A shared performance incentive scheme for health workers and their supervisors improved productivity and health outcomes and was more effective than paying the incentive to only one group
Financial incentives are believed to be a key policy tool to improve the quality of bureaucracies and public service delivery (Xu et al. 2023). Incentives are a particularly promising intervention for specialised frontline workers, whose performance maps clearly to measurable outcomes, and several papers document the positive impacts that are generated by raising the level of incentives (e.g., see Dathan et al. 2022 for health workers; Mbiti et al. 2023 and Muralidharan and Sundararaman 2011 for teachers; and Khan et al. 2016 for tax collectors). However, these frontline workers are typically embedded in a hierarchy, with a supervisor that trains and coordinates the efforts of several subordinate staff. This raises the question of how incentives should be allocated across the layers of the hierarchy. Unfortunately, there is currently very limited empirical evidence to guide policy makers on this problem.
Should incentives be mainly allocated to lower tier workers, higher tier workers, or should multiple layers be incentivised? In recent research (Caria, Deserranno, León-Ciliotta and Kastrau 2023), we shed light on this question in the context of Sierra Leone’s National Community Health Worker Program which is run by Sierra Leone’s Ministry of Health and Sanitation. While paying incentives to Community Health Workers (CHWs) is fairly common across community health programmes similar to the one we study in Sierra Leone, incentivising higher-tier workers is extremely rare (Perry 2020). Whether incentivising only frontline workers like CHWs is the best incentive allocation for maximising productivity, though, is an open empirical question.
While having high-performing health workers is an important goal anywhere in the world, improving health care provision in Sierra Leone is an especially pressing issue. In 2017 – the year before our study took place – Sierra Leone had the world’s third-highest maternal mortality rate and the fourth-highest child mortality rate (WHO 2017). CHWs play a pivotal role in Sierra Leone and often are the primary touch point with the public healthcare system for households, especially in remote rural villages. They provide a basic package of primary health care services to households in their communities and refer severe cases to the local health clinic. Supervisors do not provide health care services to households directly and instead are responsible for training and advising CHWs.
Study design
We conducted a randomised controlled trial to find out how incentives should be allocated across CHWs and their supervisors to maximise health visits and, importantly, health outcomes of the population living in the areas served by CHWs (Deserranno et al. 2023). The study took place in 372 health clinics across six of the fourteen districts of Sierra Leone. We randomly assigned each health clinic and its supervisor and CHWs to one of four groups:
- Worker incentive group, in which each CHW received a bonus of SLL 2,000 (about US$0.22 in 2019) on top of her base pay for each health visit she conducted.
- Supervisor incentive group, in which each supervisor received a bonus of SLL 2,000 on top of her base pay for each health visit conducted by a CHW under her supervision.
- Shared incentive group, in which the bonus was equally divided between supervisors and CHWs. The supervisor and the CHW each received a bonus of SLL 1,000 on top of base pay for each health visit conducted by the CHW.
- Comparison group (status quo), in which CHWs and supervisors did not receive performance-based pay.
CHWs and supervisors in all four groups also received a fixed monthly salary from the Ministry of Health and Sanitation. CHWs earned SLL 150,000 (about US$ 18 in 2019) and supervisors earned SLL 250,000 (about US$ 29 in 2019). This means a CHW who made one health visit per day could earn up to 40% of her fixed monthly salary in the worker incentive group, and up to 20% in the shared incentive group. Since each supervisor oversees an average of eight workers, their potential gains were even larger. The financial incentives were in place from May 2018 to August 2019.
Performance incentives were calculated based on self-reported data. After completing each household visit, the CHW sent an SMS to a toll-free number indicating the date of the service, the name and phone number of the patient, and the service delivered. A monitoring team back-checked this data by contacting a random 25% of households to confirm the information reported by the CHW, and sanctions were in place in case of cheating.
Key results
Despite the incentive amount per health visit being the same in the three incentive groups, sharing the incentive between CHWs and their supervisors lead to an increase in healthcare visits (as reported by households in the endline survey) which was 61% larger than the increase seen in both one-sided incentive groups. Figure 1 shows that compared to CHWs in the control group who conducted 5.3 health visits per household in a six-month period, CHWs in the shared incentive group conducted 3.4 additional health visits per household. CHWs in both the worker incentive group and the supervisor incentive groups only conducted 2.1 additional visits.
The shared incentive group also outperformed all other groups on improving health outcomes of households. Pregnant women served by CHWs in the shared incentive group were 5.8 percentage points more likely to receive at least four ante-natal visits before birth and 3.6 percentage points more likely to give birth at a health clinic rather than at home. Children were 5.8 percentage points less likely to have fevers. The effects on health outcomes for both the worker and supervisors incentive groups were lower in magnitude and not significantly different from the control group.
What can explain these large effects? We find that the key factor explaining our results is that the work performed by CHWs and supervisors is highly complementary. Without training from supervisors, CHWs are unable to correctly diagnose and treat diseases and will lose credibility in the community, resulting in fewer health visits and lower improvements in health outcomes. Similarly, the effort of supervisors in training CHWs remains ineffective if CHWs are not motivated to go out and conduct health visits. But why do supervisors in the supervisor incentive group not replicate the incentive scheme of the shared incentive group by offering CHWs a payment for each completed visit? We provide evidence that supervisors rarely make such side payments because it is too costly for them to monitor and verify the reports of CHWs, it is difficult to commit to making future payments, and there are social norms against side payments. Shared incentives work best because both CHWs and supervisors play an important role in conducting high-quality visits that generate further demand from households and the shared incentive motivates both to increase their effort. Contractual frictions prevent the shared incentive scheme from being replicated when the initial allocation of incentives is not optimal.
Takeaways
Our results are quite surprising given the popularity of frontline worker incentives and the absence of similar incentives for higher tier workers in many public sector organisations. This is also reflected in the predictions of researchers who we asked through the Social Science Prediction Platform which of the three incentive allocations would maximise the number of health visits conducted by CHWs: 51% of forecasters selected the CHW-only incentive scheme and only 28% selected the shared incentive scheme. Policy makers planning to provide incentives to frontline workers should therefore ask themselves whether to not also incentivise their supervisors or managers. The key question they should consider in their decision is whether these higher tier workers play a complementary role in the output of frontline workers similar to the role played by supervisors in Sierra Leone’s Community Health Worker Program – which we show would call for providing shared incentives – or whether they mainly engage in monitoring or administrative tasks in which case it would likely be better to only incentivise frontline workers. This question is especially relevant for organisations where vertical side payments are limited and the initial allocation of incentives is unlikely to differ from the actual allocation.
References
Besley, T, R Burgess, A Khan, and G Xu (2022), “Bureaucracy and development.” Annual Review of Economics, 14: 397-424.
Dathan, A, P Dupas, and K Macours (2022),” Pay for performance incentives for health in low- and middle-income countries.” J-PAL Policy Insights.
Deserranno, E, S Caria, P Kastrau, and G León-Ciliotta (2023), “The Allocation of Incentives in Multi-layered Organizations.” Working Paper
Khan, A Q, A I Khwaja and B A Olken (2016), “Tax farming redux: Experimental evidence on performance pay for tax collectors.” The Quarterly Journal of Economics, 131(1): 219-271.
Mbiti, I, M Romero, and Y Schipper (2023), “Improving foundational learning in Tanzania through teacher rewards.” VoxDev Article.
Muralidharan, K, and V Sundararaman (2011), “Teacher performance pay: Experimental evidence from India.” Journal of Political Economy, 119(1): 39-77.
Perry, H (2020), Health for the people: National community health worker programs from Afghanistan to Zimbabwe.
WHO (2017), Global Health Observatory.
Xu, G, E Deserranno, D Moreira, and E Teso (2023), Bureaucracy. VoxDevLit, 8.1.