Tunneling Dividends by liaoqinmei


									                                           Tunneling Dividends¡î

                                   Chi-Wen Jevons Leea , b,*, Xing Xiaob, **
               A.B. Freeman School of Business, Tulane University, New Orleans, LA 70118-5669, USA
                   School of Economics and Management, Tsinghua University, Beijing, 100084, China

                                                  November 2004

Numerous findings in the literature suggest that paying cash dividend mitigates agency problem
between majority shareholders and minority shareholders. Many common law countries require
mandatory cash dividend policy to protect minority shareholder’s interest. This paper provides
opposite evidence. We find that state dominant firms in China have high propensity to pay cash
dividend but low propensity to subscribe rights offering. Furthermore, state dominant firms often
increase cash dividend payment soon after rights offerings. As state-held shares in China are
non-tradable, giving up subscription rights and using receipts from rights offering to pay cash
dividend are equivalent to selling a portion of the non-tradable shares by the majority
shareholders to the minority shareholders. The computed selling price is about three times higher
than that of officially approved private placement. Market reacts negatively to the cash dividend
announcement of state dominant firms, but positively to others. Our findings suggest that instead
of alleviating agency problem, cash dividend might be used as a vehicle for tunneling in state
dominant firms.

JEL classification: G34, G35, G38, M41
Key words£º agency problem, cash dividends, corporate governance, rights offering, state
dominant firms

¡î We acknowledge detailed comments from Halit Gonenc in 2003 EFMA Annual Meeting, Millicent Chang in 2004
AFAANZ Annual Meeting, and Henk Berkman in 2004 CICF Meeting. We also appreciate comments from Yan Cao,
Chao Chen , Kevin Chen, S.P. Kothari, Charles Lee, Jing Liu, T.J. Wong, and the seminar participants of the 4th
Doctoral Consortium of APJAE, the 2nd Doctoral Symposium of Greater China, HKUST, National Taiwan University,
Taipei University, Tsinghua University , Tulane University, and Zhejiang University. Xing Xiao acknowledges the
financial support of National Social Science Grant (No. 03CJY005).
*Corresponding author: Chi-Wen Jevons Lee, A.B. Freeman School of Business, Tulane University, New Orleans, LA
     70118-5669, USA Tel: (504) 862-8485, Jevons.Lee@Tulane.Edu.

**Contact information of Xing Xiao: School of Economics and Management, Tsinghua University, Beijing 100084, P.
R. China. Tel: (8610)6279-5464, E-Mail: xiaox@em.tsinghua.edu.cn.
                                            Tunneling Dividends
1. Introduction

         Tunneling, defined as siphoning assets and profits out of firms for the benefit of their

controlling shareholders, has attracted attention from academic and media cycles in recent years

as an important issue of corporate governance.1 However, empirical evidences about tunneling

rarely went beyond anecdotal story telling. The study of stock market in China allows us to

collect extensive data to conduct a detailed examination of tunneling. We find an unusual

tunneling devise: the controlling shareholders use cash dividend as a medium to sell their

non-tradable shares to the minority shareholders at a favorable price.

         As a transition economy, literally all listed firms in China were carved out from

    state-owned enterprises (SOEs). The parent-SOE holds the non-tradable state shares of the listed

    firm, whereas the new shareholders in wake of IPO hold negotiable shares. The non-tradable

    shares can be traded only with private placement between institutions under special permission.2

    While selling a portion of shareholding in the U.S. can be viewed as home-made dividend

    according to Miller and Modigliani (1961), this instrument is either not available or very costly

    for controlling shareholders in China’s state-dominant firms.

         Furthermore, to assure smooth birth of stock market and out of big-daddy role in socialism,

    Chinese government sets very stringent quota on IPO and regulated every detail of the IPO

    process, such as the volume of offering, the asset valuation, and the P/E multiple for IPO

    pricing. 3 In order to increase the total amount raised from IPO under these restrictive

    regulations, parent-SOEs generally carve out the most profitable portion for listing, while

    keeping the non-performing assets and less productive employees in the parent-SOEs.

  Johnson, et al. (2000) raised the issue of tunneling with anecdotal stories. They suggested that even the legal system
in developed nations allows tunneling to take place. They also indicated that legal tunneling took a variety of forms,
including expropriation of corporate opportunity, transfer of assets from the firm by controlling shareholders at
below-market prices, etc. Bae et. al. (2002) found evidence of using acquisition as a way to explore minority
shareholder by the Chaebol. Jian and Wong (2004) found that “group-firms” in China use the related party transaction
for tunneling.
  Lack of open competition, the transaction price in private placement, which is 1.49 times of the book value, fetches
only one third of the going price for negotiable shares.
  For the reform of SOE, see Lee (1996, 2001), and Chen, Lee, and Li (2003). For the process of IPO, see Aharony,
Lee, and Wong (2000).
     The cash flow required for feeding the inefficient parent-SOE together with the restriction

on selling the state shares motivate the parent-SOE to use cash dividends to liquidate her


     Two factors affect the feasibility of liquidating with cash dividends. Firstly, other

shareholders get dividends proportional to controlling shareholder. This is a cost to controlling

shareholder when liquidating with cash dividends, and this cost is associated with shareholding

concentration. We find that both the probability of paying cash dividend and the payout ratio

decrease with the shareholding concentration of controlling shareholder.

     Secondly, liquidating with cash dividends is possible only when the listed firm has enough

cash. However, government’s strict quota and the various regulations on IPO process make

many listed firms seriously lack of working capital. In short of funds, the listed firms initiate

rights offering to raise cash and uses part of receipts to pay cash dividends. We find that listed

firms increase cash dividend payment within one year after rights offering. We also find that all

negotiable shareholders subscribe rights offering, but only less than 30% of non-tradable

shareholders do. When the listed firm uses part of rights offering receipts to pay cash dividends

to all shareholders, it is equivalent to force negotiable shareholders buy shares from

non-tradable shareholders.

     Johnson et. al. (2000) argued that controlling shareholder could tunnel minority

shareholder through legal transactions including contracts favoring the controlling shareholder.

If non-tradable shareholders sell part of their shareholding to negotiable shareholders at a

disadvantaged condition, using receipts from rights offering as cash dividend payment after

giving up the subscription right might be tunneling to negotiable shareholders that participate in

rights offering. We find that market react negatively to cash dividend announcement of state

dominant firms, but positively to others. This indicates that negotiable shareholders failed to

anticipate the subsequent cash dividend at the point of rights offering. Hence, they are not price

protected when they subscribe shares in rights offering. In this case, the subsequent cash

dividend payment is a tunneling to negotiable shareholders.
      Many researches indicate that paying cash dividend can limit the abuse of firm’s free cash

flow by managers, and alleviate agency problem between shareholder and managers (Easterbrook

(1984), Jensen (1986), Gomes (2000)). Some recent studies show that paying cash dividend can

also alleviate agency problem between controlling shareholder and minority shareholders. La

Porta et. al. (1998) studied the relationship between dividend payout ratio and firm’s growth

opportunity around the world. Faccio et. al. (2000) argue that cash dividend can mitigate

expropriation of controlling shareholder to minority shareholders. Some common law countries

use mandatory cash dividend policy as a way to protect minority shareholders (La Porta et. al.

(2000)). However, this paper provides opposite evidence. Our finding shows that cash dividend

might not limit tunneling, but might be a vehicle for tunneling instead. This challenges

mandatory cash dividend policy for protecting minority shareholders.

      The remainders of this paper are organized into six sections. Section 2 documents cash

dividend behavior in China. We discover two puzzling empirical regularities. Section 3 analyzes

institutional setting in China and develops hypotheses. Section 4 describes our sample and data.

Section 5 deals with research design. Section 6 reports empirical tests, and section 7 concludes.

2. Puzzling Cash Dividend Paying Behavior in China

    According to agency theory of cash dividend (Jensen (1983), Easterbrook (1984), Gomes

(2000)), firms payout free cash flow as cash dividend to mitigate agency problem. Agency theory

predicts that cash dividend payment increases with free cash flow, but decreases with growth

opportunity of the firm. Ambarish et. al. (1987) also derive a negative relationship between

growth and dividend payouts in a signaling model. Although La Porta et. al. (1998) argues that

the relationship between growth and dividend payout might be affected by law protection to

shareholders, they still predict that there is a positive or negative relationship between the two

according to different degree of shareholder protection.

    Table 1 shows the mean free cash flow and sales growth of dividend payers and non-payers

respectively. It is shown that there is no relationship between growth rate and cash dividend
payment. Furthermore, the mean free cash flow for both dividend payers and non-payers are

negative. These are not consistent with agency theory or signaling theory of cash dividend.

                                       [Insert Table 1 here]

    When firms pay cash dividend to convey the information that managers or majority

shareholders will not abuse free cash flow (agency theory) or that about future performance,

increase of cash dividend is interpreted as good news, and hence market will respond positively

(Ahroney and Swary (1980), Asquith and Mullins (1983)). Figure 1 shows the market reaction to

cash dividend announcement. We can see that market has no reaction to increase of dividend, but

has positive reaction to decrease of dividend. This is opposite to the prediction of agency theory

or signaling theory.

                                       [Insert Figure 1 here]

    If China’s listed firms don’t pay cash dividend to solve agency problem or convey

information about future performance, is cash dividend not prevalent in China? Table 2 shows the

frequency of cash dividend payment in China. We can see that 47.18% of listed firms pay cash

dividend in China, and the mean payout ratio is 60.17%. Both the frequency of cash dividend

payment and the payout ratio are even higher than that of the U.S. (According to Fama and

French (2001)).

                                       [Insert Table 2 here]

    Taxation is also a prevailing explanation for cash dividend. However, capital gain is tax free

for individual investors in China, whereas cash dividend is taxable on a 20% tax rate. For

institutional investors, capital gain and cash dividend are considered as taxable income

indifferently. For this reason, taxation should not be a motivation for paying cash dividend.

    The indifference of free cash flow and growth between dividend payers and non-payers,

together with the positive market reaction to decline of cash dividend lead us to ask the question:

Why do China’s listed firms pay cash dividend?
3. Institutional Settings and Hypotheses Development

3.1. Large shareholder and cash dividend

       Miller and Modogliani (1961) argue that capital gain is ‘home made’ cash dividend when

transaction cost is zero. However, La Porta et. al. (1999) indicates that the existence of a large

shareholder is pervasive. When there is a large shareholder, getting ‘home made’ cash dividend is

costly. The large shareholder have to face threaten of dilution of control right and negative

influence on market price.

       In China’s stock market, the largest shareholder holds 45.55% of shares on average. This

indicates the existence of large shareholder in China. Furthermore, most of China’s listed firms

are carved out from SOEs. The parent SOE holds the state share of the listed firm. To uphold the

fundamental doctrine of Socialism that all economic means belong to “The People,” all state

shareholdings are ruled by the State Council as non-tradable so that “no person is allowed to sell

out the Socialism.” Non-tradable shares can only be bought and sold through private placement

with special approval from the government. The non-tradable status of controlling shareholding

makes it more costly for large shareholder getting ‘home made’ dividend. Paying cash dividend

might be necessary for them.

       On the other hand, government has strict restrictions on IPO. CSRC determines which firm

can go public, the total shares to be issued, and the P/E ratio that determines issuing price.                             In

this case, the only opportunity that parent-SOEs can increase the amount of money raised in IPO

is to increase earnings. However, there are also restrictions on the conversion ratio between

carved out assets from the parent-SOEs and the shares they get.4 Hence, in order to raise more

capital from the IPO, the parent-SOEs have to carve out the assets that earn a higher return. For

these reason, parent-SOEs generally carve out the most profitable assets for public offering. The

parent-SOE keeps those money-draining social services such as kindergartens, primary schools,

dining rooms, hospital, and sometimes even high-way maintenance and local policy force. This

process is called “financial packaging” as was discussed in Aharony, Lee and Wong (2000). After

    In order to protect state owned assets, there is a restriction on the conversion ratio between the assets and the shares.
     Every one Yuan assets should be converted to at least 0.65 shares in the listed firm.
the financial packaging, the non-performing assets are left to parent-SOEs. Moreover, the

younger, more able, and better educated employees are transferred to the listed firms. The older

and less educated are kept in the parent-SOE. Hence, the financial packaging of IPO process

works extremely unfavorable against the parent-SOE. The manager and employees of

parent-SOEs buy in for the premise that the parent-SOE holds the ultimate control and the listed

firm will pay tributes back to the parent-SOE in the future. Very much like the traditional Chinese

family structure, the parent takes excruciating pain and tremendous sacrifice to give birth and to

raise a child. However, filial piety is expected in return. With shareholding non-tradable, making

the listed firms to pay cash dividend is one of the ways to fulfill parent-SOEs expectation.

3.2. The birth of stock market and the lack of working capital of parent-SOEs

     As a socialist planned economy for three decades until 1978, most market activities in China

ceased to function and most manufacturing activities were carried out by SOEs. Government

allocated funds and resources to SOEs and then distributed SOEs’ products to households by

directives. Beginning in 1978, a series of reform were adopted to revive the moribund SOE. In

1979, SOEs were allowed to retain part of their profits for future development. The fiscal budget

was replaced in 1980 by bank lending as the major financing source for SOEs. However, due to

primitive monitoring and control, the whole China was awash with bad debts by 1990, known as

the “triangle debt” problem. The ensuing liquidity crunch troubled all SOEs for more than a


     Chinese stock market was created in 1990 to transfer the government’s responsibility of

financing SOEs to financial intermediaries in the open market. To assure stable stock market

development, the central government sets stringent quota and requires tough screening for all

financial transactions in the stock market.5 The listed firms were grossly undercapitalized due to

 Prior to Dec. 1996, quota system restricts the amount of shares can be issued in each province. The local government
of province chooses the SOEs to go public, and report to CSRC for approval. In order to increase the number of firms
to be listed, the local government always limits the size of the assets going public, and tries to increase the number of
firms going public. In Dec. 1996, the quota system changed to restrict the number of firms to be listed. Response to
such change, the provincial government bundles several SOEs to go public. Restriction on issuing price is based on a
predetermined range of P/E, which is generally 15 to 20. The authorization system ceased to be effective in 2001.
Restrictions on issuing price were also canceled in 2001.
these regulations. 76% of the listed firms requested the approval of rights offering for raising

additional capital in two years after IPO (Lee and Song (2003)). They faced severe shortage of

working capital. Table 1 shows that for both cash dividend paying firms and non-payers, the

mean free cash flows are mostly negative.

3.3. Rights offering and tunneling dividend

     As many listed firms are seriously lack of working capital, the scheme of retrieving cash

from the firm by controlling shareholder is more feasible after rights offering. However, paying

cash dividends with the receipts from rights offering may do harm to negotiable (minority)

shareholders. The price of rights offering is mostly below the market price, but well above the

book value. Hence, all negotiable shareholders would subscribe, but non-tradable shareholders

often not. Table 3 shows that on average, non-tradable shareholders only subscribe 29% of the

shares allocated to them in rights offering whereas negotiable shareholders do 100%. Using the

receipt from rights offering to pay cash dividends is equivalent to ask negotiable shareholders to

pay non-tradable shareholders. 6 Managers have the final decisions on rights issuing price.

Underwriters buy all the negotiable shares left in rights issuing if there is no enough subscription.

These shares are generally sold in the market in a short period of time.

                                                  [Insert Table 3 here]

     Johnson et. al. (2000) argued that controlling shareholder could tunnel minority shareholder

through legal transactions including contracts favoring the controlling shareholder. If negotiable

shareholders cannot anticipate the subsequent cash dividend payment at the point of rights

offering, they are not price protected from being tunneling. As a consequence, using receipts

from rights offering as cash dividend payment after giving up whole or part of the subscription

right, non-tradable shareholders may ask negotiable shareholders that participate in rights

offering to buy their shares at a disadvantaged condition. In this way, non-tradable shareholders

  Actually, the biggest prey of this scheme of expropriation by cash dividends is the “People (aka State)”. The parent
SOE is the custodian for the state ownership. Cash dividends are generally consumed within the parent SOE. Cash
dividend is a tool for the custodian to liquidate his principal’s asset for his own benefit. This paper focuses on the
interaction between shareholders. We will not discuss the expropriation of the People.
may tunnel negotiable shareholders with cash dividend payment.7

3.4. Ownership concentration and dividend policy

     Parent SOE has the incentive to retrieve cash from the firm with cash dividend. As cash

dividend must be paid strictly based on proportion of shareholding, parent SOEs’ motivation for

paying cash dividend might be related to shareholding concentration.

     On the other hand, parent SOEs may tunnel negotiable shareholders with cash dividend if the

firm use receipts from rights offering as dividend. La Porta et. al (1999) and Bennedsen and

Wolfenzon (2000) suggest that monitors from other large shareholders can reduce the controlling

shareholder expropriation. However, the motive of other large shareholders to monitor the large

shareholder in control is to avoid themselves being expropriated. Other non-tradable shareholders

may have equal benefit from tunneling in the case of cash dividend, so it is also possible that they

collude with parent-SOE to tunnel negotiable shareholders. However, even if other non-tradable

shareholders collude with parent-SOEs, the presence of other free-riding non-tradable

shareholders would increase the cost of tunneling for the parent SOE. Hence, the incentive for the

parent SOE to issue cash dividends decline as the concentration of her shares decline.

     When the concentration of parent SOE’s holding is very low, the cost of tunneling with cash

dividend becomes very high. Furthermore, with no dominating shareholders, the manager of

listed firm has incentive to use cash dividends to convey information about the firm’s future

performance. The board of this independent listed firm would use the cash dividends to control

and monitor the manager, like what is going on in the U.S. To sum up, we state the following


     H1: State dominant firms(high state share concentration) pay cash dividend to retrieve cash

from the firm. Independent firms (low state share concentration) pay cash dividend to convey

   A question is that why negotiable shareholders participate in rights offering if they are tunneled. Actually not all
listed firms use receipts from rights offering to pay cash dividend. Furthermore, among those firms that initiate rights
offering before paying cash dividends, not all of the non-tradable shareholders give up their subscription rights. Hence,
there are two types of firms, and two types of actions for each type. Negotiable shareholders cannot distinguish which
type is the firm, and which type is the rights offering ex ante. Hence, it’s tunneling ex post, but not ex ante.
future performance.8

     We can test the above hypothesis from several different angles, by examining the cash

dividends policy in terms of (1) frequency of paying, (2) payout ratio, and (3) level of payment.9

Therefore we can derive following three testable hypotheses.

     H1 (a): Other things being equal, the probability of paying cash dividend increases with

shareholding concentration.

     H1 (b): Other things being equal, dividend per share and payout ratio increases with

shareholding concentration.

     The tunneling hypothesis suggests that a motive behind rights offering is to raise funds from

negotiable shareholders to pay non-tradable shareholders through cash dividends. Hence, we

have following testable hypotheses.

      H2(a): For state dominant firms, the probability of paying cash dividend,                        dividend per

share, and payout ratio increase after rights offering..

     H3 (b): There is favorable market reaction to the decline of cash dividends for state

dominant firms, but the opposite to independent firms.

4. Sample and data

     We limit our study to the period of 1996 to1999. The stock market started functioning in

China at 1990 as a social experiment. The common stocks were treated as government bonds at

the birth of stock market. Cash dividend was fixed and statutory before the CSRC was

established in 1993. In the first two years of CSRC’s existence, cash dividend payment was rather

volatile. Both the CSRC and the investors were learning their roles in the market. Hence, we

exclude the data before 1996 from our study. A stipulation was announced on Mar. 28, 2001

   Although not all listed firms are controlled by parent-SOE, privately controlled listed firms are very rare in our
sample. Other controllers include government agency and other companies. Approximately 20% of all the listed firms
are controlled by government agencies or other companies. Our tests show that there is no significant difference
regarding cash dividend paying behavior among those different types of firms.
  A few studies have explored the economic determinants on the frequency of dividend payment. Wei (2000) found
that the higher the proportion of state ownership, the higher is the frequency. Lv and Wang (1999, 2002) found that
firm size, book value, profitability, liquidity, agency cost, state ownership, and debt burden are correlated with
frequency of paying dividend.
requiring continuous three years of cash dividend as a condition for the approval for rights

offering. Because dividend payment for 2000’s profit is announced in 2001, we exclude cash

dividend payment for 2000’s profit. This method of sample selection can eliminate a large

number of confounding factors in a transition economy. The pattern of cash dividends behavior

documented in Tables 3 may be limited to the period of 1996-1999. However, the proposition that

the ownership and the control by the non-tradable state-shareholding would affect cash dividend

decisions should apply well beyond this sample period.

                                                    [Insert Table 3 here]

       Firms with negative earnings and those have B share or H share issued were deleted from our

sample. Firms with B or H shares face two or three markets simultaneously. Shareholders in

different markets may have different incentives. Our paper focuses on the game played by

negotiable and non-tradable shareholders in the domestic market.10 Regulations discourage firms

losing money to pay cash dividend. Furthermore, Chen, Lee, and Li (2003) pointed out that the

firm at loss faces the discipline action of delisting. They may use cash dividends to play a game

of totally different nature. It is beyond the scope of current study. We exclude all firms at loss in

our sample.

         Financial data and final dividend plan data used in this paper come from Genius Stock

Market Information System (GSMIS). We make necessary corrections on the data through

cross-checking the data in two websites, http://www.p5w.net.cn and http://www.cnlist.com.cn.

Preliminary dividend plan data were hand-collected from China Securities Daily and Shanghai

Securities Daily. Stock price and return data were collected from CSMAR database. The total

number of observations is 2397, of which 658 reported cash dividend on the after-tax basis11. The

     B share is share issued to foreign investors. H share is share issued in Hong Kong Stock Exchange. In early years,
firms usually had different dividend policy to different investors. For instance, they may pay cash dividends to foreign
investors, but stock dividends to domestic investors. According to New Securities Daily, Dec. 8, 2001, among the 27
firms that paid cash dividends continuously from 1996 to 2000, 12 have B or H shares issued. Firms with B or H shares
may have different motivations for cash dividends. This would be an issue beyond the scope of this paper.
   According to China’s “Individual Income Tax Law,” stock dividends are taxed at 20% rate. The taxable income is
based on one Yuan par value of the stock. The listed firm has the responsibility to collect accrued tax on the behalf of
Taxation Bureau. When firms pay cash dividend, they just deduct the tax payable from the cash dividend, and
shareholders only receive the amount after tax. However, when firms pay stock dividend, it is impossible to do so. To
average payout ratio is 60.17%, and mean dividend per share is 0.14 Yuan.

5. Methodology

     We divide all sample firms into two groups according to shareholding concentration: state

dominant firms (high state share concentration) and independent firms (Low state share

concentration). We use the difference between the largest shareholding and the second largest

shareholding to measure the concentration. Our purpose is to measure the dominance in

ownership and control by the parent-SOE. Low shareholding concentration refers to a difference

less than 15%. 12 Dummy variable CONCEN equals one for high concentration, and zero


     In Equations (1) to (3), we specify three regressions to test all hypotheses, H1 and H2,

together in terms of three measurements of dividend decision: to pay or not to pay (DIVP),

payout ratio (PAYOUT), and dividends per share (DPS). Equation (1) specifies the test related to

dividend paying decision, Equation (2) the test related to payout ratio decision, and Equation (3)

the test related to dividends per share decision.

  DIVPt = a + b 1 SHARDIFt +b 2 PREFINANCEt + b 3 PREFINANCEt×CONCENt + b 4 SIZEt + b 5

                          FCFt +b 6 ROAt + b 7 DIVPt-1 + ε                                (1)


                              b5 FCF t + b6 PAYOUTt-1 + ε t                         (2)

   DPSt= a+b1 SHARDIF t +b 2 PREFINANCEt + b 3 PREFINANCEt×CONCENt + b 4 SIZEt + b 5

         DIVPt-1 + b6 DIVPt-1¡Á CONCENt+b7 EPS t+b8EPS t¡Á CONCEN t + +ε t                               (3)

The three dependent variables in Equations (1)-(3) are specified as: DIVP equals to one for firms

paying cash dividends, and zero otherwise; PAYOUT is calculated on an after-tax basis (Dividend

per share / earnings per share); DPS is dividends per share collected from GSMIS data base.

solve the tax problem, some firms pay cash dividend to cover the tax payable induced by stock dividend. When this
happens, shareholders do not receive cash eventually. In order to eliminate the influence of tax, we calculate tax
payment on both a before-tax basis and after-tax basis.
   We also use the sum of top five and top ten shareholdings and sum of squared top five and top ten shareholdings to
measure the concentration. We also change the classification criteria for high/intermediate/low concentration. Our
results are robust with the various measurements of concentration
    FCF (free cash flow) is used to control the effect of Jensen’s Agency Theory. Normally,

dividends are paid out of profits. Hence, profitability, measured by ROA, may affect dividend

decision. We include ROA in Equation (1) as a control variable. Lintner (1956) found current year

DPS is a weighted average of DPS of the last year and EPS of the current year. Hence, we

include EPSt and DIVPt-1 in Equation (3).

6. Empirical Results

                                            [Insert Table 4 Here]

    We conduct a preliminary exploration in Table 4. First, we find an expected result that only

six out of 255 loss firms pay cash dividends. Current profitability is usually a precondition for

cash dividends under the regulation. Second, shareholding concentration is positively associated

with cash dividend paying decision. State dominant firms (high shareholding concentration) have

much higher tendency to pay cash dividends than independent firms (low shareholding

concentration). Third, firms had rights offering in previous year are more motivated to pay cash


6.1.Using cash dividend to retrieve cash from the firm and tunneling dividend

      Tables 5 to 7 report the test results on the two hypotheses from three angles,

correspondingly, the dividend paying decision, the payout ratio, and the dividend per share. The

specifications of regression were given in Equations (1)-(3) above. In all three tables, the

shareholding concentration variables, SHARDIF and CONCEN, are significant factors in various

aspects of cash dividend decisions. Hence, corporate ownership and control are the important

factor in shaping the cash dividend decision in China. The results are consistent with Hypotheses

H1(a) and H1(b). FCF is uniformly insignificant in Tables 5 and 6. Hence, the Agency Theory

articulated in Jensen (1986) is irrelevant to the case of China. These results are consistent with

Hypothesis H1(c). PREFINANCE is significant in Tables 5 and 6, but marginally so in Table 7.

Hence, the availability of cash through previous financing activities is crucial for dividend paying

decision and payout ratio decision, but less so for dividend per share. The results are consistent
with Hypothesis H2(a). Furthermore, for state dominant firms, the influence of shareholding

concentration (SHARDIF) on cash dividend payment is positively significant different from

independent firms. This is consistent with our prediction that state dominant firms are more

motivated to increase cash dividend payment after rights offering.

                                         [Insert Tables 5, 6, and 7 here]

         Regression (4) to regression (6) in Table 5 indicate that the more dominant is the parent-SOE in

shareholding, more frequent would the listed firm pay cash dividends. The result implies that the

parent-SOE desires cash dividends and uses her influence to get it. Comparing Regression (1) to

Regression (6) in Table 6, we can observe that the serial correlation in payout ratio is attributable to

shareholding concentration. Firms with low shareholding concentration show insignificant trace of

autocorrelation in payout ratio. Comparing Regression (3) with Regression (7) in Table 7, we can

observe that the serial correlation in DPS is mostly associated with firms of low shareholding

concentration. When the shareholding concentration gets higher, the level of serial correlation gets


6.2. Equivalent selling price when using receipts from rights offering as cash dividend

    When non-tradable shareholders give up all or part of the subscription right in rights offering,

their shareholding are diluted. If the firm increases cash dividend payment soon after rights

offering, the non-tradable shareholder is selling a proportion of the shares they hold to negotiable

shareholders. Table 8 reports the equivalent selling price and P/B. It is shown that non-tradable

shareholders sold 3.60% of the total shares of the firm to negotiable shareholders on average. In

return, they get 31.15 million Yuan with cash dividend. The equivalent selling price is 7.24. The

P/B is 3.74. Although this P/B is lower than the market average, which is 5.69, it is much higher

than that in private negotiation (P/B in private negotiation is 1.49 onaverage).

6.3. Information content of cash dividends

    Table 9 shows the market reaction to cash dividend announcement. In China, cash dividend
announcement is a part of the annual report. The important confounding factors include stock

dividends, earnings per share (EPS) and rights offering. We use the dummy variable STKDIVt to

control for the announcement of stock dividends, with one for announcing stock dividend and

zero otherwise.13 The dummy variable RI is to control for the announcement of rights issuing

with one for announcing rights issuing, and zero otherwise. We use the change of EPS as a

measure of the new information in EPS.14 The market reaction is measured by the cumulative

abnormal returns (CAR). Regression (4) is used to measure the refined effect of unexpected

dividends (UNEXPDIV).

             CARt = a + b1 STKDIVt + b2 RIt + b3UEXPDIVt + b4 UEXPEPSt + εt                                   (4)

     The results are reported in Panel A of Table 9. We find that the market reacts negatively to

cash dividends and positively to earnings news. However, these results are insignificant.

                                                    [Insert Table 9 here]

     In Panel B of Table 9, we set the dummy variable CONCEN to one for firms with the

difference of shareholding between the largest and second largest shareholder greater than 15%

and 0 otherwise to control for concentration of ownership and control by the parent SOE. Hence,

we have:
             CARt = a + b1 STKDIVt + b2 RIt + b3UEXPDIVt + b4 UEXPDIVt ¡Á CONCEN +
                             b5UEXPEPSt + εt                   (5)
Several interesting results can be found in Panel B. Comparing to Panel A, R-square in Panel B is

about ten times higher. Hence, controlling for shareholding concentration is important in measuring

market reactions to cash dividend news. The effect of stock dividends is still positive, but much less

significant. The effect of rights offering is still negative, but somewhat stronger. The market reactions

to cash dividend news and to earnings news become significantly positive, as is usually found in the

literature on well-developed markets in the West. High shareholding concentration makes cash

dividends unfavorable to investors.

   China’s listed firms have two forms of stock dividends. One is direct stock dividend; the other is transferring from
   capital paid in excess of par to paid-in-capital..
   For earlier works in this area, see Chen, Chen, Ni (1998), Wei (1998), Chen and Yao (2000), and Yu and Cheng
(2001). These studies did not pay attention to the institutional setting in China. Moreover, they did not properly control
confounding factors and generate conflicting results that are difficult to interpret.
6.4. Additional test

     In the previous tests, we have assumed that other non-tradable shareholders will monitor the

parent-SOEs, or their existence will increase the cost of parent-SOEs’ tunneling. However, it is

possible that other non-tradable shareholders collude with parent-SOE, or their existence has no

influence on parent-SOE’s behavior. We consider these possibilities in Table 10.

                                        [Insert Table 10 here]

     In Panel A of Table 10, we use the total proportion of shares hold by non-tradable shareholders

(NONNEGO) as a substitute of SHARDIF in regression (1) to (3). Similar to CONCEN in regression

(1) to (3), HINONNEGO here is used to divide sample firms into two groups: state dominant firms

and independent firms. HINONNEGO is a dummy variable. It equals 1 when NONNEGO is greater

than the sample median, and 0 otherwise.

     In Panel B of Table 10, the shareholding of the largest shareholder, FIRST and HIFIRST, are used.

HIFIRST equals 1 when FIRST is greater than the sample median, and 0 otherwise. With NONNEGO,

we assume that other non-tradable shareholders collude with the largest shareholder, and by using

FIRST, we assume that other non-tradable shareholders do not have any influence on parent-SOE’s


     From Table 10, we can see that the result in Table 5 to Table 7 doesn’t hold when SHARDIF is

substituted with NONNEGO. However, the results hold when FIRST is used. Although it is still

impossible to verify whether other non-tradable shareholders monitor parent-SOE or their existence

just increases parent-SOE’s tunneling, we can exclude the possibility that other non-tradable

shareholders collude with parent-SOE.

7. Conclusion

      This paper provides evidence that non-tradable shareholder of China’s listed firms use cash

dividend to tunnel negotiable shareholders.
     Non-tradable shareholder can only sell shares with private placement. This makes ‘home

made’ dividend very costly. Paying cash dividend is one of the ways for them to retrieve cash

from the firm. We find that both the probability of paying cash dividend and the payment level

are positively correlated to shareholding concentration. These finding are consistent with the

hypothesis that non-tradable shareholder use cash dividend to retrieve cash from the firm.

     With strict quato in IPO process, many China’s listed firms are seriously lack of working

capital. Paying cash dividend after rights offering is more feasible scheme. We find that firms

increase cash dividend payment after rights offering, especially those with high shareholding

concentration. We also find that non-tradable shareholders only subscribe less than 30% of the

shares allocated to them in rights offering. Through giving up subscription rights and using

receipts from rights offering as cash dividend, non-tradable shareholders sell a proportion of the

shares they hold to negotiable shareholders, at a price much higher than that in private

negotiation. In this way, cash dividend becomes a vehicle of tunneling negotiable shareholders.

Market reacts negatively to cash dividend announcement of firms with intermediate or high

shareholding concentration.

     Our findings show that among a variety form of tunneling, dividend might be one of means

of legal tunneling. This is opposite to the opinion that cash dividend can alleviate agency problem

between majority shareholder and minority shareholders.

Aharony, J., Lee C.J., and Wong T.J., 2000, Financial packaging of IPO firms in China, Journal of

    Accounting Research 38, 103-126.

Asquith, P. and Mullins D.W., 1983, The impact of initiating dividend payments on shareholder’

    wealth, Journal of Business 56, 77–96.

Bae, K., J. Kang, and J. Kim, 2002, Tunneling or value-added: evidence from mergers by Korean

    business groups, Journal of Finance 57(6): 2695-2740.

Bennedsen M. and Wolfenzon D., 2000, The balance of power in closely held corporations,

    Journal of Financial Economics 58, 113-139.

Chen, X., Lee C.J., and Li J., 2003, Chinese Tango: Government assisted earnings management,

    working paper, Tsinghua University.

Easterbrook, Frank, 1984. “Two Agency Cost Explanations of Dividends.” American Economic

    Review 74(4): 650-659.

Faccio, Mar, Larry Lang, and Leslie Young, 2001. “Dividends and Expropriation.” American

    Economic Review 91(1): 54-78.

Fama, E. F. and French K. R., 2001, Disappearing dividends: changing firm characteristics or

    lower propensity to pay? Journal of Financial Economics 60, 3-43.

Gomes, Amando, 2000. “Going Public with Asymmetric Information, Agency Costs, and

    Dynamic Trading.” Journal of Finance (April): 615-646.

Jian, M. and T.J. Wong, (2004). ‘Earnings Management and Related Party Transactions: Evidence

    from Chinese Corporate Groups’, Working paper, the Hong Kong University of Science and


Jensen, M., 1986, Agency costs of free cash flow, corporate finance, and takeovers, The American

    Economic Review 76, 323–329.

Johnson, S., La Porta R., Lopez-de-Silanes F., and Shleifer A., 2000, Tunneling, American

    Economic Review Papers and Proceedings 90, 22-27.

La-Porta, R., Lopez-De-Silanes F., Shleifer A., and Vishny R.W., 1998, Agency problems and
    dividend policies around the world, NBER working paper.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 1999. Corporate ownership around the world.

    Journal of Finance 54, 471-517.

La Porta R., Lopez-de-Silanes F., Shleifer A., and Vishny R., 2000, Investor protection and

    corporate governance, Journal of Financial Economics 58, 3-27.

Lee, C.J. and Y.H. Song., 2003, Rights issue and operating performance in China, working paper,

    Tsinghua University.

Lintner, J., 1956, Distribution of incomes of corporations among dividends, Retained earnings,

    and taxes, The American Economic Review 46, 97–113.

Miller, M. H. and Modigliani F., 1961, Dividend policy, growth, and the valuation of shares,

    Journal of Business 34, 411-433.
Table 1
Profitability, free cash flow, growth rate and cash dividend payments
This table shows mean earnings per share, free cash flow, and growth rate of sales for payers and
non-payers respectively. Payers and non-payers are determined on an after-tax basis. Free cash flow is
calculated as Cash from operations minus Cash paid for fixed assets, intangibles, and construction in
process. There was no statement of cash flow until 1998. Free cash flow after 1998 is calculated directly
using the corresponding numbers in the statement of cash flow. To calculate free cash flow before 1998,
cash from operations was replaced with Operating EBIT + ¡÷ accumulated depreciation. ¡÷ (fixed assets +
intangible assets + construction in process) was used for cash paid for fixed assets, intangible assets, and
construction in process. Growth of sales is calculated as (Salest / Salest-1) –1. F-test refers to the result of
ANOVA test between payers and non-payers. P-values are in brackets.

             Mean Earnings Per Share                  Mean Free Cash Flows                     Mean Growth of Sales
                       (Yuan)                              (Million Yuan)                               (%)
                         (1)                                     (2)                                     (3)
          Payers    Non-payers     F-test         Payers   Non-payers      F-test        Payers    Non-payers      F-test
Total                               21.73                                    1.89                                    0.05
           0.33        0.29                        -0.8        -1.43                       4.50        3.75
sample                             (0.0000)                                (0.1696)                                (0.8292)
                                     1.00                                    0.02                                    1.35
1996       0.33        0.30                       -4.12        -0.95                       7.00        3.01
                                   (0.3175)                                (0.8965)                                (0.2452)
                                    11.37                                    0.04                                  0.0516
1997       0.36        0.31                        4.19        0.69                       12.16       13.05
                                   (0.0008)                                (0.8441)                                (0.8203)
                                     3.07                                    1.03                                  0.2201
1998       0.33        0.30                       -22.81      -38.42                       0.39        2.32
                                   (0.0802)                                (0.3114)                                (0.6391)
                                    30.26                                    1.51                                  0.0804
1999       0.31        0.24                       13.12       -12.11                       1.29        -1.32
                                   (0.0000)                                (0.2190)                                (0.7768)
Table 3
Cash dividends payment and payout ratio
Percentages of total number of listed firms are shown in brackets of column (1) to (5). China tax law requires investors paying tax on stock dividends. The
listed firm is obligated to remit the dividend tax.    A major mechanic reason for paying cash dividends is to pay the tax associated with stock dividends.
Column (4) and (5) indicate respectively the number of total cash dividends payers and the number of payers for cash dividends beyond the tax for stock
dividends. Numbers in column (6), (7) and (8) are calculated on the after-tax basis. This table shows the composition of sample and cash dividend payment and
payout ratio from 1996 to 1999. Before 1996, the average frequency of cash dividend payment was 57.17%, but it was volatile. Payout ratio was lower in those
years. The average payout ratio was 39.03%.
               Total number        Number of          Number of        Number of        Number of       Mean payout       Mean dividend       Mean dividend
               of listed firms   Profitable firms     sample firms   payers (Before       payers            ratio        per share (Yuan)          yield
                     (1)               (2)                (3)             tax)          (After tax)                             (7)                 (%)
                                                                           (4)              (5)             (6)                                     (8)
Total                                 2785               2397             1131             658
                   3040                                                                                   60.17%               0.14                0.61
sample                              (91.61%)           (78.85%)          (47.18)         (27.45%)
                                       513                397             212              102
1996                545                                                                                   55.09%               0.12                0.22
                                    (94.13%)           (72.84%)          (53.40)         (25.69%)
                                       678                585             323              153
1997                733                                                                                   55.48%               0.15                0.32
                                    (92.50%)           (79.81%)          (55.21)         (26.15%)
                                       748                663             271              178
1998                834                                                                                   68.42%               0.16                0.37
                                    (89.69%)           (79.50%)          (40.87)         (26.85%)
                                       846                752             325              225
1999                928                                                                                   59.15%               0.13                1.16
                                    (91.16%)           (81.03%)          (43.22)         (29.92%)
    Table 2
    Subscription rate in rights offering

    This table shows the subscription rate in rights offering for various types of shareholders. Subscription rate
    is calculated with the number of shares subscribed in rights offering divided by the number of shares
    allocated to the specific group of shareholders. F-test refers to the result of ANOVA test between the given
    type of shareholders and the negotiable shareholders. P-values are in parentheses. The sample includes all
    firms that provide rights offering in each year of 1997-2000. Outliers are deleted.

          Negotiable           All non-tradable                                                  Top ten largest
                                                             The largest shareholder
         shareholders            shareholders                                                     shareholders
         subscription                 (2)                                                              (4)
             rate           Subscription                    Subscription                     Subscription
                                           F-Test                              F-Test                        F-Test
              (1)               rate                            rate                             rate
 Total                                       1643.58                           822.42                         1241.61
            100%                29%                              37%                              38%
sample                                       (0.0000)                         (0.0000)                        (0.0000)
                                              130.23                            47.71                          129.88
1997        100%                43%                              60%                              50%
                                             (0.0000)                         (0.0000)                        (0.0000)
                                              289.19                           160.94                          339.73
1998        100%                35%                              43%                              40%
                                             (0.0000)                         (0.0000)                        (0.0000)
                                              991.35                           479.52                          449.76
1999        100%                21%                              27%                              32%
                                             (0.0000)                         (0.0000)                        (0.0000)
                                              839.92                           409.51                          385.42
2000        100%                23%                              29%                              34%
                                             (0.0000)                         (0.0000)                        (0.0000)
Table 4
Preliminary explorations in the associations between cash dividends and profitability,
shareholding concentration and rights offering

A firm has high shareholding concentration if the differences between the shares hold by the first largest
shareholder and the second largest shareholder is greater than 70%. A firm has intermediate shareholding
concentration if the difference between the shares hold by the first largest shareholder and the second
largest shareholder is less than 70%, but greater than 15%. Payers have paid a greater than 0.01 Yuan per
share after-tax cash dividend. Chi-square refers to the result of test of frequency. P-values are in brackets.

                                                                                          Firms with rights
                               Profitability                    Shareholding               offering in the
                                     (1)                       Concentration                preceding year
                                                                      (2)                          (3)

                          Profitable        Loss
                            Firms           Firms            Low             High         Yes              No

     Payers                  785                6
                                                             118             673          287              504
 (% of the total           (25.82)         (0.20)
                                                            (3.88)          (22.14)      (9.44)          (16.58)
   Non-payers               2000               249
                                                             555             1694         695              1554
 (% of the total           (65.79)         (8.19)
                                                            (18.26)         (55.72)     (22.86)          (51.12)
   Chi-square                  80.9932***                       32.3377***                      7.7471***
    (P-value)                   (<0.0001)                        (<0.0001)                      (0.0054)

   Table 5
   To pay or not to pay: Economic determinants

   The Logit regression is based on 2397 samples over 1996-1999 that include all listed firms excepting loss
   firms and firms with B or H shares issued. The dependent variable, DIVP (dividend paying) equals one for
   firms paying cash dividends and zero otherwise. SHARDIF measures the ownership concentration of the
   parent SOE. It is calculated as the difference in shareholding between the two largest shareholders.
   CONCEN is a dummy variable that specify the sample firm into two groups according to shareholding
   concentration. CONCEN equals one for state dominant firms (firms with SHARDIF higher than 15%), and
   zero for independent firms. PREFINANCE is a dummy variable to indicate previous financing event, with
   value of one for firms had rights offering in the previous year and zero otherwise. FCF (free cash flow)
   measures agency cost. It is calculated as cash from operating activities minus cash paid in fixed assets,
   intangible assets, and construction-in-process. Total asset at the year-end is used as a deflator. ROA, the net
   income divided by total assets, measures the firm’s profitability. DIVPt-1 is the dividend paying status in the
   previous year, with one for dividend paying and zero otherwise. SIZE is the log of year-end asset.
   Chi-square values are reported in parentheses.

                                   (1)            (2)               (3)           (4)           (5)           (6)
INTERCEPT                      -1.4625***    -1.6670***        -7.1689***    -6.7750***    -7.0081***    -6.8027***
                               (255.8372)    (298.2622)        (110.3840)    (99.8958)     (105.2808)    (100.5422)
SHARDIF                                                                      0.7701***     0.7145***
                                                                             (13.0757)     (11.0942)
PREFINANCE                                                     0.4391***                   0.4184***     0.5332**
                                                               (20.1927)                   (18.1786)     (5.1095)
PREFINANCE×CONCEN                                                                                        0.4993**
FCF                            0.2037        0.2941            0.6049        0.4037        0.5250        0.4743
                               (0.2638)      (0.5031)          (2.1750)      (0.9588)      (1.6214)      (1.3187)
ROA                            0.8320***     1.0011***         0.3351***     0.4639*       0.3114        0.3913
                               (13.2685)     (18.0699)         (1.9535)      (3.8319)      (1.6729)      (2.6790)
DIVPt-1                        0.8293***     0.7975***         0.7872***     0.7393***     0.7920***     0.7711***
                               (67.3945)     (56.0763)         (56.6535)     (50.9815)     (57.0476)     (54.3711)
SIZE                                                           0.5006***     0.4533***     0.4627***     0.4575***
                                                               (68.9709)     (55.3298)     (57.2213)     (56.2352)
Chi-square                     82.1767***    296.1234***       159.4202***   153.6495***   168.1188***   158.7325***

Table 6
Economic determinants of payout ratio
The regression is based on 658 sample firms over 1996-1999 that has paid cash dividend on an after-tax
basis. The dependent variable, PAYOUTt , is the payout ratio in year t calculated on an after-tax basis
(Dividend per share / earnings per share). For definitions on explanatory variables, see Table 5. T-statistics
are reported in parentheses.

                                   (1)          (2)             (3)          (4)          (5)          (6)
  INTERCEPT                     1.6861      1.4467***       -5.2135***   -4.7350***   -4.7319***   -4.5670***
                                (14.91)     (12.99)         (3.53)       (-3.19)      (-3.19)      (-3.09)
  SHARDIF                                                   2.1363***    2.1526***
                                                            (4.60)       (4.64)
  PREFINANCE                                                             0.7741***    1.8147***    1.4512***
                                                                         (3.08)       (3.83)       (2.98)
  PREFINANCE×CONCEN                                                                   1.8702***    1.8409***
                                                                                      (3.20)       (3.16)
  PAYOUTt-1                     0.2075***   0.1998***       0.1909***    0.1914***    0.1912***    0.0494
                                (7.07)      (7.04)          (6.57)       (6.60)       (6.59)       (0.92)
  PAYOUTt-1×CONCEN                                                                                 0.1963***
  FCF                           0.4295      0.4361          0.1984       0.4449       0.4077       0.3787
                                (0.44)      (0.46)          (0.20)       (0.46)       (0.42)       (0.39)
  SIZE                                                      0.5412***    0.4832***    0.4955***    0.4914***
                                                            (4.09)       (3.62)       (3.73)       (3.71)
  Adj-R2                        0.0260      0.0892          0.0495       0.0539       0.0543       0.0588
  F                             25.33***    60.45***        24.68***     21.74***     21.90***     19.96***

    Table 7
    Economic determinants on dividends per share (DPS)
    The regression is based on 658 samples over 1996-1999 that has paid cash dividend on the after-tax basis.
    The dependent variable, DPSt , is the cash dividend per share in year t calculated on an after-tax basis.
    DPSt-1 is the cash dividend paid per share in year t-1 on an after-tax basis. EPSt refers to the earnings per
    share in year t. For definitions on explanatory variables, see Table 5. T-statistics are reported in

                               (1)          (2)          (3)          (4)           (5)          (6)          (7)
INTERCEPT                  -0.0222***   -0.0289***   -0.0296***   -0.0278***    -0.0284***   -0.0278***   -0.0281***
                           (-4.44)      (-5.63)      (-5.77)      (-5.42)       (-5.54)      (-5.43)      (-5.44)
SHARDIF                                                           0.0066***     0.0063***
                                                                  (4.11)        (3.92)
PREFINANCE                                           0.0019**                   0.0016**     0.0061***    0.0018**
                                                     (2.43)                     (2.10)       (3.45)       (2.30)
PREFINANCE×CONCEN                                                                            0.0012
FCF                        0.0028                    -0.0012      -0.0015       -0.0012      -0.0013      -0.0003
                           (1.19)                    (-0.52)      (-0.62)       (-0.50)      (-0.57)      (-0.13)
DPSt-1                                  0.2139***    0.2195***    0.2111***     0.2157***    0.2127***    0.2506***
                                        (9.23)       (9.43)       (9.13)        (9.29)       (9.15)       (5.25)
DPSt-1¡Á    CONCEN                                                                                        -0.0448
EPSt                                    0.5095***    0.4889***    0.4869***     0.4568***    0.4783***    0.2389*
                                        (5.59)       (5.12)       (5.15)        (4.78)       (5.01)       (1.83)
EPSt¡Á CONCEN                                                                                             0.3326***
SIZE                       0.0026***    0.0028***    0.0028***    0.0025***     0.0025***    0.0025***    0.0026***
                           (5.88)       (6.28)       (6.31)       (5.53)        (5.58)       (5.54)       (5.89)
Adj-R                      0.0124       0.0568       0.0584       0.0623        0.0635       0.0621       0.0605
F                          17.60***     53.93***     33.68***     36.01***      30.78***     30.08***     25.23***

 Table 8
 Computed selling price when using receipts from right offering as cash dividend
 This table reports shareholding dilution for non-tradable shareholders after giving up all or part of the
 subscription right in rights offering. The sample includes all firms paid cash dividend and provided rights
 offering in the same year during the period of 1997-2000. Computed selling price is calculated with ( (4) –
 cash paid for subscribing shares in rights offering by non-tradable shareholders) / ( (3) × total shares before
 the rights offering). Computed P/B is calculated with (5) / book value of equity before the rights offering.

             Mean              Mean             Mean           Mean cash
          non-tradable     non-tradable      decrease of       dividend         Mean
          shareholding     shareholding     non-tradable      received by     computed
          before rights     after rights    shareholding     non-tradable      selling       Mean        Market
            offering         offering           (%)          shareholders       price      computed      average
              (%)               (%)            (1)-(2)          (Million       (Yuan)        P/B           P/B
               (1)               (2)             (3)             Yuan)           (5)          (6)          (7)
              52.38            48.79             3.60            31.15           7.24         3.74         5.69
1997          56.52            53.00             3.52            41.42           7.85         2.88         4.58
1998          53.21            50.34             2.87            28.14           2.88         4.50         5.48
1999          51.39            47.26             4.12            33.20          12.58         4.71         7.82
2000          47.09            42.87             4.22            21.12           7.16         2.77         5.33

 Table 9
 Information content of cash dividend announcements
 This table shows the market reaction to unexpected dividend with the panel data from 1996 to 1999. Stock
 dividend, rights issue, and unexpected earnings are controlled. The dependent variable is CAR. CAR,t = ∑T
 AR,t, where AR is calculated with Naïve model. STKDIV is a dummy variable representing whether the
 firm plans to pay stock dividend. RI is a dummy variable that equals one for firms plan to provide rights
 issuing. UEXPDIV and UEXPEPS represent unexpected dividend per share and unexpected earnings per
 share respectively. Both of them are calculated based on random walk model. Panel A does not distinguish
 market reactions to cash dividend announcements for firms with different shareholder concentration. In
 Panel B, CONCEN is used to reflect the influence of shareholder concentration. CONCEN equals one if
 the difference in shareholding between the largest and second shareholder is greater than 15%, and 0
 otherwise. The sample includes all listed firms with data available. [-1, +1], [-3, +3], and [-5, +5] are
 windows around the cash dividend announcements. T-statistics are in parentheses.
Panel A:
Window               INTERCEPT       STKDIV            RI         UEXPDIV       UEXPEPS        Adj-R2       F
[-1, +1]             -0.0019        0.0352***     -0.0127         -0.0068       0.0752         0.0018   2.14*

                     (-0.28)        (2.50)        (-0.87)         (-0.27)       (0.88)
[-3, +3]             -0.0035        0.0382        -0.0103         -0.0104       0.0610         0.0024   2.49**

                     (-0.51)        (2.78)***     (-0.72)         (-0.42)       (0.74)
[-5, +5]             0.0023         0.0383***     -0.0185         -0.0154       0.0571         0.0029   2.79**

                     (0.33)         (2.78)        (-1.29)         (-0.62)       (0.69)

Panel B:
Window     INTERCEPT     STOCKDIV            RI        UEXPEPS       UEXPDIV        ×CONCEN       Adj-R2         F

[-1,+1]    -0.0028       0.0273        -0.0134         0.2399**      0.9601**      -0.9879*       0.0254   2.41**
           (-0.63)       (1.33)        (-1.48)         (1.98)        (2.01)        (-1.94)
[-3,+3]    0.0016        0.0592**      -0.0153         0.3099*       1.5185**      -1.6672**      0.0330   2.85**

           (0.25)        (1.99)        (-1.16)         (1.76)        (2.19)        (-2.26)
[-5,+5]    0.0042        0.0495        -0.0267*        0.3791*       1.2126        -1.5049*       0.0246   2.37**

           (0.54)        (1.39)        (-1.68)         (1.79)        (1.45)        (-1.69)

             Table 10
             Additional test: Suppose other non-tradable shareholders collude or monitor

             This table provide additional test by substituting SHARDIF in regression (1) to (3) with NONNEGO
             (Panel A) and FIRST (Panel B). NONNEGO is the total shareholding of all non-tradable shareholders.
             FIRST is the shareholding of the largest shareholder. HINONNEGO and FIRST are dummy variables.
             HINONNEGO equals 1 when NONNEGO is greater than the sample median, and 0 otherwise.
             HIFIRST equals 1 when FIRST is greater than the sample median. All other variables have the same
             meaning as they are in Table 5 to Table 7. Wald value or T value are in parenthesis.

             Panel A: Colluding non-tradable shareholders
                                          Pay or not pay                  Payout ratio                    Dividend per share
                                       (Regression (1))               (Regression (2))                     (Regression (3))

INTERCEPT                        -8.4942***       -8.3466***    -6.2422***      -6.1588***   -1.7864***     -1.7691***    -1.4794***
                                 (131.5652)       (127.9737)    (-4.12)         (-4.05)      (-7.82)        (-7.76)       (-6.32)
NONTRAD                          2.2587***        2.0808***     1.8017***       1.6664***    0.3080***      0.2917***
                                 (24.6341)        (20.5334)     (3.00)          (2.77)       (2.90)         (2.73)
HINONTRAD ×PREFINANCE                             0.5330***                     0.1993***                   0.1084**
                                                  (13.9172)                     (3.14)                      (2.11)
PREFINANCE                       0.4013***                      0.7743***                    0.0836**                     0.0939***
                                 (16.6258)                      (3.07)                       (2.34)                       (2.63)
FCF                              0.4383           0.4146        0.7625          0.7585       0.0005         -0.0018       -0.0094
                                 (1.1332)         (1.0148)      (0.79)          (0.78)       (0.00)         (-0.02)       (-0.09)
ROA                              0.3037           0.3157
                                 (1.5891)         (1.7192)
PAYOUTt-1                                                       0.1976***       0.1976***
                                                                (6.80)          (6.80)
DIVPt-1                          0.7799***        0.7791***                                  0.2104***      0.2101***     0.2108***
                                 (54.8864)        (54.6963)                                  (9.58)         (9.56)        (9.06)
DIVPt-1×HINONTRAD                                                                                                         -0.0318
EPSt                                                                                         0.3375***      0.3408***     0.2876***
                                                                                             (3.77)         (3.81)        (3.17)
EPSt×HINONTRAD                                                                                                            0.6280***
SIZE                             0.4859***        0.4846***     0.5822***       0.5826***    0.1491***      0.1488***     0.1399***
                                 (64.5759)        (64.2902)     (4.43)          (4.43)       (7.38)         (7.36)        (6.76)
Chi-square / F test              177.6769***      176.1733***   0.0474          0.0476       40.49***       40.30***      35.30***
Adj. R                                                          19.11***        19.21***     0.0825         0.0821        0.0835

            Panel B Monitoring non-tradable shareholders
                               Pay or not to pay                  Payout ratio                     Dividend per share
                                (Regression (1))                (Regression (2))                    (Regression (3))

INTERCEPT                 -7.1820***     -6.9873***    -5.3044***        -5.2211***   -0.0291***     -0.0286***        -0.0281***
                          (110.9744)     (106.2675)    (-3.59)           (-3.53)      (-5.69)        (-5.59)           (-5.44)
FIRST                     1.0625***      0.8271***     2.8290***         2.4896***    0.0099***      0.0093***
                          (15.5438)      (8.3903)      (4.80)            (4.18)       (4.68)         (4.17)
HIFIRST ×PREFINANCE                      0.5903***                       1.7531***                   0.0018
                                         (9.4746)                        (3.31)                      (1.17)
PREFINANCE                0.4152***                    0.7901***                      0.0016**                         0.0018**
                          (17.8685)                    (3.15)                         (2.08)                           (2.30)
FCF                       0.4991         0.4691        0.3845            0.3788       -0.0011        -0.0012           -0.0003
                          (1.4600)       (1.2858)      (0.40)            (0.39)       (-0.48)        (-0.51)           (-0.13)
ROA                       0.2960         0.3479
                          (1.5674)       (2.1030)
PAYOUTt-1                                              0.1906***         0.1904***
                                                       (6.57)            (6.57)
DIVPt-1                   0.7922***      0.7822***                                    0.2141***      0.2123***         0.2506***
                          (56.9593)      (55.6221)                                    (9.23)         (9.14)            (5.25)
DIVPt-1×HIFIRST                                                                                                        -0.0448
EPSt                                                                                  0.4412***      0.4551***         0.2389*
                                                                                      (4.61)         (4.76)            (1.83)
EPSt×HIFIRST                                                                                                           0.3326***
SIZE                      0.4600***      0.4563***     0.4927***         0.4990***    0.0024***      0.0024***         0.0026***
                          (56.9541)      (56.2721)     (3.71)            (3.77)       (5.31)         (5.28)            (5.98)
Chi-square / F test       171.8321***    165.7583***   0.0547            0.0552       0.0658         0.0648            0.0605
Adj. R                                                 22.06***          22.28***     31.95***       31.42***          25.23***

Panel A: Whole Sample




                                                                            Event date

                                                            unexpdiv<0                   unexpdiv>0

Panel B: Listed Firms with High and Intermediate Shareholding Concentration




                                                                            Event date

                                                                 UEXPDIV<0               UEXPDIV>0

Figure 2 Market Reaction to Cash Dividend Announcements The two figures in each
panel describe market reaction to cash dividend announcements of all the samples firms and firms with
high or intermediate shareholding concentration respectively. Market reaction is measured by cumulative
abnormal returns (CAR) in the [-20, +20] window around cash dividend announcements at t=0. CARt = ∑t
AR, where AR is calculated with Naïve model. Unexpected dividend is calculated with random walk
model on an after-tax basis. The sample firms are divided into two groups. One includes firms with
unexpected cash dividend greater than 0 (UEXPDIV>0). The other includes firms with unexpected cash
dividend less than 0 (UEXPDIV<0). The horizontal axis labels the event date. The vertical axis labels
cumulative abnormal returns. Because China’s listed firms announce cash dividend payment plans in the
annual reports, this figure shows the CAR around the annual reports announcements.


To top