MaxedOutMama
Thursday, October 19, 2006
Mortgage Vintages
I have been staring unhappily at various stats relating to delinquencies for the last several days. The long and the short of it is that early returns suggest that both the 06 and 05 mortgage "vintages" are going sour, with way too many early defaults.
It's too soon to tell what that means. There are several excellent posts and comment threads on Calculated Risk discussing the matter. Try here and here, and especially Tanta's comment. I think she explained the potential implications very well - in words yet.
Before I continue on into coma-inducing boredom, I'd just like to point out one thing: Given the metrics of a large portion of loan originations since 2004, a surprisingly high number of loans are being paid at a rate which would have in the past qualified them as being in some stage of delinquency. The only difference the new loan terms may make over the long term is that loss allowances are lower than they would have been.
Onto the staggering boredom:
Historically, the highest defaults in mortgages occur in years 3-5, with a steep drop-off after 7 to 10 years. If you want more information, try this OTS document from 2002. This is because if borrowers are correctly qualified, they can make current payments unless something material occurs to change their circumstances, and the chance of something material occurring to change their circumstances (such as job loss, illness, economic downturn) rises with time. However, traditional mortgages have required at least some downpayment, and also included small principal payments in the first few years. So with time and hopefully a bit of appreciation, most homeowners would be able to keep their homes by refinancing/workouts or sell them before they got into too much trouble.
Many things may cause financial trouble for a homeowner, but generally speaking, only being unable to sell a home for enough to clear your debt on the home causes foreclosures. Obviously adjustable-rate mortgages with lower initial terms carry more risk for earlier defaults.
According to DataQuick, in CA:
What is relatively startling is the median age of the defaulted loans cited. The borrowers had to fall into arrears within the first year of the loan, which have some pretty hefty implications. Early returns on the '06s aren't sounding positive either, with a disproportionately high rate of defaults being shown. Without knowing the exact characteristics of the borrowers and loans, it's impossible to figure the true implications. It is possible to say that everyone will be looking very carefully at their portfolios, though. Early defaults like these just blow up many expectations for mortgage performance.
To get an idea of what those implications might be, I would like to direct you to this white paper (pdf, 33 pages) available from LoanPerformance. The conclusions were overly optimistic, but the figures were pretty good as of the third quarter of 2005. If you go to the 14th page of the document (internal numbering page 13), you will see a tabulation by state of the percentage of borrowers having less than 5% and less than 15% equity. Since many markets have indeed dropped 10%, we are now concerned with the 15% equity column. These are the worst states:
Unfortunately, the proportion of borrowers having low or negative equity and ARMs is higher than the proportion of borrowers having low or negative equity and fixed rates. Last year, 17% of borrowers with ARMs had negative equity, and 39% of borrowers with ARMs had less than 15% equity. These borrowers are the likeliest to face repayment problems and the least likely to be able to work around the problem by refinancing or selling. This is also known as a "risk concentration". According to the white paper, of borrowers who had ARMS originated in 2005, 32.3% had zero or negative equity, and nearly 52% of them had less than 10% equity. Assuming even a 5% price drop if you have to sell ahead of foreclosure, and adding the cost of selling, which is running at least 7% in many of these markets between commissions and cash-back allowances at closing, that 52% of the '05 ARM class is now at severe risk for several more years.
In other words, there is reason for concern.
In other economic news, the G.17 report for September showed an industrial production drop, and it was a broad-based drop. I think we reached tipover for the general economy. Only mining was up, and mining was the sector which had been down year-over-year.
I favor a Merton-like model for default forecasting in a declining market. According to Merton models, we may be in for a very, very tough and rough slide down. I hunted around to find something free, and here's this article about portfolio forecasting:
It's too soon to tell what that means. There are several excellent posts and comment threads on Calculated Risk discussing the matter. Try here and here, and especially Tanta's comment. I think she explained the potential implications very well - in words yet.
Before I continue on into coma-inducing boredom, I'd just like to point out one thing: Given the metrics of a large portion of loan originations since 2004, a surprisingly high number of loans are being paid at a rate which would have in the past qualified them as being in some stage of delinquency. The only difference the new loan terms may make over the long term is that loss allowances are lower than they would have been.
Onto the staggering boredom:
Historically, the highest defaults in mortgages occur in years 3-5, with a steep drop-off after 7 to 10 years. If you want more information, try this OTS document from 2002. This is because if borrowers are correctly qualified, they can make current payments unless something material occurs to change their circumstances, and the chance of something material occurring to change their circumstances (such as job loss, illness, economic downturn) rises with time. However, traditional mortgages have required at least some downpayment, and also included small principal payments in the first few years. So with time and hopefully a bit of appreciation, most homeowners would be able to keep their homes by refinancing/workouts or sell them before they got into too much trouble.
Many things may cause financial trouble for a homeowner, but generally speaking, only being unable to sell a home for enough to clear your debt on the home causes foreclosures. Obviously adjustable-rate mortgages with lower initial terms carry more risk for earlier defaults.
According to DataQuick, in CA:
Trustees deeds recorded on homes totaled 3,424 during the third quarter, up 76.9 percent from 1,936 for the previous quarter, and up 362.1 percent from 741 for last year's third quarter. Trustees deeds, or actual foreclosure sales, peaked at 14,896 in second-quarter 1997, and hit a low of 636 in the second quarter of last year.Now, this is still not a huge number of defaults, although the number of foreclosures is rising much more rapidly than defaults, meaning that way too many people going into default can't sell ahead of foreclosure. In CA it takes at least a few months after the NOD (notice of default) to foreclose, so the current foreclosures relate more to the previous quarter's defaults. It looks like about 1 out of 6, which would mean that foreclosures should go up further in this quarter. The speed of the change is somewhat daunting, too, since the all-time low was reached in 2005.
...
Lending institutions sent 26,705 default notices to homeowners in the state during the three-month period ending in September. That was up 28.3 percent from 20,812 for the prior quarter, and up 111.8 percent from 12,606 for 2005's third quarter, according to DataQuick Information Systems.
The median age of the home loans that went into default last quarter was 14 months, and more than half were originated in 2005.
On primary mortgages, homeowners were a median of five months behind on their payments when the lender started the default process. The borrowers owed a median $9,829 on a median $306,000 mortgage.
On lines of credit, homeowners were a median six months behind on their payments. Borrowers owed a median $3,200 on a median $60,000 credit line
What is relatively startling is the median age of the defaulted loans cited. The borrowers had to fall into arrears within the first year of the loan, which have some pretty hefty implications. Early returns on the '06s aren't sounding positive either, with a disproportionately high rate of defaults being shown. Without knowing the exact characteristics of the borrowers and loans, it's impossible to figure the true implications. It is possible to say that everyone will be looking very carefully at their portfolios, though. Early defaults like these just blow up many expectations for mortgage performance.
To get an idea of what those implications might be, I would like to direct you to this white paper (pdf, 33 pages) available from LoanPerformance. The conclusions were overly optimistic, but the figures were pretty good as of the third quarter of 2005. If you go to the 14th page of the document (internal numbering page 13), you will see a tabulation by state of the percentage of borrowers having less than 5% and less than 15% equity. Since many markets have indeed dropped 10%, we are now concerned with the 15% equity column. These are the worst states:
ST; <5%; <15%;Not surprisingly, these states now are popping up in the lists of states with high foreclosure rates. Median prices are down nationwide according to NAR, but median prices are also overstated because of the attached cash-backs and other giveaways. So interpolate between the 5% and 15% columns to get an approximation of people who can't afford to sell unless they can bring money to the table. If they go into default, they are quite likely to lose the home.
GA; 10.8%; 27.1%
KY; 14.3%; 27.2%
WA; 15.0%; 28.1%
IL; 15.6%; 28.1%
NM; 15.0%; 28.3%
MN; 16.7%; 28.9%
IN; 17.4%; 32.7%
NC; 13.3%; 33.2%
MO; 20.1%; 33.9%
OK; 19.9%; 37.4%
WI; 24.3%; 38.5%
OH; 23.9%; 39.8%
AR; 21.3%; 40.0%
IL; 22.1%; 40.1%
MI; 24.6%; 40.6%
CO; 28.5%; 47.0%
TN; 28.0%; 47.8%
Unfortunately, the proportion of borrowers having low or negative equity and ARMs is higher than the proportion of borrowers having low or negative equity and fixed rates. Last year, 17% of borrowers with ARMs had negative equity, and 39% of borrowers with ARMs had less than 15% equity. These borrowers are the likeliest to face repayment problems and the least likely to be able to work around the problem by refinancing or selling. This is also known as a "risk concentration". According to the white paper, of borrowers who had ARMS originated in 2005, 32.3% had zero or negative equity, and nearly 52% of them had less than 10% equity. Assuming even a 5% price drop if you have to sell ahead of foreclosure, and adding the cost of selling, which is running at least 7% in many of these markets between commissions and cash-back allowances at closing, that 52% of the '05 ARM class is now at severe risk for several more years.
In other words, there is reason for concern.
In other economic news, the G.17 report for September showed an industrial production drop, and it was a broad-based drop. I think we reached tipover for the general economy. Only mining was up, and mining was the sector which had been down year-over-year.
I favor a Merton-like model for default forecasting in a declining market. According to Merton models, we may be in for a very, very tough and rough slide down. I hunted around to find something free, and here's this article about portfolio forecasting:
4. A roll rate, or state-transition, model is the final class considered. In such models, there is no particular need to estimate "defaults." Rather, the analyst uses a much more complex process in an attempt to estimate the probabilities that any particular loan will "transition" from its current state--for example, "current" or "30 days past due"--to any other state (such as foreclosure). These various probabilities of transitioning to any particular state are expressed in a transition matrix or, equivalently, a roll rate matrix. Since, theoretically, each loan faces such a probability matrix at the start of any given month, it is possible to use computer-intensive techniques to simulate the future path of each loan or a portfolio of loans. When any loan enters, through simulation, a state that could involve a loss (such as foreclosure), the analyst may apply a separately estimated loss model to estimate the present value of the losses attributable to the loan entering into the state of default. The study u ses a particular form of state-transition model--the LoanPerformance RiskModel[TM]--in all cases that compare the performance of such models vis-a-vis the performance of other models.
...
The state-transition model, however, can employ an estimated transition-probability matrix without needing to estimate explicit correlations (the actual correlations are embedded within the transition matrix). These matrices, moreover, can be based on all of the historical information--obligor characteristics such as FICO, loan characteristics such as loan type and age, and so forth--associated with each loan, and the model can employ assumptions regarding future probabilistic paths of key macro variables such as interest rates and housing prices on a state-by-state or MSA-by-MSA basis. For this reason, the state-transition mod els, as a class, are the most "complete" of the models examined.
...
3. A Merton-type model, like the OW model, does not necessarily need any cohorting to be done; it relies on an estimated PD and LGD for each loan in a portfolio. However, rather than use a loan's default correlation with other loans to help estimate the loss distribution, the Merton model relies on a stylistic view of the default process for each obligor. In this process, the obligor (home owner) is assumed to default only when, and always when, the value of the obligor's assets (the house price) falls below the value of the obligor's liabilities (the outstanding balance on the mortgage loan). Under this assumption, the analyst needs to estimate the distribution of house prices and, especially, the degree to which house prices are correlated across obligors.
Comments:
<< Home
Hi Mama; Another great post. Perhaps you should start a newsletter to make a few extra bucks. It would make you sick if you know what we were shelling out for stuff not .1% as good as this. Anyway, Does this mean that the lower credit quality sub debt of the cmo machine are going to fail? Remember, I'm a govie guy, full faith and credit.....Perhaps that could be the undoing of the hedge funds who own all this stuff. That would really bum me out!!!
# posted by 
: 6:17 PM
: 6:17 PM
Well, why do you think I do my own analysis? There's not much worth buying out there, and trying to figure out what is worthwhile demands that you pretty much do the analysis. But if I had to do this for money, I'd probably get really depressed.
I don't know if it is the same Merton, but part of the problem is that too many people are drawing conclusions that do not match their numbers due, I'm sure, to sales pressures.
Normally I would favor a state-transition model, but those don't work when you have a huge underlying change that affects a large part of the market. The matrices do encode past effects, but not future effects.
I don't know if it is the same Merton, but part of the problem is that too many people are drawing conclusions that do not match their numbers due, I'm sure, to sales pressures.
Normally I would favor a state-transition model, but those don't work when you have a huge underlying change that affects a large part of the market. The matrices do encode past effects, but not future effects.
As to bottom rung CLOs and CMOs, all I can say is that this isn't looking good. Unless you can pretty firmly count on isolation from the macroeconomy (a big no), this becomes a self-reinforcing cycle that has the potential to escalate with exceptional rapidity. It really depends on how those are sliced and who is holding them. I think some of these loan shops were making so much selling the tranches that they kept the lowest ones, thinking they could compensate for any losses with the profits.
I think the banks with big HELOC portfolios are in the worst trouble; whole loan portfolios should do better. They have way more room to bargain and compensate.
Another thing that shook me up badly is that I think mortgage rates will have to rise on risk (they are rising now), and that means the Fed will probably raise rates at least once more. The Fed is boxed in, because letting inflation expectations rise will push up mortgage rates, and rising mortgage rates create a self-reinforcing cycle too, as affordability drops due to rates and home prices have to come down even further to compensate.
I was wrong about the stimulating effects of the gas prices; after looking at the September CPI & PPI numbers it appears that so much throttled up pricing pressure was built into the system that prices for the consumer are not likely to drop much - or at least not now, when we really need it to boost the fourth quarter.
We could very well see further OPEC cuts in December and the Fed raising. I think OPEC would be idiotic to cut again, but then I do think that they are idiotic.
Post a Comment
I think the banks with big HELOC portfolios are in the worst trouble; whole loan portfolios should do better. They have way more room to bargain and compensate.
Another thing that shook me up badly is that I think mortgage rates will have to rise on risk (they are rising now), and that means the Fed will probably raise rates at least once more. The Fed is boxed in, because letting inflation expectations rise will push up mortgage rates, and rising mortgage rates create a self-reinforcing cycle too, as affordability drops due to rates and home prices have to come down even further to compensate.
I was wrong about the stimulating effects of the gas prices; after looking at the September CPI & PPI numbers it appears that so much throttled up pricing pressure was built into the system that prices for the consumer are not likely to drop much - or at least not now, when we really need it to boost the fourth quarter.
We could very well see further OPEC cuts in December and the Fed raising. I think OPEC would be idiotic to cut again, but then I do think that they are idiotic.
<< Home
