Fund Manager’s Holding Preference: Evidence from China Firms

Gao Lei Special-term Assistant Professor at Business School of Shantou University

Yin Shuxi Assistant Professor at Management School of St. Gallen University

Abstract

Based on principal-agent theory and in the context of China’s special investment environment, this paper studies Chinese fund manager’s preference in stock holding. We find that in portfolio management, fund manager pursues maximizing investor’s interests, and he also exhibits moral hazard, which hurts investor’s interests. Fund manager’s holding preference is affected by corporate governance, corporate finance, and stock characteristics at the current time, but also in the previous period (2-3 years). As the operating environment improves, fund manager’s preference leans increasingly towards the maximization of investor’s interests.



Keywords: Principal-agent problem, fund manager, holding preference, moral hazard

1 Introduction

Funds misconduct shocked China in 2000. Afterwards, the supervision authorities wanted to investigate fund managers’ behavior, and investors became skeptical that fund managers made sensible investment decisions. Accordingly, scholars carried out research on Chinese fund managers’ investment preference and performance. In fund contracts, fund managers promise to pursue long-term investment return and to maximize return. If some misconduct did occur, it means that fund managers commit moral hazard. Stock is usually the most significant and important part of investment portfolio; decisions concerning stock holding are the most important decisions taken by fund managers. By studying the features of firms held by Chinese investment funds, we can investigate whether fund managers pursue dual goals, that is, maximizing fund investors’ interest and committing moral hazard.

There are a series of literatures on fund manager’s preference in stock holding. Badrinath et al (1989) report that institutional investors prefer the large-scale firms with good past performance, highβvalue, high liquidity, and long listing period. They dislike the listed firms with high volatility. Falkenstein (1996) uses data of firms held by American open-end funds from 1991 to 1992 to study the fund managers’ holding preference. He finds that American open-end fund managers prefer stocks with high volatility and high turnover ratio. They avoid stocks with low stock price, listed at small exchanges, and stocks of no-name firms. Based on the data of firms held by institutional investors from 1988 to 1991, Eakins et al (1998) claim that institutional investors prefer to hold stocks with high market value, high asset turnover ratio, highβvalue, high asset debt ratio, and regular dividend payments. Dahlquist et al (2001) study overseas investors that hold Swedish stocks. They find that these overseas investors are mostly mutual fund and other institutional investors, which prefer firms with large-scale investment, low dividend payout ratio, sufficient cash flow, and international recognition. Brands et al (2002) investigate the stock holding preference of Australian investment funds. They find that Australian investment funds prefer to hold firms with high volatility, high market value, low transaction costs, high predictability of earnings, high turnover ratio, and market recognition. They also find that these funds have industry preference. Aggarwal et al (2003) study American mutual funds’ investment in emerging markets after East Asian financial crisis. They report that these funds prefer stocks in countries with more stockholder rights, complete legal system, and sound accounting standard. After these country-level factors are controlled, American mutual funds prefer to invest in firms with more disclosure of accounting information. Yang Dequn et al (2004) study the holding preference of Chinese funds. They find that Chinese funds prefer to firms with high earnings per share, low liquidity, high stock price, high float market value, high turnover ratio, and short listing period.

As Chinese capital market is rather distinct from other markets, especially in terms of government regulations, legal system, and credit system, Chinese fund managers might commit serious moral hazard. The present paper empirically investigates whether Chinese fund managers maximize investor’s interests or their own interests when holding stocks. We also check as internal governance and external supervision strengthens, whether fund manager’s holding preference more closely follow the principle of maximizing investor’s return.

The rest of this paper is organized as follows. Part 2 analyzes Chinese funds’ governance mechanisms and stock holding preference. Part 3 designs the research approach. Part 4 demonstrates and analyzes the empirical results. Finally Part 5 is conclusions and implications.

2 Theoretical analysis

Fund assets’ ownership is separated from the management. Thus, the basic model of fund governance is agency system. According to principal-agent theory, the information asymmetry between principal and agent causes agent (which has information advantage) to hurt principal’s interest and commit moral hazard. The ideal solution is to design an effective and complete contract. Of course, we know that is impossible. In China, fund investor’s legal status is obscure (Honghui Chen, 2003; Xiaoxing He, 2003), fund managers lack independence (Yiguo Xia, 2002; Xiaoxing He, 2003), supervision is ineffective (Xiaoxing He, 2003), and manager’s motivation is absent (Ming Lu, Haiping Jiang, 2003). Henceforth, fund governance suffers from insider control (Yihuai Zhao, Chankui Liu, 1998; Yiguo Xia, 2002). Under these circumstances, external governance is especially important. However, good external governance environment has not come into being in China. Although law and supervision institutions function somehow, fund manager’s moral hazard is still serious due to a series of institutional shortcomings.

The most important asset of fund managers is credibility. If fund managers disregard credibility and hurt investor’s interest, investors can vote by feet for self protection. The principal-agent relation between managers and investors then terminates. Although fund investors do not observe the process of manager’s behavior, they can observe the outcome of manager’s behavior. Unsatisfactory results can cause their relation to terminate. Therefore, fund managers often pursue dual goals. On the one hand, they maximize their utility. On the other hand, they have to somehow maximize the interest of investors in order to sustain the principal-agent relation. If fund governance is not complete, fund managers can commit moral hazard as long as they can keep the relationship between managers and investors.

If fund managers follow the principle of maximizing investor’s utility, they would choose stocks that meet the following criteria: comprehensive governance structure, good financial status, stock’s characteristics are suitable. As professional investors, fund managers are based on careful evaluation of listed firms. But due to information asymmetry and incomplete contract, fund managers often maximize their own utility and hurt investor’s interest. Thus, they commit moral hazard. In this case, fund managers would chose firms that would be easy to conspire with to pursue benefits. Since fund scandal took place in 2000, regulatory institutions have strengthened the governance and supervision of fund. For instance, in 2003, Securities Investment Fund Law was formally promulgated. These measures function somehow to curb fund manager’s moral hazard. As the number of funds increases, the competition among funds becomes fierce, and fund governance would gain efficiency. As China’s securities market has been sluggish since 2001, fund managers face great external pressure and have to consider investor’s interest when they choose stocks.

Based on the analysis above, this paper proposes the following three hypotheses.

Hypothesis 1: Fund managers pursue two goals. They maximize their own utility and investor’s utility. On the one hand, fund managers choose firms with sound corporate governance, financial performance, and suitable status, thus they can demonstrate that they pursue maximizing of investor’s interest. On the other hand, they choose firms that they can easily conspire with, thus they can commit moral hazard.

Hypothesis 2: To maximize investor’s utility, while choosing stocks, fund manager’s preference is affected by the corporate governance, corporate finance, and stock characteristics not only of the current period, but also in the previous period (2-3 years).

Hypothesis 3: As fund’s governance environment improves, fund manager’s preference increasingly leans towards maximizing investor’s utility.

3 Research approach

It is debatable whether fund managers pursue dual goals, whether their stock holding preference maximizes investor’s interest, and whether they commit moral hazard. It is almost impossible to know manager’s real intention. Thus, this paper investigates Chinese fund manager’s holding preference and evaluates whether manager’s behavior maximizes investor’s utility. We compare the difference in governance, finance, and stock characteristics of firms in which fund invests and firms in which fund does not invest. This way, we can get to know what kind of stocks fund prefers to hold, and evaluate whether managers maximize investor’s interest. It is difficult to test whether managers commit moral hazard in their portfolio investment. Nevertheless, if managers obviously break their promise that they would hold stocks that yield long-term stable returns, we can regard that managers commit moral hazard. If fund chooses to invest in firms that are easy to conspire with, we can also regard that managers commit moral hazard.

If fund managers maximize investor’s utility, they make their decision based on long-term observation of the listed firms. Therefore, in order to test hypothesis 2, this paper investigates whether fund’s investment preference is affected by the corporate governance, finance and stock characteristics of this year, and previous years.

It is hard to investigate whether fund increasingly leans to maximizing investor’s interest as fund’s corporate governance improves. We can not directly evaluate the improvement of fund’s corporate governance environment. In long run, a country’s institutional environment would evolve towards the direction of increased economic efficiency (North, 1981). Thus it can be argued that as time elapses, the environment for funds would improve. Thus we can use one alternative hypothesis: As time passes, fund managers increasingly lean towards maximizing investor’s interest. A good solution is to evaluate the manager’s investment preference year by year. This paper takes cross section data of every year from 1998 to 2003 and uses Logit model to test hypothesis 3.

3.1 Variable definition

Our dependent variable is whether the firm’s stocks have been held by a fund. As only a large amount of stocks would have great impact on fund, we study only the firms whose large amounts of stocks are held by a fund. We read annual reports of listed firms and tabulate the top ten shareholders of a firm. If a firm has a fund as its top ten shareholder, we regard the fund invests in the firm. Thus, some funds might have very little stocks thus they are not regarded as investors of a firm in our study. This might influence the reliability of this study. But our goal is to analyze the fund manager’s stockholding preference. So we choose only a large amount of stocks held by fund managers. Moreover, when fund managers hold only a small amount of stocks, they might conduct short term speculation. This stockholding does not reveal the manager’s investment principle. Thus, the definition of fund’s stockholding is suitable for our research purpose. In our Logit model, the dependent variable takes value 1 if the fund is top ten shareholder of the firm, and takes value 0 if otherwise.

Here we discuss the variables that measure corporate governance, corporate finance and stock’s characteristics, and the expected signs of these variables.

Variable inde_ratio is the share of outside director in the board. It illustrates the composition of the board. The higher share of outside director, the less likely that board is controlled by insiders, the higher the level of corporate governance, and the higher market value of the firm (Chongen Bai et al, 2005), and funds are more likely to hold stocks of these firms. But high ratio of outside director would make it difficult for fund managers to conspire with holding shareholders, and would decrease the probability that a fund would hold this firm’s stocks. Thus, we can not ascertain the expected sign of this variable.

Variable duality is a dummy variable. It characterizes firm’s leadership structure. It takes value 1 if board member serves concurrently as general manager, and takes value 0 if otherwise. In most Chinese firms, general mangers are appointed by holding shareholders (Donghui Shi, 2004). Thus, whether general manager and board are independent in formality can not influence the board’s independence, and can not improve the firm’s governance effectiveness. So, it is hard to predict whether firms would care about this indicator and difficult to predict the sign of this variable.

Variable board_size is the number of board members. It captures the board’s size. In a reasonable range, increased board size can help the board to improve efficiency (Jensen et al, 1990). It could improve the probability that a fund would hold this firm’s stocks. Thus we expect the coefficient of this variable to be positive.

Variable meeting is the number of meetings that the board convenes. It captures the intensity of a board’s activity. Frequently board activities might signal that the firm experience serious management and governance problems. This would decrease the chance that a fund would hold this firm’s stocks. Thus, we expect the coefficient of this variable to be negative.

Variable single is a dummy variable. It captures a firm’s stock structure. It takes value 1 if a firm has a dominant shareholder who holds no less than 50 percent shares, or less than 50 percent but more than 40 percent and the sum of shares held by from the second largest to the fifth largest shareholders. Otherwise, the variable takes value 0. When there is a dominant shareholder, he will derive great benefits from the control right and decrease the firm’s transparency. Consequently, funds are unlikely to hold stocks of this firm. However, a dominant shareholder facilitates the conspiracy of funds and firm’s holding shareholders, and would increase the chance that funds would hold this firm’s stocks. Thus, it is hard to predict the sign of this variable’s coefficient.

Variable state is a dummy variable. It captures the legal status of the holding shareholder. It takes value 1 if the holding shareholder is the state, and value 0 if otherwise. If the holding shareholder is the state, it would embezzle the firm’s assets (Zengquan Li, et al 2004), and reduce the firm’s market value (Chongen Bai et al, 2005), this would reduce the chance that a fund would hold this firm’s stocks. But state-controlled firms could get support and protection of the state. Investing in these firms could maintain contact with the state. Thus, fund would likely to hold this firm’ stocks. Overall, it is hard to predict the sign of this variable’s coefficient.

Variable group is a dummy variable. It captures the form of the holding shareholder. It takes value 1 if the holding shareholder is a corporate group, and takes value 0 if otherwise. If holding share holder is a corporate group, it would more seriously embezzle the firm’s assets (Jian and Wong, 2003), and increase the firm’s probability of failure. Thus, funds are unlikely to hold the firm’s stocks. Yet if the holding shareholder is corporate group, funds are more likely to conspire with the firm (for instance, through capital reconstruction), and thus funds are more likely to hold this firm’s stocks. Therefore, it is hard to ascertain the direction of this variable’s impact.

Variable occupy is asset appropriation. It captures the holding shareholder’s asset appropriation. The more assets the holding shareholder appropriates, the more seriously corporate governance fails. This would decrease the chance that a fund chooses to hold this firm’s stocks. We expect the coefficient of this variable to be negative.

Variable m_share is the share of stocks held by senior management over all stocks. This variable captures the incentive that managers receive. High management holding would align manager’s interest with the firm’s interest and would increase the firm’s governance efficiency. Consequently, funds would likely to hold stocks of this firm. So we expect this variable to have positive sign.

Variable audit is a dummy variable. It reveals the firm’s information disclosure. If the firm’s annual reports are Clean Opinion, the variable takes value 1. Otherwise, the variable takes value 0. Clean Opinion audit reports demonstrate high information disclosure and would increase fund’s probability to hold stocks of this firm. But information disclosure makes conspiracy between fund and firm difficult and would decrease fund’s probability to hold this firm’s stocks. Thus, we are not sure about the expected sign of this variable.

Variable bigfive is a dummy variable. It reveals firm’s information disclosure. It takes value 1 if the firm hires big five audit firms for its annual report, and it takes value 0 if otherwise. Big five audit firms are large-sized and enjoy great reputation. They issue reliable and high-quality audit opinions. Hiring big five audit firms would increase information disclosure and increase the probability that a fund would hold this firm’s stocks. But information disclosure foils the conspiracy between fund and firm, and would thus decrease the probability that funds would hold this firm’s stocks. So we are not certain about the expected sign of this variable.

Variable hbshare is a dummy variable. It captures whether the firms face the supervision of overseas legal system. It takes value 1 if the firm has issued H share or B share, and takes value 0 if otherwise. The supervision of overseas legal system would improve the corporate transparency and governance efficiency (Chongen Bai et al, 2005). Thus, funds are more likely to hold stocks of this firm. But conspiracy between fund and firm would become more difficult. Thus funds are unlikely to hold this firm’s stocks. Overall, it is hard to predict the sign of this variable.

Variable developed is a dummy variable. It captures the market condition that a firm faces. The variable takes value 1 if it is in eastern China, and takes value 0 if otherwise. Good market condition provides good external governance environment and helps a firm to improve internal governance efficiency. Thus, it would increase a fund’s probability to hold the firm’s stocks. We expect that sign of this variable to be positive.

Variable protected is a dummy variable. It describes the degree of competition on the product market. If a firm operates in a field protected by the state, the variable takes value 1, and it takes value 0 if otherwise. If competition at product market is weak, a firm faces less external pressure, and has low management and corporate governance efficiency. Thus, funds are unlikely to hold this firm’s stocks. Yet fields protected by the state are of strategic importance. Holding these firms’ stocks would meet the demand of the state. Consequently, funds would tend to hold stocks of these firms. Overall, it is hard to predict the sign of this variable.

Variable core_roa is overall asset profit ratio. It reflects a firm’s permanent earning ability. Higher ratio would increase the probability that a fund would hold this firm’s stocks. Thus, we expect this variable to carry positive sign.

Variable leverage is asset debt ratio. It reflects a firm’s debt-paying ability. High financial leverage means high financial risk and would decrease the probability that a fund would hold this firm’s stocks. We expect this variable to have negative sign.

Variable glfy_asset is management expense asset ratio. It reflects the firm’s ability to control management cost. High ratio means that the firm has low control ability and low management efficiency. It does not facilitate the rise of the firm’s value and would decrease the probability that a fund would choose to hold this firm’s stocks. Thus, we expect this variable to carry negative sign.

Variable cash_asset is cash flow asset ratio. It reflects the firm’s cash flow. High ratio means that the firm has liquidity and increases the firm’s risk management ability. It would increase the probability that a fund would hold the firm’s stocks. Thus, we expect the sign to be positive.

Variable growth is asset growth rate. High growth rate would increase the firm’s value and increase the probability that this firm’s stocks would be held by funds. Thus, we expect the sign of this variable to be positive.

Variable beta is the market risk of a firm. High market risk requires high return. If fund can beat the market, it can harvest high return. Yet if it can not beat the market, it would suffer loss. Thus, it is hard to predict the sign of this variable as it is difficult to predict the fund’s preference.

Variable sigma is standard deviation of return. It describes the fluctuation of stock return ratio. Although modern investment portfolio theory argues that only market risk is associated with investment decision, institutional investors tend to hold stocks with low standard deviation of return to avoid loss in individual stocks (Badrinath, 1989). We expect this variable to carry negative sign as funds prefer to hold stocks with low volatility.

Variable M/B is market book ratio. Low market book ratio may signal that the firm has inefficient management. But it is also possible that the firm’s value is undervalued by the market. Thus it is not possible to predict the sign of this variable.

Variable lncmv is the natural log of current market value. It captures the firm’s market size. Big firms have more public information and curb moral hazard due to information asymmetry (Falkenstein, 1996). But large size makes it difficult for managers to manipulate the stock price and to commit moral hazard. Thus, it is hard to predict the sign of this variable.

Variable age is the listing period. Newly listed firms have exceedingly high returns in three years after being listed (Li Liu et al, 2001). We expect this variable to carry positive sign as funds prefer stocks of firms that are newly listed.

Variable turnover is turnover ratio, which describes stock liquidity. In order to reduce cost, institutional investors usually avoid overdue dispersed investment. If stocks have low liquidity, large trade by institutional investors would have great impact on stock price. Henceforth, institutional investors prefer liquid stocks (Gompers et al, 2001). We expect the variable to carry positive sign.

3.2 Sample and data source

This paper uses all listed firms from 1998 to 2003 as sample. The data include stocks held by funds, corporate governance data, financial data, and data about corporate stocks. We obtain data of stocks held by funds by tabulating a firm’s ten largest shareholders. We obtain corporate governance data, financial data and stock data from CCER database.

3.3 Model

We use whether a firm’s stocks are held by funds as dependent variable. Thus, dependent variable can take only two values. That is, it takes value 1 if stocks are held by funds, and takes value 0 if otherwise. Linear regression model does not fit. Binary choice model is suitable. Common binary choice model includes Probit model, Logit model and extreme value model. This paper chooses Logit model. We use P(i,t) as the probability that stock i is held by funds at period t. We use x (i,t-j) to denote the a vector of firm i’s characteristics at period t-j. B is coefficient vector to be estimated. Then we can construct probability distribution function as follows:

This equation above shows that the probability of stock i to be held by funds in period t depends on the firm’s characteristics x(i,t-j) in period t-j. In panel model t=0, j takes value 0-3. Thus we can get four panel models, when (t,t-j) takes value (0,0), (0,-1), (0, -2) and (0, -3). In cross section model, t can take value from 1998 to 2003, and j takes value 0. Thus, we can get six annual cross section datasets. It is generally believed that j should not take value 0, as there is a time lag between the end of accounting year and the disclosure of firm’s annual report. The lag is usually 1-5 months. When funds hold a firm’s stocks, they do not have information disclosed in the firm’s annual report of the current year. But as fund managers have rather strong analytical and judgment ability, they can judge a firm’s financial situation based on the information disclosed in mid-year and quarterly reports. Thus, we think that such design is sensible. Vector of coefficients b reveals the relative impact of firm characteristics on whether the stocks would be held by funds.

Before we run regression, we should check out the possible multicollinearity among variables. We analyze each two variables. We find that correlation indexes are smaller than 0.5. Thus, we argue that the likelihood of multicollinearity is rather small. Three types of independent variables (governance, finance, and stock characteristics) describe a firm from different perspectives. And these three types bear inherent relationship. Governance variables describe corporate governance, financial indicators describe corporate performance, and variables of stock characteristics describe features at the stock market. Corporate governance is fundamental. To a great extent, it determines a firm’s financial performance. And corporate governance and financial performance would affect stock price. This is, governance is the reason behind firm’s financial performance and both governance and financial performance are the reasons behind stock price. If we put three types of variables into one model, we increase the likelihood of multicollinearity, and it is hard to explain the regression results. Thus, this paper analyzes the regressions with three types of variables separately.

4 Empirical results

4.1 Single variable analysis

This paper splits the whole sample 1998-2003 into two categories: stocks held by funds and stocks not held by funds. Then we compare the means of these two sub-samples. Table 1 lists the results of t test of the means of the two sub-samples. In terms of governance, fund managers prefer firms that do not have separated leadership, that have one dominant shareholder, are controlled by that state, whose holding shareholders are corporate groups and appropriate little assets, whose audit reports are Clean Opinion and are issued by big five audit firms, that have not issued B share or H share, that are located in eastern China, and operate in fields protected by the state. In terms of financial indicator, fund managers prefer stocks with high return on assets (ROA), low financial leverage, low overhead expenses, high operating cash flow and high asset growth rate. In terms of market indicator, fund managers prefer stocks with low volatility, low market book ratio, high float market value, short listing period, and low liquidity.

Our preliminary analysis provides some evidence for hypothesis 1. In terms of corporate governance, on the one hand, fund managers prefer firms whose holding shareholders appropriate little corporate assets, and that have high information disclosure and good market foundation. This preference is consistent with the principle to maximize the investor’s utility. Because these firms have good corporate governance, holding these firms’ stocks would yield long term return. On the other hand, fund mangers prefer stocks of firms whose holding shareholders are the state and exist as corporate group. This is against the principle to maximize investor’s utility, because these institutional arrangement can help holding shareholders to hurt small shareholder’s interest, and facilitate major shareholder’s tunnel digging (Zenquan Li et al, 2004), and increase the risk of holding these stocks. Fund managers prefer these stocks probably out of consideration of moral hazard. Investing in state-controlled firms is probably to meet the state’s demand. The existence of holding shareholder as corporate group and one dominant shareholder facilitates the conspiracy between fund and firm (such as through capital reconstruction). Fund managers avoid B share or H share perhaps because it is hard to manipulate these stocks as foreign investors are present. Fund managers prefer stocks of firms in eastern China, perhaps because most funds are located in eastern China. Thus, funds are more familiar with firms in eastern China, and the cost of negotiation and conspiracy with firms in eastern China is low. Why do fund managers prefer stocks of firms that operate in fields protected by the state? Our explanation is that fund managers do so to accommodate the demand of national policy and strategic development. In terms of financial indicator, fund manger’s preference is consistent with the principle to maximize the investor’s utility. In terms of market indicator, the preference overall follows the principle. Through the analysis above, we see that fund mangers follow dual goals. They maximize fund manager’s interest and commit moral hazard.

Table 1 single variable Analysis

Note: *, **, and *** denote statistically significant at 10%, 5%, and 1% level respectively in two-tailed t test.

4.2 multi-variable panel data regression analysis

To test hypothesis 1 and hypothesis 2, this paper use panel data and Logit model to test whether fund manager’s stocks held in a year is influenced by the governance, finance, and stock market conditions in this year, last year, two years before and three years before. The results are contained in Table 2, 3 and 4.

Table 2 shows that in all models, governance variable single, state, group, occupy, audit, hbshare, developed, and protected are statistically significant and stable. Can we thus draw conclusion that fund manager’s preference is affected by the firm’s governance in this year, last year, two years before and three years before? We think we have to do some analysis. It can be argued that fund manager’s decisions are affected by corporate governance in this year, last year, two years before and three years before. But I can also be argued that corporate governance is affected by institutional arrangement. It hardly changes after it forms. Thus, if the variables of the current period are significant, then the variables of previous periods are inevitably significant. We think that this case is possible, and we can not eliminate the impact of institutions in our model. Although most indicators of corporate governance do not change much over time, some variables (such as audit, occupy) change over time. Thus we can argue that hypothesis 2 is confirmed. Table 2 also shows that fund managers prefer stocks of firms whose board holds few meetings, with dominant shareholder, are controlled by the state, whose holding shareholders are corporate groups and appropriate little assets, whose audit reports are Clean Opinion and are issued by big five audit firms, that have not issued B share or H share, are located in eastern China and operates in fields protected by the state. These conclusions are consistent with them of single variable analysis and confirm hypothesis 1.

Table 3 shows that expect variable glfy_asset whose coefficient is not significant in current period, all other variables are statistically significant and stable. That is, financial conditions this year, last year, two years before, and three years before would affect the probability that fund managers would hold this firm’s stocks. Thus we obtain highly credible evidence to confirm hypothesis 2, because these financial indicators changes over time and problems in the corporate governance model no longer exist here. Table 3 also shows that fund managers prefer stocks with high asset profit ratio, low leverage, low management expense ratio, high cash flow ratio and high asset growth rate. These conclusions are consistent with the conclusions in single variable analysis. They provide evidence that fund managers maximize investor’s interests.

Table 4 demonstrates that only variable lncmv and age are statistically significant and stable. But we can not draw conclusion that hypothesis 2 is confirmed, because firm size and age are fixed when they are listed, and do not change over time. Table 4 also shows that fund managers prefer stocks with high market risk ratio, low volatility, low market ratio, large size, and short listing period. But they do not have preference towards stock liquidity. Fund managers prefer stocks with high market risk. This is consistent with Badrinath et al (1989). Fund managers have strong incentive to beat market and gain excessive return, but they would suffer great loss if they fail. Fund managers prefer stocks with low volatility, which confirms our expectation. Fund managers prefer stocks with low market ratio, which means that the firm is inefficient, or the firm is undervalued by the market. Based on the empirical analysis above, we can exclude the probability that the firm is inefficient. Thus, a sensible explanation is that the firm is under valued by the market. Fund has special status in financial market. Through holding stocks undervalued, the funds give a signal to the market, so that the market knows the stock’s true value. Thus the fund can benefit after the stock gains value. Therefore, fund manager’s preference maximizes the investor’s interest. Fund managers prefer to hold stocks of large firms, perhaps because large firms give more public information and can reduce moral hazard due to information asymmetry (Falkenstein, 1996). In China, information asymmetry is a very serious problem (Fengqi Cao, 1999). But on the other hand, large firm size makes it difficult for fund managers to manipulate stock price and to commit moral hazard. From this perspective, fund managers maximize investor’s utility. Fund managers prefer stocks with short listing period, which is inconsistent with the research on American mutual fund by Falkenstein (1996). Falkenstein (1996) finds out that funds prefer stocks that have been listed for a long time. This deviation might due to the difference between American and Chinese capital markets. According to Li Liu et al (2001) and Zhongguang Bai et al (2003), newly listed stocks in China have abnormally high returns in at least three years. But research on American market by Ritter (1991) shows that newly listed stocks have abnormally high return on the day of IPO, but their accumulative return are lower than market index and than reference sample in the subsequent three years. In China fund managers use the Chinese IPO effect to gain abnormally high return. Thus, we can argue that managers maximize investor’s interests.

Table 2 Multi-variable regression results of corporate governance panel data

Table 3 Multi-variable regression results of stock characteristics panel data

Table 4 Multi-variable regression results of corporate governance panel data

Note: *, **, and *** denote statistically significant at 10%, 5%, and 1% level respectively in two-tailed t test.

4.3 Multi-variable cross section regression analysis

Cross section model helps me to analyze fund manager’s preference as the model controls the impact of external environment and institutions. By analyzing fund manager’s preference under different external environment, we can test hypothesis 3.

This paper uses cross section annual data from 1998 to 2003 and Logit regression model. The results are contained in Table 5, 6 and 7. Table 5 shows that in 2003, fund managers prefer stocks of firms with high ratio of outside directors and big board. From 2000, fund managers prefer stocks of firms with high information disclosure (whose audit reports are Clean Opinion). This reveals that fund managers increasingly prefer firms with good internal governance. Table 6 shows that from 2000, funds prefer stocks from firms with low financial leverage. From 2002, funds prefer stocks from firms with low expense ratio, high cash flow asset ratio. This shows that funds prefer firms with good financial status. Table 7 shows that fund managers prefer growth stocks in 1998, yet they change to prefer value stocks in 2001. This shows that fund managers pay more attention to stocks undervalued at the market. From 2002, fund managers prefer stocks with high liquidity, which shows that fund mangers pay attention to stock’s liquidity risk. All empirical results show that over time, fund managers increasingly prefer value stocks with good financial status and low risk. Fund manager’s behavior follows the principle of maximizing investor’s utility. Thus, hypothesis 3 is confirmed.

Table 5 Regression results of corporate governance cross section data

Table 6 Regression results of corporate finance cross section data

Table 7 Regression results of corporate stock cross section data

Note: *, **, and *** denote statistically significant at 10%, 5%, and 1% level respectively in two-tailed t test.

4.4 Discussions

Although the test result of the Logit model shows that the correlation is not strong, some causal relationship might exist between some variables. The most typical example is occupy. Asset appropriation might be the outcome of some corporate governance mechanisms. When we eliminate variable occupy from our model, the significance of other variables do not change. This shows that the collinearity caused by adding variable occupy is not serious. Therefore, this logic causal relationship does not pose serious econometric problem.

We do not include variable inde_ratio into panel model. This is because before 2001, most Chinese listed firms have not adopted the outside director system. In August 2001, the Chinese Securities Regulatory Commission made the outside director system mandatory. Institutional change causes structural chance in data. Thus, the variable inde_ratio loses econometric value in panel data.

5 Conclusion and implications

From the perspective of principal-agent theory and in the context of Chinese business environment, this paper analyzes Chinese fund manager’s investment preference. We find that the firm preferred by fund managers has his features. In terms of corporate governance, the firm’s board holds few meetings. The firm has a dominant shareholder, and it is controlled by the state. The holding shareholder is corporate group and appropriates little corporate assets. The firm’s audit reports are Clean Opinion and are issued by big five audit firms. The firm has not issued B share or H share, is located in eastern China, and operates in fields protected by the state. In terms of corporate finance, fund mangers prefer the firm with high ROA, low financial leverage, low overhead expenses, high operating cash flow, and high asset growth rate. In terms of stock characteristics, fund managers are interested in the firm with high β value, low volatility, high float market value, short listing period, high turnover ratio, low market book ratio.

This paper can enhance our understanding of fund manager’s stockholding preference. Fund managers have dual goals: maximize investor’s utility and commit moral hazard. To maximize investor’s utility, fund manager’s preference is affected by corporate governance, finance and stock characteristics not only in the current period, but also in previous periods (2-3 years ago). As fund mangers commit moral hazard, the state must strengthen the supervision of fund mangers. It should also construct legal system to protect investors, perfect credit system, increase listed firm’s information disclosure, and improve the fund’s operating environment.

Reference

Aggarwal, R., L. Klapper and P. D. Wysocki, 2003, “Portfolio Preferences of Foreign Institutional Investors”, working paper, SSRN.

Badrinath, S.G., D.G. Gerald and R.K. Jayant, 1989, “Patterns of Institutional Investment, Prudence, and the Managerial ‘safety-net’ Hypothesis”, Journal of Risk and Insurance, 56(4), pp.605-630.

Brands, S., D. R. Gallagher and A. Looi, 2002, “Active investment manager portfolios and preferences for stock characteristics”, SIRCA working paper.

Dahlquist, M. and G. Robertsson, 2001, “Direct foreign ownership, institutional investors, and firm characteristics”, Journal of Financial Economics, 59(3), pp.413-440.

Eakins, S. G., S. R. Stansell and P. E. Wertheim, 1998, “Institutional Portfolio Composition: An Examination of The Prudent Investment Hypothesis”, The Quarterly Review of Economics and Finance, 38 (1), pp.93-109.

Falkenstein, E.G., 1996, “Preferences for Stock Characteristics as Revealed by Mutual Fund Portfolio Holdings”, Journal of Finance, 51 (1), pp.111-136.

Gao, L.,2004, Behavioral Finance and Chinese Stock Market, Berlin: Logos Publisher.

Gompers, P. A. and A. Metrick, 2001, “Institutional Investor and Equity Prices”, The Quarterly Journal of Economics, 116 (1), pp. 229-259.

Jensen M. and K. Murphy, 1990, “Performance Pay and Top-management Incentives”, Journal of Political Economy, 98, pp.225-264.

Jian, M. and T. J. Wong, 2003, “Tunneling and Earnings Management through Related Party Transactions: Evidence from Chinese Corporate Groups”, HKUST Working Paper.

North, D. C., 1981, Structure and Change in Economic History, New York, Norton.

Ritter, J. R., 1991, “The Long-Run Performance of Initial Public Offerings”, Journal of Finance, 46(1), pp.3-27.

Bai, Chongen, Qiao Liu, Zhou Lu, Min Song, Junxi Zhang, 2005, Empirical investigation of governance structure of Chinese listed firms, Economic Research, volume 2, pp. 81-92.

Fengqi, Cao, 2003, Strategic thinking about developing investment fund, Finance, trade, economics, volume 2, pp. 22-27.

Honghui, Chen, 2003, Shortcomings and counter measures of governance structure of contract-type securities investment fund, Finance and economic science, supplement, pp. 59-61.

Chen, Donghua, Xinyuan Chen, and Hualin Wan, 2005, Salary regulation and manager’s consumption in state-owned enterprises, Economic Research, volume 2, pp. 92-102.

Xiaoxing, He, 2003. Optimization of governance structure of Chinese contract-type securities investment fund, Economic theory and economic management, volume 11, pp. 29-33.

Li, Liu and Wende Li, 2001. Study of IPO and long term performance of Chinese stocks, Economic science, volume 6, pp. 34-44.

Lu, Ming and Haiping Jiang, 2003. Governance structure of investment fund from the game theory perspective, Eastern China economic management, volume 4, pp. 122-124.

Li, Zengquan, Zhiwei Wang, and Zheng Sun, 2004, Embezzlement and property right arrangement: Evidence from asset appropriation by large shareholders in Chinese listed firms, Accounting research, volume 12, pp. 3-13.

Liu, Shaojia, Pei Sun, and Naiquan Liu, 2003, Ultimate property right, stock ownership structure and corporate performance, Economic research, volume 4.

Liu, Junhai, 2003, Legislation and judicial coordination of legal responsibility in securities market, Modern law, volume 1, pp. 3-13.

Shi, Donghui, and Danian Situ, 2004, Empirical investigation of corporate governance and its impact on corporate performance of Chinese listed firms, World Economy, volume 5, pp. 69-79.

Tang, Qingquan, Danglun Luo, and Li Wang, 2004, Large shareholder’s tunnel digging and balancing power: Evidence from Chinese market, Collections of Third International Academic Conference on Empirical Accounting.

Wang, Guogang, 2002, Corporate type: basic choice of organizational mode of Chinese securities investment fund, Finance, trade and economics, volume 10, pp. 18-25.

Xia, Yiguo, 2002, Insider’s control risk and its preventions in Chinese securities investment fund, Market weekly journal: fortune forum, ppp. 10-13.

Yang, Dequn, Mingchao Cai, and Donghui Shi, 2004, Empirical study of stockholding characteristics of Chinese securities investment fund, Journal of Zhongnan University of Economics and Law, volume 2, pp. 68-74.

Zhao, Huaiyi and Chuankui Liu, 1998, Insider’s control and counter measure of investment fund, Investment research, volume 12, pp. 26-29.