New Orleans Brings Data-Driven Tools to Blight Remediation

The city’s successful BlightSTAT performance management approach has been the backbone of efforts that resulted in the elimination of over 15,000 blighted units from 2010 to 2015.

by Katherine Hillenbrand / October 12, 2016
New Orleans Skyline flickr/Paul Taylor

This story was originally published by Data-Smart City Solutions.

Since New Orleans began its recovery after Hurricane Katrina, the city has been tackling its problem with blighted properties in a variety of ways. Reducing blight has been one of Mayor Mitch Landrieu’s top priorities since he took office in 2010.

The city’s successful BlightSTAT performance management approach has been the backbone of efforts that resulted in the elimination of over 15,000 blighted units from 2010 to 2015, accomplished through a mix of demolition, sale, and owner repairs. BlightSTAT, along with the city’s other performance management efforts, is led by the Office of Performance and Accountability (OPA), the city’s team dedicated to improving services and delivering results through data. The OPA team is responsible for both performance and analytics, which director Oliver Wise points to as a “huge advantage” because the team is able to identify use cases from being deeply embedded in operations. The OPA, in collaboration with the city’s Department of Code Enforcement, has continued to apply the latest data-driven tools to advance the city’s goal of eliminating blight.

NUDGING FOR COMPLIANCE

The OPA team recently concluded a successful pilot effort to use behavioral science to improve voluntary homeowner compliance. When the city receives 311 complaints about potential code violations at an address, it needs to research and inspect the property before moving it forward in the process. Through What Works Cities, Bloomberg Philanthropies’ program to help 100 mid-sized cities use data and evidence to improve the lives of residents, New Orleans received technical assistance from the Behavioral Insights Team to apply “nudging” to the problem. The team tested adding a new initial step of sending homeowners a letter stating that a 311 complaint had been made about their property. The pilot was successful, and the letters meant homeowners were more likely to voluntarily bring their properties into compliance before any additional intervention by the city. The letter alone increased the occurrence of finding no violation at inspection by over 6 percentage points. This pilot program has now been fully implemented within the Code Enforcement Department, saving the city the equivalent of approximately the cost of one full-time inspector. 

A DECISION SUPPORT SCORECARD 

If the property owner does not voluntarily bring a reported property into compliance, the resulting process for handling a blighted property involves a number of steps.  When the city’s Department of Code Enforcement identifies and inspects a blighted property, it then conducts a hearing against the owners, who then have another opportunity to bring the property into compliance. If they do not and they receive a guilty judgment, the city has the legal authority to potentially demolish the house or foreclose on the property and sell it at auction. Historically, the process has come down to one director’s decision-making, honed by years of experience.  The decision is complex, as the city must take into account the property’s location, historical significance, market interest, condition and other factors. Over time, a substantial backlog of over 1,500 properties awaiting a decision had built up because the inspections and hearings were occurring more rapidly than the abatement decisions.

City administrators noticed this problem and tasked Oliver Wise and the OPA team with finding a way to distribute the decision-making responsibility without sacrificing the quality and rigor of those decisions. The process started out with low-tech teamwork: a group of subject experts from the OPA manually scored over 600 test case properties on decision factors identified through interviews of the staff. OPA was then able to hone the list of criteria, determine how they affected the outcome and begin the process of building a data-driven solution.

Working with Enigma, a data science startup, OPA tested machine learning algorithms to see if a model could be trained on that data. Based on the tests, OPA chose a logistic regression model and built an in-house tool, the Blight Scorecard, to work within the existing Code Enforcement workflow.

A screenshot of the Blight Scorecard.

 

 

The Blight Scorecard allows a mid-level supervisor to score a property and then receive a weighted recommendation between 0 and 100: 0 meaning the property should be demolished, and 100 meaning it should be sold. Staff members can now use the tool to evaluate properties, vastly increasing the speed and consistency of the process. The scorecard is a decision support tool, but is not replacing human judgment – it provides a recommendation, but the director still approves each demolition decision.  It has improved the workflow of the city’s Code Enforcement Department, enabling guilty judgments to go to supervisors first before the director. Moreover, the new process adds efficiency by removing all paper components and effectively eliminated the backlog.                                                                                             

OPA revisits the algorithm periodically. The goal is to have the algorithm make recommendations that the director nearly always accepts. They plan to tweak the weighting if there are any disparities, ensuring it will remain accurate over time.

The ongoing successes of New Orleans in addressing blight showcase the value of a creative approach that considers many kinds of tools to tackle a high-priority issue. According to Chad Dyer, Director of the Code Enforcement Department, “Using data to be smarter about our operations has been a total win for the city, for homeowners and for our neighborhoods.”