Evidence from simultaneous field experiments 6 th Workshop in Behavioral and Experimental Health Economics Oslo Dec 14 2018 Varun Gauri Julian C Jamison Nina Mažar Owen Ozier World Bank ID: 787741
Download The PPT/PDF document "Motivating bureaucrats through social re..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Motivating bureaucrats through social recognition: Evidence from simultaneous field experiments
6th Workshop in Behavioral and Experimental Health Economics Oslo: Dec 14, 2018
Varun Gauri
Julian C. Jamison
Nina
Mažar
Owen Ozier
World Bank
University
of Exeter
Boston University
World Bank
Slide2Is this…
Health economics?Development economics?Governance / organizational behavior?Field experiments?Behavioral economics?External validity / methodology?
Slide3Policy motivation
Nigeria: population around 200 million, although the exact number is unknownNigerian states are ethnically, linguistically, and culturally distinct (and sometimes heterogeneous)They also operate with a fair degree of autonomy
Slide4Policy motivation
Nigeria: population around 200 million, although the exact number is unknownNigerian states are ethnically, linguistically, and culturally distinct (and sometimes heterogeneous)They also operate with a fair degree of autonomyNigerian health system: poor outcomes, even for GDP (Dunsch, Evans, Eze-Ajoku, Macis
2017)
Slide5Policy motivation
Central and local governments often know remarkably little about how much they or others spend on primary health care
Slide6Policy motivation
Central and local governments often know remarkably little about how much they or others spend on primary health careObvious implications for efficiency and cost-effectiveness analysesBut also matters for accountability and ‘leakage’
Slide7Policy motivation
Central and local governments often know remarkably little about how much they or others spend on primary health careObvious implications for efficiency and cost-effectiveness analysesBut also matters for accountability and ‘leakage’Hence ‘real-time’ public expenditure tracking, but need to incentivize participation efforts
Slide8Research motivation
Naturally want to evaluate optimal low-cost incentives in public sector
Slide9Research motivation
Naturally want to evaluate optimal low-cost incentives in public sectorBut also contribute to recent literature on external validity: studying variations in outcomes for varying implementation of ‘exactly the same’ interventionAllcott and Mullainathan (2012)Bold et al. (2013)Banerjee, Karlan, and
Zinman
(2015)
Bates and
Glennerster
(2017)
Slide10Context
In this case we work in two quite different states of Nigeria: Ekiti and Niger
Slide11Context
Slide12Context
Slide13Context
Slide14Context
In this case we work in two quite different states of Nigeria: Ekiti and NigerLarger project introduced relatively simple forms to track income and expenditure streams at PHCsWeekly visits by enumerators to build capacity as well as check on progress, but no incentivesQualitative work suggests little existing motivation for record-keeping (e.g. belief in social benefits); health committees cared primarily about visitation
Slide15Checklist
Form C records all cash transactions at the facilityFor this study we added a scoring checklist:Was form C filled out prior to arrival?Did the treasurer check form C for accuracy?Is each of the five main sections complete?Are drug purchases / sales as recorded in form C consistent with other documentation?Can staff provide receipts / invoices to substantiate the data in form C?
Slide16Behavioral incentives
There are of course entire literatures devoted to incentivizing workers and employees, including both intrinsic and extrinsic motivatorsRelatively little of this has been carried out in severely capacity- and resource-constrained environments, especially re public service delivery
Slide17Social recognition
Considerable research shows that public / peer recognition and status is highly motivational for humans; e.g. Stajkovich & Luthans (1997, 2001)Also less likely to ‘crowd out’ intrinsic motivation than are monetary rewards (Ryan & Deci, 2000)
Slide18Social recognition
Considerable research shows that public / peer recognition and status is highly motivational for humans; e.g. Stajkovich & Luthans (1997, 2001)Also less likely to ‘crowd out’ intrinsic motivation than are monetary rewards (Ryan & Deci, 2000)Mathauer & Imhoff (2006) find this for public / private / NGO healthcare workers in Benin
Ashraf,
Bandiera
, & Jack (2014) find that stars beat money selling female condoms in Zambia
Slide19Evaluation design
Intervention is straightforward: transform the checklist score into a number between 0 and 5, corresponding to stars
Slide20Social recognition certificate
Slide21Social recognition certificate
Slide22Evaluation design
Interventions are straightforward: transform the checklist score into a number between 0 and 5, corresponding to starsSample size: 65 facilities in Ekiti; 75 in NigerMoH required same intervention in both statesFive weeks of observation at each: one pre-intervention, four during
intervention
Slide23Evaluation design
We compare social recognition to control, stratified by state.
Week
0
Week
1
Week
2
Week
3
Week
4
1
st
half of facilities in each state
Baseline
Social Recognition
2
nd
half of facilities in each state
Baseline
Comparison
Slide24Tests of balance
Slide25Results!
Slide26Results: Niger
Slide27Results: Ekiti
Slide28Tests of balance
Slide29Results (Ekiti)
Slide30Results
Slide31Results
Slide32Differences in characteristics between states
Slide33External validity
Slide34Heterogeneity
Slide35External validity
Slide36External validity
Slide37External validity
Slide38Discussion
We deduce that culture and/or systematic factors matter for the success of [health] interventionsHow do we measure these directly?What are the policy implications?
Slide39Discussion
We deduce that culture and/or systematic factors matter for the success of [health] interventionsHow do we measure these directly?What are the policy implications?Separately note that the control group (in Ekiti) is trending up, perhaps due to observation; cf literature on community-based monitoring
What would longer-term results have been?
Slide40Conclusions
We implemented an RCT testing social recognition to incentivize health workers to track expenses more thoroughlyStrong evidence that social recognition matters and drives (or at least can drive) behaviorThough facilities differ in observable characteristics across states, those differences don’t explain the differential treatment effects.
Slide41Lotteries
Idea is to stretch the efficacy of limited resources by leveraging tendency to overweight small probabilities and focus more on the prize itselfHas been used in finance (e.g. prize-linked savings) and road safety (lottery tickets given to all qualified drivers and taken away for safety infractions)In health: Volpp et al. (2008) successful financial lotteries for medication adherence and weight loss
Slide42WEEKS 1-4
WEEKS 4-8
Slide43Results: Niger
Slide44Results: Ekiti