Financial Incentives as Signals: Experimental Evidence from the Recruitment of Community Health Promoters in rural Uganda

How to cite

DESERRANNO, ERIKA. 2016. "Financial Incentives as Signals: Experimental Evidence from the Recruitment of Community Health Promoters in rural Uganda." AEA RCT Registry. AEARCTR-0001852. https://doi.org/10.1257/rct.1852-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

ERIKA DESERRANNO*

* Primary investigator

Registry ID
AEARCTR-0001852
DOI
https://doi.org/10.1257/rct.1852-1.0
First registered
Last updated
Status
Completed
Trial dates
2012-05-15 – 2014-08-15
Countries
Uganda (Kabale, Muhanga, Rukungiri, Buyanja, Kiganda, Kibito, Bundibugyo, Bwera, Adjumani, Moyo, Yumbe and Maracha, Musita, Buyala, Mayuge, Idudi, Nkonkonjeru, Kasawo and Busunju. )
JEL codes
M52, D83, D82, J21, J63, O15
Registry record
http://www.socialscienceregistry.org/trials/1852

Abstract

I study the role of financial incentives as signals of job characteristics when these are unknown to potential applicants. To this end, I create experimental variation in expected earnings and use that to estimate the effect of financial incentives on candidates' perception of a brand-new health-promoter position in Uganda and on the resulting size and composition of the applicant pool. I find that more lucrative positions are perceived as entailing a lower positive externality for the community and discourage agents with strong prosocial preferences from applying. While higher financial incentives attract more applicants and increase the probability of filling a vacancy, the signal they convey reduces the ability to recruit the most socially motivated agents, who are found to stay longer on the job and to perform better.

Experimental design

The empirical design is composed of two experiments. The first experiment (information experiment) tests the signal channel of financial incentives by collecting data on agents' perception after the job is advertised as being more or less lucrative. The effect of financial incentives on the applicant pool is analyzed in a separate experiment—the recruitment experiment—in which a candidate's decision on whether to apply is recorded without any questions on perceptions having been asked prior to that decision (and hence without distorting expectations).

The exogenous variation used in both experiments is identical: experimentally changes the way the job is advertised, i.e., exogenously change whether the position is advertised as being high, medium or low paying, while the actual incentives do not change. The high-pay treatment reveals the maximum of the earnings distribution (“CHPs earn up to 200 thousand UGX per month”), the medium-pay treatment reveals the mean (“CHPs earn an average of 30 thousand UGX per month”) and the low-pay treatment reveals the minimum (“CHPs earn at least 7 thousand UGX per month”).

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
Village level randomization (recruitment experiment), respondent level randomization (information experiment)
Clusters
15 branches (recruitment experiment), 231 rural communities across 4 branches (information experiment)
Observations
315 rural communities (recruitment experiment), 6,845 respondents (information experiment)
Arms
Recruitment experiment: 105 communities in Low-Pay Treatment, 104 in Medium-Pay, 106 in High-Pay Information experiment: 2,296 respondents in Low-Pay Treatment, 2,279 in Medium-Pay, 2,270 in High-Pay
Primary outcomes
candidate perceptions about the job before recruitment, size and composition of the application and worker pool, worker retention and performance

Cited by

No citation data available for this registration yet.

Keywords