Tax Control and corporate VAT compliance:

An empirical assessment of the moderating role of tax strategy

Maarten A. Siglé a, b
m.sigle@my.nyenrode.nl
Corresponding author, Laan op zuid 45, Rotterdam, The Netherlands

Sjoerd Goslinga c, b
s.goslinga@fsw.leidenuniv.nl

Lisette E. C. J. M. van der Hel a

l.vdhel@nyenrode.nl

Ryan J. Wilson d
ryan-wilson@uiowa.edu

a Nyenrode Business University, Straatweg 25, 3621BG, The Netherlands

b Netherlands Tax Administration1
c Leiden University, Wassenaarseweg 52, 2333 AK, Leiden, The Netherlands
d University of Iowa, Iowa City, IA 52242, USA

Paper is forthcoming in Journal of International Accounting, Auditing and Taxation

1 Maarten Siglé and Sjoerd Goslinga work for the Netherlands Tax Administration (NTA). Their contribution to this paper is
written in a personal capacity and does not necessarily reflect statements and/or opinions of the NTA. For Ryan Wilson and
Lisette van der Hel there is no declaration of interest. This research did not receive any specific grant from funding agencies in
the public, commercial, or not-for-profit sectors.

Electronic copy available at: https://ssrn.com/abstract=4636196

Highlights

• Tax authorities increasingly advocate for better tax control of large organizations.

• Organizations with conservative tax strategies utilise their TCF to prevent

unintentional tax non-compliance.

• Organizations with aggressive tax strategies utilise their TCF to minimize tax without

increasing the risk of audit adjustments.

Keywords: VAT compliance, tax strategy, internal control, tax control framework

Electronic copy available at: https://ssrn.com/abstract=4636196

Abstract: Tax control frameworks (TCF) of higher quality are seen by tax authorities and the
OECD as a prerequisite for corporate tax compliance. However, a higher quality TCF can also
enable organizations to reduce their tax burden by providing information that allows them to
make the best use of opportunities within the boundaries of the tax law. We investigate the
effect of TCF quality on tax compliance by looking at the tax strategy of the organization and
whether the organization unintentionally or intentionally fails to comply. We focus on Value
Added Tax (VAT) and test our hypotheses using a sample of large organizations, using a
combination of survey data and tax audit results from the Netherlands. Our results show that a
TCF of higher quality is positively associated with VAT compliance, resulting in both less
unintentional and less intentional errors requiring tax adjustment. For organizations with a more
conservative tax strategy, we find that the quality of the TCF does not affect the level of
intentional non-compliance. For organizations with a more aggressive tax strategy however the
level of intentional non-compliance is conditional upon the quality of the TCF, with a lower
(higher) quality TCF leading to more (less) intentional non-compliance.

Electronic copy available at: https://ssrn.com/abstract=4636196

1. Introduction

The component of internal control that assures accuracy, timeliness, and completeness

of tax returns and all other types of tax disclosures is known as the Tax Control Framework

(TCF) (OECD, 2013). The TCF supports organizations in controlling its tax risks and includes

building blocks for managing tax such as tax strategy, tax policy, and roles and responsibilities

relating to tax functions (OECD, 2016). The quality of each organization’s TCF exists on a

continuum. A high quality TCF is assumed to detect mistakes and possible tax risks, and thus,

help organizations to increase tax compliance. This assumption is central in the Organisation

for Economic Co-operation and Development (OECD) documents and reports on cooperative

compliance programs (OECD, 2013; 2016). Besides the role the TCF plays in such programs,

many tax authorities benefit from a higher quality TCF to the extent it helps improve the

efficiency and effectiveness of their tax audits. Tax authorities and the OECD view a high

quality TCF as a prerequisite for corporate tax compliance (Goslinga et al., 2019), and many

tax authorities actively advocate for large organizations (both profit and not-for-profit) to

establish and/or improve their TCF.

Depending on the tax strategy, a TCF might be directed towards tax compliance, but it

could also be directed towards facilitating more tax avoidance, utilising opportunities within

the boundaries of the tax law (Bauer, 2016; Gallemore & Labro, 2015). A better TCF helps tax

aggressive organizations stick to the letter of the law, but perhaps not necessarily to the spirit

of the law. In line with this reasoning, Blaufus et al. (2023) find that when organizations

perceive increased (audit) attention from the tax authority, they do not alter their tax planning

behaviour but, rather, invest in the quality of their TCF to (better) prevent errors. These findings

raise an important question for tax authorities: is it wise to promote high quality TCFs among

large organizations? Specifically, does a high quality TCF raise tax revenues through improving

tax compliance, or does it facilitate aggressive tax planning, and is the answer to these questions

Electronic copy available at: https://ssrn.com/abstract=4636196

conditional upon the tax strategy of an organization? Our paper is aimed at answering these

questions.

Important to this discussion is the distinction made in the tax compliance literature (e.g.

James & Alley, 2002; D’Ascenzo, 2015; Siglé et al., 2022) between unintentional and

intentional (non-)compliance. Unintentional non-compliance refers to unconscious violations

of tax laws and regulations due to ignorance about, or lack of understanding of, tax law and

regulations, or as a result of unintentional errors in the preparation of accounting information

or in calculations. In this context, advocating for a better TCF is clearly beneficial for the tax

authority because it prevents errors before they occur.

Intentional non-compliance ranges from outright tax evasion, which represents clear

non-compliance (e.g., deliberate under-reporting of income or sales), to tax avoidance, which

often involves complex tax strategies designed to exploit uncertainty in the tax code for the

benefit of the taxpayer. Tax avoidance strategies can run contrary to the spirit of the law, but

typically involve at least some legal justification for the position (Jallai, 2020). Large and

complex organizations have greater opportunity to engage in tax avoidance because of the

issues that arise from multi-state and international tax legislation (e.g., the treatment of

intangible assets). Gallemore and Labro (2015) find that organizations with greater

coordination needs arising from a dispersed geographic presence benefit more from high

internal information quality. They also find that organizations with high information quality

can achieve lower tax rates without altering their level of tax risk. In this context, a higher

quality TCF would seem to clearly benefit the organization, but the benefits for the tax authority

are less clear.

The focus of the OECD and tax authorities regarding the quality of the TCF is mainly

on so-called hard controls (i.e., the more formal and explicit internal controls in an

organization), as these controls will enable organizations to detect risks and errors. That is not

Electronic copy available at: https://ssrn.com/abstract=4636196

to say that the OECD and tax authorities neglect the role of soft controls (i.e., the more informal

controls that focus on organizational culture, such as the tax strategy), but the emphasis appears

to be on hard rather than soft controls. Through a tax strategy top management can communicate

the moral and ethical values of the organization (‘tone at the top’) (OECD, 2013). Well-

functioning soft controls can thus create an organizational environment in which intentional

non-compliance is considered non-acceptable.

To examine the implications of a better TCF for tax compliance, we use a combination

of survey data, and the results of value added tax (VAT) field audits performed by the

Netherlands Tax Administration (NTA). We use survey data to measure the quality of the TCF,

and the (self-reported) degree of tax aggressiveness to differentiate between more tax

conservative and tax aggressive organizations. This last measure allows us to examine how

(profit and not-for-profit) organizations characterize their approach to tax compliance. In this

sense it is a direct measure of organizations’ culture around taxes.

To measure VAT (non-)compliance, we use the results from the field audits as reported

by the lead tax auditors, focusing on the number of audit adjustments as well as the (scaled)

amount of audit adjustments. We use the number of audit adjustments as a proxy for more

unintentional non-compliance and the scaled amount of audit adjustments as a proxy for more

intentional non-compliance. This distinction is based on the assumption that the higher the

financial importance of a transaction (as measured in the scaled amount), the more likely it is

that the transaction was subject to more stringent internal controls and, therefore, to more

management involvement. As a result, it is also more likely that the audit adjustments concern

an issue that was intentionally created by management. Contrary to this, the issues concerning

small adjustments are less likely to have been intentionally evaluated by management. As such,

Electronic copy available at: https://ssrn.com/abstract=4636196

we use the number of small adjustments as a proxy for unintentional non-compliance. 2 To

amplify these differences, we additionally calculate the total number of adjustments below

€ 10,000 (on an individual adjustment level) as a proxy for more unintentional non-compliance

and the total scaled amount of adjustments above € 10,000 as a proxy for more intentional non-

compliance.

Many countries rely on a Value Added Tax (VAT, also called Goods and Services Tax),

as an important source of revenue (Bergman & Nevarez, 2006; Datt et al., 2017). Like other

forms of taxation, VAT can lead to various forms of non-compliance (De Mello, 2009; Schenk

et al., 2015). The EU publishes, on an annual basis, studies on the VAT GAP in the EU. These

reports state that non-compliance with VAT represents more than fraud and evasion, explaining

that “the VAT GAP also covers VAT lost due to, for example, (…) legal tax optimisation”

(European Commission et al., 2021, p. 10). VAT law provides opportunities for tax aggressive

organizations to try to minimize taxes paid. Examples of such opportunities are highly

dependent on specific regulations in national VAT law. Within the Dutch context, one way an

organization can be aggressive in its VAT strategy is by structuring intercompany transactions

in a way that maximizes VAT tax deductions. This can, for example, be done by using special

purpose vehicles and sale and lease back constructions. Because VAT is subject to both

unintentional and intentional non-compliance, the Dutch setting allows us to investigate the

interactions between the TCF, corporate culture around taxes, and intentional and unintentional

non-compliance. We assume that the organizational dynamics of VAT- and Corporate Income

Tax (CIT-)compliance are the same or similar enough to make our analyses meaningful to both

types of taxes. An advantage of studying VAT-compliance is that we can focus on a broader

set of organizations, including profit and not-for-profit organizations, where studies focusing

2
See section 3.2 for a complete discussion of the of our selection of these proxies for intentional and
unintentional tax avoidance.

Electronic copy available at: https://ssrn.com/abstract=4636196

on CIT are limited to profit organizations. From a government or tax authority perspective, both

profit and not-for-profit organizations are a point of focus when tax compliance is concerned.

We find, that a better TCF is negatively associated with the number and scaled amount

of VAT adjustments and thus, positively associated with VAT compliance. When we take the

moderating effect of the tax strategy into account, we find that the level of intentional non-

compliance is not affected by the quality of the TCF for organizations with a self-reported

conservative tax strategy. However, organizations inclined to be more aggressive in their tax

planning display more intentional non-compliance if they have a TCF of lower quality, and less

intentional non-compliance when their TCF is of higher quality. We take this result as an

indication that a high quality TCF helps organizations with a more aggressive tax strategy to

minimise tax without increasing the risk of audit adjustments.3 Finally, we do find that a better

TCF is negatively associated with more unintentional non-compliance for more conservative

as well as for more aggressive organizations.

Our study has direct relevance for tax authorities. The results can be seen as a reminder

to tax authorities not to focus solely on a high quality TCF, but to emphasize that a better TCF

must be coupled with a tax strategy aimed at compliance with both the letter and spirit of the

tax law. Our study also contributes in several ways to the academic literature. Ours is the first

study attempting to incorporate proxies for the difference between intentional and unintentional

non-compliance into the development and the empirical tests of our hypotheses, and in doing

so provides more insight into the role the TCF plays in tax compliance. Further, whereas most

previous tax compliance research focuses on corporate income tax (CIT), we focus on value

added tax (VAT), which is an equally important type of tax in terms of revenue (OECD, 2021)

but has, compared to CIT-compliance, scarcely been studied (Datt et al., 2017). Our study sheds

3
This interpretation assumes that aggressive organizations with a high quality TCF are also taking steps to avoid
VAT taxes, but their high quality TCF results in fewer adjustments. Unfortunately, our data does not allow us to
observe the ex-post VAT tax burden of the sample organizations.

Electronic copy available at: https://ssrn.com/abstract=4636196

new light on the role of intentional and unintentional tax non-compliance, as well as on the

interaction between organizational culture around taxes and TCF quality in VAT compliance.

The remainder of the paper is structured as follows. In the next section we provide

theoretical background on internal control and tax compliance and develop our hypotheses.

Then, we summarize the research method, and present the data analyses and results. In the last

section, we conclude with a discussion of our findings, their limitations, and suggestions for

future research.

2. Literature review and hypothesis development

Internal control is a process effected by the management of an organization with the aim

to provide reasonable assurance regarding the achievement of objectives related to operations,

reporting, and compliance with rules and regulations (Committee of Sponsoring Organizations

of the Treadway Commission (COSO), 1992). A TCF is that part of internal control that relates

to tax (Huiskers-Stoop & Gribnau, 2019) and is implemented and maintained by management

to achieve tax objectives, such as assuring the accuracy and completeness of the tax returns and

to identify and disclose tax risks (OECD, 2016). The term TCF was introduced after legislation

like the Sarbanes-Oxley Act (in 2002) increased governmental interest in control frameworks

in general. There is nothing novel about the substance of a TCF; all organizations always have

tax related control measures in place to be able to submit accurate and complete tax returns.

This means that every organization has, to some extent, a TCF (Hein, 2022). However, TCFs

can differ in their degree of formalisation (i.e., the degree to which tax control is an explicit and

formalised part of the overall control framework) and quality.

Defining quality of a TCF is inherently difficult because organizational controls must

adjust to the circumstances of the individual organization. A larger, more complex organization

will generally have more risk around tax compliance, which requires a larger investment in the

TCF. This fact alone does not mean that such an organization has a ‘better’ TCF, unless it is

Electronic copy available at: https://ssrn.com/abstract=4636196

compared to organizations of similar size and complexity. As a result, even when prescribing

TCFs, tax authorities and other stakeholders (e.g., OECD, 2016; Owens & Leigh-Pemberton,

2021) only provide building blocks for designing TCFs, often based on the COSO framework

(1992). Examples of such building blocks are the implementation of a clear tax strategy by top

management, a process that identifies the relevant tax risks, and monitoring of the correct

working of control mechanisms.

A high quality TCF has various benefits for organizations. First, it can help

organizations achieve higher levels of tax certainty (Gallemore & Labro, 2015; Plesner

Rossing, 2013). Second, it can protect organizations from reputational harm resulting from non-

compliance (Lanis & Richardson, 2012; Goslinga et al., 2019). Third, when a TCF is perceived

as a demand from the tax authority, for example within the context of a cooperative compliance

program, implementing a TCF can be a way to gain trust (and less audit activity) from the tax

authority (Parker & Gilad, 2011). Fourth, it can increase tax compliance by helping to prevent

unintentional non-compliance (OECD, 2014). Fifth, it can also increase tax compliance by

improving managerial decision making around tax planning strategies and thus preventing

overly aggressive tax behaviour (Bauer, 2016). From the perspective of the tax authority, these

last two benefits are the most important (Van der Hel & Siglé, 2015). Because we focus on the

effects of a higher quality TCF and thus on the relationship between the quality of the TCF and

tax compliance, whether organizations implement and/or improve their TCF for these reasons

is not relevant for our paper.

To test the role of internal control on tax compliance, we specifically focus on VAT

because it involves decision making both on an operational level and on a managerial level.

VAT compliance is related to transactions, and because the decision making regarding these

transactions takes place at the operational level, VAT is vulnerable to – from the perspective of

the management of the organization – unintentional errors. In this context, the above discussed

Electronic copy available at: https://ssrn.com/abstract=4636196

examples of building blocks of a higher quality TCF that focus on risk identification and

monitoring play an important role. At the same time, VAT compliance is related to (aggressive)

tax planning activities, a strategy decided at the managerial level. VAT avoidance can take

numerous forms to minimise tax payments, for example (re)structuring operations aimed at

maximally exploiting VAT exemptions and zero-rates, fragmenting business aimed at

maximally using certain thresholds, or manipulating the allocation of inputs to taxable supplies

(Schenk et al., 2015). In this context, the other example of a building block of a higher quality

TCF (i.e., the implementation of a clear tax strategy by top management) plays an important

role. Both types of non-compliance can be related, for example when aggressive tax planning

activities are the impetus for management to prevent unintentional errors since this could

undermine their efforts at more intentional strategic avoidance.

Better control can achieve higher VAT compliance by ensuring management is provided

with sufficient internal information to deal with the high level of complexity that is part of

making tax related decisions (e.g., Gleason et al., 2017) and because it increases the ability of

the organization to identify potential tax risks and to prevent these risks from becoming actual

errors (Siglé et al., 2022). Where tax errors do occur, a higher quality TCF detects and corrects

these errors before such errors are included in tax returns. Siglé et al. (2022) provide some

tentative empirical evidence that a higher quality TCF improves VAT compliance by

preventing unintentional errors. Regardless of whether an organization is aggressive or

conservative in approaching tax planning, we expect organizations will have less unintentional

errors when their TCF is better.

The relationship between the quality of the TCF and intentional non-compliance is less

straightforward. On the one hand, a TCF of higher quality could enable organizations to initiate

more risky tax behaviour, which could result in less compliance (and thus more audit

adjustments). On the other hand, a TCF of higher quality could help organizations to better

10

Electronic copy available at: https://ssrn.com/abstract=4636196

streamline (or even conceal) their non-compliance (and thus reduce audit adjustments). A

priori, it is unclear which of these potential effects dominates. This leads to the following

hypotheses:

H1a A higher quality TCF is negatively associated with unintentional VAT non-compliance.

H1b A higher quality TCF is associated with intentional VAT non-compliance.

An improved TCF can also be utilised for maximizing organizational value, in ways not

immediately connected with tax compliance, such as managerial decision making. However,

improved managerial decision-making can, also, be used to both decrease or increase tax

avoidance (Bauer, 2016; Gallemore & Labro, 2015). In this role, the TCF is part of the

governance mechanism that acts to align the interests of managers and shareholders (Bauer,

2016). To what end the TCF is deployed is therefore likely to depend on other variables within

the corporate governance structure, such as incentive alignment between management and

shareholders. The benefits of corporate tax avoidance accrue primarily to the shareholders

(Lanis & Richardson, 2011), whereas the costs (e.g., reputational risks, lower compensation) of

such behaviour mainly fall on the shoulders of the managers (Rego & Wilson, 2012). Rego and

Wilson (2012) find equity incentives are positively associated with tax avoidance consistent

with the idea that when managers can share in some of the benefits of tax avoidance, they are

more willing to incur the associated risks. Tax managers must walk a tightrope when balancing

these incentives. Better internal control can help tax managers find 'the sweet spot' of an optimal

level of tax avoidance (Chang et al., 2020; Chen et al., 2020). However, from the perspective

of the tax administrator, improved tax compliance on its own has very different implications

than improved tax compliance in conjunction with more effective tax avoidance.

TCF is a broad concept, and we expect the implications of a TCF for tax compliance are

a function of the components of the TCF. The OECD (2016) identifies six essential building

11

Electronic copy available at: https://ssrn.com/abstract=4636196

blocks for TCF quality that are based on the COSO model (1992) for internal control. At the

top of the COSO model, is an organization’s control environment, which focusses on soft

controls such as the tone at the top. The control environment reflects how objectives are defined

and structured within an organization. The other elements of control (risk assessment, control

activities, information and communication, and monitoring) are then critical to assuring

objectives around control are met.

Depending on the tax objectives of the organization, a TCF can be used to a smaller or

larger degree to prevent unintentional mistakes and/or to facilitate complex tax avoidance.

These objectives should be clear in the tax strategy, which sets out the organization’s standards

for tax decisions made in supporting the organizational objectives and which are communicated

through the tone at the top. Such strategies can be scaled on a continuum of tax avoidance, with

conservative strategies on the one end and highly aggressive strategies on the other (Blouin,

2014; Krupa, 2021). Organizations with a more conservative tax strategy are more concerned

about risks to the organizational reputation (Graham et al., 2014), and more likely to refrain

from abusive transactions. We assume these organizations are not looking to take on more tax

risks and would prefer to avoid adjustments to their tax return resulting from audits of the tax

authorities (audit adjustments). As a result, they are more likely to primarily use the TCF to

prevent unintentional mistakes.

Organizations with a more aggressive tax strategy are more likely to pursue abusive

transactions, such as tax shelters that are devoid of economic substance and primarily motivated

by tax savings (Penno, 2021), and relatedly, to create complex structures involving related

parties (Chen et al., 2020). Thus, organizations with a more aggressive tax strategy are more

likely to use the TCF to maximize organizational value by pursuing arguably abusive

transactions, implementing complex tax avoidance strategies, and exploiting tax loopholes. A

high quality TCF can help aggressive organizations provide the documentation and support

12

Electronic copy available at: https://ssrn.com/abstract=4636196

necessary to justify aggressive tax positions in a manner that mitigates the risk of an in-depth

audit by the tax authority and thereby reduce tax adjustments related to intentional non-

compliance. However, if the TCF is not sufficient to mitigate scrutiny from tax authorities of

those aggressive positions, aggressive organizations may exhibit more intentional non-

compliance than conservative organizations. This leads to our final hypothesis:

H2 Organizations with an aggressive tax strategy will have more (similar) intentional

VAT non-compliance than (as) conservative organizations, when their TCF is of lower (higher)

quality.

For aggressive organizations, the question is whether a better TCF is a necessary

condition to engage in aggressive tax avoidance, or whether a better TCF is simply helpful in

allowing organizations to avoid audit adjustments after aggressive strategies are employed. To

the extent the former case is descriptive, we would not necessarily expect to observe differences

in the number of audit adjustments for intentional tax avoidance among aggressive

organizations contingent on the quality of their TCF. If anything, organizations with a better

TCF may still have more audit adjustments than organizations with a poor TCF that are unable

to implement aggressive tax strategies. If the latter case is descriptive, we would expect

aggressive organizations with a better TCF to exhibit fewer audit adjustments than aggressive

organizations with a poor TCF. Thus, ultimately for aggressive organizations, the relation

between TCF quality and tax compliance is an empirical question.

13

Electronic copy available at: https://ssrn.com/abstract=4636196

3. Method

3.1 Research design and sample selection

We use a dataset created by the Netherlands Tax Administration (NTA), as part of an

internal research project.4 The NTA has made the promotion of a good TCF an important part

of its regulatory strategy. At a substantial share of the large organizations (i.e., those with whom

the NTA has concluded a cooperative compliance agreement) the NTA actively promotes

further development of the TCF. As a result, there is considerable variance amongst Dutch large

organizations regarding the maturity of their TCF. This makes the Netherlands a powerful

setting for the purposes of our study.

Given the large scale of the research project, the NTA selected the sample of 368

organizations in three parts (168 in 2014, 100 in 2016, and 100 in 2018). The research

population in each of these three years consists of about5 5,000 large (profit and not-for-profit)

organizations in the Netherlands.6 The samples were stratified by the NTA for reasons outside

of our paper, namely 1) to include enough organizations that participate in the Dutch

cooperative compliance program, 2) to include a sufficient number of organizations that only

recently met the criteria to qualify as a large organizations, and 3) to ensure an even distribution

over all five regional offices of the NTA.

We collected data for each organization in the sample in the following manner. First,

we use a survey among the senior staff responsible for tax matters at the organizations in the

sample (we refer to this as the organization survey). This survey was pre-tested by 52 randomly

selected tax managers of large organizations. These tax managers and the organizations they

4 Two of the authors of this paper are part of the NTA research team and were directly involved with the development of the
research instruments. We received permission from the NTA to perform our analyses on the data and report our results in this
paper.
5 The research population differs slightly between the three years due to mergers and bankruptcies.

6 The NTA distinguishes large organizations from other taxpayers based on the following criteria: a) turnover exceeds ten

million euros and gross wages exceed two million euros; or b) gross wages exceed eight million euros; or c) assets exceed one
billion euros. For the purposes of the internal research project, the 3,500 largest organizations of the 8,500 large organizations
in the Netherlands were excluded from the research population and thus 5,000 large organizations remained.

14

Electronic copy available at: https://ssrn.com/abstract=4636196

work for are not part of our sample. This pre-test led to various adjustments to the wording of

the questions in the survey. To protect the anonymity of the respondents, the NTA

commissioned the administration of the survey to an external agency. 7 All prospective

respondents were assured that the NTA could not link responses to the organization of the

respondents. The NTA informed all prospective respondents by telephone call and with an

official announcement letter from the director of the NTA.

We use the organization survey to collect data related to both our independent and

control variables. For each year in which the survey was conducted the response rate was at

least 58% and thus relatively high for this type of research. The NTA noted that this was to a

large degree due to the importance of the subject for the respondents as well as the great care

that was taken to protect anonymity.

Second, after the completion of the surveys the value added tax (VAT) tax returns of all

organizations were audited (regardless of participation in the survey) for the most recently

closed fiscal year (this usually concerns the year prior to the sample, so 2013 for the 2014

sample, and 2015 and 2017 for the 2016 and 2018 samples respectively). These audits were not

part of the NTA’s regular audit efforts that are based on risk driven audit selection; the research

samples were drawn at random without any risk assessment up front on the part of the NTA.

All audits were performed by experienced audit teams8 from the NTA and were fully in line

with the design, execution, and settlement of regular audits.

Results from the audits were reported in an ‘audit format’. This audit report format was

pre-tested by 17 NTA audit employees with different backgrounds, including VAT, CIT, and

statistical auditing. This pre-test led to some adjustments in the order and the wording of the

items included in this study. The final audit report format included questions related to the

7This external agency also merged the data from the three data sources into one dataset.
8The audit teams were led by lead auditors, who are usually certified public accountants (85% of lead auditors were either a
certified public accountant or in training to become one) with 12 years of audit experience on average.

15

Electronic copy available at: https://ssrn.com/abstract=4636196

findings of the audits, such as the size of audit adjustments (if any), nature and cause of these

audit adjustments9, and some general perceptions of the auditor regarding the compliance of

the organization. The NTA subjected each audit to at least one internal quality review, both

during and after conclusion of the audit. We use this audit format to collect data for the

construction of our dependent variables.

A small number of audit reports that contain inaccuracies and/or that were not fully

completed at the time we started our analyses were dropped from our analysis. Furthermore,

due to a variety of other reasons, such as bankruptcy of the organization, not all planned field

audits could be conducted. Our final sample consists of 240 large organizations of which both

the audit reports and the organizational survey were completed (see Table 1). About 18% of

these organizations are not-for-profit organizations, and about 20% participate in the Dutch

cooperative compliance program. Almost all these organizations have a yearly net turnover of

€ 10 million or more and about half have at least 100 employees.

To protect anonymity, the external agency administered the collection of data through

the organization survey as well as the audit report format. Subsequently, the external agency

connected both data sources and provided the researchers with an anonymized dataset.

3.2 Measures

3.2.1 Non-compliance

We use the results from the field audits performed by the NTA to measure the degree

of VAT non-compliance. Following Siglé et al. (2022), we divide our proxies for non-

compliance in two categories: the number of audit adjustments and the scaled amount of audit

adjustments.

9
The reported audit adjustments concern, i.e., wrong application of complex tax law, deduction of non-deductible costs (mainly
personnel related), use of wrong tariff (e.g., based on the type of services delivered), wrong application of exemptions,
intercompany transactions, etc.

16

Electronic copy available at: https://ssrn.com/abstract=4636196

 The audit methods of the NTA focus on detecting misstatements that decrease the tax

burden, i.e., the focus is on adjustments that increase the tax burden for organizations. However,

audits may also detect misstatements that lead to an overstatement of taxes due. The detection

of such overstatements is not systematic, because generally, auditors are not actively looking

for adjustments that might decrease the tax burden. This may lead to an uneven treatment across

the organizations in our sample for these kinds of misstatements. Therefore, we exclude such

misstatements in calculating the total number of audit adjustments. We divide the total amount

of adjustments by the materiality threshold used in the audit to calculate the scaled amount of

audit adjustments.10 To normalize these two variables, we categorized the scores in five clusters

of approximately equal size, although the zero-adjustment category is inevitably larger than the

other four categories. See Table 2 for descriptive statistics on the audit adjustments.

To test our hypotheses regarding the two potential roles of internal control, we create a

measure of intentional and unintentional non-compliance (cf. D’Ascenzo, 2015; James and

Alley, 2002; Siglé et al., 2022). To do so, we divide audit adjustments into proxies that are more

likely to reflect intentional non-compliance, and those that more likely to reflect unintentional

non-compliance. Following Siglé et al. (2022), we regard the number (scaled amount) of audit

adjustments as a better proxy for more unintentional (more intentional) non-compliance. This

distinction is based on the assumption that the higher the financial importance of a transaction

(as measured in the scaled amount), the more likely it is that the transaction was subject to more

stringent internal controls. As a result, it is also more likely that management is aware of this

transaction and, thus, that the audit adjustments concern an issue that was intentionally created

by management. Therefore, the higher the scaled amount of audit adjustments, the more

intentional the non-compliance is assumed to be. For example, there is one organization with

an adjustment of over € 100,000 in which an intercompany transaction from a taxable entity to

10The audit materiality is a threshold that defines when a reporting issue is important enough to warrant attention from the
audit team. The NTA uses a materiality table based on the net turnover of taxpayers to calculate the materiality threshold.

17

Electronic copy available at: https://ssrn.com/abstract=4636196

a non-taxable entity was kept outside the VAT. For this transaction, the organization was

provided with external advice in how to ‘creatively’ prevent taxation. Given the size of the

transaction and the fact that outside advice was sought, it is highly unlikely that top

management was unaware of this transaction.

We proxy for unintentional non-compliance using only the number of small audit

adjustments. The number of audit adjustments, as opposed to the amount of audit adjustments,

is more likely to reflect more unintentional non-compliance because the financial importance

of a transaction has no influence on this proxy. Including only small audit adjustments

emphasizes the degree we capture more unintentional non-compliance. For example, there are

several organizations with a large number of small adjustments (a few hundred euros) related

to incorrect application of VAT law regarding non-financial benefits for personnel. This part of

the VAT law is considered complicated to apply and is often not well understood by

organizations. Therefore, it is highly likely that top management was unaware of these errors.

The distinction between the scaled amount and number of audit adjustments is more a

difference in degrees than an absolute division between intentional and unintentional non-

compliance (that is why we label these proxies as more intentional or more unintentional). To

further amplify the differences between the scaled amount and number of audit adjustments,

we use two additional proxies to better differentiate between less or more intentional non-

compliance. These two additional proxies are variations on the two main proxies, for which we

exclude audit adjustments with either a large (for more unintentional non-compliance) or small

(for more intentional non-compliance) monetary value. The number of audit adjustments is seen

as a better proxy for more unintentional non-compliance because it does not take the amount of

the audit adjustments into account and therefore gives more weight to smaller audit adjustments

(as compared to the scaled amount of adjustments). To amplify this difference, we calculate the

total number of adjustments below € 10,000 (on an individual adjustment level) as a proxy for

18

Electronic copy available at: https://ssrn.com/abstract=4636196

more unintentional non-compliance and the total scaled amount of adjustments above € 10,000

as a proxy for more intentional non-compliance.

Like our two main proxies, we categorized the scores of these two additional proxies

(the total number of adjustments below € 10,000 and the total scaled amount of adjustments

above € 10,000) in five clusters of approximately equal size to normalize these variables. 11 To

validate our proxies for more unintentional and more intentional non-compliance, we use data

on whether the audited firm was imposed a fine based on the highest level of culpability. In the

Netherlands, this equates to a level of culpability that is known as ‘intentional’. Our two proxies

for more unintentional non-compliance are not correlated with the imposition of such a fine

(respectively rho=-.08, p=.33 and rho=.02, p=.80), while our two proxies for more intentional

non-compliance are (respectively rho=.15, p=.06 and rho=.13, p=.10).

3.2.2 Tax control framework and tax strategy

A TCF is a general concept with many facets. The OECD (2016), for example, identifies

six essential building blocks that comprise a TCF. The building blocks identified by the OECD

are derived from the model for internal control of the Committee of Sponsoring Organizations

of the Treadway Commission (COSO) (1992 and 2004). Like the OECD, we follow the COSO

framework to construct a measure of the quality of the TCF (see also Goslinga et al., 2021;

Siglé et al., 2022). Respondents answered 21 survey questions developed to assess the different

aspects of internal control as described by COSO (see Appendix A for the wording of all items).

Principle Components Analysis yielded four factors with an Eigenvalue above 1. Of the original

21 items, 5 items were dropped that decreased the reliability of the four factors. Our general

TCF measure was constructed as the average of these 16 items, which formed a reliable scale

11 Because of the relatively low number of audits with a total of adjustments above € 10,000, the zero-adjustment category is
larger for this proxy than for the other four proxies.

19

Electronic copy available at: https://ssrn.com/abstract=4636196

with a Cronbach’s alpha of 0.91 (we use the separate factors in our additional analyses below).12

We measure the aggressiveness of the tax strategy with a single-item measure (“My

organization explores the boundaries of tax legislation”, 1=Totally disagree, 7=Totally agree).

3.2.3 Control variables

We include a control for whether the organization is a for profit versus non-profit

entity 13 (cf. Goslinga et al., 2021), turnover, complexity, formal participation in horizontal

monitoring14 (all three cf. Siglé et al., 2022), and whether the organization is audited by a big

four audit firm or not (cf. Kanagaretnam et al., 2016). Non-profit is a dummy variable that

captures whether an organization is a non-profit organization (e.g., a care provider, a university,

a pension fund, etc; 1) or not (0). We measure complexity with the number of ‘fiscal

identification codes’, i.e., the number of individual fiscal entities subsumed in the organization.

The external agency involved in the administration of the surveys categorized these variables

in five classes to protect anonymity. Horizontal monitoring (HM) is a dummy variable that

captures whether the organization has closed a horizontal monitoring covenant with the NTA

(1) or not (0). Big4 auditor is a dummy variable to measure whether the organization is audited

by a big four audit firm (Deloitte, EY, KPMG, or PWC; 1) or not (0). See Table 3 for all

descriptive statistics.

12
By definition, the TCF is part of the overall control framework of an organization. From this perspective it makes sense that
for 3 of our 16 items for the quality of the TCF no mention is made of ‘tax’ or ‘fiscal’ (13 of the 16 items do mention this). We
have reperformed our primary analyses using only the 13 items that do mention ‘tax’ or ‘fiscal’. Results from these analyses
are qualitatively similar to our primary analyses.
13 The organizations in our sample come from more than twenty different industries and, thus, leave us with too few

observations per industry to include industry fixed effects. However, within the context of Dutch VAT law, the differences
between for profit and non-profit organizations are the most relevant differences related to industry and we do control for this.
14 Horizontal monitoring is a so-called cooperative compliance strategy (Goslinga et al., 2019). As part of this supervisory

strategy, the NTA actively promotes good internal control, more conservative tax strategies, and corporate tax compliance.

20

Electronic copy available at: https://ssrn.com/abstract=4636196

 The mean score of our measure for the aggressiveness of the tax strategy is quite low

(2.69), indicating that most organizations report having a more conservative tax strategy. The

mean score of our TCF variable is somewhat above midpoint (4.37).

4. Results

4.1 Correlations

The Pearson correlation results are presented in Table 4. As expected, our four proxies

for non-compliance are positively correlated. The quality of the TCF is negatively correlated

with three out of four of our proxies for non-compliance. These correlations are consistent with

our expectations in Hypothesis 1. The strength of these correlations does not seem to differ

much between our proxies based on the number of audit adjustments and our proxies based on

the scaled amount of audit adjustments. Furthermore, the aggressiveness of the tax strategy is

not correlated with any of the proxies for non-compliance.

Noticeably, non-profit organizations are confronted with a higher number and a higher

scaled amount of audit adjustments (we observe a significant relationship with all four proxies

of non-compliance). We also observe that organizations with more turnover have a more

aggressive tax strategy and that organizations participating in Horizontal Monitoring have a

TCF of higher quality. Organizations that are audited by a big four auditor are confronted with

a higher scaled amount of audit adjustments (for both proxies), are more likely to be a non-

profit organization, and are less complex. Furthermore, whether an organization participates in

horizontal monitoring is not associated with any of the proxies for non-compliance. Finally,

more complex organizations are not necessarily less compliant, but we do find that more

complex organizations have significantly less audit adjustments above € 10,000.

21

Electronic copy available at: https://ssrn.com/abstract=4636196

4.2 Multivariate analyses

4.2.1 Test of H1a and H1b

Based on our first hypothesis, H1a, we expect that organizations with a better TCF will be more

compliant and thus have fewer audit adjustments due to more unintentional errors. Based on

our second hypothesis, H1b, we expect that a higher quality TCF is associated with intentional

VAT non-compliance (without clear expectations about the direction).

We first employ two OLS regressions with our main proxies for non-compliance

(‘number of audit adjustments’ and ‘scaled amount of audit adjustments’) as dependent

variables to test these expectations. Results for these regressions are shown in Table 5. We

include the quality of the TCF, tax strategy, and the interaction between TCF and tax strategy

as predictors, and a number of control variables. In line with H1a, we find that the quality of the

TCF is significantly and negatively associated with the number of audit adjustments (ß = -.16,

p < .05). Regarding H1b, we find that the quality of the TCF is significantly and negatively

associated with the scaled amount of audit adjustments (ß = -.18, p < .01).

We argue that our two main proxies for non-compliance - the scaled amount and number

of audit adjustments - reflect a difference in degree rather than an absolute division between

intentional and unintentional non-compliance, which is why we label these proxies as more

intentional or more unintentional. To further amplify the differences between the scaled amount

and number of audit adjustments, we use two additional proxies to better differentiate between

more or less intentional non-compliance. We employ two additional OLS regressions to test

our first two hypotheses in the context of these proxies of VAT non-compliance (also shown in

Table 5). We find a significant effect of the quality of the TCF on non-compliance for both

proxies. Overall, we find support for H1a and, regarding H1b, a similar negative association

between the quality of the TCF and more intentional non-compliance. We do not find a direct

effect of tax strategy on any of our four proxies for non-compliance.

22

Electronic copy available at: https://ssrn.com/abstract=4636196

 Regarding the control variables, we find that non-profit organizations are less VAT

compliant than profit organizations. More complex organizations are also less VAT compliant,

and this effect seems stronger for more unintentional non-compliance. This finding is in line

with the intuition that more complex organizations have more difficulty with being in control.

We find that organizations that are audited by a big four auditor are less VAT compliant, but

only with regard to more intentional non-compliance. This might indicate that big four auditors

are less able to detect intentional VAT non-compliance or that they actively assist organizations

in structuring their VAT to minimise their tax burden, with a risk of higher audit adjustments

as a result.

4.2.2 Test of H2

Based on H2, we expect that organizations with an aggressive tax strategy will have

more intentional VAT non-compliance than conservative organizations, but only when their

TCF is of lower quality. The results presented in Table 5 for the scaled amount of audit

adjustments provide some support for this hypothesis. We find a significant interaction effect

between the Quality of the TCF and Strategy in the regressions of the scaled amount of

adjustments (ß = -16, p = .01) and the scaled amount of all audit adjustments above € 10,000 (ß

= -.11, p < .10).

Taken together, we find one significant interaction effect and one marginal significant

interaction effect, between the Quality of the TCF and Strategy on our proxies for more

intentional non-compliance. To interpret these interaction effects, we created Figure 1. Results

show that organizations with a more aggressive tax strategy have more intentional mistakes

than organizations with a more conservative tax strategy when the TCF is worse. However,

when the TCF quality is good, organizations with a more aggressive tax strategy have less

intentional non-compliance as compared to organizations with a more conservative tax strategy.

23

Electronic copy available at: https://ssrn.com/abstract=4636196

 To gain more insight into the role the TCF plays for more conservative and more

aggressive organizations, we split our sample into more conservative (those that score 3 or

lower on the degree of aggressiveness, n=160) and more aggressive organizations (those that

score 4 or higher on the degree of aggressiveness, n=80). We perform the same analyses

(without the interaction variable) as reported in Table 5 in these samples and report the

standardized coefficients and p values for the quality of the TCF in Table 6.

In the sample of more conservative organizations, we find significant effects of the

Quality of the TCF on both proxies for more unintentional non-compliance but none of the

proxies for more intentional non-compliance. In the sample of more aggressive organizations

the results are the opposite: we find significant effects for both proxies for more intentional

non-compliance but none of the proxies for more unintentional non-compliance. These results

suggest that the results of our primary analyses, as reported in Table 5, concerning

unintentional (intentional) non-compliance are mainly driven by more conservative (more

aggressive) organizations.

4.3 Additional analyses

4.3.1 Overstatements

As mentioned in Section 3, we excluded misstatements that lead to an overstatement of

taxes due because the audit methods of the NTA do not focus on such misstatements. However,

such misstatements can be seen as an alternative proxy for unintentional non-compliance since

it is very likely that any such misstatements in which an organization overpays on taxes is

unintentional. Therefore, we replicate our analyses for H1a using only such ‘overstatements’.

Results from these analyses (untabulated) show that the quality of the TCF is negatively and

significantly associated with both the number of audit adjustments overall and the number of

24

Electronic copy available at: https://ssrn.com/abstract=4636196

audit adjustments below € 10k (both ß = -.20 and both p = .01). This validates the claim that

organizations can use their TCF to prevent unintentional non-compliance.

4.3.2 Subscales TCF

As mentioned in Section 3, the items with which we measure the Quality of the TCF

were based upon the different aspects of internal control as described by COSO (1992 and

2004).15 The OECD (2016) used COSO and other sources to further specify the components of

the TCF. In order to identify different factors composing our general TCF measure, we

conducted Principle Components Analysis with the 16 TCF-items. This yielded 4 factors with

an Eigenvalue above 1 that together accounted for 75% of the total variance. The items

belonging to the four factors had sufficient internal reliability (Cronbach’s alphas between .79

and .92, see Table 7) and four subscales were constructed based on the unweighted means.

Comparable with COSO (1992 and 2004) and OECD (2016), we labelled these four

factors Tax Awareness, Formalised Processes, Performance Indicators, and Roles and

Responsibilities (see Appendix 1A for the corresponding items). The items of the subscale Tax

Awareness focus on a clear strategy and objectives from the top (the so-called tone at the top)

and clear insight into where the line between compliance and non-compliance (risk appetite) is

drawn. The subscale Formalised Processes focusses on a process-oriented approach in which

all processes related to tax are described and documented to create a comprehensive TCF

throughout the organization as a whole. The subscale Performance Indicators focusses on

keeping accurate track of tax risks and regularly reporting to top management. Finally, the

subscale Roles & Responsibilities focusses on making sure that responsibilities delegated to

staff are carried out correctly by personnel that is qualified to perform the tasks at hand. Table

15
Later updates of the COSO model – most importantly the change in 2017 to the Helix model instead of the cube – are notably
lacking as an inspiration for TCFs. As a result, we have not included this in our paper.

25

Electronic copy available at: https://ssrn.com/abstract=4636196

7, Panel A shows the correlations between these four subscales and their correlation with the

original TCF construct. All correlations are above .30 and highly significant (p < .01).

Given our results concerning H2, actively promoting increased TCF quality might be a

double-edged sword for tax authorities. It is interesting which of these subscales are most

relevant in the context of our hypotheses, so tax authorities could perhaps better target their

activities towards specific components of a TCF. If some subscales are more relevant in the

context of unintentional non-compliance and less so in the context of intentional non-

compliance, these might be the part of the TCF that tax authorities should aim to accentuate.

Further, more conservative taxpayers might utilize other parts of a TCF than more aggressive

taxpayers. We do not formulate formal hypotheses concerning such differences, but we do have

some a priori intuitions regarding how taxpayers with different strategic concern might make

use of the TCF.

First, aggressive tax strategies using abusive transactions, such as tax shelters that are

devoid of economic substance and primarily motivated by tax savings (Penno, 2019) and

complex structures involving related parties (Chen et al., 2020), are developed by top

management (and oftentimes supported by tax advisors). Given this, the part of the TCF that

focusses on the strategic level of the organization is likely the most helpful in supporting such

a strategy. Therefore, we expect that our subscale Tax Awareness plays an important role for

more aggressive taxpayers in streamlining their intentional non-compliance as to prevent audit

adjustments. Additionally, we expect that more aggressive organizations will try to avoid

attention from the tax authorities and, therefore, can also be expected to prevent unintentional

errors from drawing unwanted attention. As a result, Tax Awareness also helps more aggressive

organizations in reducing unintentional non-compliance. Furthermore, more aggressive

organizations can also use Formalised Processes, Roles & Responsibilities and Performance

Indicators to achieve reduced unintentional non-compliance. These parts of a TCF focus on

26

Electronic copy available at: https://ssrn.com/abstract=4636196

clear processes and responsibilities for all personnel, on ensuring that personnel are qualified

to perform the tasks at hand, and on keeping accurate track of tax risks and regularly reporting

to top management, thus helping to prevent unintentional non-compliance. We expect that these

subscales are less important for more aggressive organizations regarding more intentional non-

compliance because such intentional non-compliance is enacted at the top management level.

Second, organizations with a more conservative tax strategy that aim to prevent

(intentional and unintentional) non-compliance, can be expected to use their TCF for this aim.

The decision making regarding most transactions takes place at the operational level and top

management must ensure that potential risks are communicated to the personnel at this level,

that personnel are capable, and that they are provided with clear instructions on how to act

regarding these risks. Therefore, we would expect that the subscales Formalised Processes,

Roles & Responsibilities and Performance Indicators are relevant for more conservative

organizations in preventing non-compliance.

To test these expectations, we replicate our analyses on both the full and the split sample

whereby we substitute our TCF variable with one of these subscales at a time.16 Results from

these analyses are reported in Table 7, Panels B, C, and D. Comparing these panels we see clear

differences between more conservative and more aggressive organizations, but the results from

these analyses are not fully in line with our expectations. Regarding the subscale Tax Awareness

we find, in line with our expectations, that more tax aggressive organizations seem to have less

intentional and unintentional non-compliance when their tax awareness is higher. For more

conservative organizations, tax awareness does not affect the number nor the scaled amount of

audit adjustments. Regarding the subscale Formalised Processes we find, also in line with our

expectations, that this part of the TCF helps more conservative organizations in preventing

16Additional T-tests show that the four subscales do not differ significantly between more conservative and more aggressive
organizations (untabulated, all p > .10).

27

Electronic copy available at: https://ssrn.com/abstract=4636196

unintentional non-compliance. 17 We expected similar results for the subscale Roles &

Responsibilities but did not find this (although we did find one significant effect of this subscale

on the scaled amount adjustments, one of the proxies for more intentional non-compliance).

Regarding the subscale Performance Indicators, we expected similar effects for conservative

and aggressive organizations. In line with this, we find a significant effect in both subsamples

for one of our two proxies for non-compliance.

5. Conclusions and discussion

A high-quality tax control framework (TCF) is seen by tax authorities and the OECD as

a prerequisite for corporate tax compliance. The assumption is that a high quality TCF detects

mistakes and possible tax risks and thus helps organizations to increase tax compliance. There

is some empirical evidence that substantiates this assumption (e.g., Bimo et al., 2019; Siglé et

al., 2022). Based on this assumption, tax authorities actively promote better internal control in

organizations to increase tax compliance (e.g., OECD, 2013). However, a better TCF can also

help organizations to effectuate their desired tax strategy and several studies show that a better

TCF enables organizations to reduce their tax burden to maximize organizational value (e.g.,

Bauer 2016; Gallemore & Labro, 2015). So, an important question for tax authorities is how

internal control (the TCF) relates to tax compliance. In this paper we focus on VAT

(non-)compliance and try to answer this question by differentiating between more unintentional

and more intentional VAT non-compliance and between organizations with a more

conservative and a more aggressive tax strategy.

Our results provide empirical support for the role of a TCF in preventing unintentional

non-compliance. We hypothesize, and find, that organizations with a higher quality TCF are

more VAT compliant. When we take the moderating effect of the tax strategy into account, we

17Our results for this are only significant at the 10%-level, but given our smaller sample size for these analyses we find this
sufficient.

28

Electronic copy available at: https://ssrn.com/abstract=4636196

find that the level of intentional non-compliance is not affected by the quality of the TCF for

organizations with a conservative tax strategy. However, organizations inclined to be more

aggressive in their tax planning display more intentional non-compliance if they have a TCF of

lower quality and less intentional non-compliance when the TCF is of higher quality. We take

this as an indication that a high quality TCF can help organizations with a more aggressive tax

strategy in preventing audit adjustments, for example by providing better justification when tax

loopholes are being exploited. This interaction between strategy and tax control is in line with

Richardson, et al. (2013), who find that the interaction effect between the composition of the

board of directors (i.e., a higher proportion of independent directors on the board) and effective

internal controls reduces tax aggressiveness.

From the additional analyses it seems that tax awareness is the most important part of

the TCF in the context of our study for more aggressive organizations. Higher tax awareness

includes increased consciousness of tax compliance risks and a clearer picture of the divide

between compliance and non-compliance, with less audit adjustments as a logical result. If this

is indicative of more sophisticated tax avoidance by aggressive organizations, promoting tax

awareness is not in the interest of the tax authority. In the context of more conservative

organizations, it seems wise to promote formalised processes, and roles and responsibilities to

prevent unintentional non-compliance. The parts of the TCF that we dubbed ‘performance

indicators’ seems relevant in the context of both more aggressive and more conservative

organizations, for example by improving timeliness of communication within the organization

(Chen et al., 2020). This might indicate that actively promoting improvement of the TCF is

inherently a double-edged sword for tax authorities and should, therefore, be pursued carefully.

Thus, it is crucial to know the tax strategy, intentions and tone at the top of organizations, and

finding ways to influence these kinds of soft controls also.

29

Electronic copy available at: https://ssrn.com/abstract=4636196

 Although our findings are robust to alternative proxies for non-compliance, the

following limitations should be taken into consideration. First, our data is limited to Dutch

organizations and the Dutch tax system, and the generalizability of our findings can be limited

by specific characteristics of the Netherlands. Second, our study is based on cross-sectional

data and, thus, provides evidence on associations rather than on causality. Third, we use

different sources for the collection of our independent variables (the organization survey) and

dependent variables (audit data). While this minimizes common method bias, it might make

our study more vulnerable to type II errors. Fourth, we rely on a one item measure to capture

an organization’s tax strategy. Perhaps future research could use the effective tax rate as an

alternative proxy for tax strategy. Reporting of the tax strategy under the Global Reporting

Initiative (GRI 207) also provides an opportunity for measuring tax strategy. Fifth, our

measures for compliance depend on the quality of the audits performed by the NTA. It might

be that the NTA was unable to detect certain types of non-compliance, for example due to

advanced tax avoidance techniques used by the audited organizations. To the degree that this is

because some organizations utilized their TCF to better conceal non-compliant behaviour, this

might have affected our results regarding H1b. The NTA did, however, implement various

additional quality control checks to ensure the highest possible quality for the audits performed.

Sixth, some of our findings in the additional analyses were only significant at the 10 percent

level. While we consider this significance level, given our sample size, to be adequate, future

research is required to ascertain whether our results can be replicated in other settings. Seventh,

our sample was not fully random due to stratification of our sample. Although this stratification

was not related to any of the variables of interest, it might impact the generalizability of our

results. As our purpose is to test theory rather than to make claims that generalize to a

population, we believe the sample is relevant to address the research question (Speklé &

Widener, 2018).

30

Electronic copy available at: https://ssrn.com/abstract=4636196

 Despite these limitations, our study contributes to the literature in several ways.

Foremost, previous literature does not provide a clear picture regarding the role of internal

control in corporate tax compliance. We introduce the moderating role of the tax strategy of an

organization as a potential explanation for the contradictory findings that emerge from previous

studies. Furthermore, we are the first study to incorporate the difference between intentional

and unintentional non-compliance into the development and the empirical tests of our

hypotheses, and in doing so provide more insight into the role the TCF plays in tax compliance.

Also, whereas most previous tax compliance research focusses on corporate income tax (CIT),

we focus on value added tax (VAT), which is an equally important type of tax in terms of

revenue (OECD, 2021) but has, compared to CIT-compliance, scarcely been studied (Datt et

al., 2017).

Our study provides various avenues for future research. First, it would be interesting to

see whether our results hold for other types of taxes such as corporate income taxes and payroll

taxes. Compared to VAT, corporate income tax is managed at a higher level within the

organization, while payroll taxes generally provide fewer tax planning opportunities. These

differences suggest that the role of the TCF concerning unintentional (intentional) non-

compliance is less important regarding corporate income taxes (payroll taxes). Future research

could delve deeper into these a priori expectations. Second, besides tax authorities, more

regulatory fields, such as finance, occupational safety, food and drug safety, consumer product

safety, and environmental protection (Earnhart & Glicksman, 2015) stimulate internal control

quality to increase compliance. Future research could test our hypotheses in these settings.

Third, we find that not-for-profit organizations are significantly less compliant than for-profit

organizations. This might be due to more complex laws and regulations that face not-for-profit

organizations. Interestingly, we find no difference in the quality of the TCF between profit and

not-for-profit organizations. Perhaps not-for-profit organizations should be held to a higher

31

Electronic copy available at: https://ssrn.com/abstract=4636196

standard and have a better TCF to be better equipped to handle the more complex tax law. We

suspect that more factors play a role in this somewhat surprising result and future research could

delve deeper into this subject.

Based on our results, tax authorities would seem wise in promoting more control

through better TCFs amongst large (profit and non-profit) organizations. However, our results

also indicate that organizations inclined to be aggressive in their tax planning can use a better

TCF to minimize their risk of audit adjustments. This indicates that better control enables more

aggressive organizations to reduce their tax burden within the boundaries (of the letter) of the

law. From the perspective of organizations, these results demonstrate that implementing a well-

functioning TCF can help to effectuate their tax strategy and prevent tax errors. These results

can be seen as a reminder to tax authorities not to focus solely on a high quality TCF, but to

emphasize that a better TCF must be coupled with a tax strategy aimed at tax compliance with

both the letter and spirit of the tax law.

32

Electronic copy available at: https://ssrn.com/abstract=4636196

Tables and Figures

Table 1: Sample

Audit VAT NTA Survey Data from

Year Sample
completed completed both sources
2014 168 158 166 123
2016 100 90 98 58
2018 100 81 91 59
Total 368 329 355 240

Table 2: Descriptives of audit adjustments

Including audits Excluding audits

without adjustment(s) without adjustment(s)
Number of audits with adjustments 159 (66%) 159 (100%)
Maximum number of adjustments 23 23
Average number of adjustments 1.83 2.75
Sd number of adjustments 2.39 2.46
Maximum amount of adjustments € 719,920 € 719,920
Maximum amount of scaled adjustments 1.77 1.77
Average amount of adjustments € 26,211 € 38,821
Sd amount of adjustments € 75,774 € 90,407
Average amount of scaled adjustments .06 .09
Sd amount of scaled adjustments .20 .24
Note: Descriptive statistics from field audits of the VAT, reported in payable taxes.

Table 3: Descriptives (n=240)

Min 25% M 75% Max Sd Skwn. Kurt.

Number of audit adjustments:
- Audit adjustments overall 1 1 2.54 4 5 1.47 0.50 -1.15
- Audit adjustments below € 10k 1 1 2.25 3 5 1.35 0.77 -0.64
Scaled amount of audit adjustments:
- Audit adjustments overall 1 1 2.52 4 5 1.44 0.47 -1.17
- Audit adjustments above € 10k 1 1 1.53 2 5 1.03 2.22 4.27
Strategy 1 1 2.69 4 6 1.50 0.49 -0.95
TCF 1 3.75 4.37 5.13 7 1.09 -0.17 0.38
Non-profit 0 0 0.18 0 1 0.38 1.68 0.84
Turnover 1 2 2.31 3 4 0.87 0.34 -0.50
Complexity 1 2 3.25 4 5 1.37 -0.27 -1.15
Horizontal Monitoring 0 0 0.21 0 1 0.41 1.45 0.09
Big4 auditor 0 0 0.41 1 1 0.49 0.38 -1.88
See Appendix B for all variable definitions

33

Electronic copy available at: https://ssrn.com/abstract=4636196

Table 4: Pearson correlations (n=240)

2 3 4 5 6 7 8 9 10 11
Number of:
1 - Audit adjustments overall .93*** .75*** .48*** .01 -.15** .24*** .12* .01 -.04 .10
2 - Audit adjustments below € 10k .55*** .26*** .01 -.13** .15** .11* .07 -.03 .05
Scaled amount of:
3 - Audit adjustments overall .74*** -.02 -.15** .27*** -.06 -.07 -.07 .19***
4 - Audit adjustments above € 10k .05 -.10 .37*** .10* -.22*** -.05 .26***
5 Strategy .03 -.03 .16** .02 .08 -.02
6 TCF .00 .02 -.01 .26*** .10
7 Non-profit -.01 -.56*** -.05 .36***
8 Turnover .06 .04 .04
9 Complexity .15** -.39***
10 Horizontal Monitoring -.13*
11 Big4 auditor
*=p<.10 **=p<.05 ***=p.01 (all p-values are two-tailed)
See Appendix B for all variable definitions

Table 5: Regression analyses (n=240)

Number of audit adjustments Scaled amounts of audit adjustments

Overall Below €10k Overall Above €10k
B p B p B p B p
1a/b
TCF (H ) -.16** .01 -.13** .05 -.18*** <.01 -.13** .03
Strategy .01 .88 .01 .93 .02 .71 .06 .34
2
TCF x Strategy (H ) -.04 .50 -.01 .90 -.16** .01 -.11* .07
Non-profit .33*** <.01 .26*** <.01 .29*** <.01 .31*** <.01
Turnover of firm .10 .12 .09 .16 -.10 .13 .08 .22
Complexity of firm .22*** <.01 .23*** <.01 .16** .04 .00 .99
Horizontal monitoring -.01 .84 -.01 .86 -.02 .79 .00 .95
Big4 auditor .07 .31 .05 .47 .16** .02 .15** .03
Adj. R2 .10 .05 .12 .16
p .00 .01 .00 .00
*=p<.10 **=p<.05 ***=p.01 (all p-values are two-tailed)
See Appendix B for all variable definitions

34

Electronic copy available at: https://ssrn.com/abstract=4636196

Table 6: Additional analyses based on subsamples of tax strategy

Conservative Aggressive
organizations organizations
Effect of the quality of the TCF on: (n=160) (n=80)
B p B p
More unintentional non-compliance, i.e. the number of:
- Audit adjustments overall -.16** .05 -.12 .28
- Audit adjustments below € 10k -.17** .05 -.05 .66
More intentional non-compliance, i.e. the scaled amount of:
- Audit adjustments overall -.09 .27 -.26** .02
- Audit adjustments above € 10k -.05 .58 -.22** .04
*=p<.10 **=p<.05 ***=p.01 (all p-values are two-tailed)
The regression models are the same as reported in Table 5, with the exclusion of the interaction variable. Reported
are the standardized coefficients of the quality of the TCF
See Appendix B for all variable definitions

Figure 1: Interaction effects

35

Electronic copy available at: https://ssrn.com/abstract=4636196

Table 7: Additional analyses with subscales of the quality of the TCF (n=240)

Panel A: Pearson correlations and reliability (in diagonal) statistics for subitems of TCF Quality
Tax Formalised Performance Roles &
Awareness Processes Indicators Responsibilities
TCF Total .83*** .77*** .69*** .72***
Tax Awareness .91 .44*** .48*** .56***
Formalised Processes .87 .40** .46***
Performance Ind. .88 .31***
Roles & Resp. .79
Panel B: Test of H1 with subitems of TCF Quality for full sample (n=240)
Number of audit Number of audit Scaled amount of audit Scaled amount of audit
adjustments overall adjustments below €10k adjustments overall adjustments above €10k
B p B p B p B p
Tax Awareness -.15** .02 -.15** .02 -.12** .05 -.11* .08
Formalised Processes -.10 .14 -.08 .26 -.08 .21 -.05 .44
Performance Ind. -.09 .16 -.07 .30 -.12** .05 -.13** .03
Roles & Resp. -.10 .11 -.08 .20 -.15** .02 -.06 .37
Panel C: Test of H1 with subitems of TCF Quality for Conservative organizations (n=160)
Number of audit Number of audit Scaled amount of audit Scaled amount of audit
adjustments overall adjustments below €10k adjustments overall adjustments above €10k
B p B p B p B p
Tax Awareness -.09 .25 -.11 .17 -.02 .76 -.02 .84
Formalised Processes -.14* .10 -.15* .08 -.04 .60 .00 .95
Performance Ind. -.14* .09 -.11 .18 -.10 .19 -.10 .18
Roles & Resp. -.12 .15 -.12 .15 -.14* .08 -.03 .68
Panel D: Test of H1 with subitems of TCF Quality for Aggressive organizations (n=80)
Number of audit Number of audit Scaled amount of audit Scaled amount of audit
adjustments overall adjustments below €10k adjustments overall adjustments above €10k
B p B p B p B p
Tax Awareness -.28** .01 -21* .07 -.31*** <.01 -.23** .02
Formalised Processes .01 .96 .08 .49 -.15 .18 -.13 .21
Performance Ind. .03 .79 .03 .77 -.15 .15 -.19* .06
Roles & Resp. -.04 .73 .02 .89 -.15 .19 -.11 .32
*=p<.10 **=p<.05 ***=p.01 (all p-values are two-tailed)
All regression for Panel B and C are similar to Table 6, except for the replacement of TCF with the reported subitem of
TCF.
Reported are only the coefficients and p-values of the subitems.
See Appendix B for all variable definitions

36

Electronic copy available at: https://ssrn.com/abstract=4636196

References

Bauer, A.M. (2016). Tax avoidance and the implications of weak internal controls.

Contemporary Accounting Research, 33 (2), 449–486. https://doi.org/10.1111/1911-

3846.12151.

Bergman, M. & Nevarez, A. (2006). Do audits enhance compliance? An empirical assessment

of VAT enforcement. National Tax Journal, 59 (4), 817-832.

https://doi.org/10.17310/ntj.2006.4.04.

Bimo, I.D., Prasetyo, C.Y. & Susilandari, C.A. (2019). The effect of internal control on tax

avoidance: the case of Indonesia. Journal of Economics and Development, 21 (2), 131-

143. https://doi.org/10.1108/JED-10-2019-0042.

Blaufus, K., Reineke, J., & Trenn, I. (2022). Perceived tax audit aggressiveness, tax control

frameworks and tax planning: an empirical analysis. Journal of Business Economics,

93, 509–557. https://doi.org/10.1007/s11573-022-01116-6.

Blouin, J. (2014). Defining and measuring tax planning aggressiveness. National Tax Journal,

67 (4), 875-900. https://doi.org/10.17310/ntj.2014.4.06.

Chang, H., Dai, X., He, Y,. & Wang, M. (2020). How internal control protects shareholders’

welfare: Evidence from tax avoidance in China. Journal of International Accounting

Research., 19 (2):19-39. https://doi.org/10.2308/jiar-19-046.

Chen, H., Yang, D., Zhang, X., & Zhou, N. (2020). The moderating role of internal control in

tax avoidance: Evidence from a COSO-based internal control index in China. The

Journal of the American Taxation Association, 42 (1), 23-55.

https://doi.org/10.2308/atax-52408.

Committee of Sponsoring Organisations of the Treadway Commission (COSO) (1992).

Internal Control—Integrated Framework. Retrieved from: https://www.coso.org.

Accessed November 6, 2023

37

Electronic copy available at: https://ssrn.com/abstract=4636196

COSO (2004). Enterprise Risk Management—Integrated Framework. Retrieved from:

https://www.coso.org/guidance-erm. Accessed November 6, 2023

D'Ascenzo, M. (2015). Global trends in tax administration. Journal of Tax Administration, 1

(1), 81-100. http://dx.doi.org/10.2139/ssrn.3173515.

Datt, K., Nienaber, G., & Tran-Nam, B. (2017). GST/VAT general anti-avoidance approaches:

Some preliminary findings from a comparative study of Australia and South Africa.

Australian Tax Forum, 32 (2), 377-398.

De Mello, L. (2009). Avoiding the value added tax: theory and cross-country evidence. Public

Finance Review, 37 (1), 27–46. https://doi.org/10.1177/1091142108316588.

Earnhart, D.H. & Glicksman, R.L. (2015). Coercive vs. cooperative enforcement: Effect of

enforcement approach on environmental management. International Review of Law and

Economics, 42, 135–146. https://doi.org/10.1016/j.irle.2015.02.003.

European Commission, CASE, Poniatowski, G., Bonch-Osmolovskiy, M. & Śmietanka, A.

(2021). VAT gap in the EU – Report 2021. https://data.europa.eu/doi/10.2778/447556.

Accessed November 10, 2023.

Gallemore, L. & Labro, E. (2015). The importance of the internal information environment for

tax avoidance. Journal of Accounting and Economics, 60 (1), 149-167.

https://doi.org/10.1016/j.jacceco.2014.09.005.

Gleason, C.A., Pincus, M., & Rego, S.O. (2017). Material weaknesses in tax-related internal

controls and last chance earnings management. Journal of the American Taxation

Association, 39 (1), 25–44. https://doi.org/10.2308/atax-51511.

Goslinga, S., Jonge, J. de, Siglé, M.A., & Van der Hel, E.J.C.M. (2021). Cooperative

compliance programs: Who participates and why? Journal of Tax Administration, 6 (2),

95-121.

38

Electronic copy available at: https://ssrn.com/abstract=4636196

Goslinga, S., Siglé, M.A., & Veldhuizen, R.F. (2019). Cooperative compliance, tax control

frameworks and perceived certainty about the tax position in large organisations.

Journal of Tax Administration, 5 (1), 41-65.

Graham, J., Hanlon, M., Shevlin, T., & Shroff, N. (2014). Incentives for tax planning and

avoidance: Evidence from the field. Accounting Review, 89 (3), 991-1023.

https://doi.org/10.2308/accr-50678.

Hein, R. (2022). The tax control framework. In: R. Russo & R. Hein (eds.), Handbook Tax

Assurance (pp. 75-103). Tilburg University Press.

Huiskers-Stoop, E. & Gribnau, H. (2019). Cooperative compliance and the Dutch horizontal

monitoring model. Journal of Tax Administration, 5 (1), 66-110.

Jallai, A.G. (2020). Good tax governance: International corporate tax planning and corporate

social responsibility – Does one exclude the other? Tilburg Law School Research Paper

Forthcoming, Available at

SSRN: https://ssrn.com/abstract=3688985 or http://dx.doi.org/10.2139/ssrn.3688985.

Accessed November 6, 2023.

James, S. & Alley, C. (2002). Tax compliance, self-assessment and tax administration. Journal

of Finance and Management in Public Services, 2 (2), 27-42.

Kanagaretnam, K., Lee, J., Lim, C.Y., & Lobo, G.J. (2016). Relation between auditor quality

and tax aggressiveness: Implications of cross-country institutional differences.

Auditing: A Journal of Practice & Theory, 35 (4), 105–135.

http://dx.doi.org/10.2308/ajpt-51417.

Krupa, T. (2021). Do enterprise risk assessments affect tax aggressiveness? Evidence from U.S.

ORSA regulation. The Accounting Review, Forthcoming. https://doi.org/10.2308/TAR-

2022-0639.

39

Electronic copy available at: https://ssrn.com/abstract=4636196

Lanis, R. & Richardson, G. (2011). The effect of board of director composition on corporate

tax aggressiveness. Journal of Accounting and Public Policy, 30 (1), 50-70.

https://doi.org/10.1016/j.jaccpubpol.2010.09.003.

Lanis, R. & Richardson, G. (2012). Corporate social responsibility and tax aggressiveness: An

empirical analysis. Journal of Accounting and Public Policy, 31 (1), 86-108.

https://doi:10.1016/j.jaccpubpol.2011.10.006.

Organisation for Economic Co-operation and Development (OECD) (2013). Co-operative

compliance: A framework. Retrieved from: https://www.oecd.org.

https://doi.org/10.1787/9789264200852-en. Accessed November 6, 2023.

OECD (2014). Measures of tax compliance outcomes: A practical guide. Retrieved from:

www.oecd.org. https://doi.org/10.1787/9789264223233-en. Accessed November 6,

2023.

OECD (2016). Co-operative tax compliance: Building better tax control frameworks. Retrieved

from: https://www.oecd.org. https://doi.org/10.1787/9789264253384-en. Accessed

November 6, 2023.

OECD (2021). Tax administration 2021: Comparative information on OECD and other

advanced and emerging economies. Retrieved from: https://www.oecd.org.

https://doi.org/10.1787/cef472b9-en. Accessed November 6, 2023.

Owens, J. & Leigh-Pemberton, J. (2021). Cooperative Compliance: A Multi-stakeholder and

Sustainable Approach to Taxation. Kluwer Law International B.V.

Penno, M.C.A. (2021). A theory of the tax avoidance (tax planning) continuum (July 9, 2021).

Available at SSRN: https://ssrn.com/abstract=3883570 or

http://dx.doi.org/10.2139/ssrn.3883570. Accessed November 6, 2023.

40

Electronic copy available at: https://ssrn.com/abstract=4636196

Plesner Rossing, C. (2013). Tax strategy control: the case of transfer pricing tax risk

management. Management Accounting Research, 24 (2), 175-194.

https://doi.org/10.1016/j.mar.2013.04.008.

Rego, S. & Wilson, R. (2012). Equity risk incentives and corporate tax aggressiveness. Journal

of Accounting Research, 50 (3), 775-810. https://doi.org/10.1111/j.1475-

679x.2012.00438.x.

Richardson, G., Taylor, G., & Lanis, R. (2013). The impact of board of director oversight

characteristics on corporate tax aggressiveness: An empirical analysis. Journal of

Accounting and Public Policy, 32 (3), 68-88.

http://dx.doi.org/10.1016/j.jaccpubpol.2013.02.004.

Schenk, A., Thuronyi, V., & Cui, W. (2015). Value added tax: A comparative approach (2nd

ed.). New York: Cambridge University Press.

https://doi.org/10.1017/cbo9781107337671.

Siglé, M.A., Goslinga, S., Speklé, R.F., & Van der Hel, E.C.J.M. (2020). The cooperative

approach to corporate tax compliance: An empirical assessment. Journal of

International Accounting, Auditing and Taxation, 46.

https://doi.org/10.1016/j.intaccaudtax.2022.100447.

Speklé, R.F. & Widener, S.K. (2018). Challenging issues in survey research: Discussion and

suggestions. Journal of Management Accounting Research, 30 (2), 3-21.

https://doi.org/10.2308/jmar-51860.

Van der Hel, E.C.J.M. & Siglé, M.A. (2015). Managing compliance risks of large businesses:

A review of the underlying assumptions of co-operative compliance. eJournal of Tax

Research, 13 (3), 760-783.

41

Electronic copy available at: https://ssrn.com/abstract=4636196

Appendix A

Questionnaire items, descriptive statistics and reliability estimates for the quality of the TCF

(n=240)

Subscale Items M SD
In my organization…
…the fiscal strategy is clear. 5.26 1.46
…the fiscal targets are clear. 4.82 1.58
Tax Awareness …the fiscal targets are realistic. 4.90 1.54
(CR=.91) …the fiscal strategy contributes to compliance with tax laws and regulations. 5.02 1.69
…fiscal risks are identified. 5.07 1.41
…it is stated what fiscal risks must be avoided. 5.24 1.38
…processes are formally described (for example, in a manual). 4.15 1.94
Formalised …the descriptions of processes include tax risks. 3.21 1.71
Processes
…the descriptions of processes include (formal) internal controls. 3.86 1.89
(CR=.87)
…the monitoring of internal controls is described in a plan. 3.34 1.83
Performance … fiscal performance indicators are derived from the fiscal targets. 3.23 1.63
Indicators … fiscal performance indicators are unambiguous. 3.46 1.71
(CR=.88) …the realisation of fiscal targets is periodically reported to the board. 3.71 1.94
Tax Roles & …the roles and responsibilities of fiscal staff are clear. 4.87 1.66
Responsibilities …we invest in training and education to keep the knowledge of fiscal staff up to date. 4.37 1.84
(CR=.79) …employees in fiscal positions are competent enough to carry out these tasks. 5.42 1.34
…unambiguous fiscal targets are derived from the fiscal strategy.
…the identification of fiscal risks is updated yearly.
Not included …fiscal risks are controlled using (formal) internal monitoring.
…the correct operation of fiscal internal controls is subject to monitoring.
…the roles and responsibilities of fiscal staff are formally stated.

42

Electronic copy available at: https://ssrn.com/abstract=4636196

Appendix B

Variable lable Variable definition

Number of adjustments overall Number of adjustments in 5 categories
Number of adjustments below €10k Number of adjustments smaller than € 10,000 in 5 categories
Scaled amount of adjustments overall Amount of adjustments scaled by the materiality in 5 categories
Scaled amount of adjustments above €10k Amount of adjustments greater than € 10,000 scaled by the materiality in 5 categories
Strategy The aggressiveness of the tax strategy on a scale from 1 to 7
TCF Total Quality of the Tax Control Framework on a scale of 1 to 7
Non-profit Dummy for profit (0) or non-profit (1) organization
Turnover Turnover measured in 4 categories
Complexity Number of fiscal entities in 5 categories
Horizontal Monitoring Organization does not (0) or does (1) participate in Horizontal Monitoring
Big4 Auditor Organization is not (0) or is audited by a big four audit firm

43

Electronic copy available at: https://ssrn.com/abstract=4636196