The Great Leveler, page 50
Although a widespread move toward somewhat greater income equality is commonly invoked in the literature, population-weighted Gini coefficients speak a different language, especially if we focus on net outcomes over longer periods of time. Among six countries for which we have data going back to 1938, inequality increased in all but one between that year and 1970, and the population-weighted overall income Gini coefficient consequently went from 0.464 to 0.548. In a larger sample of fifteen countries, income inequality rose in thirteen of them between 1950 and 1970, and more moderately from 0.506 to 0.535 overall—a very high level by international standards. Notably, in two of the three countries that experienced net reductions in inequality, these improvements were effectively confined to the 1950s: in Argentina, they coincided with Juan Perón’s aggressively statist and redistributive government, and in Guatemala, they occurred during and after a bloody civil war. Venezuela is thus the main candidate for peaceful leveling through economic development, possibly joined by Chile if we accept an alternative set of inequality estimates that suggest leveling between 1930 and 1970, which would have been driven by economic and (peaceful) political change.19
Public borrowing to sustain protectionist policies and nationalized industries in the 1970s triggered debt crises in the 1980s, which came to be known as a “lost decade” during which economic growth stalled and poverty expanded. This, in turn, spurred on economic liberalization that opened up the region’s economies and furthered their integration into global markets. Inequality outcomes varied considerably among different countries, whereas in the 1980s and 1990s, the region as a whole witnessed moderate increases in the population-weighted income Gini coefficient by a little less than 2 points per decade and saw a peak around 2002.20
What all this shows is that Latin American income inequality increased under a wide variety of economic conditions: export-led growth, state-led industrialization and protectionism, economic stagnation, and liberalization. In the four countries having the longest time series, population-weighted income Gini coefficients climbed from 0.348 in 1870 all the way to 0.552 in 1990; for six countries, from 0.377 in 1913 to 0.548 in 1990; and for fifteen, from 0.506 in 1950 to 0.537 in 1990. Although this conceals local variation and flattens temporary swings, and although the precise values often remain unknown, the long-term trend could not be clearer. Inasmuch as progress is to be found, it was limited to a mere deceleration in the rise of inequality in the second half of the twentieth century. As we can see in Fig. 13.2, occasional leveling was short-lived and limited to periods of economic downturn, triggered by foreign macroeconomic crises first in Britain and then the United States in the 1900s and 1930s and, finally, by deep recession arising from both domestic and international factors in the 1980s.21
The most recent phase in the evolution of Latin American income inequality began soon after 2000. Perhaps for the first time in recorded history, inequality fell across the region. In fourteen out of seventeen countries that have produced relevant data series, income Gini coefficients in 2010 were lower than they had been in 2000. Costa Rica, Honduras, and probably Guatemala are the only documented exceptions. For the other fourteen countries, the average Gini of market income fell from 0.51 to 0.457 and the average Gini of disposable income from 0.49 to 0.439, or more than 5 points by either measure. This compression is certainly impressive in terms of both scale and geographical scope but needs to be put in proper perspective. It lowered inequality for market income from a level typical of India, a highly unequal society, to one closer to that of the United States, whereas the fall in net inequality took Latin America from Chinese and Indian heights to a Gini still 7 points above that for the United States, the undisputed inequality champion among Western countries. The effect of these changes on the exceptionally skewed distribution of Latin American incomes should thus not be overrated.22
Figure 13.2Estimated and conjectured income Gini coefficients for Latin America, 1870–1990 (population-weighted averages for four, six, and sixteen countries)
To make matters worse, since 2010 this downward trend has continued in fewer than half of the countries for which we have data (in Argentina, Bolivia, the Dominican Republic, Ecuador, El Salvador, Uruguay, and Venezuela). In those years, inequality has remained fairly stable in Brazil, Chile, Guatemala, Panama, and Peru and has begun to climb again in Mexico and Paraguay and possibly also in Honduras, where the evidence is poor. Costa Rica had always bucked the regional trend with gently rising inequality ever since the 1980s. All this raises serious questions about the causes and sustainability of the leveling that occurred in the first decade of this century: could it have been a short-lived improvement that has run its course?
It is impossible to explain this leveling as the result of Kuznetian downward pressure on inequality once countries in the region had passed some sort of inflection point in development at which economies had become rich enough for incomes to be more equitably distributed. In 2000, per capita GDP in the fourteen countries with declining inequality varied by a factor of 7.6 between the richest and the poorest one (Argentina and Bolivia, respectively). Dispersion across this wide range was quite regular even if biased toward the lower end: mean annual per capita GDP fell between $1,000 and $2,000 in five countries, between $2,000 and $4,000 in another five, and between $5,000 and $8,000 in the other four. This alone rules out the possibility that the synchronized leveling observed in the following decade was connected to levels of economic development per se. Formal testing has confirmed that notwithstanding strong economic growth in those years, the Kuznets model cannot account for most of the observed decline.23
Recent studies have identified several reasons for this process: falling skill premiums and strong foreign demand that compressed market income inequality by reducing sectoral earnings gaps, recovery from earlier unequalizing macroeconomic crises that had exacerbated poverty, strong labor markets driven by more rapid economic growth, and the redistributive effect of certain government transfers on disposable income inequality. At least in theory, the first of these factors holds particular promise as a potential peaceful driver of equalization in the longer term. Market reforms in the 1990s tended to be accompanied by an expansion of the educational system, an expansion that has since continued and increased the supply of skilled workers, which in turn lowered returns on higher-level schooling and skill premiums and thus overall labor income inequality. There is no single answer to the question whether the reduction in skill premiums owed more to improved supply or diminishing demand. In some countries, premiums shrank in response to weaker demand, as in Argentina, which casts doubt on future prospects for economic development. In El Salvador and Nicaragua, inequality fell because real (rather than just relative) earnings of workers having secondary or tertiary education declined in the face of weaker demand. El Salvador is a particularly worrying case: real wages fell at all levels of educational attainment but more so for more educated workers. This serves as a reminder that equalizing outcomes do not always arise from desirable economic developments.24
In some cases, the distributional benefits of falling skill premiums may have been bought at a high price. According to one striking finding, education is now valued so little in Bolivia that the wage premium for workers having tertiary education compared to those who underwent only primary schooling is zero. This points to an alternative or at least complementary cause of reduced skill premiums. The quality of education may have deteriorated with increased access to schooling beyond basic levels, and teaching and labor market demands may be poorly matched. This pessimistic view receives some support from evidence for negative returns on higher education in Peru and Chile owing to decreased teaching quality and for the consequences of mismatch between secondary schooling and employer demand in Argentina, Brazil, and Chile.25
Other economic factors have been more transient. Strong international demand for commodities helped rural workers narrow the wage gap to urban ones but has since abated. Some of the leveling since 2002 simply represented a recovery from a prior temporary surge in inequality that had been triggered by economic crises. The best known case is that of Argentina, where a massive economic meltdown between 1998 and 2002 plunged large parts of the population into poverty. Since then, a steady economic recovery, coupled with a shift to low-skill labor-intensive sectors that has reduced demand for skilled labor and depressed skill premiums, has disproportionately benefited the less affluent half of the population. So, too, have stronger unions and increased government transfers. Colombia, Ecuador, Uruguay, and Venezuela likewise experienced some inequality attenuation from similar recoveries. According to one estimate, if we were to exclude the equalizing effects of the recovery from crisis, the average reduction in income inequality in the first half of the 2000s would be quite modest, on the order of a single Gini point. More generally, the abatement of unfavorable short-term consequences of liberalization in the 1990s exerted a mitigating influence. Strong economic growth, averaging 4 percent per year in real terms or twice the rate of the previous decades, boosted employment but has been estimated to account for only a small fraction of the observed change in inequality. Moreover, these favorable conditions no longer apply, as annual GDP growth in the region declined for five consecutive years after 2010, from 6 percent in 2010 to a projected 0.9 percent in 2015. At the time of writing, Brazil, by far the largest economy in the region, was said to be enduring its worst recession since the Great Depression. All this casts doubt on the prospects of further leveling.26
Finally, expanded government transfers have attracted considerable publicity as a means of combating disposable income inequality. In Brazil, for example, where changes in the size, coverage, and distribution of transfer payments accounted for about half of the decline in inequality in the first decade of this century, the “Bolsa Familia” program has reached 11 million poor families. Nevertheless, compared to that found in developed countries, the actual scale of redistributive transfers in Latin America has remained very low. It is true that the presence of large numbers of impoverished households makes it possible even for relatively modest transfers (on the order of a few tenths of a percentage point of GDP) to make a difference to many people’s lives and produce equalizing effects. Yet in Western Europe, gross incomes tend to differ greatly from disposable incomes, whereas in Latin America, they hardly do so at all. Multiple reasons have been invoked. The volume of tax collection relative to GDP is small by international standards, and income taxes are particularly low. At the same time, tax evasion is rife, partly because of distrust in government and partly thanks to the large size of the informal sector. The average exemption level for income tax is about twice mean per capita GDP for the region as a whole, and in several countries, progressive rates apply only at very high income levels. Lack of state revenue thus severely limits the potential for transfers. To make matters worse, some welfare schemes are conducive to net inequality. Pensions and unemployment insurance disproportionately benefit those in the top quintile of the income distribution, primarily urban workers in formal employment arrangements, and discriminate against the rural population and those in the informal sector. Only direct cash transfers are different in that they mostly support those in the lower half of the income distribution—but they can do so only to the extent that they are not impeded by revenue constraints and offset by more regressive forms of welfare.27
Why is fiscal redistribution in Latin America so feeble? This question takes us back to the central theme of this book, the transformative power of violent shocks. As we have seen, the progressive fiscal systems of the West are firmly rooted in the two world wars, just as redistribution under communist regimes was rooted in other forms of violent upheaval. By contrast, economic development as such is not a useful indicator of the degree of fiscal redistribution. In 1950, when Western nations and Japan were busy taxing the rich and erecting ambitious welfare systems, per capita GDP (in 1990 International Dollars) ranged from $4,000 to $7,000 in Germany, France, the Netherlands, Sweden, the United Kingdom, and Canada, was closer to $2,000 in Japan, and even in the United States was not dramatically higher than in Western Europe. These values are broadly in line with leading South American economies such as Argentina and Venezuela even at that time and with a wider range of Latin American countries today: equivalent mean per capita GDP in the eight most developed substantial countries in the region was $7,800 in 2010 and averaged $6,800 in a much larger sample. By this metric, the average Argentinian, Chilean, and Uruguayan is better off now than the average American was in 1950.28
This shows that fiscal restraint in Latin American countries has not been determined by economic performance. Around the world, violent shocks have been an essential precondition for the expansion of fiscal systems, not merely in the first half of the twentieth century but also for hundreds and even thousands of years. Bloody interstate wars and transformative revolutions played a very minor role in the last two centuries of Latin America history. This helps us understand how high levels of inequality have persisted across most of the region. Various features that are specific to the region have been invoked in accounting for this phenomenon, most notably the pernicious influence of racism and colonial institutions of forced labor and slavery and the persistence of clientelism and oligarchic power. Yet what did not happen may be similarly or, arguably, even more important as we attempt to make sense of abiding differences in the sheer scale of inequality between Latin American and most other parts of the world. Against this background, it is highly questionable whether major breakthroughs in income equalization are feasible, let alone plausible.29
Policy decisions related to public spending on education, foreign investment, and tax revenues and transfers explain much of the leveling that has occurred in Latin America since the opening years of this century. More purely economic factors contributed in the form of favorable international conditions and recovery from prior crises but have proven to be more short-lived. As recovery has run its course and external demand is diminishing, further leveling would require more aggressive fiscal restructuring to improve education (considering that falling skill premiums are a mixed blessing if they stem from falling demand or poor educational outcomes) and expand redistributive transfers. It is too soon to tell whether the leveling process that commenced more than a decade ago will continue—or rather, in many cases, resume. Five or ten years from now, we will have a better sense of the sustainability of this trend.30
I conclude that the Latin American experience offers only very limited evidence for peaceful inequality attenuation and, at least for now, none at all for persistent and substantial leveling in the absence of violent shocks. During the last 150 years, phases of growing inequality have been interspersed with episodic reversals linked to external factors such as Western macroeconomic crises or, in a few cases, aggressive or violent policies. Although it is hard to disagree with Bolivia’s president Evo Morales’s maxim that “if you combine intellectual and professional capacity with a social conscience, you can change things,” the history of Latin America does little to challenge the primacy of leveling by violent means.31
What is more, none of the forces discussed in this chapter and the preceding chapter can be shown to have had a consistently dampening effect on material inequality. This is true of peaceful land and debt reform, economic crises, democracy, and economic growth. What all of them have in common is that they sometimes alleviate inequality and sometimes do not: in short, there is no even remotely uniform trend in outcomes. It is true that as modern economic development has caused the importance of human capital to rise relative to that of physical capital and as inequality in the distribution of human capital is primarily a function of the provision of education, equalizing policies regarding the latter may seem particularly promising. Even so, although investment in education, through its effect on wage differentials, may indeed serve as a viable mechanism of nonviolent leveling, it has historically been enmeshed in less peaceful processes: the documented swings in American skill premiums during the twentieth century once again underscore the importance of warfare in shaping social policies and economic payoffs. As we have seen in chapter 5, much the same applies to unionization. Redistributive fiscal and welfare policies do reduce disposable income inequality, but their scale and structure likewise tends to be tied to the legacy of violent shocks and its longer-term repercussions: the contrast between Western and East Asian inequality on the one hand and conditions in Latin America on the other reminds us of this fundamental association. Even after reviewing alternative causes of inequality compression, there is no escaping the fact that violence, actual or latent, has long been a critical catalyst for equalizing policy measures.
1 Ginis for Italy: Rossi, Toniolo, and Vecchi 2001: 916 table 6 (decline since 1881); Brandolini and Vecchi 2011: 39 fig. 8 (stability between 1871 and 1911). Italian emigration: Rossi, Toniolo, and Vecchi 2001: 918–919, 922. Positive selection among emigrants: Grogger and Hanson 2011. Mexico has been a partial exception: Campos-Vazquez and Sobarzo 2012: 3–7, and esp. McKenzie and Rapoport 2007 for the complexity of outcomes. Remittances tend to reduce inequality but only to a small extent: see, e.g., Acosta, Calderon, Fajnzylber, and Lopez 2008 for Latin America. Immigration lowered U.S. real wages between 1870 and 1914: Lindert and Williamson 2016: 180–181. Card 2009 estimates that immigration accounted for 5 percent of the increase in U.S. wage inequality between 1980 and 2000. Throughout history, migration occasionally created fairly egalitarian settler societies from scratch: examples range from ancient Greek colonists to American pioneers. However, the picture may change substantially once we take account of corresponding increases in intergroup inequality between indigenes and newcomers.
2 Alvaredo and Piketty 2014: 2, 6–7 comment on the inadequacy of the current evidence for petro-states. Note Piketty’s argument that strong economic growth in the decades after World War II was associated with falling inequality primarily because the violent shocks of 1914 to 1945 and their policy consequences had caused the rate of return on capital (after tax and wartime losses) to fall below the rate of growth: Piketty 2014: 356 fig. 10.10.
