1863 Leadership · Issue Paper No. 13
A “Low-Crime Tax”
Sixty-five percent of private prison contracts require the state to keep the beds full. Five repairs, and a referee with no stake in the game.
Abstract
More police officers prevent homicides — an additional ten to seventeen prevents one a year, and the effect is twice as large for Black victims. The same officers also generate seven to twenty-two additional arrests apiece for offenses like drug possession. That tradeoff has made this argument unresolvable for thirty years, and it dissolves the moment the petty offenses stop being offenses. This paper proposes decriminalizing conduct with no victim, with treatment capacity built first; ending license suspension for debts unrelated to driving; abolishing civil asset forfeiture; removing every financial interest — corporate, municipal or institutional — from the decision to detain a human being; and training officers to slow down, which the only randomized trial on the subject found reduced injuries to officers by thirty-six percent. We are candid that the fiscal case is weaker than its advocates claim: where legalization has actually happened, revenue exceeded forecasts and criminal justice spending did not fall at all.
Key findings
- 65 percent of private prison contracts reviewed contained occupancy guarantees — the state must keep beds filled or pay for the empty ones. Most commonly 90 percent; three Arizona contracts require 100.1
- An additional 10 to 17 officers prevents one homicide a year, twice as large an effect for Black victims — while each officer produces 7 to 22 additional low-level arrests.2
- Drug offenses account for 26 percent of all arrests in the United States — roughly 1.16 million a year, with marijuana alone at 11 percent.3
- Roughly 11 million Americans hold a suspended driver's license for unpaid fines and fees. In Tennessee, over 93 percent of those with unreinstated licenses are poor.4
- The first randomized trial of de-escalation training found use of force down 28.1 percent, citizen injuries down 26.3 — and officer injuries down 36.0. Replications are under way and not yet reported.5
- The mechanism has migrated. 90.8 percent of people in immigration detention are held in privately operated facilities under contracts carrying guaranteed minimums — up from 79 percent two years earlier.1
Section 1The question before us
A referee cannot own one of the teams. Not because referees are dishonest, but because nobody watching can tell the difference between a hard call and a bought one, and the game stops being worth watching either way. Everyone understands this about sports. We have spent forty years forgetting it about the only branch of government permitted to take a person's liberty.
Sixty-five percent of the private prison contracts one researcher could obtain contain an occupancy guarantee: the state must keep the cells filled to a stated percentage, or pay for the empty ones anyway. The most common figure is ninety percent. Three Arizona contracts require one hundred.1
A state that reduces crime owes money to a company. That is not an inference about incentives. It is a clause.
This paper is about removing financial and bureaucratic interests from decisions that should turn on nothing but whether a person did something to another person. It is not soft on crime. Two of its five reforms are designed to put more police capacity against violence, not less.
1.2 The argument was not missing. It lost.
It is customary at this point for a paper like this to confess that the people who share its premises built the system it criticizes. The honest history is more interesting than that, and less comfortable for everyone.
Milton Friedman published the case against drug prohibition in Newsweek on the first of May, 1972. He put the ethical question directly: do we have the right to use the machinery of government to prevent an adult from becoming an addict? Reason with him, he wrote. Tell him the consequences. Pray with him. But we have no right to use force.10
Seventeen years later he wrote again, this time an open letter to the newly appointed drug czar, published in the Wall Street Journal. The sentence that matters is this one:
"Your mistake is failing to recognize that the very measures you favor are a major source of the evils you deplore."
He added a prediction that has aged into a diagnosis: had drugs been decriminalized seventeen years earlier, crack would never have been invented, because it was invented precisely when the high cost of illegal drugs made a cheaper version profitable.10
The reply came twelve days later in the same newspaper. There was little if anything new in the letter, the drug czar wrote, and the legalization argument was an old and familiar one revived by a small number of journalists and academics who insist that the only solution to the drug problem is no solution.10
So the argument was not absent. It was made, in the country's most prominent business newspaper, by a Nobel laureate, twice, across two decades — and it was dismissed as unserious by the man holding the office. Then the policy he defended ran for another thirty-five years and produced the arrest numbers in Section 2.
That is the failure worth owning, and it is not a failure of analysis. The tradition that produced Friedman also produced the mandatory minimums, the forfeiture statutes and the contracting. Being right in print is not the same as prevailing, and a movement that cannot tell the difference will keep losing arguments it has already won.
Section 2The tradeoff nobody can resolve
Begin with the finding that has made this debate impossible, because everything in this paper is an attempt to escape it.
Researchers at Pennsylvania, Oregon, UCLA and Barnard measured what happens when a city hires more police. An additional 10 to 17 officers prevents one homicide per year, and in per capita terms the effect is roughly twice as large for Black victims. The mechanism is deterrence rather than incapacitation — serious-crime arrests actually fall, and the decline is four to six times greater for Black suspects.2
And each additional officer generates somewhere between 7 and 22 more arrests for low-level offenses — drug possession, liquor violations, the ordinary business of street enforcement.
So more police saves lives, disproportionately Black lives, and produces more petty arrests, disproportionately of Black people. Both halves are true. Which is why one side of this argument talks only about the homicides and the other talks only about the arrests, and why the argument never ends.
The tradeoff is real only if the petty offenses remain offenses.
Drug offenses account for 26 percent of all arrests in the United States — about 1.16 million annually, with marijuana alone making up 11 percent.3 Remove them and the additional officer still deters the homicide. They simply no longer produce the arrest.
That is the argument of this paper, and the reason its reforms belong together rather than in separate bills.
Section 3Reform I: Criminalize conduct with a victim, and stop there
The criminal law should address conduct that deprives another person of life, liberty or property. Possession and consumption should be handled as health matters. Supply should be regulated and taxed rather than prohibited — but only after treatment capacity exists, and not before.
3.1 Oregon is the reason for that last clause
Oregon's Measure 110 decriminalized possession of small quantities of all controlled substances in November 2020 — the first American state to do it. By August 2023 two-thirds of Oregonians wanted it repealed. House Bill 4002 recriminalized possession in 2024.6
The causal evidence is genuinely contested and we will not pretend otherwise. One early analysis attributed a 23 percent rise in unintentional overdose deaths to the measure. A synthetic-control study found no association. Another found the association disappears once the arrival of fentanyl is accounted for. A more recent one finds sustained increases and argues the fentanyl adjustment absorbs part of the effect. A three-year evaluation funded by the National Institute of Justice concluded there was "little to no support" for blaming Measure 110, attributing the trends to the pandemic and to fentanyl.6
Portugal is not the precedent it is usually offered as. In the year before decriminalizing, Portugal had about 10 overdose deaths per million residents. Oregon had 188. Portugal had already decriminalized in practice before 2001, and the mortality decline is probably attributable to the simultaneous large expansion of publicly funded treatment rather than to the legal change itself. Portuguese illicit drug use, meanwhile, rose from 7.8 percent in 2001 to 12.8 percent in 2022.6
The lesson is not that decriminalization fails. It is about order of operations. Portugal built the treatment system and then changed the law. Oregon changed the law and promised the treatment. The treatment was slow arriving, the public saw disorder without a visible alternative, and the reform did not survive one electoral cycle. Whatever the mortality data eventually shows, that sequence is fatal and it is avoidable.
3.2 Where the money is, and where it is not
Here we have to correct a claim one side of this argument repeats, because it will not survive scrutiny.
Miron and Waldock estimated that legalizing all drugs would save roughly $41.3 billion a year in enforcement and generate about $46.7 billion in tax revenue — the familiar $88 billion figure. An updated study put state and local prohibition spending at $29 billion and federal spending at $18 billion.7
But that same update examined what actually happened after Colorado, Oregon and Washington legalized marijuana. Tax revenue exceeded forecasts. Criminal justice expenditure fell essentially not at all.7
That finding comes from an institution that favors legalization, reporting against its own prior estimate, and it should be read as a general law of government rather than a quirk of drug policy: when a category of work disappears, budgets do not shrink. Departments reallocate.
So claim the revenue and claim the redeployment, not the savings. If a quarter of all arrest activity ceased, the country would gain enormous enforcement capacity against violence without hiring a single officer — and by Section 2's finding, deterrence operates through presence rather than through arrests. That is a real benefit, it does not depend on any budget line falling, and it is the argument that survives contact with the evidence.
The prison numbers deserve the same honesty. About 362,000 people are incarcerated for drug offenses — roughly one in five of everyone held. But in state prisons, where most prisoners are, drug offenders are 13 percent, and of 127,000 state drug prisoners only 34,382 are held for possession rather than trafficking or manufacturing. Sixty-three percent of state prisoners are there for violent offenses.8 Decriminalizing possession empties fewer cells than the rhetoric implies. Regulating supply empties considerably more.
3.3 What a legal market does that enforcement cannot
Prohibition does not remove the market. It transfers it to people willing to enforce contracts with violence, and hands them the margin.
The American precedent is exact. Alcohol prohibition created Capone and repeal ended him — not through better policing, but through competition from a legal product he could not undercut.
And it answers the objection about children better than enforcement ever has. Nobody stands on a corner selling cigarettes to twelve-year-olds. Not because cigarettes are safe — they kill more Americans than every illegal drug combined — but because there is no margin in it. The pusher is a creature of the price premium that prohibition creates. Remove the premium and the occupation disappears, which is a result no amount of enforcement has ever achieved.
We state one limit plainly. Fentanyl strains this logic, because its economics are driven by potency per gram rather than by volume, and a regulated market would have to decide what it actually sells. We do not have a confident answer for opioids and we would rather say so than pretend the cannabis case generalizes.
Section 4Reform II: Stop suspending licenses over debt
Driver's licenses should be suspended for conduct that makes a person dangerous behind the wheel, and for nothing else. Not unpaid tickets, not unpaid child support, not drug convictions, not student loans.
Roughly 11 million Americans currently hold a suspended license for unpaid fines and fees. About 40 percent of all suspensions are for reasons unrelated to driving. In Florida, failure to pay caused 77 percent of suspensions across 2012 to 2015. Twenty states have revoked licenses over student debt; in South Dakota, nearly a thousand people cannot drive because they owe money to state universities.4
The mechanism is a trap that runs itself. A person cannot pay a ticket. They lose the license. They still have to get to work. They drive, are stopped, and acquire a criminal charge and more fees — and are now further from paying than at the start. Tennessee analysis found that over 93 percent of people with unreinstated licenses are poor.4
It does not even collect the money. California replaced suspensions with income-based payment plans in 2017 and collections went up.
Washington State spent more than 70,000 staff hours in a single year processing suspensions for non-driving offenses — capacity spent on paperwork rather than on the violence in Section 2.4
This is also the reform with the most momentum and the least partisan valence. Twenty-two states and the District of Columbia have passed some version. California restored 450,000 licenses; Texas, 634,933; Virginia more than half a million. The federal Driving for Opportunity Act is bipartisan, with Republican sponsors.4
One detail shows how easily this is defeated. Several states ended debt-based suspensions but kept registration holds as the collection tool — which still gives an officer legal cause to stop the car. Daunte Wright was stopped for an expired registration.4 Reform that leaves the stop in place has changed the paperwork and nothing else.
4.1 And what should replace it
Scale fines to income. A $200 ticket is an annoyance to a lawyer and a catastrophe to a warehouse worker. Finland and Germany price penalties as a share of daily earnings, which collects more and traps fewer.
Divert before conviction, and seal automatically. A record is often a heavier sentence than the sentence. Expungement produces substantial wage gains, but take-up is low because it requires petitioning and usually a lawyer. Automatic sealing reaches people; a right to apply does not.
Schedule service around work, not against it. Community service programs commonly fail because hours fall during working hours, so participants miss shifts, drop out, and are violated into custody for non-compliance. Evenings and weekends, with the person's job treated as the fixed constraint, is the difference between a functioning alternative and a slower route to jail.
Section 5Reform III: Abolish civil asset forfeiture
Property should not be taken from a person who has not been convicted of anything. Where property is genuinely the instrument or proceeds of crime, it should be forfeited following a conviction, and the proceeds should go to a general fund that the seizing agency does not control.
Civil forfeiture has generated at least $82 billion since 2000. Across 24 states, half of all currency forfeitures are worth less than $1,678 — against an estimated $3,300 to hire an attorney to contest one. Pennsylvania's median seizure was $369; Michigan's, $423. In Philadelphia the median item taken was $600, and respondents reported seizures as small as $25 in cash and a $20 cologne set.9
In that Philadelphia sample, only about one person in four was found or pleaded guilty to any wrongdoing. Sixty-nine percent lost their property permanently. Among Justice Department forfeitures, 71 percent were completed administratively — no judge involved at all — and in three states with data, 62 to 76 percent were forfeited by default. In half the states, the deadlines before a hearing total 188 days.9
It does not fight crime. A multistate analysis found no effect on crime rates or clearance rates. What it does track is money: a one-percentage-point rise in local unemployment is associated with roughly a 9 percent increase in seizures.9 That is not law enforcement responding to crime. It is a revenue department responding to a shortfall.
And any state-level reform that ignores the federal equitable sharing program accomplishes nothing. Local police may hand seized property to federal prosecutors, have it forfeited under federal law, and receive up to 80 percent back — regardless of what their own state's law says. Closing that channel is not an optional detail; it is the reform.
Section 6Reform IV: No one may profit from the decision to detain
The state may purchase goods and services. It may never delegate the discretionary exercise of coercive power, and no party — corporate, municipal or institutional — may hold a financial interest in the outcome of a case or in the number of people held.
6.1 The quotas
In the Public Interest obtained and analyzed 62 contracts covering 77 facilities. Sixty-five percent contained occupancy guarantees. They range from 80 to 100 percent, with 90 percent the most common. Arizona holds three contracts at 100 percent; Oklahoma three at 98; Louisiana at 96; Virginia at 95.1
Where the beds go unfilled, the state pays anyway. The report calls these payments a "low-crime tax." Colorado, where crime fell by roughly a third over a decade, paid about $2 million for cells it did not need. Nashville paid $487,917 for empty beds under a 90 percent guarantee.1
Some of these clauses were not in the original agreements. They were added later, by amendment.1
That contract census was conducted in 2013 and nobody has repeated it. We flag that plainly rather than presenting a decade-old count as current. What can be established is that the practice did not stop: Arizona executed a contract amendment containing such a clause in 2017, and the Cicero Institute — which favors private provision and proposes to reform rather than abolish it — described guaranteed minimums in 2024 as the traditional structure of these agreements.1
And the centre of gravity has moved. The same mechanism now operates at far greater scale in immigration detention, where contracts carry "guaranteed minimums" paying the operator a set sum regardless of how many beds are used. As of July 2023, 90.8 percent of people held in immigration detention each day were in facilities owned or operated by private corporations, up from 79 percent in September 2021 — with average daily detention rising from 15,444 in January 2021 to 30,003, against a congressional appropriation of $2.9 billion for 34,000 beds.1
Whatever one thinks about immigration enforcement, a contract that pays a company for empty beds gives that company a financial interest in how many people are detained. That is the identical objection, at ten times the scale, in the part of the system currently growing fastest.
6.2 And the version that is not an inference
Two Pennsylvania judges accepted more than $2.6 million in payments in exchange for sentencing children to a privately operated juvenile facility. One was sentenced to more than twenty years in federal prison.1
That is not a slippery slope. It is the referee owning the team, indicted and convicted, in Luzerne County, Pennsylvania.
6.3 Where the line falls
The principle needs a workable boundary, because the easy version of this argument collapses on contact. Nobody proposes that the government manufacture its own patrol cars. Court-appointed defense counsel is usually a private attorney. Armies have contracted logistics since the Revolution. Private security guards outnumber sworn officers in this country.
The line we propose: the state may buy goods and services, and may never contract out a discretionary decision to deprive a person of liberty. Custody decisions, charging decisions, sentencing, disciplinary findings, and the use of force belong to accountable public officials. Food service may be purchased. An occupancy guarantee may not exist.
6.4 The same rule applied to governments and unions
A principle that reached only corporations would be a partisan convenience rather than a principle.
Municipalities must not fund themselves through their own courts. A town drawing a meaningful share of revenue from fines has given its police department a financial interest in stopping people, which is the forfeiture conflict wearing a different uniform. If a town needs money it should raise taxes, where voters can see the decision and reverse it.
And organized institutional interests in incarceration volume are the same problem. Corrections officer associations have lobbied against sentencing reform, most prominently in California. We do not claim they are equivalent to a lockup quota; we claim the objection is to any organized financial stake in how many people are held, and that the people making this argument do not get to notice only the corporate version.
One procedural point. Federal use of private prisons was ordered phased out in 2016, reversed in 2017, ended again by executive order in 2021, and reversed again in 2025 — four reversals in nine years. An executive order is not a reform. It is a lease. This must be statute.
Section 7Reform V: Train officers to slow down
Police are a defensive force. Force is a last resort, discharged in defense of life. The job is to deliver a person to a judge, not to be one. And de-escalation should be trained, funded and measured as a core competency rather than as a sensitivity seminar.
There is a saying in American policing: better to be judged by twelve than carried by six. It assumes a tradeoff between an officer's safety and their restraint, and the evidence says the assumption is false.
The Louisville Metro Police Department ran the first randomized controlled trial of de-escalation training in 2019 — a stepped-wedge design covering 1,049 sworn officers. Use-of-force incidents fell 28.1 percent. Citizen injuries fell 26.3 percent. And officer injuries fell 36.0 percent — the largest reduction of the three.5
The officers trained to slow down got hurt less. Not more. That is not an ethos argument, it is a finding.
We would rather make the case that way than by appealing to anyone's sense of duty. Officers are asked to absorb risk that most citizens never face, and they are entitled to policies grounded in evidence rather than in sentiment.
Replication is under way and should be watched rather than assumed. The Indianapolis Metropolitan Police Department completed the first replication across roughly 1,700 officers in 2022 and 2023 using the same stepped-wedge design, with further federally funded evaluations in Phoenix, Cincinnati and Oklahoma City and a statewide evaluation in New Jersey. The National Institute of Justice has rated the program on the strength of the Louisville trial.5 Section 8.7 sets out why we would still not call this settled.
7.1 Fewer encounters, not fewer officers
Everything above should be read alongside Section 2. We are not arguing for smaller police forces. The evidence says larger ones prevent homicides and that the benefit falls most heavily on Black communities.
What we are arguing is that every encounter between an armed officer and a citizen carries a small probability of catastrophe, and that a great many encounters are manufactured by rules that serve no safety purpose — an expired registration, a suspended license over an unpaid ticket, a quantity of a substance in a pocket. Reduce the manufactured encounters and the same officer prevents the same homicides with fewer opportunities for the interaction to end badly.
Where enforcement of a rule requires no judgment and creates no danger, it does not require an armed officer. That is worth examining honestly for routine traffic enforcement, though we note the evidence base there is thin and we do not propose it as settled.
Section 8 · The strongest case against
8.1 Oregon repealed it within four years
The first American experiment in decriminalization was reversed by the state that ran it, with two-thirds public support for repeal and endorsement from progressive officials including Portland's mayor. Whatever the causal evidence eventually shows, the political verdict was delivered quickly.6
Our answer is sequencing, and a reader may regard that as an excuse. It is at least a testable one: Portugal built treatment first and its reform has lasted twenty-five years.
8.2 The savings we are promised have not appeared
Colorado, Oregon and Washington legalized marijuana and criminal justice expenditure did not fall.7 The $88 billion figure circulating in this argument is half revenue, which is real, and half savings, which the best available natural experiment suggests may not materialize.
We report it because a reader will find it, and because a paper in this series that quoted the convenient half would deserve what followed.
8.3 Legal markets increase use
Portuguese illicit drug use rose from 7.8 percent in 2001 to 12.8 percent in 2022.6 Commercial legal markets for alcohol and tobacco produce enormous consumption and enormous harm. It is reasonable to expect more use of currently illegal substances under legality, and reasonable to weigh that against the harms of prohibition differently than we do.
We think the comparison is between a regulated market with known potency and age controls and an unregulated one run by people who adulterate their product. But more use is a real cost and we are not going to characterize it as anything else.
8.4 Fentanyl may break the model
The economics that make a legal cannabis market work — bulk, weight, visibility — do not obviously apply to a substance lethal in quantities smaller than a grain of rice and profitable at potency rather than volume. Section 3.3 says we lack a confident answer. That is a substantial gap in a paper proposing regulated supply.
8.5 More police is not a costless recommendation
Section 2 cuts both ways and we have used only one edge of it. The same research that finds 10 to 17 officers prevent a homicide finds each officer generating up to 22 additional low-level arrests. Our proposal assumes decriminalization removes most of those. If decriminalization fails politically — as it did in Oregon — a reader is left with the police expansion and not the offsetting reform, which is a worse outcome than either alone.
8.6 Public institutions have their own perverse incentives
Removing the profit motive does not produce a neutral referee. Corrections officer associations lobby for sentencing policy. Prosecutors run for office on conviction records. Police departments compete for federal grants tied to activity measures. A reader who concludes we have identified one distorting interest and left several standing is correct, and Section 6.4 concedes as much without solving it.
8.7 The de-escalation evidence has a complication we did not expect
In 2023 the Department of Justice published a pattern-or-practice finding against the Louisville Metro Police Department covering substantially the same period as the trial, concluding that the department had engaged in excessive force, searches on invalid warrants, and unlawful stops in violation of the Constitution.5
So the agency that supplied the evidentiary foundation for this training's national expansion was, during those same years, found by the federal government to be policing unconstitutionally. That does not invalidate the measured reductions — they were measured against the department's own baseline, and a 28 percent fall from a bad baseline is still a fall. But it should temper any claim that the training transforms an agency's culture.
There is a comparison worth sitting with. D.A.R.E. spread to most American school districts on the strength of enthusiasm and early impressions, and was later found not to work. A police training program with one strong trial, heavy institutional backing and a rapidly expanding footprint is in a familiar position, and the honest posture is to fund the replications and wait for them.
8.8 What we concede, and what we do not
We concede that Oregon repealed decriminalization and that the causal evidence is contested. We concede that the promised enforcement savings have not appeared where legalization occurred. We concede that legal markets likely increase use. We concede that we have no confident answer for fentanyl. We concede that the de-escalation evidence rests principally on one trial, in a department the Justice Department later found to be policing unconstitutionally. And we concede that removing corporate profit leaves other institutional interests untouched.
We do not concede that a contract requiring a state to keep prison cells filled is a defensible instrument of government, or that a police department should keep what it seizes, or that a town should fund itself by stopping its own residents. Those are not policy disagreements about how to fight crime. They are arrangements that would be recognized as corruption in any other branch of government, and they are legal here only because we stopped looking.
Section 9What we are not claiming
We are not claiming police are the problem. The evidence in Section 2 says more of them prevents homicides and that Black communities benefit most. Nothing here reduces a police budget or a police force.
We are not claiming drugs are harmless. They kill tens of thousands of Americans a year and will continue to. The question is whether the criminal law reduces that number or adds to it, and after fifty years the burden of proof has shifted.
We are not claiming this saves $88 billion. Section 8.2 explains why we think half that figure is unreliable.
And we are not claiming officers are the ones who built this. They enforce statutes they did not write, under contracts they did not sign, in towns whose budgets they did not set. The quota was negotiated by a legislature. The forfeiture fund was authorized by a legislature. The suspended license was ordered by a court. Every arrangement this paper objects to was created by someone who never had to make the stop.
Section 10The argument you can carry
The paper compressed to what a person can remember and repeat.
And the one that is indicts the argument's own side. The principle that incentives determine behavior is the oldest thing we own. We then tolerated a system where departments keep what they seize, towns fund themselves through their own courts, and companies are paid per occupied cell — and were surprised when seizures rose in recessions and prison populations proved unresponsive to falling crime. The incentives worked. They always do. Being right about the principle is worth nothing if you decline to apply it where it is inconvenient.
Section 11Conclusion
There is a version of this argument that treats police as the problem, and this is not it. The best available evidence says more officers save lives and that Black Americans benefit most from their presence. Any reform built on shrinking police forces is building against the data.
The problem is what we ask them to do, and what we have allowed other people to earn from it.
We ask an armed officer to enforce an expired registration, a suspended license over a parking ticket, a substance in a pocket. Each of those manufactures an encounter that did not need to exist, and a small fraction of encounters end in a way nobody intended. Then we let the department keep the property that officer seizes, let the town balance its budget on the tickets they write, and let a company charge the state for cells the judge declines to fill.
None of that was designed by the people in the cars. It was designed by legislatures, and it can be undesigned by legislatures.
The test is the one every American already applies to a football game. The referee may not own a team, may not bet on the outcome, and may not be paid by the play. We accept that without argument because we understand that the alternative is not merely unfair — it is that nobody believes the result.
A criminal justice system is a referee with the power to take a person's liberty. It should be held to a standard at least as high as the one we apply to a game.
Notes
- In the Public Interest, "Criminal: How Lockup Quotas and 'Low-Crime Taxes' Guarantee Profits for Private Prison Corporations" (September 2013), analyzing 62 contracts covering 77 facilities; summarized by the Office of Justice Programs, ojp.gov, and the Equal Justice Initiative, eji.org. Source for the 65 percent figure, the 80-to-100 percent range with 90 percent most common, the Arizona, Oklahoma, Louisiana and Virginia figures, the Colorado and Nashville payments, and the observation that guarantees were sometimes added by later amendment. In the Public Interest opposes privatization of public services; the contracts are primary documents and any published version should cite specific agreements directly. The contract census was conducted in 2013 and, so far as we can establish, has never been repeated — itself worth noting in a debate that turns on these terms. On the persistence of the practice, see the Arizona Department of Corrections contract amendment of April 21, 2017, catalogued in the American University Business Law Review, aublr.org; and Cicero Institute, "Aligning Profit with Outcomes in Private Prison Procurement" (2024), ciceroinstitute.org, which describes guaranteed minimums as the traditional contract structure and proposes performance-based contracting as an alternative — Arizona House Bill 2783 of 2024 being its legislative expression. The Cicero Institute favors private provision reformed rather than ended; we cite it because a source that rejects our conclusion confirms our premise. On immigration detention — 90.8 percent of daily detainees in privately owned or operated facilities as of July 2023 against 79 percent in September 2021, average daily population rising from 15,444 to 30,003, a $2.9 billion appropriation for 34,000 beds, and the use of guaranteed minimums — American Civil Liberties Union, "Unchecked Growth," aclu.org. The ACLU litigates against immigration detention; the population figures are ICE's own. These are 2023 figures and detention populations have moved substantially since — update before publication. On the Luzerne County prosecutions, see contemporaneous reporting of the convictions of Mark Ciavarella and Michael Conahan; figures for the payments vary across accounts between roughly $2.6 and $2.8 million and should be confirmed against the court record.
- Aaron Chalfin and colleagues at the University of Pennsylvania, the University of Oregon, UCLA and Barnard College, published in American Economic Review: Insights; summarized at penntoday.upenn.edu. Source for the estimate that 10 to 17 additional officers prevent one homicide annually, that the per capita effect is roughly twice as large for Black victims, that each additional officer is associated with 7 to 22 additional low-level arrests, and that the decline in index-crime arrests is 4 to 6 times greater for Black suspects. See also Chalfin and McCrary (2017) and Mello, "More COPS, Less Crime."
- FBI Uniform Crime Reporting data on drug abuse violation arrests, as compiled by the National Center for Drug Abuse Statistics, drugabusestatistics.org. Approximately 1.16 million annual arrests for sale, manufacture or possession; drug arrests including marijuana at 26 percent of all arrests, marijuana alone at 11 percent. Arrest totals fell sharply in 2020 and have since partially recovered; verify against the most recent UCR release before publication.
- Fines and Fees Justice Center and the Michigan Journal of Law Reform, "Driver's License Suspension for Unpaid Fines and Fees: The Movement for Reform," repository.law.umich.edu. Source for the 11 million figure, the 22 states plus D.C. that have passed reform, the restoration counts in California, Texas and Virginia, the registration-hold compromise, and the Daunte Wright example. On the 40 percent share of suspensions unrelated to driving, the Florida figure, the South Dakota student-debt example and the Washington State staff hours, see Civil Rights Corps, civilrightscorps.org, and its companion fiscal analysis, which is also the source for the Tennessee finding that over 93 percent of those with unreinstated licenses are poor. On California collections rising after the 2017 reform, The Pew Charitable Trusts, pew.org. On the federal bill, Senator Kevin Cramer's announcement of the bipartisan Driving for Opportunity Act, cramer.senate.gov. Several of these organizations advocate for the reform this paper endorses; readers should weigh the framing, though the underlying counts come from state motor vehicle records.
- Robin S. Engel, Nicholas Corsaro, Gabrielle T. Isaza and Hannah D. McManus, "Assessing the impact of de-escalation training on police behavior: Reducing police use of force in the Louisville, KY Metro Police Department," Criminology & Public Policy 21 (2022): 199–233, onlinelibrary.wiley.com. Stepped-wedge randomized controlled trial covering 1,049 sworn officers, evaluating the Integrating Communications, Assessment, and Tactics curriculum developed by the Police Executive Research Forum. Reductions of 28.1 percent in use-of-force incidents, 26.3 percent in citizen injuries and 36.0 percent in officer injuries, larger than contemporaneous changes in arrest patterns. On replication: the National Policing Institute conducted the first replication with the Indianapolis Metropolitan Police Department across approximately 1,700 sworn officers in 2022–2023 using the same stepped-wedge design, with further National Institute of Justice-funded evaluations in Phoenix, Cincinnati and Oklahoma City and a statewide evaluation in New Jersey, policinginstitute.org. Results were not fully published at the time of writing and should be checked. The program is rated by the National Institute of Justice's CrimeSolutions on the strength of the Louisville trial, crimesolutions.ojp.gov. On the complication in Section 8.7, the Department of Justice's 2023 pattern-or-practice findings regarding the Louisville Metro Police Department covered substantially the period of the trial. The National Policing Institute develops and evaluates this curriculum; readers should weigh its publications accordingly.
- On Measure 110's passage, repeal by House Bill 4002 and public opinion, and on the contested causal literature: Portland State University's three-year National Institute of Justice-funded evaluation, pdx.edu; and a review of the competing synthetic-control analyses at medrxiv.org, which sets out the four conflicting findings summarized in Section 3.1. On the pre-decriminalization overdose rates — approximately 10 per million in Portugal against 188 per million in Oregon — see the analysis published in the Journal of Health Economics, sciencedirect.com. On Portuguese drug use rising from 7.8 percent to 12.8 percent, and on the attribution of Portugal's mortality decline to its treatment expansion, see reporting collected by the Foundation for Drug Policy Solutions, gooddrugpolicy.org, an organization opposed to decriminalization, and the Reason Foundation's comparative analysis, reason.org, which supports it. We cite both deliberately.
- Jeffrey A. Miron and Katherine Waldock, "The Budgetary Impact of Ending Drug Prohibition," Cato Institute (2010), ssrn.com, for the $41.3 billion enforcement saving and $46.7 billion revenue estimates and their federal and state splits. Jeffrey A. Miron, "The Budgetary Effects of Ending Drug Prohibition," Cato Institute, cato.org, for the updated $29 billion state and local and $18 billion federal expenditure figures — and, critically, for the finding that legalization in Colorado, Oregon and Washington generated more tax revenue than forecast while producing essentially no reduction in criminal justice expenditure. Cato favors legalization and reports that finding against its own earlier estimate, which is why we give it prominence.
- On the 362,000 people incarcerated for drug offenses across state prisons, local jails and the federal system, Prison Policy Initiative, prisonpolicy.org. On the composition of state prison populations — drug offenders at 13 percent or about 127,000, of whom 34,382 were held for possession rather than trafficking or manufacturing, against 642,500 held for violent offenses — Manhattan Institute, manhattan.institute, which also documents the failure of states to reduce real corrections spending as populations fell. The Prison Policy Initiative favors decarceration and the Manhattan Institute is skeptical of it; the two sets of figures are consistent and we use both.
- Institute for Justice, Policing for Profit, fourth edition, ij.org, for the $82 billion total since 2000, the finding that half of currency forfeitures across 24 states are worth less than $1,678 against an estimated $3,300 attorney cost, the 71 percent of Justice Department forfeitures completed administratively, the 62 to 76 percent default rates in three states, and the 188-day median pre-hearing timeline. On Philadelphia — median item $600, median incident $1,370, one in four found or pleading guilty while 69 percent forfeited — Institute for Justice, ij.org. On the absence of any crime-fighting effect and the association between a one-point rise in unemployment and roughly 9 percent more seizures, Brian D. Kelly, "Fighting Crime or Raising Revenue?", Institute for Justice, ij.org. The Institute for Justice litigates against civil forfeiture; its data collection is the most comprehensive available and its conclusions should be read with its position in mind.
- Milton Friedman, "Prohibition and Drugs," Newsweek, 1 May 1972, p. 104, collected in An Economist's Protest (1972) and available through the Hoover Institution's Collected Works, miltonfriedman.hoover.org. Milton Friedman, "An Open Letter to Bill Bennett," The Wall Street Journal, 7 September 1989, A16 — source for the quoted sentence and for the observation regarding crack cocaine; text at web.uncg.edu. William J. Bennett, "A Response to Milton Friedman," The Wall Street Journal, 19 September 1989, web.uncg.edu. Quotations should be verified against the original published texts before publication; the versions available online are transcriptions.
A note on the author
Issue papers are published under the name of 1863 Leadership rather than an individual byline.
A note on sources
We have deliberately cited organizations on both sides of each question and said which is which: Reason and the Foundation for Drug Policy Solutions on decriminalization, the Prison Policy Initiative and the Manhattan Institute on prison composition, the Institute for Justice on forfeiture, In the Public Interest on contracting. The most important citation in the paper is note 7, where an institution that favors drug legalization reports that the enforcement savings it had previously estimated did not appear. We have given that finding more prominence than the figure it undercuts. Note 1 flags that the contract analysis is from 2013 and requires updating, and note 5 flags that the de-escalation evidence rests on a single site.
Recommended citation
1863 Leadership. “A ‘Low-Crime Tax.’” Issue Paper No. 13. September 2026. 1863leadership.org
Corrections: None to date. Errors of fact are corrected on this page within one business day of notice, with a dated note describing the change.