Industrial and Corporate Change, Volume 19, Number 6, pp. 20412072 doi:10.

1093/icc/dtq060 Advance Access published November 4, 2010

Causes for changing performance of the business groups in a transition economy: market-level versus firm-level factors in Chinaz
Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Bong-Kyo Seo*, Keun Lee** and Xiaozu Wangy

This article investigates longitudinal changes in the relative performance of business groups utilizing data on listed companies in China over a 10-year period (19942003). Using a measure of firm value in the stock market and panel regression methods, this article finds the initially superior and eventually worsening performance of group-affiliated firms compared with stand-alone firms. To explain the downward performance, this article considers several alternative hypotheses, namely, institutional development, increasing competition, diversification discount, agency costs from state-ownership, and agents asset diversion behavior. This article has found certain differences in the explanatory power of each hypothesis. While the institutional development hypothesis is somewhat weak, the increasing discount for unrelated diversification as well as serious agency costs revealed in asset diversion in the business groups can better explain the longitudinal decrease in the performance of business groups. We find that while diversification still creates values, its marginal contribution has decreased over time, and that while the state-ownership variable negatively affects the values of firms in general, it is not the cause of the worsening valuation of business groups.

*Bong-Kyo Seo, Department of Chinese Studies, Dongduk Womens University, Wolgok-dong 23-1, Seoul, Korea. e-mail: sbongk@hanmail.net **Keun Lee, Department of Economics, Seoul National University, Gwanak-ro 599, Seoul, Korea. e-mail: Kenneth@snu.ac.kr Xiaozu Wang, School of Management, Fudan University, 670 Guo Shun Road, Shanghai, China. e-mail: wangxz@fudan.edu.cn
z

Earlier versions of this article were presented in several occasions, including the 2007 ACE Conference held in Hong Kong in December 2007 and the 2005 Convention of the East Asian Economic Association held in Beijing in November 2005.

The Author 2010. Published by Oxford University Press on behalf of Associazione ICC. All rights reserved.

2042

B.-K. Seo et al.

1. Introduction
The prevalence and growth of business groups is an important phenomenon in emerging economies. A recent survey in the Journal of Economic Literature (Khana and Yafeh, 2007) indicates that this has become one of the most important issues in academic research. Business groups are defined as the collections of firms bound together in some formal and/or informal way and by an intermediate level of binding; that is, they are neither bound merely by short-term strategic alliances nor legally consolidated into a single entity (Granovetter, 1995). Based on the transaction cost economics of Coase (1937) and Williamson (1975), business groups are considered to be responses to market failures in emerging economies (Leff, 1978; Goto, 1982). This view predicts that once market institutions mature, business groups eventually disappear. Consequently, a world of more homogenous firms is conceived as the result. Several studies prove this long-term decline and eventual homogenization of firms thesis. Khanna and Palepu (2000b), Lee et al. (2008), and Zattoni et al. (2009) provided empirical verifications on the decline in the performance of business groups in Chile, Korea, and India, respectively; and attributed the decline in grouping benefits to the development of market institutions. However, more recent-studies express serious doubts about the long-term decline of business groups and point out their continuing survival and evolution (Choo et al., 2009; Lee and He, 2009; Lee et al., 2010). Korean business groups showed serious declining and worsening performance in the 1990s, as shown in Lee et al. (2008), but they underwent thorough restructuring during the financial crisis in the late 1990s and early 2000 and have been showing a stunning turnaround in the 2000s. Choo et al. (2009) and Lee et al. (2010) argue that enhanced technological capabilities and correction of the early shortcomings, such as over-investment, over the decades caused the turnaround in the post-crisis period. The rebounding of the performance of Korean business groups or Chaebols is an example of continuing evolution and survival despite radical external opening and the maturation of market institutions caused by the post-crisis reforms imposed by the International Monetary Fund (IMF). This finding is consistent with a resource-based view (Penrose, 1959) of business groups (Amsden and Hikino, 1994; Guillen, 2000), where group structures are regarded as the results of the dynamic accumulation of certain capabilities by repeated entry into new businesses, such as project execution capability. The foregoing account underscores the importance of the study of business groups as it touches upon the heterogeneity of firms in the literature, which dates back to Nelson (1991) and a recent re-visitation of Nelson (2009). While the traditional view focuses on market institutions acting as a force for selection and thus homogenization, reality shows that heterogeneous forms of firms, such as business groups, continue to survive as they adapt themselves to the changing business environment. While business groups are heterogeneous compared with the

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Causes for changing performance in a transition economy

2043

Anglo-Saxon style of specialized firms, we also see heterogeneity within business groups observed across countries in the world. While the majority of business groups are owned or managed by families, business groups in post-reform China are state-owned and managed by hired managers. Their origins are also different. They have emerged not only as a response to market failures but also as an initiative of the government (Keister, 2000; Lee and Jin, 2009). In other words, business groups in China have their own unique characteristics that reflect the socio-economic context of China while sharing some characteristics with those in other countries. These features of Chinese business groups make this subject interesting. Thus, this article takes the case of business groups in post-reform China to explore the issue of firm heterogeneity in terms of two different forces (external market environments versus firm-level behavior and strategies) shaping their evolution and changing performance. Business groups have emerged in China since the mid-1980s due to the reform and restructuring of state-owned enterprises (SOEs), which aimed at increasing scale economies and specialization (Keister, 2000; Seo, 2006; Lee and Jin, 2009). In the last two decades, many of these business groups have succeeded in becoming major players in the Chinese and world economy. The Chinese governments, both at the central and the local levels, have played important roles in the formation and development of business groups (Hahn and Lee, 2006; Lee and Jin, 2009). In terms of ownership, the state still holds the dominant position as the controller of business groups in China (Hahn and Lee, 2006). However, recently, non-government ownership has been rapidly growing (Lee and Kang, 2010). Business groups of private and other ownership forms account for $45% of a total of 2856 business groups in 2007, although their share in terms of sales is still 520%. They are also diversified, but they seem to be less diversified compared with their counterparts in other countries. In addition, a recent refocusing tendency is reported by some studies (Lee and Woo, 2002). Ownership structure of business groups in China is different from that in Korean Chaebols, which have complicated matrices of equity ownership among affiliates. Instead, similar to Italian business groups, they are taking a pyramid shape with several vertical tiers (Zattoni, 1999). While there are a number of studies on business groups in China, most are not adopting econometric analysis, except Lee and Jin (2009) which tests the origins of business groups in China. This study is a first of its kind to explain changing performance of business groups in China, with its focus on the comparison of relative explanatory power of environmental (market) and firm-level factors. This article finds the initially superior and then declining performance of business groups in China, which is similar to those observed in some emerging economies. Subsequently, it tries to explain the dynamic change in the performance of business groups from the perspective of several contending hypotheses. One of our contributions is to find that, contrary to the existing views emphasizing the importance of market institutions or external environment, firm-level factors are more important in

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2044

B.-K. Seo et al.

explaining performance change of business groups in China, which is consistent with the similar finding by Choo et al. (2009) in the case of Korean business groups. The market failure or institutional development hypothesis perceives that business groups internalize market failures to overcome imperfections in capital, labor, raw materials, components, and technology markets in emerging economies (Leff, 1978; Goto, 1982). This hypothesis also claims that the performance of business groups will decline as market institutions develop further. This article investigates whether the institutional development hypothesis can also explain the changes in the stock market performance of listed business groups in China. This article also considers another factor representing the external environment, that is, the increased level of market competition in China. This article finds that the traditional hypothesis and emphasis on external factors (institutional development or market competition) is not sufficient to fully explain the changing performance of groups. Thus, alternatives are sought at the firm level, such as agency costs and diversification behavior. We find that the increased discount for diversification as well as the asset tunneling/ diversion hypothesis can better explain the longitudinal decrease in performance of business groups. The next-section discusses several theoretical views and hypotheses on business groups, with an explicit focus on the comparison of market- versus firm-level factors. Section 3 examines the longitudinal performance or market valuation of group firms and non-group firms. Section 4 presents the idea of measuring variables in the key hypotheses and of testing the hypotheses by regression models. Section 5 provides the results of the regression analysis and some interpretation of the results in terms of the institutional features of the Chinese economy. The article concludes with Section 6 summarizing the findings and discussing the implications.

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2. Theoretical perspectives on the performance change of business groups

2.1 Impacts of the market-level factors
2.1.1 Market/institutional development hypothesis There are several theoretical perspectives on business groups. Market failure or the institutional development hypothesis is the oldest and is especially good at explaining the existence or emergence of business groups in late-comer economies. This view perceives that business groups internalize market failures to overcome imperfections in capital, labor, raw materials, components, and technology in emerging economies (Leff, 1978; Goto, 1982). When the market is inefficient, firms must pay the costs associated with using the inefficient market intermediaries. In the absence of efficient intermediaries, internal transactions within groups are more advantageous compared with market transactions (Khanna, 2000; Khanna and Palepu, 1997, 1999, 2000a).

Causes for changing performance in a transition economy

2045

The hypothesis also posits that the performance of business groups will decline as market institutions develop further. As market institutions evolve, the benefits of grouping in overcoming market failures or in reducing transaction costs in emerging economies decrease. The development of institutions also entails the maturation of market mechanisms characterized by a gradual emergence of intermediaries and a gradual reduction of ambient transaction costs. Therefore, it becomes increasingly difficult for business groups to create value by running internal labor, capital, raw material, and technology across their affiliates. In their regression models, Khanna and Palepu (2000b) described the decline in grouping benefits associated with non-diversification as a result of the development of market institutions. This market institutional development hypothesis is also supported by longitudinal performance changes in the diversified expansion of conglomerates in advanced economies. In advanced economies, diversified conglomerates were regarded as more efficient than stand-alone units in the 1960s and 1970s. However, in the 1980s, the market valuation for the diversified expansion strategy of conglomerates changed, with market investors considering the strategy as destructive to the values of firms (Lang and Stulz, 1998). Based on the recent Western experience, it is often suggested that business group strategies in emerging economies also destroy the value of firms (Granovetter, 1995). If the benefits of grouping strategy arise only from overcoming market imperfections in emerging economies as the transaction cost theory suggests, then the longitudinal performance changing from premium to discount would be inevitable. The problems facing large business groups in East Asia, such as Chaebols, after the financial crisis in 1997 are widely seen as vindications of the view that the grouping strategy of emerging economies is wrong (Smyth, 2000). The Chinese economy can provide a good example in testing the longitudinal changes in the relative performance of business groups and the factors affecting the changes arising from the development of market institutions. China experienced rapid transition from a centrally planned economy to a decentralized market economy. The transition of the Chinese economy can be considered as the process of increasing the degree of market determination of economic outcomes, that is, marketization. Market intermediary institutions have rapidly developed as the economy evolved from a centrally planned economy to a market economy. The evolution has been very successful, and the market has rapidly become more competitive. Thus, we can hypothesize that the following: as market institutions become mature over time in China, the market valuation performances of business groups should decrease. The development of market institutions in China from 1994 to 2003 is very remarkable. However, the degrees of development in different industry segments are not identical. The Chinese government or State Council seems to favor SOEs in the protected sectors during selection decisions because these firms are State monopolies and operate under the direct supervision and control of the State

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2046

B.-K. Seo et al.

Council. Therefore, the firms in these industry segments receive direct control from government sectors and have less freedom to set prices. In addition, they act more as politicians or bureaucrats rather than as businesspeople (Shirai, 2002). Khanna and Palepu (2000b) overlooked the differences in the market institution development among industry segments in their models, as they constructed an aggregate index for the national level and not for the sector level. Industry effects can be the major determinants of firm success; hence, this study defines and measures the market institution development by industry segments. This method is more sensitive to industry differences in the development of institutions. 2.1.2 Impact of increasing market competition The Chinese economy experienced rapid transition from a centrally planned economy to a decentralized market economy. Through the reforms the market has rapidly become more competitive. Increasing market competition in China can be traced to three origins. First, market-oriented reform since 1978 transformed the planned economy into a decentralized market economy characterized by excess supply rather than by supply shortages as in the past. In the Chinese case, buyers markets emerged in the mid-1980s at different paces depending on the industry (Byrd, 1987). In the 2000s, excess supply problems extended to almost all industry sectors. Second, because of the progress in national economic integration or market-driven integration in the domestic economy, the notorious provincial protectionism has been reduced. In the mid-1990s, provincial protectionism had been reduced to insignificant levels (Hahn and Lee, 2006). Third, an increasing degree of market competition in China is driven by the new and strong entry of private and foreign direct investment (FDI)-backed foreign companies. It is also important to note that China entered WTO in 2002. In the years leading to the entry and since then, China introduced many changes that made the domestic market increasingly open to competition with foreign companies. According to the resource-based theory and its application (Guillen, 2000), the benefit of grouping strategy comes from the ability to combine inputs, operational knowledge, distribution channels, and contacts with foreign groups and government quickly and efficiently in repeatedly entering a variety of industries. The capability to enter industries repeatedly entails a variety of skills. This capability is not industry specific, and thus it tends to allow for unrelated diversification. This capability is also difficult to trade because it is embodied in an organizations owners, managers, and routines. Kock and Guillen (2001) observed that protectionism is a necessary condition for the rise of local business groups in emerging economies because foreign multinationals can also take advantage of inefficient markets, unless there is protectionism. Naturally, as market competition increases over time, the benefits of grouping strategy for entering new business opportunities in emerging economies decrease because increases in market competition entail the diminishing flows of profits
Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Causes for changing performance in a transition economy

2047

created by entering new industry segments. Therefore, it becomes increasingly difficult for the grouping strategy to create value by expanding to new industry segments. If they keep entering new sectors, despite increasing market competition, they will be subject to value discount. Degree of market competition is conceptually different from the degree of market failure, or general development of market mechanisms/institutions which is the more important issue in transition economies like China. In general, we reason that a higher degree of market competition in a sector would affect negative performance of any firms in such a sector, regardless of their types. Here, we are specifically interested whether or not such negative effect is more serious in group-affiliated firms. This will be done by inserting an interaction term between a market competition variable and a dummy for group firms. While the discussion above suggests decline of the benefits from entries into new markets as one of the reasons for group firms to perform worse in higher competition sectors, there can be more reasons. In the context of China, most of the business groups are state-owned, and thus one can also reason that these state-owned groups might be performing worse in a more competitive environment. Thus, we will first test the general impact of increasing competition on groups firms by using a dummy, whereas specific impact of diversification as a firm-level factor will be tested separately by adding a diversification variable. Put differently, group firms and stand-alone firms might differ in many dimensions other than diversification. Thus, it is necessary to use a dummy to represent group firms and to see how they perform in more competitive environments. In sum, given the trend of increasing market competition in China, testing its impact on the performance of business groups is worthwhile. Thus, we hypothesize the following: if market competition becomes more intense, the market valuation performance of business group will decline.

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2.2 Impact of firm-level variables: diversification and agency costs

Groups firms and stand-alone firms are different from each other in many dimensions. Here we will focus on two factors, degree of diversification and asset diversion leading to agency costs. Given our objective of explaining the decline performance of group firms, we will add interaction of the variables representing diversification and asset diversion, respectively, with a variable representing the different period of time. If these interaction terms are showing negative coefficients, then we can infer that there are increasing discounts over time for more diversified firms or firms doing more asset diversions which are actually group-affiliated firms. 2.2.1 Increase of diversification discount over time? As our data will show later, we observe that group firms in our sample are more diversified than other firms. Then, if we can show that there is an increasing discount for diversified firms, we identify one factor for declining performance of group firms.

2048

B.-K. Seo et al.

But, we still have to provide reasons for increasing discount over time for diversified firms in China. The literature on business groups tends to reason that if a market becomes more mature, namely there is less market failure, there will be less premium, or equivalently more discount, for diversified firms. But, one can also reason that when the market becomes more competitive, it leads to more discount for diversified firms. Actually, one can argue that the degree of market competition would be more directly related to diversification performance than the degree of market failure. As the degree of competition increases in general, more diversified firms will feel more resource-constrained in responding to ever-increasing market competition in several markets. However, sources for diversification discount can come from other sources. Thus, in our analysis, we interact the diversification variable with the time variable to check the changing impacts of diversification in different time periods in China. In other words, the time variable reflects so many diverse factors (and their change over time) that may possibly affect the role of diversification. 2.2.2 Increasing impact of the agency costs and asset diversion on firm values The fourth hypothesis that explains declining grouping benefits or increasing grouping costs is based on the agency cost theory. In contrast to advanced economies, the management environment in emerging economies increases the potential agency costs associated with grouping. Higher asymmetric information and weak corporate law and lax enforcement mechanisms might allow management and large shareholders to exploit the firm more easily for their own purposes, using grouping strategy (Bertrand et al., 2002; Lins and Servaes, 2002; Baek et al., 2006). In emerging economies, the monitoring system is not perfect, and internal profit transfer is common because the corporate governance system is not perfect. For example, despite the low ownership, controlling shareholders in publicly traded firms maintain control of these poor corporate governance mechanisms. If controlling shareholders want to tunnel the firm value using grouping structures, they have incentives to mask true firm performance and conceal their private control benefits from outside investors. Previous studies have also shown that investment inefficiencies or asset appropriation arising from agency problems are important factors for the discount of diversified conglomerates in advanced economies (Scharfstein and Stein, 2000). In transition economies, the economic and political problems created by agency problems tend to increase because of reform programs. Experiences in mass privatization in Eastern Europe and the former Soviet Union show that reform programs are extremely difficult and that their outcomes have consistently fallen below initial expectations (Sachs and Woo, 1994, 1997, 2003). A corporate governance vacuum arises from the partial reform process in transition economies; hence, agency cost problems become more severe and investors discount for grouping strategy increases. Subsequently, once market investors recognize the agency problems associated with

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Causes for changing performance in a transition economy

2049

grouping structures more clearly, this may be reflected over time in value discount for group-style firms. According to Gao and Kling (2008), Tenev and Zhang (2002), Meyer and Lu (2004), and Lee and Hahn (2007), diverse forms of agency costs are serious in listed companies in China, and long-term investment or loans from listed to unlisted subsidiaries are often used as channels for tunneling or asset stripping. As discussed in Lee and Jin (2009) and Hahn and Lee (2006), business groups have emerged in China through diverse channels, such as spin-offs, M&A, and joint ventures, but these channels are often used for illicit asset stripping because property rights are not clearly defined in the transition periods of China. Group formation, often used for the illicit stripping of assets of (listed) mother-companies by affiliates, has been so pervasive in China that a popular line claims that mamma is poor but her kids are rich (Ding, 2000). Numerous anecdotes have suggested that controlling shareholders treat listed firms as cash machines, where they can withdraw money as long as they wish (Liu and Lu, 2004).1 In addition, managers want to gain further autonomy from their supervising agencies by breaking up existing enterprises to form subsidiaries and joint ventures with foreign or domestic partners. Additional motivations can include undertaking new business opportunities and shifting bad debts and surplus labor burdens to parent companies. Although the Communist Partys control over the rights to appoint and dismiss top SOE managers might have served as an important counterbalance to managerial discretion, illegal tunneling of firm profits and assets has been found in many cases. For instance, Xiongmao Eletric Company was one of the most profitable listed companies in the mobile phone industry, but it experienced financial difficulty because of the illicit diversion of company resources. For several years, the CEO diverted huge amounts of money to un-consolidated affiliates. He ordered Xiongmao to invest huge amounts of money and guarantee the loans of these affiliates as he had the absolute stake proportion in these affiliates. Affiliated companies doing real estate business are often used as means for tunneling initiated by an unlisted company in the same group that has control over the listed companies in the same business group. If market investors are smart enough, they will give lower valuation for firms associated with higher agency costs. While this reasoning is natural, our goal is not simply to verify this relationship but to show that the impact of agency costs has been changing over time to affect more negatively in the Chinese context. Thus, we will add an interaction variable between the asset diversion variable and a time variable. The negative coefficient of this term implies that the market investors have been
1

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

For example, the largest shareholder of Meierya, which was then a profitable company, colluded with other insiders to embezzle 41% of the companys total equity in 2001. In the same year, the largest shareholder of Sanjiu Pharma, one of the blue chips in China, held 96% of the listed companys total equity (Liu and Lu, 2004).

2050

B.-K. Seo et al.

learning over time and thus tended to value less and less for the firms associated with high-agency costs. In sum, we will first measure the degree of the agency costs to show the higher incidence of such costs in group firms, and will proceed to show that the impact of the agency costs on firm value had been increasing over time.

3. Measuring the performance of business groups in China

3.1 Data and performance measures
This study uses the China Stock Market and Accounting Research Database (CSMAR) from 1994 to 2003, which contains the financial statements and stock market information of all listed firms. The CSMAR was designed by the China Accounting and Finance Research Center of Hong Kong Polytechnic University based on the annual reports of listed companies. Although publicly listed companies represent only a small subset of enterprises, the database of listed companies is the most consistent, accurate, and transparent among available information on all types of firms in emerging economies. As such, many studies on emerging economies use the database of listed companies in analyzing the performance of groups (Khanna and Palpepu, 2000a, b; Lins and Servaes, 2002). In the past 10 years, the total number of listed companies in China rapidly increased. In 1994, only 291 companies were listed in the China Stock Exchange. However, in 2003, there were 1287 listed companies (780 in Shanghai and 507 in Shenzhen). This study excludes companies in the financial service industry because financial service companies have different accounting principles. As seen in Table 1,
Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Table 1 Number of listed companies in the sample

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total Panel sample Whole (unbalanced) sample companies 283 283 283 307 323 283 510 530 282 715 745 282 821 851 282 918 949 282 277 275 273 2802

1054 1132 1188 1246 8174 1088 1160 1224 1287 8448

Total number of listed 291

Note: Our balanced panel sample (used in regression) contains these 273 companies existing every year. In calculating a comparable indicator, like EXCESS values, for the financial performance of a sample firm, each firm from the balanced panel is compared with the median of non-group firms belonging to the entire unbalanced panel of firms to obtain a sufficient number of non-group firms.

Causes for changing performance in a transition economy

2051

there were 283 listed companies in 1994. Among them, 273 were still listed in 2003. Hence, our balanced panel sample contains these 273 companies existing every year. This article uses this balanced panel sample of firms in all panel regressions to make sure that we obtain a consistent picture of the trend free from selection or survival bias. In calculating a comparable indicator, like EXCESS values, for the financial performance of a firm, each firm, regardless of group- or non-group firms, from the balanced panel is compared with the median of non-group firms belonging to the entire unbalanced panel of firms to obtain a sufficient number of non-group firms. Following the method used in Lee and Woo (2002), this article first defines a business group as having at least two affiliated companies. However, we also try another criterionhaving four or more affiliated companiesin checking the robustness of regression results. There is no theoretical ground to defend either of the two criteria. In other literature, such as La Porta et al. (1999), the criteria of two or more listed affiliates was adopted. We count both listed and unlisted affiliates; hence, the criterion of four or more affiliates can be considered in a sense as closer to the criteria in the previous literature. To save space, we present the results with the first definition and then present both results when a robustness check seems important.2 In this study, affiliated companies of business groups are those that are included in the consolidated balance sheet of the parent companies each year, as their shares are more than 50% owned by the parent companies. Information on the affiliations is manually obtained from the annual reports of each company, which contain the name of affiliations and their primary business segments. According to the method of defining groups as having two or more affiliates, 81% of the companies are classified as groups in the balanced panel sample. Business groups comprise 69% of the whole sample. Industry classification is based on the China Securities Regulatory Commissions two-digit code Standard Industry Classification (CSIC) system. To avoid very small samples, this article uses a one-digit code SIC for companies in some sectors, except in manufacturing. This article finally divides the 22 industry sections: 12 one-digit code SIC (plus manufacturing industry) and 10 two-digit code SIC. The performance measures can be classified as accounting measures and financial measures. This article uses the financial market valuation measure because accounting measures can be manipulated given that the accounting system is not complete in emerging economies. In the 1990s, there were also a number of changes in accounting rules, which makes it even more difficult to compare performance using accounting measures. In addition, market valuation performance measures represent the possibility of the future growth or profitability of companies because stock market investors consider the future values of firms.
2

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

More results are also available in Seo (2006).

2052

B.-K. Seo et al.

To investigate whether business groups are valued differently from non-groups, this article employs the valuation methodology called the excess values proposed by Berger and Ofek (1995). The same methodology was used in the case of Korean Chaebols by Ferris et al. (2003). Excess value is the natural logarithm of the ratio between the real value (V) of firms and the imputed or hypothetical (IV) value of firms. The formula is as follows:   V EXCESS ln , 1 IV where
Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

V CAPITAL (market value of equity) (book value of debt) a real value (CAPITAL) of a firm in industry i:   V IV AI AI ng imputed value of a firm in industry i as a typical non-group:  AI a real value of the accounting item (sales or asset) of a firm  V the ratio of a value of capital to an accounting item (sales or asset) AI ng for the median of the non-group firms in industry i Real value or capital is the sum of the market value of equity and the book value of debts. The imputed value of any company (whether it is a group or non-group firm) can be obtained by multiplying this median (V/AI) ratio of non-group firms to the actual sales or asset of a firm. To calculate the median ratio, this article uses the median values from non-group firms from the whole sample (rather than from non-group firms belonging to a fixed number of balanced panel data) to obtain a sufficient number of non-group firms in each sector.3 Excess value is more commonly used in comparing diversified business group firms and non-group firms than other performance measures, such as ROA and Tobins q. The excess value measure has the advantage of being neutral to industry and time shocks that affect all firms similarly. Excess value is a relative measure; hence, it is not affected by trends in the stock market. If the EXCESS value of a group is more than zero, the real market value of a firm is larger than the imputed value (V4IV). This is defined as the premium of that firm. If the EXCESS value of a firm is smaller than zero, the real market value of a

The longitudinal decreasing performance of group firms has also been found using balanced panel sample data in calculating the median ratio of non-groups.

Causes for changing performance in a transition economy

2053

Table 2 Time-trend of the excess values and the differences

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

The results using the sales values Full sample 0.05 0.15 0.25 Non-groups 0.18 0.15 0.15 Groups Groupnongroup The results using the asset values 0.01 0.15 0.29 0.18* 0.19 0.16 0.05 0.03 0.23 0.20 0.14 0.11 0.20 0.11 0.17 0.17 0.16 0.09 0.20 0.17 0.10 0.22 0.14 0.06 0.17

0.01 0.13* 0.18* 0.23** 0.31*** 0.34*** 0.30** 0.32*** 0.23**

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Full sample Groups Groupnongroup

0.01 0.07 0.37 0.02 0.08 0.40

0.24 0.22 0.16 0.05 0.26 0.26

0.15 0.01 0.18

0.12 0.16 0.17

0.08 0.22 0.13

0.08 0.26 0.14

0.02 0.29 0.07

Non-groups 0.01 0.04 0.27

0.03 0.04 0.13** 0.10* 0.21*** 0.18*** 0.33*** 0.36*** 0.40*** 0.35***

*P50.10; **P50.05; ***P50.01.

firm is smaller than the imputed value (V5IV ).4 This is defined as the discount of that firm.

3.2 Examining the time trends of market performance

To determine the trends of excess value of groups, this article first uses the yearly simple performance comparison and OLS regressions. First, this study compares the excess value of group and non-group firms in each year. Table 2 shows the longitudinal changes of the simple EXCESS performance differences between groups and non-groups. The difference is the excess value of group minus non-group firms; hence, positive values indicate value premium for group firms. Table 2 shows there was value premium for group firms in 1994. However, thereafter, it turned to a negative value and increased to a bigger negative number. Second, this section tests the time trend using yearly OLS regressions after controlling for a number of basic characteristics that could affect performance differences. Following Berger and Ofek (1995), this article chooses firm size, profitability, and growth as control variables for EXCESS value regression. This article uses the
4

Specifically following Berger and Ofek (1995), extreme case is defined as the ratio between actual values and imputed value. Accordingly, an extreme case occurs when the excess value is above 1.386 or below 1.386 (i.e. actual values is either more than four times imputed or less than one-fourth imputed). In following Berger and Ofek (1995) method, the extreme case is 3.2% (asset multiplier) in the whole sample data and 3.4% (asset multiplier) in the panel sample data.

2054

B.-K. Seo et al.

SIZE (the log of total asset), ROA (net income on total assets)5 and GROWTH (increase of sales over the sales of previous year) as the control variables for multiple regressions. The group dummy variable takes the value of 1 if the firm operates as groups and 0 if non-groups. Hence, the yearly OLS regression model for EXCESS values is as follows: EXCESS  1  Group dummy2  SIZE 3  ROA 4  GROWTH Error term: 2

Table 3 shows the longitudinal downward trends of the coefficients of the group dummy in OLS regressions. The group dummy coefficients were positive in 1994 and 1995. Thereafter, they significantly turned negative. These results indicate a downward trend in the relative performance of groups or group premium changing to a discount in China measured by market valuation, which is the same in other emerging economies, such as Chile or Korea in the 1990s. Third, another commonly used method that checks the time varying tendency utilizes the interacting term of time and group dummy variables (Khanna and Palepu, 2000b; Lee et al., 2008). The time variable takes a value of 1 in year 1994, 2 in year 1995, and so on. The model specification is as follows: EXCESS  1  Group dummy2  SIZE3  ROA4  GROWTH 5  Time6  Time Group dummy Error term: 3 Following their method, this article tests the interaction term of time and group dummy (Time Group dummy) variables as the measure of time varying tendency of group premium. Table 4 presents the results of the regressions. In both results using two different definitions of business groups in terms of the number of affiliates (two or more affiliates in part A and four or more affiliates in part B), the time varying tendency variable (time group dummy) coefficients are significantly negative. The above discussion shows that with the lapse of time, the grouping premium in China based on market valuation measures tends to diminish, eventually turning to a discount. The market valuation performance measures represent the possibility of future growth or profitability. Hence, market investors believe that with the lapse of
5

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

ROA is defined as (net income)/(total assets). This is redefined as the interaction of net profit margin (NPM) and asset turnover ratio (ATR). However, some net income data need to be coded because net profit margin is below 1 or above 1. Hence, this article uses the coded net income data, which have a net profit margin between 1 and 1. In the total net income data, 1.3% of the net income data are coded. ROA NI NI SALES NPM  ATR: TA SALES TA 1

Table 3 Time-trend of the group firm dummy, using year by year OLS regression
1996 EXCESS (Sales) 0.122 (1.47) 0.021 2.36* EXCESS (Asset) 0.070 (1.54) 0.121 10.36*** 11.69*** 27.63*** 0.134 0.278 (1.17) (2.96)*** 0.053 0.130 0.060 (1.05) 0.357 39.16*** 0.142 (2.28)** 0.401 46.17*** 0.118 (1.50) 0.368 40.17*** 0.159 (1.93)** 0.337 34.65*** 0.143 (1.67)* 0.246 22.48*** 3.02* 10.19*** 6.69*** 0.030 0.125 0.092 (1.80)* (1.95)** (2.23)** (2.04)* 0.201 15.97*** 0.148 0.166 0.244 0.214 0.152 (1.15) 0.096 6.38*** 0.214 (1.65) 0.097 6.09*** 0.073 (0.59) 0.105 7.04*** 1997 1998 1999 2000 2001 2002 2003

Year

1994

1995

Variables

Group

0.196

0.004

dummy

(1.63)

(0.05)

Adjusted R2

0.079

0.080

F-value

3.76***

5.91***

Variables

Group

0.063

0.025

dummy

(1.02)

(0.49)

Adjusted R2

0.194

0.188

F-value

9.58***

15.49***

Note: Only the group dummy coefficients and adjusted R2, and F-value are presented.

Causes for changing performance in a transition economy

*P50.10; **P50.05; ***P50.01.

2055

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2056

B.-K. Seo et al.

Table 4 Multiple regressions for time varying tendency in EXCESS values

EXCESS (Sales) Regression OLS Fixed Random EXCESS (Asset) OLS Fixed Random

Results with group firms (those with two or more subsidiaries) Group dummy TIME Group dummy TIME SIZE ROA GROWTH R2 F-test Hausman test 0.059 (0.88) 0.033 (2.82)*** 0.038 (3.68)*** 0.166 (11.71)*** 0.481 (3.11)*** 0.001 (0.68) 0.082 33.93*** 0.165 (2.61)*** 0.042 (4.20)*** 0.049 (5.40)*** 0.205 (7.04)*** 0.225 (1.61) 0.005 (2.78)*** 0.526 6.43*** 0.129 (2.13)** 0.041 (4.16)*** 0.047 (5.35)*** 0.189 0.283 (2.08)* 0.004 (2.38)** 0.056 15.39*** 0.038 (0.88) 0.025 (3.43)*** 0.056 (8.57)*** 0.218 0.310 (3.03)*** 0.003 (1.89)* 0.234 131.82*** 0.092 (2.47)** 0.022 (3.76)*** 0.076 (14.31)*** 0.498 0.329 (3.76)*** 0.004 (3.54)*** 0.632 8.86*** 0.086 (2.36)** 0.024 (4.18)*** 0.071 (13.70)*** 0.419 0.317 (3.70)*** 0.004 (3.53)*** 0.306 102.38***

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

(9.02)*** (25.08)*** (31.82)*** (30.94)***

Results with group firms (those with four or more subsidiaries) Group dummy 0.269 0.300 0.281 0.095 (3.14)*** TIME Group dummy TIME SIZE ROA GROWTH R2 F-test Hausman test 0.088 (6.33)*** 0.033 (2.95)*** 0.320 (3.81)*** 0.069 (6.02)*** 0.045 (4.96)*** 0.532 (3.67)*** 0.074 (6.53)*** 0.040 (4.46)*** 0.446 (2.58)** 0.040 (6.71)*** 0.008 (1.63) 0.209

0.155 (4.71)*** 0.035 (7.43)*** 0.025 (6.60)*** 4.50

0.135 (4.17)*** 0.038 (7.80)*** 0.019 (4.87)*** 0.357

(15.53)*** (14.46)*** (14.57)*** (23.56)*** (29.47)*** (27.78)*** 0.447 0.248 0.261 0.106 0.039 0.055 (5.38)*** 0.008 (2.53) ** 0.175 87.63*** (3.41)*** 0.014 (5.67)*** 0.160 9.21*** 2.85 (3.76)*** 0.013 (5.48)*** 0.167 (2.98)*** 0.002 (1.74)* 0.293 172.76*** (1.29) 0.003 (2.65)*** 0.267 10.15*** 130.72*** (1.87)* 0.003 (2.59)*** 0.279

Causes for changing performance in a transition economy

2057

time, the grouping benefits will decrease and grouping costs will increase. Why do market investors change their belief for grouping strategies from premium to discount? We will answer this question in the next section.

4. Testing the hypotheses

4.1 Measuring key variables for hypotheses
4.1.1 Measurement of institutional development in each sector This article measures the development of capital, labor, and product market institution using the following method. First, following the literature (Khana and Palepu, 2000b; Lee et al., 2008), this study measures capital market development by the ratio of the market capitalization of listed companies to the GDP. In the Chinese context, not all stocks are traded in stock markets. In the 1990s, $60% of shares were owned by state agencies and related legal persons, and their shares were not traded in markets.6 Thus, our measure of capital market development in each industry is defined as the ratio of market capitalization of tradable shares in each industry to the GDP of each industry. TMARKET in Table 5 represents this value of capital market development in China, as an average of the values for sectors. This value shows an increasing trend over time, reaching 15% in the early 2000s. Second, for the development in the labor market, we also consider the Chinese context where many workers are employed by the SOEs. Khana and Palepu (2000b) used the school enrollment ratio as the explanatory variable for labor markets. Workers in SOEs are inflexible, and employments and dismissals are restricted. Thus, the proportion of state workers can represent labor market imperfections. This study uses the proportion of workers in non-SOEs in each sector to the total workers as the explanatory variable for labor market development in each sector. NSTATEW in Table 5 represents this value for labor market development as an average of sectoral values. It shows that the share of non-state workers represents more than 30% of the total in the early 2000s. Third, to represent the development of product markets, this study uses the percentage of the registered capital in non-SOEs in the total registered capital of the whole domestic enterprises. SOEs are part of the state planning, and SOEs allocations for inputs and products depend outside the market allocation system. NSOE in Table 4 represents this value as an average of sectoral values, and it is shown that non-state sector represents more than 60% in the early 2000s.
6

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

State-owned shares refer to shareholdings of the central and local governments, or institutions (including other SOEs) and departments designated by the State Council or by local governments. Legal person shares refer to those owned by domestic enterprises or other economic entities enjoying legal person status, who are generally promoters of the invested company.

2058

B.-K. Seo et al.

Table 5 Market institutional development in the Chinese economy (19942003)

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 TMARKET Market capitalization ratio (tradable stocks) NSTATEW Share of non-state workers (%) NSOE INSTIT Share of non-SOEs (%) Aggregate Industry A Industry B Industry C Industry D Industry E Industry F Industry G Industry H Industry J Industry K Industry L 44.2 47.7 51.4 50.3 55.5 54.9 56.3 59.7 62.8 62.8 1.2 1.0 0.8 0.6 0.2 0.1 0.6 0.9 0.3 4.8 4.3 1.0 0.6 5.1 4.7 1.3 4.4 5.0 4.5 4.2 3.4 3.1 2.9 2.8 2.7 2.6 26.5 26.7 26.5 26.2 26.6 28.6 29.2 30.0 31.3 34.4 1.7 1.4 2.8 6.0 7.3 9.1 14.1 15.2 11.8 9.8

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

3.1 3.9 4.2 4.4 4.0 2.2 0.7 0.8 1.0 1.2 1.5 2.0 3.1 4.0 0.5 1.1 0.9 1.7 2.2 3.3 4.1

1.4 5.6 5.0

2.5 4.1 4.6 3.7 3.5 2.2 1.2 0.6 0.4 0.2 4.4 4.5 4.5 3.7 2.7 2.1 1.9 1.9 1.6 1.6 4.4 4.5 4.5 3.7 2.7 2.1 1.9 1.9 1.6 1.6 1.8 1.7 1.8 1.9 2.0 2.6 0.8 3.1 1.5 3.3 2.0 3.2 2.5 3.5 3.1 1.9 1.4 1.0 0.5 0.2

1.6 0.6 0.1 0.5 0.2 0.1 0.6 0.9 1.2 1.2 3.3 3.4 3.8 3.9 3.6 3.5 3.2 2.7 2.6 2.8

Notes: TMAKRET, NSTATEW, NSOE are all calculated first at each sector, and in the above presented are their average values across sectors. TMARKET (market capitalization of tradable stocks) is calculated by CSMAR data and is (market capitalization of tradable stocks/ GDP) 100. NSTATEW (share of non-state worker) is 100(staff and workers employed in state-owned units/total staffs and workers)100. NSOE (share of non-SOE) is 100(registered capital in SOEs/registered capital in the whole enterprise) 100; INSTIT at each sector represent the degree of institutional development, which is the average of the three market (capital, labor, and product) development ratio. Before taking average, these three ratios for each sector are first standardized by taking the distance from the across-sector mean and are then divided by the standard deviation. Aggregate index are the simple average of sector values. In regressions, sector values are used; Industry A: Farming, Forestry, Animal Husbandry, and Fishery; B: Mining; C: Manufacturing; D: Utilities; E: Construction; F: Transportation and Warehousing; G: Information Technology; H: Wholesale and Retail Trade; I: Finance and Insurance; J: Real Estate; K: Social Services; L: Communication and Cultural Industries; M: Others. Sources: Calculations using data from China Statistical Yearbook; China Securities Regulatory Commission.

The next step is to calculate the index of overal market development in each sector over time. For this, we have to utilize three measures representing development of capital, labor, and product markets, respectively, which are explained above. But, these three ratios for each sector are first standardized by taking the distance from the

Causes for changing performance in a transition economy

2059

across-sector mean and are then divided by the standard deviation. For example, the labor market development indicator for a sector is defined as (the share of the non-state workers in that sectormean over sectors)/(standard deviation of the mean). After calculating such indicators for the three markets for each sector, we take a simple average as the index of the overall institutional development of a sector. Table 5 shows the institutional development of sectors over time. We use these values in regressions to represent the degree of institutional development in each sector over time. The top row of the same table also shows an aggregate measure of the national-level market development in China. This aggregate index is an average of institutional development in each of the three markets, and the development in each of these markets is represented by taking an across-sector average of the standardized value of each sectors development. As seen in Table 5, the aggregate INSTIT variables have increased from 1.2 in 1994 to 1.3 in 2003. Market institution developments are higher in manufacturing, construction, wholesale and retail trade, and real estate than other industry segments. 4.1.2 Measurement of the degree of competition in each sector It is difficult to calculate the degree of market competition in China. Due to the shortages of data, some common index cannot be calculated, such as industry CR3 (top three firm concentration ratios). As such, this article uses the sector-level profitability as the degree of market competition. Increasing market competition diminishes profitability. As a result of increasing market competition, the market turnover ratio (sale/total asset) of listed companies has slightly decreased. The gross profit margin (total profit/total asset) has also decreased, probably due to heavy competition. Table 6 presents the trends of the market turnover ratio and gross profit margin. In the regression analysis, we will use the inverse of sector-level gross profit margin to represent the degree of market competition in each sector (COMPET). 4.1.3 Measurement of the unrelated diversification of firms The unrelated diversification information is based on China Securities Regulatory Commissions two-digit code Standard Industry Classification (CSIC) system. Information on affiliate companies, such as the number of affiliates and the number of segments of affiliates, is created for this study. Information on affiliate companies comes from the annual reports of the Shanghai Securities News (zhengquanbao) and the China Securities Regulatory Commission. The annual reports provide the financial statements of parent companies and the consolidated financial statements of business groups. This article classifies business groups operating in three or more industry segments as unrelated diversified business groups. The unrelated diversification dummy variable (DIVER) is 1 if the group has more than three industry segments; otherwise, it is equal to 0.

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2060

B.-K. Seo et al.

Table 6 Turnover ratio and gross profit margin trend

1994 Turnover Mean Median Profit margin Mean Median 0.26 0.16 0.16 0.10 0.23 0.12 0.18 0.14 0.17 0.14 0.20 0.13 0.14 0.12 0.11 0.09 0.09 0.08 0.12 0.07 0.60 0.47 0.58 0.47 0.56 0.47 0.56 0.45 0.51 0.42 0.50 0.42 0.50 0.40 0.51 0.40 0.54 0.42 0.58 0.46 1995 1996 1997 1998 1999 2000 2001 2002 2003

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Note : Proportion of turnover ratio and gross profit margin is calculated using the CSMAR data. In regression, the inverse of sector-level profit margins are used to represent the degree of market competition (COMPET) in each sector.

Table 7 reports the average number of industry segments within business groups. The average number of industry segments measured by the two-digit CSIC of group organizations rose from 2.61 in 1994 to 2.77 in 2000 and decreased to 2.27 in 2003. This U-shaped tendency or the recent focusing trends of listed companies reflect the increasing competition in the local market. For instance, the taxi business, which had higher profits prior to the late 1990s, used to be one of the primary diversification business segments in the 1990s. However, the number of taxi business affiliates rapidly decreased in the 2000s. 4.1.4 Measurement of the agency costs and asset diversion of firms The complications of internal organizations are difficult to measure and finding a good proxy for agency costs is difficult as well. A simple measure is to use the ownership structure, especially the ratio of state-ownership, with the idea that agency costs would be more serious with more state-ownership. However, it is a more indirect measure of agency costs because there is no guarantee that more state-ownership would automatically lead to more agency costs. Moreover, state-ownership in China is related to other factors, such as special protection and privilege from state authorities or near monopoly in certain sectors, although some literature tends to find a negative correlation between state-ownership and performance. Thus, we initially use state-ownership as a measure of agency costs but try a more direct measure of it. Table 8 shows the share of state-owned equity by group- and non-group firms, which is $29%. There is no significant difference between the two types of firms. Thus, this factor might not be able to explain the performance difference between the two as well as its trend over time. We will verify this by regression analysis. Another measure we tried is to reflect more directly on the consequence of agents costly behavior. As discussed above and more recently in Gao and Kling (2008), we

Causes for changing performance in a transition economy

2061

Table 7 Average number of industry segments of groups and number of groups

1994 Segment 2.61 1995 2.63 1996 2.64 1997 2.61 1998 2.69 1999 2.72 2000 2.77 2001 2.51 2002 2.25 2003 2.27

Number of segments 1 2 DIVER 0 3 4 5 6 7 DIVER 1 38 63 101 62 38 8 1 0 109 38 61 99 63 38 9 1 0 111 38 60 98 63 38 10 1 0 112 47 67 114 55 42 11 3 0 111 46 63 109 59 33 20 4 0 116 44 74 118 58 30 20 9 0 117 44 73 117 60 24 21 11 2 118 60 74 134 51 31 16 4 1 103 59 88 147 65 19 4 1 0 89 56 88 144 66 18

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

4 1 0 89

Note : SEGMENT is the average number of industry segments of the parent company and affiliations. DIVER 0 means that the number of the segments is 2 or less and these firms are not diversified.

assume that asset diversion between listed parent-companies and affiliated companies represents the essence of agency costs in China. Thus, we try to represent the degree of asset diversion by the degree of long-term investment (LONGINV) in affiliated/other companies relative to the total asset of the listed companies. This measure is similar to the measure used by Gao and Kling (2008). They tried to capture the degree of related party transactions using the gap between accounts receivable and payable. Table 8 shows the figures in both group and non-group firms and their change over time. For group firms, the period average is above 23%, which is significantly higher than non-group firms (14%). Thus, this factor could be one of the factors explaining the performance difference. We will verify this by regression analysis. The table also presents the leverage ratios, which will be used as control variables in the regressions. Compared with the others, group firms are more indebted. This higher indebtedness of groups firms is also observed in other countries, such as Korea (Ferris et al., 2003).

4.2 Regression methodologies

To test the hypotheses explaining the longitudinal downward trends of grouping performance, this article uses the following regression methods. Taking advantage of the panel structure of the data, this article adopts the panel regression method with

2062

B.-K. Seo et al.

Table 8 Changes in long-term investment, debt leverage, and state-ownership in listed companies in China
1995 0.12 0.19 0.07* 0.39 0.47 0.08*** 32.97 30.48 2.49 2.74 0.97 2.07 1.96 29.97 30.10 30.33 28.49 32.71 29.14 28.26 26.54 0.08*** 0.10*** 0.07** 0.10*** 27.30 28.09 0.79 0.48 0.47 0.49 0.52 0.55 0.14*** 0.39 0.38 0.42 0.42 0.41 0.20 0.07* 0.21 0.04 0.20 0.03 0.27 0.17*** 0.27 0.12** 0.13 0.17 0.23 0.09 0.15 0.12 0.28 0.16*** 0.57 0.58 0.01 27.11 27.49 0.38 1996 1997 1998 1999 2000 2001 2002 0.19 0.29 0.10 0.60 0.65 0.06 27.43 27.85 0.42 2003 0.12 0.23 0.11** 0.68 0.65 0.02 27.06 28.25 1.19 Total 0.144 0.236 0.093*** 0.433 0.536 0.103*** 29.968 29.202 0.767

1994

LONGINV

Non-groups

0.11

Groups Difference

0.20 0.09***

Leverage

Non-groups

0.33

Groups

0.42

Difference

0.09***

STATE share

Non-groups

34.02

Groups

31.58

Difference

2.44

Notes : The difference is the values of group firms minus those of non-group firms. Its significance is *P50.10; **P50.05; ***P50.01. LONGINV is the rate of long-term investment/loan to the total asset in the previous year. STATE share is the share of the equity owned by the state in percentages. Leverage is the ratio of total debts to total asset.

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Causes for changing performance in a transition economy

2063

the error term defined as the sum of the unobservable individual effect (vi) and the remaining stochastic disturbance term ("it). eit vi "it , where vi the unobservable individual specific effect, and vi is time-invariant and accounts for any individual-specific effect not included in the regression, "it the remainder stochastic disturbance term, and "it is assumed independent of Xit for all i and t. "it $ iid (0, "2). In terms of the different assumption for the unobservable individual effect (vi), panel regression is classified as either fixed effect or random effect. Campa and Kedia (2002) used the fixed effect panel model to investigate the relative performance of diversified groups, while Khanna and Palepu (2000b) used the random effect panel model to investigate the relative performance of groups and non-groups. This article has first tried both fixed and random effect panel models, as well as OLS regression, but presents the relevant results only, after selection based on Breusch-Pagan test and the Hausman test. Actually, OLS regressions are rejected by Breusch-Pagan test, and some random effects, by Hausman test. These multiple regressions model a firms market valuation (EXCESS) as a function of group dummy, time, and typical control variables, such as size (SIZE), profitability (ROA), growth propensity (GROWTH), leverages (LEVERAGE) and age (AGE), and the key explanatory variables to test the hypotheses. The model specification is shown below as equation (6). Key variables are also entered as interacting with a group dummy or a time variable depending on the nature of the variables. Sectoral variables, such as institutional development (INSTIT) and degree of competition (COMPET), are inserted as interaction with a group firm dummy to test whether such variables negatively affect the performance of group firms. Market competition at sectors is proxied by the inverse of gross profit margins of sectors, shown in Table 6. Firm-level key variables, such as unrelated diversification (DIVER), state-ownership (STATE), and asset diversion (LONGINV), are interacted with a time variable to test whether these variables negative impacts become serious over time. As discussed above, group firms are known to have higher degrees of diversification and long-term investments. EXCESS  1  Group dummy2  SIZE3  ROA4  GROWTH 5  LEVERAGE 6  INSTIT 7  Year99 8  INSTIT Group dummy9  COMPET Group dummy 10  Year99 DIVER11  Yeaer98 LONGINV 12  Year99 STATE 13  DIVER14  LONGINV 15  STATE Error term: 4
Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2064

B.-K. Seo et al.

Time variable can be represented by either a time trend variable or a time-period dummy. We have tried both methods to get the consistent results. Given that a time trend variable assumes the linearity of the impacts, we present only the results with a time-period dummy (Year99), which takes a value of one for the period starting from 1999, and thereby divide the sample period (19942003) at the mid-point. In investigating the effects of diversification to the relative performances of groups, the endogeneity problem becomes important. Some studies insist that endogeneity problems are important in measuring diversification because the firms decision to diversify is not a random action (Campa and Kedia, 2002; Villalonga, 2004a, b). Diversification might just reflect the sample selection biases such that already low-performing firms tend to go for diversification (Stulz, 1990; Rajan et al., 2000; Graham et al., 2002; Campa and Kedia, 2002; Villalonga, 2004a, b). To control endogeneity problems using the diversification variable, this study has also tried the instrument variable method to get the consistent results.7 The results are available upon request but not presented here. Actually, to the extent that endogeneity is coming from time-invariant variable, the fixed effect results already take care of this.8

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

5. Results and discussions

Table 9 shows the results of the multiple regressions. We present both results with two different definitions of business groups (group firm defined as having two or more affiliates, and defined as having four or more affiliates). Our interest is on the coefficients of the interaction terms, such as our key variables interacting with group firm dummies or with time variables.
7

Campa and Kedia (2002) modeled a firms decision to diversify as a function of industry, firm, and macroeconomic characteristics for instrument variable regression. This article follows their method. The first set of instruments entails industry and time characteristics. This study uses industry characteristics as the instrument for diversification. Industry characteristics influence the decision to diversify (Lang and Stulz, 1998). In addition, the proportions of diversified groups in a given industry represent industry attractiveness. The higher the fraction of diversified groups, the more attractive the industry factors are to diversification. This study also captures time trends in the macroeconomic conditions and business cycles, and includes real growth rates of the GDP. The second set of instruments is firm specific. This study creates a dummy variable for the firms that issue B- or H-shares. They take the value 1 when the firm issues foreign shares, and 0 if otherwise. Listing for foreigners (B-shares) or listing on foreign exchanges (H-shares) could affect a firms decision to diversify. Another firm-specific variable we try is the share owned by the biggest shareholder which is supposed to be inversely correlated with diversity of shareholders. More diverse composition of shareholders must to be correlated with more diversification. We have tried some combinations of these instruments. This is suggested by a referee.

Table 9 Determinants of the performance change of group firms: in (I), group firms defined as having two or more affiliates; in (II), having four or more affiliates
EXCESS (Asset) (I) Random 0.189 0.018 0.007 0.369 0.469 0.003 0.131 0.033 0.164 0.642 0.004 0.011 0.239 0.061 0.449 0.479 0.015 9.547 0.190 629.59*** 4.22 2603 (3.20)*** (1.24) (0.17) (5.52)*** (4.34)*** (2.09)** (7.67)** (0.84) (2.46)** (5.56)*** (3.05)*** (3.37)*** (3.38)*** (6.69)*** (15.16)*** (7.11)*** (5.97)*** (15.92)*** 0.095 (1.68)* 0.017 (1.03) 0.009 (0.37) 0.320 (4.89)*** 0.319 (3.10)*** 0.001 (1.17) 0.111 (3.88)*** 0.007 (0.34) 0.192 (2.56)*** 0.498 (4.52)*** 0.007 (3.57)*** 0.040 (3.31)*** 0.195 (2.78)*** 0.063 (3.91)*** 0.519 (13.97)*** 0.235 (3.25)*** 0.014 (5.75)*** 11.107 (14.92)*** 0.149 9.34*** 0.048 (1.98)** 0.014 (2.39)** 0.038 (2.39)** 0.045 (1.74)* 0.029 (0.68) 0.001 (2.00)** 0.186 (16.50)*** 0.012 (0.78) 0.046 (1.60) 0.023 (0.51) 0.002 (3.50)*** 0.002 (1.19) 0.099 (3.50)*** 0.038 (6.01)*** 0.495 (33.82)*** 0.139 (5.31)*** 0.004 (4.31)*** 10.291 (35.02)*** 0.236 10.84*** (II) Fixed (I) Fixed (II) Fixed 0.020 (0.81) 0.022 (3.08)*** 0.004 (0.35) 0.124 (4.50)*** 0.074 (1.72)* 0.001 (1.65)* 0.019 (1.58) 0.009 (1.03) 0.137 (4.36)*** 0.078 (1.69)* 0.004 (4.91)*** 0.002 (1.58) 0.138 (4.69)*** 0.003 (0.48) 0.468 (30.25)*** 0.034 (1.11) 0.003 (2.73)** 9.652 (31.11)*** 0.210 9.64***

EXCESS (Sales)

(I) Fixed

Group dummy INSTIT Group dummy COMPET Group dummy DIVER YEAR99 LONGINV YEAR99 STATE YEAR99 INSTIT COMPET DIVER LONGINV STATE LEVERAGE YEAR98 AGE SIZE ROA GROWTH Constant R2 F-test Wald 2 Hausman test 2 Breusch Pagan 2 Number of observations 134.4*** 1794.9*** 2483 523.0*** 1495.5*** 2608

0.148 (2.33)** 0.022 (1.44) 0.001 (0.02) 0.302 (4.50)*** 0.431 (3.96)*** 0.001 (0.86) 0.285 (9.77)*** 0.022 (0.58) 0.138 (1.72)* 0.599 (5.14)*** 0.009 (4.61)*** 0.011 (3.23)*** 0.285 (3.88)*** 0.139 (8.53)*** 0.529 (13.81)*** 0.427 (6.30)*** 0.017 (6.30)*** 11.654 (15.18)*** 0.140 8.05***

Causes for changing performance in a transition economy

1755.7*** 2603

184.9*** 1612.8*** 2491

2065

*P50.10; **P50.05; ***P50.01.

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2066

B.-K. Seo et al.

First, let us discuss the two sector-level variables representing changes in the external environment, such as institutions and competition. We notice that the institutional development hypothesis is partly supported. Only when using the asset multiplier is the coefficient of (INSTIT Group dummy) variables significantly negative. As shown by the positive coefficient of the institution variable itself, the development of market institution itself has a significant positive relation to the market performance of all types of listed firms in China. Further, the increasing market competition hypothesis is rather weak; its interaction term with a group dummy shows a negative and significant coefficient only in one of the two asset multiplier models. Let us now turn to firm-level variables, which are interacted with a time variable. First, the diversifications time varying effect (DIVER Year99) variables represent the time trends of the diversification impacts on the excess values of firms. We notice that this longitudinal effect of unrelated diversification is significant and negative, and is also robust in all specifications. Given that group firms tend to show a higher degree of diversification as shown in Table 7, investors devaluate the firms unrelated diversification strategy more seriously with time. Thus, more diversified firms, such as group firms, have been subjected to increasing value discounts. These results can support the diversification discount hypothesis. However, it is interesting to note that the coefficients of the diversification variable itself are all positive and significant. Put together, these imply that while diversification still creates values, its marginal contribution has been decreasing over time. This pattern seems to reflect the still transitional nature of the Chinese economy. To test the agency cost hypothesis, this article uses two variables, namely the state-ownership (STATE) and the long-term investment (LONGINV). It turns out that while the state-ownership variable consistently and negatively affects the values, its interactions with a time dummy (Year99) are not significant or even positive. The unstable coefficients of this variable of state ownership are not surprising because state-ownership could have diverse and opposite implications for performance in the context of the Chinese economy. It could be associated with special protection and privileges from state authorities with near monopoly in certain sectors and with agency costs of having no real owner other than bureaucratic managers. Furthermore, as shown in Table 8, the share of state-ownership is not significantly different between group- and non-groups and thus cannot explain the performance difference between these two types of firms. We could most likely use more measures that represent ownership structure. However, as discussed above, unless the measures are significantly different between group- and non-group firms, they may not provide answers. The asset diversion behavior represented by long investment in affiliates seems to be the correct firm-level answer. As shown in the table, the (LONGINV Year99) variables are all negative and significant, except in one model. This means that the agency costs in asset diversion have important explanatory power in the declining

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Causes for changing performance in a transition economy

2067

performance of group firms, as group firms tend to have a higher degree of asset diversion (see Table 8). In sum, the regressions results confirm more explanatory power of firm-level variables, especially to the diversification discount hypothesis. The weak explanatory power of the hypothesis of institutional development is understandable considering the fact that we are dealing with a relatively short period (10 years). Although the Chinese economy went through a very rapid structural change, fundamental institutional change cannot happen during such a short period. Further, one important nature of an institution is its long-lasting tendency. We can also look for more China-specific reasons behind the results. One of the most prominent features of Chinas economic reforms was the decision to promote large enterprises and business groups of SOEs in strategic sectors (Smyth, 2000; Lee and Jin, 2009). Business groups were used to reform SOEs in China (Lee and Jin, 2009). They were the primary vehicle through which the Chinese state and large industrial enterprises redefined their ties (Keister, 2000). According to the transaction cost theory, the grouping strategy is adopted because of the differences between market transaction costs and internal transaction costs within the group/network. The transaction cost theory assumes that individual firms spontaneously choose the best efficient transaction form according to transaction costs. However, in China, the grouping strategy was used as a means to reform SOEs. In addition, political considerations are prior to the goals of economic efficiency, such as reducing transaction costs. Most especially, in the early stages of grouping, membership in business groups closely resembled the group of firms that was overseen by an administrative bureau prior to the reform period. The similarities between the former bureaus and the business groups were partially a result of the states encouragement of firms to ally themselves with other factories formerly associated with the same administrative bureau (Keister, 2000). Government bureaucrats wanted to keep their control over the formerly subsidized firms through group organizations. Hence, the grouping of separate firms having few relationships has been frequent instead of the grouping of firms with vertical and horizontal relationships to reduce transaction costs (Tenev and Zhang, 2002; Meyer and Lu, 2004). As such, it is difficult to say whether grouping in China has something to do with the reduction of transaction costs. Rather, grouping has been used as a means to solve losses in SOEs by combining the good SOEs and the ailing ones. For example, in the cases of DongFeng and The First Automobile Group, their economic conditions were negatively affected because ailing SOEs were grouped with them. In contrast, the strong impacts of the ever tightening market conditions, especially on diversified groups, are widespread in the corporate sector. In the early days of economic reform and open door policy, when the profitability of firms was more prioritized, many companies diversified to seek to new and profitable industry segments, which (in some ways) originated from business groups

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2068

B.-K. Seo et al.

(Hahn and Lee, 2006; Lee and Jin, 2009). Further, the diversification drive was short-lived in China because of the intensifying market competition. This short-lived payoff to unrelated diversification from heavy competition and low profitability has been found in the changes in the average number of industry segments of groups (see Table 7). The average number of industry segments measured by the two-digit CSIC of group organizations rose from 2.61 in 1994 to 2.77 in 2000 but fell back to 2.27 in 2003. The limited unrelated diversification and refocusing trends of listed companies since year 2000 can be associated with heavy competition in the local market. In China, the integration of the domestic market and the growth of FDIs accelerated competition. Thus, in the case of China, we can say that the diversification discount hypothesis is more suitable than the institutional development hypothesis, considering the actual experiences on the reform of SOEs and on diversification.

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

6. Summary and concluding remarks

Using data on listed companies in China over a 10-year period (19942003), this study has investigated longitudinal changes in the relative performance of business groups. Using excess value as the value performance measure and panel regression methods, this article confirms the downward trend in the relative performance of group-affiliated firms. To explain this declining performance change, this article has considered several competing or complementary hypotheses, such as institutional development, increasing competition, diversification discount, agency costs of state-ownership, and asset diversion. These hypotheses have been represented by two different groups of variables. The two variables representing changes in the external market environment are the index of marketization measuring institutional development in each sector and the inverse of profitability of sectors measuring the degree of market competition. Three variables representing firm-level behavior or strategies (i.e. unrelated diversification dummy, share of state-ownership, and a variable of asset diversion) have been used. Market-level variables test the hypothesis that changes in the external environment drive performance changes in business groups, whereas firm-level variables test the hypothesis that firm-level behaviors drive performance changes. While this article has verified the explanatory power of each hypothesis or variable, certain differences among the alternatives are revealed. First, among the two market-level variables, this article finds that increasing sector-level competition has not necessarily put business groups into a more disadvantageous position, while the maturation of market mechanisms is not much more robust in explaining the decreasing performance of business groups in China. More importantly and interestingly, this article finds that firm-level variables, such as unrelated diversification and asset diversion, are much more robust and powerful in explaining the declining

Causes for changing performance in a transition economy

2069

performance of business groups. We find that while diversification still creates values, its marginal contribution has decreased over time. This pattern seems to reflect the still transitional nature of the Chinese economy. We also find that while the state-ownership variable negatively affects the values of firms in general, it is not the cause of the worsening valuation of business groups when other things are controlled. We should probably not try to sell the results too strongly because all the explanations have their own relevance, and our proxies for the hypotheses cannot be perfect. However, if the regressions results are to evaluate the relative importance of the alternatives, we have no choice but to assign more weight to firm-level variables, especially to the diversification discount hypothesis. In summary, while overall marketization of the Chinese economy seems to have affected the performance of business groups to a certain extent, Chinese business groups are doing equally well in terms of market competition. It is primarily some firm-level wrong-behavior that produces the worsening value performance of Chinese business groups. To put it differently, unless they pursue unrelated diversification and asset diversion, markets (investors) will not necessary value them low. We believe that the declining trend in the performance of business groups in China does not mean that business groups will continue to perform badly. It is still possible for them to evolve further, which will be the topic of our future research. Korean Chaebols are fine examples of continuing evolution and learning, as they showed a stunning turn-around in the 2000s after their discouraging performance in the pre-crisis period in the 1990s. In the case of Korean Chaebols, the most notorious agency costs were in the form of excessive investment. However, they rid themselves of this problem during the crisis-and-restructuring period and thus were born again as successful global players. Similar things could happen to Chinese business groups. Too much diversification or asset diversion should be corrected. The findings of this article, combined with the recent literature on business groups, imply that diverse factors are involved in determining the performance of diverse forms of firms in different countries. Moreover, these diverse forces that keep firms heterogeneous are as strong as the forces promoting homogeneity.

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Acknowledgements
The authors would like to thank valuable comments from the two referees.

Funding
K.L. acknowledges the support of the World Class University program through the Korea Science and Engineering Foundation funded by the Ministry of Education, Science, and Technology (R32-20055) of Korea.

2070

B.-K. Seo et al.

References
Amsden, A. H. and T. Hikino (1994), Project execution capability, organizational know-how and conglomerate growth in late industrialization, Industrial and Corporate Change, 3(1), 111147. Baek, J., J. Kang and I. Lee (2006), Business groups and tunneling: evidence from private securities offerings by Korean chaebols, Journal of Finance, 61(5), 24152449. Berger, P. G. and E. Ofek (1995), Diversifications effect on firm value, Journal of Financial Economics, 37, 3965. Bertrand, M., P. Mehta and S. Mullainathan (2002), Ferreting out tunneling: an application to Indian business groups, The Quarterly Journal of Economics, 117(1), 121148.
Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

Byrd, W. (1987), Impact of the two-tier plan/market system in Chinese industry, Journal of Comparative Economics, 11(3), 295308. Campa, J. M. and S. Kedia (2002), Explaining the diversification discount, The Journal of Finance, 17(4), 17311762. Choo, K., K. Lee, K. Ryu and J. Yoon (2009), Performance change of the business groups in Korea over two decades: investment inefficiency and technological capabilities, Economic Development and Cultural Change, 57(2), 359386. Coase, R. H. (1937), The nature of the firm, Economica, 4(16), 386406. Ding, X. L. (2000), The illicit asset stripping of Chinese state firms, The China Journal, 43, 128. Ferris, S. P., K. A. Kim and P. Kitsabunnarat (2003), The costs (and benefits?) of diversified business groups: the case of Korean chaebols, Journal of Banking & Finance, 27, 251273. Gao, L. and G. Kling (2008), Corporate governance and tunneling: empirical evidence from China, Pacific-Basin Finance Journal, 16(5), 591605. Goto, A. (1982), Business groups in a market economy, European Economic Review, 19, 5370. Graham, J., M. Lemmon and J. Wolf (2002), Does corporate diversification destroy value? The Journal of Finance, 57(2), 695720. Granovetter, M. (1995), Coase revisited: business groups in the modern economy, Industrial and Corporate Change, 4(1), 93130. Guillen, M. (2000), Business groups in emerging economies: a resource-based view, Academy of Management Journal, 43(3), 362380. Hahn, D. and K. Lee (2006), Chinese business groups: their origins and development, in S.-J. Chang (ed.), Business Groups in East Asia: Financial Crisis, Restructuring, and New Growth. Oxford University Press: Oxford, UK. Keister, L. (2000), Chinese Business Groups; The Structure and Impact of Inter-firm Relations During Economic Development. Oxford University Press: Oxford, UK. Khana, T. and Y. Yafeh (2007), Business groups in emerging markets: paragons or parasites? Journal of Economic Literature, 45(2), 331372. Khanna, T. (2000), Business groups and social welfare in emerging markets: existing evidence and unanswered questions, European Economic Review, 44, 748761.

Causes for changing performance in a transition economy

2071

Khanna, T. and K. Palepu (1997), Why focused strategies may be wrong for emerging markets, Harvard Business Review, 75, 4151. Khanna, T. and K. Palepu (1999), The right way to restructure conglomerates in emerging markets, Harvard Business Review, 125134. Khanna, T. and K. Palepu (2000a), Is group affiliation profitable in emerging markets? An analysis of diversified Indian business groups, Journal of Finance, 55, 867891. Khanna, T. and K. Palepu (2000b), The future of business groups in emerging markets: long-run evidence from Chile, Academy of Management Journal, 43, 268285. Kock, C. J. and M. F. Guillen (2001), Strategy and structure in developing countries: business groups as an evolutionary response to opportunities for unrelated diversification, Industrial and Corporate Change, 10(1), 77113. Lang, L. and R. Stulz (1994), Tobins Q, corporate diversification and firm performance, Journal of Political Economy, 102, 12481280. La Porta, R., F. Lopez-de-Silanies and A. Shleifer (1999), Corporate ownership around the world, Journal of Finance, 54, 471517. Lee, G. B., M. Peng and K. Lee (2008), From diversification premium to diversification discount during institutional transition, Journal of World Business, 43(1), 4765. Lee, K., J. Kim and Y. Lee (2010), Long-term evolution of the firm value and behavior of business groups: Korean chaebols between weak premium, strong discount, and strong premium, Journal of the Japanese and International Economies, 24(3), 412440. Lee, K. and Y. Kang (2010), Chapter 8: business groups in China, in A. Colpan, T. Hikino and J. Lincoln (eds), Oxford Handbook of Business Groups. Oxford University Press: Oxford, UK, pp. 210236. Lee, K. and X. He (2009), Capability of the samsung group in project execution and vertical integration: created in Korea and replicated in China, Asian Business & Management, 8(3), 277299. Lee, K. and X. Jin (2009), The origins of the business groups in China: an empirical testing of the three paths and the three theories, Business History, 51(1), 7799. Lee, K. and D. Hahan (2007), Market competition, plan constraints, and the hybrid business groups: explaining the business groups in China, Seoul Journal of Economics, 20(4), 481505. Lee, K. and W. T. Woo (2002), Business groups in China compared with Korean chaebols, in J. Dutta (ed.), Post-crisis Challenges for Asian Industrialization. Elsevier Science Press: Amsterdam, The Netherlands. Leff, N. (1978), Industrial organization and entrepreneurship in the developing countries: the economic group, Economic Development and Cultural Change, 27, 661676. Lins, K. and H. Servaes (2002), Is corporate diversification beneficial in emerging markets? Financial Management, 31(2), 531. Liu, Q. and Z. Lu (2004), Earnings Management to Tunnel: Evidence from Chinese Listed Firms, University of Hong Kong working paper.

Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011

2072

B.-K. Seo et al.

Meyer, M. W. and X. Lu (2004), Managing indefinite boundaries: the strategy and structure of Chinese business firm, Management and Organization Review, 1, 5786. Nelson, R. (1991), Why do firms differ and how does it matter, Strategic Management Journal, 12(2), 6174. Nelson, R. (2009), Why do firms differ and how does it matter? a Revisitation, Seoul Journal of Economics, 21(4), 607619. Penrose, E. (1959), The Theory of Growth of the Firm. Oxford University Press: New York. Rajan, R., H. Servaes and L. Zingales (2000), The cost of diversity: the diversification discount and inefficient investment, The Journal of Finance, 55(1), 3580. Sachs, J. and W. T. Woo (1994), Structural factors in the economic reform of China, Eastern Europe and the formal Soviet Union, Economic Policy, 18, 101145. Sachs, J. and W. T. Woo (1997), Understanding Chinas economic performance, NBER Working Paper No. 5935. Sachs, J. and W. T. Woo (2003), Chinas economic growth after WTO membership, Journal of Chinese Economic and Business Studies, 1(1), 131. Scharfstein, D. and J. Stein (2000), The dark side of internal capital markets: divisional rent-seeking and inefficient investment, Journal of Finance, 55(6), 25372564. Seo, B. K. (2006), Time trends in the performance of business groups in emerging economies: the case of business groups in China, PhD thesis. Seoul National University. Shirai, S. (2002), Is the equity market really developed in the Peoples Republic of China?, ADB Research Paper No. 41. Smyth, R. (2000), Should China be promoting large-scale enterprises and enterprise group? World Development, 28(4), 721737. Stulz, R. M. (1990), Managerial discretion and optimal financing policies, Journal of Finanical Economics, 26, 327. Tenev, S. and C. Zhang (2002), Corporate Governance and Enterprise Reform in China; Building the Institutions of Mordern Markets. World Bank and the International Finance Corporation: Washington, DC. Villalonga, B. (2004a), Diversification discount or premium, new evidence from the business information tracking series, The Journal of Finance, 59(2), 479506. Villalonga, B. (2004b), Does diversification cause the diversification discount? Financial Management, 33, 527. Williamson, O. (1975), Markets and Hierarchies; Analysis and Antitrust Implications. The Free Press: New York. Zattoni, A. (1999), The structure of corporate groups: the Italian case, Corporate Governance: An International Review, 7, 3848. Zattoni, A., T. Pedersen and V. Kumar (2009), The performance of group-affiliated firms during institutional transition: a longitudinal study of Indian firms, Corporate Governance: An International Review, 17(4), 510523.
Downloaded from http://icc.oxfordjournals.org/ at CBS Library on December 14, 2011