Knowledge Hoarding

Last registered on August 28, 2024
View Trial History

Pre-Trial

Trial Information

General Information

Title
Knowledge Hoarding
RCT ID
AEARCTR-0014206
Initial registration date
August 19, 2024

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
August 28, 2024, 2:45 PM EDT

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

Locations

Country
Burundi
Region
Gitega, Mwaro

Primary Investigator

Name
Nicholas Swanson
Affiliation
University of California, Berkeley
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Luisa Cefala
PI Affiliation
University of California, Berkeley
Contact Investigator
PI Name
Pedro Naso
PI Affiliation
Swedish University of Agricultural Sciences
Contact Investigator
PI Name
Franck Irakoze
PI Affiliation
University of Burundi
Contact Investigator

Additional Trial Information

Status
In development
Start date
2024-08-19
End date
2024-09-30
Keywords
Agriculture, Labor
Additional Keywords
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
The diffusion of new technologies is a crucial driver for growth in low-income countries, especially in the agricultural sector. Motivated by theory, researchers and policymakers have relied on networks as a source of rapid knowledge diffusion, yet often with modest empirical results, suggesting that information transmission is not frictionless. We propose a novel mechanism that might prevent information diffusion: “knowledge hoarding.” In settings where early technology adopters (incumbents) derive rents from it, they might strategically halt its diffusion to preserve the rent their knowledge grants them. This project asks whether knowledge hoarding limit aggregate productivity gains because of distributional consequences generated by knowledge diffusion.

We study this question in the context of modern agricultural technologies in Burundi, where agents already trained in the new technology (incumbents) have a private incentive not to share knowledge in the labor market: they face higher labor demand and a wage premium from mastering the planting technologies. We design two experiments to quantify the effect of knowledge hoarding on knowledge diffusion, and to show the distributional and aggregate consequences of knowledge hoarding on labor market equilibria, technology adoption, and productivity.
External Link(s)

Registration Citation

Citation
Cefala, Luisa et al. 2024. "Knowledge Hoarding ." AEA RCT Registry. August 28. https://doi.org/10.1257/rct.14206-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
In this project, we conduct two different sets of experiments that identify different incentives that workers have to “hoard” knowledge from others. We described one set of experiments in a PAP on the AEA called “Knowledge Hoarding”. In this registration we describe the second set of experiments.

In this experiment, we vary both the identity of the individual to be trained, as well as the task (and whether it is perceived as rivalrous or not) to look at under which conditions individuals hoard knowledge.

In this experiment, we measure the effects of hoarding behavior on labor market outcomes. To do this we identify workers in each village who are skilled in row-planting/microdosage, and workers who are unskilled in these techniques. Villages are then assigned to one of four treatment arms:

1) Same village workers, rivalrous task: skilled workers are invited to a training event. At this training event, they are asked to spend the entire day with unskilled workers. At this event, they are told that they can use the time together to do different activies, one of which is teaching the other workers row-planting and fertilizer microdosage. Unskilled workers in this treatment are from the same village (labor market).

2) Same village workers, placebo task: skilled workers are invited to a training event. At this training event, they are asked to spend the entire day with unskilled workers. At this event, they are told that they can use the time together to do different activies, one of which is teaching the other workers a non-rivalrous agricultural task. Unskilled workers in this treatment are from the same village (labor market).

3) Different village workers, rivalrous task: skilled workers are invited to a training event. At this training event, they are asked to spend the entire day with unskilled workers. At this event, they are told that they can use the time together to do different activities, one of which is teaching the other workers row-planting and fertilizer microdosage. Unskilled workers in this treatment are from a different village (labor market).

4) Different village workers, placebo task: skilled workers are invited to a training event. At this training event, they are asked to spend the entire day with unskilled workers. At this event, they are told that they can use the time together to do different activities, one of which is teaching the other workers a non-rivalrous agricultural task. Unskilled workers in this treatment are from a different village (labor market).
Intervention Start Date
2024-08-19
Intervention End Date
2024-09-30

Primary Outcomes

Primary Outcomes (end points)
Our primary outcome is whether the individual is “trained” in the task taught during the training.

We measure training in two ways: 1) the duration of the training (as reported by the unskilled worker), 2) using the score from some incentivized quizzes that the unskilled workers take at the end of the event.
For the training duration, we compare the (standardized) time that the skilled worker spent training the unskilled in the task.



Primary Outcomes (explanation)
Regarding the quizzes:
In the case of the rivalrous task, the quiz consists of a timed row planting practice, where they need to plant as well as they can a plot of land of a given size. Each pocket is given a score.
In the case of the placebo task, we use a knowledge test.
Based on pilot data, we consider trained an individual who obtained at least 60% of the correct answers. The threshold was established based on piloting and correlating the scores with adoption among skilled workers. However, we will also test some skilled workers in the experiment to measure their score distribution. If the 30th percentile score among skilled workers is significantly higher or lower than our cutoff, we may modify the cutoff based on this distribution.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We sample in all villages a set of laborers who have skills in row-planting/fertilizer microdosage, as well as a set of unskilled laborers in those tasks.

Skilled workers are invited to a training event. At this training event, they are asked to spend the entire day with unskilled workers. At this event, they are told that they can use the time together to do different activities, where the task is either row-planting and microdosage, or another task, depending on the villages treatment condition. In Same-village treatment arms, unskilled workers in this treatment are from the same village (labor market). In Different-village treatment arms, unskilled workers in this treatment are from another village (labor market). Specifically, we take the same workers that were the same village workers for one task, and then use the same laborers as the laborers to be trained for the other task in the different village treatment.

We then measure the likelihood that the worker is trained (as measured by their skill in the task at the end of the event) as described above.
Experimental Design Details
To be consistent with our prior experiment, we will sample workers who are unskilled in row-planting. However, these workers may already be skilled in the non-rival task. To ensure that this does not drive our results, we will collect a baseline measure of skill in the second task and show treatment effects by those who are already skilled (or not).
Randomization Method
Computer
Randomization Unit
Village
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
Approximately 100 villages
Sample size: planned number of observations
We expect to sample 8-12 skilled workers per village, and 8-12 unskilled workers in villages assigned to the “Same village treatment”. This entails a total sample size of 1,600 to 2,400 workers.
Sample size (or number of clusters) by treatment arms
We expect to have around 25 villages per treatment arm
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
The Committee for Protection of Human Subjects - UC Berkeley
IRB Approval Date
2024-03-25
IRB Approval Number
2024-01-17047
Analysis Plan

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