Meritocracy in firms

We administer a large-scale survey with employees in Norway, working in firms with at least 5 employees. We ask about their perceptions of wage inequality at their firm (how much do the top/bottom 20% at their firm make), whether current inequalities are fair, what they believe are the key sources of unequal pay, and whether they would like to change existing inequalities. We use vignettes to classify respondents’ fairness views. Finally, we measure employees’ job satisfaction and turnover intentions.

2025-11-07

2026-07-01

Is inequality at the firm perceived as fair, is it perceived as efficient, fairness views, should the current wage inequality be changed (and how), job satisfaction, turnover (intention)

- Is inequality at the firm perceived as fair: We ask participants whether the current wage differences at their firm are fair on a 5-point likert scale from completely disagree to completely agree.
- Is inequality at the firm perceived as efficient: We ask participants whether the current wage differences at their firm are optimal for the firm’s economic outcomes on a 5-point likert scale from completely disagree to completely agree.
- Fairness views: we use two scenarios that describe inequality between two co-workers as either a function of merit or outside options. We use these scenarios to define four different fairness views (egalitarians who think that both sources of inequality are unfair, meritocrats who think that inequality is fair if it is a function of merit but not otherwise, marketarians who think that outside options are the only fair source of inequality, and libertarians who find both sources of inequality fair)
- Should the current inequality be changed: We give participants three options, my firm should reduce wage differences, should not change, or should increase wage differences. In addition, we ask whether inequality should be changed to be more reflective of individual contributions and whether it should be changed to be more reflective of an individual’s outside options.
- Job satisfaction: We ask whether overall people are satisfied with their job on a 5-point likert scale from completely disagree to completely agree.
- Turnover (intention): We ask how likely it is that they will change their employer within the next year on a 4-point likert scale from very unlikely to very likely. We will observe actual turnover in the admin data.

We measure perceptions about the extent of inequality at the firm, beliefs about the sources and consequences of inequality at their firm, a wider range of fairness judgements (own wage, CEO’s wage, capital-labor ratio), views on horizontal versus vertical inequality, transparency of wage setting, union membership, attitudes towards inequality and policies in Norway (beyond their firm), and career aspirations.

We administer the same survey to all employees. We adjust the wording of some questions to make them relevant to private vs public sector workers. We elicit employees’ fairness views through two scenarios. We randomize the order in which participants assess the fairness of the different scenarios.

Not available

The order in which scenarios are presented is randomized by a draw in the beginning of the experiment. Additionally, we randomize information about horizontal inequality (participants are aware that this is hypothetical).

Randomization is done at the individual level. For the vignettes there are two possible orders. For the information about horizontal inequality there are 6 treatment arms (no information, imagine to earn 30% more/less than the median, earn 10% more/less than the median, earn the same as the median).

No

Does not apply.

We use Norstat to contact Norwegian employees. Our sample restriction is that respondents need to be employed, over 18, work at firms with at least 5 people and cannot be employed via an employment agency. The goal is to reach the whole universe of employees signed up in the Norstat sample. Assuming average response rates, we expect to recruit around N=18,600 people.

The order of vignettes is randomized such that half of the participants see scenario 1 first, and the other half sees scenario 2 first. The assignment of information is also randomized to achieve an equal sample size in each cell.

Given the large sample size of the data collection, we will be able to conduct different heterogeneity analyses (see PAP).