The Tax Gap: The Current State of Play

There is largely consistent opinion across academic literature and amongst many national and regional tax authorities as to how the tax gap might be defined, although subtle and significant differences do remain. Mazur and Plumely (2007), for example, define the tax gap as the difference between the amount of tax that should be imposed by the tax code of a country and the amount that is actually reported and paid on timely filed returns. The UK’s HM Revenue & Customs (HMRC) in similar vein defines the tax gap as‘the difference between the amount of tax that should, in theory, be collected by HMRC, against what is actually collected.’(HMRC 2016, p. 3). The US’s Internal Revenue Service (IRS) adds a twist by defining the tax gap as ‘the difference between the tax that taxpayers should pay and what they actually pay on a timely basis’(IRS 2006). The language is similar to the one adopted by the HMRC with a slight but important difference:

the IRS introduces the notion of‘timely payment’as a factor in the consideration of the tax gap.

The International Monetary Fund’s (‘IMF’) shares core aspects of the defin-itions above, but adds an important element to understanding by suggesting that the appraisal has to be within‘the current policy framework’. (IMF 2013, p. 11). In doing so the IMF suggests that there are two aspects to the tax gap requiring consideration. One is the effect of taxpayer-non-compliance on tax revenue, a notion that is captured in the other definitions noted. The other aspect is the impact that policy choices made by legislators and regulators might have had in reducing available tax revenues. These two different aspects of tax gap are labelled by the IMF as, firstly, the‘compliance gap’caused by non-payment that results from non-compliance with tax rules and, secondly, the‘policy gap’, which refers to tax laws granting exemptions, tax liability deferrals or preferential tax rates (IMF 2013, p. 11).

The European Commission Taxation and Customs Union (TAXUD) who commissions the annual study of the EU’s VAT gap (TAXUD 2018) explicitly embraces the IMF’s concept of the‘tax policy gap’, noting that:

[T]he Policy Gap captures the effects of applying multiple rates and exemptions on the theoretical revenue that could be levied in a given VAT system. In other words, the Policy Gap is an indicator of the additional VAT revenue that a

Member State could theoretically, i.e. in case of perfect tax compliance, generate if it applied a uniform VAT rate on all goods and services. (TAXUD 2016, p. 51) TAXUD does prepare annual estimates of the policy gap with regard to VAT (TAXUD 2018). TAXUD also extends its work to the compliance gap for that tax.

These two international institutions apart it would, however, appear that the issue of tax policy gaps is largely ignored: no national tax authority appears to appraise this issue in an effective way at present.

They may be discouraged from doing so by academic opinion on the worth of measuring tax gaps: Gemmell and Hasseldine (2012, p. 17) noted, for example, that ‘There are few, if any, reliable methods of measuring direct tax gaps as conventionally defined.’ The OECD when offering another succinct definition of the tax gap as ‘the difference between tax due and tax collected’

(OECD 2017, p. 182) appear to, at least in part, share this pessimistic view when saying:

While the tax gap has intuitive attraction for both the public and political representatives, it is a difficult concept to define precisely. Estimation is also difficult as much of the tax gap is either deliberately concealed from view and/or data may be difficult tofind. The measurement and publishing of tax gaps should therefore be navigated and communicated carefully. Limitations of tax gap estimates mean they are not a good basis for explicit performance targets.

(OECD 2017, p. 181) In so doing the OECD appears to endorse three opinions. Thefirst is that tax gap appraisal is about the measurement of the efficiency of tax administrations. The second is that the‘bottom-up’approaches currently used by most tax authorities for this purpose mean that they are not especially suited to this task. The third is that tax gap methodology does not extend to policy gaps.

The difference between ‘bottom up’ and ‘top down’ methods of tax gap estimation is important. As the IMF has noted, the UK’s tax authority‘follows a pattern of employing “bottom-up”based estimates for the direct tax gaps, and

“top-down”estimates for the indirect tax gaps.’(IMF 2013, p. 9). The OECD sees merit in both approaches:

The use of tax gap measurements is becoming more common, especially for VAT, as jurisdictions increasingly see the benefits of having high level estimates of non-compliance within the tax system. Top-down methodologies that use national accounts data represent a relatively low-cost means of producing such estimates. These approaches are often associated, though, with a fairly high degree of uncertainty and therefore are of limited operational use. Bottom-up methodologies that include information from random audits, on the other hand,

can provide a more accurate picture of lost revenue across segments and tax types. (OECD 2017, p. 62)

A top-down approach uses macro-economic data to estimate the potential tax base within an economy. Taking VAT as an example, on this basis the likely VAT due on each part of consumption within national income is estimated as if no allowances or reliefs are supplied to taxpayers. Allowance is then made for the items exempted from charge as a result of policy decisions. In addition, the cost of those allowances and reliefs granted either for reasons of administrative ease or to influence taxpayer behaviour is estimated. These two estimates constitute the VAT policy gap. The estimated tax due net of the VAT policy gap is then compared with the actual yield to suggest a compliance tax gap in a top down approach. The compliance gap represents tax lost as a result of taxpayer behaviour.

VAT gap analysis of this sort is dependent upon the existence of statistics of sufficient quality on the size of the tax base derived from sources other than taxpayer records (IMF 2017, p. 33). The IMF appears to be satisfied with the quality of statistics available in the UK but the fact that there have been concerns on this issue within the EU is apparent from the fact that TAXUD’s VAT gap estimates excluded Cyprus until 2017 because of a lack of reliable national statistics (TAXUD 2017, p. 8).

In contrast to this top-down approach, a‘bottom-up’approach uses an audit sample of submitted tax returns to estimate errors found within them and then extrapolates this error rate across the whole population of submitted returns. The method does, however, leave this approach very vulnerable to estimates of tax not declared at all on tax returns not submitted by persons whose identity may not even be known. The methodology is also not good at capturing tax not paid by relatively small groups in society, such as the very wealthy. As Zucman et al.

(2017) have noted, if such groups are predisposed to evasion then resulting tax gap estimates may be very vulnerable to error. For this reason, HMRC say in their note on‘bottom-up methodologies’:

Different methods and data sources are used, depending on best available, to estimate how much tax is lost within each area. HMRC uses internal data and operational knowledge to identify areas of potential tax loss. (HMRC 2017, p. 13) The IMF notes that there is room for improvement in this approach that would enhance HM Revenue & Customs’analysis of the tax gap, including the construc-tion of bottom-up estimates for the VAT gap in order to compare results from top-down estimates (IMF 2013, p. 35). This is in part, at least, because the IMF has suggested that any tax gap estimate, and especially those based on bottom-up methods, should not be used as the sole basis for inference about taxpayer

compliance behaviour (IMF 2013, p. 44). In so doing the IMF endorses the two differing methodologies but suggests they should be used simultaneously and reconciled when possible, and not be considered in isolation. What the IMF added was that bottom-up approaches offer very limited explanation of tax policy gaps, whilst noting that‘In general top-down models can be easily extended to estimate the policy gap.’(IMF 2013, p. 48) As they added:

As top down models generally involve creating an estimate of potential revenue by modelling how the current tax applies to the tax base, modelling the policy gap would require replacing the current tax structure in the model with some normative version of the tax structure. (IMF 2013, p. 48)

This issue is referred to again later in this chapter.

This being noted, the IMF’s advice appears to have been ignored: although the tax gap reporting by the UK’s HMRC is the most comprehensive such exercise undertaken by any country in the world,‘bottom-up’methods of estimation are still relied on for all taxes but VAT and excise duties (HMRC 2019, p. 4). In addition, 19 of the 36 estimates made are reported on the basis of experimental methodologies (HMRC 2019, p. 4). If HMRC stands at the forefront of tax gap methodology, their present lack of appropriate methodologies to estimate tax avoidance and evasion in key taxes suggests there is still room for considerable progress, not least because the tax policy gap is completely ignored.

This is also the case within much academic literature on the subject. Some such literaturefitsfirmly into the framework used by tax and regulatory authorities and seeks to improve the methodologies used by them; for example, Hamilton (2015) suggests there is a need to improve sample selection used in the bottom-up methods of tax gap estimation. Others are critical of specific methods of official tax gap estimation (Slemrod and Johns 2010), whilst still using them as a basis for analysis as they are the only ones available at present (Slemrod 2007). Yet others argue that all current and conventional tax gap estimates are unreliable as they omit behavioural responses: Gemmell and Hasseldine (2012), for example, make clear their lack of faith in current methodologies.

These approaches do, however, all consider tax gap data as part of a micro-economic appraisal of tax authority efficiency. The macro-micro-economic significance of tax gap data is reflected in another body of literature that uses the concept of the shadow economy as the basis for the assessment of tax gaps (Christie and Holzner 2006; Murphy 2012; Williams and Nadin 2012; Murphy 2019). This work only appears to be replicated by the Finnish government amongst regulatory and tax agencies (FTA 2016). The Finnish tax authority’s use of shadow economy data in this way is sufficiently unusual to mean that neither the OECD nor the European Commission consider them in the list of those jurisdictions preparing tax gap estimates.

Civil society activists (for example, Henry 2012, Murphy 2006, 2008, 2011, 2012, 2014a, 2014b and 2019, and Cobham and Jansky 2015) have also been involved in estimating tax gaps, although most have focussed on international and not national dimensions of this issue. Zucman (2014, 2015) has done the same in academic literature. These civil society surveys do share in common with the estimates prepared by tax authorities a focus on tax lost as an issue in its own right.

The broader academic literature on the issue makes it clear that a wider macro-economic dimension to this issue exists, partly as a result of the high rate of non-compliance that are repeatedly found. Kleven et al. (2011)find that almost 45 per cent of those self-employed in Denmark routinely avoid taxes, broadly replicating thefindings of Advani (2017, p. 2) who suggested, based on HMRC audit data, that 36 per cent of self-assessment taxpayers who were randomly audited included errors in their tax returns that resulted in an average underpayment of £2,320, a sum equivalent to 32 per cent of the average initial tax amount declared. Advani found that the rate of non-compliance did not vary greatly with income, but it is likely that a random audit programme would not reveal the behaviour of the wealthiest. This approach is, however, used to estimate the tax gap that they create (HMRC 2019, p. 4). That the ratio of aggregate misreported income to true income is generally higher with increasing income is found in several studies (Bishop et al. 2000; Johns and Slemrod 2010; Zucman et al. 2017). There are suggestions that this is partially the case because high incomes are often received in a form more prone to misreporting (Johns and Slemrod 2010). It has also been mooted that this trend may be more than just a coincidence and that in fact financial institutions, for example in Switzerland, could specifically cater to the tax management needs of higher earners (Zucman et al. 2017) and that they might in the process assist tax non-compliance by those using their services.

This issue of non-compliance is also addressed by Zucman (2013), who esti-mates that approximately 10 per cent of the global GDP is held in tax havens (Chapter 5 presents an overview of such estimations). Whilst some of this activity is perfectly legal, many instances are not: for example, two reports by the United States Senate in 2008 and in 2014 found that before 2009 between 85 and 95 per cent of accounts owned by US entities in Switzerland at Credit Suisse or UBS were left undeclared (Zucman et al. 2017).

Of significance for this chapter is that what this work on shadow economies, tax havens and international tax gaps all makes clear, in a way that adds to the suggestions of the IMF and TAXUD on tax policy gaps, is that tax gap data has a much greater purpose than simply assisting assessment of the efficiency of tax authorities. Tax gap data could assist policy-makers and economists when con-sidering the distribution (and redistribution) of tax burden and the impact that this might have on inequality and overall economic performance. The IMF (2017, p. ix) appear to have embraced this argument:

Advanced economies with relatively low levels of progressivity in their personal income tax (PIT) may therefore have scope for raising the top marginal tax rates without hampering economic growth. Different types of wealth taxes can also be considered.

The implication is that the tax policy gap has a role in appraising macro-economic policy and that the tax compliance gap is a measure of the effectiveness of a government in delivering this policy. This chapter suggests that this is the way in which the tax gap might have greatest use in the future.