Market-level Effects of Competition: Prices, Quality, and Mechanisms

Last registered on February 24, 2022


Trial Information

General Information

Market-level Effects of Competition: Prices, Quality, and Mechanisms
Initial registration date
February 23, 2022

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
February 24, 2022, 1:31 PM EST

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



Primary Investigator

University of Sydney

Other Primary Investigator(s)

PI Affiliation
UW Madison
PI Affiliation
UW Madison
PI Affiliation
Tegemeo Institute of Agricultural Policy and Development
PI Affiliation
University of Chicago

Additional Trial Information

On going
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Conventional wisdom has it that market competition drives down prices. The effect of competition on product quality, however, is more ambiguous. This project uses a market-level experiment to identify the causal effect of market entry on agricultural input markets. We randomize the rollout of new agricultural retail stores in Kenya, where agricultural input markets are served by small-scale retail shops with spatial monopoly power. We use this experimental variation in competitive pressure to estimate the effect on prices, product quality, and firm behavior including entry and exit decisions.
External Link(s)

Registration Citation

Deutschmann, Joshua et al. 2022. "Market-level Effects of Competition: Prices, Quality, and Mechanisms." AEA RCT Registry. February 24.
Sponsors & Partners


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

Request Information
Experimental Details


The NGO is opening agro-input shops, in which they will sell high-quality hybrid seeds and fertilizers. Consumers will be able to obtain any quantity of inputs at any time during the year.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
- Input quality at incumbent stores
- Input prices at incumbent stores
Primary Outcomes (explanation)
For each of the target inputs, we obtain samples using covert shoppers and record the purchase price. The key inputs are
1) DAP (diammonium phosphate) fertilizer
2) hybrid maize seed

For DAP, we measure quality using lab tests (conducted by an ISO-certified laboratory)
For hybrid seeds, we measure quality by controlled germination rates (also conducted by a laboratory)

Our main outcome here is the price/kg at incumbent stores

Secondary Outcomes

Secondary Outcomes (end points)
- Incumbent firm demand
- Entry and exit of agro-input stores in the study catchment areas
- Firm behaviors, strategies, and beliefs
Secondary Outcomes (explanation)
Demand is elicited as part of the firm survey (number of customers per day during peak season)

Number of agro-input incumbent firms that exit between survey rounds.
Measured via a geo-tagged census of all agro-input stores in the study catchment areas

Number of new agro-input retail entrants in the study catchment areas.
Measured via a geo-tagged census of all agro-input stores in the study catchment areas

Stocking, suppliers that they source from, investments in the business, and pricing behaviors
Product mix

Firm perceptions of the competitive environment
Firm beliefs about quality concerns in the market for agricultural inputs

Experimental Design

Experimental Design
We partner with a large and established NGO that is opening agro-input stores in Kenya. The NGO provided us with a list of 100 locations where they wanted to open retail shops, and allowed us to randomize the roll-out. The randomized roll-out provides the key source of experimental variation for the study.

- Market centers: defined as the 100 locations where the NGO planned to open retail shops
- Catchment areas: defined by all locations within 10 km of the market center
- Entrants: the NGO-run stores

Each round of data collection takes place as three separate activities:
1) We conduct market censuses in all catchment areas. Specifically, we list, geo-tag, and photograph each agro-input retailer within the 10-km catchment area (i.e., all stores that sell seeds or fertilizer).
2) We employ covert (mystery) shoppers, who purchase seed and DAP fertilizer from 10 randomly sampled stores within each catchment area listed in the census. We sample the mystery-shopped stores withour replacement (i.e., if catchment areas overlap, we exclude shops already selected to be surveyed based on also falling in another catchment area).
3) We conduct firm surveys with the same stores that we sent mystery shoppers to. This exercise is separated in time from the mystery shopping activity to ensure that the covert shoppers remain covert.

Experimental Design Details
Randomization Method
We randomized market centers into early and late roll-out using pairwise randomization. We used the following variables to create the pairs:
1) An indicator of whether the entrant location is within 1 km of a main road
2) Altitude
3) Number of incumbent stores within 5 km of the entrant location
4) Population density within 10 km of the entrant location

If the NGO faces difficulties opening stores in some of their planned treatment locations (for example if they can not find an appropriate piece of land), the pairwise randomization will enable us to drop that location and it’s matched control location.

- Distance to main road: we label all roads classified as trunk or primary roads according to the Kenyan road system as "main roads"
- Number of incumbents: we proxy the number of stores in an area using a 2015 census conducted by McKinsey of all agro-input stores in Kenya
- Population density: we use population density estimates from WorldPop (2013)
- Altitude of the market centers: from a Digital Elevation Model done by the USGS Shuttle Radar Topography Mission

Using the above variables, we constructed pairs of market centers following the procedure employed by King et al. (2007), i.e.:
* Calculate the Mahalanobis distance for every possible pair of market centers within each region
* Choose the pair with the smallest distance and remove it
* Repeat these steps until all stores have been matched
Randomization Unit
Market centers
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
100 markets
Sample size: planned number of observations
2,000 fertilizer samples; 2,000 maize seed samples, 1000 firm surveys
Sample size (or number of clusters) by treatment arms
50 markets control 50 markets treatment
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


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