Open Banking could help improve financial inclusion and access to credit. If a bank or a consumer credit institution, has access to a customers bank transaction data, it can assess their ability to repay a loan in a much more efficient way.
Especially compared to traditional credit scoring, which «is sort of right on average, but it is wrong in every individual case», explained Gerald Chappell, CEO and co-founder of British fintech Abound, during our interview at the Web Summit 2023 in Lisbon.
AG. Gerald, what is wrong with traditional credit scoring?
GC. Traditional credit scoring is embedded into almost every kind of consumer lending decision in most European countries, in the US… pretty much globally. And the problem with that is it is a very crude metric for understanding someone’s credit risk.
It is sort of right on average, but it is wrong in every individual case. And this results in a lot of people being excluded from affordable and low cost credit.
So many credit decisions are still based on 1980s technology, a logistic regression of the credit facility that you have had in the past. But today we can have access to new data sources.
AG. And that is why you decided to found Abound…
GC. Before joining Abound in 2020, I was a Partner at McKinsey, where I had two roles. I was their Head of Digital Lending and their Global Head of Credit Analytics. We did a lot of work with large financial institutions and banks globally on reinventing and improving their lending businesses.
And it struck me for a very long time that consumer credit could be done a lot better, if you could remove the reliance on traditional credit scoring.
AG. How?
GC. Our idea was to look at people’s true financial situation by doing a financial X-ray of them: by taking their bank transaction data, we would be able to understand their income, their expenses, their true affordability.
And from that, we could make much better credit decisions. Traditional scoring looks at your past, but it doesn't actually know what you are and your spending behaviour, nor your financial stability.
AG. How did you turn this idea into reality? How does Abound work?
GC. We wanted to launch a service that uses bank transaction data to properly understand people and make fairer decisions. So to be able to offer lower interest rates and, basically, say yes to more people.
And we wanted to do this as a credit technology business, building this kind of underwriting capacity for banks and lenders, and providing them the technology so they could serve more customers.
But we very quickly realised that no one would believe that it would work, unless we proved it first. So we started as a lender and we built the technology for ourselves.
AG. And did it work?
GC. We've done about £200 million in lending so far in the in the UK, and that's kind of ramping up very quickly.
We started lending in March 2021, and since then we have been running at a 70% reduction in defaults versus the “credit score predicted” level of defaults.
So, for every ten defaults that we should have had, we have only actually had three.
This proves that using bank transaction data is a really much superior approach to than just using traditional traditional credit scores.
AG. Is your approach actually improving financial inclusion?
GC. Our customer base is about 20% traditional Prime customers and 80% in the “near Prime” space. That “near Prime” space includes a lot of categories, including some people that are New to Credit.
For example, people who have just graduated and are starting working. Or have just arrived in a country. They have no credit history.
Others have a disrupted credit history: they might have had credit, then they moved away and now they are back. They are facing problems in getting credit, as well.
In some cases, traditional credit bureaus estimate a very high probability of default, wehereas our model says these customers can be very strong and reliable. Some of them have good structural affordability, a stable income and can afford credit.
Thanks to open banking data, we open up to a new segment that would receive no credit, or could get it, but at unfair prices. We can give these people a much, much better proposition.
AG. Are you considering other kinds of alternative data to assess the creditworthiness of an individual, or a small business?
GC. My fundamental belief is that more data is always going to lead to better and fairer decision. So we try and use as many sources as we can.
At the moment, we use bank transaction data and credit bureau data, but we do not use credit scores. We use the actual underlying what they call facility or level keys, level information.
Within that, we also bring in a number of other sources as well. So we can bring in LinkedIn data, where we can where we can match a customer against LinkedIn. And we validate the LinkedIn profiles by comparing the employer on LinkedIn versus the salary that we can see in the open banking data.
You can bring in the education history and the job history of the customer. We also bring through land registry data as well, to know the properties the customers are living in. To some of them, you get to a point where although it would add value, the incremental value is quite marginal.
We know that some firms in other regions of the world have been doing things with mobile data and are using them to make lending decisions. It is an interesting case, in the absence of hard financial information.
But if you can access information like bank transaction data, then the incremental predictive discriminatory power that you get from alternative information is very marginal. We chose not to do that kind of thing because the benefit would be limited, and customers have significant privacy concerns about that.
AG. You said it was necessary to prove banks that it is possible to give credit using bank transactions data instead of traditional credit scoring. Do banks believe you now? Are they interested in working with you or in developing similar tools?
GC. Banks are interested in the concept, but find it hard to execute. You see, it is very difficult to drive change in a big financial services organisation. Very slow-paced, expensive project with a pretty high failure rate.
If you want to incubate a new way of doing things in a bank, you’ve got to coordinate across at least a dozen of different departments, all of which have different priorities.
Any new idea takes a long time to get off the ground, if at all.
Banks would like to change the way they take their credit decisions, and some of them will claim that they are using open banking data, or bank transaction data. But that, to my knowledge, is not really the case. They might be doing it for replacing some processes that they already have.
A typical case would be, instead of asking people to take photos of their payslips and send those in, they ask customers to connect their bank transaction data.
Using the salary income it’s not really using open banking to do credit decisioning. It is not fundamentally building your models from the ground up of bank transaction data, which is what we have done.
The banks for decades have had all the bank transaction data, but they haven’t actually used that in their models.
AG. But do you think they will, in the future?
GC. I think in 3 to 5 years time everyone will have to have to do it for two big reasons.
Number one: it works. We have shown that we get massive outperformance in terms of lower defaults, by using bank transaction data to underwrite. And you can serve a much wider addressable market doing it.
If you can serve more customers with lower credit risk, then it is a profitable business model and if you do not do that, all your competitors will.
Number two: Regulation will drive the adoption of open banking in lending. All consumer Regulators across Europe are worried about financial inclusion and responsible lending. They want to make sure that loans are only given when they are affordable.
You cannot proclaim you do responsible lending unless you look at someone’s bank transaction data.
In the UK, for example, all of the banks use Office of National Statistics data to assess the affordability for a consumer loan. It is about statistics. If you earn X, based on where you live in the country, the bank will assess the average expenditure of a resident in that part of the country and see if the loan is affordable.
Once again: it is all based on an average. It is right on average, but wrong in every individual case. So a lot of these loans then turn out to be unaffordable.
Using bank transactions data you can also intercept financial vulnerabilities, things like gambling, day trading, crypto speculiation and so on. We estimated that in the UK £2 billion is lent every year to problem gamblers. People are basically taking those loans and using them to finance gambling.
At the moment, that kind of behaviour can just be funded because no one's looking at the bank transaction data.
And I think that's a really indefensible position for the industry to be.
AG. What about the new generation of digital challenger banks? Are they culturally closer to an open banking based credit decisioning?
GC. I would definitely hope so, and for the reasons I said before. There are lot of neobanks that have built great deposit franchises, they have attracted a lot of customers. But the only way to make money in banking is on net interest margin.
One of the things we do is partner with that kind of organisations to help them launch lending.
We have an embedded lending offering, where we can basically plug into their systems and do white label lending on their behalf. Or we can just provide them the technology and they can run themselves. So that's really the credit tech part of the business that we're doing.
