Industry Must Help Clarify Big Data’s Policy Future, as FTC, Congress Continue Research, Says Brill

Despite vigorous debate and research about big data, “we may not know what the best course should be,” said FTC Chairman Edith Ramirez during opening remarks. “There is no clear path for navigating the use of big data in a way that advances opportunities for all consumers while diminishing the risks of adverse differential impact on vulnerable populations.” Panelists and presenters echoed Ramirez’s sentiment. “Big data is immature,” said Pam Dixon, executive director of the World Privacy Forum. “There is no firm scalpel-like legislative definition of big data."

FTC Chief Technologist Latanya Sweeney has been guiding a team doing big data research over the summer. Monday, Sweeney -- a prominent data de-identification researcher -- discussed some preliminary findings about how big data analytics deliver online ads. Formerly, publishers reviewed and approved ads before letting them appear, ensuring they weren’t alienating the company’s demographic, Sweeney said. Today, she said, ads are delivered through an online network, and no staff member is necessarily reviewing them.

"If there are sites out there that are more exclusive to a demographic,” said Jinyan Zang, an FTC technology and data governance research fellow, “what is the advertising experience like on those sites?” For example, Sweeney said, a website for Omega Psi Phi -- an historically African-American fraternity -- tended to receive only ads for poorly rated credit cards, and never highly rated credit cards. “Could there be the potential for disparate impact based on the type of ads that are shown or not shown?” asked Zang, who helped present the research. “What are the publishers’ rights and responsibilities?” Sweeney asked. She said in the coming weeks, her team would be releasing more research.

Sweeney’s 2014 stint as chief technologist is just one indicator of the FTC’s big data focus over the last year-plus (WID April 21 p1). The commission released in May a very granular report on the data broker industry, laying out a slate of legislative proposals to ensure transparency and consumer access to their data (WID May 28 p1). It also held a workshop on alternative scoring methods, or how new data analytics tools can be used to effect credit offers (WID March 20 p3).

The potentially unregulated scoring methods were a frequent topic at Monday’s workshop. Laws like the Fair Credit Reporting Act (FCRA) And the Equal Credit Opportunity Act (ECOA) provide some protections on how companies can make credit decisions, panelists said. “If, however, the same company lowers my credit limit based on the scores of a group in which I am a member, the application of the FCRA may be less clear,” Ramirez said. And the ECOA only applies to those seeking credit, said Leonard Chanin, a consumer finance lawyer at Morrison Foerster. “The law says it’s illegal to discriminate against an applicant,” he said. “If you have not applied for credit, the law technically does not apply to you.” This puts any “pre-credit activity,” including marketing, out of the law’s purview, he said.

FCRA’s contours were hotly debated after the FTC’s data broker report (WID June 2 p1), but ECOA has not been as prominent in the big data discussion. The law does have some teeth, said Peter Swire, a law and ethics professor at the Georgia Institute of Technology’s business school. Swire is also a former privacy official in the Clinton administration and was head of the World Wide Web Consortium-backed efforts to establish a Do Not Track standard. “The FTC has unusual enforcement power under [ECOA],” he said. “It doesn’t have to be somebody engaged in commerce” for the FTC to bring ECOA enforcement actions, he said. And the Consumer Financial Protection Bureau (CFPB) can also act, said Swire. It did this summer, ordering GE Capital Retail Bank to pay $225 million to settle charges it was not offering certain credit offers to individuals who indicated they preferred to communicate in Spanish or those with Puerto Rico addresses (http://1.usa.gov/1lpkkH8).

But there are 50 to 60 million “credit invisibles,” said Stuart Pratt, president of the Consumer Data Industry Association. Big data can help reach them, he said. Public record data sources, like utility bills, or telecom records can help credit companies find those without credit scores, but who may be quality applicants for credit. These records are “ubiquitous and deeply penetrated into communities of color,” which are often underserved, he said. “This broadens our markets for consumers to engage in a successful product."

Brill agreed in her remarks, noting the original rise of credit scores and credit bureaus in the early 20th century actually “broadened access to credit” and made credit determinations “more efficient and more objective.” But the government studied those scores closely, Brill said, examining whether credit scores were “using variables that act as proxies” for sensitive categories, such as race or age. The result was the FCRA, under which “the use of credit scores has thrived,” she said.

The government hasn’t done the equivalent research on modern alternative credit models, Brill said. And there’s not time to wait for Congress or the FTC, she said. “Companies can begin this work right now and provide us all with greater insight into and greater assurances about their models.”