Guilt Cognitions, Religiosity and Taxpayer Voluntary Disclosure

Increasing tax income by sanctions and penalties for taxpayers who do not report taxes properly requires a significant cost. One effective method is tax amnesty. Even though it is considered as an effective method, the implementation of tax amnesty has not been optimal even the income from tax amnesty is far less than the potential domestic tax. Therefore it is necessary to consider the factors driving the taxpayer to carry out a tax amnesty. Previously, taxpayers only considered economic and non-economic aspects, but there were other considerations such as guilt and driven by the level of intrinsic religiosity of taxpayers when they did not report taxes with actual conditions. This study examines whether guilt has an influence on the voluntary compliance of taxpayers driven by intrinsic religiosity. Respondents of this study are taxpayers in Indonesia using online surveys. This study was analyzed using SEM-based covariance with Smart PLS 2.0. The results of the study explained that foreseeability and intrinsic religiosity had a significant effect on tax amnesty reporting. Intrinsic religiosity only moderates foreseeability as an indicator of guilt for tax amnesty reporting.


Introduction
The level of intent did not report taxable income worldwide are increasingly high that it requires significant costs to disclose (Dunn et al., 2016). This prompted the government to find out how to increase voluntary taxpayer compliance, including Indonesia. Recent periods of Indonesia applying tax amnesty. This is stated to be more effective than the use of sanctions and penalties for taxpayers who do not report their taxes properly, because the use of sanctions and penalties requires a significant cost. In contrast, the tax authority views tax amnesty tends to be cheaper. Taxpayers are given the opportunity to correct their own mistakes before, if they have been reported correctly, but in the future errors are found, the tax authorities will impose sanctions and penalties.
The application of tax amnesty has become the choice of the world's tax authorities today. Since 2000, more than half of US states have offered one or more tax amnesty programs (Weinreb, 2009). Then followed by many developed countries such as, Greece, Ireland, Italy, Portugal and Spain and developing countries such as, Argentina, Brazil, Colombia, India, Philippines (Dunn et al., 2016). Indonesia is no exception that provides tax amnesty opportunities from 1 October 2016 until 31 March 2017 (online tax.com).
Although the application of tax amnesty is more effective than sanctions and penalties, economicbased study studies suggest that tax amnesty is not very effective in encouraging taxpayer participation, the benefits of the tax amnesty program are few, as stated by Hasseldine (1998) that the amount of American tax amnesty States range from 0.008 to 2% of state tax revenue. Similar to Indonesia, Sri Mulyani stated (in kominfo.go.id) that the number of tax amnesty reporting is still small compared to the potential for domestic taxpayers. Thus it is necessary to see what factors can increase the participation of taxpayers to carry out tax amnesty.
Dunn et al. (2016) states that many things can be considered to increase the motivation of taxpayers to carry out tax amnesty. Previously taxpayers made decisions for economic and non-economic reasons (Coricelli et al,. 2010;Maciejovsky et al., 2012), perhaps the tax amnesty program would be more effective if the authorities considered factors other than economic factors, as individual factors (Tarmidi, 2019) as stated by Dunn et al. (2016) namely the guilt factor. The results of his research stated that taxpayers feel a personal responsibility for unethical reporting when listening to an unethical action from a tax advisor. In addition, taxpayers also feel greater negative consequences when trying to avoid taxes. However, that does not mean that guilt is a major factor in increasing taxpayers in making ethical decisions.
Huffman (1988) in Vitell et al. (2007) states that religiosity is the strongest determinant in a value. Religiosity is a maturity of one's faith that shows the level of commitment and perspective on certain religious characteristics or events (Benson et al., 1993). A person's level of faith can be identified based on his adherence to religious rules by doing good deeds and leaving bad deeds. However, Batson in Hardy and Carlo (2005) states that the relationship between religiosity and ethical decision making is affected the situation of the individual. Thus, it can be concluded that religiosity does not absolutely affect individuals in making ethical decisions. Allport and Ross (1967) state that every individual has a religious orientation, such as an intrinsic and extrinsic orientation. Individuals with intrinsic orientation use religion for their main purpose. These individuals try to be in harmony and follow religious beliefs fully and they believe that life is their religion. Whereas individuals with extrinsic orientation use religion for themselves such as providing security, comfort, socialization, distraction, status, and self-justification. Z (2017) states that only intrinsic religiosity has a significant influence on ethical decision making. This is due to the control of individual behavior when they have to make one ethical decision, such as reporting taxes (Tarmidi and Nurlita, 2018). Thus, this study uses intrinsic religiosity as a moderating taxpayer guilt to be able to report the tax in real.
Research related to tax amnesty is mostly done abroad while in Indonesia there is very little researching. One was conducted by Suliyawanti (2018) to examine the influence of the taxpayer, subjective norms and morals. Suliyawanti (2018) did not test guilt by Dunn (2016) is the factor that determines the taxpayer following the tax amnesty or not. Based on the background above, this study examines whether guilt has an influence on the voluntary compliance of taxpayers that is moderated by intrinsic religiosity?

Guilt Cognitions
Guilt is a negative emotional (Kubany and Watson, 2003). Phenomenologically guilt is defined as an unpleasant feeling with a belief that accompanies that a person should think, feel, or act differently (with implications of responsibility, mistakes, and/or not justification enough) in making decisions (Kubany and Watson, 2003). Kubany and Watson (2002) conducted further structured interviews related to trauma and guilt. Some previous research (in Kubany and Watson, 2003) translates stress referred to here is in the context of stressful events in life, such as the death of a loved one, physical abuse, sexual abuse. However, Kubany and Watson (2003) illustrate contextual factors that can contribute to the magnitude of the trauma of guilt and guilt in general. Finally, a traumatic event that can be tense can be an excellent context for studying the relationship between guilt and shame, that is, an emotion whose relationship with guilt is not fully understood in a controversy.
The findings of Kubany and Watson (2003) the formulation of guilt is conceptualized as a multicomponent construct. The amount of guilt that a person will experience after being exposed to a negative event is considered a function of attendance and the magnitude of the combination of the five factors hypothesized to be the main component or determinant of error. These factors are distress, responsibility for a decision, justification for a decision, predictable consequences arising from a decision, and personal values.
Feelings of guilt occur when distress is an unpleasant feeling associated with a negative outcome, and the individual's cognition plays a role in the negative outcome. Thus, guilt will be high when someone is depressed, takes full responsibility for causing negative results, cannot justify a decision, predicts the negative consequences arising from the decision, and the act violates the individual's personal values (Dunn et al., 2016). Dunn et al. (2016) state that when extended to the tax context, it means that a taxpayer who has committed tax evasion will feel depressed, and is likely to experience guilt if he takes full responsibility for his choice to avoid, unable to justify the decision to avoid, predicted that embezzlement would be detected, and that action violated his personal values.
However, Dunn et al. (2016) only use three guilt factors, namely responsibility for a decision, justification for a decision, and estimated consequences arising from a decision. In the context of tax decision making, taxpayers have a level of responsibility for their tax reporting decisions, must be able to justify tax reporting decisions, and understand the negative consequences that may arise from inaccurate taxes.
In addition, distress is an affective state that involves feelings of unhappiness, depression, and anxiety which can result in emotional exhaustion. Whereas personal values are similar to attitude but their nature is embedded, permanent, and stable.

Religiousity Huffman (1988) in Vitell et al. (2007) states that religiosity is the strongest determinant in value.
Religiosity is a maturity of one's faith that shows the level of commitment and perspective on certain religious characteristics or events (Benson et al., 1993). A person's level of faith can be identified based on his adherence to religious rules by doing good deeds and leaving bad deeds.
Hunt and Vitell (1986) in Vitell (2009) identify the religiosity of individuals influencing specific aspects of the ethical decision making process. The results suggest that more religious individuals are clearly defined by deontological norms and each norm plays an important role in decision making. However, Batson in Hardy and Carlo (2005) states that the relationship between religiosity and ethical decision making is influenced by situations faced by individuals. Thus, it can be concluded that religiosity does not mutually influence individuals in making ethical decisions. Allport and Ross (1967) state that every individual has a religious orientation, such as an intrinsic and extrinsic orientation. Individuals with intrinsic orientation use religion for their main purpose. These individuals try to be in harmony and follow religious beliefs fully and they believe that life is their religion. Whereas individuals with extrinsic orientation use religion for themselves such as providing security, comfort, socialization, distraction, status, and self-justification. This research focuses on intrinsic religiosity, because intrinsic orientation is the main driver of indiviidu in ethical decision making. If related to tax reporting, individuals with intrinsic religiosity will be more likely to report taxes according to the actual situation. Coricelli et al. (2010) found that there are emotional costs when taxing, so that emotional arousal increases when tax avoidance is initially detected. Schwartz and Orleans (1967) in Dunn et al. (2016) found that moral circumstances have a stronger impact on compliance than the threat of punishment, and positive emotional incentives by taxing authorities are more effective in encouraging tax compliance than financial incentives, such as guilt.

Hypothesis Development
Dunn et al. (2016) consider guilt to be an important driving factor in influencing tax decision making. Grasmick and Scott (1982) found that sanctions, social stigma, and feelings of guilt, the preventive mechanism with the greatest inhibitory effect on non-compliance was feelings of guilt. Erard and Feinstein (1994) developed a tax compliance model that includes guilt, and states that guilt can bias taxpayers' perceptions about audit probability. Furthermore Cho et al. (1996) add that the 'psychological costs' of tax evasion may be high enough to prevent the economic benefits of tax evasion. Thus, there is a positive correlation between psychological stress and tax compliance, and found that psychological stress exceeds the excitement of possible gains from tax evasion (Dulleck et al., 2012). Examining the direct relationship between disclosure of guilt and forgiveness of new taxes Dunn et al. (2016) did it. So that it becomes the motivation of researchers to develop a guilty relationship with tax amnesty. Dunn et al. (2016) suggest using the personality trait variable to moderate the guilt for tax compliance with tax amnesty. However, researchers tend to use intrinsic religiosity as a motivating factor for guilt. The cause of religiosity is the strongest determinant in a value (Huffman, 1988). Kubany and Watson (2003) the formulation of guilt is conceptualized as a multicomponent construct. The amount of guilt that a person will experience after being exposed to a negative event is considered a function of attendance and the magnitude of the combination of the five factors hypothesized to be the main component or determinant of error. These factors are distress, responsibility for a decision, justification for a decision, predictable consequences arising from a decision (foreseeability), and personal values.
Feelings of guilt occur when distress is an unpleasant feeling associated with a negative result, and the individual's cognition plays a role in negative results. Thus, guilt will be high when someone is depressed, takes full responsibility for causing negative results, cannot justify a decision, predicts the negative consequences arising from the decision (foreseeability), and the act violates the individual's personal values (Dunn et al., 2016). Dunn et al. (2016) state that when extended to the tax context, it means that a taxpayer who has committed tax evasion will feel depressed, and is likely to experience guilt if he takes full responsibility for his choice to avoid, unable to justify the decision to avoid, predicted that embezzlement would be detected, and that action violated his personal values.
Dunn et al. (2016) state that when extended to the tax context, it means that a taxpayer who has committed tax evasion will feel depressed, and is likely to experience guilt if he takes full responsibility for his choice to avoid, unable to justify the decision to avoid , predicted that embezzlement would be detected, and that action violated his personal values. Crawford et al. (1990) found that higher estimates were significantly related to blaming rape victims, while Lagnado and behavior were despicable, so predictable actions were more likely to be blamed. Although researchers have found that each responsibility, justification, and foreseeability. The findings of Dunn et al. (2016) show the effect of threeway interactions in such a way that taxpayers tend to make disclosure of tax amnesty when they predict that they will be captured by tax authorities, unless they can defuse responsibility for their avoidance and justify their avoidance. This study modified the research of Dunn et al. (2016) by adding an intrinsic religiosity factor to encourage guilt to increase their tax disclosure desire in the tax amnesty program. Based on the development of the discussion above, the hypothesis can be as follows: Hypothesis 1a: Responsibility has an influence of taxpayers decision to make tax amnesty disclosures. Hypothesis 2b: Responsibility will be increasingly influential with religiosity in taxpayers decision to make tax amnesty disclosures Hypothesis 2a: Justification has an influence on taxpayers decision to make tax amnesty disclosures Hypothesis 2b: Justification will be increasingly influential with religiosity in taxpayers decision to make tax amnesty disclosures Hypothesis 3a: Foreseeability has an influence on taxpayers decision to make tax amnesty disclosures Hypothesis 3b: Foreseeability will increasingly influence religiosity in taxpayers decision to make tax amnesty disclosures.
Hypothesis 4: Religiosity has a positive influence on taxpayers decision to make tax amnesty disclosures.

Methodology of research
This research is a quantitative study with a survey research design, using a measurement tool called a questionnaire. The type of data used is cross-sectional for photographing phenomena at one particular time. Sampling of this study through nonprobability sampling, namely purposive sampling. The sample used in this study is taxpayers in Indonesia. Data were analyzed using Structural Equation Model (SEM) based on covariance with the SmartPLS program.

Respondent characteristics
The respondents of this study were taxpayers in Indonesia. The data collection period is one month using a questionnaire. Questionnaires are distributed online using Google forms, Google form links are sent through whatsApp groups. There are 175 responses from taxpayers in Indonesia, but only 164 can be analyzed. This is caused by respondents not fulfilling the requirements. Details of the number of respondents can be seen in Table 1.

Measurements
This study uses a case developed by Dunn (2016), with consideration of cases of fraud that often occur in Indonesia, especially at Educational Institutions.

Taxpayer's Voluntary Disclosure
Taxpayers Voluntary disclosure is the voluntary disclosure of taxpayers in the tax amnesty program using vignettes developed from the model of Dun et al. (2016) adapted to the language and conditions in Indonesia. The measurement scale used is the answer whether or not to report tax amnesty on a given vignettes.

Guilt Cognitions
Guilt is defined as an unpleasant feeling with the conviction that accompanies that taxpayers should think, feel, or act differently when they want to take the decision to report the tax they should. This study developed the Kubany and Watson (2003) model. The measurement scale uses a 5-point Likert scale, 1 "strongly disagree" to 5 "strongly agree". Responsibility is the responsibility of taxpayers for their tax reporting decisions, there are 6 question items but only 3 valid with Cronbach Alpha 0.790078 and composite reliability 0.888808. When interacting with religiosity, the results of the validity show that most of them are invalid, thus the researcher removes the interaction between responsibility and religiosity interactions. Justification is a justification of taxpayers from their tax reporting decisions, there are 4 question items but only 2 are valid with Cronbach Alpha 0.74123 and composite reliability 0.894317. Whereas when interacting with Cronbach's religiousity the alphabet is 0.964178 and composite reliability is 0.96302. Foreseeability is the alleged taxpayer's negative consequences that may arise from inaccurate tax returns 4 question items with Cronbach Alpha 0.804621 and composite reliability 0.870421, when interacting with Cronbach Alpha religiosity it becomes 0.981714 and composite reliability 0.982011.

Intrinsic Religiosity
Intrinsic religiosity is a taxpayer who has a high level of faith (understanding right and wrong actions). This model uses the Allport and Ross (1967) model developed by Vitell et al. (2007). This study developed a model (11 question items). The measurement scale uses a 5-point Likert scale, 1 "strongly disagree" to 5 "strongly agree". However, only 10 items were valid questions with Cronbach Alpha 0.871143 and composite reliability 0.856871. Table 2 shows the descriptive results of respondents' answers, known number of respondents, average, and standard deviation. The results of respondents' answers the average value for the variable taxpayer (TP), responsibility (TR), religiosity (TRI), and justification (TJ) above the actual average value. While the average value of actual foreseeability is below the average value of the theory. Respondents' answers when faced with cases of wealth transferred from abroad into the country, the wealth must be reported to the State and paid taxes. Most respondents chose to report this wealth, as seen from the average taxpayer (TP) of 0.87. The sense of responsibility (TR) of respondents to report the tax is also high as the average value of respondents' answers is above the theoretical average of 13.01. However, respondents' understanding of the consequences that would occur if they did not report the wealth below the actual average value, which is 9.16. Respondents tend to look for justification for what they do as the average value of actual justification is above the theoretical average value, which is 8.81. Most respondents rated that their religious control was quite high as the results of the average respondent's answer to internal religiosity was very high at 45.89. As a basic control for individual internal religiosity, it should be able to increase guilt cognition in reporting tax amnesty.

Data Analysis Descriptive statistics
The structural model is evaluated by using R2 to measure the degree of variation in the changes in the independent variable on the dependent variable and t-value to test the significance of the constructs in the structural model (Hartono, 2014). R-Square value of 0.09211 that the taxpayer constructs variability that can be explained by internal responsibility, justification, foreseeability, and religiosity is 9.211 percent. While R-Square value interaction model is 0.170879, which means that the taxpayer construct variability that can be explained by responsibility, justification, foreseeability, internal religiosity, foreseeability interaction and justification interaction is 17.09 percent. While the rest is explained by other factors not explained in this study. The value of the path coefficient or inner model shows a significant level in testing the hypothesis shown by the t-statistic. According to Hair (2010) the T-statistic value must be above 1.96 for two-tailed and 1.64 for one-tailed with alpha 5 percent and power 80 percent. Table 13 shows the path coefficient of the research model. H1a explains the relationship of responsibility to tax amnesty disclosure. The parameter coefficient is 0.199611, meaning responsibility has a positive effect on tax amnesty disclosure. The higher individual responsibility will affect tax amnesty disclosure. However, t-count shows 1.650287 smaller than t-table 1.96. This means that H1a is not accepted.
H2a explains the relationship of justification to the disclosure of tax amnesty reporting. The parameter coefficient is -0.022507, meaning justification has a negative effect on tax amnesty disclosure. The higher individual justification will influence his decision not to disclose tax amnesty. However, t count shows 0.202991 which is smaller than t-table 1.96. This means that H2a is not accepted.
H2b explains the relationship of justification to the disclosure of tax amnesty moderated by religiosity. The parameter coefficient is -0.092082, meaning justification has a negative effect on tax amnesty disclosure which is moderated by religiosity. However, t count shows 0.54787 which is smaller than t-table 1.96. This means that religiosity does not moderate justification, H2b is not accepted.
H3a explains the relationship between foreseeability and tax amnesty disclosure. The parameter coefficient is 1.779118, meaning that justification has a positive effect on tax amnesty disclosure. The higher the individual's foreseeability will affect his decision to disclose tax amnesty. T count shows 2.095622 which is greater than t-table 1.96. This means that H3a is accepted.
H3b explains the relationship between foreseeability and tax amnesty disclosure. The parameter coefficient is 1.902558, meaning that justification has a positive effect on the disclosure of tax amnesty with moderation of religiosity. The higher the individual's foreseeability will influence his decision to disclose tax amnesty with religious moderation. T count shows 2.188237 which is greater than t-table 1.96. This means that H3b is accepted. Religiosity moderates foreseeability.
H4 explains the relationship of religiosity to tax amnesty disclosure. The parameter coefficient is 1.779118, meaning that religiosity has a positive effect on tax amnesty disclosure. The higher religiosity of an individual will affect his decision to disclose tax amnesty. T count shows 1.723824 which is greater than t-table 1.64. This means that H4 is accepted.

Discussions
Respondents' perception of responsibility when faced with tax amnesty disclosure cases is quite high, as seen from the average responsibility of 13.09 is greater than the theoretical average. This shows that respondents have a sense of responsibility for reporting taxes from wealth that has been previously saved. Besides feeling guilty about what happened before. But seen from the analysis of the inner model responsibility did not have a significant effect when compared with t table 1.96. On the contrary it will have a significant effect when compared with t table 1.64 because the value of t arithmetic is responsive to tax amnesty reporting. As stated by Dunn (2016) that responsibility will have a significant effect if there are pressure conditions or negative impacts that will arise from the tax evasion.
Respondents tend to make a justification for what is done by embezzlement of wealth this can be seen from the average justification of respondents to research instruments that exceeds the theoretical average, which is 8.81. The inner model test results also show the parameter -0.022507, meaning that the justification has a negative effect on tax disclosure through tax amnesty. This is supported by the statement of Shalvi et al. (2011) the level of lies depends on the extent to which self-justification is available the greater the level of justification, the greater the incidence of lying. However, this study showed insignificant results. The same results when religiosity as intervening, but not significant. This means that religiosity does not moderate the justification for disclosing tax amnesty.
The perception of the alleged negative thing that the respondent will face when embezzling tax is not too high, this can be seen from the average respondent's answer is 9.16. The results showed no significant results on the disclosure of tax amnesty. Different when interacting with religiosity foreseeability has a significant effect on tax amnesty disclosure. Thus religiosity moderates foreseeability in expressing tax amnesty. Religiosity has a positive effect on tax amnesty reporting. This is as stated by Hunt and Vitell (1986) in Vitell (2009) that individual religiosity influences specific aspects in the ethical decision making process. The results suggest that more religious individuals are clearly defined by deontological norms and each norm plays an important role in decision making.

Conclusions and suggestions
This study aims to examine the effect of guilt factors on the disclosure of tax amnesty and intrinsic religiosity as intervening variables that encourage guilt to express tax amnesty. The results of the study explained that the alleged negative consequences (foreseeability) had a significant effect on the disclosure of tax amnesty. Religiosity significantly moderates foreseeability of tax amnesty reporting. But religiosity does not moderate responsibility and justification for tax amnesty disclosure.
This study uses a survey method with an online questionnaire, so it does not see the condition of respondents feeling guilty. For this reason, researchers suggest using experimental research methods.