Return-to-office mandates, tighter budget gates, and AI bets that need coordination have many executives pulling decisions back to headquarters. At the same time, manager engagement is sinking and frontline talent is harder to keep. That creates a live contradiction: firms need speed and judgment at the edge, yet their structures often assume the best answer sits in the center.
What decentralization actually buys you in bad times
A large comparative study of firms hit by the Great Recession shows what happens when turbulence rises and local knowledge matters more. Using data from manufacturing plants across ten OECD countries and from US establishments, researchers found that companies that had delegated more decision-making power to local plant managers before the crisis outperformed more centralized peers in the sectors most severely affected (Aghion, Bloom, Lucking, Sadun and Van Reenen, 2021). The authors tie this advantage to the greater value of local information under conditions of product churn and market volatility. When the ground shifts quickly, the people closest to customers and operations see the signal first, and they act before a central hierarchy can process it.
This is not a romantic case for anarchy. The study is careful on mechanism and alternatives. The performance edge for decentralized firms is consistent with the local-information view and remains robust when accounting for explanations such as greater fear of bankruptcy or changes in coordination needs (Aghion et al., 2021). The design move that mattered was simple and concrete: pushing decision rights from headquarters to local managers before the storm hit, so the muscle memory for acting on local data already existed when it counted.
The contemporary rhyme is obvious. Supply shocks, policy swings, and digital shifts have made 2020s markets choppy. The historical evidence suggests that in these conditions, central clearance slows you down. Delegation is not a perk. It is a risk response.
Flat is not free, and it is not the point
There is a temptation to equate distributed decision-making with abolishing management altogether. That is a category error. Self-managing organizations are a specific form that decentralizes authority radically and eliminates the manager–subordinate relationship through formal, organization-wide design (Lee and Edmondson, 2017). The literature synthesized by Lee and Edmondson is explicit on the trade: classic managerial hierarchy is well suited to reliable execution of known tasks in stable settings, while it tends to inhibit boundary-spanning problem solving, entrench rigidity, and reinforce status differences that suffocate development in less predictable work (Lee and Edmondson, 2017).
The implication for most firms is practical, not ideological. You do not need to erase managers to get the benefits of local judgment. You do need to be clear about what kind of work you are running. If the problem is repetitive and the environment steady, centralized routines and tight controls can shine. If the work crosses functions and the environment is turbulent, distributing decision rights where the information lives is more promising. The evidence for fully self-managing models is still limited, and the form is not a cure-all (Lee and Edmondson, 2017). What travels better is the logic: move authority toward the problem when uncertainty rises.
Safety is the operating system for local judgment
Even with crisp decision rights, local initiative stalls if people feel unsafe to voice problems or take smart risks. Psychological safety is the team-level belief that it is safe to take interpersonal risks such as asking for help or admitting an error (Edmondson, 1999). In a field study of 51 teams in a manufacturing company, psychological safety was associated with more learning behavior, and that learning behavior mediated the relationship with performance (Edmondson, 1999). The sample was one firm and cross-sectional, so treat it as foundational insight rather than causal proof across all settings. Still, the mechanism matters for distributed decision-making. If teams cannot surface anomalies, run small tests, and share what they learn without penalty, headquarters will keep being asked to referee decisions that should be made locally.
Leaders who want to push decisions out must therefore invest in the social infrastructure that lets local knowledge flow. That starts with routines that signal it is acceptable to flag uncertainty, with rapid review cycles that turn small bets into shared learning, and with visible follow-through when people raise issues. Authority without safety breeds silence or blame-shifting. Neither speeds adaptation.
The management crisis is a design signal
There is also a human cost to running decisions through the middle. Globally, only 20 percent of employees are engaged in 2025, and Gallup estimates the productivity loss at about 10 trillion dollars, roughly 9 percent of world GDP (Gallup, 2025). Manager engagement has dropped sharply, from 27 to 22 percent between 2024 and 2025, which makes the familiar model of the overburdened “traffic controller” even less sustainable (Gallup, 2025). The gap between typical organizations and best-practice ones is striking. In the latter, 79 percent of managers are engaged, almost four times the global average, which points to design choices rather than fate (Gallup, 2025).
Distributed decision-making is one of those choices. It changes a manager’s job from policing approvals to providing context, building capability, and removing blockers. That shift does not guarantee higher engagement on its own. It does make the role more coherent with what complex work needs and gives frontline teams the permission and clarity to act. The Gallup figures do not prove causality here, but they underscore the cost of clinging to designs that exhaust managers while slowing down responses at the edge.
A practical playbook for 2025
Start by mapping decisions by speed, reversibility, and proximity to customers. Push high-velocity and easily reversible calls to the edge. Keep one-way-door bets and cross-unit trade-offs closer to the center. Put decision rights in writing so no one waits for a ghost approval. Pair that with simple escalation triggers, not blanket sign-offs.
Create a cadence for learning. Require short after-action notes for local experiments so insights travel across sites. Make it easy for teams to ask for help without fear, and respond quickly when they do, because that is how psychological safety becomes more than a poster. The link between safety, learning behavior, and performance is a reason to treat these routines as core infrastructure, not extras (Edmondson, 1999).
Finally, pre-commit to a crisis posture. Aghion and colleagues show that firms that had already delegated before the downturn were the ones that performed better in the most affected sectors (Aghion et al., 2021). You cannot improvise trust. Decide now which approvals you will suspend, what thresholds will move to local leaders, and how you will monitor outcomes when volatility spikes. That preparation is cheap insurance.
The headline is simple. In calm seas, centralization can be efficient. In storms, it becomes a liability. Design for both states. Push decisions out when uncertainty rises, and build the safety and routines that let local information do its work.
Sources
- Aghion, Bloom, Lucking, Sadun and Van Reenen (2021). Turbulence, Firm Decentralization, and Growth in Bad Times (AEJ: Applied Economics 13). Shows that firms that delegated decision power to local managers before the Great Recession outperformed centralized peers in the hardest-hit sectors, consistent with the value of local information under turbulence. https://www.aeaweb.org/articles?id=10.1257%2Fapp.20180752
- Lee and Edmondson (2017). Self-managing organizations: Exploring the limits of less-hierarchical organizing. Defines self-managing organizations and synthesizes when hierarchy helps or hinders, grounding the article’s guidance on when to distribute authority. https://www.sciencedirect.com/science/article/abs/pii/S0191308517300059
- Edmondson (1999). Psychological Safety and Learning Behavior in Work Teams. Introduces team psychological safety and finds it is associated with learning behavior that mediates performance, supporting the need for social infrastructure when pushing decisions out. https://journals.sagepub.com/doi/10.2307/2666999
- Gallup (2025). State of the Global Workplace 2025. Reports global engagement at 20 percent, a $10 trillion productivity cost, and a sharp drop in manager engagement, framing the urgency of redesigning decision roles. https://www.gallup.com/workplace/659610/state-global-workplace.aspx