The Federal Trade Commission's (FTC) recent report, Big Data: A Tool for Inclusion or Exclusion?, released on January 6, warns businesses engaged in big data analytics that the practice poses liability risks. While the FTC does not propose new law or policy, the report advises that existing laws-including the FTC Act, the Fair Credit Reporting Act (FCRA), and the Equal Credit Opportunity Act (ECOA)-already can reach certain uses of big data.
The FTC's warning is not intended as a prohibition. Indeed, the Commission recognizes, and even encourages, the use of big data to innovate and improve the ability of businesses to serve underserved communities. Nevertheless, the FTC plans to monitor the use of big data under privacy, credit, and fair lending laws and bring enforcement actions where appropriate. Accordingly, businesses using or considering the use of big data analytics should review the FTC's report carefully and consider the risks those practices may present, especially in the privacy and fair lending areas.
The FTC's report is the product of a public workshop held in September 2014, as well as a prior FTC seminar on alternative scoring products, and it aims to educate businesses on important legal compliance questions and broader big data research issues.1
Before addressing specific legal and policy considerations, the report discusses the “life cycle” of big data and emphasizes the associated potential benefits and risks. The FTC distinguishes descriptive analytics-where the objective is to uncover or summarize patterns in data sets-from predictive analytics-where the use of statistical models generates new data. The FTC notes that predictive analytics are frequently used to draw inferences about consumers' likely choices.2 The report highlights a number of benefits that may emerge as a result, including new methods to access credit as well as better health and educational outcomes. Echoing themes from the White House's 2014 Big Data report, the FTC's report also discusses potential risks, suggesting that analytics could be used to expose consumers' sensitive information, implement new forms of price discrimination, or weaken the effectiveness of consumer choice by drawing inferences about consumers who choose to opt out of data collection.3
Applicable Consumer Protection and Equal Opportunity Laws
The FTC stresses that the report is not intended to identify gaps in the law, but rather to explain the Commission's view that various laws already apply to big data practices.4 Accordingly, the FTC notes that a number of existing consumer protection laws, including the FCRA, federal equal opportunity laws, and the FTC Act, may already apply to big data practices.
- Fair Credit Reporting Act: The FCRA applies to companies that compile, sell, or use consumer reports to make eligibility determinations for certain benefits or transactions. The FTC's report notes that companies are increasingly purchasing and using predictive big data analytics products for eligibility determinations.5
The FTC cautions that while companies are generally free to use data derived from their own customer relationships for purposes of making decisions about their customers, this calculus changes when third-party data analytics providers are involved. If an unaffiliated third party evaluates a company's own data and provides that company with evaluations for covered eligibility determinations, the third party acts as a consumer reporting agency (CRA), and both companies can be considered users of consumer reports.6 The FTC emphasizes once more that businesses can be considered a CRA under the FCRA even when they compile non-traditional information, such as data from social media platforms, for eligibility purposes.7
The report also reevaluates the FTC staff's position from its 2011 report 40 Years of Experience with the FCRA and now explains that the FTC may consider reports in which identifying information about a specific consumer has been stripped away to be consumer reports for purposes of the FCRA under certain circumstances.8 Take, for example, a consumer who provides information about her social media habits on a credit application. If a company subsequently removes any identifying information, yet still uses the underlying data to generate an aggregated analysis and make a decision affecting that consumer, the analysis would likely be deemed a consumer report, according to the FTC report. The FTC encourages businesses to be mindful of any use of big data analytics to make an eligibility determination that could be covered under the FCRA.
- Equal Opportunity Laws: The report devotes considerable attention to the applicability of federal equal opportunity laws, including ECOA, Title VII of the Civil Rights Act of 1964, the Americans with Disabilities Act, the Age Discrimination in Employment Act, the Fair Housing Act, and the Genetic Information Nondiscrimination Act, to big data use. The report notes that to prove a violation of these laws, plaintiffs must show either “disparate treatment” or “disparate impact,” based upon a protected characteristic such as race, gender, marital status, or receipt of public income.9 The report places a particular emphasis on compliance with ECOA, over which the FTC shares enforcement authority with the Department of Justice and the Consumer Financial Protection Bureau.10
Disparate impact analysis is especially important in the context of big data analytics.11 A disparate impact occurs when a company employs a facially neutral policy or practice that has a disproportionate effect on members of a protected class, absent a legitimate business need that cannot be achieved by less disparate means. For example, a business might avoid expressly screening job applications based on gender and instead use analytics to screen applicants in a way that nevertheless has a disparate impact on women. Similarly, making credit decisions based on zip codes could prove problematic if those determinations have an impact on a protected class and cannot be justified by a legitimate business necessity. Thus, in certain circumstances, facially neutral policies could run afoul of equal opportunity laws.
The FTC's report cautions that advertising and marketing practices could implicate equal opportunity laws for creditors.12 Even if credit offers are open to all who apply, marketing that affects a creditor's subsequent lending patterns or the terms and conditions of the credit received by borrowers could be cited as evidence of discrimination.
- Section 5 of the FTC Act: The FTC reiterates that businesses should take care to ensure that their use of big data analytics is done in a way that is not unfair or deceptive to consumers. The FTC recommends that companies engaging in big data analytics consider whether they (1) are honoring promises made to consumers and providing consumers with material information about their data practices, (2) have put reasonable security measures in place, and (3) are undertaking reasonable measures to know the purposes for which customers are using their data.13 At minimum, the FTC remains concerned that companies not sell or share analytics products if they know or have reason to know that the recipients could use those products for discriminatory or fraudulent purposes.
The report emphasizes that any FTC or regulatory inquiry under these laws is ultimately case- and fact-specific. The report provides a set of high-level questions focused on FCRA and ECOA requirements that companies using big data analytics should consider when navigating these laws.14
Big Data Policy Considerations
Beyond legal compliance, the report also discusses several policy considerations raised by big data. The FTC appears worried that errors and biases in data may be incorporated throughout the entire big data life cycle. The Commission cautions that adding more data may not correct inaccuracies or remove these biases.15 The FTC encourages businesses to consider the following questions when deploying big data analytics:
- How representative is your data set? Citing the Boston Street Bump application that was designed to detect potholes via a smartphone app, the FTC explains that once the app team realized that lower-income individuals were less likely to carry smartphones, they also realized their application data was not representative of road conditions across Boston.16 According to the FTC, companies should be aware of how “digital divides” may produce under- or over-representative data sets before launching products or services that could produce skewed and unfair ramifications.17
- Does your data model account for biases? The report notes that even before the widespread use of big data, computer models used to identify good job applicants could reproduce existing biases by incorporating pre-existing discriminatory actions into new decision-making. The FTC recommends that companies think carefully about how both their data sets and algorithms have been generated.18
- How accurate are your predictions based on big data? The FTC cautions that while big data has improved the ability to detect correlations among data points, it does not always explain which correlations are meaningful.19 The report highlights efforts by lenders to improve access to credit by using non-traditional indicators such as rental or utility bill payment history, but notes that there could be legitimate reasons for consumers to withhold paying or otherwise dispute a bill, which could throw off these innovative credit models.20 The report stresses that when big data analytics affect consumers' ability to complete transactions, the potential effects from inaccuracies could be substantial.
- Does your reliance on big data raise ethical or fairness concerns? The notion that big data analytics raises larger ethical issues emerged in the White House's 2014 Big Data report and continues to be of concern to the FTC. The report suggests that companies consider assessing what factors go into an analytics model and balance any predictive value against fairness considerations.21 The FTC, however, also highlights how big data can be used to expand opportunity.22
As the report states, “[T]he Commission will continue to monitor areas where big data practices could violate existing laws, including the FTC Act, the FCRA, and ECOA, and will bring enforcement actions where appropriate. The Commission will also continue to examine and raise awareness about big data practices that could have a detrimental impact on low-income and underserved populations, and promote the use of big data that has a positive impact on such populations.”23
In a concurring statement, Commissioner Ohlhausen acknowledges some of the concerns around big data analytics as legitimate, but encourages policymakers to evaluate these concerns within “the larger context of the market and economic forces companies face.”24 She notes that big data analytics may combine with a competitive market to resolve rather than exacerbate industry's misunderstandings of low-income populations. For instance, lenders could collect and analyze predictive data to market credit offerings to traditionally underserved communities. In this way, big data could allow lenders to reach populations that are difficult to score using traditional credit models.25 She recommends that the FTC test hypothetical harms with strong economic reasoning and empirical evidence as it addresses big data in the future.
The overarching message of the FTC report encourages businesses to apply big data analytics in a way that provides benefits and opportunities to consumers, while also identifying the legal pitfalls that could violate consumer protection or equal opportunity laws. Use of big data as a proxy for protected class status or in a way that results in a disparate impact based on protected characteristics without a countervailing justification based on a business necessity carries risks. Where businesses carefully consider their objectives and pressure test their practices, they can reduce exposure to discrimination claims and expand access to credit consistent with the FTC's guidance.
1 On January 10, 2016, two Consumer Financial Protection Bureau (CFPB) officials stated that no CFPB guidance is forthcoming on the proper use of big data analytics. However, Bryce Stephens, the CFPB's section chief for compliance analytics and policy, reported that the agency remains concerned about model-related questions surrounding the use of big data in marketing and lending determinations. See Chris Bruce, Bloomberg BNA Banking Daily, "Don't Expect CFPB Fair Lending Guidance on Big Data" (Jan. 10, 2016), available here.
2 Fed. Trade Comm'n, Big Data: A Tool for Inclusion or Exclusion? 4-5 (Jan. 2016).
3 Id. at 9-11. See Executive Office of the President, Big Data: Seizing Opportunities, Preserving Values 48-54 (May 2014).
4 Id. at 12-13.
5 Id. at 15.
7 Id. at 14.
8 Id. at 16, fn. 85.
9 Id. at 18, 20.
10 Id. at 18, fn. 94.
11 Id. at 19.
12 Id. at 20-21.
13 Id. at 21-23.
14 Id. at 24.
15 Id. at 25.
16 Id. at 27.
18 Id. at 29.
20 Id. at 30.
21 Id. at 31.
22 Id. at 32.
23 Id. at v.
24 Id. at A-1.
25 See id. at A-2 and fn. 4-5; see also Executive Office of the President, Big Data: Seizing Opportunities, Preserving Values at 7, 45-47 (May 2014).