Example of Policy Questions:
-how much would one additional year of schooling(x) enable young adults to increase earnings (y)?
-What would the impact be of reducing class size(x) be on students learning outcomes(y) (measured by test scores)
In recent years there has been a big movement to use natural and controlled experiments to (evidence-based evaluations)
Cross-sectional evidence: compare across countries, across regions, across individuals
Time-series/longitudal evidence: take a specific unit of observation (country, region, and person) observed at different points in time
Why Correlation does not indicate causation:
Omitted variables/selections bias: there may be third factors (z) that caused x and y
Endogeneity/ Reverse causality: causation may run from y to x
Examples: Return to schooling
High positive correlation between schooling and income could be due to unobserved ability/effort/background (omitted variable)
High negative correlation between learning and class size could be due to effect of student performance on school budgets/enrollment (reverse causality)
Oct 30, 2014
Health Policy and Development:
What is the evidence for effects of health on productivity and incomes?
Is there evidence of effectiveness of specific health interventions?
Measuring health and nutrition is tricky bias/errorrelies on anthropometric measures such as height and weight -height and weight depend on early childhood nutrition, energy expanded, etc
Hypothesis To Be tested:
Health and nutrition affects work productivity affects wages
Health and nutrition affects