Go back to menu

Big Data and the Obsolescence of Consumer Credit Reports

Managing your financial identity in the modern world

16 July 2019

Following the launch of the UK Financial Conduct Authority’s study into the credit information market this article, written by Nikita Aggarwal - lawyer and researcher at the University of Oxford, Faculty of Law and Oxford Internet Institute - highlights key weaknesses in the existing consumer credit reporting system in the UK that will need to be addressed, focusing on the inadequacies of consumer credit reports.

The consumer credit reporting system performs three main functions. First, sharing data about consumers reduces the asymmetry of information in consumer credit markets, whereby credit providers typically know less than consumers about factors affecting the latter’s creditworthiness. This helps to mitigate adverse selection and moral hazard effects that undermine market efficiency. Secondly, sharing information via centralised credit reference agencies (CRAs) improves the efficiency of the credit information market. It also enhances the assessment of consumer creditworthiness: CRAs aggregate the data that they receive from credit providers with publicly available data, such as information on bankruptcies and electoral registration, to provide credit reports and scores to credit providers. Thirdly, CRAs offer these credit reports to consumers (known in the UK as ‘Statutory Credit Reports’), allowing them to better understand, and take measures to improve, their creditworthiness. As such, consumer credit reports are an important tool for educating and empowering financial consumers, helping them to manage their financial identity and thereby improve their access to, and cost of, credit.

The Growing Obsolescence of the Consumer Credit Reporting System

These are the main functions that the consumer credit reporting system is supposed to perform. In reality, however, this system is grossly underperforming — and failing to meet the needs of consumers. As I have argued previously, we are now living in a world in which all data is credit data. Credit providers (including both mainstream banks and alternative ‘fintech’ lenders) increasingly rely on algorithmic credit scoring — leveraging Big Data and AI/ML techniques, and a wide range of ‘social’ and ‘behavioural’ data — to reach credit decisions. This has two implications: first, lenders rely less on the conventional ‘credit’ data that is shared on a reciprocal basis with other lenders and CRAs (namely, credit performance, account transaction and, more recently, utility, telecom and rental payment data). This is particularly true for ‘thin file’ or ‘no file’ borrowers, who lack conventional credit data. Secondly, Statutory Credit Reports offer consumers an increasingly under-representative picture of their creditworthiness: at least in the case of the three main CRAs (Equifax, Experian and TransUnion), their reports reflect only conventional credit data and limited additional categories (such as electoral and bankruptcy information).

This begs the question whether the existing credit reporting system is still fit for purpose. In particular, how useful can Statutory Credit Reports be for consumers in understanding and improving their creditworthiness, if they don’t reflect a large part of the information that lenders actually use to reach credit decisions?   

Not only is the scope of information covered by consumer credit reports inadequate, the lack of explanation as to how information is used by credit providers reduces its usefulness for the consumer. This is not a new problem. Statutory Credit Reports do not currently offer detailed, personalised explanation, for example, of the relative weight given by lenders to the different categories of information on their reports, nor the relative weight of that credit report as compared to other information relied upon by lenders (e.g. collected from consumers directly, or a credit report from another CRA). However, this lack of explanation becomes more pernicious in a world of Big Data and algorithmic credit scoring, where lenders rely on many more, and much more complex, types of data, that often have a much less intuitive relationship with creditworthiness. Unlike your credit history (it is fairly intuitive that if you manage your debts well or avoid bankruptcy your perceived creditworthiness will be higher), it is less intuitive how your social, behavioural data like Facebook and Uber activity correlate with creditworthiness, and more importantly, how you as a consumer can act on this information to improve your credit score.  

Modernising Consumer Credit Reporting

To avoid obsolescence, the consumer credit reporting system needs to be modernised. There are two main aspects of this system that demand attention: the inter-creditor aspect, and the consumer-facing aspect. On the inter-creditor side, consideration could be given to bringing certain forms of alternative data about consumers into the framework for information-sharing between credit providers and CRAs. However, this suggestion needs careful further analysis, in particular, taking into account the competitive dynamics of the consumer credit and credit information market, unintended consequences of over-sharing consumer information (e.g. disincentivising innovation or enabling gaming of the system), as well as the implications of new data portability mechanisms under EU data protection and payment services regulation (notably, OpenBanking).

On the consumer-facing side, consideration must be given to modernising the form and content of consumer credit reports. This could involve adding the main categories of alternative data that lenders rely upon to credit reports — for example, a social media ‘score’, based on a consumer’s social media activity. It is important to note in this regard that the main CRAs — notably, Experian —are also key players in the data brokerage market and purveyors of much of the alternative data and Big Data analysis that credit providers increasingly rely upon. Notwithstanding the questionable legality of this market, their prime position makes them well placed to reveal more to consumers about their ‘data selves’.

More radically, consideration should be given to transforming the format of the credit report. A particular weakness of existing Statutory Credit Reports is that they are static, lagging indicators of creditworthiness, and often inconsistent inter se: traditional credit data are shared with CRAs only on a monthly basis, and not all credit providers share all data with all CRAs. A possible solution could be to disintermediate CRAs and replace the Statutory Credit Report with a ledger on which all consumer data can be shared, updated and accessed in real-time, by both consumers and lenders — as for example in the Bloom credit-chain (banks, such as RBC, are evidently exploring similar solutions). A more decentralised, user-controlled data governance infrastructure, such as a credit-chain, could also facilitate the implementation of consumers’ data portability rights. 

Modernising the consumer credit report in this way is necessary to satisfy the requirements of EU and UK data protection law. In particular, consumers have a right to access all personal data being processed about them, which in relation to CRAs is treated as all personal data relevant to a consumer’s ‘financial standing’ (s 13(2) UK DPA 2018). In a world where all data is credit data, meeting this standard clearly requires CRAs to provide much more information than they currently do in consumer credit reports. Likewise, consumers have a right to receive ‘meaningful information about the logic involved’ in any decision taken using ‘automated means’ (Arts 13 and 14 GDPR). At the very least, this should require consumer credit reports to reflect more of the data that credit providers actually rely upon, as well as the relative weight placed by credit providers on different categories of data in reaching credit decisions using algorithmic credit scoring.

Conclusion

This is not an unqualified endorsement for the use of Big Data and AI/ML in consumer credit decision-making. As I have argued previously, serious attention should be given to concerns about reliability and fairness when using alternative data (notably, social media data), as well as strengthening the governance and oversight of algorithmic, data-driven tools. There are also ethical, privacy and security concerns relating to sharing personal data about borrowers, and the concentration of this information in the hands of a few CRAs. Yet, even adjusting for these concerns, credit decisions will continue to rely upon more information than is currently revealed to consumers in their credit reports. As such, there is no escaping the fact that, in the absence of reform, consumer credit reports are rapidly becoming obsolete. They must be modernised if they are to meet the needs of consumers to manage their financial identity and leverage the opportunities that flow from this, not least gaining access to credit on fair terms.

Written by Nikita Aggarwal - lawyer and researcher at the University of Oxford, Faculty of Law and Oxford Internet Institute. 

First published on the Oxford Business Law Blog, and republished here with permission.