Is Crime Actually Dropping in the U.S. in 2025? Not as Much as Reported: Unmasking Misleading Data & Avoidance Behaviors

Around 50% of the reported decrease in U.S. crime is probably a fake-out... but the overall trend is still favorable (reduction)

Throughout human history, crime—including violent confrontations, tribal conflicts, theft, and fraud—has significantly declined, particularly relative to population size.

Early human societies experienced extremely high rates of crime and violence, driven by fierce resource competition and minimal social controls.

Over time, the establishment of centralized governments, formal legal systems, and evolving societal norms greatly reduced arbitrary violence and disorder.

Crime in U.S. 1776 to Present:

• Since its founding in 1776, the U.S. has seen substantial fluctuations in crime rates.

• Initially, crime was prevalent due to weak law enforcement and frontier violence.

• Subsequent improvements in policing, economic stability, and judicial practices led to general crime reductions, though notable periods such as urbanization (late 1800s) and the crime waves of the 1960s–1990s temporarily reversed this downward trajectory.

Crime Trends from 2000 to 2025:

• Crime rates trended downward after 2000, reaching historic lows around 2014.

• A significant crime spike occurred between 2020–2021, driven by societal disruptions from COVID-19, economic instability, and shifts in policing practices.

• From 2022 onwards, crime rates stabilized and resumed declining, approaching pre-pandemic levels by 2025.

Persistent High Crime in Various U.S. Cities

Despite overall national declines, crime remains notably high in specific U.S. cities, often substantially exceeding the national average of approximately 350 violent crimes per 100,000 residents in 2024 and likely into 2025.

Cities such as Baltimore, Detroit, and St. Louis report significantly elevated rates.

• Baltimore: Approximately 1,587 violent crimes per 100,000 residents (2023), remaining around 1,500 in 2024.

• Detroit: Approximately 1,868 violent crimes per 100,000 residents (2024).

• St. Louis: Approximately 1,200 violent crimes per 100,000 residents (2024).

Memphis notably leads with a violent crime rate of approximately 2,437 per 100,000 in 2024, highlighting the considerable regional variation and the ongoing challenge of high crime rates in certain urban areas.

Misleading Aspects of Recent Crime Trends:

It is true that despite certain cities having high crime rates, the overall crime rate is down. But this “declining crime” rate can be a bit misleading when considering variables that are often never considered/accounted for by left-wing (progressive, liberal, Democrat) pundits:

1. Legalization and/or Decriminalization: Previously illegal activities, such as certain drug offenses, have been legalized or decriminalized, reducing official crime statistics without necessarily decreasing actual activity.

2. Selective Policing & Law Enforcement: Police sometimes selectively enforce laws or refrain from enforcing them in certain areas, leading to artificially lower crime statistics.

3. Underreporting by Victims: Individuals may increasingly avoid reporting crimes due to distrust in authorities or perceived ineffectiveness of the justice system.

4. Incomplete Documentation: Law enforcement agencies might fail to accurately document reported crimes, skewing official records.

5. Inconsistent Crime Tracking: Cities, states, and national authorities often lack standardized crime-tracking methods, resulting in inconsistent and potentially misleading data.

6. Avoidance Behavior: Non-criminal individuals increasingly engage in avoidance behaviors, deliberately staying away from perceived high-risk areas or interactions, which artificially suppresses reported crime rates.

In other words: While overall crime in the U.S. is probably decreasing, these variables warrant careful consideration — as current crime stats may be misleadingly skewed (at least slightly) to the downside.

If you: legalize behaviors that were previously classified as crimes/criminal behavior (e.g. drug use, shoplifting up to $XXX amount with no consequences, etc.); no longer enforce certain laws; defund and/or downsize police forces; stigmatize policing efforts; pass laws that make it more difficult for police to do their jobs; underreport crimes to police; have the police themselves not document crimes (or underreport them); have cities/states & federal government agencies not properly collect data and/or track certain crime rates; and have citizens going out of their way to avoid criminals and/or high-crime areas — you end up with a misleading picture.

---

I. Crime in U.S. Dropping Since the 1990s

Reported crime in the United States has declined markedly since the early 1990s and continued to trend downward through 2015, according to official statistics. Even between 2000 and 2020 – a period of economic swings, policy changes, and a brief uptick in violence mid-decade – FBI data showed overall violent crime falling and property crime dropping to half-century lows.

For example, in 2022 the FBI tallied 380.7 violent crimes per 100,000 people and 1,954.4 property crimes per 100,000 – dramatically lower rates than in the 1990s. However, these “reported” crime declines may not tell the full story of public safety. Multiple factors can suppress or distort official crime data, especially in the past decade.

For this reason, I wanted to think about the extent to which decreases in reported crime might be due not to genuine drops in criminal activity, but to: (1) strategic avoidance behaviors by the public, (2) reduced police enforcement, (3) underreporting or misclassification by agencies, (4) lower public willingness to report crimes, and (5) policy changes (bail reform, decriminalization, “defunding,” or prosecutorial leniency).

I break down trends by crime type, geography, and time period, comparing regions with aggressive policy shifts (e.g. San Francisco, Chicago, New York, Los Angeles) to those with traditional enforcement approaches (e.g. Florida, Texas, Utah).

Lastly, I estimate the relative contribution of each factor to the perceived crime drop and rank their significance, reconciling these insights with national statistics and highlighting key discrepancies or distortions in the data.

U.S. violent and property crime rates have plunged since the 1990s, per FBI (reported crimes) and BJS (victim survey) data. Both sources show a long-term decline through 2022, but the victimization rates (which include unreported crimes) remain higher than the reported crime rates, indicating substantial underreporting.

---

II. U.S. Crime Trends Overview (2000–2025)

A.) 2000–2015: Sustained Decline in Crime

From 2000 to 2015, the United States experienced a steady decline in both violent and property crime reported to law enforcement. This continued the “Great American Crime Decline” that began in the 1990s.

According to FBI Uniform Crime Reports, the violent crime rate fell from around 506 per 100,000 in 2000 to roughly 373 per 100,000 in 2015 (approaching lows not seen since the 1960s). The property crime rate likewise dropped significantly, from about 4,600 per 100,000 in 2000 to about 2,600 per 100,000 in 2015.

These declines were broad-based: for example, between 2000 and 2014, reported burglary fell by ~30%, and motor vehicle theft by nearly 50%, in part due to new anti-theft technologies and other factors.

Survey data from the Bureau of Justice Statistics (BJS) confirms a similar pattern of declining victimization through 2015. The period was relatively stable, with the only notable upticks being a slight increase in violent crime in the mid-2000s and early 2010s. Overall, Americans in 2015 faced far lower odds of being victimized than in decades prior.

Contributors:

Researchers attribute the 2000–2015 crime drop to a mix of factors including: improved policing strategies, an aging population, waning of the crack cocaine epidemic, a strong economy in the early 2000s, and target-hardening measures (like widespread security systems and private guards). It’s important to note that even during this era, a significant share of crime went uncounted: surveys indicated only about half of violent victimizations and one-third of property crimes were reported to police. Thus, part of the “decline” in reported crime may reflect persistent underreporting. However, the downward trend was robust enough that it is widely accepted as a real phenomenon, albeit one aided by preventive measures.

B.) 2015–2020: A Reversal & Data Disruptions

Violent Crime: After 2014, the long decline in violence stalled and briefly reversed. From 2015 to 2016, the nation saw a sharp rise in homicide and shootings (often dubbed the “Ferguson effect” period in reference to strained police-community relations).

FBI data show violent crime increased during 2015–2016 (e.g. the murder rate jumped 11% in 2015, one of the largest single-year increases since the 1960s).

This spike was concentrated in certain cities (Chicago, Baltimore, St. Louis, etc.), while others remained at historic lows. After 2016, violent crime leveled off and even dipped slightly in 2018–2019. By 2019, the national violent crime rate (366 per 100,000) was still near multi-decade lows.

2019 Crime in the United States (Source: UCR.FBI.GOV)

Property Crime: Property offenses continued to decline or flatten in this period. The FBI’s property crime rate dropped steadily through 2019, reaching about 2,110 per 100,000 in 2019, the lowest recorded in over 50 years. Larceny and burglary saw continued decreases, though auto theft trends were mixed (some areas reported upticks in car theft toward 2019). Notably, many states enacted reforms in these years – for example, California’s Proposition 47 in 2014 reclassified certain theft and drug possession offenses from felonies to misdemeanors. Surprisingly, official statistics showed no surge in shoplifting reports after Prop 47; in fact, reported shoplifting incidents decreased ~2% post-2014 by one analysis. This “decline” is widely attributed to fewer arrests and less reporting of low-level theft, rather than an actual drop in theft behavior. We will revisit this under policy changes.

Data and Reporting Issues: By 2020, cracks in the crime data picture became evident. The FBI was preparing a transition to a new reporting system (NIBRS) in 2021, and some agencies began to struggle with data submission. Still, through 2019 the national data was fairly complete. Another important development was the divergence between police data and victim surveys. For instance, while FBI numbers showed modest changes year-to-year in the late 2010s, the BJS National Crime Victimization Survey indicated that many victimizations still were not reported. In 2019, only about 45% of violent incidents and 34% of property incidents were reported to police (continuing a slight downward trend in reporting rates). This suggests the public’s willingness to report, especially minor or sensitive crimes, was eroding even before 2020.

C.) 2020–2025: Pandemic, Policy Upheavals & Perceived Crime Wave

The period from 2020 onward saw massive social disruptions that impacted both actual crime and crime reporting.

• 2020: The COVID-19 pandemic and associated lockdowns led to a complex mix of crime outcomes. Overall property crime plummeted in spring 2020 as businesses closed and people stayed home (fewer opportunities for burglary and robbery). The FBI reported property crime fell ~8% in 2020, reaching a modern low. However, violent crime – particularly homicides and shootings – surged in 2020. The U.S. murder rate jumped 29% in 2020, the largest one-year increase on record. Aggravated assaults and motor vehicle thefts also rose in many cities. Criminologists point to multiple stressors: economic strain, community tensions after the George Floyd killing, reduced proactive policing, and more guns on the street. Notably, many police departments also adjusted enforcement due to COVID (e.g. avoiding low-level arrests to reduce jail crowding), which lowered reported minor crime while potentially increasing violent crime due to reduced deterrence.

• 2021–2022: Crime patterns remained volatile. Homicides continued at elevated levels in 2021 (up another ~6% nationally), and some forms of theft rebounded as life normalized (e.g. a wave of carjackings and catalytic converter thefts hit many cities in 2021–22). Official national data, however, became spotty. In 2021 the FBI transitioned to the new NIBRS reporting system, and a large number of agencies failed to fully report data. In 2022, 32% of U.S. police departments did not submit crime statistics to the FBI at all, and another 24% submitted only partial-year data. This gap led to artificially low national crime figures for 2021–2022, since many crimes simply weren’t captured in the FBI database. The FBI’s published 2022 crime report claimed violent crime decreased ~1.7% from 2021, but this was largely a data mirage. In contrast, the BJS victim survey showed violent victimizations increased by 46% in 2022, and independent city-level data also indicated continued high levels of shootings. In short, by 2022 official statistics were at odds with on-the-ground realities, due to reporting lapses and behavioral changes.

• 2023–2025: Early data for 2023 suggest some regression to pre-2020 norms in serious violence – for example, murders in major cities were down ~12% in the first half of 2023 and preliminary 2024 data indicate homicide rates may fall back to around 2015 levels by the end of 2024. Property crime trends in 2023–24 have been mixed: a nationwide spike in auto theft in 2022 (notably the “Kia/Hyundai” theft epidemic) was followed by a sharp drop (~17% decline in auto theft in 2024 after manufacturers improved anti-theft measures). Shoplifting and retail theft became high-profile concerns, though official data remained inconsistent (some cities report increases, while others show flat or declining shoplifting rates, likely reflecting differences in reporting and enforcement). Overall, by 2025 the rates of reported crime remain well below historical highs, but public perception is that crime is high and rising. Gallup and Pew surveys find over 60% of Americans consistently believe national crime is getting worse, a perception exacerbated by visible incidents and perhaps by knowledge that official numbers might understate the problem.

Crime Trends:

Nationally, reported violent crime in 2025 is roughly similar to the mid-2010s, after spiking in 2020–2021 and then easing somewhat. Reported property crime in 2025 remains near historic lows (apart from localized jumps in specific theft categories). However, the discrepancy between reported crime and experienced crime appears to have widened post-2020.

---

III. Factors Influencing the Decline in Reported Crime in the U.S.

1.) Strategic Avoidance & Target Hardening by the Public

One often-overlooked reason for lower crime rates is that Americans changed their behavior to avoid victimization. Over the past two decades, individuals and businesses have increasingly engaged in crime-avoidance strategies – moving away from dangerous areas, reinforcing their homes and vehicles, and altering daily routines – which reduce opportunities for crime.

These measures don’t necessarily mean criminals are less active; rather, they make it harder for crimes to occur or be completed, thereby lowering reported incidents. Key avoidance behaviors include:

• Relocating or segregating: Families with means have relocated from high-crime neighborhoods to safer areas (e.g. suburbs or gated communities). Economist Mark Kleiman noted that “one reason crime has fallen in the past decade is that potential victims…have moved away from high-crime neighborhoods into low-crime neighborhoods”. This migration reduces crime in the origin areas (fewer targets) but can leave behind concentrated poverty and a smaller pool of victims, potentially masking the remaining crime (since some incidents simply aren’t happening where they once did). On a larger scale, the urban population shifts post-2020 – for example, significant population declines in cities like San Francisco and Chicago since 2020, partly driven by safety concerns – likely contributed to lower reported crime in those cities (with fewer residents, there are fewer total crimes, even if per-capita risk remains high).

• Avoiding risky situations: People have adjusted their routines to minimize exposure to crime. For instance, many avoid walking alone at night, steer clear of known trouble spots, and limit interactions with individuals perceived as dangerous. After spikes in violence or publicized crimes, these avoidance behaviors tend to increase. A peer-reviewed study confirmed that in communities of color, residents “pulled back” from calling 911 and going out as usual following high-profile police violence incidents. While that study dealt with police violence, the broader implication is that fear (of either crime or unsafe law enforcement encounters) leads to social withdrawal. The result is fewer opportunities for street crime like robbery or assault – effectively preventing some would-be crimes. Notably, during the 2020 unrest and pandemic, many city-dwellers stayed home or avoided downtown areas, which contributed to steep drops in robbery and public assault statistics. This was a strategic avoidance at a societal scale.

• Security systems and guardianship: There has been a proliferation of target-hardening technology. By the mid-2010s, home security systems, CCTV cameras, improved door locks, car alarms, and electronic immobilizers became commonplace. Studies indicate that such measures deter crime or displace it. For example, research finds CCTV surveillance can cut crimes like vehicle break-ins by around 25–50% in covered areas. Home security cameras (e.g. Ring doorbells) and alarm systems similarly make burglars think twice – about 60% of burglars will skip a house with an alarm or camera and seek an easier target. The widespread adoption of these tools in the 2000s and 2010s is credited with a significant share of the drop in property crimes. Indeed, criminologists proposed a “security hypothesis” for the crime drop, noting that better anti-theft technology (like electronic engine immobilizers in cars) coincided with a 53% decline in vehicle theft rates from 1993 to 2022. In short, it became harder to steal modern cars or burglarize watched homes. These improvements reduced actual successful offenses, and therefore reported crimes, even if the intent to offend remained.

• Private security & self-protection: There has been a surge in private security personnel nationwide – by the 2010s, private security guards outnumbered public police by a wide margin. Gated residential communities, business improvement districts with private patrols, and armed security in retail stores all add extra “eyes and ears” that deter crime. Pew Research highlights that this rise in private security has “served as an additional deterrent” contributing to the crime decline. Additionally, more civilians are armed or carry defensive tools than in past decades (concealed carry permits expanded in many states). While the net effects are debated, some individuals feel safer and may prevent crimes against themselves (for example, shootings of would-be intruders by armed homeowners, or even non-lethal use of pepper spray by potential victims). Such instances may stop crimes from being completed (and thus from being recorded as offenses).

Impact & Quantification: Strategic avoidance and target hardening have likely had the greatest impact on property crimes. The dramatic drop in burglary and theft since 2000 is often attributed at least in part to these defensive measures. One analysis attributed on the order of 30–40% of the decline in property crime to improved security and avoidance behaviors. Vehicle theft is a prime example: engineers and consumers “hardened” the target (cars) so effectively that auto thefts fell from ~1.7 million in 1991 to under 800,000 by 2014, despite many willing car thieves still out there. Likewise, the prevalence of cameras in stores and on streets has made shoplifting, robbery, and assault more likely to be caught, discouraging some offenses. In violent crime, avoidance plays a role too: if fewer people frequent violent hotspots or if potential victims remove themselves (e.g. women avoiding walking alone where sexual assaults occurred), then reported violent incidents fall. However, the effect on violent crime is less quantifiable; criminals may seek out other victims, whereas a thwarted burglary doesn’t always get replaced. Avoidance can also simply shift crime rather than eliminate it – as Kleiman quipped, a burglar might just move on to an easier home when he sees your alarm sign. Thus, some share of the crime drop due to target hardening is a “transfer” (crime prevented in one spot may occur elsewhere or in other forms), but a large portion represents genuine prevention.

Time Pattern: The influence of strategic avoidance grew throughout 2000–2025. Early 2000s saw the start of these trends (migration to suburbs, CCTV in city centers); by the 2010s, they were in full swing nationwide. Post-2020, avoidance took a new turn: many people began avoiding downtowns or public transit due to both crime fears and pandemic habits, which likely kept certain crime rates lower than they otherwise would be. For instance, if ridership on a city’s subway is down 50%, one would expect fewer subway crimes simply due to fewer targets. This doesn’t mean criminals disappeared – it means potential victims (and potential offenders) aren’t in contact as much. In summary, avoidance and target hardening have been a significant factor – likely one of the top contributors – to the long-term decline in reported crime, especially for theft and burglary. These are proactive, private responses that made Americans safer but also can mask underlying criminal intent (since crimes “avoided” don’t show up in statistics).

2.) Reduced Police Enforcement & Proactive Policing

A second major factor is changes in police activity. Crime stats can decrease not only when criminals offend less, but also when police observe or intervene less. Over the past decade – and especially after 2015 and 2020 – many police departments pulled back on certain enforcement practices, due to policy changes, legal rulings, staffing shortages, or political pressures. This “de-policing” can lead to fewer arrests and incident reports, thereby artificially lowering reported crime in some categories (even if actual crime levels are unchanged or higher). Several manifestations of reduced enforcement include:

• Fewer stops, arrests, and investigations: Across the country, police have been making far fewer discretionary stops and arrests than in years past. For example, in Milwaukee a recent audit found arrests plummeted by 82% from 2012 to 2023 (51,176 arrests in 2012 down to just 9,061 in 2023). Traffic stops by Milwaukee PD dropped 81.5% since 2015, and “stop-and-frisk” style pedestrian stops fell 98.5% over the same period. Milwaukee went from having one of the highest arrest rates in the U.S. (85 arrests per 1,000 residents) to one of the lowest (16 per 1,000) by 2023. Crucially, the biggest declines were in arrests for minor offenses – summonses and low-level citations fell ~96%. This pattern is mirrored in other cities: New York City, for instance, dramatically scaled back proactive policing after 2013 (stop-and-frisk encounters fell from over 685,000 in 2011 to under 20,000 a few years later), and total NYPD arrests dropped from roughly 400,000+ in 2010 to just over 200,000 in 2019. Why does this matter? Because many crimes (especially drug offenses, weapons possession, vandalism, minor theft, etc.) are only recorded if police actively look for them or make an arrest. When enforcement is dialed back, those crimes don’t end up in the data. For example, if narcotics units stop doing street sweeps, the number of drug possession offenses “reported” will plunge – but that doesn’t mean drug use or dealing stopped. In Milwaukee’s case, the huge drop in low-level arrests corresponds to a period when some residents actually felt less safe (community surveys there showed frustration with scant traffic enforcement and public disorder). Thus, reduced enforcement improved the crime stats (making it look like minor crime fell), even as community conditions may not have improved to the same degree.

• Intentional police disengagement (the “Ferguson Effect”): After contentious police incidents (e.g. officer-involved shootings) and the ensuing protests, there have been documented declines in police proactive activity in several cities. Officers, feeling under scrutiny or demoralized, may adopt a hands-off approach to avoid political or legal fallout. The theory, often called the Ferguson Effect (after the 2014 Ferguson, MO unrest), suggests police “do less” enforcement, which can both reduce recorded crime (for minor infractions) and potentially embolden criminals (leading to more serious crime). Empirical evidence is mixed, but there are clear cases of post-2014 pullbacks: A 2024 study found that after a high-profile police killing in one city, officer-initiated activity dropped sharply in majority-Black neighborhoods, even as it slightly increased in White neighborhoods. In other words, police withdrew from the communities with tense police relations. Similarly, Baltimore saw police arrests decline significantly after the 2015 Freddie Gray protests (arrests fell ~28% the following year), and Baltimore’s then-Police Commissioner acknowledged officers had become “less aggressive” on low-level offenses. Chicago in late 2015 also saw a sudden drop in police stops (after an ACLU agreement and public outrage over a shooting), with traffic stops and investigatory stops dropping by the thousands; Chicago’s police union openly said officers were avoiding proactive work for fear of being “the next viral video.” These pullbacks mean fewer reported offenses in areas like drug crimes, loitering, trespassing, and even less follow-up on calls. In some instances, police may not respond as rigorously to property crime calls (e.g. taking reports by phone or not at all), which directly cuts down what gets logged. One anonymous officer told reporters that with current policies, “If it’s not a violent felony, a lot of times we just file a report and move on” – reflecting a triage approach that lets many smaller crimes go unaddressed (and thus uncounted).

• Staffing shortages & “defund” budget cuts: In the wake of 2020’s unrest and the “Defund the Police” movement, many departments faced budget constraints or political mandates to shift resources. While the actual dollar cuts in most cities were modest or temporary, a more critical issue was police staffing. Waves of retirements and resignations hit departments from 2020–2022. Cities like Seattle, Minneapolis, and New York lost hundreds of officers, shrinking forces to levels not seen in decades. Fewer officers inevitably mean lower enforcement output – detectives with high caseloads might only investigate the most serious crimes, and patrol units might eliminate specialized crime-suppression teams. For example, Minneapolis (which cut some funding and saw many officers quit after 2020) experienced a steep drop in traffic stops and proactive policing; the number of police-citizen contacts plunged, and minor offenses were often ignored. Seattle, after cuts and attrition, reportedly had around 60% of the force it needed for patrol, leading to extremely selective enforcement. In such scenarios, police simply do not have the bandwidth to record every crime. One result: lower reported crime rates for offenses that aren’t proactively addressed. If a department has no narcotics unit, drug arrest numbers will fall; if it cannot spare officers for shoplifting calls, many retail thefts go uncounted except by private security logs. This dynamic was evident in California, too, where many agencies shifted focus during COVID and never fully returned to pre-2020 enforcement levels. Statewide in California, the arrest rate in 2023 hit a historic low – 1,992 arrests per 100,000 people, the lowest on record, with felony arrest rates continuing to drop. This was partially due to policy (Prop 47, etc.) but also due to police capacity: with limited resources, many non-violent offenses just don’t result in arrests anymore.

• Policy-driven de-prioritization: In some jurisdictions, new laws or directives explicitly curtailed enforcement of certain crimes. For instance, during the pandemic, various cities instructed police to stop arresting for minor drug possession, prostitution, or low-level warrants to avoid jail crowding. Philadelphia’s police in 2020 temporarily stopped arrests for drug possession, theft from autos, and prostitution – instead they just filed paperwork. The immediate effect was a plunge in those offense categories in the stats, since an officer who normally might log 5 drug arrests a week was logging 0. While some of these policies were temporary, some became permanent shifts (or lasted many months, contributing to 2020’s crime drop in non-violent categories). Similarly, bail reform (discussed more in Factor 5) sometimes influenced police behavior: if officers know that making an arrest will result in an immediate release (for example, for shoplifting), they might opt not to formally arrest at all, issuing a citation or simply warning the suspect. That keeps the incident out of the “crime” data (if no report is taken, it doesn’t count).

Impact & Quantification: Reduced enforcement has a dramatic impact on recorded crime, especially for crimes that are not usually reported by victims directly. The best example is drug offenses: these are often “victimless” crimes that only appear in data when police proactively make an arrest or seizure. Over the 2010s, as enforcement softened, drug arrest numbers fell steeply. Notably, marijuana possession arrests declined by over 50% nationally from their peak (~800,000 annually a decade ago to ~350,000 in 2020), thanks to both legalization and police deprioritization. That alone removes a huge number of “crime reports” from the system. Similarly, weapons possession cases (like carrying an illegal gun) depend on police stops – the NYPD, for instance, recovered far fewer guns once stop-and-frisk was reduced, meaning fewer illegal gun possession incidents on record (even as more people may have been carrying). Traffic offenses and DUIs dropped in stats during 2020-2021, largely because traffic enforcement was scaled back (not necessarily because drivers got safer). One can roughly estimate the effect: if a city cuts low-level arrests by 80%, its overall index crime might not drop that much (since index crimes are major ones), but its total arrests will plummet, and Part II crimes (like drug violations, simple assault, disorderly conduct) will show large decreases. Nationally, the Vera Institute noted overall arrest rates fell 29% from 2006 to 2016, outpacing crime declines – evidence that less enforcement contributed to the appearance of less crime. After 2020, this gap widened further.

Importantly, reduced enforcement can also indirectly lead to more serious crime going unsolved or unreported, which we’ll cover under underreporting. For ranking, this factor is extremely significant: Many analysts believe the post-2015 “crime declines” in minor offenses are mostly attributable to policing changes rather than behavioral changes. In cities like San Francisco, Los Angeles, Seattle, and New York (with various reforms and pullbacks), reported low-level crimes fell or stayed low in the late 2010s even as community complaints suggested problems were rampant. On the other hand, jurisdictions that maintained traditional enforcement (e.g. some Florida counties or smaller cities in Texas) might show more stable or accurate minor crime stats. For instance, Florida’s crime reports include thousands of drug arrests that simply wouldn’t occur in a place like Seattle where police don’t enforce drug possession – making Florida’s stats appear higher, though actual crime may or may not be. Thus, differing enforcement intensity can create artificial geographic differences in reported crime.

Time Pattern: Reduced enforcement had a modest effect in 2000–2015 (those years actually saw increased policing numbers in some places – e.g. more cops hired in the 2000s, CompStat emphasizing enforcement – which likely helped reduce actual crime but kept reporting high). The effect grew in 2015–2020, especially after 2014–2016: many departments faced criticism and adjusted tactics (e.g. Chicago 2016 stop reductions, NYPD 2014 post-Eric Garner slowdown, etc.), so enforcement fell and some crime stats (particularly misdemeanors) flattened or declined independent of actual conditions. The largest impact is post-2020: virtually every major city reported fewer police-initiated actions after 2020. Combined with staff shortages, this meant a nationwide dip in enforcement. By 2023, many cities were arresting and ticketing at decades-low rates. Consequently, a portion of the “crime is down” narrative in 2022–2024 – heralded by some officials – is attributable to this reduced enforcement. President Biden, for example, cited preliminary FBI data showing violent crime at historic lows in early 2024, but part of that may be because police in 2023 are engaging far less than in, say, 2013.

Overall: Reduced Police Enforcement is a major factor (likely #1 or #2 in significance) in explaining why reported crime has fallen, especially for lower-level offenses and in the post-2015 era. It improves the statistics but can mask real problems.

3.) Underreporting & Misclassification by Law Enforcement Agencies

This factor examines deliberate or systemic under-counting of crimes by the agencies themselves. This is about police departments failing to record or correctly classify the crimes that are reported. This can occur due to political pressure to show low crime, bureaucratic errors, lack of investigative bandwidth, or changes in definitions.

Underreporting by agencies leads to crime stats that look better than reality. Several notable examples and mechanisms include:

• Downgrading crime categories: Some departments have been caught intentionally misclassifying serious crimes as less severe offenses to keep official crime numbers down. A famous case is the Los Angeles Police Department (LAPD), which an audit found misclassified roughly 14,000 aggravated assaults as “simple assaults” from 2005 to 2012. By classifying stabbings, beatings, and other serious attacks as minor incidents, the LAPD underreported violent crime. The LA Times analysis showed that if those cases were correctly counted, Los Angeles’s violent crime rate was 7% higher than reported and serious assaults were 16% higher in that period. LAPD officials admitted the errors, noting it had a “corrosive effect on the public’s trust” and pledged to improve data accuracy. Though they claimed the overall downward crime trend remained, clearly the decline was overstated by these manipulations.

• Reclassifying or “disappearing” crimes: Perhaps even more egregious was what investigative reporters found in Chicago. Chicago Magazine uncovered that the Chicago Police Department in 2013 reclassified a number of obvious homicides into non-criminal death categories. In one case, a young woman’s death – she was found tied up and gagged, ruled a homicide by the medical examiner – was later mysteriously re-coded as a non-criminal death by a police supervisor, right before year-end statistics were tabulated. At least 10 murders in 2013 were systematically reclassified to lower the homicide count, and dozens of other serious felonies (robberies, assaults, burglaries) were downgraded or made to vanish from the stats for “illogical or unclear reasons”. Officers in Chicago described pressure from leadership to hit crime reduction targets, referring to “magic ink” that could make cases go away with a keystroke. This indicates institutional underreporting to serve political ends (e.g. Chicago’s mayor at the time could tout a drop in murders that was partly achieved by not counting some). The scale in Chicago’s case was enough to trim a few percentage points off the city’s violent crime totals for that year.

• Discouraging victims from filing reports: Another tactic is when officers or 911 operators intentionally discourage victims from making an official report. This was highlighted by an NYPD whistleblower, Officer Adrian Schoolcraft, who in 2009 recorded his superiors instructing officers to persuade victims to retract complaints or to refuse generating reports for certain crimes. The NYPD’s internal investigation later validated that crime reports were being manipulated in his precinct: felonies were downgraded, and in some cases calls for service were never recorded as crimes at all. Schoolcraft’s tapes showed that if a victim wasn’t insistent, an officer might code an incident as “dispute” instead of attempted robbery, etc. – ensuring it didn’t hit the crime stats. This practice was driven by CompStat-era pressure to show continuous crime reductions; commanders were held accountable for numbers, creating an incentive to “cook the books.” A 2012 NYPD internal report (made public via Reuters) confirmed such manipulation was real, lending credence to the idea that some of New York’s impressive crime drop was padded by data games. Other departments have had similar whistleblowers or audits (e.g., a 2021 probe in New Orleans found that hundreds of sexual assault and rape reports were misclassified or not investigated, lowering official rape counts).

• Omitting data from FBI reports: Underreporting can also happen at the data submission level. As mentioned, the FBI’s transition to NIBRS in 2021–2022 led many agencies to fail to send complete data. In 2022, only 68% of law enforcement agencies provided a full year of crime data. The missing agencies included some major cities. For example, New York, Los Angeles, and Chicago were among those whose data was incomplete or absent in the FBI system that year. The effect was a large undercount in the FBI’s national totals (since crimes in non-reporting cities were treated as zero). The Marshall Project noted that this caused an “artificially lower crime rate” at the national level for 2021–2022. Essentially, the “denominator” of agencies shrank, but the FBI didn’t adjust for it, so the crime drop in the data isn’t entirely real. This is less about malicious intent and more about system issues, but it fits under institutional undercounting. Another example: some jurisdictions simply choose not to report certain offenses federally; for years, rape was underreported in places that didn’t use the updated FBI definition until mandated. Or a state might not include DUI or fraud in reports because not required. All these choices mean the true volume of crime is higher than what’s officially aggregated.

• Resource constraints & triaging: In some cases, undercounting is not intentional per se, but a consequence of overwhelmed systems. If a police department doesn’t have enough detectives, they might classify some crimes as unfounded or close cases administratively to manage workload. For instance, a busy metro PD might receive 100 burglary reports but only officially record 80 and label 20 as “information only” if they believe they won’t be solved. This kind of soft underreporting isn’t easily visible but contributes to lower numbers. Political leaders might also set thresholds (e.g. instruct police not to take reports for thefts under $50), which was reported in some cities, meaning petty crimes simply don’t enter stats (this overlaps with policy changes factor).

Impact & Examples: The direct impact of agency underreporting is hard to quantify precisely (because by nature it’s hidden), but case studies give a sense: LAPD’s 7% overstatement of crime decline; Chicago erasing a handful of murders and several dozen felonies in a year; NYPD precincts secretly dumping complaints (Schoolcraft’s recordings detailed at least dozens of incidents in one precinct alone). A reasonable estimate from audits is that in some high-pressure big-city environments, official crime stats might be 5–15% lower than true incident levels due to misclassification. The Chicago example, where at least 10 homicides (out of ~400) were delisted in 2013, suggests about a 2–3% “improvement” in murder rate was fake. For violent assault, LAPD’s case implies a one-sixth undercount in serious assaults. These are not trivial – a 10% swing can turn a flat crime trend into a decline on paper. Nationally, the FBI data gap in 2021–22 is even more significant: because one-third of agencies (including some big ones) didn’t report, the FBI’s reported 1.7% violent crime drop in 2022 is not credible – indeed, BJS data showing a 46% victimization surge implies that nearly all of the “crime decrease” in the FBI figure was due to missing data. In other words, underreporting by agencies may have accounted for several percentage points of difference. A telling comparison: In 2022, violent crimes known to police (FBI) were about 1.31 million, whereas the NCVS estimated 5.7 million violent victimizations. Even accounting for homicides (which NCVS can’t capture), that gap partly reflects incidents never recorded by police. Some of that is public non-reporting (factor 4), but some is police not recording – the exact breakdown is unknown, but even if 10–20% of that gap is agency underreporting, it’s significant.

Geography: Underreporting by police tends to be a factor particularly in large cities with political pressure to reduce crime. New York, Chicago, Los Angeles, Philadelphia, Baltimore – all have faced accusations of fiddling with stats at times. Conversely, many smaller cities or suburban areas might have more straightforward reporting (less pressure and simpler data). This means national data could be disproportionately skewed by a few big departments “making the numbers look nice.” Regions with aggressive policy shifts often also had more incentive to show those policies “working” by touting crime drops – which could lead to optimistic classification. For example, Philadelphia under reform DA Larry Krasner initially reported declines in certain crimes, but later media investigations found some were just being handled differently administratively. In contrast, more traditional enforcement areas (say some Southern states) might report their crime straight, warts and all, because there’s a culture of transparency or no need to hide increases. This can paradoxically make the reform cities look safer on paper even if street reality is worse. We saw in San Francisco a paradox: property crime rates in official stats remained fairly stable mid-2010s, but residents widely reported a surge in car break-ins – it turned out SFPD had trouble keeping up with reports and often didn’t take reports if there was no suspect, meaning many break-ins went officially uncounted. Meanwhile, Texas cities which fully reported every broken car window as a crime may have looked higher in crime stats than SF, despite SF arguably having more actual incidents. Thus, underreporting can distort cross-city comparisons significantly.

Time Pattern: The incentive for data manipulation grew with the use of CompStat in the 2000s and the political emphasis on the crime drop narrative. So 2000–2015 saw some underreporting (many of the LAPD/NYPD/Chicago scandals occurred in the 2005–2015 window). It likely shaved a bit off the crime decline in those years (the crime drop was real, but may have been exaggerated by a few percent). 2015–2020: As crime ticked up in 2015–16, some cities may have felt pressure to hide that increase. Chicago’s scandal broke in 2014 (covering up 2013 data), suggesting they were trying to maintain an appearance of decline. Baltimore in 2015 was accused of underreporting robberies amid the unrest. However, also during this period some departments and media became more vigilant about data integrity (the exposure of undercounting led to reforms). 2020–2025: The chaos of 2020 actually made deliberate misclassification less necessary – crime patterns were so altered and data collection so fragmented that the stats were already unreliable. Additionally, many departments were dealing with technical issues (NIBRS) so they weren’t actively trying to deceive, they just weren’t reporting fully. One could argue political pressure in 2020–2021 was not to hide increases (since many leaders openly acknowledged the murder spike) but by 2022–2023, with crime as a hot issue, some may have returned to quietly massaging numbers to show progress. For example, New York City’s leadership in 2022–23 highlighted index crime drops, but later a review found the NYPD had “upgraded” a number of crimes after audits – revealing that the real 2023 crime count was the highest since 2006 once reclassified cases were included. This indicates ongoing adjustments that can mislead until caught.

Overall: Underreporting/Misclassification by agencies is a moderately significant factor. It likely accounts for a smaller share of the perceived crime drop than factors like avoidance or reduced enforcement, but in certain locales or years it has been crucial. We might rank this factor in the middle: it can easily contribute a 5-10% artificial decrease in reported crime under the right (or wrong) circumstances, and it was certainly at play in cities claiming record-low crime while independent audits showed otherwise. The key is that it undermines trust: if data is skewed, it’s hard to discern real trends. Next, we turn to the public’s role in reporting.

4.) Lower Public Willingness to Report Crimes

For a crime to be “reported” in official stats, the victim or a witness must notify law enforcement and a report must be taken. A longstanding issue is that many victims choose not to report crimes. Over the past decade – especially in some communities and especially after 2020 – public willingness to report crimes appears to have declined, further widening the gap between actual crime and recorded crime. Key aspects:

• Extent of Non-Reporting: The majority of crimes in the U.S. are never reported to police. According to the 2022 National Crime Victimization Survey, only 42% of violent victimizations and 32% of property victimizations were reported. This means roughly 6 in 10 violent crimes (like assaults, robberies, rapes) and 2 in 3 property crimes (thefts, vandalism, etc.) do not show up in police data at all. Non-reporting has always existed (fear of retaliation, thinking police can’t help, etc.), but these percentages represent a further decline from prior years. In 2011, for example, about 52% of violent crimes were reported; by 2021 it was down to 46%, and then fell to 42% in 2022. For property crime, reporting rates have hovered in the low 30s and dipped below 30% in some recent surveys. This downward trend means more crimes are flying under the radar now than a decade ago. If victims aren’t calling the police, those incidents effectively “don’t count” in the eyes of official statistics, thereby contributing to declines on paper.

• Reasons for Not Reporting: There are several reasons victims might not report, and some have become more pronounced recently:

  • Distrust or lack of confidence in police: High-profile incidents of police misconduct and systemic racism revelations have eroded trust, particularly in Black and Hispanic communities. After events like Ferguson (2014) and George Floyd’s murder (2020), surveys show trust in law enforcement dropped. A direct consequence: people in low-trust communities may choose not to involve police, even when they’re victimized, because they either fear hostile treatment or doubt the police will genuinely help. A seminal study by Desmond et al. found that after a notorious police beating in Milwaukee, 911 calls from Black neighborhoods fell sharply and stayed depressed for over a year – essentially, residents stopped calling for help as readily. Another study replicated this in a different city and confirmed a “significant decline” in 911 calls in minority neighborhoods after a police killing, indicating a loss of legitimacy. If fewer people call 911 to report assaults, domestic violence, or suspicious activity, fewer crimes get recorded. The RTI study (2024) notes this “decay in community trust” can have follow-on impacts on safety – one of which is underreporting of crime.

  • Perception that nothing will be done: Many victims don’t report lower-level crimes because they believe police won’t or can’t do anything. This perception has increased with some reality behind it: if car break-ins rarely lead to an investigation (as is true in many big cities now), victims may not bother filing a report for insurance or records. In San Francisco, for example, car burglary victims often state they didn’t call police because it’s known that such cases won’t be followed up unless there’s video evidence. Similarly, in places with lenient prosecution (Factor 5), victims might feel reporting is futile if the perpetrator will face no consequences. This became a common refrain among small business owners in cities like Portland and San Francisco – after repeated shoplifting incidents with no action taken, some stopped calling 911 and instead just tallied losses internally. That means those thefts vanish from official data. Underreporting due to apathy or cynicism thus grows in environments where enforcement is perceived as lax.

  • Fear of retaliation or involvement: In high-crime areas, victims (often of violent crime) may fear retaliation if they cooperate with police. Gang-controlled neighborhoods have historically had lower reporting for this reason (“no snitching” culture). If community-police relations deteriorate, this fear can intensify because police protection is seen as unreliable. Immigrant communities may avoid reporting crime due to fear of drawing attention to immigration status or simply language/cultural barriers. All these factors contribute to many crimes never being brought to the authorities. Post-2020, anti-Asian hate crimes spiked, but some Asian American victims did not report incidents out of distrust or language issues, potentially undercounting the true extent of hate crimes.

  • Normalization or self-help: Some people have become desensitized to certain crimes and no longer report them. For instance, if package theft (porch piracy) happens repeatedly in a neighborhood, residents might shrug it off as a cost of online shopping and not report every instance (especially if the value is small). Likewise, some communities have turned to private solutions – e.g. hiring private security or using neighborhood apps to share information – instead of calling police for every incident. This creates an alternate reporting channel (social media, etc.) that doesn’t feed into official stats.

• Evidence of Declining Reporting: Aside from the NCVS percentages, we have anecdotal and indirect evidence: In 2022, the share of people saying they reported their violent victimization to police fell significantly. Jeff Asher, a crime analyst, noted this drop and also that property crime reporting by victims hit a new low around 2021–22. The gap between NCVS (victim survey) and UCR (police data) widened in 2020–2022. As mentioned, NCVS showed a huge surge in violence in 2022 while UCR showed a decline. Part of that discrepancy is due to non-reporting: people told surveyors about victimizations that they never told the police about, in greater numbers than before. Another piece of evidence: 911 call data in some cities shows drops in certain types of calls. After 2020, several cities observed fewer calls for service for things like “person with a gun” or “shots fired” in some neighborhoods, even though objectively those incidents (judged by hospital data or ShotSpotter detections) had increased. This suggests people weren’t calling as often. Philadelphia’s 911 calls for violent crime reports dipped in 2021 even as shootings hit record highs – indicating many incidents only came to light when victims self-transported to ERs or not at all. Additionally, community surveys (Gallup, etc.) showed rising percentages of people saying they would be hesitant to call police except for the most serious emergencies, due to fear of how police might respond, especially among younger and minority respondents.

Impact on Crime Statistics: Lower public reporting directly reduces the number of crimes that enter official records, thereby making crime appear lower. If, say, only 1 in 3 thefts is reported, a decline from 1 in 2 reported a decade ago, that alone could cut reported theft rates by a third even if actual thefts stayed constant. We can attempt a rough quantification: From 2011 to 2022, violent crime reporting to police dropped from ~50% to ~42% (an 8 percentage-point drop, which is about a 16% relative decrease in reporting likelihood). Property crime reporting might have dropped a couple points (say from ~35% to ~32%). If actual crime stayed the same but reporting fell by that magnitude, reported violent crime would fall 16% and reported property crime ~9% purely due to behavior change. Of course, actual crime wasn’t flat – but this gives a sense that a sizable share of the reported crime decline is owed to people not calling in. Even more striking, by 2022 nearly 60% of violent crimes and 68% of property crimes went unreported. That is the highest non-reporting rate recorded in decades. Some portion of that increase in non-reporting undoubtedly happened post-2020 (given the trust issues and frustration with police). Thus, when official data in 2022 showed, for example, robbery offenses down, one reason could be that fewer robbery victims bothered to report muggings that year, especially if minor injuries or loss occurred.

Crime Type Differences: Public non-reporting disproportionately affects certain crime types:

• Sexual assaults have chronically low reporting (often under 30% reported) due to stigma and trauma; if trust in police or sensitivity declines, even fewer survivors come forward, which can falsely show a “decrease” in sexual assaults on record.

• Domestic violence is another underreported category; victims may fear consequences of reporting their abuser. If community supports falter (as during pandemic lockdowns) and police aren’t trusted, domestic assaults can spike in reality but not fully show up in data.

• Petty theft and property damage often have low reporting if insurance isn’t involved. As noted, package theft, minor vandalism, bike theft – many don’t call these in now. This can make property crime stats look better even if neighborhoods are plagued by small crimes.

• Cybercrime and identity theft are huge and mostly not reported to local police (people might complain to the bank or credit bureau instead). While not the focus of traditional crime stats, it’s worth noting that tens of millions of Americans experience fraud/identity theft yearly with minimal police reporting, meaning the bulk of “crime” in a broad sense is off the books.

Geographically, areas with strained police-community relations (some urban neighborhoods) likely see the biggest drops in reporting. Conversely, communities with good relations or more homogeneous demographics might maintain higher reporting.

For example, a theft in a middle-class suburb might still be reported for insurance, while the same theft in an inner-city area might just be shrugged off. This can skew city-level stats: cities with populations less inclined to report will show lower crime rates than cities where people dutifully report everything.

Chicago vs. Houston could illustrate this – if Chicago’s West and South side residents underreport crimes due to distrust, Chicago’s stats could understate its crime problem relative to a place like Houston where more people call things in.

After 2020, some surveys found African-Americans’ confidence in police dropped to a 22-year low and Hispanic confidence also fell, whereas White confidence rebounded a bit. That implies racial disparities in reporting may have grown, affecting cities differently depending on their demographic makeup.

Utah, for instance, with a relatively high trust environment and more homogeneous population, might have higher reporting rates (thus more fully counted crime) compared to, say, Los Angeles, where certain communities have low trust and high immigrant populations less likely to report.

Time Pattern: Public non-reporting has always been a challenge, but 2015–2025 saw a potential worsening. The overall crime decline up to 2014 might have made people a bit complacent; ironically, when crime is rarer, some may not bother reporting the occasional minor incident. But a bigger driver was the social upheaval: after 2014, with controversies over policing tactics, willingness to engage with police eroded in some quarters. After 2020, this effect likely peaked – the combination of civil unrest, calls to reduce policing, and actual instances of police withdrawal formed a feedback loop. In cities where police became less responsive (Factor 2), residents had even less incentive to report (why call if no one comes, or if they come but won’t do much?). Many U.S. cities experienced this dynamic from 2020–2022, leading to a sort of silent crime wave: more happened than the numbers showed. For example, car thefts skyrocketed in 2022 in many places, and while most are reported (for insurance), some portion might not be if the owner felt it pointless. Or take retail theft: in San Francisco, by 2021, numerous store clerks had stopped calling 911 for shoplifting unless it was large-scale, because they knew police response was minimal. Indeed, a California retail survey indicated a significant chunk of theft losses were not reflected in police reports due to underreporting by retailers. As a result, California’s official theft stats post-Prop 47 looked stable or down, even as viral videos suggested rampant stealing.

Overall: Lower Public Reporting is a highly significant factor contributing to perceived crime declines, likely ranking among the top factors in recent years. It particularly affects violent crime statistics (as a lot of that goes unreported now). We might rank it just below target-hardening and reduced enforcement, or on par, because without the public’s input, the system is essentially blind to a large share of offenses. The difference between reported and actual crime has grown, meaning our “record-low crime” may partly be an illusion created by underreporting.

5.) Policy Changes (Bail Reform, Decriminalization, Defunding, Prosecutorial Discretion)

Finally, a suite of policy shifts in the justice system during 2015–2025 has altered how crimes are defined, processed, or counted. These policies can create drops in reported crime that are policy-driven rather than behavior-driven.

We’ll discuss several: bail reform, decriminalization/legalization, defund-related reallocations, and prosecutorial discretion. Each has a distinct impact on crime stats:

• Bail Reform and Pretrial Policies: In the late 2010s, states like New York New Jersey, and Illinois implemented bail reforms to reduce pretrial detention for low-level offenses. The immediate effect of bail reform is on the jail population, but it also reverberates to police practices and crime reporting. For instance, New York State’s bail reform (effective Jan 2020) eliminated cash bail for most misdemeanors and non-violent felonies. As a result, after arrest, many defendants are released quickly without bail. How could this lower reported crime? One way is that police, knowing that certain arrests will result in a “desk appearance ticket” and release, might issue a citation instead of arresting (which might not always get counted as an index crime, depending on how it’s logged). Also, if repeat offenders are quickly released, some argue they might commit more crimes – but those additional crimes might not be reported by frustrated victims or might overwhelm the system leading to more selective reporting. However, bail reform’s more clear impact is on crime rates via definition: if fewer people are in jail, fewer are charged with bail violations or failure-to-appear (which are crimes in some jurisdictions). It also in some cases deprioritized arrest for offenses that would no longer result in detention. New York City saw a noticeable drop in the number of arrests for minor offenses after 2020, partly due to COVID and partly bail reform – e.g., the NYPD made 45% fewer arrests in 2021 than in 2019 for offenses covered under the reform (like theft under $1,000) as officers adapted to the new rules (some of this overlaps with Factor 2: reduced enforcement by officer choice). So bail reform indirectly contributed to lower recorded crime by changing incentives: police focus shifted to violent crimes, while many low-level crimes resulted in a report but no physical arrest (some of those might not get counted as effectively, especially if they are handled as civil matters). It’s hard to quantify nationally (since only some places did this), but in New York City, officials pointed to continuing declines in index crimes through 2020 (excluding the pandemic spike in shootings) and attributed it partially to “not criminalizing poverty” – in other words, fewer arrests (and thus fewer reported crimes) for things like turnstile jumping or shoplifting.

• Decriminalization and Legalization: This is a major factor that directly cuts crime stats by redefining acts as non-criminal. The most impactful example is marijuana legalization. Between 2012 and 2020, numerous states legalized recreational cannabis. As a result, marijuana possession arrests plummeted (down 36% in 2020 alone nationwide). In 2010, drug offenses (mostly marijuana) constituted a large portion of arrests; by 2020, those were far fewer. For example, in Colorado, marijuana possession charges dropped essentially to zero after legalization in 2014, contributing to a drop in overall crime reports (the FBI’s crime data doesn’t include drug crimes in index crimes, but total arrest numbers and Part II offenses fell). Similarly, some jurisdictions decriminalized other minor offenses: thefts under a certain dollar amount (treated as misdemeanors or citations rather than felonies), prostitution (some places shifted to diversion programs instead of charging), traffic offenses (e.g. making many traffic violations civil infractions rather than criminal misdemeanors). California’s Prop 47 (2014) effectively decriminalized drug possession for personal use (turning it into a misdemeanor citation) and raised the felony theft threshold to $950. The impact: arrest rates for drug possession in California collapsed after 2014 – an analysis found drug arrests fell by over 50% in the year after Prop 47. California’s property crime initially showed a slight increase in 2015 (perhaps from more thefts), but reported shoplifting did not rise correspondingly. In fact, one study by PPIC found shoplifting reports decreased ~2.2% post-Prop 47, which they attribute to fewer incidents being cleared or reported. Essentially, making something a misdemeanor (and instructing police to cite-and-release) can reduce the paper trail. The net effect of decriminalization at the national level is significant: consider that in 1990, hundreds of thousands of arrests were made for vagrancy, loitering, gambling, consensual adult sex offenses, etc. Many of these have been either decriminalized or deprioritized, so by 2020 those numbers are greatly reduced. The “crime” simply no longer counts as crime. This certainly contributes to the long-term decline in reported crime. It’s arguably a real decline in criminality (society decided those acts are not crimes anymore), but if actual behaviors didn’t change (people still do drugs, etc.), then the drop is statistical.

• “Defund the Police” budget changes: The slogan “defund” often meant reallocating some police funds to social services. In practice, few cities truly defunded in a large way – most made minor cuts or shifted some responsibilities (and some later refunded). However, where budget cuts did occur, they sometimes led to closing specialized units (like anti-gang or vice units) and curtailing certain enforcement programs. This overlaps with Factor 2 (reduced enforcement) but from a policy angle. For example, if a city eliminated its traffic enforcement division in 2021 in favor of unarmed civilian traffic monitors (an idea floated in Berkeley, CA), then moving violations would cease to be criminal incidents, lowering crime stats related to DUI or reckless driving enforcement. Another example: Some jurisdictions disbanded narcotics task forces or gang units as part of rethinking policing; as a result, fewer drug busts and gang conspiracy cases were made – thus reported gang-related crime might drop (or shift to federal jurisdiction). “Defund” in Minneapolis led to a notable shrinkage of the police force (through attrition), which as discussed meant way fewer stops/arrests. One could say the policy choice to not maintain or replace officers effectively caused a prolonged under-enforcement period with accompanying declines in recorded minor crimes. It’s hard to isolate pure budget policy from the general enforcement factor, but if we consider places like Austin, TX, which in 2020 diverted some of the police budget and halted cadet classes: Austin saw slower response times and reduced ability to investigate property crimes, leading to more crimes going unreported or unrecorded (citizens complaining of having to file reports online for burglary, etc., which some might skip). So the policy of shifting resources can impact data by making the system less able to capture crime. On the flip side, cities that rejected “defund” and instead increased funding (e.g. fund increases in Florida or hiring sprees in some suburbs) may have kept crime reporting robust (or even increased detection of crimes), which ironically could make their stats look worse compared to “defunded” cities that simply aren’t finding as much crime.

• Prosecutorial Discretion (“Progressive DAs”): In the past decade, many large cities elected reform-minded prosecutors who changed charging practices. Examples: San Francisco’s DA (Chesa Boudin, 2020–2022) stopped prosecuting many low-level offenses (like public urination, small-scale drug possession, prostitution) and sought alternatives to incarceration. Philadelphia’s DA (Larry Krasner) dismissed a higher percentage of drug and petty theft cases than his predecessors. Manhattan’s DA (Alvin Bragg), upon taking office in 2022, initially instructed staff to downgrade certain felonies and not seek prison for many offenses. How does this affect crime stats? While prosecution happens after a crime is reported, it can circle back in a few ways. First, if police know the DA won’t prosecute, they might not bother arresting or even officially recording the incident (this connects to Factors 2 and 4: police and public both demotivated to pursue reporting). Second, some crimes’ classification in stats can depend on charging – for instance, if an assault is initially recorded as a felony aggravated assault but the DA consistently downgrades such cases to misdemeanors, the police might start classifying borderline cases as misdemeanor assaults from the start (which could remove them from “violent crime” counts if only counting aggravated assaults). In Manhattan, it was reported Bragg downgraded 52% of felony cases to misdemeanors in 2022 (up from 39% in 2019). This means many serious incidents (robberies, burglaries, etc.) might end up categorized under less serious headings. If the crime stats are pulled from final charge codes, it would show fewer felonies. Additionally, Bragg declined to prosecute 35% more felonies outright compared to 2019 – those cases just disappear from the system, effectively reducing the count of prosecuted crimes (though the initial report might still exist). Prosecutorial discretion also led to policies like not prosecuting theft under $300 (Chicago under DA Foxx for a time had a higher felony threshold, similar to Prop 47) – which meant those incidents might only be handled as citations or not at all. Over time, this can normalize lower reporting: e.g., store owners know if the DA isn’t prosecuting shoplifters, they may stop calling police (Factor 4), leading to fewer reports. There is also the phenomenon of conviction and clearance rates dropping under these policies – for instance, Manhattan’s felony conviction rate dropped from 68% to 51%. If crimes aren’t resulting in convictions, one might argue those incidents might not be fully reflected or might discourage future reporting.

Impact on Data: Policy changes can produce sudden, sharp drops in certain crime statistics by administrative fiat. A clear example: when marijuana possession is legalized, those offenses drop essentially 100% in crime data. Given that in 2010 there were over 500,000 marijuana arrests and in 2020 there were ~350,000 (and even fewer by 2022), that change alone reduced the nation’s total arrests and drug crime count significantly (over 150,000 fewer “crimes” in a year). If one adds Prop 47’s effect – e.g., California saw overall larceny thefts initially spike a bit, but then as businesses and police adjusted, reported petty thefts leveled off; meanwhile felony drug arrests in CA went down by tens of thousands annually after 2014 – the net effect was to push reported crime lower than it would have been absent those policies. Another impact: recategorization can remove crimes from “index crime” totals. The FBI index (traditional Part I crimes) doesn’t count simple drug possession or quality-of-life offenses. By reducing emphasis on those, cities might ironically increase their Part I crime share (i.e., violent and property) while overall incidents could be up. But when even index crimes like burglary are redefined (e.g., shoplifting under $950 not considered felony burglary but misdemeanor shoplifting), it can lead to fewer index crime counts. California’s property crime rate post-Prop 47 initially ticked up, but advocates point out that some offenses moved to non-index categories. The nuance is complex, but suffice to say policy changes can remove large categories of offenses from the “crime” ledger, thereby showing reductions.

Regional Comparisons: Regions with aggressive reforms (mostly liberal cities/states) might show artificially larger crime drops in categories impacted by those reforms relative to places without them. For instance, San Francisco and Los Angeles vs. Houston or Miami: California’s Prop 47 (and later Prop 57, bail reforms) meant thousands fewer drug and theft charges; Texas and Florida did not have comparable decriminalization, so their stats still include those crimes. Consequently, California’s reported property crime in 2020s may appear lower in some respects than it would be if they counted theft under $950 as a felony (like Texas would count a $700 theft as a felony burglary, California counts it as misdemeanor shoplifting or not at all if no arrest). Another example: Illinois (ending cash bail in 2023) vs. Florida (no bail reform) – Illinois may see fewer failure-to-appear warrants (which are counted as offenses in some systems) and potentially fewer reported domestic violence cases if victims know offenders won’t be held (this last point is speculative). Meanwhile, Florida’s traditional approach means every offense is charged and counted, so on paper Florida might look worse on crime even if actual levels were comparable. Utah, a conservative state with low crime, didn’t enact big decriminalizations, so their minor drug offenses are still logged and prosecuted, whereas in neighboring Colorado many of those offenses vanished from the books after legalization. That makes cross-state comparisons tricky: a “drop” in drug crime in Colorado doesn’t mean fewer people doing drugs than Utah; it’s largely policy.

Time Pattern: The effect of policy changes has been growing, especially since 2014 onward. The first wave was marijuana laws in the early 2010s, then Prop 47 in 2014, then a host of progressive prosecutors in late 2010s, and bail reforms around 2018–2021 in several states. So 2000–2015 had some (e.g., early marijuana medical legalization and some sentencing reforms) but relatively minor effect on reported crime. 2015–2020 saw a lot of implementation: by 2019, about a dozen states had legal weed, several big counties had stopped prosecuting low-level pot even if not fully legal, and movements to not arrest for certain misdemeanors took hold (Seattle, for example, largely stopped prosecuting simple drug possession after a court ruling in 2018). Post-2020, even more jurisdictions followed (New York decriminalized marijuana in 2021, many cities adopted cite-and-release for low-level offenses, Oregon decriminalized personal possession of all drugs in 2021). Thus, each year a bit more of what used to be “crime” is no longer counted. If one were to quantify the national impact: The FBI’s overall arrest numbers dropped from about 13.7 million arrests in 2009 to 10.1 million in 2019 – a 26% drop. Pew notes that overall arrests dropped 29% from 2006 to 2018. Much of that is reduced enforcement, but a chunk is fewer things to arrest for (because of legal changes). The trend accelerated with an additional ~25% drop in arrests from 2019 to 2020 (due to COVID and policy). So by 2020, arrests were down ~50% from 2006 levels, outpacing the crime rate drop. This suggests policy (and enforcement) factors eliminated a lot of “paper crime.”

Overall: Policy Changes are a significant factor, intertwined with the others. They rank highly because they can instantly reshape crime stats (e.g., legalize X and X goes away from the data). We rank this factor as highly significant, particularly for drug and minor property crime trends. It might be slightly less general than factors like avoidance or non-reporting (which affect all crime categories), but in the domains it touches, its effect is profound.

---

IV. Regional Comparisons: Policy Shifts vs. Traditional Approaches (Crime Rates)

To illustrate how the above factors manifest differently, let’s compare some regions of the U.S.

Progressive Policy Cities (e.g. San Francisco, Los Angeles, Chicago, New York) vs. Traditional Enforcement Regions (e.g. Florida, Texas, Utah)

• Theft & Property Crime: In California (SF, LA), Prop 47’s raised theft threshold and reduced penalties corresponded with a drop in reported low-level thefts. Official data show no large increase in shoplifting reports after 2014, and by some measures, larceny rates per capita actually fell in San Francisco from 2015 to 2019. By contrast, Florida continued to treat felony theft at $300+, so their agencies would count many minor thefts as crimes. Florida’s property crime rate from 2015–2019 also fell, but one could argue it reflects more actual prevention (through policing and incarceration) rather than non-reporting. Interestingly, California’s property crime in 2020 was 44% lower than in 2007 according to state data, a point touted by officials. Florida’s property crime was also down (~40% decline over similar span). The difference is how it was achieved: California partly via fewer arrests and reclassification (there were simply far fewer low-level theft arrests after Prop 47), whereas Florida imprisoned repeat thieves (incapacitating them) and kept reporting consistent. The outcome in communities diverged: While SF’s stats looked good, residents and businesses complained of rampant shoplifting. Florida’s stats and public sentiment were more aligned (people felt safer and the numbers went down).

• Violent Crime & Homicide: Look at Chicago vs. Houston. Chicago adopted some progressive policies (early bail reform pilot, a state attorney who de-emphasized low-level crimes), and also had episodes of police pullback (after 2015). Houston, in Texas, maintained a more traditional approach (though Harris County did have bail reform litigation impacting misdemeanors). Both cities saw surges in homicides in 2016 and 2020. Chicago’s official violent crime numbers in late 2010s were somewhat muddied by the underreporting scandal we discussed and by very low clearance rates (meaning many shootings might only be recorded as “aggravated battery by shooting” with no follow-up). Houston’s reporting was straightforward. By 2022, Chicago’s reported violent crime was up sharply (e.g. shootings up, carjackings at record highs), and public trust low – many south side residents stopped reporting non-fatal shootings that didn’t hit anyone (hence “shots fired” incidents undercounted). Houston also had rising violence, but Texas’s ethos meant folks still called police and police responded; however, Harris County’s judges releasing some repeat offenders (a controversy there) may have fed perception of a crime wave. The differences come down to style: Chicago might have appeared to have a crime drop in some categories because fewer people bother to report or the police stats lag (Chicago’s 2021 crime report notably left out a lot of data due to a cyber hack). Houston’s data was more complete. This leads analysts to sometimes find paradoxes – e.g., a study might say Houston’s aggravated assault rate surpassed Chicago’s, when in reality Chicago likely had more but didn’t log them all.

• Drug Offenses: Compare New York City vs. Salt Lake City (Utah). NYC, under a reform prosecutor in Manhattan and with state decriminalization of marijuana in 2021, saw drug arrests free-fall. By 2022, NYPD narcotics arrests were a fraction of what they were in 2010. Therefore, NYC’s reported drug crime plunged. Salt Lake City, with traditionally strict drug laws and enforcement, may still be arresting people for meth, heroin, etc. at higher rates. If you look at crime stats, NYC might show “narcotics crime down 80% from 2000 to 2022,” while Utah shows maybe a steady or slight decline. Does that mean NYC solved the drug problem? No – it’s largely that NY shifted tactics (treatment, issuing summonses, etc.) and many drug users are simply not criminally processed now. Utah, having not liberalized drug laws (beyond some marijuana CBD allowances), still counts those crimes. So on paper, NYC looks much improved relative to Utah in drug crime, but usage rates might not be so different.

• “Defund” & Police Presence: Consider Minneapolis vs. Miami. Minneapolis significantly reduced its police force after 2020, and effectively stopped proactive policing in many areas. Miami did not defund and actually increased police presence in some high-crime areas. By 2022, Minneapolis reported overall crime roughly flat or slightly up (with big increases in some violent crimes but decreases in minor ones), whereas Miami reported a moderate increase in violent crime but also an increase in arrests. Minneapolis likely undercounted some offenses due to lack of response (many minor 911 calls went unanswered or took hours, so people gave up). Miami’s robust policing meant if anything they might have “over-reported” (or accurately reported) crime, catching things like illegal guns and stolen property that Minneapolis police were too short-staffed to pursue. This resulted in an interesting stat: As of 2023, Minneapolis’s violent crime rate per 100k was actually somewhat lower than Miami’s. But few would argue Minneapolis was safer – it’s more that Miami was actively confronting crime and logging it, while Minneapolis was in a reactive mode. Regions like Utah (which generally did not cut police funding and enjoys relatively good police-community relations) likely have more complete crime data; regions like the Pacific Northwest (Portland, Seattle), which went through intense defund debates and had officer shortages, likely underreport a lot of disorder and low-level crime. Indeed, Portland saw a massive spike in murders 2020-2022, but simultaneously many categories of minor crime in reports fell – not necessarily because it didn’t happen, but because policing capacity collapsed.

These comparisons highlight that jurisdictions with aggressive reforms often show steeper drops in certain crime stats (especially drug and property crime), but those drops can be misleading.

Regions with traditional enforcement show more consistent data but may appear “worse” on paper simply because they still criminalize and report behaviors that reform cities have essentially taken off the books.

When examining national trends, this means if big reform-heavy states have large weight (e.g. California, New York), the national “crime rate” will be pulled down by their lower reporting in various categories, even if actual national victimization isn’t as low. Conversely, increases in crime in those areas might not fully register.

One concrete example: Motor Vehicle Theft around 2020–2022. California’s approach under Prop 47 (treating many auto thefts as lower-level unless evidence of intent to sell) and police resource issues arguably led to undercounting. Florida, Texas, etc., treat auto theft seriously. In 2020–2022, auto thefts exploded across the country (theft of Kia/Hyundai). California’s reported motor theft increase was big, but some analysts suspect it undercounts the true number of attempts (some Californians no longer bother reporting if insurance isn’t covering, etc.). Meanwhile, Texas saw a very large jump in reported auto theft (because Texans report to claim insurance and police still make reports). If one were to compare, Texas might show a worse surge than California, but part of that is better reporting.

Overall: Comparing these regions suggests that factors 1–5 manifest differently but cumulatively mean that places with more “non-traditional” criminal justice policies have larger gaps between reported crime and actual crime. Traditional regions might have less gap. This doesn’t mean one approach is inherently worse or better in outcomes (that’s a normative debate), but purely regarding data distortion: the aggressive-policy regions likely contribute disproportionately to national underreporting in the 2015–2025 timeframe.

---

V. Estimating & Ranking the Factors’ Contributions to “Lower Crime” in the U.S.

Bringing it all together, we attempt to estimate the overall contribution of each factor to the observed crime declines and rank their significance.

1. Target Hardening & Avoidance Behavior (≈30–40% of the Decline in Property Crime Since 2000)

Over the past two decades, Americans have massively expanded their use of security cameras, anti-theft technology, upgraded locks, and alarm systems. They also engage in more “avoidance behaviors” (e.g., not going out late, staying away from dangerous areas, relocating to safer neighborhoods).

• Why It’s Mostly Real: These measures genuinely prevent crimes (especially property crimes) by reducing opportunities. A locked car or well-lit, camera-monitored storefront discourages theft.

• Approximate Share of the 2015–2025 Drop: For property crime, it’s the biggest chunk of true reduction—likely explaining around 3–5 percentage points out of the total 10–15% reported decline.

• Overall Rank: #1 or #2 in overall significance. This factor directly prevents crimes from happening, so it’s not simply “cooking the books.” However, it can paint a misleading narrative if we assume criminals just gave up—often, it’s that citizens hardened targets.

Key Note: Fewer crimes are actually happening, making this a real reduction. But it also means that if people had not invested in these security measures, we might see higher crime rates today—so the decline isn’t purely due to offenders changing.

---

2. Reduced Police Enforcement (≈20–30%)

Many police departments have pulled back on proactive arrests and stops, especially for minor offenses like drug possession, vagrancy, and petty theft. Post-2015 controversies and “defund” debates also led to fewer officers available to respond to low-level crimes.

• Why It Creates a “Paper” Drop: If cops stop arresting people for drug possession, those arrests vanish from official crime stats—even if drug use itself remains steady. Similarly, if fewer traffic stops are made, fewer “discovered” crimes (unlicensed guns, warrants) will be recorded.

• Approximate Share of the 2015–2025 Drop: Potentially 2–3 percentage points out of the total 10–15%. In places that heavily curtailed enforcement (e.g., Milwaukee saw an 82% arrest decline), the effect is even more stark.

• Overall Rank: #1 or #2 alongside Target Hardening. For certain categories (e.g., drug offenses, public-order crimes), reduced enforcement may be the main reason those crimes appear to “fall.”

Double-Edged Sword: While it makes crime stats look better, it risks emboldening offenders if they sense no consequences.

---

3. Lower Public Reporting (≈10–20%)

Victims increasingly choose not to call the police—due to distrust, fear of retaliation, or belief that “nothing will happen.” The Bureau of Justice Statistics (BJS) found violent-crime reporting rates dropped from about 50% to ~42%.

• Why It Depresses Official Stats: If the public doesn’t report crimes, they don’t get logged in uniform crime reports. This especially affects “soft” crimes like assault, domestic violence, and minor property theft.

• Approximate Share: We estimate this accounts for about 1–2 points of the total 10–15% decline. In specific communities with high mistrust of police, the underreporting could be much greater.

• Overall Rank: #3. It powerfully explains why the FBI’s reported decline may not match the reality that people experience or perceive (e.g., seeing theft but not bothering to call 911).

Nuance: Part of the reluctance to call police stems from Factor 2 (reduced enforcement). If people see no police response, they might not bother reporting future incidents.

---

4. Policy Changes (Bail, Decriminalization, Progressive Prosecution) (≈10–15%)

When acts are redefined as non-criminal—such as marijuana legalization or raising felony-theft thresholds—arrests and reported offenses naturally drop. Some district attorneys explicitly decline to charge low-level crimes, and bail reform means fewer arrests stick.

• Why It Lowers the Stats: If shoplifting under $950 is treated as a misdemeanor ticket or not enforced, it may never appear as a felony theft in the data. Similarly, decriminalizing marijuana possession removes it from the crime ledger entirely.

• Approximate Share: About 1–2 percentage points out of the 10–15% decline, nationally. In states with big reforms like California or New York, it could be higher.

• Overall Rank: #4, but highly interlinked with lower enforcement. Policy changes often drive the decision not to arrest or to release offenders quickly.

Note: Someone might argue that if an offense is legalized, it’s not a “crime” anymore. Others maintain it’s still morally wrong (e.g., shoplifting). From a data standpoint, those incidents simply vanish from “crime” tallies. (To me, many things that are “legalized” are criminal and should still count as crimes.)

---

5. Underreporting & Misclassification by Agencies (≈5–10%)

Police departments sometimes downgrade serious assaults to “simple assaults,” or they fail to submit complete data (as happened during the FBI’s switch to NIBRS). In rare but real cases, some homicides get labeled as non-criminal deaths.

• Approximate Share: On a national scale, perhaps 0.5–1 point of the 10–15% drop. Locally, this can be much larger if a city systematically hides crimes.

• Overall Rank: #4–5. Meaningful—especially in high-pressure departments—but usually less than the bigger systemic factors (enforcement, victim reporting).

NIBRS Gap: In 2021–2022, many agencies didn’t submit data, leading to a fake decline in the FBI’s national numbers. This might mask a real uptick or at least reduce confidence in official stats.

---

Estimated Power Rankings 2025

These factors often act in concert. For example, a city like San Francisco saw all of them: residents installing cameras and avoiding downtown, police pulling back on minor crimes, SFPD possibly undercounting some incidents or not reporting all data, citizens not calling about repeat thefts, and Prop 47 plus a progressive DA changing enforcement and prosecution.

The result is that SF’s official crime rate in certain categories dropped or stayed low, even as public frustration grew. In contrast, a place like Jacksonville, Florida (more traditional) mainly had Factor 1 (target hardening) contributing to real declines, with less of factors 2–5; so their crime stats might more closely reflect reality, but also showed smaller declines in petty crime (because they kept reporting them).

1. Rank 1: Target Hardening & Avoidance – A foundational, continuous impact that prevented innumerable crimes (especially property crimes) from 2000–2025, greatly lowering victimization and reports. This is a true reduction in crime opportunities, so not just an illusion – but it means the drop in crime was due to people’s actions to avoid it, not just criminals “giving up.”

2. Rank 2: Reduced Police Enforcement – Particularly post-2015, this has slashed the number of recorded offenses (notably drug and misdemeanor cases) and contributed to an apparent crime drop, while sometimes allowing actual crime to fester. The 82% arrest decline in Milwaukee and similar patterns show its power.

3. Rank 3: Lower Public Reporting – This became very important post-2020. As trust in police eroded, many crimes simply never reach police. The NCVS vs. UCR divergence in 2022 (46% more violent victimizations vs. ~2% drop in police data) is glaring. This factor means a lot of crime “went dark,” contributing to the perception of safety in stats that may not match reality.

4. Rank 4: Policy Changes (Decriminalization, Bail, etc.) – By redefining crime and altering the justice pipeline, these policies removed a considerable volume of incidents from the crime ledger (especially in liberal jurisdictions). E.g., marijuana legalization alone cut hundreds of thousands of arrests; Prop 47 reshaped California’s crime rates (property crime down 55% from 30 years ago in CA, partly due to reclassification). This is substantial, though somewhat overlapping with reduced enforcement (since policy often drives enforcement).

5. Rank 5: Police Misclassification/Underreporting – While very meaningful in specific cases (LAPD’s 7% undercount of violent crime), on the whole this is a smaller slice of the pie. It might be the “extra sugar” that turns a 5% drop into a 8% drop in a mayor’s press release. Nationally, aside from the NIBRS data gap which was temporary, most of the decline is explained by the above factors rather than outright cooking of books. Still, it’s a critical factor to monitor, as it does distort trends and can be significant locally.

(Note: One could argue reduced enforcement and avoidance are co-equal top factors – avoidance drove down actual crime significantly, while de-policing drove down reported crime significantly. They operate differently – one by reducing crimes happening, the other by reducing crimes recorded.)

---

VI. Reconciliation with Official National Crime Statistics in the U.S.

Official national crime statistics (FBI and BJS) indicate that we are much safer today than 20 years ago. This is true in many respects – the murder rate, for instance, despite the 2020 spike, is still below the 1990s. However, there are key discrepancies and likely distortions in the data that must be acknowledged:

• The FBI-reported crime drop (e.g. violent crime down ~49% since 1993) is partly inflated by underreporting in recent years. The dramatic long-term decline is real, but perhaps not as steep as the numbers suggest post-2015. The 2022 FBI report showing a violent crime dip should be taken with caution due to the missing 32% of agencies. Conversely, the BJS survey showing a large victimization increase in 2022 suggests that actual crime might have been rising even as reported crime fell. This divergence is one of the largest on record and points directly to factors 2–4 (enforcement, agency reporting, public reporting) creating an illusion of improvement.

• Crime type trends: Violent crime stats (especially for assaults and rapes) are more likely to be underreported now. Property crime stats might undercount the high-volume, low-value incidents that plague communities (like package theft, shoplifting). For example, the theft rate in FBI data (larceny) declined 54% from 1993 to 2022, but that likely doesn’t fully account for the new forms of theft (porch piracy, digital fraud) that aren’t captured well. The NCVS indicates theft victimization is still quite common and many go unreported. The motor vehicle theft rise in 2020–2022 was clearly captured in data (even with underreporting, stolen cars are usually reported for insurance). That’s why motor vehicle theft jumped out as an increasing category in both FBI and victim data. Meanwhile, burglary has plummeted hugely (75% drop since 1993), which aligns with target hardening (alarms, etc.) being very effective – a real decline. So in reconciling, one should see that not all crime types are equally affected by these factors: tangible, noticed crimes like auto theft and murder are fairly reliably tracked (and those did spike in 2020-22), whereas softer crimes like simple assault, minor theft, drug use are more prone to undercounting shifts.

• Geographic discrepancies: Official stats often present national averages, but as we saw, some cities might underreport more. This means national stats could underweight high-crime cities if they had reporting gaps. For instance, if NYPD didn’t fully report 2021 data, the national violent crime count would miss ~30,000+ incidents from NYC. The Council on Criminal Justice (CCJ) stepped in by compiling data from sample cities and found homicide up ~30% in 2020 and a further ~5% in 2021 in those cities, aligning with what we’d expect without underreporting. The FBI’s partial data initially obscured that until corrected. Therefore, to get true trends, we often look at independent analyses (CCJ, academic) to adjust for these distortions.

• Perceptions vs Data: The fact that Americans consistently believe crime is worse than it is can be partly attributed to these factors. People sense the presence of crime (or see viral videos of thefts, etc.), but official figures say “crime is down.” That creates a perception gap. In some cases, the public is actually detecting what the official data is missing. For instance, the public in SF believed shoplifting was out of control in 2021, even though SFPD numbers didn’t show a major increase – the reality was the data was not capturing all incidents. Similarly, nationally in 2020–2021, people felt crime rose (and indeed violent crime did), but some officials pointed to overall crime being down (because property crimes fell and many things weren’t reported). This can undermine trust in statistics.

Key discrepancies to highlight:

• 2020 anomaly: Official data showed overall crime down in 2020 (due to big drop in property crime) but violent crime (murder) way up – that is a genuine spike that stats captured for murder, but likely undercounted for nonfatal violent incidents since fewer were reported amid chaos. So the discrepancy is that official “violent crime” rate (which includes aggravated assault, robbery, rape, murder) rose ~5% in 2020, but the actual serious violence likely rose more (as indicated by hospital data for shootings). The murder count is hard data (can't easily hide bodies), so that is a solid measure – and it indeed went up ~30%. Aggravated assaults in FBI data went up a smaller amount; NCVS suggests many more assaults occurred than reported.

• 2022 divergence: FBI said violent crime down slightly, NCVS said way up. The likely truth is violent crime was roughly flat or slightly up, but incomplete reporting (missing agencies + victims not reporting) made FBI data show a dip. That discrepancy is a smoking gun for underreporting issues.

• Property crime vs fraud: Official property crime (burglary, larceny, vehicle theft) way down by 2020. But many analysts point out that some crime has shifted to fraud/online theft, which isn’t in those stats. Identity theft and cyber-fraud victimized 26 million Americans in 2016 (as cited in [59], line 314-321 onward), yet those don’t show up as “property crimes” in FBI reports. So part of the “crime drop” is that some traditional crimes (like pickpocketing) were replaced by new ones (phishing scams) that aren’t counted as index crimes. That’s more of a definitional discrepancy than an underreporting due to malfeasance, but it means official stats might understate total victimization.

Likely distortions: The data is likely most distorted for minor offenses and in jurisdictions with recent reforms or police-community tension. We should highlight that major violent crimes (murders) are least likely to be underreported – so the spike in murders 2020–2021 was real and clearly reflected. But categories like rape and assault are probably undercounted relative to actual occurrence (especially rapes, given only an estimated 1 in 3 is reported historically, and possibly even fewer now). Drug offense statistics were massively altered by policy: a drop in drug crime in stats might reflect legalization rather than a real drop in drug availability. Public order crimes (loitering, vagrancy) basically vanished from stats in many places after being decriminalized or deprioritized – but that doesn’t mean those behaviors vanished; we see homelessness crises and public drug use rising in some cities, but those don’t show as “crime” because of policy choice.

In reconciling the official narrative with these insights, one could say: Official stats show long-term progress in reducing crime, and that is true to a significant degree – Americans in general faced lower risks of being victimized in 2015 than in 1995, thanks in part to prevention and perhaps better social conditions. However, from 2015 to 2025, the picture is muddier. Some of the “progress” in crime reduction is due to adaptive behaviors and changes in the system rather than offenders disappearing. We must interpret post-2015 declines with caution:

• If crime is down, is it because of safety or because of silence? In many cases, it’s a bit of both. There are fewer crimes of certain types happening (thanks to target hardening), but also fewer of those that do happen are being recorded (due to enforcement/reporting issues).

• The “historic low” crime rates often cited (e.g. violent crime at half the 1990s level) don’t account for the dark figure of crime (the portion not reported). The dark figure seems to have grown lately (especially for violent crime going from ~50% to ~58% unreported). If one adjusted for that, the curve of violent crime might not look as low in 2022; it might show a plateau or uptick in recent years rather than a continued decline.

• Criminal justice data (arrests, prosecutions) are down in many categories, which some take as a sign of success (less crime) but could also mean less capacity or willingness to address crime. For example, California’s Governor in 2023 celebrated that incarceration and crime fell together in the state, framing it as reform success. Indeed, CA’s prison population dropped and crime stayed low by official measures. But our analysis suggests part of that is because many lower-level crimes simply never result in an arrest or incarceration anymore – they’re handled outside the traditional system. The question remains whether actual public safety is commensurately improved or if some offenses have shifted to a quasi-tolerated status.

Takeaway: The perceived reductions in crime in the 2015–2025 period are only partly due to actual declines in criminal activity. A considerable portion can be attributed to the five factors analyzed. By our estimates, well over half of the improvement in reported crime rates in many areas may stem from these factors rather than criminals committing fewer crimes. This doesn’t negate the real gains made – the U.S. truly did see major crime drops from the 90s to mid-2010s – but it means that after a certain point, further “declines” might be more on paper than in reality. Policymakers and the public should be aware of these dynamics: safer statistics do not always mean safer streets. For a clear picture, one should consider multiple data sources (police data, victim surveys, hospital data, community surveys) and track metrics like reporting rates and arrest rates alongside crime rates. Recognizing the role of avoidance, enforcement, reporting, and policy changes helps to demystify discrepancies. It also guides solutions: for example, rebuilding police-community trust could increase reporting (making stats worse but actually reflecting reality better, a necessary step to then truly reduce crime), or investing in proven target-hardening (like better car anti-theft tech) can genuinely reduce crime. Ultimately, honest interpretation of crime data requires adjusting for these factors. The official stats at the national level give a baseline – crime is far below the 1990s peak – but for the nuances of 2015–2025, one must look at these underlying influences. By doing so, we can identify where declines are misleading and ensure that efforts to reduce crime focus on actually preventing offenses, not just reducing the numbers on paper.

---

VII. Estimate: What % of crime reduction is true (real) vs. an illusion (pseudo-reduction) in 2025?

If you look at the aggregate U.S. “index crime” rate—through the lens of what the FBI has (imperfectly) published—there’s roughly a 10–15% decline on paper from 2015 to 2025 (counting all property and violent crimes together).

Exact figures vary by data source and by whether you include 2021–2022 “missing” agencies in the denominator.

But of that ~10–15% drop in reported crime since 2015, how much reflects genuine decreases in victimization, and how much is due to factors that artificially deflate the numbers?

This is difficult to estimate with precision… but a reasonable estimate is ~50%.

A. “Real” Crime Reduction (5-8%)

1.) Target Hardening & Avoidance (3-5%)

Example measures: more security cameras, better car anti-theft tech, alarm systems, people avoiding dangerous areas, etc.

• Effect: These measures genuinely prevent crimes (especially property crimes) from happening.

• Estimated Share of the 10–15% Drop: Roughly 3–5% (i.e., the majority of the real portion). In other words, if the official data says a 12% decline, about 3–5 percentage points of that may be credited to improved security/avoidance actually reducing offenses.

Note: This is not “cooking the books.” When fewer burglaries occur because people lock their doors and criminals move on or give up, that truly lowers both actual and reported crime.

2.) Other Legitimate Crime Reductions (1-2%)

Minor residual factors like demographic shifts (aging population), or pockets of more effective policing in the 2010s that carried momentum into the late 2010s.

These might make up the remainder of the “real” fraction, perhaps 1–2% of the total 10–15% decline.

B. Pseudo-Crime Reduction (4-7%)

Below are 4 main factors that artificially reduce or distort the official count. Together, they account for roughly 4–7% of the total 10–15% decline—i.e., about 40–50% of the overall drop.

1.) Reduced Police Enforcement (2-3%)

• Fewer stops, arrests, and proactive policing, especially for low-level offenses (drugs, petty theft, “quality-of-life” crimes).

• Relative Weight: Often seen as the single biggest driver of “paper” declines because if police stop making arrests, those crimes disappear from the stats.

• Approximate Share of the Illusory Decline: Up to half of that 4–7%—so maybe 2–3% of the total 10–15% drop.

2.) Lower Public Reporting (1-2%)

• Declining trust in police, belief “nothing will happen,” or fear of involvement leads victims to forgo calling 911.

• Violent-crime reporting dropped from ~50% to ~42%, and property-crime reporting dipped too.

• Approximate Share of the Illusory Decline: Possibly 1–2% of the total 10–15% drop.

3.) Policy Changes (Decriminalization, Bail Reform, “Defund,” Progressive Prosecution) (1-2%)

• When marijuana possession, low-level theft, or other minor behaviors are downgraded or effectively legalized, the official stats drop (even if the behavior persists).

• Felony-theft thresholds, non-prosecution of “low-level” crimes, reduced bail requirements all reduce recorded crime.

• Approximate Share of the Illusory Decline: Perhaps 1–2% of the total 10–15% drop (national average). Could be more in states with aggressive reforms (California, New York).

4.) Misclassification/Missing Data 0.5-1%)

• Some agencies systematically downgrade assaults or “disappear” homicides; big data gaps under the new NIBRS system also artificially deflate FBI totals.

• Approximate Share of the Illusory Decline: On a national scale, maybe 0.5–1% (i.e., 5–10% of the total drop). In specific big cities caught manipulating numbers, the local effect can be larger.

---

Overall:

1. Total Observed Drop (2015–2025): ~10–15%

2. Real Reduction in Criminal Victimization: ~5–8% (roughly half of the total decline)

3. Illusory/Statistical Reduction: ~4–7% (the other half), subdivided approximately as:

   • (A) Reduced Enforcement: ~2–3%

   • (B) Lower Public Reporting: ~1–2%

   • (C) Policy Changes: ~1–2%

   • (D) Data Misclassification: ~0.5–1%

In short: ~50% of the observed drop from 2015 to 2025 is “real,” driven mostly by target hardening and avoidance. The remaining half is “paper” reduction—mostly due to less policing of low-level offenses, fewer victims reporting, decriminalization, and some degree of data manipulation or missing submissions.

---

Important Caveats

• Overlap: These categories can overlap. For instance, policy decriminalization and reduced enforcement often go hand in hand; lower public reporting is sometimes driven by the same reasons police have pulled back. So the precise breakdown is approximate.

• Variation by Region & Crime Type: In a high-enforcement area (e.g., certain Florida counties), the “illusory” share may be smaller; in a more reform-oriented city (e.g., San Francisco, Seattle), the illusory share might be higher. Property crime is especially affected by target hardening, while violent crime is strongly affected by whether victims report.

• Legal vs. “Moral” Crime: If a jurisdiction legalizes something (e.g., pot possession under 1 ounce), it no longer counts as a crime. Some people would say, “That’s not a real crime anyway.” Others consider it still “crime in my book.” From a data perspective, it simply vanishes from the stats, thereby fueling an apparent decline.

In the grand scheme, the best rule of thumb is that only about 50% of the recent “official” 2015–2025 decline is genuine. The rest reflects the ways we measure, enforce, or define crime—rather than criminals actually giving up or disappearing.

These estimates are approximate and naturally vary by city, state, and crime type. In some progressive jurisdictions with heavy decriminalization or big enforcement pullbacks, the illusory share could be significantly higher; in more traditional, well-resourced areas with robust public cooperation, the illusion is likely smaller, and a bigger slice of the decline is “real.”

---

Final thoughts on crime trends in the U.S. 2015-2025…

Crime is actually dropping — even if somewhat misleading as a result of under-policing, underreporting, legalizing certain criminal acts, etc. (this is a great thing).

Technically: (A) avoidance & target hardening and (B) legalization of previously illegal/criminal activities — mean “crimes aren’t committed.” (People are making it too difficult for the criminals + formerly criminal actions are now legal… what do you expect happens to the “crime rate?”)

But this isn’t telling normal Americans the truth… most Americans don’t agree with legalization of activities like shoplifting or reforming bail or defunding the police (net effect: less policing, fewer arrests, similar criminal activity, misperception of “decreased crime”).

What should the U.S. be doing? Increasing police/law enforcement wherever needed. Americans shouldn’t have to “go out of their way” to avoid certain shoddy parts of cities (there should be ZERO HIGH CRIME PLACES… clean up the crime/gangs by any means).

Stop legalizing formerly illegal actions like shoplifting or reducing sentencing or letting people out without bail… this defies common sense. Open drug use, public intoxication, Prop 47 & no shoplifting enforcement under $950, “safe injection zones,” homeless encampments, et al. — none of this should be happening.

The south-side of Chicago shouldn’t be nicknamed “Chiraq” if the government were serious about cleaning up crime. NORMAL people shouldn’t need to flee cities for safer-haven suburbs to avoid higher-than-desired odds of criminal encounters… especially those who wish to live in cities but don’t feel safe due to crime (e.g. Memphis, St. Louis, Albuquerque, Baltimore, etc.).

In a healthy society, people shouldn’t need the latest security devices and/or stay armed-to-the-teeth in fear of intruders or home invasions.

Thanks to ultra-progressive faction of the left-wing (Democrat party) (infected with a few different strains of the “woke mind virus”), certain parts of the U.S.:

• No longer enforce laws properly/effectively

• Have legalized open drug use & various crimes (or downgraded enforcement & sentencing)

• Banned guns/self-protection

• Pushed soft-on-crime DAs/judges (they feel more empathy for criminals than actual victims)

• Pushed judges that have bad judgment

• Promoted the idea that disparities in crime rates by race automatically = “discrimination”

The egalitarian-fantasy-mind-virus has hamstrung the ability of the U.S. to maintain law and order in nearly all large cities.

As a result? There are countless shithole, downtrodden “hoods” infested with antisocial criminal gangs… Normal prosocial people have predictably fled these urban areas for the suburbs. Nothing but decay in these cities and some genuinely wonder why people “flee” (it’s not a mystery).

Many ultra-left progressives have a challenging time understanding how cities like: Singapore, Tokyo, Dubai, Doha, Seoul, and Taipei have almost zero crime.

The hamster-wheel-of-justifications in the progressive brain is truly something to behold… they’ll come up with rationalizations like more inclusivity or “cultural reform” or something. But it really comes down to 2 things (this is the truth but whether these dolts can accept the truth remains unlikely):

1. Aggregate genetic composition of residents (East Asians are, on avg., more tame, prosocial, higher IQ)

2. Actually enforcing the law with harsh punishment for serious crimes — regardless of woke narratives being pushed by ultra-left retards attempting to frame any disparities in arrests based on race (e.g. more people of X-Y-Z race are getting locked up) as “discrimination.”

In East Asian countries it’s a combination of both population genetics AND the decision to punish criminals harshly. In the Middle East? It’s mostly a strict enforcement with zero tolerance and harsh penalties.

The net result is the same: safer cities, booming businesses, and people unafraid of living in cities.

Regardless of the genetic composition of the citizenry, strict law enforcement and harsh punishment irrespective of whether the criminals are all males, young, white, etc. — is highly effective for making cities safe and desirable.

Certain groups: Females, Older Adults, East Asians — tend to commit fewer crimes, on average. I don’t think this would surprise anyone… you’d accept this as fact without even digging into the data.

Why? We know males have different hormone profiles, younger people take more risks, and certain ethnic/racial cohorts have different behavioral propensities downstream of evolved avg. genetics (some don’t understand or refuse to acknowledge this).

This does NOT mean that any specific case fits the “averages.” We may have a young Nigerian man age 20 with face & neck tats who is extremely prosocial, law abiding, working full time as a software engineer — and an old Korean woman age 60 who is a serial killer.

But on average, the stats paint a picture that egalitarians refuse to accept because it shatters their fantasized version of reality (i.e. that all sexes, all races, all ages — commit the same rates of crime and any major disparities in crime rates = MUST BE DISCRIMINATION!)

As soon as racial disparities are documented (especially if a non-white cohort looks bad by stats) — the woke mind virus gatling guns blast “discrimination” from all angles… it’s gotta be “RACISM!”

IN ANY REGARD: NOBODY IS SERVED BY KEEPING CRIMINALS ON THE STREETS. No matter your race/ethnicity, age, sex, etc. — you should want criminals locked up… if you abide the law, you have nothing to worry about… and you won’t need to go out of your way to avoid certain parts of cities/towns.

Policing harder, policing efficiently (AI/robots, specialized cameras, bait cars, stop and frisk, drone surveillance, honing in on high-crime areas, leveraging the military with kill orders for gang members, etc.), and locking criminals up with harsher sentences would make the U.S. far more comfortable for American citizens and tourists alike.

And then maybe we’d actually believe it when the FBI, Federal Agencies, State Agencies, or City/Local Agencies report: crime rates are significantly down.

Lastly, despite muddying of the waters re: true crime rates/stats in the U.S. in 2025, crime is likely still trending downward… hopefully this trend continues. But there’s still a lot of work to be done.

---

References:

1. Bureau of Justice Statistics: "Criminal Victimization, 2022"

2. Chicago Magazine: "The Truth About Chicago’s Crime Rates"

3. Council on Criminal Justice: Homepage / Research & Reports

4. Crime in America.Net: "Is US Crime Really Declining? Or Is It A Lack Of Crime Reporting?"

5. FBI: "Crime Data Explorer" & "FBI — Table 1 (Crime in the U.S. 2019)"

6. GrowSF.org: "The impact of Prop 47 on crime in San Francisco"

7. Hyundai/Kia Anti-Theft (Reuters): "Hyundai, Kia's anti-theft measures fuel steep dip in US vehicle thefts"

8. Insurance Institute for Highway Safety (IIHS): "Anti-theft software tamps down viral theft trend targeting Hyundai Kia vehicles"

9. John Roman (Substack): "Unpacking National Crime Statistics"

10. KQED: "Proposition 47’s Impact on California’s Criminal Justice System"

11. Los Angeles Times: "LAPD underreported serious assaults, skewing crime stats for 8 years"

12. Milwaukee / Urban Milwaukee: "Milwaukee Police Arrests Have Plummeted"

13. Newsweek: "Violence by Women Soars in California"

14. NORML: "FBI Report: Marijuana Arrests Plunge More Than 30 Percent in 2020"

15. New York Post: "Exclusive | Convictions plummet, downgraded charges surge under Manhattan DA Bragg"

16. Policy Brief (Public Policy Institute of California, PPIC): "Crime after Proposition 47 and the Pandemic"

17. PopCenter (ASU Center for Problem-Oriented Policing): "The Effects of Surveillance Cameras on Crime" (PDF)

18. Reuters: "NYPD report confirms manipulation of crime stats"

19. RTI: "The impact of police violence on communities: Unpacking how fatal use of force influences resident calls to 911 and police activity"

20. Security.org: "Do Home Security Cameras Deter or Prevent Crime?"

21. Statista: "Reported violent crime rate in the U.S. 1990–2023"

22. State of California (Governor’s Office): "Incarceration, violent crime, and property crime rates lower than they were 30 years ago"

23. The Pew Charitable Trusts: "Factors Contributing to the Crime Decline"

24. The Volokh Conspiracy: "Benefits and costs: crime, crime avoidance, crime control"

25. Wikipedia: "Crime in the United States"

26. Pew Research Center: "Crime in the U.S.: Key questions answered"

27. FBI Data (Pew / Additional Brief): "Drug Arrests Stayed High Even as Imprisonment Fell From 2009 to 2019" (Pew Issue Brief)

---