Direct Demand Registration

Last registered on December 18, 2020

Pre-Trial

Trial Information

General Information

Title
Direct Demand Registration
RCT ID
AEARCTR-0006456
Initial registration date
December 18, 2020

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
December 18, 2020, 12:19 PM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Region

Primary Investigator

Affiliation
University of Southern California

Other Primary Investigator(s)

PI Affiliation
Yale University
PI Affiliation
i.e. India
PI Affiliation
Yale University

Additional Trial Information

Status
On going
Start date
2020-09-28
End date
2021-03-31
Secondary IDs
Abstract
The most vulnerable and marginalized citizens - though eligible - are often excluded from safety net programs, even when schemes are self-targeted. The project considers India's workfare guarantee, MGNREGA, which is designed to provide guaranteed work to citizens who request, or "demand" work. Our study directly elicits demand from vulnerable populations, then provides local leaders and program administrators with aggregate information about local work demand. We randomize information provision at two levels to understand complementarities in information sharing across government tiers. We also randomize which communities' work demand is made most salient to higher-level officials, to understand the extent to which there are spillovers or crowding out in service provision. Our study will also explore the role of implementers’ beliefs about work demand in affecting safety-net access, and assess how closing perception gaps related to interest in working affect our outcomes of interest.
External Link(s)

Registration Citation

Citation
Khanna, Madhulika et al. 2020. "Direct Demand Registration." AEA RCT Registry. December 18. https://doi.org/10.1257/rct.6456-1.0
Experimental Details

Interventions

Intervention(s)
We will study the role of information at multiple levels of government hierarchy on accountability to citizens’ requests for short-term employment under India’s public works guarantee program, MGNREGA. Using an IVR survey, we will directly solicit interest in working under the program among a high-vulnerability group, migrants who have returned to their home state of Bihar after the Covid-19 pandemic-related lockdowns. We will then share information about migrant interest in work with officials at the block- (supervisory) and Panchayat-levels (village cluster) responsible for implementing MGNREGA. This information will be shared in a format that will closely mimic the internal communication shared by the Government of Bihar. We will run a second IVR survey of migrants to understand how access to the safety net program changes for migrants and utilize administrative data to understand the impact on the overall access to employment in communities. We will also explore the role of implementers’ beliefs about migrant interest in work on changes in safety net access and well-being.
Intervention Start Date
2020-10-27
Intervention End Date
2021-01-23

Primary Outcomes

Primary Outcomes (end points)
Our analysis will focus on the following families of outcomes:
1) Officials’ beliefs about the demand for work under MGNREGA.
2) Demand for work under MGNREGA.*
3) Steps taken to ensure access to work: These include number of job cards provided, additional funds requested for etc.* **
4) Work provided under MGNREGA*

Outcomes marked by a single asterisk (*) will be measured using IVR surveys with migrants. Outcomes marked by a double asterisk (**) will be measured using phone surveys with officials.
Primary Outcomes (explanation)
The aim of this experiment is to measure the impacts of information on migrant households’ demand for work under MGNREGA on the level of work provided and work received specifically by migrants and their families.

Secondary Outcomes

Secondary Outcomes (end points)
In addition to the outcomes listed above, we examine the effect on the following secondary outcomes:

1) Return Migration decisions*
2) Measures of target households’ well-being*

Outcomes marked with a single asterisk (*) will be measured using IVR surveys with migrants.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
The sampled panchayats come from 145 blocks which we will divide into two block-level intervention nodes:

A.1: Block officials not given the information
A.2: Block officials given the information

A cross-cutting intervention will randomize Panchayat within the control and treatment arms of the block-level intervention. In the treatment arm, local officials, elected and appointed, will receive information about MGNREGA demand among migrants’ families. Since only a random subsample of Panchayats will be included in the information that we will share with block-level officials in the treatment arm, we will equally divide the sample of 1,334 Panchayats across 14 districts in Bihar into six intervention nodes:

B.1: Block officials not given the information + Panchayat officials given the information-sharing product
B.2: Block officials not given the information + Panchayat officials not given the information-sharing product
B.3: Block officials given the information + Panchayat is on the block list + Panchayat officials given the information-sharing product
B.4: Block officials given the information + Panchayat is on the block list + Panchayat officials not given the information-sharing product
B.5: Block officials given the information + Panchayat is not on the block list + Panchayat officials given the information-sharing product
B.6: Block officials given the information + Panchayat is not on the block list + Panchayat officials not given the information-sharing product
Experimental Design Details
Randomization Method
Randomization was conducted in the office via a Stata program.
Randomization Unit
The intervention of sharing information with the block-level officials is at the block level. Further, a random subsample of Panchayats will be included in the list shared with the block-level officials. The intervention of sharing information with Panchayat-level officials and leaders is at the Panchayat level.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
145 clusters
Sample size: planned number of observations
Elected and contractual Panchayat-level officials form 1,334 Panchayats, contractual block-level officials from 145 blocks, and IVR calls to 403,490 unique phone numbers in 14 districts
Sample size (or number of clusters) by treatment arms
We will almost equally divide 1,334 Panchayats into six intervention nodes. Panchayats are divided into six intervention nodes such that the block-level treatment arm that does not provide any information-sharing packages to block-level officials (A1) has two Panchayat-level intervention nodes (B1, B2) and the treatment arm that provides information-sharing packages to block-level officials (A2) has four intervention nodes (B3, B4, B5, B6). Therefore, about one-third of the 145 blocks will be placed in the treatment arm that does not provide any information-sharing packages, and the remaining two-thirds will be placed in the treatment arm that provides information-sharing packages. We will reach out to 403,490 unique phone numbers; other research we have completed using IVR surveys suggests we can expect a response rate between 15% and 20%, leaving us with responses from roughly 60,524 to 80,698 migrants, almost equally split across the six intervention nodes.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We will use three sources to measure outcomes: (1) administrative data on MGNREGA, (2) data from phone surveys with Panchayat- and block-level officials, and (3) data from IVR surveys with migrants. The number of observations from the first two sources will be the same, that is, we will get data from approximately 145 blocks and 1,334 Panchayats. We expect a response rate of about 20% in the IVR survey and since we run the IVR survey in only those Panchayats that have at least 200 migrants, we expect to get responses from at least 40 migrants in a Panchayat. We followed convention in the social sciences for power calculations, using a significance level (probability of Type I error) of 0.05 and power (probability of avoiding a Type II error) of 0.8. Using administrative from the previous year from around the time when the intervention will be launched, we calculate the intra-cluster correlation in person-days worked under MGNREGA to be around 0.10. For the block-level intervention, we expect to detect a minimum effect size of 0.11 units when using administrative data and 0.03 units when using the IVR survey data. To compare any two intervention nodes for the Panchayat-level intervention, we expect to detect a minimum effect size of 0.13 units when using administrative data and of 0.03 units when using the IVR survey data.
IRB

Institutional Review Boards (IRBs)

IRB Name
Institute of Financial Management and Research (IFMR)
IRB Approval Date
2020-09-30
IRB Approval Number
IRB00007107
IRB Name
Yale University
IRB Approval Date
2020-09-16
IRB Approval Number
2000027920
Analysis Plan

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

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials