|
Post by westonkevRS on May 6, 2016 19:57:16 GMT
Previously in this thread RS stated that the expected default level was based on some 8,000,000 contracts or records. I simply wonder how this wealth of information seems to have been given second place to the new math calculation and result in a reduction of the expected defaults. I would expect that with so much historical data not much would change! IIRC the previous methodology purely looked at the actual rate at which the loans made 6-18 months previously had defaulted and assumed that this performance would be consistent with similar loans throughout the loanbook. If loans have a higher default rate in the opening period, this would overstate the likely defaults for the older loans and so an adjustment such as that made would be appropriate.
If I am correct on the methodology, RS must have significantly tightened their credit procedures as the cohort concerned appear to have a higher default rate than that now assumed for the portfolio.
- PM
The cohort used to determine the probability of default is a "12:18", so it's a full years of applications starting with ones at least 18 months old. So roughly this month that is the November 2013 to October 2014 cohort. More recent and we don't have time for performance to evolve, older and it doesn't look like recent underwriting. 12:18 is the standard within Basel II and professional risk model development. The 2014 cohort was probably RateSetter's worst performing cohort, hence I expect the probability of default estimates to remain pessimistic until we start to consider 2015 loans in about 5 months time, which have performed better (for a number of reasons, which I won't go into). Kevin.
|
|
|
Post by propman on May 9, 2016 8:31:26 GMT
IIRC the previous methodology purely looked at the actual rate at which the loans made 6-18 months previously had defaulted and assumed that this performance would be consistent with similar loans throughout the loanbook. If loans have a higher default rate in the opening period, this would overstate the likely defaults for the older loans and so an adjustment such as that made would be appropriate.
If I am correct on the methodology, RS must have significantly tightened their credit procedures as the cohort concerned appear to have a higher default rate than that now assumed for the portfolio.
- PM
The cohort used to determine the probability of default is a "12:18", so it's a full years of applications starting with ones at least 18 months old. So roughly this month that is the November 2013 to October 2014 cohort. More recent and we don't have time for performance to evolve, older and it doesn't look like recent underwriting. 12:18 is the standard within Basel II and professional risk model development. The 2014 cohort was probably RateSetter's worst performing cohort, hence I expect the probability of default estimates to remain pessimistic until we start to consider 2015 loans in about 5 months time, which have performed better (for a number of reasons, which I won't go into). Kevin. Kevin, thanks for putting me right.
Actually it looks to me like the worst performing cohort is May 2014 to July 2015 that is the worst performing cohort, although I agree that after February 2015 there is a slight improvement. So I take it that the above methodology means that you are still including the defaults from 6 months before the defaults increased appreciably. The largest 12 month cohort averages 2.97% (April 2014 to March 2015) on a simple basis or 2.92% weighted, despite only 70% repayment of these loans (ie due to the immaturity of the later loans, more defaults are expected).
I appreciate that you are limited in what you can tell us, but would you indicate whether the expected defaults are reweighted to the categorisations of the loans concerned? This seems likely as the cohort actually used appears to have a weighted average default of 2.8% rather than the 2.3% currently in use.
Many thanks
- PM
|
|
|
Post by westonkevRS on May 9, 2016 17:43:32 GMT
propmanYour analysis looks broadly right, although loan performance significantly improved from Feb 2015, not July 2015. You are probably including some institutional lending that went through a different (higher risk) underwriting process that was not protected by the Provision Fund. Due to it's unique underwriting process that we no longer provide, these loans are not included in the expected loss calculations. Yes, we "amortise" the bad debt forward to get to a final Probability of Default that is larger than the actual achieved to date. So yes we include future defaults. So a 70% amortized cohort is increased accordingly, this used to be "straight lined" but now we use a curve as per the blog explanation. Kevin.
|
|
jlend
Member of DD Central
Posts: 1,817
Likes: 1,444
|
Post by jlend on Jun 13, 2016 6:17:23 GMT
Well talking about the MR has been fun for the last few days... What's next? Ahhh the coverage rate just dropped back down to 1.3 Whoopi. We like to keep things interesting. Besides when it went up a few points we had complaints of fudging the numbers, now it's gone down a few points we'll get complaints of potential defaults being to high. All good fun. Just for the record though, we update the variables monthly. So there can be a slight movement each month. I actually expect the coverage rate to drop a little the next 4 months, and then grow through to the end of 2016. But that's an estimate, only time will tell... Kevin. Why do you think it will fall from 1.3 ?
|
|
|
Post by propman on Jun 13, 2016 12:17:43 GMT
As I understand it, rather than a subjective assessment of bad debts based on all loan history known, the estimate (the denominator of the ratio) is based upon the actual performance of a particular pre-selected 12 month cohort of loans. This is then applied to the remaining balance of loans on a category by category basis and reduced for an expected reduction in bad debts of older loans.
If the month after the selected cohort has a worse performance than the earliest month included, when the cohort is advanced a month, the bad debt estimate will increase. This will reduce the Coverage Ratio.
|
|
|
Post by westonkevRS on Jun 13, 2016 18:03:16 GMT
As I understand it, rather than a subjective assessment of bad debts based on all loan history known, the estimate (the denominator of the ratio) is based upon the actual performance of a particular pre-selected 12 month cohort of loans. This is then applied to the remaining balance of loans on a category by category basis and reduced for an expected reduction in bad debts of older loans.
If the month after the selected cohort has a worse performance than the earliest month included, when the cohort is advanced a month, the bad debt estimate will increase. This will reduce the Coverage Ratio. This is true, for the retail portfolio. But a few caveats. 1) Commercial lending doesn't have enough RateSetter own data and the real estate portfolio has never had a default. So the expected losses for these are based on credit rating expectation of default and losses based on security. 2) As RateSetter transitions to charging over the lifetime of loans, this has a negative impact on the coverage ratio. This means more and more of the loans we are writing pay nothing or a smaller than usual payment up front to the Provision Fund. This causes the Provision Fund coverage to drop simply by booking these loans, which are making up a greater proportion of the business we write. Now could can argue about the prudence of this approach vs. the negative coverage ratio connotations. In the same vein that we don't I Clyde future payment revenues from loans on payment plans. But all this is for another day... That said, credit performance hasn't been as good as I'd have liked. For example there is a new culture of borrower being encouraged to take a loan and then instantly go on a debt management plan (with fees to a provider). We've also had some lumpy commercial loans going through new arrangements. But this is all spmething constantly monitored and pricing on new loans adjusted for, it's a key part of the job and you have to expect the ebbs and flows. It all just makes using the coverage ratio not an ideal single metric for lenders to use. Kevin.
|
|
jlend
Member of DD Central
Posts: 1,817
Likes: 1,444
|
Post by jlend on Jun 14, 2016 7:01:38 GMT
As I understand it, rather than a subjective assessment of bad debts based on all loan history known, the estimate (the denominator of the ratio) is based upon the actual performance of a particular pre-selected 12 month cohort of loans. This is then applied to the remaining balance of loans on a category by category basis and reduced for an expected reduction in bad debts of older loans.
If the month after the selected cohort has a worse performance than the earliest month included, when the cohort is advanced a month, the bad debt estimate will increase. This will reduce the Coverage Ratio. This is true, for the retail portfolio. But a few caveats. 1) Commercial lending doesn't have enough RateSetter own data and the real estate portfolio has never had a default. So the expected losses for these are based on credit rating expectation of default and losses based on security. 2) As RateSetter transitions to charging over the lifetime of loans, this has a negative impact on the coverage ratio. This means more and more of the loans we are writing pay nothing or a smaller than usual payment up front to the Provision Fund. This causes the Provision Fund coverage to drop simply by booking these loans, which are making up a greater proportion of the business we write. Now could can argue about the prudence of this approach vs. the negative coverage ratio connotations. In the same vein that we don't I Clyde future payment revenues from loans on payment plans. But all this is for another day... That said, credit performance hasn't been as good as I'd have liked. For example there is a new culture of borrower being encouraged to take a loan and then instantly go on a debt management plan (with fees to a provider). We've also had some lumpy commercial loans going through new arrangements. But this is all spmething constantly monitored and pricing on new loans adjusted for, it's a key part of the job and you have to expect the ebbs and flows. It all just makes using the coverage ratio not an ideal single metric for lenders to use. Kevin. Thanks for the detailed response. Very useful background. It can be difficult to follow what the various numeric data tells me over time. There is a lot of data and metrics for lenders to try and follow and make sense of when deciding how much to lend on ratesetter. I would find it really useful to have a commentary just once a month to go with the various numeric data already provided. Something along the lines of your post. Perhaps something that could be seen when you log in and on the monthly email to lenders.
|
|
|
Post by propman on Jun 14, 2016 7:45:56 GMT
Isn't taking on a loan with the intention of going onto a Debt management plan fraud? I appreciate that it may be difficult to prove, but if someone is "encouraging them", then this sounds like something that should be investigated, particularly if widespread. With the much broadcast increases in collaboration between providers to detect and prevent fraud, shouldn't something be done to collect the evidence for prosecution?
|
|
adrianc
Member of DD Central
Posts: 9,045
Likes: 4,841
|
Post by adrianc on Jun 14, 2016 8:27:46 GMT
Isn't taking on a loan with the intention of going onto a Debt management plan fraud? Never mind that, if somebody's that close to needing a DMP, why on earth is anybody lending them more money?
|
|
toffeeboy
Member of DD Central
Posts: 507
Likes: 363
|
Post by toffeeboy on Jun 14, 2016 11:52:02 GMT
Isn't taking on a loan with the intention of going onto a Debt management plan fraud? Never mind that, if somebody's that close to needing a DMP, why on earth is anybody lending them more money? Unfortunately there are unsavoury people out there that will take large loans just prior to their credit rating going belly up as they know there is no way for the lender to get the money back. Another one is if someone is planning on leaving the country so ruining their credit rating doesn't bother them and it is very hard to track someone once they have left the country.
Going bankrupt doesn't seem to bother people nowadays, when I was growing up you wanted a good credit rating to get a mortgage but now so many people have no chance of getting on the property ladder then they don't see the need to worry about such trivial things as a good credit score.
I am sure I read that Chappell who helped rip BHS apart has been bankrupt 3 times but is still a millionaire???
|
|
adrianc
Member of DD Central
Posts: 9,045
Likes: 4,841
|
Post by adrianc on Jun 14, 2016 11:55:03 GMT
I am sure I read that Chappell who helped rip BHS apart has been bankrupt 3 times but is still a millionaire??? Three BRs plus one IVA. www.bbc.co.uk/news/business-36241660Although "is still" is not quite the same thing as "has rebuilt his finances to again be"
|
|
|
Post by westonkevRS on Jun 14, 2016 19:06:02 GMT
I would find it really useful to have a commentary just once a month to go with the various numeric data already provided. Something along the lines of your post. Perhaps something that could be seen when you log in and on the monthly email to lenders. You are not alone in expressing a wish for a form of commentary and this has been discussed. The prime issue isn't that this will provide another overhead (with an eventual message that would probably be compliant and PRd to death to the point of meaningless drivel stating the obvious), but one of providing advice. There is a danger that providing some words could be interpreted to mean safety or advice that losses are unlikely, " nothing to worry about". And then if something did go wrong, remembering that this is P2P lending with all the risks that entails with it's lack of FSCS protection, these " words" could be used against the platform to gain compensation somehow. It simply isn't worth the risk to try and provide some reassurance or a neutral update with words, when if we just provide factual numbers based on audit-able methodologies people can interpret this as they wish. Kevin.
|
|
|
Post by westonkevRS on Jun 14, 2016 19:15:39 GMT
Isn't taking on a loan with the intention of going onto a Debt management plan fraud? Never mind that, if somebody's that close to needing a DMP, why on earth is anybody lending them more money? Because it only stays on your credit file for 6 years. Some people think that credit purgatory is worth the wait for some " free" money. Despite the unsavoury nature, lack of moral compass and understated hassle it entails (e.g. doorstop collections visits, no mortgage, no mobile phone contracts, etc). Some are probably naively led by the ambulance chasers. We call these borrowers " social suicide". And as mentioned by toffeeboy there are those that plan to leave the country, we called these " flight risk". Basically things are a lot harder than 20 years ago when people tended to have the moral compass. The issue is that higher defaults due to this behaviour increases the cost of credit to everyone, and when credit is expensive then the economy cannot grow due to reduced or expensive capital for investment and research. This drags the economy down. I've noticed from working in many countries across the EMEA region that the culture to paying back credit varies hugely, and those countries with the most " relaxed" attitude to paying back debt tend to have the weakest economies. One day I'll write a blog about this, but you've got to be careful to not overly stereotype or be accused of xenophobia, I prefer to stick to the facts of default rates.... Kevin. Kevin.
|
|
adrianc
Member of DD Central
Posts: 9,045
Likes: 4,841
|
Post by adrianc on Jun 14, 2016 19:18:50 GMT
Never mind that, if somebody's that close to needing a DMP, why on earth is anybody lending them more money? Because it only stays on your credit file for 6 years. Some people think that credit purgatory is worth the wait for some " free" money. I think you're answering that from the other way round. You've answered "Why borrow more?" My question was "Why lend money to them?" For them to be at that stage, their credit record must be positively toxic.
|
|
|
Post by westonkevRS on Jun 14, 2016 19:24:35 GMT
Depends where you are in the order, RateSetter won't lend to anyone that is in difficulty or appears to be increasing (or getting ready) to move into a debt spiral. I don't want to give the impression this is a big issue for RateSetter as we are not a high risk lender. But it is just another one of the challenges faced by lenders in today's competitive and changing market.
But the sub-prime lenders charging 100% APR plus will still lend, on the basis that half of them just made a mistake.... Doh!
Kevin.
|
|