Restorative Justice Mechanisms: Restitution and Victim-Offender Mediation

Restitution is the oldest and most straightforward restorative mechanism: the offender pays the victim back for a quantifiable loss, whether that is stolen property, medical bills, lost wages or damage to belongings.

Legal theorist Randy Barnett argued in a 1977 article that restitution should be treated as a distinct paradigm of criminal justice in its own right, closer to a civil damages remedy than to punishment, because its purpose is to restore the victim's position rather than to inflict suffering on the offender.

Most systems still run restitution through the criminal court rather than a separate civil claim, because criminal conviction already establishes the facts and because a criminal order carries enforcement tools a private judgment often lacks. Restitution is therefore usually set within the ordinary sentencing procedure rather than a standalone hearing.

In the United States, the federal Mandatory Victims Restitution Act of 1996 requires courts to order full restitution for a defined set of offences, removing judicial discretion over whether to award it, though disputes over the amount remain common.

In India, Section 357 of the Code of Criminal Procedure, 1973 let a court direct that part of a fine be paid to the victim as compensation, and a 2009 amendment added Section 357A, which required every state to run a victim compensation scheme funded independently of the offender's ability to pay.

The Bharatiya Nagarik Suraksha Sanhita, 2023, which replaced the 1973 Code with effect from 1 July 2024, carries an equivalent compensation provision forward, so the underlying entitlement continues under the new code.

The practical limits of restitution are the same everywhere: it only reaches losses that can be priced, so it says little to a victim of an assault whose loss is not primarily financial, and it depends on an offender who has, or can be made to have, the means to pay. Courts routinely spread payment over instalments, and an offender's default can trigger further proceedings, which is why enforcement, not the order itself, is usually where restitution fails in practice.

England and Wales run a parallel model through the Compensation Order, available under the Sentencing Act 2020, which lets a criminal court order payment for personal injury, loss or damage without the victim filing a separate civil claim; the amount is capped by what the offender can reasonably pay within a set period, and default can lead to further enforcement action including, for adults, a fine or, in the last resort, imprisonment for wilful non-payment.

The pattern across these systems is the same: a criminal court can compensate a victim quickly and cheaply, but only up to the limit of the offender's means, which is why restitution rarely functions as full compensation for serious harm on its own.

Victim-offender mediation is a facilitated meeting in which the person harmed and the person responsible for the harm speak directly, with a trained mediator managing the process rather than deciding the outcome.

The first documented programme ran in Kitchener, Ontario in 1974, when a probation officer arranged for two young men convicted of vandalism to meet the people whose property they had damaged, an approach that was formalised soon after as the Victim Offender Reconciliation Program and later spread across North America and Europe under the broader label of victim-offender mediation.

A referral can come from a court before sentencing, from a prosecutor as a diversion from charging, or from a probation service after conviction. Before any meeting, the mediator screens both parties separately, checking that the offender accepts responsibility for the act, that the victim is participating by choice and not under pressure from the offender, family or the system, and that a face-to-face meeting will not expose the victim to further harm or intimidation.

Cases involving intimate-partner violence or a severe power imbalance are frequently screened out for this reason, or moved to indirect, shuttle-style contact instead of a direct meeting.

The session itself usually opens with the offender describing what happened, the victim describing the impact, and the mediator keeping the exchange respectful without scripting what either party says.

Where the parties reach agreement, the outcome is typically a written plan covering an apology, restitution, community service or a combination of these, which is then reported back to the referring court or agency. This is one of the few points in the process where victim participation in criminal proceedings goes beyond giving testimony to actively shaping the outcome.

Facilitator neutrality is the main safeguard against the process re-victimising the person it is meant to help; a mediator who steers the victim toward forgiveness, or who treats the meeting as complete once an agreement is signed without following up on compliance, undermines both the safety and the credibility of the process.

Programmes that track compliance after the session, rather than closing the file once an agreement is signed, report meaningfully higher completion rates than those that do not; a probation officer or coordinator who checks in on an apology, a restitution instalment or a community-service commitment catches a stalled agreement early enough to bring the case back to court, whereas a programme that treats the signed plan as the end point often only discovers non-compliance when the victim complains months later.

Several national bodies, including the UK's Restorative Justice Council, now accredit facilitators and audit case files against a published practice standard for this reason, since the quality of the safeguards described above depends entirely on how consistently an individual facilitator applies them.

Family group conferencing widens the circle from two individuals to their extended families. New Zealand built the model into law through the Children, Young Persons, and Their Families Act 1989, which made a family group conference the ordinary response to most youth offending rather than a court hearing.

The Act was drafted partly in response to Maori and Pacific communities' objections that a Western court process ignored the collective, family-based decision-making that their own communities already used, and a conference brings the young offender's family, the victim (or their representative), a youth justice coordinator and police together to agree a plan the family itself helps design.

Circle sentencing takes a related idea into adult sentencing and has a distinct origin in Canadian Indigenous justice practice. Yukon Territorial Court Judge Barry Stuart is credited with formalising the modern version in the case of R. v. Moses in 1992, where he invited the offender, victim, elders and other community members to sit in a circle and speak in turn before he passed sentence, drawing on First Nations traditions of collective conflict resolution that predate the criminal code.

The circle does not replace the judge's sentencing authority, but the judge weighs the circle's recommendation heavily and departs from it only for clear reasons.

Both models share a structural feature that plain mediation lacks: they recruit people who were not directly harmed, such as parents, grandparents, elders and neighbours, and give them a stake in whether the plan actually gets carried out. That wider group is often what makes a restorative outcome stick after the meeting ends, because compliance is monitored by people the offender sees every day rather than by a court that only checks in at a review hearing.

The trade-off is scale and consistency. A conference or circle needs available family and community members willing to participate, a coordinator skilled at managing many voices at once, and enough time to run properly, all of which are harder to guarantee in a busy urban court than in the smaller communities where both models were first developed.

Resourcing explains most of the difference in how far each model has spread. A family group conference needs one trained coordinator and a room, so New Zealand runs it as the default response to nearly all youth offending nationwide.

A sentencing circle needs elders or community members willing to give a half-day or more per case, which is workable in the smaller Yukon communities where Judge Stuart first used it but far harder to sustain in a high-volume urban court list, so circle sentencing has stayed a targeted option used selectively rather than a default procedure even in Canada.

The United Nations gave restorative justice its clearest international reference point through the Basic Principles on the Use of Restorative Justice Programmes in Criminal Matters, adopted in 2002.

The Basic Principles define a restorative process as any process in which the victim, the offender and, where appropriate, other affected individuals or community members participate actively in resolving matters arising from the offence, generally with a facilitator's help, and they set out safeguards that closely track what section 2 described: informed and voluntary consent from both parties, an offender's acknowledgement of the basic facts, confidentiality of what is discussed, and the option to withdraw and return to the ordinary criminal process at any point.

The document is explicitly non-binding, described as basic principles rather than a convention, and it leaves states free to decide where in the criminal process a restorative programme sits, whether as a diversion before charging, a parallel track during prosecution, or a factor considered at sentencing.

Its main effect has been to give national reformers a shared vocabulary and a checklist against which to design domestic schemes, rather than to impose a single model, and it complements the wider international frameworks for victim rights that developed around the same period.

Integration into ordinary procedure differs by jurisdiction. England and Wales use restorative justice largely as an adjunct to conventional sentencing or as part of youth cautions. New Zealand and, following its lead, several Australian states built conferencing directly into youth justice law rather than treating it as optional.

India has no dedicated restorative justice statute; existing tools such as plea bargaining, carried forward from Chapter XXI-A of the Code of Criminal Procedure, 1973 into Chapter XXIII of the Bharatiya Nagarik Suraksha Sanhita, 2023 with the same eligibility limits, and compounding of specified offences achieve some restorative effect, such as victim consent and negotiated compensation, without using restorative language.

This unevenness is deliberate rather than accidental: the Basic Principles were drafted to be adaptable to legal systems as different as common law and civil law jurisdictions, and the price of that flexibility is that whether a given country has any restorative mechanism at all, and how well safeguarded it is, depends entirely on domestic legislative choice.

The Basic Principles also address what happens when a restorative process fails: if the parties do not reach an agreement, or an agreement is not implemented, the case reverts to the ordinary criminal process without the failed attempt itself being held against the offender, and anything said during the restorative process is not to be used as an admission in later proceedings.

This confidentiality rule is what lets an offender speak candidly in mediation or a conference without it becoming evidence against them if the case ends up in court after all.

Evaluation research on restorative justice is more developed than for most sentencing innovations, and it converges on a consistent finding: victims who participate in mediation or conferencing report higher satisfaction with the process than victims routed through conventional prosecution, largely because they get to ask questions, describe the impact directly, and have some say in the outcome rather than being a witness to a process run entirely by others.

Effects on re-offending are more mixed and generally smaller than early advocates claimed. Several systematic reviews conducted since the early 2000s, most prominently the meta-analyses associated with criminologists Heather Strang and Lawrence Sherman, found modest reductions in repeat offending for programmes that met the direct-conference model, with the strongest effects for property and violent offences involving an identifiable victim, and negligible or inconsistent effects for other offence types.

Reviewers have been careful to note that many of the strongest studies involve self-selected volunteers on both sides, which limits how far the results generalise to offenders who would not agree to participate.

Restitution has a narrower but more measurable outcome record: compliance is the main variable, not satisfaction. Instalment plans set at a level tied to the offender's actual income, and monitored by a probation officer rather than left to the offender to self-report, close out at markedly higher rates than open-ended orders with no follow-up, which is consistent with the broader finding above that supervision after an agreement is signed drives most of the difference in whether a restorative disposition is actually carried out.

The standard critique is threefold. First, power imbalance: a victim who feels unable to say no, or an offender who agrees only to appear compliant to a court, undermines the voluntariness the entire process depends on, which is why the screening described in section 2 is treated as essential rather than optional.

Second, unsuitability for serious violence: most jurisdictions exclude or heavily restrict restorative mechanisms in cases of homicide, serious sexual assault and ongoing domestic abuse, on the grounds that direct contact risks re-traumatising the victim and that the harm is too severe for restitution or an apology to meaningfully address.

Third, net-widening: critics including sociologist Stanley Cohen warned as early as the 1980s that diversionary and community-based measures, restorative or otherwise, can expand rather than shrink state control, by drawing in minor cases that previously would have received no formal response at all.

None of these critiques argues for abandoning the mechanisms; they argue for using them selectively, with rigorous consent screening, and as a genuine alternative rather than an additional layer of supervision, which is the design challenge every jurisdiction adopting restorative justice at scale still has to solve.

Two further critiques attach specifically to the mechanisms rather than to restorative justice in the abstract. Restitution can fall unevenly on offenders with little income, turning a disposition meant to repair harm into a debt that a poorer defendant struggles to clear and a wealthier one pays off without consequence, which is one reason several jurisdictions cap orders against the offender's proven ability to pay rather than the full value of the loss.

Mediation, in turn, typically proceeds without either party having independent legal advice in the room, which speeds the process but means an offender can agree to restitution terms, or a victim can accept an apology in place of a harsher sentence, without anyone checking that the agreement is actually fair to them.