The corporate pendulum has swung back toward control. Return-to-office mandates are often paired with tighter approvals and more status checks, even as market shocks and AI upend operating assumptions. Meanwhile, the people tasked with policing these gates look depleted. Global employee engagement fell to 20 percent in 2025, and managers’ engagement dropped to 22 percent, down five points in a single year, according to Gallup’s worldwide survey (Gallup, 2025). If the front line is waiting for permission from a disengaged middle, the firm is paying an approval tax it cannot afford.
The alternative is not anarchy. It is pre-committing who decides what, close to where the work is done, with clear guardrails. The question is whether there is hard evidence that pushing decision rights out helps when conditions get rough.
What the data say about downturns and decentralization
Aghion, Bloom, Lucking, Sadun and Van Reenen studied firms across ten OECD countries using two large micro datasets and a quasi-experimental design that tracked outcomes before and during the Great Recession (Aghion et al., 2021). Their central finding is direct. Firms that had delegated more decision-making from headquarters to local plant managers before the crisis outperformed more centralized peers in the sectors hit hardest. The pattern held across alternative specifications. The authors argue that the mechanism is straightforward: under turbulence, local information becomes more valuable, so local decision rights raise performance. They also show the result is robust to competing explanations such as fear of bankruptcy or simple shifts in coordination needs (Aghion et al., 2021).
This is not a small footnote to strategy. It is a structural claim about how organizations metabolize shocks. When demand whipsaws and inputs change weekly, the time a choice spends in an approval queue is not neutral. It is lost learning. The Aghion et al. study is unusually persuasive for management research: it combines broad coverage with a difference-in-differences design, and the pattern appears where theory would predict, namely in more turbulent sectors. The implication for the 2020s, marked by supply chain volatility and rapid technological change, is clear in principle even if the study predates today’s specific shocks. Push more authority to where information lives.
Safety is the price of admission
Delegating decisions is easy to announce and hard to sustain. The behavioral precondition is people speaking up about risks, errors, and half-formed ideas without fear. Edmondson’s foundational field study of 51 work teams in a manufacturing firm coined the construct of team psychological safety as a shared belief that the team is safe for interpersonal risk-taking (Edmondson, 1999). In that cross-sectional analysis, psychological safety was associated with learning behaviors such as asking for help and experimenting, and those learning behaviors mediated the link to performance. Team efficacy was not a substitute once safety was accounted for. The message is practical. If you want faster, better local decisions, you need teams that actually surface what they see.
Treat that as a condition, not a platitude. A decentralized structure without safety often yields silence followed by escalation at the first sign of trouble, which recreates the very bottleneck you thought you removed. A centralized structure with safety is still slow, because the queue remains. You need both: decision rights at the edge, and an environment where edge teams learn in public.
Flattening is not a vibe, it is a design choice
Too many firms confuse fewer meetings with fewer layers. Lee and Edmondson offer the clearest scholarly definition of self-managing organizations: entities that decentralize authority in a radical, formal, and systematic way across the organization, eliminating the manager–subordinate relationship as a defining feature (Lee and Edmondson, 2017). Their review emphasizes boundaries. Traditional hierarchy performs reliably for known tasks in stable contexts, but it tends to inhibit cross-functional problem solving and can entrench status dynamics that stunt development. They also note that generalizable quantitative evidence on the full performance effects of radical self-management remains limited. That caveat matters. “Flat” is not a universal solvent. It is a design with strengths and trade-offs.
For most large firms, the actionable read is not to abolish managers overnight. It is to specify decision rights with the same discipline applied to financial controls. Write them down. Decide which calls are local by default, which are centralized by design, and which flip based on triggers. Clarify what information must be shared, on what cadence, and what constitutes a reversible versus one-way bet. Radical models show the direction of travel. Your job is to choose where to stop.
What to change on Monday
The evidence supports decentralization when conditions are choppy and information is local. To turn that into practice without theater, focus on four moves:
- Pre-commit guardrails. Publish decision charters that define scope, thresholds, and tripwires for key decisions in product, pricing, service recovery, procurement, and hiring. Make the default local up to a clear risk limit, then require escalation above it. This reduces decision latency without sacrificing control.
- Measure decision latency. Track time from problem identification to committed action for repeatable decision types. The goal is to make approval drag visible, not to shame individuals. Visibility is how you learn where centralization still makes sense and where it does not.
- Redefine the manager role. In the Gallup data, only 22 percent of managers reported being engaged in 2025, though best-practice firms reported 79 percent engaged managers, a descriptive gap that points to design choices rather than fixed constraints (Gallup, 2025). Move managers from approvers to coaches and integrators. Their job is to set guardrails, remove cross-team blockers, and steward shared standards, not to second-guess every call.
- Invest in safety rituals. Borrow from Edmondson’s work by institutionalizing debriefs, lightweight pre-mortems, and public logs of experiments. The point is not performative vulnerability. It is to normalize the learning behaviors that Edmondson found were associated with higher performance through better learning (Edmondson, 1999).
None of this negates the need for coordination or for some calls to stay central. As Lee and Edmondson caution, hierarchy does dependable work in stable domains, and sweeping claims about the superiority of self-management outrun today’s empirical base (Lee and Edmondson, 2017). Use a portfolio. Centralize irreversible, firm-wide bets. Delegate choices that are frequent, bounded by standards, and closest to fast-changing information. Revisit the split quarterly.
The practical endgame is not ideological. It is economic. In bad times, firms that had already moved authority out performed better where turbulence was highest, consistent with a simple information story (Aghion et al., 2021). At the same time, the engagement trends Gallup reports are a blunt reminder that the status quo feels broken for many employees and for many managers too (Gallup, 2025). Cut the approval tax with explicit decision rights and the safety to use them. You will make faster decisions, learn more per cycle, and give your managers a job worth doing.
Sources
- Aghion, Bloom, Lucking, Sadun and Van Reenen (2021). Turbulence, Firm Decentralization, and Growth in Bad Times. Firms that had delegated more decision-making before the Great Recession outperformed centralized peers in the hardest-hit sectors, consistent with the value of local information in turbulence. https://www.aeaweb.org/articles?id=10.1257%2Fapp.20180752
- Edmondson (1999). Psychological Safety and Learning Behavior in Work Teams. In a field study of 51 teams, team psychological safety was associated with learning behaviors that mediated the link to performance, highlighting a behavioral precondition for distributed decisions. https://journals.sagepub.com/doi/10.2307/2666999
- Lee and Edmondson (2017). Self-managing organizations: Exploring the limits of less-hierarchical organizing. Defines SMOs as radical, formal decentralization that removes manager–subordinate ties, and reviews where hierarchy helps and where it hinders, framing design choices and trade-offs. https://www.sciencedirect.com/science/article/abs/pii/S0191308517300059
- Gallup (2025). State of the Global Workplace 2025. Reports global engagement at 20 percent and managers’ engagement at 22 percent in 2025, with best-practice organizations at 79 percent for managers, underscoring the design gap rather than inevitability. https://www.gallup.com/workplace/659610/state-global-workplace.aspx