Beyond Profit: Ethical Research on Consumer Data
Digital platforms collect and often share data regarding our online behavior. Increasingly, consumers agree to release data in exchange for digital services. How this consumer data is used – by industry, by researchers – raises important ethical and practical questions.
In January, we convened more than 50 academics, private sector leaders, regulators, and policymakers to discuss research experiments and partnerships using digital platforms and consumer data. The Promoting Financial Inclusion and Innovation in Latin America conference – hosted by UC Berkeley Haas School of Business Lab for Inclusive Fintech (LIFT), the Schools of Business and Government at Universidad Adolfo Ibáñez (UAI), and the Center for Effective Global Action (CEGA) – examined innovative financial products, the regulations surrounding them, and research findings related to consumer behavior.
This convening allowed us an opportunity to reflect on important questions regarding how we lead rigorous research partnerships designed to improve consumer welfare.
Balancing benefits and costs
Industry leaders claim data sharing provides consumers with direct benefits, such as improved services and convenience. For example, you may benefit from ordering your groceries for delivery some days rather than shopping yourself and driving across town.
During our LIFT conference, we had a chance to learn from some industry leaders in Latin America – including Mercado Libre, Galgo, RappiCard, and Banco Estado – during the Session 4 panel and discussions. In addition, through the two days we had the chance to learn how we – and consumers – can benefit from research that relies on consumer data:
- Just-in-time messaging sent to consumers when they are making loan and payment decisions can save consumers money through lower interest rates and fees (Session 2 – Innovations in Credit presentations by Santiago Truffa, Daniel Schwartz, and Erik Berkwart).
- Lending with digital collateral can provide welfare gains for low-risk, middle-earning customers, but only for customers who had more leniency in the lock-out terms and lower penalties (Session 8 – Innovations in Credit presentation by Brett Green).
- Consumers need to know themselves as much as others know about them! Consumer overconfidence can erode the effects of informational messaging on credit terms (Session 8 – Innovations in Credit presentation by Paolina Medina) and many who could benefit from financial education are prone to underestimate its benefits and overestimate its costs (Session 6 – Financial Education presentation by Elaine Shen).
- Policies focused on early, high quality financial education can have large knowledge gains that change financial behavior for learners with spillover effects in their communities (Session 6 – Financial Education presentation by Veronica Frisancho).
But what are the costs of access to and sharing this data? Industry can certainly use information about consumer behavior and preferences to increase financial costs to consumers, such as charging more for delivery or other platform fees. There are other costs that are less obvious, which can include a loss of privacy and trust in systems when data is used for unknown or deceptive behavior.
So, when are benefits worth the cost and when are they not? Recent consumer backlash provides insight into when individuals and regulators do not think there is balance in the benefit-cost ratio. In the US, several stories about Instacart – from paying a $60 million fine to the US Federal Trade Commission for deceptive practices to investigative reporting revealing a consumer pricing experiment and the swift end to that experiment following backlash – came out around the same time as our conference and allowed us an opportunity to reflect.
Distinguishing our whys and hows
Our organizations—the UC Berkeley Haas School of Business Institute for Business & Social Impact (IBSI), the Center for Effective Global Action (CEGA), Innovations for Poverty Action (IPA) and universities in Chile — support and lead research focused on consumers and financial services (see here, here, and here). We see four key distinctions between our research and the Instacart activities discussed in December:
Focus on societal impacts
The private sector builds internal data science teams to meet company goals, often centered around maximizing profit and minimizing consumer surplus. Our research is focused on how financial services and digital platforms may improve lives, particularly for vulnerable and underserved populations, while remaining profitable. Our studies include a rigorous and systematic appraisal of the welfare impacts of interventions. This welfare analysis can also uncover situations where misperceptions – by lenders, consumers, regulators – are identified and corrected to improve both social impact and firm and/or policy performance.
Align incentives and ethics
In private sector research environments, those who will gain from profit-maximizing set the learning agenda and the research team applies their skills to meet these goals. There is often little if any additional, independent review of the ethics of the experiment to ask—even if we can do this, should we? In contrast, we build dynamic partnerships across independent academic researchers, private sector, funders, and policymakers to help align incentives. As research organizations, we contribute to the systems and best practices to deliver rigorous and ethical experiments that are transparent and reproducible—including study registration, transparent reporting, appropriate data and code sharing—and in some cases include an “IRB plus” ethical review focused on meaningful informed consent and review by subject matter experts who can better speak to contextual risks and potential harms. (Want to learn more about how we do this? Check out CEGA’s research transparency initiative, IPA’s Research Hub, or IBSI’s Open Science resources.)
Cultivate curiosity and courage
Research experiments—and the knowledge they build—can only be as good as the questions that drive them and the willingness to learn. Co-creating research agendas with government and private sector leaders is often the most meaningful way to ask productive questions about how products and services affect different populations and how policy can serve particularly vulnerable and previously neglected populations. In addition, experiments can fail in implementation or present results that contradict a previous belief or interest. Productive partnerships require a shared commitment to learning from null, disappointing, or unexpected findings. That mindset is often a precondition for eventual success.
Build trust and partnership
What builds trust and partnership? This is where the headlines around the Instacart activities and the backlash to end them can inform our work. The pricing experiment wasn’t revealed in a clear, transparent manner by its authors—it was discovered by independent groups focused on consumer protection. The goals of the pricing experiment weren’t publicly defined anywhere—they were inferred and assumed.
The research projects presented during our LIFT conference were years in the making. Years of relationship building, question refining, data sharing. In every study we have worked on, no one takes these experiments lightly. Researchers feel the weight of their responsibility to their implementation partners and the research participants. Industry analysts feel the weight of their responsibility to their company’s shareholders and their customers. Democratic policymakers feel the weight of their responsibility to citizens.
Building meaningful experiments
Nobody wants to make headlines for the wrong reasons. Not the researchers, the private sector leaders, nor the policymakers. We can and should acknowledge – the line between the work we do and private sector consumer research like the Instacart pricing experiment is thin. But it’s up to us to define and strengthen that line. It’s a long road to design and implement meaningful experiments and one that is built on our trust in our partners to ask important questions and their trust in us to carefully and responsibly answer those questions with them. We look forward to doing just that as we follow up with policy, implementation, and research partners over the next few months.
This blog is cross-posted with CEGA’s blog here – https://cega.berkeley.edu/article/beyond-profit-ethical-research-consumer-data/
