GROUP DECISION MAKING WITH UNCERTAIN OUTCOMES: UNPACKING CHILD–PARENT CHOICE OF THE HIGH SCHOOL TRACK

• Published date: 01 May 2016
• Date: 01 May 2016
• DOI: http://doi.org/10.1111/iere.12168
• Author: Pamela Giustinelli

INTERNATIONAL ECONOMIC REVIEW

Vol. 57, No. 2, May 2016

GROUP DECISION MAKING WITH UNCERTAIN OUTCOMES: UNPACKING

CHILD–PARENT CHOICE OF THE HIGH SCHOOL TRACK∗

BYPAMELA GIUSTINELLI1

University of Michigan, U.S.A.

Predicting group decisions under uncertainty requires disentangling individual members’ utilities over the conse-

quences of choice, their expectations for uncertain outcomes, and their choice process as a group. I estimate simple

Bayesian models of child–parent choice of high school track with subjective risk and unilateral or bilateral, nonstrategic

decisions, by combining families’ actual choices with novel survey information about children’s and parents’ subjective

probabilities over choice consequences, their individually preferred choices, and their decision roles. A set of policy

counterfactuals confirms the importance of introducing the beliefs and decision roles of individual members in models

and policy analysis of group decisions.

1. INTRODUCTION

A board of directors deciding to enter a new market or to develop a new product, a couple

choosing a place to live or a contraceptive method, a patient and her doctor selecting a treatment,

or an elderly parent and her children selecting a nursing home are an assorted sample of

consequential choices members of specific “groups” make jointly and in the face of uncertainty.

Revealed preference analysis of any such choice requires a strategy to separately identify (i)

the utility valuations individual agents assign to the future outcomes or consequences of choice,

(ii) agents’ subjective probabilities over the uncertain states, and (iii) the decision process or

protocol they use as a group. This is because multiple configurations of these components are

likely to be observationally equivalent while carrying different implications for prediction and

policy (e.g., Manski, 2000, 2004; Gilboa et al., 2008).2

∗Manuscript received March 2013; revised December 2014.

1I heartily thank Chuck Manski for his helpful comments on this work and for his constant encouragement and

guidance throughout my dissertation, and the other members of my dissertation committee, David Figlio, Joel Mokyr,

and Elie Tamer, for their availability and feedback. I am enormously indebted to Paola Dongili and Diego Lubian for

without their support and friendship this project would have not come into existence. I have greatly benefitted from

patient and insightful discussions with Federico Grigis and from precious inputs at different stages by the following

colleagues: Peter Arcidiacono, Hector Calvo Pardo, Ivan Canay, Matias Cattaneo, Damon Clark, Marco Cosconati,

Adeline Delavande, Jon Gemus, Aldo Heffner, Joel Horowitz, Lars Lefgren, Diego Lubian, Peter McHenry, Aviv

Nevo, Nicola Pavoni, Matthew Shapiro, Zahra Siddique, Chris Taber, Giorgio Topa, Basit Zafar, Claudio Zoli, seminar

participants at the 2009 MOOD Doctoral Workshop, Northwestern University, the VIII Brucchi Luchino Workshop,

University of Verona, the V RES Ph.D. Meeting, University of Michigan SRC, University of Michigan Economics,

University of Maryland AREC, University of East Anglia, Uppsala University, University of Alicante, University

of Konstanz, University of Bern, the 2010 MEA Annual Meeting, the 2010 EEA Summer Meeting, the University

of London Institute of Education, the 2011 AEA Annual Meeting, the 2011 Dauphine Workshop on “Beliefs in

Decision Theory,” University of Michigan CIERS, the 2012 HCEO Family Inequality Workshop on “Family Economics

and Human Capital in the Family,” 2nd Lisbon Research Workshop on “Economics, Statistics and Econometrics of

Education,” Bocconi University, Cornell University, and University of Michigan Decision Consortium, and three

anonymous referees. Warm also thanks to Stefano Rossi for useful suggestions and tireless support with the revisions,

and to Mike Gideon, Adam Karabatakis, and Amanda Sonnega for generous editing assistance. Data collection was

funded by the SSIS of Veneto, whose financial support I gratefully acknowledge together with that of the University of

Verona. All remaining errors are mine. Please address correspondence to: Pamela Giustinelli, Michigan Center on the

Demography of Aging, 426 Thompson St., PO Box 1248, Ann Arbor, MI 48106. E-mail: pamela.giustinelli@gmail.com.

2By revealed preference analysis, I refer to the practice of inferring decision rules from data on observed choices

and of using those inferences to predict behavior in other settings. An example of a decision rule for choice under

573

C

(2016) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

574 GIUSTINELLI

Absent any information about the group decision process behind observed choices, the com-

mon approach thus far has been to treat the group as a “black box” and to make strong

assumptions about the expectations of its members. In this article, I present the first empiri-

cal application that simultaneously departs from treating the family as a monolithic decision

maker and from making strong rationality assumptions about the relevant expectations for

intrafamily decision making under uncertainty. I present and estimate simple Bayesian models

of child–parent choice of high school track with subjective risk and with unilateral or bilateral,

nonstrategic decision making, by combining observed enrollment into distinct tracks for a sam-

ple of Italian 9th graders, with novel survey information about the probabilistic expectations,

individual choice preferences, and choice participation and roles of the students and their par-

ents.3The parameters of interest are the utility weights the child and the parent individually

attribute to the outcomes of choice and, for families whose members report making a bilateral

decision, the weights the child and the parent jointly use to aggregate either their beliefs or their

subjective expected utilities.

Family members hold subjective beliefs, which are directly defined over the consequences of

track choice they use to maximize their subjective expected utilities. For example, the empirical

specification of the child’s and the parent’s subjective expected utilities includes short-term

outcomes, such as the child’s enjoyment, effort, and achievement in high school, and longer

term outcomes such as the child’s opportunities and choices after high school.4This assumption

is both cognitively and descriptively plausible (Gilboa and Schmeidler, 2004), and it eases belief

elicitation. As a downside, it rules out any perceived interdependencies among the probabilities

as well as the utilities of different outcomes.

The Bayesian paradigm of choice under uncertainty implies that decision makers act upon

their beliefs as though the probabilities were known to them, ruling out ambiguity or higher

order beliefs. Additionally, Bayesian group decision making requires that group members

process and aggregate their beliefs and their utilities separately, one outcome at a time, and

that they reveal their beliefs and utilities honestly to one another, thus excluding situations with

conflict of interest among group members, risk sharing, or social planning (e.g., Hylland and

Zeckhauser, 1979; Keeney and Nau, 2011). Bayesian group decision making does not guarantee

that the group choice is efficient in a Paretian sense (e.g., Raiffa, 1968; Hylland and Zeckhauser,

1979). However, under some conditions, it can be represented by a process where individual

members aggregate their subjective expected utilities directly, yielding Pareto-efficient choices,

an implication I test in the data (Keeney and Nau, 2011).5

Enrollment of Italian students into high school tracks—general, technical, or vocational,

with additional subcategories—occurs nonselectively (“open enrollment”) by family choice at

the end of 8th grade, aided by teachers’ recommendations. Curricular specialization during

secondary education or earlier makes track choice consequential and implies a greater uncer-

tainty the younger the students at tracking. Although curricular tracking is the norm among the

OECD countries (see Betts’s, 2011 review), its implementation and institutional features tend

to vary across countries. Italian tracking has both “rigid” and “flexible” features. On the one

hand, different tracks or curricula are generally offered in separate schools, and track-switching

occurs infrequently and can be costly timewise. On the other hand, graduation certificates from

the majority of curricula, including vocational ones, enable students to continue onto college,

uncertainty is subjective expected utility maximization. By group decision process, or protocol, I mean whether and

how the utility valuations and probabilistic beliefs of the individual group members enter and drive the decision of the

group.

3Recent works point to heterogenous decision-making agency of adolescents across families and decision domains

(e.g., Lundberg et al., 2009; Dauphin et al., 2011).

4Recent works in economics of human capital suggest that academic achievement and monetary returns may not be

the only or most important drivers of educational choices (e.g., Jacob and Lefgren, 2007; Zafar, 2013).

5A choice is Pareto-efficient for the group if no other feasible choice would increase the subjective expected utility

of at least one member of the group without decreasing those of the other members (Raiffa, 1968).