Here’s a deceptively simple question: what are the expected defaults on Ratesetter’s loan book?
It’s simple because there should be a straightforward answer, even if there’s some uncertainty and subjective judgement built into the number. It’s deceptive because there seem to be two answers, one that makes everything look fine and another, which requires some calculation, that’s not so great.
Ratesetter is one of the UK’s top three online lenders, which call themselves P2P, and probably the most complicated of the lot. While the classic P2P lending model involves individuals taking individual risk, Ratesetter collectivises risk. The startup, which is backed by star fund manager Neil Woodford, does this through its provision fund, which exists to protect loan investors against defaults.
As we’ve discussed in detail before, Ratesetter now reserves the right to take interest payments and capital away from investors and put it into the provision fund, in order to make sure there are enough funds to account for loans going bad. The upshot is that investors in Ratesetter need to pay attention to the entire book of loans, rather than their individual investments. (Whisper it softly, it’s sort of like a bank.)
So how’s the provision fund doing? Well, here’s one set of data the company shows to investors on its provision fund page, as it was on Tuesday, with the relevant numbers highlighted:
As you can see, the current and expected money in the fund is larger than the expected losses. Great, even if the reliance on “contractual future income” has increased in recent months.
But there’s another set of data provided by Ratesetter, which tells a slightly different story. It’s on their statistics page, which provides “data on annual performance of loans and returns, updated automatically in real time.” Again, the relevant numbers are highlighted below:
The first thing to note is that the expected bad debt rate for 2014, 2015 and 2016 are all well above 3 per cent. Those three years account for about 90 per cent of Ratesetter’s lending to-date, which raises a question mark over the 2.8 per cent or less bad debt rate figure provided on the provision fund page.
The second thing to note is that it’s possible to calculate the total value of expected bad debt using the numbers on the statistics page. It’s a simple matter of applying the “actual lifetime bad debt rate to date” figures to the total amount lent, and then comparing that to the value implied by the “projected lifetime bad debt rate”. The difference is the total amount of defaults that are yet to happen.
It comes to £24.3m, according to our calculations, which is about £2m higher than the amount of current and expected money in the provision fund. However, according to a Ratesetter spokesperson, that’s not the whole story. They said “around £3m” of those losses are on loans not covered by the provision fund, which means they are held by institutional rather than retail investors.
We’re not the only ones to get tripped up by this nuance. Ratesetter is updating the 2.8 per cent bad debt rate on its provision fund page to 3.1 per cent because it too was including loans not covered by the provision fund in its calculation.
But, even after accounting for that, the figures on the statistics page suggest total expected bad debts of £21.3m, which is a little too close for comfort to the total value of the provision fund. It also doesn’t match with the £18.6m of expected bad debts shown on the provision fund page.
So what’s going on here? Well, it’s complicated in the kind of way that makes you wonder if any of Ratesetter’s investors actually understand what’s going on.
According to Ratesetter, the difference is because they use different methodologies to calculate the expected bad debts on each page. Bear with us here as we try to explain.
The expected bad debts on the provision fund page are estimated by looking at a 12-month cohort of loans that are at least 18 months old. So, at the moment, it reflects the performance of loans originated between June 2014 to May 2015. The company then extrapolate that performance to the entire loan book, taking into account the split between loans that are less than one year old or more than one year old. Older loans are given a lower expected loss figure, on account that they are past the point where default risk is highest.
The projected year-by-year loss numbers on the statistics page are calculated in a different, slightly weirder way. Take 2016, for example. As we’ve still got a month to go, Ratesetter does the same thing it does above in terms of looking at a past 12-month cohort of loans that are at least 18 months old. When the year is finished and no more 2016 loans are being originated, the company then looks at the actual performance of loans originated this year and adjusts its expected losses. If the loans originated in 2016 have lower defaults than in previous years, which Ratesetter believes will be the case, then the expected loss number will come down over time.
However, the key reason for the difference today, according to Ratesetter, is that on the statistics page they don’t take into account lower expected losses on older loans — this, apparently, is why the loss number is higher than the one given on the provision fund page.
The obvious point here is that it’s not particularly helpful for a company to be putting out two sets of numbers about the performance of its loans that are calculated in two different ways. Or, at least, it’s not helpful to do that without a careful explanation and comparison of the methodologies.
Ratesetter has had issues with its data in the past. In March, when we dived into their loanbook, we found that some of the loans had been mislabelled. Two weeks ago, a spokesperson for Ratesetter emailed us to let us know they were updating some of the numbers on the statistics page, specifically the figures for provision fund usage and percentage repaid. Apparently there was a problem with how they had been pulled from their database. Here’s how far they were off:
If companies are going to sell loans to retail investors, there has to be some confidence that the retail investors are able to grasp the risks involved. Regardless of the intelligence of investors, it’s impossible to do that without clear, straightforward and consistent information.
Otherwise, you’re just left trusting the platform when they say everything is fine, which, for fear of repeating ourselves, isn’t very “peer-to-peer”.
Wake me up when online lenders are done turning into banks — FT Alphaville
P2P lending rediscovers balance sheet magic — FT Alphaville
Re-setting Ratesetter’s provision fund — FT Alphaville
Re-setting Ratesetter’s default ratings — FT Alphaville
Ratesetter’s provision fund “is going to be pushed to the limit” – FT Alphaville
Peer-to-peer maturity transformation – FT Alphaville
The curious state of UK “peer-to-peer” lending – FT Alphaville
It’s lenders all the way down – FT Alphaville
Lifting the bonnet on an online lending business – FT Alphaville
Copyright The Financial Times Limited 2016. All rights reserved. You may share using our article tools. Please don’t cut articles from FT.com and redistribute by email or post to the web.