Life Expectancy, Intragenerational Redistribution, and the Alienation Effect: An Experimental Study with and without a Veil of Ignorance
How to cite
Traub, Stefan; Meemann, Christine; Rikker, Fabian. 2026. "Life Expectancy, Intragenerational Redistribution, and the Alienation Effect: An Experimental Study with and without a Veil of Ignorance." AEA RCT Registry. AEARCTR-0018304. https://doi.org/10.1257/rct.18304-1.0
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
Differences in life expectancy related to socio-economic status have a regressive redistributive effect on defined-contribution pension systems, whereby short-lived low-income groups subsidize the pensions of long-lived high-income groups. Krieger et al. (2025) demonstrate in an experiment that confronting individuals with income-dependent differences in mortality under the assumption of a veil of ignorance (VOI) reduces preferences for redistribution when redistribution involves an efficiency loss (“leaky bucket”). They refer to this phenomenon as the “alienation effect.” This study replicates and extends their experiment using a 2×2 between-subjects design with 120 participants, systematically varying the factors of positional knowledge (with and without VOI) and mortality risk structure (symmetric and asymmetric with a “leaky bucket” mechanism). The goal of the present study is to determine whether the alienation effect occurs even when income positions are known, and whether it is stronger or weaker than when distribution decisions are made from under a VOI.
Intervention
Participants are randomly assigned to one of four between-subjects treatments, which are explained below. They are then randomly assigned to groups of five. They are presented with a table containing five income levels (A through E), each associated with pension entitlements and corresponding work contributions. Using a slider to adjust a redistribution parameter lambda, they must then decide how much they want to redistribute the entitlements, where lambda=0 means no intragenerational redistribution and lambda=100 means an egalitarian distribution. There are 4 rounds per group, in which the variance and mean of the entitlements and pension payouts, respectively, are varied. At the end, the group median of lambda from one of the four rounds determines the actual redistribution or the payouts. Before participants receive the payout corresponding to their income position, they must perform a real-effort task (slider task) in accordance with the specifications in the table (higher income positions involve more sliders). Additionally, there is a certain probability that one of the group members will not receive a payout even though the task was completed (this simulates the mortality risk of the insured in a pension plan).
The four treatment variations are:
T1 Equal Risk with VOI
In T1, participants do not know their income rank within the group before they set the redistribution parameter lambda. Furthermore, the probability of receiving no payout is the same for all group members (20%).
T2 Asymmetric Risk with VOI
In T2, participants do not know their income position within the group before they set the redistribution parameter lambda. Furthermore, while the probability of receiving no payout is also 20% on average, it is lower for group members with higher entitlements (A: 33, B: 27, C: 20, D: 13, E: 7)%, with A having the lowest entitlement and E the highest.
T3 Equal Risk w/o VOI
In T3, participants know their income rank within the group before they set the redistribution parameter lambda. Furthermore, the probability of receiving no payout is the same for all group members (20%).
T4 Asymmetric Risk w/o VOI
In T4, participants know their income position within the group before they set the redistribution parameter lambda. Furthermore, while the probability of receiving no payout is also 20% on average, it is lower for group members with higher entitlements (A: 33, B: 27, C: 20, D: 13, E: 7)%, with A having the lowest entitlement and E the highest.
Experimental design
The experiment replicates and extends the design by Krieger, Meemann and Traub (2025, JEBO).
1. Participants are randomly assigned to one of four between-subjects treatments, which are explained below. T1 and T2 replicate treatments ER and SRL from Krieger et al. (2025).
2. After having been assigned to a treatment, participants are randomly assigned to groups of five. They are presented with a table containing five income levels (A through E), each associated with pension entitlements and corresponding work contributions (sliders to be solved in a real-effort slider task, Gil and Prowse (2012, AER).
3. Using a slider to adjust a redistribution parameter lambda, they must then decide how much they want to redistribute the entitlements, where lambda=0 means no intragenerational redistribution and lambda=100 means an egalitarian distribution.
4. There are 4 rounds per group, in which the variance and mean of the entitlements and pension payouts, respectively, are varied (the numbers are taken from Krieger et al. (2025).
5. At the end, the group median of lambda from one of the four rounds determines the actual redistribution or the payouts.
6. However, before participants receive the payout corresponding to their income position, they must perform a real-effort task (slider task) in accordance with the specifications in the table (higher income positions involve more sliders).
7. Additionally, there is a certain probability that one of the group members will not receive a payout even though the task was completed (this simulates the mortality risk of the insured in a pension plan).
8. After the main task, we collect additional data on their risk and social preferences, their beliefs and justice perceptions.
The four treatment variations are:
T1 Equal Risk with VOI [ER in Krieger et al. 2025]
In T1, participants do not know their income rank within the group before they set the redistribution parameter lambda. Furthermore, the probability of receiving no payout is the same for all group members (20%).
T2 Asymmetric Risk with VOI [SRL in Krieger et al. 2025]
In T2, participants do not know their income position within the group before they set the redistribution parameter lambda. Furthermore, while the probability of receiving no payout is also 20% on average, it is lower for group members with higher entitlements (A: 33, B: 27, C: 20, D: 13, E: 7)%, with A having the lowest entitlement and E the highest.
T3 Equal Risk w/o VOI
In T3, participants know their income rank within the group before they set the redistribution parameter lambda. Furthermore, the probability of receiving no payout is the same for all group members (20%).
T4 Asymmetric Risk w/o VOI
In T4, participants know their income position within the group before they set the redistribution parameter lambda. Furthermore, while the probability of receiving no payout is also 20% on average, it is lower for group members with higher entitlements (A: 33, B: 27, C: 20, D: 13, E: 7)%, with A having the lowest entitlement and E the highest.
An important note is that redistribution in T2 and T4 (with asymmetric risk) involves an efficiency loss which is due to the higer risk assigned to the lower income positions. Krieger et al. (2025) show that the alienation effect is due to the efficiency loss and that the alienation effect is correlated with certain justice beliefs and perceptions.
A technical note is that the experiment is conducted individually, and the actual group matching (Step 2) takes place only ex post, that is, after all data has been collected. Consequently, participants selected for payout will not receive their payouts until after the entire experiment has concluded.
In addition, participants are informed at the outset that, according to the random lottery incentive system, only about one in five participants will be selected for payout at the end. For these selected participants, the above payout mechanism is then applied, i.e., the group lambda, their own position, and the probability of default, and the conversion rate from Tokens to Euros determine the payout in Euros.
Design details
Cited by
No citation data available for this registration yet.