Can Your Grocery Shopping Predict Your Credit Score? | This Is Kellogg

Why It Matters

Retail‑derived alternative data can unlock credit for underserved consumers, but its use must balance financial inclusion with privacy and regulatory safeguards.

Key Takeaways

• Alternative data can create credit scores for unbanked consumers.
• Grocery shopping habits predict credit repayment behavior accurately.
• Regular purchase timing and consistent spend indicate lower default risk.
• Cigarette purchases are strongest predictor of credit default.
• Privacy concerns arise, yet data aims to broaden credit access.

Summary

The Kellogg webinar features Professor Eric Anderson discussing how alternative data—specifically grocery loyalty information—can be used to build credit scores for people who lack traditional banking histories. Anderson frames the issue as a global credit‑access problem affecting over a billion individuals, many of whom have irregular incomes and struggle to obtain formal loans.

Using a partnership with a large Middle‑East supermarket‑card issuer, the research links purchase frequency, timing, spend consistency, brand loyalty, and promotion response to repayment outcomes. Regular shopping on the same day or time slot, stable basket size, and buying the same brands all correlate with lower default rates. Machine‑learning models sift through tens of thousands of SKUs to identify the strongest signals, with cigarette purchases emerging as the top predictor of delinquency.

Anderson highlights a memorable exchange: “Cigarettes are by far the number one predictor of whether you’ll default.” He also addresses privacy concerns, noting that the exact variables used in scoring are not disclosed to consumers, and argues that the intent is to help underserved households—such as single mothers with uneven paychecks—gain access to credit.

The findings suggest lenders can tap retail data to expand credit to the unbanked, potentially reshaping underwriting practices. However, they also raise regulatory questions about data transparency and consumer consent, underscoring the need for responsible AI deployment in financial services.

Original Description

In this episode of the This is Kellogg series, Professor Eric Anderson of the Kellogg School of Management presents his research on using grocery shopping data and machine learning to create credit scores for the more than one billion people worldwide who lack access to formal credit. Drawing on studies from the UAE, Peru and the UK, he reveals how shopping habits can signal repayment potential. He also discusses the real-world challenges of moving from proof of concept to production, the regulatory hurdles of using AI in lending and how partnerships between retailers, nonprofits and banks can bridge the gap for unbanked consumers.

The This is Kellogg series showcases Kellogg thought leadership and initiatives important to the future of the school, highlighting faculty research on the pressing challenges facing global business and society.

#KelloggSchool #FinancialInclusion #AIinFinance #ConsumerAnalytics