Improving Housing Inspections

Last year, New York City’s Department of Housing Preservation and Development (HPD) received almost 542,000 complaints regarding housing conditions, including more than 334,000 complaints about “emergencies.” In response to those complaints, and its own analytic predictions about building conditions, HPD completed almost 700,000 inspections of apartments in privately owned buildings. Yet HPD issued notices of violations (a finding that conditions required correction) in only a small fraction of the inspections generated by emergency complaints. A large number of inspections thus may have been unwarranted, and a number of complaints might have been addressed more efficiently. One cost of those inefficiencies is substantial delays – an average of 12.4 days to “close” emergency complaints.

This picture immediately raises questions about whether HPD’s largely complaint-driven system for detecting housing quality problems and enforcing the housing maintenance code is the most efficient way to ensure the safety and health of the City’s renters. The issue plagues many cities across the nation and around the world. We propose to use insights and methodology from a variety of disciplines – data science, economics, health, law, machine learning, and public administration – to analyze how cities can more effectively identify housing code violations and fairly and efficiently deploy their code enforcement resources.