An investigation of the Theory of Planned Behavior and the role of Tax Amnesty in tax compliance

The individual taxpayers’ low awareness has become the main problem of developing countries in tax aspect. Thus, this study aimed to examine the determinant factors of tax noncompliance using Ajzen’s (1991) Theory of Planned Behavior as a theoretical framework. Specifically, Tax Amnesty is added to the theory’s constructs: attitude, subjective norms, and perceived behavioral control. Tax Amnesty is expected to be a moderating influence. The population of this study is individual taxpayer in KPP Pratama Sukomanunggal. Based on convenience sampling method, the number of sample in this study are 145 samples. The data was analyzed using Structural Equation Modeling (SEM) with SmartPLS.3.0 and SPSS 21. The results indicated that first, attitude and subjective norms are significantly influence behavioral intention except perceived behavioral control. Second, the model including Tax Amnesty provides a significant influence of tax noncompliance in two constructs; attitude and subjective norms. However, the interaction effect of perceived behavioral control does not appear significantly.


INTRODUCTION
Indonesia is categorized as a country, which has a low tax ratio level. In 2014, Indonesia tax ratio compared to its gross domestic product approximately 12.8% and then, 2015 decreased to 12.32%. When compared with other countries, Indonesia tax ratio remains below even with Philippines is 14%, Malaysia is 16% and Singapore is 22%. Having low tax ratio level, indicating low tax revenue collected by Indonesia government (www.kemenkeu.go.id 2016).
Tax revenue is the main revenue for covering the government expenditures. Thus, the government always does some great efforts to increase tax revenue by increasing the tax compliance. Various programs were implemented by the government to raise the awareness in paying taxes. The Directorate General of Taxation had issued Sunset Policy pro-gram in 2008, the year of Tax Amnesty in 2014, and the recently program is Tax Amnesty policy. Sunset Policy is regulated by Act 28 year 2007, conformity with article 37 of statutes, which stand for the existence of Sunset Policy. Actually, Sunset Policy is not a new program as Sunset Policy considered the lowest Tax Amnesty, which was implemented in 1984. Sunset Policy is a program of sanction amnesty for the taxpayers who report their income honestly and pay the rest of the tax before 21st March 2009. If taxpayers used this program, they would get two benefits. Firstly, they got reduction or removal of administrative sanction such as interest of annual tax return before the end of 2007. Secondly, reduction or removal of administrative sanction for unpaid tax and lower tax payment either organizational or individual. This is also open for the individual taxpayers who register themselves voluntarily to have taxpayer"s identity.
After the expiration of Sunset Policy, the level of tax compliance did not seem to increase. Besides, the lower tax revenue collection is from Individual taxpayers.
Actually, corporation has a better compliance than individual taxpayers (http://ekonomi.metrotvnews.com). It is proved by revenue from tax sector which never reached the target which set by the government since 2009 until 2015. In conclusion, the potential long-run revenue gains from Sunset Policy are relatively small. The comparison between target and realization of tax revenue is illustrated by chart presented in Figure 1.
The next implemented program was the year of amnesty for taxpayer in 2015, hopefully it would increase public awareness to pay taxes and to report their annual tax return correctly. The year of amnesty for taxpayer is a chance from government before implementing Tax Amnesty in 2016, which had many advantages; incentives gave to all kind of taxes, taxpayers who did not report amount tax correctly and report tax. Beside, the taxpayers who were late to report the tax in 2015 would not be fined or pay the rest of tax. If the taxpayer increases, the revenue from tax will increase. Surprisingly, in 2015 tax ratio decreased from 11.32% to around 11% from the previous year. So, the year of amnesty for taxpayer program did not success enough (http://www.tribunnews.com 2016).
Low of public awareness to pay tax reduces tax compliance and it has become the main problem of the state from tax aspect. Commonly, small acceptance to pay tax voluntary happens in developing countries, including Indonesia, it seems not only is encouraged by the low of ratio level of taxpayer and the number of people but also other dominant factors such as the high number of tax avoidance practices, institutional trust, capacity to implement, tax evasion, administration complexity, tax compliance, and tax morality.
It was found by Bobek et al. (2007) that social norm mostly influences taxpayers to fulfill their debt. It is also supported by Wenzel (2004) that personal ethics which is based on religion and cultural norm may have impact on the free tax compliance behavior of the fiscal exchange between taxpayers and government. Theoretically, factor have been identified in tax compliance tax knowing as Theory of Planned Behavior developed by Ajzen (1991) which extend the attitude, social norm, and perceived behavioral control create a person"s intention to perform or not to perform and behavior is the most important factors immediately determinant of personal action. Some previous study done by Bobek and Hatfield (2003) examined the noncompliance of taxpayers by using Theory of Planned Behavior, which is developed by Ajzen (1991). In their" research framework, tax compliance was formed by three factors; they are the attitude toward the behavior, subjective norms, and perceived control by adding moral obligation as a moderating variable. The results were attitude toward compliance and subjective norms affected behavioral intention.
Similar result was also reported in Bobek et al. (2007). On occupation the study of Elia (2007) proved that not only attitude, subjective norms, and perceived behavioral control influenced on tax compliance but also financial, facilities, and organization environment. However, the inconsistency of findings on the relationship between compliance and some of its determinant (Widi and Argo 2010), they failed to prove all assertion. Koch et al. (2015) reported that attitude might make less compliant, otherwise it significantly related to compliance after moderating by financial condition. Wenzel (2004) offered an analysis of norm process in tax compliance, which social norm referred to the perceived injunctive norms seemed to reduce tax compliance and no longer significant influence.
According to Ajzen (2005: 42) there are three broad categories of factors that interact with attitudes or personality traits. They are situational circumstances surrounding performance of the behavior, characteristic of the individual and secondary characteristics of the disposition. Situational factor is the different condition in some situation but consistent with each other such as regulation and poli-cy. Hence, Tax Amnesty as one of recently fiscal policy of government can classified as situational factors, which may moderate tax noncompliance, and its determinant factors.
Based on the inconsistency result of previous study and phenomenon, the researcher is going to examine noncompliant behavior toward taxpayers. In this present study extends this prior tax compliance research in two ways. First, this study use factor determinants of noncompliance as suggested by Theory of Planned Behavior. Second, it focus on situational factors, Tax Amnesty, as moderation variable will interact with construct toward behavior intention; specific behavior (attitudes), social pressures an individual feels to perform (subjective norms), the amount of control an individual perceived to engage in that behavior (perceived behavior control).

THEORETICAL FRAMEWORK AND HYPO-THESIS Theory Planned of Behavior
Theory of Planned Behavior is formulated in 1991 by Icek Ajzen. This theory suggests that a person"s behavior is determined by intention to perform the behavior. Intentions are a function of three basic determinants; one reflecting personal nature, one social in nature, and a third dealing with issues of control (Ajzen 2005: 117). All subject"s interaction are defined as well as Figure 2.
First, personal factor is the individual"s attitude toward the behavior. This attitude is the individual"s positive or negative evaluation of the particular behavior. Attitude is representation of personal norms, as actualization of self based standards or expectation for behavior that flow from internalized values (Bobek et al. 2007).
Source: Tax council website, processed.

Figure 1 Difference of Target and Realization of Tax Revenue since 2009 until 2015
Source: Ajzen (1991). Attitudes are predispositions to respond favorably or unfavorably to an object, group, places, do something in certain condition. Attitude is a person"s feeling or way of thinking that affects a person"s behavior.
Moreover, individual"s attitudes are representation of their belief that underlying the outcome. That is expectation to achieve some benefits by engaging in their behavioral intention. Attitude comes from their feelings that define the behavior. So, attitude in this study represents how the taxpayers thinking and expect to after fulfilling their tax liabilities.
The second determinant of intention is social in nature. Social in nature is person"s perception of social pressure to perform or not perform the behavior under the consideration. Social in nature deals with perceived normative prescription, this factor is termed subjective norms. Subjective norms are a reference from other peoples that influence someone to do something (Ajzen 1991). Subjective norms refer to a person"s belief about individual or groups approve or disapprove that will motivated taxpayers to behave compliant.
According to Bobek et al. (2007) subjective norm is representation of individual"s perception of the injunctive norms held by "referent" others such as family, friends, and co-workers. Taxpayers" beliefs appear from their compliance reference from surrounding such as groups, family members, employers, friends, and spouse.
Third, the determinant of intention is control represents the sense of self-efficacy or ability to perform or not perform the behavior. This factor is termed with perceived behavioral control. This control is more likely to personal deficiencies and external obstacles can interfere with performance the behavior.
Finally, all the determinant factors; attitude toward behavior, subjective norms, perceived behavioral control interact each other predict intention. Then, the intention will translate into action. Explanation about how Theory of Planned Behavior frames this research discussed in the next discussion.

Tax Compliance
Taxpayers have high contribution to manage their tax responsibilities accurately. Taxpayers should count, pay and report their tax by themselves, so tax compliance awareness is needed. Tax compliance can be defined as a condition that taxpayers fulfill all their tax liabilities Siti (2016: 38).
According to Siti (2016: 110) there are two kinds of tax compliance; formal compliance and material compliance. Formal compliance is a taxpayer"s compliance fulfils their liabilities based on tax law formally. Materiel compliance is a taxpayer"s condition fulfils all their tax liabilities substantively based on the content of tax law. Materiel compliance also includes formal compliance, so doing formal compliance does not represent materiel compliance.
Internal revenue Service defines the formal compliance in three conditions. They are filling compliance, payment compliance, and reporting compliance. These Criteria based on regulation 198/PMK.03/2013 refer to KEP-213/PJ/2003. Based on regulation 198/PMK.03/2013 taxpayer"s compliance is explained as filling accurate and complete annual tax return before the due date, the rightness of periodic tax report, paying of all tax debt (taxpayers never get notice of tax collection), not having tax arrears except it is permitted by tax officer to pay in installment, never commit a tax criminality at least for 5 years.

Tax Amnesty
Based on UU RI 17 year 2016, Tax Amnesty is an elimination of a tax liability relating to a previous tax period or periods. It will free charge from interest and criminal prosecution including administrative or punishment if they inform both taxpayer"s liabilities and assets. Every taxpayers may get the forgiveness by disclose their asset. The type of amnesty is forgiveness for tax liability until the end of fiscal year. In exchange for forgiveness, taxpayers must pay redemption voluntarily.
There are some purposes of Tax Amnesty. First, it is for accelerating the economic growth and economic restructuring by property transferring. That will impact on increasing of domestic liquidity, improved exchange rate, lower interest rate and increased investment. Second, it is for encouraging tax reformation; equitable tax system, extended data basis it make data more valid, comprehensive, and integrated. Finally, the main aim of Tax Amnesty is for increasing tax revenue.

Theoretical Framework
The main purpose of this research is to predict tax noncompliance using Theory of Planned Behavior model and examine whether the Tax Amnesty will interact with the TPB construct or not. According the objective of the research, in this study, there will be three steps.
First, test the influence of attitude toward noncompliance, subjective norms, and perceived beha-vioral control on noncompliant intention. Second is analyzing the effect of noncompliant intention tax noncompliance. Third is examining the moderation effect tax between Tax Amnesty and the constructs of Theory of Planned Behavior. Graphically, the theoretical framework is shown in Figure 3 (see the variables identification).
The previous theoretical framework shows how every variable will have interaction each other. Based on the previous study and theoretical framework, below are the hypotheses of study.

The Influence of Attitude toward noncompliance to Intention
Attitudes are predispositions to respond favorably or unfavorably to behave noncompliance. Attitude is a person"s feeling or way of thinking that affects a person"s behavior. Attitude comes from their internalized values. Various studies have demonstrated the role of personal norms for taxpaying behavior. That was done by Bobek and Hatfield (2003) and Elia (2007) that attitude toward noncompliance was significantly influent behavioral intention. The other side Widi and Argo (2010) proved that attitude toward noncompliance did not affect behavioral intention. Based on this consideration, the hypothesis is formulated as follow: H1: Attitude toward noncompliance significantly influence to noncompliant behavioral intention.

The Influence of Subjective Norms to Intention
In many condition, taxpayers behavioral is influence by other peoples. The decision of taxpayer to behave to behave compliance or non compliance is motivated by referent people. The previous study showed that subjective norms had high influence to noncompliant intention (Bobek and Hatfield 2003;Widi and Argo 2010;Elia 2007). In other hand, Widi and Bambang (2012) did not support the influence of subjective norms and noncompliant intention. Based on this consideration, the hypothesis is formulated as follow: H2: Subjective norm significantly influence to noncompliant behavioral intention.

The Influence of Perceived Behavioral Control to Intention
Perceived Behavioral Control (PBC) is outcomes of control belief. Control belief lead to the perception taxpayers have or do not have capacity to carry out the behavior. PBC refer to taxpayer"s belief about the presence or absence of factors which facilitate or impede noncompliance.
The higher control toward taxpayers, noncompliance may decrease. Empirically, the relation between perceived behavioral control and noncompliant behavioral is proven. But two previous studies oppose the influence of perceived behavioral control and noncompliant behavioral (Bobek and Hatfield 2003;Widi and Argo 2010). Based on this consideration, the hypothesis is formulated as follow: H3: Perceived behavioral control significantly influence to noncompliant behavioral intention.

The Influence of Perceived Behavioral Control to Noncompliance
Perceived Behavioral Control (PBC) is outcomes of control belief. Control belief lead to the perception taxpayers have or do not have capacity to carry out the behavior. PBC refer to taxpayer"s belief about the presence or absence of factors which facilitate Source: Ajzen (2005)  or impede noncompliance. The higher control toward taxpayers, noncompliance may decrease.
Empirically, the relation between perceived behavioral control and noncompliant behavioral is proven. But two previous studies oppose the influence of perceived behavioral control and noncompliant behavioral (Bobek and Hatfield 2003;Widi and Argo 2010). Based on this consideration, the hypothesis is formulated as follow: H4: Perceived behavioral control significantly influence to noncompliant behavior.

The Influence of Intention to Noncompliance
Intention connects the tax noncompliance and its factors determinant. Generally, intention significantly influences tax noncompliance. Based on Ajzen (2005: 101) intention can accurately predict a variety of corresponding action tendencies.
Empirically, Intention is proven related to tax noncompliance (Bobek and Hatfield 2003;Widi and Bambang 2015;Elia 2007). Based on this consideration, the hypothesis is formulated as follow: H5: Tax noncompliant intention significantly influence to noncompliance The Influence of Tax Amnesty to Theory of Planned Behavior constructs Tax amnesty as compliance driver will increase taxpayer"s compliance. Previous studies found that tax amnesty had great impact on compliance. Advocates of amnesties argue that successful amnesties were accompanied by increased enforcement following the amnesties James et al. (1990).
Generally, there are factor of various kinds can influence intention, but this influence is usually moderated by more specific condition. An obvious potential moderating variable relate to behavioral context is situational constraint. Hopefully, Tax Amnesty as a recent fiscal policy can moderate the attitude and personal trait. If an amnesty is accompanied by these features, then simple model of noncompliance will decrease.
Based on this consideration, the hypothesis is formulated as follow: H6: Tax amnesty moderates relationship between attitude and noncompliance. H7: Tax amnesty moderates relationship between subjective norms and noncompliance. H8: Tax amnesty moderates relationship between perceived behavioral control and noncompliance.

RESEARCH METHOD Population, Sample, and Sampling Technique
Population is a large collection of individuals or object that is the main focus of study. Population of this research is individual taxpayers in KPP Pratama Sukomanunggal. It is located in Bukit Darmo Golf No.1 Surabaya. To generate the population, sample is needed. Sample is a group of subjects or participants from whom data are collected. Sample of this research is expectantly more than 100 Individual taxpayers who come to tax office and voluntarily want to become respondents.
The method for choosing the subject is Convenience Sampling. Convenience sampling is one of non probability sampling method which subjects are selected. Besides, Convenience Sampling is availability or ease to access. Using Convenience Sampling, sample is more objective, valid, and reliable. Convenience Sampling usually represents the population, easy to analyze and interpret the result, give generalization possible to similar object Millan (2008: 118).

Data Collection Procedures
Related with the research questions on the background of the study, the data collection procedure is self-administered survey that primary data are collected by survey to respondents for collecting individual taxpayer"s opinion Jogiyanto (2012: 140). Taxpayers will give the opinion based on the questionnaire instruction. Then, qquestionnaire is handed out to individual taxpayers who come to tax administration office in KPP Pratama Sukomanunggal. Researcher waits for the taxpayers to fill in the questionnaire and collects the questionnaire directly.
Independent variables (X) are attitude toward non compliant behavior, subjective norms, perceived behavioral control. Dependent Variable (Y) is tax noncompliant behavior. Intervening Variable is behavioral intention. Moderating Variable is Tax Amnesty

Operational Definition and Variables Measurement
In this research, behavioral variables use Theory of Planned Behavior (TPB). All variables classify as la-tent variables. Latent variables are variables which not directly observed but inferred from other variables that are observed, Latent variable is measured using indicators called manifest variables Byrne (2010: 4). Moreover, this study needs taxpayers to give their opinion, so the research"s instrument is questionnaire and the respondent"s should answer the questionnaire as well as interview transcript.
Type of the questionnaire is interval and numerical scale by giving score; 1 until 7 in every statement. The respondents must give suitable score for each item between 1 until 7. The information of scale"s representation is explained below: Scale 1: Strongly disagree Scale 2: Disagree Scale 3: Approximately disagree Scale 4: Neutral Scale 5: Approximately agree Scale 6: Agree Scale 7: Strongly agree Tax Noncompliant Behavior Noncompliant Behavior of Individual taxpayers are noncompliance of individual taxpayers who do not obey the tax liabilities. Noncompliant behavior as latent variable is measured by IRS"s definition of tax compliance replicated by study of Brown and Mazur (2003). Every item is elaborated which refer to definition of formal compliance which regulated by 198/PMK.03/2013.
The tax compliance can be measured indirectly using indicators which consist of 7 items. They are never get fine for late reporting periodic tax return, late reporting annual tax return, never get fine for underpayment tax, never commit a tax criminality, never correct the fiscal more than 10%, feel guilty while doing tax fraudulent, and thinking that doing noncompliance is not unethical actions. The respondents must give suitable score for each item between 1 until 7.

Intention to Tax Noncompliance
Intention is a planning of someone to perform or not perform a specific behavior it is immediately determinant of their action. So, Intention represents taxpayers willing to pay tax, to comply all their tax liabilities. Measurement of intention to noncompliant behavior as latent variable uses respondent"s opinion replicate Mustikasari (2007).
There are two postulates which represent intention variable"s indicators; (1) preference and (2) decision to comply or do not comply tax regulation. The respondents must give suitable score for each item between 1 until 7.

Tax Amnesty
Tax Amnesty is an elimination of a tax liability relating to a previous tax period or periods. It will free charge from interest and criminal prosecution including administrative or punishment by informing both taxpayer"s liabilities and assets. In exchange for forgiveness taxpayers must pay a defined amount voluntarily.
There are 7 indicators to measure Tax Amnesty refer to James (2009). The questions include desire to take opportunity, considering its benefits, knowing the certain tax liabilities, good reference from others, sufficient facilities to take amnesty, and tax officer"s guidance. The respondents must give suitable score for each item between 1 until 7.

Attitude toward Noncompliance
The attitude toward noncompliance is determined by the taxpayer"s evaluation of outcomes of associated with noncompliance behavior and belief strength of these associations. According to Ajzen (2005: 124) Estimation of the attitude toward noncompliance is based on the person"s accessible beliefs about noncompliance by multiplying belief strength and outcome evaluation, and summing the resulting products. Attitudes are algebraically modeled as follows: AB= a ∑ biei.
(1) Where: AB = the attitude toward B; bi = the subjective probability that performing behavior B will lead to outcome i; ei = the evaluation of outcome i.
The measurement of attitude toward behavior is used 7-point scales which replicated Elia (2007). To measure belief strength and evaluate the outcome, respondent should decide rate of expectancy in every indicator. They are the willingness to pay less tax, allowance for tax investigating fund, feeling a loss upon tax system, the perception of tax benefit is not transparent, and the higher bribery cost than economical tax. The respondents must give suitable score for each item between 1 until 7.

Subjective Norms
Normative beliefs are the beliefs that underlie subjective norms. Normative beliefs refer to people whom motivate, give taxpayers social pressure to comply and the most referents to perform compliance or non compliance. Subjective norms can be assessed by asking respondent to judge how likely it is that most people who are important to them would support their behavior. Symbolically, Subjective norm is express in equation below: SN= a ∑ nimi.
(2) Where: SN = the subjective norm; ni = normative beliefs concerning referent i; mi = the person"s motivation to comply with referent i.
The measurement of subjective norm is used 7point scales which replicated Elia (2007) by 4 indicators to measure this subjective norm. The indicators measure the most reference people while fulfill taxpayer"s liabilities such as friends, family members, tax consultant, and tax officers. The respondents must give suitable score for each item between 1 until 7.

Perceived Behavioral Control
Perceived behavioral control becomes a function belief. That is beliefs about the presence or absence of factors that facilitate or obstruct performance compliance or non compliance. Beliefs about resources and opportunities to comply on don not comply may be as underlying perceived behavioral control. It can be measured directly by asking taxpayers whether they believe that they are capable of performing compliance non compliance and whether they believe that doing so is under other"s control. The relation between control belief and perceived behavioral control in symbolic form are showed below: PBC = ∑cipi.
(3) Where: PBC = perceived behavior control ci = control belief that a given factor i; pi = the power of factor i to facilitate or inhibit performance of the behavior.
Perceived behavioral control is measured with 3 indicator replicated Elia (2007). They are the possibility of having tax audit by tax authorities, the possibility of getting tax penalties, and the possibility of third-party reporting. The respondents must give suitable score for each item from 1 to 7.

Data Analysis Technique
This research use statistic technique in analyzing the variables. The data were processed by using SPSS 20.0 and SmartPLS 2.0. The analysis process is described as the following.

Descriptive analysis
Descriptive statistic includes the mean, variance, maximum, minimum, sum, range, and percentage. Cross tabulation is also used to understand com-prehensively about the respondent"s profile. This technique is especially used to overview the extent of tax noncompliance and its determinant"s. This descriptive analysis is analyzed by SPSS 20.0.

Structural Equation Modeling
Structural Equation Modeling is applied to answer the hypothesis. SEM-PLS specifically uses SmartPLS 2.0. Besides it, to describe the data and research subject, SPSS 20.0 is used.
There are four steps to analysis PLS-SEM;

Defining Theoretical Model
Structural equation model is based on causal relationship which assumed that every change of variable will effect to other variable. Causal relationship is strong based on theoretical justification not because analysis method which used in the research. So, the most important of structural equation model must be parsimony with concise theoretical model.

Developing The Overall Measurement Model (Outer Model)
The second step is measuring path analysis and its structural equation. Path analysis is a set relationship between exogenous, endogens variables and manifest. Outer Model defines how each block of indicators relates to its latent variables. Quality of a measurement model (outer model) can be checked by validity and reliability. Validity test is assessed by examining convergent validity and discriminant validity while reliability test is assessed through composite reliability (Chin 2014: 316). The validity and reliability test had been explained in the validity and reliability test point.

Assessing Structural Model Identification (Inner Model)
Evaluation of PLS models should apply nonparametric prediction-oriented instrument. To that extent, (1) the R-square is used for dependent latent variables, (2) Stone-Geisser is tested for predictive relevance, and (3) t-statistic is used to assess the significance of constructs.
R-square has criterion; the model is good if R2 score is 0.67, moderate if R2 score is 0.33, and weak if R2 score is 0.19 (Imam Ghozali 2011: 42). Besides looking at R-square as a criterion for predictive value, F-square can also be used to evaluate structural model in PLS. F-square determines the effect size of latent variables. F-square value indicates how big the power of latent variable is influenced in structural level. The power of latent variable is "big" if F2 is 0.35, moderate if F2 is 0.15, and weak if F2 is 0.02.

Interpretation and Model Modification
After model is accepted, the researcher may consider modifying the model. The purposed of improvement the model is to repair theories explanation or goodness-of-fit. The measurement is done by modifying indices score.
To answer whether the hypothesis is supported or not, it should compare t-value and significance of parameter, which is 5%. For the regression result, the hypothesis is supported or significant if it has t-statistic more than 1.96, but if tstatistic is less than 1.96, the hypothesis is not supported or not significant.

DATA ANALYSIS AND DISCUSSION
The Questionnaire submission is more than one month, which started from 31 th October 2016 until 2 nd December 2016. In this research, there were 150 questionnaires delivered to respondents. The entire questionnaire came back but there are 5 incomplete questionnaires. Incomplete questionnaire is out of sample because the respondents did not fill completely and give their opinion clearly. Therefore, only 145 questionnaires could be analyzed. Hence, the Respondents are divided into 4 categories. They are age, gender, formal and informal education, and the length of time became taxpayers. Table 1 explained shortly how the demographic of research subject.

Research Subject Based on Age
Looking on the age of taxpayers, the smallest number of respondents are at the age of 26, which are only 2 respondent or 1.4%. This taxpayer does not realize that they have done noncompliance. The highest numbers are 27 to 35 years old, which are 71 respondent or 49% in which 36 respondents are male and 35 respondent are female. In this age, taxpayers begin to look for the way to do noncompliance and think how they can pay smaller tax. Table 1 demonstrates that the male respondents are higher than female respondents. It means that male is more noncompliant than female. Some research supports that female is complier than female. David et al. (1994) defines female has a different attitude through ethic and ethic code. The maturity also decreases noncompliance.
After taxpayers are 27 years old, the percentage of noncompliance decreases gradually. Kohlberg model defines moral maturity influence someone in decision making, whether they do favourable or unfavourable action, indeed noncompliance.

Research Subject Based on Education
There are two kinds of educations, they are formal and informal. Data respondents based on formal and. Formal education consists of doctoral degree, master degree, bachelor degree, and senior high school. Informal education is divided into 5 categories. They are brevet, course, training, seminar, and others.
Looking on the formal education as Table 1, the most noncompliant are bachelor degree, which are 53.8% or 78 respondents in which male respondents are 46 and female respondents are 31. Individual taxpayers who have passed bachelor degree and they are between 27 until 35 years old are the highest numbers on doing noncompliance, which is 28.3%.  Moreover, on the informal education, training is the highest numbers of respondent"s informal education that they had joined. They are 74 respondents or 51%, which 43 respondents are male and 31 respondents are female. Brevet is the smallest numbers of respondent"s informal education which are 8 respondents or 5.5% in which male respondents are male and 7 respondents are female.
The higher level of education should decrease noncompliance. On the contrary, the phenomena shows bachelor degree has highest number on doing noncompliance. Thus proves that having high education does not guarantee taxpayers will more comply.

Research Subject Based on Length of Time to be taxpayers
Refers to Table 1, the highest numbers are between 11 to 20 years became taxpayers which are 40.7% or 59 respondents in which 37 respondents are male and 22 respondents are male. On the other hand, the smallest one is more than 30 years old, which are only 8 respondents or 5.5%. This can be concluded that the higher age will decrease noncompliance because the taxpayers have known how to fulfil their tax liabilities both administrative or payment process.
Next step is counting the mean and analyzing the opinion of respondents and the result is shown as follow: Based on the questions relate to tax noncompliance, individual taxpayers disagree doing noncompliance is unethical actions. They think that doing noncompliance is ethical actions because it does not bring a loss to government directly. Fraud triangle defines doing noncompliance while thinking its noncompliance is not unethical action is categorized as rationalization. So, even though the taxpayers did not admit their tax noncompliance, they cannot avoid the punishment. Table 2 indicates individual taxpayers strongly disagree if they did not correct the fiscal more than 10%. Taxpayers strongly agree they ever correct the fiscal more than 10% and it becomes habit. They are not afraid if they are investigated because tax low in Indonesia provides some choice to be used in calculation tax and Seeing on taxpayer"s intention, they very agree when they do tax noncompliance, firstly they decide to not comply with its tax regulation. Before deciding, they will incline to do tax compliance after deciding to not comply with tax regulation. The reason why they don"t comply is a feeling a loss upon tax system, the influence of family members to avoid tax liabilities, and arguing that tax authorities will not audit taxpayers who have small tax expense. Tax authorities prefer to audit corporate taxpayers than individual taxpayers. Besides,  taxpayers strongly agree that Tax Amnesty motivate them to comply.

PLS Analysis
Model evaluation in PLS-SEM is a two-steps process. First, evaluating the measurement model is called outer model evaluation. Measurement model is evaluated in the term of validity and reliability. Second, examining the structural model, it is also called inner model evaluation. Evaluation of inner model can be assessed by examining the F-square and coefficient determination (R 2 ) Ravand and Baghaei (2016).

Outer Model Evaluation
Outer Model defines how each block of indicators relates to its latent variables. Quality of a measurement model (outer model) can be identified through validity and reliability test. Validity test is assessed by examining convergent validity and discriminant validity while reliability test is assessed through composite reliability (Chin 2014: 316). The outer model evaluation is explained in Table 3.
Convergent validity should be fulfilled by executing invalid variables must from model until the entire factor loadings are above the acceptable level of 0.50 and check the quality is through average variance extracted (AVE). All of the latent variables have high average variance extracted (AVE), which is more than 0.5. AVEs of 50% mean that the construct explains at least half of the variance of its observed variables. The highest AVE is subjective norms, which 0.983. This indicates that construct of subjective norms (indicators) is able to explain 98.3% of the variance of the its observed variables.
Reliability test in PLS-SEM can be assessed by composite reliability (CR) and Cronbach"s alpha. According to the composite reliability of model evaluation which it is explained in Table 4, coefficients of composite reliable are above 0.70, so all of constructs are reliable. Then, all of the latent variables have Cronbach alpha value more than 0.70, so all variables are reliable. In conclusion, all of latent variables have supported reliability and validity. Table 3 shows that R-Square for behavioral intention before including moderating effect is 0.362 and R-Square for tax noncompliance is 0.014. It means that firstly, the ability of independent variable"s indicators to explain (attitude, subjective norms, perceived behavioral control, and Tax Amnesty) behavioral intention is 36.2%. Then, 63.8% of behavioral intention is influenced by others factors. It also indicates that the structural model is "more than moderate".

Inner Model Evaluation
Secondly, the ability of behavioral intention to explain tax noncompliance is only 0.014 or 1.4%. Moreover, 98.6% of tax noncompliance is explained by others factors which is not included in this model.
The model seems to improve after adding the moderating variable, which is Tax Amnesty. The improvement is shown by the higher R-Square, which is 0.400 or 40%.
To define whether the hypothesis is supported or not based on t-statistic. If t-statistic score is more than 1.96, it means that the hypothesis is supported. If t-statistic score is less than 1.96, hypothesis is not supported. In conclusion, the hypothesis 1, hypothesis 2, hypothesis 5, hypothesis 6, and hypothesis 7 are supported. Different results, hypothesis 3, hypothesis 4, and hypothesis 8 do not have significant effect.

The Influence of Attitude toward Noncompliance to Behavioral Intention
Attitude toward noncompliance are predispositions to respond favorably or unfavorably to behave tax noncompliance. Attitude is also defined as a person"s feeling or way of thinking that affects taxpayer"s noncompliance. Attitude is not built but it comes from taxpayer"s internal values. Theoretically, Ajzen (1991) claimed that attitude is an indication of noncompliance which elaborated in Theory of Planned Behavior. According to Theory of Planned Behavior, an attitude toward noncompliance is one of determinant factor influencing behavioral intention which indirectly will affect to tax noncompliance (Ajzen 1991). If the taxpayers have strong attitude toward noncompliance, the intention to do noncompliant behavior will be higher.
Based on the empirical result, Table 4, attitude toward noncompliance is proven significantly influence to individual taxpayer"s behavioral intention. Therefore this result of study is linear with Theory of Planned Behavior which demonstrates attitude toward noncompliance influences behavioral intention. Then, behavioral intention directly influences tax noncompliance.
As the characteristic of tax, which is enforced contribution and the purpose is for financing government operational expenditure, taxpayers will look for the way to pay tax less than their liabilities. Beside it refers to individual taxpayer"s response rate, when taxpayers fulfill their tax abilities they will consider some factors. The factors that will be considered is consist of feeling a loss upon tax system, the willingness to pay less tax, the higher bribery cost than economical tax, amount of allowance for tax investigating fund, and taxpayer"s perception of tax benefit is not transparent.
Five factors above as instruments to measure attitude toward noncompliance. Factor, which has a big effect to avoid tax abilities, is feeling a loss upon tax system while factor which has a small effect to avoid tax abilities is taxpayer"s perception. Taxpayer precept tax revenue is not used transparently. However this taxpayer"s perceptions lead to perform tax evasion or noncompliance.
There is also evidence showing that attitude do impact on taxpaying behavioral intention. In experimental study by Bobek and Hatfield (2003) to predict noncompliance behavior in paying tax, proved the sum of the attitude beliefs was always highly significant. Even though, the significance of the individual beliefs varied by scenario.
Furthermore, the finding seems consistent with Widi and Bambang (2015) and Elia (2007) which tax professional as a representative of corporate taxpayer supported tax compliance as contribution to government. The respondents agreed that attitude influenced behavioral intention. Contrast with the finding of Widi and Argo (2010) proved that attitude did not significantly influence behavioral intention.

The Influence of Subjective Norm to Behavioral Intention
This research confirmed that subjective norms influence to behavioral intention. Based on individual taxpayer"s responses, they will consider advices of family members, tax consultant, their friends, and tax officers. The first reference people who will influence individual taxpayers are family members. Family member is the most confidential people.
Taxpayers have responsibility to fulfill their need. Therefore, taxpayer will consider their advice and family member"s need to do noncompliance. On the contrary, tax consultant is the last consideration while individual taxpayers do noncompliance. One reason for justification, not all of taxpayers may have tax consultant. Theory of Planned Behavioral proposes subjective norms may have effect to behavioral intention.
Theoretically, subjective norm is one of indicators of that influence behavioral intention which indirectly influence tax noncompliance. Subjective norms are a reference from other peoples that in-fluence someone to do something (Ajzen 1991). Subjective norms refer to person who will give motivation to decide noncompliance. In many condition, taxpayers behavioral is truly influenced by other peoples.
Refers to Table 4, subjective norms are proven significantly influence behavioral intention. Therefore, this result of study is linear with Theory of Planned Behavior which indicates subjective norms influence behavioral intention. Then, behavioral intention directly influences tax noncompliance. Consistent with the result of Elia (2007), subjective norms influenced behavioral intention to do noncompliance.
In that research disclosed that tax professional"s recommendation had a high impact to corporate taxpayers to do noncompliance. Because of all the amount of tax expense is calculated by tax professional. The other study"s result that supports this finding is Bobek and Hatfield (2003). In contrast, Bobek and Hatfield (2003) did not successfully prove this expectation.

The Influence of Perceived Behavioral Control to Behavioral Intention and Tax Noncompliance
Based on the result of Table 4 empirically, proved perceived behavioral control does not influence behavioral intention. Therefore, it can be concluded that the third hypothesis and fourth hypothesis are not significantly supported. As Table 4 shows, this can happens because individual taxpayer is not afraid of some factors. They are the possibility of having tax audit by tax authorities, the possibility of getting tax penalties, and the possibility of thirdparty reporting. They only consider the advantages of noncompliance.
This result of study is contrast with Theory of Planned Behavior, which is claimed that perceived behavioral intention influences behavioral intention and tax noncompliance. Theoretically, perceived behavioral control refers to degree of controlling taxpayer"s perception they have to involve in a particular compliant behavior. The higher perceived behavioral control expectedly can decrease the behavioral intention to do noncompliance and increase tax compliance, although it does not seem effect the behavioral intention.
The prior studies also find that perceived behavioral control does not influence behavioral control (Widi and Argo 2010). Bobek and Hatfield (2003) prove that perceived behavioral control does not effect to behavioral intention but directly influence tax noncompliance. On the other hand, perceived behavioral control is significant influence both intention or tax noncompliance Widi and Bambang (2015) and Elia (2007).

The Influence of Intention to Tax Noncompliance
Based on Table 4 empirically, proved that behavioral intention has significantly influence to tax noncompliance. Therefore, it can be concluded that the fifth hypothesis is significantly supported. Behavioral intention of individual"s taxpayer indicates the inclination and decision to do tax non compliance influence to tax compliance. Both of the factors will lead individual taxpayer to do tax non compliance.
Based on Table 4 demonstrates more than half individual taxpayers decide to not comply with tax regulation. Then, they rarely agree that they incline to do noncompliance. This means that the behavioral intention is higher to do noncompliance, tax noncompliance will be higher. Hence, to decrease tax noncompliance, behavioral intention should be controlled using such regulation such penalties, low rate, and others.
This result of study also supports Theory of Planned Behavior which defines that person"s behavior is determined by intention to perform the behavior. Intentions are a function of three basic determinants; one reflecting personal nature, one social in nature, and a third dealing with issues of control.
All of the determinant factors is presented in attitude toward behavior, subjective norms, perceived behavioral control interact each other predict intention. Then, the intention will translate into action (Ajzen 2005: 117). Nurul and Titik (2010) intended to do something will increase the desire of behavior. Mostly, there are some studies that also prove there are significant influence between intention and behavior Widi and Bambang (2015); Widi and Argo (2010);Elia (2007).

Discussion of Moderating Effect
According to Ajzen (2005: 42) there are three broad categories of factors that interact with attitudes or personality traits. They are situational circumstances surrounding performance of the behavior, characteristic of the individual and secondary characteristics of the disposition.
First, situational factor is the different condition in some situation but consistent with each other such as regulation and policy. Hence, Tax Amnesty as one of recently fiscal policy of government can classified as situational factors which may moderate tax noncompliance and its determinant factors.
Second, characteristic of the individual relates with individual difference in various conditions. Individual difference is considered as moderator is based on the assumption that consistency may be expected for some individuals but not for others such as self-monitoring, private self-consciousness, need for cognition.
Third, secondary characteristics of the disposition may influence the relationship between general attitude and specific behavior for example internal structure, reflection, accessibility, vested interest, involvement, importance, confidence, direct experience. So, theoretically, the most suitable moderator between behavior and it determinant factor is situational factor. Hopefully, Tax Amnesty as compliance driver will increase taxpayer"s compliance.
Empirically, In this research Tax Amnesty is tested whether succeeded or not influence tax noncompliance. Tax Amnesty is measured by seven points. They are good reference from other people, the desire to take opportunity, consideration of its benefits, knowing the certain tax abilities, good reference from others, and the sufficient information about Tax Amnesty, sufficient facilities, and tax officer"s guidance for following Tax Amnesty. Based on descriptive statistic as Table 2 the highest factor that gives big influence is good reference from others and sufficient information about Tax Amnesty.
Based on hypothesis testing, Tax Amnesty empirically can moderate the relationship between attitude toward noncompliance and intention and between subjective norms and behavioral intention. By contrast, Tax Amnesty evidently cannot moderate the relationship between perceived behavioral control and intention. Consequently, there are 2 hypothesis is not supported which sixth hypothesis and seven hypothesis. But, hypothesis 8 is supported. Thus, the result of study can support and develop Theory of Planned Behavior.
Previous studies found that Tax Amnesty had great impact on compliance. Advocates of amnesties argue that successful amnesties were accompanied by increased enforcement following the amnesties James et al. (1990). Actually, Tax Amnesty will drive tax compliance higher but the long run effect cannot be guaranteed Koch et al. (2015). The previous study is done by James et al. (2009) proved that Tax Amnesty had a high effect on revenue but the effect is about five years after implementing Tax Amnesty.
By this, it can be concluded that Tax Amnesty is not the most suitable regulation to increase tax compliance. After implementing Tax Amnesty, gov-ernment may have program which can keep the taxpayer"s compliance.

CONCLUSION, IMPLICATION, SUGGES-TION, AND LIMITATIONS
This research purposes to examine noncompliant behavior toward taxpayers using Theory of Planned Behavior constructs. Besides that, this also analyses the moderation effect of Tax Amnesty to attitude, subjective norms, and perceived behavioral control. These research objectives are applied to the individual taxpayers by convenience sampling. From the discussion in the last chapter.
Based on the research result, there is a significant influence between attitudes toward tax noncompliance and subjective norms to behavioral intention. In contrast, perceived behavioral control does not significantly influence behavioral intention.
This study empirically does not only test the influence of perceived behavioral and behavioral intention but also the influence of perceived behavioral and tax non compliance. As the discussion above, perceived behavioral control does not significantly influence tax noncompliance. Therefore, perceived behavioral control does not significantly influence both behavioral intention and tax non compliance.
Empirically, behavioral intention is significantly influence tax noncompliance. Hence, behavioral intention can be the mediation between tax noncompliance and its factor determinant.
According to the moderating effect analysis, generally, Tax Amnesty moderates the relationship Theory of Planned Behavior construct. Tax Amnesty is proven can influence the relationship between attitude toward non compliance, subjective norms and behavioral intention. Nevertheless, the relationship between perceived behavioral control and behavioral intention does not seem to have significant relationship. As conclusion, all of the research result supports Theory of Planned Behavior but Tax Amnesty may not moderate its constructs.
However, this study is designed well; this study is not free from limitation. This limitation of study explains three important points cannot be solved by the researcher. First, variables which are used as determinant of tax noncompliance is only utilized by Theory of Planned Behavior construct. Even though, there are many variables can influence tax noncompliance. Consequently, the behavior that taxpayers portray under this method may not be a truthful representation of their actual behavior.
Second, in this study tries to improves the model. Besides using Theory of Planned Behavior construct adding Tax Amnesty as moderating variables. Moreover, the model is categorized "weak" and needs to be exercised deeply.
Third, Furthermore, the focus of this study is on individual taxpayers. So, the result of study cannot generalize with corporate taxpayers because corporate taxpayers may have different opinions, perception, and behavior from the individual taxpayers.
Based on the result of and limitation of this study, there are recommendations for government and future researcher. For government, relates to Tax Amnesty which recently implemented, as the result of this research, this regulation may change the taxpayer"s behavior. However, the government should provide sufficient facilities, clear information and guidance.
The interesting one, Tax Amnesty not only moderates the behavioral intention and its determinants factor but also decrease tax noncompliance directly. For future research is expected to conduct several improvement by utilizing other or all types of taxpayers in orders to get more comprehensive result. Moreover, not only exercises the causality, but also comparing taxpayer"s behavior between others country. Use this study limitation to improve the future research. The next researcher is expected to be able to explain comprehensively the research problem.