Friday, May 29, 2009

Reset Chart from Credit Suisse has a Major Error

Versions of the Credit Suisse reset graph have been featured in the Financial Times, used by the International Monetary Fund, and are a staple of web sites like these that are dedicated to the housing bubble. It’s fairly shocking then to see what appears to be a major error in their calculations.

Last week (here and here) I noted that something didn’t make sense about Wells Fargo’s estimates that only one third of one percent of their Option-ARM loans would recast before 2012. Reader’s comments and an email from CalcualtedRisk helped me to get my head around what was going on. In fact, Wells Fargo’s numbers are technically correct.

First for some background, when Wells Fargo purchased Wachovia it acquired about $120 billion of Option-ARM loans originated by a Northern California company called Golden West. These Option-ARMs, or “Pick-A-Payment” loans, allowed borrowers to literally pick their monthly payment from four options: 1) a payment that would pay off the loan in 15 years, 2) a payment that would pay off the loan in 30 years, 3) a payment that only covered the interest, and 4) a minimum payment that was less than the interest only payment.

The last option allowed for “deferred interest” or “negative amortization”. In essence, since not all the interest due was being paid by the borrower, the principal balance on the loan would grow each month. This was an extremely popular choice for borrowers at the peak of the real estate bubble, as Golden West noted in their 2005 year-end filing with the SEC:

In 2005, the initial monthly payment selected on almost all new loans was lower than the amount of interest due on the loans.

For obvious reasons this has been of particular concern. If the amount one owes on a loan goes up while the value of the home is decreasing, there is a heightened risk the borrower might walk away from the home leaving the bank to deal with the loss.

As the housing market has continued to deteriorate, the question on everyone’s mind is when will borrowers have to face the reality that they can't make a minimum payment forever? The Credit Suisse chart has been a guidepost for that reality check.

But interest rate resets don't pose much a problem for those with Option-ARMs when rates are low. They can continue making a minimum payment that only adjusts upwards by 7.5% once a year. The real "payment shock" comes when the loan is recast (i.e. becomes fully amortizing over the remaining term of the loan). This occurs after a contractual time limit (normally 5 or 10 years) or, if due to negative amortization, the loan value increases to 110-125% of the original principal of the loan.

For this reason Credit Suisse used the recast date in their chart above for Option-ARMs as stated by Mathew Padilla of the Orange County Register:
Credit Suisse, an analyst told me, used resets in the chart for all loans except option adjustable-rate mortgages, when borrowers can choose a minimum payment that may be less than interest owed (option ARMs are in yellow on the chart… see how they are rising!). For option ARMs it used recasts, which can happen either when the loan amount expands to a maximum allowed — often 115% or 125% of original principal — or a set period, such as five years.

If you’re paying close attention you'll notice that something doesn't add up. Wells Fargo, who holds more Option-ARMs on its books than any other institution, states in their last 10-Q filing:
Based on assumptions of a flat rate environment, if all eligible customers elect the minimum payment option 100% of the time and no balances prepay, we would expect the following balance of loans to recast based on reaching the principal cap: $4 million in the remaining three quarters of 2009, $9 million in 2010, $11 million in 2011 and $32 million in 2012... In addition, we would expect the following balance of ARM loans having a payment change based on the contractual terms of the loan to recast: $20 million in the remaining three quarters of 2009, $51 million in 2010, $70 million in 2011 and $128 million in 2012.

In short, Wells expects $56 million in Option ARMs to recast due to the loan balance reaching 125% of the value of the original loan and another $269 million to recast based on the terms of the loan. Given that we’re talking about a portfolio of over $100 BILLION of these loans, this means ESSENTIALLY NO LOANS WILL RECAST due to the negative amortization limits or contractual terms before 2012.

Both assumptions seemed suspect, yet, they are in fact true. Looking at page 55 of the Golden West 10-K from 2005 we read:

...most of our loans are scheduled to have a payment change without respect to any annual limit in order to reamortize the loan over its remaining life at the end of the tenth year or when the loan balance reaches 125% of the original amount. We term this reamortization a “recast.” Historically, most loans in our portfolio have paid off before the loan’s payment is recast.

History doesn’t look like it will be a good guide going forward but this at least clearly spells out what we are facing. If recasts don’t happen contractually for 10 years this means that the $49 billion of Golden West Option ARMs originated in 2004 will recast in 2014, and the $51 billion originated in 2005 will recast in 2015.

But take a look at the chart. Credit Suisse shows NOT A SINGLE OPTION-ARM RECAST IN 2014 (nor any in 2013 or the first three months of 2015).

Maybe Credit Suisse moved the recasts up based on the loans hitting their 125% balance caps as they state in their footnote? But again, the numbers don’t add up to what Wells states in their last 10-Q: virtually no recasts due to "reaching the principal cap" before 2012.

They also don’t add up if you model this out in excel with some actual numbers. Take a look at the table to the right (click for larger image):

Here we look at a hypothetical $500,000 Option-ARM mortgage originated by Golden West in January 2004 that now sits on Wells Fargo’s books. In order for the loan to recast before 2014 it needs to have enough negative amortization to hit the 125% limit. Wells Fargo assumes a "flat rate" environment in their calculations, but I assumed that today the CODI index (used for a majority of Golden West loans) plus margin jumps to 7% from its current level of 4.875%. At the end of the year I assume it jumps another 100 basis points to 8% and stays there until 2014.

Even with the assumption of higher rates the loan never hits its 125% cap. The loan currently has accrued about $40,000 of negative amortization. By 2014 it would accrue another $45,000, making the final balance $585,388 (115% of the original loan). In order to get the loan to hit the 125% cap by 2012 the accrual rate would need to jump today to 10% and say there for the next 3 years.*

In short, the Credit Suisse chart above looks to be incorrect. Contractually, these recasts won’t happen until 2014 and given the generous 125% cap it is unlikely negative amortization makes these loans hit that limit unless we have a major spike in interest rates. As mind boggling as it may seem, recasts for Wells Fargo’s giant Option-ARM portfolio won’t take place for another 5 years.

Of course, none of this is to say that Wells Fargo is out of the woods. They are essentially stashing away on their balance sheet tens of billions of neg-am loans that will recast into 20-year fixed rate mortgages in 2014 and 2015. Talk about a payment shock. (Although, it is important to note that the minimum payment automatically increases by 7.5% a year. If rates were to stay low this could mean that borrowers would eventually start paying more than their interest only payment and paying down their loans. My guess though is that these automatic increases will get people walking away from their homes before the recast date. Who wants to pay $3000 a month on a home that's underwater? Any way you look at it, it’s a mess.)

Also worrisome is that we’ve heard from readers that WaMu/JP Morgan is notifying Option-ARM borrowers that they are extending their minimum payments out another 5 years. Are they pushing off recasts into 2014/15 as well?

The bottom line something doens't add up when Wells Fargo predicts virtually no Option-ARM recasts before 2012, while Credit Suisse predicts no recasts after 2012. While I would guess Wells Fargo is underestimating recasts assuming a flat rate environment, the bulk of recasts do look like they will be pushed out to 2014/2015. Surely, some of these recasts need to be reflected in the Credit Suisse chart. I’ll leave it to others what this means for the housing market, but what is clear is this needs more attention.

* [To verify these numbers download this spreadsheet from CalcualtedRisk, update the index with the CODI series, and plug in your own home values.]

Update: Regarding the 125% cap, to clarify, this was for loans with a LTV of less than 85% at origination. This represents the structure of "almost all" the Option-ARM loans according the the Golden West 10-K:

A 125% cap on the loan balance applies to loans with original loan-to-value ratios at or below 85%, which includes almost all of the loans we originate. Loans with original loan-to-values above 85% have a 110% cap.

Wikinvest Wire