x

We are happy to announce that all trial registrations will now be issued DOIs (digital object identifiers). For more information, see here.
Optimizing Payment Formulas for Results-Based Financing: Evidence from a Survey and a Lab-in-the-Field Experiment in Nigeria
Last registered on July 18, 2019

Pre-Trial

Trial Information
General Information
Title
Optimizing Payment Formulas for Results-Based Financing: Evidence from a Survey and a Lab-in-the-Field Experiment in Nigeria
RCT ID
AEARCTR-0002482
Initial registration date
October 05, 2017
Last updated
July 18, 2019 10:56 AM EDT
Location(s)

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Affiliation
Center for Global Development
Other Primary Investigator(s)
PI Affiliation
World Bank
Additional Trial Information
Status
On going
Start date
2017-08-01
End date
2020-08-01
Secondary IDs
Abstract
Results-based financing (RBF) programs for health in low and middle-income countries often offer financial incentives to health care providers or workers to increase the quantity of a selected set of health care services and/or quality measures. PBF programs can induce behavioral responses in several ways, e.g., by communicating what services are considered “priority” and/or by the financial incentives attached to the included measures.

We survey health workers in Nigeria to examine whether they know about and understand an ongoing RBF program. We use an incentivized experiment to examine health workers’ response to the information and the financial incentives (rewards or penalties) implicit in RBF programs. Specifically, health workers are asked to read several partograph cases and to determine the correct actions to be taken based on this information. Within a clinic, we randomize workers into three study arms: (a) information-only; (b) information with rewards for specific activities; and (c) information with penalties for not doing specific activities. We compare health worker performance in each of these three study arms. We also examine whether performance is better for items that are included in the list, carry any financial incentive, and the value of the reward or penalty.
External Link(s)
Registration Citation
Citation
Bauhoff, Sebastian and Eeshani Kandpal. 2019. "Optimizing Payment Formulas for Results-Based Financing: Evidence from a Survey and a Lab-in-the-Field Experiment in Nigeria." AEA RCT Registry. July 18. https://doi.org/10.1257/rct.2482-3.0
Former Citation
Bauhoff, Sebastian and Eeshani Kandpal. 2019. "Optimizing Payment Formulas for Results-Based Financing: Evidence from a Survey and a Lab-in-the-Field Experiment in Nigeria." AEA RCT Registry. July 18. https://www.socialscienceregistry.org/trials/2482/history/50256
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
Interventions
Intervention(s)
In this study, health workers are asked to each read five partograph cases and to identify the correct action(s) to be taken based on this information. Individuals are randomized into three study arms: (a) information-only; (b) information with rewards for specific activities; and (c) information with penalties for not doing specific activities. Participants in all study arms are shown a printed list of tasks. Only the lists for the rewards/penalty arms have a monetary value attached to some (but not all) tasks on this list.

All participants are paid in mobile phone airtime. The rewards and penalty groups receive a base payment to which we add or subtract based on their performance.
Intervention Start Date
2017-08-01
Intervention End Date
2017-10-15
Primary Outcomes
Primary Outcomes (end points)
Proportion, binary measures and z-scores of the share of correct answers. Separately for each of the partographs and combined in indices.
Primary Outcomes (explanation)
We will use three outcomes to evaluate the effect of the three intervention arms:

1. Proportion of correct responses. If participants get all responses correct, we will instead use a binary measure of all correct versus not all correct.
2. Binary measure of “one or more” correct responses.
3. Z-score of correct responses, where the z-score is calculated as (individual score – mean score ) / ( standard deviation )

We will calculate (and examine) these outcomes for each of the five partographs. In addition, we will create equally-weighted indices across all partographs.
Secondary Outcomes
Secondary Outcomes (end points)
Binary measure whether a specific answer option (a clinical task) was selected.
Secondary Outcomes (explanation)
We will use a binary measure of whether a task was mentioned by the participant to analyze the impact of being listed, whether a positive reward/penalty was assigned, and what the level of the penalty/reward is.
Experimental Design
Experimental Design
The experiment will be conducted with health workers who are already being interviewed in primary and secondary care facilities in Nigeria as part of the impact evaluation of a performance-based financing (PBF) project. The larger PBF project has assigned three conditions at the state-level: performance-based financing, decentralized facility financing and a control condition.

The information provided to participants has several features. First, not all “correct” tasks are included on the list. Second, even in the reward/penalty arms, some tasks are not incentivized. Third, the value of the reward/penalty varies across incentivized tasks.

We will conduct two sets of analyses:

1. Comparing the effects of providing information, rewards and penalties. We will compare the outcomes under the three study arms. This analysis will be done the level of the health worker, accounting for clinic-specific effects and clustering at the clinic level.

2. Comparing the effect of (a) whether the task is included in the list given the participants (binary); (b) whether the task has a positive reward/penalty (binary); (c) the level of the reward/penalty (continuous). This analysis will be done at the level of each task included on the list, accounting for worker-specific effects and clustering at the worker level.

The primary analyses will not include additional covariates for the clinics and/or workers. In secondary analyses, we will include selected covariates at the clinic and worker levels that are described below.

In additional secondary analyses, we will examine interactions with indicator variables for the larger projects’ interventions (PBF, decentralized facility financing, control). For observations within the PBF states, we will also conduct secondary analyses comparing those who exhibited knowledge of the PBF program versus those who did not.
Experimental Design Details
Not available
Randomization Method
Randomization is done by the CAPI software in the field.
Randomization Unit
Health workers. In each clinic, one of the three workers to-be-interviewed is assigned to one of the three study arms so that all study arms are represented in each clinic.
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
1,100 health clinics, of which we expect 500 to be in RBF areas.
Sample size: planned number of observations
Up to 3 health workers per clinic who state that they know how to read a partograph. We expect 40% of health workers to report that they cannot read partographs, e.g., the community health workers affiliated with the clinics. We therefore expect a final sample size of about 2,000.
Sample size (or number of clusters) by treatment arms
About 650 health workers for each of the three study arms.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
Health Media Lab IRB
IRB Approval Date
2017-07-26
IRB Approval Number
[no number]