Settle in folks, because it's time for a tweetstorm about my new @nberpubs paper with @MeltemDaysal, Todd Elder, Judy Hellerstein, and Chiara Orsini. 1/25

https://www.nber.org/papers/w28652?utm_campaign=ntwh&utm_medium=email&utm_source=ntwg25

This paper was inspired by work by @sbaroncohen and co-authors about how #autism could be an heritable, extreme realization of certain parental traits - specifically systemizing & lack of empathizing. 2/25

We tie this to the burgeoning literature in econ on social skills (e.g. work by @ProfDavidDeming). To do this, using data from Denmark, we infer parental skills and traits from their occupations via O*Net. 3/25

A principal factor analysis reveals three primary skills which we dub 1) professional, 2) systems and ordering, and 3) communications. The latter two align well with systemizing and empathizing, respectively. 4/25

Communications skills also align with social skills as defined in Deming (2017; https://academic.oup.com/qje/article/132/4/1593/3861633?login=true). For completeness, we also look at other skills in his paper including routineness, service skills, and non-routine analytical skills. 5/25

In general, occupations line up as we would ex-ante expect. For example more service-oriented fields (e.g. police, child-care workers) are stronger on communications, while more analytical fields line up with systems/ordering (e.g. air traffic controllers, doctors). 6/25

We estimate the relationship b/w parents' skills and the likelihood of having a child w/ Autism Spectrum Disorder (ASD) while controlling for earnings history, location of birth, & other factors to test theories of how parental traits (proxied by job skills) may lead to ASD. 7/25

We then go on to see whether there is assortative mating on these skills (yes!) and whether that assortative mating could explain large increases in ASD (doubling from 0.6% to 1.3% in 15 years) in Denmark during our sample period (no!) 8/25

Finding 1: 9/25

Fathers with more professional skills (essentially salaried occupations) show no statistically significant change in children's ASD diagnoses, though for mothers it is negative. 10/25

For systems/ordering (systemizing) and communications (empathizing) skills results for fathers are consistent with @sbaroncohen theories that more systemizing leads to more ASD while more empathizing leads to less. Estimates are modest but statistically significant. 11/25

Finding 2: We try to more explicitly test the Empathizing-Systemizing (formerly Extreme Brain) theory and find some evidence that ASD rates are larger in both (empathizing and systemizing) extremes, unlike theory that suggests it would only be latter. 12/25

Even so our results that a child with two extreme systemizing (systems/ordering minus comms skills >= 2 SD) is correlated with ASD rates 35% higher relative to having both non-extreme ("balanced") parents. 13/25

Finding 3: What about skills in Deming (2017)? Substantial corrs for dads but not much for moms. Children w/ ASD are more likely to have fathers in jobs w/ more routine tasks (6% for 1 SD increase) & less likely for jobs needing more social skills (10% for 1 SD increase). 14/25

One hypothesis made by Baron-Cohen is that assortative mating on these skills could explain some of the large increase in ASD over the last quarter-century. In our context, our conclusion is that this is very unlikely for a few reasons. 15/25

First, we estimate interaction effects across parents for our skill measures and find little to indicate that the pairing of higher skilled parents makes an impact beyond the simple additive effect. 16/25

Second, while we see clear evidence of assortative mating on skills in the cross section 17/25

we find little to indicate that assortative mating is changing over time in a way that would increase ASD (if anything it would decrease it) 18/25

Back of the envelope calculation indicates that our estimates (which admittedly have a lot of measurement error due to the nature of the proxy variable) would have to be off by two orders of magnitude to even explain 1/10 of the increase in ASD we see during these cohorts. 19/25

To conclude, we find (in our minds) pretty compelling evidence that parental traits are related to child ASD but mixed support for existing theories about these relationships. We further show it is unlikely that assortative mating on these traits explains ASD increases. 20/25

This paper has been a long journey. We started working on it almost a decade (!) ago. But it's been a labor of love (given my personal experience with an #autistic child) and a pleasure working with such fantastic co-authors! 21/25

And while you're here if you are interested in more along this broad line of research check out these other papers I've done on #autism and #disabled children. (ungated versions can be found on my website https://sites.google.com/view/imberman/home): 22/25

Elder, T., Figlio, D., Imberman, S., & Persico, C. (2020). The role of neonatal health in the incidence of childhood disability. American Journal of Health Economics, 6(2), 216-250. 23/25
https://www.journals.uchicago.edu/doi/abs/10.1086/707833?casa_token=QRKTTf5HD0YAAAAA:IlfHZUpoc52k80oHWoxqmQWksOV09XK5sWtBFFLpZK5XtVB3C85mgW-QrCInjMXVOBtqvuLHrGC5

Elder, T. E., Figlio, D. N., Imberman, S. A., & Persico, C. L. (2021). School Segregation and Racial Gaps in Special Education Identification. Journal of Labor Economics, 39(S1), S151-S197. 24/25

https://www.journals.uchicago.edu/doi/full/10.1086/711421?casa_token=IBkHGDHWuzkAAAAA:C5qs5Lcrz5quY-Na-wwhRWsQXsGhF5-b_qvVN0lFBvfmr6RbDTxKGx_fpkb1JLMwZMGHQov7ahQ3

Acton, R. K., Imberman, S. A., & Lovenheim, M. F. (2019). Do Health Insurance Mandates Spillover to Education? Evidence from Michigan's Autism Insurance Mandate (No. w26079). National Bureau of Economic Research. 25/25 https://www.nber.org/papers/w26079