The Crime Paradox: America's Rates Are Plummeting, But the Debate Is More Heated Than Ever

In 2025, homicides across 35 major American cities fell 21% from the prior year, reaching a rate 25% below the pre-pandemic 2019 baseline [1]. The FBI reported that the national violent crime rate hit 359.1 per 100,000 in 2024 — the lowest since 1969 [2]. Yet in the same period, the National Crime Victimization Survey recorded sharp increases in reported victimization, retailers cited theft as a reason for closing stores, and car insurance premiums spiked 26% in a single year [3][4]. Gallup polling found that nearly half of Americans still believed crime was getting worse as recently as 2024 [5].

These contradictions are not an accident. They reflect a genuine fracture in how crime is measured, experienced, and discussed in the United States — a fracture that has become one of the most politically charged disputes in American public life.

The FBI Numbers: A Historic Decline

The FBI's Uniform Crime Reporting (UCR) program, which collects data from more than 16,000 law enforcement agencies covering 95.6% of the U.S. population, paints an unambiguous picture of recent decline [2]. In 2024, violent crime fell an estimated 4.5% from the prior year. Murder and non-negligent manslaughter dropped 14.9%, with an estimated 16,935 homicides nationally — a rate of 5.0 per 100,000 residents [2]. Property crime fell 9.0%, with motor vehicle theft declining 19.5% [2].

These declines followed a 13% drop in murder in 2023, which crime analyst Jeff Asher called "likely the largest one-year decline ever recorded" [6]. The trajectory has continued. The Council on Criminal Justice (CCJ), which tracks crime across 40 large U.S. cities, reported that the 2025 homicide rate in its study cities was 10.4 per 100,000 — 44% below the 2021 peak of 18.6 per 100,000 [1]. Thirty-one of 35 cities in the study experienced homicide decreases, with Denver, Washington D.C., and Omaha each posting declines exceeding 40% [1].

Source: FBI Uniform Crime Reporting Program

Data as of Mar 21, 2026CSV

To put these numbers in historical context: the U.S. violent crime rate peaked at roughly 747 per 100,000 in 1993 [7]. It fell steadily to a modern low of about 361 per 100,000 in 2014, rose modestly through 2016, held roughly stable through 2019, then spiked sharply in 2020 [7]. The 2024 rate of 359.1 per 100,000 is now at or slightly below the 2014 trough [2]. If nationwide data for 2025 confirms the city-level trends, the national homicide rate may approach 4.0 per 100,000 — which would be the lowest ever recorded in law enforcement or public health data going back to 1900 [1].

The NCVS Problem: A Completely Different Story

The Bureau of Justice Statistics' National Crime Victimization Survey (NCVS), which interviews approximately 240,000 people in 150,000 households about crimes they have experienced — whether or not those crimes were reported to police — has told a starkly different story [8].

While the FBI's UCR reported a 2% drop in violent crime from 2021 to 2022, the NCVS showed violent victimization rising 75% across the same period [9]. This is not a minor discrepancy. It is a chasm, and it has fueled legitimate skepticism about official crime statistics from observers across the political spectrum.

Several factors contribute to this divergence.

The NIBRS transition. In 2021, the FBI attempted to move all law enforcement agencies from the legacy Summary Reporting System (SRS) to the National Incident-Based Reporting System (NIBRS), a more detailed format. The transition was disastrous in terms of participation. In 2021, nearly 40% of agencies failed to report data. In 2022, 31% of the nation's roughly 18,000 law enforcement agencies were still missing from the database [10]. The two largest police agencies in the country — the New York Police Department and the Los Angeles Police Department — were absent from the federal data during this period [10]. Less than 10% of agencies in Florida and Pennsylvania submitted data in 2022 [10].

Participation improved to 94.3% population coverage by 2023, with agencies reporting in either NIBRS or SRS format [10]. By 2024, coverage reached 95.6% [2]. But the data gaps in 2021 and 2022 mean that year-over-year comparisons involving those years are unreliable, a point Asher has emphasized repeatedly in his analysis of FBI estimates [11].

Declining reporting rates. The NCVS captures the "dark figure" of crime — offenses that victims experience but never report to police. More than two million violent crimes go unreported every year in the United States [8]. If reporting rates to police are declining — because of distrust, perceived futility, or other factors — then FBI data based on police reports will show crime falling even if victimization is stable or rising. The NCVS is specifically designed to capture this gap, which makes its divergence from FBI data impossible to dismiss.

Methodological changes. In 2024, BJS implemented a split-sample design, concurrently administering a legacy instrument and a redesigned survey instrument [8]. This complicates direct comparisons with prior years and will take time to sort out.

Conservative critics, including researchers at the Heritage Foundation and the Washington Examiner, have argued that the FBI data specifically is unreliable and that politicians use it selectively to claim crime is falling when Americans' lived experiences suggest otherwise [12]. This argument has force precisely because of the NIBRS data gaps and the NCVS divergence — neither of which can be hand-waved away.

Defenders of the FBI data counter that the 2023 and 2024 estimates, with vastly improved coverage, show consistent declines across nearly every category, and that independent city-level data — which does not rely on the FBI system — confirms these trends [6][1]. Asher has noted that while 2021 estimates were unreliable enough to fall outside the 95% confidence interval, the 2023 and 2024 figures were produced with coverage levels comparable to the pre-NIBRS era [11].

The honest answer is that both datasets have legitimate strengths and weaknesses, and anyone who cites one while ignoring the other is telling an incomplete story.

The 2020 Spike: Who Paid the Price

Before the current decline, there was the surge. In 2020, homicides rose approximately 30% nationally — the largest single-year increase since 1960 [13]. In a sample of 30 cities, homicide rates spiked 68% between April and July of that year [13]. This was not an abstraction. It translated to thousands of additional deaths concentrated in communities that were already the most vulnerable.

The demographic burden was staggeringly unequal. Black males experienced a homicide victimization rate of 51.5 per 100,000 — compared to 13.5 for Hispanic males and 2.9 for white males [14]. Young Black people aged 18–24 were 19 times more likely to die by firearm homicide than white people of the same age [15]. The Violence Policy Center reported that Black Americans, who constitute roughly 13.4% of the population, made up more than 53% of homicide victims in 2020 [16].

The Heritage Foundation's Rafael Mangual has argued that this demographic reality is the most important and underreported dimension of the crime debate — that progressive criminal justice policies disproportionately harm the Black communities they claim to protect by reducing deterrence and allowing repeat offenders to cycle through the system [17]. This argument cannot be evaluated in the abstract; it requires examining what the data shows about specific policy interventions, which is addressed below.

What is beyond dispute is that the 2020–2021 homicide spike produced roughly 5,000 to 6,000 additional homicide deaths per year above the 2019 baseline of approximately 16,400, and that these deaths fell overwhelmingly on young men of color in a handful of cities [13][14]. The subsequent decline — however welcome — does not erase that toll.

Cities That Recovered, Cities That Haven't

The CCJ's year-end 2025 data allows a city-by-city assessment of recovery from the pandemic-era spike [1].

Cities that have returned to or exceeded pre-pandemic levels: New York, Los Angeles, Chicago, Philadelphia, Oakland, Detroit, Birmingham, Richmond, and several others posted record or historic lows in 2025 [1]. Across the study cities, homicides were 25% below 2019 levels [1]. The highest-homicide cities experienced a 36% decline from their 2019 rates, the largest improvement of any group [1].

Cities with persistent problems: Only three cities in the CCJ sample saw homicides increase in 2025: Little Rock (up 16%), Fort Worth (up 2%), and Milwaukee (up 1%) [1]. Boston and El Paso also saw modest increases [1].

Source: Council on Criminal Justice

Data as of Mar 21, 2026CSV

Beyond homicide, the CCJ data shows broad declines: aggravated assault fell 9%, gun assaults dropped 22%, robbery declined 23%, and carjacking plummeted 43% from 2024 to 2025 [1]. Compared to 2019, robbery is down 36%, gun assaults are down 13%, and domestic violence is down 19% [1].

The property crime picture is more nuanced. Motor vehicle theft dropped 27% from 2024 to 2025 but remains 9% above 2019 levels [1]. Residential burglary fell 17% year-over-year and is 45% below 2019 [1]. Shoplifting declined 10% from 2024 to 2025 and is 4% below 2019 [1]. Nonresidential burglary is essentially flat compared to pre-pandemic levels [1].

The Retail Theft Debate: Organized Crime or Moral Panic?

Few crime categories have generated more political heat and less analytical light than organized retail theft. The National Retail Federation (NRF) reported that external theft incidents increased 19% from 2023 to 2024, on top of 26% growth from 2022 to 2023 [18]. Violence during theft events rose 17%, and incidents involving weapons increased 16% [18]. Cargo theft jumped 27% in 2024, with average losses exceeding $200,000 per incident [18]. Sixty-six percent of retailers reported transnational organized crime involvement in thefts since 2024 [18].

These industry-reported figures have driven significant policy responses. Target closed nine stores across New York, Seattle, San Francisco-Oakland, and Portland in 2023, explicitly citing theft and organized crime [19]. CVS announced plans to close 900 stores by 2026 [20]. Retailers have placed increasing numbers of products behind locked displays, a practice that has itself become controversial.

The counterarguments are substantial. The Brookings Institution and the Brennan Center for Justice have challenged the retail theft narrative with data showing that, among 24 cities with consistent shoplifting reporting, shoplifting decreased in 17 of them, and overall shoplifting in 2023 remained below 2018 and 2019 levels [21][22]. Even San Francisco — the poster city for retail theft narratives — saw a 5% decline in shoplifting between 2019 and 2023 [21].

Walgreens, which had blamed rising crime for multiple San Francisco store closures in 2021, reversed itself. CFO James Kehoe acknowledged in 2023 that the company "maybe cried too much" about theft [20]. Jeff Asher's analysis showed that the Target stores closed in Portland and Seattle actually had less reported crime than stores that remained open [20].

The R Street Institute, a center-right think tank, has urged caution about organized retail crime claims, noting that the NRF's data comes from self-reported retailer surveys rather than law enforcement statistics, and that "shrinkage" — the retail industry's term for inventory loss — includes employee theft, administrative errors, and damage, not just shoplifting [23].

The conservative rebuttal: even if aggregate shoplifting numbers are stable or declining, the nature of theft has changed. Organized rings operating across multiple jurisdictions, reselling goods on online marketplaces, represent a qualitatively different problem than individual shoplifters, and one that traditional policing and prosecution are poorly equipped to address. The NRF's data on transnational involvement and cargo theft losses are not readily dismissed as overstatement, and the 42% increase in larceny-theft incidents classified as hate crimes from 2023 to 2024 — from 467 to 665 incidents — suggests that at least some categories of theft-related crime are genuinely worsening [24].

The truth likely sits between the extremes: aggregate shoplifting is not at crisis levels, but organized and violent theft has become a more significant problem in specific markets, and retailers' decisions to close stores or lock up merchandise reflect a rational response to real costs — even if those same retailers sometimes exaggerate the scale of the problem for public relations or insurance purposes.

Hate Crimes and Specific Categories of Increase

Even as overall violent crime has declined, specific categories have moved against the trend. The FBI recorded 11,679 hate crime incidents in 2024 — the second highest year on record since tracking began in 1991 [24]. While this represented a 10% decrease from 2023's total of 12,498, the long-term trend has been sharply upward [24].

Carjacking surged during the pandemic, though it has since reversed. The Council on Criminal Justice found that carjacking fell 43% from 2024 to 2025 and is now 29% below 2019 levels [1]. However, rates of both carjacking and non-violent car theft remained above pre-pandemic levels for most of the 2020–2024 period [25]. The situation in Washington, D.C. was particularly acute: carjackings spiked 97.9% in 2023, from 484 to 958 reported incidents [26].

Motor vehicle theft more broadly exceeded one million reported incidents in 2023, up from roughly 810,000 in 2019 [27]. The Hyundai-Kia security flaw, which made 2011–2021 models of those vehicles extremely easy to steal, was a significant driver [27]. Auto theft dropped 17% in 2024 — the largest single-year decline in 40 years — partly because of manufacturer fixes and law enforcement crackdowns on theft rings [27].

Car insurance premiums reflected these trends. U.S. drivers paid an average of $2,543 per year for car insurance in 2024, a 26% increase from the prior year and a 52% increase from 2021 [4]. While theft was not the only factor — repair costs, medical costs, and litigation trends also contributed — insurers in high-theft cities explicitly priced in elevated theft risk [26].

The Perception Gap: Why Do People Feel Less Safe?

For most of the post-2020 period, public perception of crime diverged sharply from official statistics. Gallup polling in 2024 found that 49% of Americans believed there was more crime than the prior year — a relatively high share given that FBI data showed consistent declines [5]. The gap was heavily partisan: 90% of Republicans said national crime rates had risen, compared to 29% of Democrats [5].

By 2025, perceptions had begun to shift. Only 49% of Americans called crime "extremely or very serious" — the lowest since 2018 [5]. Only 30% said their community had become more dangerous, down from 49% in 2024 [5]. Notably, the partisan gap reversed direction with the change of presidential administration: Republican perceptions of crime dropped sharply while Democratic perceptions of crime increased [5].

This partisan pattern — where people feel safer when their party controls the White House — is consistent across decades of polling and is among the strongest pieces of evidence that crime perceptions are heavily mediated by political identity rather than direct experience [28].

But dismissing public fear as pure partisan perception is too simple. Several factors explain why people might rationally feel less safe even as aggregate rates fall.

Media environment. Social media and local news provide a constant stream of violent crime footage that was not available in earlier decades. A single viral carjacking video reaches millions of viewers; the statistical decline in carjacking rates reaches almost nobody.

Visible disorder. Open-air drug markets, tent encampments, and public mental health crises have become more visible in many cities since the pandemic. These may not register as "crime" in FBI statistics but contribute powerfully to a feeling of disorder and danger.

Specific categories rising. As documented above, motor vehicle theft, certain types of organized retail crime, and hate crimes increased even as overall rates fell. If a person's car was stolen, it matters little to them that aggregate auto theft is down 17% nationally.

Insurance costs and store closures. People experience rising car insurance premiums and locked-up merchandise at pharmacies as tangible evidence of a crime problem, regardless of what the UCR says.

The conservative argument, articulated by Mangual and others at the Manhattan Institute, is that the perception gap reflects not ignorance but pattern recognition — that Americans are correctly identifying a deterioration in public order that aggregate statistics fail to capture because of declining police reporting, reduced prosecution, and the decriminalization of low-level offenses [17].

The progressive counterargument, advanced by researchers at the Brennan Center and the Council on Criminal Justice, is that media coverage and political rhetoric have manufactured a crime panic that is contradicted by every available dataset when examined honestly [22][1].

Both arguments contain elements of truth, which is what makes this debate so resistant to resolution.

Source: Bureau of Labor Statistics

Data as of Mar 21, 2026CSV

The Progressive Prosecutor Debate

Few criminal justice controversies have been more heated than the question of whether progressive prosecutors — district attorneys who campaigned on reducing incarceration, declining to prosecute low-level offenses, and reforming cash bail — have increased crime.

The electoral record is mixed. San Francisco voters recalled DA Chesa Boudin in June 2022, with 55% voting to remove him [29]. Los Angeles County petitioners collected over 520,000 signatures to recall DA George Gascón, falling just under 47,000 short of qualifying for the ballot; Gascón subsequently lost his 2024 re-election bid [29]. In Philadelphia, the Republican state House impeached DA Larry Krasner, though the effort was widely seen as partisan [29].

On the other side, Krasner won re-election by a wide margin in 2021. Manhattan DA Alvin Bragg, one of the most politically targeted prosecutors in the country, won his initial election decisively and has overseen a period of falling crime in Manhattan [29]. The fact that voters in high-crime cities have re-elected reform prosecutors complicates the narrative that these policies are uniformly unpopular with the communities most affected by crime.

The academic evidence. The most rigorous study to date is Petersen, Mitchell, and Yan's 2024 quasi-experimental analysis published in Criminology & Public Policy, which examined crime rates in the 100 largest U.S. counties from 2000 to 2020 using difference-in-differences regressions [30]. Their findings were nuanced: progressive prosecutors were associated with approximately 7% higher property crime rates compared to matched counties, but had "limited impact on rates of violent crime" [30]. In absolute terms, crime rates fell in jurisdictions with both traditional and progressive prosecutors [30].

The Center for American Progress, citing this and other research, concluded that "progressive prosecutors were not responsible for increases in violent or property crime before, during, or after the COVID-19 pandemic" [31]. This is a defensible reading of the violent crime findings but arguably understates the property crime results.

The conservative case against progressive prosecution rests on different evidence. The Heritage Foundation has documented that of the 30 cities with the highest homicide rates, 27 have Democratic mayors, and 14 have prosecutors that Heritage classifies as "rogue" — prosecutors who, in Heritage's account, decline to enforce significant categories of law [17]. These 14 cities accounted for 68% of homicides in the top-30 cities [17]. Heritage argues that the academic studies cited by reform advocates use overly narrow definitions of "progressive prosecutor" and cherry-pick time periods that obscure the effects of policy changes [32].

The Manhattan Institute's critique extends to bail reform studies specifically. Heritage and Manhattan Institute researchers have challenged the methodology of Cook County bail reform research, arguing that the study undercounted murders, used a narrow definition of "violent crime" that excluded many serious offenses, and only counted the first new charge against released defendants even when multiple new crimes were charged [32]. Their analysis found that the number of released defendants charged with new crimes increased approximately 45%, and new violent crime charges increased about 33% [32].

The bail reform evidence, assessed on its merits, is genuinely mixed. The Brennan Center analyzed 33 cities, comparing 22 with bail reforms to 11 without, and found no relationship between bail reform and crime rates [33]. New York's bail reform, which eliminated cash bail for most misdemeanors and nonviolent felonies, was associated with reduced recidivism according to a study by the Data Collaborative for Justice [34]. In Illinois, which eliminated cash bail entirely in September 2023, jail populations declined with no increase in failure-to-appear rates, and violent crime in Cook County has decreased since implementation [35]. New Jersey's elimination of cash bail has been followed by declining crime, and 89% of pretrial defendants showed up for court appearances between 2019 and 2024 [35].

But the Heritage Foundation's methodological critiques of individual studies are specific enough that they cannot be dismissed as mere ideological opposition. The question of whether bail reform affects crime remains genuinely contested among researchers who have examined the same data and reached different conclusions, partly because the answer depends on which crimes you count, how you define the comparison group, and what time period you examine.

The Police Staffing Crisis

One factor that both sides of the debate largely agree on: American policing is in a staffing crisis. A 2024 national survey found that 70% of agencies reported recruitment is more difficult than five years ago [36]. On average, agencies are staffed at only 91% of approved positions [36]. Large departments employed 6% fewer officers in 2024 than in 2020, although medium and small agencies have recovered [37].

The scale of the shortages is significant. As of 2024, Baltimore was short roughly 600 officers, Chicago more than 1,300, Los Angeles over 1,000, Philadelphia about 1,200, and Washington D.C. nearly 500 [36]. Sixty-five percent of agencies reported reducing services or eliminating specialized units because of staffing — up from 25% in 2019 [36].

Departments have responded by lowering hiring standards. The NYPD changed its education requirement in 2025, prompting daily applications to jump from an average of 53 to 231 [37]. Other departments have reduced or eliminated requirements for college credits, prior experience, or physical fitness benchmarks [37].

The causes of the staffing crisis are debated. Law enforcement organizations point to the post-2020 "defund the police" movement, negative public sentiment, and scrutiny of use-of-force incidents as having made policing unattractive to potential recruits [36]. Reform advocates argue that departments have struggled with recruitment for decades due to low pay, dangerous conditions, and a culture that resists accountability — and that the post-2020 moment simply exposed longstanding structural problems.

Source: Police Executive Research Forum / American Police Beat

Data as of Mar 21, 2026CSV

What is measurable: when a city loses 20–30% of its police force and cannot replace them, response times increase, proactive policing decreases, and clearance rates — the share of crimes that result in an arrest — decline. Whether this contributes to higher crime rates is empirically supported by a substantial body of research showing that police presence has a deterrent effect, though the magnitude of that effect is debated [38].

The International Comparison

The United States remains a dramatic outlier among wealthy democracies in its homicide rate. In 2021, the U.S. recorded 6.81 homicides per 100,000 inhabitants — more than three times Canada's rate of 2.07, and roughly seven times the rates in Germany, France, and the United Kingdom [39]. Japan, at 0.23 per 100,000, had a rate roughly thirty times lower than the United States [39].

The U.S. rape rate was approximately seven times higher than the European average, and the robbery rate was four times higher [39]. These gaps have been persistent across decades and are not primarily attributable to differences in reporting methodology, though reporting practices do vary.

The 2024 decline in U.S. homicides — to approximately 5.0 per 100,000 — narrows the gap modestly, but the structural factors driving the disparity (gun availability, inequality, incarceration-driven social disruption, drug markets) have not fundamentally changed. A comparison between U.S. and Canadian police-reported crime trends from Statistics Canada found broadly parallel trajectories in recent years, with both countries experiencing pandemic-era spikes followed by declines, though the U.S. rates remain substantially higher in absolute terms [40].

Who Gets Credit for Falling Crime?

Both parties have claimed credit for the post-2020 crime decline, and neither claim withstands scrutiny.

The Trump administration pointed to its emphasis on law-and-order messaging and federal law enforcement support as drivers of the decline. The Biden administration pointed to the American Rescue Plan's funding for local policing and violence intervention programs. In reality, the crime decline has been remarkably broad — cutting across cities with progressive and conservative prosecutors, cities that increased and decreased police funding, cities with Democratic and Republican mayors. This breadth suggests that the primary drivers are structural rather than political: the unwinding of pandemic-era disruptions, the receding of the fentanyl-driven overdose crisis in some markets, demographic shifts, and the natural regression from an anomalous spike.

No president controls local crime rates in any direct sense. The rhetoric of taking credit — or assigning blame — serves political purposes but does not reflect how crime actually works in a federal system where policing, prosecution, and sentencing are overwhelmingly state and local functions.

The Data We Still Don't Have

Honest engagement with the crime debate requires acknowledging what remains unknown.

The NIBRS transition, while largely complete, produced a two-year gap in reliable national data (2021–2022) that makes trend analysis across the pandemic period uncertain [10][11]. The FBI's practice of revising prior-year estimates — sometimes significantly — means that today's headlines about historic declines could be revised in future data releases [11].

The NCVS redesign in 2024 creates a break in the victimization survey's time series that will take several years to fully understand [8]. Until the redesigned instrument produces stable trend data, the divergence between the NCVS and FBI data will remain an open question rather than a resolved one.

We lack reliable national data on several crime categories that drive public anxiety: organized retail theft (for which we have only industry surveys, not law enforcement data), carjacking (which most jurisdictions do not track as a separate offense), and online fraud (which is vastly underreported and poorly captured by existing systems).

And the fundamental question of whether declining reporting rates are masking stable or rising victimization — the core tension between the FBI and NCVS data — cannot be definitively resolved with current measurement tools.

What the Evidence Supports

The following statements are supported by the available evidence, with the caveats noted above:

Overall violent crime rates, including homicide, have fallen dramatically from the 2020–2021 pandemic spike and are now at or near historic lows, whether measured by FBI data, city-level police data, or public health records [1][2].

The 2020 homicide spike produced thousands of excess deaths concentrated overwhelmingly among young Black men, representing a genuine public health catastrophe that is often lost in the political debate [13][14].

Specific categories of crime — particularly motor vehicle theft, organized retail theft operations, and hate crimes — increased even during the overall decline, and the experience of these crimes is not captured by headlines about falling aggregate rates [18][24][27].

The gap between FBI police-reported data and NCVS victimization surveys is real, methodologically significant, and not fully explained [8][9].

Progressive prosecutors do not appear to have increased violent crime, but the evidence for modest increases in property crime under their tenure is plausible and consistent across multiple studies [30][31].

Bail reform has not produced the public safety disasters that critics predicted, but individual studies have been challenged on methodological grounds that merit serious engagement rather than dismissal [33][34][32].

Police staffing shortages are severe and have measurably reduced the capacity of departments to respond to and investigate crime [36][37].

American crime rates remain dramatically higher than peer democracies, particularly for homicide, and recent improvements have not changed this structural reality [39].

Public perceptions of crime are heavily influenced by partisan identity, media consumption, and visible disorder — factors that can operate independently of actual crime rates [5][28].

The honest conclusion is not that crime is rising or falling — it is that "crime" is not a single phenomenon. Some categories are at historic lows. Others remain elevated. The data infrastructure that Americans rely on to answer the question is itself contested and incomplete. And the political incentives to declare either victory or crisis ensure that the debate will continue to generate more heat than light.