there are still banks with very small
scales of operations in Taiwan. Finally, in terms of
the number of branches, we find that the minimum
is 24 while the maximum is 175, and the standard
deviation is as high as 38.42, showing that the banks
1
Since the banks did not disclose accurate data on the IT nonperforming
loans ratio, we calculate the corporate borrowers disclosed by the banks
as the sample used in this study, and the IT loan amount is estimated by
the loan ratio multiplied by the amount of loans for the corporations
which the banks have disclosed.
Banks and Bank Systems, Volume 5, Issue 2, 2010
116
have significantly different opinions as to how many
branches should be established
1
.
To avoid the problem of collinearity, we use the
variance inflationary factor (VIF) to test the collinear-
rity of the variables. Marquardt (1970) thought that
when the VIF was greater than 10, the model would be
seriously affected by collinearity. Through the ana-
lysis, the variables of the above-mentioned models all
have VIF values under 10. Thus, we determine that
there is no collinearity that will affect the empirical re-
sults in the model. The analysis is presented in Table 4.
3.2. Empirical results. 3.2.1. The effects of lending
policy on bank performance. From Table 5, we are
able to find that for Model 1 (including Model 1-1 and
Model 1-2), both the F-test and the LM test indicate
that the panel data model performs better than the OLS
model. The Hausman test also suggests that the FE
model is more suitable for use than the RE model.
Table 4. The VIF of the independent variables
We use the variance inflationary factor (VIF) to test the
collinearity of the variables. ROA is the ratio of net gain after tax
before interest to total assets. ROE is the ratio of net gain after tax
before interest to total equity. ITRATIO is the ratio of the loans for
IT industry to total loans. CONRATIO is the ratio of consumer
loans minus housing, car, maintenance, and welfare loans to total
loans. LORATIO is the ratio of total loans to total assets. DRATIO
is the ratio of total debt to total assets. SCALE is the logarithmof
total assets. BRANCH is the logarithmof the number of branches.
TIME dummy equals one if after the middle of year 2000 and
zero otherwise. The dependent variable of models 1, 2, 3 is ROA.
The independent variables of Model 1 are ITRATIO, CONRATIO,
and LORATIO. The independent variables of Model 2 are
ITRATIO, CONRATIO, LORATIO, DRATIO, SCALE, and
BRANCH. The independent variables of Model 3 are ITRATIO,
CONRATIO, LORATIO, and TIME.
Variables Model 1 Model 2 Model 3
ITRATIO 1.034 1.324 1.142
CONRATIO 1.016 1.102 1.026
LORATIO 1.035 1.357 1.036
DRATIO 1.603
SCALE 3.434
BRANCH 3.406
TIME 1.119
Table 5. The effects of loan policy on bank
performance
The dependent variable is either ROA (Model 1-1) or ROE
(Model 1-2). ROA is the ratio of net gain after tax before
interest to total assets. ROE is the ratio of net gain after tax
before interest to total equity. ITRATIO is the ratio of the loans
for IT industry to total loans. CONRATIO is the ratio of
consumer loans minus housing, car, maintenance, and welfare
loans to total loans. LORATIO is the ratio of total loans to total
assets. We use F-statistic test (F-test), Lagrange Multiplier test
(FM-test) and Hausman test to examine whether OLS model,
fixed effect model, or random effect model are appropriate
1
In the analysis of descriptive statistics, both total assets and the
number of branches are original values, the maximumamount of total
assets is 13151.8 billion, and the minimumis 126.40 billion.
ones. ***, **, and * indicate statistical significance at the 1%,
5%, and 10% levels, respectively.
Variables
Model 1-1
ROA
Model 1-2
ROE
ITRATIO -0.02714*** -0.43191***
CONRATIO 0.028952** 0.470653**
LORATIO 0.027503** 0.401161**
F-test 4.32*** 3.47***
LM-test 24.70*** 11.73***
Hausman test 32.27*** 32.16***
Adj R-squared 0.2368 0.1865
Through Model 1-1 and Model 1-2, we can arrive at
the following empirical results: when banks raise
their ratios of loans to the IT industry, this will give
rise to a significant negative influence on the banks’
performance. There are two possible reasons for
this. When banks increase their loans to the IT
industry, even though this action may bring interest
income from the spread between deposits and loans,
banks still have to input more resources in order to
enter this industry and avoid the decreases in the
quality of the loans because the IT industry is highly
risky and it is difficult for it to provide collateral.
Besides, during the 1998-2002 period, the banks’
NPLs increase but profitability decreases; this might
be due to the the fact the bank cannot ascertain the
risk and select a good borrower; consequently, this
gives banks higher NPLs, erodes the profitability
and creates this negative correlation.
In addition, in terms of the consumer loan ratio,
although this proves that the trend in the banks’ loan
policies is correct, what is particularly worth noting is
that the market is expanding and maturing, and the
competition will be more fierce. Newcomers will
divide the existing market and banks will have to pay
more to maintain the share, such as in terms of the
marketing cost and R&D cost. The fierce competition
will also force banks to decrease interest rates and
fees, and even expand the market to risky borrowers,
such as students. In these circumstances, whether or
not increasing the consumer loan ratio will be
beneficial to the bank’s performance is certainly
worth looking into.
In terms of the allocation of bank assets, we find that
the loan ratio is significantly positively correlated with
the banks’ performance, and this conclusion is the
same as that of Molyneux et al. (1998), indicating that
banks’ loan services have higher profitability than
other investments. One of the possible reasons may be
that loan service is the main service of banks, and both
talent and experience are comprehensive; thus, banks
may rely on loans for profit. Another possible reason is
that the securities market in Taiwan has yet to mature;
the targets of investment are limited and the risks are
high, thus affecting the profitability of banks.
3.2.2. The effects of bank structural variables on
bank performance. From Table 6, we may find that
Banks and Bank Systems, Volume 5, Issue 2, 2010
117
in Model 2 (including Model 2-1 and Model 2-2),
both the F-test and LM test indicate that the panel
data model performs better than the OLS model.
The Hausman test also suggests that the FE model is
more suitable to use than the RE model.
From the influences that loan policy has on bank
performance, we find that in Model 2-1, the IT
industry loan ratio and consumer loan ratio are
significant at the 1% and 5% significance levels,
respectively. In Model 2-2, the IT industry loan ratio
and consumer loan ratio are significant at the 1% and
10% significance levels, respectively. This indicates
that when banks increase their ratios of loans to the
IT industry, the lending policy will significantly
negatively influence bank performance; when banks
increase their consumer loan ratios, their lending
policy will significantly and positively benefit bank
performance. This conclusion is the same as in the
case of Model 1. During the period of 1998-2002, the
annual business loan interest rate was 4%~6%, which
is lower than the annual consumer loan interest rate
(the annual credit card interest rate was 18%~20%,
and annual cash card interest rate was 17%~18%).
The NPLs ratio was increasing from 4.36% to 8.55%,
and majority NPLs were business bad debts. The
asymmetric return-risk distributions between business
and consumer loans made the IT industry loan
negatively related to bank performance while the
consumer loan positively related to bank
performance. These encouraged Taiwanese banks to
granting credit to consumers. The Taiwanese banks
faced the cash-credit card crisis in 2005.
Table 6. The effects of bank structure variables on
bank performance
The dependent variable is either ROA (Model 2-1) or ROE (Model
2-2). ROA is the ratio of net gain after tax before interest to total
assets. ROE is the ratio of net gain after tax before interest to total
equity. ITRATIO is the ratio of the loans for IT industry to total
loans. CONRATIO is the ratio of consumer loans minus housing,
car, maintenance, and welfare loans to total loans. LORATIO is the
ratio of total loans to total assets. DRATIO is the ratio of total debt
to total assets. SCALE is the logarithmof total assets. BRANCH is
the logarithmof the number of branches. We use F-statistic test (F-
test), Lagrange Multiplier test (FM-test) and Hausman test to
examine whether OLS model, fixed effect model, or randomeffect
model are appropriate ones. ***, **, and * indicate statistical
significance at the 1%, 5%, and 10% levels, respectively.
Variables
Model 2-1
ROA
Model 2-2
ROE
ITRATIO -0.0189032*** -0.3154417***
CONRATIO 0.0237138** 0.3780544*
LORATIO 0.0058228 0.0845423
DRATIO -0.4337637*** -6.454996***
SCALE 1.978064*** 28.77167***
BRANCH -1.423505*** -19.89049***
F-test 6.05*** 4.80***
LM-test 23.47*** 11.34***
Hausman test 134.38*** 99.94***
Adj R-squared 0.5792 0.4951
In terms of the allocation of bank assets, we find
that the loan ratio does not have a significant
influence on bank performance in Model 2; this
conclusion is different from the positive results of
Model 1. This shows that with the addition of the
bank’s structural variables, the effect of the loan
ratio on performance disappears, indicating that
other variables compared to the loan ratio are more
significant in terms of their effect on bank
performance. Therefore, in Model 2, other variables
account for the explanatory power of the loan ratio,
and the loan ratio does not exhibit a significant
result in Model 2. As a result, this study suggests
that increasing the loan ratio does not have a
significant positive influence on bank performance.
This study also suggests that the bank has to
reinforce its own crediting capability, so that it will
be able to maintain a certain level of profitability
under the situation where there is no good
investment opportunity available.
Furthermore, the empirical results show that bank
scale has a significant and positive influence on
operating performance, indicating that when a bank
expands its scale and integrates resources, talents
and service, it may increase its competitiveness and
profitability. Recently, Taiwanese banks have
expanded their scales of operations through mer-
gers. If merged successfully, the competitiveness of
domestic banks will increase due to internationali-
zation. From another perspective, when resources
merge and increase, how to allocate resources
effectively and reduce unnecessary costs will
become important issues.
We also learn that the number of branches is still
negatively correlated with bank performance,
matching the empirical results of Anderson et al.
(1982) from Model 2. This indicates that there are
generally too many branches in Taiwan, and it causes
inefficiency in operation, leaving too many unused
resources and, thereby, adversely affecting profita-
bility. In fact, in recent years, the establishment of
new branches has slowed down. Banks may have
also faced problems with having too many branches.
Conversely, in order to maintain their services and
cut costs, banks have gradually replaced normal
branches with simplified branches to improve the
banks’ financial situation by decreasing costs.
In terms of the empirical results of other variables,
we find that the debt ratio is significantly negatively
correlated with bank performance; this is the same
as in the previous literature. When the debt ratio is
too high, banks will have to monitor risk and will
not be able to engage in high-risk high-profit
businesses (Lown and Peristiani, 1996; Berger and
Udell, 1994); this also compresses the profitability
of banks. The above-mentioned considerations of
Banks and Bank Systems, Volume 5, Issue 2, 2010
118
cost lead to banks using external capital and facing
high risk, as a result of which they have to offer
higher interest rates on deposits. It may also be one
of the reasons why the debt ratio and bank
performance are negatively correlated.
3.2.3. The effects of the time factor on bank perfor-
mance. During the period from 1998 to 2002, the
financial environment of Taiwan was dramatically
changed. In 2000, the Kuomintang Party lost the
election, and the new ruling party, Democratic
Progressive Party, launched many financial reforms.
In addition, the domestic political risk adversely
affected the investment and economic environments
in Taiwan. Furthermore, many electronic companies
suffered huge losses from the Internet bubble. In
this study, we further to discuss the impact of these
changes on bank performance.
From Table 7, we may see that in Model 3
(including Model 3-1 and Model 3-2), both the F-
test and LM test indicate that the panel data model
performs better than the OLS model. The Hausman
test also suggests that the FE model is more suitable
for use than the RE model.
Table 7. The effects of the time factor on bank
performance
The dependent variable is either ROA (Model 3-1) or ROE
(Model 3-2). ROA is the ratio of net gain after tax before
interest to total assets. ROE is the ratio of net gain after tax
before interest to total equity. ITRATIO is the ratio of the loans
for IT industry to total loans. CONRATIO is the ratio of
consumer loans minus housing, car, maintenance, and welfare
loans to total loans. LORATIO is the ratio of total loans to total
assets. TIME dummy equals one if after the middle of year 2000
and zero otherwise. We use F-statistic test (F-test), Lagrange
Multiplier test (FM-test) and Hausman test to examine whether
OLS model, fixed effect model, or random effect model are
appropriate ones. ***, **, and * indicate statistical significance
at the 1%, 5%, and 10% levels, respectively.
Variables
Model 3-1
ROA
Model 3-2
ROE
ITRATIO -0.0115472* -0.1880818*
CONRATIO 0.0400379*** 0.6440054***
LORATIO 0.019804* 0.2807685
TIME -0.3392419*** -5.304663***
F-test 3.70*** 2.88***
LM-test 26.83*** 12.00***
Hausman test 18.73*** 18.30***
Adj R-squared 0.2787 0.2289
Through the empirical results of Model 3, we may
learn that since the middle of the year 2000, the
domestic and international political and economic
environment has had a significant negative influence
on bank performance. The reasons for this influence,
in terms of the political environment, may be twofold.
The first reason is that the new government was
inexperienced and had trouble with China, which in
turn gave rise to political risk and influenced the
willingness of international investors to invest in
Taiwan, thereby, resulting in a poor economic envi-
ronment. Thus, banks cannot get sufficient earnings
from either loans or investments, resulting in a signi-
ficant negative influence. The second reason is that the
new government has passed many new amendments
and new financial laws. The financial institutions
when faced with such new laws must incorporate the
changes into their organizational structures and
include new services, or else simplify their organiza-
tions after paying high costs, which may have a
negative influence on bank performance. In terms of
economics, since the middle of the year 2000, the
domestic economic environment has continued to
deteriorate, and private investors have not had the
willingness to invest. Moreover, the Internet bubble
has also impacted the world economy, resulting in
significant negative influences on the operating
performance of the domestic banking industry.
3.3. Discussions. While analyzing the influences of
lending policy on operating performance, we find that
when banks increase their IT industry loan ratios, this
will result in undesired operating performance. When
we review the background of the period under study,
we find that from 1998 to 2002 bank performance
continued to decrease, while the IT industry continued
to develop, and so the motive was generated. As the
IT industry continues to develop, the demand for
funds increases, meaning that banks may increase
their IT loan ratios; in addition, as bank performance
continues to decline, it needs to be asked whether it is
influenced by IT loans. After the analysis of this
study, we find that with the increases in loans from
the banks to the IT industry, IT loans do give rise to a
negative influence on bank performance.
Furthermore, banks earn profit through loan services
by collecting loan interest. If loan services are not
profitable, such performance will directly reflect
upon the banks’ loan interest income rate (interest
income divided by the loan amount); if the loan
interest income rate is high, it means that each
amount that banks lend out can achieve maximum
interest income. Thus, this study further analyzes
the correlation between the IT loan ratio and loan
interest income rate in order to re-examine the
above-mentioned conclusions.
By means of Pearson correlation coefficient test, we
may find that the correlation coefficient of the IT loan
ratio and loan interest income rate is -0.387, and it is
statistically significant at the 0.01 significance level
(see Panel A of Table 8), indicating that the higher
the ratio of loans to the IT industry is, the lower the
profitability will be
1
. During this period, Taiwanese
1
Banks with missing data are excluded fromthe sample. As a result, 22
domestic banks listed on the Taiwan Stock Exchange from1998 to 2002
(220 semi-annual observations) are included in this study.
Banks and Bank Systems, Volume 5, Issue 2, 2010
119
government pushed the “Two Trillion, Twin Stars”
project and arranged with domestic banks about
granting low interest rate loans to IT industry.
Therefore, the IT loans were the low return and high
risk assets for banks.
In addition, this study finds that the consumer loan
ratio is positively correlated with the banks’ operating
performance. Therefore, we use the same data to
examine the Pearson correlation coefficient for the
consumer loan ratio and the loan interest income rate.
The result shows that Pearson correlation coefficient
between the two is 0.256, and it is statistically
significant at the 1% significance level (see Panel B
of Table 8). This also proves that the higher the loan
ratio is to the general populace, the higher the
profitability will be.
To sum up, both the results of the panel data reg-
ression analysis and Pearson correlation coefficient
test show that increasing the banks’ ratio of loans to
the IT industry will negatively impact the banks’
operating performance. In addition, the consumer
loan ratio is positively related to the banks’ operating
performance. This shows that the empirical results of
this study are robust.
Table 8. The results of the Pearson correlation
coefficient test
We use a Pearson correlation coefficient test to examine the corre-
lation between the IT loan ratio and loan interest income rate. Panel
A reports the Pearson correlation coefficient test of IT loan ratio
and loan interest income rate. Panel B reports the Pearson correla-
tion coefficient test of consumer loan ratio and loan interest
income rate. *** indicates statistical significance at the 1% level.
Panel A. IT loan ratio & loan interest income rate
IT loan ratio Loan interest income rate
IT loan ratio 1*** -0.387***
Loan interest income rate -0.387*** 1***
Panel B. Consumer loan ratio & loan interest income rate
Consumer loan ratio Loan interest income rate
Consumer loan ratio 1*** 0.256***
Loan interest income rate 0.256*** 1***
Conclusions
The primary purpose of this study is to discuss bank
loan policy and its relationship with performance.
Hopefully, we are able to understand through our
analysis whether there are adjustments to the loan
ratio and the loan amount to the IT industry, and the
consumer loan ratio. This study can help banks to
improve their profitability under an undesirable
performance, or actually result in more losses.
Meanwhile, we empirically prove the influence of
different factors, such as a bank’s debt ratio and the
number of branches in relation to bank performance.
The sample consists of 27 domestic banks listed on
the Taiwan Stock Exchange from 1998 to 2002. We
find that in terms of the influence of bank loan policy
on bank performance, when banks increase their loan
ratios to the IT industry, the action is negatively
correlated with bank performance, meaning that
when banks are faced with the compression of loan
business and the transformation of traditional industry
into the IT industry, increasing the loan amount to the
IT industry does not only result in profit, but will be
adversely affected by the complexity and high risk of
the industry, and will result in losses, thus reducing
the banks’ operating performance. If the
transformation of industry is inevitable, banks must
first improve their loan quality, and offer loans to
establish the IT industry with better credentials, avoid
loan losses, and increase profitability. Furthermore,
banks may, through increasing service quality,
increase their own profit bases, and by enhancing
their negotiating capabilities and raising their loan
interest rates, reflect the costs to the banks in order to
sustain themselves. By contrast, the empirical results
demonstrate that the consumer loan ratio has a
significant and positive influence on bank perfor-
mance, indicating that with the growth of the
consumer loan market, banks can input funds in the
market as an effective use of an ideal loan policy. Yet
banks should be aware that once the consumer
market has reached its limits, the pursuit of new
clients may result in a lower standard of borrowers,
and in increased personal bad debts. The increases in
bad loans will erode the banks’ profitability.
Then, in terms of the results of analyzing the
influence of a bank’s structural variables on bank
performance, we find that both the debt ratio and the
number of branches are significantly negatively
correlated with bank performance. This conclusion
agrees with the empirical results of Berger and Udell
(1994) and Anderson et al. (1982), respectively.
Based on this conclusion, banks must decrease their
external capital and utilize more of their internal
capital to lower risk, increase the expansion of
services and positively influence performance. In
addition, banks should decrease the number of their
branches as they increase operating costs, for an
appropriate reduction or the restructuring of branches
into simplified branches will positively increase bank
performance. The empirical results show that bank
scale has a significant positive influence on bank
performance, meaning that the scales of the domestic
banks tend to be too small and the banks, hence, do
not benefit from economies of scale. Through
partially merging, resources and talents can be
integrated into better competitiveness for domestic
banks. In the meantime, this study finds that, since
the middle of the year 2000, banks’ performances
have become more undesirable, indicating that the
political and economic environment since the middle
of the year 2000 has worsened and has impacted the
survival and development of banks. If banks expect
to perform better, increasing their operating and
managerial capability will become a primary task.
Banks and Bank Systems, Volume 5, Issue 2, 2010
120
There are two limitations to this study. First, banks
did not fully disclose their IT industry loan ratios
and loan amounts; thus, this study has to calculate
these figures from the names of borrowers the banks
have disclosed to obtain the data. Second, the lack
of observations could be improved in the future; we
expect that when the data become more
comprehensive, the future related studies will be
able to include more variables to measure bank loan
policy and to contribute to this issue.
References
1. Anderson, R., Handoff, W., McCarthy, M. (1982). Bank Size and the Management of Capital Ratios. The Bankers
Management 165 (1), 64-69.
2. Arshadi, N., Lawrence, C.E. (1987). An Empirical Investigation of New Bank Performance. Journal of Banking
and Finance 11 (1), 33-48.
3. Berger, A.N., Robert., D. (1997). Problem Loans and Cost Efficiency in Commercial Banks. Journal of Banking
and Finance 21 (6), 849-870.
4. Berger, A., Udell, G. (1994). Did Risk-Based Capital Allocate Bank Credit and Cause a ‘Credit Crunch’ in the
U.S. Journal of Money, Credit & Banking26 (3), 585-628.
5. Bourke, P. (1989) Concentration and Other Determinants of Bank Profitability in Europe, North America and
Australia. Journal of Banking and Finance 13 (1), 65-79.
6. Fraser, D.R., Phillips, J .W., Rose, P.S. (1974). A Canonical Analysis of Bank Performance. Journal of Financial
and Quantitative Analysis 9 (2), 287-295.
7. Fraser, D.R., Rose, P.S. (1971). More on Banking Structure and Performance: The Evidence from Texas. Journal
of Financial and Quantitative Analysis 6 (1), 601-611.
8. Glover, J .D. (1939). Industrial Loan Policy of the RFC. Harvard Business Review 17 (4), 465-476.
9. Graddy, D.B., Kyle, R. (1979). The Simultaneity of Bank Decision-making, Market Structure, and Bank
Performance. Journal of Finance 34 (1), 1-18.
10. Hannan, T.H. (1991). Bank Commercial Loan Markets and the Role of Market Structure: Evidence from Surveys
of Commercial Lending. Journal of Banking and Finance 15 (1), 133-149.
11. J oyce, W.B. (2001). Return and Reward: Bank Performance and CEO Compensation. American Business Review
19 (2), 93-99.
12. J ordan, J .S. (1998). Problem Loans at New England Banks, 1989 to 1992: Evidence of Aggressive Loan Policies.
New England Economic Review (J an/Feb), 23-38.
13. Kuo, H.C. (2001). Strategic Decisions for the Management of Banks’ Credit Granting System. International
Journal of Services Technology and Management 2 (3/4), 313-325.
14. Liew, C.K. (1970). Lending Behavior and the Nature of Default Loans in the Case of the Savings and Loan
Industry. Journal of Finance 25 (3), 701-702.
15. Lown, C., Peristiani, S. (1996). The Behavior of Consumer Loan Rates During the 1990 Credit Slowdown.
Journal of Banking and Finance 20 (10), 1673-1694.
16. Marquardt, D.W. (1970), Generalized Inverses, Ridge Regression, Biased Linear estimated, and Nonlinear
Estimation. Technometrics 12 (4), 591-612.
17. Miller, S.M., Noulas, A.G. (1996). The Technical Efficiency of Large Bank Production. Journal of Banking and
Finance 20 (3), 495-309.
18. Molyneux, P., Remolona, E., Seth, R. (1998). Modeling Foreign Bank Performance and Lending Behavior.
Financial Markets, Institutions and Instruments 7 (4), 26-41.
19. Pringle, J .J . (1974). The Capital Decision in Commercial Banks. Journal of Finance 29 (3), 779-795.
20. Sigler, K.J., Porterfield, R. (2001). CEO Compensation: Its Link to Bank Performance. American Business Review
19 (2), 110-114.
21. Kwan, S.H. (2002). Operating Performance of Banks among Asian Economics: An International and Time Series
Comparison. Federal Reserve Bank of San Francisco, Working Paper.
22. Webster, A.L. (1997). The Impact of Technological Change on Bank Performance. Journal of Economics and
Finance 21 (3), 41-47.
23. Zhang, J . (1996). Measurement and Determinants of X-efficiency, Technological Change and Profitability in
Banking. Unpublished Ph.D. Dissertation, The Temple University.
doc_915213685.pdf
This study aims to investigate the impact of IT loans on the performance of banks. Using panel data covering the period from 1998 to 2002, the results of the empirical analysis indicate that such IT loans have a significant negative influence as banks increase their loan ratios to the IT industry.
Banks and Bank Systems, Volume 5, Issue 2, 2010
108
Hsien-Chang Kuo (Taiwan), Lie-Huey Wang (Taiwan), Yi-Hsun Lai (Taiwan), Shuhui Yu (Taiwan),
Chinfu Wu (Taiwan)
Loan policy and bank performance: evidence from Taiwan
Abstract
This study aims to investigate the impact of IT loans on the performance of banks. Using panel data covering the
period from 1998 to 2002, the results of the empirical analysis indicate that such IT loans have a significant negative
influence as banks increase their loan ratios to the IT industry. The implications are that, during a macroeconomic
downturn, banks will suffer losses if they abruptly offer loans to the IT industry and do not handle such business
appropriately. Besides, this study also finds that there is a positive correlation between the consumer loan ratio and
bank performance, suggesting that it is beneficial for banks to exploit the consumer loans market. Conversely, we find
that the changes in the political and economic environment have a negative effect on the performance of banks during
the year 2000, which lies in the middle of the sample period.
Keywords: loan policy, bank performance, IT industry, IT loan, consumer loan.
JEL Classification: C23, G21, G32.
Introduction
©
The role of a bank in a financial market is that of a
financial intermediary, which makes use of loan and
deposit services to effectively channel the idle funds of
the general public into valuable production and other
investment projects. As indirect finance is still the
major funding channel for enterprises, banks through
their intermediation bridge the gap between the supply
and demand for funds, thereby reducing the costs of
exchange, and efficiently supervising the capital
utilization of the enterprises. In addition, banks play a
key role in the development of the overall economy
of the society as it creates profit for itself.
In recent years, we have observed that the financial
institutions in Taiwan have faced the double
predicament of a decline in credit business and an
increase in nonperforming loans (NPLs). For example,
in February 2001, the total loans of the Bank of
Taiwan, Taiwan Cooperative Bank, First Commercial
Bank, Hua Nan Commercial Bank, and Chang Hwa
Commercial Bank slid to a new 22 month-low. As for
the loan-deposit ratios in recent years, we find that the
loan-deposit ratio for all banks was 87.90% in 1997,
but was only 72.66% in 2002, indicating that the banks
have not been doing well in the loan market.
In the 1990s, the lifting of restrictions on the
establishment of new banks in Taiwan, along with
existing financial institutions being allowed to engage
in banking business, resulted in an increasingly large
number of banks and especially fierce competition in
Taiwan’s financial markets. However, in recent
years, as the investment environment has deterio-
rated, traditional industries have hollowed out and the
incomes of the general populace have decreased,
there has been a significant decline in the ability of
industry to purchase or upgrade mechanical equip-
ment and people have been less willing to consume.
©Hsien-Chang Kuo, Lie-Huey Wang, Yi-Hsun Lai, Shuhui Yu, Chinfu
Wu, 2010.
This has resulted in a fall in the demand for loans. In
addition to the influence of the external environment
upon the demand for loans, the government’s request
for risk controls has also indirectly frustrated the loan
decisions of banks that have higher debt ratios (Lown
and Peristiani, 1996; Berger and Udell, 1994). In
terms of the high NPLs ratios, the statistical data
clearly show that the banks in Taiwan are facing high
pressure in this regard. For instance, the NPLs ratio
for Taiwan’s financial institutions greatly increased
from 3% in 1995 to 6.84% in 2002, reaching a peak
of 8.78% in the first quarter of 2002. Using the
domestic banks (see Table 1) as an illustration, there
is no doubt that the domestic banks are characterized
by low profitability and high NPLs ratios.
From the perspective of industries, we find that
Taiwan’s industries are undergoing a transformation
from traditional industries to high-technology indust-
ries (IT industries). In the 1990s, the market value of
listed IT firms was only 2.7% of total listed firms; in
1995, it had increased to 13.4%; and in 1999, it was
54.2%. From 1994 to 1999, the average annual growth
rate of the IT industry was 19.03%, and the ratio of its
market value among manufacturing industries increa-
sed from 15.83% to 28.99%; moreover, it ranked fifth
among the world’s leading IT industries. This transfor-
mation was a showcase for the rapid transformation
and the superiority of Taiwan’s development in the IT
industry. Furthermore, the government in Taiwan has
actively allocated resources and industrial strategies
into its “Two Trillion, Twin Stars” project. On the one
hand, it is enhancing the existing superior industries
(including the semiconductor and TFTLCD
industries), and expects that the output production
value of these industries will reach a total value of two
trillion NT dollars, thereby making it the third largest
IT industry in the world; this is the so-called “Two
Trillion” project. On the other hand, it aims to continue
to build upon the superior research, development talent
and social environment that Taiwan has as a founda-
tion, in order to develop global star industries, such as
Banks and Bank Systems, Volume 5, Issue 2, 2010
109
its digital content industry and bio-technology indust-
ry, which is referred to as the “Twin Stars” project. As
the government and the general public set these plans
into action, and inject resources into the “Two Trillion,
Twin Stars” project, it is expected that the process of
transforming Taiwan’s industry will be accelerated.
Table 1. Financial data of domestic banks in Taiwan
(1998-2002)
Financial data of domestic banks in Taiwan from 1998 to 2002.
The data are taken from Statistics Office of the Bureau of
Monetary Affairs, Ministry of Finance, Financial Statistics and
Financial Statistics Index.
Item 1998 1999 2000 2001 2002
Annual total assets
(million NT)
175,479 192,601 207,751 217,408 220,970
Return on total assets 0.7 0.54 0.46 0.26 -0.47
Net income rate 9.57 8.01 7.27 4.5 -10.57
Nonperforming
loans ratio
4.37 4.88 5.34 7.48 6.12
Since the transformation of Taiwan’s industries
started to take place, the IT industry has gradually
replaced traditional industry as the primary driving
force of economic development. Through the
extension of loans to the IT industry, banks can
generate profits, sustain themselves, and allocate
further funds to the industry, which will enable the
industry to gain enough funds for investment and
will help develop the economy as a whole.
Nevertheless, since the industry started to be
transformed, the profitability of banks has decreased
as a whole (see Figure 1).
From Figure 1 we can see that during the industrial
transformation and the development of the IT
industry, the profitability and loan quality of the
banks continuously worsened. It may have been
caused by the special characteristics of the IT
industry. While the IT industry has experienced a
very rapid pace of change, a big requirement for
investment and a long return period compared to the
traditional industry, it has encountered much greater
risk than traditional industries. Meanwhile, many
firms in the new technology industries do not have
enough collateral, and therefore, banks have to
worry about NPLs and bad debts when lending to
the IT industry. The loss may lead to potential
bankruptcy. Therefore, the dilemma between profi-
tability and potential loss causes banks to hesitate in
their lending policies.
-200
-100
0
100
200
300
1995 1996 1997 1998 1999 2000 2001 2002
Year
LINE 1
LINE 2
LINE 3
Notes: The data are taken from Financial Statistics Monthly, Republic of China (Taiwan), published by the Economic Research
Department, Central Bank of the Republic of China (Taiwan), and the Taiwan Economic J ournal Data Bank. Line 1 is the domestic
banks’ earning deflator; line 2 is the domestic banks’ NPL ratio deflator; and line 3 is the IT firms’ market value to all listed firms’
market value deflator (the deflator is 100 in 1995).
Fig. 1. Domestic banks and IT industry development
When banks make the maximization of profitability
their target, loan policy takes into consideration the
competitiveness and resources of a bank to determine
how many resources it will allocate to its credit
departments, or how much funding each credit
branch (e.g., in terms of business loans, consumer
loans, etc.) can have. Under limited bank resources,
the quality of the bank’s loan policy determines the
ability of the bank to achieve its goal. This study does
not evaluate the quality of bank loan policy by
observing the history of its strategy. It rather analyzes
the profitability of different loan combinations, and
the kinds of borrower that can help the bank achieve
maximum profit, as it seeks to further discuss
whether a bank should reallocate its business
resources and change its loan policy to achieve or
accelerate the achievement of its goals. This study
analyzes the effects of loan policy toward the IT
industry on the banks’ operating performance, and
hopes to answer the question as to whether banks in
Taiwan should increase their loans to the IT industry
in order to maximize their profits.
This study seeks to clarify whether loans to the IT
industry can become the primary source of profit for
banks from the perspective of protecting the financial
Banks and Bank Systems, Volume 5, Issue 2, 2010
110
markets and promoting economic development. If the
results show that the increase in the flows of capital
between financial institutions and the IT industry will
boost the profitability of financial institutions, it
becomes clear that this kind of loan policy should be
implemented. The reinforcement of the loan
relationships between banks and the IT industry will
be significant for the stability of the financial market
and the development of the economy. On the
contrary, if the results show that increasing loans to
the IT industry will result in a loss, banks should
either be more actively seeking other sources of profit
or reevaluating and readjusting the existing structure
of interactions with industries in order to increase
profitability and reduce their bad debts.
In addition, this study also discusses the loan ratio in
relation to a bank’s assets: even though loan
services are the main sources of income of banks,
banks can still allocate their resources. Besides
loans, banks also handle bonds and securities
investments. In particular, since the rise of the bond
market, the bonds that banks possess have increased
significantly (from 1998 to 2002 that growth has
almost doubled). It, thus, needs to be asked whether
allocating assets to loan services would result in
profitability. The inclusion of the loans to total
assets ratio in this study helps us to understand
whether banks’ resources in terms of loan services
are beneficial for creating the banks’ profitability.
The remainder of this article is organized as follows.
Section 1 discusses previous studies, highlighting
those studies related to loan policies and the banks’
performance. Section 2 explains the variables that
this study adopts and introduces the methodology
and the model specifications. Section 3 presents the
analysis and discusses the empirical results. The
final section provides the conclusions.
1. Literature review
Since the loan spread is the basic source of profit for
banks, the banks’ loan policies are closely tied to
their operating performance. Many scholars in
discussing the optimal loan policies make the
maximization of the banks’ profitability their
primary goals.
Pringle’s (1974) assumption was based on the
borrowers and interest rates remaining unified, so that
banks could earn profit by controlling the amount of
loans and investing in government bonds, to set the
target function as in the case when banks pursued the
goal of maximizing private wealth. Pringle deduced
that the optimal loan policies (the amount of the
loans), would be affected by the spread of deposits
(minus the risk-free rate of interest), the spread of
loans, the amount of the deposits in the next period,
and the capital of the bank. The study emphasized
that the risk-free rate of interest did not influence the
bank’s decision regarding the amount of the loans.
Graddy and Kyle (1979) also started with the
equation on the balance sheet, and concluded that the
amount of loans, deposits, bank capital and labor cost
were factors that were significant to a bank’s
profitability. In addition, Molyneux et al. (1998),
under the assumption that managers of banks pursued
to maximize profitability, minimize risk, and
maximize expected utility, deduced that the two
major factors influencing a bank’s amount of loans
were the profitability of the loans and the risk faced
by the bank itself.
Besides making theoretical deductions, scholars
have also empirically discussed the influence of the
amount of loans on a bank’s operating performance.
Taking Molyneux et al. (1998) as an example, in
that paper they discussed how loan behavior affects
bank performance, utilizing the data for foreign
banks in the United States from the first quarter of
1990 to the third quarter of 1992. With the 2SLS
method, they found that with the return on total
assets (ROA) as the proxy variable for operating
performance, when the amounts of the loans
increase, the bank’s performance is significantly
positively affected. Graddy and Kyle (1979) made
use of the sample data for 463 commercial banks in
1974, and adopted OLS and 3SLS estimations to
find that, when banks raised the amount of the loans
in their assets, at the 0.01 significance level, doing
so would have a significantly positive influence on
the labor costs ratio (labor cost to total assets).
Kwan (2002) examined the banks in 7 Asian
countries (Hong Kong, Indonesia, South Korea,
Malaysia, Philippines, Singapore, and Thailand)
from 1992 to 1999, using regression analysis, and
empirically found that when banks increased the
amount of their loans in earning assets, at the 0.01
significance level, they would significantly and
positively influence the labor cost ratio (the labor
cost to earning assets). This conclusion is in
agreement with Graddy and Kyle (1979).
In financial markets, the conditions of borrowers are
not identical; therefore, in addition to the banks’
decisions as to the amounts of the loans, the targets of
the loans and their condition are also important
factors related to bank performance. By under-
standing the borrower, the risk premium can be
ascertained, and the profit erosion from bad debts can
be decreased. Basically borrowing can be categorized
as consisting of business loans and consumer loans.
With differences in time periods, regions, and
subjects of study, scholars categorize the amounts of
the loans, and then discuss the influence of different
types of loans on bank performance. For example,
Fraser et al. (1974) used American banks in 1969 and
1970 as the study sample, and applied canonical
Banks and Bank Systems, Volume 5, Issue 2, 2010
111
correlation analysis to find that the composition of
loans (the ratio of different types of loans including
real estate, agriculture, consumer and business
loans) would significantly affect the bank at the 0.01
significance level.
In addition, Arshadi and Lawrence (1987), based on
Graddy and Kyle’s (1979) research, divided the
factors that influenced bank performance into seven
types, including cost, structure, loan composition,
deposit composition, regional factors, economic
conditions and scale. The loan composition type
categorized the total amount of loans into real-estate
loans, agricultural loans, consumer loans and
business loans. Through the categorization, their
study empirically analyzed new banks in America
from 1980 to 1984 (the purpose of the study was to
discuss the influence of the factors on bank
performance after the new banks had entered the
market), using canonical correlation analysis. It was
found that the above-mentioned factors did exhibit a
significant correlation with operating performance.
Furthermore, different types of loan and bank
performance had different degrees of correlation. For
example, the study empirically found that agricultural
loans and business loans were significantly
negatively correlated with operating performance,
and the authors reached the conclusion that loan
composition directly influenced bank performance.
The meaning of a functional policy under a loan
policy is that, when the bank makes a loan policy
decision to allocate resources to different
departments, each department can, based on the
direction of the policy, design functional policies to
assist in the execution of the loan policy and attain
profitability. For instance, by observing the
borrowers, it may be possible to decrease risk and
loss associated with the loan. Glover (1939) once
categorized the factors that should be taken into
consideration in lending into nine items, including
expecting a future surplus, the financial statement of
the borrower, additional conditions, such as
collateral, the business management of the borrower,
the business operations of the borrower, the purpose
of the loan, the relationship with the borrower, the
scale of the borrower’s business, and the term of the
loan, and stated that the banks should reinforce their
understanding of these factors to increase executive
capability. In addition, Kuo (2001) further introduced
a support system to the loan policy, as a result of
which the data of the borrower (such as the financial,
managerial, and economic aspects) were put into the
system to scientifically increase the quality of the
loans and the profitability of banks.
In addition to the above-mentioned discussions of
loan policy and bank performance in terms of the
amounts of the loans and the types of loans, there
are other studies that regard NPLs as an explicit
factor of the severity of the loan policy, the reason
being that the amount of the loans is sometimes
determined by the external factors of the
environment, and not necessarily by the willingness
of the manager to take risks. However, when a bank
chooses a high-risk loan policy, it should reasonably
reflect on the higher amount of bad loans, and create
a positive influence on the performance of the bank.
This is referred to as the policy hypothesis by
scholars. Jordan (1998) took the new banks in
England from 1989 to 1992 as his sample to test the
hypothesis. The study used cost efficiency and profit
efficiency as proxy variables for bank performance.
Through examination, he found that the amount of
bad loans did have a significant positive correlation
with profitability, proving that the effect of the policy
hypothesis did exist. In other words, assuming that
the hypothesis is valid, when banks have high NPLs,
it means that they have easier loan policies, or are
more willing to engage in loan services. The above-
mentioned hypothesis indicates the existence of this
effect, but it is worth noting that Berger and Robert
(1997) once created three different hypotheses as to
the causes of bad loans, namely, bad luck, bad
management, and skimping and moral-hazard.
Skimping meant that, in the pursuit of high
performance, the resources on monitoring were
reduced, which resulted in a high NPLs ratio.
Although the phenomenon of a high NPLs ratio and a
high return did occur, its direction of influence was
directly opposite to that of the policy hypothesis; the
two hypotheses did not contradict each other.
Meanwhile, interest rates on loans can reflect the
preferences of a bank’s lending policy. Liew (1970)
thought that when interest rates for loans increased,
profitability would increase, and would cause a bank
to increase its preference for risks, thus resulting in
an easier loan policy (when interest rates for loans
increase, the debtor’s risks also increase, resulting in
a higher probability of default). Conversely, Lown
and Peristiani (1996) thought that when banks tried
to reduce loans, they would also increase interest
rates to increase the difficulties for the borrowers.
Nonetheless, in an empirical analysis of the
relationship between loans and interest rates,
Hannan (1991) found that whatever the amount of
the loan was, the interest rate and the amount of the
loan exhibited a significant negative correlation.
The empirical finding of Lown and Peristiani (1996)
exhibited significant positive correlations in small
banks, and insignificant negative correlations in
large banks, indicating that the two loan policies had
a mixed influence on the interest rate.
From the previous literature we may clearly see that
there are many different factors related to the
differences in banks’ lending policies. Different
Banks and Bank Systems, Volume 5, Issue 2, 2010
112
lending policies (severeness, composition) have
different effects on bank performance. However, the
previous literature rarely studies the influence of loan
policies on bank performance in a country that is
transforming its industries from traditional industries
to IT industries. When we find that Taiwan faces the
problem of low performance for banks and the
transformation of industry, we consider the loan
policies of banks towards the IT industry and their
relationship with performance. On the one hand, it
discusses the aspects that the previous studies have
rarely touched upon; on the other hand, it provides
appropriate suggestions for banks in the face of
today’s dilemma. We trust that this study will be
valuable and will contribute to related studies.
2. The method
2.1. Variables. 2.1.1. Bank performance variables.
The purpose of this study is to discuss the
relationship between loan policies and bank
performance. There are many factors involved in
measuring bank performance. The performance
referred to in this study is operating performance,
i.e., banks are evaluated from the perspective of
profitability. The important consideration for loan
policies concerns whether banks in Taiwan will
increase the ratio of loans to the IT industry, which
has higher risk. In terms of the explanatory variable,
the return on total assets (ROA) and the return on
equity (ROE) serve as the proxy variables for bank
performance (Arshadi and Lawrence, 1987; Bourke,
1989; Webster, 1997; J ordan, 1998; Molyneux et
al., 1998; J oyce, 2001; Sigler and Porterfield, 2001).
2.1.2. Loan policy variables. The loan policy
variables include the loan policies of banks toward
the IT industry, the consumer loan ratio, as well as
the total loans to total assets ratio. Through the design
of the above-mentioned variables, the analysis is
expected to indicate whether increasing the loan ratio
and amount for the IT industry, increasing the
consumer loan ratio, and utilizing assets in loan
services can effectively increase the bank’s
profitability, i.e., the analysis of the adjustment of the
loan policy following the transformation of industry.
(1) IT loan ratio
While discussing the relationship between
borrowers and bank performance, the purpose is to
understand the bank’s allocation of funds for loans.
When the borrowers are different, there will be
differences in terms of bank performance. In terms
of the loans, many previous studies used “the loans
for specific borrowers to total loans” as a proxy
variable for this decision (Fraser and Rose, 1971;
Fraser et al., 1974; Arshadi and Lawrence, 1987).
Some scholars have also elected to use “the loans
for specific borrowers to total assets” as the proxy
variable (Graddy and Kyle, 1979). The difference is
that the latter variable takes into consideration the
bank’s asset allocation policy and loan policy at the
same time. This study uses “the loans to the IT
industry to total loans” (IT loan ratio) as the proxy
variable. Its function is to discuss whether the IT
industry has a more significant influence than other
borrowers when banks select borrowers. Of course,
when banks do not have other ideal borrowers,
whether or not lending to the IT industry can result
in profitability is also an important issue.
(2) Consumer loan ratio
With the decreases in the traditional industry’s
demand for funds, banks can try to increase
profitability not only through increasing loan
services for the IT industry, but also by extending
the consumer loan market. Increasing the consumer
loan ratio is also one of the banks’ main loan
policies in recent years. The consumer loans have
continued to reach new highs in recent years, and
the consumer loan ratio has increased from 32.81%
in 1997 to 37.61% in 2001. It is mainly caused by
not only traditional personal loan services, but also
the proliferation of credit cards and cash cards.
Since Makto Bank first started to provide this
service, 30 financial institutions have begun to offer
cash card services. Compared to the high risk and
uncertainty of loans in the IT industry, whether
increasing amount of consumer loans and the
consumer loan ratio (consumer loans to total loans)
should be included in banks’ lending policies to
increase profitability or not will be empirically
analyzed in this study. In this study, in order to
analyze the influence of the newly-launched
consumer loan services on bank performance, the
consumer loans in this study will exclude mortgages
as well as car, housing and welfare loans for the
sake of accuracy in this study.
(3) Total loans to total assets ratio (Loan ratio)
To examine the allocation of bank assets, this study
also applies “total loans to total assets” (the loan
ratio) as the proxy variable for the asset allocation
policy. The meaning of this variable involves
measuring the allocation of bank assets, and it also
reflects the risk faced by the bank. When this ratio is
higher, it means that banks will be more enthusiastic
in offering loans to the general public instead of
letting them remain unused or using them to
purchase securities, which have much higher risk.
Under the basis of high risk and high returns,
increasing the total loans to total assets ratio should
have a positive influence on bank performance;
however, if the risk management is ineffective,
which causes the NPLs ratio to be too high and
erodes profitability, the result will be negative.
Molyneux et al. (1998) empirically found that the
Banks and Bank Systems, Volume 5, Issue 2, 2010
113
foreign banks in America were in a situation where
the loan ratio was significantly negatively correlated
with the ROA.
2.1.3. Bank’s structure variables. The bank’s
structural factors in this study include the debt ratio,
the bank scale, and the number of bank branches.
Their purpose is to analyze whether the high debt
characteristic of the bank industry and bank scale
influence the profitability of banks. In addition, in the
financial environment of Taiwan, where there are
large numbers of banks and bank branches, it needs
to be asked whether increasing the number of bank
branches has a negative influence on profitability, or
impacts the operation of banks because of high cost.
(1) Debt ratio
On the one hand, the debt ratio represents a bank’s
consideration of cost for its sources of funds, mea-
ning that when the bank uses more external capital
its profitability will be more limited. On the other
hand, when the debt ratio is too high, the bank will
have to monitor its risk and cannot engage in high-
risk high-return services (Lown and Peristiani,
1996; Berger and Udell, 1994). It also compresses
the profitability of a bank. Under this theory, the
debt ratio should be negatively correlated with bank
performance.
(2) Bank scale
When a bank has a larger scale of operations, it
means that it has more resources and services for its
consumers, and it can also decrease its costs through
scale economies; its size is, thus, beneficial for
profitability. However, under poor management,
when the bank cannot cover its huge costs, its
oversized scale of operations will adversely affect
the bank’s performance. In terms of empirical
studies, Miller and Noulas (1996) found that larger-
scale banks had lower operating performance,
showing that bank scales were negatively correlated
with their performance. Meanwhile, Zhang (1996)
categorized the scales of banks by dividing total
assets into large, medium and small, and found that
when the assets scale was too big or too small, it
would become non-efficient, indicating that only the
medium scale would be beneficial for bank
performance. This study uses total assets as the
proxy variable for bank scale.
(3) Number of branches
The number of branches, which can increase the
competitiveness of a bank, represents the bank’s
ability to provide service and convenience. Under
this theory, the number of bank branches will be
positively correlated with the expected profitability
of the bank. Arshadi and Lawrence’s (1987) study
also shows that there is a positive correlation. On
the other hand, there are a total of 3,068 branches of
financial institutions in Taiwan. With a limited
market, the overabundance of branches will render
the operation inefficient, result in unused resources,
and affect the profitability of the banks adversely.
Anderson et al. (1982) found that having too many
branches would negatively influence bank
performance. Therefore, this study includes the
number of bank branches as an explanatory variable.
2.2. Major events during the study period. During
the period from 1998 to 2002, the financial
environment of Taiwan underwent major changes.
In the middle of the year 2000, for the first time in
Taiwan’s political history the ruling party changed.
The original ruling party, the Kuomintang (KMT,
the Chinese Nationalist Party), lost the election, and
the Democratic Progressive Party (DPP), which had
never had the experience of ruling before, became the
ruling party. On the one hand, the previous ruling
party had launched many financial reforms and, on
the other hand, the inexperience of the new
government was reflected on its indecisive policies,
such as those related to the nuclear power plants and
the reform of the farmers’ and fishermen’s associa-
tions. Meanwhile, the lack of communication or a
relationship between the new government and China
also increased the domestic political crisis. Using a
study by Business Environment Risk Intelligence
(BERI) as an example, the analysis shows that the
investment environment in Taiwan in 2003 was at the
lowest risk level 1A and was ranked 4
th
globally,
while Taiwan’s political risk was ranked 11
th
globally
(the lower the rank, the less the risk), indicating that
the political environment of Taiwan adversely
affected the investment environment in Taiwan. In
addition, in 2000, after the ruling party had changed,
Taiwan’s political risk score decreased by 6, and its
rank declined to 16
th
globally. The political risk
indicator did not return to its former level until 2002.
In addition to political factors, during this time,
Taiwan’s economic environment also underwent
drastic change. From the fourth quarter of 2000
onwards, Taiwan’s GDP growth rate began to slide
down. In the second, third and fourth quarters of
2001, there was even negative growth, which had
not occurred in 20 years. Furthermore, the private
investment growth rate and the investment amount
also began to decrease in 2001, showing that the
economic environment had undergone significant
changes since the middle of the year 2000.
In particular, there was the bursting of the Internet
bubble in the year 2000. With this unexpected turn
of events in the development and profitability of the
electronic business market, investors in this market
encountered huge losses. Amazon.com, for example,
Banks and Bank Systems, Volume 5, Issue 2, 2010
114
saw its stock price fall from 113 USD in 1999 to 40
USD in 2000. A number of well-known internet
companies also suffered huge losses, which
impacted both the domestic and foreign economic
environment.
To sum up, after the middle of the year 2000, the
following questions are raised. Do the increases in
political risk, the changes in the domestic economic
and investment environment, and the effects of the
Internet bubble on the international economic
environment negatively influence the domestic
economy? Do those influences result in a loss of
profitability, or a loss of performance indirectly
through the bank’s business relationship with
various corporations?
This study discusses the impact of such influences on
bank performance as a result of these drastic changes.
Thus, we set a dummy variable, referred to as TIME,
before the middle of the year 2000 as TIME =0, and
after the middle of the year 2000 as TIME =1, in
order to analyze the impact of the change in the
ruling party on Taiwan’s financial sector.
The operating definitions of the variables in this
study will be shown in Table 2.
2.3. Statistical method and model specifications.
The sample in this study consists of banks covering
the period from 1998 to 2002, with half a year as a
period, or a total of 10 periods, including cross-
sectional and time series panel data. Since banks
differ from each other, we should not directly
adopt the ordinary least squares (OLS) method, but
through testing should select the most appropriate
model for the data based on the OLS model,
namely, the fixed effects model (FE model) and the
random effects model (RE model), to avoid
estimation bias. The distinctive feature of the FE
model is that, by using panel data, the intercept
term will change because of the different
characteristics of banks, but the error term of the
sample does not change with time, indicating that
the error term remains constant at different times,
i.e., during the observation period, the individual
sample is reflected by the intercept term of the
regression equation. Therefore, the FE model is
also referred to as the dummy variable method; at
this time, the intercept changes only with the
differences among banks, but not with time. The
RE model, by contrast, mainly assumes that the
individual differences in the sample are random,
and are independent of the independent variable. It,
therefore, shows that the differences in terms of the
individual samples will be reflected in the error
term in the regression equation. Hence, the RE
model can also be referred to as the error compo-
nent model.
Table 2. Operating definitions of variables
In this table we present the operating definitions of return on assets
(ROA), return on equity (ROE), high-technology industries loan
ratio (IT loan ratio, ITRATIO), consumer loan ratio (CONRATIO),
loan ratio (LORATIO), debt ratio (DRATIO), bank scale (SCALE),
number of branches (BRANCH), and time dummy (TIME).
Variable Operating definition
Return on assets (ROA)
(Net gain after tax before interest ÷ Total
assets) ×100%.
Return on equity (ROE)
(Net gain after tax before interest ÷ Total
equity) ×100%.
IT loan ratio (ITRATIO) (The loans for IT industry ÷ Total loans) ×100%.
Consumer loan ratio
(CONRATIO)
[(Consumer loans minus housing, car,
maintenance, and welfare loans) ÷ Total loans]
×100%.
Loan ratio (LORATIO) (Total loans ÷ Total assets) ×100%.
Debt ratio (DRATIO) (Total debt ÷Total assets) ×100%.
Bank scale (SCALE) Logarithm of total assets.
Number of branches
(BRANCH)
Logarithm of the number of branches.
Time factor (TIME)
Dummy variable: before the middle of year 2000,
TIME = 0; after the middle of year 2000, TIME = 1.
To compare which of the models would be more
appropriate for the data, we have to test each of them.
The comparison between the OLS model and the FE
model can be achieved by means of the F-test, the F-
test being a distribution that corresponds to F (N-1,
NT-N-K-1). The comparison between the OLS model
and the RE model can be achieved by means of the
Lagrange Multiplier (LM), where the LM is a chi-
square distribution with one degree of freedom. When
the FE model and the RE model are both superior to
the OLS model, we should select from the FE model
and the RE model, and use the Hausman test to test
whether the intercept term and the independent
variable are statistically correlated. If the intercept term
and independent variable are not statistically correla-
ted, we should adopt the RE model; if, alternatively,
the intercept term and independent variable are
statistically correlated, we should adopt the FE model.
Through the above-mentioned hypothesis tests, we
select the most appropriate model for the data type
from the OLS model, FE model and RE model, in
order to decrease the estimation bias. This will be
helpful to the accuracy of this study.
This study mainly discusses the effect of loan
policy, the bank’s structural factors and changes of
the politico-economical environment in 2000 on
bank performance. Based on the above-mentioned
variable specifications, we select the appropriate
variables for the analysis. This study constructs
three models for analysis.
The purpose of the first model (Model 1) is to
discuss the relationship between borrowers, resource
allocation and bank operating performance. Thus, in
Model 1, we select three variables as independent
variables: the IT loan ratio (ITRATIO), the consumer
loan ratio (CONRATIO), and the loan ratio
Banks and Bank Systems, Volume 5, Issue 2, 2010
115
(LORATIO). The dependent variables are denoted
by operating performance (OP) in equation (1).
When we use ROA as the proxy variable for bank
operating performance, it is shown in Model 1-1. In
addition, this study attempts to again treat ROE as
the proxy variable for bank performance, and places
it in Model 1-2 as the dependent variable for the
model to analyze and compare it with Model 1-1.
Through the specification of Model 1, we expect
that after the empirical analysis is completed, the
result can help banks to select borrowers and design
loan policies, in order to increase their profitability.
it it i it
CONRATIO ITRATIO OP
2 1 0
it it
LORATIO
3
. (1)
In Model 2, in addition to the same independent va-
riables as in Model 1, i.e., ITRATIO, CONRATIO,
and LORATIO, we further incorporate the bank’s
structural variables, including the debt ratio
(DRATIO), bank scale (SCALE), and the number of
branches (BRANCH), into the model to discuss the
influences that these variables have on the bank’s
profitability, and thereby, increase the explanatory
power. In Model 2, we use ROA (in Model 2-1) and
ROE (in Model 2-2) separately to represent the
factors, and to measure, discuss and compare the
influence of loan policy and structural factors on the
bank’s performance:
it it t i it
CONRATIO ITRATIO OP
2 1 0
it it it
SCALE DRATIO LORATIO
5 4 3
it it
BRANCH
6
. (2)
The purpose of Model 3 is to discuss whether the
bank’s performance exhibits significant change
before and after the middle of the year 2000, and
whether the influence the borrowers have on the
bank’s operating performance has changed
significantly after this time period. Thus, in Model 3,
in specifying the independent variables, we include
the three variables, the same as in Model 1; we also
include the middle of the year 2000 as the time
factor. The model with ROA as the dependent
variable is Model 3-1, and the model with ROE as the
dependent variable is Model 3-2. Through the
specification of the model, we can analyze the
influence that the time factor has on the bank’s
operating performance.
it it i it
CONRATIO ITRATIO OP
2 1 0
it it it
TIME LORATIO
4 3
. (3)
3. Results
3.1. Data source and descriptive statistics. Our
data are collected from the “Financial Report
Database” in the Taiwan Economic Journal (TEJ ),
and a part of the undisclosed data is deduced from
the estimation
1
. The sample consists of all domestic
financial institutions and small to medium
commercial banks listed on the Taiwan Stock
Exchange from 1998 to 2002. Banks with missing
data are excluded from the sample. As a result, a total
of 27 listed banks (270 semiannual observations) are
included in this study. The descriptive statistics of the
variables are presented in Table 3.
Table 3. Descriptive statistics of variables
The sample consists of 270 observations on Taiwanese listed
banks from1998-2002. Data are taken fromthe “Financial Report
Database” in the Taiwan Economic Journal (TEJ). ROA is the
ratio of net gain after tax before interest to total assets. ROE is the
ratio of net gain after tax before interest to total equity. ITRATIO
is the ratio of the loans for IT industry to total loans. CONRATIO
is the ratio of consumer loans minus housing, car, maintenance,
and welfare loans to total loans. LORATIO is the ratio of total
loans to total assets. DRATIO is the ratio of total debt to total
assets. SCALE is the logarithmof total assets. BRANCH is the
logarithmof the number of branches.
Variables N Minimum Maximum Average Std. deviation
ROA 270 -2.44 1.45 0.19 0.68
ROE 270 -57.59 15.64 1.72 10.53
ITRATIO 270 0.00 52.59 3.80 11.06
CONRATIO 270 0.00 40.89 50.18 4.63
LORATIO 270 41.51 78.55 65.84 6.22
DRATIO 270 88.81 97.28 92.29 1.64
SCALE 270 4.84 7.18 5.81 0.70
BRANCH 270 3.18 5.16 3.98 0.54
From Table 3, we may find that, among the observa-
tions, banks have huge differences in profitability.
Taking ROA as an example, the maximum is 1.45%,
and the minimum is -2.44%. We find that banking is
not an “easy to operate” industry and slight mistakes
can result in huge losses.
Moreover, with regard to the IT loan ratio and the
consumer loan ratio, we may find that the loan
policies differ significantly among banks. From the
significant differences in the banks, we may find
that domestic banks do not have a general consensus
on the loan policy for optimal profitability; this is
also why this study is valuable.
In addition, from the loan ratio, we may learn that
all of the samples treat loaning as the main business,
and the differences are slight. In terms of the assets
scale, the maximum is ten times the minimum,
showing that there are still banks with very small
scales of operations in Taiwan. Finally, in terms of
the number of branches, we find that the minimum
is 24 while the maximum is 175, and the standard
deviation is as high as 38.42, showing that the banks
1
Since the banks did not disclose accurate data on the IT nonperforming
loans ratio, we calculate the corporate borrowers disclosed by the banks
as the sample used in this study, and the IT loan amount is estimated by
the loan ratio multiplied by the amount of loans for the corporations
which the banks have disclosed.
Banks and Bank Systems, Volume 5, Issue 2, 2010
116
have significantly different opinions as to how many
branches should be established
1
.
To avoid the problem of collinearity, we use the
variance inflationary factor (VIF) to test the collinear-
rity of the variables. Marquardt (1970) thought that
when the VIF was greater than 10, the model would be
seriously affected by collinearity. Through the ana-
lysis, the variables of the above-mentioned models all
have VIF values under 10. Thus, we determine that
there is no collinearity that will affect the empirical re-
sults in the model. The analysis is presented in Table 4.
3.2. Empirical results. 3.2.1. The effects of lending
policy on bank performance. From Table 5, we are
able to find that for Model 1 (including Model 1-1 and
Model 1-2), both the F-test and the LM test indicate
that the panel data model performs better than the OLS
model. The Hausman test also suggests that the FE
model is more suitable for use than the RE model.
Table 4. The VIF of the independent variables
We use the variance inflationary factor (VIF) to test the
collinearity of the variables. ROA is the ratio of net gain after tax
before interest to total assets. ROE is the ratio of net gain after tax
before interest to total equity. ITRATIO is the ratio of the loans for
IT industry to total loans. CONRATIO is the ratio of consumer
loans minus housing, car, maintenance, and welfare loans to total
loans. LORATIO is the ratio of total loans to total assets. DRATIO
is the ratio of total debt to total assets. SCALE is the logarithmof
total assets. BRANCH is the logarithmof the number of branches.
TIME dummy equals one if after the middle of year 2000 and
zero otherwise. The dependent variable of models 1, 2, 3 is ROA.
The independent variables of Model 1 are ITRATIO, CONRATIO,
and LORATIO. The independent variables of Model 2 are
ITRATIO, CONRATIO, LORATIO, DRATIO, SCALE, and
BRANCH. The independent variables of Model 3 are ITRATIO,
CONRATIO, LORATIO, and TIME.
Variables Model 1 Model 2 Model 3
ITRATIO 1.034 1.324 1.142
CONRATIO 1.016 1.102 1.026
LORATIO 1.035 1.357 1.036
DRATIO 1.603
SCALE 3.434
BRANCH 3.406
TIME 1.119
Table 5. The effects of loan policy on bank
performance
The dependent variable is either ROA (Model 1-1) or ROE
(Model 1-2). ROA is the ratio of net gain after tax before
interest to total assets. ROE is the ratio of net gain after tax
before interest to total equity. ITRATIO is the ratio of the loans
for IT industry to total loans. CONRATIO is the ratio of
consumer loans minus housing, car, maintenance, and welfare
loans to total loans. LORATIO is the ratio of total loans to total
assets. We use F-statistic test (F-test), Lagrange Multiplier test
(FM-test) and Hausman test to examine whether OLS model,
fixed effect model, or random effect model are appropriate
1
In the analysis of descriptive statistics, both total assets and the
number of branches are original values, the maximumamount of total
assets is 13151.8 billion, and the minimumis 126.40 billion.
ones. ***, **, and * indicate statistical significance at the 1%,
5%, and 10% levels, respectively.
Variables
Model 1-1
ROA
Model 1-2
ROE
ITRATIO -0.02714*** -0.43191***
CONRATIO 0.028952** 0.470653**
LORATIO 0.027503** 0.401161**
F-test 4.32*** 3.47***
LM-test 24.70*** 11.73***
Hausman test 32.27*** 32.16***
Adj R-squared 0.2368 0.1865
Through Model 1-1 and Model 1-2, we can arrive at
the following empirical results: when banks raise
their ratios of loans to the IT industry, this will give
rise to a significant negative influence on the banks’
performance. There are two possible reasons for
this. When banks increase their loans to the IT
industry, even though this action may bring interest
income from the spread between deposits and loans,
banks still have to input more resources in order to
enter this industry and avoid the decreases in the
quality of the loans because the IT industry is highly
risky and it is difficult for it to provide collateral.
Besides, during the 1998-2002 period, the banks’
NPLs increase but profitability decreases; this might
be due to the the fact the bank cannot ascertain the
risk and select a good borrower; consequently, this
gives banks higher NPLs, erodes the profitability
and creates this negative correlation.
In addition, in terms of the consumer loan ratio,
although this proves that the trend in the banks’ loan
policies is correct, what is particularly worth noting is
that the market is expanding and maturing, and the
competition will be more fierce. Newcomers will
divide the existing market and banks will have to pay
more to maintain the share, such as in terms of the
marketing cost and R&D cost. The fierce competition
will also force banks to decrease interest rates and
fees, and even expand the market to risky borrowers,
such as students. In these circumstances, whether or
not increasing the consumer loan ratio will be
beneficial to the bank’s performance is certainly
worth looking into.
In terms of the allocation of bank assets, we find that
the loan ratio is significantly positively correlated with
the banks’ performance, and this conclusion is the
same as that of Molyneux et al. (1998), indicating that
banks’ loan services have higher profitability than
other investments. One of the possible reasons may be
that loan service is the main service of banks, and both
talent and experience are comprehensive; thus, banks
may rely on loans for profit. Another possible reason is
that the securities market in Taiwan has yet to mature;
the targets of investment are limited and the risks are
high, thus affecting the profitability of banks.
3.2.2. The effects of bank structural variables on
bank performance. From Table 6, we may find that
Banks and Bank Systems, Volume 5, Issue 2, 2010
117
in Model 2 (including Model 2-1 and Model 2-2),
both the F-test and LM test indicate that the panel
data model performs better than the OLS model.
The Hausman test also suggests that the FE model is
more suitable to use than the RE model.
From the influences that loan policy has on bank
performance, we find that in Model 2-1, the IT
industry loan ratio and consumer loan ratio are
significant at the 1% and 5% significance levels,
respectively. In Model 2-2, the IT industry loan ratio
and consumer loan ratio are significant at the 1% and
10% significance levels, respectively. This indicates
that when banks increase their ratios of loans to the
IT industry, the lending policy will significantly
negatively influence bank performance; when banks
increase their consumer loan ratios, their lending
policy will significantly and positively benefit bank
performance. This conclusion is the same as in the
case of Model 1. During the period of 1998-2002, the
annual business loan interest rate was 4%~6%, which
is lower than the annual consumer loan interest rate
(the annual credit card interest rate was 18%~20%,
and annual cash card interest rate was 17%~18%).
The NPLs ratio was increasing from 4.36% to 8.55%,
and majority NPLs were business bad debts. The
asymmetric return-risk distributions between business
and consumer loans made the IT industry loan
negatively related to bank performance while the
consumer loan positively related to bank
performance. These encouraged Taiwanese banks to
granting credit to consumers. The Taiwanese banks
faced the cash-credit card crisis in 2005.
Table 6. The effects of bank structure variables on
bank performance
The dependent variable is either ROA (Model 2-1) or ROE (Model
2-2). ROA is the ratio of net gain after tax before interest to total
assets. ROE is the ratio of net gain after tax before interest to total
equity. ITRATIO is the ratio of the loans for IT industry to total
loans. CONRATIO is the ratio of consumer loans minus housing,
car, maintenance, and welfare loans to total loans. LORATIO is the
ratio of total loans to total assets. DRATIO is the ratio of total debt
to total assets. SCALE is the logarithmof total assets. BRANCH is
the logarithmof the number of branches. We use F-statistic test (F-
test), Lagrange Multiplier test (FM-test) and Hausman test to
examine whether OLS model, fixed effect model, or randomeffect
model are appropriate ones. ***, **, and * indicate statistical
significance at the 1%, 5%, and 10% levels, respectively.
Variables
Model 2-1
ROA
Model 2-2
ROE
ITRATIO -0.0189032*** -0.3154417***
CONRATIO 0.0237138** 0.3780544*
LORATIO 0.0058228 0.0845423
DRATIO -0.4337637*** -6.454996***
SCALE 1.978064*** 28.77167***
BRANCH -1.423505*** -19.89049***
F-test 6.05*** 4.80***
LM-test 23.47*** 11.34***
Hausman test 134.38*** 99.94***
Adj R-squared 0.5792 0.4951
In terms of the allocation of bank assets, we find
that the loan ratio does not have a significant
influence on bank performance in Model 2; this
conclusion is different from the positive results of
Model 1. This shows that with the addition of the
bank’s structural variables, the effect of the loan
ratio on performance disappears, indicating that
other variables compared to the loan ratio are more
significant in terms of their effect on bank
performance. Therefore, in Model 2, other variables
account for the explanatory power of the loan ratio,
and the loan ratio does not exhibit a significant
result in Model 2. As a result, this study suggests
that increasing the loan ratio does not have a
significant positive influence on bank performance.
This study also suggests that the bank has to
reinforce its own crediting capability, so that it will
be able to maintain a certain level of profitability
under the situation where there is no good
investment opportunity available.
Furthermore, the empirical results show that bank
scale has a significant and positive influence on
operating performance, indicating that when a bank
expands its scale and integrates resources, talents
and service, it may increase its competitiveness and
profitability. Recently, Taiwanese banks have
expanded their scales of operations through mer-
gers. If merged successfully, the competitiveness of
domestic banks will increase due to internationali-
zation. From another perspective, when resources
merge and increase, how to allocate resources
effectively and reduce unnecessary costs will
become important issues.
We also learn that the number of branches is still
negatively correlated with bank performance,
matching the empirical results of Anderson et al.
(1982) from Model 2. This indicates that there are
generally too many branches in Taiwan, and it causes
inefficiency in operation, leaving too many unused
resources and, thereby, adversely affecting profita-
bility. In fact, in recent years, the establishment of
new branches has slowed down. Banks may have
also faced problems with having too many branches.
Conversely, in order to maintain their services and
cut costs, banks have gradually replaced normal
branches with simplified branches to improve the
banks’ financial situation by decreasing costs.
In terms of the empirical results of other variables,
we find that the debt ratio is significantly negatively
correlated with bank performance; this is the same
as in the previous literature. When the debt ratio is
too high, banks will have to monitor risk and will
not be able to engage in high-risk high-profit
businesses (Lown and Peristiani, 1996; Berger and
Udell, 1994); this also compresses the profitability
of banks. The above-mentioned considerations of
Banks and Bank Systems, Volume 5, Issue 2, 2010
118
cost lead to banks using external capital and facing
high risk, as a result of which they have to offer
higher interest rates on deposits. It may also be one
of the reasons why the debt ratio and bank
performance are negatively correlated.
3.2.3. The effects of the time factor on bank perfor-
mance. During the period from 1998 to 2002, the
financial environment of Taiwan was dramatically
changed. In 2000, the Kuomintang Party lost the
election, and the new ruling party, Democratic
Progressive Party, launched many financial reforms.
In addition, the domestic political risk adversely
affected the investment and economic environments
in Taiwan. Furthermore, many electronic companies
suffered huge losses from the Internet bubble. In
this study, we further to discuss the impact of these
changes on bank performance.
From Table 7, we may see that in Model 3
(including Model 3-1 and Model 3-2), both the F-
test and LM test indicate that the panel data model
performs better than the OLS model. The Hausman
test also suggests that the FE model is more suitable
for use than the RE model.
Table 7. The effects of the time factor on bank
performance
The dependent variable is either ROA (Model 3-1) or ROE
(Model 3-2). ROA is the ratio of net gain after tax before
interest to total assets. ROE is the ratio of net gain after tax
before interest to total equity. ITRATIO is the ratio of the loans
for IT industry to total loans. CONRATIO is the ratio of
consumer loans minus housing, car, maintenance, and welfare
loans to total loans. LORATIO is the ratio of total loans to total
assets. TIME dummy equals one if after the middle of year 2000
and zero otherwise. We use F-statistic test (F-test), Lagrange
Multiplier test (FM-test) and Hausman test to examine whether
OLS model, fixed effect model, or random effect model are
appropriate ones. ***, **, and * indicate statistical significance
at the 1%, 5%, and 10% levels, respectively.
Variables
Model 3-1
ROA
Model 3-2
ROE
ITRATIO -0.0115472* -0.1880818*
CONRATIO 0.0400379*** 0.6440054***
LORATIO 0.019804* 0.2807685
TIME -0.3392419*** -5.304663***
F-test 3.70*** 2.88***
LM-test 26.83*** 12.00***
Hausman test 18.73*** 18.30***
Adj R-squared 0.2787 0.2289
Through the empirical results of Model 3, we may
learn that since the middle of the year 2000, the
domestic and international political and economic
environment has had a significant negative influence
on bank performance. The reasons for this influence,
in terms of the political environment, may be twofold.
The first reason is that the new government was
inexperienced and had trouble with China, which in
turn gave rise to political risk and influenced the
willingness of international investors to invest in
Taiwan, thereby, resulting in a poor economic envi-
ronment. Thus, banks cannot get sufficient earnings
from either loans or investments, resulting in a signi-
ficant negative influence. The second reason is that the
new government has passed many new amendments
and new financial laws. The financial institutions
when faced with such new laws must incorporate the
changes into their organizational structures and
include new services, or else simplify their organiza-
tions after paying high costs, which may have a
negative influence on bank performance. In terms of
economics, since the middle of the year 2000, the
domestic economic environment has continued to
deteriorate, and private investors have not had the
willingness to invest. Moreover, the Internet bubble
has also impacted the world economy, resulting in
significant negative influences on the operating
performance of the domestic banking industry.
3.3. Discussions. While analyzing the influences of
lending policy on operating performance, we find that
when banks increase their IT industry loan ratios, this
will result in undesired operating performance. When
we review the background of the period under study,
we find that from 1998 to 2002 bank performance
continued to decrease, while the IT industry continued
to develop, and so the motive was generated. As the
IT industry continues to develop, the demand for
funds increases, meaning that banks may increase
their IT loan ratios; in addition, as bank performance
continues to decline, it needs to be asked whether it is
influenced by IT loans. After the analysis of this
study, we find that with the increases in loans from
the banks to the IT industry, IT loans do give rise to a
negative influence on bank performance.
Furthermore, banks earn profit through loan services
by collecting loan interest. If loan services are not
profitable, such performance will directly reflect
upon the banks’ loan interest income rate (interest
income divided by the loan amount); if the loan
interest income rate is high, it means that each
amount that banks lend out can achieve maximum
interest income. Thus, this study further analyzes
the correlation between the IT loan ratio and loan
interest income rate in order to re-examine the
above-mentioned conclusions.
By means of Pearson correlation coefficient test, we
may find that the correlation coefficient of the IT loan
ratio and loan interest income rate is -0.387, and it is
statistically significant at the 0.01 significance level
(see Panel A of Table 8), indicating that the higher
the ratio of loans to the IT industry is, the lower the
profitability will be
1
. During this period, Taiwanese
1
Banks with missing data are excluded fromthe sample. As a result, 22
domestic banks listed on the Taiwan Stock Exchange from1998 to 2002
(220 semi-annual observations) are included in this study.
Banks and Bank Systems, Volume 5, Issue 2, 2010
119
government pushed the “Two Trillion, Twin Stars”
project and arranged with domestic banks about
granting low interest rate loans to IT industry.
Therefore, the IT loans were the low return and high
risk assets for banks.
In addition, this study finds that the consumer loan
ratio is positively correlated with the banks’ operating
performance. Therefore, we use the same data to
examine the Pearson correlation coefficient for the
consumer loan ratio and the loan interest income rate.
The result shows that Pearson correlation coefficient
between the two is 0.256, and it is statistically
significant at the 1% significance level (see Panel B
of Table 8). This also proves that the higher the loan
ratio is to the general populace, the higher the
profitability will be.
To sum up, both the results of the panel data reg-
ression analysis and Pearson correlation coefficient
test show that increasing the banks’ ratio of loans to
the IT industry will negatively impact the banks’
operating performance. In addition, the consumer
loan ratio is positively related to the banks’ operating
performance. This shows that the empirical results of
this study are robust.
Table 8. The results of the Pearson correlation
coefficient test
We use a Pearson correlation coefficient test to examine the corre-
lation between the IT loan ratio and loan interest income rate. Panel
A reports the Pearson correlation coefficient test of IT loan ratio
and loan interest income rate. Panel B reports the Pearson correla-
tion coefficient test of consumer loan ratio and loan interest
income rate. *** indicates statistical significance at the 1% level.
Panel A. IT loan ratio & loan interest income rate
IT loan ratio Loan interest income rate
IT loan ratio 1*** -0.387***
Loan interest income rate -0.387*** 1***
Panel B. Consumer loan ratio & loan interest income rate
Consumer loan ratio Loan interest income rate
Consumer loan ratio 1*** 0.256***
Loan interest income rate 0.256*** 1***
Conclusions
The primary purpose of this study is to discuss bank
loan policy and its relationship with performance.
Hopefully, we are able to understand through our
analysis whether there are adjustments to the loan
ratio and the loan amount to the IT industry, and the
consumer loan ratio. This study can help banks to
improve their profitability under an undesirable
performance, or actually result in more losses.
Meanwhile, we empirically prove the influence of
different factors, such as a bank’s debt ratio and the
number of branches in relation to bank performance.
The sample consists of 27 domestic banks listed on
the Taiwan Stock Exchange from 1998 to 2002. We
find that in terms of the influence of bank loan policy
on bank performance, when banks increase their loan
ratios to the IT industry, the action is negatively
correlated with bank performance, meaning that
when banks are faced with the compression of loan
business and the transformation of traditional industry
into the IT industry, increasing the loan amount to the
IT industry does not only result in profit, but will be
adversely affected by the complexity and high risk of
the industry, and will result in losses, thus reducing
the banks’ operating performance. If the
transformation of industry is inevitable, banks must
first improve their loan quality, and offer loans to
establish the IT industry with better credentials, avoid
loan losses, and increase profitability. Furthermore,
banks may, through increasing service quality,
increase their own profit bases, and by enhancing
their negotiating capabilities and raising their loan
interest rates, reflect the costs to the banks in order to
sustain themselves. By contrast, the empirical results
demonstrate that the consumer loan ratio has a
significant and positive influence on bank perfor-
mance, indicating that with the growth of the
consumer loan market, banks can input funds in the
market as an effective use of an ideal loan policy. Yet
banks should be aware that once the consumer
market has reached its limits, the pursuit of new
clients may result in a lower standard of borrowers,
and in increased personal bad debts. The increases in
bad loans will erode the banks’ profitability.
Then, in terms of the results of analyzing the
influence of a bank’s structural variables on bank
performance, we find that both the debt ratio and the
number of branches are significantly negatively
correlated with bank performance. This conclusion
agrees with the empirical results of Berger and Udell
(1994) and Anderson et al. (1982), respectively.
Based on this conclusion, banks must decrease their
external capital and utilize more of their internal
capital to lower risk, increase the expansion of
services and positively influence performance. In
addition, banks should decrease the number of their
branches as they increase operating costs, for an
appropriate reduction or the restructuring of branches
into simplified branches will positively increase bank
performance. The empirical results show that bank
scale has a significant positive influence on bank
performance, meaning that the scales of the domestic
banks tend to be too small and the banks, hence, do
not benefit from economies of scale. Through
partially merging, resources and talents can be
integrated into better competitiveness for domestic
banks. In the meantime, this study finds that, since
the middle of the year 2000, banks’ performances
have become more undesirable, indicating that the
political and economic environment since the middle
of the year 2000 has worsened and has impacted the
survival and development of banks. If banks expect
to perform better, increasing their operating and
managerial capability will become a primary task.
Banks and Bank Systems, Volume 5, Issue 2, 2010
120
There are two limitations to this study. First, banks
did not fully disclose their IT industry loan ratios
and loan amounts; thus, this study has to calculate
these figures from the names of borrowers the banks
have disclosed to obtain the data. Second, the lack
of observations could be improved in the future; we
expect that when the data become more
comprehensive, the future related studies will be
able to include more variables to measure bank loan
policy and to contribute to this issue.
References
1. Anderson, R., Handoff, W., McCarthy, M. (1982). Bank Size and the Management of Capital Ratios. The Bankers
Management 165 (1), 64-69.
2. Arshadi, N., Lawrence, C.E. (1987). An Empirical Investigation of New Bank Performance. Journal of Banking
and Finance 11 (1), 33-48.
3. Berger, A.N., Robert., D. (1997). Problem Loans and Cost Efficiency in Commercial Banks. Journal of Banking
and Finance 21 (6), 849-870.
4. Berger, A., Udell, G. (1994). Did Risk-Based Capital Allocate Bank Credit and Cause a ‘Credit Crunch’ in the
U.S. Journal of Money, Credit & Banking26 (3), 585-628.
5. Bourke, P. (1989) Concentration and Other Determinants of Bank Profitability in Europe, North America and
Australia. Journal of Banking and Finance 13 (1), 65-79.
6. Fraser, D.R., Phillips, J .W., Rose, P.S. (1974). A Canonical Analysis of Bank Performance. Journal of Financial
and Quantitative Analysis 9 (2), 287-295.
7. Fraser, D.R., Rose, P.S. (1971). More on Banking Structure and Performance: The Evidence from Texas. Journal
of Financial and Quantitative Analysis 6 (1), 601-611.
8. Glover, J .D. (1939). Industrial Loan Policy of the RFC. Harvard Business Review 17 (4), 465-476.
9. Graddy, D.B., Kyle, R. (1979). The Simultaneity of Bank Decision-making, Market Structure, and Bank
Performance. Journal of Finance 34 (1), 1-18.
10. Hannan, T.H. (1991). Bank Commercial Loan Markets and the Role of Market Structure: Evidence from Surveys
of Commercial Lending. Journal of Banking and Finance 15 (1), 133-149.
11. J oyce, W.B. (2001). Return and Reward: Bank Performance and CEO Compensation. American Business Review
19 (2), 93-99.
12. J ordan, J .S. (1998). Problem Loans at New England Banks, 1989 to 1992: Evidence of Aggressive Loan Policies.
New England Economic Review (J an/Feb), 23-38.
13. Kuo, H.C. (2001). Strategic Decisions for the Management of Banks’ Credit Granting System. International
Journal of Services Technology and Management 2 (3/4), 313-325.
14. Liew, C.K. (1970). Lending Behavior and the Nature of Default Loans in the Case of the Savings and Loan
Industry. Journal of Finance 25 (3), 701-702.
15. Lown, C., Peristiani, S. (1996). The Behavior of Consumer Loan Rates During the 1990 Credit Slowdown.
Journal of Banking and Finance 20 (10), 1673-1694.
16. Marquardt, D.W. (1970), Generalized Inverses, Ridge Regression, Biased Linear estimated, and Nonlinear
Estimation. Technometrics 12 (4), 591-612.
17. Miller, S.M., Noulas, A.G. (1996). The Technical Efficiency of Large Bank Production. Journal of Banking and
Finance 20 (3), 495-309.
18. Molyneux, P., Remolona, E., Seth, R. (1998). Modeling Foreign Bank Performance and Lending Behavior.
Financial Markets, Institutions and Instruments 7 (4), 26-41.
19. Pringle, J .J . (1974). The Capital Decision in Commercial Banks. Journal of Finance 29 (3), 779-795.
20. Sigler, K.J., Porterfield, R. (2001). CEO Compensation: Its Link to Bank Performance. American Business Review
19 (2), 110-114.
21. Kwan, S.H. (2002). Operating Performance of Banks among Asian Economics: An International and Time Series
Comparison. Federal Reserve Bank of San Francisco, Working Paper.
22. Webster, A.L. (1997). The Impact of Technological Change on Bank Performance. Journal of Economics and
Finance 21 (3), 41-47.
23. Zhang, J . (1996). Measurement and Determinants of X-efficiency, Technological Change and Profitability in
Banking. Unpublished Ph.D. Dissertation, The Temple University.
doc_915213685.pdf