Collaborating in strategic experimentation

Last registered on September 14, 2020
View Trial History

Pre-Trial

Trial Information

General Information

Title
Collaborating in strategic experimentation
RCT ID
AEARCTR-0005503
Initial registration date
September 14, 2020

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 14, 2020, 7:37 AM EDT

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

Locations

Country
Netherlands
Region
Amsterdam

Primary Investigator

Name
Katharina Brütt
Affiliation
Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status
In development
Start date
2020-09-15
End date
2020-12-31
Keywords
Other
Additional Keywords
strategic experimentation, team work, free-riding, informational public good, learning, beliefs, bandit problem
JEL code(s)
D81, D83, D91
Secondary IDs
Abstract
This experiment studies how experimentation incentives are shaped by our strategic environment. In particular, this study looks at incentives for experimentation in a laboratory experiment where a successful breakthrough represents a public good, as in Bonatti and Hörner (2011). Varying the observability of experimentation effort as well as whether experimentation is strategic, I investigate how experimentation in collaborative team projects is encouraged. The aim is to further test the theory's underlying assumptions that i) people respond to the information revealed by others’ actions in a manner consistent with Bayesian updating and ii) strategically consider the effect of their actions on others’ beliefs. This project furthermore examines how behavioral motives, such as reciprocity, drive behavior in the context of strategic experimentation.
External Link(s)

Registration Citation

Citation
Brütt, Katharina. 2020. "Collaborating in strategic experimentation." AEA RCT Registry. September 14. https://doi.org/10.1257/rct.5503-1.0
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)
Intervention Start Date
2020-09-15
Intervention End Date
2020-12-31

Primary Outcomes

Primary Outcomes (end points)
The study has two sets of primary outcomes. The first is the level of experimentation, measured by the share individuals invest in a project of unknown quality. The second are subjects beliefs, specifically subjects beliefs about how likely the project is of high quality, their partner's beliefs about the project's quality and the investment share of their partner.
Primary Outcomes (explanation)
See the analysis plan for a detailed explanation of the primary outcomes.

Secondary Outcomes

Secondary Outcomes (end points)
Measures of risk aversion, reciprocal behaviour, and social value orientation.
Secondary Outcomes (explanation)
See the analysis plan for a detailed explanation of the secondary outcomes.

Experimental Design

Experimental Design
This experiment investigates strategic experimentation where a successful breakthrough represents a public good. The experiment uses four treatments, varying the observability of experimentation and whether agents' experimentation is strategic or non-strategic.
Experimental Design Details
The experiment will consist of four treatments, varying the observability of experimentation and whether subjects invest in the same or separate projects.

The strategic treatments (UnobsStrategic and ObsStrategic) closely follows a two-stage variant of the model by Bonatti & Hörner (2017). This two-stage model is specified as follows: There are two agents i = 1,2 that can choose to invest effort x_(i,t )∈[0,1] at t = 1,2 in a project with unknown quality. Doing so entails a private cost of c(x_(i,t )), with c′(x_(i,t )) > 0 and c′′(x_(i,t )) > 0. Both agents get a payoff of Y from that project if a breakthrough occurs. A breakthrough terminates the project. Whether a breakthrough occurs depends on the quality of the project, which can be high or low, and on the effort the two agents invest in that project. p is the common prior that the project is of high quality. Conditional on the project being of high quality, the probability that a breakthrough occurs in period t is given by (x_(1,t )+x_(2,t ))/2 and thus increasing in the effort invested. If the project is of low quality, there will never be a breakthrough. Second-period payoffs are discounted by a common discount factor δ. In UnobsStrategic, agents do not observe their partner’s invested share in the first period. In ObsStrategic, this is observable.

The non-strategic treatments (UnobsNonStrategic and ObsNonStrategic) differ from the strategic treatments, as the two agents work on two different projects. Both agents get a payoff of Y if at least one of the two projects has a breakthrough. This means that agents have the same payoff externality as in the strategic treatments. Conditional on the project being of high quality, the probability that a breakthrough occurs in period t is given by x_(i,t )/2 for the project i is working on. This ensure that the marginal expected payoff from investing is also constant between treatments.

For the experiment, the following parameterization is chosen for all treatments: Y=13, δ=1, c(x_(i,t ))= 2x_(i,t)^2
Randomization Method
Computerized randomization
Randomization Unit
Subjects are first randomly assigned to a matching group. Every period, a subject is randomly assigned to one of the other subjects in the matching group. The randomisation ensures that subjects are not paired to each other in two consecutive periods.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
64 clusters
Sample size: planned number of observations
384 subjects
Sample size (or number of clusters) by treatment arms
96 subjects by treatment. One matching group consists of 6 subjects, so 16 matching groups (clusters) per treatment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Economics & Business Ethics Committee (University of Amsterdam)
IRB Approval Date
2020-01-28
IRB Approval Number
EC 20200128020125
Analysis Plan

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

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials