Recent advancements in artificial intelligence and machine learning, along with the evolution of large-scale data storage, access, and processing technologies, have ignited interest among financial institutions in novel data reservoirs for credit scoring. These innovative sources encompass bill payment histories for phone, utility, and streaming services, transaction records from various financial accounts like checking, savings, and money market accounts, and rent payment histories. This interest is twofold — driven by the pursuit of profit, including the generation of new accounts, and the noble aim of enhancing social welfare by extending credit access to individuals devoid of conventional credit scores.

A recent study from the University of Notre Dame reveals that frequent visits to the grocery store might hold the key to demonstrating creditworthiness. Titled “Utilising Grocery Data for Credit Assessment,” the forthcoming research in Management Science by Joonhyuk Yang, Assistant Professor of Marketing at Notre Dame’s Mendoza College of Business, in collaboration with Jung Youn Lee from Rice University and Eric T. Anderson from Northwestern University, sheds light on this intriguing prospect.

The research team partnered with a multinational conglomerate operating in multiple cash-reliant, developing countries across Asia and Africa. The conglomerate, the data sponsor, owns a credit card issuer and a large-scale supermarket chain. This unique arrangement facilitated merging data from these two domains, enabling the observation of behaviours from a sample of 30,089 consumers.

Their methodology involved transforming raw data into a refined set of inputs while filtering out credit risk indicators embedded within grocery data. Yang explains, “Our approach was prompted by insights gleaned from discussions with the data sponsor’s manager, who stressed the need for a strategic summarisation of key data elements into meaningful variables. Merely inundating the problem with vast amounts of data sans structure is unlikely to yield results.” This sentiment echoes similar sentiments expressed by managers from leading U.S. banks, highlighting the challenge of efficiently leveraging extensive consumer data for loan decision-making.

The study found that recurrent grocery shopping habits provide discernible signals of credit risk. For instance, purchases of cigarettes or energy drinks correlate with a higher likelihood of missing credit card payments or defaulting. Conversely, buying ‘good’ or healthy groceries, such as fresh milk or vinegar dressings, is associated with consistent and timely credit card bill payments.

Drawing from a vast body of literature on habits, the researchers constructed variables measuring the consistency in consumers’ purchasing patterns. This approach was facilitated by the nature of grocery items as non-durable necessities, leading to frequent and recurrent choices by consumers.

Moreover, the study demonstrates that an individual’s grocery purchases can elucidate their payment behaviour, even after controlling for socio-demographic variables and credit scores. Yang elaborates, “Using item-level survey ratings, we found suggestive evidence that purchasing healthier yet less convenient food items predicts responsible payment behaviours. Additionally, there is a robust correlation between the consistency in various dimensions of grocery shopping and timely credit card bill payments.”

Furthermore, cardholders who consistently pay their bills on time exhibit particular behavioural patterns, such as shopping on the same day of the week, maintaining similar expenditure levels across months, and showing brand and product category loyalty.

Through simulations of hypothetical credit scoring and decision-making processes, the research team illustrates how grocery data can provide valuable insights into credit risk, leading to enhanced credit outcomes for deserving individuals and increased profitability for lenders. For instance, the inclusion of grocery data substantially enhances default predictive accuracy for individuals lacking credit scores, resulting in improvements ranging from 3.11 to 7.66 percentage points.

However, the study also identifies scenarios where the use of grocery data adds minimal incremental value, thereby underscoring the limitations of this new data source. Yang notes, “The incremental benefit of grocery data diminishes sharply when traditional credit scores or relationship-specific credit history are available.” These findings elucidate when lenders might find it advantageous to collect, acquire, and utilise alternative data sources.

These findings have profound implications for financial institutions. Leveraging grocery data for credit scoring presents an opportunity to tap into a vast, untapped market segment. By extending credit to consumers currently underserved by the traditional credit system, lenders can expand their customer base and enhance profitability while contributing to financial inclusion and social welfare.

More information: Jung Youn Lee et al, Using Grocery Data for Credit Decisions, Management Science. DOI: 10.2139/ssrn.3868547

Journal information: Management Science Provided by University of Notre Dame