Tax effect and government regulation effect on dividend policy by OLS

In this section, first, we use OLS to test the timing structure change and incentive of paying dividend, by using dummy variable and imputation tax credit rate,

respectively. We use dividend payout to be dependent variables to estimate the change of dividend policy over the sample period from 1994 to 2008. For these regressions, we include the product term from the ownership structure factors and dummy variable of tax reform, to test H3, whether investors have incentive to force managers to pay more dividends.

Under H1 and H2, we expect that in order to maximum the investors’ benefit, firms will pay more dividends after imputation tax system. And the higher imputation tax credit rate that firms have, the higher dividend firms will pay to investors. The evidence in Table 4 is consistence with the prediction of H2 but not H1, we find evidence that firm with higher imputation tax credit rate will pay more cash dividend and stock dividend, the regressive coefficients are positively to cash dividend and stock dividend;

the difference between cash dividend and stock dividend is negatively correlated to imputation tax credit rate, it indicates that firm will pay more stock dividend rather than cash dividend. But it’s not significant to the tax effect dummy variable on cash dividend, in other words, we have no evidence to prove that firms have propensity to pay more cash dividend after imputation tax system. But to stock dividend, it’s significant to be negatively correlated to the tax effect dummy variable. We infer that balanced dividend payout policy and abolishment of lagged taxation regulation result in the negatively correlated between stock dividend and the imputation tax system.⁷

Alli et al. (1993) used the proportion of shares owned by institutional investors to capture different tax clientele effect on different marginal tax rate of shareholders. Here we use proportion of share owned by investors which are institutional or individual to capture clientele effect after imputation tax system. Under H3, we expected that investor ask for higher dividend payout since they will gain tax credit from the new tax reform.

The higher proportions of share owned by institutional investor and individual investors, the more right they can execute to force managers pay more dividends because of the tax incentive.

From table 4 we found that both proportion of institutional and individual investors are negatively correlated to the imputation tax system even thought they are

7 On 2000, The Securities and Futures Bureau in Taiwan that advocate the balanced dividend policy, which promote firms paying stock dividend with cash dividend, to avoid the value of firm being affected by diluted earning per share; at the same year, lagged tax stock dividend has been canceled .

significantly correlated. From Jianguo and Nont researched on 2009, there is strong evidence to support that in New Zealand stock market the corporate dividend payment is closely tied with management share holdings. The higher the management share holding, the lower the dividend payout ratio, which is consistent with the agency cost hypothesis reported in Chen and Steiner (1999). But there is no propensity to manager to affect dividend payout.

4.2.2 Government regulation on dividend policy

In this part we test several predictions derived from substitution hypothesis. Follow Dittmar (2000), he found that firms usually use share repurchase to substitute pay cash dividend in U.S. and Grullon and Michaely (2002) found that cash spend on

repurchasing share has been increasing higher than cash dividends. In this section, follow the test as above, by using dummy variable and share repurchase per net income, respectively, to test H3 and H4.

Under H4, we expect that firm will prefer to use share repurchase to substitute cash dividend payout. For firms that exclusively use repurchase to pay out cash to

shareholders, earnings are likely to explain repurchase in a manner similar to traditional relation between dividends and earnings. By table 4, we found that firms will not reduce the cash dividend payment after the government regulation change. And the correlated coefficient between cash dividend and share repurchase per net income is negative, it is not consistence with the prediction of H3 , we find no evidence to said that firm will use share repurchase to substitute cash dividend payout, in other words, there is no

substituted effect but complementary effect between cash dividend and share repurchase.

We thought the reasons are that the level of free cash flow of firms can afford paying cash dividend and repurchasing share, simultaneously. But we found that there is

negative correlated between share repurchase and stock dividend payout.

We infer that there is the same reason as the circumstance in tax reform, On 2000, The Securities and Futures Bureau in Taiwan that advocate the balanced dividend policy, which promote firms paying stock dividend with cash dividend, to avoid the value of firm being affected by diluted earning per share so that firms will reduce stock dividend payout but increase cash dividend payout. According to the empirical result, that firm will not use share repurchase to substitute cash dividend payout in Taiwan since the market environment is quit differ from U.S. and the purposes of share repurchase in Taiwan are not focus on distributing earning to shareholders. Overall, if firms have enough free cash flow then they will make effect will transfer to complementary effect.

DeAngelo (2002) showed the same result.

[Insert Table 4 here]

To prevent the interaction between tax reform and government regulation, we divide the sample into two sub-period sample, 1994 to 1999 for imputation tax

system、1998 to 2003 for share repurchase. From Table 5 that we found cash dividend is significant to be positively correlated to the dummy variable of imputation tax system.

The proportion of share own by institutional investor become insignificant to dividend payout. And the results on the sub-period form 1998 to 2003 are identical to the whole period sample; there is no substituted effect between share repurchase and cash

dividend.

[Insert Table 5 here]