The Trump administration recently announced that it would terminate Justice Department consent decrees in Minneapolis and Louisville — two cities placed under federal oversight after the murder of George Floyd and the police killing of Breonna Taylor, respectively. Predictably, the reaction was swift and furious, not least because the five-year anniversary of Floyd’s death is today. Civil rights groups, progressive mayors, and Democratic officials accused the administration of turning its back on police accountability.

But Trump is right. These consent decrees should end — not because misconduct never occurred, but because the federal response is based on bad science with disastrous consequences for public safety.

Consent decrees are legally binding agreements in which cities agree to reform their police departments under federal supervision. They often follow DOJ “pattern or practice” investigations, which allege systemic misconduct — especially racial discrimination. Minneapolis and Louisville were both accused of such patterns, and in both cases, the DOJ relied heavily on statistical claims to support the charge of racial bias.

Yet those claims collapse under scrutiny. The DOJ’s 2023 Minneapolis report alleges that the police department engages in “discriminatory policing” against black residents. The evidence? Black individuals are subjected to use of force and vehicle searches at rates higher than their share of the population. After adjusting for a few race-neutral factors — such as time of day, location, or whether a weapon was involved — the DOJ still found a small residual disparity, which, they argued, was proof of racism.

This is statistical malpractice. As social scientists have long warned, disparities are not discrimination. To infer bias, we must rule out all relevant race-neutral explanations — a standard the DOJ fails to meet. In fact, the DOJ’s own report quietly shows that 86% of the racial disparity in use of force and 93% of the disparity in vehicle searches is explained by non-racial factors. The small remainder is far too modest, and far too ambiguous, to justify sweeping federal intervention. The most prudent interpretation is to assume the observed disparities would disappear entirely in the presence of a more comprehensive set of statistical controls.

Instead of following the facts to their parsimonious conclusion, Biden’s DOJ imposed a burdensome consent decree requiring layers of reporting, federal monitoring, and dramatic changes in policy and practice. Officers pulled back. Proactive policing dropped. Violent crime surged. Communities — especially minority communities — suffered the consequences.

We’ve seen this before. A landmark study by economists Tanaya Devi and Roland Fryer found that DOJ investigations launched in the wake of viral police incidents led to almost 900 excess homicides and 34,000 additional felonies. The problem wasn’t reform — it was deterrence. Officers feared becoming the next viral video or federal case study, so they disengaged. Policing fell. Crime rose.

Minneapolis saw this pattern in real time. After George Floyd’s death and amid federal scrutiny, officer morale cratered and staffing collapsed. The number of sworn officers in the city dropped by nearly a third. Violent crime, homicides, and carjackings spiked. The very communities allegedly harmed by police were now left more exposed to violence — with no recourse and no reassurance.

In New Jersey, a similar story unfolded on the Turnpike. After accusations of racial profiling, state police scaled back enforcement. The result was a spike in traffic fatalities. When independent researchers had studied the data, they found that racial disparities in stops were fully explained by differences in speeding and driving behaviour — not bias. Findings from this methodologically superior study were conspicuously absent from the report informing the New Jersey Attorney General’s Office of Public Integrity and Accountability.

This is what happens when policy follows narrative instead of evidence. Consent decrees, meant to protect, have too often done harm. They rely on junk statistics, fuel public distrust, demoralise police, and ultimately make cities less safe.

None of this excuses actual police misconduct, which remains a real problem. But any reform must be rooted in truth. The DOJ should investigate actual abuses, not statistical fictions. And the federal government should not micromanage local departments based on weak evidence and ideological zeal.

The Trump administration is right to pull the plug on these decrees. Minneapolis and Louisville deserve accountability, but they also deserve safety. And that starts with restoring the rule of law, not distorting it.