Worker Heterogeneity and Firm Productivity

Last registered on June 10, 2019

View Trial History

Pre-Trial

Trial Information

General Information

Title

Worker Heterogeneity and Firm Productivity

RCT ID

AEARCTR-0004270

Initial registration date

June 07, 2019

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

June 10, 2019, 10:42 PM EDT

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

Locations

Country

India

Region

Primary Investigator

Name

Arkadev Ghosh

Affiliation

Duke University

Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status

On going

Start date

2019-05-03

End date

2020-03-31

Keywords

Firms & Productivity

Additional Keywords

JEL code(s)

Secondary IDs

Abstract

This project will be a Randomized Controlled Trial (RCT) aimed at studying the impact of worker heterogeneity (demographic) on team production in a production plant in India. The plant employs over 900 workers. This study will experimentally vary exposure of workers to non-coreligionists, co-workers from other castes and social groups (by creating worker teams that vary on these dimensions) in order to study whether greater integration at workplace can positively affect perception and in turn impact productivity. We will complement high frequency production data from the firm with a range of baseline and endline survey measures of out-group preferences and perceptions to answer this question.

External Link(s)

Registration Citation

Citation

Ghosh, Arkadev. 2019. "Worker Heterogeneity and Firm Productivity ." AEA RCT Registry. June 10. https://doi.org/10.1257/rct.4270-1.0

Former Citation

Ghosh, Arkadev. 2019. "Worker Heterogeneity and Firm Productivity ." AEA RCT Registry. June 10. https://www.socialscienceregistry.org/trials/4270/history/47934

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

Intervention Start Date

2019-05-03

Intervention End Date

2020-03-31

Primary Outcomes

Primary Outcomes (end points)

- High frequency daily team level output data obtained from the firm

- Different measures of implicit and explicit out-group perceptions and prejudice across treatment and control arms

Primary Outcomes (explanation)

- measures of out-group perceptions and prejudice will be constructed using survey data in order to analyze whether greater integration with individuals from other groups have an affect on beliefs
- there will be direct measures of social distance, for example through survey responses to questions on preferences towards marriage outside one's religion and caste
- there will also an indirect measure using an implicit association test to elicit implicit prejudice

These outcomes will be compared across treatment and control groups at baseline and endline.

Secondary Outcomes

Secondary Outcomes (end points)

Secondary Outcomes (explanation)

Experimental Design

The experiment involves randomizing workers in the firm into teams that vary on ethnic dimensions (religion and caste) in order to analyze the effect of ethnic team composition on daily firm productivity. Team production will be studied over a period of time. The hypothesis we want to test through this intervention is whether greater contact with non-co ethnics can positively affect perception, reduce prejudice and improve firm production. Understanding temporal variation in differences across teams is also an important objective of this study. Production data will be complemented with baseline and endline survey data to understand mechanisms driving productivity differences across groups should they exist.

4 different production lines will be part of the experiment. Within each production line three distinct teams will be randomly created, a control group, and two different treatment arms (treatment1 and treatment2). Therefore, there will be 12 different groups.

Experimental Design Details

The firm will continue to operate in the usual manner, and all workers will face exactly the same incentives (in terms of benefits, pay and work hours) that they otherwise would. The only change will be the team compositions. Each production line in the firm has 3 separate teams that work each shift (morning, afternoon and night) per day. Teams move from morning to afternoon and finally to the night shift every week. The intervention will target the demographic composition of these teams.

Within each production line, 3 distinct groups will be randomly generated. The first group will be composed of only Hindu workers (treatment 1), the second will be a randomly generated mixed group with Muslim workers comprising of 15%-20% of total workers (control) 6. Finally, the third group will be composed of approximately 40% Muslims (treatment 2). Randomization will ensure that caste composition of Hindu workers across the teams is not different, which is important for identification.

The intervention will likely involve 4 different production lines and therefore (4 x 3) = 12 different groups (4 control groups, 4 treatment1 and 4 treatment2).

Randomization Method

Done through STATA on a computer.

Randomization Unit

Individual

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

n/a

Sample size: planned number of observations

400 individuals (at the very least) will be part of the experiment and will be surveyed. Output data is recorded daily at the firm at the team level. The experiment will run for at least 5 months in 4 production lines, and therefore a minimum of (20 days per month x 5 months x 12 groups) = 1200 output data points are expected to be observed.

Sample size (or number of clusters) by treatment arms

- At least 400 individuals will be part of the experiment and will be surveyed approximately 133 in control, 133 in treatment1 and 133 in treatment2.
- Team level output data will be generated on a daily basis for 5 months (duration of intervention) at the very least. Considering 4 groups in each of the arms (control, treatment1 and treatment2) there will be 400 data points per arm.

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Supporting Documents and Materials

IRB