An evaluation of the effect of the 1986 Tax Reform Act on risk-adjusted measures of corporate tax equity.

Lirely, Roger L. ; Welker, Robert B. ; Little, Philip L. 等

ABSTRACT

This paper examines the effects of the Tax Reform Act of 1986 (TRA86) on the relationship between corporate taxes and ability to pay taxes using two dimensions of tax equity, horizontal equity and vertical equity. We define a corporation's ability to pay as windfall profits, which because it explicitly adjusts for differences in the risk associated with corporate investment, is a better measure of economic income than financial statement measures prevalent in prior research. We find that TRA86 had ambiguous effects on corporate tax equity and, thus, may not have achieved its intended purpose of enhancing corporate tax equity under an ability-to-pay criterion.

INTRODUCTION

The U.S. Congress sought to tax more closely the economic income of corporations (U.S. Congress, House 1985, 59, 306-7; U.S. Congress, Senate 1985, 518-520) by modifying the corporate tax provisions as part of the Tax Reform Act of 1986 (TRA86). These modifications were designed, in part, to increase corporate tax equity by ensuring that corporations pay federal income taxes in proportion to their abilities to pay tax (U.S. Congress, House 1985, 55; U.S. Congress, Senate 1985, 6). Prior studies (Kern & Morris, 1992; Manzon & Smith, 1994; Shevlin & Porter, 1992) provided evidence of the effects of TRA86 on the equality of corporate tax burdens. Each of the prior studies used financial statement data to estimate corporate ability to pay tax, including pretax income (Manzon & Smith, 1994; Shevlin & Porter, 1992), adjusted pretax income (Kern & Morris, 1992) and estimates of operating cash flows (Kern & Morris, 1992; Shevlin & Porter, 1992).

Accounting-based financial statement data may, however, be unsuitable for comparatively evaluating the abilities of corporations to pay a tax. One of the complications associated with applying an ability-to-pay standard to corporations is that the ability of a corporation to pay a tax depends, in part, on investment risk. For instance, a change in tax law to equate the tax burdens of companies with similar financial statements but different investment risks may structurally alter the relative competitiveness of corporations in capital markets, disadvantaging those with higher risk and advantaging those with lower risk. Because investment risk is masked in accounting-based financial data, the degree to which accounting-based measures of ability-to-pay appropriately reflect relative differences in investment risk among corporations is indeterminate.

The purpose of the study is to determine the effects of TRA86 on corporations' ability to pay taxes using a risk-adjusted measure of income. We depict the ability-to-pay corporate taxes in terms of windfall profit, which we define as the excess (deficiency) of realized returns over (under) the amount necessary to compensate investors for the risk of their investment (e.g., Pigou, 1928, 177). The explicit treatment of risk as a cost is more consistent with a notion of economic income.

We evaluate ability-to-pay on the basis of whether TRA86 increased or decreased horizontal and vertical equity. Horizontal equity is the variation of tax paid that remains unexplained after controlling for windfall profits or losses (Anderson, 1985; 1988; Enis & Craig, 1990; Grasso & Frischmann, 1992; Pierce, 1989). We consider TRA86 to have increased horizontal equity if this unexplained variation of tax is lower after the enactment of TRA86. Vertical equity is defined with respect to the rate of change in the relationship between average taxes and windfall profit (Musgrave & Thin, 1948; Pigou, 1928). A tax system is progressive (regressive) when the change in average tax for a unit change in windfall profit increases (decreases) as windfall profit increases. We deem TRA86 to have greater vertical equity if its enactment either increased the level of progressivity or decreased the level of regressivity.

BACKGROUND

Congress sought to increase corporate tax equity by extensively modifying the provisions that govern corporate taxation. The ten provisions with the greatest impact on corporate tax revenues, as projected by the Joint Committee on Taxation (Joint Committee, 1987), are listed in table 1. Congressional committees identified several of these provisions as modifications designed to increase corporate tax equity by more closely taxing corporate economic income (e.g., U.S. Congress, House, 1985, 306; U.S. Congress, Senate, 1985, 518-519). The House Ways and Means Committee discussion of the repeal of the investment tax credit and the revision of the cost recovery system exemplifies this mind-set:

The committee is very concerned about the significant inequities that result from the amount and concentration of these tax benefits ... corporations who have considerable amounts of economic income are permitted to pay little or no tax by using the credit and ACRS (accelerated cost recovery system) (U.S. Congress, House, 1985, 145). The Senate Finance Committee Report (U.S. Congress, Senate, 1985, 7) contains similar language.

While the House Ways and Means Committee and Senate Finance Committee expressed their equity concerns in terms of economic income, equity provisions of TRA86 focus on financial statement income. For instance, one unprecedented change enacted under TRA86 for tax years 1987 through 1989 required corporations to compare financial statement income and adjusted taxable income. If financial statement income sufficiently exceeded adjusted taxable income, corporations were required to add one-half of the excess to alternative minimum taxable income. For tax years after 1989, the adjustment is based on the difference between adjusted current earnings (a tax-based concept) and adjusted taxable income. The use of accounting income for benchmarking, rather than economic income, suggests that TRA86 may be taxing a corporation on the basis of its economic risk.

A corporation with high economic risk may be required to pay a tax on an economic cost of doing business, a tax that it may not have an ability to pay. Thus, while TRA86 added provisions to enhance equity in terms of accounting income, it may have not enhanced equity in terms of economic income.

A commonly accepted approach for evaluating corporate tax equity is to examine the distribution of corporate effective tax rates following newly enacted legislation. Several post-TRA86 studies examined the effects of TRA86 on corporate tax equity or, more generally, on the distribution of corporate effective tax rates as a result of the changes enacted by TRA86. Table 2 lists these studies and provides the effective tax rate measure and data source for each study. For a sample limited to 250 of the largest U.S. corporations, the Citizens for Tax Justice (CTJ, 1988; 1989) compared effective tax rates for the years 1981-85 to those from 1987 and 1988. CTJ (1988) reported increased U.S. federal effective tax rates following TRA86 and stated that TRA86 had "created a more level playing field" (p. 4) for corporations by closing the gap between corporate effective tax rates and that "tax reform is working" (p. 9). The General Accounting Office (GAO, 1990) reported that worldwide effective tax rates increased from 27.8 percent in 1986 to 35.5 percent in 1987 and concluded that increased corporate effective tax rates were consistent with the changes introduced in TRA86 (GAO, 1990, 3). Gupta and Newberry (1992) found that worldwide and U.S. federal effective tax rates increased immediately following TRA86, then leveled off by 1988. However, none of these studies conducted statistical analyses.

Shevlin and Porter (1992) performed a statistical analysis to determine the relative effect of increases in the corporate taxable income base, decreases in statutory tax rates and changes in financial statement income on post-TRA86 effective tax rates. Their results suggested that the observed increase in effective tax rates following TRA86 was due in large part to increases in corporations' taxable income base. Although their interest was not specifically related to the distribution of effective tax rates as a result of TRA86, Kern and Morris (1992) provided evidence that the differences between the effective tax rates of small versus large corporations decreased following TRA86. Finally, Manzon and Smith (1994) examined the distribution and variation of effective tax rates before and after two tax acts, the Economic Recovery Tax Act of 1981 and TRA86. Their results suggested that TRA86 increased corporate effective tax rates and reduced the variation of effective tax rates between corporations in different industries.

ASSESSMENT OF EQUITY

Equity is achieved under an ability-to-pay principle when individuals sacrifice equally through payment of tax (Musgrave & Musgrave, 1973, 198). Because equity is a construct defined in terms of individual taxpayers, assessing corporate tax equity is problematic. Prior studies have addressed this problem by imputing corporate taxes to individuals based upon their relative share of corporate income. Tax equity was then assessed at the individual level by examining the relationship between individuals total tax burden including imputed corporate taxes and their ability to pay taxes (e.g., Browning & Johnson, 1979; Feldstein, 1988; Pechman, 1985; Wallace et al., 1991). While this approach is suitable for examining the overall impact of tax reform on tax equity, it does not allow one to isolate the equity effects of entity-specific provisions.

The use of windfall profit as a measure of ability to pay tax allowed us to "impute" corporate income and taxes to shareholders while also allowing us to isolate the effects of the TRA corporate tax reforms on tax equity. We defined windfall profit (y) as the change in the value of corporate equity during a fiscal year above (positive windfall profit) or below (negative windfall profit) the amount of change that provides a normal return to investors. The change in value was estimated as the difference between the daily compounded increase or decrease in the value of the investment from the beginning to the end of a fiscal year and the daily compounded increase in value that one would expect given the corporation's investment risk. In short, windfall profit is the unexpected increase in shareholder wealth, that is, the economic income that accrued to the shareholder from investment in the corporation, during the corporation's fiscal year.

There are two dimensions of equity under the ability to pay principle. Horizontal equity requires equal sacrifice by individuals with equal abilities to pay. It is typically measured as the degree of variability in taxes paid across individuals entities with equal incomes. Perfect horizontal equity is denoted by zero variability of tax across individuals with equal incomes.

Because some variability of tax is inevitable, studies of horizontal equity have generally been constrained to evaluating whether tax system features or tax reform increased horizontal equity by reducing tax variability. The general approach is to assemble tax-paying entities into groups of approximately equal incomes, then examine the difference in the variability of tax liabilities (Anderson, 1985; 1988; Enis & Craig, 1990; Madeo & Madeo, 1981; 1984; White & White, 1965) or of tax rates (Pierce, 1989; Ricketts, 1990) within groups. The present study employs a similar approach. We measure horizontal equity as the variation in the tax burden of corporations that remains unexplained after controlling for windfall profits. We considered TRA86 to have increased (decreased) horizontal equity if the unexplained variation is less (greater) in the post-TRA86 period than in the pre-TRA86 period.

Vertical equity requires equal sacrifice by individuals with unequal abilities to pay. A taxpayer's sacrifice is typically measured in terms of taxes paid and vertical equity is typically evaluated in terms of progressivity, the degree with which individuals at higher income levels pay a greater proportion of income in taxes (e.g., Musgrave & Thin, 1948; Pigou, 1928). A tax system is progressive (regressive) when those with higher incomes pay a greater (lower) proportion of their income in tax and proportional when everyone pays the same proportion of income irrespective of income level. Since variability in taxes paid is inevitable at each income level, progressivity is typically estimated using the average proportion of tax paid by individuals at each income level. Consistent with this general approach, we consider TRA86 to have increased (reduced) progressivity if the rate of change in the slope of the relationship between average taxes paid by corporations is more (less) positive after TRA86 than it was before TRA86.

PRE-TRA86 AND POST-TRA86 TEST PERIODS

We assessed the effect of TRA86 on ability-to-pay by comparing the change in equity measures between two fiscal years, a pre-TRA86 year and a post-TRA86 year. The pre-TRA86 year was a fiscal year that began on or after July 1, 1982, and concluded no later than May 31, 1984. The July 1 starting date was selected for convenience. It coincides with 1983 fiscal year data compiled on Compustat PC Plus, which served as a principal source of data for the study. Fiscal years that concluded later than May 31, 1984 were excluded to ensure that estimates of investment risk and effective tax rates were not correlated with stock market responses to TRA86. For instance, the May 31 date is well before September 1984, which Givoly and Hayn (1991) identified as the date in which the first information about TRA86 (when the Chairman of the House Ways and Means Committee endorsed tax reform) entered the market.

The post-TRA86 year was a fiscal year that began on or after July 1, 1988, and concluded no later than May 31, 1989. Again, we chose the July 1 starting date because it coincides with Compustat PC Plus data. Earlier fiscal years were not considered to ensure that figures (e.g., tax, income, and risk measures) used for estimating the equity measures fully reflected the provisions of TRA86. For example, Scholes et al. (1992) showed that corporations responded to reduced tax rates enacted under TRA86 by accelerating expenses into 1986 and deferring income into 1987.

MEASURES

Our measure of corporations' tax burden was current worldwide taxes paid (T), which is the sum of currently payable federal, state, and foreign income taxes. Stickney and McGee (1982) and Bernard and Hayn (1986) argue that worldwide taxes paid is a more meaningful measure of tax burden than other measures, because income taxes paid to one governmental unit can often be offset against either taxable income or taxes payable to another governmental unit. Since current worldwide taxes paid is not reported in financial statements and corporate tax return information is not publicly available, we estimated current worldwide taxes paid by subtracting deferred tax expense from total tax expense (in Compustat PC Plus mnemonics, TOTTAX less TXIDT). This approach is consistent with prior research (Omer et al., 1991; Shevlin & Porter, 1992; Stickney & McGee, 1982).

The value of the corporation at a given time is the product of share price and number of shares outstanding. We estimated windfall profit as the difference between the value attained at the end of a fiscal year and the value that was expected for the fiscal year given the corporation's beta risk. Notationally, the difference can be expressed as:

 [[empty set.sub.ij] = ([N.sub.ij] x [P.sub.ij ]x [R.sub.ij]) -
 ([N.sub.ij] x [P.sub.ij] x E[(R).sub.ij]) (1)

where:

[N.sub.ij] is the number of shares outstanding at the beginning of fiscal year j for corporation, i;

[P.sub.ij] is the price of a share of common stock on the first day of fiscal year j;

[R.sub.ij] is the daily compounded actual return for the fiscal year; and

E[(R).sub.ij] is the expected daily compounded return for the fiscal year given the corporation's beta risk. The daily compounded actual return, R, and the daily compounded expected return, E(R) for corporation i in year j were defined as:

 R = [P(1+[r.sub.d])] - 1, (d = 1 ... D) (2)

 E(R) = [[ETH](1+E([r.sub.d])] - 1, (d = 1 ... D) (3)

where [r.sub.d] is the actual return for trading day d (d = 1, ... D, the number of trading days in the fiscal year) and E([r.sub.d]) is the expected return for trading day d given the corporation's beta risk.

Estimation of the capital asset pricing model (Lintner, 1965; Mossin, 1966; Sharpe, 1964; see Jensen, 1972, 5, for model assumptions) provided an estimate of the expected return E([r.sub.d]). Beta coefficients were estimated for the pre-TRA86 test period using daily stock returns from the 150 trading days immediately preceding each corporation's fiscal year, whereas betas for the post-TRA86 test period were estimated using daily returns from the 150 trading days immediately following each corporation's fiscal year. The value-weighted return of all corporations on the NYSE-Amex and Nasdaq for the relevant estimation period provided a proxy for the market return, [R.sub.M,d] (Mendenhall & Nichols, 1988). The daily return of treasury bills that were approximately thirty days from maturity provided a proxy for the risk-free return. For the few holidays each year that banks were closed but stock markets were open, we estimated risk-free returns using the mean of the "asked" treasury bill discount for the day preceding the holiday and the day following the holiday. To ensure that estimated betas fully captured all of the differences between the risk-free return and the market return, the intercept of the capital asset pricing model was constrained to equal zero (Lintner, 1965).

DATA SOURCES AND SAMPLE SELECTION

We obtained income and tax data from the 1989 Standard and Poor's Compustat PC Plus, price data for estimating daily returns from the 1994 CRSP database, and treasury-bill information for estimating risk-free returns from The Wall Street Journal. The starting point for sample selection was the 8,706 U.S. corporations listed on Compustat PC Plus. The following corporations were eliminated from the initial sample, in the order presented:

1. Corporations (n=6859) with missing data on Compustat PC Plus in
either test period.

2. Foreign corporations, because foreign corporations are taxed only
on U.S.-sourced income, whereas U.S. corporations are taxed on
worldwide income. As a result, comparisons between U.S. and foreign
corporations are difficult.

3. Corporations with greater than a three percent change in shares
outstanding during the test periods (n=964), because material changes
in shares outstanding confounds the calculation
of equity measures.

4. Corporations that merged with, acquired, or divested another
corporation or discontinued significant operations during the time
frame of the study (n=367). Mergers, acquisitions, and divestitures
were obtained from financial statement footnotes on Compustat PC Plus.

5. Corporations with missing CRSP data (n=157).

Next, we randomly divided the final sample, which contained 359 corporations, into two groups, a pre-TRA86 group (n=180) for estimating equity measures in the pre-TRA86 test period and a post-TRA86 group (n=179) for estimating the post-TRA86 test period equity measures.

To evaluate whether the samples were representative of the Compustat PC Plus population, corporations in the pre- and post-TRA86 samples were pooled and the mean size and industry representation of corporations in the pooled sample were compared to those of U.S. corporations listed on Compustat PC Plus. A significant mean difference was found with respect to three size variables: stockholders' equity (t=3.95, p<.001), income (t=3.07, p<.01), and sales (t=2.59, p<.01). In each case, the Compustat corporations were larger than corporations in the pooled pre- and post-TRA86 samples. As to industry representation, a significant difference was found ([c.sup.2](5) = 69.43, p .01 [pounds sterling]). Manufacturing (52% vs. 37%) is over-represented and financial services (13% vs. 26%) is underrepresented in the pre- and post-TRA86 groups.

CENTERING THE PRE- AND POST-TRA86 GROUPS

The research design, assessing equity before and after the enactment of TRA86, requires that corporations in the pre-and post-TRA86 groups reasonably generalize to the same population of corporations. The randomization of corporations into two distinct groups, a pre-TRA86 group and a post-TRA86 group, was performed with this end in mind. However, the two groups may not generalize to the same population because of uncontrolled differences (e.g., economic factors) between the pre- and post-TRA86 periods attributed to the passage of time.

To assess generalizability, we applied an F-test to determine whether the standard deviations of windfall profits and its derivative measures for the pre-TRA86 group was similar to standard deviations for the post-TRA86 group. The results of the F-test comparison are reported in panel A of table 3. With one exception, the beta coefficient, the standard deviations were significantly greater in the post-TRA86 group, suggesting dissimilar economic conditions in the two test periods.

The range of windfall profit ($-3,528.80 million to $2,470.73 million) for corporations in the pre-TRA86 group was considerably smaller than the range of windfall profit ($-27,867.95 million to $5,599.65) for corporations in the post-TRA86 group. For comparability, it is essential that estimates of horizontal and vertical equity for the pre- and post-TRA86 test periods were computed for similar ranges of windfall profit. To achieve greater comparability, we adopted the following rule: corporations were removed from the pre-TRA86 (post-TRA86) group if no corporation in the post-TRA86 (pre-TRA86) group had windfall profit within 20 million dollars. Application of this rule eliminated 11 corporations from the pre-TRA86 group and 20 corporations from the post-TRA86 group. Panel B of table 3 shows standard deviations for the centered samples. We could not reject (F(158,167) = 1.08, n.s.) a null hypothesis of no difference in standard deviation of windfall profits between the pre-TRA86 group (n=168) and post-TRA86 group (n=159). However, the standard deviations of three constituent measures of windfall profits (API, shares outstanding, and share price) were significantly different, suggesting that the centering approach may not have fully centered the samples with respect to economic conditions.

After centering the samples, the pre-TRA86 group and post-TRA86 group were compared to see if disparities existed with respect to size characteristics and industry makeup. Table 4 reports means of size variables (total assets, total liabilities, stockholders' equity, market value, net income, and sales) for each group. The groups compare favorably in terms of the size variables. Table 5 compares the groups in terms of industry representation. For industry dispersion testing, the corporations in three industries, agriculture and forestry, transportation and communication, and utilities, which were few in number, were combined into one industry group. The difference in industry representation between the pre-TRA86 group and post-TRA86 group is non-significant ([c.sup.2](5) = 3.36, n.s.).

RESULTS

Vertical and horizontal equity of the tax system were assessed by modeling the curvilinear relationship between taxes and windfall profits with the following regression model:

 [T.sub.i] = [a.sub.0] + [a.sub.1][y.sub.i] +
 [a.sub.2] [y.sup.2.sub.i] + e (4)

With respect to vertical equity, [a.sub.2] > 0 indicates a progressive tax system, [a.sub.2] < 0 indicates a regressive tax system, and [a.sub.2] = 0 indicates a proportional tax system. Horizontal equity is indicated by the distribution of the error term ([e.sub.i]) at a given level of windfall profits ([y.sub.i]). A less diffused error term suggests greater horizontal equity.

As a preliminary step, the pre- and post-TRA86 samples were pooled and equation 4 was estimated with the pooled sample. Two tests were applied to the distribution of errors ([e.sub.I]).

First, we tested the regression errors to determine whether outliers were present in the data set and removed three corporations that had error terms in excess of three standard deviations from the regression mean (e.g., Bernard & Hayn, 1986). Table 6 contains statistics relating to the three outlying corporations, each of which was in the post-TRA86 sample. Since OLS is particularly sensitive to the presence of outliers, analyses were conducted both with and without the three outlying corporations in the post-TRA86 sample.

Second, we employed the Goldfeld-Quandt test for homoscedasticity (Goldfeld & Quandt, 1965) to assess the degree of heteroscedasticity in the distribution of error terms. We anticipated heteroscedasticity because of a size bias in the measurement of windfall profit and taxes. In comparison with small firms, large firms tend to have higher levels of windfall profit simply because of their greater market value. In addition, large firms tend to have higher levels of accounting income and pay a greater amount of tax because of their greater earning power. Thus, we expected error variances to explode in opposite directions of y=0, that is, both as positive y becomes more positive and as negative y becomes more negative. The pre-TRA86, the post-TRA86 sample with outliers, and the post-TRA86 sample without outliers were each tri-sectioned on y, and regression model 4 was estimated with each sub-sample. Heteroscedasticity was assessed by examining the differences in mean square error (MSE) among the trisections of each sample. The test results are presented in table 7. An F-statistic was computed by comparing the trisection with the highest MSE with the trisection with the lowest MSE. It is significant for each sample, indicating the presence of heteroscedasticity. The anticipated explosion in the error variance is evident in the test results for the pre-TRA86 sample and the post-TRA86 sample without outliers. For these samples, the MSE is least for the middle trisection, which contains values of y closest to zero.

Because of the presence of heteroscedasticity, we assessed horizontal equity by sectionalizing the pre-TRA86 sample and the post-TRA86 sample into six income subsections. The six subsections were deemed to be large enough to maintain adequate sample size and small enough to reduce the influence of the heteroscedastic error on test results. For each subsection, we obtained a MSE for the pre-TRA86 sample and a MSE for the post-TRA86 sample by estimating the regression model in equation 4. We applied an F-test to test the null hypothesis that the MSE for the pre-TRA86 sample is equal to the MSE for the post-TRA86 sample. Significance suggests that the distribution of residual error changed in that subsection as a result of TRA86. The results are reported in table 8.

With outliers excluded, we found a significant difference in MSE for the each income subsection except the "<-$100MM" and the ">$100MM" subsections. In the subsections where significance was found, the results are mixed. The MSE for the post-TRA86 sample exceeded that of the pre-TRA86 sample for three subsections, the "-$100MM to -$50MM" subsection (1129 versus 90), the "-$50MM to $0" subsection (111 versus 72) and the "$50 to $100MM" subsection (794 versus 75). However, for the "$0 to $50MM" subsection, the MSE for the pre-TRA86 sample exceeded that of the post-TRA86 sample (106 versus 59). The absence of a consistent pattern of change in MSE suggests that TRA86 did not have an all-encompassing affect on horizontal equity. However, the mixed character of the results, both in terms of significance and direction, precludes the drawing of precise conclusions.

We evaluated the change in vertical equity (progressivity) between the pre- and post-TRA86 test periods with a Chow test (Chow, 1960). The preferred approach is to estimate equation 4 separately for the pre-TRA86 and post-TRA86 samples (the unrestricted model) and estimate equation 4 with the pre-TRA86 and post-TRA86 samples combined (the restricted model). The Chow test compares the sums of squared error of the unrestricted model with the sum of squared error of the restricted model to test the null hypothesis that the pre- and post-TRA86 samples have the same set of regression coefficients.

An assumption of the Chow test is a homoscedastic distribution for regression error term, which, as previously stated, does not hold for the pre- and post-TRA86 samples. Two steps were made to restore homoscedasticity. First, the three outliers were excluded from the post-TRA86 sample. Second, we transformed income (y) data and taxes (T) data into natural logarithms and estimated the following regression model:

 [logT.sub.i] = [d.sub.0] + [d.sub.1] ([logy.sub.i]) + [d.sub.2]
 [([logy.sub.i]).sub.2] + [e.sub.i] (5)

With the transformed data, the Goldfeld-Quandt test did not reject the null hypothesis of homoscedasticity (for the pre-TRA86 sample, largest F (43,43) = 1.11, and for the post-TRAl86 sample, largest F(46,46) = 1.21).

The regression coefficients in equation 5 cannot be interpreted in the same manner concerning progressivity as their counterparts in equation 4. For instance, while [a.sub.2] = 0 in equation 4 indicates a proportional tax system, [d.sub.2] = 0 in equation 5 indicates a proportional tax system with regard to the elasticity of taxes. Whether the system is progressive or regressive would depend additionally on the sign of the [d.sub.1] coefficient, with a positive sign suggesting an upward sloping logarithmic function and progressivity and a negative sign suggesting a downward sloping logarithmic function and regressivity. While the interpretation of regression coefficients [d.sub.1] and [d.sub.2] is less obvious, the transformation of the data does not detract from the intent of the study. We are interested in the change in progressivity between two test periods, rather than the degree of progressivity in any one period. The application of the Chow test with the transformed regression data answers the question as to whether the structural relationship between taxes (T) and income (y) across levels of income is different between the two test periods.

An additional requirement of the Chow test is that the variance of the pre-TRA86 sample and the variance of the post-TRA86 sample be reasonably similar in order to estimate the restricted (pooled) model without violating regression assumptions. The null hypothesis of no difference in MSE between the pre-TRA86 sample and the post-TRA86 sample is not rejected (F(153,165) = 1.07). Chow test results are reported in table 9. The null hypothesis of similar structure is not rejected (F(3,318) = .84, n.s.). Thus, we are unable to conclude based on the Chow test that vertical equity in the pre-TRA86 test period is different than the vertical equity in the post-TRA86 test period.

Using the estimated coefficients for the pooled sample, the relationship between taxes (T) and income (y) is graphed in figure 1. The curve is U-shaped, centered around the zero point. The U-shaped curve reflects the size bias of the tax system. Large corporations tend to have greater accounting income and therefore pay more tax than small corporations. They also tend to have the greatest market value and therefore the greatest positive or negative economic income. As a consequence of these tendencies, the finding that corporations with high negative incomes also pay a substantial amount of tax is not surprising.

[FIGURE 1 OMITTED]

The U-shaped curve indicates both equity and inequity of the tax system. For corporations with y>0, the curve is upward sloping at a gradual increasing rate, indicating progressivity and a modest degree of vertical equity. However, for corporations with y<0, the slope of the curve is negative. Corporations with negative incomes are paying only slightly less average tax than corporations with a like amount of positive income. This finding is a predictable manifestation of taxing systems that tax corporations based on accounting income and that do not explicitly consider investment risk in establishing tax policy. It also suggests inequity under an ability-to-pay standard. Corporations with negative income do not have the same ability to absorb the tax burden as corporations with positive income without being harmed in capital markets.

The analysis rests on several potentially restrictive assumptions. As a next step in the analysis, we examined the sensitivity of the results to several of the assumptions made in the estimation of equity measures. The following assumptions were either relaxed or re-specified, and the test statistics were recomputed to determine whether the change in assumption would have led to a different set of conclusions:

1. The CAPM was estimated without constraining the intercept to zero.

2. U.S. Federal taxes paid was substituted for worldwide taxes paid.

3. The 20 million-dollar threshold used for centering the sample was
changed to 10, 30, 40, and 50 million dollar thresholds.

Each of the sensitivity tests provided similar results.

CONCLUSION

Several studies (e.g., CTJ, 1988; 1989; GAO, 1990; Gupta & Newberry, 1992; Kern & Morris, 1992; Manzon & Smith, 1994; Shevlin & Porter, 1992) have examined the relationship between effective tax rates and measures of ability to pay before and after TRA86 to ascertain whether TRA86 achieved its intended purpose of enhanced corporate tax equity. The present study extends on this research in two ways. First, it evaluates the effect of TRA86 on horizontal and vertical equity of the corporate tax system. Some prior studies (e.g., CTJ, 1988; 1989; GAO, 1990; Gupta & Newberry, 1992) provided only descriptive evidence about effective tax rate distributions before and after TRA86. The primary research interest of other studies (e.g., Kern & Morris, 1992; Manzon & Smith, 1994; Shevlin & Porter, 1992), while providing evidence regarding the distribution of effective tax rates before and after TRA86, was not tax equity. Second, it measures ability-to-pay in terms of windfall profit, the amount of yearly gain in shareholders' wealth after deducting the return that compensates shareholders' for a normal investment return. We argue that windfall profit, because it adjusts for investment risk, is closer to the notion of economic income and to ability to pay than previously applied measures derived from financial statement information. The normal investment risk of a corporation, which economists consider to be a cost in the computation of economic income, is outside the scope of the measurements of financial statement information. Consequently, financial statements may yield incomplete measures of economic income, which may lead to misleading conclusions concerning a corporation's ability to pay a tax and thereby distort depictions of tax equity.

Our findings do not provide clear indications that TRA86 enhanced either horizontal or vertical corporate tax equity. Horizontal equity was assessed separately for six income groups. Depending on the income group studied, TRA86 was found to enhance equity, to reduce equity, and to have no effect on equity. Because of the mixed findings, we are unable to conclude that TRA86 had a general positive or negative effect on horizontal equity. Vertical equity was assessed by examining the structural relationship between average taxes and windfall profits. The structural relationship that existed after TRA86 was not different than the structural relationship that existed before TRA86. This finding suggests that TRA86 had a relatively small, if any, impact on vertical equity.

Past studies of the effectiveness of TRA86 (CTJ, 1988; 1989; GAO, 1990; Gupta & Newberry, 1992; Kern & Morris, 1992; Manzon & Smith, 1994) have generally concluded that the enactment of TRA86 achieved its purpose by improving upon the distribution of effective tax rates between large and small corporations and between industries. Our results raise the possibility of an opposite conclusion. In the deliberation process that led to TRA86, Congressional committees repeatedly alluded to corporate economic income as the ability-to-pay criterion for evaluating corporate tax equity. To the extent that windfall profit surrogates economic income, we are unable to conclude that TRA86 fulfilled goals of enhanced corporate tax equity within the Congressional committees' conception of the term equity.

The findings should be evaluated in light of limitations. First, our measure of ability to pay, windfall profit to shareholders, is an incomplete measure of economic income. It assumes that shareholders are the individuals who bear the burden of corporate taxes. Whether corporations can pass corporate taxes to factors other than shareholders (e.g., to debt holders, labor and consumers) has been the subject of considerable theoretical debate and empirical research with no definitive conclusions (see the discussion in Wallace et al., 1991). In the absence of definitive evidence regarding the incidence of corporate taxes, this study assumed that shareholders bear the entire burden of corporate taxes.

Second, we operationalized ability to pay in terms of shareholders' wealth from investment in the corporation, which assumes that the ability of shareholders to pay corporate taxes is derived solely from their investment in a corporation. This assumption was necessitated by a lack of data regarding shareholders' other sources of income (see Browning & Johnson, 1979; Feldstein, 1988; Pechman, 1985; Wallace et al. 1991 for methodologies that have attempted to address this limitation).

Third, the use of the capital asset pricing model to compute shareholder wealth assumes that a corporation's market beta risk captures the aggregated risk assumed by its individual shareholders. This assumption would be met, for instance, if each shareholder's investment included only shares in the corporation, which is implausible. However, the assumption is necessary, since we did not have access to the investment portfolios of individual shareholders.

REFERENCES

Anderson, K. E. (1985). A horizontal equity analysis of the minimum tax provisions: An empirical study. The Accounting Review, 60(2), 357-371.

Anderson, K.E. (1988). A horizontal equity analysis of the minimum tax provisions: 1976-1986 Tax acts. The Journal of the American Taxation Association, 10(2), 7-25.

Anderson, K. E., T. J. Hill, Jr. & D. P. Murphy. (1995). Horizontal equity and implicit taxes. The Journal of the American Taxation Association, 17(2), 89-100.

Bernard, V. L. & C. Hayn. (1986). Inflation and the distribution of the corporate income tax burden. National Tax Journal, 39(3), 171-187.

Browning, E. & W. Johnson. (1979). The Distribution of the Tax Burden. Washington, D.C.: American Enterprise Institute.

Chow, G. (1960). Tests of equality between sets of coefficients in two linear regressions. Econometrica, 28 (July): 591-605.

Citizens for Tax Justice (CTJ). (1988). The Corporate Tax Comeback. Washington, DC: Citizens for Tax Justice.

Citizens for Tax Justice. (1989). The Resurgence of Business Investment & Corporate Income Taxes: It's Working, But.... Washington, DC: Citizens for Tax Justice.

Enis, C. R. & D. L. Craig. (1990). An empirical analysis of equity and efficiency attributes of degressive forms of a flat tax. The Journal of the American Taxation Association, 11(1), 17-33.

Feldstein, M. S. (1988). Imputing corporate tax liabilities to individual taxpayers. National Tax Journal, 41(2), 37-59.

Givoly, D. & C. Hayn (1991). The aggregate and distributional effects of the Tax Reform Act of 1986 on firm valuation. Journal of Business, 64(July), 363-392.

Goldfeld, S. M. & R. E. Quandt (1965). Some tests for homoscedasticity. Journal of the American Statistical Association, 60, 539-547.

Grasso, L. P. & P. J. Frischmann (1992). Measuring horizontal equity: A regression approach. The Journal of the American Taxation Association, 14(2), 123-133.

Gupta, S. & K. Newberry (1992). Corporate average effective tax rates after the Tax Reform Act of 1986. Tax Notes, 55(May 4), 689-702.

Jensen, H. L. (1972). The foundations and current state of capital market theory. In M.C. Jensen (Ed.) Studies in the Theory of Capital Markets, New York: Praeger Publishers.

Kern, B. B. & M. H. Morris. (1992). Taxes and firm size: The effect of tax legislation during the 1980s. The Journal of the American Taxation Association, 13(1), 80-95.

Lintner, J. (1965). The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. Review of Economics and Statistics, 47(February), 13-37.

Madeo, L. A. & S. A. Madeo. (1984). The equity and motivating effects of the maximum tax. The Journal of the American Taxation Association, 5(1), 40-49.

Madeo, S. A. & L. A. Madeo. (1981). Some evidence of the equity effects of the minimum tax on individual taxpayers. National Tax Journal, 34(4), 457-65.

Manzon, Jr., G. B. & W. R. Smith. (1994). The effect of the Economic Recovery Tax Act of 1981 and the Tax Reform Act of 1986 on the distribution of effective tax rates. Journal of Accounting and Public Policy, 13(Winter), 349-62.

Mendenhall, R. R. & W. D. Nichols. (1988). Bad news and differential market reactions to announcements of earlier-quarters versus fourth-quarter earnings. Journal of Accounting Research, 26(Supplement), 63-86.

Mossin, J. (1966). Equilibrium in a capital asset market. Economica, 35(February): 74-82.

Musgrave, R. A. & P. Musgrave. (1973). Public Finance in Theory and Practice. New York: McGraw-Hill.

Musgrave, R. A. & T. Thin. (1948). Income tax progression, 1929-48. Journal of Political Economy, 56(December), 498-514.

Omer, T. C., K. H. Molloy & D. A. Ziebart. (1991). Measurement of effective corporate tax rates using financial statement information. The Journal of the American Taxation Association, 13(1), 57-72.

Pechman, J. A. (1985). Who Paid the Taxes, 1966-1985? Washington, D.C.: The Brookings Institution.

Pierce, B. J. (1989). Homeowner preferences: The equity and revenue effects of proposed changes in the status quo. The Journal of the American Taxation Association, 10(1), 54-67.

Pigou, A. C. (1928). A Study in Public Finance. London: MacMillan and Co.

Ricketts, R. C. (1990). Social growth versus income tax reform: An analysis of progressivity and horizontal equity in the federal tax system in the 1980s. The Journal of the American Taxation Association, 11(1), 34-50.

Scholes, M. A., G. P. Wilson & M. A. Wolfson (1992). Firms' responses to anticipated reductions in tax rates: The Tax Reform Act of 1986. Journal of Accounting Research, 30(Supplement), 161-91.

Sharpe, W. F. (1964). Capital asset prices: A theory of market equilibrium under conditions of risk. Journal of Finance, 19(September), 425-442.

Shevlin, T. & S. Porter. (1992). "The corporate tax comeback in 1987" Some further evidence. The Journal of the American Taxation Association, 14(1), 58-79.

Stickney, C. P. & V. E. McGee (1982). Effective corporate tax rates: The effect of size, capital intensity, leverage, and other factors. Journal of Accounting and Public Policy, 1(Winter), 125-152.

U.S. Congress, House Ways and Means Committee (House, 1985). House Ways and Means Committee Report to Accompany H.R. 3838 (S.Rep. No. 99-426). 99th Cong., 1st Session.

U.S. Congress, Joint Committee on Taxation (Joint Committee, 1987). General Explanation of the Tax Reform Act of 1986. Washington, DC: U.S. Government Printing Office.

U.S. Congress, Senate Committee on Finance (Senate, 1985). Report of the Senate Finance Committee to Accompany H.R. 3838 (S.Rep. No. 99-426). 99th Cong., 2nd Session.

U.S. General Accounting Office (GAO, 1990). Tax Policy-1987 Company Effective Tax Rates Higher Than In Prior Years. Washington, D.C.: U.S. Government Printing Office.

Wallace, S., M. Wasylenko & D. Weiner, (1991). The distributional implications of the 1986 tax reform. National Tax Journal, 44(2), 181-198.

White, M. & A. White. (1965). Horizontal inequality in the federal income tax treatment of homeowners and tenants. National Tax Journal, 18(3), 225-239.

Roger L. Lirely, Western Carolina University

Robert B. Welker, Southern Illinois University at Carbondale

Philip L. Little, Western Carolina University

Table 1
Corporate Tax Provisions of TRA86

Provision Estimated Revenues
 (in $Billions)

Repeal of investment tax credit 119
Lowering of statutory tax rates (116)
Changes in inventory capitalization rules 32
Stricter application of the alternative minimum tax 22
Disallowance of the completed contract method 10
Changes in the foreign tax credit computation 9
Changes in ACRS 8
Repeal of the reserve for bad debt for nonfinancial
 corporations 7
Stricter application of the installment method 7
Limitation on deductions for meals, travel, etc. 6

* Source: U.S. Congress, Joint Committee of Taxation (5/8/87)

Table 2
Summary of TRA86 Studies

Study Effective Tax Rate Measure Data Source

CTJ, Federal Income Tax Liability, Financial
1988; 1989 U.S. Net Income Before Taxes statements and
 Forms 10-K

GAO, 1990 Federal Income Tax Liability, Financial
 U.S. Net Income Before Taxes statements

Gupta & 1. Worldwide Income Tax Liability, Compustat,
Newberry, 1992 Net Income Before Taxes NAARS, and
 2. Federal Income Tax Liability, Forms 10-K
 Net Income Before Taxes
 3. Foreign Income Tax Liability,
 Net Income Before Taxes

Kern & Morris, 1. Worldwide Income Tax Liability, Compustat and
1992 Operating Cashflows Value Line
 2. U.S. Federal Income Tax Liability,
 Net Income Before Taxes

Manzon & Smith, 1. Federal Income Tax Liability, Compustat
1994 Net Income Before Taxes
 2. Worldwide Income Tax Liability,
 Net Income Before Taxes

Shevlin & 1. Federal Income Tax Liability, Compustat
Porter, 1992 Net Income Before Taxes
 2. Federal Income Tax Liability,
 Operating Cashflows

Table 3
Standard Deviations of Variables Used in Estimation of Windfall
Profits: A Comparison between Pre- and Post-TRA86 Groups

Panel A: Results for the full sample (n = 180 for pre-TRA86 group
and n = 179 for post-TRA86 group)

Windfall Profit Variables Pre-TRA86 Post-TRA86 F-Statistic for
 Group Group Difference in
 Std. Dev.

Windfall Profit ($MM) 386.04 2180.17 31.90 ***
Abnormal Performance Index .63 .32 3.88 ***
Beta Risk Coefficient .51 .46 1.23
Shares Outstanding (MM) 120.72 158.53 1.72 ***
Share Price ($) 9.73 14.55 2.24 ***

*** p < 0.001

Panel B: Results for the centered sample (n = 168 for pre-TRA86
group & n = 159 for post-TRA86 group)

Windfall Profit Variables Pre-TRA86 Post-TRA86 F-Statistic for
 Group Group Difference in
 Std. Dev.

Windfall Profit ($MM) 102.52 106.39 1.08
Abnormal Performance Index .63 .32 3.88 ***
Beta Risk Coefficient .49 .44 1.24
Shares Outstanding (MM) 91.08 61.16 2.22 ***
Share Price ($) 8.85 11.48 1.68 ***

*** p < 0.001

Table 4
Mean (Std. Dev.) Size of Corporations
in Pre-TRA86 and Post-TRA86 Groups

 Size Variables Pre-TRA86 Post-TRA86
 Group (n=168) Group (n=159)

Total Assets ($MM) 927.99 (6215.69) 1133.85 (4952.44)
Total Liabilities ($MM) 737.30 (6050.12) 846.35 (4415.25)
Stockholders' Equity ($MM) 190.69 (513.39) 287.49 (700.55)
Market Value ($MM) 312.17 (760.98) 562.44 (1672.70)
Net Income ($MM) 21.36 (49.30) 36.09 (124.25)
Sales ($MM) 534.25 (1357.58) 792.19 (2160.49)

 Size Variables Significance of
 Difference in
 Mean
Total Assets ($MM) n.s.
Total Liabilities ($MM) n.s.
Stockholders' Equity ($MM) n.s.
Market Value ($MM) n.s.
Net Income ($MM) n.s.
Sales ($MM) n.s.

Table 5
Industry Distribution of Corporations
in the Pre- and Post-TRA86 Windfall Samples

Industry SIC Range Pre-TRA86 Post-TRA86
 (n = 168) (n = 159)

Agriculture and Forestry 0100 - 0971 1 0
Mining, Extraction and
 Construction 1000 - 1799 15 18
Manufacturing 2000 - 3999 90 79
Transportation and
 Communication 4000 - 4899 2 3
Utilities 4900 - 4991 6 10
Wholesalers and Retailers 5000 - 5999 23 15
Financial Services and
 Insurance 6000 - 6799 20 23
Service 7000 - 8999 11 11

Table 6
Outlier Statistics

Taxes T Income y No. of SDs from
 Regres. Mean

 260 113.88 3.89
 447 .08 6.63
 884 272.29 14.41

Table 7
Results of Testing for Heteroscedasticity

Sample Pre-TRA86 Group Post-TRA86 Group
Trisection With Outliers

 Range of Y MSE Range of Y MSE

1 -692.76, 19.13 -684.14 23.00
 .38 -6.08
 (n=56) (n=53)

2 -.36, 14.88 10.28 -6.01, 3.20 123.12
 (n=56) (n=53)

3 15.04, 506.48 19.86 6.05, 517.19 58.43
 (n=56) (n=53)

Test F(53,53) F(50,50)
 statistic = 1.93 = 5.35
 (a) p < .01 p < .001

Sample Post-TRA86 Group
Trisection Without Outliers

 Range of Y MSE

1 -684.14, 23.22
 -6.83
 (n=52)

2 -6.08, 2.76 3.89
 (n=52)

3 3.20, 517.19 25.52
 (n=52)

Test F(49,49)
 statistic = 6.56
 (a) p < .001

(a) In the test, the tri-section with the highest MSE is
compared with the tri-section with the lowest MSE
(Goldfeld & Quandt, 1965)

Table 8
Test of Horizontal Equity (a)

y Subsections Pre-TRA 86 Post-TRA86 Sample with Outliers
 ($MM) Sample Included

 n MSE n MSE F p <

 < -100 7 4334 8 2711 1.60 n.s.
 -100 to -50 5 90 7 1129 12.54 .05
 -50 to 0 52 72 71 111 2.13 .01
 0 to 50 76 106 54 14429 136.12 .00
 50 to 100 18 75 7 794 10.59 .00
 > 100 10 2791 12 17228 6.17 .00

y Subsections Post-TRA86 Sample with Outliers
 ($MM) Excluded

 n MSE F P <

 < -100 8 2711 1.60 n.s.
 -100 to -50 7 1129 12.54 .05
 -50 to 0 71 111 2.13 .01
 0 to 50 53 59 1.8 .05
 50 to 100 7 794 10.59 .00
 > 100 10 3116 1.12 n.s.

(a) Degrees of freedom for the numerator and
denominator of the F-test are n-3.

Table 9
Results of the Chow Testing
General Regression Model:
log[T.sub.i] = [d.sub.0] + [d.sub.1] ([logy.sub.i]) +
[d.sub.2] [([logy.sub.i]).sup.2] + [e.sub.i]

 Estimated
 Coefficient
 (t-Statistic)

 Pre-TRA86 Post-TRA86 Combined
 Variable Sample Sample * Samples *

[d.sub.0] -0.24 (-1.19) -0.23 (-1.11) -0.23 (-1.62)
[d.sub.1] 0.05 (1.05) 0.03 (0.58) 0.03 (1.01)
[d.sub.2] 0.10 (6.76) ** 0.13 (8.00) ** 0.12 (10.44) **
Mean Square Model 80.77 107.63 185.41
Mean Square Error 3.21 3.22 3.21
Adjusted [R.sub.2] 0.22 0.29 0.26
Model [F.sub.2,165] 25.20 ** 33.39 ** 57.83 **

*--without outliers

**--p < 0.001

Chow Test Statistic: F(3,318) = .84, n.s.