External finance and dividend policy: a twist by financial constraints

Transcription

1 Accounting and Finance 56 (2016) External finance and dividend policy: a twist by financial constraints Zhong He a, Xiaoyan Chen b, Wei Huang c, Rulu Pan d, Jing Shi b,e a School of Finance, Jiangxi University of Finance and Economics, Nanchang, China b School of Economics, Finance and Marketing, RMIT University, Melbourne, VIC, Australia c School of Risk Management and Insurance, University of International Business and Economics, Beijing, China d UQ Business School, University of Queensland, Brisbane, QLD, Australia e International Institute for Financial Studies, Jiangxi University of Finance and Economics, Nanchang, China Abstract This study assesses distorting effect of financial constraints on the inverse relationship between internal and external finance by examining impact of an exogenous financing shock (i.e. a regulation released in China in 2008) on dividend policies in a quasi-natural experimental setting. Our result shows that in the absence of the regulation, the inverse relationship holds. However, the relation is twisted by the 2008 regulation. Compared with unconstrained firms, financially constrained firms are more willing to pay dividends and are more restrained to reduce cash dividends after the regulation, despite the fact that their external financing capacities are further constrained. Key words: External finance; Dividend policy; Financial constraint JEL classification: G32, G35 doi: /acfi Introduction While corporate internal funds and the use of external finance normally follow a negative relationship as suggested by theories, high levels of financial We would like to acknowledge the constructive comments and suggestions of Larry Li and Qiaoqiao Zhu. We have also benefited from and are grateful for comments by seminar participants at RMIT University, Jiangxi University of Finance and Economics, Australian National University, University of Queensland, and the 2014 Asian Finance Association Annual Meeting. Received 5 August 2016; accepted 16 September 2016 by Tom Smith (Editor).

2 936 Z. He et al./accounting and Finance 56 (2016) constraints may twist such a negative relationship. Almeida and Campello (2010) find that financially constrained firms do not necessarily reduce their demand for external finance following increases in internal funds. Furthermore, dividend increases by financial constrained firms are not rare in practice (Pathan et al., 2016) that is, financially constrained firms may not cut dividends to increase internal funds as a result of constrained availability of external finance. This study aims to confirm that it is the high levels of financial constraints that distort the relationship between internal funds and external finance by examining the changes in dividend policies of Chinese firms following an exogenous financing shock which strengthens financial constraints across all firms including already constrained firms. The pecking order hypothesis (Myers, 1984; Myers and Majluf, 1984) suggests an inverse relationship between internal funds and the use of external funds that is, a firm will reduce its usage of external finance following an increase in its internal funds (profitability). In addition, the residual dividend policy theory (Preinreich, 1932; Sage, 1937; Miller and Modigliani, 1961) states that a firm pays out excess cash (internal funds) as dividends only after all profitable investments are financed. Thus, financially constrained firms that is, firms with higher external financing costs and constrained access to external finance, are expected to adopt conservative dividend policies with the aim to maintain or increase internal funds to cover the shortage of external finance. Therefore, dividend increases by financially constrained firms appear to be a puzzling phenomenon within the framework of the established theories. We propose that firms may increase or initiate dividends (decrease internal funds) to enhance or prevent further deterioration of their external financing capacities when their use of external finance is highly constrained. Firms, particularly financially constrained firms, have incentives to use dividends to mitigate information asymmetry, establish reputation and thus reduce future financing costs. Firms can choose to pay higher dividends to develop good reputations and reduce future financing costs (La Porta et al., 2000). It is well documented in the literature that seasoned equity offering (SEO) announcements lead to negative market reactions due to the existence of information asymmetry (Korajczyk et al., 1991; Bayless and Chaplinsky, 1996). However, it is found that dividend announcements can serve to reduce potential information asymmetry (Howe and Lin, 1992; Gunasekarage and Power, 2002; Khang and King, 2006; Li and Zhao, 2008; Cheng et al., 2009). Prior studies (for example, Li et al., 2006; Pathan et al., 2016) try to test the link between financing capacity and dividend policies but are subject to the endogeneity problem as a result of the unclear causality between financing decisions and dividend policies that is, while financing capacity is dependent on dividend policies, dividend policies are also dependent on the sufficiency of finance (see also Chan et al., 2016). This study addresses the endogeneity problem by utilising a quasi-natural experimental setting formed by the release of the regulation Decisions on Amending Provisions on Cash Dividends by

3 Z. He et al./accounting and Finance 56 (2016) Listed Firms by the Chinese Securities Regulatory Commission (CSRC) in October Given its coerciveness, this regulation can be regarded as an exogenous shock to firms capacity to acquire external finance that is, it essentially imposes an extra layer of financial constraints on all the firms, that is no more equity finance for firms with unqualified dividend pay history. The regulation imposes a minimum requirement of cash dividend pay history (no <30 percent of the average realised annual distributable profits over the past 3 years) on firms seeking equity finance, which is the first time that the CSRC formally emphasises the importance of cash dividends over the other distribution types. The impact of the regulation is expected to be more pronounced for already financially constrained firms. We refer to Whited and Wu (2006) and measure the degree of financial constraints by constructing a synthetic Whited-Wu index (WW index, hereinafter), according to which we categorise our sample into financially constrained and unconstrained firms. We find that, in the absence of the 2008 regulation, constrained firms are less willing to pay cash dividends than unconstrained firms; when they do pay, they tend to pay lower dividends, which is consistent with the common belief. With the enforcement of the 2008 regulation, firms, no matter financially constrained or unconstrained, are forced to choose from two options: restraining dividend payments as a response to aggravated financial constraints or paying dividends to attain the permit for equity finance raising. While the former option is consistent with the implication of the established theories, the latter one supports our proposition that is, high levels of financial constraints cause firms to pay dividends to alleviate financial constraints and thus enhance their external financing capacities, which are financing eligibilities in this case. We find that, after the implementation of the 2008 regulation, firms are more likely to pay cash dividends to qualify for the external financing permit. What s more, financially constrained firms are even more motivated by the regulation to pay cash dividends as the increase in the propensity of paying dividends is much more significant for financially constrained firms. Furthermore, financially constrained firms are also more restrained to reduce dividends than their unconstrained peers following the regulation. As a robustness check, we also perform the test by employing three alternative measures of financial constraints in the literature (i.e. firm size, state ownership and sales growth) and our conclusion remains unchanged. In addition, we confirm empirically that the motivation of financially constrained firms to pay cash dividends is to enhance their external financing capacities. The basic intuition is that if firms are motivated by the 2008 regulation to pay cash dividends for the external financing permit, they will raise equity finance from the public once they are eligible for the permit. As a result, we should observe an increase in the probability of equity financing by eligible firms after the regulation. Financially constrained firms should be more

4 938 Z. He et al./accounting and Finance 56 (2016) likely to raise equity finance once becoming eligible as they are more motivated to pay cash dividends than unconstrained firms. Our results confirm that financially constrained firms have a higher propensity of raising equity finance among all the eligible firms in the post-regulation period. Our study contributes empirical evidence on the relationship between corporate external financing capacity and corporate dividend policy, particularly for financially constrained firms. While corporate dividend policies and the use of external finance normally follow a relationship as indicated by established theories such as the pecking order hypothesis that is, low external finance normally accompanied by low cash dividends, the deterioration of corporate external financing capacities tends to twist such a relationship and induces cash dividend payments, that is strategic dividends for the purpose of enhancing external financing capacity. Our findings greatly enrich the existing knowledge of the real effects of financial constraints and introduce deeper insights into the relationship between financial constraints and corporate behaviours. The rest of the article is organised as follows: Section 2 provides the background information and develops the hypotheses; Section 3 describes the data, the sample and the classification of subsamples; Section 4 presents our empirical results; and Section 5 concludes the article. 2. Background information and hypotheses development 2.1. The 2008 regulation in China While firms in developed economies generally have stable dividend polices, Chinese firms have fewer dividend initiations and lower dividend payments (Shao and Lin, 2004) as most of them are experiencing rapid growth with a focus on capital accumulation and expansion. Recognising the fact that the interests of minority shareholders are not properly protected, CSRC released a series of regulations over the past a few years with the aim to encourage cash dividend payments by listed firms. From October 2008, a firm that seeks equity finance is required to have a total cash dividends of more than 30 percent of the average annual realised distributable profits over the past 3 years, following the release of the Decisions on Amending Provisions on Cash Dividends by Listed Firms (the 2008 regulation) by CSRC. Although this is not the first regulation that includes dividend pay history as a part of the requirements for external financing eligibility, this is the first regulation that specifies the type of dividends (cash dividends) as well as the amount of dividends (30 percent of the average realised annual distributable profits over the past 3 years) that need to be paid. The regulation imposes a stringent financial restriction on all the firms no more equity finance if no satisfactory dividend pay history,

5 Z. He et al./accounting and Finance 56 (2016) which further deteriorates the external financing capacities of already financially constrained firms Hypothesis development In the absence of the 2008 regulation normal situation Modigliani and Miller (1958) argue that in a frictionless external capital market, firms can raise sufficient funds for all positive NPV (net present value) investment projects at a cost equal to that of internal capital. Therefore, external finance can fully cover the insufficiency of internal funds investments and growth do not depend on the availability of internal funds. However, once capital market imperfections are introduced, internal funds and external finance are no longer perfect substitutes. For example, the information asymmetries between investors and management increase the cost of external capital relative to internal funds (Myers and Majluf, 1984), and thus, internal funds are preferred over external finance, which is the main argument supporting the pecking order hypothesis. The wedge between the costs of internal and external funds represents the extent to which a firm is financially constrained. Therefore, the importance of cheap and readily available internal funds increases with the increase of firm financial constraints. In the literature of financial constraints, dividend payout ratio is commonly used as an a priori criterion to identify costly external finance (see, for example, Fazzari et al., 1988; Kaplan and Zingales, 1997; Almeida et al., 2004; Almeida and Campello, 2010; Denis and Sibilkov, 2010; Pathan et al., 2016). Specifically, Whited and Wu (2006) treat dividend as an important determinant of the WW financial constraints index. The general conclusion is that financially constrained firms tend to pay lower cash dividends and also have a weaker desire to pay. External finance is more expensive for financially constrained firms so they tend to retain most of the low-cost internal funds for investments. If financially constrained firms increase dividends, in addition to being exposed to more financial distress risk due to diminished cash balances, 1 they could also experience underinvestment in the product market and become less competitive with tighter finance. 2 For these reasons, financially constrained firms should avoid increasing dividends which leads to negative financial market reactions and deteriorates firm operating performance in both short term and long term (Chen and Wang, 2012; Pathan et al., 2016). Therefore, in the absence of the 2008 regulation, we propose the following hypotheses: 1 See Opler et al. (1999), Kaplan and Zingales (2000), Lamont et al. (2001), Almeida et al. (2004), Whited and Wu (2006), Denis and Sibilkov (2010) for a detailed discussion. 2 See Chevalier and Scharfstein (1994); Campello (2003); Almeida et al. (2004).

6 940 Z. He et al./accounting and Finance 56 (2016) H1a: Before the enforcement of the 2008 regulation, financially constrained firms have a lower propensity to pay cash dividends than unconstrained firms. H1b: Before the enforcement of the 2008 regulation, financially constrained firms pay lower cash dividends than unconstrained firms A twist by the 2008 regulation The 2008 regulation represents an exogenous shock to corporate external financing capacities that is, no more equity finance for Chinese firms without a satisfactory cash dividend pay history. The deterioration of financing capacities may cause firms to adjust their cash dividends: downwards if they intend to reserve internal funds to cover for insufficient external funds; upwards if they use cash dividends to enhance their corporate financing capacities, that is to be eligible for equity finance in terms of cash dividend pay history. While it is established in the literature that reduced availability of external finance causes firms to increase internal funds, high levels of financial constraints may lead to different firm behaviours. First of all, the deterioration of external financing capacities may not depress constrained firms demand for external funds because constrained firms have higher opportunity costs of investments and tend to reserve internal funds for future investments (Almeida and Campello, 2010). Instead, with deteriorating external financing capacities, firms particularly financial constrained firms have the incentive to enhance their external financing capacities through paying dividends. SEO announcements usually lead to negative market reactions due to the existence of information asymmetry (Korajczyk et al., 1991; Bayless and Chaplinsky, 1996). The dividend literature provides compelling evidence that relative to non-payers, information asymmetry is a less serious problem for dividend payers (Howe and Lin, 1992; Khang and King, 2006; Li and Zhao, 2008). Therefore, the market reacts less negatively to SEO announcements made by dividend payers than those made by non-payers (Booth and Chang, 2011). Moreover, Deshmukh (2005) shows that issue costs are adversely associated with dividend yield. In addition, in countries with weak legal protection of minority shareholders, firms pay higher dividends to develop a reputation that investors favour to gain easier and cheaper access to external funds (La Porta et al., 2000). As the protection of minority shareholders is relatively weak in China (Chen and Xu, 2001), Chinese firms are more motivated to use dividends to establish a good reputation and thereby reduce future financing costs, especially when firms are financially constrained. Using a sample of listed Chinese firms, He et al. (2016) examines why firms pay stock dividends and find that the Chinese firms use stock dividend to attract analysts attentions and to keep the firm s stock price within its acceptable trading range. Wei and Jiang (2001) conduct a survey on

7 Z. He et al./accounting and Finance 56 (2016) dividend policies of Chinese listed firms and find that the current and future finance needs are considerations behind dividend decisions. Li et al. (2006) report that financing capacity is the most important determinant of dividend policies, particularly for non-soes, which are normally considered more financially constrained than SOEs. Therefore, we hypothesise that: H2a: After the enforcement of the 2008 regulation, financially constrained firms are more motivated (by the regulation) to pay cash dividends, compared to unconstrained firms. H2b: After the enforcement of the 2008 regulation, financially constrained firms are more reluctant to reduce cash dividend (due to strengthened financial constraints by the regulation), compared to unconstrained firms. The 2008 regulation imposes another layer of financial constraint (restriction) on all the firms that is, the accumulated cash dividends paid by a firm over the past 3 years must be no less than 30 percent of the average realised annual distributable profits for the firm to raise equity from the public. If firms, particularly financially constrained firms, pay dividends to meet this prerequisite for equity issuance, we should observe a post-regulation increase in their propensities to raise equity finance once they are qualified and the increase should be more pronounced for qualified financially constrained firms. Therefore, we hypothesise that: H3: In the post-regulation period, qualified financially constrained firms are more motivated to raise equity finance from the public than qualified unconstrained firms. 3. Data, sample and subsamples 3.1. Data and sample Our sample contains 8,758 firm-year observations from 941 non-financial firms listed on the Shanghai and Shenzhen Stock Exchanges from 2003 to Data are sourced from the China Stock Market and Accounting Research (CSMAR) database. The original sample covers all Chinese non-financial firms listed on the Shanghai and Shenzhen Stock Exchanges during the period from 2003 to We choose this sample period because a consistent and unified set of accounting standards were not adopted in China until 2000 and the controlling shareholder data are unavailable before The following data filters apply to our sample: first, delete firm-year observations with missing values and firm-year observation with non-positive values for total assets, capital stock or sales; second, exclude firms with reported short-term debt greater than reported total debt or with reported changes in capital stock more

8 942 Z. He et al./accounting and Finance 56 (2016) than the sum of reported sales of capital goods and reported depreciation following Whited and Wu (2006); third, remove firms that are delisted, under Special Treatment (ST) or Particular Transfer (PT) to avoid firms already in financial distress; 3 fourth, drop dividend observations with a nonzero beforetax cash dividend but a zero after-tax dividend, following the approach in Lee and Xiao (2004) and Huang et al. (2011). 4 Appendix summarises all the variables. Two dividend variables are used in this study: CDPayer is an indicator that equals 1 if a firm pays a cash dividend in the current financial year and 0 otherwise; Payout is cash dividend payout ratio defined as total cash dividends divided by EBIT (earnings before interest and tax). To capture the effect of the 2008 regulation, we define a dummy variable, Regulate, which equals 1 if it is after the release of the 2008 dividend regulation and 0 otherwise. Concerning certain extreme values, Payout is winsorised at the top and bottom one percent and SG (sales growth) at the top one percent of their respective distributions Classification of financially constrained and unconstrained firms There are two common approaches to estimate the extent to which a firm is financially constrained: the Q-theory model and the Euler equation, both of which build upon the dynamic optimisation model of investment. We employ the approach of Whited and Wu (2006) an Euler equation approach, and develop a Chinese version of the Whited-Wu (WW) index to measure the level of financial constraint. 5 In the Whited and Wu (2006) model, financial constraint is interpreted as the shadow cost of raising new equity and measures how costly external equity finance is relative to internal finance. Following the approach of Whited and Wu (2006), we re-parameterise the shadow cost of equity finance using Chinese data as a function of observable firm characteristics. Unlike Whited 3 China has an investor warning system. If a firm has a negative net income for two consecutive years, the CSRC gives a ST to the shares of the company. The daily limit of price change for ST firms is 5 percent. If a firm has consecutive losses for 3 years, it is further classified as an ST* firm and faces delisting if it continues to lose money. If a firm has negative net income for three consecutive years, the trading of its shares is suspended. PT is a trading service offered for suspended stocks on every Friday. Cash dividend payments are restricted for firms with ST, ST* and PT status. 4 All dividends, including stock dividends, are taxed at 10 percent from 2005 in China. As listed firms need to withhold the dividend tax, many firms pay cash dividends simply to cover the tax payable induced by stock dividends. 5 The Q-model is criticized for its significant measurement error and identification problems (Kaplan and Zingales, 1997; Erickson and Whited, 2000; Alti, 2003) and is more data-demanding. In addition, the market valuation of capital, a proxy for the marginal Q, is difficult to estimate in many developing countries due to the imperfections of financial markets (Huang et al., 2016).

9 Z. He et al./accounting and Finance 56 (2016) and Wu (2006), we do not include a dividend-paying indicator in the function because this study explicitly studies dividend policy. Instead, we include a state ownership indicator, considering that SOEs in China have an inherent advantage in receiving external finance (Qian, 1994; Kornai, 1998; Tian, 2005; Jia et al., 2013; Chen, 2015; Haveman et al., 2016). For instance, Chen (2015) also confirms privilege of state-controlled enterprises to access equity financing. Following Whited and Wu (2006), 6 we apply GMM and estimate the nonlinear Euler equation in first differences to eliminate potential firm fixed effects. Our instruments include all the Euler equation variables and also inventories, depreciation, current assets, current liabilities and tax payments. All of the instruments are normalised by total assets. Unlike Whited and Wu (2006), we do not include either the ratio of dividends to total assets (because dividends are explicitly studied here) or the average ratio of cash flows to assets over the past three periods (given its high correlation with other variables) to measure profitability. Instead, we measure profitability using a dummy that equals 1 if the ratio of cash flows to assets in the previous year is positive. All of the instrumental variables are lagged two periods in the GMM estimation as the Euler equation is estimated in first differences, Table 1 reports the estimation results of the GMM estimation of the Euler equation. Column (1) contains the estimates from the most general specification of the model, with all the nine financial health variables included. Column (2) contains the estimation results of the model if we drop the constant (because it has the smallest t-statistic). All coefficients of the explanatory variables, except for LTD and NA, 7 have the expected signs. For example, the positive coefficient of SOE suggests that state-owned listed firms are less likely to be financially constrained, which is consistent with the expectation that stateowned listed firms possess greater access to external financing (Qian, 1994; Tian, 2005). The negative sign on the LogTA indicates that small firms are subject to greater financial constraints, which essentially captures the welldocumented size effect (see, for instance, Beck et al. 2005). The combination of the negative coefficient of SG and the positive coefficient of ISG is as expected as firms in high-growth industries but with low individual sales growth are eager to invest more so more likely to be constrained (Whited and Wu, 2006). The negative coefficients of Cash and CF are also consistent with the prediction that is, firms with high levels of cash holdings and sufficient cash flows are financially healthier and less likely to be financially constrained. Finally, the extent of financial constraints is greater for firms in low-debt capacity industries, confirmed by the negative coefficient of IDAR. Therefore, the 6 Equation (10) of Whited and Wu (2006). 7 This is not unexpected, as Chinese listed firms have a strong preference for equity finance (Huang and Zhang, 2001) and that the industry of security analysts is still at its early stage of development (Zhang and Lv, 2007).

10 944 Z. He et al./accounting and Finance 56 (2016) Table 1 Euler equation estimates Parameter Expected sign (1) (2) LTD ( ) ( ) SOE ( ) ( ) SG ( ) ( ) LogTA ( ) *** ( ) ISG (0.0002) (0.6016) Cash ( ) ( ) CF ( ) ( ) NA (0.0002) (0.2458) IDAR ( ) ( ) Constant (0.0000) l *** ( ) *** ( ) a ( ) ( ) a * (1.7302) * (1.8608) a (1.4603) (1.5188) l *** (6.9429) *** (8.1986) MKT ( ) ( ) SMB (0.1705) (0.1707) HML (1.1048) (1.1580) Observations 5,272 5,272 J statistic p-value The table reports the estimation results of an Euler equation based on a sample of all nonfinancial listed firms in the Chinese stock market during the period of The Euler equation is given by equation (10) of Whited and Wu (2006). The nonlinear GMM estimation is carried out with the model in first differences with twice lagged instruments. a i is the investment adjustment cost parameter, and l is a markup. LTD is the ratio of the long-term debt to total assets; SOE is an indicator variable that equals to one if the firm s ultimate controlling shareholder is a state-owned enterprise and zero otherwise; SG is the firm s annual sales growth; LogTA is the natural log of total assets; ISG is the firm s industry sales growth; Cash is the ratio of cash, marketable securities and short-term investments to total assets; CF is the ratio of cash flows to total assets; NA is the number of analysts following the firm; and IDAR is the firm s industry debt-to-assets ratio. MKT, SMB and HML are the Fama French factors on market, size and book to market. T-statistics are reported in parentheses. The value of the J statistic and the p-value of the J-test on the model specification are reported in the last two rows. ***denotes significance at 1% level, **denotes significance at 5% level, *denotes significance at 10% level following formula for the WW financial constraints index is constructed based on the estimates in Column (2): k it ¼ 0:1477SOE it 0:0654SG it 0:0508LogTA it þ 0:018ISG it 0:1287Cash it 0:0076CF it 0:3651IDAR it : ð1þ

11 Z. He et al./accounting and Finance 56 (2016) Where SOE is an indicator variable that equals 1 if a firm s ultimate controlling shareholder is a state-owned enterprise and 0 otherwise; SG is annual sales growth; LogTA is the natural log of total assets; ISG is industry sales growth; Cash is the ratio of cash, marketable securities and short-term investments to total assets; CF is the ratio of cash flows to total assets; NA is the number of analysts following a firm; and IDAR is industry debt-to-assets ratio. The fitted value of k it is the WW financial constraints index for firm i at time t. A higher value of k it suggests a higher degree of financial constraints. We sort all the firms in our sample into ten deciles based on the values of their WW indices. Firms with the lowest WW index values are placed in Decile One, and firms with the highest WW values are placed in Decile Ten. Following the approach in Chen et al. (2007) and Chen and Wang (2012), we assign firms to groups based on their WW deciles of the financial year before the release of the 2008 regulation. Firms in the top 3 WW index deciles are classified as being more financially constrained, while firms in the remaining deciles are classified as being less financially constrained (or more financially unconstrained). 8 Table 2 presents descriptive statistics of the full sample and the financially constrained and unconstrained subsamples. More than half of the firms in the sample pay a cash dividend during the sample period, and the average payout ratio is approximately 17 percent. Regulate has a mean value of 0.531, which suggests that our sample is balanced with respect to the number of observations before and after the regulation. In addition, financially constrained and unconstrained firms display significant differences in firm characteristics: first, financially constrained firms are less likely to pay cash dividends and pay less than unconstrained firms (0.535 vs , vs , respectively), which is consistent with the common belief in the literature; second, state-owned firms are less likely to be financially constrained, which is not surprising considering that the privilege of state-owned enterprises to access external finance in China; third, financially constrained firms typically operate in high sales growth industries but have low individual sales growth; fourth, financially constrained firms are in industries with low debt capacity; finally, financially constrained firms are usually smaller, have less cash flow and analyst coverage. 4. Empirical results 4.1. Univariate analyses Table 3 shows the changes in firm dividend policies after the 2008 regulation through a univariate analysis, with firms categorised into the financially 8 Our sorting scheme is analogous to that in Lamont et al. (2001), except that we use the WW index instead of the KZ index. The conclusion remains unchanged if we define financially constrained firms as those in the top 20 percent (as in Chen and Wang (2012)) or 40 percent (as in Whited and Wu (2006)) of the WW index.

12 946 Z. He et al./accounting and Finance 56 (2016) Table 2 Summary statistics Full sample Financially constrained Financially unconstrained Variable N Mean SD Min Max N Mean SD Min Max N Mean SD Min Max CDPayer 8, , , Payout 8, , , Regulate 8, , , State 8, , , SOE 8, , , LTD 8, , , SG 8, , , LogTA 8, , , ISG 8, , , Cash 8, , , CF 8, , , NA 8, , , IDAR 8, , , Leverage 8, , , Volatility 8, , , The table reports summary statistics of the key variables. All the variables are defined in Appendix. The full sample consists of 8,758 firm-year observations during the period of , representing 941 non-financial listed firms in the Chinese stock market. For every sample year, we sort all the sample firms into deciles based on the values of their WW indices, where Decile 1 represents the lowest values and Decile 10 the highest. We then assign firms to groups based on their WW deciles of the financial year before the announcement of the regulation (i.e. 2008). Firms in the top 3 WW index deciles are classified as financially constrained; firms in other deciles as financially unconstrained.

13 Z. He et al./accounting and Finance 56 (2016) Table 3 Changes in dividend policy after the 2008 regulation Financially constrained Financially unconstrained Constrained-unconstrained Variable Before After Before After Difference in change Panel A: The propensity to pay cash dividends CDPayer 0.469*** (32.00) 0.589*** (44.63) 0.685*** (80.01) 0.719*** (91.42) Change (After Before) 0.120*** (6.08) 0.034*** (2.85) 0.086*** (3.89) Observations 1,159 1,390 2,940 3,263 Panel B: The cash dividend payout ratio Payout 0.168*** (24.12) 0.143*** (28.58) 0.200*** (51.89) 0.152*** (51.74) Change (After Before) 0.025*** ( 3.00) 0.045*** ( 9.11) 0.020** (2.17) Observations 1,159 1,389 2,940 3,252 This table reports the changes in firm dividend policies after the 2008 regulation for financially constrained and unconstrained firms, respectively. Panel A shows the results on the propensity to pay cash dividends and panel B on the payout ratio. We assign sample firms to groups based on their WW deciles of the financial year before the 2008 regulation. Financially constrained firms are firms in the top 3 deciles of the WW index. Financially unconstrained firms are firms in other deciles. CDPayer is an indicator variable that equals one if the firm pays a cash dividend in the current financial year. Payout is defined as cash dividends divided by EBIT. The changes in CDPayer and Payout are calculated on a basis of after minus before, and the difference in changes is calculated on a basis of constrained minus unconstrained. T-statistics are reported in parentheses. ***, ** and * denote significance at 1, 5 and 10 percent levels, respectively.

14 948 Z. He et al./accounting and Finance 56 (2016) constrained and the unconstrained groups. Panel A reports the propensities of paying cash dividends by financially constrained and unconstrained firms, respectively. We find that firms, no matter financially constrained or unconstrained, are more likely to pay cash dividends after the 2008 regulation, with the increase in the propensity of paying cash dividends significantly more pronounced for financially constrained firms. This finding implies that the imposed financial constraints (by the regulation) motivate firms particularly already financially constrained firms to pay cash dividends in order to enhance their external financing capacities, which supports the hypothesis H2a by demonstrating that firms with higher external financing costs are more eager to obtain financing eligibility through paying cash dividends. Panel B reports the changes in cash dividend payout ratios of financially constrained and unconstrained firms, respectively. Despite the fact that firms are more likely to pay dividends, they reduce their cash dividend payout ratios after the regulation, which confirms that firms are inclined to increase internal funds to cover the insufficiency of external funds a negative relationship between internal funds and the use of external finance as suggested by the literature. However, financially constrained firms reduce their cash dividend payout ratios by a significantly smaller amount compared to unconstrained firms, implying financially constrained firms are restrained to decrease cash dividends considering the stronger financial constraints after the regulation, which supports the hypothesis H2b Multivariate cross-sectional regression analyses Propensity of paying cash dividends: before and after the 2008 regulation To further investigate the separate and joint effects of the 2008 regulation and firm financial constraints on firms willingness to pay dividends, we apply a difference-in-difference approach and estimate a Logit model on firms propensities to pay cash dividends. Columns 1 3 in Table 4 present the results. The dependent variable is CDPayer, which equals 1 if the firm pays a cash dividend in the current financial year, and 0 otherwise. A financial constraint indicator Constraint is defined as 1 if the WW index of a firm is in the top 3 deciles before the 2008 regulation and zero otherwise. The coefficient of Regulate is positive no matter which specification of the model is employed, implying that firms are motivated by the 2008 regulation to pay dividends. The negative coefficient of Constraint, which does not change with different model specifications, indicates that financially constrained firms are less likely to pay cash dividends relative to unconstrained firms in the absence of the 2008 regulation, which support the hypothesis H1a. To examine the joint effects of the 2008 regulation and the existing financial constraints, we include an interaction term of Regulate and Constraint (Regulate*Constraint). The significantly positive coefficient of the interaction

15 Z. He et al./accounting and Finance 56 (2016) Table 4 Effect of financial constraints and a financing shock: propensity to pay cash dividends Independent variable All firms Most financially constrained firms Most financially unconstrained firms (1) (2) (3) (4) (5) (6) (7) (8) (9) Regulate 0.162*** (2.92) Constraint 0.901*** ( 12.69) Regulate* Constraint 0.321*** (3.29) 0.260*** (3.73) 0.781*** ( 8.01) 0.331*** (2.65) Leverage 1.803*** ( 10.57) Volatility 9.954*** ( 8.94) Lag1CDPayer 2.170*** (38.17) Constant 0.779*** (19.60) 0.579*** (5.06) 0.335*** (4.12) 0.347** (2.40) 4.173*** ( 10.25) 8.851*** ( 5.07) 0.257*** (3.64) 0.483*** (6.02) 0.123** ( 2.08) 0.589*** (5.74) 1.658*** ( 5.49) *** ( 6.34) 2.009*** (20.11) 0.719*** (5.77) 5.301*** ( 7.28) *** ( 3.56) (0.31) (0.46) (0.74) 1.927*** ( 5.78) 8.109*** ( 3.54) Firm fixed effect NO NO YES NO NO YES NO NO YES Observations 8,752 7,762 5,041 2,549 2,247 1,580 2,642 2,345 1,338 Likelihood , Ratio (v 2 ) Prob > v ( 0.70) 1.218*** (18.06) 2.247*** (19.40) 0.929*** (4.14) (1.08) 4.667*** ( 5.84) ( 1.25) 0.436*** (3.16) The table reports Logit estimation results on firms propensity to pay a cash dividend. We assign the sample firms to groups based on their WW deciles of the financial year before the 2008 regulation. Firms are classified as the most financially constrained if they are in the top 3 WW index deciles. The most unconstrained firms are firms in the lowest 3 deciles. The dependent variable is CDPayer, which is an indicator variable that equals one if the firm pays a cash dividend in the current financial year. Regulate is a dummy variable that takes a value of one if the current financial year is after the 2008 dividend regulation and zero otherwise; Constraint is an indicator variable that takes a value of one if the firm s WW index is in the top 3 deciles and zero otherwise (Constraint is not included on its own in the presence of the firm fixed effects, as it is firmspecifically measured based on the WW decile for the financial year before the 2008 regulation); Regulate*Constraint is the interaction between Regulate and Constraint. See Appendix for definitions of all other variables. Z-statistics are reported in parentheses. ***, ** and * denote significance at 1, 5 and 10 percent levels, respectively.

16 950 Z. He et al./accounting and Finance 56 (2016) term Regulate*Constraint indicates that financially constrained firms are more motivated by the regulation to pay cash dividends that is, a larger postregulation increase in their propensities of paying cash dividends than unconstrained firms, which further supports Hypothesis H2a. In addition to Regulate, Constraint and Regulate*Constraint, we incorporate a number of control variables suggested in the literature, for example leverage (Myers, 1984), volatility (Chay and Suh, 2009) and lagged dividend status (Fama and French, 2001). Our conclusion remains unchanged with these controls taken into consideration. In addition, the coefficients of all the controls have expected signs. For example, leverage is negatively related to propensities to pay cash dividends; the positive coefficient of lagged dividend status confirms that dividends are sticky (Lintner, 1956; Myers, 1984) Cash dividend payout ratio: before and after the 2008 regulation Similarly, to examine the separate and joint effects of the 2008 regulation and firm financial constraints on the size of cash dividends, we employ a similar difference-in-difference framework and estimate a multivariate Tobit model on cash dividend payout ratios. Columns 1 3 of Table 5 report the estimation results. Regulate is significantly negative, implying that the 2008 regulation causes firms to reduce the size of cash dividends. Similarly, the coefficient on Constraint is significantly negative, confirming that financially constrained firms pay lower cash dividends than unconstrained firms in the absence of the 2008 regulation consistent with the hypothesis H1b. However, the significantly positive coefficient of the interaction term Regulate*Constraint, which essentially offsets the significantly negative coefficient of Regulate, supports the hypothesis H2b by showing that already financially constrained firms barely decrease their dividend payments while the regulation, that is an extra layer of financial constraints on every firm, causes the other firms to pay lower cash dividends. With the extra financial constraints (the regulation), financially constrained firms are reluctant to decrease cash dividends to increase internal funds because they are too financially constrained to risk any further deterioration of their external corporate financing capacities as a result of cutting dividends. To control for the effect of other potential explanatory variables, we regress Payout against the same set of control variables, as displayed in Column (3). The effects of Regulate, Constraint, and Regulate*Constraint still persist. Leverage and Volatility have negative impacts on cash dividend payout ratios, respectively, while lagged dividend status (Lag1CDPayer) has a positive impact, 9 consistent with expectation. 9 The lagged dividend payout ratio is not included as a control because using lagged payout ratios as an explanatory variable is akin to regressing the dependent variable on itself (Fama and French, 2001).

17 Z. He et al./accounting and Finance 56 (2016) Table 5 Effect of financial constraints and a financing shock: dividend payout ratio Independent variable All firms Most financially constrained firms Most financially unconstrained firms (1) (2) (3) (4) (5) (6) (7) Regulate 0.032*** ( 5.17) 0.043*** ( 5.94) 0.020*** ( 2.94) (0.47) (1.17) 0.061*** ( 6.84) 0.041*** ( 4.82) Constraint 0.058*** ( 8.33) 0.080*** ( 7.77) 0.057*** ( 5.46) Regulate* Constraint 0.041*** (2.94) 0.026** (1.99) Leverage 0.415*** ( 24.66) 0.460*** ( 10.92) 0.392*** ( 16.67) Volatility 1.153*** ( 9.59) 1.879*** ( 6.74) 0.768*** ( 4.10) Lag1CDPayer 0.215*** (32.72) 0.257*** (17.06) 0.174*** (16.15) Constant 0.138*** (27.88) 0.144*** (27.00) 0.215*** (17.91) 0.027** (2.25) 0.138*** (5.31) 0.187*** (28.96) 0.261*** (14.92) Observations 8,740 8,740 7,750 2,548 2,246 2,637 2,340 Likelihood ratio (v 2 ) , Prob > v The table reports Tobit estimation results on firms cash dividend payout ratios. We assign the sample firms to groups based on their WW deciles of the financial year before the 2008 regulation. Firms are classified as the most financially constrained if they are in the top 3 WW index deciles. The most unconstrained firms are firms in the lowest 3 deciles. The dependent variable is Payout, defined as cash dividends divided by EBIT. Regulate is a dummy variable that takes a value of one if the current financial year is after the 2008 dividend regulation and zero otherwise; Constraint is an indicator variable that takes a value of one if the firm s WW index is in the top 3 deciles and zero otherwise (Constraint is firmspecifically measured based on the WW decile for the financial year before the 2008 regulation); Regulate*Constraint is the interaction between Regulate and Constraint. See Appendix for definitions of all other variables. T-statistics are reported in parentheses. ***, ** and * denote significance at 1, 5 and 10 percent levels, respectively. We do not include the firm fixed effects in the models as there does not exist a sufficient statistic allowing the fixed effects to be conditioned out of the likelihood and the unconditional fixed-effects estimates are biased (Honore (1992)).

18 952 Z. He et al./accounting and Finance 56 (2016) Subsample analyses To explicitly examine the impact of financial constraints on firms responses to an exogenous financing shock (the 2008 regulation), we analyse the subsamples of the most financially constrained group and the most financially unconstrained group, respectively. We categorise financially constrained and unconstrained firms based on their WW Decile of the financial year before the 2008 regulation. Firms are classified as most financially constrained if they are in the top 3 WW index deciles. The most financially unconstrained firms are those in the lowest 3 deciles. Columns (4)-(9) in Table 4 report the Logit estimation results on the propensities of paying dividends for the more financial constrained and less financially constrained subsamples, respectively. For firms with more financial constraints, the coefficients on Regulate are significantly positive. Yet, for less financially constrained, it is insignificant. This finding suggests that the 2008 regulation (an extra layer of financial constraints) significantly motivates financially constrained firms to pay cash dividends while firms that are most unconstrained are not as motivated further confirming the hypothesis H2a. With the regulation further dampens their external financing capacity, financially constrained firms are motivated to pay dividends to remove the adverse influences of the regulation. In contrast, unconstrained firms are not as motivated. Columns (4)-(7) of Table 5 report the Tobit estimation results on cash dividend payout ratios for the two subsamples. For the most constrained group, the coefficient on Regulate is insignificant. In contrast, for the most unconstrained group, the coefficient on Regulate is significantly negative. The finding implies that unlike firms in the most financially unconstrained group who reduce their cash dividends significantly (likely for the purpose of saving more cash to meet investment needs) when the regulation dampens corporate external financing capacities, financially constrained firms are reluctant to reduce their cash dividends consistent with the hypothesis H2b and indicating that they are now too constrained to risk any further deterioration of external finance capacity The causality between paying dividends and acquiring finance The 2008 regulation imposes a minimum requirement of cash dividend pay history on firms seeking equity finance. Based on this requirement, an indicator variable Qualified is defined, which is 1 if a firm meets the minimum dividend requirement for a specific year, that is accumulated cash dividends distributed over the past 3 years more than 30 percent of average realised annual distributable profits, and 0 otherwise. In addition, an equity finance indicator, Finance, is defined as 1 if a firm conducts a SEO or a rights issue in the current financial year, and 0 otherwise.