An Ethiopian employer assumed a costly application process screened out weak candidates — a field experiment found it was screening out the strongest ones

the problem

An employer's hiring process required applicants to travel to a test site and give up paid work time to apply, a cost the employer treated as a routine, roughly equal tax on every applicant

background

what everyone would do

The standard assumption behind a costly, demanding hiring process is that the cost itself does useful work — it screens out unmotivated or unqualified candidates, so only genuinely committed applicants clear the bar, and a harder process therefore produces a better, more filtered applicant pool. Under that assumption, removing the cost should only make the pool larger and, if anything, worse on average, since it lets in exactly the less-motivated candidates the friction was supposedly filtering.

what they saw

Abebe, Caria and Ortiz-Ospina saw that the friction wasn't screening for motivation or quality at all, it was screening for patience and lack of alternatives — and those are close to the opposite of what a talented candidate typically has. A high-ability jobseeker with strong outside options doesn't need to wait out a slow, costly, low-odds application process; they take the first reasonable offer that comes along instead, meaning the very people an employer most wants to hire are the ones with the least reason to endure the friction. Removing the cost wasn't lowering the bar, it was removing a bar the employer never realized it had erected specifically against its own best prospects.

the move

Rather than assume the application cost was screening for quality by filtering out unmotivated or unqualified candidates, the researchers tested the assumption directly: reimbursing applicants for their actual application and test-day costs, and comparing the resulting applicant pool's measured ability against the pool that faced the standard costly process.

why it works

By reimbursing the actual application and test-day costs, the intervention removed the specific factor that was disproportionately deterring high-ability candidates — since those candidates' opportunity cost of waiting is what made the friction expensive for them specifically, not a lack of interest in the job itself, eliminating the cost directly addressed the actual mechanism driving them away rather than trying to make the job posting more appealing in some unrelated way. The structural model's finding that application cost was positively correlated with jobseeker ability, driven by dynamic selection (patient, lower-ability candidates stick around through a slow process; impatient, higher-ability candidates don't), explains precisely why the resulting pool wasn't just bigger but measurably better on average — the incentive didn't attract new low-quality applicants who wouldn't otherwise have applied so much as it retained high-quality candidates who otherwise would have dropped out for a faster opportunity elsewhere. The second experiment's finding that local recruiters themselves underestimated this effect is itself important evidence: it shows the mistaken 'friction equals screening' intuition isn't a rare blind spot, it's the default assumption practitioners hold even when directly asked to predict the outcome of the exact intervention being tested.

the payoff

The employer that offered the small monetary incentive for applying attracted a larger and, on average, more talented applicant pool — not merely a bigger pool of the same average quality. A structural model estimated that the cost of making a job application was large and positively correlated with jobseeker ability, driven by dynamic selection: higher-ability jobseekers have better outside options and less patience for a slow, low-odds hiring process, so they are the ones most likely to give up and take another offer rather than wait out a costly, friction-heavy application. A second experiment found that local recruiters themselves systematically underestimated how much application incentives would improve their applicant pool, suggesting the mistaken 'cost equals screening' assumption is widespread among the people actually running hiring processes, not a one-off oversight.

where it breaks

This mechanism depends on the target population genuinely having a wide range of ability levels with a real correlation between ability and outside options — in a labor market or context where all candidates face similarly limited alternatives regardless of skill, the specific dynamic-selection effect this study documents wouldn't operate the same way, since there'd be no differential patience between high- and low-ability applicants to unlock. It also depends on the friction being cost-based rather than genuinely diagnostic — a process that costs applicants time or money but reveals no real information about their qualification is pure deadweight loss ripe for removal, but a test or hurdle that is actually informative about candidate quality (a skills assessment, a work sample) serves a different function entirely and removing it would lose real signal, not just unlock hidden talent. And the fix here was a targeted, modest cost reimbursement rather than eliminating all screening — an employer that swings too far the other way, removing every filter along with the cost, risks flooding the pool with genuinely unqualified applicants rather than retaining the specific high-ability candidates the friction was silently excluding.

what came after

Published as Abebe, Caria & Ortiz-Ospina, 'The Selection of Talent: Experimental and Structural Evidence from Ethiopia' (American Economic Review, 111(6), 2021), the study is cited in labor economics and recruitment-practice literature as evidence that application friction doesn't screen candidates neutrally — it disproportionately filters out the most talented applicants, who have the least patience for a slow or costly process precisely because they have better alternatives, inverting the intuition most employers hold about what a demanding hiring process accomplishes.

references

1. [1]The Selection of Talent: Experimental and Structural Evidence from EthiopiaAmerican Economic Review 111(6): 1757-1806, 2021aeaweb.org
2. [2]The selection of talent: Experimental and structural evidence from EthiopiaInternational Growth Centre, 2021theigc.org

keep it

same kind of clever

• 2009 · Airbnb (Brian Chesky, Joe Gebbia)Airbnb's New York bookings weren't stalled by its algorithm — they were stalled by bad photos, so the founders rented a camera and fixed the listings by hand, one apartment at a time.
• 2012 · Buc-ee'sBuc-ee's turned the gas station bathroom — the part every competitor treats as pure cost — into the actual reason to stop there
• 2006 · Cedars-Sinai Medical CenterCedars-Sinai got doctors to wash their hands by photographing the bacteria growing on their own palms