ten countries, ten phaseouts

if childhood lead pushes adult violence and self-harm at the modal age, each country's outcome decline should show up roughly a cohort-age after that country's own phaseout cliff, not all synchronised in the 1990s. japan 1986. canada 1990. south korea 1993. germany and the us 1996. mexico 1997. uk 1999. france 2000. australia and italy 2002.

this is a natural-experiment view of the cohort claim. ecological association, not individual causation. outcomes are the OWID mirror of UNODC harmonised homicide and the WHO Mortality Database age-standardised suicide series. cohort-lead curves are constructed indicative shapes scaled to the US ~16 ug/dL plateau, pinned at each country's published phaseout year.

If lead does what we say it does, the turn should happen at each country's own cohort offset, not all at once

Ten countries, ten phaseout dates, spread across the 1986 to 2002 window: Japan 1986, Canada 1990, South Korea 1993, Germany and the United States 1996, Mexico 1997, the United Kingdom 1999, France 2000, Australia and Italy 2002. If childhood lead pushes adult violence and self-harm at the modal age the cohort produces them, each country's outcome decline should show up roughly a cohort-age after that country's own cliff, not all synchronised in the 1990s. This is a natural experiment view of the cohort claim. Ecological association, not individual causation. Data source for outcomes is the OWID mirror of UNODC harmonised homicide and the WHO Mortality Database age-standardised suicide series. Cohort lead curves are constructed indicative shapes, scaled to the US ~16 ug/dL plateau and pinned at the country's published phaseout year.

1. Metric

2. Small multiples, one country per panel

Outcome rate per 100k (left axis, solid)   Constructed cohort-lead curve at birth year (right axis, dashed)   Country phaseout window
Outcome years are plotted, but each year is also a birth year: birth_year = year minus modal age, which is 21 for homicide and 40 for suicide. The cohort-lead curve underneath is read by birth year, so the dashed line and the solid line meet on the same x-tick by construction. Each panel uses its own y-scale; absolute baselines vary by an order of magnitude across countries and forcing a shared scale would crush the small countries flat.

3. All ten overlaid, normalised to each country's own pre-decline peak

Same outcome series, each rescaled so that country's own peak in the visible window equals 100. The point of this card is the SHAPE of the decline and the LAG between countries, not the level. If cohort lead does what the literature says, the lines should fall in rough proportion to how long ago each country's leaded-petrol cliff happened. Japan's cliff is decades older here, so its curve is more settled. Italy and Australia, both 2002, are at the recent end and show the most incomplete declines. Mexico is the visible outlier on homicide: its cartel-era rise from 2007 swamps any cohort signal in the data, which is exactly what an honest panel should show rather than hide.

4. What to read into this, and what not to

The visual claim is the lag-shifted decline pattern, not the absolute levels. Ten countries with very different homicide baselines (Japan under 1 per 100k, Mexico over 25 at peak) and very different post-war histories all show, at least in the bulk of the panels, a downward bend that lines up better with each country's own cohort offset than with any single global year. This is consistent with the lead-cohort hypothesis. It is not proof: lots of things changed in these countries over the same decades (policing, drug markets, abortion access, demographic composition, alcohol policy, healthcare, and in Mexico's case a cartel war beginning in 2006-2007 that dominates the homicide panel). The defensible read is corroborative shape, not isolated causation.

Honest gaps are flagged in the panels. UNODC cross-country homicide starts only in 1990, so the long pre-decline ramp is not in the data for any country. Germany's reunified series starts at 1990 for both metrics, so the Germany panel cannot show its own pre-1990 plateau. Suicide's modal age (40) means much of the relevant cohort is born before the petrol era for Japan in particular, so suicide is the weaker of the two outcomes for this kind of comparison. Suicide is a sequestered series in the US corpus for related reasons and is shown here for transparency, not as the headline.

Find lead in your own home

Ten countries phased out leaded gasoline. Lead paint is still in the kitchen.

Get the Drip Kit, $50 Read the full report

Phasing out leaded gas drove cohort childhood blood lead to under 1 ug/dL. Legacy lead paint and lead-pigment imports keep showing up in dishes and toys. FluoroSpec finds it. The Drip Kit is the entry option.