Get small loans anywhere, anytime.

Lost but not yet found — female borrowers!

Lost but not yet found — female borrowers!

For the last year, Paylater, Nigeria’s first mobile lending app, has been playing hide & seek with potential borrowers across the country. Stats we have seen show that only 8% of Nigerians have access to credit and thus the other 80+ million others should be ‘seeking’ us out! So far we have had more luck with men, who make up 75% of our loans. We decided to delve into our data to see if there were any clues to this imbalance and go beyond the general truisms that we hear all the time:

  • Women make better borrowers — they don’t like owing and they use money productively unlike men who are likely to squander funds on gambling, drinking and all manner of nefarious activities.
  • Women won’t borrow as much as they are very conservative — they like to live within their means so borrowing is a last resort.

These are the views we hear all the time but, without any judgement of the prevailing sentiments, our view is that 1) the fact that Nigeria’s population is approximately 50% women implies that we are not reaching a significant part of our market and if we solve for this, we get an immediate 25% increase in our numbers and 2) that if it’s true that women are more credit worthy, then we want more of them in our lives as it would imply an increase in our loan book AND a decrease in our default rate — consumer finance nirvana!


Commercial interests aside though, we were haunted by feedback from a small survey we did last year on borrowing habits. One of our respondents Hadiza (not her real name) said she regularly borrowed money from her friends and family members to fund a small trading business but stopped immediately she got married.

Pause and think about why marriage would stop her from borrowing. I bet many reading (certainly in Nigeria) would jump to the conclusion that as soon as she got married, her husband stepped in to support her financially — either to give her the credit needed or with his benevolence she no longer needed to borrow. Wrong & Wrong. Hadiza actually earned more than her husband so, whilst there was still a need to access credit, she felt that her family would look down upon her new husband as not being able to “support his wife” as expected of all ‘good husbands’.

One of our main goals at Paylater was to provide convenience and privacy to borrowers — no agent visits, no trips to the bank, just you filling out your application in the toilet, on the bus, or whilst listening to your boyfriend drone on about his latest business idea. In that regard, given the numbers we are seeing we have failed to either find the thousands of Hadizas that we believe are out there, or our app is not meeting their needs. Or the user experience, or our messaging, our branding……..We don’t know, hence the motivation for our journey.

Context

We took a sample size of approx 40k borrowers during the early part of this year, of which as mentioned 25% were women to try and see if there were material differences in their demographics, application behaviour and credit performance. The first observation was that our population diverged significantly from EFINA’s 2016 survey which showed that ratio of men to women in the formal financial system was about 1.6:1 vs our sample at 3:1.

efina Paylater

 

Furthermore, the age distribution of our sample differed significantly from EFINA’s with the bulk of Paylater customers (>60%) between the ages of 26 and 35.

efina 2 Paylater

Application Behaviour

We recently put restrictions in place on the number of applications in a given period but for this sample period it was a free for all. Paylater adheres to strict affordability criteria and so applicants would change their applications to try and ‘game our system’ . Interestingly, there were 3.8 applications per successful loan application for women compared to 4.1 for men. For us, there doesn’t seem to be any material difference between these two numbers and so we conclude that there was equal ‘hunger’ for credit.

Furthermore, both groups seem to apply during the same time of the day (and yes we did look at variations in the day as well — little change).

 

Infographic Paylater

 

Definitely no bulge in the afternoon indicating house mothers with time after dropping the kids or a spike in the evening when those drunk gamblers needed more cash to fuel their vices. Perhaps an indication of the similarity in professions/lifestyle of the applicants? Or that many biases are just that without any data to provide supporting evidence. More food for thought.

Demographics

For all intents and purposes, the make up of both sets of borrowers were similar in average salary earned. To be clear, these salaries were not verified by us so there may be inherent bias. Our view however is that all liars are the same and we assume that even if there is some ‘creative lift’, the effect will be consistent across both groups. Happy to discuss this assumption offline 🙂

 

Infographic Paylater

 

So, apart from the group of 55 years and older where women appear to have ‘hammered’ more than men (we view this as an anomaly in the data and are investigating it), it seems like there isn’t much divergence. Delving deeper, we also tested for distribution across the salary bands and observed no discernible difference.

 

Infographic Paylater

 

As a final check, we tested for bias in our approval process and found that the ratio of male to female application was the same in the approval rates.

Aha — something of interest!!!

The only area of significance divergence came in the looking at the geographic source of loans. For context, most of our marketing is done using Google Adwords and Facebook, with a focus on age and interests. So the divergence in distribution of loans across different regions is unexpected and subject to current research.

 

Infographic Paylater

 

If anyone has any clues as to this, we would love to talk about it. For now, we are not making any assumptions, just looking for golden nuggets of insight!

Performance

So to the real heart of the beast — performance across the sexes. If we found that women were significantly less risky than men, then the pivot to just lending to women was about to happen. And the results were…..tadaaaaaaa:

Infographic Paylater

So let’s start that we don’t know what to make of it. On the whole, women performed better but perhaps the jury is still out. We are getting some of the eggheads in the company to bring out their t-tables and calculate the significance of this result.

Conclusions and further work

So because it’s Friday and the bars are about to open, we are going to rush to some general conclusions that may be wrong and may not hold the test of time.

  • We believe that there may be some truth in the stereotype of women being more credit averse. Application behaviour between the two groups were similar across salary bands, time and day of application etc with the key difference being the penetration in their respective ‘banked’ populations
  • We do not believe that there is sufficient evidence that there is lower credit risk in lending to women. More work needs to be done on this but on the face of it, both groups showed equal likelihood to pay (or not — that is the question)
  • For the socio-economists, the differences in application trends across the different regions may or may not be supporting evidence of long-held beliefs regarding literacy levels, internet penetration, and economic empowerment, to name but a few.
  • Finally, the big take away for us is that more work needs to be done in understanding the imbalance in loans between the sexes. All the developers at Paylater are male (female developers please get in touch — we need you) and we don’t discount the possibility of inadvertent design bias. We don’t know but, lately, we have become convinced of the need for sustained customer research which is a big hole in our armory. Proponents of Human-Centered-Design (HCD) we love you and want to get in touch.

These are our views and we guarantee it will change in short time. But we want to start the discussion as to how credit can reveal data and impact lives. Please get in touch either with ideas, money or customers.

Till next time.



Leave a Reply