Helping ensure integrity in a multibillion-dollar program
How KPMG helped improve a city’s response to the migrant crisis while saving money
Helping ensure integrity in a multibillion-dollar program
How KPMG helped improve a city’s response to the migrant crisis while saving money
Client
Department of Investigation
Industry
State and local government
Primary Goal
Detect and respond to fraud, waste, and abuse
Technology
Python integrated with Microsoft Power BI™
By the middle of 2023, the number of asylum seekers entering the United States had strained the resources of municipal governments across the nation. To deal with the influx, one of America’s largest cities established an Asylum Seekers Initiative, budgeting more than $10 billion to care for the new arrivals through 2026. To guard against fraud, waste, abuse, or other irregularities, the city’s Department of Investigation asked KPMG to serve as the program’s Integrity Monitor. To date, KPMG has improved program efficiency; expanded the city’s fraud, waste, and abuse prevention and detection capabilities; and saved taxpayers millions of dollars.
Client transformation journey
We’re not your typical integrity monitors—it's not a ‘gotcha’ exercise. We're here to make sure that guests are getting what they're entitled to from the providers that the city has hired. We’re ensuring accountability and compliance with regulations, and—ultimately—leading to cost savings for taxpayers.
Tom Stanton
Principal, KPMG LLP
KPMG had served as the city’s Integrity Monitor in previous crises and had the experience and the resources to manage large, multipart engagements. In addition, the firm had a strong background in the construction and healthcare sectors, and both would likely be major beneficiaries of the awarded contracts. Along with the right people and skills, KPMG came in with the right roadmap.
KPMG realized that it would need people to deal with the city agencies to get permissions to visit the sites holding the refugees. It would also need people to inspect the sites and interview the guests—“guests” is the term of art—many of whom spoke no English. The information they gathered would have to be segmented into standardized categories, where individual answers could be coded and analyzed. An analytics team, well versed in artificial intelligence (AI) and natural language processing, would be needed to develop the models that processed the data and provided reports.
KPMG quickly assembled project management teams, field teams, and analytic teams. Three project management officers (PMOs) dealt with city agencies, scheduling and tracking site visits. Site teams visited the guests where they were housed. On the analytics side, KPMG selected data scientists from its Lighthouse and tax evaluation teams. Approximately 15 people worked scraping data from contracts and invoices and developing programs to mine it.
KPMG used an efficient, risk-based approach to prioritize and schedule site visits. The site work was both abstract and deeply personal, as it probed individual decisions to develop aggregate insights.
Four KPMG field teams of two people each went out every day to inspect the 400 shelters, hotels, apartments, and other facilities housing the new arrivals. The teams were equipped with an internet-connected 80-item checklist that followed the guests beginning with how they arrived—by bus, from a hotel—and getting all relevant information.
Then the teams examined conditions in the facility, beginning with fire safety. They checked to see that fire drills were conducted, that extinguishers were up to date and the buildings up to code. They monitored the surrounding area to ascertain cleanliness and ensure the property was free of vermin; and they looked for threats to children, from broken window locks to proper separation between families and single adults in sleeping arrangements.
From the site visits, KPMG learned that food waste was a significant problem. At some sites, up to 70 percent of the ordered food was thrown away. There was nothing wrong with it per se, but it was out of touch with the religious restrictions or dietary preferences of the new residents. This waste was not only an economic issue, but also a fire hazard: residents would smuggle hot plates and air fryers into the facility and try to do their own cooking. Indeed, contraband of all kinds remains an ongoing concern, and KPMG has recommended several measures, such as equipping security guards with scanning wands, to interdict weapons and drugs.
When KPMG brought the food waste issue to the attention of city agencies, the agencies acted. They ordered providers to reduce the amount of food they were buying and, going forward, to track the amount of food purchased and the amount thrown away. Now, with statistical data, the city could see the effects of better management. Between mid-2023 and March 2024, thanks in large part to the efforts of KPMG, the city has saved an estimated $15 million in food waste at the shelters alone.
In addition, the city surveyed the guests to find out exactly what type of food they preferred for breakfast, lunch, and dinner. Their preferences drive the city’s choice of vendors. If a given vendor can’t provide desirable foods based on survey results, then the city will find another. The wide number of ethnicities represented in the shelter population has often required the city to reach out to small, specialized providers. Every item in every purchase can be identified and tracked.
The city lacked a database that allowed it to compare information across all city and noncity agencies easily. So, the key priority from the points of view of both KPMG and the city was to establish and connect a central data repository. To do so, KPMG needed to capture all contract information—including updated extensions and amendments—from all the subcontractors involved, and to combine and track everything in a database it would design.
KPMG used natural language processing to “scrape out” the scopes and budgets of contracts and invoices and employed data analytic models—some AI and some Python code—to deconstruct the invoices. With optical character recognition, KPMG was able to take the data behind the invoices to spot high-level anomalies, such as a supplier’s billing in excess of 24 hours on a given day. In addition, KPMG was able to compare invoice charges to contract stipulations. If a firm was billing $60 an hour when a contract mandated $25, for example, then KPMG alerted the city agency for follow-up and remediation.
At the same time, AI helped KPMG convert its on-site checklists into reports. These reports provide the city’s DOI with a three-page synopsis of each particular site, bringing added clarity to decision-making.
Tomorrow's government today
Emerging technologies in government
Transforming government spend using data-driven analysis
Efficiency meets innovation: optimize, integrate, and invest in smart spend visibility
Transforming the future of Medicaid
How social, technological, economic, and regulatory trends are shaping the future of Medicaid