How Earned Time Credits Work in Federal Prison, and Who Is Left Out

A bill's name is a press release. A bill's cross-references are the law. If you want to know what Congress did, skip the title, skip the signing ceremony, skip the op-eds, and go find the subsection that says "except for a prisoner serving a sentence for a conviction under any of the following provisions," followed by a list long enough to need its own zip code.

That subsection sits at the center of the First Step Act of 2018, the most significant federal sentencing and prison law in a generation, signed on December 21, 2018. The act is routinely described in slogans: "criminal justice reform," "second chances," "soft on crime." The slogans are all wrong in the same way. They describe a mood. The act is a machine, and it has parts. Let us take the machine apart and see what it does, who it helps and who it deliberately leaves out.

What an earned time credit is

A federal prisoner serves most of his sentence. There is essentially no federal parole for newer offenses, a point that surprises many Americans: the Sentencing Reform Act of 1984 set in motion the end of the old Parole Commission system and replaced it with fixed terms followed by supervised release. A federal prisoner can shorten his term in one traditional way, called good conduct time. Before 2018 the Bureau of Prisons applied a formula that gave up to 47 days off per year of the sentence imposed. The First Step Act fixed the arithmetic so that it counts up to 54 days per year, which was what Congress had intended in the first place. That change is separate from what this post is about, though people confuse the two.

Earned time credits are the new thing. They are a reward for participating in programs, not for staying out of trouble. Under the act, which added a new chapter to Title 18 of the United States Code (Chapter 229, beginning at section 3631), the Bureau of Prisons must offer "evidence-based recidivism reduction programs" and "productive activities," and a prisoner who successfully participates earns credit toward an earlier transfer out of a prison cell.

Define the pieces:

• Evidence-based recidivism reduction program (EBRR): a group or individual activity that has been shown to reduce recidivism and is designed to address a prisoner's assessed needs. Think cognitive behavioral therapy, vocational training, substance abuse treatment, education.
• Productive activity: a group or individual activity that is designed to prepare a prisoner for success upon release and that is not an EBRR program. Institutional work assignments can count.
• Risk level: a classification, minimum, low, medium or high, that the Bureau assigns using a risk and needs assessment tool.

The assessment tool: PATTERN

The statute required the Attorney General to develop a "risk and needs assessment system." The Justice Department's answer was a tool named PATTERN, an acronym that stands for Prisoner Assessment Tool Targeting Estimated Risk and Needs. A government acronym that spells a real word is a rare event, and I suppose we should congratulate whoever worked backward from the word.

PATTERN is a scored instrument that uses factors such as age, criminal history, disciplinary record inside, program participation and, for certain items, the nature of the offense, to put each prisoner in a risk category for general recidivism and for violent recidivism. The Bureau reassesses prisoners over time. That matters, because the statute lets the risk category change the credits a prisoner earns, and because the whole system is only as good as the instrument that sorts people.

The tool has drawn criticism from both directions. Some reformers argue that factors like criminal history can embed past inequities. Some skeptics on the right argue that any predictive tool is a bet, that the bet can be wrong in the wrong direction, and that a man misclassified as low risk who then commits a violent crime has made the program look reckless. Both complaints are fair questions to put to any algorithm that decides someone's freedom. The relevant point for citizens is that the sorting step is statutory, not optional: the credits cannot be earned without it.

The arithmetic of the credits

Here is the rule, stated exactly as the statute states the numbers.

1. A prisoner who is eligible and who successfully participates in EBRR programs or productive activities earns 10 days of time credit for every 30 days of successful participation.
2. A prisoner who has been assessed as minimum or low risk of recidivism, and who has kept that level in two consecutive assessments, earns an additional 5 days for every 30 days, for a total of 15 days per 30.

Make that concrete with a worked example, labeled as hypothetical. Picture a prisoner who is eligible, participates for a full year of programming and is not low risk. Twelve months of 30 days each is roughly twelve periods. At 10 days per period, that is 120 days of credit. A similar prisoner who is low risk for two consecutive assessments earns 15 days per period, so 180 days.

Now, a crucial point that the slogans leave out. These credits do not simply subtract from the sentence the way good conduct time does. They are applied to move the prisoner out of a prison cell earlier, into one of two settings: prerelease custody (halfway house or home confinement) or a longer stretch of supervised release. The statute caps the supervised release piece at 12 months of credit. In practice, a prisoner who has banked 365 days of credits could finish the last year of his term in a halfway house or at home under supervision, or have up to a year taken off the front end of his supervised release term, according to the statute's terms for which prisoners qualify.

The mechanics matter because they change the argument. A conservative who hears "early release" pictures a door opening. The act's actual design is closer to a step-down: more supervised, structured time in the community, under the Bureau's control and with conditions, before the sentence formally ends. You can still say it is too generous. But you cannot say it is a pardon.

Who is left out

Now the line that deserves a zip code. The act bars a prisoner from earning credits if he is serving a sentence for any of a long list of offenses, set out in 18 U.S.C. 3632(d)(4)(D), which runs to dozens of subparagraphs. The list is the political price of the bill, and it is where the real choices were made. Read in outline, it covers:

• Violent crime: among them murder, many categories of assault and kidnapping, certain robbery and carjacking offenses, and a range of firearms offenses tied to violence.
• Sex offenses: including sexual abuse and exploitation of children and a range of related crimes.
• Terrorism and national security offenses: such as terrorism, espionage and related crimes.
• Certain drug offenses, including offenses involving fentanyl above specified quantities, along with other listed trafficking provisions.
• Offenses against the justice system and certain other federal crimes, each specifically listed by statute number.

The exact list is long, and it has been amended since. You should read it in the United States Code itself, on the House Office of the Law Revision Counsel's website, rather than trust any summary, including this one. The principle is easy to state, though: the people the public fears most are the people the credits exclude.

There is a second group excluded for a different reason. A prisoner who is the subject of a final order of removal under immigration law cannot apply earned credits. The theory is that a person who will be deported on release has no community to be stepped down into. Reasonable people can argue about that provision, and it is worth knowing it is there.

A third wrinkle is subtler. Even a prisoner convicted of an excluded offense can still participate in programming. He just does not earn the time credit for it. That means the Bureau still gets the rehabilitative value of the classroom or the workshop, and a Congress worried about releasing violent offenders still gets its guardrail.

The other half of the act: sentencing changes

The credits get the attention, but the First Step Act also changed sentencing law at the front end, and those changes are the ones that most divided conservatives. There are three.

The safety valve. For many federal drug offenses a mandatory minimum applies. A "safety valve," in 18 U.S.C. 3553(f), lets a judge sentence below the minimum for a defendant who meets a list of conditions. Before 2018, the criminal history condition was strict: a defendant with more than one criminal history point was shut out. The act loosened it, so that a defendant with a modestly longer record, though not one with certain serious prior offenses, can qualify. The judge still decides whether to use the valve, and the defendant must still be truthful with the government. This expanded the pool of people who can be sentenced below the mandatory minimum, and it returned some discretion to judges.

The end of "stacking." Section 924(c) of Title 18 punishes using or carrying a firearm during a violent crime or drug trafficking crime. A first offense carries a mandatory minimum of five years or more depending on conduct. A "second or subsequent conviction" carried a 25-year mandatory minimum, to be served consecutively. Under a 1993 Supreme Court decision, Deal v. United States, a "second" conviction could be one charged in the same indictment as the first. That produced the notorious result of a defendant with several counts in one case being sentenced to a first count of five years and a string of 25-year add-ons for each further count. The First Step Act amended the statute so that the 25-year enhancement applies only where the earlier 924(c) conviction has become final. In other words, a prior conviction must be a prior conviction. This is, I think, one of the most plainly sensible changes in the bill. The text of a statute that says "second" should mean second.

Reductions in some drug enhancements. The act reduced certain mandatory enhancements for repeat drug offenders. Notably, it cut the 20-year enhancement for certain prior drug felonies to 15 years, and the mandatory life enhancement to 25 years, and replaced the term "felony drug offense" with "serious drug felony" and "serious violent felony" with defined limits. Congress also made the Fair Sentencing Act of 2010 retroactive, so that people sentenced under the old crack cocaine thresholds could ask a court for a reduction. We pick up that history in Mandatory Minimums Divide Conservatives, and Both Sides Have a Point.

Note that most of these sentencing changes were not made retroactive, with the crack provision as the major exception. The stacking reform, for example, applies to defendants not yet sentenced at enactment. It does not reopen old cases automatically.

The steelman and the critics on the right

State the skeptic's case fairly. It runs like this.

A conservative critic of the First Step Act says: our first duty in criminal law is to protect the public, and this act shortens the time that serious offenders spend behind a wall. Mandatory minimums were passed in response to real crime waves and real abuses of judicial discretion. Risk tools are guesses. Early transfer to home confinement depends on a Bureau that has a poor record of managing its own programs, and the first victim of a failure will not be a Bureau official but a citizen. Some Republican senators opposed the bill for exactly these reasons, Senator Tom Cotton prominent among them, arguing that it would release dangerous offenders and weaken tools prosecutors use to dismantle criminal organizations. Prosecutors, who rely on mandatory minimums as leverage to win cooperation, were among the loudest worriers.

That is a serious argument, made by serious people who care about victims. It deserves a serious answer.

1. The excluded list is the answer to the first worry. The credits do not apply to violent crime, sex offenses or terrorism. The prisoner who earns them is, by definition, serving time for something that Congress did not put on the list.
2. Most federal prisoners come home regardless. The question is not release versus no release. It is whether the last year is spent under structure and supervision, with a job or a program behind it, or on a bus with a check and a bad idea. The recidivism numbers in What the Recidivism Numbers Really Say About Second Chances show how front-loaded the risk is after release, which is an argument for supervised step-down rather than against it.
3. The credits can be taken away. A prisoner who stops participating stops earning, and one who commits serious infractions can lose credits the Bureau has already awarded. The incentive runs in the direction conservatives usually like: reward conduct, not time served.
4. Mandatory minimums are the legislature's choice, and so is changing them. A republic does not outsource its punishments to prosecutors. If Congress set a 25-year term that sweeps wider than anyone intended, Congress can narrow it. That is not weakness; it is the job. The Constitution gives Congress the power to define crimes and fix punishments, and it can take back some of what it delegated to charging decisions.

The honest verdict is that the critics were right about one thing: the act's success depends on execution. A good statute run by an agency that cannot staff programs, assess prisoners on schedule or keep track of credits is a statute that fails in the dull way that government programs fail. Congress writes the law, and an agency decides whether it works.

How the act passed, and what it tells conservatives

The Senate passed the bill on December 18, 2018, by a vote of 87 to 12, and the House followed days later by a wide margin. President Trump signed it on December 21. A bipartisan vote that large, on a bill that touched mandatory minimums, tells you the coalition was unusual: religious conservatives and fiscal conservatives joined civil libertarians and reform groups on the left. Some on the right came for the fiscal reason that prison is expensive, and some for the moral reason that every man is more than his sentence. The act's drafters courted both and made their offer concrete with that list of exclusions.

The coalition is a useful lesson for citizens. The act got through because its authors were specific. They did not promise "reform." They promised credits at stated rates, a risk tool, an excluded list and a few narrow fixes to sentencing, and every one of those could be read, argued over and voted on.

What to watch

Here is a short checklist for a citizen who wants to follow the law as it works rather than as it is advertised.

• Read the list. The offense exclusions in 18 U.S.C. 3632(d)(4)(D) tell you more than any press release. If the list is revised, someone has made a decision worth noting.
• Watch the tool. The Justice Department's risk instrument is reviewed and updated. Ask how often, by whom, and with what evidence on accuracy.
• Watch the capacity. Credits mean little if programs are full. Appropriations for programming, staff and halfway house beds decide whether the machine runs.
• Watch the victims. A good reentry system tells victims what is happening and respects their right to be informed. A related guide is What to Do if You Are a Crime Victim, Your Rights Step by Step.
• Watch for an honest evaluation. The act required reporting. A law that measures itself is rare; hold it to that.

The case in brief

1. The First Step Act of 2018 created earned time credits: 10 days per 30 days of successful programming, plus 5 more for prisoners who hold a minimum or low risk level, applied to earlier prerelease custody or supervised release.
2. A risk tool named PATTERN sorts prisoners, and a long statutory list of violent, sexual, terrorism and certain drug offenses shuts the people the public fears most out of the credits.
3. The act also widened the sentencing safety valve and ended the "stacking" of 25-year firearm enhancements for counts charged in the same case, a text-based fix that makes "second" mean second.
4. The conservative critics were right that execution decides everything, and wrong that the act is a blanket release: it is a conditional, supervised step-down with a list of exclusions.
5. A law like this lives or dies on staffing, honest assessment and public reporting, which are exactly the things citizens can ask about.

What you can do: read 18 U.S.C. 3632(d)(4)(D) on the Law Revision Counsel's website and then ask your member of Congress how the Bureau of Prisons is reporting its progress on credits and program capacity.