Download presentation
Presentation is loading. Please wait.
Published byLaura Stokes Modified over 8 years ago
1
Discussant Comments: Crime and Specialized Populations Jens Ludwig Georgetown University & NBER
2
Butcher & Piehl: Immigration Could hardly be more timely Careful, well done analysis Convinces me immigrants are less crime-prone –(Disclaimer: Discussant fits their definition of “immigrant”) Paper offers plausible explanation for two seemingly contradictory facts: –Immigrants have lower earnings than native born… –But immigrants are less criminally active
3
Potential concerns Measurement error w/ Census, but they address that convincingly in my view Deportation of immigrant arrestees –But show similar patterns hold for immigrants who are citizens (not subject to deportation) Main suggestion: More on role of geography in the paper as part of understanding changes over time in immigrant offending patterns?
4
Place, continued Ratio of immigrant / native incarceration (T2): –1980: ~1/3 –1990: ~1/2 –2000: ~1/5 Are immigrants disproportionately concentrated in areas where surge and ebb in crack epidemic most pronounced? (big cities in “blue states”) Butcher and Piehl show for 20 largest cities, crime drops more in places w/ more immigration Add more on geographic distribution of immigrants to census analysis of changes?
5
Paper raises new questions about BCA of immigration policy Put into framework of Cook (1986) model for “market for criminal offenses” “Price” = net return to crime (loot minus expected penalty) vs Q of crime Supply of offenses increasing function of P “Demand” decreasing function of P (as crime increases, citizens take more protective measures)
6
How does immigration affect market for criminal offenses? Even if immigrants less likely to participate in crime, shifts out supply-of-offenses? Which way does immigration shift “demand for offenses” schedule? –Immigrants less loot (D shifts in) –But perhaps less likely to spend on protective measures and/or call police (D shifts out)
7
Market for offenses, cont’d Framework helps highlight that immigration could have implications for: –Offending and victimization rates for native- born Americans Butcher-Piehl across-city analyses seem informative about how this nets out Put this into context of larger benefit-cost analysis of diff’t immigration policies? (Maybe separate paper, but important)
8
Youth work and crime I will discuss three papers: –Staff, Osgood, Schulenberg, Bachman, Messersmith (HLM) –Apel, Brame, Bushway, Haviland, Nagin and Paternoster (propensity score matching) –Apel, Brame, Bushway, Paternoster, Sweeten (IV – BONUS DISCUSSANT COMMENTARY)
9
Commonalities across papers All careful, well done, interesting All pay attention to heterogeneity in effects of youth employment on crime –Seems very important for this issue (think @ motivations for poor vs middle class youth) All rely on youth self-reports @ behavior –How worried should we be about this? –(Discussant scarred from comparing MTO self-reports on arrests to admin data…)
10
Identification, Part 1: Staff et al: Use HLM to control for individual fixed effects in MTF data - But why do some youth change their work & crime status while others do not? Apel et al: Propensity score matching - But why do some youth w/ same observed p-hat (Xs) choose to work and others don’t? Both represent step forward in literature but Qs about causality necessarily remain
11
Identification, Part 2 Most intriguing analysis (at least to me) is Apel et al. IV paper that got dropped Uses variation in teen work from variation in state child-labor laws Concern that these laws shift youth from formal to informal work? Key is two age thresholds, opposite treatment effects of state laws on teen work so lets us net out state effects even w/ state law instruments
12
Opposite effects on youth crime w/in states at 2 age thresholds Age 14-1516-1718+ Maine18 hrs max (Federal) 20 hrs maxUncapped Colorado18 hrs max (Federal) 40 hrs maxUncapped
13
Don’t give up on this paper! (Ex) Show table with a row per state, four columns, %Δ or (rank in % Δ) for: –Change teen hours work 14-15 vs 16-17 –Change youth crime 14-15 vs 16-17 –Change hours work 16-17 vs 18 –Change youth crime 16-17 vs 18 (Columns 1 & 2 and 3 &4 should be negatively correlated, and so should 2 & 4) Have a nice story here about what’s driving variation (natural experiment) plus opportunity to rule out some forms of confounding state heterogeneity
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.