Lecture 4- Mental Health in Childhood and Human Capital (Currie and Stabile,2009) Flashcards
What is the main goal of the paper?
To study the long-term effect that certain mental disorders in childhood have on future outcomes such as education and delinquency
What were the contributions of this paper to existing literature?
They explored to see if the effect of mental disorder as they researched how a child’s long term outcomes differs for families with differing levels of income or based on the level of the mother’s education.
They use “screener “ questions that they asked to all children in the sample not just those with mental disorder.
They found that mental disorders have a large negative effect on education outcomes.
Describe the data-set
Many surveys were conducted on a nationally representative large sample in Canada of over 22,000 children with their family. The surveys started from 1994 and were bi-annual until 2002.
This was called the NLSCY
( National Longitudinal Survey of Children and Youth)
They also looked at other survey data in the US from NLSY that surveyed 6000 young men and 6000 young women from the ages of 14 to 21. This began in 1979
What was the main regression?
Oucome I= a0+BMental Health 94+ LamdaX94 ei
Mental 94 is a vector of an index of scoring of 3 child mental disorders in 1994
X is a vector of background characteristics (child age, income level, sex)
This OLS model estimates the effect that mental health in 1994 has on future outcomes
What was the alternate regression?
Outcome i f= a+BMental94 i f +LambdaZ94 i f+uf +e i f
Why was the first regression : Oucome I= a0+BMental Health 94+ LamdaX94 +ei
inadequate?
And how was the alternate Outcome I f= a +BMental94 f +LambdaZ94f+ uf + e f
It assumes that your mental health outcome in 1994 effects your outcomes today, we can’t assumes causality as this wasn’t a randomized control trial, we did not even use regression discontinuity or dif-in-dif
We use the alternate regression which vector Z omits all characteristics that are common for both siblings, which means that for example two siblings coming from a household where there is a stigma towards mental health is controlled for. Family fixed effects are controlled for by uf.
We then compare B in the first regression and B in the alternate and if they are very similar, we estimate that family fixed effects have very little impact on outcomes today and the first equation doesn’t suffer from OVB.
What is the combined mental health score?
It averages levels of mental health across several mental disorders.
Why is combined mental health score not significant in determining educational outcomes when hyperactivity is controlled for?
Because hyperactivity is the main disorder that causes negative outcomes so when it is accounted for mental disorders as a whole don’t have an impact on mental disorders.
Did having a higher income reduce likelihood of mental disorder negatively impacting education outcomes?
We expected it to reduce negative outcomes, however in the fixed effects regression having a higher income had no effect
