Estimating Preferences from Strictly Concave Budget Restrictions

How to cite

Gerhardt, Holger; Suchy, Rafael. 2022. "Estimating Preferences from Strictly Concave Budget Restrictions." AEA RCT Registry. AEARCTR-0009837. https://doi.org/10.1257/rct.9837-3.1

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

Holger Gerhardt* , Rafael Suchy

* Primary investigator

Registry ID
AEARCTR-0009837
DOI
https://doi.org/10.1257/rct.9837-3.1
First registered
Last updated
Status
Completed
Trial dates
2022-11-22 – 2022-12-07
Countries
Germany (North Rhine-Westphalia)
JEL codes
C91, D01, D81, D90
Registry record
http://www.socialscienceregistry.org/trials/9837

Abstract

We propose an experimental method, Strictly Concave Budget Restrictions, to determine preference parameters precisely with the lowest possible number of observed choices. Our approach extends a method that is commonplace in economics for estimating preference parameters: the analysis of choices that were made subject to linear budget restrictions. Compared to other elicitation methods, budget-based approaches allow for relatively precise estimation of the preference parameters that govern individuals’ choices. This is the case, however, only as long as individuals choose interior allocations. It has turned out across numerous experimental studies that the majority of choices from linear budgets are corner allocations. The main idea of Strictly Concave Budget Restrictions is to encourage interior choices by making corner allocations extremely unattractive. Just like with linear budgets, the estimation of preference parameters is based on rates of substitution, such that simple and transparent estimation strategies are available, and the method is applicable to various domains (e.g., risk, time, and social preferences). In a series of experiments, we investigate whether our method indeed drastically reduces the frequency of corner allocations and whether it improves, compared to linear budget restrictions, the ability to estimate parameters on the individual level and to characterize interindividual heterogeneity.

Intervention

We investigate a novel method: estimating preference parameters from decisions under strictly concave budget restrictions. This method extends an existing method: estimating preference parameters from decisions under linear budget restrictions.

We compare the performance of the novel method to the performance of the existing method. To do so, the participants of our study make intertemporal decisions and decisions under risk under both strictly concave budget restrictions and linear budget restrictions. The order of the type of budget restrictions that our participants face is randomized between participants.

Experimental design

We propose an experimental method, Strictly Concave Budget Restrictions, for the estimation of preference parameters, in particular parameters governing choices under risk and intertemporal choices.

Our approach extends a method that is commonplace in economics for estimating preference parameters: the analysis of choices that are made subject to linear budget restrictions. Compared to other elicitation methods, budget-based approaches allow for relatively precise estimation of the preference parameters that govern individuals’ choices. This is the case, however, only as long as individuals choose interior allocations. It has turned out across numerous experimental studies that the majority of choices from linear budgets are corner allocations. Our central idea to overcome this issue is to make the budget restriction a strictly concave function, thereby making corner allocations unattractive.

We conduct an online study consisting of two experiments. In the first experiment, we estimate time preferences, and in the second experiment, we estimate risk preferences, using both linear and strictly concave budget restrictions. We will investigate whether our method reduces the frequency of corner allocations, compared to linear budget restrictions, and whether it improves the ability to estimate preference parameters on the individual level, thereby also improving the characterization of interindividual heterogeneity.

Design details

Randomization method
Within-subject design: The order in which the two types of budget restrictions are presented to the participants is randomized by the computer. So is the order of the various budget restrictions of each type.
Randomization unit
Individual level.
Clusters
150 participants.
Observations
200 participants: 100 participants in the study on the elicitation of time preferences and 100 participants in the study on the elicitation of risk preferences. Of the 100 participants in the time preference study, 50 will also participate in the risk preference study.
Arms
All treatments are within-subject. In the time preference study, each participant will make 82 decisions of which one half involve linear budget restrictions, while the other half involves strictly concave budget restrictions. In the risk preference study, each participant will make 46 decisions of which one half involve linear budget restrictions, while the other half involves strictly concave budget restrictions. The number of decisions and the number of participants was chosen such that we can compare our results to those of Choi et al. (2007, https://doi.org/10.1257/aer.97.5.1921)—who collected 50 decision each from 93 subjects—and to those of Andreoni and Sprenger (2012, https://doi.org/10.1257/aer.102.7.3333)—who collected 45 decisions each from 97 subjects.
Primary outcomes
The combination of payments that a participant chooses from the budget restriction offered in the respective round.

Papers from this trial

  1. Holger Gerhardt and Rafael Suchy. “Estimating Preference Parameters from Strictly Concave Budget Restrictions.” ECONtribute Discussion Paper No. 336, September 6, 2024. https://www.econtribute.de/RePEc/ajk/ajkdps/ECONtribute_336_2024.pdf. https://www.econtribute.de/RePEc/ajk/ajkdps/ECONtribute_336_2024.pdf

Cited by

No citation data available for this registration yet.

Keywords