Income Inequality and Mortality in Canada and the USA

Income Inequality and Mortality in Canada and the United States Nancy Ross Statistics Canada The “Big Idea” • Wealthier is healthier for individuals • But, despite more wealth, the richest nations are not the healthiest • Measures of income distribution do a better job accounting for variations between industrialized countries in life expectancy and mortality (controlling for GNP per capita) (Rodgers 1979, Wilkinson 1992, Duleep 1995) How Can Inequality be Related to Mortality? • Indirectly via underinvestment in human resources – low spending on public goods/infrastructure – poor social relationships (weak social capital) • Directly via one‟s appraisal of place in the social order – social position affords an increased „host defence‟ (Whitehall) – hassles and frustrations of daily living more likely to become pathological when assessed against others‟ more enviable situations The U.S. - Canada Comparison • U.S. income and mortality data come from Kaplan et al., 1996 (states) and Lynch et al., 1998 (metropolitan areas). • Canadian data from the 1991 census. Mortality data (1990-92) from Statistics Canada‟s National Mortality Data Base. • Inequality measured as the proportion of total household income (post-transfer but pretax) going to least well-off half - “ the median share” Working-Aged Male (25-64) Mortality by Median Share U.S. States and Canadian Provinces 800 U.S. States with weighted linear fit (from Kaplan et al., 1996) Canadian Provinces with weighted linear fit (slope not significant) Rate per 100,000 Population MS 675 LA SC AL FL 550 TX CA PEI 425 MN SASK QUE NH NS NB NFLD MAN BC ALTA ONT Mortality Rates Standardized to the Canadian Population in 1991 300 0.18 0.20 0.22 0.24 WAMWeightedCan&US June 16, 1999 2:40:26 PM Median Share of Income Working Age (25-64) Mortality by Median Share U.S. and Canadian Metropolitan Areas 600 NewOrleansLA AugustaGA U.S. cities (n=282) with weighted linear fit (from Lynch et al. 1998) Canadian cities with weighted linear fit (n=53) (slope not significant) FlorenceSC Rate per 100,000 Population PineBluffAR 500 NewYorkNY MonroeLA 400 BryanTX LosAngelesCA Shawinigan SiouxCityIA PortsmouthNH 300 McallenTX Montreal Vancouver Toronto AppletonWI Barrie Oshawa Mortality Rates Standardized to the Canadian Popluation in 1991 200 0.15 0.19 0.23 Median Share of Income 0.27 WAWeightedCan&US June 29, 1999 12:53:18 PM Metropolitan Area Regression Results Age Group Model Intercept Median Share (%) -19.73 -19.35 -18.18 -2.49 -2.43 -2.26 -21.71 -21.80 -20.92 -20.58 -18.03 -14.16 -15.09 -14.82 -13.84 Median Income* (in thousands of dollars) Country Dummy Adjusted R2 Infants US only US only US & CAN w/ dummy 1341 1386 1358 110 116 113 848 838 826 5255 5547 5490 1110 1141 1127 -1.6 -1.5 -0.20 -0.30 0.40 0.20 -10.50 -11.20 -1.10 -1.30 -280 -18 -67 -399 -91 .03 .02 .27 .11 .11 .35 .33 .34 .51 .01 .03 .16 .13 .12 .34 Children and Youth WorkingAge Elderly US only US only US & CAN w/ dummy US only US only US & CAN w/ dummy US only US only US & CAN w/ dummy All Ages US Only US Only US & CAN w/ dummy Conclusions • The relationship may be nonlinear or it‟s not universal • Income a stronger determinant of life chances in the U.S. than in Canada • Likely that a more equitable distribution of social resources in Canada protects population health Next Steps in the Research • Analysis of earned income inequality • Analysis of urban residential economic segregation and its relationship to health outcomes • Development of an educational disparity index