Spatial and economic patterns in life expectancy among US States

ABSTRACT

This study investigates the determinants of gender-specific life expectancy across US states over the period 1995–2007. We employ a production function specification where life expectancy depends on health expenditure, income, education and lifestyle variables, allowing for spatial effects. Empirical results suggest that education attainment and health expenditure are the main factors behind improving longevity, whereas smoking bears a strong negative influence. For robustness purposes, we also use health spending as well as education criteria, apart from geographical ones to model interstate spillovers. In the former case, states with similar health expenditure are ‘neighbors’ and affect positively the life expectancy process. If education is applied instead of health spending together with geographic proximity, the spatial correlation is insignificant, i.e. education ‘neighbors’ do not affect life expectancy. Our findings do not imply significant gender differences regarding health production. The results suggest that health care policy will have to focus on wider economic and social considerations, like education and lifestyle changes, except medical care provision in order to exploit the full potential for life expectancy improvements of the US population.

KEYWORDS:

• Life expectancy
• health spending
• state level
• education
• spatial model

JEL CLASSIFICATION:

• I12
• H51
• H70
• C31
• C33

Acknowledgments

We would like to thank V. Sarafidis for useful comments and suggestions and seminar participants in DePaul University, University of Exeter and University of Illinois at Urbana-Champaign. The usual disclaimer applies.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The applicability of this CD test along with other similar tests is discussed in Hoyos and Sarafidis (2006).