25 Minute Version of
Angus Deaton’s
Inequalities in Income & Inequalities in Health for
Public Finance II Spring 08 University of Texas, Austin Nick Mastronardi
Background • Working paper 1999 • For conference Increasing Inequality in America - Inequality Increasing? - What kinds? - Effects?
Motivation • Applied Welfare Economics – Income, Consumption = usual components – More than goods & services? • If so, what else • Relationships between factors
• Health Effects Welfare – Correlated with Income – Not perfect
Motivation II • UN Human Development Index – Health (Life Expectancy) – Income – Education (Literacy)
• European Region of WHO – Reduce Health Inequality by 25%
• Economic Policy is Public Health Policy
Abstract • Measure for Inequality in Health? • Standard Income Inequality Tools? • Relationship – Health, Health Inequality, Income, Income Inequality?
Overview • Def of Health Inequality – Mortality – Life Expectancy
• Relationship w Income Inequality • Propose Model st – Individual Health Depends on Individual Income relative to Group Mean Income
• Testable Implications
Result For the Proposed Model • Income Inequality Effect Average Health – No Significant Effect
• Income Effect Health – Gradient • if sig & pos, then income is protective
– Depends on Ratio • Between Group Income Inequality • Within Group Income Inequality
II. Lit Review • Wilkinson (1996) – Income Inequality threatens Health – Theoretical Underpinnings? – Challenge Hypothesis
Lit Review (cont) • Measuring Health Inequality – Le Grand (1987) • Years lived (lag, quality) • 50, 49, 45 paradox • Transfers for life?
Lit Review (cont ii) • Adler et al (1993) – Access to Health Care not effect Mortality – What does? Income Inequality? – If so, how?
Lit Review (cont iii) • Preston & Milo – Education Effect Health (Mortality) • Machinery work through Income 𝑦𝑖 − 𝑦 = 𝛽 𝑠𝑖 − 𝑠 + 𝜀𝑖 𝐻𝑖 − 𝐻 = 𝛼 𝑦𝑖 − 𝑦 + 𝜇𝑖
– Health Hazards • Surgeon General & Teachers
III. Benchmark Empirics • Aggregate Income Inequality and Individual Health = Bad – Not robust • Inequality Measures • Aggregation Fallacies • Controls – race, …
• Data – Industrialized Nations, Acceptable Quality – World Bank, World Development Indicators – Gini Coefficients after-tax Income, Life Expectancy • No Significant Correlation (pg 20)
IV. Model • Wilkinson Hypothesis – Income Inequality Effect Mean Health – Mechanism? (AZ Prop Taxes, Man in NY)
• Deaton 𝐻 = 𝛼 + 𝛽(𝑦 − 𝜇) 𝐸 𝐻 𝑦 = 𝛼 + 𝛽(𝑦 − 𝐸 𝜇 𝑦 )
Model cont 𝑦|𝜇~𝑁(𝜇, 𝜎𝜀2 ) 𝜎𝜀2 ≡ 𝑚𝑒𝑎𝑠𝑢𝑟𝑒 𝑜𝑓 𝑖𝑛𝑒𝑞𝑢𝑎𝑙𝑖𝑡𝑦
𝜇~𝑁 𝜇, 𝜎𝜇2 𝜎𝜇2 ≡ 𝑚𝑒𝑎𝑠𝑢𝑟𝑒 𝑜𝑓 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑔𝑟𝑜𝑢𝑝 𝑖𝑛𝑒𝑞𝑢𝑎𝑙𝑖𝑡𝑦 ∴ 𝐸 𝜇 𝑦 = 𝑎 + 𝑏𝑦
&
𝜇 =𝑦−𝜀
𝛽𝜎𝜀2 𝐸 𝐻 𝑦 =𝛼+ 2 (𝑦 − 𝜇) 𝜎𝜀 + 𝜎𝜇2 𝛽𝜎𝜀2 𝐸 𝐻 𝑦 =𝛼+ 2 𝑦 − 𝜇 − 𝛽𝜃𝜎𝜀 𝜎𝜀 + 𝜎𝜇2
Empirics cont • Mortality and Income Inequality – – – –
Countries with lots of good data 1981-1993 following Birth Cohorts March CPS, Berkeley Mortality Database More micro, but still problematic • Action at a distance • Cohort = head of household • Inequality Measure = Variance
– Mortality Rates increase w age – Inequality increase w age
Empirics cont ii • Condition on Cohorts Age in Population – Need better age-profile of mortality • Could identify effects of Income Inequality better
• Treat Men separate from Women • Instrument for Permanent Income • Significantly more protective
V. Data Analysis • Rising Inequality lowers Mortality Rates – Actually, not significant • Likely time lags? • Cumulative Exposure to Inequality?
• Instrument for Permanent Income • Significantly more protective 𝛽𝜎𝜀2 𝐸 𝐻 𝑦 =𝛼+ 2 𝑦 − 𝜇 − 𝛽𝜃𝜎𝜀 2 𝜎𝜀 + 𝜎𝜇
VI. Caveats • Endogeneity • Inequality Measures • Controls
VII. Summary • Angus Deaton – “Little Support for the Idea that Inequality is a Health Hazard at the National Level” • Income Inequality on Health: No Effect
– “There is a Well-Defined Health Gradient” • Income on Health: Significant, Positive • Gradient Depends on Income Inequality – Group with High Dispersion, then High Returns to Income – Income is Protective of Health
Questions
Nick comments •
Govt to Decrease Inequality, but what kind? – Income • •
which metric General but high moral hazard consequences –
Distortion of labor supply
– Health • •
how, not accessibility More direct, Less moral hazard –
•
Still lack of inceentive to invest in health
less short-term effectiveness
– Education • • •
Information diffusion Less direct, least moral hazard Longest term solution
– …
•
Optimal commodity of Redistribution by Government = Information. – Contains all the answers agents could ever want to achieve any goals they would have. – It provides the keys to vast warehouse of tools to serve the needs of all individuals desiring to cultivate their own potential. – Procurement of library services for public use, Subsidies and at least mild de-tiering regulation of internet, subsidies of information Databasing Firms, Librarians and search cost routines. •
Google Medical Records