Giuseppe RUSSO | Progetti

INEQUALITY OF OPPORTUNITY IN HEALTH

This project contributes to the literature by proposing a decomposition-based approach to measure inequality in objective health that captures Roemer’s distinction between circumstances and effort. The method fully conditions on circumstances by splitting the sample according to type and then estimating separate regressions of health outcomes on effort for each sub-sample. This non-parametric approach allows the model to be fully saturated in the way that it handles the circumstance variables. Using linear regression at this step generates a heterogeneous set of regression coefficients that it is used in a regression-based decomposition of total inequality in the biomarkers. Note that this does not require additive separability of circumstances and effort and allows interactions between them (through heterogeneous slopes) which is relevant for the assessment of the “fairness gap” in the spirit of Fleurbay and Schokkaert (2009, 2012). To retrieve the relative contribution of circumstances and effort to the total inequality, a decomposition of the Gini coefficient with heterogeneous responses proposed by Jones and Lopez-Nicolas (2006) is employed and an extension of this method is developed to complement the standard Gini with an Inequality of Opportunity Gini that measures inequality relative to the most disadvantaged type, in the spirit of the “fairness gap” principle. The decomposition method identifies five normatively-relevant decomposition terms: a direct and an indirect (through effort ) contribution of circumstances to the total inequality, the contributions of within and between-type variation in effort to the total inequality and the contribution of randomness and luck. The between-type term is new in the equality of opportunity literature and it takes into account the contribution of the systematic variation in efforts by types on the overall inequality in health. This might be relevant in order to distinguish a “pure individual” responsibility from a “group responsibility” arising, for instance, by social contagion or social norms An original element of the empirical model is the use of biomarkers as outcome variables and as proxies of relevant effort variables such as smoking, diet and physical activity. Biomarkers are characteristics that are ‘objectively measured and evaluated as indicators of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention’ . They are measured on a continuous scale associated with an increasing or decreasing risk (depending on the biomarker) of a disease state and they are often highly correlated with mortality (Rosero-Bixby and Dow, 2012; Sattar et al., 2009; Gruenewald et al., 2006). A key advantage of using biomarker data is having a measure of health which is free of reporting bias. This is particularly relevant given the possible presence of systematic reporting behaviour across individuals sharing the same set of circumstances. Indeed, previous empirical investigations show the presence of a systematic variation in reporting behaviour across socio-economic groups (e.g., Sen, 2002) which may bias the estimates of the equality of opportunity in health in a significant way.