Do Fairness Perceptions Matter in Tax Compliance? The Effect of Procedural × Distributive Justice Interaction on Intention to Pay Taxes

This study examined the interactive effect of procedural and distributive justice on intention to pay taxes. Data for the study were gathered through a cross-sectional survey administrated to category C taxpayers in Addis Ababa and some surrounding Oromia zones. A hierarchical multiple regression and PROCESS macro regression analyses results revealed that taxpayers’ intention to voluntarily comply with tax obligations depends on their perceptions of distributive justice and their emotional attachment with the tax authorities. Moreover, procedural justice was found to be a moderator of the effect of distributive justice on intention to voluntarily pay taxes.

both forms of the Procedural Justice × Distributive Justice interaction lead to high procedural justice or high distributive being sufficient to lead to positive responses to the authority in the form of will to defer to the authority's decisions. In other words, negative responses appear only when procedural justice and distributive justice are both low.
Various theories explain the effect of the Procedural Justice × Distributive Justice interaction on responses to authorities and the collective they represent. More specifically, referent cognitions theory (Folger, 1987), the instrumental model of justice (Thibaut and Walker, 1975), and the group value model (Lind and Tyler, 1988) offer explanations as to why the Procedural Justice × Distributive Justice interaction predicts cooperative responses of individuals to collective authorities. Brockner and Wiesenfeld (1996) and De Cremer et al. (2010) note that the explanations by the aforementioned theories can be understood as emphasizing a sense-making process. The argument of the sense-making process is that individuals interpret and assign meaning to events and encounters, mainly in response to negative, unexpected, or ambiguous circumstances (see Jones and Skarlicki, 2013;Weick, 1995). Before they assign meaning, people consider other aspects of their encounters. For example, unfair (vs. fair) outcomes are perceived as negative; this makes people want to make sense of their surroundings by asking the following question: "Can I understand something about why this negative experience materialized and how bad the future will look like?" Procedural justice offers a partial answer to this question, in the sense that low procedural justice conveys information that unfair outcomes are results of an unfair decision-making procedure and are thus more likely to continue into the future. Fair procedures, on the other hand, communicate that the unfair outcomes result from chance, or at least from a less stable factor, and are therefore less likely to continue into the future (Seidikides, De Cremer, Hart & Brebels, 2010). This leads to a relatively strong positive effect of distributive justice on responses to authorities when procedural justice is low (vs. high). The above arguments culminate in the following hypothesis. High procedural justice leads to high intention of taxpayers to voluntarily comply with taxation when distributive justice is low (vs. high).
This study explores the interactive effect of the two justice components on intention to voluntarily comply with taxation, controlling for some demographic variables and trust in the tax authority, as shown in figure 1 below.

Method 3.1 Respondents
Data were gathered from a sample of 761 category "C" taxpayers (i.e., taxpayers whose tax liabilities are determined on the basis of standard assessments) in Addis Ababa and surrounding Oromia zones. Given that the tax authority does not allow any third party to access taxpayers' information and that it is impossible obtain the number of Category "C" taxpayers, it was not possible to use probability sampling techniques. Consequently, a convenient sample of 812 taxpayers (412 from Addis Ababa City administration and 400 from Oromia zones around Addis Ababa) were drawn to fill a self-administered questionnaire. In respect of the risk of nonrepresentativeness of the sample, the large sample size and the fact that behavioral studies often utilize convenience Procedural justice Intention to comply with taxation  Gender  Age  Education  Income level  Business years  Trust in the tax authority Distributive justice sampling without serious threat to their conclusions (see Leary, 2012) guarantee the validity of the results drawn from this study.
The study sample consisted of 716 category "C" taxpayers in the Ethiopian capital, Addis Ababa and Oromia zones surrounding Addis Ababa, namely Lege Tafo, Sululta, and Burayu. A questionnaire prepared in English and translated into Amharic language was administered to the respondents. Of the respondents, 53.5% were male, 37.8% were female, and 8.8% did not indicate their gender. Respondents were between 20 and 60 years of age (M age = 39.8, SD age = 8.23). With regard to the highest level of education, 23.5% had completed elementary education only, 21% had completed high school education, 21.8% had a 2-year college diploma, 13.3% had a bachelor's degree, 2% had completed a master's degree and above, and 18.4% did not indicate their highest level of education. With respect to their annual earnings, 42.6% of the respondents reported that they had annual earnings of 20,000-40,000 Ethiopian Birr, 12.8% had earned 40,000-60,000 Birr, 9.7% had earned 60,000-80,000 Birr, 7% had earned 80,000-100,000 Birr, 8.7% had earned 100,000-120,000 Birr, 6.4% had earned more than 120,000 Birr, and 12.8% did not indicate their annual earnings. Among the respondents,24.8% reported that they had less than 2 years of experience with the tax authority, 27.8% had 2-6 years, 20% had 6-10 years, 12.1% had 10-20 years, 9.1% had more than 20 years of experience, and 6.2% did not disclose their years of experience with the tax authority. The demographic control variables were coded as: (a) income range of respondents, 1 = 20,000-40,000 Birr, 2 = 40,000-60,000 Birr, 3 = 60,000-80,000 Birr, 4 = 80,000-100,00 Birr, 5 = 100,000-120,000 Birr, and 6 = more than 120,000 Birr; years of experience with the tax authority, 1 = less than 2 years, 2 = 2-6 years, 3 = 6-10 years, 4 = 10-20 years, and 5 = more than 20 years; highest level of education attained, 1 = completed elementary school, 2 = completed high school, 3 = completed college diploma, 4 = completed first degree, 5 = completed master's degree, and 6 = completed a PhD.

Procedure
The study employed a printed questionnaire and unstructured interview to collect data. The data were collected the over a period of three months, January-March, 2018. A total of 812 questionnaires were distributed to category "C" taxpayers in Addis Ababa and three Oromia zones surrounding Addis Ababa. Accompanying the questionnaire was a cover letter that explains the intent of the study and guarantees strict confidentiality of responses. Seven hundred and ninety-six questionnaires were returned. Of these, thirty-five respondents skipped a significant number of questions and were therefore removed from the dataset. Accordingly, a total of 761 usable questionnaires were included in the analysis.

Measures
Procedural justice was measured with a 7-item scale developed and validated by Colquitt (2001). The study preferred to use this scale over scales that were used before in the tax compliance literature because those other scales include items that do not accurately refer to aspects of decision-making procedures, and often overlap with other constructs. For instance, Murphy's (2004) measure of procedural justice of the tax authority contains items that measure trust. An example item that overlaps with trust is "The tax office treats people as if they can be trusted to do the right thing." The Colquitt (2001) measure, however, is exclusively based on Leventhal's (1980) and Thibaut & Walker's (1975) concept of procedural justice and measures the most common procedural justice criteria: voice, bias suppression, accuracy, consistency, and adherence to norm. Therefore, this measure is an improvement over existing procedural justice measures in the tax compliance literature. The Colquitt (2001) items were slightly adapted to fit the context of procedural justice of the tax authority. Item examples (preceded by the stem "The following items refer to the procedures used to arrive at tax-related decisions.") are "I have been able to express my views and feelings during those procedures" and "Those procedures have been free of bias" (1 = strongly disagree, 7 = strongly agree). The items were averaged into a procedural justice scale. The complete adapted scale is found in Appendix A on page 23.
Distributive justice of the tax authority was measured with a 5-item scale from Verboon and van Dijke (2007) that measures both taxpayers' horizontal and vertical equities in addition to exchange equity with the government. Item examples are "The use I make of all kinds of social services reflects in a proper way the taxes I pay" and "Regarding social services I get little return for my tax money" (reverse coded) (1 = completely disagree, 7 = completely agree). We averaged these items into a distributive justice scale.
Cognition-as well as affect-based trust were included as control variables. Cognition-based trust was measured with the 6-item scale developed by McAllister (1995). The scale items were slightly adapted to fit the context of trust in the tax authority. Item examples are "The tax officials approach their job with professionalism and dedication" and "Given these officials' track record, I see no reason to doubt their competence and preparation for their job" (1 = completely disagree, 7 = completely agree). We averaged the items into one scale. We measured affect-based trust with the 5-item scale from McAllister (1995). We slightly adapted the items to fit the context of trust in the tax authority. Item examples are "The tax officials and I have a sharing relationship; we can both freely share our ideas, feelings, and hopes" and "I can talk freely to these officials about difficulties I am having regarding tax and know that they will want to listen" (1 = completely disagree, 5 = completely agree). We averaged the items into one scale.
Intention to voluntarily pay taxes was measured with the 10-item scale 1 from Kirchler and Wahl (2010). Item examples (preceded by the stem "When I pay my taxes as required by the Ethiopian tax laws and regulations, I do so...") are "…because to me it's obvious that this is what you do" and "…to support the state and other citizens" (1 = completely disagree, 7 = completely agree).We averaged the items to create a voluntary tax compliance index. Table 1 below shows that all the variables of interest in this study (i.e. distributive justice, procedural justice, affect-based trust, and cognition-based trust) are associated with respondents' intention to pay taxes voluntarily with procedural justice only moderately related to it. These associations are in line with prior findings (see, for instance, Wenzel, 2002Wenzel, , 2003Gobena and van Dijke, 2017 for association of procedural and distributive justice with voluntary tax compliance, and Murphy, 2004;van Dijke, 2016, 2017 for correlation of trust with voluntary tax compliance). Of the demographic control variables, only gender is significantly related with intention to pay taxes voluntarily, female taxpayers exhibiting more voluntarism compared to their male counterparts. This is in line with a prior finding documented by Wenzel (2002

Regression Analyses
Two regression analyses were conducted in order to examine the main and interactive effects of the variables considered as determinants of category "C" taxpayers' intention to pay their taxes voluntarily. The first is a multiple regression in which all the independent variables were included in the analysis and regressed on the intention of the responded taxpayers to voluntarily comply with their tax obligations. Table 2 below depicts the results of the regression analysis. As table 2 shows, gender, age, and years of business experience, considered separately, significantly predicted intent to pay tax voluntarily. Of course, while gender and age negatively affected intent to voluntarily pay tax, years of business experience positively predicted it. The negative coefficients of gender and age mean that female respondents reported to have greater intention to voluntarily pay their taxes than their male counterparts and older respondents have lesser intention to voluntarily pay their taxes than their younger counterparts. The positive coefficient of years of business experience means that the longer a respondent was in business, the more s/he intended to voluntarily pay their taxes. Contrary to the anecdotal claim that the taxpayers complain about their assessed tax because it is beyond what they can afford to pay, the income of the taxpayers did not significantly predict their intention to voluntarily pay tax. Of the variables of interest, distributive justice and affect-based trust significantly predicted the intention to pay taxes voluntarily positively.
The second one was regression with PROCESS macro (Hayes, 2013;model 1) which was meant to examine the interactive role of procedural and distributive justice on the taxpayers' intention to voluntarily pay their taxes. Table 3 depicts the interactive effect of the two elements of justice on intention to pay taxes voluntarily. .03 -.02 .08 .99 (+1 SD from the mean) -.02 -.08 .03 Table 3 shows that while distributive justice of the tax authority was sufficient to stimulate the intention to voluntarily pay taxes in its own right, the main effect of procedural justice did not predict the intention to voluntarily pay taxes. The interactive effect of procedural and distributive justice, which is of paramount interest to this study, negatively stimulated the taxpayers' intention to voluntarily pay taxes. The negative sign of the interaction effect means that procedural justice predicts intention to voluntarily pay taxes only when distributive justice of the tax authority is perceived to be low in the eyes of the taxpayers. When distributive justice is at its high, taxpayers do not need to look back on to the procedures (i.e., the decision-making processes) that led to the distribution of the favorable outcomes, namely tax burdens and tax-funded benefits. This finding is in line with the argument in the sense-making analysis of justice that people tend to look around for justifications of negative and unfavorable encounters to learn what the future holds (see Brockner & Weidenfeld, 1996 for overview). This result concurs with many findings documented in organization behavior (e.g., Shapiro, 1991).
The interactive effect of procedural and distributive justice on intention to pay taxes voluntarily is depicted in figure 2 below.

Discussion
Taxpayers' compliance with relevant tax laws often involves a fundamental social dilemma in which the taxpayers may decide to voluntarily cooperate with the tax authorities' decisions (for example, by paying their taxes) presuming that the tax money would be used by the government for mutual benefits with their fellow citizens. Alternatively, taxpayers may fear abuse of power by the tax authority (or the entire government) in which case the tax money would be used for non-productive government expenditure and may even be embezzled (see for instance, Alm, Jackson, & McKee, 1993). The effect of this dilemma is heightened by citizens' lack of information on whether the tax authority (and the entire government system) can be trusted not abuse their power. In such circumstances, taxpayers often resort to procedural fairness information that they use a heuristic to guide their decisions to comply with or evade taxes (see Van Dijke and Verboon, 2010;van Dijke, 2016, 2017). This study examined (1) the main effects of trust in the tax authority, procedural and distributive justice of the authority, and exchange equity of the authority and (2) the interactive role of procedural and distributive justice on intention to voluntarily pay taxes. The main effects of distributive justice and affect-based trust affected intention to voluntarily pay taxes. The interactive role of the tow justice components-procedural and distributive-also stimulated intention to voluntarily pay taxes. In the following sections, implications and limitations of the research are discussed.

Theoretical and Practical Implications
This study revealed that it was not what category C taxpayers anecdotally cite as a major determinant factor for their grievance over their 2017 tax assessment, namely their income falling short of the amount they are assessed to pay in taxes. Instead, their perception of distributive justice and emotional attachment with the government stood as stimulators of intention to voluntarily pay taxes. In addition, the interactive effect of procedural and distributive justice on intention to pay taxes voluntarily was significant. Owing to the significant interactive role of procedural and distributive justice on intention to pay taxes voluntarily, this study contributes a noble finding to the tax compliance literature extending the interactive role of the two types of (i.e., procedural and distributive) justice that has extensively surfaced in the organization behavior studies (see e.g., Shapiro, 1991;Brockner et al., 1994;Brockner and Wiesenfeld, 1996;De Cremer, 2005;De Cremer et al., 2010). As to the knowledge of the author, this finding has never been documented in the tax compliance context by any prior study and is hence a new significant contribution.
This study also contributes to tax administration and tax policy design. One tax administration as well as High Distributive Justice policy implication of the study is that female taxpayers reported that they are more compliant with their tax obligations than their male counterparts, as a result of which enforcement efforts need to be geared toward male than female taxpayers. A second practical contribution of the study is that older respondents indicated that they have lesser intention to voluntarily pay their taxes than their younger counterparts, and this informs tax authorities and tax policy makers that more needs to be done to get the willingness of elder people to pay their taxes voluntarily. A third contribution is that perceived fair distribution of tax burdens and tax-funded benefits (i.e., distributive justice) and emotional attachment of taxpayers with the tax authorities are associated with high level of intention to voluntarily pay taxes.
A fourth and more pervasive contribution of this study is that tax authorities would be informed about the interactive effect of two justice components-procedural and distributive justice-in stimulating voluntary cooperation of citizens with authorities. Specifically, a perceived low level of distributive justice of the tax authority is shown to be buffered by a high rather than low procedural justice.

Limitations and Future Research Direction
As is the case with all research, this research is not free from limitations. The first limitation is that the data used in this study is self-reported data on a highly sensitive subject about which respondents are not believed to provide genuine answers. This is presumably because taxpayers are less likely to confess their tax evasion as they fear legal actions following their self-report (see, for instance, Torgler, 2008;Wenzel, 2002;Murphy, 2004). Hence, other controlled methods such as experimental method are recommended to replicate the results of this study. The second limitation is that data for the study was gathered through a one-time cross-sectional survey and hence it would not warrant causal claim between the study variables. Consequently, other research with a longitudinal or experimental design might warrant causal relationship between the variables. The third limitation is that the sample respondents in this study were not randomly selected and thus will not be representative any population. Random selection of respondents was not possible because tax authorities view taxpayers' information as highly confidential and the researcher had to resort to convenient sampling whereby any accessible category C taxpayers from the geographic areas specified for this research were included. Therefore, the results must be cautiously viewed in a way they are less likely to be generalized to any population outside the sampled respondents.