Causalidade entre renda e saúde: uma análise através da abordagem de dados em painel com os estados e os municípios brasileiros

Abstract

The income and life expectancy increase and also, poverty and mortality rate reduction, indicate an improvement of social welfare. Therefore, to understand the relation between income and health is considered to be of fundamental importance. In the theoretical literature, such as, Sala-i-Martin (2005), Weil (2005) and Chen (2008), the causality relationship between income and health is presented as bidirectional. This dissertation has as main objective to analyze causality relationship between income and health, seeking to control the potential differences of this relation over the Brazilian territory. In this case, three Granger causality tests to panel data, proposed respectively by Holtz-Eakin, Newey and Rosen (1988), Granger and Huang (1997), and Hurlin and Venet (2004) and Hurlin (2004, 2005), are applied to a Brazilian States database in the period from 1981-2007. The first two approaches are applied to a database with the counties in Brazil in the period of 1970-2000. To the Brazilian states, the results of Holtz-Eakin, Newey and Rosen (1988) method shows bidirectional causality for complete sample (Brazil), for the group of states with the highest incomes (South-Central) and for the group of states with lower income (North – Northeast). The results of Granger and Huang (1997) test shows unilateral causality from income to health for Brazil, unilateral causality from income to health for the south-central States and non-causality relationship between income and health for the North-Northeast state´s group. Yet, the proposed test by Hurlin and Venet (2004), and Hurlin (2004, 2005) the evidences are clearer for the causality, in a way from health to income for the three cases examined. In general, either the full sample or the division by regions and income groups, the results of the two tests applied to the database with the counties in Brazil, show evidence of a bi-causal relationship between income and health. However, the results presented here are not all consensual.