Introduction

Debates over tax reform and its impact on state capacity, social cohesion, income inequality, and economic stability have proliferated among academic and political circles across Latin America over the last decades (^{Bird and Zolt, 2013}; ^{Jiménez, 2015}; ^{Lustig et al., 2013}). Taxation is usually considered among the tools that governments can alter to promote development and, by focusing on alternative or competing policies, is simplified to its economic, institutional, or social effects. Understandably then, some of these debates are consigned to technical audiences. But neglecting broader questions on the legitimacy of fiscal arrangements, shared obligations, public trust, and perceived state capacities, risks undermining the functioning of tax systems and their stability over time (^{Bergman, 2009}; ^{Campbell, 2009}; ^{Lewis and Mitchell, 2008}). What most policy disputes assume is that taxation is a core element of democracy, underpinning a healthy relationship between civil society and the state that transcends the information gap between policy maker and citizen (^{Levi, 1989}; ^{Ross, 2004}; ^{Tilly, 2009}). This assumption easily translates into a perspective that it is possible to alter taxation without affecting citizen trust, legitimacy, and cooperation with the fiscal pact. The basic tenet of this link is the assumption that taxpayers know the scale and reach of their obligations: how much tax they pay and what taxes are for. Material and political considerations stir tax awareness, but not only self-interest drives compliance as citizens contribute willingly when they perceive taxes to be legitimate (^{Leroy, 2008}).

In this article, we argue that Chilean discussions over tax reform overlooked individual-level perceptions over taxes. We recast the assumption of well-informed and strategically engaged taxpayers by focusing on individual-level awareness of taxes, i.e. recognition of taxes paid, and how they might relate to taxpayers’ commitment to the state and their fellow taxpayers. We use data from a nation-wide survey conducted in Chile between June and July of 2014, a few months after a protracted debate over the government’s star fiscal reform project had monopolised the public agenda. We assess whether respondents are conscious of their tax obligations, if they can “see” the taxes they are required to pay, and if this visibility is indeed associated, as fiscal sociology contends, to their moral attitudes towards the state and other taxpayers at large. In short, our main objective is to understand the visibility of the tax structure, and how this visibility could affect taxpayers’ attitudes toward taxation. After showing that the Chilean state has formalised, through the tax system, a cooperative and visible relationship only with an elite, we discuss our findings in terms of the potential social outcomes of such a disparate relationship between state and society.

We first place competing explanations of tax awareness within the institutional context of the Chilean tax system. In this way, we place Chile within the Latin American region helping us to frame the discussion of results from a comparative perspective. Then we present the data along with the analytical strategy that organises this study. Next, we analyse the results of a set of models predicting awareness of income and vat taxes. By focusing in these two set of taxes we can study the individual characteristics associated to “seeing” a direct and an indirect tax. After critically describing the gap between tax structure and visibility that we find, we end by discussing how tax awareness might relate to moral commitment and preferences for redistribution. A final section wraps up the results and poses two questions for future research in the region.

Institutional Context and Individual-Level Perceptions

The wider context of the Chilean tax system along with its recent tax reform, is a good setting to problematize how taxes link state and citizen. Among Latin American countries, the Chilean state usually tops comparative tables on efficiency, accountability, the rule of law, and transparency (^{IMF, 2013: 23-24}; ^{Transparency International, 2016}; ^{World Bank, 2014}). This institutional framework resonates in case studies and comparative work on taxation that analyzes the reasons behind lower levels of tax evasion, trustier taxpayers, and a more consensual process of long-term tax reform in Chile than in the rest of Latin America, particularly Argentina (^{Bergman, 2003}, ²⁰⁰⁹; ^{Boylan, 1996}; ^{Sánchez, 2011}). Using data previous to the 2014 tax reform, table 1 reinforces this point by summarizing three features of the Chilean tax system. Excluding social contributions, the Chilean central government raised a comparable level of revenue to that of wealthier societies (column 1). Time spent filling and paying taxes, a measure of transparency and friendliness toward investors, is not necessarily high (column 2); and what is more, the cost of collecting taxes looks favorable in comparative light (column 3). For instance, while France collects around 22 per cent of taxes relative to GDP, Chile collects around 19 per cent. And while a French firm might, on average, spend less hours than a Chilean in declaring, computing and paying taxes, the costs of collection seem to be relatively higher in France than in Chile. These numbers review the state of taxation before the 2014 reform, and are important to highlight the institutional context of the discussion. For our purposes this is more important than the effects of the reform, which are not yet clear (for instance, tax revenue might have fallen in some years such as 2016, while time to prepare taxes and costs of collection have remained stable; see references in table 1). Overall, the Chilean tax system has worked well for political stability, for entrepreneurs, and for institutional development (^{Portes and Smith, 2010}; ^{Wormald and Cárdenas, 2014}). Hence, one could expect a fluid interaction between state and citizen that could inform potential pathways for tax reform in the rest of Latin America.

Yet, what makes Chile a somewhat ideal case is that this institutional picture is not altogether rosy. Research has uncovered a more questionable side to Chilean taxes. It has been known for some time that taxes barely touch the distribution of income (^{Engel, Galetovic, and Raddatz, 1999}), while more recent studies emphasize that the tax system has perhaps been too friendly to investors and the wealthy, spurring economic growth in detriment to social cohesion and economic fairness (^{Atria, 2014}; ^{Fairfield, 2010}; ^{Fairfield and Jorratt, 2014}). Recent cases of tax dodging and dubious connections between large corporations and emblematic political actors have unsettled public trust (^{AdimarkGfk, 2015; CEP, 2015}). In addition to academic concern at the top, new social movements and social demands on the ground betray a few cracks in the erstwhile technical and political consensus (^{Labarca, 2016}; ^{Somma, 2012}).

In this context, Michelle Bachelet, heading a broad center-left coalition, won convincingly the 2013 presidential election at the back of an ambitious legislative agenda. Central among the new proposals was a flagship tax reform bill to increase revenue for education expenditures and social equity. The proposal was sent to Congress for debate in early April 2014 and was approved in September amid continued controversy in the media and growing anxiety among economic actors. Two issues were (and still are) heatedly contested: the increase of the top corporate tax rate and the elimination of tax exemptions to reinvested undistributed profits (^{biobiochile.cl, 2014}; ^{La Tercera Online, 2014}; ^{La Segunda online, 2014}; ^{Libertad y Desarrollo, 2014a}, ^2014b; ^{Beyer, 2014}). Together, the government expected, they would raise revenue, limit evasion, and improve equality. Detractors predicted a fall in investment and economic growth. Despite the far-reaching consequences for economic and political life, bargaining between political and economic actors left little room to question public perceptions and attitudes toward taxation. These issues are all the more important as the current Government, headed by a center-right coalition, is proposing a new tax reform.

The Chilean Tax System and Individual-Level Awareness

Aggregate comparisons could suggest that Chilean taxes inspire a healthy dose of support. As Mary ^{Douglass (1994: 60-66)} noted, institution survival is conditional on the moral commitment of their members. If the Chilean state were efficient in collecting taxes, keeping corruption down, and formalising citizen cooperation, we would anticipate that, through a mix of deterrence and mutual trust, Chilean taxpayers acknowledge and accept their share of the burden, and expect their fellow taxpayers to pay their taxes while counting on an effective state (^{Martin, Mehrotra and Prassad, 2009}; ^{Bergman, 2009}; ^{Steinmo and Bergman, 2018: 284)} . It is because they know this state of affairs and find it legitimate, that they are willing to cooperate with tax officials.

As the consensus on taxation became contested, we might argue instead that taxes became salient not only in political debates but among citizens’ everyday preoccupations. In particular, that people would be more conscious of the type of taxes and the quantities they paid as questions on the legitimacy of the tax system multiplied. To alter the fiscal consensus amounts to questioning the fundamentals of the social contract (^{Martin, Mehrotra, and Prasad, 2009}). Not surprisingly, the incoming governing coalition denounced the inequities of the existing fiscal system and saw tax reform as a first step to mend social cohesion.¹

However, it is also plausible that the Chilean tax system might work comparatively well because of its opacity, and not because of the conscious moral commitment it encourages. As in many developing countries, the Chilean tax system was completely overhauled through reforms in 1980-1990 that sought to stabilise the economy and increase revenue through indirect taxes on consumption (such as the vat) that promote cooperation through hiding tax obligations (^{Bird and Gendron, 2007}; ^{Mahon, 2004}). As indirect taxes were implemented, the fiscal system might have worked efficiently without requiring a fully engaged taxpayer.

Tax awareness hinges on the visibility of tax obligations. This visibility is a function of the structure of taxation, usually defined by whether taxes are direct or indirect. The social effects of indirect and direct taxes have long been studied (^{Campbell, 1993}; ^{Martin and Prasad, 2014}). Direct taxes, for instance on income and wealth, are levied straight on taxpayers who fill a tax return on the amount they earn or have and are assessed accordingly, usually at progressive rates. Hence, these taxes are visible to the taxpayer. Visibility would make citizens aware of their individual efforts and encourage them to hold the state accountable for these contributions. Accountability strengthens democracy in the long-run (^{Ross, 2004}; ^{Tilly, 2009}), but this assumes in turn an informed and active citizen: not only institutional tools are in place for making the state accountable to civil society, but citizens themselves actively use them (^{Bräutigam, 2008}; ^{Martin, Mehrotra and Prasad, 2009}). In contrast, indirect taxes are invisible in the sense that they are hidden in consumption goods, usually levied at a flat rate. Such indirect taxes as the vat result in regressive effects over income inequality. However, indirect consumption taxes are often linked to institutional stability and income redistribution precisely because obligations remain hidden while the state relies on a more robust, diverse, and efficient tax-base (^{Kato, 2003}; ^{Wilensky, 2002}).

Table 2 summarises the structure of the Chilean tax system, and shows the prevalence of indirect consumption taxes. The Chilean pattern follows the structure of taxation of several Latin American countries where invisible taxes on consumption of goods and services are acquiring increasing relevance in comparison to other sources of revenue. In table 2, Chile sits around the middle between two cases, Mexico, where almost 62 per cent of its revenue derives from indirect taxes on consumption, goods and services, and Colombia, where the proportion falls to around 21 per cent of total revenue. The structure of Chilean indirect taxes is heavily influenced by the vat. The Chilean vat is not registered in the invoice so remains largely hidden from the taxpayer, except for more expensive consumer goods (e.g. cars). Given that the debate surrounding Bachelet’s tax reform raised questions on the role of the rich and corporations, vat and income tax rates, among others, we might expect that tax obligations became salient as public opinion was confronted by disagreements over the fiscal pact.

Besides tax structure, another plausible explanation for the opacity of the system is that taxpayers themselves do not need to be aware of their taxes. It is a learned routine partly based on what they expect to get back. Together with indirect taxation, reforms in the decade of 1980 gave control over most of social insurance to private agents. Individuals are conscious of what they pay to privately run insurance firms, instead of expecting the state to provide cover for risk and old age. That is what Chilean (formal) workers see deducted from their monthly pay check. In table 2, Chilean social contributions stand out as a minor component of total tax revenue. As reciprocity (e.g. taxes for insurance) between state and citizen faded, taxpayers spent little time worrying about taxes because direct taxes and (public, not private) contributions were not a visible component of their daily routines (^{Atria, 2014}). Some evidence of this phenomenon is found in studies touching on tax-based redistribution. Recent research in Argentina and an official report on pension reform in Chile confirm that people form their insurance preferences taking into consideration the performance of private pension funds, demanding more public intervention as private insurance loses credibility (^{Carnes and Mares, 2013}; ^{Comisión Asesora Presidencial sobre el Sistema de Pensiones, 2015}). Calls for publicly provided insurance might then bring taxes back to the fore as criticism engulfs private insurance firms.

It would be far too naive to analytically reduce direct taxes to democracy -through tax awareness-, or indirect taxes to welfare -through a wider fiscal base. Answers to which form of taxation is better are not only difficult because options are seldom straightforward; tax systems are messy and conflate both types of taxes, not to mention their possible causal effects. Focusing on Latin America these questions become even more problematic. Research on these two broad social effects has often looked into the experience of wealthier nations, where welfare states increasingly collect revenue through indirect taxes to fund progressive social spending (^{Kato, 2003}; ^{Steinmo, 1993}), but long after having enacted a mass direct income taxes (^{Scheve and Stasavage, 2016}). What is interesting of Chile and Latin America is that reforms in the 1980-1990 were mounted on top of relatively underdeveloped direct tax systems (^{Bird and Gendron, 2007}; ^{Mahon, 2004}), which remained “class-based” as they accrued to higher incomes. Indirect taxation was not introduced to support a growing welfare state but to address a debt crisis amid macroeconomic adjustment.

As social insurance became partly privatised, indirect taxation was enacted to offset the absence of a wide-base, mass, income tax system, and to stabilise the economy (^{Mahon, 2004}). It is difficult to find a link between indirect taxation and the growth of the welfare state in Latin America (^{Haggard and Kaufman, 2008}). Mass taxes in Latin America are levied only on consumption goods. Income taxes remain class-based, levied only on high incomes and have historically protected wage earners (^{Lieberman, 2003}; ^{Marshall, 1939}; ^{Sánchez Román, 2012}). Taxation, in contrast to social spending, rarely determines the outcome of an election campaign.

As fiscal systems cut across different levels of analysis, encompassing tax structure and individual attitudes, we can analytically describe this article’s data through three competing hypotheses: economic, political, and civic explanations of tax salience. In this article, we confirm that tax structure does impact salience. As long as indirect and direct taxation in Latin America correspond to mass and class taxes respectively-the former levied on the whole of society, the latter targeted to the wealthy-taxes should be more salient among the rich and well-off, those who work in formal jobs, and who are required to fill tax returns. After clarifying this point, we want to know what individual-level characteristics are associated with the visibility or invisibility of taxes. We could expect a positive association of structural factors, such as social class, together with partisanship and civic commitment, with recognizing payment of both direct and indirect taxes.

A first hypothesis is as follows. As income and wealth increase, individuals are more likely to fill an income tax return and be subject to a wider variety of direct taxes (regardless of their type). However, more than income, we are interested in the objective economic opportunities that individuals face, i.e. how their income or wealth are structured. Their class position in the labour market will affect their likelihood to “see” a tax, deduct social (public or private) contributions, and hence recognize payment. Being higher up in the class structure should be associated to seeing and paying income taxes (e.g. through formal employment) but not necessarily the vat, which everybody is subject to pay when consuming goods and services.

Secondly, political self-identification could drive the chance to ‘see’ taxes up. Research confirms that individual attitudes towards redistribution relate to demands for more or less taxation. Politically, taxes remain a divisive issue. During the Chilean presidential campaign, centre-left-leaning intellectuals argued against the regressivity of indirect taxation and called for higher tax rates on corporations and the wealthy (^{Jelvez, 2014}; ^{Martner, 2014}; ^{Rivera, 2014}; ^{Vidal, 2011}). They highlighted that the vat fell disproportionately on the poor and that marginal tax rates were not progressive enough. Alternatively, right-leaning intellectuals claimed the tax system worked well and that proposed changes in the fiscal contract were detrimental to the economy. They defended lower corporation taxes to boost growth, while some right-wing politicians called to slash indirect petrol duties (^{El Mercurio Online, 2014}; ^{Larraín, 2014}; ^{Libertad y Desarrollo, 2014a}, ^2014b). Although the reasons diverge, both “extremes” in the political spectrum ought to be sensitive to the taxes they pay as they bear on their policy preferences. Naturally, the mechanism behind visibility should be different for the left and the right. Being to the left ought to increase awareness of the regressive nature of the vat, and hence to see it. But it could also be that people toward the left favour more social expenditures, and might be alert to other forms of direct taxation as well. Perhaps they would be more likely to see taxes as they wish more of them or want to put on record their own contribution. The converse argument would improve awareness of right-leaning taxpayers: as they wish less of them, feel oppressed by taxation, or as they believe in a minimal state and individual enterprise, they might be more conscious of tax payment, either direct or indirect. Despite different mechanisms, we would expect a positive association between awareness of tax payment and identification to both political poles. Partisanship intensifies commitment with public issues, and taxation remains a core issue of modern democracies.

Thirdly, scholars remind us of the links between taxation and state formation in terms of civic commitment. Taxes have historically been one of the main republican obligations that states unconditionally require from their citizens, while income taxation became a tool to generalise sacrifice (^{Scheve and Stasavage, 2016}). Those with a deep sense of republican commitment would be more likely to recognize payment. Individuals with a high degree of support for compulsory civic duties (e.g. voting, military service, paying taxes) should be more likely to see taxes and expect citizen cooperation. This “we are all in it together” attitude condenses the traditional link between civic obligations and state formation. This core assumption of modern institutions has been difficult to establish in Latin American, particularly as the state has been able to fund itself through foreign debt and natural resources (^{Centeno, 2002}). While partisanship disputes taxation as a fair instrument for redistribution and social policy, republican values hold taxation as an end in itself: as a normative bond between citizen and state. Traditional duties toward the state include paying taxes, military service, and compulsory voting; they manifest an upholding of civic obligations that feed into a shared sense of nationhood (^{Lieberman, 2003}). These values were rooted in the school curriculum of many Latin American states, including Chile (^{Serrano, Ponce de León and Rengifo, 2012}). Republicanism expresses permanent features of civic commitment to the state, to hold certain duties as equally mandatory.

These three hypotheses map into economic, political, and civic models of tax awareness. The three are plausible and not incompatible with each other. In the next section, we will analyse what type of individuals ‘see’ their taxes, and how salience relates to tax structure. Once we study how the visibility of taxes is structured, we will examine these results in the context of their alleged consequences, in particular whether recognising tax payment leads to more critical attitudes toward tax evasion and pressure for redistribution.

Latin American countries have offered a fertile ground to study preferences for redistribution, tax rates, and levels of tax morale (^{Torgler, 2005}). In this paper we are concerned with a previous step. Before forming an objective interest in paying more or less taxes, deciding whether to cooperate with tax officials, or supporting the fiscal pact, we simply want to know if individuals ‘see’ the taxes they are meant to pay given the fiscal structure. We explore the gap between tax structure and awareness through individual-level lenses. But while Chile is an excellent case to study tax obligations given the efficiency of its institutions and recent reforms, we should be cautious. Taxation is indeed complex; many levels interact. Awareness of tax obligations distills a rich institutional history which, until recently, sought to exclude the public from its design. Hence, taxes have seldom been salient or a powerful motive for political mobilization. Fiscal protest is extremely rare in the Chilean context, with one exception that we will consider later. As a result, respondents might rightly underestimate what taxes they pay while answering a long survey.

Data and Analytical Strategy

Data are taken from the ^{Encuesta Nacional Bicentenario (2014)}, a yearly survey administered to a representative, randomly selected, sample of the Chilean population since 2006. The target population is comprised of the 18 year-old or older individuals living both in urban and rural areas. We use the 2014 wave, conducted between June and July, which included a battery of questions on taxation amid the coming debate on fiscal reform. Note that the survey was conducted during the heat of the debate, so we could expect added interest over taxation (through the reform the Government expected to increase revenue to fund a flagship educational reform and improve social expenditures).² During the interview, respondents were asked a list of seven types of taxes (income, property, a car permit, vat, a special tax on petrol, alcohol and tobacco, and others) and whether they paid any of them during the last year. We look into two indicators to assess the respondent’s odds of recognising direct and indirect taxes payment: a direct tax on income, and an indirect tax on consumption, the vat. This will allow us to know what individual characteristics condition tax awareness.

The analytic strategy is divided in three parts. In the first we evaluate the visibility of direct and indirect taxes through descriptive information. In the second, we fit payment of income tax and vat to individual-level covariates to measure the impact of economic structure, political partisanship, civic commitment, and individual controls. As the respondent is confronted with a binary choice over income tax and vat payment, the response pattern follows a binomial distribution. Hence, we study the likelihood of paying a tax through a logistic regression as explained in the following equation:

Where indicates having paid the income or the vat tax. Economic structure, political partisanship, and civic commitment represent the main parameters of interest as they capture our three main explanations for tax payment and visibility. Vector X_1i contains control variables and e corresponds to the error term.

Finally, in the third part, we compute the probability of recognising payment of these taxes and evaluate them against questions on tax morale, perceived fairness, and policy preference. In the last section we discuss the probable consequences of ‘seeing’ taxes and suggest questions for further research.

Table 3 summarises the variables used in the analyses. We employ two dependent variables: the respondent’s recognition of having paid income tax and value-added taxes during the last year. Around 23 per cent of the sample claimed to have paid direct income taxes and 36 per cent claimed to have paid the vat during the previous year (2013). Together they provide information on the visibility of two highly contentious and well researched direct and indirect taxes. We confirm that tax structure does impact salience but as fiscal systems cut across different levels of analysis, individual factors interact with the tax structure and contribute to make taxes more or less visible.

We look into the respondent’s economic position through information on occupation and education. Occupation illustrates the respondent’s position in the labour market. We re-codified the original question, which asked a range of seven possible situations from unemployment to “high executive in a big company” into a class scale differentiating between formal (i.e. contract) and informal employment status, and manual and service occupations. Unfortunately, we could not assert the contract status of manual skilled workers, hence the category conflates formal and informal workers. We also re-codified education to the highest level attained by the respondent from less than primary to some or complete university. Missing information from both variables reduced the sample from 2012 to 1 771 individuals.

We tried different measures of political partisanship from an original ten-point scale where the respondent could choose any number from 1, left, to 10, right. We ran different models with the original scale and a variety of codifications. We finally settled for a simple left, non-partisan, and right categorisation where 1-4 were recorded as left, 5 and non-responses as non-partisan, and 6 to 10 as right. Non-responses were re-coded as non-partisan in order to keep a large number of cases that would have otherwise been lost (646). Analyses with or without these cases did not alter the parameters and coefficients of the models.

Finally, we tried two measures to capture civic commitment: whether the respondent considered that military service and voting should be compulsory. Both variables had a similar effect in the models. While conscription best expresses the historical link between shared sacrifice and state building, we settled for compulsory voting as voluntary voting had just been introduced in 2012 and remains a contentious issue.

It might seem odd to explain the acknowledgement of paying or not paying taxes from attitudes (partisanship, civic commitment). In fact, stratification research usually explains attitudes from structural variables, such as class, and seldom employs norms as independent variables. The rationale behind this is that class determines, to some extent, our beliefs and attitudes. Unfortunately, we do not have structural information about partisanship and civic commitment (whether the respondent voted, signed up for military service, or belongs to a political party). We make this move because paying direct and indirect taxes presupposes a different experience which is conditional on the tax structure. In this regard, we want to know if the structural dimensions of economic life correspond to this very important structural dimension of the social contract.

Seeing Taxes

We begin by describing recognition of tax payment. Figure 1 shows that 24.4 per cent recognise income tax payment, which roughly coincides with the 23 per cent of adults that was subject to income tax payment during that year according to the Internal Revenue Service (SII). The SII computes payment over the total individual tax assessment. Calculated over a wider base, for instance considering every adult over 18 years of age, those subjected to income tax payment would only be around 16 per cent of the adult population in 2014 (^{INE, 2016}; ^{SII, 2016}). This might suggest that direct taxes, which are levied on wealthier citizens, most of them in formal jobs, should be perfectly visible for those who are subject to pay them. The figure also shows that consumption taxes are effectively hidden: more than 60 per cent of respondents claim not to have paid them. Crucially, taken together, more than a third of those surveyed could not recognize to have paid at least one type of tax during the previous year.

Figure 1 Acknowledgment of Paying Taxes by Type of Tax 

As we hinted above, one reason for this pattern is history: the Chilean state, along with many Latin American countries, never generalised the income tax (^{Biehl and Labarca 2018}). It remains a class tax to this day. From its inception it sought to protect the consumption capacity of wage-earners (^{Marshall, 1939}). The debt crisis and macroeconomic adjustment of the decades of 1970 and 1980 were addressed by generalising consumption taxes, with different rates of success across the region (^{Bird and Gendron, 2007}). Research suggests it is easier to implant and generalise indirect consumption taxes as they do not require protracted bargaining, political compromise, nor are they technically costly to implement (^{Bräutigam, 2008}; ^{Cetrángolo and Gómez, 2010}; ^{Eaton, 2001}; ^{Moore, 2008}). Hence, they lose salience among public opinion.

Other questions in the survey might point toward a second reason for tax salience. The survey asks which type of tax hurts the most. A specific tax to petrol was the most salient and negatively assessed by the respondents (despite, it could be argued, of its progressive outcomes and positive environmental externalities). Almost 45 per cent of the sample singled it out as the most hurtful. As it disincentives the use of cars, the tax becomes a source of resentment among a growing middle class that seeks to avoid deficient public transport systems. Because it successfully punishes consumption, one of its aims, truck drivers usually rally around this tax, and in 2008 staged a nation-wide strike in the only experience of fiscal protest in recent times (Astudillo and Caro, 2008; ^{El Mercurio Online, 2008}). It is particularly resented outside Santiago, in provinces where public transport is less developed (where over 50 per cent of respondents consider it the most hurtful). There’s a political cost -one that perhaps politicians should continue to bear as it has positive environmental externalities- that makes the tax salient, but it is a feeble ground on which to base sustainable, long-term, political mobilization.

Getting into the factors that could explain individual identification of taxes and the likelihood of recognising payment, tables 4 and 5 show the results of five logistic models that assess the economic, political, and civic explanations of acknowledging income tax and vat payment respectively. Table 4 reports the odds of recognising payment of the income tax. As expected, class and education effects stand out. In all models, those in the richest occupations are more likely to recognize payment of the income tax. Formal unskilled (models 3, 4, and 5) and the service sector workers (models 1 and 5) also tend to see the tax more than informal unskilled workers (who by the nature of their employment are not taxes). However, the effect is only consistently strong and significant in the upper service occupations, and this holds when changing the reference category. This structure remains when we control for education, which reflects the stratification of the tax. Upper service workers are more likely to fill an annual tax return, appeal for tax deductions, and expect annual tax rebates. Workers who approximately earn below usd 900 per month are not subject to fill a tax return. A class income tax produces class visibility.

Along with occupation, we tried different codifications of income as a proxy for class. Although it improved the models’ predictive power, this strategy had important shortcomings. While results confirmed the association between socio-economic position and acknowledgment of paying income taxes, it meant losing more than 300 cases that did not report an individual income. More substantively, the interpretation lost coherence as occupation mediates the opportunities and responsibilities faced by the respondent while income is conditional on several unobservable factors.

Model 2 complements social class with education. The income tax is particularly felt by those with some years of university education, who are always significantly and strongly more likely to see the tax than those with less than primary education. In all models, those with complete primary education show lower odds to see the tax than the least educated but the effect is small. Overall, both upper service and university-educated workers are almost five times more likely to pay the income tax than informal and the least educated workers respectively. Models 4 and 5 introduce political and civic variables. Both explanations barely affect the class models, and have no distinct effect over the visibility of taxes. One alternative is to think that political attitudes are themselves affected by class (^{Chan and Goldthorpe, 2007}) and that respondents do not connect these different obligations. These models confirm a class structure and stratified visibility of the Chilean income tax.

Individual controls add another nuance to class explanations. Age and age squared are included to measure working life dynamics. The probability of acknowledging income tax payment increases with age (as the respondent enters the labour market and, with experience, gets better paid jobs) but only to a certain point, where the effect decreases. Interestingly, women report fewer odds of recognising income tax payment. This is surprising as under Chilean legislation it is the individual (e.g. not a couple) who must fill a tax return (^{McCaffery, 2009}). Again, this could be associated with participation in the labour market as women are less likely than men to work and hold formal jobs (^{Sehnbruch, 2006}), yet these results persist even after controlling for occupation and education. The same gender bias is present in the visibility of the vat, which poses further questions on the association of gender and taxation. Individual controls support labour market accounts of income tax awareness.

Table 5 reports vat visibility. This tax offers a completely different scenario. Everybody pays vat when purchasing goods or services. The tax is salient in public debates on inequality and regularly hits the news given its regressive nature. Asking for vat payment should reveal the public scope of a generalised tax. However, model 1 reveals that visibility of the vat is stratified just as the income tax. Workers in formal manual occupations or in the service sector are more likely to report paying the vat, than informal manual workers. Given that skilled workers can work in informal jobs, it is not surprising that their odds of seeing the vat are not statistically different than those of informal unskilled workers. What is surprising is to find a role for contract that highlights the mediation of labour market dynamics. These effects do not disappear when controlling for education (model 2), where the most educated consistently see the tax. The same effect emerges when running the models with income instead of occupation (which improves the accuracy of the models, but renders the interpretation more difficult). In other words, the visibility of the vat is also stratified: richer and more educated workers who are more likely to pay the income tax are also more likely to acknowledge payment of vat.

Political and civic controls (models 3 and 4) barely affect the stratification of the vat’s visibility. Compared with the income tax, class structure looms over the vat in a more subtle way. Visibility is not polarised between rich and poor, but between those working under formal conditions and those in informal jobs. The mediation of a contract is indeed puzzling, but seems to give substance to labour market explanations.

Compared with the income tax, age loses predictive power. In a way, the stratification of the vat is more closely related to a labour market position than to a working life cycle. Even if the effect is small, women are again less likely to see the tax. If contract mediates tax visibility, then the fact that women are more likely than men to work in informal jobs might be part of the explanation. But as the models control for class and education, the result remains puzzling. Research regularly points out that women carry most of the household chores in Chile, which would make the vat an important consideration when shopping or paying household bills (^{INE, 2009}). Hence, we could have expected women to be equally or more likely to see the vat than men.

Both vat and income tax payment follow class: wealthier and more educated individuals are exposed to a larger quantity of taxes, including direct taxes, and are therefore more likely to distinguish between them and recognise payment. The pattern of responses is not irrational nor does it reflect mere ignorance. On the contrary, it is consistent with the type of tax system that Chile has. However, vat invisibility leaves many questions open: models are less predictive, stratification operates in a more nuanced way, and despite being contentions, its visibility is not driven by political identification.

Consequences of Tax Visibility

Tax policies are highly sophisticated instruments that reveal the many competing motives for growth, redistribution, and public spending present in society. Not surprisingly, debates over tax reform are heatedly contested among experts. In Latin America, inflation and macroeconomic adjustment were behind tax reform in the past, and some might argue it would be irresponsible to ask misinformed citizens with inconsistent preferences to set tax rates or social policy goals. These citizens could end choosing taxes that disincentive long-term investment or expenditures without a stable fiscal base. But as Mahon noted when reflecting on past Latin American reforms, those efforts to broaden the tax base and stabilise the economy have

[…] a more fundamental claim to importance as well: major reforms of tax law and administration, complete with new penalties for evasion, amount to a real reconstitution of a core area of the state, with implications for its relationship with civil society (^{Mahon, 2004: 4}).

In the end, the success of any tax reform is conditional on citizens’ willingness to pay taxes, and questions of tax morale and cooperation are fundamental to sustain any form of taxation in the long run.

This begs the question of whether this fiscal structure affects taxpayers’ attitudes and expectations. For instance, one line of research documents that paying taxes make us more stringent when assessing other people’s tax compliance (^{Bergman, 2009}). A way to explore this is to see if acknowledging payment is associated with expectations over tax evasion. From model 5 in tables 4 and 5, we computed the probability of recognising tax payment and recoded them in low (0-0.3), medium (0.3-0.59), and high (0.6-1), which yield 74, 19, and 7 per cent for low, medium, and high categories respectively. For the vat these are 38, 53, and 9 per cent. According to our results, those who recognise paying the income tax or the vat are more likely to call for a jail sentence for those who evade taxes (Figure 2): 47 per cent of those with low probability but 54 per cent of those with a high probability of seeing the income tax think that evasion should be punished by jail. The percentages for seeing the vat are similar (45 and 53 per cent respectively). Equally, those who recognise paying taxes are more likely to never justify tax evasion (results not shown). Recognising tax payment seems associated to censuring non-compliance.

Figure 2 Attitudes toward Tax Evasion According to Recognizing Tax Payment 

Even if paying taxes might reflect some form of consent -one pays because one finds the system legitimate- recognizing payment in Chile does not necessarily translate into trust in political institutions. Respondents that acknowledge paying taxes are no more likely to trust Congress, Government, or the Judiciary than those who do not recognise paying taxes. What is more, over 65 per cent of those who pay and do not pay taxes hold the view that the state wastes public revenue. Naturally, the reasons of why Chileans mistrust their state run deeper. But perceptions over taxes reveal a fragile bond between state and citizen: while the Chilean state might rank as efficient and capable, it is deeply mistrusted by those who pay taxes and who could, arguably, hold a higher stake on institutional development.

If material conditions mediate tax visibility, we might be able to detect redistribution preferences, which, as we discussed above, could reveal the type of exchange confronting state and citizen. Paying might be conditional on what one expects to get back. Respondents were asked whether the state should focalise expenditures to help those in need or should spend the same on everybody regardless their needs, on a ten point scale from 1 “target help” to 10 ‘spend the same on everyone’. Figure 3 plots the predicted probabilities of acknowledgment of paying income and vat taxes against preferences for focalised or universal redistribution. Recognising payment makes no difference in preferring targeted or universal expenditures.

Figure 3 Preferences for Redistribution According to Probability of Paying Taxes 

Although the previous “no results” are informative, fiscal sociology suggests that overall tax effort might lead to expect increased social spending: it is because we feel that we pay too much that we could demand redistribution (^{Beramendi and Rueda, 2007}; ^{Kato, 2003}). The issue then might be one of degree rather than form, such as the salience of the petrol tax discussed before. We can explore this further by plotting those who claimed that they had paid “much” and “too much tax” over the previous year, close to 40 per cent among those who paid at least one tax, against their preferences for redistribution (Figure 4). Although, not significant, the direction of the association confirms expectations that those who feel to have paid too much tax are more likely to support universal, instead of targeted, social expenditures. That is, they might expect something back as a result of their contribution (and the perceptions they form about those contributions).

Figure 4 Preferences for Redistribution among Those Who Declare to Pay “Much” and “Too Much” Tax 

The question remains open. Figure 4 could also reveal that some people feel that they receive too little for what they expect in return for their taxes. Hence, their expectations are frustrated and would deter them from supporting a broader welfare state. And although a pattern emerges, differences are significant up to a point. The association between opinions should be treated with care. We began the article asking whether taxpayers see their taxes before forming opinions on fiscal policy. Taxes are largely invisible, and those who actually see their taxes display a set of characteristics determined by the labour market and, to some extent, gender. It is for this set of respondents that recognising tax payment seems to define expectations over their fellow taxpayers’ behaviour and over state policy. Further, it is impossible to know if recent discussions over tax reform in the media affected tax salience. If they had a positive effect, then the survey results are likely to be transient and overestimate tax awareness. However, we do not possess data to explore this point given that this set of questions was included only for the 2014 wave of the survey.

Discussion and Future Research

Our results call for further research into the conditions that make certain taxes visible and the mechanisms that link tax behaviour to attitudes towards public institutions and citizenship.³ While this study suggests that tax visibility is strongly associated to class position, we should be careful to interpret causality in strictly material terms. Studies in a different area could shed some light on this issue. For instance, the introduction of voluntary voting in Chile has often been interpreted as the result of class stratification, whereby the rich vote and the poor become marginalised from citizenship. In this article we suggest causation is more complex. Without neglecting the role of social class, those who are subject to paying an explicit and visible tax, the personal income tax, are also more likely to acknowledge payment of a mass, implicit and invisible, vat. The stratification of the income tax and the stratification of vat visibility raise the question of what type of citizen is strategically and consciously engaged with the fiscal pact of the country.

In addition, the article suggests that questions on how to alter taxation to improve equality and raise revenue should be taken together with that of the wider social effects that such policies may produce. These questions involve the relationship between state and citizen, including the quality of democracy. For instance, progressivity of the fiscal system should not be the only consideration when altering the fiscal pact, as the relationship between the different groups in society and the state does not seem to be linear. If the question of progressivity disregards the existence of class-taxes, whereby only the wealthier sections of society are likely to develop an explicit bond with the state, it is probable that those same groups will gain leverage over state action. As Campbell points out, “institutions may create potentially contradictory incentives and opportunities for action” (^{Campbell, 2004: 174}). Hence class taxes may not lead to more egalitarian outcomes if they deepen the association between state and elites, undermining democracy along the way.

This last point is particularly useful to think of future research in taxation within the Latin American context. Questions on what type of fiscal reforms the region needs in order to improve tax revenue, state capabilities, and social equality should not dissociate themselves from questions on possible second-order or unintended consequences in society. The connection is all the more relevant in light of ^{Mahon’s (2004)} study of tax reform in Latin America between 1977 and 1995. His paper finds that a main driver of fiscal reform across the region is macroeconomic adjustment in response to inflation crises. For instance, the introduction of the vat in many Latin American countries was the direct result of a fiscal crisis (^{Bird and Gendron, 2007}; ^{Harberger, 1989}). They were technically efficient and produced revenue at a low collection cost. Partly, this had to do with their invisibility, with introducing them tacitly in people’s daily lives. At the same time, their objective was the consolidation of revenue given a wider goal of macroeconomic adjustment. Consequently, these reforms lacked an explicit social goal that could provide them with legitimacy and support over time. In this article, we have sought to understand how individuals are actually capable of seeing the taxes they are meant to pay, like the vat. These technical changes were spurred by the necessity to sustain revenue levels and keep the state afloat. But the unintended consequence is that they were introduced invisibly into the fiscal pact, probably undermining the social support they needed. Bergman’s comparative research of taxation in Argentina and Chile is cautionary how changes, even if institutional capacities are comparable, could tension the relationship between institutions and society, leading to a loss of trust and cooperation (Bergman, 2003, 2009). Even if the region should pay heed to calls from international organizations and experts to improve taxation, we need country-specific research to inform the likely social outcomes of these reforms, including social cohesion and political stability (^{Boylan, 1996}; ^{Durand and Thorp, 1998}). We are well aware that this call is indeed ambitious and problematic. The issue goes beyond technical considerations and collective action problems (e.g. evasion, cooperation, and free-riding) because it involves the construction of nationhood through the visible inclusion of citizenship in the fiscal pact (^{Lieberman, 2003}).

In recent years, researchers on Latin American taxation have added similar perspectives in order to include more complexity in the analysis of taxation. This research has seen taxation in a wider institutional and social ecology. Recently, ^{Ondetti (2017)} shows how an historical perspective can help us understand and explain the “anomaly” of Mexican business groups’ aversion to state intervention (tax reform among others), leading to a more contested relationship between business groups and the state. In doing this, his article highlights the causal role of a historically constructed ideology and the need to avoid taking groups’ preferences as given. On the other hand, comparative research by ^{Fairfield (2010)} shows the dissimilar use of instrumental and structural power by Chilean and Argentinian business groups in dealing with their respective state’s attempts to improve taxation. In this analysis we see a different relationship between business elites and the state than in Mexico. Finally, a recent volume by ^{Atria, Groll and Valdés (2018)} has put in place a framework to understand taxation in the region from a relational, historical, and transnational perspective in an attempt to understand the complexity of tax policy, their successes and failures, and their prospect for the future.

Answers from the Chilean survey we have analysed, then, leave two broad questions open for future research that apply specifically to Latin America. First, what does this tax structure mean in the context of proposed tax reforms across the region? While most advanced democracies discuss how to reduce taxes and focalise expenditures, boost growth and attract foreign investment, Chile along many Latin American countries is having a discussion on how to increase taxation to fund a wider array of public goods without first discussing how to generalise domestic taxes, together with ways of spurring cooperation and consent. Second, what does this structure mean for the functioning of the state? Despite much debate in political and academic circles, respondents are not altogether conscious of their tax responsibilities but exhibit well-formed opinions on public policy and state use of revenue. Should respondents care of state or institutional development and accountability if they believe they do not pay taxes? While opacity has worked well for the efficiency of the tax system and provided the state with capable institutional tools to collect and administer revenue, the Chilean state seems to be funded largely through taxes but without taxpayers.