Party System Institutionalization and Reliance ... - Wiley Online Library

Journal of International Development J. Int. Dev. 30, 274–301 (2018) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/jid.3347

PARTY SYSTEM INSTITUTIONALIZATION AND RELIANCE ON PERSONAL INCOME TAXATION IN DEVELOPING COUNTRIES 1

ARMIN VON SCHILLER1,2*,† Deutsches Institut für Entwicklungspolitik, Bonn, Germany 2 London School of Economics, London, UK

Abstract: This paper explores the effect of party system institutionalization on the relevance of personal income taxation in the tax composition. Based on a fiscal contractualism approach, I argue that institutionalized political party systems increase the capacity of political actors to commit credibly to fiscal contracts agreed with wealthy taxpayers. Consequently, especially where bureaucratic capacity is low, institutionalized political party systems should have a strong effect on the share of the tax burden that wealthy taxpayers are willing to accept. The analysis of panel data between 1990 and 2010 for a sample of over 90 countries supports this hypothesis. Party system institutionalization has a significant and strong positive effect on the relevance of the personal income taxation where bureaucratic capacity is low. At high levels of bureaucratic capacity, the effect disappears. The findings strongly support the claim that, particularly in developing countries, where bureaucratic capacity tends to be limited, taxation is best understood as a problem of credible commitment. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd. Keywords: party system institutionalization; personal income tax; bureaucratic capacity; fiscal contractualism JEL Classification: D78; H24; H30

*Correspondence to: Armin von Schiller, Deutsches Institut für Entwicklungspolitik, Tulpenfeld 6, 53113 Bonn, Germany. E-mail: [email protected] † I would like to express my gratitude to Luis Camacho, Luciana Cingolani, Jörg Faust, Stefan Leiderer, Arndt Leininger, Carlos Scartascini, Christian von Haldenwang and anonymous reviewer at the IDB, UNU-WIDER and the Journal of International Development for constructive comments. The paper also benefited from comments by seminar participants at the workshops ‘Political Institutions and Inclusive Development’ (Hertie School of Government, November 2015), ‘New Politics of Taxation’ (ECPR-Joint Sessions 2015) and ‘Taxation and Revenue Mobilization in Developing Countries’ (UNU-WIDER and the ICTD, March 2016) as well as the panel ‘The Politics of Taxation’ at the EPSA Annual Conference 2015. Special thanks go to Miguel NiñoZarazúa for sharing a harmonized version of the WIID dataset including observations for countries in 5-year averages. I am especially indebted to Mark Hallerberg and Mark Kayser for providing thorough feedback on previous versions of this paper.

© 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and the content is offered under identical terms.

Party System Institutionalization and PIT 275 1

INTRODUCTION

Taxation is high on the development agenda. The international development community, which for a long time had been focused on the spending side of the budget, is increasingly engaging in lively discussions about taxation. All actors now recognize the centrality of effective tax systems to inclusive and sustainable development, democratic governance and state-building (Fjeldstad, 2014; United Nations, 2015; World Bank and IMF, 2015; Mosley, 2015). Although ‘virtually every scholar agrees that taxes involve politics’ (Gould & Baker, 2002: 91), historically, political scientists have not paid much attention to the topic. This has changed recently, and the rapidly growing literature on the political economy of taxation is bringing politics more systematically into the tax game (von Haldenwang & von Schiller, 2016). Most of the literature in this field has been focused on the overall level of tax revenue collection and the occurrence of tax reform.1 Tax composition, understood as the relevance of each tax type in the tax mix, however, remains widely understudied.2 The absence of consistent body of literature on this topic is surprising, as it is crucial to understand how the burden of financing the state is distributed among different social groups. Following standard tax incidence assumptions, we can expect that the stronger the relevance of progressive taxes in the tax composition, the higher the burden carried by wealthy taxpayers will be. Interestingly, developing countries’ poor tax collection records seem to be connected to the poor performance of progressive tax types (OECD/AfDB/UNECA, 2010). In general, tax systems in developing countries rely heavily on indirect taxes and most prominently on the value added tax (hereafter VAT). Direct taxes—and especially those considered to have a more progressive tax incidence, such as the personal income taxation (hereafter PIT)—in many cases play a marginal role. Economic and administrative considerations partly explain and justify a stronger emphasis on more regressive tax types. However, as many scholars point out, from a political perspective—especially in democratic developing countries—the high level of inequality and the low performance of more progressive tax types is a fact ‘hard to reconcile with the workhorse model in political economy, that is, the median voter model of redistributive politics’ (Ardanaz & Scartascini, 2013: 1637). I argue that the concept of fiscal contractualism (Moore, 2008; Timmons, 2005) can help to explain this puzzle. Many developing countries do not have sufficient bureaucratic capacities to efficiently use coercion to tax their citizens. As a result, governments in these countries have a strong incentive (and need) to foster “quasi-voluntary tax compliance” (Levi, 1989). This incentive should be particularly strong when facing wealthy taxpayers. On the one hand, wealthy taxpayers can be considered the actors that are best equipped to evade taxation and resist it using political means. On the other hand, given the high convergence of economic and political power, governments will be reluctant to use coercion on actors whose support is crucial to remaining in power.3

1

This literature is mostly based on case studies (e.g. Gillis, 1989; Lora, 2007; Thirsk, 1997). For large N studies on OECD countries, see Basinger and Hallerberg (2004). Outside the OECD, the number of studies is even lower and mostly focus on Latin America (Focanti, Hallerberg, & Scartascini, 2013; Mahon, 2004). 2 Here again the focus has been on case studies, exceptions being Mahdavi (2008), Kenny and Winer (2006) and Timmons (2010a). 3 On difficulties and strategies to tax the wealthy, see Fairfield (2013). © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

276

A. von Schiller

Following fiscal contractualism, finding agreement between the government and the wealthy on the exchange of services for tax contributions would benefit both. However, defining and sustaining fiscal agreements is not an easy task. In particular, if governments cannot credibly commit to taxpayers that benefits will be provided in exchange for tax contributions, a defensive stance towards taxation will prevail. This is why taxation should be approached at least ‘partly as a game of credible commitment rather than a game of pure coercion’ (Timmons, 2010a: 211). In this paper, I concentrate on the dimension of commitment and examine the degree to which institutionalized political party systems, by enhancing the capacities of political actors to make credible commitments to wealthier taxpayers, influence the share of the tax burden that these taxpayers pay. More precisely, my understanding of party system institutionalization emphasizes the stability of the party system over time. The intuition behind my argument is simple: The relationship between taxpayers and the government faces classical principal-agent problems. The capacity of the political system to sustain agreements over time and to disincentivize opportunistic behaviour by political leaders is a core concern for wealthy taxpayers. Institutionalized party systems can mitigate these problems by encouraging clearer and more stable expectations of the behaviour of political actors in the future. Parties that endure tend to have more institutional strength and more closely attached voters (Ezrow, Tavits, & Homola, 2013; Tavits, 2012). Also, political parties in stable political party systems tend to have more moderate positions (Gallagher, Laver, & Mair, 2011) and are less personalistic (Mainwaring & Torcal, 2006: 221). Finally, stable political party systems tend to have less pronounced policy shifts (Scartascini, Stein, & Tommasi, 2013). All these characteristics make the political environment for wealthy taxpayers more reliable and credible and, therefore, enable that they accept to pay higher amounts of taxes. In a first step, I discuss the specificities of developing countries and how, because of limited bureaucratic capacity, the factors enabling credible commitment are particularly important in explaining contributions by wealthy taxpayers in these countries. The theoretical discussion leads to the hypothesis that party system institutionalization most prominently affects wealthy taxpayers’ acceptance of taxes on them where bureaucratic capacity levels are low. In a second step, I examine this hypothesis empirically. Strongly restricted availability and poor quality of public finance data are a major restriction in this field, especially when focusing on developing countries. In recent years, major initiatives have tried to close this gap. In this context, I employ the recently launched Government Revenue Dataset (Prichard, Cobham, & Goodall, 2014) that not only provides more reliable data than previous initiatives but also allows for a much broader empirical basis than normally used in the studies on the political determinants of taxation because of its comparatively large geographic coverage.4 The results provide support for my hypothesis. At low and middle levels of bureaucratic capacity, party system institutionalization has a strong positive effect on the relevance of PIT in the tax composition. This effect disappears where bureaucratic capacity levels are high. The results are robust to different specifications although the degree of heterogeneity 4

Note that as a robustness check, I also run the analysis using IMF (2012) data that the IMF Fiscal Affairs Department kindly shared with me. There are certain arguments supporting that the Government Revenue Dataset (GRD), which combines data from several major international databases, is the most reliable alternative to use when aiming at crossnational analysis (Prichard, 2016). Therefore GRD is prioritized, and results in the main text rely on data from this source.

© 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 277 is high. The analysis of subsamples indicates that the effect is particularly reliable in democratic settings and where government parties have a programmatic orientation. The broader significance of these findings is worth considering. First, the results suggest that short-term unilateral promises have limited prospects for getting wealthier taxpayers to pay more taxes in developing taxes. The acceptance of a bigger share of the tax burden hinges on the existence of more institutionalized political systems that are able to make credible, long-term commitments. Second, the results of this paper imply that focusing on the question of what states need in order to tax may obfuscate the crucial question of what citizens need in order to accept paying taxes, an aspect that receives far less attention in academic and policy debates. The paper proceeds in six sections. After this introduction, Section 2 summarizes the empirical literature on the determinants of tax composition. Section 3 presents the argument and the main hypothesis. Section 4 elaborates on the empirical approach. Section 5 discusses the results while section 6 concludes.

2 EMPIRICAL LITERATURE ON THE DETERMINANTS OF TAX COMPOSITION Most of the work on tax composition and the PIT has been focused on economic and administrative variables. One main result is that level of economic development heavily influences the level of tax collection and the tax composition. Developing countries rely more on corporate and trade taxes than developed ones.5 Closely related, in the process of economic globalization, developing countries have had far more problems replacing the associated decrease of trade taxes and have performed this most prominently by exploiting VAT. Furthermore, most studies conclude that—given the lack of an alternative tax base and the administrative challenge presented by ‘hard to collect taxes’ (e.g. the PIT)—relying on VAT and trade taxes is a rationally sound and economically efficient strategy. Political variables are since recent years receiving a rapidly growing attention in this debate. This process has been driven by the growing interest in the political economy dimension of public finance. Two main arguments dominate the discussion about the link between political factors and tax composition. The first one stresses the effect of political ideology. The second one focuses on the effect of political institutions.6 The relevance of domestic political institutions to economic policies in general, and taxation in particular, is well researched (Bird, Martínez-Vázquez, & Torgler, 2008; Cheibub, 1998; Kenny & Winer, 2006). The approach relies on the idea that political institutions affect the incentives for politicians to use particular tax instruments and react to external pressures or shocks in a certain way. In terms of empirical analysis, the variables that have received most attention are the number of veto players (Hallerberg & 5

For literature focusing on developing countries, see Aizenman and Jinjarak (2009), Agbeyegbe, Stotsky, and WoldeMariam (2006) and Baunsgaard and Keen (2010). The debate on developed economies has been centred on the idea of globalization-induced convergence. Excellent overviews of the debates can be found in Basinger and Hallerberg (2004), Plümper, Troeger, and Winner (2009) and Swank and Steinmo (2002). 6 This literature review is focused on the effect of political variables and does not discussed the literature focusing on economic and admistrative variables. An overview on the diverse determinants of taxation that have been analysed can be found in Hallerberg and Scartascini (2017). For the interest of this paper, it is relevant to highlight that existing studies show how bureaucratic capacity, which is a component of the interaction term analysed in this paper, has a significant positive effect on PIT increasing tax reforms (Hallerberg & Scartascini, 2017). © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

278

A. von Schiller

Basinger, 1998), the degree of party dominance (Steinmo & Tolbert, 1998), regime type (Cheibub, 1998; García & von Haldenwang, 2015),7 the level of decentralization (Kenny & Winer, 2006) and legislative malapportionment (Ardanaz & Scartascini, 2013). Two reasons explain why de facto most of these empirical studies are restricted to developed countries. First, data availability and quality are negatively correlated with level of development. Hence, the more specific the variable tested, the lower the number of developing countries in the sample. In addition, testing institutional arguments in developing countries also entails the problem, which in many cases, informal institutions play a stronger role than formal ones (Therkildsen, 2001: 111). As a result, assuming that the effect of formally similar institutions will be homogeneous across large samples will be difficult, as these same institutions will operate very differently in different countries. Partly overcoming this problem, much of the most influential work on institutions and taxation in developing countries has a strong focus on historical legacies and path dependencies. Using an historical political economy perspective, the characteristics of contemporary tax performance are explained by the constraining effect of sociopolitical patterns that emerged in the past (e.g. Mkandawire, 2010; Thirsk, 1997). Specifically with regard to the performance of the PIT, Lieberman (2003) offers the most prominent contribution. In his book, Lieberman argues that the historical pattern of interaction between the state and upper-middle classes is a crucial factor in understanding the consolidation of different ‘tax states’ and illustrates this by comparing the experience of the ‘adversarial tax state’ in Brazil with the ‘cooperative tax state’ in South Africa. As for partisanship, the literature provides strong evidence that it affects fiscal policies, tax composition and the distribution of the tax burden among social groups. However, much of the evidence is again based on developed countries, and interestingly, the research comes to partly counter-intuitive results. Against common expectations, several studies show that left-leaning governments tend to rely more heavily on regressive taxes. Several explanations have been proposed for this empirical puzzle. Kato (2003) argues that, in the context of globalization, left-leaning governments use indirect taxation more intensively because they have no alternative to finance the high levels of public expenditure that they aim at. In this line, Beramendi and Rueda (2007) emphasize the relevance of the institutional setting in which left-leaning governments operate and show how, in the absence of corporatism, left-leaning governments are associated with a strong use of progressive tax instruments, whereas in the presence of corporatism, they exploit more regressive taxes. Also, Cusack and Beramendi (2006) have shown how left-leaning governments rely more heavily than right-leaning ones on taxing labour. Analyses on the relationship between partisanship and taxation outside the developed economies are scarce and heavily concentrated on Latin America. In addition, existing studies also show inconsistent results. For instance, whereas Hart (2010) finds evidence that in Latin America, left-leaning governments tax less than right-leaning ones, the findings of Stein and Caro (2013)—based on a broader sample and newer data, as well as additional indicators for partisan ideology—contradict these results. 7

Works by Bird et al. (2008), Kenny and Winer (2006) and Profeta and Scabrosetti (2010) have deepened this analysis by exploring the effect of democracy on the performance of particular tax types. The results indicates that democracies tend to rely more strongly on more progressive tax types.


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 279 Timmons (2010a) offers an alternative perspective on the relationship between partisanship and taxation. In his model, partisanship serves primarily as a device for credible commitment towards voters. Timmons underlines that only if governments are credible will taxpayers accept paying more taxes in expectation of benefits in the future. As parties are most credible to their platforms, partisanship should result in the higher performance of taxes set on them. Timmons’ analysis of 18 OECD countries supports this hypothesis. Left-leaning governments tend to tax the poor more heavily but also focus spending on policy areas considered to be more beneficial to them. By contrast, rightleaning governments rely more on progressive tax types but also spend more on policies that can be expected to have a regressive incidence. In the following section, building on Timmons’ contribution, I develop my argument and hypothesis. In essence, I argue that the idea that taxation is a game of credible commitment holds for developing countries, especially when it comes to taxing the wealthy and in countries where bureaucratic capacity is low. I also claim that, given the specific characteristics and challenges that most developing countries share and face, the institutionalization of the political system rather than the ideology of the party in government is the factor that will foster credibility.

3 THE ARGUMENT: INSTITUTIONALIZED PARTY SYSTEMS AND THE RELEVANCE OF PROGRESSIVE TAX TYPES Governments need coercion in order to extract taxes from citizens not willing to pay them. Using coercion is costly and leads to suboptimal equilibria for both governments and taxpayers (Levi, 1989; Timmons, 2005, 2010a). Hence, governments and taxpayers have incentives to enter fiscal agreements. Governments prefer to exchange taxes for services in order to avoid the costs of coercion and gain efficient access to more revenue. For their part, taxpayers prefer fiscal agreements because they lead to predictable benefits in exchange for contributions. Not complying could be an option, but the potential sanctions in the case of being prosecuted might be high, and these payments would not be linked to any benefits. Thus, as long as the fiscal agreement is better than any outside option for both actors, exchanging contributions for state services should represent a self-enforcing equilibrium (Timmons, 2010a). Precisely, governments in developing countries should have strong incentives to foster this type of agreements, especially with the wealthy taxpayers. In many developing countries, bureaucratic capacities tend to be limited. As a result, monitoring and sanctioning those who are best positioned to evade taxation, namely, the wealthy, can be expected to be particularly costly and inefficient.8 In addition, regardless of whether the capacity to use coercion exists, the actual threat to use it against wealthy taxpayers may lack credibility. Because of the concentration of economic and social power, small groups of wealthy taxpayers can be assumed to control much de facto political power (Acemoglu & Robinson, 2006). This disproportionate political influence makes the government’s threat to coerce wealthy taxpayers weak. 8

For a comprehensive discussion on how bureaucratic capacities influence the pool of policies available for a country, see Chuaire, Scartascini, and Tommasi (2014). Their analysis has an emphasis on the ability to respond to risks and uncertainties introduced by the volatility of international markets, but the argument can be transferred easily to other policy areas such as taxation.


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

280

A. von Schiller

The condition for wealthy taxpayers to enter into a fiscal-exchange agreement and voluntarily accept paying taxes in the absence of strong coercion mechanisms is the expectation of getting something valuable in exchange for taxes (Berens & von Schiller, 2016).9 If, as is commonly assumed, taxation equals redistribution, wealthy taxpayers should not expect any benefits from taxation. As a result, there would be no scope for quasi-voluntary compliance, and threat of sanctions would be the only element driving tax contributions. I argue that assuming that taxation equals redistribution is questionable. In fact, I consider that, in developing countries, the scope for wealthy taxpayers to benefit from state action financed by their tax contribution is comparatively large. Presumably, most policies at lower levels of development will not focus on redistribution but rather on aspects such as infrastructure development, security and basic state services. These are all aspects that offer tremendous benefits to wealthy taxpayers.10 Consequently, I claim that the common opposition of wealthy taxpayers to taxation is not primarily the consequence of lack of potential benefits associated with taxes but rather because of the low credibility that many governments and political systems in developing countries have in providing valuable public services in exchange for them. The process leading from revenue collection through to public policy design and implementation to the realization of expected benefits is long and involves many steps (Archer, 1989: 419). Consequently, intertemporal calculations play a crucial role in evaluating the expected utility of paying taxes (Berens & von Schiller, 2016). Based on the arguments mentioned earlier, the major challenges that developing countries face when trying to tax the wealthy becomes evident. The bureaucratic capacity of developing countries tends to be low. Consequently, the threat of sanctions, where credible, will also be low and not being able to force taxpayers into compliance, enhancing ‘quasi-voluntary compliance’ by convincing citizens of the benefits of taxation becomes the main alternative. However, political systems in developing countries also tend to have acute problems of credibility (Lupu & Riedl, 2013; Scartascini et al., 2013). Most of these credibility problems can be conceptualized as principal-agent problems. In particular, I highlight two: the problem of potential opportunistic behaviour by politicians and the problem of the political sustainability of agreements. First, wealthy taxpayers have a legitimate concern about the willingness of current governments to follow through and use higher tax contribution as agreed.11 In principle, the threat to destabilize the government by withdrawing support should be enough to keep the political leaders in line. However, the credibility of this threat is itself limited. Destabilizing a government is costly for the wealthy, and alternative political scenario might even be worse than existing ones. Also, the threat hinges on the ability of the wealthy to successfully coordinate between them successfully. Overall, these problems increase the possibilities for governments to

9 In addition to the requisite that ‘rulers will keep their bargains’, Levi (1989: 53) underlines the requisite that ‘the other constituents will keep theirs’. I consider this element not to be highly relevant for my argument because, in contrast to Levi, I assume that rulers do have the capacity to offer positive selective incentives to taxpayers. 10 Timmons (2005) elaborates on the argument that, in many instances, a well-financed state is a precondition for achieving the provision of goods that taxpayers cannot privately provide efficiently. Supporting the idea that elites might benefit from a better-financed state, surveys on the main obstacles to investment and private sector development consistently show that high tax rates are not the main concern but rather other aspects such as infrastructure, education levels, policy stability and security (World Economic Forum, 2013). 11 Note that it is irrelevant whether the misuse aims at increasing social welfare or increasing private rents through corruption and patronage. The key is that the content of the explicit or implicit fiscal agreement is not respected.


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 281 act opportunistically without fearing to be sanctioned. Anticipating this situation, the incentives for wealthy taxpayers to contribute will be reduced further. Second, wealthy taxpayers will be concerned about the capacity of the political system to sustain agreements over time. As Scartascini et al. (2013: 8) point out, ‘while some countries seem capable of sustaining most policies over time, in others policies are frequently reversed, often in response to minor changes in political winds’. Although this concern applies to all public policies, it is exacerbated in the context of fiscal agreements due to the long-time horizons involved. In addition, it is safe to say that concerns about policy instability are stronger in developing countries than in developed ones.12 I contend that institutionalized party systems can help mitigate the problems of policy sustainability and political opportunism via two mechanisms. First, institutionalized party systems reduce the scope for opportunism by enhancing party discipline and sharpening the content profile of policy options (e.g. Mainwaring, 1998). As party structures become stronger vis-à-vis leaders, party support is necessary for politicians to win elections. Furthermore, in order to be credible to voters in general, and wealthy taxpayers in particular, parties need to preserve their labels. Therefore, they must show that the party structure is able to monitor and sanction leaders that do not honour agreed policies and assure continuity in policy terms (Keefer, 2013). This reduces party leaders’ room for manoeuvre and scope for opportunistic behaviour. Second, institutionalized party systems also enhance centripetal forces in political systems and the capacity to find consensus on core policies. In addition, the durability of party structures increases incentives for both politicians and wealthy taxpayers to invest in constructive relations (Doner & Schneider, 2000). Most importantly, institutionalized political party systems create an environment that mimics repeated interaction, which fosters cooperation among parties (Hallerberg, Scartascini, & Stein, 2009: 296) and between politicians and powerful social groups (Schneider, 2010: 242–244). As a result, in institutionalized political systems, concerns about policy sustainability should be lower, and a change of the governing party is less likely to change significantly the rules of the game. Thus, in countries characterized by institutionalized political party systems, the credibility of the political system to commit to fiscal agreements should be higher at a systemic and individual governmental level.13 In this line, we should expect party system institutionalization to have a positive effect on the amount of taxes that wealthier taxpayers would voluntarily accept to pay. We should however expect the states’ level of bureaucratic capacity to moderate this effect. The lower the level of bureaucratic capacity, the more attractive the option to evade taxes becomes. Consequently, the capacity of a government to credibly commit to fiscal agreements in order to convince the wealthy to contribute becomes even more crucial in such contexts. By contrast, in states where bureaucratic capacity levels are high, the credibility of the political system should not affect tax contributions so significantly. High levels of bureaucratic capacity create sufficiently strong negative incentives for wealthy taxpayers to pay taxes. Assuming that governments will rely predominantly on progressive tax types to extract revenue from wealthy taxpayers, higher tax contributions by the wealthy should be mirrored by a more 12

Of course, the level of uncertainty is a problem of degree. Developed countries also face problems of uncertainty. Yet, the level of uncertainty tends to be vastly greater in developing democracies, and uncertainty is present in different dimensions (Lupu & Riedl, 2013). 13 This argument resonates with the argument made by Gehlbach and Keefer (2011) on ruling-party institutionalization as a mechanism to increase private investment in autocracies. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

282

A. von Schiller

relevant role of progressive taxes in the tax composition. This discussion leads me to the following testable hypothesis: The lower the level of bureaucratic capacity, the stronger the effect of party system institutionalization on the relevance of progressive taxes in the tax composition.

4 EMPIRICAL APPROACH 4.1

Main Variables

To test my hypothesis, I analyse a dataset including more than 90 countries over a period of 20 years (1990–2010).14 The time constraint results from poor data reliability before 1990. This is not only an issue regarding tax collection data, but also concerns other indicators, most prominently ‘Bureaucratic Capacity’. The main dependent variable is ‘relevance of progressive taxes in the tax composition’. This variable is operationalized as PIT as a share of total tax. Making general incidence assumptions about different tax types is a highly contested issue, but there is broad agreement that personal income taxes tend to be progressive.15 Data are provided by Government Revenue Dataset (Prichard et al., 2014).16 I use PIT as a share of total tax collection rather than as a share of gross domestic product (GDP), because my main interest is to analyse the effect on the relevance of progressive taxes in the overall tax effort rather than the performance of a particular tax. There are two interacting independent variables of main interest: party system institutionalization and bureaucratic capacity. As already outlined, my understanding of party system institutionalization emphasizes the stability and continuity of political party systems over time. In line with Mainwaring and Torcal (2006: 206), I understand an institutionalized party system as one ‘in which actors develop expectations and behaviour based on the premise that the fundamental contours and rules of party competition and behaviour will prevail into the foreseeable future’. To capture this dimension, I employ the indicator ‘Party Age’ from the Database of Political Institutions (Cruz, Keefer, & Scartascini, 2016), which codes the average age of the two biggest parties in government and the main party in the opposition.17 Thus, high values for party age denote continuity in party alternatives and repeated interaction in party competition. Compared with alternative measures for party system institutionalization, this indicator is available for a much broader geographical area and offers longer time series.18 14

A list of countries included in the sample can be found in Appendix A. Information on the sources and definition of the main variables as well as summary statistics and a correlation matrix can be found in Appendices B, C and D. 15 The debate on the incidence of different tax types is broad. Since the path-breaking paper by Shah and Whalley (1991), much of the debate has focused on the consumption tax. Most studies, however, support the idea that personal income taxation tends to be fairly progressive, especially if compared with other tax instruments. 16 Data used correspond to the June 2016 version. The Core Dataset (‘Merged’) is used in order to include as many countries as possible. 17 Country-year observations for which the age of one party in the subset was not available were excluded from the sample. The variable is logged for linearity issues. 18 For an overview on different measures of party system stability, see Marinova (2015). Tavits (2012) discusses indicators to measure parties’ organizational strength. She highlights how organizational continuity has also been used as an indicator to measure organizational strength but uses more specific indicators to measure disaggregated dimensions of this concept. As she admits for various, her appealing approach is difficult to implement on a larger cross-national basis. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 283 The second key independent variable is bureaucratic capacity, which I proxy with the indicator ‘Bureaucratic Quality’ from the ICRG Database (PRS Group, 2012). Compared with other alternatives, this source also offers broader geographic and time coverage.

4.2

Control Variables

As outlined in Section 2, many variables affect tax collection and tax composition. Based on the literature, I have included the most prominent variables. To control for economic structure, I include the variable ‘Agriculture, value added as percentage of GDP’ (World Bank, 2012). To capture the development level, I use GDP per capita in constant U.S. dollars (World Bank, 2012). Trade openness might also influence the tax composition by forcing governments to de-emphasize the use of certain taxes. This aspect is captured by the variable ‘Economic Globalization’ (Dreher, 2006). I include the variable ‘Urban Population’ to control for the fact that urban citizens are easier to monitor and tax. I further control for alternative sources of revenue that might disincentivize taxation by constructing a ‘Non-Tax Revenue’ variable, which is operationalized as the difference between tax revenue and overall revenue. Finally, I include the Gini Index to control for the fact that higher levels of inequality might increase the pressure to use more progressive taxes (Solt, 2009).19

4.3

Econometric Approach

Formally, the baseline model for the analysis is Y i;t ¼ β0 þ β1 logðParty AgeÞi;t þ β2 Bureacratic capacityi;t þ β3 logðParty AgeÞi;t Bureacratic capacityi;t þ β4 X i;t þ εi;t where I is the country index and t is the year index; Y is PIT (as share of total tax collection) our main dependant variable of interest. Log(Party Age) and Bureaucratic capacity are the main variables of interest, and Log(Party Age)*Bureaucratic capacity is the interaction term discussed earlier; X is the vector of the controls; and ε is a random error. The main interest of the analysis lies on β1 + β3*Bureaucratic capacity as these coefficients show the effect of Party Age at different level of bureaucratic capacity. Based on this baseline model, and taking into account the panel structure of the data, I estimate a random-effects and fixed-effects specification, the latter including country and year fixed effects. Tax performance and party institutionalization are phenomena that tend to vary little over time within countries. Thus, in comparison with fixed-effects approaches, using a random-effects approach is attractive, as it makes use of the between variation. However, if time-invariant characteristics of countries not controlled for in the model are systematically correlated with the included variables, the estimations will be biased. This is likely to happen when analysing taxation as including all potential relevant controls in the specification (e.g. labour composition, institutional inheritance, informality levels or political culture) is hardly feasible. Fortunately, most of these variables can be 19 The Gini Index capturing market-income inequality (before taxes and transfers) is included. Solt (2009) presents diverse estimations for this Gini Index. I use the average of all available estimations.


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

284

A. von Schiller Table 1.

Main results

Dependant variable: Personal income taxation as share of total tax

Party Age (log) Bureaucratic capacity Party Age (log)*Bureaucratic capacity Gini Index GDP p.c. (log) Trade openness Urban pop. (%) Agriculture (V.A.) Non-tax revenue Constant 2

R N No. of countries

RE model

FE model

1.243 (1.42) 1.058 (1.43) 0.135 (0.52) 0.0145 (0.14) 1.921* (1.87) 0.0475 (1.39) 0.0674 (0.82) 0.0369 (0.51) 0.0269 (0.23) 2.407 (0.26)

1.932* (1.96) 1.599* (1.94) 0.374 (1.33) 0.0150 (0.14) 3.765* (1.80) 0.000963 (0.02) 0.00163 (0.01) 0.0962 (1.21) 0.0975 (0.79) 46.51** (2.21)

0.453 1161 95

0.462 1156 95

t statistics in parentheses. ***p < 0.01; **p < 0.05; *p < 0.1. The RE model includes robust standard errors. Both FE models include clustered standard errors by country. Results for country and year fixed effects not displayed.

expected to vary little over time, which makes it feasible to control for them using fixed effects. Thus, to minimize the risk of omitted variable bias, in a second stage, I estimate a model that includes country fixed effects. Year fixed effects are included to control for temporary effects such as global economic crises.20

5 RESULTS Table 1 presents the results of two model specifications.21 Column 1 corresponds to the GLS random-effects model without country and year fixed effects while Column 2 shows the results of the fixed effects model including year fixed effects. The results for both specifications point in the same direction, and the control variables show the expected signs. Party system institutionalization consistently shows a positive association with the relevance of PIT in the tax composition. Also, in both models, the interaction term is negative.22 Figure 1 illustrates the interaction effect.23 The solid sloping line represents the marginal effect of party system institutionalization. The dashed lines delimit the 90 20

All models include heteroskedasticity-consistent and autocorrelation-consistent standard errors. Nonstationarity does not appear to be a problem. The Lagram-Multiplier test for serial correlation suggests concerns about autocorrelation. Clustered standard error should solve this problem (Hoechle, 2007). 21 To test the robustness of the results, I run the analysis also employing IMF (2012) data. The results can be found in Table A1. Figures presented in the main text can be found in the appendix based on model specification employing IMF data instead of the GRD. 22 The Hausman test and the Wald test for time fixed effect indicate that the random effects specification should be rejected in favour of the fixed effects specification including both time and country fixed effects (Hausman test, p-value = 0.00; Wald test, p-value = 0.00). Consequently, I employ the fixed effects model presented in Table 1 as my core model. Further analysis in this article is consequently based on this model. 23 The construction of these graphs relies on the code provided by Brambor, Clark, and Golder (2006). © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

25 20 15 10

Percent of observations

2 1

|

|

|

|

2

3

4

0

-1

|

5

0

Marginal effect of Party Age (log) on PIT as % of GDP

3

30

4

35

Party System Institutionalization and PIT 285

1

Bureaucratic capacity

Figure 1.

Marginal effect of party system institutionalization on the relevance of personal income taxation (PIT) in tax composition.

per cent confidence interval. The downwards slope suggests that the effect of party system institutionalization is weaker the higher the level of bureaucratic capacity. The fact that the lower dashed line crosses the zero line shortly before the highest level of bureaucratic capacity is reached indicates that the effect is not significant at this level (4). It is also noteworthy, however, that, as the confidence intervals overlap at all levels of bureaucratic capacity, the analysis provides no evidence that the marginal effects of party system institutionalization are significantly different from each other at different levels of bureaucratic capacity. Thus, the marginal effect is not necessarily smaller at higher levels of bureaucratic capacity, but it is statistically significant only at low levels. Overall, Figure 1 supports the hypothesis that bureaucratic capacity moderates the effect of party institutionalization. Given that developing countries tend to be characterized by lower levels of bureaucratic capacity, the results suggest that party system institutionalization has a significant and strong effect mainly in these countries.24 In terms of size of the effect, one standard deviation in party system institutionalization at a level of bureaucratic capacity of zero and one is estimated to lead respectively to an increase of 18.9 per cent and 13.5 per cent in the relevance of PIT in the tax composition. The size of the effect decreases significantly at higher levels of bureaucratic capacity. At a level of bureaucratic capacity of two and three,

24 More details can be found in Table A3. Using IMF data, the corresponding estimated effects are even higher. These calculations are based on the respective fixed effects models.


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

286

A. von Schiller

the size of the estimated effect of one standard deviation is respectively at 8.5 per cent and 3.4 per cent.25,26 Given the construction of my dependent variable, the greater relevance of PIT might not be driven by stronger collection of that tax but by lower collection levels of others. This would contradict my causal chain, as rather than encouraging the performance of the more progressive PIT, higher levels of party system institutionalization would be connected to lower performance of more regressive tax types. Table 2 examines this aspect by running the preferred specification, using the collection data of other tax types (Corporate, General Sales and Trade) as well as total tax collection as dependent variables.27 The results presented in Table 2 provide additional evidence supporting the hypothesis. Party system institutionalization only has a statistically significant effect on the collection of PIT and not on other the collection of other taxes or overall tax collection.28 The results presented in Table 1 also remain robust to a number of additional model specifications. One concern related to the control variables is that the use of Solt’s SWIID dataset implies relying on imputed data, something that might lead to major measurement error (Jenkins, 2015). Against this backdrop and in order to validate the results, I re-estimated the model using real data provided by the WIID dataset (UNU-WIDER, 2017). Employing this source for the Gini Index implies substantial loose of degrees of freedom because of the inexistence of country year observations.29 Even taking this into account, the results using the WIID dataset and 5-year average for all control variables provide further evidence for the hypothesis. The interaction between Party Age and bureaucratic capacity behave very similarly to the one in the main estimation although it is important to note that the effect of Party Age misses statistical significance at the different levels of bureaucratic capacity.30 Furthermore, the inclusion of additional control variables such as ‘Age of the largest government party’, ‘Years in office of the chief executive’, ‘Years in office of the party

25

Please note that in the core model is a level-log model. An increase of 1 per cent in Party Age is associated with an effect of (β1 + β3*Bureaucratic capacity)/100 on PIT collection as share of total tax collection. In the case of a bureaucratic capacity of zero and relying on the fixed effects specification, the results are that an increase in 10 per cent in Party Age is associated to an increase in 0 and 2 points in PIT as share of total tax collection measure. 26 As a robustness test, I specify a log model. Although the interaction term behaves similarly to the ones presented as main specifications (β1 is positive and β3 is negative), the estimated effect of Party Age only reaches statistical significance at a bureacratic capacity of 3. In terms of size of the effect, at a level of bureacratic capacity of 3, a 1 per cent increase in Party Age is estimated to lead to an increase by 0.25 per cent in the PIT as percentage of total tax collection measure. Using IMF data, the estimated effects are stronger and also statistically significant at lower level of bureacratic capacity. Results can be provided upon request. 27 To focus on the raise in absolute tax collection, in this case, the dependent variable is measured as a percentage of GDP and not as a share of total tax. 28 To test the robustness of the results, I run the analysis also employing IMF data (IMF, 2012). The results can be found in Table A2. 29 Note that using the 5-year averages also adresses the concern that results might be driven by shor- term noise produce by mesurement error in particular country year observations. Employing 5-year averages also smoothes potential business cycle effects and reduces measurement error in general. 30 See result in Table A4. Note that a specification employing the SWIID dataset and 5-year averages of all variable leads to similar results as those achieved using the WIID dataset. This increases our confidence in the results of the main specification. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid


0.39 1156 95

0.401** (2.01) 0.195 (0.87) 0.0839 (1.24) 0.00705 (0.42) 0.540 (1.04) 0.0114 (1.02) 0.0220 (0.50) 0.0214 (1.40) 0.0297 (1.06) 8.880 (1.59) 0.10 1148 94

0.00101 (0.43) 0.00178 (0.67) 0.000384 (0.38) 0.0000914 (0.48) 0.0157** (2.41) 0.000113 (0.47) 0.000296 (0.82) 0.0005*** (2.89) 0.00120 (1.42) *** 0.151 (2.64)

CIT

0.24 1340 103

0.199 (0.85) 0.178 (0.78) 0.00169 (0.02) 0.00927 (0.67) 0.557 (0.98) 0.0375*** (4.22) 0.0300 (0.98) 0.00802 (0.51) 0.0189 (0.98) 2.257 (0.41)

0.625 (1.53) 0.754* (1.74) 0.231* (1.74) 0.0140 (0.74) 1.228 (1.46) 0.0689*** (2.98) 0.0693 (1.09) 0.0844** (2.21) 0.143*** (4.60) 12.90* (1.66) 0.26 1388 105

TRD

GS

GRD revenue data

Tax collection of different tax types as percentage of GDP

0.39 1483 106

0.161 (0.32) 0.118 (0.19) 0.0719 (0.35) 0.000290 (0.01) 0.514 (0.54) 0.0402 (1.24) 0.0350 (0.41) 0.0461 (1.07) 0.118** (2.48) 11.93 (1.45)

TOTAL TAX

t statistics in parentheses. *0.1; **0.05; ***0.01. All models include country and year fixed effects. Standard errors are clustered by country. The model specification corresponds to fixed effects model presented in Table 1. Personal income taxation (PIT); Corporate income taxation (CIT); General sales tax (GS); Trade tax (TRD); Total tax collection (TOTAL TAX).


2

Party Age (log) Bureaucratic capacity Party Age (log)*Bur. cap. Gini Index GDP p.c. (log) Trade openness Urban pop. (%) Agriculture (V.A.) Non-tax revenue Constant

PIT

Table 2.


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

288

A. von Schiller

of chief executive’, ‘Regime Durability’, ‘Regime type transition’ and ‘Participation in major violent conflict’ does not change the results remarkably.31 Another source of concern is that the results might be driven by particular observations. To deal with that, I estimate a jackknife model to verify that the results presented in Table 1 are not dominated by observations in one specific country. Also, the fact that there are only very few observations at the lowest level of bureaucratic capacity might be driving the results. To discard this, I re-estimate the analysis, excluding those observations with a value of zero for bureaucratic capacity. Also, the sensitivity of the results to excluding extreme values of Party Age is tested. None of these robustness tests call the results into question.32 Finally, there is the reasonable concern that the estimations might unmask high degree of heterogeneity among subsamples of countries. For the sake of my argument, two subsamples appear to be are particularly relevant: There are good reasons to expect that whether a country is a democracy and whether the government has a clear economic programme are context factors that should moderate the relationship between party system institutionalization and the collection of PIT. From a theoretical perspective, it seems reasonable to expect that the effect of party institutionalization might differ between democracies and autocracies. In autocracies, electoral competition and the right to found political parties is restricted, so that higher levels of party system institutionalization, as operationalized in this paper, will predominantly capture the capacity of the main government party to survive politically.33 As a result, in autocracies, high party system institutionalization indicates a consolidation of the autocratic regime rather than, as conceived in Section 3, the stability and continuity of the political party systems and party competition. However, in both political settings, it is possible to imagine party institutionalization having an effect on tax composition, although through different causal mechanisms. Whereas a positive effect of party system institutionalization in autocracies might be connected to stronger capacity to coercively tax population outside their power base, in democracies the enhanced collection of progressive taxes will be connected to higher levels of credibility of the political actors and higher levels of quasi-voluntary compliance by wealthy taxpayers. To analyse this aspect, I re-estimate the main specifications, including a triple interaction. In addition to the interaction between Party Age and bureaucratic capacity, I

31

These correspond to the variables gov1age, yrsoffice, prtyin in the Database of Political Institutions (Cruz et al. 2016), durable in the Polity IV Dataset (Marshall, Gurr, & Jaggers, 2014), democracy_trans (Boix, Miller, & Rosato, 2013; version 2.0) and a conflict dummy in the country based on sideA and Int (Themnér & Wallensteen, 2012). The rationales behind the inclusion of the different additional control are the following: I include ‘Age of the largest government party’ to control for the possibility that the effect attributed to party system institutionalization might not be related to the age of the parties in the party system but most prominently to the age of the main party in government. Similarly, ‘Years in office of the chief executive’ and ‘Years in office of the party of chief executive’—by capturing the continuity in power of individual politicians or parties— control for the risk that continuity in power, rather than in the system, is key to credibility. The durability of the political regime, the occurrence of major regime transition and the participation in a major violent conflict are included to take into account the possibility that, rather than the institutionalization of parties, the absence of major events in the broader political scene might be the factor driving the results. Estimation results can be provided upon request. 32 Results can be provided upon request. 33 Gehlbach and Keefer (2012) analyse the effect of institutional specificities of autocracies and show their differential effects on private investment. However, taking this into account goes beyond the scope of this paper. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 289 Autocracies 50 40

8

30 20

4

|

2

3

4

0

0


Figure 2.

|

|

0

1

|

|

|

2

3

4

0

1

|

-2

0

|

10

2

20 10

|

-2

|


6

40 30

6 4 2

personal income taxation in the tax composition

0

Marginal effect of partyage (log) on the relevance of

8

50

Democracies


Marginal effect of Party Age at different levels of bureaucratic capacity in democracies versus autocracies

also interact these variables with an electoral democracy dummy.34 Figure 2 illustrates the interactions in the context of each political regime. Figure 2 strongly suggests that the intuition presented and discussed in this paper holds better in electoral democracies.35 Whereas in democracies, the positive effect of party institutionalization decreases as bureaucratic capacity increases, the opposite is true for autocracies. A possible explanation is that taxpayers will only be able to resist coercive taxation of weak autocracies, not of strong ones. Moreover, a strong dictator will have few constraints on extracting as many resources as possible from his or her political opponents. By contrast, democratic governments will restrain themselves from using their full coercive potential to tax political opponents, as it would not only be illegal but also politically costly. Another aspect that could potentially influence the strength of the causal mechanism linking party system institutionalization and reliance on PIT is the nature of the parties involved. Political parties differ greatly in their internal organization and mobilization strategies towards citizens (Keefer, 2013). Programmatic parties in particular need to develop credibility, as per definition their main mobilization instrument—programmatic appeals—involves ‘an intertemporally drawn out exchange in which broad categories of voters receive benefits that often accrue with delay and indirectly’ (Kitschelt & Kselman, 2013: 1455). Consequently, compared with clientelist or machine parties that tend to avoid committing themselves to specific policy goals, for programmatic parties, it is fundamental to avoid being labelled as incoherent and erratic (Keefer, 2013). 34

To identify electoral democracies, I use two variables of the DPI dataset—fraud and Legislative Index of Electoral Competitiveness. I code as democracies countries in which Legislative Index of Electoral Competitiveness value is higher than 6 (‘largest party got less than 75 per cent’), and the fraud variable is zero (meaning no fraud was identified). 35 Regression tables are available upon request. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

A. von Schiller 6

50

Largest government party is not programatic

45 40

5 0

1

|

|

|

2

3

4

|

|

0

1

|

|

|

2

3

4


0

-2

5 0

-2

5

-1

|

10

0

15

1

20

2

25

30

30 25 20 15


3

35

4

40 35

4 3 2 1

personal income taxation in the tax composition

0

| -1

Marginal effect of partyage (log) on the relevance of

5

45

6

50

Largest government party is programatic

10

290


Figure 3. Marginal effect of Party Age at different levels of bureaucratic capacity when the largest government party is programmatic versus when it is not

Non-programmatic parties are institutionally condemned not to be credible, and being part of the political system for a long time should not change this remarkably. In this line, the effect of party system institutionalization on the reliance on PIT should be stronger where programmatic parties are in power. To analyse this claim, similarly to the analysis mentioned above, I estimate a model including a triple interaction between Party Age, bureaucratic capacity and a dummy variable coding whether the largest party in government is programmatic.36 Figure 3 illustrates how the interaction behaves when the largest government party is programmatic or not. The results support the intuition asserting that whether the government has a reasonable clear programme is key for Party Age to have an effect on the relevance of PIT in the tax composition. If the largest government party is programmatic, the effect of party system institutionalization is significant at all levels of bureaucratic capacity below the highest. By contrast, if the largest government party is not programmatic, the effect of party system institutionalization is not significant at any level. 6 CONCLUSION Timmons (2010b: 191) underlines that: “social scientists have expended considerable energy trying to answer who gets what, when and how. Far fewer papers have addressed the flipside of the question—who pays what, when, and how.” 36

Following Keefer (2011), I code a party as programmatic if GOV1RLC from the Database of Political Institutions is equal to 1, 2 or 3 (Cruz et al., 2016)—that is, if the party can be considered to have a specific and clearly recognizable orientation with respect to economic policy. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 291 The present paper represents an effort in this direction. Its most important contribution is to focus on the peculiarities of developing countries and the specific challenges they face when it comes to making wealthy taxpayers pay taxes. In opposition to other approaches, the present paper has not set the strength of the state’s bureaucratic capacities at the core of this challenge; rather, accepting that these capacities are weak, the paper analyses the circumstances under which wealthy taxpayers would accept a higher tax burden in expectation of benefits in exchange. The central argument of this article is that problems associated with low credibility of the political system are crucial to understanding the low relevance of progressive taxes in many developing countries. The results support the argument: Where there are low levels of bureaucratic capacity, party system institutionalization has a strong positive effect on the relevance of PIT in tax composition. The results are robust to different specifications, and the effect is particularly reliable in democratic settings and when government parties are programmatic. Relying on common tax incidence assumptions, this implies that, in developing countries, where levels of bureaucratic capacity tend to be low, wealthy taxpayers assume a bigger share of the tax burden in the presence of credible institutions able to sustain fiscal agreements. In this sense, the analysis has strong policy implications as it highlights how in order to support the development of higher performing and fairer tax systems it is crucial to move the emphasis from merely exploring what governments need in order to tax to questions addressing what taxpayers need in order to accept being taxed. Taxation should not be approached merely as a technical issue, and there is a need to understand how sociopolitical institutions shape the feasibility of tax systems. To make substantial and sustainable improvements, governments need to convince citizens, and especially wealthy ones, that paying taxes is reasonable. Only this will lead to more ambitious fiscal contracts, and to achieve this goal, the ability of core political institutions to credibly commit and sustain fiscal contracts over time is crucial. This paper shows that party systems are one of these pivotal institutions.

REFERENCES Acemoglu D, Robinson JA. 2006. De facto political power and institutional persistence. American Economic Review 96(2): 325–330. Agbeyegbe TD, Stotsky J, WoldeMariam A. 2006. Trade liberalization, exchange rate changes, and tax revenue in Sub-Saharan Africa. Journal of Asian Economics 17(2): 261–284. Aizenman J, Jinjarak Y. 2009. Globalisation and developing countries—a shrinking tax base? Journal of Development Studies 45(5): 653–671. Archer W. 1989. Risk, politics and tax reform: lessons from some latin america experiences. In Tax Reform in Developing Countries, Gillis M (ed). Duke University Press: Durham. Ardanaz M, Scartascini C. 2013. Inequality and personal income taxation: the origins and effects of legislative malapportionment. Comparative Political Studies 46(12): 1636–1663. Basinger SJ, Hallerberg M. 2004. Remodeling the competition for capital: how domestic politics erases the race to the bottom. American Political Science Review 98(2): 261–276. Baunsgaard T, Keen M. 2010. Tax revenue and (or?) trade liberalization. Journal of Public Economics 94(9–10): 563–577. Beramendi P, Rueda D. 2007. Social democracy constrained: indirect taxation in industrialized democracies. British Journal of Political Science 37: 619–641. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

292

A. von Schiller

Berens S, von Schiller A. 2016. Taxing higher incomes: what makes the high-income earners consent to more progressive taxation in Latin America? Political Behavior 39(3): 703–729. Bird RM, Martínez-Vázquez J, Torgler B. 2008. Tax effort in developing countries and high income countries: the impact of corruption, voice and accountability. Economic Analysis and Policy 38(1): 55–71. Boix C, Miller M, Rosato S. 2013. A complete data set of political regimes, 1800–2007. Comparative Political Studies 46(12): 1523–1554. Brambor T, Clark WR, Golder M. 2006. Understanding interaction models: improving empirical analyses. Political Analysis 14(1): 63–82. Cheibub JA. 1998. Political regimes and the extractive capacity of governments: taxation in democracies and dictatorships. World Politics 50(3): 349–376. Chuaire MF, Scartascini C, Tommasi M. 2014. ‘State capacity and the quality of policies’. Working Paper IDB-WP-532. Washington, DC: Inter-American Development Bank. Cruz C, Keefer P, Scartascini C. 2016. Database of political institutions codebook, 2015 update (DPI2015)’. Inter-American Development Bank. Updated version of Thorsten Beck, George Clarke, Alberto Groff, Philip Keefer, and Patrick Walsh, 2001. ‘New tools in comparative political economy: the database of political institutions’. World Bank Economic Review 15(1): 165–176 (September). Cusack TR, Beramendi P. 2006. Taxing work. European Journal of Political Research 45(1): 43–73. Doner RF, Schneider BR. 2000. Business associations and economic development: why some associations contribute more than others. Business & Politics 2: 261–288. Dreher A. 2006. Does globalization affect growth? Evidence from a new index of globalization. Applied Economics 38: 1091–1110. Ezrow L, Tavits M, Homola J. 2013. Voter polarization, strength of partisanship, and support for extremist parties. Comparative Political Studies 47(11): 1558–1583. Fairfield T. 2013. Going where the money is: strategies for taxing economic elites in unequal democracies. World Development 47: 42–57. Fjeldstad O. 2014. Tax and development: donor support to strengthen tax systems in developing countries. Public Administration and Development 34: 182–193. Focanti D, Hallerberg M, Scartascini C. 2013. ‘Tax reforms in Latin America in an era of democracy’. Working Paper IDB-WP-457. Washington, DC: Inter-American Development Bank. Gallagher M, Laver M, Mair P. 2011. Representative Government in Modern Europe. McGraw-Hill: New York. García MM, von Haldenwang C. 2015. Do democracies tax more? Political regime type and taxation. Journal of International Development 28: 485–506. Gehlbach S, Keefer P. 2011. Investment without democracy: ruling-party institutionalization and credible commitment in autocracies. Journal of Comparative Economics 39(2): 123–139. Gehlbach S, Keefer P. 2012. Private investment and the institutionalization of collective action in autocracies: ruling parties and legislatures. Journal of Politics 74: 621–635. Gillis M. 1989. Tax Reform in Developing Countries, 1. Duke University Press: Durham. Gould AC, Baker PJ. 2002. Democracy and taxation. Annual Review of Political Science 5(1): 87–110. Hallerberg M, Basinger S. 1998. Internationalization and changes in tax policy in OECD countries: the importance of domestic veto players. Comparative Political Studies 31(3): 321–352. Hallerberg M, Scartascini C. 2017. Explaining changes in tax burdens in Latin America: do politics trump economics? European Journal of Political Economy 48: 162–179. Hallerberg M, Scartascini C, Stein E. 2009. The budget process as a political arena. In Who Decides the Budget? A Political Economy Analysis of the Budget Process in Latin America, Scartascini C, Stein E (eds). Harvard University Press: Cambridge. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 293 Hart A. 2010. Death of the partisan? Globalization and taxation in South America, 1990–2006. Comparative Political Studies 43(3): 304–328. Hoechle D. 2007. Robust standard errors for panel regressions with cross-sectional dependence. Stata Journal 7(3): 281–312. International Monetary Fund (IMF). 2012. Revenue Data for IMF Member Countries. Washington, DC: International Monetary Fund. Jenkins SP. 2015. World income inequality databases: an assessment of WIID and SWIID. The Journal of Economic Inequality 13(4): 629–671. Kato J. 2003. Regressive Taxation and the Welfare State: Path Dependence and Policy Diffusion. Cambridge University Press: Cambridge, United Kingdom. Keefer P. 2011. ‘Collective action, political parties and pro-development public policy’. Policy Research Working Paper 5676. Washington, DC: World Bank. Keefer P. 2013. ‘Organizing for prosperity: collective action, political parties and the political economy of development’. Policy Research Working Paper 6583. Washington, DC: World Bank. Kenny LW, Winer SL. 2006. Tax systems in the world: an empirical investigation into the importance of tax bases, administration costs, scale and political regime. International Tax and Public Finance 13(2–3): 181–215. Kitschelt H, Kselman DM. 2013. Economic development, democratic experience, and political parties’ linkage strategies. Comparative Political Studies 46(11): 1453–1484. Levi M. 1989. Of Rule and Revenue. University of California Press: Berkeley. Lieberman ES. 2003. Race and Regionalism in the Politics of Taxation in Brazil and South Africa. Cambridge University Press: Cambridge, United Kingdom. Lora E. 2007. Trends and outcomes of tax reform. In The State of State Reform in Latin America, Lora E (ed). Stanford University Press: Stanford. Lupu N, Riedl RB. 2013. Political parties and uncertainty in developing democracies. Comparative Political Studies 46(11): 1339–1365. Mahdavi S. 2008. The level and composition of tax revenue in developing countries: evidence from unbalanced panel data. International Review of Economics and Finance 17(4): 607–617. Mahon JE. 2004. Causes of tax reform in Latin America, 1977–95. Latin American Research Review 39(1): 3–30. Mainwaring S. 1998. Party systems in the third wave. Journal of Democracy 9(3): 67–81. Mainwaring S, Torcal M. 2006. Party System Institutionalization and Party System Theory after the Third Wave of Democratization. Sage: London, United Kingdom. Marinova DM. 2015. A new approach to estimating electoral instability in parties. Political Science Research and Methods 3: 265–280. Marshall MG, Gurr TR, Jaggers K. 2014. Polity IV Project: Political Regime Characteristics and Transitions, 1800–2013. Center for Systemic Peace: Vienna, Virginia. Mkandawire T. 2010. On tax efforts and colonial heritage in Africa. Journal of Development Studies 46(10): 1647–1669. Moore MP. 2008. Taxation and state-building in developing countries: capacity and consent. In Between Coercion and Contract: Competing Narratives on Taxation and Governance, Moore MP (ed). Cambridge University Press: Cambridge, United Kingdom. Mosley P. 2015. Fiscal Composition and Aid Effectiveness: A Political Economy Model. World Development, 69: 106–115. Organisation for Economic Co-operation and Development (OECD), African Development Bank (AfDB) and United Nations Economic Commission for Africa (UNECA). 2010. African Economic Outlook. OECD: Paris, France. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

294

A. von Schiller

Plümper T, Troeger VE, Winner H. 2009. Why is there no race to the bottom in capital taxation? International Studies Quarterly 53(3): 761–786. Prichard W. 2016. Reassessing tax and development research: a new dataset, new findings, and lessons for research. World Development 80: 48–60. Prichard W, Cobham A, Goodall A. 2014. The ICTD government revenue dataset. ICTD working paper 19. Brighton, UK: International Centre for tax and Development, Institute of Development Studies Profeta P, Scabrosetti S. 2010. The Political Economy of Taxation: Lessons from Developing Countries. Edward Elgar Publishing: Cheltenham, United Kingdom. PRS Group. 2012. International Country Risk Guide Researchers Dataset. PRS Group: East Syracuse. Scartascini C, Stein E, Tommasi M. 2013. Political institutions, intertemporal cooperation, and the quality of public policies. Journal of Applied Economics 16(1): 1–32. Schneider BR. 2010. Business politics and policymaking in contemporary Latin America. In How Democracy Works: Political Institutions, Actors, and Arenas in Latin American Policymaking, Scartascini C, Stein EH, Tommasi M (eds). Inter-American Development Bank: Washington, DC. Shah A, Whalley J. 1991. Tax incidence analysis of developing countries: an alternative view. World Bank Economic Review 5(3): 535–552. Solt F. 2009. Standardizing the world income inequality database. Social Science Quarterly 90(2): 231–242. SWIID Version 4.0, September 2013, 90, 231, 242 Stein EH, Caro L. 2013. Ideology and taxation in Latin America. Working Paper IDB-WP-407. Washington, DC: Inter-American Development Bank. Steinmo S, Tolbert CJ. 1998. Do institutions really matter? Taxation in industrialized democracies. Comparative Political Studies 31(2): 165–187. Swank D, Steinmo S. 2002. The new political economy of taxation in advanced capitalist democracies. American Journal of Political Science 46(3): 642–655. Tavits M. 2012. Organizing for success: party organizational strength and electoral performance in postcommunist Europe. Journal of Politics 74: 83–97. Themnér L, Wallensteen P. 2012. Armed conflicts, 1946–2011. Journal of Peace Research 49(4): 565–575. Therkildsen O. 2001. Understanding taxation in poor African countries: a critical review of selected perspectives. Forum for Development Studies 28: 99–123. Thirsk WR. 1997. Tax Reform in Developing Countries. World Bank: Washington, DC. Timmons JF. 2005. The fiscal contract. World Politics 57: 530–567. Timmons JF. 2010a. Taxation and credible commitment: left, right, and partisan turnover. Comparative Politics 42(2): 207–227. Timmons JF. 2010b. Taxation and representation in recent history. Journal of Politics 72: 191–208. United Nations. 2015. Addis Ababa Action Agenda. United Nations: New York. UNU-WIDER. 2017. World Income Inequality Database (WIID3.4), January 2017 von Haldenwang C, von Schiller A. 2016. The politics of taxation: introduction to the special section. The Journal of Development Studies. 52: 1685–1688. World Bank. 2012. World Development Indicators. World Bank: Washington, DC. World Bank and IMF. 2015. From Billions to Trillions: Transforming Development Finance, Post2015 Financing for Development: Multilateral Development Finance. World Bank and International Monetary Fund: Washington, D. C. World Economic Forum. 2013. The Global Competitiveness Report 2013–2014. World Economic Forum: Davos, Switzerland. © 2018 UNU-WIDER. Journal of International Development published by John Wiley & Sons, Ltd.

J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid


APPENDIX A A. LIST OF COUNTRIES INCLUDED IN THE ANALYSIS Albania, Algeria, Angola, Argentina, Armenia, Australia, Austria, Bangladesh, Belarus, Belgium, Botswana, Brazil, Bulgaria, Burkina Faso, Cameroon, Canada, China P.R., Colombia, Republic of Congo, Costa Rica, Cote d’Ivoire, Croatia, Cyprus, Czech Republic, Denmark, Dominican Republic, Egypt, Estonia, Ethiopia, Finland, France, Gabon, Germany, Ghana, Guatemala, Guinea-Bissau, Guyana, Honduras, Hungary, Iceland, India, Indonesia, Ireland, Italy, Jamaica, Japan, Kazakhstan, Kenya, Republic of Korea, Latvia, Lithuania, Madagascar, Malawi, Malaysia, Mali, Malta, Moldova, Mongolia, Morocco, Mozambique, Namibia, Netherlands, New Zealand, Niger, Norway, Pakistan, Panama, Papua New Guinea, Peru, Philippines, Poland, Portugal, Russian Federation, Senegal, Slovak Republic, Slovenia, South Africa, Spain, Sri Lanka, Sweden, Switzerland, Tanzania, Thailand, Togo, Trinidad and Tobago, Tunisia, Turkey, Uganda, Ukraine, United Kingdom, United States, Uruguay, Venezuela, Vietnam and Republic of Yemen

B. MAIN VARIABLES’ DEFINITION AND SOURCES Variable Relevance of progressive taxes in the tax composition. Party Age (log)

Definition

Source

Personal income taxation as a share of total tax collection. ‘tax individs’ (Tax on individuals) divided by ‘taxes ex sc’ (Taxes excluding social contributions).1 Average age of the two biggest parties in government and the main party in the opposition.

Government Revenue Dataset (Prichard et al., 2014)


Bureaucratic quality

Gini Index GDP p.c. (log)

Market income inequality GDP per capita: GDP per capita (const. 2000 US$)

Trade openness

KOF Economic Index: KOF Index on Economic Globalization. It works mainly with two sets of data: data on actual flows and data on restrictions Urban population as percentage of total population

Urban pop. (%)

Agriculture (V.A.)

Value added of agricultural sector as percentage of GDP

Non-tax revenue

Difference between tax revenue and overall revenue collected

Political regime

Countries are coded as being a democracy if the value for Legislative Index of Electoral Competitiveness is higher than 6 (‘largest party got less than 75%’), and the fraud variable is zero (meaning no fraud was

Database of political institutions (Cruz et al., 2016) ICRG database (PRS Group, 2012) Solt (2009) World Development Indicators (World Bank, 2012) KOF index of globalization Dreher (2006) World Development Indicators (World Bank, 2012) World Development Indicators (World Bank, 2012) Government Revenue Dataset (Prichard et al., 2014) Database of political institutions (Cruz et al., 2016) (Continues)


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

296

A. von Schiller (Continued)

Variable

Definition identified). All other countries are coded as being autocracies. Following Keefer (2011), party in government is code as programmatic if GOV1RLC from the Database of Political Institutions is equal to 1, 2 or 3—that is, if the party can be considered to have a specific and clearly recognizable orientation with respect to economic policy

Programmatic party

1

Source

Database of political institutions (Cruz et al., 2016)

To construct the variable using IMF Data for robustness purposes the variables employed are ‘indiv’ and ‘total tax’.

C. SUMMARY STATISTICS

PIT/total tax collection (%) Party Age (log) Bureaucratic cap. Gini Index GDP p.c. (log) Trade openness Urban pop. (%) Agriculture (V.A.) Non-tax revenue

Obs.

Mean

Std. dev

Min

Max

1156 1156 1156 1156 1156 1156 1156 1156 1156

21.77742 3.277124 2.642734 44.05701 8.274285 63.16248 59.93981 11.93575 5.425859

14.41749 .9553974 1.176777 7.271032 1.63868 17.90621 21.42085 12.58088 3.8642

0 0 0 26.3217 4.793603 18.89 9.16 .0363742 .1890841

56.23382 5.062595 4 72.39214 10.64314 97.64 100 62.3827 35.75251

Calculations based exclusively on observations used in the fixed effects model presented in Table 1 (N = 1156).

D. CORRELATION MATRIX Party Age (log) Party Age 1 (log) Bureaucratic 0.4642 cap. Gini Index 0.0476 GDP p.c. 0.4848 (log) Trade 0.3709 openness Urban 0.3717 pop. (%) Agriculture 0.3285 (V.A.) Non-tax 0.1403 revenue


Gini Index

GDP p.c. (log)

Trade openness

Urban pop. (%)

Agriculture Non-tax (V.A.) revenue

1 0.1364 0.8548

1 0.1730

1

0.6774

0.0519

0.7570

1

0.6215

0.1831

0.7988

0.6519

0.0959 0.8623

0.6861

0.7380

0.2871

0.2415

0.7107 0.2954

0.0997

0.3314

1 1 0.3178

1

Calculations based exclusively on observations used in the fixed effects model presented in Table 1 (N = 1156).


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 297 E. TABLE A1: RESULTS MAIN SPECIFICATIONS USING IMF DATA Dependant variable: Personal income taxation as share of total tax IMF revenue data

Party Age (log) Bureaucratic capacity Party Age (log)*Bureaucratic capacity Gini Index GDP p.c. (log) Trade openness Urban pop. (%) Agriculture (V.A.) Non-tax revenue Constant

RE model

FE model

2.230* (1.88) 1.259 (0.95) 0.242 (0.57) 0.0437 (0.48) 3.306*** (2.80) 0.0155 (0.37) 0.0523 (0.73) 0.0625 (0.61) 0.0793 (0.97) 22.57** (2.10)

4.246*** (2.94) 2.636* (1.79) 0.955** (2.00) 0.0725 (0.66) 2.911 (1.12) 0.0721 (1.32) 0.201 (1.15) 0.190 (1.52) 0.0203 (0.27) 15.48 (0.58)

0.569 1251 95

0.048 1244 95

2


t statistics in parentheses. The RE model includes robust standard errors. FE model includes clustered standard errors by country. Results for country and year fixed effects not displayed. ***p < 0.01. **p < 0.05. *p < 0.1.

25 20 15 10


5 4 3 2 0

5

1

Marginal effect of Party Age (log) on PIT as % of GDP

6

30

7

8

35

F. FIGURE A1. MARGINAL EFFECT OF PARTY SYSTEM INSTITUTIONALIZATION ON THE RELEVANCE OF PERSONAL INCOME TAXATION (PIT) IN TAX COMPOSITION (ESTIMATION USING IMF DATA).

|

|

|

|

2

3

4

0

-1

|

1



J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

298

A. von Schiller

G. TABLE A2: TAX COLLECTION OF DIFFERENT TAX TYPES AS PERCENTAGE OF GDP (ESTIMATION USING IMF DATA). IMF revenue data

Party Age (log) Bureaucratic capacity Party Age (log)*Bur. cap. Gini Index GDP p.c. (log) Trade openness Urban pop. (%) Agriculture (V.A.) Non-tax revenue Constant

PIT

CIT

GS

TRD

TOTAL TAX

0.814** (2.43) 0.476 (1.31) 0.197* (1.72) 0.0321 (1.49) 0.402 (0.56) 0.0311** (2.48) 0.00771 (0.21) 0.0304 (1.14) 0.00201 (0.11) 2.772 (0.42)

0.230 (0.53) 0.194 (0.59) 0.0179 (0.15) 0.0115 (0.56) 2.498*** (3.30) 0.00284 (0.11) 0.0356 (0.70) 0.0723* (1.79) 0.00760 (0.26) 17.31** (2.57)

1.799 (0.71) 4.017* (1.68) 0.955 (1.10) 0.0794 (0.68) 6.305 (1.26) 0.190* (1.89) 0.550* (1.97) 0.547** (2.36) 0.357** (2.04) 57.16 (1.25)

0.546 (1.13) 0.328 (0.90) 0.0667 (0.51) 0.00209 (0.14) 0.644 (0.95) 0.0388*** (3.63) 0.0559 (1.24) 0.0272 (0.85) 0.00539 (0.52) 3.687 (0.60)

0.317 (0.45) 0.110 (0.12) 0.0400 (0.16) 0.0888** (2.03) 2.210 (0.96) 0.0559 (1.05) 0.0725 (0.53) 0.0595 (0.46) 0.0738 (0.96) 1.088 (0.05)

0.01 1244 95

0.01 1279 97

0.00 1429 101

0.18 1316 102

0.34 1437 102

2


t statistics in parentheses. *0.1; **0.05; ***0.01. All models include country and year fixed effects. Standard errors are clustered by country. The model specification corresponds to the fixed effects model presented in Table A1. Personal income taxation (PIT); Corporate income taxation (CIT); General sales tax (GS); Trade tax (TRD); Total tax collection (TOTAL TAX).

H. TABLE A3: EFFECT OF ONE STANDARD DEVIATION OF PARTY AGE AT DIFFERENT LEVEL OF BUREAUCRATIC CAPACITY GRD data Level of bureaucratic capacity 0 1 2 3 4

Marginal effect

Average value of PIT as % of total tax

Increase of the share of PIT as % total tax revenue

1.8 1.5 1.1 0.8 0.4

9.7 10.9 13.2 22.3 37.9

18.9 13.5 8.5 3.4 1.1 (Continues)


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 299 (Continued) GRD data Level of bureaucratic capacity IMF data Level of bureaucratic capacity 0 1 2 3 4

Marginal effect Marginal effect 4.0 3.1 2.2 1.3 0.4

Average value of PIT as % of total tax

Increase of the share of PIT as % total tax revenue

Average value of PIT as % of total tax 9.7 9.8 12.8 20.8 37.8

Increase of the share of PIT as % total tax revenue 41.4 31.8 17.2 6.3 1.1

I. TABLE A4: RESULTS OF THE FIXED EFFECTS MODEL USING WIID DATASET FAS SOURCE FOR THE GINI INDEX VARIABLE37 Dependant variable: Personal income taxation as share of total tax GRD data Party Age (log) Bureaucratic capacity Party Age (log)*Bureaucratic capacity Gini Index GDP p.c. (log) Trade openness Urban pop. (%) Agriculture (V.A.) Non-tax revenue Constant

3.969 (1.65) 2.504 (1.57) 0.793 (1.33) 0.00976 (0.16) 2.440 (0.68) 0.0127 (0.22) 0.0614 (0.43) 0.0584 (0.49) 0.318 (1.61) 22.77 (0.84)

2


0.161 302 93

IMF data 4.686 (1.64) 3.544 (1.12) 1.179 (1.23) 0.113 (1.60) 1.524 (0.40) 0.0483 (0.67) 0.0982 (0.61) 0.225 (1.09) 0.325* (1.76) 17.93 (0.56) 0.346 319 89

t statistics in parentheses. * 0.1; ** 0.05; *** 0.01. Both models include country and 5-year period fixed effects (not displayed). Standard errors are clustered by country. GDP, gross domestic product; PIT, personal income taxation; GRD, Government Revenue Dataset. *p < 0.1.


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

300

A. von Schiller

J. FIGURE A2: MARGINAL EFFECT OF PARTY AGE AT DIFFERENT LEVELS OF BUREAUCRATIC CAPACITY ON POLITICAL REGIMES (ESTIMATION USING IMF DATA) 50

10

12

8

40

40

10 8

30 4

20

2 0

20


30

6

6 4 2

1

|

|

2

3

4

|

|

0

1

|

|

|

2

3

4

10

0

|

10

|



0

0

-4

-4

-2

0

| -2

Marginal effect of partyage (log) on the relevance of personal income tax in the tax composition

12

Autocracies 50

Democracies


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid

Party System Institutionalization and PIT 301 K. FIGURE A3: MARGINAL EFFECT OF PARTY AGE AT DIFFERENT LEVELS OF BUREAUCRATIC CAPACITY WHEN THE LARGEST GOVERNMENT PARTY IS PROGRAMMATIC VERSUS WHEN IT IS NOT (ESTIMATION USING IMF DATA) 10

50

Largest government party is not programatic

7

40

8

30

30

25

4

20

2

3

25 20 2

3

4

10

0

|

|

0

1

|

|

|

2

3

4


0

-4

0

-4

-3

-3

5

5

-2

10

|

15

1 1

|

-1

15 0

|


5

35

35

6

40

7 6 5 4 3 2 1 0

|

-1

|

-2

Marginal effect of partyage (log) on the relevance of personal income taxation in the tax composition

8

45

45

9

9

50

10

Largest government party is programatic


SUPPORTING INFORMATION Additional Supporting Information may be found online in the supporting information tab for this article.


J. Int. Dev. 30, 274–301 (2018) DOI: 10.1002/jid