A Deeper Dive into California’s “Urban Crime” Rates

A recent report by the Center on Juvenile and Criminal Justice (CJCJ) has been making headlines with a bold claim that “California urban crime declined in 2020 amidst social and economic upheaval.” While the CJCJ report does not convey direct falsehoods, the interpretations are misleading without proper context.

First, the report examines the crime rate in California’s 72 largest cities (i.e., cities with populations in excess of 100,000) over time from 2010-2020 and from 2019-2020 (p. 2). A key finding from this section is that crime rates fell an average of 7% throughout 2020 to a “historically low level.” On its face, this sounds promising. However, this statement is a bit misleading, mostly because not all of these cities experienced a crime drop. In fact, 24 cities actually showed an increase in reported crime rates, lead by Santa Maria (+68%), Fremont (+24%), and Fresno (+15%). Of the 48 cities experiencing a decrease in crime rates, the leading cities were Santa Clara (-31%), Vacaville (-28%), and San Bernardino (-24%). Further, only some crime types fell, like robbery (-15%) and larceny/theft (-16%). Other types of crime increased, including homicide (+34%), assault (+10%), and motor vehicle theft (+10%). Given that crime didn’t fall universally, the trends are difficult to interpret. The findings in California tend to mirror nationwide estimates that also show decreases in robbery (-9%) and larceny/theft (-16%) alongside increases in homicide (+30%), assault (+6%), and motor vehicle theft (+13%).

The authors note that California was already “enjoying a period of record-low crime,” from 2010-2020, but this statement might be misleading. Among the 72 cities, overall crime rates did fall 14% between 2010 and 2020, but not for all crimes, and not for all cities. On average across the 72 cities, there were decreases in burglary rates (-36%), robbery rates (-30%),  and larceny/theft rates (-14%). However, at the same time, there were increases in assault rates (+18%), homicide rates (+15%), and motor vehicle theft rates (+4%). The authors argue that in comparison to 2010, “most communities were safer in 2020.” While the definition of “safe” is subjective, it seems crude to automatically imply that most communities were “safer” when homicides and assault rates had increased during this same time. The changes in California from 2010-2020 mirror some, but not all, of the larger nationwide trends seen from 2010-2019: at a national-level, there were decreases in rates for burglary (-51%), robbery (-31%), and larceny/theft (-18%), but homicides increased by 15%. For rates of assaults and motor vehicle thefts though, national trends from 2010-2019 actually indicated a slight decrease (-1% and -2.4%, respectively) in contrast to the increases seen in California (+18% and +4%, respectively).

When looking at different cities within California, changes in crime rates from 2010-2020 vary quite a bit, ranging from -46% in Visalia to +22% in Irvine (p. 12). This suggests there is a lot more going on at the local level that is affecting crime rates. The authors agree that “crime rates and trends varied widely from city to city,” but they don’t elaborate on how this could impact their implications. The report acknowledges the “unusually wide variation” but dismisses it as being due to the pandemic exerting “both upward and downward pressures on the crime landscape.” While true — the pandemic likely did have myriad impacts on crime — this seems like an overly simplistic explanation that in reality is complicated by things like geography, population density, police relations, and local politics. These types of factors have been known to influence crime rates in the past and are very important for understanding crime trends, especially when rates seem erratic. This concept is elaborated on in the FBI’s guide to using the Uniform Crime Report.

Authors use the modest reductions in crime from 2010-2020 as a justification for large scale criminal justice reforms, including the 2011 passage of Public Safety Re-alignment (AB 109) and 2014 passage of Prop 47 (p. 2). In sum, authors claim that these types of reforms are not a factor driving increased crime. They justify this based on a few things, the first one being that the 2020 increase in crime was due to “abnormal conditions that year.” While 2020 certainly was an abnormal year, that doesn’t necessarily mean that prior legislation didn’t impact crime rates. Evaluating the impact of these reform efforts is difficult, and even the most rigorous studies have a hard time doing so. Part of the reason for this is because when numerous reforms take place, it is difficult to isolate the impact of each one individually. Second, there are many other factors impacting crime rates that are difficult to control for, making it even more difficult to isolate the impacts resulting from policy alone. Thus, it seems a bit bold to conclude that these reforms had no impact on crime rates when the analysis used to make that conclusion did not use a rigorous approach in isolating these effects. According to the authors, the wide variation in crime rates among cities is further support that state-level policies did not impact the crime rate. The argument behind this is that wide variation from city to city is more likely reflective of local practices rather than state policies. In other words, state-level reforms likely did not increase crime because if they did, increases would be more similar across different cities. The issue with this interpretation is two-fold. First, just because local policies and practices have an effect on crime rates does not necessarily mean that the state-level policies had no impact. Second, the exact thing that AB 109 did was transfer people from state prisons to local custody and supervision, therefore impacting local services. In this way, it is possible that state-level reforms had downstream impacts at the local level, and could have potentially had an impact on the variation across cities.

Regarding the impact of state-level reform on crime, the authors contend that the types of crimes increasing during 2020 were “not the ones targeted by recent criminal justice reforms and are unlikely to reflect lagging effects of law changes.” However, Prop 47, passed in 2014, reclassified several drug and property offenses from felonies to misdemeanors, which means that the large reduction in property offenses could be related to a documentation issue rather than a prevalence issue. If some crimes were previously felonies and are now being classified as misdemeanors, obviously the numbers and rates for these felony crimes would decrease due to the changes in statute. In fact, property crimes actually did not decline between 2010-2013 and only started declining after the passage of Prop 47. Thus, it seems more plausible that decreases in property crime are reflective of how crimes are being documented rather than actually reflecting “real” drops in the prevalence of crime. Further, decreases in robbery, burglary, and larceny were also apparent in nationwide trends, suggesting that California is not an anomaly in this regard. If anything, the fact that California’s motor vehicle theft rates increased sets them apart from nationwide trends. A more rigorous research study from Temple University actually supports this idea, with findings revealing that AB109 was associated with increases in larceny/theft and Prop 47 was associated with increases in motor vehicle theft.

Relatedly, the authors point out that despite increases in homicide rates, “Los Angeles [county]’s homicide rate remains near historic lows” (p. 5). However they fail to mention what the historic low is, providing no context for this finding. They use the erratic changes in Los Angeles county’s homicide rate as further justification that criminal justice reforms are not responsible for the trend. Unfortunately, they base this portion of the findings on Los Angeles county, not Los Angeles city. This is problematic because county-level crime rates were not used in earlier portions of the report. Statistics presented in the first part of the report are based on city-level statistics while statistics presented in the Los Angeles-only section are based on county-level statistics. County-level and city-level crime trends are very different from each other, namely because county-level data tend to reflect wider geographical regions than city-level data do. The two really should not be used interchangeably for this reason. Los Angeles county’s changing pattern of homicide may reflect erratic patterns in homicide, but this finding is somewhat superfluous to the other findings of the report and it does not strengthen the argument.

The report’s main findings center around this idea of “urban crime,” which is loosely defined based on the crime rates of California’s largest 72 cities, i.e., those with populations exceeding 100,000. Changes in crime from 2000 through 2020 were calculated based on crime rates as reported to the Uniform Crime Report. The reliance on crime rates, rather than crime counts, is ideal because rates account for population size whereas counts simply report on the frequency of crime. However, the authors do not rely on rates consistently. When exploring the preliminary 2021 data, authors attempt to examine changes in crime relative to the same time period of 2020 for California’s seven major cities (p. 3). To assess changes from 2020-2021, the authors calculated the percent change in crime counts (aka, the number of times a crime occurred) rather than calculating the percent change in crime rates (aka, the number of times a crime occurred per 100,000 people). This makes it somewhat confusing to compare 2021 to 2020 as currently presented.

The difference between a rate and a count is important to understand for two reasons: 1) population size has a direct impact on the count (frequency) of crimes occurring in an area, and thus needs to be accounted for, and 2) small changes in crime counts can sometimes translate to large percentages that can be misleading. For example, an increase from 5 homicides to 10 homicides is technically a 50% increase in the number of homicides. In a population of 100,000 though, this would translate to a 5% increase in the homicide rate, or the number of homicides per 100,000. In a city with 500,000 people, an increase from 5 to 10 homicides would translate to a 1% increase in the homicide rate. Rather than using rates to examine preliminary 2021 data, the CJCJ report shows percent changes in the counts of different crimes. The only real finding from the 2021 data is that “trends are erratic and inconsistent,” but ironically enough, the way results are tabulated make trends look even more erratic than they actually are. For example, per the report, percent changes in homicide counts ranged from -10% in San Diego to +135% Oakland. It is possible to convert this to a rate by obtaining the UCR data (cited in the references section, p. 10). Upon doing this, it appears the percent changes in the homicide rate only range from -1% in San Diego to +7.87% in Oakland. Even after converting the numbers to a rate, the early 2021 trends do still appear somewhat erratic, albeit less so.

Overall, the CJCJ report appears technically accurate, but there are several reasons why these findings might be misrepresenting the data, as discussed above. At best, the report’s results appear overstated and should be interpreted with caution given these concerns.