At two-thirty on a hot afternoon, a city can look as if it is working normally. Buses move. Deliveries arrive. Construction continues. Street vendors keep selling. The emergency is visible only if you know where to look: the mason whose hands have begun to shake, the sanitation worker who has stopped sweating, the woman in a tin-roofed home deciding whether to spend scarce money on water or electricity. Now imagine that the city has done many sensible things. The weather department has issued an alert. The health department has circulated advice. An NGO has distributed water. A company has changed working hours at one site. A donor has funded a cooling centre.
Every project may be successful. The problem can still remain unsolved. This happens because institutions are usually designed to own activities. Someone owns the forecast. Someone owns the clinic. Someone owns labour compliance. Someone owns the grant. Almost no one owns the chain of events that determines whether a worker is protected. The chain matters. A forecast must become a decision. The decision must become permission to act. The action must be affordable. The worker must trust it. The employer must not punish it. The health system must recognise heat illness. Each link can work while the chain fails. We often respond by asking for a stronger lead institution. Sometimes that is right. But climate risk rarely respects a single institution’s authority. A municipal commissioner cannot redesign every workplace. An employer cannot cool an informal settlement. A health department cannot replace lost wages. A community organisation cannot change building codes.
So the real choice is between two imperfect approaches. The first is to build another excellent project: clear scope, measurable outputs, one accountable organisation. This is efficient. It can move quickly. It is also legible to funders. The second is to organise around the whole outcome. That means several actors accepting distinct responsibilities, agreeing on triggers, sharing information, and adapting when one part fails. This approach is slower at the beginning. It creates coordination costs. It can also solve what no project can solve alone. My judgment is that complex resilience problems need distributed action with a clearly named steward of the outcome. The steward is not a super-agency. It does not control everyone. Its job is more practical: keep the whole problem visible, notice broken links, convene the actors who can repair them, and make failure difficult to hide. Consider an illustrative occupational heat programme in a South Asian city. The weather service could define risk thresholds. Employers and worker groups could agree on rest, shade, hydration and work-hour protocols. The health system could track symptoms and treatment. Social protection agencies could test temporary income support when work must stop. Researchers could compare which measures actually reduce illness and income loss. A city climate cell could hold the response together. No single actor does everything. One actor remains responsible for asking whether workers are safer. This distinction changes how money should be used.
A conventional grant might pay for ten cooling centres. A field-oriented grant might still fund cooling centres, but also finance the protocols, data agreements, worker feedback, employer adoption and public accountability that determine whether the centres are used by the people at greatest risk. It also changes what counts as success. Counting alerts or water stations is easy. The harder questions are whether workers changed behaviour, whether employers allowed them to, whether heat-related illness fell, and whether protection reached people outside formal employment. The same pattern appears elsewhere. In urban flooding, one agency manages drains, another roads, another land use, while households live with the combined result. In agriculture, one programme supports seeds, another credit, another market access, while the farmer experiences one livelihood. In early warning, technology can detect danger even when institutions cannot convert warning into action.
The phrase “own the problem” can be misread as a call for more control. I mean the opposite. Good problem ownership starts by recognising interdependence. It asks who is best placed to do each part and what must connect those parts. There is a danger here. “Ecosystem building” can become a dignified name for meetings. Coordination can consume money without changing decisions. A steward can claim broad responsibility while avoiding measurable accountability. The remedy is to make the chain explicit. For every actor, specify the decision they control, the action expected, the trigger for action and the evidence that it happened. Convening has value when it changes one of those things. I would change my mind about this approach under two conditions. First, if the problem has a stable technical solution and one institution has the authority and capability to deliver it, distributed governance may add needless friction. Replacing a broken pump does not require an ecosystem. Second, if comparative evidence across several cities showed that multi-actor stewardship consumed more time and money without improving health, income or coverage, I would favour narrower delivery models.
But for problems like heat, flooding and livelihood insecurity, fragmentation is often part of the hazard. The absence of ownership is not an administrative detail. It is one of the reasons risk persists. A useful first question for any new initiative is therefore not “What will our project deliver?” One practical tool is an ownership map built around decisions rather than logos. List the moments when risk must become action: issuing an alert, changing a shift, opening a clinic protocol, releasing emergency cash, checking employer compliance. Beside each decision, name the actor with authority, the people affected, the information required and the consequence of delay. Empty cells reveal the real design problem. This map also disciplines the steward. It prevents a coordinating organisation from absorbing every task. Its success is visible when decisions move to legitimate owners and connections keep working without constant intervention. Over time, the steward may become smaller because the system has become stronger. That is a better outcome than building a permanent centre that everyone must depend on. It is: “If every funded activity succeeds, who will know whether the problem was actually solved?”
Leave a comment