IMPACT NIGERIA IMPACT NIGERIA

IMPACT NIGERIA IMPACT NIGERIAIMPACT NIGERIA

Context

Midwife attrition is a major challenge in Nigeria; though there are few concrete statistics, many midwives leave their jobs after a short period of time. This attrition affects the supply of health professionals qualified to provide antenatal care and assist with deliveries. The Government of Nigeria and its development partners have observed this challenge over years of implementing Maternal and Child Health programs, and has identified midwife attrition as a crucial obstacle to achieving success in these programs.

Nigeria has a critical need for innovative and effective programmes that can reduce midwife attrition and affect other factors that contribute to maternal and neonatal health. While maternal mortality decreased from 1,100 to 545 maternal deaths per 100,000 live births between 1990 and 2008, the country still accounts for some 15% of global maternal deaths despite having only 2% of the world’s population. Lack of access to health care is considered a major contributor to these high levels of maternal mortality: 42% of pregnant Nigerian women receive no skilled antennal care during their pregnancy, and 61% of childbirths take place with no skilled attendant present.

From 2012 to 2015, the Government of Nigeria launched the Subsidy Reinvestment and Empowerment Programme Maternal and Child Health Project (SURE-P MCH). This national effort, funded by the reduction of the fuel subsidy, sought to improve the health of mothers and babies in underserved areas. An important focus of SURE-P MCH was increasing the supply of health care in these areas by deploying a total of 3,158 midwives, plus additional clinical staff, to 1,000 government primary health care facilities (PHCs) in all of Nigeria’s 36 states. Despite these efforts, a survey conducted about 1

IMPACT NIGERIA

year after the start of SURE-P MCH found that only 5% of supported PHCs had the recommended four midwives on staff, and 11% had no midwives at all.

Intervention

As part of SURE-P MCH, the government of Nigeria included an impact evaluation to investigate strategies for reducing midwife attrition in government PHCs, and to learn more about how to increase retention. This study, which uses a cluster-randomized controlled trial study design, analyses the impact of monetary and non-monetary incentives provided to reward

INCENTIVIZING MIDWIFE RETENTION IN NIGERIAN PRIMARY HEALTH CENTRES

Key Points

• Findings from this experimental impact evaluation show that, on average, economically meaningful incentives can serve to reduce midwife attrition.

• Incentives are, however, less effective for highly intrinsically motivated persons.

• Together, this suggests that the effectiveness of incentives depends both on the design of the incentive and on the characteristics of the person that receives it.

consistent attendance of a midwife at her assigned PHC. The impact evaluation study was led by Development Impact Evaluation (DIME) at the World Bank, Imperial College London, and University College London, with support from the Bill & Melinda Gates Foundation and the Strategic Impact Evaluation Fund.

Midwives hired by SURE-P MCH and working in SURE-P MCH-supported facilities were enrolled in this study and corresponding intervention through a baseline survey conducted at all 500 Phase I SURE-P MCH-supported facilities between September and December 2013. After responding to a detailed questionnaire, some midwives were told that they were eligible to receive an incentive if they had consistently good attendance at work, defined as uninterrupted service over a period of three months. The intervention was intended to last one year, delivering up to four rounds of incentives to qualifying midwives.

Midwives were randomly assigned to one of four study groups, with all midwives working within the same SURE-P MCH cluster of four PHCs always assigned

to the same study group. One group was eligible to receive a quarterly non-monetary incentive such as a uniform, calendar, or clock for their consistent attendance. Another group of midwives was eligible to receive a quarterly monetary incentive of 30,000 naira per quarter, a value equal to 25% of their salary from the Federal Government over the same period. A third group of midwives were eligible to receive both the monetary and non-monetary incentives together each quarter. A fourth group of midwives was not informed of the intervention; these midwives served as a control group for the impact evaluation.

SURE-P MCH project staff evaluated midwife attendance each month. Midwives who were enrolled in the intervention and present in their facility for the months of December 2013, January 2014, and February 2014 were eligible to receive the first set of incentives. Any midwife who was recorded as absent for one month became ineligible for all future incentive distributions. After the first incentive distribution, two additional incentive rounds were distributed to qualifying midwives.

Impact Evaluation

The intervention was implemented as a cluster-randomized controlled trial. The study enrolled 1,285 SURE-P MCH midwives from 500 facilities across Nigeria’s 36 states and Federal Capital Territory (FCT). 125 clusters of midwives were assigned randomly to one of the three incentive groups or to the control group. The cluster is composed of all midwives working in four SURE-P MCH-supported PHC facilities that report to a common referral hospital.

All 1,285 midwives were administered a structured questionnaire between September and December 2013. This questionnaire included modules covering personal and family history, education, burnout, work conditions, assets and other revenues, measures of motivation, and other key issues. A follow-up round of data was collected from all of the midwives, including those who had stopped working for SURE-P MCH, between December 2014 and February 2015. In addition, attendance data was recorded by SURE-P MCH project staff for the duration of the intervention.

IMPACT NIGERIA IMPACT NIGERIA IMPACT NIGERIAIncentivizing Midwife Retention in Nigerian Primary Health Centers2

IncentiveGroups

Monetary Incentive

(30,000 naira)

Both Incentives

(30,000 naira plus uniform,calendar, or clock)

Non-‐Monetary Incentive

(uniform, calendar, clock)

No incentives

(control)

Study Questions

This randomized control trial seeks to answer three primary research questions:

1. Does providing an incentive to SURE-P MCH midwives (either money, goods, or both) to reward attendance reduce rates of attrition?

2. Are monetary and non-monetary incentives substitutes or complements? Is their impact increased when they are provided in combination, rather than separately?

3. What are the causal mechanisms through which incentives work, or fail to work?

This impact evaluation innovates in several ways. First, prior impact evaluations have not been designed to compare the relative effects of monetary and non-monetary incentives. Second, this evaluation is conducted in a real-world, workplace-based setting on an already-recruited workforce performing difficult and critical work. Furthermore, the evaluation is designed to test for various causal mechanisms through which incentives may work, or fail to work.

Data Analysis

The data was analyzed using a linear probability model that included covariates to improve the precision of the estimates. Three levels of covariates were included: midwife-specific covariates such as work experience and burnout; PHC-specific covariates such as number of midwives and level of security in the area; and cluster-specific covariates such as socioeconomic status and level of women’s

education in that geographic area. Further analysis was carried out to test the validity of causal mechanisms which are thought to make incentives work or fail, namely information implied by the incentive and the interaction of different forms of incentives with self-image and social norms.

Key Findings and Policy Implications

The findings presented here are based on an analysis of the likelihood that a midwife will leave her facility within nine months of the incentive scheme being launched in December 2013.

• Midwives in the control group who were not eligible for any incentive had a nine-month dropout rate of 31%

• Eligibility to receive the quarterly non-monetary incentive (either alone, or together with a monetary incentive) did not have any effect on reducing attrition.

• Midwives eligible to receive the quarterly monetary incentive (either alone, or together with a non-monetary incentive) were 6.3 percentage points less likely to drop out. This is equivalent to a reduction in midwife attrition of 20%.

• Midwives that received both the monetary and non-monetary incentive together were 7.3 percentage points less likely to drop out over a period of nine months. This is equivalent to a reduction in midwife attrition of 24%. But we cannot conclude that this is different from the effect of receiving the monetary incentive alone.

IMPACT NIGERIA IMPACT NIGERIA IMPACT NIGERIA3Incentivizing Midwife Retention in Nigerian Primary Health Centers

Midwife Incentives

Monetary Incentive:31 clusters

Non-monetary Incentive:31 clusters

Both Incentives:32 clusters

Control:31 clusters

These findings show that, on average, economically meaningful incentives can serve to reduce midwife attrition.

It is important to understand not only whether incentives work, but also why they work (or fail to do so). One reason incentives may fail to work is that they “crowd out” intrinsic motivation (in which case their effect on attrition would be limited). In fact, for midwives with high levels of intrinsic motivation receiving both types of incentives has no effect on the likelihood that they will drop out over a nine-month period. And for highly intrinsically motivated midwives eligibility to receive non-monetary incentives alone actually leads to an increased likelihood of drop-out.

These findings show that there is important variation that is hidden by the average impact, and that the effect of any individual may be related to their individual characteristics. For example, incentives are less effective for highly intrinsically motivated individuals, and may even backfire.

Incentives may also affect attrition through conveying implicit information about work as a SURE-P midwife. For example, midwives that were assigned to receive either a monetary or non-monetary incentive (in isolation or together) expected that their workload would increase following the introduction of the incentive scheme. Likewise, those receiving a monetary incentive expected that the manager would be more demanding in the future. But these same midwives also expected that they would enjoy life in their community more after the introduction of the incentive scheme.

These findings show that it is important to be mindful of the implicit information conveyed through

any incentive scheme. Communication strategies may need to be adopted to limit unintended negative consequences of this information, whether or not it is correct.

Finally, incentives may change outcomes through changing the social norms. For example, midwives eligible to receive the non-monetary incentive stated, on average, that the minimum acceptable length of service was 1.3 months longer than did the control group, for whom on average the minimum acceptable length of service was 13.9 months.

Together, the results suggest that all incentives are not created equal, and that their effectiveness depends not only on the design of the incentive itself but also on the characteristics of the person receiving the incentive. Incentives can be a powerful tool to support midwives and other public sector works operating in challenging environments, but designing an effective incentive scheme will require detailed knowledge of the workforce and some degree of customization to account for individuals’ varying motivations and preferences.

This impact evaluation was implemented as a collaboration between the Nigerian Ministry of Health, the World Bank Development Impact (DIME) team, University College London and Imperial College Business School. For more details on these results, please contact DIME (dime@worldbank.org).

IMPACT NIGERIA IMPACT NIGERIA IMPACT NIGERIAIncentivizing Midwife Retention in Nigerian Primary Health Centers4