Credit scoring in Nigeria, with Jes Freemantle

In the very first episode of HTLMTS, Raymond Anderson mentioned a consultant who’d served as something of a mentor to him – Jes Freemantle. So it feels fitting that, to wrap up the year and the first season of the show, we have Jes himself on, talking about his current project building bureaus scores in Nigeria, as well as a career spent as a scorecard pioneer.

Jes runs Stratus Credit Consultancy and can be found via LinkedIn

You can learn more about Credit Registry and their SMARTScore at https://www.creditregistry.ng/category/smartscore/

You can learn more about myself, Brendan le Grange, on my LinkedIn page 

If you have any feedback, questions, or if you would like to participate in the show, please feel free to reach out to me via the contact page on this site.

Regards,

Brendan

The full written transcript, with timestamps, is below:

Raymond Anderson 0:00

We had a consultant for many years with Standard Bank, his name was Jes Freemantle, and it was possibly a decade, if not a bit more, that he was guiding us. He was for all intents and purposes, my mentor. I learned a lot from him...

Brendan Le Grange 0:26

If that voice sounded familiar, congratulations. You're both an observant and a loyal listener, because that was a clip from my very first episode with Raymond Anderson, essentially introducing today's guest for me, because if someone of the stature Raymond calls Jes a mentor, that I surely had to learn more, this is How to Lend Money toSstrangers with Brendan le Grange.

Jes Fremantle, owner of Stratus Credit Consultancy, welcome to the show. As I said in my intro, it's great to have you on to close the loop that Raymond opened in my first episode, but not just for that, you've been building scorecards for over 35 years now, technically working out of your home base in Nottingham, but you've travelled far and wide for your craft. So before we get into your latest project in Nigeria, let's start by looking back at your career. How did you get into credit scoring? And where has it taken you?

Jes Freemantle 1:44

Sure. Well, we have to wind the clock all the way back to 1985. I had just left York University with a maths and stats degree, and to be honest, I didn't have a clear idea of what I wanted to do with my life. And I had about four or five interviews lined up shortly after graduating. Some of those were with large groups like the Mars Group, a tobacco company I remember, even my backup plan was the McDonald's burger chain as a trainee manager.

But I had one other interview with a company at the time who were called CCN, now known as Experian, and have been for many decades. When I went into that interview, I didn't really know what it was they did, because back in the mid 80s, this whole infrastructure of credit decisioning was quite a hidden world, it wasn't in the public domain. So I remember being sat in the foyer picking up brochures, desperately trying to work out what this company did before I went into my interview. But what really swung it for me was more of a personal connection with the man who interviewed me, Roger Aubrook, I really thought to myself, I can work with this guy. And it was on that personal level that I decided to accept their offer.

It was very early in the days of credit scoring, and Experian or CCN, as it was, that wasn't a part of their first intake. Their first in-take of staff in this area, were very high calibre people that had come from Oxford and Cambridge University. And in some respects, I think I was a bit of a champion challenge, a test of what what happens if we take on someone who didn't go to Oxford and Cambridge because I didn't, I was at York.

So that's how I started, and I was extremely lucky in several ways. Firstly, the people that were training me were extremely good. They were, in my mind, the forefathers of this discipline. And the training at that point was very informative was very much hands on, you were thrown into projects, I would sit next door to people, Charles Taylor was one who was very instrumental in training me, Michelle Lundy, Simon Harben. So I really learnt from a hands on point of view.

And I think I was also lucky that it was the beginning of the credit scoring discipline. And because back then in time, you have to remember, we didn't even have automated databases. So projects would involve gathering data from application forms. So you really had to get sort of down and dirty with the weakest link if you liked - where the data comes from. We had to deal with lots of issues regarding inconsistency in the data, you know, some lenders would have 20 different types of application form. And some questions would feature on some forms, but not others. I also remember that, you know, we had a computer called PerkinElmer that we were building models on that was the size of a small submarine where you would submit a regression job on a Friday evening. And you'd be lucky if it completed by Monday morning.

Brendan Le Grange 4:36

I think we used to have long queries like that when I started that it was because we hadn't learned how to index tables yet and our code wasn't great. You at least you've had a proper excuse

Jes Freemantle 4:46

Yes, but I think whenI compare the environment I grew up in compared to the environment that people coming into the industry now, with the lack of tools, everything was written by base code. We were literally dealing with the physical paper forms that were coming in to create our samples. I think it was a benefit, although it's very, very inefficient compared to modern automation. It gave you an insight into what you were doing in a way that I don't think happens so easily today.

But anyway, I started off as an analyst, being trained to build scorecards, and then eventually, I was promoted through various stages into having a team of about 30 or so people underneath me who I was managing. I was responsible for their Scandinavian market, but I had this sort of growing unease that the more I was going down that progression, the more I was becoming detached from the technical work, which is what I really loved the most.

So after being at Experian for about 10 years, just under 10 years, I saw an opportunity to go freelance - because back then, the banks were starting to take credit scoring in-house, they were no longer as comfortable to have the likes of Experian and FICO developing models on their behalf. So they were starting to create their own internal teams and I could see that there was going to be a training requirement. And at that time, the main credit bureaus were a bit reluctant to train lenders to do what they were offering to the market themselves. And this is 1995. It was good timing, from my point of view.

I was full of the over-confidence of youth, because I resigned from my job at Experian, and hadn't at that point secured funding for what I wanted to do. Now I didn't need much funding, I needed to buy a car and a laptop to survive, I thought maybe up to six months. And if it failed, I would go back with my tail between my legs and and hopefully get another job. But I needed about £10,000. So I approached my bank having already resigned, and I gave my business case to my local bank manager and I explained what I was going to be doing, creating these automated systems for making credit decisions that essentially eventually would make his job rather redundant.

He rejected me. So I didn't have a line of funds at that point to launch the business. So I was very lucky that my first client, a company called Ikano, the holding company from Ikea, agreed to part-fund me upfront. If it wasn't for that, and in particular, a gentleman called Jens Rom, I'm not quite sure what I would have done.

So anyway, I got I got over that hurdle, and started consulting, to begin with in territories where I was known already, particularly Scandinavia in the UK. But eventually that took me into other markets - in particular to South Africa. And that started off as a best practice audit, I was called into assess where they were, where they needed to be, and to create a sort of a five year plan for the people, the tools, the databases, they needed to get to sort of first class territory. So I did a an assignment along with another consultant in slightly different space to me called Helen McNab, who has a company called ScorePlus out of Bristol in the UK. She dealt with more of the sort of operational side of things, whereas I was dealing with the credit scoring side - which the management of the bank then asked me to help to implement.

I was going down to the Johannesburg eight times a year to begin with, I would come in each trip, and basically just be given licence to roam around and to sit next to people see what they were doing. Talk to them, look at the documentation, they were producing, the way that they were approaching the credit scoring problem. And then at the end of each trip, I would write a report with all of my findings. It continued much longer than than I ever thought it would, it went on for definitely 10 years, maybe 12 years. The whole point of it was it was a knowledge transfer exercise that they would eventually become free of me. That happened about six or seven years ago, that contract finished. And I guess for me, it was the most satisfying engagement of my professional life as an independent.

So after that, I took a little bit of a hiatus. And then I was approached by an ex-colleague of mine, who was a Nigerian national who had worked for Experian back in the 90s. And he introduced me to a gentleman called Taiwo Ayedun, who had founded one of the three credit bureaus in Nigeria. And they were looking to do all the value added services the likes of Experian/ Equifax have been offering for years. So it was a bit of a departure for me in terms of - South Africa was in a way, although it's a developing market, it's still fairly westernised, if that's the right word, whereas Nigeria is much more at the beginning of that journey.

Brendan Le Grange 9:50

No, I know what you mean, much of the South African banking and borrowing infrastructure is very well developed. I think because of our history and the fact that our income is more unevenly distributed, more of the addressable market sits in that first world part of the economy. So it's quite easy, for the traditional big banks at least, to largely target that. Whereas from my outsider's view of Nigeria, the only way to compete there is by trying to overcome those infrastructure and systematic issues that you think of when you think of a developing market.

Jes Freemantle 10:22

Well, there's a separation between formal lending and informal lending - around half of adult Nigerians possess a bank account. Within that half, there's something of a gender bias, about a quarter of females have bank accounts. And that's something that my client is keen to redress, and to then encourage lenders to embrace female borrowers more than they have them. So there's a lot of informal lending, whether that's within families or local associations, which are really off the radar from, from a credit bureau point of view, they don't have reporting, a lot of it will be paper based, you've then got five or six main commercial banks, who think it's fair to say a little bit slow to embrace mass consumer lending. I think one of the problems that's held them back a bit is that a lot of their savings deposits are converted into government bonds, and they get good returns on the savings so there hasn't been a huge incentive to sort of act like a retail bank would do. In the UK, for example of balancing saving and borrowing activities. And obviously, you know, to create those borrowing activities requires a lot of investment, you know, a lot of infrastructure, a big project. And if the financial incentive to do so isn't there, then it's, I think, held them back.

But that has created something of a void that other players have come into particularly fin tech companies, those that are routed overseas, coming in, there's a lot of a lot of activity in that space.

Brendan Le Grange 12:01

I think it's always great to hear when fintechs are taking off. But especially now, because that first wave of African FinTech, what I think of as the M-Pesa cohort of innovators, they were often finding innovative ways to solve infrastructure problems. So leveraging SMS technology to deliver micro payments and basic banking services, because physical banking networks were too expensive to run, they solved a lot of problems. But those problems were fairly specific to developing countries, which limited their export'ability. But this current wave built around smartphones doesn't have that same issue. So I think the folk that solve modern financial problems in Lagos can just as easily solve them in London. But the flip side to that is the entrance of foreign fintechs, which you've mentioned, and it sounds like there may be a fairly unique situation brewing in Nigeria, with these big banks who have been sitting on the sidelines, coming up against fintechs that are not just more nimble than they are, that may actually have more lending experience.

Jes Freemantle 13:07

I totally agree that we are seeing signs of change. One of the things that has kind of held them back a bit has been a reluctance to lend. It's not just a question of lending to non existing customers, there's a question of lending to your own customers where you don't receive the salary into a cheque account. That has been a very key rule that a lot of the main banks have apply, that they'll only lend to customers where they receive salaries. So, how to lend money to strangers, well, that's very relevant. A lot of the customers haven't been lending to actually not strangers, they just don't receive their salaries. So they've been overly cautious and not embracing risk. There's an acceptable level of risk. And that concept has been quite hard to swallow.

Brendan Le Grange 13:58

You're listening to How to Lend Money to Strangers with Brendan le Grange. I'm speaking to Jes Fremantle. And if you're enjoying the content, now is the perfect time to hit that little plus button to subscribe. Now back to the show...

Jes Freemantle 14:13

In terms of other credit products, I mean, very limited penetration of credit cards at the moment, a couple of players but that's still very early doors, not much in the way of buy now pay later that we might see in the Western markets. I think partly because the retail landscape is much more fragmented. There aren't so many large major brands, as we'd find in the UK.

Brendan Le Grange 14:37

That's an interesting take on it. I hadn't actually thought of it that way. I've been talking to a buy now pay later in Ghana who use a hybrid model. You go into a physical store and when you're ready to purchase, you take out your phone and open the app and go through the application for financing. And I'd figured it was to overcome the fact that most retail in Ghana is physical still, but you now that you mentioned it, the other side of that is also the negotiation, right? If you want to do partner deals and buy now pay later in the UK, you can cover a big chunk of the market by going to maybe 10 huge brands, whereas in Ghana, it might take you 10,000 negotiations. And it sounds like that's in Nigeria, too. We are seeing the emergence of completely standalone buy now pay laters, like Zilch, but yeah, it's an interesting space to watch.

Let's turn the focus back to you, though, as you said, you were approached by Taiwo to help build out the bureau's core products at Credit Registry. And as you also said, it's quite a different data landscape there. So how has it been getting the SMARTScore off the ground in Nigeria?

Jes Freemantle 15:45

It wasn't so much a case of getting it off the ground, because my client already had a an existing bureau score, but it had been a while since it had been updated. So my first task was really to revamp their bureau score. And that involves developing initially a retrospective environment where you could go back in time and gather the data that was present at that point in time. I draw upon my global experiences to create this sort of list of characteristics, we've got about 400 different characteristics. And we released the new version of SMARTScore about 18 months ago. So its still the early days, but it's definitely, particularly in the thick file segment, a very predictive tool.

I'm not entirely certain if this is novel or not. But one of the features of the score that that I developed is that there's a clear split between thick file and thin file, the SMARTScore that we offer comes in two separate ranges, so a two digit score between 10 and 99 is a rank order of the thin file population and a three digit score between 100 and 999 is the rank order of the thick file population. A score of 99 is not worse than a score of 100, because it's in a totally different space. So we are pretty open in terms of where we've got sufficient data to come up with a strong estimate of risk. In other markets, I've seen situations where there isn't that distinction, every everyone gets a three digit score, and yet, you'll find that there's a certain score value that's a constant in the scorecard, and you got a big clump of cases scoring on that particular score, who are essentially your thin file population. So I prefer it as an approach that's more transparent.

In terms of the the infrastructure that is used, the software that the credit bureau runs off, it's definitely of first world quality. It's licenced from the US and it's at least as good or possibly even better, in some regards, to what I've seen even in the UK - in terms of the ability to access data, lenders ability to inspect data is excellent. We've created a - and I say we, my client - has created a database of over 60 million entities. You know, whether those are consumers, some of those also SMEs.

The regulated commercial banks are obliged to supply data to us so we have 100% coverage there, the micro finance sector is more voluntary. So we have a proportion of the microfinance data on our books, rather than the entirety of it. And we also have a data roadmap in terms of future data assets that we're looking to acquire, be that through telco or retail or other types of financial institution. Data quality in Nigeria is clearly an issue, and we are to an extent dependent upon the quality of the data that's provided to us, but that's also been an aspect of my work, how you control that how you identify irregular, what looks like irregular data or spurious data.

So we've created a first world infrastructure, but it's been an uphill battle to get lenders to embrace the use of bureau data and credit scores and bureau scores to make mass decisions. In Nigeria, it's a legal requirement that you perform two credit inquiries, but that's not to say that you must make good use of the data you're given, as long as you know, you've met your legal obligations. So there hasn't really been an appreciation of how that data can help you improve your decision making. So that's the journey that I've been trying to help my clients to realise - it's been as an educational upskilling, a sort of training project as much as much as a hands on rebuild the bureau score project.

Brendan Le Grange 19:36

Yeah, that's often a problem in a new market. People can become so used to the idea of a credit bureau as a tool just to identify that small but expensive group of consumers who should be declined, that they build their businesses around that. And then the only way that they can actually leverage all these extra insights you're providing is by updating their business is by building new risk-based strategies.

Jes Freemantle 20:00

Yes, yes. So yeah, my emphasis, I guess, has changed in the early part of my contract I was hands to my computer, you know, designing the retro environment and developing the scores. And since that's been done, my emphasis has been very much on client engagements, because it is holding the country back. The reason why I joined this contract - in some respects Nigeria's, call it one of the most challenging markets in the world for someone who does my work. But when when I first had the discussions with the founder, he was educated in the US, although he's Nigerian, and when he was in the United States he became aware of this infrastructure regarding credit decisioning. His curiosity was why why what's happening here? And he was like, well, why don't we have this in Nigeria? Is this is this what's holding us back? We don't have these tools to enable people to trust each other, to transact. So that was my conversations with him, it became clear that he is very much coming from the right place in his heart, he cares about his country a great deal. And he wanted to help his country progress by having the sort of risk assessment infrastructure in place. So that's why he founded the bureau.

I think one of the cultural changes that's required is the willingness to invest money in order to save money, that thinking hasn't really been that prevalent in Nigeria, in the past at least. Curiously, that's the first time I've ever encountered a situation where lenders have questioned, well, why are we paying for a credit inquiry if we ended up rejecting that customer? Why would we want to pay for that? Which kind of misses the point of the protection that screening for risk gives you.

Another thing that's been holding back the main banks has been a tendency to operate at a regional or local level, rather than to be centralised. And just basically too many manual processes, the amount of time it takes to make a credit decision, measured in days rather than minutes. There are several factors which are which are holding back the main banks in terms of mass lending, and none of them are easy solutions,

Brendan Le Grange 22:09

Jes, it's Nigeria now nut you've built scores all over the world in familiar and exotic locations, and over a period of time that almost spans the entire lifetime of the concept. What has this taught you about the science and art of scorecard building?

Jes Freemantle 22:25

Yes, when you think about the scorecard development process, they're really sort of three main components to those the statistical process that you're using. So that might be logistic regression, or random forest, probably up to 10 that I've encountered. There's the human oversight, the analyst who's performing the work. And then there's the data. Of those three components, in my opinion, others may disagree with me, but in my opinion, the statistical methodology is the least important. Now this kind of flies in the face a little bit of modern FinTech in the sort of artificial intelligence machine learning where increasingly human intervention is being removed from the process.

But I would say that credit scoring, actually is a form of machine learning. It's highly supervised machine learning. And what worries me if I look at the general long term trend is that the human involvement is gradually being phased out. And that may be possible in warehousing projects or facial recognition projects, but the data that we work with is heavily affected by big business processes, which sit upstream of it, whether that's marketing, or collections, for example.

So changes in those environments, change the data. And without having that context, that thinking behind, well this predictive pattern I'm seeing, can I trust it? Is it the result of some marketing strategy, some product change, some limit change, some collections change - that kind of human context thinking, I don't think AI will ever get to that point. It will tend to trust the data that it sees in isolation from what led to that data being there. So I'm firmly of the belief that the skill of the human analyst and the data and the types of data that you get hold off, that's where you're going to get your benefits, rather than some new statistical technique. I mean, I've tried lots of different statistical techniques. And broadly, they give very similar results. That's not where you're going to get the quantum leap.

For me, that's my opinion. There are others who would agree and some would disagree. A good analogy for me is that the statistical method is a tool at the end of the day, it's a tool like a spade. Now, you'll get experts in spade design, who will know exactly how you're moulded to handle and forge the blade and the woods and actually If you're too much of an expert in space, you don't think about the hole you need to dig. And that's that's the kind of skill that I think is we're essentially we are using a tool to dig a hole, possibly a bad analogy in terms of the hole, but it's the application of the tool - that is the important thing. And if you think too much about the tool itself, you don't think about the hole you're trying to take.

But I think in the long run, unfortunately, the machine is going to win. Because there'll be less people like me, who've known any other way. So maybe 20/ 30/ 40 years time, there won't be people with enough knowledge to question whether this is the right way to go.

Brendan Le Grange 25:43

Jes, thank you so much. It has been a pleasure and an education. If anyone listening needs help building their own scores or mentoring their scorecard teams, how can they reach out to you to learn more?

Jes Freemantle 25:55

While I'm on LinkedIn, you can search for me on my my my company will come up if you inquire against Jes Fremantle. Sometimes I'm down as Jeremy, my original name, so my internet footprint is kind of split between Jes and Jeremy but anyone who Google's my name will find me.

Brendan Le Grange 26:11

And thank you for listening. And not just to this episode, but to the first season of How to Lend Money to Strangers, which is now wrapped up. The first show of the new season will drop on Thursday 6January 2022, a deep dive into ConfirmU, the new business I've joined to gamify access to credit among the unbanked. So until then, have yourself a great holiday season and make sure you've subscribed so you don't miss it.

How to Lend Money to Strangers is written hosted and edited by mysel,f Brendan le Grange, and recorded outside the actually not that rainy city of Maidstone, England. The theme tune and show music is by Iam_wake you can find show notes, written transcripts more in depth articles and details on how to book me for speaking engagements at www.howtolendmoneytostrangers.show.

Hi, it's me again, just in case you've had your full of lending talk. Did you know that I've also published two pulpy action adventure thrillers. Drachen and Butterfly Hill are both available as ebooks, paperbacks and audio books from Amazon and other online retailers are not Shakespeare but they're not expensive either. And BestThrillers.com called Butterfly Hill, an exotic, utterly captivating heist novel featuring one of the genres most entertaining detectives