Final Version for Developing Economies (SSCI)
Ownership and Non-performing Loans:
Evidence from Taiwan’s Banks
*Jin-Li Hu, Yang Li, and Yung-Ho Chiu
We first derive a theoretical model to predict that the relation between non-performing loan ratios and government shareholdings can be downward-sloping, upward-sloping, U-shaped, and inversely U-shaped. An increase in the government‟s shareholding does
facilitate political lobbying. On the other hand, the private shareholding will induce more non-performing loans to be manipulated by corrupt private owners. We adopt a panel data set with 40 Taiwanese commercial banks during 1996-1999 for empirical analysis. The rate of non-performing loans decreases as the government shareholding in a bank goes higher (up to 63.51 percent), while the rate thereafter increases. Banks‟ sizes are negatively related to the rate of non-performing loans. Rates of non-performing loans are shown to steadily increase from 1996 to 1999. Banks established after deregulation, on average, have a lower rate of non-performing loans than those established before deregulation.
Key words: mixed ownership, political lobbying, corrupt private sector, immature civil
society, Asian financial crisis, risky output
JEL classification: C22, C51, L33
Correspondence: Jin-Li Hu, Institute of Business and Management, National Chiao-Tung
University, 4F, 114, Sec. 1, Chung-Hsiao W. Rd., Taipei City 100,
Taiwan. E-mail: firstname.lastname@example.org; URL:
http://www.geocities.com/jinlihu; FAX: 886-2-23494922.
* Jin-Li Hu is Associate Professor, Institute of Business and Management, National Chiao-Tung University. Yang Li is Associate Professor, Department of Applied Economics, National University of Kaohsiung. Yung-Ho Chiu is Professor, Department of Economics, Soochow University. The theoretical part is from a research project by the first author, which is financially supported by Taiwan‟s National Science Council (NSC-90-2415-H-032-006). We are indebted to Hung-Ju Chen, Hsiao-Ru Chen, Wei-Kai Chu, Shih-Rong Li,
Fumiko Takeda, Leonard F.S. Wang, and Ya-Hwei Yang for helpful comments. The authors are also grateful to seminar participants at the Taiwan Economic Association Meeting, the Conference on Asian Financial Crisis, IV, and National University of Kaohsiung. The usual disclaimer applies.
Loans are the major output provided by banks, but loan is a risk output - there is always an ex ante risk for a loan to finally become non-performing. Non-performing loans (NPLs) can be treated as undesirable outputs or costs to a bank, which will decrease the bank‟s performance (Chang 1999). The risk from NPLs mainly arises when the external economic environment becomes worse off such as during economic depressions (Sinkey and Greenawalt 1991). Since the 1997 Asian financial crisis, NPLs have been swiftly accumulating in many Asian economies (Chang 1998; Lauridsen 1998; Robison and Rosser 1998; Wade 1998). Controlling NPLs is hence very important for both an individual bank‟s overall performance (McNulty et al. 2001) and an economy‟s financial environment.
The 1997 Asian financial crisis has had a great impact on Taiwan. Listed companies are now reportedly facing a series of their own financial crises. As a result, stock collaterals pledged to banks have depreciated steeply due to a drastic fall in local stock prices caused by the financial crisis. At the same time, the real estate prices have also dropped, with many investors facing a hard time with their own insolvency problems. As a result, the NPL ratio of Taiwan‟s financial institutions jumped from 4.18 percent at the end of 1997 to 7.48 percent in June 2002.
As Economist reported (Nov. 11, 2000), bad loans among Taiwan‟s domestic banks
rocketed to new highs and a local financial crisis is very imminent. The New York Times
(Dec. 5, 2000) and Business Week (Dec. 11, 2000) cited Salmon Smith Barney that the ratio
of NPLs among listed banks in Taiwan amounted to more than 6 percent, and because of the narrow definition in official NPLs statistics, it could in reality be as high as between 10 to 15 percent. On December 6, 2000 Standard & Poors also revised its outlook on Taiwan from „stable‟ to „negative.‟ According to official statistics by Taiwan‟s Bureau of Monetary Affairs, the NPL ratio in Taiwan is in fact rising very fast. Many researchers are warning that Taiwan may have a looming banking crisis (Montgomery 2002).
Most existing literature finds that state-owned banks are vulnerable to political lobbying and administrative pressure, resulting in a higher NPL ratio. Walter and Werlang (1995) find that state-owned financial institutions underperform the market, because their portfolios concentrate on NPLs indebted by the state. They take Brazil and Argentina as examples. Jang and Chou (1998) adopt the ratio of non-performing loans to total loans as the measure of risk. They then use 1986-1994 data of 13 Taiwanese banks for an empirical study. The average risk-adjusted cost efficiency of the four provincial government-owned banks is the lowest among the sample banks.
The famous Coase Theorem says that the assignment of property rights (ownership) will not affect economic efficiency as long as the transaction cost is zero (Coase 1960; Cheung 1968, 1969). However, the real world is imperfect and the transaction cost can be sufficiently high. In an imperfect world with high transaction costs, ownership does matter to economic efficiency, making different ownership types be associated with different transaction costs (Cooter and Ulen 2000). In this case, we can change the conduct and the corresponding performance by changing ownership (Stiglitz 1974, 1998). Therefore, privatization may help a bank resist political lobbying and administrative pressure and hence reduce its politics-oriented loans.
After the Conservative Party led by Margaret Thatcher won the 1979 election, the U.K. started to privatize its public enterprises with full effort. The privatization experience there has since become an example followed up by many developed and developing countries. One of the main objectives of privatization is to improve the efficiency of a public enterprise (Bishop et al. 1994). Most countries fulfill privatization through the transfer of ownership, but during the process of privatization, 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 participation.” In the early 1990s, Taiwan began to
pursue privatization of its own public enterprises in order to enhance competition and economic efficiency across all industries.
Deregulation in Taiwan‟s banking industry consists of two major aspects: Privatization of public enterprises and entrance opportunity. During the past 12 years, 9 state-owned banks have been privatized, 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. Taiwan‟s government in 1991 released the Commercial Bank Establishment
Promotion Decree in order to relieve the legal entrance barriers to its banking markets. Twenty-four new commercial banks were established afterwards, bringing the total number of domestic commercial banks in Taiwan to forty-eight by 2002. Taiwan‟s government is still
trying to make its banking markets more competitive for public, mixed, and private banks.
In an imperfect (but real) world, the public ownership may help improve a bank‟s performance. Bureaucratic power here becomes more important to productivity in a more centralized, constrained, or imperfect economic environment. Tian (2000) explicitly models bureaucratic power and degree of market perfection into a Cobb-Douglas production function.
His model predicts that in an imperfect economic environment a mixed enterprise maximizes social surplus by balancing the bureaucratic procurement power and the manager‟s incentive.
The major goal of a private enterprise is of course profit maximization. However, for public enterprises, profit maximization is never the primary goal. Public enterprises are required to achieve particular social ends, such as reducing the unemployment rate, promoting economic development, etc. Most governments set up mixed enterprises so as to combine the economic efficiency of private enterprises with a socio-political goal of public enterprises.
Eckel and Vining (1985) provide the first step towards analyzing mixed enterprises‟ performance. They suggest that there are 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 investment from the government 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 enterprises, while the third advantage is flexibility whereby mixed enterprises achieve both profitability and social goals more efficiently 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) further 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 further point out (P. 225): “Different ownership conditions affect the extent to which
mixed enterprises engage in profit maximization, socio-political goal maximization, and managerial utility maximization (or a combination). They also affect the degree of conflict between one owner and another, and between an owner and management.” They predict that mixed enterprises have more 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.
Corruption is not unusual in many countries. According to the Global Corruption
Report, annually investigated and reported by Transparency International (2003), corruption is still a worldwide phenomenon, especially in developing countries. People pay bribes to buy licenses, jobs, and votes, and to reduce taxes as well as for lenient enforcement, etc. (Tullock
1996). Bribery does take place in a corrupt society and as Lui (1996) summarizes, corruption has three important aspects: (a) It is a rent-seeking activity induced by deviation from the perfectly competitive market. (b) It is illegal. (c) It involves some degree of power. With the existence of corruption, the market is no longer perfectly competitive.
The public sector is absolutely not the only corruptible sector in society, because the private sector can also be corruptible. In many developing countries, the civil society is still immature and it is a long way from learning to achieve a lifestyle of democracy and rule of the law (Finkel et al. 2000; Johnson and Wilson 2000). People are not used to legally
contracting and democratic decision-making. As a result, the private sector also resorts to informal connections and illegal means for seeking economic rents. In this case, 100% privatization of a public bank may not be able to decrease its NPL ratio. For example, in Taiwan many financial institutions manipulated by families and/or local political factions have higher rates of NPL ratios. In this case, government shareholding may help complement their weak internal control.
We will explain how government shareholding affects civil corruption and lobbying and hence NPL ratios. A panel data set of 40 banking firms in Taiwan during the period 1996-1999 is used for estimation. This paper is organized as follows: Section II provides the theoretical model. Section III consists of the data source, econometric modeling, and empirical results. Section IV concludes this paper.
II. The Theoretical Model
Three essential factors should be taken into account to determine the NPL ratio and ownership: political lobbying, civil corruption, and joint ownership. Interest groups engage in political lobbying in order to affect administrative decisions. The state-owned banks monitored by both the administrative and legislative branches are more vulnerable to political lobbying than private banks. In a country with a corrupt private sector, private banks easily become a family-owned business, illegally supplying risky loans to enterprises controlled by the same family. Mafias and local political factions can also control financial institutions for illegal money laundering and for money borrowing. Interaction between public and private owners can also affect the loan quality: If they check and balance each other, then the default risk can be reduced. However, if they collude with each other, then the default risk will be increased.
In our model, there is a bank under S (0 ? S ? 1) portion of the government
shareholdings. A bank makes loans to either the public sector or the private sector (or both). Therefore, for any bank the sum of loans ratio to public and private sectors must be exactly one. In every society with limited loans, public and private sectors compete for bank loans. Without losing generality, we assume that initially these two sectors equally split the loans of a bank.
The public sector puts political pressure on this bank, in order to gain loans so as to fulfill policy targets or to save enterprises with good political connection. The extra loans ratio gained by political lobby is B. The extra benefit of politically-gained loans ratio to the
public sector is RB, where the parameter R > 0 represents the marginal benefit of the public ?
sector to increase its loans ratio.
Political lobbying becomes more effective in obtaining a loan as the government share increases. It is reasonable to assume that there is a marginally increasing political lobbying cost function. Without loss of generality, the political cost function can be expressed as ？！2(1；S)B. The parameter ？ is strictly positive and a higher ？ corresponds to a higher 2
difficulty in political lobbying. The effectiveness of political lobbying is strictly increasing with the government stock share, with the parameter ！ > 0, while the political lobbying cost is
marginally increasing with the private stock share and gained loans.
In a corrupt civil society, internal control decreases with the government stock share. That is, in a society that lacks civil self-discipline, government regulation may help complement the deficiency in a bank‟s internal control. The extra loans ratio here gained by civil corruption is b. The extra benefit of loans gained through civil corruption to the private sector is r?b. The parameter r > 0 is the marginal benefit of the private sector to gain extra loans.
Private corruption becomes more effective in obtaining loans as the private stock share increases. Without loss of generality, the civil corruption cost function can be expressed as ；，2Sb. The parameter ； is strictly positive and a higher ； corresponds to a higher difficulty 2
in gaining loans through civil corruption. The civil corruption cost is marginally increasing with the government stock share and gained loans, where the parameter ， > 0.
11The loans ratios gained by the public and private sectors are + B ； b and + b ； 22
B, respectively. This is a lobby game between the public and private sectors to gain extra
loans. The strategic interaction between the public and private sectors will affect the equilibrium loan ratios: Both sectors will evenly split the loans under equal (or zero) efforts to gain extra loans. A sector with a relatively higher effort will gain a higher loan ratio.
Therefore, the net benefit of the government concerning this bank is
1？！2G(B) = R?( + B ； b) ； (1；S)B. (1) 22
Benefit of Political Lobbying Cost of Political Lobbying
The net benefit of the private sector concerning this bank is
；1，2g(b) = r?( + b ； B) ； Sb (2) 22
Benefit of Civil Corruption Cost of Civil Corruption
The expected net benefit maximization problems of the public and private sectors concerning this bank are
Max1？！2 E[G(B)] = R?(+B；b) ； (1；S)B, (3) B22
Max；1，2 E[g(b)] = r?(+b；B) ； Sb. (4) b22
We first solve the two sectors‟ net benefit maximization problems and obtain equilibrium
extra loans gained by political lobbying and civil corruption from this bank:
(，Rr**[B, b] = . (5) ,)?！，(1S)S；；？?：
！，The second-order conditions are ；？(1；S) < 0 and ；；S < 0, which always hold under our
*parameter setup. Note that B strictly increases with R, but strictly decreases with ？. Note
2?B(S)?B(S)RR；！；1；！；2that = > 0 and = > 0. That is, B ！(1；S)！(1;！)(1；S)2?S？？?S
is a strictly convex function of S for all ！ > 0. As long as the increasing marginal costs (the
decreasing returns) assumption is imposed, then B is a strictly convex function of S. Similarly,
2?b(S)rr?b(S)；，；1；，；2we also have = > 0 and = > 0. That is, ，S，(1;，)S2?(1；S)；；?(1；S)
b is a strictly convex function of (1；S) for all ， > 0. As long as the increasing marginal costs
(the decreasing returns) assumption is imposed, then b is a strictly convex function of (1；S).
Not every case of loans gained by lobbying will necessarily turn out to be non-performing, but part of these lobbying-gained loans do turn into non-performing: A proportion ， of loans to the public sector and a proportion ~ of loans to the private sector
will become non-performing, making up the total amount of non-performing loans. The variable U is a non-negative random variable with the mean > 0, representing the U
stochastic non-performing loan ratio. Therefore, this bank‟s NPL ratio caused by political
1?( + B ； b) and the non-performing loans ratio caused by civil lobbying (NPPL) are ，2
1corruption (NPCC) is ?( + b ； B). Moreover, there is a joint ownership effect on this ~2
，1；，bank‟s NPL ratio (NPJO): ：?S(1；S), with 0 < ， < 1. The coefficient ： is positive if two
sectors act collusively to obtain loans, and is negative if two sectors check and balance each other. Note that the joint ownership effect becomes zero if the bank is purely public (S = 1)
or purely private (S = 0).
To sum up, we can express this bank‟s total non-performing loans ratio (TNPL) function
111；，，TNPL = ，?(+B；b) + ~?(+b；B) + ：?S(1；S) + U 22
= NPPL + NPCC + NPJO + U. (6)
This bank‟s expected total non-performing loans ratio is
11，1；，E(TNPL) = ，?(+B；b) + ~?(+b；B) + ：?S(1；S) + U22
= NPPL + NPCC + NPJO + . (7) U
Substituting (5) into (7), we can express this bank‟s expected non-performing loans ratio can
be explicitly expressed as a function of the government stock share:
，1；，1E(TNPL(S)) = (，+~) + (，；~)?[B(S) ； b(S)] + ：?S(1；S)+. (8) U2
?b(S)?[B(S)；b(S)]?B(S)Note that = + > 0; that is, without the joint ownership ?(1；S)?S?S
effect, the total nonperforming loans ratio is strictly increasing [decreasing] with government‟s shareholdings if ， ； ~ > [<] 0.
Figures 1 to 6 depict the relation between the government‟s stock share and the TNPL
ratio from Equation (8). Without the joint ownership effect, the expected NPL ratio will be strictly increasing (when ， ； ~ > 0) or decreasing (when ， ； ~ < 0) with the government‟s
stock share: The first case is when the public sector has a higher NPL ratio and the associated NPL ratio ranking is: public, mixed, and private banks (see Figure 1). The second case is when the civil sector has a higher NPL ratio and the associated TNPL ranking is:
private, mixed, and public banks (see Figure 2). In both cases, our theoretical model predicts
that a mixed bank on average will have the medium TNPL and the total NPL ratio is either
upward or downward-sloping in government stockholdings.
Insert Figure 1 to 6 here
Without the joint ownership effect, the expected NPL ratio will be strictly increasing or decreasing with the government‟s stock share: The first case is when the public sector has a
higher NPL ratio and the associated NPL ratio ranking is: public, mixed, and private banks (see Figure 1). The second case is when the civil sector has a higher NPL ratio and the associated TNPL ranking is: private, mixed, and public banks (see Figure 2). In both cases, our theoretical model predicts that a mixed bank on average will have the medium TNPL and
the total NPL ratio is either upward or downward-sloping in government stockholdings.
When the joint ownership effect on the NPL ratio is negative and its magnitude is sufficiently large, then a mixed bank may have the highest NPL ratio, and the relation between the NPL ratio and government shareholdings is U-shaped (Figures 3 and 4). That is, the mixed bank ownership then minimizes the NPL ratio by balancing the political lobbying pressure and civil corruption. When the joint ownership effect on the NPL ratio is positive and sufficiently large, then a mixed bank may have the highest NPL ratio, and the relation between the NPL ratio and government shareholdings is inversely U-shaped (Figures 5 and 6). That is, the mixed bank ownership then maximizes the NPL ratio by suffering the collusion between the public and private owners. From the above discussion, we obtain the following propositions:
[Proposition 1] Without the joint ownership effect, a bank’s NPL ratio is strictly decreasing
or increasing with respect to government shareholdings.
[Proposition 2] If the joint ownership effect on NPL ratios is negative and its magnitude is sufficiently large, then the relation between a bank’s NPL ratio and government
shareholdings is U-shaped.
[Proposition 3] If the joint ownership effect on NPL ratios is positive and sufficiently large, then the relation between a bank’s NPL ratio and government shareholdings is inversely
III. Empirical Analysis
Our data set consists of 40 Taiwanese commercial banks (all established before 1996) during the period of 1996-1999. In 1996 this data set consists of 4 public commercial banks (where the government‟s shareholding in a bank is almost 100%), 10 mixed commercial banks (where the government‟s shareholding in a bank ranges from 1%-99%), and 26 private
commercial banks, giving a total of 40 commercial banks in our sample set. Because Taiwan‟s government continued the process of privatization, by the end of 1999 there were 2 public commercial banks, 10 mixed commercial banks, and 28 private commercial banks. The data sources are financial releases and public statements and Taiwan
Economic News Service reports.
When analyzing the panel data, ordinary least squares (OLS) estimators may be inconsistent and/or meaningless if heterogeneity exists across firms (Hsiao 2002). The fixed- and random-effects models can take into account the heterogeneity across firms by allowing variable intercepts. The choice among these three models is based on some statistical tests: F-test (the OLS model vs. the fixed-effects model), LM test (the OLS model vs. the
random-effects model), and the Hausman test (the random-effects model vs. the fixed-effects
model). We will employ these three tests to choose the best model to perform our empirical analysis. The dependent variable is the rate of NPLs of commercial banks.
As shown by our theoretical model, state-owned banks monitored by both the administrative and legislative branches are easily distorted by interest groups, which engage in heavy political lobbying. The government share may hence be positively related to the rate of non-performing loans. However, private banks in the corrupt private sector can easily become family-owned businesses, which may supply risky loans to enterprises controlled by the same family. This indicates that private banks might possibly have higher rates of non-performing loans. The joint ownership effect depends on whether or not the two types of owners check and balance each other. These three effects suggest that a downward-sloping, upward-sloping, U-shaped, or inversely U-shaped effect may exist for government shareholding on the NPL ratio. In other words, mixed banks might have the highest, medium, or lowest rate of non-performing loans. We will hence include the linear and quadratic terms of government shareholding in the empirical model. Coefficients of the linear and quadratic terms can be used to check the effects of government shareholding on the