Jonathan Cohen

Runner, biker, and (free) food enthusiast.

Formerly @MITEcon, @nberpubs, @NorthwesternU, and @PremierMontrose.

[Jonathan Cohen @jonpcohen]

@LuedtkeAllison Here's an email template I've sent out to undergrads after they finish the intro course to highlight research + get connected with PhD students.

I tried to highlight diversity w/ topics and authors
(and have succinct 1-line summaries @SayWhatYouFound)

https://t.co/vuumcVHl4L

[Jonathan Cohen @jonpcohen]

@captgouda24 7. Shape restrictions

Paper: Dutz et al unpublished WP
Idea: In Manski fashion, gradually introduce theory-based restrictions to tighten set identification (eg, sign income/substitution effects, etc).

[Jonathan Cohen @jonpcohen]

@captgouda24 6. Randomize response incentives

Paper: Dutz et al '25 RESTUD
Idea: Like Behaghel, have a response instrument. Unlike them, have multiple! Monetary incentives + outreach effort. Tour de force reviewing various nonresponse correction methods.

[Jonathan Cohen @jonpcohen]

@AlecStapp @captgouda24 FWIW I reviewed these and other related papers. I agree with the overall take of rent decreases, but it’s important to not sweep under the rug the overall take of price increases.

My overall takeaway is the cold take of needing to focus on increasing new supply.

[Jonathan Cohen @jonpcohen]

[1/16] Catching from @just_economics, here's my read

All: prices ↑

Chang (RF event study), BF (structural search), Coven (structural BLP): rents ↓, consistent w/ segmented markets S&D

Hanson (RF event study) and GQZ (RF shift-share): rents ↑, but w/ data & design drawbacks

[Jonathan Cohen @jonpcohen]

@nealemahoney @mattyglesias @weakinstrument My read of the evidence is that prior should be rent decrease. (Caveat I missed Gurun et al that you cited).

The rent decrease mechanisms are (1) reallocate owner-occupied to rental and (2) more overall new supply. 1st principles is that 7-year BTR window disincentivizes (2).

[Jonathan Cohen @jonpcohen]

[1/16] Catching from @just_economics, here's my read

All: prices ↑

Chang (RF event study), BF (structural search), Coven (structural BLP): rents ↓, consistent w/ segmented markets S&D

Hanson (RF event study) and GQZ (RF shift-share): rents ↑, but w/ data & design drawbacks

[Jonathan Cohen @jonpcohen]

@Birdyword @ArmandDoma It depends what FICO segment the card attracts.

https://t.co/nRtncgOBNx

https://t.co/cxJhOMGEGc

[Jonathan Cohen @jonpcohen]

@captgouda24 There were several other papers apart from these two that came out around the same time. My take was that all showed price increases to some degree, but the higher quality evidence showed rent decreases.

[Jonathan Cohen @jonpcohen]

[1/16] Catching from @just_economics, here's my read

All: prices ↑

Chang (RF event study), BF (structural search), Coven (structural BLP): rents ↓, consistent w/ segmented markets S&D

Hanson (RF event study) and GQZ (RF shift-share): rents ↑, but w/ data & design drawbacks

[Jonathan Cohen @jonpcohen]

@Afinetheorem @Jabaluck @jasonfurman Timely methods paper for that question!

[Nicolaj Thor @nicolajthor]

@jamespstratton and I put out a working paper a few days ago on exactly this question! We propose a way to answer it using a simple causal analogue of the R² ("Causal R²").🧵(1/12). Paper: https://t.co/FgD8fFIEYj

[Jonathan Cohen @jonpcohen]

@paulnovosad It’d be nice to correct for pre-trends by instrumenting for an unaffected confound proxy (eg # of enrolled classes, behavioral issues, etc) with policy leads, rather than extrapolating a linear trend (Freyaldenhoven, Hansen, Shapiro ‘19 AER)

https://t.co/TzwhZDSF1r

[Jonathan Cohen @jonpcohen]

@KelseyTuoc My take is to follow the local time.

(Regretfully typing this from a Boston to Chicago flight that lands 7pm CT.)

[Jonathan Cohen @jonpcohen]

@SashaGusevPosts @KelseyTuoc I’m curious to hear author @Econ_Geoff’s take.

Outcomes are selective school apps/enrollments. Clearly relevant to ed trajectories, less obviously sufficient stats for benefit.

(I get you’re making a broader policy point equating absence of evidence w/ evidence of absence.)

[Jonathan Cohen @jonpcohen]

@arpitrage The biweekly survey is plotted as a 6-mth moving avg. A chunk of the drop off is from a late July outlier.

In addition to the past 2-wk Q, there’s a next 6-mth Q. That has less of a drop off. And its higher level suggests expectations of more adoption.

https://t.co/WUjiue9tty

[Jonathan Cohen @jonpcohen]

@instrumenthull @tramir Does this also generalize to OLS with binary D?

Like IV w/ binary D gets complier X’s and (un)treated potential outcomes using appropriate interactions w/ D or 1-D.

Do those same interactions in the single OLS equation get “untreated effective sample” X’s and POs?

[Jonathan Cohen @jonpcohen]

@instrumenthull @MPaallysaho A nice example is Bhuller et al

https://t.co/WrtDBeLTkg

[AEA Journals @AEAjournals]

Forthcoming in AER: Insights: "Disemployment Effects of Unemployment Insurance: A Meta-Analysis" by Jonathan P. Cohen and Peter Ganong. https://t.co/1hQbIoEqsQ

[AEA Journals @AEAjournals]

Forthcoming in AEJ: Applied Economics: "Skill Depreciation during Unemployment: Evidence from Panel Data" by Jonathan Cohen, Andrew C. Johnston, and Attila Lindner. https://t.co/vmsqOp817w