If Cancer Had Dropped Like This, We'd Know Why. Why Doesn't Anyone Study the Crime Decline?

Cancer mortality in the United States is down 34% since 1991. We spend roughly $7.2 billion a year through the National Cancer Institute trying to understand why and accelerate the trend. Violent crime is down about 60% over the same window. We spend roughly $30 million a year through the National Institute of Justice. The two declines saved a comparable number of lives. The two research budgets differ by a factor of about 240. That is not an accident, and it has consequences.

The Comparison That Should Be Obvious

Imagine a parallel universe where U.S. cancer mortality had fallen 34% in a generation, but nobody had seriously investigated why. No epidemiology. No mechanism studies. No randomized trials of which interventions were actually doing the work. Just a vague consensus that something is going on and a faint hope it would continue.

Nobody would tolerate that. The premise is absurd. The American Cancer Society publishes an annual report with thousands of pages of methodological detail. The Surveillance, Epidemiology, and End Results program (SEER) — a federally funded cancer registry — has tracked every cancer diagnosis in catchment areas covering nearly half the U.S. population since 1973. Mechanism is reverse-engineered from a half-century of pooled clinical data. We can tell you, in numerical detail, how much of the mortality decline came from reduced smoking, how much from earlier screening, how much from better treatment, and how those weights differ by cancer site.

For the crime decline, we have none of that. We have a small number of working academics, a handful of think-tanks doing heroic part-time work with patchwork data, and a single former data scientist named Jeff Asher publishing one of the most cited running estimates of national homicide. That is not a critique of Jeff Asher — he's doing the work everyone else should have been doing — it's a description of how thin the bench is.

The Funding Numbers

The asymmetry is concrete and it is easy to look up.

A few notes on the comparison. NCI is not the only place cancer is studied — NIH cancer-related funding across other institutes adds billions more, and private philanthropy and pharma push the real total well past $20 billion a year. NIJ is similarly not the only crime-research funder — the Bureau of Justice Statistics gets about $35 million, and CDC's violence-prevention work adds modest sums. Including those still leaves the comparison off by roughly two orders of magnitude.

Why the Gap?

There are at least four reasons, none of them sinister, and all of them additive.

1. Crime data is owned by 18,000 agencies, not one

Cancer reporting in the United States is centralized. Hospitals, pathologists, and registries feed into a small number of standardized data systems run by CDC and NCI. Crime reporting, by contrast, is owned by roughly 18,000 individual law enforcement agencies, each with its own records-management system, its own categorization conventions, its own decisions about what to publish, and its own political incentives about what to make visible. The FBI's Uniform Crime Reports is a voluntary aggregation of that patchwork. The transition to NIBRS — the more granular successor — is still incomplete after more than a decade of effort and several major cities are not reporting at all in 2026. You cannot build SEER-quality mechanism research on top of a data layer that fragmented. The data layer has to come first.

2. Crime sits in the Department of Justice, not Health

Cancer research sits at NIH, an agency whose explicit job is to understand and reduce a category of human suffering through science. Crime research sits at the Department of Justice, an agency whose explicit job is to enforce federal law. These are not the same mission. NIJ is a small research office inside a massive enforcement bureaucracy, and the research budget is forever competing with operational law enforcement priorities that are politically louder and electorally more legible. That is a structural disadvantage no amount of advocacy can fully overcome.

3. Nobody wants the answer to be inconvenient

Cancer is a politically clean problem. Nobody is rooting for cancer. Crime is not. A serious mechanism study of the 1990s decline would have to weigh the contribution of mass incarceration, broken-windows policing, lead abatement, demographics, abortion access, economic boom, immigration, and the unwinding of the crack epidemic. None of those answers is comfortable for one political coalition or the other, and the honest answer is some weighted combination of all of them. The same is going to be true of the current decline. Researchers feel the chill on both sides.

4. Cancer has constituencies; crime declines don't

Every major cancer site has an organized patient advocacy community, a disease-specific research foundation, and a constituency of survivors who lobby Congress. The constituency for understanding a crime declineis essentially nobody. Victims' groups are typically organized around the existence of specific crimes, not the absence of others. The avoided harms are invisible, which means the avoided-harm research has no political base. This is the same dynamic that makes public-health prevention chronically underfunded relative to acute treatment.

What a Serious Effort Would Look Like

If we treated the crime decline as a real research problem, the program would look something like this.

• A SEER-equivalent for incidents. Mandatory, standardized, geocoded, near-real-time incident reporting from every law enforcement agency in the country, with a small federal grant to fund the records-system upgrades needed to make compliance feasible. NIBRS is the right direction but chronically underfunded. SEER took 50 years to build and is now indispensable.
• A National Crime Mortality Index.Treat homicide as the public-health outcome it is. Link homicide records to the National Death Index. Build the equivalent of a cancer-survival statistic for places — homicide rate, trajectory, and demographic breakdown — published with the same regularity and the same scientific rigor as ACS's annual report.
• Funded mechanism studies. Comparative natural experiments — the kind of work Brennan Center and Council on Criminal Justice are doing now on a shoestring — but with NIH-scale grant programs, peer review, replication requirements, and pre-registered hypotheses. Specifically test competing theories of the 1990s decline against the 2023–2025 one. Some mechanisms should re-validate; some should not.
• A standing crime-decline research consortium. The cancer-research apparatus co-ordinates dozens of cooperative groups working on a common problem. Crime research is structurally individual: one professor, one grant, one paper. Pooling resources around the decline question would get a more honest answer faster.
• Open address-level data as a public good. The mechanism work above cannot happen on top of city-aggregated UCR rollups. It requires block-level, near-real-time, standardized incident data. Some of this can be built in the private sector — SpotCrime, OpenDataDC, OpenJustice and others are doing it now — but it should not only be a private good. The cancer registry is a public good for a reason.

None of this requires a new agency. SEER didn't require one. It required sustained funding, a standards-setting body, and a political consensus that the decline was worth understanding. The first two are budget items. The third is the harder problem.

What We're Doing About It in the Private Sector

We won't wait for federal research funding to catch up. SpotCrime has spent more than a decade building the data layer the public sector should have built: standardized incident ingestion from thousands of agencies, address-level geocoding, public access through SpotCrime.com, and a commercial API for builders who need it as infrastructure.

That data is increasingly the substrate that academic researchers, journalists, and operational risk teams are actually using. It's not a replacement for a federal research program — it shouldn't have to substitute for one — but in the absence of a SEER-equivalent for crime, the private and open-data ecosystem is what's holding the question together. The crime decline of the next decade will be understood, if it's understood at all, partly because that infrastructure exists.

A Decline We Don't Understand Is a Decline That Can Reverse

The asymmetry between cancer research and crime research is the kind of thing that's easy to dismiss as "well, of course, they're different problems." They're different problems in detail. They are the same problem in shape: a large mortality decline, a generational improvement in human welfare, a mechanism we don't fully understand, and a future that depends on whether the mechanism continues to hold.

We invested $7 billion a year, every year, in understanding cancer, and the result is one of the great public-health success stories of the modern era — incremental, hard-won, never finished. The crime decline could be the same kind of story. We are choosing, by the size of our research investment, not to find out.

The cancer-research apparatus didn't build itself. People decided it should exist. The crime-research apparatus is a decision we have not yet made.