|
Field
Trial Status
|
Before
in_development
|
After
on_going
|
|
Field
Last Published
|
Before
November 07, 2016 03:43 PM
|
After
December 20, 2017 01:42 PM
|
|
Field
Intervention (Public)
|
Before
|
After
This randomized control trial (RCT), which takes place at scale in two large Indian states, investigates how wage payment delays to workers in the government's large workfare program are affected by decreasing costs of acquiring management-relevant information by lower-level bureaucrats and reducing the costs of monitoring by higher-level bureaucrats.
|
|
Field
Primary Outcomes (End Points)
|
Before
Our main outcomes of interest are average days taken to pay wages, both overall and by payment sub-step, and MGNREGA activity (expenditure, participants, days worked).
|
After
Our main outcomes of interest are the mean and absolute average deviation in the time to payment for payment sub-steps under the purview of block and district officials, as captured through administrative data used in our system.
|
|
Field
Experimental Design (Public)
|
Before
We will evaluate PayDash using an RCT in the states of Chhattisgarh, Jharkhand, and Madhya Pradesh. The three study states together are composed of 98 districts and 704 blocks. We conducted a pilot roll-out in one district in each of our study states in summer 2016. The full-scale intervention will be implemented across the set of 95 non-pilot districts beginning in fall 2016.Our evaluation design has four district-level treatment arms (block PayDash; district PayDash; district+block PayDash; and control). Approximately 23-24 districts, covering roughly 171 blocks, will be randomly assigned to each treatment arm. Within each state, random assignment of districts into arms is stratified on above/below median: average days to payment across transactions and total transactions over the previous year. We will conduct heterogeneity analysis of treatment impacts by officer-specific traits including intrinsic motivation.
|
After
We randomized access to PayDash at the district level to the following treatment arms:
1.) Control: Both block and district-level officials overseeing the wage payment process must rely on status quo information sources about payment processing. These sources are readily available but time consuming to compile and examine on a regular basis.
2.) District only PayDash access: Only the highest level of supervisors across the districts receive access to the PayDash platform, ensuring they are better able to monitor payment activity and delays managed by lower-level block officials.
3.) Block only PayDash access: Only the lower-level managers (block officials) receive access to PayDash. Information in the platform helps these officials identify and process payments in their jurisdiction.
4.) District + Block PayDash: Both district and block-level MGNREGA officials have individualized access to the platform.
|
|
Field
Randomization Unit
|
Before
We randomly assign two cross-cutting treatments at the district level: provision of PayDash to (1) district officials and (2) block officials.
|
After
A randomized control trial across multiple states will be conducted in which we provide the platform to different levels of the bureaucratic hierarchy responsible for program administration.
Across the three states, we will randomly assign two cross-cutting treatments at the district level: provision of PayDash to (1) district officials and (2) block officials.
|
|
Field
Planned Number of Clusters
|
Before
98 districts
|
After
73 districts. The randomization strategy includes providing access to the platform to all lower block-level officials in randomly selected districts. In addition to the control arm, our study includes districts where only district-level officials have access to the platform (TD), districts where only lower block-level officials have access (TB), and districts where both district and block-level officials have access (TDB).
|
|
Field
Planned Number of Observations
|
Before
Approximately 8,400 block-months.
|
After
73 districts, over 12+ months
|
|
Field
Sample size (or number of clusters) by treatment arms
|
Before
704 blocks: approximately 176 blocks for each of 4 treatment arms (Block PayDash, District PayDash, District+Block Paydash, and Control)
|
After
In Madhya Pradesh, 50 districts were split equally across four treatment arms. In the second state, Jharkhand, the four treatment categories were randomly assigned across 23 districts in approximate proportions of Control: 1/3; TD: 1/6; TB: 1/6, TDB: 1/3.
|
|
Field
Power calculation: Minimum Detectable Effect Size for Main Outcomes
|
Before
|
After
In Madhya Pradesh, the four treatment categories were randomly assigned across 50 districts, excluding a pilot district, in approximately equal proportions, while across 23 districts in Jharkhand we assigned approximately twice as many districts to the District+Block PayDash and control arms than to the District and Block PayDash arms. This sample size powers us to detect a minimum effect size of 0.187 standard deviations (SD) for comparisons of control to Block+District PayDash (translating into delay decreases of 1.97 days or more in the steps of the payment process under officer purview), of 0.205 SD for comparisons of either control or Block+District PayDash to either Block or District PayDash, and of 0.217 SD for comparisons of Block and District PayDash.
|
|
Field
Secondary Outcomes (End Points)
|
Before
|
After
We also will examine MGNREGA-related outcomes available from the MGNREGA MIS, including work requested, number of person-days worked, and total wage and materials expenditure. Contingent on funding availability, we will conduct endline surveys and qualitative interviews with officials to understand changes in their management practices, time allocation, and professional networks, and to obtain additional platform design feedback useful to scale-up.
|