Modelling, scoring & decisioning: here’s why lenders need alternative data

Portfolio management has played a crucial role in every successful financial organisation throughout history. But, gone are the days of relying on outdated and limited datasets for credit scoring. Today's world requires something more.
Published on
June 28, 2023
Author
Category
Finance & Fintech

Since the dawn of civilization, credit has been a cornerstone of economic activity. From the ancient Mesopotamian grain loans to the birth of the modern credit card system in the 20th century, the mechanics of credit have continuously evolved to better meet the changing needs of society.

Today we stand at the precipice of another significant shift in the credit landscape, one that leverages the power of real-time current account transaction data, and reduces the risks of depending on limited and non-representative datasets. Furthermore, due to economic shifts on a global scale, the potential unlocked for lenders to make better data-driven risk decisions has happened just in time.

The evolution of consumer credit scoring

The modern concept of consumer credit came into existence in the late 19th century with the establishment of retail installment payments. By the mid-20th century, credit cards were introduced, paving the way for a new era of consumer borrowing, and encouragement of a buy-now, pay-later mindset as part of the post-war boom times.

With this increasing ease of access to credit,the need for a system to assess creditworthiness became more crucial than ever. So, in the 1950s, the first credit bureaus began to emerge, collecting and selling data about consumers' borrowing and payment habits. A decade later, the Fair Isaac Corporation (FICO) introduced the credit-scoring model, that still serves as the basis for most of our current credit rating systems.

This model, primarily based on five factors - payment history, amounts owed, length of credit history, credit mix, and new credit - was a massive leap forward in assessing credit risk. Finally lenders had some means of comparing risks of default between different applicants.

However, the traditional credit-scoring model does have some major drawbacks in its predictive validity, and deserves better inputs, to serve the importance of its outcomes. For example, there is a simple lack of consistency between data sets, with some agencies only reporting negative data, and bureaus themselves sharing data incompletely on a basis of reciprocity.

As Clare McCaffery, CCO at DirectID explained:

That's fine for people like me who've had a mortgage forever and got a credit card, and know how the system works. But if you're new to credit, then there is no history on you.

This data fragmentation typically disadvantages certain groups systematically, including young people, and mobile/migrant populations.

Today such credit invisibles with thin files may be highly paid mobile professionals, for example, who relocate to a new area where their skills are sought after - and where they have no borrowing history. They also include people with diverse and complex income streams, who are innately resilient and flexible money managers as a result. Many of these may be highly creditworthy individuals who just don’t check the boxes, and represent lost business for lenders.

Furthermore, even when consumers dutifully accumulate a good credit score, it’s essentially always out of date anyway, as McCaffery continued:

One of the other issues is that data is always contributed a month after the effects. So if someone defaults on their mortgage, then that data is only contributed by HSBC a month after that default happened. It’s not real-time current up-to-date data… Also, lenders contribute data at different times of the month, so it's not as the default happens. It could be they're getting data from HSBC on the 1st of the month, then data from Royal Bank of Scotland on the 15th. So, it's a little bit messy.

Messing further with the data: local and global shocks

Unforeseen events such as job loss, illness, or divorce can dramatically alter an individual's financial condition, but these changes may not be reflected in their credit score for months, even years.  

And of course, unforeseen events can happen on a global, rather than individual scale. The seismic shock of the COVID-19 pandemic disrupted consumer activity in every market, and drastically altered consumer spending behaviour.

Large-scale lay-offs created a new class of consumers who, despite possessing historically good credit, have become high-risk due to sudden job loss. Meanwhile, to combat economic contraction,governments worldwide injected unprecedented fiscal stimulus into their economies. While these measures were crucial in staving off financial catastrophe for many households, they also introduced lasting anomalies in credit data.

A study by the Federal Reserve Bank of New York found that credit card debt fell by a record $76 billion in Q2 2020, largely due to stimulus measures and reduced spending opportunities[1].

However, this reduction in consumer debt didn’t correlate with an improved ability to service new loans, as it was largely contingent on temporary government aid rather than an increase in stable income. In the UK, credit scores trended upwards during the pandemic[2] - indicating a growing disconnect between credit data and actual credit risk during this period, and systematically invalidating the dataset on which such assessments depend.

The need for alternative data

While the WHO talks of a post-pandemic world, other geopolitical shifts seem to be with us to stay. That includes the sheer pace of change. A further legacy of the COVID era was certainly to highlight the fragility and interconnectedness of our global economy and supply chain. From war on the edge of Europe, to a big ship stuck in a small canal, financial ripples are felt fast. So when it comes to making profitable lending decisions, relying on distorted and outdated sources of consumer creditworthiness, is a risk too far.

Open banking, the practice of sharing financial information electronically, securely, and only under conditions that customers agree to, has emerged as a potential solution to these problems. Analysis of direct spending information enables lenders to access real-time current account transaction data, offering a comprehensive, up-to-date view of an individual's financial health.

This real-time transaction data provides a wealth of information that goes beyond the traditional credit score. It allows lenders to assess a borrower's income, expenses, savings behaviour, and overall money management skills, as well as immediately signalling changes in their situation which could impact on their ability to repay. Moreover, it can capture the financial picture of those credit invisibles, potentially opening up credit opportunities for this underserved population.

DirectID’s APIs enable credit applicants to consent to share real-time current account data with their prospective lenders, unlocking valuable insight into their real ability to pay, and thus achieve a fair and appropriate service.

As McCaffery explained:

You can tell an awful lot about the consumer and how they interact with society from their bank statement… It’s in real time, completely up to date - DirectID can allow lenders to update and refresh the data up to 4 times per day, so you can seethe delta between anything that’s occurred through the day, as well as the final balance.

Lenders can see if there is any other income, such as dividends or child support payments, as well as seeing whether someone can really afford things. Where they’re really spending their money. And you can get this view directly from the consumer, without them having to access credit or go into debt, to generate a traditional credit score.

A new portrait of a new consumer

Compared to the broad brushstrokes of traditional credit scores, what emerges is a photorealistic high-resolution image of the customer, and all their income and outgoings. This is particularly important for those seeking to understand and serve the new generations of consumers, who not only lack detail on their borrowing file, they may also be running up new forms of debt - such as through the retail By Now Pay Later apps like Klarna or Zilch.

These point-of-sale credit apps don’t presently report to the credit bureaus, and can represent a significant chunk of future funds committed, over one brief period of binge shopping. Now that Apple is rolling out a native BNPL scheme offering[3] the option to run up consumer debt at the tap of a screen will soon be available to vastly increased numbers of consumers of all demographics, further driving the need for up-to-the-moment detail on exactly what spending is going on.

It is to be hoped that the insights that open banking provides will also help some to become more responsible and educated consumers, using budgeting apps which tap into the same sources. And third-party debt management providers can also analyse financial data holistically, to help people pay off debts quickly and efficiently.

Meanwhile, by taking ownership of their transaction data and managing its access consents, consumers can take an active part in a more transparent and accessible fintech landscape, whether they seek peer-to-peer lending or other non-traditional forms of funds for business innovation and similar ventures. They can become more mindful of their financial habits, and develop better money management skills. There will be greater incentive to do so when, instead of taking months to correct a poor credit score, they can instead demonstrate real-time responsibility and prudence which reflects rapidly in better access to attractive financial services.

New opportunities for lenders

The use of real-time transaction data in lending decisions offers more than one benefit for lenders, beyond the obvious one conclusion that increased accuracy could lead to lower default rates,reducing the cost of credit losses for lenders.

By being able to serve 'credit invisibles', lenders can tap into a vast, previously unreachable market. According to Experian data in 2022[4], over 5 million British consumers alone are excluded from the best rates and deals, offering a huge opportunity for lenders - once they are derisked, in a data-driven way.

McCaffery explained:

To demonstrate this we create a swap-set for them, using data from consumers they declined [for failing a traditional credit check], as well as those they accepted. We can directly calculate the loss to their portfolio, not only from those we predict are likely to default, but the lost revenue from all those potentially reliable borrowers they turned down.

Going further with the bank transaction data supplied, this can then be modelled against known credit score data, reading a differential over the bureau-issued scoring. This new swap-set effectively scores the credit score, creating predictive ratings for anticipated deviation from the traditionally derived risk assessment.

A further competitive edge to consider: using data-driven modelling for consumer credit scoring is simply faster, than anyother way. With real-time access to financial data, lenders can potentially make instant credit decisions, reducing the wait time for consumers and improving the overall customer experience, according to research from Deloitte.[5]

When decisions are reached promptly, it reinforces perceptions of closeness and cooperativeness with the brand inquestion, making the consumer more likely to deepen their financial relationship with that institution in future, rather than shopping around.

It’s time to change the dataset

The advent of open banking standards and real-time transaction data provides lenders with an unprecedented opportunity to revolutionize their credit underwriting processes, moving away from the constraints of traditional credit rating agencies, and transitioning to real-time data-driven decision-making.

All the critiques of traditional credit-scoring systems indicate a pressing need for change. The use of real-time transaction data not only addresses these critiques, but also offers substantial benefits to both lenders and borrowers.

By embracing this data-driven lending approach, lenders can make more accurate, faster credit decisions, and tap into unserved new market segments. Simultaneously, borrowers benefit from improved access to credit, personalized lending, and greater financial transparency. It's a win-win scenario that heralds a new era of credit underwriting, characterized by increased inclusivity, efficiency, and customization.

From the Mesopotamian grain loans to data-driven lending, the credit industry's evolution reflects society's continuous march towards progress. As we navigate this digital transformation, it's evident that those who adapt will be at the forefront of the next chapter in the history of consumer credit, whatever changes global geopolitical and economic circumstances have in store.

[1] https://www.federalreserve.gov/aboutthefed/files/newyorkfinstmt2020.pdf/

[2] https://www.consumerfinance.gov/about-us/blog/office-of-research-blog-credit-score-transitions-during-the-covid-19-pandemic/

[3] https://www.apple.com/newsroom/2023/03/apple-introduces-apple-pay-later/

[4] https://www.experianplc.com/media/latest-news/2022/meet-the-5-million-credit-invisible-brits-still-at-risk-of-exclusion-from-the-financial-system/

[5] https://www2.deloitte.com/content/dam/Deloitte/us/Documents/financial-services/us-consumer-experience-in-banking.pdf/

Stay up to date with the latest DirectID news

Get the latest news, thought leadership, product updates and customer stories straight to your inbox - subscribe to our blog today.