#460 2019 · BRAC Uganda (recruitment field experiment by Erika Deserranno) · Nonprofit / community health
BRAC found the best health workers by advertising the job's low pay, not its high pay
the problem
An NGO hiring door-to-door health workers had no cheap way to tell a mission-driven applicant from one just chasing the paycheck
background
BRAC needed to recruit thousands of Community Health Promoters across rural Ugandan villages -- unsupervised, commission-paid agents who visit households to sell subsidized health products and refer sick villagers to clinics. The job's real output depends on effort no manager can watch: whether a promoter actually knocks on doors on a slow week or just waits for easy customers to come to her. Interviews and screening tests can check whether someone says the right things, not whether she'll do the work when nobody's checking, and formal screening at that scale is itself expensive per hire.
The standard playbook for filling a hard-to-monitor role is to raise the advertised pay -- a bigger number pulls in more applicants and, the assumption goes, better ones, since money is what draws talent to any job. BRAC's own recruitment materials, like most NGOs', advertised the position by naming the most attractive end of its real earnings range.
what everyone would do
Advertise the most attractive pay you can to pull in the largest, most competitive applicant pool -- the standard recruiting assumption that a bigger number gets you more and better candidates for any job, including one where the actual work happens out of sight.
what they saw
A wage isn't just compensation, it's information: candidates read the advertised pay as a clue to what kind of job this really is, and a high number reads as 'this is a job for the money,' which is exactly the wrong signal to send when the job needs someone who'll do the work whether or not anyone's counting.
the move
Researchers working with BRAC's actual recruitment campaign randomized what the job ad revealed about pay across 315 villages: some ads named a figure near the top of promoters' real earnings distribution, others named the average, others the low end -- all describing the identical job, identical training, identical commission structure. Nothing about the job itself changed, only what applicants were told to expect from it.
why it works
Advertising high pay widens the applicant pool by attracting people motivated primarily by income, but for an unsupervised, effort-dependent role that pool is adversely selected against the trait that predicts performance -- prosocial motivation -- because a high number signals a transactional job and screens IN people who see it that way. Advertising the low or average end of real pay instead signals a job whose reward is largely non-monetary, which selectively discourages paycheck-first applicants and leaves a smaller pool skewed toward people who stay for the mission, and who then visit more households and quit less because their motivation didn't depend on the number in the first place.
the payoff
Ads promising higher pay did pull in more applicants, but the effect ran the wrong way on quality: candidates who applied to high-pay ads were significantly less likely to have prior volunteer health experience or say they valued community service, and more likely to report they wouldn't donate to a health charity at all. Two years later, promoters recruited under the high-pay framing had dropped out at a markedly higher rate -- each standard-deviation increase in a candidate's measured prosocial motivation was associated with an 8-percentage-point lower dropout rate, and prior volunteer experience cut dropout by 11 points. Promoters hired through the low-pay-framed ads visited more households, ran more prenatal checks, and reached more vulnerable families over the following two years.
where it breaks
It only works when the job's real output actually depends on intrinsic motivation that supervision can't substitute for -- for a role that's easy to monitor and pay for output directly (piece-rate factory work, metered sales), advertised pay has no adverse-selection cost and a bigger number just gets you more capable applicants. It also assumes candidates have some other job option; in a market where nobody can afford to be picky about mission fit, everyone applies regardless of the ad's framing and the signal stops discriminating between applicant types.
what came after
Published as "Financial Incentives as Signals: Experimental Evidence from the Recruitment of Village Promoters in Uganda" (American Economic Journal: Applied Economics, 2019), the paper won the UniCredit & Universities Foundation's best paper award and the Royal Economic Society's John R. Hicks dissertation prize, and is now cited widely in the personnel-economics literature on job-ad framing as a low-cost screening tool for roles where intrinsic motivation is the thing that actually predicts performance.
references
- [1]Financial Incentives as Signals: Experimental Evidence from the Recruitment of Village Promoters in UgandaAmerican Economic Journal: Applied Economics (American Economic Association), 2019aeaweb.org
- [2]Does higher pay attract better applicants? Evidence from a Ugandan NGOWorld Bank Blogs (Development Impact), 2018blogs.worldbank.org