Are Elites Meritocratic and Efficiency-Seeking? Evidence from MBA Students

Our intervention consists of experimentally manipulating two key factors that may influence redistributive decisions: (1) the process by which income inequality is generated between two hired workers, and (2) the efficiency costs of redistribution.

For the first intervention, we randomly vary whether the earnings of workers recruited through an online labor market are determined by performance or luck. In the "performance" treatment, the winner of each worker pair is determined based on their relative productivity in the encryption task. In the "luck" treatment, the winner is determined by a random coin flip, completely disconnecting earnings from effort or merit.

For the second intervention, we introduce an efficiency cost to redistribution. In the "efficiency cost" treatment, redistribution reduces the total income available (for every $1 taken from the winner, the loser receives only $0.50), creating a trade-off between equality and efficiency.

2023-11-15

2024-10-30

Our key outcome variables of interest are:

1. The level of inequality implemented by MBA students across different experimental conditions, measured by the Gini coefficient of the final earnings distribution.
2. The elasticity of redistribution with respect to the source of inequality, measured by the difference in implemented inequality between the performance and luck conditions.
3. The elasticity of redistribution with respect to efficiency costs, measured by the difference in implemented inequality between the efficiency cost and luck conditions.
4. The distribution of fairness ideals among MBA students, classified as egalitarian, libertarian, meritocratic, or other, based on their redistribution patterns across conditions.

These outcomes allow us to compare MBA students' distributive preferences with those of the general population and examine how future elites weigh fairness considerations against efficiency costs when making redistributive decisions.

We use two complementary approaches to measure fairness ideals:

a) Between-subject approach: Using only the first redistributive decision of each spectator to avoid order effects, we classify spectators based on their single observed choice. Egalitarians equalize earnings regardless of source, libertarians maintain initial inequalities, and meritocrats redistribute based on the source of inequality.

b) Within-subject approach: Using multiple decisions from the same spectator across different conditions, we create individual-level classifications based on the pattern of choices. This approach provides more information about each spectator's behavior at the potential cost of sequential decision bias.

We conducted an incentivized laboratory experiment with MBA students to measure their fairness and efficiency preferences. The experiment followed the impartial spectator paradigm with three stages:

1. Production Stage: Workers recruited from an online labor platform completed an encryption task.

2. Earnings Stage: Workers were randomly paired, with winners receiving $6 and losers $0. We experimentally randomized how winners were determined through two mechanisms:
- "Performance" treatment: Winner determined based on relative task performance
- "Luck" treatment: Winner determined by random coin flip

3. Redistribution Stage: MBA students acting as impartial spectators chose how to redistribute earnings between worker pairs across three experimental conditions:
- Luck-based earnings with no redistribution cost
- Performance-based earnings with no redistribution cost
- Luck-based earnings with efficiency costs of redistribution (every $1 taken from the winner only increases the loser's earnings by $0.50)

Our design implements two levels of randomization: First, we randomly assign worker pairs to have their earnings determined by either performance or luck. This randomization allows us to causally identify how the source of inequality affects redistribution decisions. Second, each MBA student makes decisions for all three treatment conditions, with the presentation order randomized to mitigate sequence effects. This randomized controlled design enables us to causally estimate how both the source of inequality and efficiency costs impact the level of inequality implemented by future elites. The spectators could redistribute any amount from $0 to $6 in $0.50 increments, with their decisions determining workers' final payments.

We employed two separate randomization procedures in this study:

1. For randomizing worker earnings conditions (performance vs. luck), we used a Stata random number generator to assign worker pairs to treatment conditions.

2. For randomizing the order of treatment conditions presented to MBA student spectators, we utilized the Qualtrics built-in randomizer function, which automatically assigned each participant to one of the six possible presentation orders of our three conditions (luck, performance, efficiency cost).

Both randomization procedures were fully automated and conducted via computer algorithms without researcher intervention in the assignment process.

Individual MBA student participants

No

N/A

In the main sample, we use data from 271 individual spectators (MBA students) making 813 redistributive decisions (3 decisions per spectator).

Workers: We recruited 876 workers from an online labor platform who were randomly assigned to treatment conditions. Among these workers, 282 (32.19%) participated in the "performance" condition where winners were determined by task performance, 322 (36.76%) participated in the "luck" condition where winners were determined by random assignment, and 272 (31.05%) participated in the "efficiency cost" condition.

Spectators: Our sample consists of 271 MBA students from an elite university who each made decisions for all three treatment conditions.