Prospect Theory Goes Public: Experimental Evidence on Cognitive Biases in Public Policy and Management Decisions

• Published date: 01 November 2018
• DOI: http://doi.org/10.1111/puar.12960
• Date: 01 November 2018

828 Public Administration Review • Novem ber | Dece mber 2 018

Public Administration Review,

Vol. 78, Iss. 6, pp. 828–840. © 2018 by

The American Society for Public Administration.

DOI: 10.1111/puar.12960.

Nicola Bellé

Sant’Anna School of Advanced Studies, Italy

Paola Cantarelli

CERGAS Bocconi, Italy

Paolo Belardinelli

Bocconi University, Italy

Prospect Theory Goes Public:

Experimental Evidence on Cognitive Biases in

Public Policy and Management Decisions

Paolo Belardinelli is a PhD candidate

at Bocconi University, Milan, Italy. His

research focuses on behavioral public

administration.

E-mail: paolo.belardinelli@unibocconi.it

Paola Cantarelli is a postdoctoral

scholar at CERGAS Bocconi, Bocconi

University, Milan, Italy. Her research focuses

on work motivation and behavioral human

resources in public organizations.

E-mail: paola.cantarelli@unibocconi.it

Nicola Bellé is assistant professor at the

Sant’Anna School of Advanced Studies, Pisa,

Italy. His research focuses on behavioral

public management and administration.

E-mail: nicola.belle@santannapisa.it

Abstract: This article tests a broad range of cognitive biases branching out from prospect theory in the context of

public policy and management. Results illuminate systematic deviations from rationality. In experiments 1 through

5, the framing of outcomes influenced decisions across policy and management domains. In experiment 6, public

employees were prone to an anchoring bias when setting standards for responsiveness. Experiment 7 shows that public

workers tend to put more effort into activities that affect higher percentages of beneficiaries, even if the absolute

number of affected clients is constant. Experiments 8 and 9 suggest that public employees are more likely to stick to

a suboptimal status quo as the number of superior alternatives increases. Experiment 10 provides evidence of an

asymmetric dominance effect: decisions changed when a decoy was present. This article contributes to behavioral public

administration by replicating and extending previous trials.

Evidence for Practice

• Being humans rather than robots, civil servants are prone to cognitive biases that may hinder public service

provision.

• As these cognitive biases are systematic rather than random, the architects of public organizations and services

should account for them when designing the architecture of jobs and tasks.

• Supposedly irrelevant factors that predictably affect decisions include the framing of outcomes, anchors,

multiple alternatives, and decoy options.

Behavioral sciences suggest that public servants’

judgments may be systematically biased under

certain circumstances. This conflicts with

decision-making models that are based on expected

utility theory, which features rational agents (Bernoulli

1954). In this article, we investigate how public

managers and employees actually make decisions

by putting to an experimental test a number of

cognitive biases that branch out from or are associated

with prospect theory (Kahneman 2011; Kahneman

and Tversky 1979). From an epistemological

standpoint, our work aims to experimentally test

descriptive models “that accurately portray human

behavior” (Thaler 2015, 30) in the context of public

management and policy. We do so by replicating and

extending previous randomized controlled trials.

Whereas behavioral public administration (e.g.,

Grimmelikhuijsen et al. 2017) studies tend to focus

on one or a few cognitive limitations at a time and

rely on samples of citizens (e.g., Baekgaard 2017;

Baekgaard and Serritzlew 2016; Barrows et al. 2016;

Jilke, Van Ryzin, and Van de Walle 2016; Marvel

2016; Olsen 2017), this article tests the effects of a

broad range of cognitive biases across multiple policy

areas and managerial tasks. To this end, we conducted

10 randomized controlled trials to explore how

framing, anchoring, proportion dominance, status

quo, and asymmetric dominance affected the decisions

of 600 Italian public workers from different industries.

We selected these cognitive biases starting from a

more comprehensive list and narrowing it down to the

ones that we believe have more direct implications for

public policy making and public management.

Theoretical Background

Before the 1940s, expected utility theory was the

dominant model used to describe decision making.

At its core, expected utility theory features a rational

decision maker who has clear and comprehensive

knowledge of the environment, a well-organized

system of preferences, and excellent computational

skills to allow for the selection of optimal solutions

(Bernoulli 1954). Starting with the work of Herbert

Simon in the 1940s and 1950s, scholars have

contended instead that decision makers are endowed

with bounded rationality. Simon (1947, 1956)

claimed that decision makers do not optimize but

rather “satisfice.” Literally a combination of “satisfy”

and “suffice,” the satisficing principle suggests that

Research Article