Thin Business Networks and Supply-Chain Frictions: The Impact of a Mobile Phone Networking Application on Small Businesses in Tanzania

Last registered on September 16, 2019

Pre-Trial

General Information

Title

RCT ID

AEARCTR-0004430

Initial registration date

September 13, 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

September 16, 2019, 1:56 PM EDT

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

Last updated

September 16, 2019, 2:08 PM EDT

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

Locations

Country

Tanzania, United Republic of

Region

Primary Investigator

Name

Jessica Rudder

Affiliation

University of Chicago

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Brian Dillon

PI Affiliation

Cornell University

Contact Investigator

Additional Trial Information

Status

On going

Start date

2019-06-10

End date

2020-02-28

Keywords

Firms & Productivity, Other

Additional Keywords

Firm networks, Supply chains

JEL code(s)

L14, L22, D83, O12, O13, D22, R12

Secondary IDs

Abstract

Identifying and alleviating constraints to improve productivity for small and medium enterprises (SMEs) is critical for increasing output and improving incomes in rural communities in low-resource settings. Yet little is known about the specific supply-chain bottlenecks that prevent business owners and entrepreneurs from expanding their operations. This research asks if lowering search costs along the supply chain decreases market frictions by improving small firms ability to source and sell goods in their local area. The proposed field experiment leverages the scale-up of an existing program where small firms can opt to list their sector, location, and phone number in a digital phone book accessible by anyone with a mobile phone. The results will inform researchers and policymakers about what types of technology can move the needle on a firms’ ability to invest in their businesses and improve their operations.

External Link(s)

Registration Citation

Citation

Rudder, Jessica and Brian Dillon. 2019. "Thin Business Networks and Supply-Chain Frictions: The Impact of a Mobile Phone Networking Application on Small Businesses in Tanzania." AEA RCT Registry. September 16. https://doi.org/10.1257/rct.4430-2.0

Sponsors & Partners

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

Request Information

Experimental Details

Interventions

Intervention(s)

The program targets 3 types of participants linked through urban-to-rural supply chains: upstream urban suppliers, rural firms, and downstream rural consumers. The intervention focuses on the middle link of the supply chain: rural firms. These rural firms located in small to medium sized commercial centers are split into a control group and a treatment group with two variations - 1) a business listing in a digital phone book that targets upstream firms, and 2) a business listing in a digital phone book that targets downstream consumers. Random assignment generates variation in the likelihood that rural firms communicate with upstream or downstream contacts.

Intervention Start Date

2019-09-18

Intervention End Date

2020-01-29

Primary Outcomes

Primary Outcomes (end points)

Number of supplier contacts,
Number of customer contacts,
Firms conducts business outside their local market,
Firm engages in price search activities

Primary Outcomes (explanation)

The primary outcome variables concentrate on outcomes related to firm contact with suppliers and customers. During follow-up phone surveys, we will ask whether new suppliers and customers called or came to the storefront. Although treatments are designed to increase either upstream contacts or downstream contacts, these outcomes will be measured for all firms to allow comparison between treatment arms.

The other primary outcome variables - conducting search outside their local market and engaging in price search - are designed to pick up changes in the geographic area over which the firm conducts its business.

Secondary Outcomes

Secondary Outcomes (end points)

Changes in input and output prices of select goods,
Sourcing costs efficiency,
Bargaining activities,
Responses to stock-outs,
Marketing Activities,
Business performance index

Secondary Outcomes (explanation)

Sourcing Efficiency: Firms’ upstream inputs sourcing activity will be used to construct a variable that measures the extent to which treated firms improve productivity in their input market.

Bargaining activities measure the extent to which firms search for price and information that are useful for business decisions.

Stock-outs measures whether customers request relevant goods that the firm does not have in stock. If stock-outs decrease as a result of treatment, it provides evidence that increasing network opportunities could improve availability of goods in local market areas.

The business performance index will be a composite measure of sales, number of customers, number of workers, and hours worked.

Input and output prices: Respondents will be asked to select three primary goods and asked to provide the price they paid for that good. Those items will then be stored and asked again during follow-up surveys in order to learn whether the firm sources or sells goods at different prices.

Experimental Design

First, a sample of communities was drawn by selecting 20 medium-sized villages located between three large city hubs: Dodoma City, Singida City, and Manyoni town. Communities with a population above 3,000 people (per the 2012 Tanzanian census) were eligible. Qualifying communities were selected after stratifying on population and distance to the nearest city. After selecting communities, the research team traveled to each location and invited all businesses to enroll in the digital phone directory and conducted the baseline survey. Firms that agreed to participate became the baseline sample and were randomly divided in the control and two treatments at the individual level, after stratifying on firm sector, market size (number of firms per village), gender of the firm owner, and a measure of how much the firm preferred to have their business listing shown within their local market versus the city market (see PAP for a full explanation).

Experimental Design Details

Randomization Method

Randomization done by computer after collecting baseline outcomes.

Randomization Unit

Individual (firm) level

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

497 firms offered to list business in directory, 413 uptake offer to participate

Sample size: planned number of observations

413 firms

Sample size (or number of clusters) by treatment arms

Upstream treatment: 138
Downstream treatment: 138
Control: 137

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

Power calculations are estimated assuming a level of 0.8 and setting size to .05. With a sample of 413 firms, and 138 per treatment arm, we are powered to pick up changes greater or equal to 0.16 standard deviations. For the first primary outcome (phone calls with upstream and downstream contacts), this translates to a minimum detectable effect size of .51 more phone calls, a 18% increase over the baseline mean of 2.84 phone calls.

Supporting Documents and Materials

IRB