Urban Bias in Latin American Life Expectancy
Jenny Garcia, Institut National d''études Démographiques INED
When mortality study takes in consideration different countries, the most traditional approach is through national context, since undoubtedly intervention policies are often inherent in national states. However, analysis of synthetic mortality indicators and indexes at the national level as a whole could obscure differentials relating to particular populations, especially where disparity takes an important place as in Latin America. Analyzing, gaps in living conditions, many studies of Latin American countries have highlighted their unequal development along spatial lines, due to a high concentration of goods and services in cities that have left behind rural areas. This is not in fact a new idea; it was in 1977 that Michael Lipton introduced the Urban Bias Thesis as a framework to understand economic and social singularities in developing countries. Following this overall approach, this research proposal contemplates to answer: Is there a differential in Latin American life expectancy patterns during the first decade of the twenty century when considering urban bias? To answer this question, data from 2000 and 2010 from seven Latin American countries have been selected: Brazil, Chile, Colombia, Ecuador, Mexico, Peru and Venezuela.
Data and Methods
Selected countries represent diverse situations in terms of life expectancy at birth and proportion of urban population. To simplify analysis and to assure comparability, definition of urban area has been attached to city. In this way “urban” is recognized as a continuum category instead of a dichotomous concept. Hence, three types of cities are recognizable: main cities (more than 500,000 inhabitants), medium cities (20,000 to 499,999 inhabitants), and small cities and purely rural areas together (less than 20,000 inhabitants).
Mortality data comes directly from health secretaries (Brazil, Mexico, Peru and Venezuela) or National Statistics Institutes (Chile, Colombia and Ecuador). While population estimations come from national statistics offices in each country. Rates are matched clustering the smaller administrative units captured in both databases that resemble cities. In this case it is the equivalent of a county level. In summary, 10,156 counties from eight countries are split into the three city-size groups mentioned before.
Due to ensuring a standard of data quality is priority, two major issues are taken into consideration on mortality data: coverage errors (underreported deaths) and quality errors (age misreporting). Coverage errors are determined and corrected using a combination of death distribution methods for adult mortality and indirect infant mortality methods (children ever born), both via population censuses. Annual specific mortality rates are estimate using a p-splines model for Poisson distribution.
Along with annual rate, life tables are computed for the three city-size clusters in each country to get life expectancy at all age. Furthermore, life expectancy decomposition technics are applied, specifically Andreev’s method (1982), both at 2000 and 2010 data (Exceptionally, due to data availability Peruvian data analysis it is done from 2003 to 2013). Results come as differentials found by sex, age and causes of deaths between the city-size groups and through time.
When considering all countries together, it does not seem to be any sex differential regarding cities size. For both sex, it is in two age groups in which internal differential in life expectancy is concentrated: infants (less than 1 years old) and older ages (65+). So, it is the advantage of these two groups in the main cities that contribute the most to the gaps. A third age group pops up in small cities and rural areas by its disadvantage, the adult group 35 to 64 years.
Single country analysis shows a close relation between city-size group gap and the level of life expectancy achieve nationally. The biggest the manifested gap the lowest would be life expectancy in the country. Even taking preliminaries results, the three countries identified as those with the highest levels of urbanization (Chile and Venezuela) show the smallest gaps in life expectancy at all age, especially at birth. In contrast Brazil, which urbanization levels are still high but manifests low life expectancy at birth, keeps gaps between cities similar to Ecuador and Peru.
Mexico and Colombia are the ones showing more particular patterns by age groups. Colombia looks to have a wider gap between main and medium cities, in which medium cities’ life expectancy is closer to small cities and rural areas. Moreover, infant mortality does not make big contribution to the differences compare with other age groups. On the other hand, Mexico is the only country with high levels of life expectancy at birth to show gap between cities but without any clear age group preference.
Further analysis is needed, especially on life expectancy trends at all age, nevertheless a relation between gaps introduce by urban bias in life expectancy is recognizable. Data from 2010 will be incorporated in coming months, in order to establish trends.
Presented in Session 1189: Mortality and Longevity