The perceived fairness of taxing social security benefits: the effect of explanations based on...

ABSTRACT

In this study, we construct explanations for the taxation of social security benefits based on previously identified dimensions of fairness (exchange, horizontal, and vertical equity). We then conduct an experiment to examine whether providing senior citizen taxpayers with

explanations increases the perceived fairness of taxing social security. The results indicate that for those subjects with the greatest self-interest (subjects currently taxed on a portion of their social security benefits), the exchange equity explanation had the most consistent positive effects on both acceptance of the explanation and on the perceived fairness of taxing social security benefits. On the other hand, for those subjects not currently taxed on their social security benefits, the vertical equity explanation was more likely to be accepted than either the exchange or horizontal equity explanation. However, while these subjects agreed with the vertical equity explanation, it did not increase their fairness perceptions. These find ings illustrate how important it is for tax policy makers striving to increase perceptions of fairness to carefully consider and develop explanations for tax provisions.

INTRODUCTION

In the United States, a topic that generates extensive political debate and concern is how to increase perceptions of tax fairness. Thus, various ways of increasing the perceived equity of our tax system have received legislative and academic attention. One approach to increase perceived equity is to reform the tax system to make it fairer (Mastromarco 1998). Others have suggested that helping taxpayers better understand the present system may be the best way to increase perceived fairness (White et al. 1990; Christensen et al. 1994). Using this second approach, we specifically examine whether the perceived fairness of taxing social security benefits is affected by providing an explanation for the provision. As debates on the viability of the social security system continue, efforts to increase perceived fairness for taxing a portion of social security benefits continue to be important.

The effectiveness of explanations in changing perceptions of fairness was previously examined by providing taxpayers with the legislative justification for a specific tax law change (Wartick 1994). In general, she found evidence that providing an explanation affected perceived fairness, but the results also suggested that effectiveness might be dependent upon the nature of the explanation. Interestingly, although perceived fairness was the study's focus and fairness is often an argument advanced by the Treasury Department and Congress for specific tax provisions, neither of the legislative explanations examined were based on fairness. Building upon research that identifies multiple facets of equity (e.g., Jackson and Milliron 1986; Gerbing 1988; Christensen and Weihrich 1996), we examine the effect of three alternative explanations that appeal to various dimensions of equity on perceived fairness of taxing social security benefits. Since research has found that taxpayers' policy judgments are also based on ec onomic self-interest (e.g., Hite and Roberts 1991), we examine the effect of the three alternative explanations on the perceived fairness of taxing social security benefits for both taxpayers who are currently taxed on their social security benefits (high self-interest) and taxpayers who are not currently taxed (low self-interest). Thus, in addition to applying prior theoretical and empirical findings to the taxation of social security benefits, the results of this study contribute to the developing literature on how explanations affect perceived fairness.

To examine these issues, we asked a sample of senior citizens to review a brief description of the provision governing the taxation of social security benefits and to rate the fairness of this provision. Subjects were randomly assigned to one of four groups, with three of the groups receiving an explanation based on fairness arguments for the taxation of social security benefits (horizontal, exchange, or vertical equity (1)) and one group receiving no explanation. Data were also gathered from the subjects in the three explanation treatments to determine their agreement with the explanation provided.

The results indicate that for those subjects most directly affected by the social security tax provision, and thus with the greatest self-interest (subjects currently taxed on a portion of their social security income), the exchange equity explanation had the most consistent positive effects on both acceptance of the explanation and on the perceived fairness of taxing social security benefits. On the other hand, for those subjects not currently taxed on their social security income, the vertical equity explanation was more likely to be accepted than the exchange or horizontal equity explanations. However, while the subjects without taxable social security benefits agreed with the vertical equity explanation, it did not increase their perceived fairness of taxing social security benefits.

LITERATURE REVIEW AND HYPOTHESES

A number of studies have been conducted to examine whether information delivered to taxpayers can improve the perceived fairness of tax provisions. Several of these studies found a positive effect for general tax education on perceived fairness (White et al. 1990; Christensen et al. 1994). Other researchers have investigated the impact of exposure to brief, singular messages. Roberts (1994) found that subjects who viewed a public service announcement rated the tax system as fairer than subjects who did not view the announcement, with a cognitive announcement more effective than an affective announcement. McGraw and Scholz (1991) found similar results. Subjects who viewed a message designed to explain the principles underlying the Tax Reform Act of 1986 viewed the tax system as fairer than those who viewed either no message or a message designed to explain strategies for reducing taxes after the tax reform. Wartick (1994) examined whether providing subjects who were negatively affected by a tax law change with the legislative justification for the change mitigated perceived unfairness. The results were unambiguous for only one of the two tax provisions she examined. Although she did not test this directly, she suggests that the legislative justification for the repeal of the personal interest deduction did not affect the fairness ratings of adversely affected subjects because they did not accept the explanation as an adequate justification for the tax law change.

The idea that the explanation must be perceived as an adequate justification is consistent with referent cognitions theory (RCT). RCT is an extension of relative deprivation and distributive and procedural justice theories and predicts that when individuals experience outcomes less favorable than their own prior outcomes, they will judge the underlying procedures as unfair (Folger 1984, 1987). The perceived unfairness is mitigated, however, when an adequate justification for the procedure exists. In contrast, further research has shown that when fairness explanations are perceived not only as inadequate but as a tool of manipulation, the results of providing the explanation are markedly negative (as opposed to no effect) (Greenberg and Omstein 1983).

In summary, RCT predicts that an explanation must be perceived as an adequate explanation in order to have a positive effect on perceived fairness. Prior research on the effect of explanations used those published in the Joint Conference report from Congress (Wartick 1994). Wartick notes, however, that such explanations are the result of a process filled with compromises. In many cases, Congress would be able to justify a provision with a number of different explanations, each having a potentially different effect on the taxpayers' perception of the adequacy of the justification and consequently on their fairness perceptions. While it is clear that there are potentially numerous explanations available to justify tax provisions, fairness is often an argument advanced by the Treasury Department and Congress. Thus, a direct appeal to one of the various dimensions of fairness previously identified in the literature may be an effective means to positively affect taxpayers' fairness perceptions.

An early review by Jackson and Milliron (1986, 137) suggested that:

tax fairness seems to involve at least two dimensions. One dimension appears to involve the equity of the trade--the benefits received for the tax dollars given. The other dimension appears to involve the equity of the taxpayer's burden in reference to that of other individuals (i.e., taxpayers' perceptions of the horizontal and vertical equity of the tax system).

While more recent examinations of fairness have identified a larger number of dimensions (Gerbing 1988; Christensen and Weibrich 1996), they often include the two dimensions suggested by Jackson and Milliron (1986). For example, Christensen and Weihrich (1996) identify five dimensions as a result of performing factor analysis on their survey data. Although horizontal equity did not emerge as a dimension in their study, exchange equity with the government and special provisions for the wealthy (vertical equity) did. (2) While the previous literature suggests that taxpayers recognize the various types of tax equity, the results of these studies also reveal differences in the importance placed on each type of equity.

Within the tax domain, the multidimensionality of tax equity provides a setting where multiple explanations, all based on fairness, could be provided for a single provision. We chose to examine the issue within the context of the taxation of social security benefits for several reasons. First, pretests indicated that taxpayers view this provision as unfair. (3) Additionally, others have suggested that taxpayers are largely unaware of the detailed workings of the social security provision (Brannon 1993), so perceptions about the fairness of this provision may be sensitive to the information provided. Finally, several explanations are not only plausible for this particular provision, but have all been advanced as part of the legislative process, allowing us to compare the effect of different types of explanations. We developed three explanations for taxing social security benefits based on various congressional committee reports, each appealing to a different dimension of fairness. The language from the reports , as well as the explanations provided to subjects for each type of equity, appear in Table 1.

The first explanation is based on exchange equity and reflects the taxpayer's perception of the fairness of the exchange between the taxpayer and the government. The exchange equity perspective for taxation of social security benefits points out that the tax revenue collected from taxing social security benefits enhances the viability of the system. This appears to be an argument recently advanced, as extending the taxation of benefits is sometimes mentioned as one of the avenues for rescuing the social security system (Steuerle 1995).

The second explanation reflects an appeal to horizontal equity, or the idea that taxpayers with the same ability to pay taxes should be taxed the same. With regard to the taxation of social security benefits, the theoretical argument is that taxation is defensible because it equates the treatment of social security benefits to other similar payments (e.g., pensions) that a taxpayer might receive.

The third dimension of fairness examined is vertical equity, or the value judgment concerning the relative tax burdens of taxpayers with differing abilities to pay. Explanations for the taxation of social security benefits based on vertical equity typically explain that taxation of benefits does not apply to low-income taxpayers (see, for example, Pollack 1994).

While theory does not provide the basis to make a directional hypothesis on the relative effectiveness of the three alternative explanations, past empirical results eliciting fairness dimensions and the relative importance placed on each (e.g., Gerbing 1988; Christensen and Weihrich 1996) provide the basis for predicting that the three explanations will not be equally accepted and therefore may have differing effects on fairness perceptions. Stated formally in alternative form:

H1a: The three legislative explanations for taxing social security benefits based on exchange equity, horizontal equity, and vertical equity will have differing effects on taxpayers' agreement with the explanation provided and the perceived fairness of taxing social security benefits.

As described earlier, we chose to examine the social security tax provision for a variety of reasons. An additional advantage of examining this provision is that since some taxpayers are subject to taxation on their benefits, while others are not, this provision also allows us to examine the effects of self-interest on fairness perceptions. Self-interest may be an important consideration since prior research has found that taxpayers' policy judgments are based on economic self-interest as well as fairness considerations (e.g., Hite and Roberts 1991). It could be argued that those subjects who currently have a portion of their social security benefits subject to taxation have the greatest self-interest and may react differently to the three alternative equity explanations than those subjects with a lesser self-interest (benefits are not currently subject to taxation). Thus, the following additional hypothesis is proposed:

H1b: The taxpayers' agreement with the explanation provided and perceived fairness of taxing social security benefits will differ depending upon whether their social security benefits are subject to taxation.

While H1b and H1b compare the relative effectiveness of the three equity explanations, it is also important to examine the effectiveness of the explanations compared to a control group that does not receive any explanation. RCT would predict that if the subjects perceive the explanation as an adequate justification for taxing social security benefits, fairness perceptions might be increased by the explanation. However, other research (Greenberg and Ornstein 1983) suggests that if the subjects do not accept the explanation and in fact perceive the explanation as manipulative, the effect of the explanation may be negative. It is unclear a priori whether the subjects will accept or not accept each of the three equity explanations. Further, it is likely that acceptance of an explanation may also be affected by the subjects' self-interest with the tax provision (i.e., current taxability of benefits). Thus, the following hypotheses compare the effect of the three equity explanations on the perceived fairness of tax ing social security vs. a control group that does not receive an explanation:

H2a: Taxpayers who are provided with an explanation for taxing social security benefits will perceive the fairness of taxing social security benefits differently than those taxpayers who are not provided with an explanation.

H2b: The taxpayers' perceived fairness of taxing social security benefits will also differ depending upon whether their social security benefits are subject to taxation.

METHOD

Subjects

Senior citizens who were members of a professional men and women's club at a senior center in the Northeast participated in this study. The subjects were asked to review a brief description of the tax law with regard to the taxation of social security benefits, and then provide fairness perceptions related to this provision. A total of 123 subjects were asked to complete the research instrument, resulting in 93 usable responses. (4) Demographic data collected from the subjects and summarized in Table 2 suggest that they were experienced taxpayers, both in regard to the social security tax provision and the overall tax system. Specifically, 100 percent of the subjects had previously received social security income and 46 percent of the subjects have a portion of their social security benefits subject to taxation. Approximately 28 percent reported that they normally prepare their own tax return; 65 percent either use a paid preparer or receive help from a friend or relative; and 3 percent indicated that they ha d never filed a federal tax return (4 percent of the subjects were unsure or did not respond to the question). Females comprised 61 percent of the sample; 39 percent were males. The median age group was 75-84 years old.

Task and Procedures

All subjects were first presented with a brief, simplified explanation of the mechanics for determining the portion of social security income that is taxable for a single and a married taxpayer. In addition, the subjects were randomly assigned to one of four conditions. In three of the conditions, subjects received an explanation justifying the provision governing the taxation of social security benefits based on either exchange, horizontal, or vertical equity. Subjects in the fourth condition received no explanation. The information that the subjects received was identical, except for differences in the wording of the explanation provided to the subjects in the three explanation treatments (as shown in Table 1). Each of these explanations was similar in terms of length, containing slightly over 50 words.

In order to test the research hypotheses, the participants were asked to rate the fairness of taxing social security benefits. The fairness perception for the taxation of social security benefits was measured on a seven-point Likert scale with endpoints labeled "extremely unfair to tax" (-3) and "extremely fair to tax" (+3), and the midpoint labeled "neither fair nor unfair" (0). After completing the instrument, subjects completed a background questionnaire. As part of this questionnaire, subjects who received an explanation for the taxation of social security benefits were asked to rate their acceptance of the explanation by indicating their extent of agreement with the explanation. Agreement was measured using a seven-point Likert scale with endpoints labeled "Strongly Agree" (1) and "Strongly Disagree" (7). All of the subjects in our experiment were also asked if a portion of their social security income was subject to taxation. Forty-three subjects responded yes to this question, with the remaining 50 sub jects responding no.

An executive director from the senior center administered the instrument to the subjects. (5)

To help ensure that subjects would not feel forced to participate and to minimize demand effects, the executive director stressed that participation was voluntary, anonymity was assured, and subjects could withdraw from the study at any time. In addition, the subjects were advised that if a question made them feel uncomfortable, they did not have to respond to that question. Most subjects completed the instrument in about 30 minutes.

RESULTS

Relative Effectiveness of Different Equity Explanations (H1a and H1b)

Hypothesis 1a predicts that the three provided explanations will have differential effects on the taxpayers' agreement with the legislative explanation provided and the perceived fairness of taxing social security. Hypothesis 1b further predicts that these perceptions will also differ depending upon whether the subject's social security income is subject to taxation or not. Because the two dependent variables related to H1a and H1b are highly correlated (Pearson Correlation Coefficient = -.336, p = .005), a MANOVA was used to test these hypotheses. The perceived fairness of taxing social security benefits and the agreement with the explanation provided are the two dependent variables. The independent variables represent the type of explanation received (Exchange, Horizontal, Vertical) and whether the subject's social security income is subject to taxation (taxable, nontaxable).

Using data from the subjects in the three explanation conditions, the MANOVA results reported in Table 3 indicate a significant main effect for the type of explanation provided (F = 5.56, p = .000). There was no main effect for self-interest (whether the subject's social security was subject to taxation); however, the interaction between the type of explanation provided and whether the subject's social security was subject to taxation (Explanation x Taxable) was significant (F = 5.76, p = .000). (6)

To further examine these findings, we ran separate univariate ANOVAs for each dependent variable. For agreement with the explanation as the dependent variable, the type of explanation had a significant main effect (F = 4.06, p = .022) while the main effect for the taxability of the subject's social security income was not significant. However, the interaction of the two variables was again significant (F = 7.59, p = .001). Similarly, when the univariate ANOVA was run with the perceived fairness of taxing social security income as the dependent variable, once again, only the main effect for the type of explanation and the interaction between the type of explanation and the taxability of social security had significant effects (F = 6.42, p = .003 and F = 4.50, p = .015, respectively). These results support the first two hypotheses since the explanations were found to have differential effects on the taxpayers' agreement with the legislative explanation provided and the perceived fairness of taxing social securi ty (H1a), depending upon whether the subject's social security income is subject to taxation (H1b).

To better understand the differential effects of the explanations and interpret the significant effect of the interaction we examine the effect of the explanations separately on those taxpayers with and without taxable social security benefits. Panels B and C of Table 3 provide the means and standard deviations for the preceived fairness of taxing social security and the mean agreement with the explanation provided by tax status (taxable or nontaxable). To analyze the different effects of the explanations on agreement with the explanation provided and the perceived fairness of taxing social security, we ran comparisons between each of the explanations, using Scheffe tests to control for Type I errors due to multiple comparisons, by tax status.

As indicated in Panel B of Table 3, subjects with taxable social security benefits were more likely to accept the exchange equity explanation (mean 4.00) than either the horizontal equity (mean = 6.13) or the vertical equity explanations (mean = 6.11). The comparisons indicate that these differences were significant (p = .004 and p = .012 for exchange vs. horizontal and vertical equity, respectively). As expected from the similar means, the horizontal and vertical equity explanations did not differ in their acceptance. Further, with a mean of -.56, the exchange equity explanation had a more favorable effect on the perceived fairness of taxing social security income than both the horizontal equity (mean = -2.73) and the vertical equity explanations (mean = -2.78). The comparisons once again indicate that these differences were significant (p = .002 and p = .004 for the exchange vs. horizontal and vertical explanations, respectively). There were no differences between the horizontal and vertical equity explanat ions. In summary, the exchange equity explanation was the most effective of the three explanations for those taxpayers with taxable social security benefits.

Analysis restricted to those subjects who are not subject to taxation on their social security benefits provides differing results. For these subjects, the results of the comparisons indicate that subjects were more likely to accept the vertical equity explanation than both the exchange equity (mean of 3.69 vs. 5.64, p = .054) and horizontal equity explanations (mean of 3.69 vs. 5.92, p = .017). The exchange and horizontal equity explanations did not differ in their acceptance. However, the vertical equity explanation did not have a significant positive effect on the perceived fairness of taxing social security benefits. The comparisons indicate that the effect on the perceived fairness of taxing social security income for the vertical equity explanation was no different than the exchange equity explanation (mean of -2.69 vs. -1.82, p = .361) or the horizontal equity explanation (mean of -2.69 vs. -1.54, p = .151). There were also no differences between the horizontal and exchange equity explanations.

In summary, while the subjects without taxable social security benefits were more likely to accept the vertical equity explanation than the other two equity explanations, the vertical equity explanation did not have a significant positive effect on the perceived fairness of taxing social security benefits. Thus, while agreement with an explanation may be a necessary condition for having a positive effect on fairness perceptions, it may not be a sufficient condition. Further, these results provide some general guidance to tax policy makers since they suggest that policy makers should focus on crafting acceptable explanations for taxpayers most directly affected by a tax provision, and not be as concerned about those taxpayers who are not directly affected.

Comparison of Equity Explanations to No Explanation (H2a and H2b)

Hypothesis 2a predicts that the perceived fairness of taxing social security benefits will differ between those taxpayers who are provided with a legislative explanation for the taxation of social security and those who are not. Hypothesis 2b further predicts that this fairness perception will also differ depending upon whether the subject's social security income is subject to taxation. An ANOVA model was used to test the two hypotheses. The perceived fairness of taxing social security benefits serves as the dependent variable. The independent variables represent the type of explanation received (Exchange, Horizontal, Vertical) or not received (Control), and whether the subject's social security income is subject to taxation (taxable, nontaxable).

The ANOVA results reported in Table 4 (Panel A) indicate a significant main effect for the type of explanation provided (F = 4.08, p = .009). The main effect of the taxability of the subject's social security income was not significant, but as with the earlier analysis, the interaction between the type of explanation provided (or not provided as in the control group) and whether the subject's social security income was subject to taxation was significant (F = 2.90, p=.04). (7) To further investigate this interaction, Panels B and C of Table 4 report the results of comparisons between each of the explanations and the control group. Dunnet's post hoc comparison method was used since this method is specifically designed for comparing experimental groups against a control group (Shavelson 1996).

Panel B of Table 4 compares the mean perceived fairness of taxing social security for each of the explanations to the control group for those subjects with taxable social security. The mean of the perceived fairness of taxing social security for the exchange equity explanation was -.56 while the mean of the control group was -1.80. However, the comparison indicates that this difference was not significant (p=.117). Further, the perceived fairness of taxing social security for both the horizontal equity explanation (mean of -2.73) and the vertical equity explanation (mean of -2.78) were not significantly different than the control group (p = .215 and p = .265, respectively). (8)

Panel C of Table 4 presents the comparisons for those subjects without taxable benefits. Similar to the subjects with taxable benefits, there are no significant differences between the control group and any of the three provided explanations.

CONCLUSIONS AND IMPLICATIONS

While prior research suggests that providing taxpayers with explanations for tax provisions may affect fairness perceptions, other research has found that these perceptions are also based on economic self-interest. We extend these two areas of research by examining the effectiveness of explanations based on various dimensions of equity. The explanations for the taxation of social security benefits used in this study were based on exchange equity, horizontal equity, or vertical equity. Further, since the subjects in our experiment had varying degrees of self-interest related to the social security provision (some were subject to a tax on their social security benefits, while others were not), we are also able to examine the joint effect of self-interest on fairness perceptions.

The experimental results suggest two primary findings. First, for those subjects with the greatest self-interest (currently taxed on a portion of their social security income), the exchange equity explanation had the most consistent positive effects on both acceptance of the explanation and on the perceived fairness of taxing social security benefits. For taxpayers with taxable social security benefits, the exchange equity explanation could be viewed as the only explanation that provides both a fairness and economic self-interest rationale. Thus, future tax researchers should examine whether explanations constructed on both fairness and economic self-interest may be an effective strategy for increasing taxpayers' acceptance of tax provisions that have directly affected them.

Our second primary finding relates to the subjects without taxable social security benefits. These subjects were more likely to accept the vertical equity explanation than either the exchange or horizontal equity explanations. This result further demonstrates the importance of considering self-interest in studying fairness perceptions. Similar to the exchange equity explanation for those subjects with taxable social security benefits, the vertical equity explanation was the only one of the three fairness explanations that provided the subjects without taxable social security benefits with an economic self-interest rationale. However, while the subjects without taxable social security benefits may have agreed with the vertical equity explanation, it did not increase their perceived fairness of taxing social security benefits. As described earlier, this result suggests that while agreement with an explanation may be a necessary condition for having a positive effect on fairness perceptions, it may not be a suff icient condition.

As with any experimental study, these findings must be interpreted in light of the limitations to which the study is subject. The first limitation relates to the subjects used in this study. Since the tax provision examined in our study was the social security tax provision, we used senior citizens as subjects. The decision to use this subject pool was designed to enhance both the internal and external validity of the study, since these taxpayers are the primary group currently affected by this provision. However, since the age and income of our subjects are not representative of the general taxpayer population, findings from this study may not generalize to other groups of taxpayers. Further, other variables that may affect perceptions of fairness, like total income and education level, were not measured as part of our research instrument. A second limitation relates to the relatively small sample sizes in some of our cells. Because we randomly assign participants to treatments and then partition the sample according to whether their social security benefits are subject to income taxes, we are unable to ensure equal sample sizes. Thus, the resulting smaller sample sizes in some cells and the relatively conservative tests for multiple comparisons reduce the likelihood of finding significant differences. Future researchers may wish to explore the effects of the various explanations using more powerful research designs. A third limitation is that we examine a single source of income. Ideally, we would have been able to ask the subjects for perceptions of fairness for all the items they face on their tax return. However, given the variety of items on returns it would have been difficult to administer our treatments with each of the various provisions that taxpayers might face. Nevertheless, future researchers could extend this literature by examining the effects of explanations for a broader range of provisions. Finally, we examine fairness perceptions using a provision for the taxation of income. Since some studies have shown that taxpayers respond differently to income items than deductions, our results may not generalize to deduction provisions (Harris and Associates 1988, 75-76).

Despite these limitations, the study offers several implications for policy makers and researchers. While some researchers found an increase in fairness perceptions following the Tax Reform Act of 1986 (Kinsey and Grasmick 1993), policy makers are once again beginning to suggest that changes to the current tax system are needed to address the negative perceptions of taxpayers. For example, Donmoyer (1998) recently reported that the negative perceptions of taxpayers helped ensure the passage of the IRS Restructuring and Reform Act of 1998. Given the difficulty and cost of addressing these perceptions with tax legislation, some IRS officials have previously tried direct appeals to specific groups of taxpayers (Donmoyer 1997). The results of this study suggest that for senior citizens who are directly affected by the social security tax provision, an appeal to the exchange equity dimension may offer a successful strategy for increasing fairness perceptions. This finding may have increasing importance if extendin g the taxation of social security benefits once again is recommended for rescuing the social security system (Steuerle 1995). Thus, more senior citizen taxpayers may be directly affected by the social security tax provision in the future. Also, while other tax researchers have previously shown that messages targeted to taxpayers can be effective (see, for example, Roberts [1994] for the potential use of public service announcements and Wartick [1994] for the potential use of legislative explanations), the present study suggests that efforts of this type may not be effective if taxpayers are not affected by the tax provision.

TABLE 1

EXPLANATIONS BASED ON TYPES OF EQUITY

Type of
Equity      Language from Congressional Report

Exchange    ...by taxing social security
Equity       benefits and appropriating these
             revenues to the appropriate trust
             funds, the financial solvency of
             the social security trust funds
             will be strenthened. (U.S. Senate
             Finance Committee 1983, 25)









Horizontal  The committee believes...that
 Equity      taxing a portion of social
             security benefits will improve
             tax equity by treating more
             nearly equally all forms of
             retirement and other income that
             are designed to replace lost
             wages.... (U.S. Senate Finance
             Committee 1983, 25)







Vertical    ...increasing the maximum amount
Equity       of Social Security benefits
             included in gross income for
             certain higher income
             beneficiaries...will enhance both
             the horizontal and vertical equity
             of the individual tax system...To
             limit the effect of this provision
             to taxpayers with a greater
             ability to pay taxes, the present-
             law income thresholds are
             maintained. (U.S. House Budget
             Committee 1993, 654)


Type of
Equity      Explanation Provided to Subjects

Exchange    Congress decided to tax social
Equity       security income for some taxpayers
             in an effort to enhance tax
             equity. Congress believes that the
             additional tax revenues generated
             from the taxation of a portion of
             social security benefits can be
             used to strengthen the financial
             solvency of the social security
             trust funds. Thus, by taxing a
             portion of social security
             benefits, the likelihood that
             retired taxpayers will continue to
             receive their social security
             benefits is increased.

Horizontal  Congress decided to tax social
 Equity      security income for some taxpayers
             in an effort to enhance tax
             equity. Congress believes that
             social security benefits are
             intended to replace lost wages,
             which is very similar to other
             forms of retirement income (for
             example, a pension from an
             employer). Since these other forms
             of retirement income are subject
             to tax, Congress believes that a
             portion of social security
             benefits should also be subject to
             tax.

Vertical    Congres decided to tax social
Equity       security income for some taxpayers
             in an effort to enhance tax
             equity. Congress believes it is
             appropriate that taxpayers with
             higher incomes pay tax on a
             portion of their social security
             benefits. Congress also believes
             that it is not appropriate for
             lower-income taxpayers to pay such
             a tax. Therefore, Congress
             established the income thresholds
             described above in order to tax
             higher-income taxpayers.

TABLE 2

DEMOGRAPHIC DATA FOR SAMPLE

                                  Percentage of Participants (a)

Age
  No Response                                  2
  65-74                                       38
  75-84                                       50
  over 85                                     11

Gender
  Female                                      61
  Male                                        39

Previously Received Social
 Security Income
  Yes                                        100

Taxable Portion of Social Income
  No                                          54
  Yes                                         46

Amount of Social Security Income
  No Response                                  3
  $0-$4,999                                   19
  $5,000-$9,999                               41
  >$10,000                                    37

Previous Tax Returns
  Prepare own return                          28
  Prepared by spouse, friend, or
   relative                                   14
  Use paid preparer                           51
  Never filed                                  3
  Unsure/no response                           4

(a) Totals may not add to 100 percent due to rounding.

TABLE 3

MANOVA RESULTS FOR AGREEMENT WITH EXPLANATION AND PERCEIVED FAIRNESS OF
TAXING SOCIAL SECURITY FOR RETIRED TAXPAYERS (COMPARISON OF THREE
EXPLANATIONS)

Panel A: Manova Results

MANOVA (n = 70)
Effect                 df  Wilk's Lambda     F  Prob.

Explanation            4        .72       5.56  .000
Taxable                2        .99        .37  .691
Explanation x Taxable  4        .72       5.76  .000

Panel B: Means and Standard Deviations for Subjects with Taxable Social
Security

Agreement with Explanation: Explanation

             n   Mean  S.D.

EXCHANGE     9   4.00  2.06
HORIZONTAL  15   6.13   .99
VERTICAL     9   6.11  1.17
Total       33   5.55  1.66

Fairness of Taxing Social Security: Explanation

             n   Mean  S.D.

EXCHANGE     9   -.56  2.35
HORIZONTAL  15  -2.73   .59
VERTICAL     9  -2.78   .44
Total       33  -2.15  1.60

Panel C: Means and Standard Deviations for Subjects without Taxable
Social Security

Agreement with Explanation: Explanation

             n   Mean  S.D.

EXCHANGE    11   5.64  1.86
HORIZONTAL  13   5.92  1.44
VERTICAL    13   3.69  2.25
Total       37   5.05  2.09

Fairness of Taxing Social Security: Explanation

             n   Mean  S.D.


EXCHANGE    11  -1.82  1.78
HORIZONTAL  13  -1.54  1.66
VERTICAL    13  -2.69   .85
Total       37  -2.03  1.52

TABLE 4

COMPARISONS WITH THE CONTROL GROUP FOR THE PERCEIVED FAIRNESS OF TAXING
SOCIAL SECURITY

(BY TAX STATUS)

Panel A: ANOVA Model (n=93)

Dependent Variable: Fairness of Taxing Social Security

Source                 df  Sum of Squares  Mean Square   F    Prob.

Explanation            3       25.31          8.44      4.08  .009
Taxable                1         .00           .00       .02  .881
Explanation x Taxable  3       18.01          6.01      2.90  .040

Panel B: Comparisons with the Control Group for Subjects with Taxable
Social Security

Experimental Group  n   Mean   S.D.  Comparison  Significance (a)

CONTROL             10  -1.80  1.40
EXCHANGE             9   -.56  2.35    -1.24         .117
HORIZONTAL          15  -2.73   .59      .93         .215
VERTICAL             9  -2.78   .44      .98         .265

Total               43  -2.07  1.55

Panel C: Comparisons with the Control Group for Subjects without Taxable
Social Security

Experimental Group  n   Mean   S.D.  Comparison  Significance (a)

CONTROL             13  -2.00  1.68
EXCHANGE            11  -1.82  1.78     -.18         .984
HORIZONTAL          13  -1.54  1.66     -.46         .786
VERTICAL            13  -2.69   .85      .69         .526

Total               50  -2.02  1.55

(a) Significance levels are based on Dunnet's post hoc comparison
method.

Submitted: February 1999

Accepted: March 2002

(1.) Musgrave and Musgrave (1989, 223) state that "taxation according to ability to pay calls for people with equal capacity to pay the same, and for people with greater ability to pay more. The former is referred to as horizontal equity and the latter as vertical equity." Exchange equity has been described as "the taxpayer's perception of the equity of the exchange between him and the government, i.e., between the taxes he pays and the benefits he receives" (Mason and Calvin 1978, 78). The three explanations used in this study were developed consistent with wording from the Senate Finance Committee (1983) and House Budget Committee (1993) reports (see Table 1) regarding the taxation of social security benefits. Thus, while the resulting explanations have a high degree of external validity, they may not represent the best possible explanations consistent with the respective dimensions of fairness.

(2.) The other three dimensions identified by Christensen and Weibrich (1996) are more closely related to the overall tax system than an individual provision, making them less relevant to the intervention tested with the present study. These dimensions include: (1) tax rate structure (2) personal fairness, and (3) overall fairness.

(3.) One hundred one part-time adult students completed a pretest instrument that asked them to rate eight provisions using a six-point Likert fairness scale ranging from -3 to +3. The mean fairness rating for taxing social security was -1.35, which was the lowest mean rating of the eight provisions and significantly less than all other provisions.

(4.) Thirty of the participants did not provide complete data as to their agreement with the explanation provided, perceived fairness of taxing social security benefits, or whether their social security benefits were subject to taxation. Since the MANOVA model used to examine H1a and H2b requires complete data for each of these variables, and the ANOVA model used to examine H2a and H2b requires complete data for the perceived fairness of taxing social security benefits and the taxability of social security benefits, these subjects could not be included.

(5.) We pilot-tested the research instrument with four senior citizens from the senior center. After making changes suggested by these participants, we were advised that the research instrument was clear and should be understood and completed by the participants in the study. We were also advised that the participants in the study would feel more at ease if the director of the senior center, rather than one of the researchers, administered the instrument.

(6.) Additional analyses were performed using the subjects' responses to the demographic and additional questions as covariates in the MANOVA model used to examine H1a and H1b. The demographic variables included age, gender, amount of social security income, political party, tax return preparer, and prior tax training. None of these variables were significant as covariates. However, two of the additional questions were significant when included as covariates. The first question asked the subjects, "For what you receive in benefits and services from the federal government, your income taxes are ...." The subjects were asked to respond to this question on a seven-point Likert scale with endpoints labeled "Far too high" (1) and "Far too low" (7). The second question asked the subjects, "In general, how concerned are you that social security benefits will be substantially reduced during your retirement?" The subjects were asked to respond to this question on a seven-point Likert scale with endpoints labeled "Not Concerned" (1) and "Very Concerned" (7). However, the results and conclusions for the relative effectiveness of the three equity explanations do not change, and in fact are substantially similar when the responses to these two questions are included as covariates. Therefore, since the inclusion of these two questions as covariates reduces the sample size for H1a and H1b from 70 to 61 subjects, the primary analysis presented in Table 3 does not include these covariates.

(7.) Additional analyses were performed using the subjects' responses to the demographic and additional questions (described in footnote 6) as covariates in the ANOVA model used to examine H2a and H2b. The only significant covariate was the subjects' response to the question "For what you receive in benefits and services from the federal government, your income taxes are...." Since the inclusion of this covariate reduces the sample size for the tests for H2a and H2b and the results do not change, the primary analysis presented in Table 4 does not include this covariate.

(8.) As explained in the literature review, previous literature suggests that acceptance of an explanation as an adequate justification is a requisite condition for a positive effect on fairness perceptions. Further, an explanation that is perceived as inadequate and a tool of manipulation may have a negative effect on fairness perceptions. However, the previous tax literature on explanations does not provide a theoretical foundation to predict the explanations used in the current study with which subjects would agree or disagree. As a result, the post hoc comparisons reported here are two-tailed tests that also control for the effect of Type I errors across three comparisons. These conditions produce relatively conservative tests that would require large differences to be significant. While the differences in means for fairness in the current study are not large enough to be significant under these conditions, future researchers may wish to reexamine this issue with more powerful research designs.

REFERENCES

Brannon, G. 1993. The strange precision in the taxation of social security benefits. Tax Notes (March 29): 1803-1804.

Christensen, A., S. Weihrich, and M. Gerbing Newman. 1994. The impact of education on perceptions of tax fairness. Advances in Taxation 6: 63-94.

-----, and -----. 1996. Tax fairness: Different roles, different perspectives. Advances in Taxation 8: 27-61. Donmoyer, R. 1997. Sophisticated taxpayers obliged to behave honorably, Richardson says. Tax Notes (March 24): 1507-1508.

-----. 1998. With last blast of fanfare, Clinton signs IRS reform. Tax Notes (July 27): 415-416.

Folger, R. 1984. Perceived injustice, referent cognitions, and the concepts of comparison level. Representative Research in Social Psychology 14 (2): 88-108.

-----. 1987. Reformulating the preconditions of resentment: A referent cognitions model. In Social Comparison, Social Justice, and Relative Deprivation, edited by J. C. Masters, and W. P. Smith. Hillsdale, NJ: Lawrence Erlbaum Associates.

Gerbing, M. 1988. An empirical study of taxpayer perceptions of fairness. Ph.D. dissertation, The University of Texas at Austin.

Greenberg, J., and S. Ornstein. 1983. High status job title compensation for underpayment: A test of equity theory. Journal of Applied Psychology 68 (May): 285-297.

Harris and Associates. 1988. 1987 Taxpayer opinion survey. Doc. 7292, 1-88. Washington, D.C.: Department of the Treasury.

Hite, P., and M. Roberts. 1991. An experimental investigation of taxpayer judgments on rate structure in the individual income tax system. The Journal of the American Taxation Association 13: 47-63.

Jackson, B., and V. Milliron. 1986. Tax compliance research, findings, problems, and prospects. Journal of Accounting Literature 5: 125-166.

Kinsey, K., and H. Grasmick. 1993. Did the Tax Reform Act of 1986 improve compliance? Three studies of pre and post-TRA compliance attitudes. Law and Policy 15 (4): 293-325.

Mason, R., and L. Calvin. 1978. A study of admitted income tax evasion. Law and Society Review 13 (Fall): 73-89.

Mastromarco, D. 1998. The "fair tax" and tax compliance: An analytical perspective. Tax Notes (April 20): 379-387.

McGraw, K., and J. Scholz. 1991. Appeals to civic virtue versus attention to self-interest: Effects on tax compliance. Law and Society Review 25: 471-498.

Musgrave, R., and P. Musgrave. 1989. Public Finance in Theory and Practice, New York, NY: McGraw-Hill Book Company.

Pollack, 5. 1994. Farewell to tax reform: The 1993 Tax Act in historical perspective. Tax Notes (August 22): 1081-1091.

Roberts. M, 1994. An experimental approach to changing taxpayers' attitudes towards fairness and compliance via television. The Journal of the American Taxation Association 16 (Spring): 67-86.

Shavelson, R. A. 1996. Statistical Reasoning for the Behavioral Sciences. Needham Heights, MA: Simon and Schuster.

Steuerle, G. 1995. Taxing social security recipients. Tax Notes (January 30): 739-740.

U.S. House Budget Committee. 1993. Committee Report on the Omnibus Budget Reconciliation Act of 1993. House Report No. 103-111. Washington, D.C.: U.S. Government Printing Office.

U.S. Senate Finance Committee. 1983. Committee Report on the Social Security Amendments of 1983. Senate Report No. 98-23. Washington, D.C.: U.S. Government Printing Office.

Wartick, M. 1994. Legislative justification and the perceived fairness of tax law changes: A referent cognitions theory approach. The Journal of the American Taxation Association 16 (Fall): 106-123.

White, R., A. Curatola, and W. Samson. 1990. A behavioral study investigating the effect of knowledge of income tax laws and tax policy on individual perceptions of federal income tax fairness. Advances in Taxation 3: 165-185.

James J. Maroney and Timothy J. Rupert are Associate Professors, both at Northeastern University, and Martha L. Wartick is an Associate Professor at the University of Northern Iowa.

The authors gratefully acknowledge comments from the editor, the two anonymous reviewers, John A. Barrick, Jean C. Bedard, Carol M. Fischer, Michael L. Roberts, Louise E. Single, and participants from the 1999 Northeast Regional AAA meeting, the 1999 Annual Meeting of the American Accounting Association and the 2000 ABO Conference.