Strategic experimentation among microenterprises

Last registered on March 06, 2024


Trial Information

General Information

Strategic experimentation among microenterprises
Initial registration date
February 26, 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
March 06, 2024, 3:32 PM EST

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


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

Request Information

Primary Investigator

UC Berkeley

Other Primary Investigator(s)

Additional Trial Information

In development
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
A large body of macroeconomic evidence suggests that lower innovation explains a lack of income convergence between poor and rich countries. But the microeconomic mechanisms through which poverty distorts innovation are not well understood. This study aims to test whether risk aversion and supply chain uncertainty reduce the experimentation of microenterprises below socially efficient levels. The study also pilots a secondary intervention aimed at understanding the role of non-excludability. I aim to test whether these frictions bind using a randomized controlled trial (RCT) in which small retailers are offered a stock of a new product which prior work suggests has unmet demand. I plan to randomize whether enterprises have the ability to return units of a first order or guaranteed access to a permanent supplier. I also plan to pilot exclusive access to stock of the product, but this is secondary because of natural expansion of the market. I will examine whether each intervention increases experimentation with the new product and whether the treatments induce higher permanent entry into the market, which would indicate that experimentation is inefficiently low absent the interventions. These treatments are designed to test a model of firm experimentation that generalizes to other microenterprises, and a set of secondary mechanism experiments may be completed later to rule out confounds. This study takes place in the context of a new motorcycle helmet, yielding large potential public health benefits from the research.
External Link(s)

Registration Citation

Killeen, Grady. 2024. "Strategic experimentation among microenterprises." AEA RCT Registry. March 06.
Sponsors & Partners

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

Request Information
Experimental Details


This study aims to test whether small firms experiment efficiently with new products. Specifically, the null hypotheses are that risk aversion, supply uncertainty, and (as a secondary outcome) non-excludability do not affect enterprises' likelihood of experimenting with a new product. To test this hypothesis, we aim to include two primary treatment arms. The first targets risk aversion by allowing firms to return unsold products from a first batch. The second provides long-run supply certainty by offering to facilitate a relationship between the firm and the product manufacturer at the end of the study. This combination of interventions also aims to test whether bandit models of experimentation accurately predict behavior in low and middle income settings. The study also aims to address whether barriers to experimentation leads to too few firms in the market, measured via an increase in the number of sustained entrants. Third, we may pilot an intervention aimed at examining non-excludability that would include a commitment not to supply helmets to firms nearby treated enterprises, but this intervention is secondary and may be dropped if it creates sampling challenges.

In addition to these main interventions, at the endline survey we aim to offer a subset of non-entrants sales data from other firms to test whether information alone is sufficient to induce non-experimenters to enter the market.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
1.) Experimentation with new product sales, measured via initial uptake. This will be measured from surveys and administrative data.
2.) Sustained entry into the market, measured via subsequent product orders. This will be measured from surveys and administrative data.
3.) Beliefs about the profitability of the new products. This will be measured from surveys.
4.) Enterprise revenue, costs, and profits. This will be measured from surveys.

The market for helmets saw large expansion between the pilot and main experiment, making it difficult to avoid including some shops near existing sellers. Hence, heterogeneity based on the number of existing sellers in an area may be important to analysis.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Control: Shops can place helmet orders with the study team for the duration of the intervention, with order sizes of 5 or more.
Risk treatment: Control + shops have the option to return unsold units of their first order for a refund during a follow-up survey.
Supply treatment: Control + the study team offers to facilitate a relationship with the helmet manufacturer at the end of the intervention.
Secondary information treatment: At the end of the intervention, a subset of non-experimenters are provided with anonymized helmet sales data from other shops.
Secondary excludability treatment: Study team commits not to recruit/supply nearby shops.

The risk and supply treatments are primary and will be cross-randomly assigned to half of enterprises each. The secondary information treatment will be assigned to half of non-experimenters at the endline survey. The secondary excludability will be offered to a random subset of enterprises at the start of the experiment if it is maintained.
Experimental Design Details
Not available
Randomization Method
Randomization will be done by a computer. Randomization will be stratified by an indicator for having one or more employees, geographic area, and an indicator for being above average distance to the nearest known existing helmet seller, measured via GPS.
Randomization Unit
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
1000 firms
Sample size: planned number of observations
Around 1000 firms
Sample size (or number of clusters) by treatment arms
- About 250 control, 250 risk treatment, 250 supply treatment, 250 both
- The exact numbers will vary slightly since randomization is stratified
- If it is maintained, 50-150 will also be assigned to the excludability treatment
- The information treatment depends on the number of non-adopters, but will include half of the non-adopters
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Institutional Review Boards (IRBs)

IRB Name
UC Berkeley Committee for Protection of Human Subjects
IRB Approval Date
IRB Approval Number