In principle, the middle sectors should be naturally interested in participating in the social contract. According to the median-voter model (see Downs, 1957) if inequality is high before taxes and public expenditure, as it is in Latin America, democracy should lead governments to raise revenue and effect significant redistribution. However, while democracy may be a necessary condition for this, it may not be sufficient even in theory.

Personal preferences towards redistribution stem from numerous sources. Attitudes are affected by individual history, in the form of mobility experiences and perceptions regarding mobility (Piketty, 1995). The organisation of the family matters, as do national and regional cultural and social values (surveyed by Alesina and Giuliano, 2009). Furthermore, the potential beneficiaries of redistributive policies may take into account the effects of taxation on the labour-leisure decisions of their fellow citizens when voting, choosing as a result to limit the size of government and the degree of redistribution (Meltzer and Richards, 1981).

Social beliefs about the degree of fairness in social competition also matter (Alesina and Angeletos, 2005). If a society believes that it is a meritocracy – individual effort determining income – and that everybody has the right and opportunity to enjoy the fruits of individual effort, it will choose low levels of redistribution and taxes. In fact, even the disadvantaged may vote for low levels of redistribution if they think that in the future they or their offspring could progress to the point that they would become net losers under such a policy (Bénabou and Ok, 2001). Societies with high mobility, or more precisely where people think that there is high mobility, may therefore opt for low levels of redistribution. This is the "prospect of upward mobility" (POUM) hypothesis. Conversely, in societies perceived as low-mobility the median-voter model is more likely to hold with a majority voting for more redistribution.10

All of these factors may be temporary though. Hirschman (1973) spoke of a "tunnel effect" of disadvantaged and middle-sector individuals willing to accept and support high (or even increasing) levels of inequality during the early stages of development. He likened this to people staying in the slow lane of a traffic jam in a tunnel, which they will do only as long as they keep their faith in future progress – that at some point their lane will start to move faster. Government credibility, risk aversion and expectations therefore play crucial roles.11

Przeworski (2007) adds an additional and challenging dimension. Even where governments are elected with a mandate to equalise rents and set out to do so, they may fail. Modern redistributive policies mainly aim to equalise human capital by investing in health and education, in contrast to the past’s focus on redistribution of land or productive assets. Such redistribution may not result in an equalisation of outcomes since, as Chapter 3 has shown, the same educational system may produce very different outcomes depending on the socio-economic background of the pupils. In other words, equalisation of opportunities may not be enough. Furthermore, if voters are aware of these weak effects, they will attach low value to publicly provided services and hence have low willingness to fund them.

The data

Among the few rigorous empirical studies in this area, Profeta and Scabrosetti (2008) find that democracy in the region has no significant effect on either the level of taxation or its progressivity. One factor behind this is low institutional capacity, especially in tax administration. Another is the low quality of democracy, which remains vulnerable to populism, as well as "termites" who erode the tax base and "devoradores" who capture social expenditure, using the language of Elizondo and Santiso (2009). To this can be added inefficiencies in the tax and expenditure systems, with both tending to benefit the high-income population disproportionately (see Breceda et al., 2008, and OECD, 2008a). Torgler (2005) highlights the low level of tax morale in Latin America, which ultimately undermines willingness to pay taxes. Finally, Gaviria (2007) argues that the high demand for redistribution and the weak support for market outcomes in Latin America in the late 1990s and early 2000s stem from pessimistic views on social justice, equality of opportunities and mobility.

Empirical research does however highlight the crucial part education plays in fostering support for taxation.

The same study supported the view that people who feel they (or those near to them) have benefitted from social mobility or who are more optimistic about future mobility tend to think that good citizens should pay taxes, and that current levels of taxation are not too high. They also tend to disapprove of tax evasion, although this result is statistically weaker. A similar result holds for belief in meritocracy: the proposition that taxes are too high is rejected by the majority of people who think that success depends on hard work rather than connections, or those who believe that a poor person in their country can become rich by working hard.

Together these results do not support the POUM hypothesis for the region. It seems that risk aversion and the demand for social insurance against downward mobility dominate the POUM effect.

The final piece of the jigsaw is the link between better public services, better institutions, and higher tax morale. Satisfaction with health-care and educational provision reinforce the view that good citizens should pay taxes and, in general, reduce the share of the population that thinks that taxes are too high (the results are weaker for pensions). Similarly, satisfaction with the functioning of democracy increases tax morale, as do lower levels of perceived corruption. On preferences for redistribution – unfortunately – no clear result emerges.12

Latin Americans with higher education (controlling for other socio-economic factors) are less tolerant about tax evasion and are less likely to think taxes are too high. This result highlights a potentially important role for education in fostering social responsibility among citizens.

Reinforcing the social contract

The social contract may be weak, but these results show how it could be reinforced. A catalyst may be improvements in the quality of public services and institutions – including political reforms

These results can be calibrated against the ECosociAL 2007 survey. This found that only a minority of Latin Americans think that the disadvantaged or middle sectors have a good chance to progress – meaning access to university, home ownership, or establishment of a business13 – that foster greater satisfaction with the functioning of democracy. Improvements in those areas should allow for higher levels of taxation in return – the relationship of citizens with their government, after all, is not just one of coercion but also based on trust.14 This virtuous circle may be consolidated by promoting education which has a positive effect on all the social attitudes measured, albeit one that takes time.15 It also found that households in the region were exposed to many of the risks that can break the social contract and undermine social integration, such as crime, labour insecurity, and poor or absent health-care cover. However, at the same time, Latin American citizens have strong beliefs in the value of effort, in the benefits of education, and in the shared responsibility of the state and the individual – backed by a willingness to pay more taxes to finance social insurance. All in all, the results are an indication of a potential basis for a stronger social contract in Latin America, with the middle sectors playing an important role in its consolidation.