BEPS Policy Failure—The Case of EU Country-By-Country Reporting¹

In 2013, the European Union included a requirement that the EU-based banks publish a limited form of country-by-country reporting (CBCR) in the revised Capital Requirements Directive IV (CRD IV) [15]. As noted in this article, the objective was to “allow stakeholders to gain a better understanding of the structures of financial groups, their activities, and geographical presence and help to understand whether taxes are being paid where the actual business activity takes place” [18]. The initial objective of the research that underpins this article was to test whether this objective could be fulfilled by checking whether reliable estimates of profit misallocation by the reporting banks could be prepared based on the data they published. The objective appeared reasonable, given the stated objective of CRD IV in this regard. In practice, this research objective could not be fulfilled as planned: the data published as a result of the CRD IV requirements could not support that objective. This article explores how and why this happened, and what can be done about it.

Global corporations, and their growth in power and dominance, have raised increasing concerns among academic researchers [2, 25] and a whole field of corporate governance research has emerged in the past two decades [32, see, e.g.]. This research suggests that in their efforts to externalize costs, firms have been determined to reduce taxes, seeing taxation as a major burden and cost to the business, rather than an opportunity to repay states for vital infrastructure services and legal protections [2, 6, 56]. In a similar vein, their power has increasingly led to corporate boards seeing regulations of any kind as a cost or a burden to their profit-making purpose [30], and firms have actively shopped globally for minimal regulations and constraints [51]. This has eroded the tax base of countries and led to a race to the bottom.

Given the significant rates of corporation taxes charged on company profits, ranging from 0% to more than 40% of profits [31], these taxes have been a key target for minimization. Organizations such as Tax Justice Network [56] and movements such as Occupy Wall Street have had a significant impact in exposing the significant levels of corporate tax avoidance through media outlets and by direct campaigns and public rallies. They have suggested that income from high-tax jurisdictions is being shifted to low-tax jurisdictions, making it very difficult for nation states to collect fair taxes—leading to a major “tax gap” [47]. As a result, transnational regulatory organizations have been pressurized to respond. Under instruction from the G20 and G8, the Organization for Economic Cooperation and Development (OECD) took charge of the initiative to tackle global tax avoidance, through the base erosion and profit-shifting (BEPS) initiative [9, 39, 40]. The primary purpose of this exercise was to identify the levels of such tax avoidance and the locations used to pursue it and to use the resulting transparency to encourage local tax authorities to effectively police and recover taxes that should legitimately fall due within their jurisdictions.

Among the measures adopted as a part of the BEPS process was a form of CBCR [40]. This was explicitly derived from recommendations made by civil society groups [33, 37]. In this context, it is important to note that CBCR is based on accounting and not tax data [37]. The purpose of CBCR is to indicate whether the risk of BEPS exists and not to, in itself, be the basis for taxation assessment. That said, there is now a growing awareness that accounting data based on most existing accounting standards, including those issued by the International Financial Reporting Standard Foundation that are used by most multinational corporations, are not suitable for the appraisal of many taxation issues [53]. This is partly by design: the International Financial Reporting Standard foundation states that they are not intended for this purpose [26, para 1.10]. This might explain why every country adjusts accounting numbers when determining tax charges [53]. The issue is compounded by the fact that accounting rules and practices also vary internationally and that even where there are international standards, their interpretation is often variable, making the implementation and enforcement of tax rules, technically very complex and difficult to enforce [48, 53]. The consequence is that there are very serious and frequently intractable technical problems in determining what a fair taxation liability for a multinational corporation might be, and how this can be apportioned among states in an equitable manner. As a result Sikka and Murphy [53] proposed a whole new conceptual framework for tax accounting, something that has never been attempted before. In the current and likely continuing absence of the adoption of such a standard, there are serious problems for the enforcement of tax rules, given the power and resources of these giant corporations and the lack of any global tax monitoring authority. The chance of CBCR succeeding has, then, to be appraised within this context. CBCR combines financial reporting with a tax methodology in an attempt to identify the consequences of BEPS. What it cannot do is overcome the inherent deficits in the accounting of a multinational corporation if that accounting data are in itself not fit for tax reporting purposes. Although similar point has been made elsewhere [e.g. 19], our additional contribution is in using specific data from reports published by banks for the period 2013–2017. This research shows how these deficits were ignored in the policy design stage, leading to a significantly detrimental outcome from a regulatory and enforcement perspective.

In the light of these various concerns, this article addresses two issues and then outlines the research method adopted. The first is the development of CBCR and the motivations for it, including the appraisal of tax gaps. Next, it considers the motivation for the adoption of CBCR by regulatory authorities, concentrating in particular on the use made of it by the European Union to appraise the tax affairs of banks. The research method and objectives for the third and final part are then outlined. The findings from the research reveal significant problems with the new disclosures and their accuracy and reliability. The quality of the data published as a result of the EU regulation and the failings within it are considered before; lastly, conclusions are drawn on the apparent failure of this process to date.

The perception of a rising global tax gap between the corporate tax that should have been collected and tax that was actually paid, and evidence that developing countries are adversely impacted by the aggressive behavior of multinational corporations, led to calls for CBCR [33, 34, 35]. A form of CBCR was proposed by the United Nations in the 1970s but fell by the wayside under pressure from the Organization for Economic Cooperation and Development [12, 59]. The idea was independently revived in a new form in 2003 as a means of revealing the financial performance of a multinational corporation in each country in which it operated, something which was not then (or now) reported in the annual financial statements of those corporations [33, 36]. This most elementary form of accountability was strongly resisted by accounting standards setters [58].

It was hoped that this transparency of information would lead to better knowledge and empowerment for local tax authorities, especially in developing countries, to collect taxes that were rightly due to them [36]. Accounting was seen in this context as a tool for better tax enforcement and regulation despite the general problem with transparency regimes in achieving behavioral change [21]. One of the first global initiatives to require such transparency was created by the European Union in 2013 as part of the Capital Requirements Directive IV (CRD IV) regulations that were intended to improve the ability of the European banks to survive another global financial crisis [15]. Given the hopes and aspirations for that regulation, the present study was conducted to analyze the information revealed and to see whether substantial tax gaps were revealed.

The original intention of the research that underpins this article was to deliver a measure of the corporation tax gap estimated based on the reporting of major EU banks required under Article 89 of the CRD IV of the European Union that was adopted in June 2013 [15]. The objective of CRD IV was to “lay down rules concerning .. . access to the activity of credit institutions and investment firms [and to provide] supervisory powers and tools for the prudential supervision of [these] institutions by competent authorities” [15, Article 1]. In the process of doing so, the European Union included provision that the regulated institutions should

disclose annually, specifying, by Member State and by third country in which it has an establishment, the following information on a consolidated basis for the financial year (a) name(s), nature of activities and geographical location; (b) turnover; (c) number of employees on a full time equivalent basis; (d) profit or loss before tax; (e) tax on profit or loss; (f) public subsidies received.

Their stated purpose in creating this legislation was to

allow stakeholders to gain a better understanding of the structures of financial groups, their activities and geographical presence and help to understand whether taxes are being paid where the actual business activity takes place. Mandatory country-by-country reporting is an important element of the corporate responsibility of institutions towards stakeholders and society and will help to restore trust in the banking sector.

In that context, the original objective of this work can be seen as being consistent with the stated policy objective for the CRD IV Article 89 disclosures. The rationale stated by the European Union is based on the transparency and corporate responsibility, with restoring trust thrown into the mix, although no mention is made in the legislation on how this will be monitored and enforced and what the penalties are for noncompliance with CBCR requirements.

The research objective was then to calculate a “tax gap” estimate for each country for which reporting was made based on the principles of unitary taxation. Unitary taxation apportions the total group profit of a multinational corporation to jurisdictions based on a formula [11, 46, 47]. The classic apportionment formula used in unitary taxation is described as the Massachusetts apportionment [11]. This apportions total group profit based on a formula that gives equal weighting to third-party sales, number of employees, and assets in a location. CBCR was designed to provide the information for this purpose. Proponents of both CBCR and unitary taxation suggest that unitary taxation is a more equitable method of apportioning the total taxable profits of a multinational corporation to the locations where it trades than that offered by the arm's length pricing methodology OECD [49]. Although the use of unitary taxation has not been agreed upon by any global body, it was felt that CBCR information would both open the way toward global unitary taxation and achieve a reduction of tax avoidance and the tax gap relating to corporation tax in the meantime. As such there were significant hopes for this regulation at the time that it was enacted. Given how recent this data set is, there is hitherto little research on the implications of this new evidence.

The global regulation of tax has been a complex arena, with no central power or authority to codify, enforce, and punish tax evaders and avoiders [46]. The OECD initiative on BEPS was set against this background [39]. The demand for reform needs to be understood in the context of the period. In December 2012, the UK House of Commons Public Accounts Committee held public, and humiliating hearings into the tax affairs of Google, Amazon, and Starbucks [45], which were widely covered in the global media. Prime Minister Cameron responded to the subsequent public outrage in a speech at the World Economic Forum in January 2013 in which he promised action [10]. The response was the adoption, largely under pressure from the UK development NGOs, of a call for CBCR in the communiqué of the G8 Summit held at Lough Erne, Northern Ireland, in June 2013 [23], although, as was apparent at the time, there was little real understanding of what this meant [3].

The European Union was also very concerned about the tax practices of multinational corporations. Its original focus was on tax payments in the extractive industries [58, p. 1177], and directives requiring limited disclosures on a CBCR for that sector were passed in June 2013 [16, 17]. On the same day as these were passed, the European Union was also considering the revised Capital Requirements Directive (CRD IV) for banks and other financial institutions and intermediaries. Almost as an afterthought, in great haste, and with almost no consultation,

One of the authors of this paper, Richard Murphy, was telephoned by an MEP involved in negotiations the night before this Directive was passed to ask what should be included in it as the opportunity for enactment had arisen that day.

There were skeptics of this CBCR and its effectiveness from the outset. Evers et al. [19] demonstrated that neither consolidated or individual financial statements nor other existing data sources seem to be an appropriate basis for providing such country-specific information. They identified technical flaws in the quality and reliability of data, which would hamper effective tax policing. They also questioned the lack of a theoretical foundation in the definition of CBCR and the benefits of this information would not outweigh the costs of gathering and monitoring the information. Instead, Evers et al. [19] suggested that tax legislators should limit profit-shifting by enforcing tax rules and closing gaps in tax law. The evidence now available is that at least some of their data concerns may have been justified, and this is discussed below.

Other recent researches have investigated effects of the implementation of CRD IV on banks. It suggests an increase in taxes paid, a decrease in profit-shifting, and no change in returns. While recent evidence by Overesch and Wolff [43] suggested that European multinational banks increased their tax expenses relative to unaffected other banks after CBCR became mandatory and Joshi et al. [29] found a significant decrease in the income-shifting activities by the financial affiliates in the post-adoption period, Dutt et al. [14] did not find significant abnormal returns for the banks affected by the political decision to include a CBCR obligation. Brown et al. [7] investigated the CBCR data and found that there is evidence of abnormal revenues and wages in tax havens, something that we would expect to see in Groups practicing tax avoidance—CBCR exposes the scale and existence of tax havens, information that was not available before. Fatica and Gregori [20] used the CRD IV data and found that the bulk of profit-shifting takes place among subsidiaries, as foreign-to-foreign tax differences matter significantly more that home-to-foreign differentials. What is different about the present study is that it examines in detail the quality of data provided by CBCR and finds this to be seriously flawed and unreliable in a variety of ways.

The aim of the current research was not to test whether or not each bank for which a report could be located had profit shifted, or not. It was, instead, to test whether or not there appears to be systemic evidence of misreporting of tax liabilities by the banks subject to the CRD IV regime, which the European Union has implied that the data should make possible [18]. To test this hypothesis, a data set developed by a team of researchers at the Czech Republic's Charles University has been used. These data collated the Article 89, CRD IV reports published by 46 different banks for a period of 5 years (2013–2017, although not all published reports for 2013). These data have already been reported on for other purposes [27] and the data are publicly available online via Open Knowledge International [42]. This specific data set is similar to some previously used data sets such as those used by Bouvatier et al. [8], Fatica and Gregori [20], Oxfam [44] but is larger in terms of years and banks covered. The banks used for research purposes are listed in Table 1. It will be noted that some banks from the ranking are not included in the data set: despite best efforts, their CRD IV data could not be found on public record despite the fact that it is a legal requirement that this information be published: it would appear that they have chosen not to comply.

To test the hypothesis, a form of formula apportionment was applied to the data published by the banks sampled to determine whether their profit reporting appeared to be consistent with the location of the economic substance of their activities. If it was consistent, it was presumed that base erosion and profit-shifting was not taking place, and vice versa. As, however, the CRD IV Directive only reports information on some of the variables required for unitary apportionment based on the Massachusetts formula, a restricted form of unitary apportionment had to be undertaken. For example, the data published on turnover are in total, and not for third party sales as would ideally be required for formula apportionment, and there is no asset data required by CRD IV. However, this still permits a formula apportionment: if equal weighting is given to the two variables on the economic substance of activities for which data are available (turnover and full time equivalent employees), then the European Union's objective of “understand[ing] whether taxes are being paid where the actual business activity takes place” is capable of being tested. It is this methodology that the empirical research reported in this article is based upon.

The basis of calculation used was to collect the data for all banks and to then aggregate this, that is, the variables reported by each bank for each country in which they operated were aggregated for each year to create aggregated totals for all banks in the sample by country by year. An average of these totals by country was then prepared by totaling the yearly data and dividing by five. The logic for doing this was to overcome the issues noted with tax liability reporting being on inconsistent cash and accruals bases. Over time, tax paid on a cash basis should approximate to tax accrued if the accrual reporting is accurate: the averaging process over a reasonable time period should, then, have eliminated, as far as was possible, the impact of the apparent data disparities arising because of poor regulation. The process is, then, intended to improve the quality of conclusions drawn. The resulting averaged aggregated data were then reapportioned to countries based on a restricted unitary apportionment formula. This apportioned profits to states with half the allocation being based on the location of turnover and half on the location of staff. The resulting profit was then compared with the reported aggregate average profit for the jurisdiction to note a gross reallocation. This was then subject to valuation for tax purposes at the headline rate of tax applicable in the country in question or as otherwise noted below.

The data were initially sorted for the purposes of the research by bank and then by year. What quickly became apparent was that there were significant issues with the data. Three problems appeared most important. The first was that in some, but not all, cases, the turnover and profit data reported were inconsistent with that reported by the entity as a whole. This was because what might be termed a “bottom-up” basis for disclosure has been adopted by some, but not all, banks reporting Article 89 CRD IV data. When this approach has been adopted, the local accounts of the bank in question have been used as the basis for CRD IV reporting purposes, that is, this approach starts with subsidiary level reporting and uses that as the basis for CBCR. This, however, often results in intra-group transactions being reported more than once, usually because profit distributions from companies low in the corporate hierarchy reappear as income received, and so as profit arising, when accounted for in intermediate holding companies. These intermediate holding companies are common in some locations, such as Luxembourg, for example

This is an issue noted from the first time that CRD IV CBCR took place. For example, see Barclays Bank plc 2013 country-by-country report [4] and commentary upon it [38] that highlighted that high levels of turnover reporting in Luxembourg appeared to arise for this reason. It would appear that dividend income of intermediate holding companies is reported as turnover not infrequently.

This has in turn given rise to a second problem. This is the tendency of some banks to report profits or losses as arising in “other,” unspecified, jurisdictions for CRD IV reporting purposes. It is of course possible that some disclosures described as such may actually refer to otherwise unspecified locations, for example, those that the bank in question consider immaterial for separate reporting, even though this appears contrary to the requirements of CRD IV. More likely, these disclosures might also represent (or do at least approximate to) the income that is potentially double counted that the “bottom-up” basis of preparation gives rise to, as previously noted. Overall the aggregate disclosure does appear to suggest this, but this cannot be confirmed on a bank-by-bank basis.

The third issue is that there has been a difference in interpretation between countries when transcribing into local legislation the CRD IV requirement that tax on profit or loss be disclosed. For example, the United Kingdom interpreted this demand as requiring the disclosure of cash paid in settlement of corporation tax liabilities during the course of the year, following a precedent set by the European Union when previously requiring CBCR for companies operating in the extractive industries. Other countries, such as France, more reasonably interpreted this demand as requiring the disclosure of the corporation tax liability that might be owed in respect of profits declared during the course of a reporting period. The difference is significant: most of the corporation tax paid during the course of any accounting period relates to profits arising in earlier periods and this sum will, therefore, not relate to the profit declared in the current period. The cash paid in respect of corporation tax during a period might then be significantly different to the sum that might be due on the profits arising during the course of the period in question.

The consequence of these differences is that a lack of comparability arises for three reasons. First, the tax paid declared in some countries cannot be readily compared with the profits declared in those same countries because they are stated on different accounting bases. This problem is exacerbated in those locations, such as the Nordic states, where as Table 2 shows, tax reported for CRD IV purposes is as disclosed in the income statement of the reporting bank, meaning that the disclosure in question includes deferred tax provisions, whereas reporting in the United Kingdom is intended to exclude such items but clearly does not always do so despite that fact, as the reporting of Standard Chartered and the Nationwide Building Society reveals. Consistency to ensure comparability is a key quality required of all accounting data, and it is absent in these cases. It is possible that aggregation for a period of time should eliminate at least some of these differences, although this cannot be guaranteed, most especially if losses also arise during a period. Second, comparison of Article 89 CRD IV data among countries where differing bases of accounting apply is not necessarily possible in this case, making the drawing of conclusions from this information much harder. Third, because tax reporting in company accounts is always undertaken, in the first instance, by comparing liabilities owed in respect of the period on profits arising during the course of that same period, there is, as a result of this reporting anomaly, a risk that the CRD IV reporting of banks in places such as the United Kingdom might not compare with the audited financial results of the banks in question in such places. This risk also arises when a “bottom-up” basis for reporting, starting with local accounts rather than from group consolidated accounts, is used.

It is stressed that these various bases of reporting are not, in themselves, wrong. A top-down approach has merit in offering a readily transparent reconciliation with the published audited accounts, while, in contrast, a bottom-up approach might provide better quality information to assist the appraisal of where the economic substance of transactions really arises, which is the objective of CBCR. Similarly, reporting both tax provisions in a profit and loss account and tax paid is useful, not least because of the comparison between the two that is enabled, although, unfortunately, CRD IV does not require both, unlike the OECD CBCR requirement studied recently by Cobham et al. [13] and Garcia Bernardo et al. [22]. The point is that there is not error on display here, but that there is instead a lack of precision in defining the required disclosures that has given rise to the preparation of inconsistent data that are undermined the objective of this process, which was to reliably indicate whether profit-shifting was taking place, where and by whom on a consistent and comparable basis. Requiring disclosure of tax provided in both the profit and the loss account and paid as shown by the cash flow, and the reconciliation of other variables to the published accounting data if prepared on a bottom-up basis, would overcome most of these problems. Some banks have appreciated the merits of such reconciliations and voluntarily provide them: Barlcays Bank [5] being a notable example, but the fact that they are an exception does reinforce this point.

To compensate for these issues of comparability, and because companies are required to report the corporation tax that they pay in a period in their cash flow disclosure under International Financial Reporting Standards, it should be expected that the total CRD IV tax cash paid should, on whatever basis it is reported, broadly reconcile with these data in the financial statements instead. To test this last hypothesis, and the potential scale of misreporting that might arise from the use of bottom-up accounting approaches for CBCR, the data reported by both Nordic and the UK banks in their Article 89 CRD IV reports were compared with the similarly described disclosures made in their audited financial statements for the same apparent periods. In each case, the comparison was restricted solely to the matter required to be disclosed by Article 89 reporting. The comparison was undertaken for each year from 2013 onwards if data were available for that year and for 2014–2017 in every case. The complete results are presented in Appendix 1, with that for 2017 being as shown in Table 2, split between the two regions.

As will be noted for 2017, and as Appendix 1 also makes clear for other years, there are differences of significance between the two sources in the case of many of these banks and in both areas. In particular, although it would appear that Nordea Bank, Svenska Handelsbanken, Lloyds Banking Group plc, and the Royal Bank of Scotland Group plc did almost certainly prepare their CRD IV reporting on what might be called a “top-down” basis (i.e., they started from the consolidated accounting data and attributed it to its country of origin) to ensure that the CRD IV disclosures made reconciled almost precisely with their audited accounts, and the Nationwide Building Society, DNB, and Swedbank might also have largely adopted this approach; the other banks that reported appeared not to do so. They did, instead, appear to adopt either a “bottom-up” approach or some other basis of accounting, with what might best be described as substantial differences in overall disclosure between the audited financial statements and the CRD IV reports arising as a result, most especially with regard to the reporting of turnover. These differences appear irreconcilable in some cases based on the disclosures made. It is surprising that, on occasion, these differences even extend to the number of employees. It should be noted that the differences on cash flow should be treated with caution: that in CRD IV, data are taken as being the same as profit and loss data when no other information is available because this disclosure is meant to represent tax paid [18].

As a consequence and in an attempt to counter the resulting possible distortions, a second aggregation was undertaken for the sample of all banks for which data have been collected. This aggregation created a single set of data for all the banks for all the reporting periods. The resulting effective tax rates reported for each of the 144 jurisdictions (plus one “other” location) for which data were collected is reported by year and in sample aggregate in Appendix 2. As is apparent from that data, the variations in reported effective tax rates implicit in Article 89, CRD IV data are substantial. The effective tax rate is calculated as the ratio of tax declared to declared profits for these purposes. What Appendix 2 also makes clear is that other ratios, such as average turnover per employee and average profit per employee, also produce anachronistic reporting based on these data. Some is due to the small level of activity, but much is not, while the average of more than 12,000 employees located in unknown jurisdictions makes no sense at all.

To check the credibility of reported variations in the calculated rates and effective tax rates, these were compared with two recent publications reporting on those rates. The first was from the OECD, published in January 2019 [41], which data set provides forward-looking or law-based effective tax rates for 70 of the jurisdictions in which banks reported the presence in their CRD IV data. The second, in this case backward-looking or data-based effective tax rates and thus similar to the estimates presented in this article, is by one of the authors of this article [28], which refers only to multinational companies within the European Union and covers the years 2011–2015. For the sake of comparison statutory headline tax rates for all the jurisdictions that had data reported for them by banks subject to Article 89, CRD IV disclosure were also noted. One data source for this was the OECD [41]. Another was the list published by KPMG [31] supplemented where data were missing for jurisdictions for which banks had disclosed data by information produced by other global professional services firms (mainly EY and PricewaterhouseCoopers). The resulting data are noted and compared in Appendix 3. There is surprising alignment between the effective tax rates reported by the OECD and headline tax rates. In contrast, the effective tax rates reported by Janský [28] showed greater variation, with some marked differences on occasion. Those from the CRD IV data appear to bear little relationship to other reported rates in a great many cases: the possible reasons for this have already been noted.

Despite these concerns about data quality, it was decided to prepare tax gap estimates based on the CRD IV data. This was because of the original objective of this work. It was also the purpose for which CBCR was designed [36]. In addition, the European Union had stated that working out whether such gaps might arise was one of their intended purposes that these data were intended to facilitate [18]. The unitary method for apportioning profits to jurisdictions used has already been noted. The tax gap estimate was prepared based on the averaged aggregated (i.e., all bank) data for the 5-year period for which the sample of banks reporting CRD IV data supplied information. These data were used to then suggest average aggregated misallocated profits. To estimate the tax impact of these misallocations across the sample as a whole, headline tax data were used because of the uncertainties and discrepancies noted in effective tax reporting and because effective tax rate data were only available for about half the countries for which data were reported. The effective tax rate data based on CRD IV data appeared too unreliable to use. The OECD reported headline tax rate was the preferred choice of tax rate used for this purpose. When such data were not available a rate secured from KPMG or another professional services firm was used instead. When no rate was available, an average corporate income tax rate of 24%, based on the KPMG data, was used instead. For the sake of comparison, a second tax gap estimate was then prepared for the EU member states alone. In this case, effective tax rate data from Janský [28] were used, with comparison then being made to the tax gap data for those same EU states based on their headline tax rates. To determine tax gaps, profits over- and underreported by jurisdiction are noted separately, which means that the tax gained or lost is noted separately by jurisdiction as a consequence. The results sorted in order of overall tax losses from profit shifting to those gaining from the process are presented in Appendix 4.

As that appendix notes, the process of profit-shifting is, by definition, a zero sum game: the net gains and losses must be equal because it is the profit of a single entity that the unitary apportionment reallocates for the purpose of preparing the tax gap estimate. This, however, is not true of the tax gains and losses resulting from those relocated profits. As expected, the data show that the countries suffering losses from profit-shifting lose more than those gaining appear to win from the process. Using the sample as a whole, and, therefore, by implication relying on headline tax rates to represent effective tax rates, what is surprising is that total losses, expressed in terms of tax revenues, amount to €7.37bn but the gains are not much less, at €6.47bn, implying a net worldwide gain for these banks of approximately €0.9bn as a result. What the evidence from the tax gap estimate for the EU members states does, however, make clear is that using what is thought to be much more credible data on effective tax rates has a significant impact on this calculation. Using headline tax rates, the losses of the EU member states to profit-shifting by the EU banks amounts to €5.31bn and the gain to €3.18bn, at a net cost of €2.13bn. However, when the costs of the same profit-shifting are estimated using more credible effective tax rates both figures fall to €4.79bn and €1.63bn, respectively, but the net cost rises considerably to €3.16bn as a result. The main impact is to be seen in those locations with superficially high corporation tax rates but low effective rates: Luxembourg is a prime example; it has the third highest overall gain from profit misallocation (being ranked behind Hong Kong and, rather surprisingly, Sweden, in this regard) but a substantial overall difference between nominal and effective tax rates. In this respect, the findings replicate and support those of Brown et al. [7].

Overall, it is apparent that some expected jurisdictions, such as Luxembourg, Hong Kong, Belgium, and Ireland, are gaining from profit misallocation, but so are many other states that are not recognized as tax havens. Indeed, many locations thought to be tax havens hardly feature in the misallocations: Jersey is the most notable to do so, while the Cayman Islands and the British Virgin Islands are hardly noticeable, based upon these data. As expected then, CBCR poses as many questions as it answers while unambiguously suggesting that profit-shifting does create significant costs for many states, of which the largest three to suffer are Italy, Spain, and the United Kingdom, in that order. The question remains though as to whether the available data can sustain these conclusions.

As the evidence presented in this article shows, the objectives stated by the European Union for the CRD IV CBCR disclosures [18] have not been met, at least as they might have desired. It is not possible to reliably appraise whether profits have been appropriately apportioned by the reporting banks to the jurisdictions in which they operate. Most especially, it has not been possible to determine whether tax is appropriately paid by each of them in each such location. This is the consequence of a number of noted failings inherent in the CRD IV regulation and in the way in which it has been implemented by member states and individual banks. As has been noted, some of these failings result from the way in which Article 89 of CRD IV was added to that directive in considerable haste. However, as the noted data on effective tax rates that have also been derived from accounting data also reveal, some of these problems are not peculiar to the CRD IV CBCR data. It would appear that currently available accounting data, and the methods by which it is generated and reported, do not provide sufficiently robust data for the purposes of analyzing the appropriateness of the tax payments made by multinational corporations. These defects could not be overcome by auditing the CRD IV data, or incorporating it into the statutory accounting framework of the companies in question: they are instead implicit in the design of the regulation and the limited scope of the data demanded. Although the motivation of those involved in this process was undoubtedly well intentioned, the outcome was less than optimal.

A number of important lessons need to be drawn from this research. First, those regulating corporate disclosures required for the purposes of appraising the appropriateness of tax payments must understand the need to require sufficient relevant, reliable, comprehensive, and comparable data to ensure that this task can be fulfilled. They must in that case seek to ensure that sufficient data are available for this purpose. The OECD version of CBCR does, for example, include seven key variables to appraise the appropriateness of profit apportionment, including data on the location of tangible asset investment [40]. The inclusion of a more comprehensive data set for CRD IV, when the requirement was already known [33], would have assisted its effectiveness.

Thereafter, it has to be appreciated that securing the regulation is in itself an insufficient process: specific guidance on its interpretation is required to ensure that its consistent application occurs in practice. Failure to do this will guarantee inconsistencies, and so a lack of comparability, within any resulting data because it would seem (as the Nordic and UK case studies included in this article make clear) multinational corporations are inclined to interpret reporting requirements in any way that suits their purpose unless specifically directed in their use.

Third, the CRD IV data did not appoint a regulator to oversee and enforce the quality of the information supplied as a result of the demands made by Article 89. This was an obvious failing, and one that followed on from the extractive industry's directive, that cannot be replicated in any future regulation that shares the objectives of this regulation. A mechanism to monitor reporting and to require its correction has to be established if regulation of this sort is to be effective.

These matters are of current significance: the European Commission still has an extant proposal for the public reporting of CBCR data by all large multinational corporations operating within the European Union. This proposal has been stalled by the European Council at present. The lessons from CRD IV must be taken into account before it progresses further. Effective accounting regulation is essential in the fight against tax abuse. As yet, it would seem that regulators have not learned how to deliver it. The result is that although some [e.g. 20] suggest that CRD IV reporting has been of benefit in the fight against tax avoidance, this survey shows it could achieve much more. A similar conclusion has, at least as far as data issues are concerned, been reached by Cobham et al. [13] with regards to initial reporting of OECD-based CBCR data. What appears clear is that tax avoidance by profit-shifting will not be beaten until reliable accounting underpins the effort.