Experimental Evidence from Sierra Leone
A challenge:
Quality of service provision is low in many developing countries
Recruiting and incentivizing service providers is difficult
A case: post Ebola epidemic in West Africa
Local delivery models since 1990s: locally recruited actors, to deliver interventions or to do government tasks
Popular, …. but is little known when and how they work
Here: focus on agricultural extension workers
Largely unpaid, using entrepreneurial worker model
Agripreneur programs (World Bank, etc), but part of government system
Promising work on improving service provision:
Recruitment
Community monitoring
Financial incentives
We show:
Theoretically:
Empirically:
Local agents may improve the targeting of and satisfaction with social programs, (Alatas et al. 2012; Basurto, Dupas, and Robinson 2020)
Applicatons: environmental regulators (Duflo et al. 2018), agricultural extension officers (Dal Bó et al. 2021), microentrepreneurs (Hussam, Rigol, and Roth 2022), taxation (Balán et al. 2022)
In fragile states: can local embedded agents act as complements to the state?
By and large a success:
Education: (Muralidharan and Sundararaman 2011), Kahn et al (2014), Glewwe et al (2010), De Ree et al (2010), (Leaver et al. 2021)
Health status: Miller et al (2012), (Mohanan et al. 2021)
Service delivery: Gertler and Vermeersch (2012), Desserano et al (2022)
Service provider inputs: teachers (Duflo et al, 2012, Leaver et al. (2021)), nurse attendance (Banerjee and Duflo (2006))
View that bottom-up pressure by empowered citizens is key to improved service delivery
Maybe even more effective when government can’t monitor behavior
Effective in health if baseline is low: (Björkman and Svensson 2009; Christensen et al. 2021) vs (Raffler, Posner, and Parkerson 2024)
Work in education is more mixed: Banerjee et al. (2010); Barr et al. (2012); (Pradhan et al. 2014); (Molina et al. 2016)
Original enthusiasm: “Power to the People” (Björkman and Svensson 2009)
Intensive treatment
Assess quality of services provided in clinic/community-specific report cards
Disseminate report cards to community and to clinic staff
Develop joint action plans
Interface meetings between clinic staff and community
A small sample (n = 50)
And amazing results:
Infant weights significantly higher
Under-5 mortality 33% lower
Immunization rates higher
Staff absenteeism and waiting times at clinics lower
We study these questions in the context of a randomized agricultural intervention in Sierra Leone.
Core design elements (more on context to come)
| treatment | label |
|---|---|
| T1 | Agent is selected by community (T1 = 1) or by state (T1 = 0) |
| T2 | Community can impose social sanctions on agent (T2 = 1) or not (T2 = 0) |
| T3 | Agents receive pay for performance (T3 = 1) or not (T3 = 0) |
After Francois (2000). A worker has type \(\theta = (\eta,\mu)\) where:
Drawn according to:
| \({\mu}{\backslash}\eta\) | \(\underline{\eta}\) | 1 |
|---|---|---|
| 0 | \(\frac{1+ \rho}{4}\) | \(\frac{1- \rho}{4}\) |
| \(\overline{\mu}\) | \(\frac{1- \rho}{4}\) | \(\frac{1+ \rho}{4}\) |
So \(\rho\) governs the correlation between embeddedness and motivation: embedded people are possibly more effective, but are they more likely to be highly motivated?
So state is more likely to select workers with high intrinsic motivation if and only if \(\rho < 0\).
Workers choose effort \(e\)
Performance depends on effort and embeddedness: \(\pi = e \eta\)
Stochastic performance signal \(\sigma\), which is high (\(\sigma=1\)) with probability \(\pi\) or low (\(\sigma=0\)) with probability \(1-\pi\)
Cost of effort: \(c(e) = \frac{e^2}{2\kappa}\) (alternative use of time, with wage elasticity of effort \(\kappa\))
\[s=(s(0),s(1)) \in [-T_m, T_m]^2.\]
Interpretable as commitments (additional) to reward (\(s(\pi)<0\)) or punish (\(s(\pi)>0\)) performance.
in line with Community Action Plan in experimental intervention (in CM)
Payoff to villager:
\[u(\pi, s) = \pi - \gamma s.\]
In crowding out version we have \(u(\pi, s) = \pi - s(\gamma - \xi T_p).\) (and replace \(s\) with \(-r\) to indicate rewards rather than sanctioning interpretation)
Under pay-for-performance (\(T_p=1\)) and high performance signal (\(\sigma=1\)), the worker receives a payment normalized to 1
In addition:
Putting it all together:
\[w(e;s,\sigma;\theta) = \sigma T_p + \pi\mu - \phi s(\sigma) T_m - \frac{e^2}{2 \kappa},\]
The sequence of events is as follows:
An equilibrium is a pair of strategies \(\{ s,e(s;\theta) \}\) such that
\[e(s;\theta) \in \arg \max_{e \in [0,1]} \mathbb{E}_{\pi } \Big\{ w(e;s,\sigma;\theta) \Big\}\] for \(T_m\)\[s \in \arg \max_{\{ (s(0),s(1)) \in [-T_m,T_m]^2 \}} \mathbb{E}_{\theta,\pi} \Big\{ u(\pi,s) \mid \hat{e}(s;\theta) \Big\}\]
\[e(s^*;\gamma,\theta) = \begin{cases} \kappa \eta (\mu + T_p + 2 \phi \zeta T_m) & \mbox{if} \quad \gamma =0 \\ \kappa \eta (\mu + T_p) \quad & \mbox{otherwise}. \end{cases}\]
Full closed form solution for all treatment conditions:
| \((T_m,T_p) | T_s\) | State selection (\(T_s=0\)) |
Community Selection (\(T_s=1\)) |
|---|---|---|
| Control \({\pi^*(T_s,0,0)}/{\kappa}\) | \(\frac{\overline{\mu}}{2} \frac{1 + \rho + \underline{\eta}^2(1-\rho)}{2}\) | \(\frac{\overline{\mu}}{2} (1 + \rho)\) |
| P4P \({\pi^*(T_s,0,1)}/{\kappa}\) | \(\frac{\overline{\mu}}{2} \frac{1 + \rho + \underline{\eta}^2(1-\rho)}{2}+ \frac{1+ \underline{\eta}^2}{2}\) | \(\frac{\overline{\mu}}{2} (1 + \rho)\) \(+1\) |
| CM \({\pi^*(T_s,1,0)}/{\kappa}\) | \(\frac{\overline{\mu}}{2} \frac{1 + \rho + \underline{\eta}^2(1-\rho)}{2} +(1+ \underline{\eta}^2) \phi \zeta\) | \(\frac{\overline{\mu}}{2} (1 + \rho) +2 \phi \zeta\) |
| P4P + CM \({\pi^*(T_s,1,1)}/{\kappa}\) | \(\frac{\overline{\mu}}{2} \frac{1 + \rho + \underline{\eta}^2(1-\rho)}{2} +(1+ \underline{\eta}^2) \phi \zeta + \frac{1+ \underline{\eta}^2}{2}\) | \(\frac{\overline{\mu}}{2} (1 + \rho) +2 \phi \zeta +1\) |
H1: Village selection increases performance.
H2: Community monitoring increases performance.
H3: Pay-for-performance increases performance.
H4: Positive interaction village selection and pay-for-performance.
H5: Positive interaction village selection and community monitoring.
Community monitoring and pay-for-performance could be positive (if independent) or negative (if crowding out)
Poor service provision: mortality is 10% of live birthsqualk
Threat of zoonosis is high and expected to increase
Animals are essential for livelihoods (90% keep animals), comprising >50% of income
Low information and poor/risky animal husbandry practices
Animal diseases are highly contagious
Large impacts on households and largely preventable
In the wake of the Ebola epidemic, together with the GoSL, we implement a “One Health” intervention, addressing animal and human health
363 villages across 7 chiefdoms in Kono district
Implemented from July 2017 to July 2019
Recruitment of Community Animal Health Worker: CAHW
21-day training, implemented by veterinarians
Tool kit and drugs for treatment of basic animal diseases
Creation One Health platform
CAHW:
Strong enthusiasm for the program: reporting throughout the period and beyond
No impact on human health, but impacts on intermediary outcomes
From planner’s perspective it matters if we can improve CAHW performance:
Recruitment
Community monitoring
Pay-for-performance
Community selection: MAF facilitators hold village meetings to nominate and select the CAHW candidate
“Business as usual”: CAHWs are recruited by state authorities in collaboration with Paramount Chiefs
Eligibility requirements, ability tests, language, residence
Training before assignment to incentives
Key design element: double listing
Community Monitoring
CAHW and community develop joint action plans
Interface meetings between ministry staff and community:
Pay-for-Performance
Note: rank ensure revenue neutrality but introduces small risk of non-independence / collusion
Community monitoring and pay-for-performance implemented as designed
The recruitment treatment not so much
Only in 100 villages do we have true double-listing (in many other cases either same person listed twice or no one listed by one party)
Implication: smaller sample and non-experimental interactions
Small amount of attrition
Measuring number and detail of disease reporting, worker availability: SD = 0.19
Both incentives shift reporting behavior
Correlation between performance and observed effort
No change in animal health, stocks, deaths
Increase in reporting quality
Surveillance performance driven by events, effort and ability
Records contain both Type 1 and Type 2 errors
Both matter for containing pandemics
We measure the quality of CAHW reporting
Practical: Central location to assess CAHW
Village visit: MAF representatives visit the CAHW’s village
Sick animal observation: Verify if sick animals had been reported
Form follow-up: Verify reported sick animals
(Less happy with raw data distributions)
There is strong relationship between surveillance quality and performance
Some of these results challenge the model more than others:
stan\[e_{ij} = \alpha_{j} + \kappa \times p(T_s, T_m, T_p | \bar{\mu}, \underline{\eta}, \rho, \nu, \phi, \zeta) + \epsilon_i\]
Wide distributions, but show movement. \(\mu\) moves, small range on \(\eta\), \(\rho\) is flat
Posterior distributions on model parameters
Two distinct types of experiments possible:
In what kind of world (for what parameter values) are different types of interventions most likely to be optimal?
Given some type of policy is likely to be in place, what kind of parameters would one want to alter in order to improve outcomes (so: thinking of parameters as possible sites of intervention)
Both can be seen from next figure.
Comparing packages
Where and when might we expect stronger selection effects?
The Björkman and Svensson (2009) effect.
The payment effect.
Comparing packages
Program
Evidence for two distinct institutional strategies
Increase institutional capital that can become even more critical during crisis
Tradeoff between P4P and CM?
Paper
Created new positions in villages
Sustained engagement
- Data on disease control, response, knowledge, preventative practices
- Covid knowledge
- Social relations, trust
- Animal Health
| Symbol | Name | Interpretation |
|---|---|---|
| \(\mu \in \{0, \overline{\mu}\}\) | (Intrinsic) motivation | Strength of motivation |
| \(\eta \in \{\underline{\eta},1\}\) | Embeddedness | Strength of a worker’s ties to the community |
| \(\theta = (\mu,\eta)\) | Worker’s type | |
| \(\rho \in [-1,1]\) | Type correlation | \(Corr(\mu,\eta)\) |
| \(\phi\) | Social sensitivity | The worker’s susceptibility to social sanctioning |
| \(\gamma \in \{0,\overline{\gamma}\}\) | Guilt | The villager’s psychological cost of social sanctioning |
| \(\zeta \in [0,1]\) | Strength of village-level norms | Probability of low-guilt villager |
| \(\kappa\) | Productivity | Effort elasticity to wage |
| \(e \in [0,1]\) | Effort | How hard the worker works |
| \(\pi \in [0,1]\) | Performance | Outcome of worker’s activity (\(\pi= e \eta\)) |
| \(\sigma \in \{0,1\}\) | Performance signal | Public signal of performance (\(\Pr(\sigma=1)=\pi\)) |
| \(s(\sigma) \in [-1,1]\) | Social sanctioning plan | Villager commits to it before observing \(\sigma\) |
