Is wealthier always healthier? The impact of national income level, inequality, and poverty on public health in Latin America

Abstract

Despite findings indicating that both national income level and income inequality are each determinants of public health, few have studied how national income level, poverty and inequality interact with each other to influence public health outcomes. We analyzed the relationship between gross domestic product (GDP) per capita in purchasing power parity, extreme poverty rates, the gini coefficient for personal income and three common measures of public health: life expectancy, infant mortality rates, and tuberculosis (TB) mortality rates. Introducing poverty and inequality as modifying factors, we then assessed whether the relationship between GDP and health differed during times of increasing, decreasing, and decreasing or constant poverty and inequality. Data were taken from twenty-two Latin American countries from 1960 to 2007 from the December 2008 World Bank World Development Indicators, World Health Organization Global Tuberculosis Database 2008, and the Socio-Economic Database for Latin America and the Caribbean. Consistent with previous studies, we found increases in GDP have a sizable positive impact on population health. However, the strength of the relationship is powerfully influenced by changing levels of poverty and inequality. When poverty was increasing, greater GDP had no significant effect on life expectancy or TB mortality, and only led to a small reduction in infant mortality rates. When inequality was rising, greater GDP had only a modest effect on life expectancy and infant mortality rates, and no effect on TB mortality rates. In sharp contrast, during times of decreasing or constant poverty and inequality, there was a very strong relationship between increasing GDP and higher life expectancy and lower TB and infant mortality rates. Finally, inequality and poverty were found to exert independent, substantial effects on the relationship between national income level and health. Wealthier is indeed healthier, but how much healthier depends on how increases in wealth are distributed.

Introduction

A seminal piece by Pritchett and Summers (1996) found increases in per capita GDP had positive health effects. In their cross-national analysis they found that a 5% increase in GDP led to an average of 1% decrease in infant mortality rates. At the national level, as a country grows wealthier, it has more resources to spend on health-promoting social programs, such as public sanitation, potable water, and health awareness initiatives. At the individual level, individual consumers have more income that they can spend on healthy foods and medical care, which could plausibly translate into improved individual and, as a result, aggregate health statistics.

What role do poverty and inequality play in a potential wealth–health relationship? Is wealthier always healthier? At the national level, Deaton (2003) showed that the effect of each additional dollar on health is weaker among richer countries than for poorer countries. Anand and Ravallion (1993) suggested that much of the positive health effects from increases in GDP occur because economic growth decreases poverty. Thus, the relationship between increases in GDP and public health should be stronger for poorer countries, where growth lifts more people out of poverty (Dollar & Kraay, 2002). However, how poverty rates within a country affect public health has not been addressed. Additionally, we know of no studies offering empirical evidence on how poverty rates modify the relationship between national income level and health.

The role of income inequality in determining public health has been hotly debated. Wilkinson’s (1992) much-cited seminal article on income distribution and life expectancy found evidence that both increases in the share of income going to the poorest 60% of the population (r = 0.80, p < 0.05) and decreases in the level of relative poverty (r = −0.73, p < 0.01) were strongly correlated with increases in life expectancy. Since then, many studies have sought to investigate whether and how income inequality affects public health (Chiang, 1999, Kaplan et al., 1996, Kawachi et al., 1997, Lynch and Kaplan, 1997, Lynch et al., 1998, Lynch et al., 2000, Marmot, 2002, Marmot, 2005, Waldmann, 1992, Wilkinson, 1994). In their review of the literature on income inequality and population health, Wilkinson and Pickett (2006) found that over 70% of the 168 analyses reviewed report that health is worse in societies with greater income inequality.

Theories explaining the mechanisms through which income inequality affects public health broadly fall into three schools. First, the psychosocial interpretation contends that poor health stems from the perception of others above oneself in the social, and particularly income, hierarchy (Marmot, 2002). This perception in turn generates negative mental states such as stress and depression that have detrimental effects on health (Wilkinson, 1996). Lynch et al. (2000), however, criticized the psychosocial interpretation as being imprecise and fraught with conceptual and empirical problems. Instead, they advanced a “neo-materialist” interpretation, which contends that “health inequalities result from the differential accumulation of exposures and experiences that have their sources in the material world” (Lynch et al., 2000, p. 1202). Finally, the “social capital” interpretation – a synthesis of the psychosocial and neo-materialist interpretations – emphasizes the income inequality’s role in preventing individuals from building and maintaining social capital, which in turn leads to poor health (Kawachi and Kennedy, 1999, Kawachi et al., 1997).

Research on inequality and health has been extended to the developing world, including Latin America, and has produced similar findings. Casas, Dachs, and Bambas (2001) presented a number of observations on income inequality and health within countries. For example, they noted that the child mortality rate was five times higher for families earning less than $50 USD per month than for families earning over $150 USD per month in Pelotas, Brazil. They also found in Peru in 1996 that the infant mortality rate for the poorest quintile of the population was almost five times larger than that of the richest quintile. Another study of Latin American indigenous peoples – disproportionately among the poor in every Latin American country – found that they had significantly higher mortality and morbidity rates than their non-indigenous counterparts (Montenegro & Stephens, 2006). Mexican men aged 50 and older from higher income brackets and socioeconomic backgrounds were reported to have better health than men from the same age group of lower income and socioeconomic positions (Smith & Goldman, 2007).

There is also a spatial dimension to income and health inequalities in Latin America, as Rojas (1998) and Orihuela-Egoavil (1993) noted. The Mexican municipality in the lowest decile of per capita income had an average of 2 hospital beds per 10,000 individuals between 1990 and 1996, as compared with 15 hospital beds per 10,000 individuals for the municipality in the highest decile of per capita income (Casas et al., 2001). Astraín, del Carmen Pría, and Ramos’ (1998) study of differences in living conditions and mortality in the Cuban province of Camagüey found high mortality rates among the poorest areas of the province. Szwarcwald, Bastos, Viacava, and de Andrade (1999) showed that in Rio de Janeiro the highest homicide rates are found in municipalities with the greatest income inequality. Frank and Finch (2004) observed the importance of geographic and socioeconomic location in infant mortality rates in Mexico from 1986 to 1996, finding that poorer areas have higher infant mortality rates than more affluent areas. Consistent with evidence in high income countries, the descriptive research on Latin America indicates that both impoverished and highly unequal regions display greater health disadvantage.

Such evidence corroborates recent findings from developing countries. Salti (2010) found in South Africa that relative deprivation was a key predictor of mortality. Kondo et al. (2009) published their own meta-analysis on studies of inequality and health, including data from developing nations such as Chile and China. Consistent with the findings of Wilkinson and Pickett (2006), the authors found evidence linking income inequality with adverse health outcomes.

Methodological and empirical criticism of evidence linking income inequality and public health has emerged in recent years. Beckfield (2004) reproduced past studies with more data and found significant income inequality effects on infant mortality and life expectancy detected under ordinary least squares regression either disappeared or became markedly weaker under fixed effects regression, which controls for unobserved heterogeneity, or country-specific factors that could distort results. Similarly, Mellor and Milyo, 2001, Mellor and Milyo, 2003 challenged evidence of the detrimental effects of income inequality on health in the United States by adding a variety of controls to their analysis and finding that income inequality has no significant effect on health. Pearce and Davey Smith (2003) presented examples where both income inequality and public health have been on the rise concurrently, such as New Zealand. In their analysis of Sweden, Gerdtham and Johannesson (2004) found evidence supporting the claim that increases in income improve health but none confirming that income inequality is detrimental to health. Finally, Lynch, Davey Smith, Harper, Hillemeier, Ross et al.’s (2004) and Lynch, Davey Smith, Harper, and Hillemeier’s (2004) comprehensive studies found that, with the minor exception of a few studies of the United States, evidence showing a negative association between income inequality and public health is sparse and inconsistent. Deaton’s (2003) comprehensive analysis concluded that “it is not true that income inequality itself is a major determinant of population health” (p. 151).

Despite the attention devoted to the relationship of GDP, poverty and inequality to public health, we are aware of no studies that examine whether the positive effects of economic development on health are modified by inequality or poverty. Does the impact of GDP on health depend on how it is distributed? That is, does the GDP-health association vary with levels of and changes in inequality and poverty? Pritchett and Summers (1996) claim that “wealthier is healthier,” but is this always the case? It is plausible that the public health gains associated with increasing wealth crucially depend how that wealth is distributed, especially in view of evidence that the income has greater effects on health among resource-poor groups and countries.

Previous studies in this area have been primarily concerned with the developed world and have utilized cross-sectional data. Studies of developing countries that use longitudinal data rarely include the number of years this study considers (covering up to 513 country-years), relying on much smaller sample sizes of highly aggregated data, sometimes as few as 10–15 data points. This study, then, is unique in that it focuses on a less developed region of the world and employs longitudinal data and time series regression methods to examine the relationship between economic variables and public health over time, while also considering the interaction between national income level, poverty, and inequality.

The principle statistical tool used in this analysis is fixed effects regression modelling, or a ‘within-country’ analysis, following the approach and recommendation of Beckfield (2004). GDP, poverty, and inequality are separately regressed on infant mortality, life expectancy, and tuberculosis (TB) mortality. Infant mortality and life expectancy are standard measures of public health and economic development. We use TB mortality as an example of infectious disease that is prevalent in Latin America that should disappear as countries become wealthier and undergo the epidemiological transition, and also because TB has long been viewed as an indicator of societal health (Stuckler, King, & Basu, 2008). Next, to test whether the wealth–health association depends on how income is distributed, GDP is regressed on each health variable, but this time the sample is stratified into periods of increasing, decreasing, or decreasing or constant inequality and poverty, providing a test of effect modification.