No, Seriously, Decriminalizing Drugs Kills People

A new study should lay to rest the debate over legalization.

Courtesy Patrick T. Fallon/AFP via Getty Images

Is banning drugs a bad idea? For decades, libertarians and progressives have insisted that prohibition creates more harm than it prevents, producing needless arrests and prison sentences while pushing users into the shadows where overdose is more likely. That argument won some purchase among the public: as of 2019, a majority of Americans endorsed making drug crimes into civil offenses instead of felonies, treating them more like minor traffic violations than something meriting prison time.

In 2021, the legalizer coalition got to run their experiment. That year, Oregon and Washington both effectively decriminalized the possession of small amounts of all controlled substances, replacing potential prison sentences with a regime of minimal fines. Proponents promised a steep decline in overdose deaths and improvement in public safety. But did that happen?

Not quite. As I have argued, public disorder and crime surged in the wake of decriminalization. Now new research brings additional data to bear on the topic, finding that decriminalization caused over 1,000 excess deaths per year—a dismal cost to a misguided experiment.

The new paper by economists David Hall, Benjamin Hansen, and Kyutaro Matsuzawa exploits the fact that only two states decriminalized, while possession remained illegal in the remaining 48. The researchers use this to construct what’s called a “synthetic control,” a weighted average of the non-decriminalization states designed so that trends in overdose deaths match those in the states where decriminalization happened. In theory, this approach allows researchers to envision the path Oregon and Washington would have followed if decriminalization hadn’t happened. The difference between that path and what actually happened is taken to be the effect of decriminalization.

The recent paper isn’t the first to attempt this approach. Two dueling papers, both using about a year of data but making slightly different methodological choices, investigated the question a three years ago. One found that decriminalization caused an increase in overdose deaths in both Oregon and Washington; the other found no increase in Oregon, and an ambiguous result in Washington.

The following year, a third paper, written in part by proponents of decriminalization, also found that decriminalization in Oregon and Washington preceded an increase in overdose deaths. But the authors found that effect went away when they controlled for the spread of fentanyl in both states—a concurrent event which, they asserted, was causally unrelated to decriminalization but which was actually to blame for the increase in deaths.

The new paper extends this debate in two important ways. First, it adds an additional two years of data. Second, it looks at several different measures of the spread of fentanyl into the drug supply, challenging the conclusion that fentanyl, rather than decriminalization, is behind the increase in overdose deaths.

Source: Hall, Hansen, and Matsuzawa 2026

Using the expanded data, Hall, Hansen and Matsuzawa estimate the effect of decriminalization in Washington and Oregon relative to their synthetic control. They find that decriminalization increased overdose deaths by 0.806 deaths per 100,000 people in Oregon and 0.732 per 100,000 in Washington. These represent a 38 percent and 30 percent increase over baseline, respectively.

In other terms, Oregon saw an extra 1,186 deaths over three years, while Washington saw an additional 1,895, both substantial figures relative to the baseline number of overdose deaths. Notably, the effects of both decriminalizations get bigger as time goes on—implying that the earlier estimates were necessarily missing big effects by using limited data.

The authors then turn their attention to the “fentanyl spread” story. That argument has always struck me as dubious: it implies both that decriminalization did not directly cause an increase in overdose deaths, and that it also was totally unrelated to the spread of fentanyl in Oregon and Washington, specifically. This “total coincidence” theory is un-parsimonious—it is entirely possible, indeed plausible, that decriminalization precipitated or exacerbated the spread of fentanyl to Oregon and Washington, in turn driving the increase in deaths.

As it turns out, the fentanyl story does not hold up empirically. Investigating it, Hall, Hansen, Matsuzawa first observe that they find no equivalent increase in overdose deaths in California, Nevada, and Idaho using the same methods. That undermines the idea that the spread of fentanyl to the West Coast generally caused the increase in overdose deaths in Oregon and Washington specifically.

They then look at various measures of the spread of fentanyl. The original paper preferred one: the share of reports containing fentanyl to the National Forensic Laboratory Information System, a data system that tracks police drug seizures. But Hall, Hansen, and Matsuzawa argue that the increasing fentanyl share is driven by a decrease in the number of non-fentanyl drugs reported—a decrease in the denominator, rather than an increase in the numerator. In other words, it’s not so much of a measure of the increase in fentanyl as it is a measure of the decrease in drug policing activity that was caused by decriminalization.

The authors then show a variety of ways that the “fentanyl spread” effect can be made to disappear using reasonable assumptions. Using other measures—such as fentanyl reports per capita or fentanyl’s share of opioid-related drug reports—they find that decriminalization does appear to increase deaths. In other words, the original fentanyl finding is highly sensitive to which measures are used, among other factors.

Finally, the authors extend the third paper’s analysis with another year of data. When they do so, they find that decriminalization causes a statistically significant increase in overdose deaths, even when using the original paper’s preferred methodological approach.

All this adds up to is a simple conclusion: decriminalizing drug possession caused a very large increase in drug overdose deaths in Washington and Oregon. That increase is not the result of a coincidental wave of fentanyl; it’s the result of a deliberate policy change.

None of that would come as a surprise to the people of Oregon and Washington, whose representatives dumped decriminalization amid mass popular backlash. If you visited Portland during decriminalization, as I did in 2023, you saw the effects with your own eyes: a massive open-air drug market dominating much of the downtown, with police unable to do anything other than write tickets and refer people to a treatment hotline no one called.

Implicitly and explicitly, critics of drug prohibition have long argued that drug control hurts people, rather than the drugs themselves. If it weren’t for prohibition, they maintain, everyone would have access to clean, safe substances of their choosing, and stigma-free options for treatment the minute they wanted to get out.

Decriminalization was an experiment in whether prohibition really was what was causing our drug crisis. Yet not a single study has been able to show that it reduced overdose deaths; now, the best evidence seems to show conclusively that it caused them to explode. Add to that the surge in crime, and it’s hard to see decriminalization as anything but an abject failure.

At a certain point, therefore, we need to admit: it’s not prohibition that’s causing our drug crisis. To paraphrase James Carville, it’s the drugs, stupid. Stop pretending otherwise.