post by Ana Rita Pena (2019 cohort)
For the past year, I have collaborated with the Centre for Responsible Credit (CfRC); a charity working to improve consumer credit regulation and lending practices with a particular focus on lower-income households.
Our collaboration made possible with an EPSRC Horizon CDT Impact Grant, has explored whether trusting and contextualising an individual’s financial situation offers a potential alternative to the traditional ‘data surveillance’ approach of credit scoring. Could borrowers be trusted to be honest about their financial circumstances, and, if so, could this information accurately predict their default risk?
I first became aware of the CfRC’s work at the end of my first year of PhD when I was about to start working with the credit card company, Capital One UK. While at Hallward Library, I discovered a book titled ‘Britain’s Personal Debt Crisis’, written by CfRC’s Chief Executive, Damon Gibbons. The book outlined the evolution of the UK’s credit industry, and following further research about the Centre, I was drawn to their holistic and innovative approaches. In particular, their FlexMyRent (‘FMR’) project caught my attention. The FMR scheme provided social housing tenants with the opportunity to personalise their rent payments over the course of a year: paying less rent at times when their finances were tight, and more when things were, relatively, less stressed. This flexible payment option was aimed at helping tenants to manage their cash flow without having to use credit (which for this group of consumers would likely incur very high and potentially predatory interest rates and charges). Application to the scheme was based on a proposed rent payment plan (how much the tenant would pay each month), rent account details and tenant answers to a questionnaire about their wider financial circumstances and ‘support needs’. There was no use of credit reference agencies or third-party data.
The information provided was risk-assessed and fed into decisions by Housing Officers regarding admittance to the scheme. This application process has some similarities to underwriting decisions within the credit industry. Specifically, information about the applicant is used to create a measure of default risk (failure to repay in line with the agreement). In the credit sector, the risk assessment (usually based on someone’s credit score, which is in itself a risk measurement) combined with the lender’s credit policy, which sets their risk appetite, and determines which products – if any – are offered.
The most significant difference between these two application processes is the provenance of the financial data. In traditional credit applications, the data is much more extensive and detailed, related to past payment behaviour, and combined with public information (e.g., county court judgments, time on the electoral roll etc.) For the FMR scheme, the data is directly requested from the applicant and is a lot less granular, i.e. “How often, in the past three months, have you had money left over after you have paid for food and other essentials including bills and credit repayments?”.
This alternative approach to risk assessment based on voluntary disclosure of an applicant’s financial information, was what caught my attention and made me interested in wanting to work with the scheme’s data. I thought it had a lot of potential to benefit consumers. The exploration of possible alternatives to traditional approaches complemented my PhD work well, where I was mainly studying and critically reviewing the current credit system.
The first step was to have an initial chat with CfRC. This served to enquire about their interest, for me to get a better understanding of the FMR scheme, and to discuss potential research inquiries that would particularly benefit from the expertise I had developed throughout my PhD.
Based on the initial conversations, I decided to apply for a project using Machine Learning (ML) techniques, as these hadn’t been used on the dataset before and could help identify unexpected behaviours. I also thought that an approach based on ML analysis of the data could be appealing to both the industry and the regulator (Financial Conduct Authority). As with my PhD, I also wanted to include an element of qualitative work to complement the quantitative analysis.
I planned a project based on two phases: an initial data analysis using ML models, and a subsequent series of interviews with applicants to gather rich and complementary data. To be able to put together the budget, I created a timeframe for the project and estimated the hours needed to complete the tasks within this. As I would be engaging tenants in the qualitative phase, it was also important to create the draft topic guide, estimate the timeframes of the interviews and include adequate remuneration for research participants.
The most challenging part of the application process was understanding how I could evidence and measure the impact of the work. Working with CfRC, we defined the outputs from the project (a freely available report) and defined some initial metrics concerning its dissemination – such as the number of views and references to the report. However, the desired outcomes are to be achieved over a much longer period. The ambition for the project is to create interest in a ‘trust-based’ alternative to credit scoring and to advocate for further exploration, which could lead to changes in policy and practice. As such, the wider the distribution of the report, the greater the chance of other institutions and stakeholders following up on the work done.
As a start, I am now working with CfRC to plan and hold a stakeholder event at the University of Nottingham in March 2025. The event will provide an initial opportunity to share our findings and discuss potential next steps. We hope to engage academic, industry, regulatory, and third-sector stakeholders and to catalyse longer-term working relationships on this topic. A link to express your interest in attending can be found at the end of this blog.
The impact activity I proposed aims to create a positive impact in the credit industry by fostering global economic performance, increasing the effectiveness of policy, and enhancing the quality of life for consumers. The activity builds on results from my PhD and the FMR Project. The output of this activity describes a new approach to default prediction based on trust and disclosure by applicants, providing an alternative avenue for the future of default prediction and risk analysis.
Reflecting on my experience, I found applying for and subsequently carrying out the impact grant’s work, has been a great development opportunity. I learned how to create a grant application and lead a research project in a much more independent manner. But more importantly, it has given me the confidence to keep pursuing projects that I deeply believe are important (to me and others). I have since applied this in other areas of my life — looking out for funding opportunities and new partners or figuring out ways to make projects happen with the resources already available. I cannot overstate how much I recommend this experience. It has been incredibly fulfilling to apply the skills I have been developing for the last five years to work that I truly believe can have a long-lasting impact on society for the better.
Stakeholder Event:
Insights from the FlexMyRent trial – Wed, Mar 19, 2025 at 10:30 AM