COVID-19 amplified racial disparities in the US criminal legal system

The study investigates how the unprecedented pandemic-era reduction in the US prison population altered its racial composition. Despite an overall decline in incarcerated people during 2020, the authors observed a sharp, temporary increase in the share of incarcerated Black (and Latino) individuals across nearly all states, reversing a decade-long trend of decreasing Black incarceration shares. The central research question asks which mechanisms—changes in admissions, releases, and/or sentence-length distributions—explain the observed spike in racial disparity during COVID-19. The work is motivated by long-standing racial inequities in the US carceral system and aims to provide data-driven insight to inform equitable policy responses during large-scale disruptions.

Prior research documents that the USA has the world’s highest incarceration rate with pronounced racial disparities. Before the pandemic, the proportion of incarcerated Black people had been declining for about a decade while the proportion of incarcerated white people had been rising, amid a broader shift toward class-driven disparities (e.g., increasing admissions among low-educated white individuals). Disparities in admissions were smaller than disparities observed in the standing prison population: Black–white disparity ratios are closer to 2:1 in admissions versus approximately 6:1 in the total incarcerated population. Reports from the Bureau of Justice Statistics (e.g., Prisoners in 2019 and 2020) and recent syntheses on racial inequality in crime and justice provide context on these trends. The pandemic’s disruptions to courts, jail-to-prison transfers, and release policies, alongside documented racial differences in sentence lengths, create a theoretical basis for expecting an increase in racial disproportionality when admissions slow substantially.

The authors assembled a new, public dataset of prison demographics from Departments of Corrections across all 50 states and the District of Columbia, spanning more than 20 years and over 9,000 records. Data collection involved manual retrieval from public reporting portals and numerous Freedom of Information Act requests to obtain state-level information on police encounters, court proceedings, incarcerated populations, and in some cases monthly admissions and releases by race. Demographic time series were analyzed at national and state levels to quantify changes before, during, and after the onset of COVID-19. For 18 states with sufficient granularity, the authors compared monthly admissions and releases by race (normalized to pre-pandemic baselines and slopes) to assess whether changes in flows could explain observed stock (population) shifts. They also examined specific case studies: Florida court statistics (showing reduced trials and dispositions, increased dismissals, and a higher share of white defendants among dismissals after March 2020) and Arkansas executive-order-based early release eligibility (2,143 individuals; evidence of disproportionate white releases given eligibility criteria and sentence classifications). The primary explanatory mechanism tested was differential sentence lengths by race: when admissions fall sharply, longer average sentences for Black (and in some states Latino) individuals imply that the remaining prison population will become more racially disproportionate even if releases proceed. State comparisons (e.g., Illinois vs. Texas) and exceptions (e.g., Maine, Maryland, Missouri, Oregon, Wyoming) were used to probe heterogeneity consistent with sentence-length distributions and the prevalence of shorter-term sentences. Visual analyses included time series of total incarcerated counts, racial composition percentages, and normalized admissions/releases; supplementary figures provided state-level trajectories and additional robustness checks.

• The US state prison population declined by at least 17% between March 2020 and July 2021, from approximately 1.23 million to 1.02 million. Every state experienced a decline, with reductions ranging from 5.8% (Nebraska) to 37.2% (New Jersey); the median state’s population fell by 18% from pre-pandemic levels.
• Despite overall declines, the share of incarcerated Black people increased sharply in 2020 across nearly all states and at the national level; Latino representation also increased. In most states, the spike was temporary and trended back toward pre-pandemic levels by 2021.
• Monthly releases fell during much of the pandemic: between February and May 2021, releases were roughly 70% of pre-pandemic levels. If admissions had remained constant, reduced releases would have increased total population, which did not occur—indicating releases were not the driver of the overall population decline.
• Admissions dropped substantially due to widespread court closures (all states except Nebraska), disrupted trials, and reduced jail-to-prison transfers starting around April 2020.
• Racial disparities in admissions and releases alone do not explain the national increase in Black and Latino population shares: in 18-state analyses, post-onset admissions had a larger white proportion, and releases were relatively commensurate, which would predict the opposite of the observed stock effect.
• The most consistent mechanism is racial disparity in sentence lengths: with sustained reductions in new admissions, groups serving longer average sentences (often Black, and in some states Latino) remain overrepresented, driving the spike in their population share.
• Case studies: Florida circuit criminal courts showed sharp declines in trials and dispositions and increased dismissals post-March 2020; among dismissals, the share of white defendants increased. Arkansas executive orders made 2,143 individuals eligible for early release; the overlap between eligibility criteria and sentence classifications resulted in disproportionate releases of white individuals.
• Five states (Maine, Maryland, Missouri, Oregon, Wyoming) did not clearly show the national pattern; these systems tend to have relatively fewer Black incarcerated individuals or a higher prevalence of shorter-term sentences (<2 years), attenuating the sentence-length mechanism’s effect on composition.
• The dataset is publicly released on Zenodo to support further research.

The findings indicate that the pandemic functioned as a stress test for the US carceral system, revealing how structural features—especially sentencing disparities by race—shape population composition under disrupted intake conditions. When court closures and reduced jail transfers sharply lowered admissions, the standing stock of incarcerated individuals increasingly reflected racial disparities in sentence lengths, producing an abrupt rise in the share of Black and, in some states, Latino incarcerated people. While certain state-specific changes in admissions or targeted releases (some of which were themselves racialized) contributed locally, national patterns align most strongly with the sentence-length mechanism. These results address the research question by distinguishing mechanisms that affect flows (admissions, releases) from stock dynamics (population composition), demonstrating that even temporary disruptions can rapidly amplify existing inequities. The work underscores the ethical and policy implications: reducing disparities requires addressing sentencing inequities, ensuring equitable admissions and release practices, and designing emergency procedures (e.g., during pandemics) that avoid exacerbating racial disproportionality.

This study documents the largest, fastest single-year decline in US state prison populations during the COVID-19 pandemic and shows that, paradoxically, racial disparities in prison composition intensified, with increased shares of Black and Latino incarcerated individuals. The evidence points to racial differences in sentence lengths as the primary mechanism behind the spike in disparity under reduced admissions, with localized contributions from court disruptions and targeted, sometimes racialized, release policies. The authors contribute a new multi-decade, multi-state dataset (publicly available on Zenodo) and a framework for interpreting stock–flow dynamics in incarceration during large-scale disruptions. Future research should: (1) refine sentence-length analyses across offense types and jurisdictions; (2) integrate higher-resolution admissions, pre-trial, and jail transfer data; (3) examine causal pathways linking court operations to racialized outcomes; (4) evaluate decarceration and emergency release policies for equity impacts; and (5) explore reforms to sentencing and court practices that mitigate disparities during routine and crisis conditions.

• Data heterogeneity and reporting gaps: States differ in definitions, race categories (e.g., Michigan reports only “white” vs. “nonwhite”), and reporting frequency/quality; monthly admissions/releases by race were available for only 18 states.
• Observational design: The study identifies plausible mechanisms but cannot establish causal effects of court closures, admissions changes, or releases on racial composition.
• Measurement constraints: Poor standards for jail-to-prison transfer reporting and incomplete court processing data limit precise attribution of mechanisms, especially for pre-trial and plea dynamics.
• Temporal scope: The spike in disparities was often temporary and began to reverse by 2021; longer-term post-pandemic trajectories may differ.
• Generalizability across systems: Federal, local jail, and unique state systems (including those with many short-term sentences or small Black incarcerated populations) may exhibit different dynamics.
• Policy heterogeneity: Executive orders and eligibility criteria for releases varied by state and may confound cross-state comparisons of release impacts.