Commodities, Poverty and Income Inequality in Latin America
This week I took part in a panel sponsored by the Nobel Foundation, which had as its main attraction a speech by Nobel Laureate Eric Maskin. Prof. Maskin talked about the impact of globalization on income distribution, emphasizing what he sees as an apparent paradox: that greater trade integration has increased income inequality at the national level, rather than lowering it, as predicted by the Ricardian theory of comparative advantage and, more precisely, the Heckscher–Ohlin model.
Not being an expert in world income distribution, I focused my talk on the remarkable decline in poverty and income inequality throughout Latin America over the last decade. This was also a period in which the region integrated more deeply into the world economy. In particular, the commodity supercycle fostered a large expansion in exports, which caused local currencies to strengthen and imports to boom. This, in turn, attracted large inflows of foreign direct investment, further deepening international integration.
Overall, my sense is that the commodity boom was an important driver behind the improvement in social indicators: for one, it provided governments with a windfall to pay for the rise in income transfers and pensions; for another, the appreciation of the exchange rate benefitted nontradable over tradable sectors, boosting the output of sectors intensive in low-skilled labor. Moreover, it was behind the acceleration in output and employment growth, which was key to reduce poverty.
Although I do not dwell on this, I believe that increased financial integration was also important, by boosting domestic credit and asset prices.
A recently released World Bank report highlights Latin America’s progress in reducing poverty.[i] Moderate poverty, defined as living on less than $4 a day, fell from 42 percent of the population in 2000 to 25 percent in 2012. Extreme poverty, defined as life with less than $2.5 a day, fell from 9.9 percent in 2002 to 5.0 percent in 2010. Progress was not, however, uniform across the region. While poverty declined substantially in what the report calls Southern Cone Extended and the Andean Region, progress in Mexico and Central America was almost nonexistent (Figure 1).[ii]
About two-thirds of the decline in poverty stemmed from improvements in labor income (Table 1). The rise in employment -- with the ensuing rise in labor force participation and a fall in the unemployment rate -- and higher labor income accounted for, respectively, 25 percent and 44 percent of this decline. Remarkably, income transfers, including conditional cash transfers, answered for just 9 percent of the drop in poverty in the region.
Table 1: Contributions of main drivers behind poverty reduction in 2003-12
Labor Income |
68% |
Men labor |
42% |
Share of occupied |
14% |
Labor income |
28% |
Female labor |
26% |
Share of occupied |
11% |
Labor income |
16% |
Other incomes |
32% |
Pensions |
14% |
Transfers (public and private)* |
9% |
Other non-labor incomes |
10% |
Source: World Bank (2014). (*) Private transfers
are basically remittances from abroad.
The World Bank study also reveals that about 68 percent of the fall in poverty between 2003 and 2012 resulted from economic growth, while the other 32 percent arose from income redistribution. The relative contributions of growth and falling income inequality were not, however, homogeneous throughout Latin America. In the Andean Region, growth answered for 83 percent of the decline in poverty, while in Mexico and Central America it accounted for 53 percent. In the Southern Cone, growth was responsible for 64 percent of the fall in poverty and redistribution for the remaining 36 percent.[iii]
Although it accounted for a just a third of the fall in poverty, the decline in inequality in Latin America in the last decade was nonetheless remarkable, widespread, and contrasting with the previous decade’s worsening of distribution (Table 2). Better wage distribution was the main driver behind the decrease in income inequality (World Bank, 2012).[iv] In particular, the wage premia earned by better educated workers declined vis-à-vis that of low-skilled workers.[v]
Table2: Gini coefficient (in percent)
Country |
1990 |
2002 |
2011 |
Country |
1990 |
2002 |
2011 |
Argentina* |
44.0 |
47.6 |
40.2 |
Honduras |
51.8 |
51.6 |
52.4 |
Bolivia |
44.1 |
55.0 |
42.9 |
Mexico** |
47.2 |
49.1 |
46.4 |
Brazil |
56.8 |
55.4 |
50.1 |
Nicaragua* |
53.5 |
49.9 |
43.3 |
Chile |
51.7 |
51.5 |
48.5 |
Panama* |
51.8 |
52.2 |
48.0 |
Colombia |
48.0 |
50.9 |
48.9 |
Paraguay |
36.9 |
53.0 |
50.5 |
Costa Rica |
42.6 |
45.9 |
47.0 |
Peru |
52.7 |
53.7 |
46.9 |
Ecuador |
47.9 |
51.3 |
44.9 |
Uruguay |
46.3 |
48.3 |
45.8 |
El Salvador** |
47.6 |
47.8 |
44.1 |
Venezuela |
40.9 |
43.7 |
37.4 |
Guatemala*** |
55.1 |
52.3 |
53.4 |
Source: Soltz (2013).[vi] (*) Last year is 2012; (**) 2010; and (***) 2006.
I have difficulty attributing this decline in wage premia to greater international integration. Regarding both trade and foreign direct investment, Latin America integration into the world economy increased more in the 1990s than in the 2000s (Figures 2 and 3). And, as shown in Table 2, the last decade of the XX century saw an increase in inequality. So it is hard to see how greater openness may have improved distribution in wage earnings in the last decade, but not in the previous one. Conversely, how can one attribute the rise in inequality in the 1990s to greater international integration, while accepting the opposite outcome in the 2000s?
Three other explanations seem more promising:
i. An increase in the supply of skilled workers; in particular, a significant expansion in the proportion of workers with secondary and tertiary education;
ii. A decline in the average quality of tertiary education, as a consequence of the large and fast expansion in enrollments;
iii. An increase in the demand for low-skilled workers, vis-à-vis that for skilled workers.
All these three forces were at play in Latin America in the first decade of this century, but it is hard to tell apart their individual effects (Lustig, Lopez-Calva and Ortiz-Juarez (2013) and importance. My view on this, though, is the following:
(a) The rise in average schooling was probably an important driver of lower inequality. But average years of education of the labor force (15 years old and more) in Latin America increased by 1.23 year in the 1990s, against a more modest rise of 1.07 year in the 2000s.[vii] Moreover, educational inequality declined in both decades by a similar amount – 2 percentage points in the Gini coefficient (World Bank, 2012). It seems hard to reconcile this with the fact that income inequality rose in the 1990s and declined in the 2000s. At the very least, it was not only education.
(b) The deterioration in the mean quality of tertiary education is consistent with anecdotal evidence from Brazil. It is explained by a combination of supply expansion concentrated on private, lower quality education and a decline in the average quality of students entering university. This hypothesis could in principle be tested by looking at the dispersion of wage premia amongst university graduates.
(c) The change in the composition of labor demand is also consistent with evidence from Brazil. Thus, almost all the acceleration in output growth between 1998-2004 and 2005-11 was due to faster growth in financial intermediation, construction, commerce, transportation, public utilities and other services. That is, non-tradable sectors that are mostly intensive in low-skilled labor (compared, for instance, with manufacturing, whose share in total employment declined).
Both growth and the decline in inequality have slowed down since 2010, signaling more modest reductions in poverty in the future. In particular, after dropping continuously between 2001 and 2010, Latin America’s Gini coefficient has stalled around 0.52 percent after 2010. As the commodity boom recedes and countries strive to lower their external imbalances, it is likely that their exchange rates will weaken further, so that resources move from nontradable to tradable sectors.
This will limit the scope for lowering inequality as it was done in last decade. Further reducing education inequality may be a way to counter the effect of the economic slowdown and the restructuring of demand. But, in my view, more will have to be done. In particular, I believe that two issues will enter the agenda.
First, but possibly more difficult, is adopting less regressive taxes. Thus, while government transfers help reduce inequality, taxes overall raise it, so that at the end of the day fiscal policy has only a modest role in improving distribution in Latin America. This is shown in Table 3, where I report the difference between the Gini coefficient calculated before taxes and transfers and afterwards. Thus, fiscal policy has a relevant contribution to lower inequality only in Brazil, Guatemala and Uruguay. Even in these cases, though, this contribution is much lower than in developed countries. And this feature has not changed over the last two decades.
Table 3: Difference between market Gini and net income Gini1 (in percentage points)
Country |
1990 |
2002 |
2011 |
Country |
1990 |
2002 |
2011 |
Argentina |
1.8 |
1.0 |
1.1 |
Panama* |
1.3 |
1.8 |
1.1 |
Bolivia |
1.9 |
1.9 |
1.8 |
Paraguay |
1.2 |
1.5 |
1.2 |
Brazil |
4.0 |
4.3 |
3.7 |
Peru |
- 0.3 |
0.8 |
0.5 |
Chile |
1.6 |
1.4 |
1.5 |
Uruguay |
4.7 |
4.9 |
4.2 |
Colombia |
0.9 |
0.5 |
1.5 |
Venezuela |
1.6 |
2.0 |
1.8 |
Costa Rica |
1.2 |
1.3 |
1.3 |
||||
Ecuador |
1.8 |
- 0.9 |
1.8 |
France |
13.5 |
18.3 |
15.2 |
El Salvador** |
2.0 |
1.4 |
1.5 |
Germany |
15.9 |
19.3 |
19.6 |
Guatemala*** |
2.6 |
2.3 |
4.4 |
Japan** |
6.0 |
5.3 |
6.9 |
Honduras |
1.9 |
2.1 |
1.8 |
United Kingdom* |
11.7 |
12.4 |
11.7 |
Mexico** |
1.2 |
2.3 |
2.3 |
United States |
9.1 |
9.0 |
9.3 |
Nicaragua* |
1.3 |
1.7 |
1.2 |
Source: Soltz (2013).[viii] (*) Last year is 2012; (**) 2010; and (***) 2006.
(1) Net income Gini accounts for taxes and government transfers.
Second, enlarge and improve the provision of basic public services, which go disproportionately to the poor. Thus, when one accounts for the provision of public services, notably public health and education, the Gini coefficient falls substantially. According to Lustig, Pessino and Scott (2013), the Gini coefficients of Argentina, Bolivia, Brazil, Mexico, Peru and Uruguay decline by, respectively, 8.8, 6.0, 9.1, 4.4, 2.0, and 6.3 percentage points when in-kind transfers are monetized and counted as sources of income.[ix] I would speculate that public housing programs, such as Minha Casa, Minha Vida in Brazil, have similar effects.
[i] World Bank, Social Gains in the Balance: A Fiscal Policy Challenge for Latin America and the Caribbean, 2014.
[ii] Southern Cone Extended comprises Argentina, Brazil, Chile, Paraguay and Uruguay; Andean Region, Bolivia, Colombia, Ecuador and Peru; and, Central America, Costa Rica, Dominican Republic, El Salvador, Guatemala, Honduras, Nicaragua, and Panama
[iii] This is consistent with GDP growth accelerating from 2.4% p.a. in 1994-2003 to 5.4% in 2003-12 in the Andean region, against accelerations from 2.0% to 4.5% in the Southern Cone, and from 2.4% to 3.0% in Mexico and Central America.
[iv] World Bank, The Labor Market Story Behind Latin America’s Transformation, 2012. According to Lustig, Lopez-Calva and Ortiz-Juarez (2013), better wage distribution accounted for about 45 percent of the decline in income inequality, followed by transfers (14 percent) and the faster increase in the adult population (12 percent). Nora Lustig, Luis Lopez-Calva and Eduardo Ortiz-Juarez, “Deconstructing the Decline in Inequality in Latin America”, Tulane University, Working Paper 1314, 2013.
[v] More to the point, the higher return to primary, secondary and tertiary schooling vis-à-vis no education or incomplete primary education.
[vi] Frederick Solt, 2013, "The Standardized World Income Inequality Database", http://hdl.handle.net/1902.1/11992 Frederick Solt [Distributor] V10 [Version]
[vii] Robert Barro and Jong-Wha Lee, “A New Data Set of Educational Attainment in the World”, 2011. http://bit.ly/1oMJxsG)
[viii] Frederick Solt, 2013, "The Standardized World Income Inequality Database", http://hdl.handle.net/1902.1/11992 Frederick Solt [Distributor] V10 [Version]
[ix] Nora Lustig, Carola Pessino and John Scott, “The Impact of Taxes and Social Spending on Inequality and Poverty in Argentina, Bolivia, Brazil, Mexico, Peru and Paraguay”, Tulane University, Working Paper 1313, 2013.
{jcomments on}