Thread by @douglasharris99

New paper w/ Jon Mills abt coll financial aid, incl #freecollege. edworkingpapers.com/index.php/ai21…. We build on past res espec @dynarski @jscottclayton @saragoldrickrab @rkelchen & others, but our study is fairly different & relevant to Biden #AmericanFamiliesPlan. #EconTwitter #highered

We studied The Degree Project (TDP) for 10 years (& counting) using a randomized trial. 1/2 of 9th graders in Milwaukee were promised $12,000 for college, enough to cover all tuition/fees at the local 2y college ("free" 2y). Same funds could also be used at 4y colleges

There were strings attached. Students had to meet multiple merit req during HS: Obtain 2.5 GPA, 90% class attendance, on-time HS grad, fill out FAFSA. Our study & results are unusual in at least 6 ways:

“First, this is one of two studies to estimate the effects of financial aid [incl free college] on college enrollment and attainment in the U.S. using RCTs.” (See also: @metrics52) TDP inc 2y degrees by 3 perc pts or 25% from baseline. Seems driven by #freecollege design/framing

“Second, this is the first study to examine financial aid effects on a wide range of high school outcomes.” This is key. Easy to forget that coll fin aid could inc college expectations & effort during HIGH SCHOOL. So, possible benefits of aid even if students don’t attend college

We find effects on low-cost, college-related behaviors like applying to more colleges & partic in college access prog, but no effects on academic metrics. Many poss reasons for lack of academic effects, espec lack of info & need for sustained effort & support.

“Third, the early commitment design, combined with our rich annual data, allow us to test for unusual and important forms of effect heterogeneity, especially the effect timing.” No surprise to parents like me w teens: students procrastinated & responded mostly at the end of HS

“Fourth, this is one of only a handful of studies to examine the effects of aid on life outcomes beyond education.” We find no effects on these outcomes yet, but it’s probably too earlier for most of them. See also: @scottclayton @carlson @goldrickrab @kelchen

“Fifth, we designed the experiment and data collection to learn how financial aid design and implementation may affect student outcomes.” The merit reqs undermined the program, keeping students from receiving funds w/o getting students to “work harder.”

Also, on implementation, counselor communication w students varied a lot & this may have influenced program effects. Counselors also expressed frustration w students who weren’t meeting the reqs. Again, merit reqs seem counterproductive.

“Sixth…we outline and test theories about student responses to financial aid.” Some of the basic econ theory seems to fit, but they’re not “adolescent econometricians” either. Behavioral econ (i.e., #psychology) seems quite useful in explaining some resp

Also, see longer version of this thread in blog from @BrookingsEd: brookings.edu/blog/brown-cen…

Many thanks to our partners—students and counselors, #Ascendium for funding the aid + admin the prog, @MilwaukeeMPS, & large res team: R. Farmer-Hinton, @johndiamondphd, T. Blakely-Reavis, B. Carl, D. Kim, H. Lustick, & K. Rifelj. (Kelly worked on this for 10 years straight!).

Many others have studied similar fin aid progs (alphab): @davidautor @TimBartik @devencarlson @ckcarruthers @SarahCohodes @ProfDavidDeming @MartaLachowska @bterrylong @JoshuaSGoodman @BHershbein @mmilleradams @lauraperna1 @turnersarahe & more (espec not on Twitter)

Finally, thanks to @IESResearch for research funding. The base of research on this & so many other #highered topics has improved exponentially bc of their support. And don’t forget to thank your program officers, too. They have very difficult jobs.