CONCLUSIONES Y RECOMENDACIONES
7.4 Fuentes electrónicas
The dividend payout ratio regression results shown in Table 2.3 seem to support the traditional view on dividend taxation. However, those regression estimates are biased due to endogeneity because corporations that want to distribute more dividends will select the S form. An instrumental variable which is related to organizational form but unrelated to dividend payout ratio in theory is necessary for unbiased estimation.
However, despite the rapid expansion of the S form since the end of the 1980s, very little empirical research has been done on the choice between C and S corporation form due to the lack of appropriate data. The primary research interest focuses on how various financial characteristics such as leverage ratio and Net Operating Loss (NOL) are related to the choice of S versus C organizational form. Even though some of them are flawed by defective research designs, the general results are that corporations respond to different tax treatment by selecting different organizational forms.
Plesko (1994) takes a sample of newly incorporated C and S corporations in 1983 which met the S form requirements from the 1984 Statistics of Income corporation file.
He compares the financial characteristics of S and C corporations to find which financial characteristics affect a firm’s decision to choose the C rather than the S form. His results show that larger, more profitable firms with higher compensation, fewer retained
earnings, and lower debt-to-asset ratio are more likely to be C corporations. However, some of the results are counterintuitive. First, firms with higher debt-to-asset ratios should choose to be C corporations because high leverage provides an interest shield to reduce CIT (interest payments are deductible from corporate taxable income) and should reduce the incentive to choose the S form. Second, fewer retained earnings implies higher dividend distribution and more double taxation, which should be associated with a higher probability of choosing the S form. As Plesko himself points out, there is an endogeneity problem with his analysis because he analyzes the impact of those characteristics
conditional on the corporations’ already having chosen their organizational forms. The choice of C versus S forms might have already had an impact on the financial
characteristics.
Ayers et al. (1996) examine a sample of firms from the National Survey of Small Business Finances conducted in 1988 and 1989. Their sample includes partnerships and sole proprietorships, as well as C and S corporations. They apply multinomial logistic regression to examine the impact of certain characteristics on organizational form. Their results suggest that if we only compare C and S corporations, corporations with higher losses and higher corporation tax payments (if they have chosen to be C corporations) are more likely to be S corporations. Their explanation is that loss could help shareholders to offset their other personal income, so corporations with higher losses will choose the S form; and if potential CIT payments are low, the corporate income will be taxed at a
lower CIT rate rather than the maximum personal income tax rate, so shareholders will choose to hide their profit in C corporations. However, their analysis is also troubled with the endogeneity problem and their findings should be viewed with caution.
To avoid the endogeneity problem, some scholars rely on the times when tax rules change and collect information before and after the change. The Tax Reform Act of 1986 (TRA86) provided a good chance for analysis. Plesko (1995) collects a sample of C corporations eligible to convert into S in 1986 and examines the impact of certain corporation characteristics on the decision to convert between 1986 and 1988. His variables capture a number of financial and legal factors, including gross corporate tax payment, undistributed income, interest paid, and carryover tax attributes. His regression results are consistent with the expectation that tax shields reduce the incentive to convert into the S form. In particular, debt has a negative impact on the conversion from C to S (contradictory to his 1994 finding).
Before 1986, there was a loophole that C corporations could purchase capital assets with retained earnings, use the cost of purchase to deduct corporate taxable income, and sell those assets after they elected the S status later so that they could avoid CIT. Then the IRS created a built-in-gain tax which taxes the sales of capital assets if S corporations sell the assets within a certain period of time after conversion. Omer, Plesko and Shelley (2000) focus on a sub-sample of firms in the natural resource industry from the same data as in Plesko (1995) did to examine the impact of such built-in-gain tax on the decision of C to S conversion. They conclude that the potential for built-in-gain realization significantly reduces the probability of converting from the C into the S form, conforming to their predictions.
The Small Business Job Protection Act of 1996 allowed banks to convert to S corporations for the first time. Hodder et al. (2003) use logit regression on a series of banks’ tax-relevant variables before the allowed conversion date to see whether banks made a conversion between 1996 and 1998. Consistent with the tax incentive theory, they find that banks are more likely to convert when conversion saves dividend taxes, avoids alternative minimum taxes, and minimizes state income taxes, and they are less likely to convert when conversion nullifies Net Operating Loss carry-forwards and creates potential penalty taxes on unrealized gains existing at the conversion date.
In spite of inherent defects in some of the research, the general finding conforms to the theoretical prediction that shareholders respond to corporate income tax and choose to elect the S status in order to pay fewer taxes. Therefore, the literature supports the argument that corporations with a higher CIT rate are more likely to elect the S form. If I can prove the relationship between CIT rate and organizational forms with my data, I can use CIT rate as an instrumental variable for unbiased regression of dividend payout ratio, presuming that CIT rate is not correlated with dividend payout ratio.