Calculating the savings from reforming nonviolent Three Strikes sentences

Estimating the financial effect of ending life sentences for nonviolent repeat offenders requires more than multiplying a prison population by an annual corrections figure. A credible estimate identifies who would be eligible for resentencing, how many years of imprisonment could be avoided, what services would replace custody, and when the government would actually receive the saving. The same framework can be used to examine California’s Three Strikes reform while presenting figures in a way that makes sense to an Australian audience.

The central question is fiscal, but the calculation also reflects a policy choice: retain severe penalties for people convicted of dangerous or violent crimes while reconsidering life terms for non-serious, nonviolent offences. A useful model should therefore separate prison savings from public-safety assumptions, distinguish immediate budget relief from long-term avoided costs, and show its workings clearly enough for journalists, advocates and voters to test.

Define the eligible population carefully

Begin with the number of people serving a life sentence under the relevant Three Strikes rule whose triggering offence meets the proposed definition of nonviolent and non-serious. This is the eligible population, not the entire prison population and not every person with a prior conviction. A defensible estimate should be based on sentencing records, offence codes, custody status and any exclusions written into the reform.

The initial count should then be adjusted for practical eligibility. Some prisoners may have disqualifying convictions, outstanding cases, immigration holds, medical restrictions or other legal barriers. Others may already be close to release. If 1,000 people appear eligible but 12 per cent are excluded after case review, the model should use 880 people rather than the headline figure.

A useful spreadsheet has one row per person or cohort and records age, current sentence, projected release date, offence category, expected resentencing date and likely supervision period. If individual records are unavailable, group people by remaining sentence: for example, zero to five years, six to ten years, and more than ten years. That is less precise, but still stronger than applying one average to everyone.

The reform background can help explain the policy rationale and the distinction between nonviolent cases and offenders considered dangerous. The savings model should preserve that distinction rather than treating every third-strike sentence as financially interchangeable.

Estimate the years of custody avoided

The most important quantity is the number of prison years that would no longer be served. For each eligible person, calculate the difference between the expected years in custody without reform and the expected years under the proposed resentencing pathway.

A simple formula is:

Avoided custody years = baseline prison years − revised prison years

Suppose a person has 18 years remaining under a life term and would receive a revised sentence resulting in four more years in custody. The gross reduction is 14 prison years. If 20 eligible people have the same profile, the cohort produces 280 avoided custody years. The real calculation should account for parole eligibility, credits, likely release dates and the time required for court applications.

Timing matters. A person released six months after a law changes creates a different first-year budget effect from someone released after a resentencing hearing two years later. Build a calendar that assigns each avoided year to a financial year. This is particularly important when lawmakers or campaigners cite a projected saving over a decade rather than a single annual figure.

A conservative model can include three scenarios. The low case may assume slower court processing, fewer eligible prisoners and later release. The central case can use the best available administrative data. The high case can assume quicker hearings and greater take-up, but it should not be presented as the expected result. Scenario analysis prevents a disputed assumption from being mistaken for a guaranteed outcome.

Use the full cost of imprisonment

The annual cost of keeping someone in prison should include the expenses that genuinely change when the population falls. Depending on the jurisdiction, these may include accommodation, food, health care, custody staff, transport, utilities, rehabilitation programmes, administration and facility maintenance. A published average prisoner cost is a starting point, not automatically the amount saved for every release.

Separate variable costs from fixed costs. Fewer prisoners may reduce food and health expenditure quickly, while a prison wing, lease, security contract or staffing establishment may remain in place for years. The near-term saving is therefore often lower than the average cost per prisoner. Over a longer period, closures, redeployment or avoided construction can create larger structural savings.

The core calculation is:

Gross prison saving = avoided custody years × annual avoidable cost per prisoner

For illustration, assume 280 avoided custody years and an avoidable annual cost of US$65,000. The gross saving would be US$18.2 million before replacement costs, inflation and timing adjustments. This is an example for demonstrating the method, not a claim about the actual California cost of incarceration.

Australian readers will recognise why location and accounting practice matter. A figure drawn from California should not be transferred directly into Australian dollars or compared casually with the cost of a prison bed in New South Wales, Victoria or Queensland. Convert the currency using a stated exchange-rate date, then identify whether the source uses an average cost, marginal cost or total system cost. A Melbourne-based analyst might present the result in both US dollars and Australian dollars, with the conversion clearly labelled rather than quietly changing the apparent size of the saving.

Subtract the costs of release and supervision

Ending a life sentence does not make every related expense disappear. People released under a revised sentence may require parole supervision, housing support, drug and alcohol treatment, mental-health care, employment assistance, transport or electronic monitoring. Some of these services are essential to a safe transition and should be included in the net-savings estimate.

Use a second formula:

Net public saving = gross prison saving − resentencing cost − release support − supervision cost − expected additional justice costs

Resentencing costs can include judicial hearings, legal representation, record review and victim notification. Release support may be a one-off expense, while parole supervision can continue for several years. If the gross saving is US$18.2 million and the combined costs of court processing, support and supervision total US$5.4 million, the net public saving is US$12.8 million before considering any change in reoffending.

The final term requires care. A model should not assume that release causes crime, nor should it assume that every released person will remain offence-free. Instead, compare the expected justice-system cost under the reform with the cost under continued imprisonment. That may include police, courts, legal aid, custody after a new conviction and victim-support spending. Use evidence from comparable resentencing or parole programmes where available, and publish the uncertainty around the estimate.

Australian service prices can make the replacement-cost question concrete. A person supervised in Sydney may face different housing and transport expenses from someone in regional New South Wales; a participant in a Melbourne treatment programme may encounter a different provider market from one in Perth. GST treatment, state funding arrangements and Commonwealth programmes can also affect the amount that belongs in a government budget. These details do not change the California legal question, but they help Australian readers understand why “cost per released prisoner” is never a universal number.

Discount future savings and test assumptions

Savings received ten years from now are worth less than savings received this year. Apply a discount rate to future cash flows, state the rate, and show both nominal and present-value results. The calculation is:

Present value = future saving ÷ (1 + discount rate) raised to the number of years

If a saving of US$10 million arrives in year five and the chosen annual discount rate is 3 per cent, its present value is approximately US$8.63 million. A public-sector evaluation may use a prescribed rate, while a campaign estimate may show a range such as 2 to 5 per cent. The important point is transparency.

Inflation also needs separate treatment. A model using constant 2026 dollars should say so and use real costs. A nominal model should increase prison, parole and service costs consistently. Mixing today’s prisoner cost with future wages and rent produces an attractive-looking number that cannot be audited.

Test the variables that have the greatest effect on the result. These usually include the eligible population, average years avoided, marginal prison cost, release timing, supervision expense and any assumed reoffending rate. A sensitivity table can show the net saving if each variable moves by 10 or 20 per cent. For example, if the estimate remains positive even when release support costs rise sharply and avoided years fall, the result is more robust than a single optimistic forecast.

Local market comparisons should be handled with discipline. A corrections budget in Canberra, Brisbane or Adelaide may use different labour, property and health-service assumptions from one in California. Use Australian figures only as context or for a separate domestic scenario, not as evidence that California would save the same amount. State and territory budget papers, auditor-general reports and corrections annual reports are more reliable than a newspaper estimate with no cost definition.

Present a result that can be checked

A strong public estimate reports the population, avoided years, annual cost, release costs, discount rate, time horizon and sensitivity range in plain language. It should state whether the result is gross or net, whether it measures government expenditure or wider social benefit, and whether prison closure savings are included. These distinctions prevent a large theoretical saving from being advertised as money immediately available for another programme.

It is also helpful to separate three outcomes. The first is a cash-flow estimate showing what a government might spend or save each year. The second is a present-value estimate over a chosen period, such as ten or twenty years. The third is a broader economic estimate that may include employment, health, family and community effects. The third category can be relevant, but it should not be merged with the corrections budget.

A worked example might read as follows: 600 eligible prisoners, an average of 11 custody years avoided, US$65,000 in annual avoidable prison cost, and US$12,000 per person in release and supervision expenses. Gross avoided custody would equal 6,600 prisoner-years, or US$429 million before timing. Release-related costs would total US$7.2 million if paid once per person, with additional annual supervision costs added separately. A ten-year discounted cash-flow schedule would then determine how much of the gross figure arrives within the period and what the net present value is.

The policy context should accompany the arithmetic. The purpose of reform is not simply to reduce a line in a corrections budget; it is to reserve the harshest penalties for serious threats while providing a lawful route to relief for qualifying nonviolent strikers. Information about the change in treatment of simple drug possession as a third strike is available in this sentencing explainer, which can be read alongside the financial assumptions rather than substituted for them.

For an Australian audience, the clearest presentation might show the original US-dollar calculation, an Australian-dollar conversion dated to the day of publication, and a note that California’s law, courts and corrections system are distinct from those in Australia. That format respects local readers who think in AUD while avoiding a misleading comparison with state prison budgets in Sydney, Melbourne or elsewhere.

The final worksheet should contain separate lines for eligible people, avoided custody years, variable prison cost, fixed-cost exclusions, resentencing, support, supervision, reoffending risk, inflation, discounting and timing. Enter the best-supported figures first, then produce low, central and high cases and publish the assumptions beside each result. Start by obtaining the relevant sentencing and corrections datasets and building that cohort-by-year worksheet.