IJE Advance Access originally published online on February 28, 2006
International Journal of Epidemiology 2006 35(3):623-632; doi:10.1093/ije/dyl026
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Article |
Peripherality, income inequality, and life expectancy: revisiting the income inequality hypothesis
1 Centre de recherche de Centre Hôpital de Université de Montréal, 3875 St Urbain, Montréal, QC, Canada H2W 1V1
2 Centre for Health and Policy Studies, Department of Community Health Sciences, University of Calgary, 3330 Hospital Drive NW, Calgary, AB, Canada T2N 4T1
* Corresponding author. Spencer Moore, PhD, MPH, Centre de recherche du CHUM, Axe santé des populations et épidémiologie sociale, 3875 St Urbain, 3e étage, porte 3-30, Montréal, QC, Canada H2W 1V1. E-mail: spencer.moore{at}umontreal.ca
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Background Recent criticisms of the income inequality and health hypothesis have stressed the lack of consistent significant evidence for the stronger effects of income inequality among rich countries. Despite such criticisms, little attention has been devoted to the income-based criteria underlying the stratification of countries into rich/poor groups and whether trade patterns and world-system role provide an alternative means of stratifying groups.
Methods To compare income-based and trade-based criteria, 107 countries were grouped into four typologies: (I) high/low income, (II) OECD membership/non-membership, (III) core/non-core, and (IV) non-periphery/periphery. Each typology was tested separately for significant differences in the effects of income inequality between groups. Separate group comparison tests and regression analyses were conducted for each typology using Rodgers (1979) specification of income, income inequality, and life expectancy. Interaction terms were introduced into Rodgers specification to test whether group classification moderated the effects of income inequality on health.
Results Results show that the effects of income inequality are stronger in the periphery than non-periphery (IV) (–0.76 vs –0.23; P < 0.05). An incremental F-test confirmed significant differences in the coefficient subsets between the two groups (F2,101 = 6.31; P < 0.01).
Conclusions Cross-national analyses of income inequality and population health have assumed (i) income differences between countries best capture global stratification and (ii) the negative effects of income inequality are stronger in high-income countries. However, present findings emphasize (i) the importance of measuring global stratification according to trading patterns and (ii) the strong, negative effects of income inequality on life expectancy among peripheral populations.
Keywords Trade, income inequality, global health, life expectancy
Accepted 31 January 2006
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Central to cross-national studies of the effects of income inequality on population health has been the hypothesis that the effects of absolute income give way to those of relative income at higher levels of national income.1–4 This shift in the supposed relationship of life expectancy and income inequality between groups has been posed as a question of structural stability with respect to absolute income level.5 Critics of the income inequality hypothesis have stressed the inconsistent evidence confirming stronger effects of income inequality among high-income countries.1,5–9 Yet, the typologies used to classify countries into high/low income have rested primarily on one of two factors: (i) the level of GDP/c, usually over $5000 per year,2–5 or (ii) membership in the Organisation of Economic Cooperation and Development (OECD).10
Despite research on a variety of measurement and methodological issues, little or no attention has been given to alternative measures for assessing power and global stratification, and their possible effects on the income inequality–health association. The following study compares two income-based with two trade-related typologies to investigate whether alternative measures of power and stratification help elucidate the potential effects of income and income inequality on life expectancy. Studies of international trade networks draw attention to the hierarchical structure of the global system and the overall power that countries possess within that system,11,12 including their ability to protect public health.13 Within a world-systems framework, the disproportionate power and influence that core countries exercise over the periphery does not simply rest on income, or GDP/c, but on a system of asymmetrical trade relationships and an international division of labour.14
Four different country typologies are compared—(I) high/low income, (II) OECD membership/non-membership, (III) core/non-core, and (IV) non-periphery/periphery (Tables 1
–4). Each typology is tested separately to determine whether there is a significant shift in structural stability between groups in each typology. Findings could help focus attention on the effects of international trade and global stratification on population health.
|
|
|
|
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Data and sample
All data were gathered from United Nations or World Bank databases. Global analyses of national-level health and income inequality that include both developed and developing countries are complicated by cross-national variations in such elements as census and surveillance infrastructures and procedures, survey methods, and measurement instruments.15,16 These cross-national variations can affect data quality on income distribution and life expectancy, and thus the overall reliability of results.15,16 Within these limitations, the study sought to improve data validity by taking averaging values and comparing data and results using alternative sources. National GDP per capita is based on purchasing power parity values and represent a country's average GDP/c during the 1990s.17 The decade-long average was used to reduce the effects of any short-term aberrations. The Gini coefficient was used to indicate the level of income inequality. These data were unavailable for each country in the same year; therefore, to be included, Gini coefficient data had to come from the decade 1991–2000 and be acceptable for a World Bank/United Nations database.18 Gini coefficients were calculated on the basis of income or consumption data and range from 0 (perfect equality) to 100 (perfect inequality). So that income- and consumption-based coefficients are more comparable for cross-national analysis, the study follows Deininger and Squire's recommendation and other analyses to adjust consumption-based Gini coefficients upwards 6.6.19,20 Life expectancy for 2002 is based on the average from both World Bank and United Nations Development Programme sources.18
To be included in the study, a country had to have complete data on GDP/c, income inequality, and life expectancy and had to be assigned to a core, semi-peripheral, or peripheral role in a study of the international trade in capital-intensive commodities.21 In total, 107 countries are included.
The typologies
Countries were classified within four separate typologies: (I) high income/low income, (II) OECD membership/non-membership, (III) core/non-core, and (IV) non-periphery/periphery. To be included in the high-income category of typology (I), countries had to maintain an average GDP/c > $5000 during the 1990s. This division replicates previous high/low income classifications.5 In total, 54 countries were classified as high income and 53 countries as low income. For Typology II, a country was classified as an OECD member if it was a member prior to the 1990s. Limiting membership to pre-1990 members allowed the typology to remain consistent with previous studies. The sample included 23 OECD member and 84 non-member countries [All 23 OECD countries are found in the high income (Typology I), and non-periphery groups (Typology IV).]
Country assignments in Typologies III and IV are based on results from Moore et al.21 Moore et al. use circa 2000 international trade data to construct a network data matrix recording the presence (1) or absence (0) of a tie between countries in the trade of capital-intensive commodities. Block modelling techniques were used to measure the degree of equivalence between countries, cluster those with similar patterns of trade, and assign countries to core, semi-peripheral, and peripheral roles. Countries with a pattern of dense horizontal and vertical (downward) ties to other countries form the core, while countries with sparse horizontal and dense vertical (upward) ties are more peripheral.11,12,21 For Typology III, countries that had been classified as core were assigned to the core; non-core countries were those having semi-peripheral or peripheral roles. In sum, 29 countries were assigned to the core and 78 countries to the non-core. For Typology IV, non-peripheral countries were those belonging to the core or semi-periphery; peripheral countries were those assigned exclusively peripheral roles. In total, 61 countries were assigned to the non-periphery and 46 to the periphery. The specific difference between Typologies III and IV is the assignment of semi-peripheral countries (There are 33 semi-periphery countries: Argentina, Belarus, Bulgaria, Chile, Croatia, Colombia, Costa Rica, Côte d'Ivoire, Egypt, Estonia, Hungary, Indonesia, Iran, Jordan, Kuwait, Latvia, Lebanon, Lithuania, Luxembourg, Mexico, Morocco, New Zealand, Oman, Pakistan, Peru, Philippines, Poland, Romania, Slovakia, Slovenia, Tunisia, Ukraine, and Venezuela.).
Interaction effects and group comparisons
After testing different model specifications, Rodgers' (1979) preferred model for estimating the effects of income and income inequality on life expectancy is
 |
k is the error term.5,22 Quadratic terms are included in the equation to correct for the curvilinear nature of the income–life expectancy relationship. Previous tests exploring alternative estimation procedures have shown no significant differences between Rodgers' estimation and alternative forms.5
Following Rodgers' (1979) preferred specification, the analysis first tested for evidence of a changing relationship between national income and income inequality in the pooled sample.
 |
Replicating test procedures used by Gravelle et al.,5 four separate analyses, corresponding to the four typologies, were then conducted to test for differences in the effects of income inequality between groups in each typology. These analyses were conducted using ordinary least squares (OLS) regression through the following four steps. First, life expectancy was regressed on GDP/c (linear and quadratic terms), the Gini coefficient, and a dummy variable representing group membership:
 |
 |
 |
 |
 |
Several regression diagnostic procedures were used to check for the presence of influential outliers, multicollinearity, and heteroscedasticity in the different models. Diagnostics flagged Botswana in Typologies I and II, and Côte d'Ivoire and South Africa in Typologies III and IV as potentially influential outliers. Since there was no conceptual justification and follow-up tests, excluding the cases did not alter the significance of the findings (Removal of the two cases in the non-periphery/periphery typology increased coefficient differences between non-periphery and periphery.); the two countries were kept. To test for spatial autocorrelation, a Moran's I was computed for each model's residuals, using all neighbours and a weighting scheme based on the inverse of the squared distance. None of the Moran's I was significant, thus showing no spatial autocorrelation globally in the residuals of any of the models. To manage heteroscedasticity while testing for interaction effects, results are reported using a heteroscedasticity-consistent standard error (HCSE) estimator of OLS parameter estimates.24 SAS statistical software was used for analyses, and the HC3 macro for computing HCSE estimators is in Hayes.25
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Summary statistics shown in Table 5 reveal differences between groups within each typology in life expectancy, GDP/c, and income inequality. The greatest differences in GDP/c between groups are those based on OECD membership or non-membership; the greatest differences in income inequality are also based on OECD membership; while the greatest difference in life expectancy is between non-peripheral and peripheral groups.
|
Regression results are found in Table 6. Pooled analyses show a significant positive effect of income on life expectancy and support findings showing a significant negative effect of income inequality on life expectancy (Model 1).2,10,22,26–28 Differences between present findings and previous studies could relate to a different sample composition and size, or alternative data sources. Pooled results support findings that have demonstrated the lack of interaction effects between income and income inequality (Model 2).5
|
Group comparison tests on each typology confirm Gravelle et al.'s5 finding of no difference in the relationship between life expectancy, GDP/c, and income inequality in high-income and low-income countries (Models 5, 6a, 6b). Results also demonstrate that the relationship between life expectancy, GDP/c, and income inequality is no different within core and non-core groups. No significant dummy variables (Model 3) or interaction effects between country classification and national income (Model 4) were found among any typologies and are, therefore, not reported.
Results from both Models 5 and 6 do, however, suggest that the effects of income inequality on life expectancy are different within OECD and non-OECD countries (Typology II), and non-peripheral and peripheral countries (Typology IV). [To verify these findings, they were retested in three ways. First, the non-periphery/periphery typology was reconstructed using alternative core-semiperiphery-periphery country assignments.12 This resulted in a smaller overall sample size (n = 87) but tests confirmed present findings by showing evidence of the stronger effects of income inequality in the periphery. Second, an alternative source of Gini coefficients, WIDER,29 was used to measure income inequality. WIDER data support the significance of the interaction effect within the non-periphery/periphery but not the OECD/non-OECD typology. Third, instead of an HCSE estimator, a weighted least squares approach in which the groups in each typology were weighted according to the inverse error variances of the bivariate regression of each group on life expectancy was used to confirm findings.30 Results support original findings.] Since no significant differences exist in the results of Models 5 and 6, Model 6 is reported. A follow-up incremental F-test, comparing the original (Model 1) and expanded (Model 6) equations, confirms the significance of coefficient differences in income inequality between non-peripheral and peripheral groups (F2,101 = 6.31; P < 0.01) but not between OECD and non-OECD groups (F2,101 = 1.15; P > 0.05).
As shown in Figures 1 and 2, the interaction is disordinal, meaning that the regression lines for both groups intersect within the range of Gini values reported. Eliminating from the equation the similar effects of income on life expectancy across the two groups, which is 
, the final predicted specification for life expectancy in non-peripheral countries is:
 |
|
|
Typology IV is the only one that exhibits a significant interaction effect with income inequality and a significant amount of incremental explained variance due to the interaction of peripherality and income inequality. Excluding the confounding effects of factors other than income, the overall effects of income inequality on life expectancy appear much stronger among peripheral countries. Although there is a sampling error associated with the calculation of points of intersection,31 the life expectancy of non-peripheral countries would be equal to peripheral countries at a Gini of roughly 45.0. As income inequality rises above 45.0 (i.e. toward greater inequality), life expectancy within peripheral countries falls below non-peripheral countries.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
While the quality of international data, the presence of other possible confounding variables, and the cross-sectional nature of the study limit the possible inferences that can be drawn, the criteria used to stratify countries in cross-national analyses appear to matter for the significance of research into the effects of income inequality on health. When GDP/c is used to stratify countries, present findings confirm earlier studies showing no significant difference in the relationship between life expectancy, income, and income inequality in high-income and low-income countries. When the criteria shifts to trading patterns and world-system role, results suggest that there is a significant difference in the relationship between life expectancy, income, and income inequality in non-peripheral and peripheral countries. There is a negative effect of income inequality on health in the non-periphery and periphery, but results show the significantly stronger effects of income inequality in the periphery.
This finding contrasts sharply with the present understanding of the income inequality hypothesis. Two main reasons have been advanced for the hypothesized stronger effects of income inequality among high-income countries: (i) high-income countries have undergone the epidemiological transition and experienced a shift away from infectious diseases toward chronic diseases and (ii) the curvilinear shape of the GDP/c–life expectancy relationship implies that absolute income has greater effects among low-income countries, which can be seen in Figure 3. While a discussion of the epidemiological transition model is beyond the scope of the present essay, recent research has argued that the epidemiological transition is not as unidirectional as postulated in the classical model.32–35 Furthermore, as Wilkinson later maintains, it is more accurate to argue that income inequality matters both pre-epidemiological and post-epidemiological transition, but income plays less of a role after the transition.4 Second, the curvilinear relationship between income and life expectancy raises concerns regarding the use of aggregate data for cross-national analyses. While the curve suggests that absolute income may have stronger effects in low-income countries, this is not evidence that income distribution has less of an effect. Indeed, scholars have suggested that the non-linear effect of income on mortality depends upon the distribution of income.27
|
Why would the populations of peripheral nations stand at an increased risk to the effects of income inequality on health than those in the non-periphery? The findings do not diminish the role or potential importance of neo-material or psychosocial explanations in understanding the effects of income inequality on health within countries.36 Instead, they only underscore the importance of such factors for peripheral countries as well. More importantly, present findings highlight the significance of global political–economic mechanisms and structures, and their interaction with national policies and institutions. Global policies and structures, including international trade agreements and neo-liberal philosophies,37,38 set the parameters of a population's exposure and vulnerability to disease and illness.39,40
The relative insensitivity of the high-low income typology may be due in part to its incapacity to measure multidimensional aspects of power, such as degree of national sovereignty, and assess the global political-economic mechanisms that affect population health. For example, comparing Typologies I and IV, 12 countries, including China, India, and Indonesia, change positions from low-income (Typology I) to non-periphery (Typology IV); five countries move from high-income to the periphery. (Belarus, China, Cote d'Ivoire, Egypt, India, Indonesia, Jordan, Morocco, Pakistan, Peru, Philippines, and Ukraine move from low income to non-periphery; Botswana, Namibia, Panama, Trinidad and Tobago, and Uruguay move from high income to periphery.) Despite the fact that countries such as China and Indonesia have a low GDP/c, they are not lacking in economic power or global influence. Having a low GDP/c is, thus, not equivalent to being peripheral. Countries in the periphery may also have a low GDP/c, but peripherality represents more than a low national income; peripherality reflects a historical process of economic and social marginalization in which peripheral countries have been excluded and disempowered from more symmetrical trading relationships.14 Understood within the premises of classical world-systems theory, these asymmetrical trading ties develop alongside the creation of an international division of labour. Unequal development, a global stratification of production and labour, and the dominance of core and semi-peripheral nations over the periphery are seen to characterize the global system.41 While various premises of world-systems theory may be disputed, the theory highlights among other elements the influence of global political and economic processes on national outcomes.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Conventionally framed, cross-national studies on the effects of income inequality have sought those effects primarily among high-income countries, assuming that relative income differences matter more in societies with high levels of economic development. In the process, it has been assumed that the main ingredient in the formula to improve the health and well-being of populations in developing countries was national economic growth.42 What becomes unintentionally obscured in such a formulation is attention to the patterns of economic growth within developing countries,42 and the global structures that affect economic growth and health. Framed instead around a country's role in the global system, present findings highlight the need for greater attention to the patterns of economic growth in peripheral countries and the global processes that marginalize those countries.
In this regard, the study does not argue that there is no variation among non-peripheral or peripheral countries in their power and influence in global affairs. It does suggest that there is a significant difference in the power of non-peripheral countries such as the United States, China, and Brazil and peripheral countries such as Albania, Ethiopia, and Kazakhstan to shape the content and implementation of international trade agreements and, thus, maintain a higher degree of national sovereignty over the global forces that impact the health of their populations.
KEY MESSAGES
|
|
Acknowledgments |
|---|
I would like to thank Alan Shiell, Lindsay Bradshaw, Ana Teixeira, Yan Kestens, Mark Daniel, and IJE's anonymous reviewers for their helpful comments and suggestions for the manuscript. Responsibility for any mistakes or oversights in the contents of the manuscript, however, rests with me in full.
I would also like to thank the following funding agencies for their support. Early work was supported through a fellowship from the Alberta Heritage Foundation for Medical Research; later work was supported through fellowships from the Fonds de la recherche en santé Québec and the Canadian Institutes of Health Research.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
1 Lynch J, Davey Smith G, Harper S Hillemeier M, Ross N, Kaplan GA et al. Is Income Inequality a Determinant of Population Health? Part 1: A Systematic Review. Milbank Q 2004;82:5–99.[CrossRef][ISI][Medline]
2 Wilkinson, R.G. Income distribution and life expectancy. BMJ 1992;304:165–68.[ISI][Medline]
3 Wilkinson, RG. The epidemiological transition: from material scarcity to social disadvantage? Daedalus 1994;123:61–76.[Medline]
4 Wilkinson RG. Unhealthy Societies: The Afflictions of Inequality. London: Routledge Press, 1996.
5 Gravelle H, Wildman J, Sutton M. Income, income inequality and health: what can we learn from aggregate data? Soc Sci Med 2002;54:577–89.[CrossRef][ISI][Medline]
6 Judge K. Income distribution and life expectancy: A critical appraisal. BMJ 1995;311:1282–87.
7 Judge K, Mulligan JA, Benzeval M. Income inequality and population health. Soc Sci Med 1998;46:567–79.[CrossRef][ISI][Medline]
8 Lynch JG, Davey-Smith G, Hillemeier M, Shaw M, Raghunthan T, Kaplan G. Income inequality, the psychosocial environment, and health: comparisons of wealthy nations. Lancet 2001;358:194–200.[CrossRef][ISI][Medline]
9 Mackenbach JP. Income Inequality and population health. BMJ 2002;324:1–2.
10 Macinko JA, Shi L, Starfield B. Wage inequality, the health system, and infant mortality in wealthy industrialized countries, 1970–1996. Soc Sci Med 2004;58:279–92.[CrossRef][ISI][Medline]
11 Snyder D, Kick E. Structural position in the world system and economic growth, 1955–1970: a multiple-network analysis of transactional interactions. Am J Sociol 1979;84:1096–126.[CrossRef]
12 van Rossem R. The World system paradigm as general theory of development: a cross-national test. Am Sociol Rev 1996;61:508–27.[CrossRef]
13 Shaffer E, Waitzkin H, Brenner J, Jasso-Aguilar R. Global trade and health. Am J Public Health 2005;95:23–34.
14 Wallerstein I. The Modern World System: Capitalist Agriculture and the Origins of the European World Economy in the Sixteenth Century. New York: Academic Press, 1974.
15 Moran T. On the theoretical and methodological context of cross-national inequality data. Int Sociol 2003;18:351–78.[CrossRef]
16 Deaton A. Commentary: The convoluted story of international studies of income inequality and health. Int J Epidemiol 2002;31:546–49.
17 World Bank. World development indicators. Available at: http://www.worldbank.com/data/wdi2004/index.htm (Accessed on March 2004).
18 UNDP. United Nations Development Programme. Available at: http://www.undp.org (Accessed on March 2004).
19 Deininger K., Squire L. A new data set measuring income inequality. World Bank Econ Rev 1996;10:565–91.
20 Ram R. Further examination of the cross-country association between income inequality and population health. Soc Sci Med 2006;62:779–91.[CrossRef][ISI][Medline]
21 Moore S, Teixeira A, Shiell A. The health of nations in a global context: trade, global stratification, and infant mortality rates. Soc Sci Med doi:10.1016/j.socscimed.2005.12.009.
22 Rodgers G.B. Income and inequality as determinants of mortality: an international cross-section analysis. Popul Stud 1979;33:343–51.[CrossRef]
23 Aiken LS, West S. Multiple Regression: Testing and Interpreting Interactions. London: Sage, 1991.
24 Hayes A, Cai L. Using heteroscedasticity-consistent standard error estimators in OLS regression with applications to moderated multiple regression. Available at: http://www.jcomm.ohio-state.edu/ahayes/hcse.pdf (Accessed on January 2005).
25 Hayes AF. Statistical Methods for Communication Science. Mahwah, NJ: Lawrence Erlbaum, 2005.
26 Legrand J. Inequalities in health—some international comparisons. Eur Econ Rev 1987;31:182–91.[CrossRef][ISI]
27 Duleep HO. Mortality and income inequality among economically developed countries. Soc Secur Bull 1995;58:34–50.[ISI][Medline]
28 De Vogli, R, Mistry R, Gnesotto R, Cornea GA. Has the relation between income inequality and life expectancy disappeared? Evidence from Italy and top industrialised countries. J Epidemiol Community Health 2005;59:158–62.
29 World Institute for Development Economics Research (WIDER). World Income Inequality Database. United Nations University. Available at: http://www.wider.unu.edu (Accessed on November 2005).
30 Overton R. Moderated multiple regression for interactions involving categorical variables: a statistical control for heterogeneous variance across two groups. Psychol Methods 2001;6:218–33.[CrossRef][ISI][Medline]
31 Jaccard J, Turrisi R. Interaction Effects in Multiple Regression. Thousand Oaks: Sage Publications, 2003.
32 Wahdan M. The epidemiological transition. East Mediterr Health J 1996; 2. Available at: http://www.emro.who.int/Publications/EMHJ/0201/02.htm (Last Accessed on April 27, 2005).
33 Carolina M, Gustavo L. Epidemiological transition: model or illusion: a look at the problem of health in Mexico. Soc Sci Med 2003;57: 539–50.[CrossRef][Medline]
34 Smallman-Raynor M, Phillips D. Late stages of the epidemiological transition: health status in the developed world. Health Place 1999;5:209–22.[Medline]
35 Cook I, Dummer T. Changing health in China: re-evaluating the epidemiological transition model. Health Policy 2004;67:329–43.[CrossRef][Medline]
36 Lynch J. Income inequality and health: expanding the debate. Soc Sci Med 2000;51:1001–05.[CrossRef][ISI][Medline]
37 Gilbert T, Gilbert L. Globalisation and local power: influences on health matters in South Africa. Health Policy 2004;67: 245–55.[Medline]
38 Coburn D. Income inequality, social cohesion, and the health status of populations: The role of neoliberalism. Soc Sci Med 2000;58:41–56.
39 Goldblatt D, Held D, McGrew A, Perraton J. Economic globalisation and the nation-state: shifting balances of power. Alternatives 1997;22:269–85.
40 Woodward D, Drager N, Beaglehole R, Lipson D. Globalization and Health: a Framework for Analysis and Action. Bull World Health Organ 2001; 79: 875–881.[ISI][Medline]
41 Kick E, Davis B, Lehtinen M, Burns T. World-system position, national political characteristics and economic development outcomes. J Pol Mil Sociol 2000;28:131–55.
42 Gershman J, Irwin A. Getting a grip on the global economy. In Kim J, Millen J, Irwin A, Gershman J (eds). Dying for Growth: Global Inequality and the Health of the Poor. Maine: Common Courage Press, 2000, 11–43.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||