Get Informed

Subscribe to our newsletters for regular updates, analysis and context straight to your email.

Close Newsletter Signup

The Incalculable Costs of Mass Incarceration

Prisons carry enormous, perhaps impossible to measure social costs—but when assessing the system fiscally, reformers should focus on staffing salaries instead of the number of incarcerated people.

Rafael Belincanta/EyeEm

The Incalculable Costs of Mass Incarceration

Prisons carry enormous, perhaps impossible to measure social costs—but when assessing the system fiscally, reformers should focus on staffing salaries instead of the number of incarcerated people.


Every year states spend about $50 billion to lock up over 1.3 million people, or about $35,000 per prisoner per year. Although individual state averages obviously vary, statistics like these suggest that even small cuts in prison populations could yield significant fiscal returns, and big cuts something massive. The Brennan Center, for example, recently argued that releasing 576,000 low-risk inmates could save $20 billion per year (which is just $35,000 times 576,000—a calculation others make as well).

But this is the wrong way to think about prisoners and costs. Measuring costs this way both significantly overstates what we fiscally save with each person we divert from prison while simultaneously understating the social costs that such a diversion avoids. Fiscal savings don’t come from reducing inmate populations—they come from reducing staffing. And the social costs of prisons and jails have little to do with budgets and far more to do with the physical, emotional, mental, and other harms incarceration imposes on inmates, their families, and their communities.

Whether we are trying to understand how decarceration frees up funds to be spent elsewhere, or whether prison is socially cost-benefit justifiable, using the average cost per prisoner—a common metric—is simply mistaken.

Salaries, not food

No state saves $35,000 when it cuts its prison population by one person because a significant part of that $35,000 is incurred regardless of whether that one person is behind bars. The person’s departure does not lead to a reduction in the heating or water bill or—most important—the number of correctional officers on staff. Wages and benefits consume about two-thirds of the $50 billion we spend on prisons, if not more.

So, the incremental, marginal savings from reducing the prison population by one inmate is much less than the average cost suggests.  According to a recent review by the Urban Institute these marginal savings might be only 12 percent of the average cost of a prisoner, or closer to $4,200. My own effort to determine what state departments of correction reported found comparable estimates.

Of course, states do not release just one prisoner at a time. And certainly if the number of prisoners released crosses a critical threshold, the marginal savings might jump as well. A state could release enough prisoners for it to close a wing of a prison or consolidate facilities, which would save on heating and water costs—and might allow for a reduction in staffing. And if the state can cut staffing, it can free up a lot more money.

But that “if” depends on cutting the prison wage bill, and here the story gets troubling. There are plenty of examples of states that close prisons without comparable reductions in staffing. Pennsylvania once closed two prisons and laid off three guards. When Michigan closed its Pugsley Correctional Facility in 2016, it was reported that it laid off 16 officers and 35 staff members, but all 51 either declined new positions or said they did not want to work elsewhere; in effect no one was involuntarily laid off.

Correctional officer unions are among the most most powerful (and least discussed) political entities in the criminal justice system. They work hard to oppose job loss, even as prison populations decline.

Digging deeper into the data, however, an even more intriguing picture emerges. Since 2010, the number of people employed nationwide as correctional officers has actually fallen in step with declining prison populations, to the point that the nationwide ratio of guards to prisoners has held fairly steady. (The data on staffing are only available at the national level, so state-specific ratios most likely show more variation.) Yet as staffing numbers have declined, total spending on wages and salaries has risen by 16 percent, and most likely more because the Bureau of Labor Statistics data does not account for overtime.

But the failure to pare back total spending on correctional officers might be a good thing, especially in the short run. While prison reform is often framed as a way to free up funds to spend elsewhere, we also need to make sure that prisons remain safe places for those we continue to lock up. If we cut spending on salaries and wages too quickly, states will find it hard to staff their prisons. South Carolina, for example, often highlights its savings through criminal justice reforms. But the state has over 600 vacant guard positions and that lack of manpower precipitated a riot in April that left seven inmates dead when the few guards on duty were unable to quickly re-establish control. Last week, a brutal murder at the Columbia Correctional facility in Florida revealed seemingly enormous problems with understaffing. 

In other words, just because the bulk of prison spending is on wages does not necessarily mean that we should aggressively try to cut payroll. But it does mean that efforts to fund other programs from reductions in prison spending will fail if we don’t. This is not an abstract concern. An effort in Indiana to divert state prisoners to local alternatives apparently ran into trouble in no small part because of confusion over average versus marginal savings.

When talking about the fiscal cost of prisons, we frame it inaccurately as cost per prisoner. It’s really more cost per staff member. Putting it that way not only emphasizes where cost savings come from more accurately, but help center correctional officers as among the largest stakeholders in the system—and thus highlights their significant incentive to fight against reform.

The real costs of prisons

The other problem with focusing on $35,000 per prisoner, or $50 billion per year, as the “cost” of prison is that it does not actually measure the real costs of incarceration, which are often in social harms. These costs are borne by prisoners and their families that provide no benefit to others. People are physically and sexually assaulted in prison. Mental health issues arise or worsen in prison. Prison is a vector of illness and STDs, which not only harms the inmates directly, but imposes avoidable costs on our health care system. The risk of death from a drug overdose rises sharply upon release. Prison exposure leads to elevated unemployment rates (and thus surely greater demands on governmental support programs), and in the short to medium term reduces life expectancy as well. Family members incur huge costs to visit people locked up in distant institutions, or suffer from not being able to see them at all (like the nearly 2,000 prisoners from Hawaii who are sent to Arizona to alleviate overcrowding in the islands’ prisons or the nearly 4,500 prisoners from Washington, D.C. who are turned over to the federal Bureau of Prisons and housed all over the country). Collect calls are expensive, as is supplying loved ones with money for commissary supplies. Children grow up apart from their parents, partners from each other. In some places, so many men are in prison that it alters the nature of dating and relationships for the entire community.

We have no idea how large these costs are, but they are surely staggering. Since 1978, there have been nearly 13.5 million new admissions to prison. Not all of these are unique people—but most of them are. Contrary to popular wisdom, it appears that about two-thirds of all people admitted to prison do not return, which means those 13.5 million admissions represent something on the order of 9 million to 10 million unique individuals. And the costs experienced by those millions in turn hurt millions more family members and friends, and thousands of communities.

Of course, those same family members and friends are also disproportionately the victims of crime—both crime and punishment are geographically densely concentrated. So any discussion of costs of prison has to account for any of its putative benefits as well (which, to be clear, are low), since those costs and benefits are felt by roughly the same communities. But most cost-benefit analyses appear to compare the marginal reduction in crime to average cost of incarceration (that $35,000 number again). Putting aside the average-marginal problem, the average social cost of locking someone up most likely dwarfs the average fiscal cost, and likewise for the marginal costs. So we are grossly understating what locking a person up costs us, by focusing on something that mostly measures the wages earned by guards, not the far greater, and far more unambiguous social costs borne by prisoners, their families, and society.

We think about the costs of incarceration poorly. By looking at the average fiscal cost, not the marginal, we convince ourselves that decarceration will produce more immediate savings than it will. At the same time, by ignoring the social costs of prison, we make it seem far cheaper than it really is.