Ownership Structure and Corporate Performance: Evidence of Pharmaceutical and Food and Beverage Industries in the

Ownership Structure and Corporate Performance: Evidence of Pharmaceutical and Food and Beverage Industries in the

Chinese Mainland from 2001 to 2008

Lu Gan (1546414) University of Groningen

Faculty of Economics and Business

L.Gan@student.rug.nl

Abstract

We subdivided ownership structure into ownership concentration and owner identity. We tested their impacts on firm performance under different performance measures (accounting profit rate and Tobin's Q). We also test the impacts of ownership concentration with different regression methods (the OLS and the 2SLS). We find significant bell-shaped parabolic relation between ownership concentration and firm performance. There is an insignicant U-shaped parabolic relation between managerial ownership and firm performance. The main owner identities (family holding and government holding) hinder firm performance. The similar results hold in industrial sub-sample tests, however, the results in the pharmaceutical industry show more stability and significance than those in the food and beverage industry.

Keywords: Ownership structure, Corporate performance, Chinese Mainland,

Pharmaceutical industry, Food and Beverage industry

1. Introduction

1.1 Research Area and Research Topic

The relationship between firm ownership structure and firm performance has been the topic of an important and ongoing debate in the corporate finance research area. This debate has a long history. Berle and Means (1932) suggest that there should be an inverse correlation between the diffuseness of firm shareholdings and firm performance. However, decades later, Demsetz (1983) argues that the ownership structure of a corporation should be considered as an endogenous outcome of decisions that reflect the influence of shareholders. An ownership structure should maximize the shareholder return, no matter if it is diffused or concentrated.

Therefore, there should not be a systematic relation between variations in ownership structure and variations in firm performance. Later on, Leech (1987) applies a Simple Game Theory to explain the relation between ownership concentration and firm performance. His research shows that ownership concentration may strongly affect firm performance. However, the final effects depend on the trade-off between the benefits and the costs ownership concentration brings to the firm. Furthermore, research findings also vary among different countries. For instance, Yabei and Shigemi (2008) find a U shaped relation between the shareholder concentration and the economic performance in Japan. However, Thomsen et al. (2000) study the largest European companies and find a positive relation between the two variables, but the effect levels off when the shareholdings are high. Here in the paper, we keep on studying the relation between ownership concentration and firm performance by using the most recent evidence (from 2001 to 2008) from the Chinese Mainland.

However, ownership concentration is not the only measure of ownership structure. There has

also been a group of researchers doing researches on owner identity, such as Sun et al. (2002)

and Conett et al. (2007). Owner identity matters mainly due to the principal-agent problem

(Jensen and Meckling, 1976; Jensen, 1986) that controlling and owning separation within a

firm influences the firm performance. Thus, the influence gained from owner identity on firm

performance is also a valuable research aspect when studying ownership structure. Here in

this paper, we will do investigation on the influence of certain owner identities on firm

performance as well. In other words, in this study, we focus on the relation between firm

ownership structure and firm performance, where the ownership structure is actually classified into two dimensions: ownership concentration and owner identity.

1.2 Background and Context of the Research

Since the Chinese Mainland launched the opening policy in the nineteen seventies, one of the key developments in the mainland has been the ownership structure development, which is mainly about transferring from state ownership to private ownership. The first stage started during 1978 and 1991 from the development in the countryside and finally gave the birth to small village factories owned partially or fully by farmers. The next stage was from 1992 to 1997, which was the further development that mainly focused on modernizing the ownership structures of large and middle-sized state-owned firms. Then the third stage was from 1998 to 2001, which was about the pluralism of major ownership structures. Many large companies gain more capital by applying shareholding ownership while the government may still be the largest shareholder (XinHua Net, 2008). Furthermore, the entry into the WTO in 2001 has deepened the process of the pluralism of ownership structures. The general policy is focusing on the efficiency of ownership, instead of on a certain ownership type (state ownership or private ownership). There is a tendency for state ownership to withdraw from common competitive fields. Instead, the government has been encouraging the variability of ownership since China’s entry into the WTO (KejianHome, 2009).

However, the similar political guildlines actually have affected companies differently during

the past several years. Some firms benefit from the ownership structure development. For

example, Northeast Medicine saved an unprofitable product (¥20 million loss per year) by

cutting labor requirement and material costs after it changed the original state ownership

structure into a joint-stock system (Company Annual Report, 1999). At the meantime, 3658

state owned companies went bankrupt during the ownership structure development from 1994

to 2005 (Fei, 2008). The inconsistent impacts of the ownership structure development on firm

performance make it necessary, important and interesting to study the relation of ownership

structure and firm performance once more, especially as China anticipates in more and more

international affairs with increasing impacts after its entry into the WTO in 2001. Specifically, we will compare the pharmaceutical industry and the food and beverage industry. The two industries are both common competitive industries, where the government plans to withdraw its original ownership smoothly through the ownership structure development. The similar guidelines for developing make them more comparable. On the other hand, different industrial features make the comparison more apparent, such as different knowledge requirements, different price ranges and so on.

1.3 Purpose of the Research

According to the statement above, the purpose of this paper is to investigate the relation between firm ownership structure (ownership concentration and special owner types, both in terms of the largest shareholding portion) and firm performance by applying the evidence from the Chinese Mainland after her entry into the WTO (2001) to 2008. In this paper, we also intend to investigate the possible industrial differences by comparing the evidence from two industries, namely, the pharmaceutical industry and the food and beverage industry.

1.4 General and Specific Research Questions

In order to meet the purposes above, we define our general research question: What is the relationship between ownership structure and firm performance in the Chinese Mainland after her entry into the WTO? To classify this research question into two dimensions, we have three specific research questions: 1) What is the impact of ownership concentration (the largest shareholdings) on firm performance? 2) How do the management shareholdings affect firm performance? 3) Do different identities of the largest shareholder also affect firm performance differently? The last specific research question is about the comparison between the two industries: Is the relation between ownership structure and firm performance different between the pharmaceutical industry and the food and beverage industry?

1.5 Contribution of the Research

There are some main contributions in this paper. First, as mentioned above, we classify ownership structure into two dimensions: ownership concentration and owner identity. We intend to find the relationship between ownership concentration and firm performance as well as the impact of different owner identities on firm performance.

Secondly, the research is based on the most recent evidence (from 2001 to 2008) from the Chinese Mainland. Historically, studies are done mainly in some developed countries and regions, such as the United States (Anderson and Reeb, 2003), the European Union (Thomsen et al., 2000) and Taiwan (Lin et al., 2008). Due to the difficulties of collecting data and language differences, the Chinese Mainland actually has not yet received much attention.

However, it is highly beneficial to do research on the Chinese Mainland as it has been growing dramatically during the past several decades and its roles in the world economy and politics have become even more important after her entry into the WTO.

Thirdly, many authors do research studies just based on one specific industry, for example, Mudambi and Nicosia (1998) work on the UK financial service industry specifically. Many other researchers use many industry dummies in their regression models to find specific industrial effects, for example, Leech and Leahy (1991) include 23 industry dummies to study the largest British companies. However, this paper focuses specifically on the pharmaceutical industry and the food and beverage industry. These two industries indeed have many differences. However, their similarities make the following comparison more meaningful. For instance, the historically state-owned structure gives them similar starting points in privatization and the ownership structure development. Hence, we are able to find out whether the relationship differs even in industries with similarities.

The final contribution is about the methodology development. First, we use different items to

present firm performance, namely, accounting profit rate and Tobin’s Q to see if the

conclusions are consistent. Secondly, we apply the Simultaneous Equation Model that

considers firm performance as an endogenous variable of ownership concentration. Large

shareholders restructure their ownership based on their future expectations for firm

performance. We do both the OLS regression and the 2SLS regression for testing the effect of ownership concentration in this paper.

1.6 Research Structure

The paper is structured as follows: In the next chapter, we will first introduce the main theoretical frameworks (the Simple Game Theoretical Approach and the agency theory). Then we will present previous studies on the relation between ownership structure and firm performance. Grounded on the early studies, we will set up our own hypotheses for this paper.

Thereafter, Chapter 3 will be about our data and methodology. After the data description, we will first define the variables in this paper. This will be followed by the introduction of our other regression method besides the OLS: the 2SLS (the Simultaneous Equation Model). The final section of this chapter will be the presentation of our regression models. Then in Chapter 4, we will present and analyze the results. Finally, a brief conclusion will wrap up the whole paper including some extensive discussion of the implementations and the future research recommendations.

2. Literature Review

In this chapter, we will first introduce two basic frameworks for our research, namely, the Simple Game Theoretical Approach and the agency theory. Then we will go through the previous relevant studies to build up our hypotheses, in order to answer the research questions of this study.

2.1 Main Theoretical Framework

2.1.1 Simple Game Theoretical Approach (SGTA)

The SGTA addresses the relation between ownership and firm performance in a theoretical

model. Leech (1987) develops a model to study the relation between the controlling power

and the shareholdings. Furthermore, he constructs a theory about how ownership

concentration may influence firm performance. His model starts with a Simple Game Theory.

There is an arbitrary N-person with a set of players N = {1, 2,…N} and a characteristic function V (.) which associates with each coalition T  N . The characteristic function satisfies the following conditions:

(i) V (  )  0 ,   1 , 2 , 3 ...., N , the expected value of participating in the game for each player.

(ii) V ( S  T )  V ( S )  V ( T ), for all sets of S , T  N

Shareholder voting can be conceived in terms of a simple game where coalitions are marked into "winning" or "losing". When V ( T )  1 , T is "winning" and when V ( T )  0 , it is "losing".

Voting games are that the players possess different numbers of votes and "winning" is defined as the majority votes.

Several shareholders need to form a leading coalition to win the majority votes, so that they finally have the ability to organize their voting power to enforce wealth maximization. In Leech's paper (1987), he focuses on ownership concentration of a coalition which is measured

by 





i

i P n

P

1

) ( )

( , where P (i ) is the member's proportional shareholdings and  ^{P (i )  1 .}

)

 (n is the power index (Shapley-Shubik Power Index, Banzhaf power index or the degree of control of Cubbin and Leech). The index can be gained by each coalition of n members.

Figure 1 presents the power curves corresponding to two concentration curves. First, the

straight line AB shows the equal distribution among players. Curve AD is the power curve of

a coalition of N/2 members, where P ( N / 2 )  1 / 2 and  ( N / 2 )  1 . If the concentration

increases while the N is constant, the concentration curve becomes curve BC. In this case, the

controlling coalition reduces its members to n ' to keep the same value of P equaling to 1/2

and the same value of θ equaling to 1. The new power curve is curve EF. Therefore, the

conclusion is that ownership concentration reduced the size of a controlling coalition and

increased its power.

Figure 1: Concentration curve and Power curve, adapted from Leech (1987), pp. 230

Leech also considers the situation when the concentration increases while N reduces as the largest shareholder extends by buying out others. Figure 2 shows the initial concentration curve AB. Then after the buyout of large holdings, N reduces to N

and the concentration curve shifts left. The new concentration curve is CD. The corresponding power curve is JK instead of the initial FG. However, the concentration curve becomes CE when the total holdings become N

, when the largest shareholder increases the same amount holdings by taking over only the lower end of the size distribution with the power curve of HI. The conclusion is the enhancement of the largest shareholder's power is greater when the shareholder adds a group of large holdings instead of requisiting small holdings even when the concentration enhancement is the same. However, it may be difficult to compare these two extension options in terms of increasing the power of the coalition in general if HI is steeper than JK.

Figure 2: Changes in Concentration curve and Power curve with changing N, source: Leech (1987), pp. 230

As mentioned earlier, ownership concentration provides coalitions voting power and the

ability to enforce wealth maximization. Hence, ownership concentration may benefit the

entire firm when it leads to more focuses on the shareholder wealth increasing. However, how ownership concentration influences firm performance actually depends on the comparison of the potential benefits and the possible associated costs. Leech (1987) also lists some costs that may occur, to complete his theory of ownership concentration. Examples would be the information costs of monitoring and evaluating policies regarding to investment, financial structure, R&D, marketing, as well as costs to improve the accountability to shareholders and the fixed costs of organizing a controlling coalition.

Leech's (1987) model and theory provide a theoretical foundation of our research on the relation between ownership concentration and firm performance. The ownership concentration (presented by the largest owner shareholdings) may reduce the size of the controlling coalition while increase its power to enforce wealth maximization for the entire firm. It may also affect the ease and the costs of coalition formation, organization and communication, which will also influence the firm performance. Therefore, ownership concentration may strongly affect firm performance; however, the final effects depend on trade-off between the benefits and the costs it may bring to the firm.

2.1.2 Agency Theory

Concerning the influence of managerial ownership on firm performance, we use the

well-known agency theory as the framework of our research. It is assumed that managers and

shareholders have fundamentally different and conflicting interests in a company. The utility

of shareholders depends completely on the market value of firm. However, managers are

willing to pursue other goals at the expense of profits, such as high compensation, low effort

levels, expense preference, empire building and so on. Then under these assumptions, the

benefit accruing to shareholders from control is the difference between the maximum

attainable value of the firm and whatever its value would be under unrestricted management

control, a gross allowance for the costs of organizing and controlling (Leech, 1987). One

typical explanation for the benefits is found in the studies of the diversification strategy. Top

managers have personal interests in diversification due to risk aversion, expense preference

and empire building. However, concentrated coalitions may go against diversification to increase the shareholder value. Demsetz and Lehn (1985), Admati et al. (1994), Bolton and von Tadden (1998) all argue that high ownership concentration limits the diversification and reduces the owners' tolerance toward risks. Hill and Snell (1989) and Hoskisson et al. (1994) both find that large blockholder ownership relates negatively to product diversification. Bergh (1995) states that ownership concentration relates positively to the sales of unrelated business units. Obviously, the restructuring also associates with organizing costs. When the costs are greater than the benefits, shareholder constraint is inoperative. In other words, the management constrain appears when the benefits are greater than the associated costs.

All in all, managerial ownership may also affect firm's final performance through a trade-off of benefits and costs. The benefits can be the controlling of loss from diversification or the increasing of motivation to improve firm value by unifying the goals of firm value maximization (Burkart et al., 1997). Costs may occur, such as the costs of monitoring the management or the inefficacy associated with the lack of management knowledge and experience.

The paragraphs above theoretically explain why and how ownership concentration (the largest owner) and owner identity (managerial ownership) relate to firm performance. We will also try to find empirical supports to explain both the relations in the following section.

2.2 Empirical Studies

2.2.1 Ownership Concentration (the Largest Shareholder)

The research on ownership concentration and firm performance has a long history. As mentioned earlier, Berle and Means (1932) state a positive relation between ownership concentration and firm's accounting profitability, which was later found also by Cubbin and Leech (1983). However, Demsetz (1983) considers ownership concentration as an endogenous outcome of shareholders' decisions, which is closely related to firm performance.

Hence, profit maximization may require either diffused or concentrated ownership. Demsetz

and Lehn (1985) find that ownership concentration is insignificantly affecting firm's accounting profitability when other variables are controlled. Later, the Demsetz-Lehn model has been affirmed by evidences from different countries and regions: Swedish evidence by Bergstrom and Rydqvist (1990), South African evidence by Gerson and Barr (1996) and European evidence by Pedersen and Thomsen (1999). However, in Pedersen and Thomsen's (1999) paper, the influence from the ownership structure depends on system effects.

According to the study of Kole (1994), there is even evidence of reverse causality that performance affects ownership rather than the other way round. Lehmann and Weigand (2000) do research on German corporations and find indeed a negative effect of ownership concentration on firm performance. Nevertheless, by applying the stock market data, scholars still often find a positive relation between the two variables. Lloyd et al. (1987) find that for owner-controlled companies, the ratio of market value to sales is higher when ownership concentration is higher. Zeckhouser and Pound (1990) and Leech and Leahy (1991) reach the similar conclusion by applying an owner influence index and the PE ratio as a dependent variable respectively. Some other studies find parabolic shaped relationship between the largest shareholdings and firm performance. For instance, Thomsen and Pedersen (2000) find a bell-shaped relation between the largest shareholdings and the asset returns, a bell-shaped relation between the largest shareholdings and the market-to-book ratio, but no effect of the largest shareholdings on the sale growth. Referring to the Thomsen and Pedersen model and theory framework, we consider that firm performance is the outcome of the trade-off between costs and benefits. In other words, the benefits will finally decrease when the costs are beyond a certain level. Hence, we follow the paper of Thomsen and Pedersen (2000) to build up our first hypothesis: Firm performance is a bell-shaped (first increasing, then decreasing) function of the ownership proportion of the largest owner.

2.2.2 The Managerial Ownership

Referring to the relation between managerial ownership and firm performance, the results are

relatively constant. Most of the previous studies show parabolic shaped relations between

these two variables. Morck et al. (1988), McConnell and Servaes (1990), Hermalin and

Weisbach (1991), Holderness et al. (1999) all find that increasing managerial ownership when it is low can increase firm value but at higher levels of managerial ownership firm value decreases with the managerial ownership increase. Precisely, McConnell and Servaes (1990) even find out that Tobin's Q increases with insider shareholdings up to some 40 percent of total shares outstanding and decreases beyond that level. Here the bell-shaped relation goes with our analysis of the agency theory. We consider that managerial shareholdings can introduce the benefits of increasing management efforts and decreasing perquisite consumption.

However, the effect of this provoke will decrease as management ownership increases.

Managers who have no or little shares probably value the increased shareholdings more than those who have already had quite some shares. The marginal benefits of the management shareholding increase may diminish as the management ownership increases. Hence, here we have our second hypothesis: Firm performance is a bell-shaped (first increasing, then decreasing) function of the management ownership proportion.

2.2.3 Type of the Largest Owner

Regarding to the largest owner identity, first we follow the paper of La Porta et al. (1999) to classify (ultimate) owners

into five types: 1) a family or an individual, 2) the government, 3) a widely held financial institution, 4) a widely held corporation or 5) miscellaneous.

We here follow the theoretical framework used by both Hansmann (1988; 1996) and Pedersen and Thomsen (1999) to compare the benefits and the costs of ownership for each ownership type. In this framework, each firm is considered as a set of contracts with different types of stakeholders. When ownership is assigned to one stakeholder, this stakeholder will have the costs of ownership but will not have the costs of market contracting, such as the costs of making contracts, the costs of information transferring and so on. The opportunity costs of changing ownership assignment are the sum of the increasing costs of ownership and of market contracting. Hence, the optimal ownership type (j) should minimize the transaction

1

According to La Porta et al. (1999), a corporation has a ultimate owner if the shareholder's (in)direct voting rights in the firm

exceed 20 percent. We apply the same owner categories to define the final owner type, but the largest shareholder is not

necessarily the ultimate owner defined by La Porta et al. (1999)

costs, which equals the sum of the costs of ownership (CO) and the costs of market contracting (CC). ^Min （ CO

^  i ^ j CC

） by j, where i is an index of the firm's stakeholders.

According to the previous studies, the costs of ownership (OC) include the costs of monitoring and risk bearing (Jensen and Meckling, 1976) and the costs of collective decision making (Hansmann, 1988). The costs of market contracting (CC) include information costs (Arrow, 1975), the traditional losses due to market power distortions (Tirole, 1992), and the ex post transaction costs associated with asset specificity (Williamson, 1985). This kind of transaction costs incurred by one particular stakeholder or a group of stakeholders can be reduced or be avoided if these stakeholders are owners of the firm since they can internalize their transactions with the firm (Thomsen and Pedersen, 2000).

Family ownership usually refers to double roles: as the owner and the manager of the firm.

Therefore, family ownership may reduce the agency problem through the integration of the principal and the agent. However, Fama and Jensen (1983) and La Porta et al. (1998) consider that family owners may derive their private benefits from running the company at the expense of the minority shareholders. The conflict between large and small shareholders is called expropriation. In earlier studies, there were supports of the expropriation hypothesis. For instance, Johnson et al. (1985) find the stock market reacts favorably to the unexpected death of the CEOs with large shareholdings. However, many recent papers (Anderson and Reeb, 2003;

Martinez et al., 2007) state that there is positive relation between family ownership and firm

performance. Maury (2006) explains the positive relation as follows: the high shareholder

protection reduces the opportunity of expropriation. Thus, the effect of reducing the

majority-minority shareholder agency problem may be more obvious. The Chinese Mainland

has oriented itself on the German law system, which does not provide very strong legal

shareholder protection (La Porta et al., 1998). Hence, we expect that when the largest owner

is a family (individual), the firm performance is worse.

Government ownership internalizes the relationship between the government and company, which functions as an institutional alternative to regulation (Thomsen and Pedersen, 2000).

Shepherd (1989), Laffont and Tirole (1993), Hart et al. (1996) suggest that governments are likely to focus on political goals besides maximizing profits, for example, the employment rate.

Hence, government dominated firms are expected to have lower performance. Nevertheless, some authors suggest that government ownership brings in advantages in terms of credit, liquidity, or the capital the costs of capital since governments are relatively wealthy (Thomsen and Pedersen, 2000). Sun et al. (2002) also support this argument: in the Chinese Mainland, government ownership has a positive impact on partially privatized government-owned enterprises. Thus, we expect the same relation here: when the largest owner is the government, the firm performance is better.

Corporate ownership internalizes the firm’s transaction with input suppliers through the vertical integration of firms along the value chain (Williamson, 1985). The typical examples would be the Japanese Keiretsu, the French cross holding structures and the Swedish business groups (Kester, 1992; Gedajlovic and Yoshikawa, 2005 and Charkham, 1994). The integration among firms through corporate ownership leads to both benefits and costs. The benefits include high input specificity, shorter transaction time, lower transaction costs and better knowledge transfers. However, the costs may be the loss of flexibility and the risk of deficient mutual monitoring (Kester, 1992). Here we set up our hypothesis as follows: when the largest owner is another corporation, the firm performance is better.

Institutional ownership internalizes firm's financing process. It provides firm advantages of

financing, such as lower lending requirements (lower risk aversion) and longer lending

periods. In addition, institutional owners specialize in portfolio investment and have at arm

length relationship with the firm. They focus on the financial results (share value and share

liquidity) even more than other owners do. This is why Cornett et al. (2007) argue that

institutional shareholders are good monitors, who have been increasingly willing to use their

ownership rights to pressure managers to act in the best interests of shareholders. The positive

relation between institutional ownership and firm performance was also supported by the

works of Levin and Levin (1982), McConnell and Servaes (1990) and Nickel et al. (1997).

Therefore, we expect the same relation as well: when the largest owner is an institution, the firm performance is better.

2.2.4 The Industrial Difference

As mentioned earlier, most studies treat industry as a control dummy, such as works from Leech and Leahy (1991), Thomsen and Pederson (2000) and Grosfeld (2006). Therefore, industry can be an influencing factor. Other papers deal with a certain industry, for example, the paper of Cole and Merhran (1998). However, there are bare studies about the pharmaceutical industry, the food and beverage industry or the comparison of these two industries. Hence, it is difficult to set up directly concreted hypotheses on the industry comparison based on previous studies. We will test the three hypotheses in each industrial sub-sample to see if the test results are consistent in the two industries.

2.3 Hypotheses

Summing up, the four hypotheses in this study we try to prove are:

H1: Firm performance is a bell-shaped function of ownership concentration.

H1a: When ownership concentration level is low, there is a positive relationship between it and firm performance.

H1b: When ownership concentration level is high, there is a negative relationship between it and firm performance.

H2: Firm performance is a bell-shaped function of the management ownership.

H2a: When the management shareholding level is low, there is a positive relationship between it and firm performance.

H2b: When the management shareholding level is high, there is a negative relationship

between it and firm performance.

H3: The effects of the different largest shareholder types on firm performance are different.

H3a: When the largest owner is a family (individual), the firm performances worse.

H3b: When the largest owner is the government, the firm performances better.

H3c: When the largest owner is a corporation, the firm performances better.

H3d: When the largest owner is an institution, the firm performances better.

H4: Between the pharmaceutical industry and the food and beverage industry, the relationship between ownership structure and firm performance does not differ.

3. Data and Methodology

In this chapter, we will first describe the data samples and present how we collect our data.

Then we will define the variables in our research based on the preceding research papers. In section 3.3, we will present our research methodology and regression models.

3.1 Data

In this research, we focus companies in the Chinese Mainland. We collect the most recent data

(2001 to 2008) from both Shanghai Stock Exchange (the SSE) and Shenzhen Stock Exchange

(the SZSE). We choose totally 70 pharmaceutical companies and 42 food and beverage

companies that went public in the Chinese Mainland. We especially pick these two industries

because first they have many differences, such as their target markets, product life cycles,

labor skill requirement, and technology developing speed and so on. Meanwhile, they do have

some quite important similarities. As mentioned earlier, in the Chinese Mainland, the

government originally held most of the pharmaceutical and the food and beverage firms,

especially for firms that are large enough to go public. They were usually the large or

middle-sized companies, so that the government started the ownership structure development

mainly in the second development stage from 1992 to 1997 by launching the shareholding

system. In addition, one of the most common privatization methods used in both these two

industries is the management buyout. According to the findings of Grosfeld (2006), with

privatization, for the previously state-owned firms, ownership concentration is sensitive to the legacy of the privatization methods. Therefore, firms in these two industries mostly have the similar starting points as previously state-owned companies and the similar development schedules or methods for the whole process of the ownership structure development. All these similarities make the two industries more comparable so that the conclusion of the comparison is more meaningful and more interesting.

We choose merely firms that went pubic before January 1

2002, so that the numbers of listing years are relatively similar. Nevertheless, we still control the number of listing years later in this research. This data selection rule also makes our data samples relatively complete that we make sure that there are available observations for each company from 2001 to 2008.

We then further leave out one pharmaceutical company and five food and beverage companies, which have multiple business scopes besides pharmaceuticals and food and beverage. We go through the company annual reports to collect data about the largest shareholding portion. We count the sum of the managerial shareholding portions. We also gain data about the after tax net income, the book value of equity and the book value of liabilities. We calculate the market value of equity at each year-end with the market price we collected from the database of a real time stock trading system. Based on all these data, we then calculate accounting profit rate

and Tobin’s Q

. Both these two ratios are used to measure firm performance. The reasons for this indicator choice will be discussed later. We calculate the logarithm of sales

, the yearly sales growth ratio, the debt-equity ratio and the number of years listing. Concerning the largest shareholder

identity, we investigate the actual holdings and the controlling structure of shareholders to identify the final owner identity. A shareholder has X percent indirect control over firm A if (1) it directly controls firm B, which directly controls X percentage voting rights in firm A; or (2) it directly controls firm C, which then controls firm B (or through a control chain with a sequence of firms leading to firm B, each of which has control over the next one), which directly controls

2

Accounting profit rate= after tax net income/book value of equity

3

Tobin’s Q= (market value of equity + book value of liability)/(book value of equity + book value of equity)

4

Sales here mean prime operating revenue since the prime business shows the industrial difference the most.

5

According to most of the annual reports, there is little information about the coalition among large shareholders.

X percentage of the votes in firm A (La Porta et al., 1999). If the largest shareholder has an ultimate owner

, we count the ultimate owner type as the largest shareholder identity. If there is no ultimate owner for the largest owner, we count the firm type as the largest shareholder identity. We collect data of the group enterprise instead of the single mother company since it is difficult to separate some enterprises due to the cross holdings among branch firms.

3.2 Variables Dependent Variable

The dependent variable is firm performance. There are a number of proxies to present the performance of a firm, such as ROA, ROE, growth rate, and Tobin's Q. We follow Demsetz and Villalonga (2001) to choose accounting profit rate and Tobin's Q as the measures of firm performance. There are two major differences between these two measures. The first difference relates to the time perspective. Tobin's Q indicates a firm’s growth opportunity and is based on investors' evaluation of its future profitability, so it is forward-looking. However, accounting profit rate is backward looking. A high Tobin's Q indicates success in the sense that a firm has deployed the investments and the firm is valued more by the market than its book value. The second difference is about accounting problems in measuring firm performance. The profit rate is measured by the accountant, who has been constrained by standards and accounting practices. For instance, different methods of depreciation may change the profit rate. However, Tobin's Q is measured by investors as well and will be affected by their estimation toward the future.

Independent Variables

Ownership concentration (OC) is presented by the largest shareholdings here. According to earlier studies, for instance, Grosfeld (2006), ownership concentration refers to the largest voting right. However, due to the difficulty of data collection, we cannot obtain detailed information about the voting right distribution. Hence, we assume here the shareholdings

6

Ultimate owner is the shareholder who (in)directly has more than 20 percent votes in the firm (La Porta et al. 1999). If a firm

has more than one ultimate owner, we only investigate the holding structure of the largest one, which is usually also the

controlling shareholder.

directly reflect the voting rights.

Management ownership (MO) is the portioned sum of the shares directly held by the management team members, such as directors, supervisors, board members and so on.

Owner identity (OI) is classified into four ownership categories: family (individual) owners, government owners, corporation owners and institution owners. Therefore, we give different dummy value for different owner identities. If the largest shareholder is a family or an individual, Fam=1, otherwise Fam=0; if the largest shareholder is the government, Gov=1, otherwise Gov=0; if the largest shareholder is a corporation, Cor=1, otherwise Cor=0; if the largest shareholder is an institution, Ins=1, otherwise Ins=0.

Control Variables

Industry is treated as a control variable when we test the whole data sample. The dummy value for the pharmaceutical industry is one; the dummy value for the food and beverage industry is zero. Concerning the concrete industrial differences, we also test the data from two industries separately.

Size was found to relate negatively to ownership concentration (Demsetz and Lehn, 1985).

This is mainly due to risk diversification and wealth limitations, which make it more expensive to gain large equity portion in larger firms (Grosfeld, 2006). However, size may also be a signal of reputation and potential performance. Hence, we have to control this variable when do a research on the relations between ownership and firm performance. Like Grosfeld (2006), we measure size as the logarithm of the firm's sales.

Growth is another control variable. Following the logic of Thomsen and Pederson (2000), we use yearly sales growth rate (current sales/last year sales-1) to present the firm’s growth.

Leverage rate is another control variable. This variable was also controlled in Thomsen and

Pederson's (2000) work. Here we use the debt-to-equity ratio to present the leverage level.

Year dummies should also be controlled since there are institutional changes from the years 2001 to 2008. The policies of the ownership structure development are relaxing to facilitate privatization or the shareholding system. The yearly macro-environment changes are the same to all firms.

3.3 Methodology and Regression Models

We do regressions by applying both the OLS and the 2SLS. Under the OLS, the dependent variable is accounting profit rate or Tobin's Q. The independent variables are the largest shareholding portion, the management shareholding portion and the largest owner identity.

Due to the empirical bell-shaped relation between ownership concentration and firm performance, the independent variables include both ownership concentration and the square from it. The control variables are industry dummy, year dummy, size, growth and leverage.

Equation 1 (Ownership Concentration)

 ^







   OC  OC  ControlVar iables 

Y

*

*

* (1)

y debt/equit leverage

1;

- sales year sales/last current

growth

; log(sales) size

2008;

for 8 2002,..., for

2 2001, for 1 dummy year

industry;

beverage and

food for 0 industry, ical

pharmaceut for

1 dummy industry

: Variables Control

portion ng

shareholdi largest

: OC

Q

s Tobin' or rate profit accounting :















 Y

Equation 2 (Management Ownership)

 ^







   MO  MO  ControlVar iables 

Y *

*

*

1

2 ¹

1

(2)

y debt/equit leverage

1;

- sales year sales/last current

growth

; log(sales) size

2008;

for 8 2002,..., for

2 2001, for 1 dummy year

industry;

beverage and

food for 0 industry, ical

pharmaceut for

1 dummy industry

: Variables Control

portion ng

shareholdi management

: MO

Q s Tobin' or rate profit accounting :

















Y

Equation 3 (Owner Identity)













      

 ^{Fam Gov Cor Ins}  ^{ControlVar iable}

Y

*

*

*

*

* (3)

y debt/equit leverage

1;

- sales year sales/last current

growth

; log(sales) size

2008;

for 8 2002,..., for

2 2001, for 1 dummy year

industry;

beverage and

food for 0 industry, ical

pharmaceut for

1 dummy industry

: Variables Control

n institutio an

is owner largest the

if 4, Ins

n corporatio a

is owner largest the

if 3, Cor

government the

is owner largest the

if 2, Gov

l) (individua family

a is owner largest the

if 1, Fam

Q s Tobin' or rate profit accounting :























 Y

Then we test the regression considering ownership concentration as an endogenous variable of the firm performance. In other words, we here apply the Simultaneous Equation Model (Demsetz and Villalonga, 2001). Investors change their shareholdings according to firm performance. Therefore, firm performance may change ownership concentration by changing investors’ expectation, which makes Tobin’s Q a better indicator of firm performance under the 2SLS test. This proxy choice is also supported by Grosfeld (2006).

Equation 4: Equations in the 2SLS test

Tobin’s Q= f (constant, leverage, growth, size, year, ind, OC, OC

) )

Q s Tobin' ing, years_list constant,

( g

OC  (5)

4. Results

In this chapter, we will first present the descriptive statistics for the full data sample and the

industrial sub-samples respectively. Then we will present and analysis the regression results

based on the full data sample in Section 4.2. Afterwards, we will compare and analysis the

results on the two industrial samples. We will provide both the theoretical fitness and practical

explanation of our results.

4.1 Descriptive Statistics

The entire sample includes data of 70 pharmaceutical companies and 42 food and beverage companies with 8 years (2001 to 2008). However, the book value of equity was negative according to the annual reports for eight companies in certain years. We treated these negative data as unavailable data. Theoretically, the book value of equity should not be negative if the firm has not gone bankrupt. In addition, usually when the book value of equity is negative, the after tax net profit is also negative, which leads to a positive accounting profit rate. These positive rates present the firms' performance in a wrong direction, which may introduce large biases during regressions. Hence, it is better to treat these data as missing data. Then concerning accounting profit rate and Tobin's Q, we have 871 observations left in the entire data sample, 541 observations left in the pharmaceutical sample and 330 observations left in the food and beverage sample.

Obs.No. Mean Median Max Min St.Dev.

Accounting Profit Rate 871 (0.01) 0.06 3.16 (16.77) 0.68

Tobin_q 871 2.50 1.93 20.91 0.83 1.90

OC 896 0.41 0.40 0.76 0.05 0.17

Man 896 0.00 0.00 0.06 0.00 0.00

Fam 222

Gov 607

Cor 65

Ins 34

Size 896 8.87 8.91 10.41 5.91 0.59

Growth 896 0.28 0.13 45.52 (0.91) 2.01

Leverage 871 1.67 0.81 161.06 0.06 6.04

Table 1: Descriptive Statistics Summary for the Full Sample

According to Table 1, the mean accounting profit rate is -1% for the entire sample. Tobin's Q fluctuates more than accounting profit rate since its standard deviation is greater. This is mainly because Tobin's Q is a measure that includes the market expectations on a firm. It relates to a firm’s market value, which is indeed more variable than its accounting measures.

On average, the largest shareholder holds 41% shares of a firm. However, the maximum case counts for 76% of the firm's entire shareholdings while the minimum case counts for only 5%.

The management shareholdings are quite small with the average shareholdings of less than

1% and the maximum case of 6%. For most companies in most years, the largest ultimate shareholder is the government or an individual. The shareholdings of the largest shareholder and the management team present the typical ownership structure in the Chinese Mainland.

Going back to history, most of companies (large enough to go public) in these two industries were held first by the government. Under this circumstance, the management gained payment generally according to the positional rankings. The incentive for management to hold firm shares was not very strong. Later on, as the government reduces or even withdraws its shareholdings, individual gradually becomes the largest ultimate shareholder. Either they assign management teams that usually receive negotiated (fixed) payments or it is difficult to distinguish owning and managing. Referring to corporation and institution shareholdings, there are only a few observations (see Table 2 and Table 3) when we go up along the holding structures to find the ultimate owner. Either the government or an individual ultimately holds most of corporations and institutions.

Phar.Firm 70

Obs.No. Mean Median Max Min St.Dev.

Accounting Profit Rate 541 (0.03) (0.01) 3.16 (16.77) 0.85

Tobin_q 541 2.37 1.90 14.52 0.83 1.64

OC 560 0.39 0.39 0.75 0.05 0.16

Man 560 0.00 0.00 0.06 0.00 0.01

Fam 161

Gov 346

Cor 42

Ins 11

Size 568 8.83 8.85 10.22 5.91 0.61

Growth 568 0.25 0.13 30.19 (0.91) 1.45

Leverage 541 1.90 0.88 161.06 0.06 7.44

Table 2: Descriptive Statistics Summary for the Pharmaceutical Sample

Food Firm 42

Obs.No. Mean Median Max Min St.Dev.

Accounting Profit Rate 330 0.01 0.01 0.48 (3.07) 0.30

Tobin_q 330 2.68 2.02 20.91 0.84 2.23

OC 336 0.43 0.41 0.76 0.10 0.16

Man 336 0.00 0.00 0.01 0.00 0.00

Fam 61

Gov 261

Cor 23

Ins 23

Size 336 8.94 8.94 10.41 6.78 0.55

Growth 336 0.31 0.12 45.52 (0.81) 2.64

Leverage 330 1.34 0.73 27.44 0.08 2.88

Table 3: Descriptive Statistics Summary for the Food and Beverage Sample

According to Table 2 and 3, the pharmaceutical industry and the food and beverage industry have quite different features in terms of ownership structure. The average accounting profit rate is negative in the pharmaceutical industry, while it is positive in the food and beverage industry. The maximum and minimum accounting profit rates both belong to the pharmaceutical industry. The different standard deviations also show that the fluctuation of accounting profit rate is greater in the pharmaceutical industry than that in the food and beverage industry. However, Tobin's Q varies more in the food and beverage industry, which indicates the market prices for the food firms may fluctuate more than those of the pharmaceutical firms. This is also very logical because the market expectations toward the pharmaceutical firms are usually connected with knowledge improvement or medicine development. The time frame may be much longer than with new food or beverage product launch. The largest owner shareholdings are similar in both industries in terms of mean value, maximum and minimum value and even standard deviations. They also have similar ownership structure history, so that the management shareholdings are both counted for small relative percentage value. However, the maximum case in the pharmaceutical industry is much bigger than that in the food and beverage industry. This may indicate an important difference between the two industries that running a pharmaceutical business probably requires more special expertise than running a food and beverage business. Management groups in the pharmaceutical industry generally have more specific expertise than those in the food and beverage industry. Actually, the privatization in the pharmaceutical companies is usually accomplished by some technicians or pharmaceutical experts within the firms. After they become the largest ultimate owners of firms, they can still manage the running business with their pharmaceutical expertise, which makes the difference between owning and managing become vague. This special feature is also indicated by data of the owner identity.

There are more individuals as the largest ultimate shareholder in the pharmaceutical industry

than in the food and beverage industry in terms of both absolute number and relative ratio (161 and 61 observations, 28.75% and 18.155% of the total industrial observations).

We also run correlation test to see if there is serious autocorrelation problem which needs to be dealt with before running regressions. According to Table 4, 5, and 6, for all three samples, the autocorrelation between variables is not serious. Though the correlation between accounting profit rate and leverage rate is a bit high, they present two different aspects of firm and are two quite different variables. Therefore, we can not omit one for another.

GROWTH ACCOUNTING_PROFIT_RATE LEVERAGE SIZE TOBIN_Q

GROWTH 1.00 0.04 0.00 0.03 0.09

ACCOUNTING_PROFIT_RATE 0.04 1.00 (0.33) 0.19 0.06

LEVERAGE 0.00 (0.33) 1.00 (0.06) (0.03)

SIZE 0.03 0.19 (0.06) 1.00 (0.05)

TOBIN_Q 0.09 0.06 (0.03) (0.05) 1.00

Table 4: Correlation Summary for the Full Sample

GROWTH ACCOUNTING_PROFIT_RATE LEVERAGE SIZE TOBIN_Q

GROWTH 1.00 0.05 (0.02) 0.06 0.27

ACCOUNTING_PROFIT_RATE 0.05 1.00 (0.23) 0.18 0.05

LEVERAGE (0.02) (0.23) 1.00 (0.09) (0.02)

SIZE 0.06 0.18 (0.09) 1.00 (0.22)

TOBIN_Q 0.27 0.05 (0.02) (0.22) 1.00

Table 5: Correlation Summary for the Pharmaceutical Sample

GROWTH ACCOUNTING_PROFIT_RATE LEVERAGE SIZE TOBIN_Q

GROWTH 1.00 0.07 0.05 0.00 (0.02)

ACCOUNTING_PROFIT_RATE 0.07 1.00 (0.25) 0.28 0.11

LEVERAGE 0.05 (0.25) 1.00 0.00 (0.07)

SIZE 0.00 0.28 0.00 1.00 0.12

TOBIN_Q (0.02) 0.11 (0.07) 0.12 1.00

Table 6: Correlation Summary for the Food and Beverage Sample

4.2 Results from the General Sample Equation 1: Ownership Concentration

The OLS test for Equation 1 shows that ownership concentration does not have significant impact on firm performance (measured by accounting profit rate) after controlling other variables (see Table 7). This result fits the result of Demsetz and Lehn (1985) very well.

However, the hypothetic bell-shape relationship does exist. On the other hand, when performance is measured by Tobin's Q ratio, ownership concentration has significant effect (10% confidence level) on firm performance. The bell-shape relation between the two variables is still valid under this circumstance. However, the R

is a bit lower in this case, which may be because the discrete pattern (huge time intervals among observations)

of yearly based data. Therefore, we can conclude that there is a bell-shaped parabolic relation between ownership concentration and firm performance. Under both two performance indicators, firm performance increases with largest shareholdings up to around 40 percent

of total shares outstanding and decreases beyond that level. This is reasonable because increasing concentration when most of the shareholders are small can effectively reduce the costs of monitoring and evaluating decisions regarding to investing, financial restructuring and R&D.

It should also be able to reduce the costs of improving shareholders' account checks easily. At this moment, the benefits gained from cost reduction may apparently increase a firm’s performance. However, when the concentration is too high (over 40 percent), the marginal benefits gained from cost reduction probably decrease. Meanwhile, an agency problem between large and small shareholders may also occur. Then large shareholders may expropriate small shareholders, who may have the veto rights on some decisions benefit the entire company (Armour et al., 2009). Therefore, it is reasonable that beyond a certain level of ownership concentration, firm performance will decrease with the increase of ownership concentration.

Dependent Variable Accounting profit rate Tobin's Q

Independent Variables Coefficient Probability Coefficient Probability

7

The data bias is more apparent for Tobin's Q ratio since with the influence of expectations, the fluctuation of Tobin's Q is supposed to be bigger than that of accounting measures within the same time interval .

8

33% under the appliance of accounting profit rate and 40% under the appliance of Tobins’ Q

OC

OC

R

Table 7: Statistics Results of Ownership Concentration for the Full Sample

Equation 2: Managerial Ownership

Both tests under the appliance of accounting profit rate and Tobin's Q show an insignificant U-shape parabolic relation between managerial ownership and firm performance.

Dependent Variable Accounting profit rate Tobin's Q

Independent Variables Coefficient Probability Coefficient Probability

Man

Man

R

Table 8: Statistics Results of Managerial Ownership for the Full Sample

This result goes against our hypothesis and the traditional agency theory. However, it indeed fits with the findings of Lin et al. (2008)’s research on non-financial companies in Taiwan that there is a significant positive relationship (alignment) between ownership and firm performance when the net ownership

is high and an insignificantly negative relation (entrenchment) between the two when the net ownership is low.

This phenomenon can be explained by the typical management shareholding style in the Chinese Mainland. Although nowadays management also owns some shares, not all the managerial shareholdings are counted as management compensation with the motivation purpose. Many managers trade their firms’ shares (relatively small amounts compared to their salaries) through the second stock market to speculate on their internal information. The random managerial shareholding pattern may explain the consistent insignificance of the

9

Net ownership= ownership - collateralized shares

impacts. In contrast, most of management compensation is in terms of negotiated payments (probably based on the position rank) that are actually not highly dependent on firm value. In addition, different position ranks can also mean different levels of perquisites, especially in a government-dominated company. In our sample, besides the government-dominated firms, there are also many family-owned firms. In such firms, the owning and managing is difficult to distinguish. Therefore, the principal-agent agency problem is little; instead, the majority-minority shareholder agency problem is serious. Hence, managerial shareholdings increase at lower levels probably will not motivate the managers to behave toward the firm value increase through solving the traditional agency problem.

According to Lin et al. (2008), with certain levels of shareholdings, the consumption of perquisites by managers may outweigh their losses from the firm value reduction. With effective control of a firm, managers are able to determine profit distribution among shareholders. Therefore, the entrenchment effect of managerial ownership comes from outright expropriation. One example would be the controlling managers’ benefit from self-dealing transactions where they transfer profits to other companies that they control. On the other hand, when the managerial ownership is beyond a certain level, the gains from firm value increase become a dominant part of the managers' compensation. Then managers will act more rationally to focus on profit maximization and firm value increase. Thus, the positive relation (the alignment effect) between the managerial ownership and firm performance dominates when the managerial ownership is high.

Equation 3: The Largest Owner Type

The full sample statistics show negative influences of all four kinds of owner identities

(family, government, cooperation and institution) when both accounting profit rate and

Tobin's Q as the proxies of firm performance. All factors show significant impacts except the

institutional owner identity when accounting profit rate is used as the proxy of firm

performance.

Dependent Variable Accounting profit rate Tobin’s Q

Independent Variables Coefficient Probability Coefficient Probability

family -0.805876 0.0503 -8.110548 0.0000

government -0.816028 0.0467 -8.024248 0.0000

corpoeration -0.882337 0.0333 -8.252347 0.0000

institution -0.615022 0.1420 -6.677295 0.0004

R

Table 9: Statistics Results of the Largest Owner Type for the Full Sample

Referring to family ownership, our hypothesis is not rejected. This result fits the majority-minority shareholder agency theory that expropriation on minority shareholders from large shareholders (usually large family and individual shareholders) may reduce firm performance. According to Maury (2006), high shareholder protection reduces the expropriation opportunities. Hence, the negative relation probably becomes more apparent in countries with relatively poor shareholder protection. The law system in the Chinese Mainland is drawn from the German law system, where the shareholder protection is relatively low compared to countries belonging to the common law system, such as the UK and the US (La Porta et al., 1998).

Referring to the government identity impact, our hypothesis is rejected. This result indicates

that the costs of government holding outweight its benefits. This explanation fits the social

system in the Chinese Mainland: the government has more aspects to consider rather than

only considering profit maximization, for instance, the employment rate. Due to some

political needs of vanity projects, firms also tend to be expanded without considering the

organization optimization. Then the redundancy within the government dominated firms

reduces their efficiency as well as the motivation of employees. On the other hand, the

possible advantage on the capital cost reduction has limits as well. Thus, the research shows

that the government being the largest owner actually brings firm more harms than benefits in

the Chinese Mainland.

In our sample, it seems that most of the corporation ownership cannot internalize firms’

transactions with their suppliers through the vertical integration of firms along the value chain.

Instead, the owning corporations are usually large groups that own several similar firms.

These similar sub-firms can with different product families be oriented to different markets.

However, the supplying relationship among them is not very apparent. In other words, the corporation owner probably cannot provide many benefits in terms of increasing input specificity, shortening transaction time and lowering transaction costs. Nevertheless, the coordination (such as controlling effort and capital) of the controlling group is very likely needed, which sacrifices individual firm’s flexibility. Therefore, the corporation ownership displays a negative impact on firm performance. As to the institution owners, the results do not support our hypothesis either. Though both industries can be considered as capital-intensive (Taggart, 1993 and International Labor Organization, 1998), financial institutions are probably not preferred as the largest owner. Both industries have high requirements of specialized knowledge of running the business, which common financial institutions may not assess. From the perspective of financial institutions, it is also very risky to invest huge amounts of money (as the largest owner) to these two industries due to the uncertainty of R&D results, especially when banks themselves as investors do not have sufficient professional knowledge of process controlling. Hence, the institution owner type is not very common in our sample either. The hypotheses on the cooperation owner type and the institution owner type are rejected. However, the lack of data on the corporation owner type and the institution owner type may also lead to biases of our results.

However, the results do not indicate all owner types definitely hinder the firm value increase

since the negative impacts from family holding and government holding may be actually due

to their too high concentration. Hence, the key to improve firm performance is again about

limiting ownership concentration to an optimal level. After ownership concentration is limited

to a certain level, family holding and government holding may be positively related to firm

performance. After all, the final impact depends on the trade-off between the relevant costs

and benefits.