|
Post by diversifier on Jul 26, 2017 14:04:02 GMT
I thought I understood what RS were doing, but events over the past few days have really made me think deep (maybe) & confused me...... Perhaps someone can help - or is this just *really* obvious to more sophisticated people than me?
My understanding: When we buy e.g. in 1-yr market we can't match directly to a 1-yr contract, both because there are other non-investable borrowing durations, and also it can't be time-aligned. The market-term is only *indicative* of the real term & interest-rate of the contracts which run under the hood, so RS separate out into principal & interest repayments. The "interest" is calculated by RS based on matched%. The "borrower bid" is just a synthetic value calculated effectively on the PF's behalf to manage it's expected default liability. The principal remaining on a particular contract after 1-yr is "some fraction" of the original, but *assumed* bought by the market if you want it, independent whether you Sellout before or after 1-yr. The "market-term" really only determines the Sellout fee at the time you do so. This is all fair enough (including the Sellout fee) - although I didn't understand it until I went into a darkened room, pondered for a bit, and bashed a spreadsheet.
My questions/suspicion: Does RS's ability to do maturity transformation allow them to select the borrower market-term to maximise RS profit? The amount of borrowing going onto each market doesn't square at all with the average origination term of 26 months It seems that RS is putting lots of e.g. 5-yr real borrowers into the Rolling Market @3.3% rather than 5-yr @ 4.8% or 1-yr, and pocketing the difference....[*]deliberately provocative, see below. This causes an apparent lack of 1-yr, 3-yr, 5-yr borrowers, exactly what we're seeing.
The "normal" yield curve expresses investors time-demand for money. If I invest money on Rolling & investor liquidity dries, it seems RS T&C's may not return the principal as normal after a month but roll it over to re-finance *borrowers contracts* who are committed up to 5yrs (average 19-month-remaining). But still at the original 1-month rate ! I now suspect that the very flat yield curve (1.5% premium for 5-yr money) prices in a liquidity freeze on Rolling, rather than a borrowers conspiracy on longer terms. That's not a problem, as long as I understand and accept that the Rolling rate *actually* prices in up to 19-month maturity.
[*] As a individual I wouldn't do this (borrow short, lend long, the mainstay of standard banking) because of the liquidity risk. But the investors bear the liquidity risk, not RS! Crucially, what does the emotive phrase "RS profit" translate to? Does it really mean "the cheapest way the *PF* can source money from investors"? Because that's OK in principle - as an algorithm it ought to do that, but suggests that it doesn't have the correct constraints to tradeoff the implied liquidity risk to investors versus return. Re-phrased, the only metric of PF sufficiency I've seen is coverage ratio. But the PF just sums over all maturities without considering liquidity risk between maturities! Once maturity transformation is allowed, *of course* that system will converge to flat-yield-curve zero-nominal-maturity indeterminate-actual-maturity.
|
|
registerme
Member of DD Central
Posts: 6,163
Likes: 5,977
|
Post by registerme on Jul 26, 2017 19:08:53 GMT
I like the analysis but I can't tell you if it's correct or not because I simply don't know (I am fairly sure that some of it is, but I suspect that some of it may not be). There's been a fair amount of discussion about what goes on under the RS hood, but unfortunately it's been spread far and wide across many threads in this sub-forum so you'll have to do some digging....
If I can find the time I'll see if I can root anything relevant out.
|
|
mary
Member of DD Central
Posts: 698
Likes: 711
|
Post by mary on Jul 26, 2017 20:03:12 GMT
I thought I understood what RS were doing, but events over the past few days have really made me think deep (maybe) & confused me...... Perhaps someone can help - or is this just *really* obvious to more sophisticated people than me? My understanding: When we buy e.g. in 1-yr market we can't match directly to a 1-yr contract, both because there are other non-investable borrowing durations, and also it can't be time-aligned. The market-term is only *indicative* of the real term & interest-rate of the contracts which run under the hood, so RS separate out into principal & interest repayments. The "interest" is calculated by RS based on matched%. The "borrower bid" is just a synthetic value calculated effectively on the PF's behalf to manage it's expected default liability. The principal remaining on a particular contract after 1-yr is "some fraction" of the original, but *assumed* bought by the market if you want it, independent whether you Sellout before or after 1-yr. The "market-term" really only determines the Sellout fee at the time you do so. This is all fair enough (including the Sellout fee) - although I didn't understand it until I went into a darkened room, pondered for a bit, and bashed a spreadsheet. My questions/suspicion: Does RS's ability to do maturity transformation allow them to select the borrower market-term to maximise RS profit? The amount of borrowing going onto each market doesn't square at all with the average origination term of 26 months It seems that RS is putting lots of e.g. 5-yr real borrowers into the Rolling Market @3.3% rather than 5-yr @ 4.8% or 1-yr, and pocketing the difference....[*]deliberately provocative, see below. This causes an apparent lack of 1-yr, 3-yr, 5-yr borrowers, exactly what we're seeing. The "normal" yield curve expresses investors time-demand for money. If I invest money on Rolling & investor liquidity dries, it seems RS T&C's may not return the principal as normal after a month but roll it over to re-finance *borrowers contracts* who are committed up to 5yrs (average 19-month-remaining). But still at the original 1-month rate ! I now suspect that the very flat yield curve (1.5% premium for 5-yr money) prices in a liquidity freeze on Rolling, rather than a borrowers conspiracy on longer terms. That's not a problem, as long as I understand and accept that the Rolling rate *actually* prices in up to 19-month maturity. [*] As a individual I wouldn't do this (borrow short, lend long, the mainstay of standard banking) because of the liquidity risk. But the investors bear the liquidity risk, not RS! Crucially, what does the emotive phrase "RS profit" translate to? Does it really mean "the cheapest way the *PF* can source money from investors"? Because that's OK in principle - as an algorithm it ought to do that, but suggests that it doesn't have the correct constraints to tradeoff the implied liquidity risk to investors versus return. Re-phrased, the only metric of PF sufficiency I've seen is coverage ratio. But the PF just sums over all maturities without considering liquidity risk between maturities! Once maturity transformation is allowed, *of course* that system will converge to flat-yield-curve zero-nominal-maturity indeterminate-actual-maturity. From my experience, I think you are mostly correct, obviously we don't know exactly, but when the Rolling rate peaks you often find that a repayment occurs very quickly. I once got 16% - which is obviously unsustainable, and it was quickly repaid. On other threads, people have commented that their matches have been repaid in a few hours when they got a late in the day match. It seems to me that this is an abuse of the P2P model, and maybe who RS keep failing to achieve full FCA compliance!
|
|
ceejay
Posts: 970
Likes: 1,149
|
Post by ceejay on Jul 27, 2017 7:32:51 GMT
Well, I don't know either, but I'd like to! It seems to me that what you've described must apply to the rolling market.
But I'm not so sure about the one-year and five-year markets. I'd really like to know, though, because a major attraction of the alleged one year market is that it limits liquidity risk for the lender: if I put money in there and it is matched to a real loan, then I "know" (subject to defaults) that I will get my money back then and I'm not dependent on liquidity at that time. It shouldn't matter if P2P has gone down the drain and all the lenders have disappeared (a significant systemic risk in P2P) - I will be repaid.
However, if the model you describe is correct and the one year loans are synthetic, the liquidity risk is real and I really ought to be made aware of that.
|
|
|
Post by investor1925 on Jul 27, 2017 7:54:05 GMT
Right now, 08:45 there are NO borrower offers on the 5 year market & £2.2 million of lender's money.
On the rolling market the borrowers & lenders are virtually matched at £2 million.
If some of the borrowers offers on the rolling market are indeed 5 year "rolling" loans, one of 2 things may happen.
1. When the rate on the rolling exceeds the 5 year, the borrowers offer will be pulled & placed on the 5 year, so the borrowers total will reduce, thus reducing rolling rates over time.
2. If a "rolling" 5 year loan is taken on the rolling market at above what the 5 year rate is presently, it will be paid back very quickly, probably within hours, then be placed on the 5 year market.
The same scenario may work for the 1 year market, where rates are already lower than the rolling market.
could be interesting
|
|
jlend
Member of DD Central
Posts: 1,817
Likes: 1,444
|
Post by jlend on Jul 27, 2017 10:11:51 GMT
Well, I don't know either, but I'd like to! It seems to me that what you've described must apply to the rolling market. But I'm not so sure about the one-year and five-year markets. I'd really like to know, though, because a major attraction of the alleged one year market is that it limits liquidity risk for the lender: if I put money in there and it is matched to a real loan, then I "know" (subject to defaults) that I will get my money back then and I'm not dependent on liquidity at that time. It shouldn't matter if P2P has gone down the drain and all the lenders have disappeared (a significant systemic risk in P2P) - I will be repaid. However, if the model you describe is correct and the one year loans are synthetic, the liquidity risk is real and I really ought to be made aware of that. Suggest you have a good read of the Lender Agreement. For example: 4.8. An offer to lend may be matched with all or part of a loan which has a term that differs from your offer. In cases where the term is longer than the term you selected, we will use our best efforts to transfer the loan to another lender at the end of the term you selected. You should be aware however that our ability to do so will be subject to there being funds available on the Exchange at the required rate. If there are insufficient funds available on the Exchange, your money will be returned to you as and when the borrower repays their loan, which may take longer than the term you selected. In cases where the term is shorter than the term you selected, your investment may be repaid early. If this happens, the sum paid back to you will be dealt with in accordance with the Re-Investment settings you have selected in your RateSetter Account.Ratesetter has been going 7 years now so it's not new news, although perhaps not everyone realises.
|
|
|
Post by diversifier on Jul 27, 2017 11:43:32 GMT
Thanks all, It seems people agree that RS are allowed to select the apparent borrower term as something other than naive assumption of nearest available; and that many but not all investors realise this (I hadn't before this!)
What RS actually choose to do is opaque, but I could try to infer from fast-repayment or borrower match amounts. If we see that, and see that it tends to equalise the term-rates, then I can draw my own inference. Right now (12:21), 1-yr unfroze slightly, and although the last match was 4.2%, I find it significant that there is borrower money @3.8%. And looking at Rolling lender offers, the volume starts about 3.7% and there's an odd hole at 3.8%. That does seem suggestive that RS took a trance of "borrower offer" money that they were prepared to stick at 3.8%, and shared it between terms. Although the 1-yr yield is bizarrely inverted, the prediction is that 5-yr really can't invert in equilibrium because that's the maximum real borrower term. Difficult to prove, because inversion is crazy unless you believe interest rates are going down.
|
|
robski
Member of DD Central
Posts: 772
Likes: 462
|
Post by robski on Jul 27, 2017 12:18:15 GMT
Rate setter always match rates as the day goes by. I suspect this is loans people ask for now, Ie want to draw down immediately, I mean why wouldn't you if you are buying a car or a phone or something
The bulk normally matches around mid day, this is the rolled over amounts released around midnight.
I thought the majority of the 1 year fixed are bullet type loans, so actually made for 1 year, so no interim payments and all at the end, where as the 5 (and now defunct 3) year loans were loans written up to that length (eg you can get a 4 year term on the 5 year product). The rolling I believe is where people have sold out from longer terms and these then become rolling loans that are matched monthly
But overall its a bit of a lottery lending on any product with rate setter. You may get locked in, however if this happens i think they will find they face severe restrictions on their ability to lend so whilst it wouldn't cause the collapse directly I think they may have severe funding issues, and hence I think they are very unlikely to do this unless its absolutely the only option left to them
|
|
ceejay
Posts: 970
Likes: 1,149
|
Post by ceejay on Jul 27, 2017 12:39:49 GMT
But overall its a bit of a lottery lending on any product with rate setter. You may get locked in, however if this happens i think they will find they face severe restrictions on their ability to lend so whilst it wouldn't cause the collapse directly I think they may have severe funding issues, and hence I think they are very unlikely to do this unless its absolutely the only option left to them I don't suppose anyone thinks that RS would indulge in failing to pay off, say, a 1 year loan at the end of its term unless they felt they absolutely had to. But I don't think that's the point. I think it's about risk scenarios, and how lenders can best understand and manage the risks. The scenario I'm thinking of here is where P2P generally has taken a battering and lenders are heading for the hills. Could be any number of causes for this, maybe not starting at RS, but that doesn't matter: the issue is that as a lender I'd like my money back, and a conscious trade-off that I'd like to be able to make is return vs. period: if I lock my money up for longer, I expect a higher return. And if I've decided to go for a lower return at (say) a one year period then I'd be mightily pissed off if at the end of the year I couldn't get my money back - not because the borrower has defaulted (that's another risk entirely) but because the borrower I'm matched with has another year to run but there are no new lenders to take over the loan. And in this scenario, BTW, the PF is probably irrelevant as it would already have been spent covering bad loans...
|
|
robski
Member of DD Central
Posts: 772
Likes: 462
|
Post by robski on Jul 27, 2017 13:35:38 GMT
But overall its a bit of a lottery lending on any product with rate setter. You may get locked in, however if this happens i think they will find they face severe restrictions on their ability to lend so whilst it wouldn't cause the collapse directly I think they may have severe funding issues, and hence I think they are very unlikely to do this unless its absolutely the only option left to them I don't suppose anyone thinks that RS would indulge in failing to pay off, say, a 1 year loan at the end of its term unless they felt they absolutely had to. But I don't think that's the point. I think it's about risk scenarios, and how lenders can best understand and manage the risks. The scenario I'm thinking of here is where P2P generally has taken a battering and lenders are heading for the hills. Could be any number of causes for this, maybe not starting at RS, but that doesn't matter: the issue is that as a lender I'd like my money back, and a conscious trade-off that I'd like to be able to make is return vs. period: if I lock my money up for longer, I expect a higher return. And if I've decided to go for a lower return at (say) a one year period then I'd be mightily pissed off if at the end of the year I couldn't get my money back - not because the borrower has defaulted (that's another risk entirely) but because the borrower I'm matched with has another year to run but there are no new lenders to take over the loan. And in this scenario, BTW, the PF is probably irrelevant as it would already have been spent covering bad loans... So its the wrong product for you then, as clearly per their T&Cs you have no guarantee. Just like from what i can tell all P2P Zopa about to scrap their fund so you could get locked in for years My understanding is that all loans would be pooled should they have to not repay capital and that way everyone would be getting some return not pot luck. The closest model I can see is ZOPA, Ive got plenty of 5 year loans written by them at 2.x and 3.x%, I have no more control there. Really if you don't like the product you shouldn't invest in it. Its not a savings account its investment, and as such terms like I want my money back on this set date guaranteed no longer apply. All IMHO.
|
|
jlend
Member of DD Central
Posts: 1,817
Likes: 1,444
|
Post by jlend on Jul 27, 2017 14:10:03 GMT
But overall its a bit of a lottery lending on any product with rate setter. You may get locked in, however if this happens i think they will find they face severe restrictions on their ability to lend so whilst it wouldn't cause the collapse directly I think they may have severe funding issues, and hence I think they are very unlikely to do this unless its absolutely the only option left to them I don't suppose anyone thinks that RS would indulge in failing to pay off, say, a 1 year loan at the end of its term unless they felt they absolutely had to. But I don't think that's the point. I think it's about risk scenarios, and how lenders can best understand and manage the risks. The scenario I'm thinking of here is where P2P generally has taken a battering and lenders are heading for the hills. Could be any number of causes for this, maybe not starting at RS, but that doesn't matter: the issue is that as a lender I'd like my money back, and a conscious trade-off that I'd like to be able to make is return vs. period: if I lock my money up for longer, I expect a higher return. And if I've decided to go for a lower return at (say) a one year period then I'd be mightily pissed off if at the end of the year I couldn't get my money back - not because the borrower has defaulted (that's another risk entirely) but because the borrower I'm matched with has another year to run but there are no new lenders to take over the loan. And in this scenario, BTW, the PF is probably irrelevant as it would already have been spent covering bad loans... Liquity risk in general is not unique to RS in the P2P world on these sort term type acounts. It is possible to get locked into the AC QAA and 30 day account's if there are not enough funds although this has not happened. The RS and AC setups are not the same but a liquity risk has always been there on these short term accounts.
|
|
|
Post by diversifier on Jul 27, 2017 15:11:40 GMT
"So its the wrong product for you then, as clearly per their T&Cs you have no guarantee. "
I think that's a half-truth. I agree that the T&Cs don't provide a guarantee.
But RS do clearly advertise separate *products* - the Rolling Rate, 1-year & 5-yr. It seems that a fairly subtle consequence of those T&Cs, is that under risk scenarios those separate products perform identically. The apparent key distinction of term, is actually no distinction at all, the yield-curve has gone nearly flat or inverted. This may be due to an internal policy/algorithm that we have no information about. So, on what basis does the investor select between Rolling & 5-yr?
Any investment can lose money, and RS are reasonably clear about the overall risks. But this is less clear-cut. I guess what I'd like is for RS to tell us the average duration of the underlying contracts on the separate markets Rolling, 1-yr, 5-yr . Think of it like a bond-fund. You expect to know the duration, and its quoted on Morningstar or whatever. Duration isn't part of the T&C's of the bond fund, and can certainly vary without notice. But you wouldn't be very happy if the manager had undisclosed cross-holdings between their separate funds with widely different durations. That *is* relevant market information.
I'm thinking you *might* respond: no, then still P2P isn't the right investment class. Metrics like duration are conventional for bond-funds, but there just aren't conventions like that for P2P it's a different investment class. You may be right.
|
|
ceejay
Posts: 970
Likes: 1,149
|
Post by ceejay on Jul 27, 2017 15:37:12 GMT
.. And if I've decided to go for a lower return at (say) a one year period then I'd be mightily pissed off if at the end of the year I couldn't get my money back - not because the borrower has defaulted (that's another risk entirely) but because the borrower I'm matched with has another year to run but there are no new lenders to take over the loan... Liquity risk in general is not unique to RS in the P2P world on these sort term type acounts. It is possible to get locked into the AC QAA and 30 day account's if there are not enough funds although this has not happened. The RS and AC setups are not the same but a liquity risk has always been there on these short term accounts. I understand this form of liquidity risk in relation to RS's Rolling product, also the AC QAA and 30 day accounts (as it happens I am invested in all of these!). The question now, though, is whether RS's 1 year account is one of these "short term accounts" or whether it is what it actually claims to be, at least in their marketing, which are one year term loans with a known end date. We can see that their Ts and Cs allow them to deviate from this principle - what I'd like to know is whether they are in fact doing so. Much as diversifier has requested in the preceding post, in fact.
|
|
robski
Member of DD Central
Posts: 772
Likes: 462
|
Post by robski on Jul 27, 2017 16:11:06 GMT
"So its the wrong product for you then, as clearly per their T&Cs you have no guarantee. " I think that's a half-truth. I agree that the T&Cs don't provide a guarantee. But RS do clearly advertise separate *products* - the Rolling Rate, 1-year & 5-yr. It seems that a fairly subtle consequence of those T&Cs, is that under risk scenarios those separate products perform identically. The apparent key distinction of term, is actually no distinction at all, the yield-curve has gone nearly flat or inverted. This may be due to an internal policy/algorithm that we have no information about. So, on what basis does the investor select between Rolling & 5-yr? Any investment can lose money, and RS are reasonably clear about the overall risks. But this is less clear-cut. I guess what I'd like is for RS to tell us the average duration of the underlying contracts on the separate markets Rolling, 1-yr, 5-yr . Think of it like a bond-fund. You expect to know the duration, and its quoted on Morningstar or whatever. Duration isn't part of the T&C's of the bond fund, and can certainly vary without notice. But you wouldn't be very happy if the manager had undisclosed cross-holdings between their separate funds with widely different durations. That *is* relevant market information. I'm thinking you *might* respond: no, then still P2P isn't the right investment class. Metrics like duration are conventional for bond-funds, but there just aren't conventions like that for P2P it's a different investment class. You may be right. I don't think its a half truth sorry. They clearly advise in the T&Cs that under normal circumstances rolling (approx 1 month), 1 year lending and 5 year lending all have the risk of lock in should funds not be available to replace if you you don't want to carry on before the loan is fully repaid. The rolling particularly is assuming liquidity. If there is a super liquidity issue then the whole of RS will change. Whilst they state this has never happened, its a risk inherent with the type of lending. If you can't accept that risk, or you don't believe the premium on the rate you are receiving supports that risk you don't invest. The model relies on an element of unknown, if today you put a £1k deposit and it matched to a just liquidated loan with 4.5 of 5 years remaining you may go, nope not for me, and immediately ask to cash that one back. The whole system would fail with too much visibility. I am pretty certain the 1 year loans are indeed that, because, 1) they don't write many the volume is low, 2) the rates were for some time lower than rolling and 5 year yet they remained being written at low volume. If RS were matching all types of loan to which ever fund was cheapest they would have been using all that low % 1 year money, but they didn't. I have also seen large round sums on this product, like £500k that appear then match some minutes later, still convincing me these are large round bullet loans not 5 year car loans. But most of that is speculation and observations. RS are (rightly in my opinion) fairly black box. The yield curve is a function of investors more than RS, although plenty refuse to accept that. Rolling tends to increase or decrease due to excess or shortage of liquidity, and the setup of investors. The higher the amount set to reinvest at "market rate" the faster or slower this is reflected. 1 year is probably the most volatile. 5 year can be volatile in reaction to spikes as well. It really is a unique asset class, there is no fixed term in relation to product, but in fact a rough assumption that most of the time you get to make a new decision after a month on rolling, or a year on 1 year, or each month on a small percentage of a 5 year.
|
|
jlend
Member of DD Central
Posts: 1,817
Likes: 1,444
|
Post by jlend on Jul 27, 2017 16:29:04 GMT
Liquity risk in general is not unique to RS in the P2P world on these sort term type acounts. It is possible to get locked into the AC QAA and 30 day account's if there are not enough funds although this has not happened. The RS and AC setups are not the same but a liquity risk has always been there on these short term accounts. I understand this form of liquidity risk in relation to RS's Rolling product, also the AC QAA and 30 day accounts (as it happens I am invested in all of these!). The question now, though, is whether RS's 1 year account is one of these "short term accounts" or whether it is what it actually claims to be, at least in their marketing, which are one year term loans with a known end date. We can see that their Ts and Cs allow them to deviate from this principle - what I'd like to know is whether they are in fact doing so. Much as diversifier has requested in the preceding post, in fact. Will loans less than 5 year be funded from the rolling and one year markets? RateSetter a year ago. Hi JLend - money on the 5 Year market will continue to fund amortising loans with terms of 4 and 5 years. Money on the 1 Year market funds 1 year loans, and the rest (including some 4 and 5 Year loans) is funded by money on the Rolling market. www.ratesetter.com/blog/article/3-year-product-changesIf you post questons on one of the RateSetter blog's on their website they respond in a few days. I've always found that works now they don't have a unofficial rep on this forum. Hope that helps.
|
|