Understanding the Context of COVID-19 Relief Spending

RS21
RS21 Blog
Published in
5 min readDec 7, 2020

--

Posted by Tim Farkas, RS21 Data Scientist

If $3 trillion dollars sounds like a lot to spend all at once, you’re right on the money.

In response to the COVID-19 pandemic, the U.S. Congress passed the landmark CARES Act and a flurry of related legislation early in 2020, infusing the American economy with a stimulus to mitigate unprecedented strain on families, businesses, and healthcare systems.

Access the tracker online at covidtracker.pogo.org.

Given the enormous scale of funding and extraordinary haste of disbursement, concerned citizens — and numerous media outlets supporting them — have taken an acute interest in where the coronavirus relief funds have gone.

To improve transparency, we teamed with the Project On Government Oversight (POGO) and FuseLab Creative to build the COVID-19 Relief Spending Tracker, an intuitive data portal into the expansive COVID spending universe.

“As an organization, POGO is committed to increasing government transparency, and with this tracker, we hope to provide a much-needed window into the several trillion dollars the government is spending to address the pandemic and aid those working to hold our government accountable,” said Danielle Brian, Executive Director of POGO.

Other trackers exist, but the POGO COVID-19 Relief Spending Tracker is unique among them. Like other spending trackers, our tool assembles the spending data in one place, allowing detailed search and rich filters to explore relief spending in real time.

But we wanted to provide investigators with data about the context in which relief spending funds were allocated. That’s why we overlaid spending data on unemployment, population size, and racial demographics.

With these auxiliary datasets, you can go beyond asking questions about where funding went, and which government agencies were responsible. You can seek a more nuanced understanding, answering questions like “Did relief spending go to the places that needed it most?” and “What do the communities who benefited from the stimulus look like?”

Follow along below to see the tool’s most powerful features, and how to go about exploring the data.

To improve transparency of how funds were allocated, the COVID-19 Relief Spending Tracker overlays spending data on unemployment, population size, and racial demographics.

Exploring Community Indicators and COVID-19 Relief Spending

The core of COVID-19 Relief Spending Tracker is, of course, the spending data. You can look at the most fundamental aggregations, permitting a comparison of how much states received in government contracts versus loans, for example. And you can isolate spending in counties, or even ZIP codes, with an interactive map loaded with toggleable layers.

As expected, a table view allows you to zero in on — and download — raw spending data of interest, filtering away millions of records not pertinent to your specific query.

Table view allows for side-by-side comparisons of data based on a specified set of filters. Data can be downloaded for offline research as well.

But the most valuable aspects of POGO’s tracker don’t come from the spending data alone. Although knowing the sum total of spending in your home county might be a nice piece of trivia, what’s really useful is to understand the context in which COVID funds are distributed.

To achieve this, we curated datasets on community demographics, focusing on some of the basics: population size, ethnic and racial composition, and unemployment.

You can explore spending in relation to community demographics in a few ways. Most simply, you can look at the raw numbers for demographics on a map for states, counties, and ZIP codes, just as you can explore spending itself.

Unemployment in July 2020 for all counties in contiguous U.S. Data from Bureau of Labor Statistics.

Going a step further, we wanted the map visualizations to help you analyze the data in real time. So we overlaid community demographics on spending maps and added scatterplots. The scatterplots visually relate spending in each geography to the concentration of minorities and unemployment in those communities.

Take a look for yourself: Is racial composition in Oregon related to relief funding received?

Scatterplot showing relationship between relief spending and concentration of minority demographics in all Oregon counties.

It sure looks like there’s a clear positive relationship between the amount of funding allocated and the proportion of residents identifying as non-white, non-Hispanic. We don’t know the cause of the relationship, of course, but it’s evidence against the notion that funds are allocated away from counties with high proportions of minority residents, at least in Oregon.

But what’s going on with that one outlier with extreme spending per capita, Crook County? The COVID-19 Relief Spending Tracker is the perfect tool for exploring anomalies like this.

Looking for clues on Crook County, we can use the tool to dig a little deeper. We find the unemployment rate to be middle of the road, as is the population size, and it’s not where the state capital is, either. Turning to the Table View, we find there aren’t any single records with extreme values of assistance, but there are an unusually high number of smaller assistance records coming from the Small Business Administration, many of which have the recipient names withheld (this is common, actually).

Table view of assistance in Crook County, OR.

There’s no smoking gun here, but using the Relief Spending Tracker in this way leads us to a more refined question: Why so much spending on small businesses in Crook County, of all places?

New Features, New Insights

New features to augment existing analytics capabilities are already in the works.

Each news cycle comes with at least a few chronicles of fraud related to COVID relief spending. Malfeasance ranges from lies about how many employees will be supported with funding to collusion between application reviewers and applicants themselves.

To help prioritize investigative efforts into fraud, we’re adapting machine learning algorithms, designed by RS21 to detect fraud in Medicare billing, to focus on spending anomalies in the COVID spending data. Analysts and investigators using the tool would be alerted to spending records that are similar to known cases of fraud, or even just cases that are downright unusual.

Access the tracker online at covidtracker.pogo.org.

The COVID-19 Relief Spending Tracker serves as an evolved model for what government oversight and transparency can mean. It goes above and beyond the raw data, filters, and aggregations you’d expect from a tool like this, sporting features that allow analysts to drill down on interrelations among a diversity of datasets. In so doing, we enable a rich understanding of the context and consequences of an historic spending effort, and better hold accountable the government we rely on to support us through times of crisis.

ABOUT RS21

RS21 is a rapidly growing, global data science company that uses artificial intelligence, design, and modern software development methods to empower organizations to make data-driven decisions that positively impact the world. Our innovative solutions are insightful, intuitive, inspiring, and intellectually honest.

--

--

RS21
RS21 Blog

RS21 is revolutionizing decision-making with data + AI. We believe the power of data can unleash human potential and make a better world. Visit www.rs21.io.