Freeriding and new product adoption

How to cite

Killeen, Grady; Swanson, Nicholas; Cefala, Luisa; Ntawiratsa, Redempteur; Naso, Pedro. 2025. "Freeriding and new product adoption." AEA RCT Registry. AEARCTR-0016091. https://doi.org/10.1257/rct.16091-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Grady Killeen* , Nicholas Swanson , Luisa Cefala , Redempteur Ntawiratsa , Pedro Naso

* Primary investigator

Registry ID
AEARCTR-0016091
DOI
https://doi.org/10.1257/rct.16091-1.1
First registered
Last updated
Status
In development
Trial dates
2025-06-09 – 2026-06-30
Countries
Burundi ()
Registry record
http://www.socialscienceregistry.org/trials/16091

Abstract

This study investigates the role of business stealing concerns and freeriding in retailers' decision to adopt new products. Motivated by a model of strategic experimentation, we investigate whether small retail firms in low and middle-income (LMIC) countries under-adopt new products because of strategic considerations and positive externalities. We implement a randomized controlled trial in Burundi to test these hypotheses in which we assign markets to have an exclusive retail partner of a new good or not and examine whether this increases product adoption. We combine this with pricing experiments to estimate demand and evaluate the tradeoffs between greater technology adoption and monopoly rents.

Intervention

Exclusive contracts: Markets are randomly assigned to an exclusive status or not. In non-exclusive (control) markets, 5 shops are offered stock of a new cookstove non-exclusively, with access for the duration of the study. Other shops in the market may also acquire them. In exclusive (treatment) markets, we approach 5 shops and elicit interest in a contract that gives the shop exclusive access to stock of the product for 6 months. Among those interested, one is then selected at random and the offer is implemented.

Pricing experiments: Midway through the experiment, we conduct randomized pricing experiments to estimate demand, implemented through randomized price increases/decreases or subsidies for the stoves.

Specifically, at the 6 month mark we randomize exclusive markets that took up stoves to a status of either (1) 2 more months of exclusivity or (2) competition. Then across all markets that took up stoves, with stratification by initial treatment status, we randomly assign half of markets to receive about a 20% price discount for 2 months. We offer areas with and without the discounts stock on consignment during this period to eliminate confounds from risk aversion. On top of this we conduct BDM elicitations of demand from 5 consumers in markets that adopted. This variation allows us to test if markets that adopted under exclusivity are less likely to be collusive overall by looking at whether cost passthrough is higher in areas where exclusivity was ended vs control markets. We also plan to estimate competitive conduct in control markets, treated markets, and overall.

Experimental design

Exclusive contract access will be randomized at the market level, stratified by province and market size, determined by below or above median firm counts within the province from a census of firms. We will census all shops and sample 5, prioritizing those who in screening are interested in principle in stocking cookstoves or good industry fits and randomly selecting among those interested. We sample markets 3 kilometers or more apart to limit spillovers.

Firms are permitted to stock beginning after the baseline survey, and we will deliver approximately once per week during the study as we get orders.

We conduct follow up surveys beginning 2 months after baseline, with high frequency pricing surveys between as the budget permits.

Pricing experiments are implemented 3-6 months after baseline.

We may also conduct spillover surveys with unsampled firms 3-6 months after baseline if the budget permits. The aim of these is to measure information spillovers across firms and across markets.

Similarly, we may conduct experiments to study information diffusion during a follow up, namely by informing all shops in some markets about the goods existence to exogenously generate a scenario where knowledge has fully diffused. If we conduct these diffusion experiments, we are focused on how they affect competition/collusion, sales, and prices.

Design details

Randomization method
Computer. We plan to stratify randomization by province and market size (below vs above median firm counts within the province).
Randomization unit
Market
Clusters
Approximately 220 - 300 markets. We sample all possible markets in visited provinces, subject to the 3km spacing and prioritizing larger markets.
Observations
1,000-2,000 (5 per market)
Arms
Equally split between exclusive and non-exclusive (approximately 110-150 in each)
Primary outcomes
Firm level new cookstove adoption. Market level new cookstove adoption. Cookstove price in markets with a seller. Cookstoves sold at the firm and market level. Consumer surplus (1(entrant in market) * area under demand) We are also interested in heterogeneity of primary effects with respect to market size, measures of collusion in the market, and the presence of substitute cookstoves.
Secondary outcomes
Enterprise revenue and profits. Price elasticity (from pricing experiments) Price and quantity of substitutes (other cookstoves)

Cited by

No citation data available for this registration yet.

Keywords