I have a new WP out today with Jason Baron, Joe Doyle, and Natalia Emanuel, on discrimination in foster care placement www.dropbox.com/scl/fi/l6nft... This time we study the whole united states, and look at heterogeneity over time and space A few key takeaways 👇 - ThreadSky

instrumenthull.bsky.social • 101 days ago

I have a new WP out today with Jason Baron, Joe Doyle, and Natalia Emanuel, on discrimination in foster care placement

https://www.dropbox.com/scl/fi/l6nftxqeqi9y4qgxa3jrj/ncands_nov7.pdf?rlkey=6dvfrbsdss5heqgjfjp4xvbs3&raw=1

This time we study the whole united states, and look at heterogeneity over time and space

A few key takeaways 👇

Comments

tj-hedin.bsky.social•99 days ago

Really cool paper Peter. Can you help me interpret Figure 2a? Is the takeaway supposed to be that the UD is higher for White children with a higher mean risk, but non-monotonic wrt mean risk for Black children? Or just that it's >0 for all mean risk levels? (or a secret third thing?)

edonavot.bsky.social•100 days ago

Thanks. Cool to see this project develop! I’m still in awe of the identification strategy 👍

instrumenthull.bsky.social•101 days ago

First, across the US, Black children are removed from home by CPS at significantly higher rates than white children with identical potential for future maltreatment (as proxied by re-investigation)

This unwarranted disparity is much larger among kids with maltreatment potential than without

instrumenthull.bsky.social•101 days ago

Second, the extent of UD nevertheless varies significantly across states

Characteristics that predict discrimination include:

- The lack share of the population
- The racial composition of social workers
- Rates of single parenthood
- Welfare generosity
- Inequality in general

leontrout.bsky.social•101 days ago

Im interested in reading this later!

Preview question, did urban-rural differences get tested and how do you handle within-State variation?

instrumenthull.bsky.social•101 days ago

I don't think we looked at urban-rural directly, but we could!

All our UDs are state-level. Our previous paper looked at within-state correlates in Michigan though https://academic.oup.com/qje/article/139/3/1611/7616082

instrumenthull.bsky.social•101 days ago

Black* share, whoops!

jmorenocruz.bsky.social•101 days ago

So sloppy.

instrumenthull.bsky.social•101 days ago

Third, unwarranted disparity in foster care placement has been declining over time -- driven by a dramatic decrease in disparities among children with potential for future maltreatment

This decline reflects a decrease in Black placement rates, following many years of policy reform

instrumenthull.bsky.social•101 days ago

This paper builds on our recent @qjeharvard.bsky.social developing tools to estimate UD w/ quasi-random caseworker assignment (in Michigan)

https://academic.oup.com/qje/article/139/3/1611/7616082

We build on this by showing how UD can be bounded and distilled to a single point estimate, even w/o quasi-random assignment

instrumenthull.bsky.social•101 days ago

That's the paper in a nutshell! Thanks for reading; comments/questions very welcome

Comments

Posting Rules

Reply