Tax Haven or Conduit?
A Discourse Analysis of Dutch Media and Politics
Martijn van der Hoek
Master thesis Political Science – Political Theory specialization June 2020
Supervisor: dr. Gordon Arlen Second reader: dr. Vivienne Matthies-Boon
1
Table of Contents
Table of Contents ... 1
Introduction ... 2
Theoretical Framework ... 3
Literature on Tax Havens ... 4
Discourse on Tax Havens ... 11
Methodology ... 15
Media Discourse ... 16
Political Discourse ... 18
Performing the Analysis ... 18
Limitations ... 19
Ethical Considerations ... 19
Results of the Discourse Analysis ... 20
Media Discourse ... 20
Political Discourse ... 35
Discussion ... 40
2 Introduction
Following a 2013 debate in the Dutch parliament on tax treaties, parliamentarians voted by a margin of 124 to 26 that, henceforth, the Netherlands ought no longer to be referred to as a tax haven (Tweede Kamer, 2020a). The negative connotations around the term have meant that, historically, even states who are widely considered to be tax havens have shunned the term. Nonetheless, that such a decision is made by parliamentary vote raises more questions than it answers, for example, over what makes a country a tax haven. But more so, what is so pressing about the Dutch situation that this vote was held and overwhelmingly supported? These questions are interesting for a range of both academic and socio-political reasons. In academic research tax havens have become a frequent object of study in the past few years, particularly within economics and related fields (e.g., Zucman, 2015; Palan, Murphy and Chavagneux, 2010). Research in these fields has focused primarily on getting a better
understanding of tax havens, both by investigating their history and generating new data about their current importance. In political theory the topic has gathered some attention as well, where central questions concern the normative implications of tax competition (e.g., Dietsch, 2015; Risse and Meyer, 2019). However, there is very little research with an explicit focus on what debates over tax sheltering look like in practice, and which dynamics exist in the
domestic politics of tax havens. Questions over which stakeholders are involved in such debates and how they attempt to make themselves heard largely fall outside the scope of existing research. Similarly, it is an open question whether the conclusions drawn in this academic research are reflected in popular discourse, and why or why not that is the case. In short, there appears to exist a significant gap in the academic study of tax havens that this thesis attempts to address. More details about this gap and how this thesis relates to existing research is discussed in the theoretical framework.
From a social and political perspective this thesis can also be of value in a number of ways. Firstly, by looking at Dutch discourse on the country’s tax structures and financial services industry and connecting those to the academic research on these topics, certain disparities between the two can be found. If those disparities are significant, that may mean there are problems of democratic accountability in Dutch politics. To elaborate, in order for citizens to make informed decisions in a democratic context, it is necessary for them to have access to information about public affairs and their consequences. If it thus turns out that the state of the Dutch tax system is not accurately reflected in public discourse, then the ability to make informed decisions is harmed. The same goes for normative problems that may arise from the
3
existence of tax havens. These have both domestic and international consequences. Within tax havens, one effect of tax avoidance is the creation of a rent-seeking industry consisting of tax lawyers and consultants which advise on and construct the legal-financial structures that are used to avoid taxes (e.g, Saez and Zucman, 2019, pp.73-76). Internationally, the money that flows into tax havens goes untaxed (or undertaxed) in the country of origin; this leads to problems of injustice, and can also contribute to uneven development between states. On this dimension as well a possible lack of public knowledge obscures these issues from being discussed and politically addressed.
To determine whether any of this is the case in the Netherlands, it is necessary to chart first of all what popular discourse on this topic looks like in the country. To do so, this thesis asks the following research question: ‘What are the defining characteristics of the Dutch discourse about the status of the Netherlands as a tax haven, as it exists in media and party politics?’ To answer this question in full the thesis is broken up into four main sections. First, an extensive theoretical framework that draws on existing literature, then a discussion of the research design and its justification, followed by two sections with the results of the discourse analysis: one on the media, and one on party politics. The discussion at the end of the thesis goes over the implications of the findings and suggests the types of future research that could be valuable.
Theoretical Framework
Given the fact that this thesis touches upon a number of fields of study, and that its topic is relatively unexplored, the theoretical framework does not consist of a single model, but rather collects and discusses the implications of a broad range of literature that is relevant. Where possible hypotheses that can be drawn from this research are put forward. The discussed literature, broadly speaking, falls into two categories. The first half of this literature review looks at the existing research on tax havens. At the heart of this thesis, for example, is the simple question of what a tax haven is; the academic research on this is discussed first. Other central questions in this first section are about the different ways tax havens can be
conceptualized, what the scale of the problem of tax havens is, what their consequences are, and where the Netherlands fits into this academic discussion. The second section of the literature review forms the bridge between the tax haven literature and the discourse analysis
4
in this thesis. There, I discuss (the lack of) research on the discourse about tax havens, and some normative and political implications findings about this discourse might have.
Literature on Tax Havens
One of the central features of the academic literature on tax havens (alternatively called offshore financial centers, or OFCs) is that there is no universally agreed upon definition of what a tax haven is (Palan, Murphy and Chavagneux, 2010, p.17). There are a few reasons for this, but two stand out. Firstly, tax havens come in different shapes and sizes (e.g., Garcia-Bernardo, et al., 2017 classifies ‘sinks’ and ‘conduits’ as two different types). Secondly, there exists a boundary drawing problem. While there is agreement on the types of policies and structures generally characteristic of tax havens, no objective measures for how egregious such characteristics must be to meet the ‘haven’-threshold exist (Palan, Murphy and
Chavagneux, 2010, pp.21-22). Moreover, even if one would draw such a boundary it would need redrawing over time as jurisdictions change their policies and become less or more haven-like, and existing havens adapt their strategies. Such changes can take place over relatively limited timespans; Sikka (2003, p. 366) argues that in roughly 20 years “the number of OFCs has doubled from about 30 to more than 60”, although again, this depends on one’s definition. In short, there is a definitional ambiguity at the core of existing research.
A second question concerns the existence of the aforementioned term ‘offshore financial center’. Palan, Murphy and Chavagneux (2010, p.24) argue that although OFCs and havens have different historical origins “today it is difficult to distinguish in practice between the activities of tax havens and those of OFCs” and over time “[t]he term OFC [has become] a polite reference to tax havens”. Effectively, the two can be used interchangeably. However, the normative associations around them are important for the discourse analysis in this thesis; one of the reasons that the use of the term tax haven can be controversial in popular discourse is because of its negative connotation.
Despite OFCs and tax havens thus lacking a precise definition, recent research has established that within these categories a useful classification can be made. By studying capital flows at the levels of individual firms, Garcia-Bernardo et al. (2017, pp.1-6) demonstrate that OFCs generally fall into one of two categories: sinks or conduits. Sinks “are jurisdictions that attract and retain foreign capital” (p.4). In other words, they are the jurisdictions that capital
ultimately ends up in, whereas conduits are “jurisdictions that act as intermediate destinations to sink-OFCs” (p.4). For this reason conduits are also referred to as “pass-through
5
jurisdictions” (Fernandez & Wigger, 2016, p.418). Another important finding for the discourse analysis is merited only a single sentence in Garcia-Bernardo et al. (2017, p.2): “territories in [the Sink-OFC] category are usually characterized as tax havens, such as the British Virgin Islands, the Cayman Islands and Bermuda”. That is to say, while most academic research says that the terms tax haven and OFC effectively refer to the same jurisdictions, in practice it is the case that the term tax haven is applied predominantly to sinks, and not to conduits.
Thinking about Tax Havens
When looking at the literature it becomes clear that, even though there is no universal
consensus on what a tax haven is, a few schools of thinking about the topic can be identified. First is a theoretical approach, which starts by conceptualizing what a tax haven is and then looks at real-world jurisdictions to see which of them most closely fit this description. The second approach does the opposite, and begins by looking at economic data in order to identify jurisdictions that stand out from the rest, and then builds a definition around those cases. Third is what Palan, Murphy and Chavagneux (2010, p.45) call the “consensual approach”, meaning that “the more authorities believe that a state serves as a tax haven, the more likely it is to be one”. All three have strengths and weaknesses worth discussing. The theoretical approach has a clear advantage when it comes to argumentative rigor. Rather than looking at a messy world, the adoption of an ideal type allows for the possibility of disagreements to be hashed out before the term is applied, and theoretically allows different scholars to converge on the same definition. In practice this is also the case: a few identifiers of this ideal type stand out across different academics who have adopted the term. For example, Sikka (2003, p.366) lists “low/zero taxes on business or investment income, no withholding taxes, light and flexible incorporation and licensing regimes, light and flexible regulatory regimes and flexible use of trusts and other special corporate vehicles”. Similarly, Palan, Murphy and Chavagneux (2010, pp.30, 33, 35) list “low or nil taxation”, “secrecy provisions” and “light and flexible incorporation”. The straightforward nature of such ideal types mean they also make their way into popular discourse; as one of the articles in this thesis’ discourse analysis states: “there is a scientific consensus that a tax haven combines no or very low taxes with secrecy” (Financieele Dagblad, 2019a). Although no source is
provided for that claim, these are the same two main components as in the previous definition. The advantages of this approach are clear; the definitions are elegant and simple to
6
As is to be expected with an idealized theory, the difficult part is in its application to the real, non-ideal world1. More concretely, each of the main elements (taxation, secrecy, and
incorporation) can be problematized in practice. To start with, there is the boundary problem already discussed. I.e., how much taxation is ‘low’? This depends on varying factors such as profit margins, the existing tax rates in similar jurisdictions, and so on. Secondly, even if such definitional boundaries are established, application can remain difficult. That is because there exist significant gaps between de jure tax rates and rates that are effectively paid, for example. Powerful actors, primarily multinational corporations and ultra-wealthy individuals, have the means to exploit difficult tax codes and international agreements in order to pay less tax than legislators intended (see Saez and Zucman, 2019 for an extensive historical study of how these discrepancies are created and grow). Additionally, these actors are aided in lowering their tax rates by what Winters (2011, p. 217) calls the “Income Defense Industry” (or
alternatively, the avoidance industry). In tax havens, this is a “highly specialized professional services industry consisting of consultants, marketing experts, lawyers, accountants and tax avoidance experts, offering tailor-made services” (Fernandez and Wigger, 2016, p.414). The effects are large: According to Zucman (2014, p.131) the effective tax rate paid by US companies “has been reduced by a third”, which has costs the government “$200 billion” in revenue. In short, because of the work done by the Income Defense Industry the link between the tax rates stated by governments and actual rates paid by the wealthy is muddied. As such, the work of identifying low-tax jurisdictions is more difficult, and careless application of an idealized standard may miss all but the most obvious tax havens, shielding others from scrutiny.
When it comes to secrecy and ease of incorporation the avoidance industry plays a similar role. Zucman (2015, p.19) provides an instructive example of how Swiss banks, forced by international political pressure to limit secrecy rules, got around this by shifting assets to shell corporations outside of the country. And even in jurisdictions where ownership is more clear and transparency required by law the complexity of corporate structures is itself a tool of significant opacity; Garcia-Bernardo et al. (2017, p.1) illustrate this by pointing out that “HSBC is composed of at least 828 legal corporate entities in 71 countries”. Thus, while ideal type definitions may be very suited for, for example, normative discussions about tax havens (e.g., Dietsch, 2015), their application is difficult. There are some ways to address this
1 An in-depth discussion of ideal theory falls far outside this thesis’ scope, but see for example
7
problem, Zoromé (2007) attempts to do so by arguing that what is unique about havens is their focus on the provision of financial services to foreigners, and then uses economic data (rather than the examination of laws or policies as stated in words) to conclude which states fit the picture. But this ultimately runs into the same problem as other theory-first approaches, namely, what justifies the exclusive focus on the provision of financial services to foreigners to begin with?
The data-based approach does not set out from first principles but rather begins by economically analyzing existing jurisdictions and then looks at which ones stand out empirically. Within this approach, there is both research using micro-level and macro-level data. Zucman (2015, pp.36-40) exemplifies the latter, by relying primarily on disparities in states’ national accounts balances. On the other hand, Garcia-Bernardo et al. (2017) are on the far end of the micro-side, looking at individual firms and then accumulating data about those. Both of these have their own strengths and drawbacks (the macro approach is relatively straightforward in producing global estimates about the volume of capital involved, while the micro shines a better light on details about capital flows), but for purposes of this thesis their differences are not of crucial importance; rather, both provide useful data. Also noteworthy about this data-approach, in part because of the secrecy and opacity mentioned earlier, is that the actual data necessary for this research has historically been sparse. It is only in recent years that this type of research is flourishing, and much research is still to be done. For the purpose of this thesis the data approach is more valuable than the theoretical one, primarily because research of this kind can objectively point out where capital goes, even when the relevant jurisdictions do not look like ideal-type havens. Concretely, one of the questions for the discourse analysis is whether the debate in the Netherlands accurately
reflects the (objectively measurable) reality about the Dutch position in offshore networks; the data answers half of this question. The downside of the data approach is that the interpretation and categorization of results may be difficult, as the results may not match concepts in
ordinary language by which they are easily described. The earlier point about the label ‘tax haven’ generally being used for sink OFCs and not conduit OFCs is a clear example of this. However, this is not a knock on the data approach, but instead shows that these ordinary language concepts may need rethinking. This theorizing is important work, and if there is a major weakness in the data-first approach it is that the theoretical implications of the results are sometimes treated in minimal or limited form.
8
The third way of identifying tax havens is what Palan, Murphy and Chavagneux (2010, p.45) described as the consensual approach: “the more authorities believe that a state serves as tax haven, the more likely it is to be one”. They use this list as a shorthand for the most clearly identifiable tax havens, but beyond these simple practical applications I would argue that this approach is methodologically lacking. There are three clear problems with it. First is
arbitrariness, more specifically, which authorities’ lists of tax havens are considered and which are not? Secondly, the methodological rigor of the consensual list depends on how the examined lists were constructed. In this case, these come from non-academic sources, which raises further questions about the accuracy of their methods. And third, perhaps most
important for this thesis, there is a political problem with this approach. To elaborate, the in- or exclusion of jurisdictions on lists of tax havens constructed by states, NGOs and
international organizations is in many cases the result of (geo-)political considerations. One clear example of this dynamic comes from an article in this thesis’ discourse analysis: when in 2009 the US government planned to list the Netherlands as a tax haven, this was made undone after lobbying attempts by the Dutch government (Financieele Dagblad, 2011). Another example can be found in Garcia-Bernardo et al. (2017, p.9): Both Luxembourg and the Netherlands, which the paper clearly identifies as tax havens, have historically been left off the EU’s list of tax havens, while non-member states with similar structures have been included. In the case of the Netherlands, the argument that conduits should not be considered havens may be an attempt at an explanation. For Luxembourg, however, this argument does not fly, as it is a clearly identifiable sink OFC and such an obvious case that Zucman (2015, p.86) calls it “the tax haven of all tax havens”. These problems should thus disqualify the consensual approach from academic use, but its dynamics are nonetheless important to consider in the discourse analysis.
Tax Havens in Global Context
To illustrate the context in which this research is taking place, three short points are worth making. First, that tax havens play an important role and are a significant feature of the contemporary global economic system. In fact, despite criticism towards tax havens and a list of policy initiatives to reduce their effectiveness, they have continued to flourish (Palan, Murphy and Chavagneux, 2010, p.226). Second, the volume of wealth flowing through tax havens is enormous. Zucman (2015, p.35) estimates that “8% of households’ financial wealth is held in tax havens”, which numerically comes down to roughly $8 trillion. Moreover, signs indicate that this number is still increasing. Third, the owners of this wealth are not uniformly
9
distributed across the world, but stark differences exist between countries. Alstadsæter, Johannesen and Zucman (2018, p.89; 95) show that in some Gulf and Latin American states, wealth with the equivalent value of more than half of these states’ GDPs flows into tax havens, whereas on the lower end of the spectrum countries like South Korea, Finland, Indonesia and Poland lose less than five percentage points worth of GDP.
Is the Netherlands a Tax Haven?
The question of whether the Netherlands is by academic definitions a tax haven depends to some degree on the definition being used. On the data-driven accounts that I have argued are more suitable to this thesis’ purposes the answer is quite clear. As mentioned,
Garcia-Bernardo’s micro-level account identifies the Netherlands as the main conduit OFC in the EU. From the macro perspective the conclusion is the same: Zucman (2014, p.128) shows how the Netherlands has over time become the leading OFC for American companies to book
corporate profits in. In short, on data-driven accounts the Netherlands can be considered a tax haven.
On the theoretical approach the answer is less equivocal, and the Netherlands may be missing from certain lists. The primary reason for this is that, on paper, the Netherlands does not obviously fit the traditional image of a zero-tax jurisdiction whose economy is dominated by the financial services industry. To recall my criticism of the theoretical approach, the
discrepancies between listed corporate tax rates and effective avoidance on the ground are missed by some theory-based approaches. However, this is not universally so, and even in older research the Netherlands may be included (e.g., Hampton, 1996, p.4). Van Dijk, Weyzig and Murphy (2006, p.14) explain this phenomenon effectively:
The Netherlands is clearly not a ‘pure’ tax haven. It could not be. It has a significant commercial base of its own which it has to tax in order to sustain its government expenditure. This is incompatible with the secretive, tax evasion assisting haven. However as this report will show, the Netherlands clearly is a country which is characterised by a preferential harmful tax regime.
In other words, to notice that the Netherlands is a tax haven on the theoretical approach requires one to dig deeper and looks specifically at its offshore sector. Fernandez and Wigger (2016, p.419) point out how this sector in the Netherlands is characterized by “tax exemption laws, a liberal withholding tax regime for royalties and interest income; [an] extensive network of bilateral double taxation and investment treaties [and a] highly competitive
10
professional-services industry specialized in tax and regulatory arbitrage”. In other words, even if capital only passes through the Netherlands and is ultimately held elsewhere, the country’s offshore sector looks upon closer examination much more like that of the ideal haven. To summarize the two approaches, by academic standards the Netherlands may thus be considered a tax haven, and accounts that do not consider it a haven are generally limited in their application to a handful of jurisdictions approaching the ideal type. Nonetheless, the existence of these idealized accounts remains relatively prevalent, and so it may be
hypothesized that similar perspectives will be encountered in the discourse analysis to argue that the Netherlands should not be considered a tax haven.
Economic Consequences
One of the reasons why the term tax haven has a negative connotation to so many is that the effects of offshore finance are seen as harmful. Saez and Zucman (2019, pp.88-109) provide an overview of some of these effects. One of these is the exacerbation of inequality, both from a generally distributional perspective, and also when breaking down different types of
income. These two effects follow from the same dynamic. In general, the wealthier someone is, the more likely they are to derive their income from capital, rather than labor. Because income from labor, generally in the form of salaries, is linked to a specific job (in a usually fixed or relatively fixed location) it is relatively hard to hide and financially immobile (see Saez and Zucman, pp.106-108 for an example). For capital this is much less so; Zucman (2015) already showed how the moving of assets was a central part of the attempt to remake the Swiss offshore sector after their banking secrecy laws were changed. Thus, the wealthy individuals who derive their income from capital can more easily dodge taxes, which
exacerbates inequality in incomes and wealth, and shifts a larger portion of states’ tax burdens to labor income.
For states who decide to go down the route of becoming a tax haven, whether this is the result of a deliberative political process or simply of economic pressures from powerful actors, one of the consequences is that rent-seeking is encouraged. This dynamic is fairly straightforward; the shifting of assets that is a central part of avoiding taxes only moves economic activity on paper. If Google moves its proprietary software to a holding company in a different country this has no actual effect on where its users and employees are. For a tax haven, however, it is precisely the attraction of these assets that provides them with income. Even if effective tax rates are lower in havens, these assets might still be taxed in a limited form. Additionally, the offshore sector generates its income through doing the work of shifting such assets.
11
Ultimately, the haven is thus provided with economic incentives to, on paper, hoover up as much foreign economic activity as is possible, without contributing to further economic growth.
To the extent that academic research suggests positive effects of the existence of tax havens, these are mostly limited to the offshore sector itself. Rose and Spiegel (2007) create an economic model to simulate the effects of offshore finance. Although they broadly share the rent-seeking conclusion, they do point out that OFCs “still have unintended positive
consequences. In particular, the presence of OFCs enhances the competitiveness of the local banking sector” (p.1329), and based on their model, these positive effects may outweigh the economic costs created elsewhere. Nonetheless, this model relies strongly on the economic effects created between countries in close proximity, and so their conclusions may only apply in limited form to the more extensive avoidance structures, which can make use of many jurisdictions at once.
To sum up the most important findings of the economic research, tax havens play an important role in today’s global economic structures. They generally come in two types: sinks, and conduits. By academic standards the Netherlands, a textbook example of a conduit, may fairly be considered a tax haven. However, in ordinary language this label is applied more often to sinks, which more closely match what tax havens are historically thought to look like. As such, it may be hypothesized that this distinction makes its way into the discourse analysis and can play an important role in the Dutch debate.
Discourse on Tax Havens
One of the major reasons that the topic of interest for thesis has so far gone unaddressed in the literature is that it touches upon, and in some way falls between, different fields of study without clearly or exclusively relating to any specific one. The fact that Garcia-Bernardo et al. (2017) pointed out how the academic definition of a tax haven may not line up with its use in ordinary language, for example, is no impediment to the economic dimension of their
research. As for the other fields this topic touches upon, research in political theory on tax havens focuses almost exclusively on their normative dimensions. Risse and Meyer’s (2019) discussion of tax competition in the context of global justice is a good example. Their focus is primarily on the normative effects of tax competition between sovereign states. Although the existence of tax havens and their function are of importance to their analysis, the domestic politics of tax havens are not conceptualized. In fact, in one of their examples they “suppose”
12
that a poor country “becom[es] a tax haven” (Risse and Meyer, 2019, p.496). The domestic politics of this process are a black box.
A very similar way of thinking can be seen in Dietsch (2015, p.2), who premises his
normative discussion on the fact that “Some states design their fiscal policy in order to attract, or to ‘catch,’ capital from abroad”. These types of premises are, in and of themselves, not overly problematic as it is obviously true that the economic pressures of global capitalism play an important role in shaping global regimes of taxation, thus leading to the creation of tax havens. However, the Netherlands and many other tax havens do not create their fiscal policies in a depoliticized vacuum focused only on the attraction of capital. While those incentives certainly play an important role, they are still part of a larger democratic process. The actors involved in this process are described in more detail in research from the field of political economy (e.g., Frieden, 1991, Palan, 2003), but there similar problems exist. That is, the focus remains on the material dimension rather than the political and discursive one. Also worth mentioning are lesser known papers on the history of specific tax havens: Van Fossen (2002) discusses Norfolk Island, while Rawlings (2004) describes how Vanuatu became a tax haven. However, their implications with regards to the Dutch situation are limited, as these are both cases where tiny economies came to be dominated by offshore services, to the point where local populations find themselves with a bigger material interest in the continued existence of these sectors. In the Dutch case, as discussed in Fernandez & Wigger (2016) the offshore sector is only one highly specialized sector of a bigger economy. Given that the Dutch economy is not entirely built off this sector, it can also be hypothesized that its
continued existence is more easily politicized, as the consequences of shrinking the sector are smaller than in the cases of tiny jurisdictions dominated by offshore business.
To summarize, the fields in which tax havens are a prominent object of study only minimally discuss the domestic politics of tax havens. From a substantive perspective, this is the gap that this thesis attempts to address.
Similar Research
From a methodological point of view, this thesis stands in a long tradition of media studies and discourse analyses on politics. Fairclough & Fairclough (2012, pp.117-234) provide a few clear examples of such studies, and demonstrate how an analysis of the arguments present in political rhetoric are at the core of a productive discourse analysis on politics. Although a
13
discourse analysis on politics is nothing new, a discourse analysis focusing on the domestic politics and media of offshore business in an OFC is to my knowledge relatively unique. The closest study similar to this thesis is Micek, Działek and Górecki (2011). These authors study the growth of the offshore sector in Kraków using a mixed methods approach, to compare the actual economic indicators about the effects of this sector with the promises made about its arrival in Kraków. For this latter part they analyzed newspaper articles from a five year period, in addition to materials from business agencies (p.1654). They find that “[t]he main discourse is dominated by the perspective of a country which promotes service centres and an optimistic vision of the creation of new employment for highly qualified employees” (p.1656). A specific example they name which may be relevant for thesis’ discourse analysis is the rhetorical device of the “Polish Silicon Valley” (p.1656), which is used as a positive analogy for what the Kraków offshore sector is supposed to represent. Upon critical examination, however, the authors conclude that not only does the analogy fail in a number of important ways, the economic data they gather shows that the promises made about the project were significantly exaggerated. Even though there are moderately positive effects in some economic indicators, these are not in proportion to the larger discourse (pp.1665-1666). Similar dynamics can thus be expected in this discourse analysis, i.e., the offshore sector may be presented as being of greater importance than it is in order to legitimate its further existence, regardless of the accuracy of such statements.
This relates closely to studies about framing and the effects of the way text and speech are presented in media and politics. Both positivist and more theoretical approaches to this subject are worth mentioning here. There is a strong consensus in the social sciences that media framing can have significant effects on the way consumers interpret an issue, and the associations they make as a result (Schuldt and Roh, 2014, p.530). Despite the fact that framing is a well-studied phenomenon, critically applying it to media readings can be difficult. To illustrate, Groeling (2013, pp.137-139), points out that one of the problems in identifying media bias is the fact that the audience cannot know which stories were selected for publication and why, let alone what the set of all possible stories is. As such, the positivist approach is more suited to look at audience reception and decision making. This does not just apply to media consumption; Kelly’s (2012) Framing Democracy shows that the behavioral models of subjects created in the social sciences can also be used to understand citizens’ decision making in political contexts. Although these models generate some useful insights,
14
their application to a critical discourse analysis is limited, as such analyses focus on the broader socio-political structures in which discourses function.
Thus, it is of importance for this thesis to consider some larger structural accounts of how the media functions, and what its relation to politics is like. One frequently cited model in this area is Herman and Chomsky’s (2002) Manufacturing Consent. In this work they sketch out a model of the media’s functioning based on its “ownership and control, dependence on other major funding sources (notably, advertisers), and mutual interests and relationships between the media and those who make the news and have the power to define it and explain what it means” (p.1). They then attempt to demonstrate the efficacy of this model through a number of case studies based on which they conclude that “the expectations of this model are realized, and often considerably surpassed, in the actual practice of the media in a range of crucial cases” (p.298). Although Manufacturing Consent is written based on the practices of the American press, the basic media structure is, materially, much like that of the Dutch
newspapers discussed. I.e., the basic profit model is identical, that of selling subscriptions to an audience and linking that audience to advertisers. The relation between the media and experts or politicians is also similar across the two contexts; journalists both have a role in critically examining those groups, and are at other times reliant on them for information. As such, the model and its implications should be kept in mind in the discourse analysis. One implication of the model is that, from the attitude the press takes towards the offshore sector, it may be possible to make inferences about their influence in various sectors of the press. This provides an interpretive framework for the media section of the discourse analysis. For the other part of the discourse analysis, which concerns the party-political side, it is useful to introduce a handful of concepts from Stanley’s (2015) How Propaganda Works. First is propaganda2, and specifically undermining propaganda (p.179):
Undermining propaganda is a claim that is presented as embodying a political ideal, but that is in the service of the kind of goal that tends to undermine that ideal. What this means is that the success of undermining propaganda depends on two things. First, it depends on people having beliefs that are resistant to the available evidence, the evidence that reveals the tension between goal and ideal. Secondly, since undermining
2 Note that while the term propaganda is historically associated with authoritarian contexts, Stanley’s
15
propaganda conceals a contradiction of sorts, the beliefs that are resistant to evidence must themselves be flawed in some way.
This last sentence relates to what Stanley (2015, p.180) calls “flawed ideology” or flawed ideological belief. He argues that such beliefs can be the result of existent socio-political inequalities, and over time become ingrained in a manner that is difficult to change.
Connecting these concepts to the discussion of the Netherlands as a tax haven, the ambiguity around the ordinary language use of the term tax haven is of importance. Particularly, the hypothesis can be put forward that some actors may deny that the Netherlands is a tax haven, and that its offshore businesses instead embodies the ideals of, for example, a free market and internationally competitive economy. However, as the section on economic consequences showed, the evidence for such claims is at best unclear, and so such claims might be
considered an example of what Stanley (2015) calls flawed ideological belief. Additionally, his notion that inequalities can solidify such beliefs may be linked to the disparities in effective taxation by those businesses and individuals that make use of offshore services and the rest of society. That is, the former might appear as more economically competitive than they are by virtue of their tax avoidance, but this can nonetheless reinforce the belief that their competitiveness is the sole reason for their wealth in the first place.
Stanley’s (2015) conceptual framework can thus be substantively related to the topic, but methodologically also aligns in some ways with the worldview of Critical Discourse Analysis. As Richardson (2007, p.27) points out, the fact that “language use may be ideological” is one of the central principles of Critical Discourse Analysis.
To summarize this section, the domestic politics of the Netherlands as a tax haven and the domestic politics of tax havens more generally has only scarcely been an object of academic study. There is thus a large gap in the existing literature which this thesis attempts to address, although the method by which it is doing so (i.e., Critical Discourse Analysis) is
well-established. To provide the context for this thesis and make it legible in relation to existing research, the theoretical framework is thus made up of studies from economics, political economy, political theory and social scientific studies on the media.
Methodology
To answer the research question a Critical Discourse Analysis, or CDA (see Fairclough, 2010), was performed. Critical Discourse Analysis as a method developed primarily in the
16
fields of linguistics and the humanities, and generally has a strong focus on the link between language and broader social structures (Richardson, 2007, p.26). Wodak (2001, p.2) describes CDA’s common use as follows:
The term CDA is used nowadays to refer more specifically to the critical linguistic approach of scholars who find the larger discursive unit of text to be the basic unit of communication. This research specifically considers institutional, political, gender and media discourses (in the broadest sense) which testify to more or less overt relations of struggle and conflict.
Therefore, a Critical Discourse Analysis is well suited to the topic of this thesis; media and political discourse are its two constitutive parts. Moreover, the central question of this thesis, the status of the Netherlands as a tax haven in discourse, is a site of political contestation with both social and material consequences.
To expand on this, CDA “sees both concrete social events and abstract social structures as part of social reality” (Fairclough, 2010, p.74). That is to say, even though CDA takes
discourse at its object of study, it does not say that social reality exists exclusively in the text. Instead, discourse plays a twofold role. Firstly as a representation of pre-existing material relations, and secondly as a platform for the reproduction of future social relations. Both materialist and idealist readings of social reality are thus compatible with Critical Discourse Analyses (Richardson, 2007, p.28-29).
Media Discourse
The media discourse analyzed consists of a sample of articles from five major Dutch newspapers spanning a period of ten years. These papers are the Algemeen Dagblad (AD),
NRC Handelsblad (NRC), the Telegraaf, the Volkskrant (VK), and the Financieele Dagblad
(FD). These papers were chosen for a number of reasons. The first four papers mentioned were used to ensure that the sample was representative of the larger Dutch media discourse; these are the four most popular papers in the Netherlands by size of readership. Additionally, the editorial leanings of these four papers cover much of the mainstream political spectrum in the Netherlands. The NRC and Volkskrant historically lean to the left in their editorial lines, while the Telegraaf has a right-wing slant. The Algemeen Dagblad does not have an explicit editorial line, and is effectively somewhere in between the aforementioned. The inclusion of these papers should thus ensure that the sample is representative of the debate within the Netherlands, and covers the most influential voices within it. The Financieele Dagblad was
17
not included because of the size of its readership (smaller than the other four but still
substantial), but because it is the biggest platform in the Dutch business press. Given the topic of tax sheltering, the inclusion of the FD might thus be used to reveal possible differences between the popular and business press, if those exist. The time period from which articles were sampled spans from 2010 to early 20203. This time period is broad enough to provide an overview of the discourse in general, but also reveal possible shifts over time, if those are present.
Having set these constraints on timespan and sources, keyword searches around the terms of ‘tax haven’ and ‘the Netherlands’ were performed on LexisNexis to create the sample, which consists of 855 articles. Below, Table 1 shows the breakdown of articles by year, whereas Table 2 shows the distribution across sources. Although Critical Discourse Analysis is a qualitative method, these numbers in themselves do generate some insights; these are discussed at the start of the next chapter.
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 20204
19 38 62 163 91 67 113 112 73 87 30
Table 1 – Number of articles by year
AD FD NRC Telegraaf VK
71 178 230 125 251
Table 2 – Number of articles by newspaper
As CDA does not constitute a single method, but rather encompasses a number of approaches, there is no single conventional or prescribed way in which data collection takes place (Meyer, 2001, pp.23-24). The choice for a keyword search was made in order to start with a large sample, the most useful articles within the sample could then be identified with more precision during the coding stage. Although CDAs generally focus on the quality of
interpretation rather than representativeness of the sample, the lack of prior research on this topic meant that a big sample could serve as a solid baseline for the research.
3 Specifically, articles until the 9th of May 2020 were included. Given the project’s time constraints
articles after this date could not be included.
18
Political Discourse
For the analysis of party political discourse, the initial plan was to examine parliamentary discourse (debates, memos, etc.) over the 2010 to 2020 timespan exclusively. However, a preliminary reading of this material showed that it had a few limitations; the discussion of the Netherlands as a tax haven was often tangential, and many documents are technical in nature. This is an interesting finding in its own right, but does not get to the core of the questions this thesis attempts to answer. To provide a more complete picture of party political discourse, the section instead focuses on a handful of prominent debates in which the nature of the
Netherlands as a tax haven is explicitly discussed, and combines this with an analysis of party programs. Specifically, the timespan being studied covers three elections (2010, 2012 and 2017) of the Dutch House of Representatives. For each of these elections, the programs released by participating parties were analyzed. Discussions of party politics and statements from politicians and parties in the sampled media articles were also used for this stage of the analysis.
Worth noting is also that, while the discourse analysis is formally separated into sections (media and politics), there is obviously frequent interaction between the two. That is, much of the political debate takes place in response to media publications, which may lead to more reporting, and so on. Moreover, the party politics of the Netherlands as a tax haven are not excluded to parliamentary debate and also influence media discourse. Although the results are thus presented in two separate sections, this separation is more muddy in practice. The finer implications and details of this interaction will be discussed with the results of the discourse analysis.
Performing the Analysis
Critical Discourse Analyses can examine texts at a number of levels, e.g., vocabulary,
grammar, text structure, the formality of language, etc. (Fairclough, 2010). Given the research topic, and the centrality of the term ‘tax haven’ this discourse analysis focuses mostly on the layer of vocabulary. On top of this, the political nature of the examined texts requires
extensive attention to be paid to types of argumentation (Fairclough & Fairclough, 2012). To elaborate, the analysis looks at what kind of arguments are made (and omitted), which
assumptions they rest on, and what this can tell us about the issue more broadly. The analysis thus focuses mostly on the content of discourse, and less on the form. Nonetheless, the form still plays into the analysis in some significant ways (consider party programs, which can be
19
structured in any way the authors want it to, and be of any length; very different from newspaper articles), and this is discussed where relevant.
As with the data collection, CDA does not prescribe a single manner in which data ought to be analyzed. But here as with the sampling, the lack of prior research meant that a structured approach was most suitable, so that the interpretation of the discourse was more likely to be complete. Thus, having gathered the samples to be analyzed, a preliminary reading was done, which in combination with the used academic literature informed a coding scheme for the analysis. The coding scheme’s major categories concerned the types of arguments being made, the presence of normative claims, the context in which articles are published (i.e., the preceding event), and the types of articles (e.g., opinion pieces, reporting, readers’ letters), and so on. This coding scheme proved mostly exhaustive, although some in-vivo codes and other minor codes were added in the case of unexpected findings that did not fit elsewhere into the scheme. MAXQDA software (VERBI Software, 2020) was used for the coding itself.
Limitations
The main limitation of this analysis is the fact that the thesis focuses on one specific case: the discussion of the Netherlands as a tax haven. The findings of the thesis can therefore not be straightforwardly applied to environments outside of the Netherlands. Although this limitation cannot be fully addressed unless a more comparative approach is taken, it can still be
mitigated in some ways. First, the fact that articles were selected through a keyword search meant that a number of articles in sample mention both the Netherlands and tax havens, but discuss other countries as havens. This provides a pseudo-comparative perspective, i.e., it reveals the implicit assumption that are made in defining tax havens outside the Netherlands. For a full comparative perspective more research is of course still needed, and suggestions to this end are made in the discussion. Additionally, there exist bodies of literature on both Dutch politics and other tax havens, these literatures in combination with this thesis may be used to generate further hypotheses.
Ethical Considerations
The source material for the discourse analysis consists exclusively of documented text and speech that exists in the public sphere, and was put there intentionally either by journalists or politicians, both in their professional capacities. As such, issues of confidentiality or
anonymity that often are part of qualitative research designs are not a problem that has to be confronted in this thesis. Moreover, this thesis is primarily interested in larger social
20
structures rather than the pinning down of any specific participant’s viewpoint. The biggest ethical consideration for the discourse analysis concerns accuracy, as the source material is written in Dutch, but the thesis written in English. Therefore, precise attention was paid to translation. Additionally, where source material is quoted in its translated form, the original Dutch is presented in footnotes.
Results of the Discourse Analysis
Media Discourse
In this section, I first discuss the structure of media discourse on a very general level,
returning to the numbers in tables 1 and 2, and include a discussion on the form of the debate (what structures it, why are articles published, and so on). Following this, I look more closely at the status of the Netherlands as a tax haven in media discourse. Is the country labeled a haven, and why or why not? Additionally, I discuss the main arguments in this debate, and I go over the normative dimensions of the discourse.
Interpreting the numbers
As shown by Table 1, the volume of articles published on this topic has ballooned over the last decade. Where previously a relatively marginal discussion, it has in a few number of years become a major topic in Dutch media discourse, and if current trends holds appears to be here to stay. Both the volume of articles and the longevity of the debate can be explained by a number of dynamics that are discussed more deeply later in the discourse analysis, but are worth explicating here in short.
One way in which the texts reveal that this discourse is a relatively new one is by looking at previous events referred to in the articles. For example, the 2016 release of the Panama Papers is referred to in an article a year after the fact as context for ongoing parliamentary discussion about tax avoidance (Giebels, 2017). A number of these events from the last decade have taken up a place in the discourse as historical markers which serve to ground the discussion. By contrast, the only pre-2010 event that attains this prominence in the discourse is an announcement by the Obama administration in May 2009 that the Netherlands will be part of a list of tax havens used by US companies. This led to severe pushback from the Dutch government, and ultimately the Netherlands was taken off the list before it went into effect or had any legal consequences (Financieele Dagblad, 2011). Nonetheless, this is referred to in
21
articles as late as December 2017, and is mentioned more frequently in the early 2010s (e.g., Tamminga, 2012 and NRC Handelsblad, 2013a; Kok, 2017). Thus, while the growth in the number of articles alone strongly suggests the increasing prominence of this debate, the discourse analysis itself provides qualitative support for this claim by showing how events from the 2010s (in addition to the one 2009 event just discussed) have become historical referents on this topic in a way that previous events have not. These referents provide a common basis around which the discourse is structured over time and through which it can be made sense of.
That is not to say, however, that it is only around these events that discussion of the topic takes place. Rather, this set of referents becomes more prominent in articles over time, so that individual articles are more often framed in the context of this larger debate, as opposed to being presented as separate incidents. The latter is more common early on in the time period. A good example is a 2011 article about HSBC. The article, written in response to reporting in the British press concerns HSBC employees who are registered as working for a Dutch HSBC subsidiary, despite performing little to no real economic activity in the Netherlands,
supposedly to avoid taxes (Telegraaf, 2011). The text states that “the company is said to use a tax haven to pay the least amount possible, namely the Netherlands”5, which rather than
outright denying or affirming the label instead stylistically suggests a sense of surprise at the fact that the British press would use this label for the Netherlands. Moreover, the article closes by mentioning that Dutch experts on the topic do not see the HSBC structure as a major issue as it concerns “tax avoidance” rather than outright “evasion”6, so that the legality of this
specific instance is the ultimate framing device. This as opposed to the Dutch role in larger structures of tax avoidance, the laws and rules that make this possible, and the normative dimension of the subject, all of which becomes more prominent over in articles over time. However, and this may be connected to the numbers in Table 2, the articles in the discourse analysis do show that the pace at which this discussion reaches prominence in the different newspapers differs and is politically mediated. For example, a 2010 opinion piece in the left-leaning NRC does label the Netherlands a tax haven7, in addition to discussing some of the legal structures used to avoid taxes and the harmful economic consequences of this avoidance
5 Trans. mine: “Het concern zou een belastingparadijs gebruiken om zo weinig mogelijk af te hoeven
dragen, namelijk Nederland.”
6 Trans. mine: “Er is sprake van belastingontwijking en niet van ontduiking.”
7 “Het is fijn om een belastingparadijs te zijn. Maar Nederland betaalt twee keer”, trans. mine: “It is
22
(Tamminga, 2010). It is only in later years that similar arguments pop up with the same frequency and prominence in the Telegraaf (e.g., Telegraaf, 2017a) and Algemeen Dagblad (e.g., Boogaard, 2015). Although the discourse thus becomes mainstream over the years, this is in part the result of an active attempt by authors such as Tamminga to make it salient, and the pejorative use of the tax haven label plays a big role in this. This explains the disparity in articles between the different papers, and moreover lines up with the cleavages in party politics that are the topic of the next chapter.
As for the longevity of the debate, the fact that the volume of articles appears to have stabilized over the last few years is noteworthy in its own way. That is to say, while some debates have been continually present in liberal democracies (e.g., over the appropriate level of social spending), the debate over the desirability of having a significant offshore sector (and whether this makes a country a tax haven) is not one of those debates at the core of all polities. In fact, it is very well imaginable that at some point consensus is reached on the topic, thus turning the debate into a more minor one.
However, there is another factor which explains the continued prominence of this discourse and is inherent to the topic itself. Namely, addressing the offshore sector requires a serious revision of Dutch laws and policies, and this would thus be a medium- to long-term project that requires extensive technical knowledge. For example, where payments on interests and royalties from the Netherlands to other countries had previously been tax-free, the
government announced in 2017 that (starting in 2021) there will be a withholding tax on such payments if they are transferred to some of the most low-tax jurisdictions in the world (based on de jure rates) (Financieele Dagblad, 2020). The lack of regulation on these payments had gathered criticism for years and shows up throughout the sample, but it is representative of the difficult nature to change tax policies; it takes more work than a single discretionary action from a government minister. As such, the sites of conflict within the discourse easily become sites of sustained conflict, thus explaining the drawn out nature of the debate.
The Structure of the Debate
The reasons why articles are published vary greatly, but the start of media cycles generally have to do with the release of new information. This term can be interpreted broadly. One clear example is the leaking of previously unknown information about tax havens and their users. The Panama Papers, Paradise Papers, and Luxleaks are all good examples of this category (Financieele Dagblad, 2016; Kooiman, 2017; Alonso, 2014). Other examples are the
23
releasing of reports about tax havens, and in some cases books on the topic lead to news coverage as well (e.g., Haegens, 2019). Additionally, judicial decisions about the legality of specific offshore constructions can generate news coverage (Berentsen, 2015). Next to these major categories a few miscellaneous events can situationally lead to news coverage (e.g., end-of-year reports by major companies).
Depending on the perceived importance of these initial articles more news may be generated, usually in a fairly distinct pattern. For topics perceived to be of minor importance news coverage can end after a single article article (e.g., Telegraaf, 2011). In longer news cycles, the initial articles are often followed up on after one or a few days by an article in which reactions to the initial news are described. Particularly the reactions by politicians get priority in these articles, which already reveals an interesting impulse. Namely that the press defers to politicians (implicitly or explicitly) on the question of whether this new information should lead to a policy response, and if so, what it should be. This reveals not only an interesting power dynamic between the media and politics on the topic, but also incentivizes politicians to respond with headline-grabbing statements. Jonker and Cats’ (2016) interview with GroenLinks (the left-liberal green party) leader Jesse Klaver is a good example; he starts the interview by arguing that tax avoidance contributes to global issues ranging from poverty and hunger to terrorism8, and in response to a question about labeling the Netherlands a tax haven he answers that it should unashamedly be called one.
Beyond these initial reactions from politicians, articles may also be written about expert perspectives on the event at hand. These experts can range from academics to those represent the offshore sector. However, these expert contributions appear more prevalently in the three papers that publish most on the topic (NRC Handelsblad, De Volkskrant, and FD). For the most prominent cases (examples being the Panama Papers, or a long-lasting legal dispute between Starbucks and the EU that involved a tax ruling between the company and the Dutch government), these news cycles become more diffuse and drag on. For example, Driessen (2017) discusses a number of parliamentary hearings about tax avoidance that were scheduled in response to the Panama Papers’ release, but only took place six months after that fact. By that time, such hearings can become their own news event, so that coverage picks up again, or attention may fade away if it does not lead to new policy.
8 “Als je naar grote wereldproblemen kijkt als armoede, honger of terrorisme, kom je uit bij
24
An important question is which actors determine the extent to which events are perceived as ‘important’ and thus deserving of news coverage. Even though it is ultimately up to
journalists and their editors to decide whether or not to write certain articles, the structure of these news cycles suggest that the other participants in the discourse play an important role. The aforementioned interview with Klaver illustrates this dynamic quite well; the newspapers in this sample generally do not publish long interviews with politicians unless they generate new or controversial insights, and for an obvious reason: their available word count (at least in print) is limited and thus scarce. But this works the other way around too, stakeholders strategically attempt to claim space in the papers for themselves. In some cases this can be observed quite literally. A NRC Handelsblad (2013b) article covers the release of two reports about the offshore sector. The first comes from SEO, who wrote their report on behalf of a lobbying organization of Dutch offshore businesses, whereas the second comes from Somo, an NGO which has repeatedly criticized the Dutch offshore sector and emphasized its harmful consequences. A Somo representative in the article mentions how they purposefully chose to release their report on the same day, because the two reports come to different conclusions and both want their perspective to be covered. They are successful in this; the first half of the article covers the SEO report, and the second half discusses Somo. Nonetheless, some choices still have to be made; the main claim of the SEO report (that businesses residing in the
Netherlands only on paper generate 3 billion euros worth of GDP) becomes the article’s headline.
Editorial Decisions
Given that the use of the term ‘tax haven’ as a descriptor for the Netherlands is contested in the press, one important question is how the newspapers deal with this contestation; do they use the label, and if so under which conditions? The picture painted by the sample is very consistent. None of the newspapers describe the Netherlands as a tax haven in their reporting. However, it is not the case that the label’s contested nature is itself the reason that it is being passed over as descriptor, because other countries are non-controversially called tax havens in reporting. Reijn (2016) serves as a good example of this dynamic: the article covers how IKEA used a Dutch subsidiary which exploited transfer pricing rules to avoid more than a billion euros in taxes over a five year period. Even though this Dutch construction is at the core of the article, the only mention of a tax haven comes later, when discussing an avoidance structure involving IKEA’s holding company. Some proceeds of this holding flow through
25
Liechtenstein, described as a “classic tax haven”9. In some cases these disparities can show up in the same sentence, Vogels (2019a) describes how “Ireland and the Netherlands, and tax havens like Bermuda of Panama are popular destinations for incorporation”10. This does not
mean that the term never shows up in reporting about the Netherlands; it is used regularly when quoting sources who describe the Netherlands, or in the form of questions. That is, sources are asked directly whether the Netherlands is haven, as in Trommelen (2012), and the article is headlined “4 questions about tax haven the Netherlands”11. However, that is about
the furthest articles will go in their reporting; the application of the label is an open question. In articles that do not fall under traditional reporting the label can be used freely. For
example, some opinion writers do label the Netherlands a tax haven without the slightest hint of controversy (Jansen, 2015; Wagendorp, 2016). And additionally, readers’ letters in which the phrase is used are similarly published without question (Algemeen Dagblad, 2020). In other words, the fact that the argument for the Netherlands being a tax haven is repeatedly made is something the newspapers are well-aware of. In fact, the existence of this debate itself is a type of meta-observation that is frequently referred to, even in the more traditional
reporting articles. Driessen (2014) metaphorically describes a relocation of the Starbucks head office from the Netherlands to the UK as providing “new ammunition in the continuing debate over ‘Tax Haven Holland’”12. Additionally, the fact that this debate has gone on for
roughly a decade now is only reinforcing these meta-observations, with a recent article describing as an “almost existential question”13 for the Netherlands (Witteman, 2019).
What this ultimately leads to is that, upon an initial reading of articles like these, one might get the impression that the editorial line is one of neutrality: because the discussion is ongoing, the newspapers refrain from coming down on either side of it. But, as shown, the label is freely used in reporting that discusses other tax havens around the world. In particular, those that look like the ‘ideal type’ havens with low tax rates and strong secrecy provisions. Thus, when discussing these jurisdictions, there implicitly is a working definition being used by these articles. If we extend the use of this definition to the Netherlands it is clear why the label is not applied; the Netherlands does not obviously and immediately match the traditional
9 Trans. mine: “Een klassiek belastingparadijs”
10 Trans. mine: “Ierland en Nederland, en belastingparadijzen als de Bermuda's of Panama zijn
populaire vestigingslanden.”
11 Trans. mine: “4 vragen over belastingparadijs Nederland”
12 Trans. mine: “nieuwe munitie in de slepende discussie over ‘Nederland Belastingparadijs’” 13 Trans. mine: “een bijna existentiële discussie”
26
image of a haven. But if that is the case, then other aspects of the discourse become difficult to explain. Specifically, one would then expect the meta-points over the importance of the discussion to be framed as being of less importance, as the papers have already chosen a definition that functions regardless of what the discussion says, but this is not the case.
Additionally, there is another type of rhetoric that is hard to square with such a reading. When in 2017 the EU Parliament voted not to include the Netherlands and other EU member states on a list of tax havens, this was covered through the frame that the Netherlands now officially was not a tax haven (Financieele Dagblad, 2017). Although it is true that such votes have may have legal implications, the frame of officiality is discursively stronger, and implies some authority on behalf of the EU to decide which countries are havens and which are not. Moreover, if there is a party that can legitimately claim the authority to take an official decision on whether the Netherlands is a haven, then that should end the discussion; this is evidently not the case.
Rather than being able to square these different frames, it seems in practice to be the case that these editorial decisions shift responsibility away from newspapers themselves, by placing the burden of judgment on external parties (either the abstract entity that is ‘the ongoing debate’, or to existing institutions). This approach might be defensible from the perspective of
neutrality. Functionally, however, one side of the discussion is in practice being listened to. That is, as long as the discussion is ongoing these papers are actively choosing not to describe the Netherlands as a haven. The burden is thus placed on those who point out that it is to convince other social actors, and ultimately the papers themselves.
Arguments in Popular Discourse
The main arguments for considering the Netherlands a tax haven fall into one of three categories. The first category has to do with the structures of avoidance (and in some cases evasion) that the Netherlands is a part of, in other words, that the Netherlands is an important pass-through jurisdiction. In these arguments the language used can at times be relatively ambiguous; terms like ‘pass-through jurisdiction’ or ‘conduit OFC’ with origins in academic research are used scarcely. Instead one is more likely to see the argument that the Netherlands facilitates tax avoidance (NRC Handelsblad, 2019; De Groot and Broekhuizen, 2013). Or in
27
more direct phrasing some state that large sums of money are being ‘channeled through’14 the Netherlands (van der Ploeg, 2019; Savelberg, 2017). However, while the facilitation of tax avoidance is pointed to primarily by those who consider the country a haven, the fact that the Netherlands is a pass-through jurisdiction does not necessarily lead to this conclusion. For example, Braaksma (2015) mentions that Google “channeled almost 10 billion in royalties from the Netherlands to tax haven Bermuda”15, and the CPB (an independent advisory body
to the government) claims explicitly that the country is “not a tax haven, but is a pass-through jurisdiction”16 (van Alfen and Mikkers. 2013). The fact that this metaphor of money being
‘channeled’ through the country is used so widely in the discourse (it shows up more than 300 times in the sample), shows how the role of the Netherlands in these financial networks is well-known; the actual disagreement is over what the discursive implications of this fact are. The second argument, in some cases used in conjunction with the first, is based on the sheer volume of money passing through the Netherlands, or relatedly, the number of individuals and companies incorporated in the Netherlands despite having little to no real economic activity in the country. The argument thus is that these parties are only incorporated here for purposes of tax avoidance. These arguments referring to volume are made most frequently in response to the release of NGO or academic research about tax havens. The aforementioned NGO Somo has made this argument repeatedly, listing the Netherlands as a major perpetrator when it comes to tax avoidance (Boogaard, 2015). Oxfam is another NGO that shows up frequently to make this argument. Not only do their spokespeople successfully position themselves among the group of experts that journalists ask for comments throughout the sample, they also release their own research on the topic. These reports routinely generate news coverage, and usually rank the Netherlands as one of the biggest tax havens in the world. To illustrate, in 2017 Oxfam research showed the Netherlands ranked only behind Bermuda and the Virgin Islands on a ranking of tax havens (Berentsen, 2017).
The previous section discussed how newspapers seem to look for other authorities to determine whether the Netherlands is a tax haven, and Oxfam’s strategy plays into this dynamic very well. The fact that their own research is at the basis of their claims provides
14 The terms ‘wegsluizen’ or ‘doorsluizen’ are used in Dutch. The verb comes from ‘sluis’, the Dutch
for an aquatic lock, which captures very well these flows of money are restructured in different jurisdictions to move them around more easily.
15 Trans. mine: “Google heeft in 2013 bijna EUR 10 miljard aan royalty's via Nederland naar
belastingparadijs Bermuda gesluisd.”
