r00lish67
Member of DD Central
Posts: 2,692 
Likes: 4,048
|
Post by r00lish67 on Jul 10, 2018 10:20:52 GMT
Hi Matthew , I've just been having a long overdue pore through the latest LW loanbook, and I'm struggling to reconcile the high-level statistics on the risk/return stats page with what's in the loanbook. Wondering if you can steer me as to where I may be going wrong. I'm focusing on the 'actual arrears rate' and 'actual lifetime bad debt rate' figures from the 3rd table down on the page. Taking just the 2017 cohort and using the latest loanbook extract, by my calculation the loans issued (excluding cancelled loans) in that year total £43,392,540. Of that, £234,148 (0.54%) is classified as defaulted and £1,863,698 (4.29%) is classified as late, making for a total of 4.84% of the loans issued as what I'll call 'troubled loans' According to your high level statistics table, the 2017 actual bad debt (defaults less recoveries) is 1.6% and the actual arrears rate is 1.3%, making for a total of 2.9% troubled loans i.e. your defaults stats appear lower than in the loanbook, but the lates way way higher. Accepting that this isn't a perfect analysis, given that there is no 'amount recovered' field in the loanbook (can this be added?), and that I'm taking all late loans rather than just those over 45 days late - there still seems to be quite a big discrepancy here as there seems to be 60-70% more troubled loans by value than the high level statistics suggest. I've had similar issues reconciling 2015-16 also. Can you clarify how the 2017 ABDR and Arrears figures on your page have been calculated? Thanks Edit: If any other statistically minded LW investors fancy helping out, by all means do! |
|
r00lish67
Member of DD Central
Posts: 2,692
Likes: 4,048
|
Post by r00lish67 on Jul 10, 2018 11:38:51 GMT
I'd also like to pick up one more statistic that concerns me. For the 2015 loan cohort only: The 'Estimated lifetime bad debt rate (at origination)' in your stats table is 2.8%. The 'Actual lifetime bad debt rate (to date)' is also 2.8%. The 'Forecast lifetime bad debt rate (projection)' is 2.6%. Looking at the loanbook, the total amount lent for 2015 excluding cancelled loans was £14,052,198. The principal outstanding in defaulted loans is £561,919 (4.0%) . The principal outstanding in late loans is £306,071 (2.18%). So: 1) Why is your ALBDR stat for this cohort stated as 2.8%, but the loanbook suggests that a full 4.0% of the 2015 loan principal is outstanding in default? 2) Your forecast LBDR being lower than your actual LBDR suggests to me that you expect net recoveries to exceed net defaults for this cohort going forwards. With a further 2.18% of late loans to deal with, some of which will presumably default, is this realistic? This is quite concerning to me. My only guess as to what might be incorrect here is if recoveries for some reason aren't being counted within column N ('principal outstanding') of the loanbook - but if they aren't, then it would make assessing loan performance impossible from the outside looking in, and we really need that adding in. Again, hoping you can clarify  |
|
|
|
Post by Matthew on Jul 10, 2018 12:25:40 GMT
Hi Matthew , I've just been having a long overdue pore through the latest LW loanbook, and I'm struggling to reconcile the high-level statistics on the risk/return stats page with what's in the loanbook. Wondering if you can steer me as to where I may be going wrong. I'm focusing on the 'actual arrears rate' and 'actual lifetime bad debt rate' figures from the 3rd table down on the page. Taking just the 2017 cohort and using the latest loanbook extract, by my calculation the loans issued (excluding cancelled loans) in that year total £43,392,540. Of that, £234,148 (0.54%) is classified as defaulted and £1,863,698 (4.29%) is classified as late, making for a total of 4.84% of the loans issued as what I'll call 'troubled loans' According to your high level statistics table, the 2017 actual bad debt (defaults less recoveries) is 1.6% and the actual arrears rate is 1.3%, making for a total of 2.9% troubled loans i.e. your defaults stats appear lower than in the loanbook, but the lates way way higher. Accepting that this isn't a perfect analysis, given that there is no 'amount recovered' field in the loanbook (can this be added?), and that I'm taking all late loans rather than just those over 45 days late - there still seems to be quite a big discrepancy here as there seems to be 60-70% more troubled loans by value than the high level statistics suggest. I've had similar issues reconciling 2015-16 also. Can you clarify how the 2017 ABDR and Arrears figures on your page have been calculated? Thanks Edit: If any other statistically minded LW investors fancy helping out, by all means do! Hi r00lish67 Pleased to see someone's downloading the loanbook! I believe the primary reasons for the variance between your figures above are as follows: 1. The loanbook includes loans funded by institutional investors, who do not have the benefit of the Shield (refer to "Shield coverage" column in the downloadable loanbook to filter), whereas the summary stats provided on the website relate only to loans funded by retail investors. We exclude institutional data from the summary stats primarily because it would give a false representation of the loss rates affecting anyone looking at that page i.e. retail investors. Institutional investors tend to take slightly more risk than we allocate to our retail investors, hence overall default rate is not representative of the risk for retail investors. 2. The loanbook includes a loan as "Late" when any amount is overdue, whereas the website stats report arrears in accordance with P2PFA definitions. What this means is that loans with Direct Debit issues etc are treated as arrears when in reality the majority of these bring their accounts up to date quite quickly. Hope this helps.
|
|
|
|
Post by Matthew on Jul 10, 2018 12:41:41 GMT
I'd also like to pick up one more statistic that concerns me. For the 2015 loan cohort only: The 'Estimated lifetime bad debt rate (at origination)' in your stats table is 2.8%. The 'Actual lifetime bad debt rate (to date)' is also 2.8%. The 'Forecast lifetime bad debt rate (projection)' is 2.6%. Looking at the loanbook, the total amount lent for 2015 excluding cancelled loans was £14,052,198. The principal outstanding in defaulted loans is £561,919 (4.0%) . The principal outstanding in late loans is £306,071 (2.18%). So: 1) Why is your ALBDR stat for this cohort stated as 2.8%, but the loanbook suggests that a full 4.0% of the 2015 loan principal is outstanding in default? 2) Your forecast LBDR being lower than your actual LBDR suggests to me that you expect net recoveries to exceed net defaults for this cohort going forwards. With a further 2.18% of late loans to deal with, some of which will presumably default, is this realistic? This is quite concerning to me. My only guess as to what might be incorrect here is if recoveries for some reason aren't being counted within column N ('principal outstanding') of the loanbook - but if they aren't, then it would make assessing loan performance impossible from the outside looking in, and we really need that adding in. Again, hoping you can clarify This answer is similar to that above. The total amount lent for 2015, excluding cancelled loans and excluding loans funded by institutional investors, is £11.3m. Of this, £0.2m has been "Defaulted" in our system (1.8%) and a further £0.2m is currently "Late" (1.7%). Of the late loans, some have already been reflected in the ALBDR due the P2PFA definition of default being met. Actual recoveries received are deducted from column N and clearly future expected recoveries are not. However, on the defaulted (and late) book, many loans are on long-term repayment arrangements, either directly with us or through third party debt management companies, and therefore we're receiving a significant amount of recoveries on these loans albeit over a longer than expected timeframe. We are satisfied that our assumptions around future recoveries etc are reflective of actual recovery rates as this is something we monitor on a monthly basis. Apologies that this is a bit confusing - we are trying to share as much data as possible while also meeting the P2PFA disclosure requirements. Hope this helps anyway.
|
|
r00lish67
Member of DD Central
Posts: 2,692
Likes: 4,048
|
Post by r00lish67 on Jul 10, 2018 12:56:27 GMT
Thanks Matthew . So, I've now done the same analysis taking into account including only loans protected by the shield. The % figures change very little for the 2017 cohort. I take your point re: late loans, however that doesn't explain why your high level stats report significantly higher defaults than the actual data. Also, (Yes it does! ) whilst you're right that the loanbook doesn't distinguish >45d late loans, that seems unlikely to me to fully explain the difference between 4.81% late in the loanbook and 1.3% late in the stats. If true, that would suggest some 3.51% (£1.5m) of loans are <45d late - is that true? What would really help is if we could see the calcs as to how your high level stats reach figures of 1.6% for defaults and 1.3% for actual arrears in 2017. Can you share those with us? Analysis repeated with updated figures: "Taking just the 2017 cohort and using the latest loanbook extract, by my calculation the loans issued (excluding cancelled loans) in that year total £42,973,053. Of that, £234,148 (0.54%) is classified as defaulted and £1,834,368 (4.27%) is classified as late, making for a total of 4.81% of the loans issued as what I'll call 'troubled loans' According to your high level statistics table, the 2017 actual bad debt (defaults less recoveries) is 1.6% and the actual arrears rate is 1.3%, making for a total of 2.9% troubled loans i.e. your defaults stats appear lower than in the loanbook, but the lates way way higher" Edit: Struck through the first point above, as I now realise this is because a chunk of the late loans are regarded as defaulted due to the P2PFA standards. |
|
r00lish67
Member of DD Central
Posts: 2,692
Likes: 4,048
|
Post by r00lish67 on Jul 10, 2018 13:12:24 GMT
This answer is similar to that above. The total amount lent for 2015, excluding cancelled loans and excluding loans funded by institutional investors, is £11.3m. Of this, £0.2m has been "Defaulted" in our system (1.8%) and a further £0.2m is currently "Late" (1.7%). Of the late loans, some have already been reflected in the ALBDR due the P2PFA definition of default being met. Actual recoveries received are deducted from column N and clearly future expected recoveries are not. However, on the defaulted (and late) book, many loans are on long-term repayment arrangements, either directly with us or through third party debt management companies, and therefore we're receiving a significant amount of recoveries on these loans albeit over a longer than expected timeframe. We are satisfied that our assumptions around future recoveries etc are reflective of actual recovery rates as this is something we monitor on a monthly basis. Apologies that this is a bit confusing - we are trying to share as much data as possible while also meeting the P2PFA disclosure requirements. Hope this helps anyway. Not at all, that makes absolute sense, and I'm largely reassured by it as the 2015 is much more affected than 2017 by switching off the shieldcoverage flag. Glad I'm not an institutional investor! I still am a bit perplexed by the 2017 situation though.
|
|
|
|
Post by Matthew on Jul 10, 2018 15:04:23 GMT
Thanks Matthew . So, I've now done the same analysis taking into account including only loans protected by the shield. The % figures change very little for the 2017 cohort. I take your point re: late loans, however that doesn't explain why your high level stats report significantly higher defaults than the actual data. Also, (Yes it does! ) whilst you're right that the loanbook doesn't distinguish >45d late loans, that seems unlikely to me to fully explain the difference between 4.81% late in the loanbook and 1.3% late in the stats. If true, that would suggest some 3.51% (£1.5m) of loans are <45d late - is that true? What would really help is if we could see the calcs as to how your high level stats reach figures of 1.6% for defaults and 1.3% for actual arrears in 2017. Can you share those with us? Analysis repeated with updated figures: "Taking just the 2017 cohort and using the latest loanbook extract, by my calculation the loans issued (excluding cancelled loans) in that year total £42,973,053. Of that, £234,148 (0.54%) is classified as defaulted and £1,834,368 (4.27%) is classified as late, making for a total of 4.81% of the loans issued as what I'll call 'troubled loans' According to your high level statistics table, the 2017 actual bad debt (defaults less recoveries) is 1.6% and the actual arrears rate is 1.3%, making for a total of 2.9% troubled loans i.e. your defaults stats appear lower than in the loanbook, but the lates way way higher" Edit: Struck through the first point above, as I now realise this is because a chunk of the late loans are regarded as defaulted due to the P2PFA standards. Hi r00lish67In order to recalculate the P2PFA-defined default figures, you'd need to see the number of days late for each of the loans in arrears, which is not something we currently show in the loanbook. Regarding the 2017 stats, of the £1.8m currently showing as "Late" in the loanbook, approximately £1.4m is <2 payments down and a further £0.25m has already been included in the default stats (due to P2PFA definition of 120 dpd). The recovery rates on 1 and 2 downs are very high and a fairly small proportion of these would go on to default. Note also that all of this, including expected losses arising from these arrears, is reflected in the forecast loss rate of 4.8% for the 2017 cohort. Thanks
|
|
benaj
Member of DD Central
N/A
Posts: 5,591
Likes: 1,735
|
Post by benaj on Nov 15, 2018 12:29:53 GMT
My experience after 6 months (2k):-
Total loans acquired: 86
Earning: 5.6% (including sign up bonus)
Loans active: 83
Loans settled early / cancelled: 3
Loans in arrears: 0
Actual defaults: 0
It seems really good so far for a platform lending from 7.46% to 26.11%. But I do wonder other lenders experienced something similar with a higher number of loans.
|
|
|
|
Post by carol167 on Nov 15, 2018 13:38:02 GMT
My experience since July 2014 :-
Classic Account : Settled 82
Cancelled 2 Active 202 Overall % currently = 5.77
ISA Account : Settled 32 Cancelled 2 Active 408 Overall % currently 5.44
All 5 year only.
Well happy so far.
|
|
r00lish67
Member of DD Central
Posts: 2,692
Likes: 4,048
|
Post by r00lish67 on Nov 15, 2018 16:15:55 GMT
I don't know if there's too much value in us looking at our own piece of the pie, when we have the complete picture at hand in their full loan book and know that none of us (to date) will have lost anything and that our rate will be an average of whatever we've signed up to. As with Ratesetter/Growth Street, we're all going to do exactly as per the tin with LendingWorks unless and until very apparent bad stuff happens. I do invest with LW, but I've always been unsettled by the graphs here: The 2014 cohort is fine, and probably so small now as to be inconsequential. The 2015 cohort has made quite a dart above target defaults in recent months, and seems to show a very implausible forecast immediate recovery in a month or so's time. The 2016 cohort actual performance graph shows actual defaults soaring above the expected rates, at what looks like 3.8% actual versus approx 2.1% expected - a huge differential. The 2017 cohort is already well above target, and the gap is already forecast to be widened further. Actually yeah, why am I still investing here? Matthew usually comes along and says reassuring things  On the plus side, the Bad Debt Coverage Ratio despite having dropped in recent months is still not too bad and higher than when I first tracked it a year or so ago. |
|
IFISAcava
Member of DD Central
Posts: 3,692
Likes: 3,018
|
Post by IFISAcava on Nov 15, 2018 18:26:55 GMT
I don't know if there's too much value in us looking at our own piece of the pie, when we have the complete picture at hand in their full loan book and know that none of us (to date) will have lost anything and that our rate will be an average of whatever we've signed up to. As with Ratesetter/Growth Street, we're all going to do exactly as per the tin with LendingWorks unless and until very apparent bad stuff happens. I do invest with LW, but I've always been unsettled by the graphs here: The 2014 cohort is fine, and probably so small now as to be inconsequential. The 2015 cohort has made quite a dart above target defaults in recent months, and seems to show a very implausible forecast immediate recovery in a month or so's time. The 2016 cohort actual performance graph shows actual defaults soaring above the expected rates, at what looks like 3.8% actual versus approx 2.1% expected - a huge differential. The 2017 cohort is already well above target, and the gap is already forecast to be widened further. Actually yeah, why am I still investing here? Matthew usually comes along and says reassuring things On the plus side, the Bad Debt Coverage Ratio despite having dropped in recent months is still not too bad and higher than when I first tracked it a year or so ago. But also APR lent at is up (7.7% in 2014 rising to 11.8% in 2018) so presumably higher return cancels out higher default rate? Or more than cancels it out hence increased rates currently?
|
|
puddleduck
Member of DD Central
Posts: 537
Likes: 489
|
Post by puddleduck on Nov 15, 2018 18:45:54 GMT
I don't know if there's too much value in us looking at our own piece of the pie, when we have the complete picture at hand in their full loan book and know that none of us (to date) will have lost anything and that our rate will be an average of whatever we've signed up to. As with Ratesetter/Growth Street, we're all going to do exactly as per the tin with LendingWorks unless and until very apparent bad stuff happens. I do invest with LW, but I've always been unsettled by the graphs here: The 2014 cohort is fine, and probably so small now as to be inconsequential. The 2015 cohort has made quite a dart above target defaults in recent months, and seems to show a very implausible forecast immediate recovery in a month or so's time. The 2016 cohort actual performance graph shows actual defaults soaring above the expected rates, at what looks like 3.8% actual versus approx 2.1% expected - a huge differential. The 2017 cohort is already well above target, and the gap is already forecast to be widened further. Actually yeah, why am I still investing here? Matthew usually comes along and says reassuring things On the plus side, the Bad Debt Coverage Ratio despite having dropped in recent months is still not too bad and higher than when I first tracked it a year or so ago. But also APR lent at is up (7.7% in 2014 rising to 11.8% in 2018) so presumably higher return cancels out higher default rate? Or more than cancels it out hence increased rates currently? I would interpret that completely differently - I would suggest rising rates are an indicator of lower quality borrowers, and thus a higher risk of defaults. You could interpret it either way, glass half full, or half empty. I suspect my view is more realistic.
|
|
r00lish67
Member of DD Central
Posts: 2,692
Likes: 4,048
|
Post by r00lish67 on Nov 15, 2018 18:57:41 GMT
But also APR lent at is up (7.7% in 2014 rising to 11.8% in 2018) so presumably higher return cancels out higher default rate? Or more than cancels it out hence increased rates currently? I would interpret that completely differently - I would suggest rising rates are an indicator of lower quality borrowers, and thus a higher risk of defaults. You could interpret it either way, glass half full, or half empty. I suspect my view is more realistic. Certainly we've seen an awful lot of pain with Z+ borrowers, high-rate asset backed borrowers, and possibly also D/E borrowers on FC (although less sure about the net return on those). I don't have a full range of stats, would be interested if anyone does - my instinct certainly says that high-rate borrowers tend not to be necessarily good value in the end for P2P lenders..
|
|
bigfoot12
Member of DD Central
Posts: 1,817
Likes: 816
|
Post by bigfoot12 on Dec 10, 2018 8:31:13 GMT
Matthew, thank you for your reply (on another thread), I was hoping for a clarification on the difference between Actual arrears rate and Actual lifetime bad debt rate (to date). If you could refer to the 2016 cohort I would appreciate it. (Or if anyone else can point me to an explanation that would be good.)
|
|
bigfoot12
Member of DD Central
Posts: 1,817
Likes: 816
|
Post by bigfoot12 on Dec 12, 2018 12:21:49 GMT
Matthew , thank you for your reply (on another thread), I was hoping for a clarification on the difference between Actual arrears rate and Actual lifetime bad debt rate (to date). If you could refer to the 2016 cohort I would appreciate it. (Or if anyone else can point me to an explanation that would be good.) , anyone?
|
|
