Brazil's Tax System in Comparative Perspective

This section provides a synthetic overview of Brazil's tax system in comparison to Latin America as a whole and the three comparative case studies more specifically, examining both the overall tax burden (including subnational government taxes and contributions to public social security programmes) and the sources of tax revenue.

During the most recent five years for which data are available, Brazil's average tax burden of 33.9 per cent of GDP was clearly the heaviest of any Latin American country (see Table 1). Uruguay had the second highest, at 28.9 per cent, followed by Argentina at 28.2 per cent. Chile's tax burden was significantly lower, at 19.2 per cent of GDP. The regional average was 18.5 per cent with Brazil and 17.7 per cent without it.

Table 1. Tax Burdens in Latin America, 2009–2013 (or Most Recent Five Years) (% of GDP)

Sources: CEPALSTAT and CIAT-IDB.

Brazil's status as the most heavily-taxed country in Latin America is not new, but it is not a historical constant either. Table 2 depicts the evolution of the tax burdens of the more developed Latin American countries, a category that includes all the most heavily-taxed countries, since 1950. The figures are decadal averages. In the 1950s Brazil and the Southern Cone countries clearly had the heaviest tax burdens. Figure 1 compares these four countries using annual data. In the early 1950s Brazil's tax burden was substantially lighter than Uruguay's, somewhat lighter than Argentina's and slightly heavier than Chile's. However, by the early 1980s Brazil had risen to the top of the ranking. This outcome reflected both Brazil's own rapid revenue growth in the 1950s and the second half of the 1960s and the stagnation of revenues in the other countries. In Argentina stagnation was evident by the late 1950s, while in Uruguay and Chile it appeared in the 1960s and 1970s, respectively. Since the 1980s Brazil has remained atop the regional ranking, despite significant increases in Argentina and Uruguay, especially in the last decade.

Sources: See Table 2.

Figure 1. Evolution of the Tax Burden in Brazil and the Southern Cone, 1950–2009 (% of GDP)

Table 2. Tax Burdens in Middle-income Latin America, 1950–2009 (% of GDP)

¹ Figure for the 1950s excludes subnational taxes and social security. Figure for the 1960s excludes subnational taxes.

² Figures for the 1950s and 1980s exclude subnational taxes.

³ Pre-1990 figures exclude some municipal taxes.

⁴ Figures includes some non-tax revenues from the state oil company. Pre-1990s figures exclude social security. Figures for the 1950s, 1960s, 1990s and 2000s exclude subnational taxes.

Sources: Data for 1990–2009 are from CEPALSTAT and CIAT-IDB. For earlier decades, the sources are:

Argentina: Oscar Cetrángolo and Juan Carlos Gómez Sabaini, ‘Política tributaria en Argentina. Entre la solvencia y la emergencia’, ECLAC, 2007.

Brazil: Instituto Brasileiro de Geografia e Estatística (IBGE), Estatísticas do Século XX, 2006.

Chile: José Díaz, Rolf Lüder and Gert Wagner, ‘La república en cifras’, Pontificia Universidad Católica de Chile, 2010; and World Bank, ‘Current Economic Position and Prospects of Chile’, 1961, 1966 and 1970; and ‘Chile: An Economic Memorandum’, 1984.

Colombia: William Easterly, ‘The Macroeconomics of the Public Sector Deficit: The Case of Colombia’, World Bank, 1991; and World Bank, ‘Economic Position and Prospects of Colombia’, 1973 and 1979; and ‘Colombia: Economic Development and Policy under Changing Conditions’ 1983.

Costa Rica: World Bank, ‘Current Economic Position and Prospects of Costa Rica’, 1965, 1977 and 1980; ‘Economic Report on Costa Rica’, 1974; Fernando Herrero Acosta, El sistema tributario costarricense (San José: Controlaria General de la República, 2002).

Mexico: Alberto Díaz-Cayeros, Federalism, Fiscal Authority, and Centralization in Latin America (New York: Cambridge University Press, 2006).

Peru: ECLAC; World Bank, ‘Current Economic Position and Prospects of Peru’, 1957 and 1965; ‘An Appraisal of the 1966–1967 Public Investment Program Of Peru’, 1965; ‘Peru: Long-Term Development Issues’, 1979; and ‘Peru: Major Development Policy Issues and Recommendations’, 1981.

Uruguay: Ulises García Reppetto, Universidad de la República, personal communication.

Venezuela: World Bank. ‘Economic Position and Prospects of Venezuela’, 1961; ‘Recent Economic Developments in Venezuela’, 1965; ‘Memorandum on Recent Economic Developments of Venezuela’, 1972; ‘Economic Memorandum on Venezuela’, 1985; and ‘Venezuela: Decentralization and Fiscal Issues’, 1992.

Brazil's contemporary tax structure, or the relative contribution of different broad categories of taxes, is not particularly exceptional (Table 3). As in most other countries in the region, taxes on production and consumption (i.e. indirect taxes) contribute the largest share, followed by more progressive income and property (i.e. direct) taxes and social security contributions. Brazil stands out mainly in terms of the weight of its social security contributions, which is well above the regional average. Among the Southern Cone countries, Argentina and Uruguay have tax structures similar to Brazil's, while Chile differs more substantially in that social security contributes only a very small share of revenues.

Table 3. Latin American Tax Structures, 2009–2013 (or most recent five years) (%)

Sources: CEPALSTAT and CIAT-IDB.

In contrast, Brazil is quite unusual in terms of how the burden is distributed among different levels of government. Subnational governments generate 30 per cent of tax revenues. Argentina, where the subnational contribution is half that amount, is the most similar case.

From a historical perspective, the weight of indirect taxes in Brazil has tended to decline, dropping from an average of 66.7 per centFootnote ⁹ of all tax revenues in the 1950s to 44.2 per cent in 2008–12. Unfortunately, the most comprehensive historical source for Brazilian tax revenues does not separate direct taxes from social security contributions. However, other sources suggest that both categories have grown in importance. For example, direct taxes levied by the central government increased their share of total tax revenues from an average of 13.1 per cent in 1956–64 to 26.2 per cent in 2008–12.Footnote ¹⁰ Social security contributions rose from 13.4 per cent of total tax revenues to 25.4 per cent during the same period.Footnote ¹¹

Data limitations pose major obstacles to a systematic comparison of the historical evolution of Brazil's tax structure to the rest of the region, but the basic pattern of heavy reliance on indirect taxes clearly follows the general regional trend.Footnote ¹² The Southern Cone countries have diverse historical tax structure trajectories, but there is one clear, general difference relative to Brazil. Specifically, social security contributions emerged as major share of taxes earlier in the Southern Cone. By the 1950s they approached or exceeded 30 per cent of total tax revenues in all three countries, a share far greater than that of Brazil.Footnote ¹³ While in Brazil social security continued to increase its share during subsequent decades, in the other countries the trend was reversed. The share of social security in total tax revenues peaked in Argentina in the early 1950s, in Uruguay in the mid-1960s and in Chile in the early 1970s. The change was most dramatic in Chile, where it plummeted from nearly 40 per cent to only 10 per cent by the late 1980s. Brazil's anomalous trajectory in this regard is a key reason why its tax burden ended up surpassing those of the Southern Cone countries.

General Theories of Taxation Level and the Brazilian Case

A wide range of theories purport to explain variation in taxation level. Economic and institutional theories tend to suggest that Brazil should have light or moderate taxation and thus provide little insight into this case. Perhaps the major exception is the scholarship on the political economy of natural resources, which provides a clear answer to the riddle of Brazil's heavy taxation, but one that is obviously incomplete. Certain theories focusing on the role of specific collective actors also illuminate important aspects of the Brazilian case. However, they either contradict or overlook others.

Differences in taxation level have been attributed to a variety of economic variables. Scholars have often found a positive correlation between a country's level of development and its tax burden and have proposed various arguments to explain it. However, this line of research is not relevant here, since there are several Latin American countries with a per capita GDP similar to or higher than Brazil's.Footnote ¹⁴ Economists have also viewed the size of the farm sector as inversely related to the tax burden, since agriculture is hard to tax.Footnote ¹⁵ However, the share of agriculture in Brazil's GDP is almost identical to the regional average, so this variable would seem to provide little leverage.Footnote ¹⁶ Another theory asserts that taxation levels are a function of trade openness because open economies are more vulnerable to external shocks and are thus pressured to develop extensive social safety nets, which require revenues.Footnote ¹⁷ Once again, the theory's prediction contrasts with reality in this case, given that Brazil has one of the least trade-intensive economies in the region.Footnote ¹⁸

Finally, with regard to economic variables, it has been noted that inflation may serve as a substitute for taxation, in that governments print money to cover fiscal deficits and consumers pay the bill in the form of higher prices. In Latin America, use of the ‘inflation tax’ has been common.Footnote ¹⁹ It is clear, moreover, that the ‘Real Plan’ stabilisation programme adopted in 1994 helped boost Brazil's tax revenues by reducing inflation.Footnote ²⁰ However, this variable sheds little light on Brazil's exceptionally heavy tax burden today, since Brazil's inflation over the last decade has been roughly average for the region.Footnote ²¹

There is also a substantial literature that relates differences in taxation and spending to political institutions. Some of it highlights variables on which Brazil does not differ significantly from most other countries in Latin America, such as presidentialism or the electoral system. Brazil does stand out with regard to another institutional variable, federalism. However, scholars have usually maintained that robust federalism leads to low taxation,Footnote ²² and Brazil is the most fiscally decentralised country in Latin America. Stein argues that decentralisation is associated with larger public sectors in Latin America, mainly because subnational governments often generate little revenue of their own and thus become a burden on the national budget. However, this ‘vertical fiscal imbalance’ is in fact relatively modest in the Brazilian case.Footnote ²³ Brazil's subnational governments not only account for a larger proportion of total spending than in most other countries, but they also obtain much more of their revenue from their own taxes, rather than transfers.Footnote ²⁴

Another institutional perspective relates the size of the public sector to the nature of the political regime. There is a venerable tradition of arguing that democracy promotes equality by allowing the lower classes to press for redistribution, which implies a larger state.Footnote ²⁵ Here again, though, the theory provides little insight into this case. Brazil's current democracy emerged at roughly the same time (during the 1980s) as that of several other countries in Latin America, including some with light taxation (e.g. the Dominican Republic) and some with heavier taxation (e.g. Argentina). Historically, moreover, Brazil has not been especially democratic, even by the modest standards of Latin America.Footnote ²⁶

A more nuanced version of this argument affirms that under democracy the level of taxation is positively correlated with income inequality, because relatively poor people naturally desire more redistribution and use their votes to pressure authorities accordingly.Footnote ²⁷ Studies focusing on Latin America have been critical of this notion and some even argue that inequality is inversely related to taxation.Footnote ²⁸ Nonetheless, it is the cornerstone of some influential contemporary studies.Footnote ²⁹

At first glance, this would appear to be a promising explanation of Brazil's heavy taxation, since income inequality studies have consistently portrayed Brazil as among the most unequal countries in the region. However, it faces some obstacles. Perhaps most important, public opinion data do not support the idea that poorer Brazilians are more favourable to redistribution than richer ones.Footnote ³⁰ For example, several surveys conducted since 1997 show that business owners, independent professionals and high executives are all more likely to describe the distribution of income as ‘very unjust’ than agricultural workers or the self-employed and street vendors. A 1996 poll asked whether the government should reduce inequalities between rich and poor. Farmers were the most likely to strongly agree, but high executives and independent professionals were both more likely to take this position than either low-level employees or the self-employed and street vendors.

In addition to the focus on formal institutions like democracy and federalism, there is also a literature that links current taxation levels to largely informal institutions inherited from the past. It explores the origins of state capacity and incorporates taxation as a component of that capacity. A major theme of this literature, based mainly on the European experience, is the positive impact of involvement in external war on the state's ability to tax. Footnote ³¹ However, scholars who have explored this argument in the Latin American context have found it unconvincing.Footnote ³² Moreover, Brazil does not stand out from other countries in the region in the extent or character of its involvement in foreign wars.

Kurtz proposes an alternative historical account tracing variation in state strength to the outcomes of two ‘critical junctures’: the transition to independence and the political rise of non-elites in the early twentieth century. Relatively strong states arose during the post-independence decades when two conditions were present: labour was not legally servile, and regional elites established a working relationship at the national level. The results of the first juncture partly determined that of the second one: where the state was already weak, the emergence of new political actors only weakened it further. Paradoxically, this account would appear to predict a relatively weak state, and thus light taxation, in Brazil, due to the persistence of legal slavery for more than six decades after independence. Unfortunately, however, the author does not explicitly discuss how his theory (which is grounded in case studies of Argentina, Chile, Peru and Uruguay) applies to Brazil.

Informal institutions are also the focus of a body of scholarship that emphasises tax compliance and its determinants, arguing that the ability to tax depends on widely diffused citizen perceptions of the legitimacy and efficacy of the state.Footnote ³³ The hypothesis that springs from this view is that Brazil's tax burden is heavy because its citizens have a positive perception of the state and its ability to enforce the law and thus tend to fulfil their tax obligations. However, the available data do not suggest that Brazilian tax compliance is particularly high. For example, recent Latinobarometro surveys asked respondents to estimate the percentage of their fellow citizens who pay their taxes in full. In Brazil the average estimate was 54.5 per cent, compared to a regional average of 53.1 per cent.Footnote ³⁴ More damaging are data from the World Bank's Enterprise Survey for 2005–6, which included a question that assessed corporate tax compliance by asking business executives to estimate compliance in their sector. Out of nine Latin American countries, Brazil had the second lowest rate.Footnote ³⁵

While the literature discussed above offers little help solving the puzzle examined here, some other approaches do contribute significant pieces. The scholarship on the ‘rentier state’ notes that the existence of plentiful natural resources controlled by the state tends to suppress taxation.Footnote ³⁶ Since taxation is challenging, leaders with access to resource rents tend to rely on them and forego heavier taxation. It is unsurprising from this perspective that the Brazilian state taxes heavily, since its resource rents are lower than those of several other countries in Latin America.Footnote ³⁷ Nevertheless, the importance of this variable should not be exaggerated, since even the total public revenues of major natural resource producers like Bolivia and Venezuela are far lower than those of Brazil.Footnote ³⁸ Thus, the high tax burden in Brazil reflects an exceptionally large public sector, rather than simply a need to compensate for a lack of non-tax revenues.

Works that emphasise the role of specific political actors in shaping tax policy also shed light on certain features of the Brazilian case, but they cannot account for others. A number of scholars have argued that the magnitude of the tax burden is a function of the strength of organisations, particularly unions and left-leaning parties, representing lower-class interests.Footnote ³⁹ Where such organisations are strong, taxes tend to be higher. This perspective parallels the ‘power resources’ school of welfare state research, which sees working and salaried middle-class political organisation as the key determinant of social spending under democracy.Footnote ⁴⁰

Some aspects of Brazil's contemporary reality seem to fit this view. Since 2003 the presidency has been held by the left-of-centre Partido dos Trabalhadores (Workers' Party, PT), which has strong ties to labour. Moreover, Brazil's workforce is currently among the most unionised in Latin America, trailing only that of Argentina.Footnote ⁴¹ Nevertheless, closer inspection reveals limitations to this explanation. In particular, Brazil's tax burden was already the highest in the region even before the PT came to power. In fact, this distinction goes back to the era of conservative military rule (1964–85). Pro-worker organisations, as argued below, have indeed contributed to rising taxation in recent decades by defending a system of heavy social spending. However, Brazil's initial rise to the status of the most heavily-taxed society in the region occurred for reasons other than popular pressure for redistribution.

A number of recent works on Latin America also emphasise the effect of the political cohesion of economic elites on tax policy. Ironically, though, they differ on the crucial question of whether cohesion facilitates or impedes taxation. Comparing attempts to increase direct taxation in Argentina and Chile, Fairfield argues that the greater cohesion of Chilean business explains why such efforts were less successful in that country.Footnote ⁴² Meanwhile, Schneider finds that elite cohesion is correlated with heavier taxation in Central America, since elites who are united enough to impose their own political project on the state are also more willing to provide revenues to fund it.Footnote ⁴³ Lieberman and Flores-Macías reach similar conclusions, the first with regard to differences in direct taxation between South Africa and Brazil and the second in reference to the approval of a dedicated tax in Colombia to fund security forces.Footnote ⁴⁴

At least when compared to other Latin American countries, the Brazilian case would seem to favour Fairfield's position, since it combines heavy taxation with what has sometimes been characterised as a weakly organised business class, especially in terms of the lack of strong encompassing, or multi-sector, associations.Footnote ⁴⁵ Moreover, elite resistance to tax increases in Brazil has traditionally been limited, although it has begun to stiffen in recent years in response to the unprecedented level of taxation.Footnote ⁴⁶

While Fairfield's version of the elite cohesion perspective draws attention to a significant aspect of the Brazilian case, it neglects others. Most clearly, it overlooks the central role played by popular pressures for social spending in driving rising taxation in recent decades. In addition, it obscures the fact that cohesion itself is shaped by the perceived need to act collectively. In Brazil, as Schneider points out, the traditional lack of business unity reflects in part the ‘comparatively hospitable political environment in which Brazilian business flourished’ during much of its history.Footnote ⁴⁷ The analysis developed below stresses that the relative friendliness of the state to elite interests was crucial to the initial construction of strong tax capacity precisely because it avoided a unified elite counter-reaction.

Interpretations of the Brazilian Case

There is no previous work that systematically explores why Brazil's tax burden is heavier than those of other Latin American countries, but a substantial body of scholarship has developed on the Brazilian tax system, particularly in recent decades. It offers some potentially relevant arguments but also leaves crucial questions unanswered.

The studies that most resemble the present one are Melo's comparisons of Brazil and Argentina, which seek to explain the vast gap in tax burdens (about 10 percentage points) between the two countries as of the early 2000s.Footnote ⁴⁸ He argues that this difference stems from Argentina's chronic political instability, which has shortened leaders' time horizons and discouraged investment of political capital in building a strong tax system. Instability is, in turn, largely a product of Argentina's dysfunctional federalism, which promotes conflict between the national government and the provinces. In contrast, a more functional federal system helped the Brazilian state to incrementally raise its tax capacity over the course of decades.

One of the problems with this account is that Argentina's tax burden has not been chronically light but, rather, quite volatile. Even as Melo's texts were being prepared, it was rising to unprecedented levels, substantially closing the gap with Brazil's. Moreover, in the 1950s under Juan Perón Argentina had a considerably heavier tax burden than Brazil (see Figure 1). A focus on chronic institutional weakness cannot explain this trajectory. A second and related problem is that this perspective ignores the crucial role of largely labour-based populism in polarising Argentine society and provoking military intervention and instability. Finally, Melo's focus on Brazil's comparative political stability leads him to underemphasise the change in the causal forces behind rising taxation since the 1980s. The analysis developed below, highlighting the dynamic interplay between the social class bases of public sector growth and the preferences for democracy, provides a better foundation for understanding the trajectory of taxation in both Argentina and Brazil.

While this comparative account of the roots of Brazil's heavy tax burden is unconvincing, in-depth analyses of the Brazilian case developed by Melo and other scholars do furnish important clues. A central theme of this literature is the robust growth of revenues since the early 1990s. While acknowledging the role of the Real Plan, most analyses view this increase mainly as a consequence of the new constitution drafted in 1987–8.Footnote ⁴⁹ The 1988 constitution substantially expanded the federal government's commitments to social spending, especially in the area of retirement pensions. It brought rural workers into the general pension system for private sector workers on a non-contributory basis, established the legal minimum wage as the floor for all public pensions, and upgraded the pension system for public employees at all levels of government. These provisions caused pension spending in both categories to surge.

At the same time, however, the constitution increased the proportion of revenues from the federal income and consumption taxes transferred to states and municipalities, diminishing the resources available at the federal level without a proportional reduction of spending responsibility. Federal authorities responded by creating new ‘contributions’, a legal category of tax earmarked for a particular purpose and not shared with state governments, and increasing the rates of existing contributions. The resulting boom in federal revenues was reflected in a sustained rise in the tax burden.

This account implies that Brazil's heavy tax burden may be attributed to an exceptional case of institutional change. Of course, Brazil is by no means the only country in Latin America that has drafted a new constitution in recent decades. Between 1978 and 2009, 16 new constitutions were approved.Footnote ⁵⁰ However, Brazil's constitution does stand out in terms of the extent to which it deals with specific policies, rather than with institutional structures alone.Footnote ⁵¹ This characteristic has been attributed to the decentralisation of the initial drafting process, which was entrusted to two dozen committees, each of which tried to incorporate proposals submitted by civil society. Decentralisation, in turn, reflected the fact that the constitution was written in the midst of a democratic transition, when society was highly mobilised and grievances from two decades of military rule were still fresh.Footnote ⁵²

Consistent with this perspective is the fact that increased public spending in democratic Brazil has been propelled mainly by social outlays, especially pensions. Between 1990 and 2010 social spending as a share of GDP increased by almost 10 percentage points, an amount equivalent to the entire increase in the tax burden during that period.Footnote ⁵³ Easily the most important driver of that growth was the state pension system for private sector workers. The benefits paid out by this system increased from 2.5 per cent of GDP in 1988 to more than 7 per cent in 2007.Footnote ⁵⁴ Although it eventually levelled off, spending on pensions for federal workers also increased rapidly during the first decade after the ratification of the new constitution. Overall spending on public pension programmes is currently roughly 11 per cent of GDP.

Nevertheless, this account begs two crucial questions. First, why was Brazil's tax burden the highest in Latin America even before 1988? This issue is important because, had the post-constitution tax increase departed from a level close to the regional average, the current burden would be far lower than it actually is. Clearly, the tax reform undertaken in 1966 under military rule, which contributed to a major increase in revenues, played a role. Analyses of that reform suggest that it reflected a determination to overcome a fiscal crisis and to use state intervention to accelerate growth; social considerations were secondary.Footnote ⁵⁵ This is a useful insight, which will be developed later, but this literature does not put Brazil's military era reform in comparative perspective and thus does not shed light on why other countries did not keep pace with Brazil.

The second question is why Brazil's leadership has not done more to attenuate the spending pressures rooted in the constitution, instead of relying on continual increases in taxation. Constitutional amendments passed in 1998 and 2003 did chip away at the benefits awarded under both the general pension system and the special system for civil servants, but they fell well short of major reform.Footnote ⁵⁶ The fact that constitutional change requires larger majorities than regular legislation is not a satisfying answer, since the rules for amending the constitution are not especially demandingFootnote ⁵⁷ and Congress has already amended it more than 80 times. Indeed, constitutional amendment has become almost a routine aspect of governance.Footnote ⁵⁸ More often than not, amendments on fiscal matters have actually contributed to higher taxation by, for example, creating new taxes or allowing contributions to be used as general revenue.Footnote ⁵⁹ Furthermore, the constitution itself mandated a ‘revision’ five years after ratification in order to correct problems that might emerge. During the 1993 revision, amendments could be passed by simple majority, yet Congress failed to pass any amendment reducing spending.

Thus, there are good reasons to doubt that the fiscal provisions of the 1988 constitution, represent a sufficient explanation of the increase in Brazil's tax burden in recent decades, much less of Brazil's long-standing status as the most heavily-taxed country in the region.

Brazil's Emergence as Latin America's Most Heavily-Taxed Country

The next two sections attempt to fill in the gaps in this puzzle by examining how Brazil became the most heavily-taxed country in the region by the early 1980s, gradually surpassing each of the three Southern Cone countries, and how it has maintained that status during the subsequent three decades. The present section focuses on answering the first of these questions, while the next one addresses the second.

The argument offered in both sections stresses the interaction between the social class basis of state expansion and the value attached to democracy among key political actors. The source of support for public sector growth is important because it shapes the character of taxation and spending and thus the distribution of benefits from state action. When pressure comes mainly from popular sectors, the resulting policies are likely to contradict elite interests by prioritising redistribution (at least towards more organised and politically savvy groups) over accumulation. In contrast, when the impetus comes from economic elites, or political elites closely tied to them, policies are likely to favour profits and investment. The strength of pro-democracy norms is significant because it affects the likelihood that economic elites and their allies will respond to adverse situations by engineering a transition away from democracy that helps them impose their own policy preferences.

The argument crafted in this section is that Brazil's tax burden surpassed those of the Southern Cone countries between 1950 and 1980 because the expansion of the state in the former was driven less by popular pressures and more by the economic elite's own support for state-led development. As a result, Brazil's business leaders and other conservatives were less inclined to view the state with hostility and to seek to reduce its role. This characteristic was crucial because, during these decades, the priority attached to preserving democracy in Latin America among domestic political elites and foreign actors was relatively low. The Cold War rivalry and its intensification after the Cuban Revolution meant that other objectives, including anti-communism, took precedence.Footnote ⁶⁰ Economic or political crises could relatively easily lead to a democratic breakdown, which would in turn provide a favourable context for elites to implement their favoured policies.

Thus, although all four countries experienced periods of conservative military rule during this period, in Brazil the military regime sought to deepen state economic intervention, while in the Southern Cone the predominant objective was to roll it back. This difference was reflected in the tax burden, which increased substantially in Brazil but stagnated or declined in the other countries, in part due to changes in social security.

The differences in economic policy between the Brazilian military regime and its counterparts in Argentina, Chile and Uruguay were marked. In all three Southern Cone countries authorities sought to reduce the state's presence through trade, financial, fiscal, labour market and other reforms. Chile's reforms, which followed the 1973 coup d’état against Salvador Allende, were the deepest, extending beyond market liberalisation to the privatisation of many public enterprises and the social security system. The tax burden stagnated during the 1970s and then dropped dramatically in the 1980s, mainly as a result of the pioneering 1981 pension reform, but also due to deep cuts in income taxation. When the armed forces left power in early 1990 Chile's tax burden, at 16.5 per cent of GDP, was less than two-thirds its pre-coup level.

Attempts to roll back what was viewed as an overextended public sector began earlier but proceeded more fitfully in Argentina. The brief (1955–8) military regime established after the overthrow of Juan Perón sought to reverse what its leadership saw as Perón's statist excesses. Argentina's tax burden fell sharply, thanks in part to a decline in social security revenues.Footnote ⁶¹ The cleavage that developed after the coup between statist peronistas and more liberal anti-peronista forces led to marked instability in fiscal policy and the tax burden over the next two decades.Footnote ⁶² In 1976 Argentina's armed forces undertook their fourth coup d’état since 1955, but this time they were more determined to bring about radical change. Although a leftist insurgency played a crucial role in triggering the coup, regime leaders and their civilian allies viewed the statist development model forged under Perón as a crucial cause of Argentina's economic and political woes.Footnote ⁶³ They thus advanced a series of deep liberalising reforms. The 1976–83 regime did not reduce the overall tax burden, presumably because it already dropped to such a low level (12 per cent of GDP) that it was obviously inadequate to the country's needs, but it did lighten the burden specifically on business and the wealthy.Footnote ⁶⁴

In Uruguay efforts to create a leaner state began in the 1960s in response to protracted economic stagnation. The changes were fiercely resisted by beneficiaries of the country's extensive welfare system and protected industrial sector. In the face of rising protest and political violence, in the late 1960s authorities began to impose a series of authoritarian controls. The process culminated in 1973, when the president shuttered Congress. Under military rule authorities took measures to enforce fiscal discipline and also implemented certain structural reforms.Footnote ⁶⁵ The generous social security system underwent changes aimed at containing costs and easing the burden on employers. Trade, finance and the labour market were liberalised. These reforms were milder than in Argentina and, especially, Chile, but they followed the same liberalising, anti-import substitution trend and they managed to halt the expansion of the state. The Uruguayan tax burden in the 1980s was nearly identical to what it had been in the decades before the decline of democracy (see Table 2).

While authoritarian rule brought the stagnation or reversal of tax burden growth in the Southern Cone, it had the opposite effect in Brazil. The initial priority of the new authorities in 1964 was short-term stabilisation, but in 1966 they set in motion an ambitious tax reform whose basic purpose, as discussed earlier, was to endow the state with the resources and incentives needed to stimulate economic growth.Footnote ⁶⁶ The tax burden, which had averaged 17.2 per cent of GDP during the decade prior to the reform jumped to 24.7 per cent in the ten years following it.Footnote ⁶⁷ Although income tax collection increased, on balance the reform was not progressive. After spiking upward during the late 1960s, the tax burden stabilised at about a quarter of GDP during the remainder of the military era. Higher tax revenues were part and parcel of a strongly interventionist policy featuring subsidised credit, infrastructure investments, tax incentives and direct state involvement in production.Footnote ⁶⁸

The difference between the economic policies of the Southern Cone dictatorships, on the one hand, and Brazil's, on the other, can best be understood as a function of the social class dynamics of state expansion in the decades preceding the military seizures of power. In all three Southern Cone countries the growth of the public sector was propelled to a very substantial extent by the demands of non-elites for socio-economic benefits, including pensions, healthcare and jobs. Those who received these benefits were typically not the poorest segments of the population. Rather they belonged to social groups, generally urban, whose salaried positions facilitated organisation into labour unions and professional associations and whose income and political knowledge made them relatively unsusceptible to petty clientelism. They included, for example, factory workers, bank clerks, teachers and civil servants. They exerted their influence both through the threat of labour action and their electoral importance to parties, especially those with a populist or leftist bent, including the Peronista Party in Argentina, the Communist and Socialist parties in Chile and the Colorado Party (especially its left-leaning batllista current) in Uruguay.

The influence of these groups was reflected in the large scale attained by social security. As Mesa-Lago has argued, the expansion of Latin American social security systems owed much to the influence of organised ‘pressure groups’.Footnote ⁶⁹ Although the initial appearance of these systems catered to elite groups, their subsequent growth reflected the rising power of organised non-elite sectors, such as the ones mentioned above. Social security may not have been redistributive on a society-wide basis, due to the exclusion or marginal inclusion of the poor, but it did succeed in redistributing resources from employers to organised workers. As mentioned earlier, in all three Southern Cone countries social security revenues at their pre-coup peak approached or exceeded a third of total tax revenues. In Chile and Uruguay, where democracy broke down later than in Argentina, they reached 10 per cent of GDP.

These systems came to be seen by economic elites and their allies as imposing an excessive burden on businesses and the state. As a result, although the military takeovers of the 1960s and 1970s were justified mainly in terms of fighting communism and armed insurgency, regime supporters took advantage of the authoritarian situation to advance reforms that would impose limits on social spending and erode the structural foundations of the groups, including labour, that defended it. These reforms were costly to some businesses, but economic stagnation and political crisis created conditions under which their resistance could be overcome.

Compared to the Southern Cone, in Brazil the pre-coup expansion of the state was less a product of popular pressures for redistribution, which were weaker. Their relative debility was reflected in the social security system, which was far smaller. In 1955 and 1960 social security accounted on average for only 14 per cent of total public spending. In comparison, spending on transportation and utilities, which reflected the state's commitment to rapid economic growth, made up 18 per cent.Footnote ⁷⁰ Data from the 1960s suggest that Brazilian social security expenditures relative to GDP were roughly a third of those in Chile and Uruguay, which implies a much smaller share of total spending.Footnote ⁷¹ Prior to the 1955 coup, Argentina's social security spending probably also exceeded Brazil's by a large margin.Footnote ⁷²

The relative weakness of popular pressures was also reflected in the lower level of organisation of the workforce. Circa 1960 union members were 10.0 per cent of the economically active population in Brazil, compared to 19.1 per cent in Uruguay, 19.3 per cent in Chile and 31.8 per cent in Argentina.Footnote ⁷³ The Chilean and Argentine labour movements were the strongest in Latin America during most of the twentieth century and had prominent roles in the politics of their respective countries.Footnote ⁷⁴ Uruguay's unions were generally less militant, but competition for worker votes played an important part in the growth of the welfare state and labour conflict escalated when authorities sought to restrain spending and wages beginning in the 1960s.Footnote ⁷⁵

That urban-based pressure groups would be weaker in Brazil is hardly surprising given the country's socio-economic characteristics. Although parts of southern Brazil were relatively industrialised, their influence was counterbalanced by the extreme backwardness of the northern part of the country, which retained a large population. As Table 4 demonstrates, Brazil circa 1960 was substantially poorer, less industrialised, more rural and less educated than the Southern Cone. These features hindered popular political organisation and the development of progressive parties.

Table 4. Key Development Indicators, 1960 v. 1980

Sources: Pablo Astorga, Ame R. Berges and Valpy Fitzgerald, ‘The Standard of Living in Latin America during the Twentieth Century’, Economic History Review, 58: 4 (2005), pp. 788; MOxLAD, op. cit.; Programa Regional del Empleo para América Latina y el Caribe, Mercado de Trabajo en Cifras, 1950–1980, 1982.

Admittedly, Brazil's 1964 coup was in part a reaction to concerns that left-leaning politicians, especially embattled president João Goulart, would use the promise of redistribution to stoke lower-class mobilisation.Footnote ⁷⁶ However, the crisis was essentially conjunctural. Popular sectors had never gained enough influence to substantially shape the character of the state.Footnote ⁷⁷ Brazil had not seen the rise of a powerful leftist or populist coalition of the kind that arose in each of the Southern Cone countries prior to their military coups. There was no real Brazilian analogue to peronismo, battlismo or Allende's Unidad Popular alliance. Getúlio Vargas, the closest thing to a major populist figure, was associated with two parties, but the strongest was the conservative Partido Social Democrático (Social Democratic Party, PSD). Goulart, who represented the more progressive pro-Vargas party, the Partido Trabalhista Brasileiro (Brazilian Labour Party, PTB), became president by accident, when the anti-Vargas Jânio Quadros unexpectedly resigned that office, handing it to his vice president. In sum, despite a brief scare, Brazil's elite never saw its control of the state challenged by popular mobilisation and therefore did not feel compelled to use the military regime to reverse its effects.

Rather, public sector growth both before and during the regime was propelled mainly by a broad consensus, spanning public and private sector elites, on the benefits of an activist state. Although state-expanding initiatives generally originated from the state, they were widely accepted and sometimes even applauded by business. In their study of Brazilian industrialists between the 1930s and the 1970s, Diniz and Boschi conclude that ‘our research did not reveal an industrial elite opposed to state intervention in the economy. In some cases [the elite] even justified that intervention by the necessity of the state filling gaps the private sector could not’.Footnote ⁷⁸ Similarly, in his work on Brazilian interest groups, Schmitter emphasises the lack of conflict occasioned by the sustained expansion of the state's economic role. Elites of all major sectors found they could profit from the interventionist policies that emerged under Vargas and did not seriously resist them.Footnote ⁷⁹ Other works that analyse the business-state relationship in Brazil during this period also characterise it as basically collaborative and mutually supportive.Footnote ⁸⁰

Some explicitly comparative studies have also highlighted the breadth of elite acceptance of state intervention in Brazil. For example, in her book on post-World War II industrial policy in Argentina and Brazil, Sikkink underscores the far greater agreement that existed in the latter country regarding the state-led development model. In contrast to Argentina, where an on-going struggle between liberals and interventionists contributed to a zigzag pattern of policy-making, in Brazil ‘liberalism never took root … the way it did in Argentina. The real debate in Brazil was not between the liberal model and the planning model but within the developmentalist camp between cosmopolitan and national developmentalists'.Footnote ⁸¹ Moreover, the latter debate, centring on the role of foreign investment and relations with the International Monetary Fund, did not reflect a fundamental cleavage either within the state or the broader society.

Perhaps the most significant exception to the rule of private sector acceptance of state intervention during this period was the mid-1970s desestatização campaign in which prominent business figures publicly decried the rapid growth of the public sector under military rule. However, as scholars have noted, this initiative did not enjoy massive support from the business community and reflected frustration with a lack of political access as much or more than it did opposition to intervention per se.Footnote ⁸²

Brazil's Persistence atop the Tax Burden Ranking

Since its emergence as Latin America's most heavily-taxed society in the early 1980s Brazil has maintained that status, but the reasons behind its persistence are not the same as the ones that explain its rise. Popular pressures for redistributive spending became a more important factor in the growth of the Brazilian state during this period, just as they stagnated or declined in the other countries. It is within this context of rising political pressure for social outlays that the origins and effects of the 1988 constitution must be understood. At the same time, the strengthening of democracy across the region made it less likely that the growth of such pressures in Brazil would culminate, as it had earlier in the Southern Cone, in an authoritarian regime capable of checking state expansion.

The liberalising, anti-industry thrust of economic policy among the Southern Cone authoritarian regimes of the 1970s tended to weaken the groups, especially organised labour, whose demands had previously helped propel public sector growth and to cast doubt on the statist policies they had advocated. At least for a time, these effects helped check the growth of the tax burden, even after the return to democracy in the 1980s and early 1990s.

Chile, again, presents the clearest example. The country's labour movement was devastated by repression, deindustrialisation and privatisation. Democracy brought only a partial recovery.Footnote ⁸³ Union density since the 1990 transition has remained below 10 per cent. A nominally centre-left coalition has governed for all but four years, but that coalition came to power through an unwritten pact in which its leaders pledged to maintain the free market model Pinochet's free market model.Footnote ⁸⁴ The governments of the Concertación used targeted social spending to reduce poverty, but the political consensus regarding the economic model (especially among business elites) combined with the weakness of labour, led them to avoid more expensive measures.Footnote ⁸⁵ In a quarter century since the 1990 transition the tax burden as share of GDP has increased by only 2–3 percentage points.

Similarly, the liberalisation policies of Argentina's 1976–83 dictatorship delivered a harsh blow to domestically-oriented industries and associated labour unions.Footnote ⁸⁶ Although the regime's repression and military adventurism were eventually rejected by society, its economic reforms initiated a long-term transition away from import substitution by shifting the balance of power toward more internationalised businesses and weakening the peronista unions.Footnote ⁸⁷ Argentine union density remained the highest in Latin America but never returned to pre-1976 levels. These changes, combined with the advent of hyperinflation in the late 1980s, created the conditions for a new round of liberalisation, including tariff reduction, privatisation of public enterprises and partial privatisation of social security. Ironically, the president behind these reforms, Carlos Menem, was a peronista, albeit one lacking strong ties to labour. Tax revenues initially rose as part of the government's stabilisation efforts, but Menem's economic teams did not seek to push the tax burden beyond about 20 per cent of GDP.

Uruguay is an intermediate case, since the liberalising reforms of the military era were not as profound as in Argentina or Chile. However, even in Uruguay the reform process undercut the protected industrial sector and initiated a gradual but sustained transition away from import substitution. Trade liberalisation continued after the 1984 democratic transition and intensified in the early 1990s.Footnote ⁸⁸ The latter period also brought reforms in other areas, including partial privatisation of social security, a move long resisted by unions and pensioners.Footnote ⁸⁹ Labour organisation experienced a brief revitalisation during the regime transition, but began to fade again in the second half of the 1980s, especially among private firms.Footnote ⁹⁰ By 2000 union density was less than half its mid-1980s level. Uruguayan authorities during the 1990s and early 2000s were generally inclined toward a compact state and did not seek a higher tax burden. Largely as a result, Uruguay's tax burden stagnated at about 22 per cent of GDP.

In recent years all three countries have shown signs of moving in a more statist direction. The trend has taken on its clearest expression in Argentina, where a coalition led by a leftist faction of peronismo has overseen a major increase in spending and taxation. Since 2005 Uruguay has been governed by the Frente Amplio, a centre-left coalition that campaigned for a return to Uruguay's welfarist tradition. In part due to a 2006 tax reform, Uruguay's tax burden has edged upward to around 30 per cent of GDP. In Chile, finally, the current centre-left government has advanced a series of progressive changes including a significant tax increase. Albeit with different characteristics in each country, this trend reflects unease with the ‘neoliberal’ model. However, none of these countries seems likely to reach Brazil's level of taxation in the near future. In Argentina, where the tax burden has surged most rapidly, the current level may be difficult to sustain, in part because of heavy reliance on highly distortive export taxes.

The Brazilian case contrasts sharply with these in that military rule brought socio-economic structural changes that, ironically, ultimately increased popular sector influence. Also, the regime's relative success with state-led development helped to attenuate the domestic appeal of the liberal reform wave of the 1980s and 1990s, abetting the efforts of labour and left parties to preserve a statist system.

As argued earlier, popular forces were weaker in Brazil than in the Southern Cone pre-1964 largely because Brazil was less developed. However, until the early 1980s, when Brazil became mired in the region-wide debt crisis, the regime's consistent policy of state-led industrialisation brought extraordinary growth and socio-economic transformation. Brazil's economic growth rate in the 1970s was three times the Southern Cone average.Footnote ⁹¹ As can be seen in Table 4, between 1960 and 1980 Brazil became substantially wealthier and more industrialised, urban and literate.

The political correlates of these changes included a denser civil society, a more informed public, a larger state apparatus and the gradual decline of rural-based conservative parties that in the past had counterbalanced urban-based forces. One of the striking early manifestations of this transformation was the emergence of a stronger labour movement and a leftist party, the PT, closely associated with it. By the early 1980s union density had more than doubled relative to 1960, exceeding 20 per cent of the work force.Footnote ⁹² In the late 1970s Brazil experienced a series of strike waves that shook the foundations of the military regime and opened a cycle of protest that would eventually encompass many other groups.Footnote ⁹³ Leaders of this movement helped to found it in 1980 and created what became Brazil's major labour confederation, the Central Única dos Trabalhadores (Unified Workers' Central, CUT) in 1983.

Over the next three decades urban-based parties established a clear predominance in Brazilian politics and the PT gradually emerged as the country's most powerful party at the national level. The Partido da Social Democracia Brasileira (Party of Brazilian Social Democracy, PSDB), a centrist party with strong ties to the urban middle class, controlled the presidency from 1995 to 2002. Since 2003 the PT has governed as the leader of a centre-left coalition. In recent years the PT has improved its electoral performance in rural areas, largely because of social policy initiatives, but the party came to power mainly by appealing to urban voters attracted to its message of social equity and clean governance. Although the PT's largest ally, the centrist Partido do Movimento Democrático Brasileiro (Party of the Brazilian Democratic Movement, PMDB) derives some strength from small-town clientelism, the conservative parties most dependent on this type of linkage have lost ground.

In other words, the socio-economic changes of the military era also helped to create a popular sector more closely resembling those of the Southern Cone before the 1970s. At the same time the continuation and comparative success of state-led development under military rule preserved an ideological climate more favourable to the demands of non-elites for state-sponsored social protection. Although Brazil did liberalise its economy during the 1990s, the lingering credibility of state intervention helped mitigate the domestic impact of the wave of liberal reformism that diffused across Latin America. As Brooks has argued, ‘For all its difficulties, Brazil's industrialisation effort provided citizens with street-level knowledge of effective public action and expectations that the state can and should play an important role in domestic allocative functions.’Footnote ⁹⁴

The construction and implementation of the 1988 constitution must be understood in light of this context. Left parties were weakly represented in the constituent assembly, in part because the PT still lacked a national structure. However, the PMDB, which controlled a majority of seats, harboured many progressive politicians with ties to popular civil society.Footnote ⁹⁵ Despite suffering some key setbacks, left-leaning delegates succeeded in inserting numerous provisions in the constitution that implied greater social spending, including the integration of rural workers into the general pension system, the establishment of the minimum wage as the floor for pensions, and a guarantee of free access to healthcare. The reforms to the pension system for public functionaries, meanwhile, reflected the political weight of this largely urban, white-collar group. Public employment grew substantially under military rule and politicians of all major parties viewed public jobs and associated benefits as a patronage resource.Footnote ⁹⁶

The generous spending provisions of the constitution no doubt reflected a propitious conjuncture for demand-making. Yet had they only reflected that context they probably would have been seriously curtailed in subsequent decades, given the relative ease with which the constitution has been altered. The fact that there has not been a setback reflects the strength of the forces supporting reforms. Attempts to make major changes to both public pension systems have been frustrated by a broad coalition of legislators, unions and pensioners' groups.Footnote ⁹⁷ The PT and other left parties have been the staunchest defenders of the status quo, especially under President Fernando Henrique Cardoso (1995–2002) of the PSDB, who attempted a major pension reform but was forced to settle for a minor one. President Luiz Inácio Lula da Silva (2003–10) of the PT secured his own modest reform of the pension system for public employees, but only at the cost of provoking acute intra-party tensions.

That the left would be a defender of the public employee pension system may seem ironic, since that system is widely seen as regressive. However, it makes sense if one considers the weight of this group within the CUT and PT. Although combative unionism of the democratic transition era sprung largely from industry, the sluggish growth of this sector in subsequent decades, combined with the steady expansion of public sector unionism, has increased the importance of public employees. For example, a study of union membership in 2005 and 2011 showed that public employees had the highest unionisation rate of any professional category in both years.Footnote ⁹⁸

At the same time that they have defended the contributory pension systems, left and centre parties have expanded social assistance, especially non-contributory pensions and the conditional cash transfer program Bolsa Família. They have also steadily increased the real value of the minimum wage, which raises the purchasing power of the lowest pensions, both contributory and non-contributory.Footnote ⁹⁹

Mainly because of pensions, social spending increased its share of total public spending from 48.9 per cent in 1990–1 to 73.6 per cent in 2008–9.Footnote ¹⁰⁰ Although social assistance has grown faster than any other social spending category, it remains only a small fraction (roughly 15–17 per cent) of pension outlays. In contrast to social spending, public investment in infrastructure and other areas has tended to decline under democracy, dropping from an average of 3.5 per cent of GDP in 1969–84 to 2.5 per cent in 1985–2007, despite an increase in overall public spending.Footnote ¹⁰¹

The coalition behind these changes, comprised of public and private sector pressure groups allied with largely urban-based left-of-centre parties, is similar to ones that had previously helped to drive public sector growth in the Southern Cone, and the ideological setting of relative confidence in the state's ability to spur development is also similar. What is different in the Brazilian case is that the efforts of this coalition have not faced an authoritarian backlash. The reason lies in the broader political context, and especially the change that has occurred in the strength of pro-democracy norms.

As Mainwaring and Pérez-Liñan have documented, normative commitment to democracy among political elites in Latin America increased markedly beginning in the late 1970s, as did the willingness of powerful foreign actors, especially the US government, to support or at least countenance democracy. These changes reflected growing repugnance for the repression and state-sponsored violence that had occurred under authoritarian rule, as well as the decline and ultimate collapse of the Soviet Union, which lowered the perceived costs of tolerating leftist political influence. Foreign and domestic transformations interacted to give Latin American democracies greater resilience. In the past, political leaders might have readily seized upon an economic crisis or episode of political conflict to suspend democracy and impose their own policy preferences, but the growing value attached to democratic politics per se served to ‘inoculate competitive regimes from breakdown’.Footnote ¹⁰² In other words, democracy has been transformed into something approaching ‘the only game in town’.Footnote ¹⁰³

A number of situations have arisen in Brazil during the last three decades that would plausibly have invited military intervention had they occurred earlier in the country's history, including the constituent assembly's approval of a markedly left-leaning first draft of the new constitution in 1987, Lula's near-victory in the presidential election of 1989, and his actual victory in 2002. Each of these episodes sowed panic among business leaders and their conservative political allies because of the perceived threat of redistributive, anti-business change. The fact that democracy was not suspended meant that the constitution preserved at least some of the leftist (and pro-social spending) features of the initial draft, that the PT was able to continue expanding as an electoral force and exercise influence through the legislature, and that it was ultimately able to capture the presidency and use that position to further increase revenues and spending.