AMERICANDIVIDENDFUND EST·MMXXVI American Dividend Fund Est. July 4, 2026 · A nonpartisan policy laboratory

Proposal № 095 of 250  ·  Released October 6, 2026

The Bail Table

On any night, roughly 400,000 Americans are in jail without having been convicted of anything, most of them because they could not pay a sum a judge set in ninety seconds. New Jersey ended money bail in 2017 and its pretrial jail population fell 44 percent with no rise in missed court dates or new arrests. Illinois ended it in 2023 and rural jail populations fell by a quarter. Freedom before trial should turn on a finding, not a bank balance: a hearing in 48 hours, a published risk table the judge must address on the record, and no one held for being poor.

The CovenantShare on X

The problem

The American jail holds two populations that the public tends to confuse. One is serving a sentence. The other, the larger one, has been convicted of nothing and is waiting for a trial. Of the roughly 650,000 people in local jails on a given day, about 70 percent are unconvicted, on the order of 400,000 people, and the great majority of them are there for one reason: a judge set a money bail and they could not raise it. The sum was set, in most courtrooms, in a hearing of a minute or two, often without a lawyer, from a schedule that keys the amount to the charge and not to the person.

Money bail sorts by wealth, not by risk. A dangerous defendant with money walks out; a harmless one without it stays in for weeks, loses the job, sometimes the apartment and the children, and then, facing a trial from a cell, pleads guilty to get out. The research on this is consistent: pretrial detention, holding everything else constant, raises the probability of conviction and of a later offence. The country is manufacturing the outcomes it fears, at about $100 a day a person.

Two states have run the alternative at scale. New Jersey ended money bail in 2017, replacing it with a hearing, a risk assessment and a rule that detention requires a finding. Its pretrial jail population fell 44 percent in two years, from 8,899 to 4,976, with no increase in failures to appear or in new charges while released, and a later study found no rise in gun violence. Illinois abolished cash bail in September 2023; in the first year, average jail populations fell 14 percent in urban counties and 25 percent in rural ones, with 89 percent of those still jailed held on a judicial finding. The federal courts have run on the same principle since the Bail Reform Act of 1984.

№ 061 put the sentence on a table. This proposal puts the decision that comes before it on one.

The proposal

The Bail Table: in every jurisdiction that takes federal justice money, release before trial is the default. Detention requires a hearing within 48 hours of arrest, with counsel, at which the prosecution must show on the record that no condition of release will reasonably assure appearance or safety. A published risk table, validated annually with its error rates by race and by charge disclosed, informs the finding and does not make it. Money may not be a condition of release. A detained defendant's trial begins within 90 days.

How it would work

  1. Release is the rule. A person arrested is released on their own recognisance, with conditions where needed, unless the prosecution moves for detention. Most arrests, for most offences, end at the station with a court date.
  1. Forty-eight hours to a hearing. Where the prosecution moves for detention, the hearing is within 48 hours, the defendant has counsel, and the prosecution carries the burden. Detention is ordered only on a written finding that no condition of release, supervision, monitoring, a no-contact order, would reasonably assure the person's appearance and the safety of others. The order says why, and it is appealable.
  1. The table. Each jurisdiction adopts a risk table: a published, validated instrument that scores the likelihood of flight and of a new offence from the record, with the validation study and the instrument's error rates by race, by charge and by county published every year. The table is advisory. A judge who departs from it says so on the record; a judge who follows it says so too. № 061's principle applies: the tool does not decide, and neither does the judge in private.
  1. No money. No release condition may require payment. The bail bond industry, which collects a nonrefundable premium of about ten percent from the families of the accused and returns nothing, ends in the jurisdictions the rule covers. Where a defendant is a genuine flight risk, the answer is supervision or detention, not a price.
  1. Ninety days. A detained defendant's trial begins within 90 days of the detention order, or the defendant is released, unless the defence asks for more time. A person held on a finding of danger is held on the state's promise to prove it soon.
  1. The condition. The rule binds the federal courts by statute, and the states as a condition of the Byrne and COPS grants of № 056. The states that have already done this receive the money as now; the rest choose.

The numbers

The detained. About 400,000 people on a given night, about $100 a day each: on the order of $15 billion a year in jail costs for people convicted of nothing, before the cost of the jobs, homes and custody arrangements they lose.

What New Jersey shows. A 44 percent fall in the pretrial population with no rise in failures to appear or new charges. If the national experience were half that, about 90,000 fewer people in jail on a given night, and about $3 billion a year in jail costs, offset by the cost of hearings, counsel and supervision, which is a fraction of it.

The industry. Commercial bail collects premiums on the order of $2 billion a year from the families of defendants, and the money is not returned when the defendant appears. That is a private tax on being accused, and the Table ends it.

The hearings. A 48-hour hearing with counsel for every detention motion is work the courts do not all do now; the cost is real and it is what the Constitution's promise of due process was always supposed to cost.

The honest objections

"Dangerous people will be released and someone will be killed." The strongest objection, and the data from New Jersey and Illinois answer it more directly than any argument: no measurable rise in new offences, in failures to appear, or in gun violence. The reason is that the Table does not release the dangerous; it detains them on a finding, which money bail does not do. Under money bail, a dangerous defendant with a rich family is released tonight. Under the Table, the prosecution asks a judge to hold them, and the judge can.

"Risk tables encode racial bias." They can, and item 3 is written for that reason: the instrument is published, validated annually, its error rates by race disclosed, and it is advisory. A judge's unaided instinct, which is the current instrument, is not published, not validated and not disclosed, and the research finds it does worse. A visible tool with a known bias is a tool that can be corrected; an invisible one is not.

"The bail industry employs people and posts bonds the state cannot." It employs people to collect a premium from the poor for a service the federal courts have done without since 1984. The industry's own argument, that defendants appear because bondsmen chase them, is contradicted by the states that abolished it and saw no change in appearance rates.

"Prosecutors will move for detention in every case and the 48-hour hearing will collapse." Some will try, and the burden of proof, the written finding and the 90-day clock are the discipline. A prosecutor who moves for detention on every case must prove danger on every case within 48 hours and try every one within 90 days, which no office can do, and which is the point.

"This is federal interference in state criminal procedure." It is a condition on federal money, as № 056 and № 058 are, and a state may decline it. The Constitution's guarantee of due process is not a state option, and the Table is a statement of what it requires before a citizen convicted of nothing is locked up for want of money.

Sources

  • Bureau of Justice Statistics, Jail Inmates in 2023, on the unconvicted share of the jail population
  • New Jersey Courts, Criminal Justice Reform annual reports; MDRC, Evaluation of Pretrial Justice System Reforms That Use the Public Safety Assessment (New Jersey); Drexel University, on gun violence after New Jersey's reform (2024)
  • Loyola University Chicago, Center for Criminal Justice, The First Year of the Pretrial Fairness Act (2024), jail population and detention findings
  • Dobbie, Goldin and Yang, "The Effects of Pretrial Detention on Conviction, Future Crime, and Employment," American Economic Review (2018)
  • Bail Reform Act of 1984, 18 U.S.C. § 3142
  • Proposals № 056 (The Clearance Rate); № 058 (The Use-of-Force Register); № 061 (The Sentencing Table); № 096 (The Public Defender Floor, forthcoming)