discussion on inequality of opportunity in adult health in colombia
TRANSCRIPT
XX Latin American and Caribbean Economic Association (LACEA) Annual
Meetings
XXIII Annual Meetings of the LACEA / IADB / WB/ UNDP Research
Network on Inequality and Poverty (NIP)
Inequality of Opportunity in Adult Health in Colombia
By Johanna Fajardo-Gonzalez, University of Minnesota
Discussant: Tito Armando Velasco, ARU Foundation
Santa Cruz de la Sierra
October 14, 2015
Discussion remarks
Introduction
Overview of the study
Empirical analysis: methodological approach, variable, sample,
measures
Usefulness of ex-ante approach for policy design
Introduction
Contribution to “flourishing empirical research field based on
advanced analysis of detailed data from individual households”
Significance of this type of research in development economics
recently highlighted by the Royal Swedish Academy of Sciences in
its recognition to the new Laureate in economics, Angus Deaton
(Sveriges Riksbank Prize)
Introduction
Contribution to the equality of opportunity literature in developing countries by
providing relevant evidence in health sector of Colombia that partitions inequalities
into inequalities in circumstances. They are supposed to be subject of policies
oriented to “equalize opportunities rather than outcomes”
These unjustifiable inequalities in the words of Ravi Kanbur are the very same
inequalities in circumstances discussed by the author (while justifiable inequalities
would be inequalities in efforts to which she refers)
Overview
Study clearly in line with the work of John Roemer (1998), Ricardo Paes de Barros
(2009) and other followers of the ex-ante approach adopted to identify given
circumstances beyond individual control that must be compensated by policies in
order to achieve equality of opportunity in specific outcomes
Main findings provide indication that early life circumstances, related to household
socioeconomic status and parental educational attainment (urban areas),
influence equality of opportunity in adult health status in Colombia
Overview
Scope of this cross-sectional study for 2010 is confined to estimate two measures of inequality of
opportunity (dissimilarity and a Gini-opportunity indexes) in adult health and the relative contribution of a
number of circumstances (in the variation in the dissimilarity index using Shapley value decomposition)
It uses a self-reported variable for health status reported in 2010 Colombian Living Standards and Social
Mobility Survey (LSSM)
Not a causal analysis that identifies determinants (early life circumstances) of adult health outcomes; but
rather a correlation study that stablishes the relationship (association) between early life circumstances
and adult health status, and identifies as well if the circumstances have direct or indirect effects on the
outcome of interest
Clear research question in that regard: “Which particular early life circumstances have a salient long-term
association with observed inequality of opportunity?”
Therefore, today’s discussion may be oriented on either:
Pertinence and robustness of the empirical methods applied on the data set, or
Usefulness of the embraced ex-ante approach for policy design purposes
Empirical analysis
Methodological approach
While author acknowledges some disadvantages of the stochastic dominance analysis (in
terms of “splitting the sample into many different groups, and the difficulty to implement a
test with multiple circumstances simultaneously”), its application on a cross section dataset
is risky in terms of confidence
Still, methodological approach presents more relevant disadvantages than the author
recognizes. E.g., the resulting measures depend on normative judgments that usually are
conflictive. They end up being very sensitive in inequality of opportunity. Nevertheless the
author uses state of the art techniques, such as non-parametric tests, in order to maximize
the confidence of results, but the doubt still persists
Empirical analysis
Self-reported variable for health status
Author points out the presence of psychological issues in the manner that individuals
perceive and report their health status
Psychological issues related not only to adjustment and adaptation in health perceptions but
possibly as well to memory, confidence and corruption. Obviously not available data about
these issues in the LSSM database, but maybe inequality of opportunities is correlated to
institutional crisis, corruption practices in services, etc.. In this regard, new experimental
economics and neuro economics tools may be useful to consider to strengthen this study
Empirical analysis
Survey sample
Dynamics of inequality tend to be different in urban and rural sub samples
Recent evidence shows that there is a lack of research around suburban zones
Estimates using urban sub samples may shed some more light on the issue of
inequality of opportunity, particularly in barrios periféricos of Colombia. If LSSM data
and sample allow, consider to try to come up with a criteria to formulate at least
indirect evidence about suburban zones
Empirical analysis
Direct measures of inequality of opportunity
Author sustains “the calculation of the dissimilarity index first requires the
estimation of a logistic regression model to obtain the predicted probability of
achieving a good or excellent health status”
However, paper doesn’t demonstrate why logistic regression is better than probit
regression model or other categorical type model. “Is it because the log of the
odds can be explained far more easily than probit? Or is it related to the purpose of
the model (e.g., predicting an outcome vs. estimating an effect)?”
Empirical analysis
Direct measures of inequality of opportunity
Dissimilarity and a Gini-opportunity indexes. Author suspects it is necessary more
evidence to support her research results
Consider micro simulations obtained from the sample data in order to find robust
results, particularly for the Gini-opportunity index. Explore how sensitive are
measures to a slight variations in the data. Moreover, explore how sensitive are them
to a different objective functions among rural and urban individuals
Usefulness of the ex ante approach for policy design
Recall the work of last year of Ravi Kanbur and Adam Wagstaff (2014), who raised critics
to the inequality of opportunity approach, particularly in education and health sectors.
Surprisingly, relevant critics were not addressed by the author. Four recall points for our
discussion:
“Implicit variable weighting leads to different measures of inequality of
opportunity”. It would be good to explicit and highlight the weighing method approach
“Adding more circumstance variables and categories within them will reveal a
different dissimilarity index”. If so, potential problem when comparing measures
over time and identifying adequate policies oriented to compensate unjustified
inequalities (inequalities in circumstances)
Usefulness of the ex ante approach for policy design
In many cases “normative measure of inequality of opportunity is determined by
casual properties of data availability”
Two fundamental issues of principle that questions the inequality of opportunity
approach:
“What happens when luck determines difference between opportunity and outcomes?”
“What happens when one person’s effort becomes another person’s circumstance?”)
Thank you