Not a Dhaka Problem: The Country Breathes the Worst Air of the Six

Executive Summary. The Bangladeshi population breathes a mean annual PM2.5 concentration of 69.25 micrograms per cubic metre (2023), the highest of the six Asian economies compared here and 2.8 times the peer median of 24.78. The gap to that median, 44.47 micrograms, is on its own larger than the total exposure of Indonesia, Thailand or Vietnam. The usual explanation does not fit. Bangladesh is the least urban of the six, 32.68% urban in 2024 against a peer median of 39.17%, and the two most urban countries in the group record the two lowest exposures. What Bangladesh has instead is density: 1,319 people per square kilometre, 4.12 times the peer median of 320.2, with no comparator close. The urban build-out is also still ahead, urban population growing 2.92% in 2025, the fastest of the six. Premature mortality from non-communicable disease, 18.1% in 2021, is second lowest of the six. That is a matter of timing, not an all-clear.

The dirtiest air of the six

In 2023 the mean annual PM2.5 exposure of the Bangladeshi population was 69.25 micrograms per cubic metre. The rest of the group: Pakistan 58.37, India 54.05, Vietnam 24.78, Thailand 23.49, Indonesia 20.84. Bangladesh ranks first of six on an indicator where first is worst, and the group splits cleanly in two, the three South Asian countries above 54 micrograms, the three Southeast Asian ones below 25.

The peer median is 24.78 micrograms. Bangladesh's reading is 2.8 times that, and 44.47 micrograms above it.

One caveat belongs here and not in a footnote. This indicator is a mean annual exposure for a whole population: an average across the country, not a reading from any particular street. The fact pack behind this essay carries no city-level figure, so nothing that follows is a statement about Dhaka in particular. That is the first reason the framing has to change. The number that ranks Bangladesh worst of six is a national number, and a national number does not yield to a capital-city policy.

Source: World Bank World Development Indicators, PM2.5 air pollution, mean annual exposure (EN.ATM.PM25.MC.M3), 2023.

The country is not urban enough for the standard story

The standard account of air pollution in a developing economy runs through cities. People concentrate, vehicles and industry concentrate with them, exposure climbs with the urban share, and incomes and enforcement eventually pull it back down. On that account Bangladesh should sit in the middle of this group. It sits at the top, and it is not urban.

The urban share was 32.68% in 2024, the lowest of the six, below the peer median of 39.17% and below every comparator: Thailand 61.87%, Indonesia 58.75%, Pakistan 39.17%, Vietnam 38.49%, India 35.38%. Run the two rankings side by side and they point opposite ways. The two most urban countries in the group, Thailand and Indonesia, record the two lowest exposures, 23.49 and 20.84 micrograms.

Six countries in one year is not a test of anything. Income, fuel mix, vehicle stock, enforcement capacity and physical geography all move together with urbanisation, and a cross-section this small cannot separate them; the association could reverse with a different set of comparators. What it does do is close off the easy story. Bangladesh cannot account for its exposure by pointing at its cities, because it has proportionally fewer city dwellers than any country it is being compared with.

What Bangladesh has instead is density

Population density was 1,319 people per square kilometre in 2023, first of six and not narrowly. India follows at 483.7, then Pakistan 321.1, Vietnam 320.2, Indonesia 148.6, Thailand 140.3. Against a peer median of 320.2, Bangladesh is 4.12 times as crowded, a gap of 999 people per square kilometre.

That is a country average, and it includes every hectare the statistics count as rural. 173.6 million people (2024) live on it. The consequence for exposure is arithmetic rather than ideological: at this density there is no dilution and no distance. Whatever burns in Bangladesh burns nearer to more people than the same source would anywhere else in this group. The pack identifies no sources, so this essay names none, but the geometry holds regardless of what the sources turn out to be.

Land scarcity is the same fact seen from the other side. Forest covers 14.49% of the land (2023), second lowest of the six, with only Pakistan lower at 4.67%. The peer median is 38.69%, so Bangladesh holds about 0.37 times the median share, in a group where Indonesia is at 47.72% and Vietnam 47.45%. The Bangladeshi share has barely moved either, 14.75% in 2000 to 14.49% in 2023, a fall of 0.26 percentage points, or 1.78% of the 2000 level. This is not a claim that trees scrub particulates. It is a claim about room. The most crowded country of the six also holds one of the smallest reserves of unbuilt land, and unbuilt land is precisely the resource a state spends when it wants to put combustion somewhere other than beside housing.

Source: World Bank WDI, population density, people per square kilometre of land area (EN.POP.DNST), 2023.

The build-out is still ahead

If Bangladesh's exposure were mainly a product of its urbanisation, the position would be worse than it looks, because the urbanisation is mostly in the future.

The urban share rose from 7.72% in 1971 to 32.68% in 2024, 4.234 times the 1971 level, a gain of 24.96 percentage points or 323.4%. The transition is not finished. Urban population grew 2.92% in 2025, the fastest of the six and 1.66 times the peer median of 1.76%, ahead of Pakistan 2.455%, Indonesia 1.871%, India 1.76%, Vietnam 1.482% and Thailand 1.425%. The countries with the cleanest air here are the ones whose urban transitions are largely complete: Thailand, 61.87% urban, is adding urban residents at 1.425% a year; Indonesia, 58.75% urban, at 1.871%.

That leaves Bangladesh in the least comfortable position available in this group. Worst exposure of the six now, on the smallest urban base, with the fastest urban growth still to come, inside an economy of US$450 billion growing 4.22% in 2024. Whatever cities add to particulate exposure, Bangladesh has mostly not added it yet.

Source: World Bank WDI, urban population as a share of total population (SP.URB.TOTL.IN.ZS), Bangladesh against comparators, 1971-2024.

The mortality data has not caught up, which is not reassurance

The obvious objection is that none of this shows up where it should. Premature mortality from cardiovascular disease, cancer, diabetes and chronic respiratory disease, measured as the probability of dying between the exact ages of thirty and seventy, was 18.1% in Bangladesh in 2021. That is second lowest of the six: only Thailand is lower, at 14.5%. Pakistan records 25.5%, India 23.6%, Indonesia 21.9%, Vietnam 20.1%. Bangladesh sits 3.8 percentage points below the peer median of 21.9%, at 0.83 times it.

So the country with the worst exposure has the second best record of premature death from the disease groups most associated with it. Take the pairing seriously before explaining it away, then be precise about what it cannot mean.

It cannot mean the exposure is harmless, because the two readings are neither contemporaneous nor commensurate. The exposure figure is 2023, the mortality figure 2021, and more importantly a mid-life mortality probability summarises conditions a cohort has already accumulated. Fine-particulate harm accumulates too, on a horizon longer than the two years separating these readings.

Nor is it a clean cross-country test. Six countries, one year, and every other determinant of premature adult mortality varies across them and is absent from this data: tobacco use, diet, obesity, hypertension detection and treatment, and the completeness of death registration itself. A country that registers fewer adult deaths, or assigns fewer of them to a named cause, scores better on this indicator without anyone living a day longer. The pack carries no registration-completeness measure, so this essay cannot rule that out and does not.

What the mortality figure does establish is narrow and worth stating exactly: Bangladesh's health accounts are not yet recording a penalty for the group's worst exposure. Whether that is because the penalty is small or because it is still in transit is not something six national averages can settle.

What would change this conclusion

The argument is that exposure in Bangladesh is a national, density-driven condition rather than a metropolitan one, and that the larger part of the pressure is ahead rather than behind. Two tests would falsify it.

First, if a district-level exposure series showed the national mean of 69.25 to be carried by a handful of urban districts, with most of the country far below it, the framing here is wrong and a metropolitan policy is the correct one. A single national average cannot distinguish "everywhere is bad" from "a few places are terrible", and this reading assumes the first.

Second, if the national figure fell below Pakistan's 58.37 in a future release while urban population growth stayed above the peer median of 1.76%, then density and growth are not doing the work claimed for them here, and source mix or enforcement was the binding constraint all along.

Three moves follow.

1. Publish exposure by district, on a fixed schedule, beside the national mean. Owner: Department of Environment. Success signal: an annual district series that accounts for the 69.25 national figure, which converts the first test above from an argument into a measurement.
2. Write exposure into the planning of the towns that do not exist yet. The urban share is 32.68%, lowest of the six, and urban population is growing at 2.92% a year, fastest of the six. Most of the urban Bangladesh of the 2050s is unbuilt. Owner: Local Government Division, with the municipal and town planning authorities. Success signal: statutory land-use plans, published and dated, for the fastest-growing secondary towns before their populations double, separating industrial from residential land.
3. Turn the forest share from a flat line into a rising one. It went 14.75% in 2000 to 14.49% in 2023. Owner: the Forest Department. Success signal: a reading above the 2000 level of 14.75%, which the series has not produced in any year since 2000, and which would mean the most crowded country of the six had begun buying land back rather than only spending it.

The counterargument

The strongest objection is that the exposure figure is not a policy result at all. Bangladesh sits in a basin where dust and particulates transported from beyond its borders load the air whatever is done inside them; the indicator is a modelled national mean rather than a monitor network; and the health data, 18.1% and second lowest of the six, says the damage is not reaching working-age deaths. On this reading the ranking is an artefact of geography and method, and treating it as an indictment misallocates attention.

Concede the first half. This pack contains no source attribution, no monitor readings and no transboundary decomposition, so nothing here rebuts the geography claim on evidence, and it should not pretend otherwise. If a large share of the 69.25 originates outside the country or outside human activity, the domestic lever is smaller than the rank suggests, and the honest position is that these six indicators cannot size that lever.

The second half does not survive. Exposure is what harms, whatever its provenance; a person breathing 69.25 micrograms breathes the transported fraction and the domestic fraction alike, and the 44.47 microgram distance to the peer median has to be lived with either way. And the mortality defence proves less than it claims, for the reason set out above: this is a country with the lowest urban share of the six and the fastest urban growth, which is to say a country whose exposure was recorded before most of its urban stock was built. A clean bill of health at that point in a transition describes the timing, not the safety.

---

Data sources: World Bank World Development Indicators, retrieved from the BDPolicyLab data lake, 2026-08-10. Comparators are India, Pakistan, Indonesia, Thailand and Vietnam; "peer median" is the median of the comparators reporting in the stated year, excluding Bangladesh. All six countries report every indicator used here, so each comparison is across six. PM2.5, population density and forest area are 2023, non-communicable disease mortality 2021, urban share 2024 and urban population growth 2025, being the latest year available in each series. These are cross-country associations in a panel of six and establish no causation.

Sources

1. World Bank WDI, PM2.5 air pollution, mean annual exposure (micrograms per cubic meter), EN.ATM.PM25.MC.M3: https://data.worldbank.org/indicator/EN.ATM.PM25.MC.M3
2. World Bank WDI, Mortality from CVD, cancer, diabetes or CRD between exact ages 30 and 70 (%), SH.DYN.NCOM.ZS: https://data.worldbank.org/indicator/SH.DYN.NCOM.ZS
3. World Bank WDI, Population density (people per sq. km of land area), EN.POP.DNST: https://data.worldbank.org/indicator/EN.POP.DNST
4. World Bank WDI, Urban population (% of total population), SP.URB.TOTL.IN.ZS: https://data.worldbank.org/indicator/SP.URB.TOTL.IN.ZS
5. World Bank WDI, Urban population growth (annual %), SP.URB.GROW: https://data.worldbank.org/indicator/SP.URB.GROW
6. World Bank WDI, Forest area (% of land area), AG.LND.FRST.ZS: https://data.worldbank.org/indicator/AG.LND.FRST.ZS