Banking on Data Science: Big Data Analytics of Consumer Behaviours Across Financial Spaces

The purpose of my research is to learn about human behaviour from mass financial transaction data such as credit card or current account statements. While traditional economic models of consumer behaviour assume that rational consumers continually make optimal choices, alternative models suggest that consumers tend be guided by heuristics or rules of thumb. The project will use data sets held at Warwick and Nottingham, including transactions from 1.4 million credit card users and trading records from 200,000 private investors provided by major UK financial institutions. It will investigate how consumers make financial decisions in practice using large datasets.
Potential topics for PhD chapters include:

Simultaneous co-holding of assets and debts. Why do people co-hold savings and debts, when they could pay down their debts and save interest payments? This behaviour apparently ignores a simple arbitrage opportunity. Are mental accounting and limited attention responsible?

Using financial products in a manner inconsistent with their primary purpose. Why, and how, do people misuse financial products (e.g., using a credit card for cash withdrawals when a cheaper loan is available)? Understanding 'consumer product fit', is important given the variety of saving and borrowing products.

Biases and over-reactions to losses and gains. Detailed investor data allows us to observe how investors react to losses and gains. What are the psychological processes underlying this economic behaviour?

Machine learned personality. Can we use mass transaction data as a record of the behaviours people engage in to abstract personality measures? How does this machine-learned personality compare against traditional consumer segmentation and credit scoring?