Real-effort laboratory experiment on the effects of production uncertainty on agent effort allocation among two inputs.

How to cite

Phipps, Aaron. 2021. "Real-effort laboratory experiment on the effects of production uncertainty on agent effort allocation among two inputs.." AEA RCT Registry. AEARCTR-0007665. https://doi.org/10.1257/rct.7665-1.0

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

Aaron Phipps*

* Primary investigator

Registry ID
AEARCTR-0007665
DOI
https://doi.org/10.1257/rct.7665-1.0
First registered
Last updated
Status
Completed
Trial dates
2016-10-17 – 2018-08-01
Countries
United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/7665

Abstract

Incentives in complex jobs — like education and health— have had mixed results in developed countries. Both these jobs have complex production functions. One way to describe job production complexity is by allowing agents to have uncertainty about how their inputs translate into productivity. When coupled with large output incentives, such uncertainty can induce agents to inefficiently shift their effort allocation to inputs with lower uncertainty, even at the cost of reducing average output. This experiment tests that hypothesis using a two-input real effort lab experiment. Participants are paid based on the number of easy and hard math questions they answer correctly. In treatment, I increase the uncertainty about the marginal payoff for inputs. The experiment is intended to test whether increased marginal uncertainty induces agents to inefficiently switch their effort allocation.

Intervention

Participants use a web browser to engage in addition tasks while a proctor controls the flow of the experiment in an administrative dashboard. The basic innovation is to present participants with a choice of two possible tasks (inputs). In the allowed time, they attempt to successfully complete either task as many times as possible. The two tasks have different difficulty and financial payoffs.

Experimental design

Each participant is presented both a control session and a treatment session (in randomized order). In treatment, the payoffs for each input are randomized. There are two treatment arms: one in which the easy question's payoffs have high variance, and the other in which the hard question's payoffs have high variance.

Design details

Randomization method
Randomization done by a computer at the beginning of each laboratory session. Randomization ensures equal split between treatment arms and the control/treatment session order. Math problems are randomly generated prior to the experiment so that all participants have the same problems.
Randomization unit
Randomized at the individual level within each experimental session. No clustering required because all participants experience treatment and control.
Clusters
60 individuals.
Observations
60 individuals.
Arms
30 with treatment Easy High Variance 30 with treatment Hard High Variance All 60 participants have control and treatment.
Primary outcomes
I measure how participants allocate their time between the two inputs. Based on their performance, I know which input is most effective for a participant. I can then see if the treatment (production uncertainty) induces them to inefficiently switch away from their effective input.

Cited by

No citation data available for this registration yet.

Keywords