Return-to-office debates grab headlines, but the deeper strain is quieter: only one in five employees worldwide report being engaged, the lowest share since 2020. Gallup estimates the shortfall costs about 9 percent of global GDP in lost productivity, roughly 10 trillion dollars. Manager engagement has also slid, dropping 9 percentage points since 2022, with the sharpest decline in the past year (Gallup, 2025). In that climate, leaders reach for pay transparency as a fairness fix that costs little cash. It is not a costless move. Transparency changes what people compare.
What a clean test shows
A rare natural field experiment inside a large Southeast Asian bank offers unusually strong guidance. Zoë Cullen and Ricardo Perez-Truglia randomly informed employees about actual salaries and tracked behavior using administrative data, incentivized surveys, and information-provision experiments (Cullen and Perez-Truglia, 2022). Two facts stand out.
First, people are badly wrong about their bosses’ pay. On average they underestimate the manager’s salary by 14.1 percent, with an average absolute error of 28 percent. By contrast, they are closer on peers’ pay, with an average absolute error of 11.5 percent and no systematic bias (Cullen and Perez-Truglia, 2022). Second, the direction of effort depends on the comparison. When employees learned that their boss earned more than they had thought, they worked more. A 10 percent increase in perceived manager salary raised hours worked over the next 90 days by 1.5 percent, a precisely estimated effect in this pre-registered randomized setting (Cullen and Perez-Truglia, 2022). The same study also reports that discovering peers earn more than expected had the opposite effect on motivation and related behaviors. Vertical comparisons tend to spur, horizontal comparisons can sour.
This asymmetry is not a quirk of survey mood. It is behavior measured in the firm’s own systems over time, inside a single company. The setting is a bank in one country, so replication in other contexts remains a fair question. As evidence about transparency and incentives goes, however, this is as clean as field data gets.
Why structure amplifies the reaction
How a firm is organized shapes what comparisons mean. In conventional hierarchies, status differences are built into the org chart. Self-managing organizations deliberately remove the boss–subordinate relationship, decentralizing authority in formal and systematic ways across the company. That structural choice is the field’s canonical definition of self-management (Lee and Edmondson, 2017). The same review notes that managerial hierarchies are strong at reliable execution in stable settings, but they tend to reinforce status gaps that can stifle development and cross-boundary problem solving (Lee and Edmondson, 2017). Pay numbers do not float in a vacuum. They signal where power sits and which ladders are worth climbing.
Viewed through that lens, the bank experiment’s pattern is intuitive. In a traditional hierarchy, learning that a boss earns more than expected can be read as evidence that moving up is valuable, so extra effort feels rational. Learning that a peer earns more can trigger a different story: similar roles, different rewards, lower perceived fairness. In flatter, self-managing systems where the formal boss–subordinate tie is absent, the vertical signal may be weaker. The horizontal comparison may dominate unless roles, contribution metrics, and decision rights are exceptionally clear. That is an interpretive point, not a causal claim. The field experiment was run in a conventional firm, and the review of self-management is a theory-driven synthesis rather than an experiment (Lee and Edmondson, 2017). The implication is practical all the same. Transparency works through social comparison, and structure sets the frame for how those comparisons land.
Safety is the buffer
Publishing pay without a forum to make sense of it is a recipe for rumor and retreat. Amy Edmondson’s foundational field study introduced the construct of team psychological safety and linked it to learning behaviors and performance via learning as a mediating mechanism (Edmondson, 1999). In 51 teams within a single manufacturing firm, higher psychological safety was associated with more speaking up, asking for help, and admitting errors. Team efficacy, once psychological safety was accounted for, did not explain the learning behaviors that drove performance (Edmondson, 1999). The study is cross-sectional and confined to one company, so it does not prove causation across settings. It does clarify the mechanism leaders need. Sensitive information yields improvement only when people believe the team is a safe place to take interpersonal risks.
Pay data is as sensitive as it gets. If you want the productive version of transparency, pair it with spaces where employees can question, challenge, and propose fixes without career penalty. That is where learning behaviors arise. Absent that, horizontal comparisons harden into grievance and turnover intentions, which is precisely the risk flagged in the bank experiment’s peer findings (Cullen and Perez-Truglia, 2022).
A design checklist that tilts to the right edge
The evidence does not tell you to hide pay. It tells you to choreograph what is revealed, to whom, and with what support. Five moves follow from the research.
- Lead with vertical clarity. Publish manager and level ranges, promotion criteria, and the capabilities required to progress. In the bank experiment, higher perceived boss pay raised effort, and the mechanism plausibly runs through expected returns to advancement (Cullen and Perez-Truglia, 2022).
- Stage horizontal exposure. Share peer ranges in bands tied to role and contribution, not names. Where individuals see unexpected gaps, offer a structured process to reconcile role scope, market data, and performance rubric. That channels the horizontal shock into a learning conversation rather than a verdict on fairness.
- Build psychological safety into the rollout. Convene team sessions that model inquiry, error acknowledgement, and help-seeking. The classic field study shows learning behaviors rise with safety, and those behaviors link to performance through learning, not bravado (Edmondson, 1999).
- Align with your structure. In self-managing settings, clarify decision rights for compensation and progression since the boss–subordinate frame is absent. The scholarly map of self-management is explicit about decentralizing authority formally and systematically. Treat pay governance the same way to avoid ad hoc status games (Lee and Edmondson, 2017).
- Watch the meters that matter. If engagement is already fragile and managers are themselves disengaging, you need tight feedback loops. Track turnover intentions, internal mobility, and request patterns before and after transparency moves. The global engagement slump and the fall in manager engagement are signals that any change will meet a stressed system (Gallup, 2025).
The practical read is straightforward. Pay transparency is not a virtue signal. It is an intervention in social comparison. If you want more effort, design transparency to emphasize vertical pathways and fair progression, then soften the horizontal blow with norms and forums for sensemaking. That approach respects how people actually react to pay information. At a moment when engagement is scarce and management attention is stretched, that realism is the difference between lighting a path and lighting a fuse.
Sources
- Cullen and Perez-Truglia (2022). How Much Does Your Boss Make? The Effects of Salary Comparisons, Journal of Political Economy 130. A natural field experiment in a Southeast Asian bank shows employees underestimate boss pay and that higher perceived boss pay raises subsequent hours worked, while peer comparisons can demotivate. https://doi.org/10.1086/717891
- Lee and Edmondson (2017). Self-managing organizations: Exploring the limits of less-hierarchical organizing, Research in Organizational Behavior 37. A theory-driven review defines self-managing organizations and explains how hierarchies reinforce status differences that shape how pay signals are received. https://www.sciencedirect.com/science/article/abs/pii/S0191308517300059
- Edmondson (1999). Psychological Safety and Learning Behavior in Work Teams, Administrative Science Quarterly 44. A field study links team psychological safety to learning behaviors that mediate performance, underscoring the need for safe forums when sharing sensitive pay data. https://journals.sagepub.com/doi/10.2307/2666999
- Gallup (2025). State of the Global Workplace 2025. Global engagement fell to 20 percent, with an estimated cost of about 10 trillion dollars in lost productivity and a 9 percentage-point drop in manager engagement since 2022, framing the urgency and risk of any pay-transparency move. https://www.gallup.com/workplace/659610/state-global-workplace.aspx