Philadelphia murder rates rise due to lenient sentences sought by progressive prosecutor Larry Krasner

As the progressive prosecutor movement grows in popularity, we see more and more policy changes that reduce penalties for certain crimes. One of the common themes is de-prosecution, or the discretionary decision to not prosecute certain criminal offenses. Another aspect of de-prosecution involves reducing the severity of punishment for individuals who are prosecuted. The movement came about due to the belief of many progressives that mass incarceration actually increases crime through supposed “criminogenic” effects. That is, they believe that people who serve long periods of time in prison will adapt to that culture and learn certain behaviors that will make them worse criminals. However, opponents argue that de-prosecution policies don’t hold offenders sufficiently accountable, and will only encourage more crime as offenders learn that there are little to no consequences for their behavior.

In Philadelphia, de-prosecution began in 2015 with District Attorney Seth Williams. This resulted in a substantial decline in both new cases prosecuted and sentencings (particularly for drug possession, drug trafficking, and felony possession of firearms), a trend that accelerated when District Attorney Larry Krasner took office in 2018. At the annual Federalist Society Convention last year, Krasner boasted that his policies are “on the side of the data,” vehemently denying that de-prosecution increases crime. However, a 2022 study published in Criminology and Public Policy refuted Krasner’s claims. The study, conducted by Thomas Hogan, revealed a causal link between de-prosecution and increased homicides in Philadelphia.

Hogan used data from other large cities to create a “synthetic Philadelphia” that could be used for comparison. The data were extracted from federally maintained law enforcement statistics, administrative state- and county-level data, standard economic and demographic factors, and public-source data regarding prosecutiorial regimes. Administrative court data were used to track the number of prosecutions while Sentencing Commission data were used to track the number of sentences. Homicide data was gathered from supplemental homicide reports (SHR) collected by the FBI’s Uniform Crime Report (UCR). Homicide data were very detailed and included information on victim, offender, and case characteristics.

Drawing from the largest 99 cities in the United States, Hogan used various sources of publically available data (e.g., agency websites, funding reports, statements to media, etc.) to categorize city prosecutor offices as progressive, traditional, or somewhere in the middle (see Table 1 of the original article for a listing of cities and how they were categorized). Offices were coded as “progressive” based on a checklist representing the 15 most salient factors associated with progressive prosecutors (per past research). To qualify as progressive, the prosecutor was required to meet at least 10 of the 15 factors. These factors included things like stated intention to prosecute fewer cases, preferences to end cash bail, support for eliminating mandatory minimums, or the belief that the criminal justice system is racist (see the full checklist in the Appendix of the original article).

Among the cities that were identified as “traditional” (i.e., those without de-prosecution policies), a subset of cities were selected for the comparison group if they were similar to Philadelphia on certain factors (e.g., population size, median household income, existing crime rate, etc.). Groups were matched based on their data from 2010-2014, i.e. the “pre-period,” when there were normal volumes of prosecutions and sentencings. By putting this information into an algorithm, Hogan  generated a synthetic comparison group that was statistically similar to Philadelphia in terms of baseline crime/homicides, clearance rates, number of prosecutions/sentencings, median income, population, and other factors. The only difference between groups was that synthetic Philadelphia did not experience de-prosecution during the “post-period” (i.e., 2015 to 2019). In other words, synthetic Philadelphia estimated the impact on homicides if de-prosecution had not occurred in Philadelphia. Changes over time could then be compared between the two groups to determine whether reductions in new prosecutions and sentencings impacted homicides. From a research standpoint, this is a rigorous design that is strong enough to determine causal effects (ranked as a 3 out of 5 on the Maryland Scientific Methods scale).

The years 2010-2014 represent the “pre-period,” before de-prosecution began in Philadelphia. The years 2015 through 2019 represent the “post-period” (2020 and 2021 were intentionally excluded so that the pandemic did not complicate results). The comparison between the pre-period and post-period allowed for a “time series analysis,” which is an analytic method used to examine temporal trends before and after a specific cut point. The time series model tested whether the number of homicides changed significantly over time, comparing Philadelphia with synthetic Philadelphia. There were two cut points: post-2015, after de-prosecution began, and post-2018, when de-prosecution was accelerated by Larry Krasner.

During Seth Williams’ term as district attorney, sentencings declined by 39% (from 7,252 to 4,423) while prosecutions declined by nearly 22% (from 14,401 to 11,304). This trend accelerated in 2018 after Larry Krasner took office. During Krasner’s term, sentencings declined by more than 50% (from 4,423 to 2,195), and prosecutions declined by nearly 14% (from 11,034 to 9,514).

During the entirety of the post-period (2014-2019), sentencings dropped by 70% (from 7,252 to 2,195), and prosecutions dropped by 34% (from 14,401 to 9,514). This equates to about 1,011 fewer sentencings per year and and about 815 fewer prosecutions per year. At the same time, homicides in Philadelphia increased by more than 30% (from 248 to 356). Meanwhile, all major demograpic trends (e.g., median income, population size) remained stable. In addition, the overall length of sentences did not change by large degree. Philadelphia’s tendency to sentence below the standard range of the sentencing guidlines did not change much throughout the study period.

Time series results revealed a statistically significant difference in homicides between Philadelphia and synthentic Philadelphia. In other words, de-prosecution in Philadelphia in the mid-to-late 2010s was causally linked with a statistically significant increase in homicides, estimated at an additional 75 homicides per year from 2015 to 2019. When looking at the post-Krasner years (i.e., 2018 and 2019), the impact was even greater, with estimated increases of more than 100 homicides per year. During the entire five year post-period, there were 373 additional homicides in Philadelphia relative to synthetic Philadelphia. Reductions in sentencings were strongly correlated with homicides, particularly regarding sentencings for felony gun crimes. The relationship between homicides and number of prosecutions was also evident, though not as strong as the former. On a scale from 0-1, with 0 representing no correlation and 1 representing a complete correlation, the correlation for felony gun crime sentencings was 0.74, compared with 0.68 for overall sentencings, and 0.54 for prosecutions.

Figure 1 below (retrieved from the original study) gives a visual illustration of the association between new cases prosecuted, sentencings, and homicides.

Conclusion

De-prosecution policies in Philadelphia were significantly associated with increases in homicides, amounting to about 75 additional homicides per year. Reductions in sentences for felony gun crimes were the primary driver of this relationship. This indicates that there is probably a trade off between de-prosecution policies and the downstream impact on public safety.

In Philadelphia, de-prosecution was aimed at both felony and misdemeanor offenses but with an emphasis on cases involving drug posession, drug trafficking, and federal gun posession. However, not all de-prosecution policies look the same. Future research should seek to look into the types of progressive policies that are being enforced (e.g. do they pertain to misdemeanors, felonies, or both) and the decision point where the intervention occurs (e.g., pretrial, conviction, sentencing) to better understand whether some types of reforms are more harmful than others.