Applying Big Data Solutions to Big Social Problems
Los Angeles has now synchronized all 4,500 traffic signals across the city’s 469 square miles. It’s the first major city worldwide to achieve this feat, using magnetic sensors at every intersection; cameras; and a central computer system that monitors cars, buses, bicycles, and pedestrians. These innovations have increased the average speed of traffic by 16 percent and reduced delays at major intersections by 12 percent. Coupled with other initiatives that alert drivers to congestion, synchronization has brought an estimated annual savings in fuel and time of $1.3 billion.
Can we not apply the same “big data” solutions to major social policy problems?
One opportunity may be found in the Supplemental Nutrition Assistance Program (SNAP), formerly the Food Stamp Program, which serves about 47 million individuals per month, providing about $75 billion annually in nutritional support. Program participants redeem their benefits at authorized retailers using an electronic benefit transfer (EBT) card. Can the data generated by participants through card swipes be used to help serve these households better?
Many participants continue to have unmet nutritional needs; about one-quarter of SNAP households still experience very low food security, coping with their constrained resources by eating less and/or relying on lower quality foods. Can we better target such households for other forms of assistance—in particular, other federal benefits for which they may already qualify? Can real-time data be used to provide early warning signals of household distress?
A recent USDA study involving in-depth interviews with 90 SNAP participants in Boston, Houston, Indianapolis, and Riverside said: “One of the most striking observations in the current research is how many households fail to budget money with which to buy food during the time at the ‘end of the month’ when SNAP benefits typically run out.” Indeed, the absence of a plan for stretching resources to the next month—or for accessing additional resources from families, friends, or other networks—was a distinguishing feature among those who were food insecure.
Here’s an idea that builds on USDA’s current use of EBT transaction data nationally to analyze SNAP benefit redemption patterns. Such data could be used to identify SNAP households that redeem more than 90 percent of their monthly benefit within the first week (as 20 percent do nationally) or those who exhaust their benefit within the first week (as 17 percent do nationally). These households very likely will have a hard time affording food at the end of the month. But that impending hardship could be alleviated by ensuring that these households receive other benefits that they qualify for (such as Medicaid, Supplemental Security Income, or the Earned Income Tax Credit) or that they get priority in receiving discretionary program support (such as energy assistance or child care subsidies).
This kind of targeted outreach would require information systems that can capture EBT transactions data in real time and can generate early warning signals to the appropriate program office with adequate protection of client privacy.
It may seem fanciful, but if Los Angeles can synchronize the city’s traffic signals, is it that far-fetched for Los Angeles County—with more than one million SNAP participants among its residents—to institute such an early warning system?
Grocery Store image from Shutterstock