The reliable objection to distributing decisions is that it works for a software startup and nowhere that carries real risk: public money, scale, a downturn. The objection is testable, and tested, it does not hold as a general rule. It holds as a conditional one, which is more useful.
The hardest proof comes from the unlikeliest place: a city budget
Sonia Goncalves (2014) studied Brazilian municipalities that adopted participatory budgeting between 1990 and 2004, letting ordinary residents decide part of how public money was spent. The result is not rhetorical. Municipalities that used participatory budgeting shifted a larger share of spending toward sanitation and health, matching allocation to local preference, and that shift came with a measurable reduction in infant mortality. Distributed decision-making, inside the state, with an effect counted in children's lives. It is the case Ricardo Semler long argued from intuition, now carrying an econometric panel behind it.
The mechanism is information, not virtue
Why does pushing decisions outward work? Because the person near the problem often holds information the center does not. Aghion, Bloom, Lucking, Sadun and Van Reenen (2021) found this in the least forgiving setting available, a recession. Firms with more decentralized decision-making performed better through the turbulence of the Great Recession, precisely when conditions changed too fast for information to travel up a hierarchy and orders to travel back down in time. Decentralization is not a fair-weather indulgence. It is an adaptation advantage when the ground moves.
The pattern is not confined to a crisis. Comparing firms across countries, Bloom, Sadun and Van Reenen (2009) found systematic links between decentralization and productivity, with the payoff to pushing authority down shaped by trust and context. The logic is older still. Studying strategic decisions in high-velocity environments, Eisenhardt (1989) showed that the teams which decided both faster and better were not the ones that centralized. They consumed more real-time information and distributed the analysis. Speed and quality of decision are not opponents. Centralization is what usually sacrifices both.
The antidote to romanticism: autonomy has preconditions
Here is the honesty that separates evidence from a motivational stage. Distributing decisions does not work everywhere or always. Hanushek, Link and Woessmann (2013) analyzed school autonomy across more than a million students in 42 countries over four waves of the PISA tests. The finding is uncomfortable for both sides of the argument: autonomy raised achievement in developed, high-performing systems and lowered it in developing, low-performing ones. Autonomy is neither good nor bad in the abstract. It depends on institutional capacity and accountability. Grant it without those preconditions and you have moved the problem, not solved it.
The individual side has preconditions too. In a meta-analysis of empowerment, Seibert, Wang and Courtright (2011) found psychological and team empowerment linked to performance and satisfaction, while mapping the conditions that make empowerment translate into results rather than confusion. Autonomy that lands on people without the context or capability to use it does not become ownership. It becomes anxiety.
Someone still has to own the call
Decentralizing is also not the same as dissolving responsibility. In their study of decision effectiveness, Rogers and Blenko (2006) found that organizations perform better when decision roles are explicit: who recommends, who gives input, who decides, who executes. The opposite of rigid hierarchy is not a free-for-all where no one is accountable. It is a clear map of who decides what, pushed to the level closest to the problem. Autonomy without defined roles produces the orphaned decision, and the orphaned decision is the one that never gets made.
The practical read
The right question is never centralize or decentralize in the abstract. It is whether a given decision sits at the level that holds the information and has a clear owner. The evidence is robust across wildly different settings: from a city's budget to a firm fighting a recession, moving the decision toward the person who lives the problem improves the result (Goncalves, 2014; Aghion et al., 2021). The same body of evidence warns that autonomy without installed capability and accountability can make things worse (Hanushek, Link and Woessmann, 2013). Mature self-management is exactly this pairing: decisions at the edge, with open information and named responsibility. It is not surrendering control. It is placing control where it works.
Sources
- Goncalves (2014). The Effects of Participatory Budgeting on Municipal Expenditures and Infant Mortality in Brazil. Municipalities using participatory budgeting redirected spending to sanitation and health and saw lower infant mortality. https://www.sciencedirect.com/science/article/abs/pii/S0305750X13000156
- Aghion, Bloom, Lucking, Sadun and Van Reenen (2021). Turbulence, Firm Decentralization, and Growth in Bad Times. Decentralized firms outperformed during the Great Recession, when conditions changed too fast for centralized control. https://www.aeaweb.org/articles?id=10.1257%2Fapp.20180752
- Bloom, Sadun and Van Reenen (2009). The Organization of Firms Across Countries. Links decentralization to productivity across countries, shaped by trust and context. https://www.nber.org/digest/dec09/international-differences-decentralization-and-productivity
- Eisenhardt (1989). Making Fast Strategic Decisions in High-Velocity Environments. Teams that decided faster and better used more real-time information and distributed analysis rather than centralizing. https://acawiki.org/Making_fast_strategic_decisions_in_high-velocity_environments
- Hanushek, Link and Woessmann (2013). Does School Autonomy Make Sense Everywhere? Autonomy raised achievement in high-capacity systems and lowered it in low-capacity ones; it depends on accountability. https://www.sciencedirect.com/science/article/abs/pii/S0304387812000661
- Seibert, Wang and Courtright (2011). Antecedents and Consequences of Psychological and Team Empowerment. A meta-analysis linking empowerment to performance and satisfaction under the right conditions. https://pubmed.ncbi.nlm.nih.gov/21443317/
- Rogers and Blenko (2006). Who Has the D? Performance improves when decision roles are explicit and pushed close to the problem. https://www.bain.com/insights/who-has-the-d/