Reference Points and the Tradeoff between Risk and Incentives

46 Pages Posted: 19 May 2022 Last revised: 6 May 2025

See all articles by Thomas Dohmen

Thomas Dohmen

University of Bonn

Arjan Non

Maastricht University

Tom Stolp

Maastricht University

Abstract

We conduct laboratory experiments to investigate basic predictions of principal-agent theory about the choice of piece rate contracts in the presence of output risk, and provide novel insights that reference dependent preferences affect the tradeoff between risk and incentives. Subjects in our experiments choose their compensation for performing a real-effort task from a menu of linear piece rate and fixed payment combinations. As classical principal-agent models predict, more risk averse individuals choose lower piece rates. However, in contrast to those predictions, we find that low-productivity risk averse workers choose higher piece rates when the riskiness of the environment increases. We hypothesize that reference points affect piece rate choice in risky environments, such that individuals whose expected earnings would exceed (fall below) the reference point in a risk-free environment behave risk averse (seeking) in risky environments. In a second experiment, we exogenously manipulate reference points and confirm this hypothesis.

Keywords: incentive, piece-rate, risk, reference point, laboratory experiment

JEL Classification: D81, D91, M52

Suggested Citation

Dohmen, Thomas and Non, Arjan and Stolp, Tom, Reference Points and the Tradeoff between Risk and Incentives. IZA Discussion Paper No. 14835, Available at SSRN: https://ssrn.com/abstract=4114387

Thomas Dohmen (Contact Author)

University of Bonn ( email )

Regina-Pacis-Weg 3
Postfach 2220
Bonn, D-53012
Germany

Arjan Non

Maastricht University ( email )

P.O. Box 616
Maastricht, Limburg 6200MD
Netherlands

Tom Stolp

Maastricht University ( email )

P.O. Box 616
Maastricht, Limburg 6200MD
Netherlands

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
32
Abstract Views
328
PlumX Metrics