Paying for Financial Flexibility: A Natural Experiment in China

Transcription

1 Paying for Financial Flexibility: A Natural Experiment in China Zhiqiang Wang Weiting Zhang School of Management, Xiamen University ; Development Research Center, Shanghai Stock Exchange wtzhang@sse.com.cn [Abstract] Prior literature suggests that firms with higher marginal value of financial flexibility should have lower payout. This study uses the introduction of a recent CSRC (China Securities Regulatory Commission) regulation on payout levels to examine whether firms with higher marginal value of financial flexibility tend to pay more dividends to meet the dividend requirement. Consistent with the notion that financial flexibility plays an important role in financial decisions, we find that firms tend to cater the regulation for qualification of equity financing when the marginal value of financial flexibility is high. Under this institutional background in China, firms have to pay dividends for extra financial flexibility, which comes from the qualification of equity financing. Our analysis suggests that financial policies of Chinese listed firms are essentially influenced by regulations, and financial flexibility is the key factor to decide how firms cater for regulations. [Key words] Quasi-mandatory Payout Regulation; Catering Strategy; Financial Flexibility; Equity Financing

2 Ⅰ. Introduction Graham and Harvey (2001) identify financial flexibility as an important factor for financial decisions. This survey has spawned a literature on the mechanism in which financial flexibility can influence financial decisions, such as dividend policy. Gamba and Triantis (2008) use a calibrated model to simulate financial decisions and estimate the value of financial flexibility. They find that firms with higher marginal value of financial flexibility will be more inclined to reserve financial flexibility or more reluctant to pay dividends if financial flexibility indeed increases firm value. Lie (2005) argues that dividend increase can convey information that firms have sufficient financial flexibility, such as high cash holdings or low debt ratios. Jagannathan et al. (2000) find that firms with higher permanent operating cash flows tend to choose to pay dividends, while firms with higher temporary non-operating cash flows are more likely to repurchase shares. They conclude that financial flexibility can account for different choices in financial decisions. Using listed firms from 23 countries over the periods, Rapp et al. (2012) find that firms tend to reduce or even omit payouts if financial flexibility is valuable. Several studies report that Chinese listed firms are reluctant to pay dividends, especially in cash. To ensure dividends payments for financing in open market, the CSRC, the regulator of capital markets in China, has issued a series of regulations since Firms that plan to seek external financing in the future have to fulfill the requirement on dividend payouts. Since these regulations do not have any impact on those will not consider external financing, they are also known as quasi-mandatory. The CSRC does not set a specific level of payout until CSRC set accurate minimum payout level for the first time, and requirement is cumulative dividends (including cash dividends and stock dividends) paid during three years before public financing

3 should not be less than 20% of average net income over the same three-year period. In 2008, CSRC modified this regulation by increasing the respective payout level to 30%, and excludes the option for stock dividends as well. These Quasi-mandatory Payout Regulations have connected the payout level with the qualification for external financing, which means that if firms would like to get higher financial flexibility, they have to sacrifice some financial flexibility at first. For firms who have higher marginal value of financial flexibility, they will be more likely to pay dividends for the minimum requirements in order to get the possibility of qualifying for external financing. The CSRC issued the Quasi-mandatory Payout Regulations in year 2006 and 2008, imposing a minimum level on dividend payouts for listed firms that wish to obtain refinancing in the future. This paper uses the introduction of this regulation as exogenous event to examine the influential mechanism of financial flexibility on financial decisions. We firstly try to figure out how the probability for catering is influenced by financial flexibility by using Logit regression, and then we detect the efficiency of the regulation by comparing performance of firms who finally get equity financing. We find consistent evidence that firms with higher marginal value of financial flexibility have higher probability to cater for regulations, which means paying more dividends at present to get opportunities of external financing in the future. However, these regulations seem not so effective when we pay attention to future performance for those firms who actually get equity financing. Regulations supposed to improve the quality of equity financing firms get the result of worse future performance overall. We attribute the Regulation Paradox here to that these minimum requirements for cash dividends set by regulations limit chances for firms with higher growth but little cash.

4 We contribute to the literature in two ways. First, this paper provides new evidence for the relationship between financial flexibility and dividends policy. Previous studies suggest that higher level of financial flexibility, which is generally interpreted as lower marginal value of financial flexibility, will result in lower payout level. However, under specific institutional background in China where the payout level is explicitly linked to financing qualifications, financially inflexible firms tend to have stronger preference for meeting the requirement thus they become more likely to pay cash dividends. We provide direct evidence that financial flexibility indeed influences financial decisions making. Second, prior studies report that firms pay dividends for many reasons, including sending signals of insider information about manager s view about prospects, alleviating agency problems associated with free cash flow, historical reasons and so on. Signaling theory argues that insiders choose payout levels to convey private information about firm s future prospects to investors. In this way, it will be easier for firms that paid dividends to be successful in external financing. For agency concerns, Easterbrook (1984) contends that firms can overcome agency problems by paying dividends. If paying dividends help alleviate agency problems, external financing will also be easier for them. However, these studies do not directly examine the relationship between financing decisions and payout behavior. We complement the literature by providing evidence on the relationship between financing decisions and dividend policy. The remainder of this proceeds as follows. Section 2 describes institutional background and develops empirical hypotheses to be tested. Section 3 introduces research design, including data, variables of interest, and model specification. Section 4 presents main empirical results. Section 5 concludes.

5 II. Institutional background and hypothesis development According to provisions of regulations, only those firms who would like to finance externally will be influenced by the regulations, so we should identify these firms in the first place. However, the decision process of firm is a black box, so we can hardly observe it and define it directly. Why don t we just use firms who have succeeded in refinancing? Because the approval of public financing in China are operated with Approval System, which means CSRC would review applications from firms and decide whether the application is successful or not. Not all applications will be approved by CSRC even if they meet all requirements. It makes actual financing behavior different from firms potential purpose of financing. Considering that the dividend policy of Chinese listed firms is rather flexible, it is more probable for firms who plan to refinance to meet the minimum requirements of regulations. So we can use the minimum payout level as a benchmark to define financing purpose indirectly. When the payout level of firms reach the minimum requirement for financing (including positive earnings in all of the three years before financing), then they are considered as having the purpose of external financing and they comprise our sample 1. The regulations were published in 2006 and 2008, respectively. If the regulations indeed influence the decision making process of management in payout level, then dividend behavior of firms should be abnormal in these two specific years. Because the requirement of payout level is the cumulative level over the three-year period, so we can judge whether the dividend behavior in 1 To avoid the possible bias caused by our definition of refinancing purpose, we also draw the pattern of payout behavior for those firms who really get external financing in the year after the regulation, which are 2007 and 2009 respectively. The results show that these firms share the same pattern with our sample firms with financing purpose, and the catering behavior is even worse. (see Appendix Ⅰ)

6 the year when the regulation was released is abnormal or not by comparing to that of past two years in the same three-year period respectively. We calculate the ratio of dividends paid this last year (2006 or 2008) to the cumulative level of three-year period (that is, 2004~2006 or 2006~2008) respectively. If it shows the same pattern in the regulation year compared to that of the past two years, then the consistency of dividend behavior will make the ratio not more than 33.33% approximately. If we assume a 10% increase in dividend policy, than the ratio can not be more than 36.50%. The patterns of the ratio are shown as follows: The Ratio of Dividends Paid in 2006 to Those of Three-Year Period 30% 20% 10% 0% % 80% 60% 40% 20% 0% Interval Probability Cumulative Probability Figure 1 Ratios of Catering Firms in 2006 The Ratio of Dividends Paid in 2008 to Those of Three-Year Period 20% 15% 10% 5% 0% % 80% 60% 40% 20% 0% Interval Probability Cumulative Probability Figure 2 Ratios of Catering Firms in 2008 Note: The number in the left side of the figure shows the probability of firms getting into respective interval of ratios (the critical value is included in the interval of larger ratio). The number in the right shows the cumulative

7 probability of firms in ratios. In Figure 1, we can see that in 2006, firms with a ratio higher than 40% account for approximately 40% of the whole sample, and there are 20% firms who have paid more than 60% cumulative dividends in the last year. For the period of 2008, we expect the abnormal payout behavior to be less severe because it already includes the year of 2006, which means they should already cater in 2006, thereby increasing their accumulated payout level. Specifically, nearly 40% of firms paid more than 40% cumulative amount in the last year, and firms with the ratio higher than 90% have reached 10% of the whole sample. That can be explained by the change in provisions of 2008 s regulation. In developed capital markets, dividend smoothing and consistency are typical characteristics of dividend policy for listed firms. However, Wei (2000) indicates that the management of listed firms in China has seldom set long-term goals for dividend payments, so dividend policy rarely follows the same pattern in time series. Te dividend policy seems to adapted to external environment to a larger extent. The operations of Chinese capital market are always directed by regulations, so in most occasions, this kind of adaption represents as the response to regulations. From those figures above, we can clearly see that in the years when regulation were published, most firms paid larger amounts of dividends compared to those of former period, showing that the dividend behavior of Chinese listed firms is a response to these regulations to some extent. It can be taken as preliminary evidence that the dividend behavior of firms with purpose of external financing is indeed influenced by the regulation. This behavior is similar to the concept of catering in Baker and Wugler (2004), with the difference being that Chinese listed firms cater to regulations rather than investors. In this paper, the concept of catering, and catering strategy is

8 referred to as the strategy that firms with financing purpose pay dividends to meet the minimum requirement of Quasi-mandatory Payout Regulation. When their payout level meets the minimum requirements, then we treat them as catering, otherwise, they will be regarded as non-catering. Here come our questions: How do firms decide whether or not to take the catering strategy? If firms really cater, then will these regulations be effective in protecting investors and increasing the value of society as a whole? III. Research Design 1. Data The primary data source CSMAR, RESSET and Wind Database. We call the years 2007 and year 2009 as event years, year 2006 and year 2008 as decision years, and the three years leading up to event years as the observation period, respectively. We start with all firms listed in A shares of Chinese stock market. We exclude those firms with titles such as ST or PT 2, firms that operate in the finance industry, and firms with incomplete or abnormal data 3. As a result, we get our sample. In 2006, there are 780 firms as a sample in total, which is 799 for 2008.All data comes from CSMAR, RESSET and Wind Database. 2. Variables (1) Marginal Value of Financial Flexibility Gamba and Triantis (2008) have proved that the value of financial flexibility is decided by the cost of external financing, the cost of cash holdings, potential growth opportunities and 2 The title ST or PT means firms do not have positive earning through consecutive two or more years. These firms always have abnormal financial behavior styles. 3 When firms indeed get qualifications of equity financing, but their respective payout ratio does not reach the level that quasi-mandatory regulations of paying dividend require. We treat them as abnormal data. This kind of data does not reach 5% of the whole sample.

9 reversibility of capital. Clark (2010) further indicates that the marginal value of cash depends on the same factors that influence the value of financial flexibility and the marginal value of financial flexibility as well. In addition, the direction of the marginal value of cash influenced by the factors are the same as those for the marginal value of financial flexibility, that is, the increasing function of potential growth opportunities and the cost of external financing, the decreasing function of the cost of holding cash and the reversibility of capital. Hence, Clark (2010) considers the marginal value of cash as an efficient proxy for the marginal value of financial flexibility. The marginal value of cash measures the sensitivity of shareholder value change to change in cash holdings. When we regress the market value of equity on the change in cash holdings, the coefficient should be the marginal value of cash, representing the increased value by increasing one unit of cash holding. In this paper, we use this method to estimate the marginal value of cash. The dependent variable of our regression model is the cumulative abnormal return (CAR). Following Clark (2010), we estimate monthly CAR by using the Fama-French three factor model and 12 monthly CAR yields a yearly CAR. The estimation model specification is referred to Faulkender and Wang (2006): C C C C E NA R r r r r L r r i, t i, t 1 i, t i, t i, t i, t i, t i, t 4 5 Mi, t 1 Mi, t 1 Mi, t 1 Mi, t 1 Mi, t 1 Mi, t 1 I D C NF r r r r L r (1) i, t i, t i, t 1 i, t i, t 10 i, t Mi, t 1 Mi, t 1 Mi, t 1 Mi, t 1 Where, X it, is the change in variable X i from time t-1 to time t for firm i; C is cash and marketable securities; E is earnings defined as earnings before interest and taxes but before depreciation and amortization; NA is total assets net of cash; I is interest expense; D is the total common dividend expense; L is market leverage; NF is net financing defined as the sum of

10 increase in long-term debt and equity issues; and M is the market value of equity. According to the definition of the marginal value of cash, the coefficients of variables interacted with C it, M it, 1 composite the marginal value of cash, so it will be expressed as follows: C MVOC r r r L ^ ^ ^ it, 1 i, t i, t Mit, 1 As shown clearly in figure 1 and figure 2, whether firms decide to take the catering strategy all depends on the data of the year before the event year, which was also the year when firms were supposed to engage in equity financing. So when we seek to discern firms equity financing behavior in 2007 and 2009 respectively, the year for making decisions should be 2006 and We use data of year 2006 and 2008 to estimate equation (1) respectively. The following table summarizes main results: Table 1 Major Empirical Result for Model (1) Intercept ** C it, M it, (2.23) *** (3.94) *** (5.59) *** (5.00) C M L C M i, t 1 i, t it, i, t 1 i, t 1 C it, M it, *** (-3.02) * (-1.85) (-0.46) *** (-3.33) Observations Adjusted R % 47.53% Note: Figures in the brackets are T Statistics; *** represents the significance at the level of 1% (two tail); ** represents the significance at the level of 5% (two tail); * represents the significance at the level of 10% (two tail). These results allow us to calculate the marginal value of cash in 2006:

11 MVOC C L i,2005 i,2006 i,2006 Mi,2005 We notice that, in 2008, the coefficient of C M it, 1 it, 1 was not significant, indicating that the marginal value of cash did not reflect the influence of cash holdings, so we just use ^ ^ r1 r3 L it, to calculate the marginal value of cash that year, that is: MVOC L i,2008 i,2008 (2) Control Variables In reviewing the literature, we discover that the debt capacity is always assumed to be finite (DeAngelo, DeAngelo and Whited, 2010; Clark, 2010). Under this assumption, the marginal value of financial flexibility will be indirectly influenced by the firm s spare debt capacity because of the component decided by leverage. However, in China, the leverage always shows large differences across industries and companies. For example, Mura et. al (2007) has defined firms with spare debt capacity as financially flexible. That confirms our thought to put the spare debt capacity into the model control for its possible influence. It is calculated as Max(0, industrial average leverage-current leverage of firm), and the leverage is book value based. Trade credit is also another way of financing and it reflects additional financing ability gained from bargaining power. We measure it by constructing the ratio of net financing amount through business credit to sales, where the net financing amount through business credit is all terms payable less all terms receivable. Other control variables include ownership concentration (the proportion of shares held by top 10 shareholders) and firm size (natural logarithm of total assets). There are two other special control variables within the special institutional environment in

12 China. The first is the budget constraints in state-owned enterprises (SOEs) 4. Given that Chinese four largest banks are owned by the central government, SOEs always enjoy more intangible convenience in debt financing compared to private firms. Lin and Tan(1999) has proved that specific situation in China, that is, SOEs often face soft budget constraints than private companies, what makes it easier for SOEs to get loans externally. We therefore use a dummy variable as a control. The second factor is how firms act when facing potential policy risk. Quan et.al (2010) points out that in the background of Quasi-mandatory Payout Regulation, it is more difficult for firms to make a decision about dividend payment behavior, so they will just mimic other firms in the same industry in order to avoid policy risk. For this reason, we control for the industrial payout level. In addition to the payout level for an industry, we also control for firm s growth 5. We define growth as average growth rate in sales during the observation period. In keeping with prior literature, we control other factors which may influence performance, such as size, prior profitability, and ownership concentration. Variables in this paper are shown as follows: Table 2 Definitions of Variables Variable Type Variable Label Definitions Dependent Variable Catering strategy Cater When the payout level of the firm reaches the minimum requirement of payout level, then CATER=1, or else CATER=0. 4 When the terminal controller of firms belong to local governments, or State-owned Assets Supervision and Administration Commission of the State Council (SASAC), both in the central government level or local government level, then firms will be regarded as SOE. 5 We also consider using M/B ratio to measure the growth opportunity. However, M/B ratio not only measures the growth opportunities but also valuation of the firm from market. Because financial flexibility can increase the firm value, so they are highly correlated theoretically. We have done Pearson correlation test as a result (see Appendix Ⅱ), and the result shows that in both sample periods, the correlation coefficient between M/B and the marginal value of financial flexibility are higher than 0.5, and significantly correlated. So we tend not to use M/B ratio to measure growth opportunities here.

13 SD If firms have paid stock dividends during observation period, then SD=1, or else SD=0. SEO When firms get equity financing, including Equity Financing seasoned equity offering and rights offering in the event year, SEO=1; or else SEO=0 ROE_ST ROE of the year after catering year Future Profitability ROE_LT The average ROE of the year after catering year and the next year Financial Flexibility MVOC The marginal value of financial flexibility Major Independent Variable Control Variables Spare Debt Capacity SDC Max (0,industrial average leverage-current leverage of firm) The Growth rate of Profits g The growth rate of net profits during observation period Business Credit BCF (Amount payable-amount Financing receivable)/sales State-owned SOE When ultimate controller is state-owned, Ownership SOE=1; or else SOE=0 Ownership OWNCON Total shares held by top 10 Concentration 10 shareholders/total shares in decision year Industrial Payout INDPAY Industrial average of payout level Level Historical ROE_lag ROE of the catering year Profitability Industrial Return on INDROE Industrial average of ROE Equity Size of Total Assets SIZE Natural logarithm of total assets in decision year 3. Model Specification First, we are concerned about factors that firms will consider in the payout decisions. To examine our financial flexibility hypothesis, we use the following logistic regression model to compare the explanatory power of each potential determinant: Cater MVOC SDC g BCF OWNCON 10 SOE INDPAY SIZE The regulation released in 2006 includes stock dividend as one option for firms to reach the minimum requirements, so firms that belong to this sample period will face the choice between

14 stock dividends and cash dividends when they would like to cater. Compared to cash dividends, stock dividends can keep cash in the company, maintaining a higher level of financial flexibility. Thus we expect that firms with higher marginal value of financial flexibility are more likely to pay stock dividends to satisfy the requirement. We use firms who follow a catering strategy and estimate the following logistic regression model: SD MVOC SDC g BCF OWNCON 10 SOE SIZE A firm s decision to cater can to some extent reflect the firm s willingness to finance. However, whether or not the firm will be able to refinance also depends on whether it can be certificated by the CSRC. As seen in our theoretical analysis, when the probability of equity financing is higher for firms with higher marginal value of financial flexibility, the Approval System is efficient. When firms who get public financing have better performance, the allocation of capital can be rational, and the benefits of investors can be protected. According to theoretical analysis before, firms with higher marginal value of financial flexibility are motivated to cater, and the increase of financial flexibility can increase firm value as a result (Gamba and Triantis, 2008), so the total increase in firm value for the whole society will be larger when CSRC makes it more likely for firms with higher marginal value of financial flexibility to get refinancing qualifications. In addition to social value, we also concern about the ex post performance. Only when the marginal value of financial flexibility is positively related with firm s performance, it will be more efficient for CSRC if it gives higher probability of approving qualifications to those firms with higher marginal value of financial flexibility, considering both the social value and investor protection. So we have to figure out at first whether the marginal value of financial flexibility has significantly positive relationship

15 with firm s performance. We therefore continue to use firms taking the catering strategy as the sample, and construct a logistic regression model as follows: SEO MVOC SDC g BCF OWNCON 10 SOE SIZE Finally, we test the efficiency of CSRC in Approval System of equity financing qualifications further and the rationality of dividend paying measures. We try to find out the future profitability of catering firms and those who really get qualifications of equity financing. We use those ex post tests to judge the efficiency of this regulation and the efficiency CSRC plays in the Approval System. It is worth noting that Quasi-mandatory Payout Regulations just set the minimum dividend paying ratio, which means the regulations use this minimum level rather than an accurate dividend paying ratio to verify firms, so we should use the dummy variable of catering strategy rather than accurate dividend paying ratio to identify the rationality of dividend paying measures. Another concern comes from the sample we will use here. In model (2), we focus on the influential factors in the decision process of catering strategy. However, we use the dummy variable of catering strategy here in order to see whether there is any difference in performance between firms who pay inadequate dividends to reach the minimum level and firms who have reached it, to further prove whether this dividend paying measure can distinguish a good company from a bad company. We should thus exclude firms who did not pay during the observation period, selecting only firms with a dividend paying ratio larger than 0 as a sample, and construct a multiple linear regression as follows: ROE _ ST Cater MVOC SEO g ROE _ Lag OWNCON 10 INDROE SIZE ROE _ LT Cater MVOC SEO g ROE _ Lag OWNCON 10 INDROE SIZE

16 IV. Empirical Results 1. Descriptive Statistics and Preliminary Findings Table 3 represents the descriptive statistics for sample firms in different sample periods. We can clearly see from the table that in 2008, the mean and maximum of the marginal value of financial flexibility were larger, and so is the standard deviation, indicating a larger difference compared to those of At the same time, firms in 2008 had almost exhausted their debt capacity with comparatively higher leverage, but the amount of financing through business credit does not show large difference. The growth for firms during the period of 2004 ~ 2006 has smaller mean and standard deviations than those in the period of 2006~2008, showing that firms have a higher growth rate and larger difference when the event year is Measure 1 is larger and more volatile than measure 2 on average, suggesting that when dividend paying measures consist only of cash dividends paid, it is more difficult for firms to cater to the policy. The ROE of two periods are nearly the same. Table 3 Descriptive Statistics Variable Sample Period N Mean Median Max Min SD Measure ~ Measure ~ MVOC SDC BCF g 2004~ ~ ROE_ST ROE_LT 2007~ ~

17 2. Catering Strategy and Financial Flexibility If the qualification of financing in the open market can be considered as a way of increasing financial flexibility, then firms can get higher financial flexibility and as a result get the value of it. The decision of catering hence depends on whether firms should get this increase in financial flexibility or not. Meanwhile, how strong the catering motive is also depends upon the value of financial flexibility. The greater the value, the stronger catering motives are for those firms with financing purpose. The question then becomes how to decide the value of this financing option. According to the definition in Gamba and Triantis (2008), financial flexibility represents the ability of a firm to access and restructure its financing at a low cost. For the acquaintance of financing qualifications means that the firm can have chance to get public financing, it can be treated as one kind of resources of financial flexibility. Getting this financing qualification can thus increase firm s level of financial flexibility. Its value should represent the firm value increased by increasing the level of financial flexibility. The marginal value of financial flexibility is defined as the value increased for every unit increase of financial flexibility level, so the value of financing option should depends on the marginal value of financial flexibility as a result. What should be pointed out here is, with the increase of financial flexibility, respective costs (including the absence of tax-shield benefits and agency costs caused by large amount of cash holding) will gradually increase, which makes the marginal value of financial flexibility decrease as the level of financial flexibility increases. In prior literature, the level of financial flexibility was decided both by cash holdings and leverage (Graham and Harvey, 2001; Lemmon and Zender, 2008; Clark, 2010). But under Chinese

18 institutional background, whether or not firms satisfy the requirement of equity financing becomes one of factors that also decide the level of financial flexibility. For Quasi-mandatory Payout Regulation, the qualification requirements of external financing contain the requirement of minimum payout level, which means dividend behavior can indirectly decide how financially flexible the firm is. When firm has larger amount of cash holdings, lower leverage or satisfying minimum requirements, its level of financial flexibility is higher; while lower cash holdings, higher leverage or dissatisfaction of minimum requirements leads to lower levels of financial flexibility. If the level of financial flexibility is finite, these three factors may substitute for each other. That is, the firm will decide whether or not take the catering strategy to get extra financial flexibility based on the marginal value of financial flexibility, which reflects the present level of cash holdings, leverage and payout behavior. All in all, we predict that the firm with higher marginal value of financial flexibility will have stronger motives to get the financing option, and at the same time exhibit a stronger preference for catering strategy. We also notice that the measure of payout level in 2006 counts the amount of stock dividends, so firms within this sample period will face the choice between stock dividends and cash dividends when they would like to cater. Compared to cash dividends, stock dividends can reserve cash in the company, maintaining the level of financial flexibility. We would expect that firms with higher marginal value of financial flexibility would be more likely to pay stock dividends to satisfy the respective requirement. If this hypothesis can be supported by empirical evidence, then we can ensure our judgment of financial flexibility. Table 4 Probability of Taking Catering Strategy

19 Sample Period Model Model (3) Model (4) Model (3) Dependent Variable Logit(Cater) Logit(SD) Logit (Cater) Intercept *** (53.70) *** (7.17) *** (57.87) MVOC *** (22.25) SDC * (2.79) g (0.25) BCF * (2.63) SOE (0.39) OwnCon *** (8.23) ** (4.01) (0.17) (0.09) (0.04) ** (5.14) * (3.56) *** (20.43) (0.92) *** (8.98) (0.30) * (2.92) *** (9.06) SIZE *** (53.03) ** (5.65) *** (47.76) INDPAY (0.45) *** (9.90) Observations Likelihood Ratio Pr>ChiSq < < Note: Figures in the brackets are Wald Chi-Square Statistics; *** represents the significance at the level of 1% ( two tail); ** represents the significance at the level of 5% ( two tail); * represents the significance at the level of 10% ( two tail). In table 4, we can see that, when the decision year is 2006, the coefficient of MVOC is significantly positive, which suggests that firms with higher marginal value of financial flexibility are more likely to take the catering strategy. The spare debt capacity is negatively correlated with the likelihood of catering, indicating that firms which have exhausted most of their debt capacity are more willing to take the catering strategy. This is still consistent with our former expectation. When firms have almost exhausted their debt capacity, they will depend more on equity financing, giving them no other choice than to cater. But we have to realize that the correlation coefficient of

20 MVOC is still larger, and more significant, representing more explaining power in the behavior of catering. It can be explained as firms still take MOVC as their first priority to consider. In addition, as we know that the cost of business credit financing is lower than debt financing and equity financing, the significantly positive correlation coefficient can be regarded as evidence to support the argument that firms always tend to increase their level of financial flexibility from another perspective. When it comes to 2008, firms decision of catering strategy just shows significantly positive relationship with MVOC, while its relationships with debt capacity and business credit financing are no longer significant. However, the growth rate shows a significantly negative influence on the probability of catering; that is primarily because of the change of standard in payout ratio. In 2008, the regulation sought to put stricter requirements for paying dividend by excluding stock dividends as one option for firms with purpose of equity financing to reach the minimum level. But as we have seen from table 4, those firms with higher growth rates are more constrained by this change in types of dividend paying as they are always cash strapped. This makes it more difficult for them to reach the minimum level within the context of 2008 s regulation, which further constrains their equity financing. In this situation, 2008 s regulation is harder to be effective in social resource allocation; for it constrains the financing opportunities for those firms with higher growth but less cash, and it does not accord well the original purpose of the regulation, hence we see here a Regulation Paradox. There are also two more interesting findings about SOEs and the industrial mimic tendency. SOEs are more willing to take the catering strategy when limited to cash dividends only. When

21 regulations change, the industrial effect, represented by the average level of dividend paying cross industries, begins to gain power in explaining catering strategy. This confirms our surmise about the responsive mode for firms in the environment of policy change while policy directs the developments of the Chinese capital market. They will mimic other firms in the same industry to protect themselves from uncertainty risk in changes of policy. In both periods, we have seen significantly positive correlation coefficients in the control variables, such as ownership concentration and size. They can be explained intuitively. The larger the firm size, the more resources firms control, making it easier for them to satisfy the minimum requirements. When ownership is more concentrated, they can response to policy changes more quickly. After discussing what decides firm s choice of catering in payout level, we have to focus on what kind of dividends firms choose to cater. The empirical results have shown that financial flexibility is still the first priority when firms choose stock dividends to cater. The higher the marginal value of financial flexibility, the larger the value of keeping financial flexibility, and the stronger the motives to pay stock dividends rather than cash dividends when firms have already decided to cater. SOEs show less willingness to cater through paying stock dividends; the same is seen in firms with higher ownership concentration. We have seen that firms with higher marginal value of financial flexibility are more likely to cater; the question that remains is whether the probability of equity financing is also higher for firms with a higher marginal value of financial flexibility. In developed capital market, the level of marketalization is higher, which means that firms financing behavior is largely reflect their financing purpose. However, under the background of

22 Approval System in Chinese capital market, financing purpose is different from financing behavior in some occasions. Research on the mechanism of financing behavior in Chinese capital market should be more appropriate by looking financing purpose rather than actual financing behavior. So we do not discuss financing theories a lot in this paper, we research on financing purpose by detecting catering strategy instead. While researching on financing purpose, what can be interesting is the variable of spare debt capacity. When there is spare debt capacity in the firm, that is, the leverage of firm is lower than industrial level. Pecking order theory predicts that the entire deficit should be financed with debt, and if we assume roughly that firms take industrial level as their target leverage, than trade off theory predicts in the same way too. Those two theories both predict that firms should increase their leverage when they have spare debt capacity. However, our results show that spare debt capacity can explain financing purpose indeed, but financial flexibility can explain more. When the marginal value of financial flexibility is higher, firms would still prefer to equity financing. It makes firms financing preference changes with the level of financial flexibility when their spare debt capacity is same. Table 5 Probability of Actual Equity Financing Event Year Model Model (4) Depend Variable Logit(SEO) Intercept (2.29) ** (4.21) MVOC (1.04) (1.73) SDC ** (5.41) g * (2.66) *** (7.83) (0.19)

23 BCF * (2.56) (0.33) SOE ** (4.14) (1.56) OwnCon (2.41) (2.11) SIZE (0.78) * (3.10) Observations Likelihood Ratio Pr>ChiSq Note: Figures in the brackets are Wald Chi-Square Statistics; *** represents the significance at the level of 1% ( two tail); ** represents the significance at the level of 5% ( two tail); * represents the significance at the level of 10% ( two tail). Table 5 shows some different results from prior ones. In both periods, firms actual behavior of equity financing has no significant relationship with MVOC, providing some evidence in inefficiency of approval process directed by the CSRC. The coefficient of the dummy variable SOE is significantly negative. Reviewing prior empirical results, we have found that SOEs are more willing to pay cash dividends but less willing to actually get equity financing even when they have reached the minimum requirements of regulations, suggesting that SOEs do indeed face looser constraints in debt financing compared to other firms. The correlations showing different directions in two sample periods exist in the variable growth and spare debt capacity. In 2007, the growth showed significantly positive influence on the probability of catering, while it turned to be negative in This confirms our prior suggestion of a Regulation Paradox. Another variable showing opposite correlations is spare debt capacity. In 2007, firms with less remaining debt capacity are more likely to finance in equity, which means that firms who have actual equity financing have relatively lower leverage. In 2009, however, the

24 situation was reversed. How did these differences arise? We should have further tests about future performance. 3. Catering Strategy, Equity Financing and Firm Performance After the analysis of decision mechanism, we concern more about the economic consequences of catering strategy. That is how these catering strategies of listed firms influence their performance and further the efficiency of entire market. According to our theoretical analysis, when there is no Approval System in equity financing process, which means that firms with stronger motives to finance are more likely to succeed in getting external financing. Then those firms who are taking the catering strategy will have a higher marginal value of financial flexibility, and thus prefer to equity financing more (Clark, 2010). Subsequently, the value addition caused by the increase in the level of financial flexibility through equity financing will be larger. Through this process, firms with higher value can get more funds in financing. We can say that resource optimization can be reached when firms decide their payout strategy for their own sake. But when CSRC joins into this process, whether it will be still effective in resources allocation depends on whether firms with higher marginal value of financial flexibility can be more probable to get qualifications for equity financing. If the probability of equity financing for firms with higher marginal value of financial flexibility is higher, then total value added by increasing the level of financial flexibility will be higher. In this situation, we still can say that the new policy is effective, and vice versa. In addition, we would like to consider whether or not these measures for payout in Quasi-mandatory Payout Regulation can identify firms with good performance from those with bad performance, in order to judge the reasonability of these measures and their signaling function for investors. Only when firms with higher level of payout measure show better performance after

25 the equity financing, we can say that these measures indeed protect benefits of investors. We have shown that, theoretically, a firm s performance will be significantly and positively correlated with the marginal value of financial flexibility. If that is the case, it can be regarded as inefficiency of the CSRC when there is no significantly positive correlation. But we still need more empirical evidence to prove this. Table 6 Determinants of Future Performance Dependent Variable ROE_ST ROE_LT Model (5) (6) Intercept *** (-3.74) Cater (1.59) *** (-3.50) (0.33) *** (-4.76) (0.91) *** (-4.3) (1.09) MVOC *** *** *** 0.027*** (4.10) (6.48) (4.25) (6.65) g ** ** ** (2.08) (2.08) (1.54) (1.87) SEO *** *** (2.86) (0.96) (3.08) (0.40) ROE_Lag *** *** *** *** (24.06) (11.13) (19.40) (9.87) OwnCon ** *** (0.15) (2.56) (1.03) (3.06) INDROE *** *** (-1.12) (4.89) (-0.63) (5.39) SIZE *** (3.40) *** (2.88) *** (4.65) *** (3.11) Sample Period Observations Adj R % 34.90% 51.36% 33.54% Note: Figures in the brackets are T Statistics; *** represents the significance at the level of 1% ( two tail); ** represents the significance at the level of 5% ( two tail); * represents the significance at the level of 10% ( two tail). We have seen clearly in table 6 that the future performance of a firm is positively correlated

26 with the marginal value of financial flexibility. For both sample period, the variable CATER shows insignificant sign, indicating that the minimum requirements of Quasi-mandatory Payout Regulation are not so effective to differentiate firms with good performance from those with worse performance. So, the threshold of payout level for equity financing is ineffective. Comparing the efficiency of these two regulations, we conclude that it is more efficient for regulating in 2007 than in It is supported by evidence below: First is about the sign of g. For both periods, the growth rate shows a significantly positive correlation with firms future performance. But we notice in table 6 that in 2007, the term g shows significantly positive sign, which mean firms that actually get equity financing approved have comparatively higher growth rates. Considering the relationship between growth rate and future performance, firms who get financing perform better than others in the same period. However, in 2009, respective sign has changed into negative, which means that firms with higher growth were less likely to get approval for equity financing. With the relationship is still positive, firms who get financing do not significantly perform better for other in the same period. More directly, we find out that when the event year is 2007, the correlation coefficient of the dummy variable SEO is significantly positive, indicating that firms who engage in equity financing perform better. This correlation becomes insignificant in So in comparison, the efficiency of regulating in 2007 is better. Why is there a difference in the efficiency of regulating between the two periods? There are two possible explanations. One explanation arises from the perspective of the correlation between leverage and performance. In the calculation formula for the marginal value of financial flexibility, it is negatively correlated with the market value based leverage. Hence, firms with lower leverage

27 will have a higher marginal value of financial flexibility, and further better performance. In the event year of 2007, the probability of equity financing was higher for firms with comparatively lower leverage, while the situation reverses in As a result, the sample of equity financing in the event year of 2007 was more likely to perform better than that of The other possible explanation is the difference between rules of regulations s regulation made it more difficult for firms with a high growth rate to cater. As a result, firms with lower growth rate became qualifying for equity financing, and in keeping with the positive correlation between firms performance and growth, it led to the situation of worse performance in firms that actually qualified for equity financing. All in all, we think that the measure of dividend payment regulations is indicative of firms performance, and its signaling function is coming through catering decision for firms. The Approval System implemented by CSRC was more effective in protecting investors in In both periods, the firm value and performance always show significantly positive correlation with the marginal value of financial flexibility, providing empirical evidence that financial flexibility can increase firm value and is a good indicator of firm performance. 4. Robustness Checks We also try other variable and model specifications and results are qualitatively similar. Alternative variables we consider include: (1) Use cash and leverage to represent the level of financial flexibility in order to avoid the possibility of bias related to our measure of the marginal value of financial flexibility. (2) Use the accurate payout ratio of firms instead of dummy variable CATER. (3) Measure the future performance of firms as the average performance of the event year

28 and after. Ⅴ. Conclusion This paper uses regulations issued in 2006 and 2008about payout threshold as the institutional background to examine relationship between firms payout choice and equity financing behaviors directly and further investigate whether these regulations can justify the performance of the firm and how effective they are in protecting investors benefits. We find that firms with higher marginal value of financial flexibility are more likely to cater for regulations by changing their dividend policy, which indicates that financial flexibility is the key factor in understanding catering behavior in paying dividend under this special institutional background. This finding can also explain the potential relationship between financing behavior and payout decisions. As a developing market, regulations still plays a major role in the capital market. Chinese listed firms must cater to these regulations to qualify for equity financing, thus making the decision of paying dividend connected with financing behavior. The larger the marginal value of financial flexibility, the higher the probability of taking catering strategy in dividend policy. We also find that the actual probability of equity financing is determined by spare debt capacity. The inconsistency between the behavior of catering and actual equity financing represents the difference between the firm s potential financing purpose and the approval process executed by CSRC. When we research further the future performance of firms after equity financing, we find that the marginal value of financial flexibility can better signal the firm s future performance than measures of payout level in Quasi-mandatory Payout Regulations, and maintenance of financial flexibility is helpful in increasing firm s value. As a consequence, the insignificant correlation