I find it interesting when factoids that are already in the public domain get treated as if they are news. Stephanie Pomboy, as reported by Barron’s Alan Abelson (hat tip Barry Ritholtz) tells us that there are a lot of adjustable rate mortgages that have no equity. And, of course, if housing prices fall, more fall into the negative equity terrain.
I’ll be the first to grant that that’s not a healthy situation, particularly since consumer optimism and spending is closely tied to their personal balance sheets. But the Barron’s article has a nifty table that relies heavily on a LoanPerformance database. That same database, plus other another large mortgage database was mined exhaustively by Chris Cagan of American CoreLogic in a study published last March (free registration required). Using end of 2006 valuations, his analysis pointed up the extent of negative equity in gorier detail than the Pomboy analysis. And while his analysis found lower negative ARM equity than she did as of year end 2006 (see page 6) of 6.9%, versus her 17.0% (as of “a year ago”), he found it to be more sensitive to price declines than she did. But this analysis was buried in the middle of a very long paper with lots of tables and charts.
Why such a big difference between the two estimates? Cagan added the second database to make the mix of properties more representative of the country as a whole. The initial database was heavily weighted towards coastal states. He also applied a proprietary valuation methodology:
The value of each property was computed as of the end of 2006 using not one but two proprietary computer models (automated valuation models or AVMs) which built upon the prices of comparable properties sold in a neighborhood and made adjustments for feature differences, location value, and time trends to estimate the valuation of each subject property. A proprietary algorithm was used to assign weights and priorities to the two valuations of each property in generating a composite estimate of its market value. To compute mortgage debt, the amounts of the current first and second mortgages on each property at the time of their origination were included in this database. To take the side of conservatism, it was assumed that all home equity lines of credit (HELOCs) were drawn to their maximum limit, and that no paying down of principal was made on any first or second mortgages since they were recorded. The equity in each property was defined as the current market value, minus the total first and second mortgage debt at origination. The resulting large valuation database was then analyzed to identify any potentially vulnerable properties.
Now as anybody who has worked with models knows, one little assumption can make a big difference in outcomes, so without getting into the guts of his model, it’s hard to assess the merits of his analysis versus Pomboy’s. But he clearly made an effort to be thorough.
But the specious part of Pomboy’s approach is that she assumes that all homes with negative equity will go into foreclosure. That’s wrongheaded (and the fact that she made such a fundamental mistake makes me skeptical of the rest of her analysis). New York went through a nasty housing decline in the 1990-1991 recession, and plenty of people with negative equity didn’t bail. If you could afford the payments, you stayed put and waited out the bear market. No one wants a mortgage default on their credit record if they can at all avoid it.
As a result of understanding this dynamic, Cagan’s conclusions weren’t as dire as one might expect. His analysis went on the premise that a homeowner had to have both equity stress and payment stress to be at risk of foreclosure, which meant either a high level of interest payments or a significant reset (note sets tiers, for example deeming those who experience a less than 25% increase but have inadequate equity to be at 10% risk of default, while a 25% to 50% increase is assumed to lead to a 40% probability of default if the property lacks enough equity to support a refinance or sale). He broke out resets by year, by interest rate (initial and post reset), by equity level, etc. It’s quite an impressive piece of analysis, and I wish someone other than Cagan had run sensitivities on it (I wrote a long post disagreeing with some of his assumptions) So the negative equity story, while grim, is only one piece in a more complicated equation.
I’ll give you Barron’s first, then the relevant bits from the American CoreLogic report, dated March 19, 2007. From Barron’s:
After modest reflection, any disinterested observer can’t help but find that accompanying table quite alARMing. It’s from a recent MacroMavens report, the handiwork of the incomparable Stephanie Pomboy, whose rants and raves we’ve had the pleasure of occasionally sharing with you. What its blood-curdling numbers depict is that the woes of mortgage lenders are not, as so widely believed, confined to the beleaguered subprime contingent, but are casting a much larger and chillier shadow.
More specifically, the table shows all too clearly that an astounding percentage of adjustable-rate mortgages already are underwater, and it estimates how much equity would be wiped out if home values decline by 5%, 10% and 15% and translates the corresponding losses into dollars.
As Stephanie comments: “Based on the share of ARMs in some state of negative equity at the end of last year and the decline in home prices so far in 2007, a stunning $693 billion in mortgage loans are already in the red. Assuming lenders are able to recover 70% of those assets — which seems optimistic given the massive amount of housing inventory yet to be unwound — that means mortgage lenders are already grappling with $210 billion in outright losses.”
And that, she points out, is merely the direct hit. Thanks to what she nicely dubs the “divine miracle of leverage,” the total financial exposure to these claims is many multiples of that. To which we say, ugh!
What’s more, Stephanie notes, these horrendous losses are coming at a time when the financial sector is “uniquely unprepared to withstand them.” Commercial banks, she points out, have let their loan-loss provisions sink to 20-year lows while increasing their exposure to real estate to record highs. Mortgages, she reckons, account for a tidy 55% of total bank loans — and that doesn’t include the trillion dollars worth of mortgage-backed securities on bank balance sheets.
So much for the myth that banks have cleverly “offloaded” their real estate risk.
Now to the American CoreLogic report (page 55 for those who want to see it in context). This may be hard to read because I haven’t yet figured a way to turn charts within pdf documents into cropped images. Apologies. So I’ll give you the text first and the awkward chart second:
These equity distributions were computed from valuations as of December 2006. Table 29 shows the important impact of future price changes (for all years of first reset).
In much of that table, there is an approximate two for one correspondence between price changes and equity percentages. For instance, a value decline of 5% lowers about 10% of many reset-year properties into negative equity: