How do Firms Grow? Understanding Knowledge Sharing in Small Firms

Last registered on March 22, 2022


Trial Information

General Information

How do Firms Grow? Understanding Knowledge Sharing in Small Firms
Initial registration date
June 16, 2021

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 17, 2021, 2:34 PM EDT

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

Last updated
March 22, 2022, 2:36 PM EDT

Last updated is the most recent time when changes to the trial's registration were published.



Primary Investigator

University of Warwick, Warwick Business School

Other Primary Investigator(s)

PI Affiliation
Pennsylvania State University

Additional Trial Information

In development
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Knowledge sharing within firms could emerge as a result of human capital spillovers and agglomeration (Lucas Jr, 1988) or interaction among peers (Krugman, 1991). The literature documents that a larger group with better skills is more likely to have the opportunity to exchange ideas, thus increasing knowledge sharing and resulting in an overall larger production outcomes at the group level. However, measuring the effects of interaction using observational data may be subject to selection bias.
While event studies and instrumental strategies have helped establish the importance of intra-group interactions for economic growth, the mechanisms through which social interactions affect knowledge diffusion and productivity remain ambiguous.\\
This study aims at filling this knowledge gap by leveraging a Randomized Controlled Trial (RCT) that will be embedded in a longitudinal study, and will involve approximately 2,000 women in the Indian State of Chhattisgarh. In collaboration with the Chhattisgarh State Rural Livelihood Mission (CGSRLM), we will encourage the creation of new firms through the Self-Help Groups (SHGs) and Village-Level Organization (VLOs) network set up by CGSRLM in rural villages. Importantly, we will randomize the size of these business groups and will administer business training to all groups via smartphones.

Our experimental design allows us to carefully study:

1) How new knowledge diffuses within firms sorted by size;

2) How firms’ production and growth process are impacted by the diffusion of new knowledge.

The RCT and the longitudinal study will be used to assess the causal impact of and the mechanisms through which interaction affects knowledge diffusion and firm growth. Detailed information on individual characteristics, social networks and interactions collected via three rounds of socioeconomic surveys (baseline, midline and endline) will allow us to quantitatively assess how firm members interact with each other, exchange information and make decisions.
To the best of our knowledge, this is the first study to focus on how group size and group members' skill composition impact knowledge diffusion and group performance among small-medium firms. It is also the first study of its kind to encourage the creation of new businesses and to track them through time in order to assess their growth and sustainability.
External Link(s)

Registration Citation

Barboni, Giorgia and Elisa Giannone. 2022. "How do Firms Grow? Understanding Knowledge Sharing in Small Firms." AEA RCT Registry. March 22.
Experimental Details



Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
Entrepreneurship, productivity and business growth are the key outcomes of this study. We will capture these metrics both in the short-term (after 4 weeks from the start of the intervention) and medium-term (approximately 8 weeks from the start of the intervention). Several indicators will be used to assess how productivity and business growth differ across treatment arms, including: changes in occupation; number of finished goods/products at the group level and per group member; time to produce a certain amount of goods; group revenues; group profits.

We are also interested in measuring the extent to which groups introduce innovations in their production process with respect to the business notions learnt during the business training: whether they expand the range of produced goods (e.g. they add new colors/scents to the soap bars, or they start producing a new good); whether they purchase new equipment that was not mentioned during the business training.

In order to dig into the mechanisms leading to business growth, we will map social interactions of group members (who talks to whom, who takes the lead on specific tasks, etc.). We will measure social interactions through specific questions in the socio-economic surveys both at baseline, midline and endline. However, self-reported measures may be subject to recall bias. To overcome this issue, we will collect innovative, real-time data on personal interactions through sensors of social proximity (‘tags’). We are currently piloting the viability of these tags in our context.

The last set of outcomes we are interested in measuring is women's empowerment and agency. We expect that changes in women's entrepreneurship and productivity may translate into improved decision power within the household and the community.
Primary Outcomes (explanation)
Whenever possible, we will aggregate our outcome measures into standardized indexes following Kling, Liebman, and Katz (2007).

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We will implement a Randomized Controlled Trial whereby we will create business groups across 200 villages in the Indian State of Chhattisgarh, of randomly assigned size: small (4/5 members); medium (8/9 members) and large (14/15 members) groups.
We will recruit approximately 2,000 women in the experiment. In each village under study, women will be invited to attend a smartphone-based business training. Then, based on the treatment arm the village has been randomly assigned to, women will be encouraged to start a production activity (namely, producing soap bars) either in small groups (Treatment 1), medium (Treatment 2) or large groups (Treatment 3). All treatment arms will receive the business training -- in this sense, we will not have a ``pure'' control group to test for the impact of training.\\

Three rounds of socio-economic surveys in the form of structured, individual interviews (approx. 60 minutes long) will be performed at different points in time across approximately 8 weeks: before the training is delivered, and 4 and 8 weeks after the training has been delivered, respectively. This will allow us to causally estimate the effects of the group size on outcomes like women’s productivity, entrepreneurship, and empowerment.\\

The business training will follow a “blended” (in-person + digital) approach. After a first in-person meeting with a trainer that will explain the aims of the business training, including the importance of acquiring business skills for empowerment, a video will be shared with women via their smartphones explaining them how to produce soap bars and encouraging them to engage in a group production. After the in-person training kick-off, women are expected to meet in person with the other members of the training group to produce soap bars.

Once the group are formed and operating for the soap production, the design will also include a cross-randomized intervention to test for information diffusion within a group. Specifically, we will ``seed'' information into groups that might be potentially useful to the households (e.g., consumption goods; financial services; health services and products). We will test whether the information shared in larger groups is more likely to spread and/or to increase consumption / use than in smaller groups.
Experimental Design Details
Randomization Method
We will identify and match triplets of villages based on minimum distance and then randomly assign them to either Treatment 1, 2 or 3. Randomization will be done in office by a computer using Stata software.
Randomization Unit
We will randomize at the village level. We expect to form one (two, at maximum) business groups per village.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Approximately 200 villages.
Sample size: planned number of observations
2,000 women across approximately 200 villages in the Mahasamund district of the Indian State of Chhattisgarh.
Sample size (or number of clusters) by treatment arms
67 villages (Treatment 1); 67 villages (Treatment 2); 66 villages (Treatment 3). There will be no pure control (see Experimental Design Section)
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number
Analysis Plan

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

Request Information


Post Trial Information

Study Withdrawal

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

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials