The education–suicide mortality gradient

ABSTRACT

Using the fifth release of the National Longitudinal Mortality Survey, I examine the role of educational attainment and self-reported health on 6- and 11-year suicide mortality risk in the United States. I first replicate the original results reported by Hamermesh and Soss. . Then, augmenting the Hamermesh model with initial educational attainment and self-reported health status, I find that years of education significantly raises suicide mortality risk in the US after controlling for initial self-reported health. This result is robust to regression specification, replication and the inclusion of covariates.

KEYWORDS:

• Suicide
• education
• health
• mortality
• human capital

JEL CLASSIFICATION:

• I12
• I21
• C21

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ https://www.cdc.gov/injury/images/lc-charts/leading_causes_of_death_age_group_2014_1050w760h.gif.

² In cross-country analysis, Noh (2009) finds suicide is only positively related to unemployment in high GDP per capita countries.

³ The 11-year follow-up sample (see appendix A.1) also demonstrates this behaviour.

⁴ $suicid{e}_{i,t+k}=\alpha +\beta incom{e}_{i,t}+\gamma employmen{t}_{i,t}+\zeta ag{e}_{i,t}+{\epsilon }_i$.

⁵ The observed behaviour is unlikely the result of sampling bias since subsample sizes for respondents in ‘poor health’ are large – 25,533 in the 6-year sample and 13,369 in the 11-year sample and summary statistics are similar.

⁶ This phenomenon is concentrated within the middle-aged (45–64 y/o) male subpopulation, consistent with Case and Deaton (2017).

⁷ For example, using Specification 3, 6y sample: high school graduates have suicide risk 0.00058, some college 0.00061, college graduates 0.00069 (p < 0.01 for all estimates).

⁸ CPS technical documentation attachment 17: https://www2.census.gov/programs-surveys/cps/techdocs/cpsoct93.pdf.

⁹ Rising ‘perceived burdensomeness’ is evidenced by significant interactions between education and age, unemployment and being disabled/unable to work – possibly indicating the more educated are more profoundly affected by negative life events than the lesser educated.

¹⁰ However, in China and the US, Li, Li, and Cao (2012); Nock and Kessler (2006), respectively, find less education is, on average, connected with higher likelihood of suicide attempts.