I’ve been thinking about LEEP since I saw their $1.66 cost estimate (in expectation) to avert a case of lead exposure in children. But I suspect there’s a category of hard-to-analyze but valuable benefits that lead exposure elimination has that are not part of their analysis: the connection between lead and anti-social behaviors such as crime.

This 2022 meta-analysis seems very dour on the lead-crime hypothesis being as important as other literature suggestions, but still generates a prediction that dropping lead exposure led to 7-28% of the drop in crime in America post-1980s. This chart seems to indicate that could be about 1/100,000 fewer people getting murdered a year, for instance. That might sound like a small change, but it’s an annual rate, not an absolute number of murders. Let’s use LEEP’s estimate they advanced things by 8 years as a starting point to convert that annual rate to a total marginal difference.

How big was the exposure change that led to the crime reduction in the US? It is currently at ~2% and this paper seems to indicate it was essentially everyone in the US. LEEP says 1/3 children are exposed to lead, but this is going to disproportionately in lower income countries, and if the US had nearly universal exposure, perhaps that’s true of lower income countries today. Perhaps almost all of the children in Malawi have lead exposure, but I think it is likely a smaller dose due to the removal of leaded gasoline. It’s likely non-trivial to get super compelling information here. Let’s half the expected impact here to be conservative about scope comparison issues (I think this is possibly unneeded).

Imagine you get a group of 100,000 people in a Malawi-like situation, it might take (8 years * (100,000 people / (45 year life expectancy) = new people per year open to childhood lead poisoning) * $1.66 = ~30k to get that earlier lead-free-ness for the cohort. And that would probably result in 1/100,000 * 8 years / 2 (for conservative comparisons) = 4 fewer murders (forgetting other crimes for a moment). This makes the intervention, of saving 1 life for ~$7,500, reasonably competitive with Malaria interventions without considering lost income or direct health benefits at all.

My error bars are huge. Maybe 5x-ish around dose-dependent lead-crime response alone, which I don’t have an obvious way of addressing. But I think this shows it could be a material factor in the analysis of the impact of these programs.