Redesigning Financial Services: Inclusive Credit Scoring Nick Henry Professor/Co-Director, Centre for Business in Society Coventry University CfRC 2018, London, 25 April
The Presentation Henry, N. and Morris, J. (2018) Scaling Up Affordable Lending: Inclusive Credit Scoring A Report to Responsible Finance. Funded by Oak Foundation. Available under Policy and Research/Publications at http://responsiblefinance.org.uk/ The process of credit scoring The problems with credit scoring: unscoreables, invisibles, thin data, etc. Why these problems? Why does it matter? Inclusive credit scoring approaches: included, upscored and appropriate Case Studies
The process of credit scoring A recognised standard industry approach to assess risk, as the basis for underwriting, charging interest & ensuring appropriate lending in line with business/organisational objectives Seen as an efficient, cost effective, statistically validated, objective and independent assessment of lending risk and an enhancement to the credit analysis process An innovative development that has supported widespread expansion of consumer credit across populations and geographies (although, also associated with rise of sub-prime lending). Led by a handful of global companies utilising similar data sources & scoring models Recent years seen growing attempts to educate the consumer to manage their credit score ( your data self ) But when you look under the hood
The problems with credit scoring Dimensions of exclusion for LARGE numbers of consumers and which are NOT directly related to creditworthiness. unscoreable, invisible, underestimated, marginal, etc Experian (2013): 66m, 25% USA adult population unscoreable 10m were prime/near-prime; 40% home owners; largest segment professionals ; income distribution same as those scored! USA Credit Score Competition Act (2017): 26m housing market credit invisibles, especially impacts African American, Latino, young PWC (2016): 10-14m, 25% UK adult popn near prime due to minor blemishes thin : lack of applications, inconsistent address histories, etc. Closer look: 43% never missed a payment in 3 yrs. Only 8% had credit cards, from one of only four providers; underserved Transunion (2016) 300 US institutions: 87% declined could not be scored ; 73% agreed creditworthy customers without access to credit
Why these problems? Life histories and lifestyles today ( structural ) migration and portability; precariat and the gig economy (volatility); millennials and delayed purchasing (car, home, insurance, cards); changing financial behaviours. 20 th century black boxes in 21 st Century realities hence build your credit history/data self and avoid thin files Exclusion of data sources and indicators of the (financial and consumption infrastructures of the) new consumers: mobile phones, utilities, rental payments Poor, out of date and inaccurate data: names, addresses, payments FTC (2013): in USA, 20% had error on their credit reports FCA (2017): compared credit reports of 1.2m HCSTC over 18 months. Differences in reported debt: 1.6bn total, 24%; 1,200 median per consumer; 1.6m items; 21%.
Why does it matter? Consumer Access to credit: declined, reduced choice, underserved Terms and conditions: higher costs, poorer terms and conditions, underestimated Risk: less appropriate and/or riskier lending channels Aqua (2014) Average extra cost per year for 5.2m UK middle income households of a poor credit score was 1,770 Credit providers: Inefficient risk-based pricing and underwriting in relation to business objectives (and market segmentation); increased costs to overcome information gaps Policy makers: viable but unmet credit needs, inefficient markets, greater levels/ gradations of detriment and exclusion; inappropriate lending; - reduced economic and social benefit and increased economic and social costs
Inclusive credit scoring: comprehensive, alternative and inclusive variety of responses to produce enhanced access to consumer finance, including different data sources, alternative data and wider sets of financial and non-financial analyses. Be reflective of and capture creditworthiness within contemporary lifestyles Taken forward by existing credit reference agencies, new entrants, providers, social mission organisations, fintechs, etc. widespread drive (to achieve appropriate and fair finance) Mainstream alternative data: data closely resembling baseline credit data ( payment information from businesses ). Thickens files, potential upscoring, some ability to score Fringe alternative data: consumer behaviour, non-financial, consumer contributed social media, psychometric consumer may or may not be aware of harvesting (GDPR!). Greatest inclusion potential, make scoreable (young, migrant, emerging economies) In 2016, 34% of 317 lenders reported using alternative data in some form ( new markets )
Inclusive credit scoring: comprehensive, alternative and inclusive PERC (various in USA): 1 in 4 of population credit invisible, socially stratified; Minimal inclusion of utility alternative data has substantial impact on visibility, increased credit scores, credit score tier, increased acceptance against existing portfolios ID Analytics (USA): 10 lenders, 2012-2016, across lending markets thin file, no-hit : 10 40% eligible dependent on market, with no increase in risk Top 10 credit card: 5% additionally included at no greater risk; 224,000 individuals Sub-prime/marginal: better and consistent credit management compared to files.
Inclusive credit scoring: comprehensive, alternative and inclusive The aim of inclusive credit scoring approaches is not to lower the bar for credit ratings but, rather, for the full and fair consideration of the widest possible of the population for credit based on accurate and timely provision of data on financial health, ability and propensity to pay back (and, thus, creditworthiness and affordability of borrowing). Will support fairer access and financial inclusion but greater visibility and thicker files may lead to scoring inclusion but continued market exclusion although would support mission-orientated and inclusion-driven policy Devil is in the detail of data source, data collection, metric and indicator: TEST!.
Report Case Studies (see Hand Out) Case Study 1: Thickening through VantageScore Case Study 2: Thickening the Aire way Case Study 3: Deepening through Experian s Trended Data Case Study 4: Comprehensive Credit Reporting in New Zealand CASE STUDY 5: Alternative data: Rental Payments, Big Issue Invest, Experian and the Rent Recognition Challenge Case Study 6: A New World of Social Media. Case Study 7: Credit Where Psychometrics is Due AND Open Banking
Thank you for listening Nick Henry: nick.henry@coventry.ac.uk Centre for Business in Society: http://www.coventry.ac.uk/research/areas-of-research/businessin-society/