So grateful!! Your advice and guidance were invaluable.

And HUGE thanks to my dissertation advisor @tghilarducci.bsky.social and outside advisor @aaronsojourner.org! What a journey.

My only regret is not working a pun into the title. Something about "drivers" of the EITC? Could never get it quite right.

Accepted paper: www.journals.uchicago.edu/doi/10.1086/...

Preprint: ofdavis.com/eitc.pdf

Kleven paper: www.henrikkleven.com/view/researc...

Looney paper: www.brookings.edu/articles/ret...

Addendum: There’s still disagreement over the EITC. Kleven’s 2024 “Reappraisal” paper made a big splash. Adam Looney of Brookings and Univ of Utah has a new wp challenging the consensus. My paper falls on the other side of that debate, but I’m keeping up with all of it.

So there we go, JMP accepted and slated to be published about 4 years after I defended. The editor and reviewers at JOLE were great. All the comments were on point and they made it a better paper.

Not everything I find tells a clean EITC→cars→work story. You might expect employment to go up around tax refund season if the EITC helps people get (back) to work. But the estimated EITC impacts don’t quite have the expected seasonality. Maybe the mechanism isn’t so immediate?

One more bonus analysis: The 2009 EITC expansion. This one was smaller, aimed only at tax filers with 3+ children. The results here are less robust. There’s suggestive evidence that 2009 expansion boosted employment in car-heavy areas. But not the most significant results.

The simulated-EITC methodology has other uses. One is state EITCs. About half of US states supplement the federal EITC. Leveraging within-state differences in EITC exposure helps better estimate the impacts. Results align with the natl expansion. And no welfare reform confounding!

The paper has robustness galore. The big worry is policy confounding: the 1990s EITC expansion coincided with Clinton-era welfare reforms, which also focused on work. I find the results hold even when you restrict to pre-1996, or to states that didn’t do reforms early.

More results! The above used CPS. Other datasets measure different things. SIPP tells us about car ownership. Decennial Census long-forms asked about commuting. In both, the EITC appears to boost car ownership and automobile commuting (bonus finding: employment results hold).

Results! EITC impacts vary systematically across metros based on how much workers need cars. Where most commuters drive, EITC boosts employment more. In public-transit-heavy locales, the impact is smaller. This very much supports the personal-transport mechanism.

My main innovation is a simulated EITC measure. The EITC measure most studies use is the maximum possible EITC by year and family size. Instead, I simulate the EITC HHs would get based on past local incomes differences and national trends. Thus heterogeneous EITC exposure.

But comparing high- and low-car areas has a problem: places with lots of public transit are weird. NYC/DC/SF are not like ATL/STL/DFW. The biggest difference is income. High-wage (subway-using) metros get less out of a national EITC expansion because higher income→less EITC.

There's this 1995 Philadelphia Enquirer article that perfectly encapsulates the empirical strategy. Public transit and cars are substitutes, and tax refunds enable used-car purchases. A sincere thanks to the workers behind the 1995 SEPTA strike for prompting this illustrative news item.

My basic empirical strategy is to compare how the big 1990s expansion of the EITC played out in places with vs without abundant transportation. If the personal-transport mechanism matters, the EITC should have smaller employment-boosting impacts in places where cars matter less.

Rather than acting on distant and uncertain incentives, workers may respond to the tax refunds right when they get them. Buying or fixing a car helps a worker stay employed. Surveys and empirical research show that a large share of EITC dollars go to car dealers and mechanics.

Most studies find that the Earned Income Tax Credit lifts employment (exceptions exist). But what’s the mechanism? Eligible workers are often unaware of the EITC or ill-informed about its rules. Even with complete info, many recipients face unpredictable earnings.

tldr: I studied how the Earned Income Tax Credit affects labor supply. The specific mechanism: helping people own cars. Tax refunds give households a jolt of liquidity that many use on personal transport. I show that EITC expansions boosted both car ownership and employment.

A happy update: My job market paper is now accepted at the Journal of Labor Economics. Since I never did a proper thread on it, here goes.

(Disclaimer: The views expressed in this paper are solely my own and do not reflect those of my employer, NYC IBO.)

startup naming conventions going full Dickens

Does the Whale’s Magnitude Diminish?

6/ The unemployment rate for 20-24 year olds is falling sharply - March it was the lowest it's been since early 2024.

Less of an improvement in the unemployment rate for people in their late teens.

How A.I. Helped One Man (and His Brother) Build a $1.8 Billion Company

www.nytimes.com/2026/04/02/t...

It must say something about the nature of the American heath industry that the weight-loss/ED/hair-meds segment is so ripe for a completely vibe-coded business. Also notable that the real labor of the service (doctors to prescribe the drugs) is outsourced.

Some wild details in this NYT story on what is kind of sort of the first one-person billion-dollar company -- a middleman for GLP-1s. Purportedly $20k in token costs to get to 9-digit revenue. AI everything, except when you call customer service and get routed to the CEO.

excited to read this

JOLTS release once again confirming that the no-hire, no-fire labor market continues. Also lines up with UI claims, which also show no signs of layoffs picking up.

Also norms might change more slowly than labor market outcomes. Women taking more "greedy" and high-status jobs over time and delaying marriage might be consistent with relatively stable LFPR ratio.

A fuller picture? www.pewresearch.org/social-trend...

A dispiriting finding in the fascinating new working paper on household ChatGPT use (just written up in the Wall Street Journal): increased chatbot usage is associated with decreased news consumption.
arxiv.org/pdf/2603.03144