Mexico

What the record shows

The Mexican record is the most heavily studied in this atlas, and that is what makes it uncomfortable reading. Most national records here rest on what an operator says about itself; this one rests disproportionately on peer-reviewed evaluation, because Mexico City has been sampling its air since 1966, running the automatic RAMA network since 1986, and publishing enough for four decades of economists and atmospheric scientists to test it. The result is a record where the flagship instrument is also the best-documented disappointment of its genre. Being measurable is not the same as being right — it is the precondition for finding out you were wrong.

The case that carries this is Hoy No Circula, the oldest plate-based driving restriction in the atlas. Since November 1989, after an ozone emergency, a car's last plate digit has kept it off the road one weekday a week, and a compulsory six-monthly emissions test assigns a hologram that decides how hard the rule bites. It is not a gesture: it has real legal machinery behind it, and the machinery is anchored to a measurement. Since 2016, a reading of 155 ppb of ozone on the metropolitan network triggers a Phase I contingency declared by CAMe, the federal-state commission for the Valley of Mexico — and the declaration binds: the driving ban doubles to run from 05:00 to 22:00, factories must cut ozone-precursor emissions by 40 per cent, sites over 5,000 square metres stop building, and the Tula refinery is capped at 75 per cent of process capacity. The threshold has only ever been tightened.

And thirty-seven years on, the peer-reviewed evidence says the ban does not clean the air. Davis found no improvement across pollutants and specifications, together with evidence that the restriction raised the total number of vehicles in circulation and shifted the fleet toward high-emission cars; his later work found the 2008 Saturday extension had virtually no discernible effect across eight pollutants, and no sign that drivers switched to metro, bus or light rail; Gallego, Montero and Salas measured the workaround directly, an initial drop in car use followed by a rise in the household vehicle stock within eleven months. Mexico City's air did get dramatically better — particulate pollution is down 60.4 per cent since the ProAire programme began in 1990 — but that is the long-run policy mix of unleaded fuel, catalytic converters, fuel switching and transit, not the emergency lever. A measured value with legal force is only as good as the lever it pulls, and this is the atlas's clearest demonstration that a perfectly disciplined trigger can be wired to the wrong one.

The same distinction — the instrument versus the thing that actually did the work — runs through the transport entries. Metrobús opened along Avenida Insurgentes in June 2005 and has peer-reviewed before-and-after measurements most BRT corridors never get: commuter exposure down by up to 45 per cent for carbon monoxide, 54 per cent for benzene and 30 per cent for fine particles, with a later differences-in-differences study finding reduced ambient CO, NOx and PM10. Those gains were bought with the physical swap of roughly 260 minibuses and 90 buses for about 98 articulated vehicles on one avenue. The open real-time feed arrived thirteen years later. Cablebus makes the point geographically: an independent ITDP and Inter-American Development Bank review found travel times cut by almost 45 per cent at an operating cost of 8.73 pesos per passenger, below Metrobús, the Metro and the city's buses — but the two lines climbing into hillside neighbourhoods carry the riders, while the flat park line runs at well under half its design capacity. A cable car earns its place by terrain, not by novelty.

Where the city instruments without a consequence attached, the record is thinner. C5 is the largest camera network this atlas documents anywhere: 113,814 street cameras in March 2026, watched from seven regional centres, most on masts that also carry a loudspeaker, a strobe and a panic button, and that relay the national seismic alert from about 13,900 poles. Its peer-reviewed findings are split, one 2025 study linking the programme to a significant fall in pedestrian robbery and a 2023 study finding no effect on delivery robberies. The number that frames it comes from Data Cívica and R3D: about 60 per cent of crimes happen within 200 metres of a C5 camera, and police use the footage in under 1 per cent of investigations. Coverage is not evidence, and the binding constraint is the institution that has to consume the data. The water utility shows the same shape in reverse: it has cut the network into metered sectors since 2019, built more than half of them, missed its 2024 completion target, still puts losses at around 40 per cent of supply — and the public dashboard of network readings promised for the end of 2022 never appeared. Residents get a rationing-schedule lookup instead of the measurements, while a 2023 study of 2,595 households found 73 per cent in the five eastern boroughs receive water only intermittently.

Two entries here are unambiguously good, and both are about access rather than sensing. One card now opens the Metro, Metrobús, light rail, trolleybus, city buses, Cablebús, the bike share and the Toluca train; the magnetic paper ticket the Metro had sold since 1969 was retired in March 2024, and ordinary bank cards have worked at every station since that October. Taking cash out of the gate turned out to be far easier than integrating the fare: riders still pay each operator separately, there is no cap and no pass, and stored value quietly expires 300 days after its last use. Ecobici, meanwhile, is Latin America's largest bike share and logged over 22 million rides in 2024, up 83 per cent year on year, with about six in ten riders using it to get to work. If the entries below agree on one lesson, it is that Mexico City has been unusually willing to let itself be counted — and the reward for that willingness is a record honest enough to show which of its own instruments never worked.