These statistically and causally quantified estimates are particularly important since recent debates have been fueled by individual anecdotes or general references to increases in crime, rather than establishing causal increases in crime because of bail reform, over and above what would have happened in the absence of bail reform. Advanced data analysis and causal inference can inform policy on timely and pressing questions of national impact.

Since New York City is a particularly urban environment, the study focuses on impacts on New York City by comparing to crime rates in large municipalities in the United States. To do so, co-author Andrew Koo merged openly available weekly data from 30 major cities to obtain the finest-grained dataset of crime rates. The study considers drug crimes, robberies, burglaries, theft, and assault.

Importantly, it is not enough to simply compare crime types before and after the reform in New York City. Though crime may be higher after the reform, this can be explained by a shared rise in crime—for example, due to national economic trends—that would have occurred anyway in the absence of bail reform. Instead of comparing crime rates in New York before and after the reform, the researchers use state-of-the-art causal methodology, the synthetic control method, to compare the before-and-after trends in crime rates of New York City to a “synthetic” New York City that emulates what would have happened to New York in the absence of the intervention. This “synthetic” New York is a certain weighted average of other large municipalities. This method lets researchers isolate the impacts of bail reform specifically on crime.

Although the analysis focuses on New York City, a growing body of evidence shows that bail reform doesn’t significantly increase crime overall. These results add to this body of evidence and have implications nationwide for pretrial policy. In fact, a recent injunction temporarily ended Los Angeles County’s use of cash bail on constitutional grounds, while in Illinois, the implementation of a recent bail reform act was blocked on legal procedural grounds.

Although the specific findings of this study are on bail reform, it exemplifies how causal inference can rigorously inform executive decision-making about the actual impacts of difficult decisions on real outcomes. This is relevant both in business and policy. It can be difficult to randomize policy or important business decisions. Despite this, advanced data analysis can handle these situations in analyzing the impacts of decisions without randomization.