Wage-pension tradeoff
How to cite
Guillemyn, Inés; Deschacht, Nick; Vujić, Sunčica. 2025. "Wage-pension tradeoff." AEA RCT Registry. AEARCTR-0015780. https://doi.org/10.1257/rct.15780-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
The objective of this study is to estimate worker's willingness to pay for current over future income. We do so by analysing worker's choice of job through an online discrete choice experiment. Using fictitious job offers, offering different wage-pension bundles, we estimate how much in current wage workers are willing to forego for an increase in pension benefits.
Intervention
The intervention is the random assignment of job offers.
Experimental design
Estimating the monetary trade-off between wages and pension benefits comes with a number of identification problems. These problems arise from unobservable characteristics, and a lack of data on individual's outside options. We try to overcome these identification issues by conducting a discrete choice experiment (DCE) using fictive job offers, each offering different combinations of wages and pensions.
Each job offer included in our study has the same attributes, but at different levels, including pension benefits and wages. Additional attributes which varied were commuting time and the possibility to work from home. These attributes were added to the vignettes, as to not reveal the purpose of our study.
Respondents were instructed to assume that the job offers were similar to each other and that they included similar tasks as in their current job. The wage offers which were included in our vignettes were a percentage of their actual wage, which they had to report earlier in the survey. Respondents who did not wish to share information about their wage, were asked to provide their wage in income brackets. Pension benefits were included by multiplying respondent's wage with a randomized replacement rate.
The product of all possible vignettes resulted in a vignette universe of 128 possible combinations. To reduce the number of vignettes respondents had to evaluate, and to minimize the variance of the parameters which we wish to estimate, we used a D-efficient experimental design. Using the D-efficient design, we drew 30 vignettes from the full vignette universe and 15 choice sets. The D-efficient design should result in a combination of vignettes which are not correlated to each other, and have a balanced distribution of attributes and levels. Job offers were randomly assigned to respondents.
Design details
Cited by
No citation data available for this registration yet.