This article was downloaded by: [National Chiao Tung University 國立交通 大學]
On: 27 April 2014, At: 17:05 Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
The Service Industries
Journal
Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/fsij20
Ownership and production
efficiency: Evidence from
Taiwanese banks
Yang Li Professor a , Jin-Li Hu Associate Professor b
& Yung-Ho Chiu Professor c a
Institute of Economics and Management , National University of Kaohsiung , Taiwan b
Institute of Business and Management , National Chiao-Tung University
c
Department of Economics , Soochow University , Taiwan
Published online: 25 Jan 2007.
To cite this article: Yang Li Professor , Jin-Li Hu Associate Professor &
Yung-Ho Chiu Professor (2004) Ownership and production efficiency: Evidence from Taiwanese banks, The Service Industries Journal, 24:4, 129-148, DOI: 10.1080/0264206042000275235
To link to this article: http://dx.doi.org/10.1080/0264206042000275235
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy,
completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
Ownership and Production Efficiency:
Evidence from Taiwanese Banks
Y A N G L I , J I N - L I H U and Y U N G - H O C H I U
In the early 1990s, Taiwan began her deregulation trend in order to enhance competition and economic efficiency across all industries. We derive a theoretical framework to predict possible rankings in technical efficiencies of public, mixed, and private banks. A panel data set with 43 Taiwanese banks during 1997 – 1999 is used for empirical analysis. We then apply a translog distance function to estimate technical efficiencies. The relation-ship between technical efficiency and government shareholding is also examined. Empirical results show that a public bank in Taiwan can improve its technical efficiency by mixed ownership at a diminishing rate. Moreover, banks in Taiwan on average performed worse after the 1997 Asian financial crisis.
INTRODUCTION
After the Conservative Party led by Margaret Thatcher won the 1979 election, the UK started to privatise public enterprises with full effort. The UK privatisation experience has since become an example followed by many developed and developing countries. One of the main objectives of privatisa-tion is to improve the efficiency of a public enterprise [Bishop et al., 1994]. Most countries fulfil privatisation through the transfer of ownership; however, during the process of privatisation, the government may not transfer all of its shareholdings. As a result, private and public sectors will jointly own an enterprise. Boardman et al. [1986] define a mixed enterprise as ‘encompassing various combinations of government and private joint equity
Yang Li is Professor, Institute of Economics and Management, National University of Kaohsiung, Taiwan. Jin-Li Hu is Associate Professor, Institute of Business and Management, National Chiao-Tung University. Yung-Ho Chiu is Professor, Department of Economics, Soochow University, Taiwan.
The Service Industries Journal, Vol.24, No.4, July 2004, pp.129–148 ISSN 0264-2069 print=1743-9507 online
DOI: 10.1080=0264206042000275235 # 2004 Taylor & Francis Ltd.
participation’. In the early 1990s Taiwan began to pursue privatisation of its public enterprises in order to enhance competition and economic efficiency across all industries.
Deregulation in Taiwan’s banking industry consists of two major aspects: Privatisation of public enterprises and entrance opportunity. During the past 11 years, nine state-owned banks have been privatised, including Chang Hwa Commercial Bank, First Commercial Bank, Hua Nan Commercial Bank, Taiwan Business Bank, Taiwan Development & Trust Corporation, Farmers’ Bank of China, Chiao Tung Bank, Bank of Kaohsiung, and Taipei Bank.
In 1991 Taiwan’s government released the Commercial Bank Establish-ment Promotion Decree in order to relieve the legal entrance barriers to banking markets. Twenty-five new commercial banks were established afterwards, bringing the total number of domestic commercial banks in Taiwan in 1999 to 43. Taiwan’s government is still trying to make banking markets more competitive for public, mixed, and private banks.
Bank efficiency has received much attention in the existing literature. In earlier studies most efficiency literature focused on estimating the functional characteristics and economies of scale and scope, e.g. Bell and Murphy [1968], Hunter and Timme [1986], Berger et al. [1987], Ferrier and Lovell [1990], Berger and Humphrey [1991], McAllister and McManus [1993], Rhoades [1993], and Huang [1998, 2000], etc. Recently, the emphasis on bank efficiencies has shifted to the decomposition into allocative and technical efficiencies. There are two methods to estimate efficiencies of the sample banks: parametric and non-parametric approaches. Most researchers use DEA (data envelopment analysis) as a non-parametric approach to esti-mate the total productivity and efficiency of banks: e.g., Berger et al. [1987], Berger and Humphrey [1991], Oral and Yolalan [1990], Favero and Papi [1995], Sherman and Ladino [1995], and Miller and Noulas [1996]. Recently, Resti [1997], Bhattacharyya et al. [1997], Chen and Yeh [2000], and Huang and Wang [2002] apply both parametric and non-parametric approaches to estimate bank efficiencies.
The major goal of a private enterprise is profit maximisation. However, for public enterprises, profit maximisation is never the primary goal. Public enterprises are required to achieve particular social ends, such as reducing unemployment rate, promoting economic development, etc. Most govern-ments set up mixed enterprises, intending to combine economic efficiency of private enterprises with a socio-political goal of public enterprises.
Eckel and Vining [1985] provide the first step to analyse mixed enterprises’ performance. They suggest three reasons for converting public enterprises to mixed enterprises: First, mixed enterprises easily achieve higher profitability and social goals at a lower cost than public
enterprises. Second, mixed enterprises have less bureaucratic restrictions than public enterprises. Third, mixed enterprises need less capital invest-ment from the governinvest-ment than public enterprises. Boardman et al. [1986] also point out that mixed enterprises have three major advantages in comparison with public enterprises. The first advantage is that mixed enterprises demand less capital cost than public enterprises. The second advantage is that mixed enterprises are more efficient than public enter-prises, while the third advantage is flexibility such that mixed enterprises achieve both profitability and social goals and are more efficient than public enterprises.
Boardman et al. [1986] indicate that the conflict of interest between shareholders and managers reduces mixed enterprises’ performance. Boardman and Vining [1991] discuss the effect of government vis-a`-vis private ownership on the internal management of an enterprise. They argue that public ownership is inherently less efficient than private ownership since public banks lack a sufficient incentive and generate higher cost inefficiencies. They argue that ‘different ownership conditions affect the extent to which mixed enterprises engage in profit maximisation, socio-political goal maximisation, and managerial utility maximisation; it also affects the degree of conflict between one owner and another’. They further predict that mixed enterprises have high owner conflict and poor performance – the worst of both worlds. However, more empirical evidence is required to judge whether or not mixed enterprises have the highest inefficiencies.
Chiu et al. [2002] establish a theoretical model with an incentive problem to show that an increase in the government stock share decreases the man-ager’s effort to reduce cost inefficiency. The theoretical model predicts that when the inefficiency caused by owner conflict is sufficiently small, the public, mixed, and private enterprises have the highest, medium, and lowest cost inefficiency, respectively. However, when the inefficiency caused by owner conflict is sufficiently large, the ranking of cost inefficiencies, from the highest to the lowest, becomes mixed, public, and private enterprises. They apply the model of Battese and Coelli [1995] to simultaneously estimate the parameters of stochastic frontier and the inefficiency model. The panel data of 74 manufacturing firms in Taiwan during 1996 – 1997 are used to estimate the translog cost function and the cost inefficiency function. According to their empirical findings, the rankings of cost inefficiencies in Taiwan, from the highest to the lowest, are mixed, public and private enterprises.
There are two approaches used in the empirical analysis of mixed enter-prises, which are performance and efficiency. Before 1994, most empirical ana-lyses focused on evaluating a mixed enterprise’s performance by employing
profit and productivity. For example, Boardman and Vining [1989] consider the 500 largest non-US industrial firms as compiled by Fortune magazine in 1983, including 419 private enterprises, 23 mixed enterprises, and 58 public enterprises. They apply the approach of OLS (ordinary least squares) to estimate the performance of private, mixed, and public enterprises, and find that the performance of mixed and public enterprises is worse than that of private enterprises. Moreover, the profitability, and productivity of mixed enterprises are no better than, and sometimes even worse than, those of public enterprises.
Vining and Boardman [1992] confirm that ownership plays an important role in determining corporations’ technical efficiency and profitability. They randomly chose a sample set of 249 private enterprises, 93 mixed enterprises and 12 public enterprises from 1986 data on 500 non-financial corporations in Canada. The OLS approach is applied for estimation. Their result is that the technical efficiency and profitability of public and mixed enterprises on average are worse than those of private enterprises. Furthermore, the technical efficiency and profitability of public enterprises on average are worse than those of mixed enterprises.
After 1994, researchers changed their interest from performance analysis to efficiency analysis when studying a mixed enterprise. For example, Ehrlich et al. [1994] use a translog cost function model to estimate and make a comparison of cost inefficiency among public, private, and mixed enterprises. They use the panel data of airline companies in 23 countries during the period 1973 – 1983. Both the long-run and short-run cost ineffi-ciencies of private enterprises are lower than those of public and mixed enterprises. In the short run, there is no significant difference in the cost inefficiency of public and mixed enterprises. However, the long-run cost inefficiencies of public enterprises are higher than those of mixed enterprises.
The efficiency of mixed enterprises, especially in mixed banks, has received limited attention. Most existing research on mixed enterprises’ efficiency focuses on cost efficiency, but not on technical and allocative efficiencies. It is well known that public and mixed enterprises achieve socio-political objectives at the price of a higher cost inefficiency [Chiu et al., 2002]. The goal of mixed and public enterprises is neither profit maximisation nor cost minimisation. Therefore, it is not adequate to judge the performance of public and mixed enterprises by cost efficiency. While public and mixed enterprises have a higher allocative inefficiency, because of government regulations on employment and procurement, if cost ineffi-ciency is decomposed into technical and allocative efficiencies, then public and mixed enterprises are still likely to perform better than private enterprises in some aspects.
This article employs parametric approach methodologies to establish a benchmark measure for a bank’s technical efficiency. We will also explain how government shareholding affects a manager’s incentive to offset inefficiencies. The panel data set of 43 banking firms in Taiwan during the period 1997 – 1999 is used to estimate the translog distance function. This article is organised as follows: The next section provides the theoretical model. The third section explains the construction of the model and the data source. The fourth section consists of the empirical results, followed by a concluding section.
THEORETICAL FOUNDATION
Three essential factors should be taken into account to determine the technical efficiency rankings in public, mixed, and private banks: Agency cost, owner conflict, and bureaucratic power. Chiu et al. [2002] establish a principal-agent model to show that an increase in the government stock share decreases the manager’s effort to reduce cost inefficiency. A similar story can be applied to technical efficiency. Denote the government stock share by S with 0 S 1. The manager’s payoff becomes less correlated with the enterprise’s profit as the government stock share increases and hence the manager will input less effort to offset stochastic technical inefficiencies. Therefore, agency cost inefficiency (ACI) is strictly increasing with the government stock share and can be expressed by the following function:
ACI(S) ¼ a1Sa; (1)
with a1. 0,a. 0, and @ACI(S)/@S . 0.
Owner conflict occurs due to different goals of the public and private owners. We use a Cobb-Douglas function to measure the owner conflict inefficiency (OCI) as a function of the government stock share:
OCI(S) ¼ a2Sb(1 S)1b, (2)
with a2. 0 and 0 ,b, . Owner conflict inefficiency first increases and then
decreases with the government stock share. This is because a more diversified ownership structure increases ownership conflict.
Bureaucratic power becomes more important to productivity in a more centralised, constrained, or imperfect economic environment. Tian [1997, 2000] explicitly models the bureaucratic power and degree of market perfec-tion into a Cobb-Douglas producperfec-tion funcperfec-tion. Following Tian’s model,
we may rewrite the bureaucratic power inefficiency (BPI) as a function of government stock share:
BPI(S) ¼ a3(1 r)(1 S)g, (3)
with a3. 0, 0 r1,g. 0. A higher value of parameterrreflects a higher
degree of economic freedom, decentralisation and market perfection. Note that @BPI(S)/@S , 0 withr= 1. In many developing countries, bureaucratic power helps much with procurement, asset acquisition, subsidies, franchises, etc. For instance, Taiwan’s government has a yearly quota to subsidise housing loans to people who work in government or public schools. Private banks are excluded from such loans projects promoted by the government. The eligible people have to go to some public banks to receive these sub-sidised loans, which usually cannot fully satisfy their monetary borrowing requirements. The remaining amount of money required to be borrowed will come from the same public bank at a regular interest rate. These loans are a little less risky since people who work in the public sector are highly
F I G U R E 1
T H E C A S E W I T H A H I G H M A R K E T I M P E R F E C T I O N
credible, mainly because of their steady income. The bureaucratic power has no effect on the technical efficiency at all if the market is perfect (r¼ 1).
Summing up, the total technical inefficiency (TTI) function can be written as: TTI(S) ¼ ACI(S) þ OCI(S) þ BPI(S) þ U
¼a1Saþa2Sb(1 S)1bþa3(1 r)(1 S)gþU: (4)
The variable U is a non-negative random variable with mean U . 0, representing the stochastic technical inefficiency. Therefore, the expected total technical inef-ficiency is:
E(TTI(S)) ¼ ACI(S) þ OCI(S) þ BPI(S) þ E(U)
¼a1Saþa2Sb(1 S)1bþa3(1 r)(1 S)gþ U: (5)
From Equation 5, we find that all kinds of orderings in technical efficiencies among public, mixed, and private enterprises can take place. For example, public enterprises have lower technical inefficiencies versus mixed and private
F I G U R E 2
T H E C A S E W I T H A H I G H M A R K E T I M P E R F E C T I O N A N D A S E R I O U S I N C E N T I V E P R O B L E M
enterprises when the market is imperfect such that bureaucratic power is import-ant to productivity (see Figure 1). Mixed enterprises will have lower technical inefficiencies than others when agency cost is high and bureaucratic power is important to productivity (see Figure 2). Private enterprises will have lower tech-nical inefficiencies than others when agency cost is high and the market is rela-tively perfect (see Figure 3).
EMPIRICAL MODEL
Construction of the Empirical Model
Research on bank efficiency has two major streams: The first stream is the parametric approach, and the other stream is the non-parametric approach. Efficiency can be evaluated from the perspective of three types of inefficien-cies, namely cost, allocative, and technical inefficiencies. Cost inefficiency is also labelled overall inefficiency, representing the gap in the ratio of minimum cost to actual cost. Allocative inefficiency occurs when banks do not employ the least costly combination of inputs to produce output, whilst technical inefficiency refers to the situation arising from a bank’s failure to operate at its efficient production frontier.
F I G U R E 3
T H E C A S E W I T H A R E L A T I V E L Y P E R F E C T M A R K E T
The parametric approach, which is generally concerned with the produc-tion or cost funcproduc-tion base, focuses on an estimaproduc-tion of the funcproduc-tion’s charac-teristics, whilst also undertaking a measurement of the scale of economics, under the assumption that all banks are operating efficiently. Following Farrell’s [1957] introduction of the frontier production function to measure efficiency, many researchers further developed the concept of the stochastic frontier production function [Aigner et al., 1977; Meeusen and Broeck, 1977]. Pitt and Lee [1981] and Schmidt and Sickles [1984] extend the stochastic frontier model to panel data, but they assume that the technical efficiency was invariant for individual firms. The advanced model, proposed by Cornwell et al. [1990] and Battese and Coelli [1992, 1995], allows us to estimate time-varying efficiency levels.
If efficiency varies, then it is natural to seek determinants of efficiency variation. Early researchers applied a two-stage approach to analyse effi-ciency effects in terms of appropriate explanatory variables [Pitt and Lee, 1981]. The first of these stages includes the specification and estimation of the stochastic frontier function and the prediction of the technical (or cost) efficiency. The second stage involves the specification of the regression model for the predicted technical (cost) efficiency. However, this two-stage procedure consists of inconsistent assumptions regarding the identical distribution of efficiency effects in the two estimation stages. Kumbhakar et al. [1991], Huang and Liu [1994], and Battese and Coelli [1995], etc., follow this by adopting a single-stage approach in which explanatory vari-ables are incorporated directly into the efficiency error component.
As discussed above, the public enterprises are required to achieve particu-lar social ends. The behaviour assumptions of profit maximisation and/or cost minimisation are unlikely to be valid in public and mixed banks. Hence, this study focuses only on technical efficiency. Furthermore, banks are multi-output industries. Traditional methods model multi-output tech-nology by a dual-cost function, and this is apparently inappropriate. Distance functions allow us to characterise the structure of production technology when multiple inputs are used to produce multiple outputs without the need to specify a behaviour objective such as cost minimisation or profit maximisa-tion. An output distance function takes an output-expanding approach to the measurement of the distance, which is the maximal proportional expansion of the output vector, given an input vector.1
According to Shephard [1970], the output distance function can be defined as follows: Do(X, Y) ¼ min u: Y u [ P(X) , (6)
where P(X) is the output sets of production technology, describing the sets of output vectors that are feasible for each input vector X. That is,
P(X) ¼ {Y: X can produce Y}: (7)
This gives the minimum amount by which an output vector can be deflated and still remain producible with a given input vector. The output distance function Do(X, Y) is non-decreasing, positively linearly homogeneous and
convex in Y, and non-increasing in X [Kumbhakar and Lovell, 2000]. Note that Do(X, Y ) 1 if Y belongs to the production possibility set of (Y [ P(X))
and that Do(X, Y) ¼ 1 if Y belongs to the frontier of the production possibility
set of X.
The statistical formulation of the output distance function defined in Equations 6 and 7 can be specified as:
Do(X, Y) ¼ f (X, Y,d)ev (8)
Heredis a vector of unknown coefficients to be estimated and v is the random disturbance term intended to capture the measurement error and statistical noise and is assumed to be iid N(0,sv2).
An appropriate functional form f(.) in Equation 8 would ideally be flexible, easy to calculate and permit the imposition of homogeneity. The translog form does satisfy the above criteria and has been used by a number of authors [Lovell et al., 1994; Grosskopf et al., 1996; Coelli and Perelman, 2000].2 The translog distance function with M outputs and J inputs is specified as: ln Don¼a0 XM m¼1 amln ymnþ 1 2 XM m¼1 XM k¼1 amkln ymnln yknþ XJ j¼1 bjln xjn þ1 2 XJ j¼1 XJ h¼1 bjhln xjnln xhnþ XM m¼1 XJ j¼1 lmjln ymnln xjnþvn, n ¼ 1, . . . , N, (9)
where n denotes the nth firm in the sample. The restriction of linear homo-geneity in outputs requires:
XM m¼1 am¼1, XM k¼1 amk¼0, m ¼ 1, . . . , M; and XM m¼1 lmj¼0, j ¼ 1, . . . , J:
Furthermore, the restriction of symmetry requires:
amk ¼akm, m, k ¼ 1, . . . , M, andbjh¼bhj, j, h ¼ 1, . . . , J:
One basic problem in estimating Equation 9 is that the dependent variable ln Donis unobservable. Fortunately, we can solve this problem by imposing
the linear homogeneity in outputs [Fa¨re and Primont, 1995]. That is,
ln Don yMn ¼a0þ X M1 m¼1 amln ymnþ 1 2 X M1 m¼1 X M1 k¼1 amkln ymnln y kn þX J j¼1 bjln xjnþ 1 2 XJ j¼1 XJ h¼1 bjhln xjnln xhn þX M1 m¼1 XJ j¼1 lmjln ymnln xjnþvn, n ¼ 1, . . . , N, (10) where y
i ¼yi/yM. Equation 5 can be rewritten as
ln yMn¼a0þ X M1 m¼1 amln ymnþ 1 2 X M1 m¼1 X M1 k¼1 amkln ymnln y kn þX J j¼1 bjln xjnþ 1 2 XJ j¼1 XJ h¼1 bjhln xjnln xhn þX M1 m¼1 XJ j¼1 lmjln ymnln xjnþvnln Don n ¼ 1, . . . , N: (11)
We then replace the unobservable component 2ln Donby a non-negative
random variable un. The latter is assumed to be independently distributed,
truncated at zero of N(m,s2
u), and independently distributed of vn.
The predicted value of the output distance for the nth firm, ^Don ¼
exp(un), is not directly observable since un only appears as part of the
composed error term, 1n¼vnþun. The conditional expectation of
un, given 1n¼vnþun, can be used to obtain the predicted value of the
output distance function. The output distances would hence be predicted as:
^
Don¼E½exp (un)j 1n: (12)
Technical efficiency can be estimated by using the property that the output distance function coincides with the Farrell output-oriented measure of tech-nical efficiency [Kumbhakar and Lovell, 2000]. Equations 11 and 12 can be estimated by the maximum likelihood method [Coelli and Perelman, 2000].
Data Collection and Choice of Outputs and Inputs
This study uses data from four public enterprises (where government share-holding in an enterprise is 100 per cent), 15 mixed enterprises (where govern-ment shareholding in an enterprise ranges from 0.1 to 99.9 per cent), and 24 private listed companies for the period 1997– 1999, giving a total of 43 bank enterprises in our sample set. The data sources are financial releases and public statements and Taiwan Economic Journal database.
There are three types of banking output: the provision of loan services (including business and individual loans), portfolio investment (mainly government securities and shares, along with public and private enterprise securities), and other real revenues. There are three types of input, namely bank staff, fixed assets, and total deposits. We refer to total bank deposits (NT$ thousand) as being accounted for by current deposits, time deposits and savings deposits. Since the data cover three years, we have deflated some variables, including three outputs, fixed assets and total deposits, by CPI (1996 ¼ 1.00). Table 1 describes the definition and explanation of variables.
This study applies the model proposed by Battese and Coelli [1995] to estimate the parameters of the distance function and the efficiency model
T A B L E 1
D E F I N I T I O N A N D E X P L A N A T I O N O F V A R I A B L E S
Variable Definition
Y1 Real loan services (including business and individual loans) (NT$ billion)
Y2 Real portfolio investment (mainly government securities and shares, along
with public and private enterprise securities) Y3 Other real revenues
L Total number of employees K Real fixed assets (NT$ thousand) M Real total deposits (NT$ thousand) t Time periods subtracting 1996 S Percentage of government shareholding SSQ S S/100
Notes: We divide firms’ Y1, Y2, Y3, K, and M by CPI (1996 ¼ 1.00).
Sources: Financial releases and public statements from each company; Taiwan Economic News Service.
simultaneously. This method helps avoid inconsistent assumptions regarding the identical distribution of efficiency effects in the two-stage approach. The technical efficiency effects are defined by
unt ¼f0þf1t þf2Sntþf3SSQntþvnt, (13)
where unt ¼ln Dont and the random disturbances vnt are assumed to be
independently distributed as truncated at (f0þf1t þf2Sntþf3SSQnt) of
N(0,s2
u). In other words, the non-negative random variable unt is assumed
to be independently distributed as truncations at zero of N((f0þf1t þ f2Sntþf3SSQnt),s2u). The computer software package Frontier 4.1 is used
to estimate the parameters of the distance function and the efficiency model.
EMPIRICAL RESULTS
The empirical results from estimating the output distance function are presented in Table 2, while the underlined parameters are calculated under the homogeneity condition. We apply the likelihood ratio test for separability between inputs and outputs: H0: lmj¼ 0, m ¼ 1, . . . , 3, j ¼ 1, . . . , 3. The
value of the likelihood ratio test is 35.65, which is far away from the critical value 18.5476 (¼x2
(6), 0:005). Therefore, the input – output separablility model is
rejected. This is reasonable since efficiency measures take both outputs and inputs into account. We are unable to judge whether the input usage is efficient without the output vector; and vice versa. The estimated g¼ 0.999 is significantly greater than zero, suggesting that the term untshould be treated
as a random variable.
The estimated coefficients for the technical efficiency function are of particular interest in this study and are presented in Table 3. All estimated coefficients are statistically, significantly different from zero at the one per cent level. The negative coefficient of the time variable suggests that commercial banks in Taiwan, on average, performed worse after the 1997 Asian financial crisis. This result compares to Huang and Wang [2002], in which they find cost efficiencies of Taiwanese banks turned worse during 1982 – 1997.
The quadratic effects of the coefficients of government shareholding on the technical efficiency imply that the technical efficiency increases as the government shareholding in a bank goes higher up to 42.3 per cent, while thereafter it decreases. When the government share is greater than 42.3 per cent, the technical efficiency of a commercial bank then increases with privatisation. However, when the government share achieves 42.3 per cent, its technical efficiency then decreases with privatisation.
The estimated efficiencies of each bank are listed in Table 4. The ranking of overall mean efficiency, from the highest to the lowest, is mixed banks (0.958), public banks (0.953), and private banks (0.926). This ranking is the same as that in 1997 and in 1999, but the situation was different in 1998. The mean efficiency then, from the highest to the lowest, is mixed banks (0.94086), private banks (0.93241), and public banks (0.93058). In summary, mixed banks have the highest level of technical efficiency among Taiwanese commercial banks. Furthermore, technical efficiencies of the
T A B L E 2
E S T I M A T E D R E S U L T S O F T H E O U T P U T D I S T A N C E F U N C T I O N
Variable
Estimated
parameter Standard error t-Value Constant 1.85709 0.81310 2.28396 ln y1 0.63129 0.45302 1.39353 ln y2 0.23401 0.34487 0.67940 ln y3 0.13470 – – ln y1ln y1 0.24123 0.08131 2.96643 ln y1ln y2 20.23522 0.05783 24.06726 ln y1ln y3 2 0.00601 – – ln y2ln y2 0.24913 0.08581 2.90311 ln y2ln y3 2 0.01391 – – ln y3ln y3 0.01992 – – ln L 21.27553 0.26915 24.73910 ln K 1.09694 0.49133 2.23258 ln M 20.87372 0.69310 21.26060 ln L ln L 0.13864 0.08272 1.67593 ln L ln K 1.04354 0.20361 5.12518 ln L ln M 20.56680 0.24370 22.32584 ln K ln K 0.40776 0.25376 1.60687 ln K ln M 21.37187 0.40602 23.37883 ln M ln M 1.32040 0.58638 2.25180 ln y1ln L 20.20805 0.06194 23.35898 ln y1ln K 20.81638 0.18208 24.48377 ln y1ln M 0.70932 0.22550 3.14553 ln y2ln L 20.10255 0.04567 22.24553 ln y2ln K 0.53094 0.23644 2.24558 ln y2ln M 20.20531 0.22937 20.89509 ln y3ln L 0.31060 – – ln y3ln K 0.28544 – – ln y3ln M 0.50401 – – s2¼s v 2þs u 2 0.08229 0.01570 5.24033 g¼su2/s2 0.99993 0.00005 0.21844 105 Significant at 10% level. Significant at 5% level. Significant at 1% level.
Note: Parameters in italics are calculated under the homogeneity condition. N ¼ 129. Log-Likelihood ¼ 214.27866.
(completely or partially) government-owned banks are higher than those of purely private-owned banks.
Most existing research on the efficiencies of mixed enterprises focuses on cost efficiency, but not on technical efficiencies. For example, Vining and Boardman [1992] and Ehrlich et al. [1994] show that private enterprises demonstrate higher cost efficiencies than public and mixed enterprises. However, the behaviour assumption of cost minimisation is unlikely to be valid in public and mixed banks. Indeed, cost efficiency may not be appro-priate for evaluating public, mixed, and private banks. By focusing only on technical efficiencies, we find that mixed banks have higher technical efficien-cies than private and public banks. Public banks also have higher technical efficiencies than private banks. Therefore, mixed ownership helps with improving the technical efficiency of public enterprise.
Our empirical findings show that mixed banks have the highest technical efficiency and private banks the lowest technical efficiency. This may be because Taiwan’s banking markets are not perfect, in which bureaucratic power still plays an important role in improving efficiency. Mixed banks also provide a better incentive scheme than public banks. If the owner conflict is not too severe, then mixed banks benefit from balancing bureaucratic power and the internal incentive scheme.
Before privatisation, public banks already possessed a long-run social reputation and connection. In Taiwan, internal control in private banks is some-times less robust. For example, a few new commercial banks were involved in scandals of embezzlement and risky loans to the bank owners’ family businesses. The government shareholdings provide more internal control measures for these commercial banks. Under the compulsory deposit insurance system in Taiwan, the government is the de facto final insurer in banking markets, helping people feel more confident if the government also owns the bank.
Public and mixed banks on average are larger than private banks. This is because Taiwan’s banking markets were mainly occupied by a few public banks for half a century after World War II. The public and mixed banks now thus enjoy benefits of economies of scale which help improve
T A B L E 3
T H E E S T I M A T E D R E S U L T S O F T H E E F F I C I E N C Y M O D E L
Variable Estimated parameter Standard error t-value Constant 2.15052 0.48837 4.40351
T 20.49712 0.11861 24.19119
S 0.03333 0.00864 3.85992
SSQ 20.03940 0.01027 23.83604 Significant at 1% level.
T A B L E 4 E F F I C I E N C Y A S S E S S M E N T 1997 1998 1999 Bank TE ST (%) TE ST (%) TE ST (%) 1 0.928768 42.25 0.916926 30.93 0.937141 29.41 2 0.93539 61.94 0.938285 36.98 0.994451 36.64 3 0.996354 43.38 0.936543 41.42 0.960024 41.17 4 0.995598 45.07 0.911618 44.16 0.984038 44.16 5 0.959909 1.17 0.960063 1.17 0.955278 0 6 0.999098 0.65 0.980026 1 0.976615 0 7 0.995294 0.02 0.942812 0 0.939225 0 8 0.949654 0.06 0.889374 0.13 0.878086 0 9 0.990624 0.02 0.982668 0 0.99399 0 10 0.99651 0.94 0.985848 0.53 0.977127 0 11 0.99367 2.17 0.97235 2.06 0.99438 0 12 0.997084 59.65 0.989215 59.65 0.995202 45.12 13 0.995544 60.48 0.936399 60.44 0.997162 36.03 14 0.930938 21.72 0.904327 22.73 0.904783 20.25 15 0.78499 0 0.772911 0 0.844461 0 16 0.975687 0 0.94958 0 0.977951 0 17 0.974075 85.71 0.967952 85.71 0.974468 85.71 18 0.992446 0 0.967773 0 0.987002 0 19 0.995694 48.08 0.995412 46.7 0.995119 28.46 20 0.895705 0 0.834995 0 0.90226 0 21 0.940985 0 0.913105 0 0.858113 0 22 0.956894 0 0.918885 0 0.893239 0 23 0.899398 0 0.885602 0 0.925868 0 24 0.837692 0 0.750683 0 0.894997 0 25 0.997411 0 0.99098 0 0.850542 0 26 0.956415 0 0.976833 0 0.974625 0 27 0.976443 0 0.966543 0 0.975642 0 28 0.850051 0 0.828421 0 0.935708 0 29 0.932304 0 0.875092 0 0.868435 0 30 0.992572 0 0.992214 0 0.814178 0 31 0.996191 0 0.952888 0 0.882783 0 32 0.963714 100 0.917523 61.22 0.930933 48.86 33 0.928083 0 0.995649 0 0.92665 0 34 0.9363 0 0.889554 0 0.950693 0 35 0.940881 100 0.838705 100 0.980732 100 36 0.972335 100 0.963233 100 0.472279 100 37 0.959936 100 0.995591 100 0.507995 100 38 0.961033 0 0.943208 0 0.995432 0 39 0.996991 0 0.943636 0 0.981266 0 40 0.978671 0 0.982159 0 0.941432 0 41 0.868218 0 0.992013 0 0.99635 0 42 0.996064 0 0.988512 0 0.972688 0 43 0.917233 0 0.978716 0 0.880588 0 Public Banks 0.96403 – 0.93058 – 0.96387 – Mixed Banks 0.96418 – 0.94086 – 0.96911 – Private Banks 0.94539 – 0.93241 – 0.89991 – Overall efficiency of public banks, 0.95828; of mixed banks, 0.958051; of private banks, 0.926073. Notes: TE ¼ technical efficiency; ST ¼ government stock.
their technical efficiencies. Consequently, in Taiwan the government stock-holdings have positive effects on technical efficiency improvement.
CONCLUDING REMARKS
In the early 1990s, Taiwan began her deregulation trend in order to enhance competition across all industries and to promote economic efficiency. We apply the concept of a stochastic frontier function to evaluate the efficiency of banks and to investigate the relationship between government shareholding and technical efficiency. Banks are in a multi-output industry and traditional methods model the multi-output technology as a dual-cost function. This is, however, an inappropriate setup since the behavioural assumption of cost minimisation is unlikely to be valid in public and mixed banks. Distance functions allow us to characterise the structure of production technology when multiple inputs are used to produce multiple outputs without the need to specify a behaviour objective such as cost minimisation or profit maximisation.
Most existing research on mixed enterprises’ efficiency focuses on cost efficiency. Nevertheless, the goal of mixed and public enterprises is neither profit maximisation nor cost minimisation. It is well known that public and mixed enterprises achieve socio-political objectives at the price of higher cost inefficiency. Therefore, it is not adequate to judge the performance of public and mixed enterprises by cost efficiency. While public and mixed enterprises may have a higher allocative inefficiency due to govern-ment regulations on employgovern-ment and procuregovern-ment, if cost inefficiency is decomposed into technical and allocative efficiencies, public and mixed enterprises are likely to perform better than private enterprises in some aspects.
We apply a translog distance function to estimate technical efficiency and to examine the relationship between technical efficiency and government shareholdings. The main findings are as follows: First, public banks in Taiwan can improve technical efficiencies by mixed ownership at a diminishing rate. Second, mixed banks have a higher technical efficiency than private and public banks. Third, banks in Taiwan on average performed worse after the 1997 Asian financial crisis. Finally, the model hypothesis on input – output separability is rejected.
The loan as an output of the bank is risky. In this article we do not take loan quality into account. However, default loans do reduce a bank’s output and/or increase its cost and hence reduce the estimated efficiency indexes. Introdu-cing loan quality may alter the rankings in estimated efficiency. In the future loan quality can be incorporated into the model, making the estimation more accurate. Furthermore, a comparison between different estimation
approaches, such as stochastic cost frontiers and DEA, has been a worthy direction for research in this field.
ACKNOWLEDGEMENTS
We are grateful to seminar participants at the Taiwan Economic Association Conference, Taiwan Financial Association Conference, and National Sun Yat-Sen University for helpful comments. Authors gratefully acknowledge financial supports from Taiwan’s National Science Council (NSC-89-2415-H-260-004-SSS and NSC-89-2415-H-032-024). The usual disclaimer applies.
NOTES
1. A distance function can be specified with either an input orientation or output orientation. The output distance function is selected over the input distance function, because it would be easier for a public or mixed bank to expand output rather than to reduce the usage of inputs. 2. The Cobb-Douglas functional form, which is one of the most popular functional forms in production analysis, only satisfies the latter two points, because of its restrictive elasticity of substitution and scale property. Moreover, it is not an appropriate model of a firm in a competitive industry since it is not concave in output dimensions [Klein, 1953].
REFERENCES
Aigner, D., C.A.K. Lovell and P. Schmidt, 1977, ‘Formulation and Estimation of Stochastic Frontier Production Function Models’, Journal of Econometrics, Vol.6, pp.21 – 37. Battese, G.E. and T.J. Coelli, 1992, ‘Frontier Production Function, Technical Efficiency and Panel
Data with Application to Paddy Farmers in India’, Journal of Productivity Analysis, Vol.3, pp.153– 69.
Battese, G.E. and T.J. Coelli, 1995, ‘A Model for Technical Inefficiency Effects in a Stochastic Frontier Production Function for Panel Data’, Empirical Economics, Vol.20, pp.325 – 32. Bell, F. and N. Murphy, 1968, Costs in Commercial Banking: A Quantitative Analysis of Bank
Behavior and Its Relation to Bank Regulation. Research Report No. 41, Boston: Federal Research Bank.
Berger, A.N. and D.B. Humphrey, 1991, ‘The Dominance of Inefficiencies over Scale and Product Mix Economics in Banking’, Journal of Monetary Economics, Vol.28, pp.117 – 48. Berger, A.N., G.A. Hanweck and D.B. Humphrey, 1987, ‘Competitive Viability in Banking:
Scale, Scope and Product Mix Economies’, Journal of Monetary Economics, Vol.20, pp.501– 20.
Bhattacharyya, A., C.A.K. Lovell and P. Sahay, 1997, ‘The Impact of Liberalization on the Productive Efficiency of Indian Commercial Banks’, European Journal of Operational Research, Vol.98, pp.332– 45.
Bishop, M., J. Kay and C. Mayer, 1994, ‘Introduction: Privatisation in Performance’, in M. Bishop, J. Kay and C. Mayer (eds), Privatisation and Economic Performance, Oxford: Oxford University Press, pp.1 – 14.
Boardman, A.E. and A.R. Vining, 1989, ‘Ownership and Performance in Competition Environ-ment: A Comparison of the Performance of Private, Mixed, and State-Owned Enterprises’, Journal of Law and Economics, Vol.32, pp.1 – 33.
Boardman, A.E. and A.R. Vining, 1991, ‘The Behavior of Mixed Enterprises’, Research in Law and Economics, Vol.14, pp.223 – 50.
Boardman, A.E., C. Eckel and A.R. Vining, 1986, ‘The Advantages and Disadvantages of Mixed Enterprises’, in A.R. Negandhi, H. Thomas and K. Rao (eds), Multinational Corporations and State-owned Enterprises: A New Challenge in International Business, Research in International Business and International Relations, Greenwich, CT: JAI, pp.221 – 44.
Chen, T.-Y. and T.-L. Yeh, 2000, ‘A Measurement of Bank Efficiency, Ownership and Productivity Change’, Service Industries Journal, Vol.20, No.1, pp.95 – 109.
Chiu, Y.-H., J.-L. Hu and Y. Li, 2002, ‘Cost Inefficiency of Taiwan’s Mixed Enterprises’, in Tsu-tan Fu, Cliff J. Huang and Knox Lovell (eds), Productivity and Economic Performance in the Asia-Pacific Region, Cheltenham: Edward Elgar Publishing, pp.208 – 28.
Coelli, T.J., 1995, ‘Simulators and Hypothesis Tests for a Stochastic: A Monte Carlo Analysis’, Journal of Productivity Analysis, Vol.6, pp.247– 68.
Coelli, T.J. and S. Perelman, 2000, ‘Technical Efficiency of European Railways: A Distance Function Approach’, Applied Economics, Vol.32, pp.1967 – 76.
Cornwell, C., P. Schmidt and R.C. Sickles, 1990, ‘Production Frontiers with Cross-sectional and Time-series Variation in Efficiency Levels’, Journal of Econometrics, Vol.46, pp.185– 200.
Eckel, C.C. and A. Vining, 1985, ‘Elements of a Theory of Mixed Enterprises’, Scottish Journal of Political Economy, Vol.32, pp.82 – 93.
Ehrlich, I., G. Gallais-Hamonno, Z. Liu and R. Lutter, 1994, ‘Productivity Growth and Firm Ownership: An Analytical and Empirical Investigation’, Journal of Political Economy, Vol.102, pp.1006 – 38.
Fa¨re, R. and D. Primont, 1995, Multi-output Production and Duality: Theory and Applications, Boston: Kluwer Academic.
Farrell, M., 1957, ‘The Measurement of Productive Efficiency’, Journal of the Royal Statistical Society: Series A, Vol.120, pp.253 – 81.
Favero, C.A. and L. Papi, 1995, ‘Technical Efficiency and Scale Efficiency in the Italian Banking Sector’, Applied Economics, Vol.27, pp.385 – 95.
Ferrier, G.D. and C.A.K. Lovell, 1990, ‘Measuring Cost Efficiency in Banking: Econometric and Linear Approach Evidence’, Journal of Econometrics, Vol.46, pp.229 – 45.
Grosskopf, S., K. Hayes, L. Taylor and W. Weber, 1996, ‘Budget Constrained Frontier Measures of Fiscal Equality and Efficiency in Schooling’, Review of Economics and Statistics, Vol.79, pp.116– 24.
Huang, T.-H., 1998, ‘Estimating Scale and Scope Economies in Taiwan’s Banking Sector Using a Stochastic Frontier Cost Function (in Chinese)’, Taiwan Economic Review, Vol.26, pp.209– 41.
Huang, T.-H., 2000, ‘Estimating X-Efficiency in Taiwanese Banking Using a Translog Shadow Profit Function’, Journal of Productivity Analysis, Vol.14, pp.225 – 45.
Huang, C.J. and J.T. Liu, 1994, ‘Estimation of a Non-Neutral Stochastic Frontier Production Function’, Journal of Productivity Analysis, Vol.5, pp.171 – 80.
Huang, T.-H. and W.-H. Wang, 2002, ‘Comparison of Economic Efficiency Estimation Methods: Parametric and Non-Parametric Techniques’, The Manchester School, Vol.70, No.5, pp.682– 709.
Hunter, W.C. and S.G. Timme, 1986, ‘Technical Change, Organizational Form, and Structure of Bank Productivity’ Journal of Money, Credit and Banking, Vol.18, pp.152 – 66.
Klein, L.R., 1953, A Textbook of Econometrics, New York: Row Peterson.
Kumbhakar, S.C. and C.A.K. Lovell, 2000, Stochastic Frontier Analysis, New York: Cambridge University Press.
Lovell, C.A.K., S. Richardson, P. Travers and L.L. Wood, 1994, ‘Resources and Functionings: A New View of Inequality in Australia’, in W. Eichhorn (ed.), Models and Measurement of Welfare and Inequality, Berlin: Springer-Verlag, pp.787– 807.
McAllister, P.H. and D. McManus, 1993, ‘Resolving the Scale Efficiency Puzzle in Banking’, Journal of Banking and Finance, Vol.17, pp.389– 405.
Meeusen, W. and J.V. van Broeck, 1977, ‘Efficiency Estimation from Cobb-Douglas Production Function with Composed Error’, International Economic Review, Vol.18, pp.435 – 44. Miller, S.M. and A.G. Noulas, 1996, ‘The Technical Efficiency of Large Bank Production’,
Journal of Banking and Finance, Vol.20, pp.495– 509.
Oral, M. and R. Yolalan, 1990, ‘An Empirical Study on Measuring Operating Efficiency and Profitability of Bank Branches’, European Journal of Operational Research, Vol.46, pp.282– 94.
Pitt, M.M. and L.F. Lee, 1981, ‘The Measurement and Sources of Technical Inefficiency in the Indonesian Weaving Industry’, Journal of Development Economics, Vol.9, pp.43 – 64. Resti, A., 1997, ‘Evaluating the Cost Efficiency of the Italian Banking System: What Can Be
Learned from the Joint Application of Parametric and Non-Parametric Techniques’ Journal of Banking and Finance, Vol.21, pp.221 – 50.
Rhoades, S.A., 1993, ‘Efficiency Effects of Horizontal (in-Market) Bank Mergers’, Journal of Banking and Finance, Vol.17, pp.411– 22.
Schmidt, P. and R.C. Sickles, 1984, ‘Production Frontier and Panel Data’, Journal of Business and Economic Statistics, Vol.2, pp.367 – 94.
Shephard, R.W., 1970, Theory of Cost and Production Functions, Princeton, NJ: Princeton University Press.
Sherman, H.D. and G. Ladino, 1995, ‘Managing Bank Productivity Using Data Envelopment Analysis (DEA)’, Interfaces, Vol.25, pp.60 – 73.
Tian, G., 1997, Breaking Up is Hard to Do: The Theory of Property Rights in Transitional Economies, Tamkang Chair Lectures Series 114, Taipei.
Tian, G., 2000, ‘Property Rights and the Nature of Chinese Collective Enterprises’, Journal of Comparative Economics, Vol.28, pp.247 – 68.
Vining, A.R. and A.E. Boardman, 1992, ‘Ownership versus Competition: The Causes of Government Enterprise Inefficiency’, Public Choice, Vol.73, pp.205 – 39.
