On Sunday, the National Disaster Management Authority put the season's toll at 107 dead and 344 injured since 26 June. Khyber Pakhtunkhwa has lost 55, Punjab 36. Forty-seven of the dead were children. Four hundred and forty-seven houses were damaged, and 269 heads of livestock were gone. Almost all the coverage has gone to the water. Around 400 square kilometres of the Lahore and Gujranwala divisions are under water, and an official at the Punjab Irrigation Department told Dawn the expanse now resembles "an inland sea." The Ravi is at medium flood level. The Met Office expects a fresh wet spell from Wednesday across the upper catchments. All of that is real, and all of it deserves reporting.
But it is not what killed most of the 107. By NDMA's own accounting, house and roof collapses caused roughly 57% of this season's deaths. Flooding accounts for about 18%. Most people who have died this monsoon did not drown in a river. They were crushed at home, usually at night, usually asleep.
That number changes the shape of the problem. A river breaching its banks is a hydrology question, and hydrology questions get answered with barrages, spurs, embankments and long arguments with Delhi about water releases. A roof falling in is a procurement question. It gets answered with cement, a raised plinth, a shorter roof span, and a mason who has been trained for two days.
The map of the dead is a map of building materials
Take two provinces. KP has 5.86 million households and has buried 55 people this season. Punjab has 19.84 million households and has buried 36. Per household, the death rate in KP is more than five times Punjab's. Now look at the 2023 census. In Punjab, 80.8% of homes are pakka and only 7.3% kacha. In KP, 55.6% are pakka, 19.6% semi-pakka, and 24.8% kacha. Close to half of KP's housing stock is something other than a permanent structure.
I want to be careful, because correlation is not the whole story. KP gets cloudbursts on steep ground, and a wall of water carrying boulders will flatten a reinforced house as easily as a mud one. Buner proved that in August 2025: more than 200 dead in a district that took over 150 millimetres of rain in a single hour. Terrain matters. Rainfall intensity matters. Nobody forecasts a cloudburst.
But the mechanism recorded against most of this year's deaths is not a torrent. It is a roof. Nationally, the census counted 38.29 million households. Two-thirds are pakka. Around 19.9%, roughly 7.6 million households, are kacha. At the census average of 6.3 people per household, that is close to 48 million Pakistanis sleeping under something not rated to hold water. Another 4.8 million households are semi-pakka. In rural Sindh, the kacha share is 61%. In Baluchistan it is 63.6%, and fewer than one in five homes in the province is pakka at all. That is the exposure. It does not recede when the rivers do.
You want to be buying in March and April, when the market is soft, when you can hold a supplier to a contract and inspect what arrives; instead, the country buys in August, on the spot market, mid-flood, competing with every district administration and every relief organisation for the same trucks on the same washed-out roads.
Sindh built the only counter-argument we have
After 2022, the Sindh government set up Sindh Peoples Housing for Flood Affectees and did something unusual: it gave the money to the households instead of to contractors. Each family receives a grant of Rs 300,000, paid directly into their bank account in tranches tied to construction milestones. Money at plinth level, money at lintel level, money once the roof is on. The design is a 250 square foot core unit built to a multi-hazard standard, with a raised plinth, a rainwater collection point and a twin-pit latrine. Local masons are trained in the standard. The household holds the budget and hires the labour, which keeps the money circulating in the village.
The scale is genuine. Sindh assessed 2.1 million damaged homes affecting 12.6 million people, at a total requirement of Rs 550 billion. A review meeting of the Sindh Rural Support Organisation in February this year put over 1.25 million houses under construction and 500,000 completed. Jacobabad district alone has finished 81,356. Qambar Shahdadkot has crossed 70,000.
None of which is a press release for the PPP. The party has governed Sindh for eighteen years, and rural Sindh's 61% kacha share is its own record, not an inheritance. SPHF is a repair job on a hole the provincial government helped dig. Verification has been slow, tranche disbursement has stalled for thousands of families, and the programme's own published figures do not reconcile with each other, which should trouble anyone who has ever tried to audit a public sector company in this country.
The model is still right, and nobody has copied it. KP, which has recorded the highest death toll in each of the last two monsoons, still treats post-disaster housing as compensation: a cash payout after the fact, no standard attached, no requirement that the rebuild be better than the thing that collapsed. Rebuild-as-was is the default. So the same house comes down next August, on the same family.
Run the numbers before saying we cannot afford it.
Converting every kacha household in Pakistan to the Sindh standard, at Rs 300,000 each, comes to about Rs 2.3 trillion. At current rates, roughly $8 billion. It is a large number, and it would take a decade. Here is the other column. The 2022 floods caused $14.9 billion in damages and $15.2 billion in economic losses, with reconstruction needs assessed at $16.3 billion. Housing alone accounted for $5.6 billion of that damage. In 2025, 1,037 people died, and more than 229,700 houses were damaged or destroyed. Rebuilding just that single year's damaged stock to the Sindh standard would cost around Rs 69 billion, about $250 million.
So the full national retrofit costs roughly 1.5 times what one bad season did to housing alone in 2022, and about a quarter of what that season cost the economy overall. We are already paying for these houses. We pay for them after they fall, at emergency prices, with a funeral attached.
We run this procurement backwards
This is the part I think about because of the day job. Everything resilient housing needs is a commodity with a seasonal price curve: cement, galvanised sheeting, treated timber, bamboo, binding wire. You want to be buying in March and April, when the market is soft, when you can hold a supplier to a contract and inspect what arrives. Instead, the country buys in August, on the spot market, mid-flood, competing with every district administration and every relief organisation for the same trucks on the same washed-out roads. Every year we rediscover that tents cost more during a tent shortage.
Labour works the same way. Training masons in multi-hazard construction is cheap but slow, and it has to happen in the dry months, or it does not happen at all. The structural thing SPHF got right was moving the spending upstream, into materials and skills, before the season rather than during it.
What would actually help
Four things, none of which needs a new law or a donors' conference.
Publish the cause-of-death breakdown. NDMA already collects it. The daily situation report should carry a standing district-level line for collapse against drowning against electrocution against landslide. Right now that 57% figure surfaces once in a wire story and vanishes. It should be the headline number every morning.
Attach a standard to the money. Provincial compensation for a destroyed house should be conditional on the rebuild meeting a published minimum: plinth height, roof span, wall thickness, anchorage. Sindh has already written the document. KP, Balochistan and Punjab can adopt it this winter for the cost of printing.
Extend the owner-driven model. Rs 300,000 released in tranches to a household with a trained mason attached is faster and cheaper than any contractor-led scheme, and it survives a change of government better than a mega-project does.
And buy in April. Nothing stops a cloudburst. But of the 107 people this country has lost since 26 June, most were killed by something we know how to build better, at a price we have already shown we can pay. Forty-seven of them were children. The rivers will be back. So will the roofs.