Association for Public Policy Analysis & Management

Crime imposes substantial pecuniary and social costs on society. In the United States alone, property crimes totaled approximately $50 billion from 2013 to 2016. These estimates understate the social cost of crime because they do not account for additional public and private investments in security to mitigate the risk of being victimized. Examples of security investment encompass a range of public and private expenditures, such as the installation of security cameras, alarms, or outdoor lighting. In the standard economic model of crime (SEMC), criminal incentives are determined by the risk of detection, the economic returns to crime, and the best available outside options to criminal behavior. While intuitive for simple changes, the precise channels through which security investments alter criminal incentives are complex and, consequently, the model’s predictions for these strategies are non-obvious a priori. Hence, assessing the net benefits associated with specific security investments will require a detailed investigation of the potential channels through which criminal behavior is affected.

We provide a novel empirical assessment of the effectiveness of street lighting technology as a criminal deterrence strategy through changes in visibility. Our study uses variation from a natural experiment in Los Angeles. From 2009 to 2012, Los Angeles carried out a program converting a majority of the city’s street light inventory to light emitting diode (LED) technology. The exogenous introduction of LED technology provides useful variation for identifying the effect of visibility on criminal activity because LEDs outperform older lighting technologies in a range of visibility metrics. Using a difference-in-differences (DD) identification strategy, we exploit variation in the location and intensity of retrofits to test channels through which visibility affects criminal activities.

Initially, we only consider property crimes because, under the SEMC, increased visibility could increase both the expected costs and benefits of crime. It is unclear which channel is dominant. Our preliminary results indicate the elasticity of property crime rates with respect to LED intensity is 0.1. So, a 1% increase in the intensity of LEDs corresponds to a 0.1% increase in property crime rates. This result is robust to a variety of treatment definitions and specifications. Next, we consider the impact of visibility on violent crimes, where visibility could increase the expected costs of committing a crime but unlikely influences the expected benefits. We find that a 1% increase in the intensity of LEDs corresponds to a modest 0.03% decline in violent crime rates.

Our study illustrates the complex relationship between visibility and crime and suggests security investments in outdoor lighting may produce unintended consequences. On one hand, our results suggest that security investments in innovative lighting technologies may reduce violent crime rates. On the other hand, we find improved visibility from outdoor lighting technologies may enable criminals to more easily identify high-value targets and, consequently, increase property-related crimes. More rigorous analyses are important for informing policymakers by uncovering the precise impact of security investments with unclear effects. In future extensions, we explore the impact of visibility on the spatial and temporal distribution of criminal activity.