Vous êtes sur la page 1sur 15

Available online at www.sciencedirect.

com

Journal of Corporate Finance 14 (2008) 118 132 www.elsevier.com/locate/jcorpfin

Ownership structure, cash flow, and capital investment: Evidence from East Asian economies before the financial crisis
K.C. John Wei a,, Yi Zhang b
a

Department of Finance, Hong Kong University of Science and Technology, Clearwater Bay, Kowloon, Hong Kong b Department of Finance, Guanghua School of Management, Peking University, Beijing 100871, China Received 21 August 2007; received in revised form 31 January 2008; accepted 14 February 2008 Available online 26 February 2008

Abstract Using financial and ownership data from eight East Asian emerging markets before the Asian financial crisis, we document that while the sensitivity of a firm's capital investment to its cash flow decreases as the cash-flow rights of its largest shareholders increase, this sensitivity increases as the degree of the divergence between the control rights and cash-flow rights of the firm's largest shareholders increases. We interpret the results to be consistent with the free cash-flow hypothesis, which postulates that too much free cash flow in the hands of entrenched managers is likely to lead to overinvestment. This is particularly true for firms with the greatest divergence between the largest shareholders' control rights and their cash-flow rights and for firms with lower profitability. 2008 Elsevier B.V. All rights reserved.
JEL classification: G32; G34 Keywords: Investment-cash flow sensitivity; Capital investment; Pyramids; Voting rights; Cash-flow rights; East Asia

1. Introduction The effect of cash flow on capital investment has been extensively studied with the role that financing constraints play receiving the most attention. In a perfect market without asymmetric information or financial constraints, a firm's cash flow should not affect its capital investments. Instead, capital investments should be solely determined by the

The previous versions of the paper were titled Ownership structure, cash flow, and corporate investment: evidence from East Asian emerging markets and Corporate governance, overinvestment and the Asian financial crisis. The authors appreciate helpful comments and suggestions from Dolly King, Yrjo Koskinen, and seminar participants at Hong Kong University of Science and Technology, Peking University, the 2005 Chinese Finance Annual Meetings, where it was awarded the third prize for best papers, the 4th NTU International Conference on Economics, Finance and Accounting in Taipei, the 2006 JBF Conference in Beijing, the 2006 China International Conference in Finance, where it was named the winner of the best paper award, and the 2006 European Finance Association meetings in Zurich. The authors also wish to thank Jeffry M. Netter (the editor) and an anonymous referee for insightful comments and suggestions and Dr. Virginia Unkefer for editorial assistance. John Wei acknowledges financial support from an RGC Competitive Earmarked Research Grant of the Hong Kong Special Administration Region, China (HKUST6014/01H) and Yi Zhang acknowledges financial support from the National Science Foundation of China (70603001). Corresponding author. Tel.: +852 2358 7676; fax: +852 2358 1749. E-mail addresses: johnwei@ust.hk (K.C.J. Wei), zhangyi@gsm.pku.edu.cn, zhangyi.05@hotmail.com (Y. Zhang). 0929-1199/$ - see front matter 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jcorpfin.2008.02.002

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

119

firm's investment opportunities. However, in the real market, although firms tend to invest more following increases in their stock prices, cash flow is a better predictor than stock prices of a firm's capital investments.1 There are two competing explanations for the positive relation between cash flow and corporate investments. The first explanation is based on the agency costs of free cash flow as suggested by Jensen (1986). Jensen shows that managers have a tendency to overspend their free cash flow on unprofitable projects for their own private benefits. This free cash-flow hypothesis suggests that the positive relation between cash flow and investment is basically a symptom of overinvestment. Firms tend to overinvest, not because external capital is too expensive, but because internal capital is too inexpensive. The alternative explanation is based on asymmetric information. For example, Myers and Majluf (1984) show that the cost of external funds is more expensive than is the cost of internal funds due to asymmetric information problems. This asymmetric information hypothesis argues that the positive relation between cash flow and investment is typically a symptom of underinvestment. Firms tend to pass up some positive net present value projects because the cost of external capital is too high compared with the cost of internal capital. The empirical results from Fazzari et al. (1988), Hoshi et al. (1991), and other follow-up studies seem to support the asymmetric information hypothesis. These studies show that financially constrained firms tend to have higher investment-cash flow sensitivities. Further, Lamont (1997) and Shin and Stulz (1998) document that capital investments by segments of a diversified firm depend on the cash flow from other segments. These results suggest that because the cost of external capital is significantly more expensive than the cost of internal capital, financial constraints are an important factor when a firm makes its investment decisions. However, Kaplan and Zingales (1997) provide both theoretical arguments and empirical evidence that investment-cash flow sensitivities are not good indicators of financial constraints. Whether or not investment-cash flow sensitivities are valid measures of financial constraints is still the subject of debate.2 The existing studies on investment-cash flow sensitivities seldom consider the overinvestment argument caused by the agency problem, which creates conflicts between managers and shareholders (Jensen and Meckling, 1976).3 Recent studies suggest that since large shareholders have strong incentives to maximize the value of the stocks they own, large shareholders can help to overcome this agency problem.4 However, large shareholders are also associated with negative entrenchment effects.5 The empirical results from Morck et al. (1988) and McConnell and Servaes (1990) appear to substantiate both the enhancement and entrenchment effects of large shareholders. They find an inverse U-shaped relation between managerial equity ownership and firm valuation. That is, firm valuation initially increases as managerial ownership increases. However, after a certain point, firm value starts to decrease as managerial ownership increases, because managers become entrenched and start to pursue their private benefits at the expense of outside investors.6 Hadlock (1998) argues that if the positive investment-cash flow relation is caused by a managerial preference to overinvest internal funds, then the positive investment-cash flow sensitivity should decrease as the alignment of interests between managers and shareholders increases. On the other hand, if the positive investment-cash flow relation is caused by a managerial preference for underinvestment, the positive investment-cash flow sensitivity should increase as the alignment of interests between managers and shareholders increases. Using managers' ownership data from 1975 as his measure of the alignment of interests between managers and shareholders along with financial data from 19731976, Hadlock (1998) finds an inverse U-shaped relation between managerial ownership and investment-cash flow sensitivities for the U.S. firms in his sample. He interprets the results to be consistent with the underinvestment hypothesis caused by asymmetric information. In this paper, we extend Hadlock's (1998) work to use the information on both large shareholders' cash-flow rights and the divergence between large shareholders' control rights and their cash-flow rights to disentangle the positive
See, for example, Fazzari et al. (1988) and Morck et al. (1990). Also see Hubbard (1998) for an excellent review of this literature. See, for example, Cleary (1999), Fazzari et al. (2000) and Kaplan and Zingales (2000). 3 There are studies that empirically test agency problems and the overinvestment hypothesis from different angles. For example, Harvey et al. (2004) find that debt mitigates agency problems associated with overinvestment. Dittmar et al. (2003), Pinkowitz et al. (2006), and Kalcheva and Lins (in press) find evidence of overinvestment by entrenched managers related to cash holdings. 4 For example, Shleifer and Vishny (1997, p. 754) point out, Large shareholders thus address the agency problem in that they have both a general interest in profit maximization, and enough control over the assets of the firm to have their interest respected. 5 Again, Shleifer and Vishny (1997, p. 758) argue, The large shareholders represent their own interests, which need not coincide with the interests of other investors in the firm, or with the interests of employees and managers. 6 Stulz (1988) has formally modeled this negative effect of large shareholders on firm value, which predicts an inverse U-shaped relation between managerial ownership and firm value.
2 1

120

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

enhancement effect from the negative entrenchment effect of large shareholders. Using the information on both control rights and cash-flow rights more effectively distinguishes between the two competing explanations for why cash flow affects capital investment than does using the information on managerial cash-flow rights alone. Further, the existing literature suggests that while the positive enhancement effect is related to the cash-flow rights of large shareholders, the negative entrenchment effect is more associated with the control rights of large shareholders. However, as pointed out by Claessens et al. (2002), it is difficult to separate the incentive effect from the entrenchment effect of large shareholders using U.S. firms, since stocks in U.S. firms are widely held and there is little divergence between control rights and cash-flow rights. In contrast, in East Asian emerging economies, many firms are owned and controlled by single large shareholders via pyramid ownership structures. Many of the firms in these pyramid structures exhibit high ownership concentration and high levels of divergence between the control rights and cash-flow rights of large shareholders.7 Thus, East Asian corporations provide an excellent sample with which to study the effect of the separation between the cash-flow rights and control rights of large shareholders on corporate investment. The implication of Hadlock's (1998) results suggests that, if firms typically overinvest their internal capital, the investment-cash flow sensitivity should decrease as the cash-flow rights of the large shareholders increase and this sensitivity should increase as the divergence between the control rights and cash-flow rights of the large shareholders increases. On the other hand, if asymmetric information typically raises the cost of external funds, we expect to observe the opposite results. If the capital market is perfect and the availability of internal funds does not affect investment choices, then managerial ownership structures should not have any effect on the relation between cash flow and corporate investment. Using financial and ownership data from eight East Asian emerging markets during 19931996, our empirical results support the overinvestment hypothesis caused by the agency costs of free cash flow. More specifically, the investment-cash flow sensitivity is negatively related to the cash-flow rights of the largest shareholders, while this sensitivity is positively associated with the divergence between the control rights and cash-flow rights of the largest shareholders, especially among firms with lower returns on assets (ROA). Our results not only support the overinvestment hypothesis, but also provide evidence for the positive enhancement effect related to the cash-flow rights and the negative entrenchment effect associated with the control rights of the largest shareholders on corporate investment. Our results complement those of Claessens et al. (2002), Lins (2003), and Joh (2003). These studies show that firm valuation is negatively associated with the separation of cash-flow ownership from control. Moreover, our evidence of overinvestment by East Asian corporations immediately before the Asian financial crisis may contribute to our understanding of this devastating financial crisis that began with the devaluation of the Thai baht on July 2, 1997. Weak corporate governance, especially the divergence between the control rights and cash-flow rights of large shareholders, has been cited as one of the causes of the crisis. Several studies have examined the impact of corporate governance on firm performance during the East Asian financial crisis of 19971998. Johnson et al. (2000) find that measures of corporate governance, particularly the effectiveness of protections for minority shareholders, explain the extent of the exchange rate depreciation and the stock market decline during the crisis better than do standard macroeconomic measures. Mitton (2002) finds that corporate governance and disclosure are positively related to firm performance during the time of the crisis (19971998). Lemmon and Lins (2003) also offer similar results. The rest of this paper is organized as follows. Section 2 discusses our hypothesis development. Section 3 describes our data and empirical specifications. Section 4 reports our baseline results, while Section 5 presents the robustness checks. Section 6 concludes the paper. 2. Hypothesis development 2.1. The overinvestment hypothesis resulting from the agency costs of free cash flow The overinvestment hypothesis proposed by Jensen (1986) argues that managers have a tendency to overinvest internally generated funds. Hadlock (1998) further argues that if the positive investment-cash flow relation is caused by a

7 See La Porta et al. (1999) and Claessens et al. (2000) for detailed discussions of ownership structures around the world and Asian firms, respectively. Also see Claessens et al. (2002) for evidence on the incentive and entrenchment effects of the largest shareholders on firm value in East Asian corporations.

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

121

managerial preference to overinvest internal funds, there should be a negative relation between the investment-cash flow sensitivity and the alignment of interests between managers and shareholders. In East Asia, ownership is concentrated as opposed to being diffused as in the U.S. In addition, in East Asian economies, many firms are owned and controlled by single large shareholders via pyramid ownership structures. In these economies, the nature of agency problems shifts away from the conflicts of interest between managers and shareholders to the conflicts of interest between controlling owners or large shareholders (who happen to be managers in most cases) and minority shareholders. In this case, there are two effects on the overinvestment problem: the enhancement effect related to cash-flow rights and the entrenchment effect associated with the control rights of large shareholders (Claessens et al., 2002). The overinvestment problem is alleviated when large shareholders have a high level of cash-flow rights, since their interests are more aligned with those of minority shareholders. Conversely, the overinvestment problem is aggravated when there is a high level of divergence between the control rights and cash-flow rights of large shareholders, since their interests are less aligned with those of minority shareholders. The above discussion leads to the following testable hypothesis: Hypothesis 1. If large shareholders have a preference for overinvestment resulting from the agency costs of free cash flow, the investment-cash flow sensitivity decreases as the level of large shareholders' cash-flow rights increases, while this sensitivity increases as the degree of divergence between large shareholders' control rights and their cash-flow rights increases. 2.2. The underinvestment hypothesis resulting from asymmetric information problems in the capital markets There is an alternative view based on asymmetric information problems. Myers and Majluf (1984) show how a firm may underinvest in the presence of asymmetric information problems in the capital markets. Hadlock (1998) further suggests that if the positive investment-cash flow sensitivity is caused by a managerial preference for underinvestment, there should be a positive relation between the investment-cash flow sensitivity and the alignment of interests between managers and shareholders.8 We can extend Hadlock's argument to infer that the underinvestment problem is aggravated when large shareholders have a high level of cash-flow rights or when there is little divergence between the control rights and cash-flow rights of large shareholders. The above discussion leads to the following testable hypothesis. Hypothesis 2. If large shareholders have a preference for underinvestment resulting from asymmetric information problems in the capital markets, the investment-cash flow sensitivity increases as large shareholders' cash-flow rights increase and decreases as the divergence between large shareholders' control rights and cash-flow rights increases. 3. Data description and empirical specification 3.1. Sample selection We start with all firms in East Asia from 19911996 that appeared on the Worldscope tape. Regarding ownership data, we focus on ultimate ownership. We use ownership data assembled by Claessens et al. (2000).9 Claessens et al. (2000) collected 1996 data on the ownership of corporations in Hong Kong, Indonesia, Japan, Korea, Malaysia, the Philippines, Singapore, Taiwan, and Thailand. Their main source was Worldscope, supplemented by other sources that describe ownership structures. From a complete sample of 5284 publicly listed corporations in the nine East Asian countries, ownership data were collected for 2980 firms. The procedure of identifying ultimate owners is similar to the one used by La Porta et al. (1999). An ultimate owner is defined as the shareholder who has the determining voting rights of the company and is not controlled by anybody else. If a company does not have an ultimate owner, it is classified as widely held. To economize on the data collection task, the ultimate owners' voting right level is set at 50% and not traced any further once the level exceeds 50%. Although a firm can have more than one ultimate owner, we focus on the largest ultimate owner. As our definition of ownership relies on both cash-flow rights and voting control rights, the cash-flow

Dybvig and Zender (1991) show that the underinvestment problem is alleviated when large shareholders have few cash-flow rights. The dataset is also used by Claessens et al. (2002) and Fan and Wong (2002). We thank Joseph Fan and Larry Lang for their generousness in providing us with their ownership data.
9

122

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

rights that support the control by ultimate owners are further identified. Firm-specific information on pyramid ownership structures and crossholdings is used to draw the distinction between cash-flow rights and control rights. In our analysis, we use a subset of these firms. First, we exclude from the sample all Japanese firms. Most Japanese firms have dispersed ownership structures, and ownership and management are separated far more often in Japan than in other East Asian economies. The most important shareholders in most Japanese firms are widely held financial institutions, again unlike in many other economies in the region. These financial institutions and their affiliated firms often work together to influence the governance of their corporations. Thus, including Japan in our sample would be less useful for disentangling the incentive and entrenchment effects of concentrated ownership and control. Second, we follow the literature to exclude firms that operate in financial industries (SIC 60006999) and regulated utilities (SIC 49004999), since their investment behavior should be different from the investment behaviors of other industries. We then merge the ownership data with the Worldscope financial data. We use the Worldscope database instead of the PACAP database, because PACAP does not provide firms' capital expenditures. Though capital expenditures could be inferred from other accounting data, different accounting standards in each country would lead to incomparable measures of capital expenditures. There are many firms with missing capital expenditures in the Worldscope database. We dropped these firm-years from our sample. We also dropped firmyears with missing values on variables such as assets, lagged assets, lagged Tobin's Q, and lagged sales. The merger of the 1996 ownership data with the 19931996 financial data requires us to assume that the ownership and control structures of the firms did not change substantially during that period. This is a reasonable assumption, since the economic and political conditions were relatively stable during our sample period. These sample selection criteria result in a final sample of 2555 firm-years from 994 corporations in eight East Asian economies over the period of 19931996.10 To include as many observations as possible, we use unbalanced panel data in our analyses, although we also analyze a smaller balanced panel dataset in the robustness checks. Broken down by economies, the sample covers 458 firm-years for 193 firms from Hong Kong, 288 firm-years for 101 firms from Indonesia, 477 firm-years for 202 firms from Korea, 390 firm-years for 127 firms from Malaysia, 133 firm-years for 60 firms from the Philippines, 397 firm-years for 143 firms from Singapore, 194 firm-years for 91 firms from Taiwan, and 218 firm-years for 77 firms from Thailand. 3.2. Summary statistics of ownership structures Panel A of Table 1 documents the largest shareholders' control rights, cash-flow rights, and the difference between control rights and cash-flow rights in eight East Asian economies. The definition of cash-flow rights and control rights follows Claessens et al. (2000) and Claessens et al. (2002). The mean of the largest shareholders' control rights is 29.31% and the median is 28.00%. The mean of the largest shareholders' cash-flow rights is 24.44% and the median is 24.00%. The mean of the difference between control rights and cash-flow rights is 4.88%. This is not a large difference because over half the firms in the sample do not exhibit a separation between control rights and cash-flow rights. A quarter of the firms have a difference between control rights and cash-flow rights that is larger than 9%. The maximum of the difference between control rights and cash-flow rights is 38%. This shows that the difference between control rights and cash-flow rights of the largest shareholders is very significant in over a quarter of East Asian corporations. 3.3. Summary statistics of firm characteristics Panel B of Table 1 presents summary statistics of firm characteristics for the entire sample and for sub-samples based on the largest shareholders' cash-flow rights quartiles. We note that investment, cash flow, and asset size vary significantly across quartiles based on the largest shareholders' cash-flow rights. However, interpreting the variations in the summary statistics on the reported variables across cash-flow quartiles as evidence for or against the presence of underinvestment or overinvestment may be problematic or premature. We do not make any inferences directly based on these summary statistics.

The two extreme percentiles of firm-year observations of cash flow/assets, investment/assets, and lagged sales/assets are eliminated from the sample. Our results remain unchanged without this elimination procedure.

10

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132 Table 1 Summary statistics Panel A: control rights and cash-flow rights of the largest ultimate owners # of firms Mean SE 12.33 Minimum 0.00 1st quartile 22.00 Median 28.00 3rd quartile 36.00

123

Maximum 63.00

Control rights (%) 974 29.31 Cash-flow rights (%) 974 24.44

12.40

0.00

15.00

24.00

32.00

59.00

The difference between control rights and cash-flow rights (%) 974 4.88 7.61 0.00 Panel B: summary statistics of relevant variables Variable Investment/Assets (%) Tobin's Q Cash flow/Assets (%) Sales/Assets (%) Market cap (in million $) Total assets (in million $) Debt ratio (%) ROA (%) Current ratio ($) Number of observations All firms 6.52 1.30 8.04 67.10 201.20 284.10 28.10 5.48 1.33 2555 Quartile 1 6.63 1.22 6.26 67.60 243.50 449.00 31.70 4.95 1.31 643

0.00

0.00

9.00

38.00

Quartile 2 7.11 1.31 8.05 70.30 222.53 298.60 27.20 5.70 1.34 621

Quartile 3 5.45 1.32 8.28 59.50 198.50 249.90 25.60 5.20 1.34 699

Quartile 4 7.14 1.37 9.74 68.90 137.80 185.80 27.60 6.51 1.33 592

Panel A of this table reports ownership structures (in percent) of the largest ultimate owners in East Asian firms. To be included in the sample, a firm must have at least one year of capital expenditures and other financial data in the Worldscope database between 1993 and 1996 and have lagged financial data as well. The ultimate ownership data is obtained from Claessens et al. (2000). Cash-flow rights and control rights are based on common stock ownership. There are a total of 974 firms for the 1996 in the sample. Panel B reports summary statistics of relevant variables. All reported figures are the medians and are calculated over the complete set of firm-years from 1993 to 1996. Investment/Assets is capital expenditures over assets. Tobin's Q is the market value of equity plus assets minus the book value of equity over assets. Cash flow/Assets is operating cash flow over assets. Sales/Assets is sales over assets in the previous year. Market cap is the market value of equity. Total assets is the total assets of the firm. Debt ratio is total debt over asset. ROA is operating income over assets. Current ratio is current assets over current liabilities. Firms with the cash-flow rights of the largest ultimate owners at or below the 25th percentile is classified as Quartile 1 firms, above the 25th percentile and below the 50th percentile as Quartile 2 firms, at or above the 50th percentile and at or below 75th percentile as Quartile 3 firms, and above the 75th percentile as Quartile 4 firms. and denote the median difference of the corresponding variable between the current ownership quartile and the next larger ownership quartile at the 10% and 5% significant levels, respectively, using a Wilcoxon Rank-Sum test. In the Quartile 4 column, this denotes the test of the difference between Quartile 4 and Quartile 1.

3.4. Empirical specification To investigate the role of ownership structures on investment-cash flow sensitivities, we use the same basic approach as Fazzari et al. (1988) and others. These authors regress investment on Tobin's Q, cash flow, and other control variables. They interpret differences in the investment-cash flow relationship between different groups of firms as evidence of financial constraints. Hadlock (1998) interacts cash flow with managerial cash-flow rights to examine the managerial enhancement effect on investment. To examine the managerial entrenchment effect on investment, we interact cash flow with the difference between the largest shareholders' control rights and their cash-flow rights (denoted as Divergence or Control minus ownership). Based on the discussion above, we use the following regression model to perform our tests: Iit Ai;t1 b0 b1 Qi;t1 b2 CFit CFit CFit b3 Cashrighti;t1 b4 Divergencei;t1 BYit eit ; Ai;t1 Ai;t1 Ai;t1 1

where I is investment, A is assets, Q is a measure of Tobin's Q, CF is cash flow, Cashright is the largest shareholder's cash-flow rights, Divergence is the largest shareholder's control rights minus his/her cash-flow rights, and Y is a set of

124

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

Table 2 Investment regression results from sub-samples grouped by the level of cash-flow rights and by the level of separation between cash-flow rights and control rights Cash-flow rights N Tobin's Q Coefficient (SE) [t-stat] Cash flow/assets Coefficient 0.408 0.326 0.272 0.370 (SE) (0.064) (0.037) (0.048) (0.063) [t-stat] [1.11] [1.70] [0.43] Adjusted R2

Panel A: portfolios sorted by the largest shareholders' cash-flow rights Quartile 1 643 0.0065 (0.0045) Quartile 2 621 0.0049 (0.0045) [1.96] Quartile 3 699 0.0053 (0.0040) [0.21] Quartile 4 592 0.0011 (0.0048) [1.21] Panel B: portfolios sorted by the largest shareholders' control minus ownership =0 1559 0.0053 (0.0042) N0 996 0.0116 (0.0082) [0.73]

0.035 0.145 0.052 0.053

0.247 0.516

(0.032) (0.064)

[3.92]

0.078 0.053

This table reports the regression results of capital investment/assets on Tobin's Q, cash flow/assets, sales/assets, and time fixed effects. The regression coefficients on sales/assets and the year dummies are not reported. In panel A, all firms identified in 1995 are sorted into quartiles based on the largest shareholders' cash-flow rights. Firms with the cash-flow rights of the largest ultimate owners at or below the 25th percentile are classified as Quartile 1 firms, above the 25th percentile and below the 50th percentile as Quartile 2 firms, at or above the 50th percentile and at or below 75th percentile as Quartile 3 firms, and above the 75th percentile as Quartile 4 firms. In panel B, all firms are separately sorted into two groups based on the difference between the largest shareholders' control rights and cash-flow rights as control minus ownership. A firm is designated as in the control minus ownership = 0 group if its control minus cash-flow rights is zero. A firm is designated as in the control minus ownership N 0 group if its control minus cash-flow rights is greater than zero. The regressions are performed separately for each group of firms using the 19931996 panel data. Year and four-digit code industry dummy variables are included in all regressions. Panel A reports the result based on cash-flow rights sorting, while panel B based on control minus ownership sorting. SE is standard error. t-statistics (t-stat) in panel A test the hypothesis of no difference between the coefficients in each quartile and in Quartile 1. t-statistics in panel B test the hypothesis of no difference between the coefficients in the control minus ownership = 0 group and in the control minus ownership N 0 group.

control variables.11 If firms face asymmetric information problems (i.e., the underinvestment hypothesis), we expect 3 to be positive and, if the entrenchment effect is important, we expect 4 to be negative. If firms face agency problems from free cash flow (i.e., the overinvestment hypothesis), we expect 3 to be negative and, if the entrenchment effect is important, we expect 4 to be positive. If the capital market is perfect and there is no agency problem, we expect both 3 and 4 to be zero. 4. Regression results 4.1. Regression results based on portfolios sorted by the largest shareholders' ownership We begin with simple tests of Hypotheses 1 and 2 based on portfolio analysis. To conduct the tests, we first assign all firms in our sample to quartiles according to the largest shareholders' cash-flow rights (Cashright). We also separately assign all sample firms to two groups according to their Control minus ownership (Divergence) values. Firms with Divergence values equal to zero are assigned to the Control minus ownership = 0 group and those with Divergence values greater than zero are assigned to the Control minus ownership N 0 group. We then estimate the following simple investment equation separately for each group: Iit CFit b0 b1 Qi;t1 b2 BYit eit : Ai;t1 Ai;t1 2

The results from the 19931996 panel data are reported in Table 2. Panel A reports the results based on sorting of the largest shareholders' cash-flow rights. Under this sorting, capital investment is sensitive to cash flow in all quartiles. In addition, there is an association between cash-flow rights and investment-cash flow sensitivities. The estimated coefficient on cash flow falls from 0.408 in Quartile 1 to 0.326 in Quartile 2, and further down to 0.272 in Quartile 3. Thus, firms in which the largest shareholders have extensive cash-flow rights have smaller investment-cash flow

11

Baker et al. (2003) and Shin and Stulz (1998) also use assets in the denominator in their studies.

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

125

sensitivities than do firms in which the largest shareholders have limited cash-flow rights. A formal test shows that the difference in the estimated coefficient on cash flow is statistically significant between Quartile 1 and Quartile 3 at the 10% level with a t-statistic of 1.70. The results imply that the overinvestment problem is alleviated in firms in which the largest shareholders have extensive cash-flow rights, which is consistent with the overinvestment hypothesis suggested by the agency problems of free cash flow. However, the estimated coefficient on cash flow in Quartile 4 increases to 0.370 from 0.272 in Quartile 3. This may suggest that the entrenchment effect appears to become more significant as the largest shareholders' control rights increase along with their cash-flow rights. However, we need to examine the cash-flow rights and control rights simultaneously to disentangle the two effects. In summary, the results in panel A of Table 2 show that there is a U-shaped relationship between investment-cash flow sensitivities and the cash-flow rights of the largest shareholders, which is opposite to the finding by Hadlock (1998) in U.S. firms. This perhaps suggests that the overinvestment hypothesis is more appropriate in describing investment activities in East Asian countries than the underinvestment hypothesis is. Panel B of Table 2 reports the results based on the control minus ownership sorting. The results indicate that there is indeed a strong relationship between Control minus ownership and the effect of cash flow on investment. The estimated coefficient on cash flow increases from 0.247 in the Control minus ownership = 0 group to 0.516 in the Control minus ownership N 0 group. Thus, firms that have a separation between the largest shareholders' control rights and their cash-flow rights have more than double the sensitivity of investment to cash flow than do firms that have no such separation. A formal test shows that the difference in the estimated coefficients on cash flow is statistically significant between the two groups at the 1% level with a t-statistic of 3.92. These results imply that the overinvestment problem is aggravated in firms with a separation between the largest shareholders' control rights and their ownership, which is also consistent with the overinvestment hypothesis. 4.2. Baseline regression results The results from the portfolio analyses in the last sub-section support the overinvestment hypothesis. In this section, we use the entire sample for a formal test of our hypotheses. Since the ownership data are collected from 1996, we use the 19931996 panel data. The regression results from Eq. (1) are reported in Table 3. Model 1 of Table 3 reports the estimates from a standard investment-cash flow regression. The estimated coefficient on lagged sales/assets is positive

Table 3 Regression of investment on Tobin's Q, cash flow, and the largest shareholders' ownership structures Independent variable Intercept Sales Tobin's Q Cash flow Cash flow Cash-flow rights Cash flow (Control minus ownership) Country and industry dummies Adjusted R2 Number of observations Yes 0.071 2555 Yes 0.064 2555 Model 1 0.066 (7.39) 0.015 (0.32) 0.007 (1.89) 0.316 (10.57) Model 2 0.069 (7.63) 0.016 (0.36) 0.007 (1.92) 0.540 (5.73) 0.955 ( 2.50) Model 3 0.065 (8.11) 0.017 (0.37) 0.006 (1.67) 0.426 (4.05) 1.340 (3.07) Yes 0.064 2555 Model 4 0.067 (7.43) 0.017 (0.38) 0.007 (1.72) 0.426 (4.05) 0.633 ( 1.57) 1.097 (2.37) Yes 0.061 2555

This table reports fixed-effect regression results of capital investment on Tobin's Q, cash flow, interactions of cash flow with the largest shareholders' cash-flow rights and with the difference between the largest shareholders' control rights and cash-flow rights (denoted as control minus ownership), controls, the country dummies, and the four-digit industry code dummies in 19931996. Investment is defined as capital expenditures over assets. Sales is the total sales over assets in the previous year. Tobin's Q is defined as the market value of equity plus assets minus the book value of equity over assets. Cash flow is operating cash flow over assets. Cash-flow right is the largest shareholders' cash-flow rights. Control minus ownership is the difference between the largest shareholders' control rights and cash-flow rights. Firm fixed effects and year dummy variables are included in all regressions. t-statistics are in parentheses. and denote significance at the 10% and 5% levels, respectively.

126

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

but insignificant. Both coefficients on lagged Tobin's Q and on cash flow have the expected positive signs and are significant, which is also consistent with previous studies that use U.S. data (e.g., Fazzari et al., 1988). Model 2 includes a term that interacts cash flow with the largest shareholders' cash-flow rights (Cash flow Cashflow rights). The coefficient of interest in this case is the coefficient on the interaction term (3 in Eq. (1)). The estimated coefficient on Cash flow Cash-flow rights is 0.955, which is negative and significant at the 5% level with a t-statistic of 2.50. This result shows that the sensitivity of investment to cash flow falls rapidly as the largest shareholders' cash-flow rights increase. This result strongly supports the overinvestment hypothesis that is heightened by the largest shareholders' enhancement effect on investment. Model 3 includes a term that interacts cash flow with Control minus ownership (Cash flow (Control minus ownership)). The coefficient of interest in this case is the coefficient on the interaction term (4 in Eq. (1)). The coefficient on Cash flow (Control minus ownership) is 1.34, which is positive and significant at the 1% level with a t-statistic of 3.07. This implies that the sensitivity of investment to cash flow increases rapidly as the degree of separation between the largest shareholders' ownership and their control rights increases. This result is also consistent with the overinvestment hypothesis that is heightened by the largest shareholders' entrenchment effect on investment. Model 4 reports results from our full model by interacting cash flow with both cash-flow rights and Control minus ownership. We use this specification as our baseline specification for any comparisons below. The coefficients of interest in this case are the coefficients on the interaction terms (3 and 4 in Eq. (1)). The coefficient on Cash flow Cash-flow rights is 0.633, which is negative and close to significant at the 10% level. The coefficient on Cash flow (Control minus ownership) is 1.097, which is positive and significant at the 5% level with a t-statistic of 2.37. The results demonstrate that the sensitivity of investment to cash flow simultaneously decreases as the largest shareholders' cash-flow rights increase and it increases as the degree of separation between the largest shareholders' ownership and their control rights increases. The results are consistent with the overinvestment hypothesis, which is heightened by the largest shareholders' enhancement and entrenchment effects on investment. This suggests that using East Asian firms allows us to detect both the enhancement and entrenchment effects of the largest shareholders on investment. These effects are difficult to identify in typical U.S. firms. Overall, the results in Table 3 support the overinvestment hypothesis caused by the agency costs of free cash flow using our East Asian sample. They do not support the underinvestment hypothesis caused by asymmetric information problems as documented by Hadlock (1998) on a U.S. sample. Our results and approach offer a much clearer picture of the impact of ownership structures on the investment-cash flow sensitivity than previous studies have been able to show. In particular, we disentangle the enhancement effect from the entrenchment effect on investment. When large shareholders care more about shareholder value, the agency costs of free cash flow are less severe and the overinvestment problem also becomes less severe. On the other hand, when the degree of separation between ownership and control rights is great, the overinvestment problem is aggravated. 5. Robustness checks and additional tests 5.1. Alternative requirements for panel data To work with a large sample, we use an unbalanced panel dataset in the above analyses. More specifically, we include all firms in the sample that have at least one firm-year in the span of 19931996, provided that all the data required for analysis are available. This sampling procedure might bias our results. In this section, we examine whether our results are robust to the requirement of specific numbers of firm-years for a firm to be included in the sample. First, we exclude firms with only one firm-year observation in 19931996. Column 1 of Table 4 reports the estimated coefficients based on this reduced sample of 2292 firm-years. The results remain almost unchanged. We then further exclude firms with only one or two firm-year observations. Column 2 of Table 4 reports estimated coefficients based on this further reduced sample of 1786 firm-years. The results change slightly. The coefficient on Cash flow Cash-flow rights becomes significant at the 10% level and the significance of the coefficient on Cash flow (Control minus ownership) reduces to the 10% level. Finally, we include only firms with all four firm-year observations. Column 3 of Table 4 reports the estimated coefficients based on this balanced panel data sample of only 1408 firm-years. The coefficients on all variables do change, but the signs on all variables remain the same. The coefficient on Cash flow Cash-flow rights improves to become significant at the 5% level. In sum, our results remain virtually unchanged or are even stronger when we impose more stringent restrictions on firm-year observations.

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132 Table 4 Regression of investment on Tobin's Q, cash flow, and the largest shareholders' ownership structures: alternative sample size Independent variable At least 2 firm-years (Column 1) Intercept Sales Tobin's Q Cash flow Cash flow Cash-flow rights Cash flow (Control minus ownership) Country and industry dummies Adjusted R2 Number of observations 0.065 (7.33) 0.002 (0.38) 0.007 (1.72) 0.426 (4.05) 0.633 ( 1.57) 1.097 (2.37) Yes 0.061 2292 At least 3 firm-years (Column 2) 0.064 (6.76) 0.000 ( 0.02) 0.008 (1.84) 0.473 (3.99) 0.864 ( 1.87) 1.170 (2.33) Yes 0.059 1786

127

4 firm-years only (Column 3) 0.060 (5.70) 0.001 ( 0.26) 0.011 (2.23) 0.576 (4.14) 1.29 ( 2.37) 1.240 (2.18) Yes 0.056 1408

This table reports fixed-effect regression results of capital investment on Tobin's Q, cash flow, interactions of cash flow with cash-flow rights and with control minus ownership controls, the country dummies, and the four-digit industry code dummies for the 19931996 panel data. Investment is defined as capital expenditures over assets. Sales is the total sales over assets in the previous year. Tobin's Q is defined as the market value of equity plus assets minus the book value of equity over assets. Cash flow is operating cash flow over assets. Cash-flow right is the largest shareholders' cash-flow rights. Control minus ownership is the difference between the largest shareholders' control rights and cash-flow rights. Column 1 reports the unbalanced panel data regression results when the sample includes 731 firms for a total of 2292 firm-years with each firm having at least two firm-years. Column 2 reports the unbalanced panel data regression results when the sample includes 478 firms for a total of 1786 firm-years with each firm having at least three firm-years. Column 3 reports the balanced panel data regression results with 352 firms for a total of 1408 firm-years with each firm having all four firm-years. Firm fixed effects and year dummy variables are included in all regressions. t-statistics are in parentheses. and denote significance at the 10% and 5% levels, respectively.

Since the East Asian financial crisis took place in mid 1997, the year 1996, which is just prior to the crisis, may be different from other years, and thus may influence our results. To test this possibility, we run the baseline regression with the panel data from 19921995. The results (unreported) are similar to those on data from 19931996. Though not reported, we also run a country-by-country analysis. The results are consistent with the whole sample analysis in general (with correct signs for a majority of countries), but are weaker due to fewer observations. 5.2. Robustness checks: different sets of control variables Our results in Table 3 may be sensitive to the exclusion or inclusion of certain control variables. In this section, we report robustness checks to validate our results. First, sales are highly correlated with Tobin's Q, which may explain why the coefficient on Tobin's Q is not large in Model 1 of Table 3. Model 1 of Table 5 reports the estimated coefficients excluding sales from our baseline result reported in Model 4 of Table 3. After excluding sales, the coefficient on Tobin's Q is positive and significant, but it does not change much. The other coefficients also do not change much. Our results continue to support the overinvestment hypothesis caused by the agency costs of free cash flow when excluding sales from the regression. We also replace sales/assets with sales growth in the regression. The result (unreported) is essentially the same. Second, it is possible that the investment-Tobin's Q relationship is not monotonic. Models 2 and 3 in Table 5 report the results with the squared term of Tobin's Q included. Model 2 excludes sales, while Model 3 includes sales. The results from both Models 2 and 3 indicate that the inclusion of the squared Tobin's Q does not change the coefficient on Cash flow Cash-flow rights or the coefficient on Cash flow (Control minus ownership) too much except that the former becomes significant at the 10% level, compared with our baseline result in Model 4 of Table 3. Third, a potential explanation for our results in Table 3 is that the largest shareholders' ownership may be a proxy for omitted variables that affect the investment-cash flow sensitivity. It is possible that the results reflect a negative correlation between firm size and the largest shareholders' ownership. To control for this possibility, we include a term

128

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

Table 5 Regression of investment on Tobin's Q, cash flow, and the largest shareholders' ownership structures: robustness checks Independent variable Intercept Sales Tobin's Q Tobin's Q squared Cash flow Cash flow Cash-flow rights Cash flow (Control minus ownership) Cash flow Ln(Size) Cash flow Tobin's Q Year and industry dummies Adjusted R2 Number of observations Yes 0.062 2555 Yes 0.053 2555 Yes 0.052 2555 Yes 0.056 2555 0.427 (4.06) 0.632 ( 1.56) 1.095 (2.37) 0.007 (1.74) 0.008 (2.07) 0.0012 (2.36) 0.422 (4.02) 0.700 ( 1.73) 1.160 (2.51) Model 1 0.068 (8.18) Model 2 0.063 (7.52) Model 3 0.062 (6.86) 0.002 (0.35) 0.008 (2.06) 0.0012 (2.63) 0.422 (4.02) 0.701 ( 1.73) 1.164 (2.54) Model 4 0.068 (7.47) 0.017 (0.38) 0.012 (1.72) 0.444 (4.13) 0.677 ( 1.66) 1.078 (2.32) 0.036 ( 0.84) Model 5 0.058 (5.28) 0.002 (0.34) 0.012 (2.32) 0.506 (4.42) 0.678 ( 1.66) 1.040 (2.24) 0.042 ( 0.99) 0.024 ( 1.56) Yes 0.055 2555

This table reports robustness checks for the fixed-effect regressions of investment on Tobin's Q, cash flow, interactions of cash flow with cash-flow rights and with control minus ownership, controls, the country dummies, and the four-digit industry code dummies for the 19931996 panel data. The regression also includes the four-digit industry dummies. Investment is defined as capital expenditures over assets. Sales is the total sales over assets in the previous year. Tobin's Q is defined as the market value of equity plus assets minus the book value of equity over assets. Cash flow is operating cash flow over assets. Cash-flow right is the largest shareholders' cash-flow rights. Control minus ownership is the difference between the largest shareholders' control rights and cash-flow rights. Ln(Size) is the natural logarithm of assets. t-statistics are in parentheses. Firm fixed effects and year dummy variables are included in all regressions. and denote significance at the 10% and 5% levels, respectively.

interacting cash flow with the logarithm of the firm's book value of assets (Cash flow Ln(Size)). The result is given in Model 4 of Table 5. The coefficient on Cash flow Cash-flow rights changes very little from 0.633 to 0.677, but it becomes significant at the 10% level. The coefficient on Cash flow (Control minus ownership) changes from 1.097 to 1.078 and is still highly significant. The coefficient on Cash flow Ln(Size) is negative but insignificant. In addition, it is possible that our results reflect a positive correlation between a firm's growth potential and the largest shareholder's ownership. To control for this possibility, we include a term interacting cash flow with the firm's Tobin's Q (Cash flow Tobin's Q) in addition to the term interacting cash flow with firm size (Cash flow Ln(Size)). In Model 5, the coefficient on Cash flow Tobin's Q is negative but it is insignificant. The coefficients on Cash flow Cash-flow rights and Cash flow (Control minus ownership) change very little, compared with the results from Model 4. The results from both Models 4 and 5 indicate that our conclusion about the overinvestment hypothesis caused by the agency costs of free cash flow is robust to the inclusion of cash flow interacted with firm size and with Tobin's Q. In sum, the results in Table 5 indicate that our overinvestment conclusion is robust to a number of alternative specifications with different sets of control variables that may affect our results. That is, our results supporting the overinvestment hypothesis caused by the agency costs of free cash flow, which are heightened with the enhancement and entrenchment effects of the largest shareholders' ownership structures, are very robust to various empirical specifications. 5.3. Comparison of overinvestment problems between low ROA and high ROA groups The overinvestment problem should be a more important issue in firms with a lower return on investment than in firms with a higher return on investment. To test this hypothesis, we classify all firms into two groups. The low ROA

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

129

Table 6 Regression of investment on Tobin's Q, cash flow, and the largest shareholders' ownership structures: low ROA group versus high ROA group Independent variable Low ROA group (Column 1) Intercept Sales Tobin's Q Cash flow Cash flow Cash-flow rights Cash flow (Control minus ownership) Cash flow Ln(Size) Cash flow Tobin's Q Country and industry dummies Adjusted R2 Number of observations 0.066 (4.07) 0.001 (0.16) 0.013 (1.52) 0.646 (2.67) 1.190 ( 1.73) 1.910 (2.44) 0.023 ( 1.64) 0.035 ( 0.52) Yes 0.058 1155 High ROA group (Column 2) 0.042 (2.11) 0.018 (1.03) 0.009 (1.26) 0.589 (3.14) 0.822 ( 1.43) 0.239 (0.35) 0.024 ( 0.56) 0.015 ( 0.86) Yes 0.048 1400

This table reports fixed-effect regression results of capital investment on Tobin's Q, cash flow, interactions of cash flow with cash-flow rights and with control minus ownership, controls, the country dummies, and the four-digit industry code dummies for the two groups classified based on returns on assets (ROA). The 19931996 panel data is used. Investment is defined as capital expenditures over assets. Sales is the total sales over assets in the previous year. Tobin's Q is defined as the market value of equity plus assets minus the book value of equity over assets. Cash flow is operating cash flow over assets. Cash-flow right is the largest shareholders' cash-flow rights. Control minus ownership is the difference between the largest shareholders' control rights and cash-flow rights. Column 1 reports the regression results for the low ROA firms with an average four-year ROA of less than 0.05. The low ROA group includes 456 firms for a total of 1155 firm-years. Column 2 reports the regression results for the high ROA firms with a four-year average ROA of greater than or equal to 0.05. The high ROA group includes 538 firms for a total of 1400 firm-years. Firm fixed effects and year dummy variables are included in all regressions. t-statistics are in parentheses. and denote significance at the 10% and 5% levels, respectively.

group contains firms with a four-year (19931996) average ROA of less than 0.05, while the high ROA group contains firms with a four-year average ROA of greater than or equal to 0.05.12 The results are reported in Table 6. For the low ROA group, the coefficient on Cash flow Cash-flow rights is negative and significant at the 10% level. The coefficient on Cash flow (Control minus ownership) is positive and significant at the 5% level with a t-statistic of 2.44. The results demonstrate that in firms with a low ROA, the sensitivity of investment to cash flow simultaneously decreases with the largest shareholders' cash-flow rights and increases with the degree of separation between the largest shareholders' ownership and their control rights. The results are consistent with the overinvestment hypothesis. In the high ROA group, the coefficient on Cash flow Cash-flow rights is negative but not significant. The coefficient on Cash flow (Control minus ownership) is positive but not significant. This suggests that the overinvestment problem caused by the agency costs of free cash flow is more severe in firms with lower returns on investments.13

We also use different ROA cut-off points to classify firms into low and high ROA groups. The results are essentially similar to those reported in Table 6. 13 Previous research has documented that large shareholders in general and the separation of ownership and control rights in particular are usually associated with family ownership. It is possible that investment behavior is different between large shareholders in a family and large shareholders not in families. To test this hypothesis, we divide the sample into two groups: firms with family owners as the largest shareholder (about 70% of our sample) and firms with non-family owners (including the state, or widely held corporations and financial institutions) as the largest shareholders. The unreported results suggest that overinvestment caused by the agency costs of free cash flow is more severe in firms with family owners than in firms with non-family owners as the largest shareholders.

12

130

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

5.4. Discussion of endogeneity issues The empirical tests conducted above implicitly assume that the levels of the largest shareholders' ownership and control rights are exogenously set. An issue that might arise is the possibility of a reverse causality in terms of the impact of the divergence between ownership and control rights on the investment-cash flow sensitivity. Suppose that the largest shareholder faces an asymmetric information problem. He/she might then decrease his/her cash-flow rights, but increase the level of the divergence between his/her control rights and cash-flow rights. On the contrary, suppose that the largest shareholder faces the free cash-flow problem. He/she might then increase his/her cash-flow rights, but decrease the level of the divergence between his/her control rights and cash-flow rights. We could then find that as the level of cash-flow rights becomes larger and/or the divergence becomes smaller, the investment-cash flow sensitivity decreases. Hence, this result would not tell us much about the possible entrenchment effect of the separation between ownership and control. However, it seems unlikely that firms can change their cashflow rights and control rights quickly and frequently in light of temporary free cash flow or asymmetric information problems. In fact, La Porta et al. (1999) report that the ownership structures of the top 20 to 30 East Asian firms are relatively stable over time. As a result, the endogeneity issue would not be important in this case. 6. Conclusions The high degree of separation between control rights and cash-flow rights in East Asian corporations provides an excellent sample with which to study investment-cash flow sensitivities in differentiating the agency costs of free cash flow from asymmetric information problems in capital markets. This separation may be impossible to detect in U.S. corporations. We find that the investment-cash flow sensitivity decreases with increases in the cash-flow rights of the largest shareholders but it increases with the degree of divergence between the control rights and cash-flow rights of the largest shareholders. All these results are consistent with the overinvestment hypothesis caused by the agency costs of free cash flow, but are inconsistent with the underinvestment hypothesis caused by asymmetric information problems. Furthermore, our results also demonstrate the enhancement and entrenchment effects of the largest shareholders' ownership and control on corporate investment. Our study contributes to the debate on whether the positive association between cash flow and investment is due to the largest shareholders' preferences to overinvest or to underinvest. The previous literature does not provide a good instrument to differentiate between these two explanations. We believe that using the largest shareholders' cash-flow rights and their control rights provides a better instrument in distinguishing between these two explanations than does using managerial cash-flow rights alone. The reason is simple. While it is not clear how we can separate the entrenchment effect from the enhancement effect based on the largest shareholders' cash-flow rights alone, the degree of divergence between the largest shareholders' control rights and their cash-flow rights in East Asian corporations does provide an excellent measure of the largest shareholders' entrenchment. Our results are different from those reported by Hadlock (1998). Firms in both the U.S. and East Asia exhibit strongly positive investment-cash flow sensitivities. However, our findings reveal that the reasons for the positive investment-cash flow sensitivities are different. While Hadlock's data support the underinvestment hypothesis caused by asymmetric information problems in U.S. corporations, we document evidence that is consistent with the overinvestment hypothesis caused by the agency costs of free cash flow in East Asian corporations. The difference between our results and Hadlock's may be the result of differences in institutional factors. It is well known that the U.S. economy is characterized by dispersed ownership structures, effective capital market governance, and stronger protections of shareholders. In contrast, the East Asian economies are distinguished by family firms, state-owned firms, pyramid and concentrated ownership structures, weak capital market governance, and expropriation of minority shareholders by dominant shareholders. These more fundamental factors may contribute to the differences in the agency problems that firms in the two groups of economies face. Previous literature, for example, La Porta et al. (1999) and Claessens et al. (2002), points out that East Asia and other countries outside the U.S. or the U.K. have different agency problems. In fact, the agency problem in the U.S. centers on the conflicts of interest between managers and shareholders, while the agency problem in East Asia is defined by conflicts of interest between large shareholders and minority shareholders. We provide further evidence to support their arguments that the level of divergence between the control rights and cash-flow rights of the largest shareholders does have an entrenchment effect on corporate overinvestment.

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

131

Since the Asian financial crisis, people have often argued that one of the major reasons for the crisis was overinvestment by Asian corporations due to too much cheap capital flowing into the region before the crisis. They have further argued that the overinvestment problem was caused by poor corporate governance mechanisms in these countries, especially in regard to the significant divergence between the control rights and cash-flow rights of the largest shareholders (see, for example, Stiglitz (1998), Greenspan (1999), Rajan and Zingales (1998), and Yellen (1998)). However, up to today, no direct empirical evidence has been offered to support these arguments. Our paper provides evidence that these conjectures are correct. More specifically, we find that firms in East Asian emerging economies showed the symptom of overinvestment before the Asian financial crisis (i.e., during our sample period from 19931996). This is particularly true for firms with the greatest divergence between the largest shareholders' control rights and their cash-flow rights and for firms with lower ROA. Since the Asian financial crisis, the quality of corporate governance mechanisms in the East Asian region has been substantially improved via government regulations, such as the requirement that independent directors serve on company boards, and the scrutiny of shareholder rights activists. However, the problems of concentrated ownership structures and the divergence between the largest shareholders' control rights and their cash-flow rights have not been resolved. Whether or not the improvement in the quality of corporate governance mechanisms alone has changed the investment behaviors of East Asian corporations is left for future research. References
Baker, M., Stein, J.C., Wurgler, J., 2003. When does the market matter? Stock prices and the investment of equity-dependent firms. Quarterly Journal of Economics 118, 9691005. Claessens, S., Djankov, S., Fan, J.P.H., Lang, L.H.P., 2002. Disentangling the incentive and entrenchment effects of large shareholders. Journal of Finance 57, 27412771. Claessens, S., Djankov, S., Lang, L.H.P., 2000. The separation of ownership and control in East Asia corporations. Journal of Financial Economics 58, 81112. Cleary, S., 1999. The relationship between firm investment and financial status. Journal of Finance 54, 673692. Dittmar, A., Mahrt-Smith, J., Servaes, H., 2003. International corporate governance and corporate cash holdings. Journal of Financial and Quantitative Analysis 38, 111133. Dybvig, P.H., Zender, J.F., 1991. Capital structure and dividend irrelevance with asymmetric information. Review of Financial Studies 4, 201219. Fan, J.P.H., Wong, T.J., 2002. Corporate ownership structure and the informativeness of accounting earnings in East Asia. Journal of Accounting and Economics 33, 401425. Fazzari, S., Hubbard, R.G., Peterson, B.C., 1988. Financing constraints and corporate investment. Brookings Paper on Economic Activity 1, 141195. Fazzari, S., Hubbard, R.G., Peterson, B.C., 2000. Investment-cash flow sensitivities are useful: a comment on Kaplan and Zingales. Quarterly Journal of Economics 115, 695705. Greenspan, A., 1999. Lessons from the global crises. Address to the World Bank Group and International Monetary Fund, September 27, 1999. Hadlock, C.J., 1998. Ownership, liquidity, and investment. RAND Journal of Economics, 29, 487508. Harvey, C., Lins, K., Roper, A., 2004. The effect of capital structure when expected agency costs are extreme. Journal of Financial Economics 74, 330. Hoshi, T., Kashyap, A., Scharfstein, D., 1991. Corporate structure, liquidity, and investment: evidence from Japanese industrial groups. Quarterly Journal of Economics 106, 3360. Hubbard, R.G., 1998. Capital-market imperfections and investment. Journal of Economic Literature 36, 193225. Jensen, M., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review 76, 323329. Jensen, M., Meckling, W., 1976. The theory of firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, 305350. Joh, S.W., 2003. Corporate Governance and firm profitability: evidence from Korea before the economic crisis. Journal of Financial Economics 68, 287322. Johnson, S., Boone, P., Breach, A., Friedman, E., 2000. Corporate governance in the Asian financial crisis. Journal of Financial Economics 58, 141186. Kalcheva, I., Lins, K.V., in press. International evidence on cash holdings and expected managerial agency problems. Review of Financial Studies, forthcoming. Kaplan, S., Zingales, L., 1997. Do investment-cash flow sensitivities provide useful measures of financing constraints? Quarterly Journal of Economics 112, 169216. Kaplan, S., Zingales, L., 2000. Investment-cash flow sensitivities are not valid measures of financing constraints. Quarterly Journal of Economics 115, 707712. Lamont, O., 1997. Cash flow and investment: evidence from internal capital markets. Journal of Finance 52, 83110. La Porta, R., Lopez-De-Silanes, F., Shleifer, A., 1999. Corporate ownership around the world. Journal of Finance 54, 471518. Lemmon, M.L., Lins, L.V., 2003. Ownership structure, corporate governance, and firm value: evidence from the East Asian financial crisis. Journal of Finance 58, 14451468.

132

K.C.J. Wei, Y. Zhang / Journal of Corporate Finance 14 (2008) 118132

Lins, K.V., 2003. Equity ownership and firm value in emerging markets. Journal of Financial and Quantitative Analysis 38, 159184. McConnell, J., Servaes, H., 1990. Additional evidence on equity ownership and corporate value. Journal of Financial Economics 27, 595612. Mitton, T., 2002. A cross-firm analysis of the impact of corporate governance on the East Asian financial crisis. Journal of Financial Economics 64, 215241. Morck, R., Shleifer, A., Vishny, R., 1988. Management ownership and market valuation: an empirical analysis. Journal of Financial Economics 20, 293315. Morck, R., Shleifer, A., Vishny, R., 1990. The stock market and investment: is the market a side-show? Brookings Papers on Economic Activity 2, 157215. Myers, S., Majluf, N., 1984. Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13, 187221. Pinkowitz, L., Stulz, R., Williamson, R., 2006. Does the contribution of corporate cash holdings and dividends to firm value depend on governance? A cross-country analysis. Journal of Finance 61, 27252751. Rajan, R.G., Zingales, L., 1998. Which capitalism? Lessons from the East Asian crisis. Journal of Applied Corporate Finance 11, 4048. Shin, H.H., Stulz, R.M., 1998. Are internal capital markets efficient? Quarterly Journal of Economics 113, 531552. Shleifer, A., Vishny, R.W., 1997. A survey of corporate governance. Journal of Finance 52, 737783. Stiglitz, J., 1998. The role of international financial institutions in the current global economy. Address to the Chicago Council on Foreign Relations, February 27, 1998. Stulz, R., 1988. Managerial control of voting rights: financing policies and the markets for corporate control. Journal of Financial Economics 20, 2554. Yellen, J., 1998. Lessons from the Asian crisis. Speech at Council on Foreign Relations, New York.