Current Inequality, Future Prospects, and Trust: Evidence from a Laboratory Experiment

How to cite

Stringhi, Alessandro; Stimolo, Marco. 2025. "Current Inequality, Future Prospects, and Trust: Evidence from a Laboratory Experiment." AEA RCT Registry. AEARCTR-0017352. https://doi.org/10.1257/rct.17352-2.1

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

Alessandro Stringhi* , Marco Stimolo

* Primary investigator

Registry ID
AEARCTR-0017352
DOI
https://doi.org/10.1257/rct.17352-2.1
First registered
Last updated
Status
Completed
Trial dates
2025-12-01 – 2026-04-01
Countries
Czechia (), Italy ()
JEL codes
C92; D31; D63.
Registry record
http://www.socialscienceregistry.org/trials/17352

Abstract

This paper compares two stylized environments in a pre-registered experiment based on a Trust Mini Game played over two independent periods. One environment features a smaller current endowment gap, implemented in the Low Inequality treatment. The other features better ex ante prospects under unchanged current inequality, implemented in the Rank Reversal treatment, where a Poor receiver may become Rich in period 2, and vice versa. The two treatments are calibrated to imply the same expected dispersion of endowments across the two periods of the game. We therefore focus on whether first-period behavior responds more strongly to a smaller current gap or to better prospects ex ante, before any rank reversal is realized.

Intervention

Between-subjects design based on a trust mini-game with dichotomous actions and heterogeneous endowments. Senders hold a median endowment (“Middle Class”), while Receivers hold either a lower (“Poor”) or higher (“Rich”) endowment relative to Senders. The trust mini-game is repeated in two blocks of 10 periods each. The design employs three treatments, in which the dispersion of the endowment distribution is manipulated (see Instructions for further details).

High Inequality (T0). In the first treatment, both blocks maintain high levels of endowment inequality. This setting represents a control condition with a symmetric but highly dispersed endowment distribution. The difference between “Rich” and “Poor” Receivers and the “Middle-class” Sender is equal to ±4 ECUs.

Low Inequality (T1). In the second treatment, endowment inequality in both blocks is halved compared to T0. The endowment distribution becomes less dispersed, with the difference between “Rich” and “Poor” Receivers and the “Middle-class” Sender reduced to ±2 ECUs.

Rank Reversal (T2). In the third treatment, the level of inequality within each block matches T0, but at the beginning of the second block there is a 50% chance of rank reversal for Receivers: half of the “Poor” become “Rich”, and half of the “Rich” become “Poor”. The expected individual endowment across blocks matches that of T1, resulting in a lower average dispersion of the endowment distribution compared to T0.

Experimental design

The experiment is conducted in the lab using oTreeg. Participants are assigned to one of three between-subjects treatments (T0: High Inequality, T1: Low Inequality, T2: Rank Reversal) and randomly to one of two roles, which remain fixed throughout: role A (“Middle-class” Sender) and role B (Receiver).

In each treatment, role-A participants receive a fixed endowment of 6 tokens in every period. Role-B participants are evenly split into a Low and a High endowment bracket. In T0 and T2, these brackets are (2, 10) tokens; in T1 they are (4, 8) tokens. The experiment consists of two blocks of 10 periods each (20 periods total). In every period, each A is randomly and anonymously matched with a B, under a typed-stranger protocol such that the same A–B pair does not meet more than once within a block. A is informed whether the matched B is Low or High; each participant knows their own endowment and the payoff structure.

In each period, A chooses between Stop and Continue, and B simultaneously chooses between Take and Share. If A chooses Stop, both players earn their initial endowment for that period. If A chooses Continue and B chooses Share, A’s payoff is twice A’s initial endowment (12 tokens) and B’s payoff is twice B’s initial endowment (treatment- and bracket-specific). If A chooses Continue and B chooses Take, A earns 0 tokens, while B earns twice their initial endowment plus 2 tokens. Apart from the endowment levels, the game structure is identical across treatments. The only additional difference is that in T2, between block 1 and block 2, 50% of Low-B participants are randomly reassigned to the High bracket and 50% of High-B participants to the Low bracket; in T0 and T1, brackets are fixed across blocks.

In each period, while making the action decision, A also reports a first-order belief (0–5) about how many B participants in the same bracket as the B they are currently paired with (i.e., “Poor” or “Rich”) will choose Share in that period. B reports a second-order belief (0–5) guessing the estimate made by the matched A in that period. At the end of the experiment, one period from each block is randomly selected for payment of choice-based earnings, and beliefs are additionally incentivized with a fixed payment per correct report.

Design details

Randomization method
Block randomization: In each experimental session, participants will be assigned to the same treatment. The number of sessions of each of the three treatment groups will be the same to ensure the same sample size. Sessions order will be randomized.
Randomization unit
Experimental session.
Clusters
No less than 12 laboratory sessions.
Observations
240 individual participants, organized in 12 laboratory sessions with exactly 20 subjects per session.
Arms
treatment arms with individual randomization: – T0 High Inequality: 80 subjects (4 sessions) – T1 Low Inequality: 80 subjects (4 sessions) – T2 Rank Reversal: 80 subjects (4 sessions) Total planned sample: 240 subjects.
Primary outcomes
Senders’ Trust (Block 1, by Receiver endowment): For each Sender, the proportion of decisions in the first block (10 decisions) in which the action “Continue” is chosen, computed separately for rounds in which the matched Receiver is “Poor” and “Rich”. Receivers’ Trustworthiness (Block 1, by own endowment): For each Receiver, the proportion of decisions in the first block (10 decisions) in which the action “Share” is chosen, computed separately for “Poor” and “Rich” Receivers. First-order Beliefs (Block 1): In each decision of block 1, Senders state an integer belief on a 0–5 scale about how many Receivers in the same bracket (“Poor” or “Rich”) will choose “Share” in that round. Primary belief outcomes are the distribution (e.g., mean) of these first-order beliefs in block 1, by Receiver bracket and treatment. Second-order Beliefs (Block 1): In each decision of block 1, Receivers state an integer belief on a 0–5 scale guessing the matched Sender’s estimate. Primary belief outcomes are the distribution (e.g., mean) of these second-order beliefs in block 1, by Receiver bracket and treatment.
Secondary outcomes
We measure the same outcomes as for block 1, but for the second block: – Senders’ trust: proportion of “Continue” choices. – Receivers’ trustworthiness: proportion of “Share” choices. – First-order and second-order beliefs (0–5 scale), defined as in block 1. As in the primary outcomes, these measures are computed by Receiver endowment (“Poor” vs “Rich”). In addition, in the mobility treatment (T2) we further condition on the four possible endowment paths across blocks (Poor→Poor, Poor→Rich, Rich→Rich, Rich→Poor).

Materials

Cited by

No citation data available for this registration yet.

Keywords