Zopa+ loanbook performance, a statistical analysis

However, if you analyse the portfolio by APR paid by the borrower, there has been a clear shift.

The Loanbook has a field called 'Lending Rate' which I've assumed to be the APR paid by the borrower. Or at least I hope it is, because at the maximum of 31% on 5 year loans would get very scary if fees was added on top....

The following chart shows the minimum, average and maximum lending rate on Zopa 5-year loans (all lender products):



In April 2017 the minimum rate was 3.1% APR (and this tallies with what you see on MoneySuperMarket), the average was 10.5% APR, but the highest has dramatically increased to 31% APR.  The maximum (and obviously therefore the average) are dramatically up on early 2016

So depending on how you classify "sub-prime", many bankers would argue that Zopa have moved into sub-prime territory for some borrowers. Personally I do classify 25%+ APR on a 5-year loan s sub-prime (it's different for different terms, just for the argument).

All of the rates have crept up in the last couple of months slightly, this could be a blip or as a reaction to actual loan performance (as hinted by lenders).

Kevin.

Looking at the high level bad debt (I'll break this down later), there is a clear trend towards higher bad debt that you'd associated with the higher risk grades and APRs.

I've used the fields available in the loanbook, and created the following fields;

> bad debt balances that are late and default, divided by original loan amount. I've called this "bad rate", although I appreciate some of the lates will cure (but again, some goods will go bad).
> forecast bad debt, which is a simple ratio of the current "bad rate" factored by the balance paid down. This is prudent, or harsh, because it assumes bad debt will constant through the life of loans (which it isn't)
> Defaults on Collected £, which is effectively of the money paid back; what amount didn't come back and got defaulted.


On an annualised based:




2017 data is too immature. But 2016 indicates a final bad debt rate somewhere between 4.5% and 10% depending on which methodology. I'll hazard a guess around 6-7% of loans will default, but only time will tell.

If you just look at 2015 to 2017;




It's clear the escalation started in 2015. Which coincides with the overall increase in volumes of loans matched, i.e. I'll hazard a guess this was done by going down the risk curve (rather than new channels of borrowers). Which is fine, as long as the loans are priced correctly, which considering the expertise they have I think they probably are.

However the one lesson that is clear is that the risk appetite has increased, and as a result we can all expect much higher defaults that historically. And for those in Zopa+, this might be uncomfortable. And we are reliant on Zopa getting the pricing and specialised underwriting right on this new segment, which despite their 12 year history is an area of lending they've not done before.

Kevin.



P.S. Here is the raw data in the charts.

When I look at actual returns, my numbers make the performance look better than I expected through the credit crisis, but actually worse more recently (never negative, it should be noted, i.e. 100% of capital has always been returned);




This isn't segmented by product or my estimate of risk, but simply compares the Loan Amount to the Principal Collected+Interest+Outstanding Balances (excluding late and default). Although it's unclear from the data how this return would be reduced by Zopa fees (i.e. are the numbers in the LoanBook gross or net of their fees)

It isn't annualised, just shows the returns in percentage (the green line) from the year the money was lent to now. So for example, overall cash lent in 2010 has returned 11.6% to now (I'd guess over a mix of terms from 12 - 60 months, so cash was probably out for an average of around 36 months - so 11.6% is a 3 year return for the 2010 vintage). 

More recent vintages are clearly returning less. Although it's difficult to say how 2016 or 2017 will eventually return, but it's fair to say the returns are lower than those halcyon days, or that's what the loan implies. Only loans to 2011 are fully amortised (i.e. finished, and performed confirmed)

Kevin.

P.S. Here's the data:

That sounds about right. Although complicated because you would have been getting your capital back, and so depends how you reinvested.

Really difficult to understand these stats. My Safeguard seems to be about 4.8%. Pre-safeguard I could figure out if I really tried, but didn't seem too bad at the time. Plus is still too early to tell (5.8% maybe currently after a couple of bad months and ignoring the early months before defaults were possible). Fees were initially on lenders and then moved to borrowers, and there were early adopters reduced fees/now bonuses, and a 'rate promise' at one point. But overall less than 4% APR for the 2010 tranche seems rather low to me from my records. 

Thanks for the very detailed analysis Kevin.  Of course, people will react emotionally when they see that their defaults are more than their interest in any given month and I suppose this is what is defined as 'noise' in this thread.  The bottom line is that anyone investing in Z+ is told very clearly that they lose all their money (almost certainly, anyway) and have no comeback if there's a default on a particular loan.  I think it might be time for a bit of extra warning from Zopa along the lines of Seedrs, in which I have some cash.  There, the warnings are very stark and actually say that most startups fail and that you should only have a certain % of your portfolio with them.  Maybe a recommendation that no more than 20% should be in Z+ and the rest in Classic and/or Access?  I just wonder how many investors have put everything into Z+ and are about to be hit hard if/when there's a downturn.  The other 'noise' concerns Z+ loans which have never paid back a penny and the number of loans for cars - probably two other increasing trends by the look of things. 

Long term Zopa lenders were used to taking all the risk in the pre-safeguard days (and through the downturn in 2008/2009), and many didn't like the safeguarded model. Zopa plus is just a return to the norm for us. But I agree new lenders who are used to some protection on their loans may not understand the risks completely. I'm also concerned that many of them don't understand the risks of P2P in general in a downturn, even in platforms with supposedly protected funds.

Edit: It's a pity the old Zopa forum has gone, it would be interesting to look back at the threads on defaults and losses from pre-safeguard days, very similar concerns I seem to remember.