Seeing Through Data

Originally posted Jan. 11, 2012 on nakedcapitalism.com, here http://www.nakedcapitalism.com/2012/01/michael-olenick-10-million-shadow-inventory-says-housing-market-is-a-long-way-from-the-bottom.html

“Shadow inventory,” the number of homes that are either in foreclosure or are likely to end up in foreclosure, creates substantial but hidden pressure on housing prices and potential losses to banks and investors. This is a critical figure for policymakers and financial services industry executives, since if the number is manageable, that means waiting for the market to digest the overhang might not be such a terrible option. But if shadow inventory is large, housing prices have a good bit further to go before they hit bottom, which has dire consequences for communities, homeowners, and the broader economy.

Yet estimates of shadow inventory, and even the definition of what constitutes shadow inventory property, vary widely. For example, the Wall Street Journal published a Nov. 11, 2011 article, “How Many Homes Are In Trouble?” where values varied from 1.6 million (CoreLogic), to “about 3 million” (Barclays Capital), to 4 million (LPS Applied Analytic), to 4.3 million (Capital Economics), to LPS Applied Analytics, to between 8.2 million and 10.3 million (Laurie Goodman, Amherst Securities).

Why do these numbers vary so much? Even though CoreLogic is generally considered to have one of the best databases of loans, its estimates of loan performance and odds of default are based on credit scores, which is a badly lagging indicator. Laurie Goodman is seen by many as having the most carefully though out model, even though industry insiders are keen to attack her bearsish-looking forecast.

I have a large database of my own, and am familiar with housing and mortgage information sources. I’ve come up with my own tally of shadow inventory and have also tried to analyze — OK — take a stab at – what I call “shadow liability,” meaning the amount of money taxpayers, investors, banks, will be lose if those homes are liquidated. Assumptions using those terms are also in the attached spreadsheet. My analysis comes up with a total close to that of Goodman’s range, 9.8 million using a narrower definition than Goodman’s of what constitutes shadow inventory.

Put more simply, things are actually worse than any of the prevailing estimates indicates, although Goodman is very close to the mark. Current loss experience suggests that this figure is staggering, easily in the $1 trillion range.

Why aren’t those losses more visible yet? Well, evidence suggests that servicers are stalling the foreclosure process, not taking title to and selling these houses. For the lenders, such delay likely allows them avoid the write-offs of both the negative equity as well as the worthless second liens. More generally, it keeps the trillion dollar losses hidden. Lenders aren’t acknowledging their stall tactics, however. When people notice how slowly foreclosures are progressing from initial steps to resale, lenders point at their foreclosure fraud related dysfunction. Lenders conveniently don’t mention that such dysfunction was self-induced, instead blaming borrowers and courts.

My Methodology

My data comes from several sources. Default information is from the October, 2011 LPS Mortgage Monitor. Housing information, including the number of houses with mortgages, comes from the US 2010 US Census and the 2009 Statistical Abstract. Median home prices — the likely value of the loans that are either in foreclosure or will be soon, is from the FHFA; specifically the Q2, 2006 state-by-state median home prices, when many of the bubble loans were written. Note: these prices are used to approximate the principal value of the loans, not what the properties are currently worth.

Because not all this data overlaps entirely some extrapolation was necessary; when required to extrapolate I tried to do so conservatively. An example is how I arrived at the number of mortgages in the US, a step on the way to calculating the number of mortgages in default.

The first step was figuring out how many housing units with residential mortgages America has. According to the 2010 census, America is home to 131.7 million housing units. Of these, 76.4 million are owner-occupied, 37.5 million are rental units, and the remaining 17.2 million are vacant, and the remaining 600K are houseboats or other exotic housing. Of the 37.5 million rentals, some are in apartment buildings that would be financed with commercial mortgages, not residential ones. Commercial loans are structured differently than residential loans, and are easier to renegotiate, so they I’ve excluded from this analysis.

To be conservative—to exclude more loans as commercial than actually are, rather than risk leaving commercial loans in the analysis—I’ve assumed that any building with 5 or more housing units is in a building that either has a commercial loan or no mortgage at all.

According to the National Multi-Housing Council, using 2011 Census data, has determined that nationally, 42% of renters live in buildings with 5 or more units. Applying that percentage to the 37.5 million rental units, and subtracting that from total renters, I end up with 21.8 million rental housing units that could have residential mortgages.

In total, then, I have 76.4 million owner-occupied homes, 21.8 million residential rental units, 17.2 million vacant homes (which includes, among other things, vacation homes and abandoned ones) and 16.3 million other, mainly units in commercial properties. All in all I end up with just over 115 million homes that could have a residential mortgage on them. But how many of them? Well, the Census reports that in 2010, 68% of owner-occupied units had at least one mortgage. I used this same 68% for investment (residential rental) properties and vacant (primarily vacation and abandoned single family homes) properties.

I believe this 68% figure is appropriate for two reasons. First, a person who has a mortgage on their own home is unlike to buy a vacation house or an investment property with cash. Indeed, even a homeowner living free and clear in their own home might need a mortgage to buy second property. So assuming the mortgage rate for investment and vacation homes is the same as owner occupied surely understates the number of mortgages. Second, the mortgage rate on abandoned homes surely is nearly 100%; why abandon a home if it’s not in foreclosure?

Using that math, I came up with 78.6 million mortgaged properties. This figure is substantially higher than many other estimates, including Goodman’s Amherst study, though the likely reason is that the census data the analysis relies upon is relatively new. Goodman’s study uses 53.7 million mortgaged homes, though the census reports 52.2 million owner-occupied loans alone, in additional to rental properties, mobile homes, and vacant properties. Given that the census cost $13 billion to produce — an amount no private organization could afford — and 2010 results were not available at this level of granularity until relatively recently, I would not be surprised to see upward revisions to other base housing unit figures in the future.

To estimate shadow inventory, I used the delinquency data from LPS Analytics. They add up loans that are delinquent, loans that are in foreclosure, then come up with a state-by-state percentage of “non-current” — loans that are, or are likely, to end up in foreclosure. There is some ambiguity in LPS’ figures; specifically the definition of “delinquent,” and whether they are counting homes or loans.

To illustrate a potential problem with these assumptions, let’s take a theoretical example of 100 houses. Let’s assume 68% have mortgages, a figure from the census, so 68 houses have mortgages. Then let’s assume these homes are in FL and 22.9%, or 23 houses, are either in foreclosure or likely to end up there soon. I’m assuming this means that 45 houses are current, 23 houses in trouble, and 32 houses paid-off, though I concede that it could mean 12 houses with two mortgages are in trouble, 32 are paid off, and 56 are fine.

This methodology differs from Goodman’s, which relies upon predicting both likely defaults and re-defaults for non-sustainable modifications, as well as a small number of homes likely to strategically default as liquidations begin and home value plummets. Conversely, my model assumes all 90-day delinquent loans will result in foreclosure and liquidation — and I’ve yet to see enough good-faith modifications to assume otherwise — whereas Amherst’s believes the figure is likely to be 80-90%. However, I do not allow for strategic defaults, which more than offsets my skeptical assessment of the mortgage mods now begin offered (my assumption is that when people default suddenly, it is really an anticipatory default: the borrower could see he was going to hit the wall, but defaulted before he was completely broke. Given the job market costs of having a foreclosure or bankruptcy on your credit record, I don’t regard that as a bad assumption). Goodman has three buckets of current loans that she anticipates will produce defaults: badly underwater loans (loan to value ratios of over 120%), moderately underwater loans (LTVs of 100% to 120%) and loans with equity borrowers will default upon anyway (LTVs less than 100%). She estimated those three groups would produce eventual foreclosures of 2.8 to 3.7 million of her total. Thus my somewhat smaller tally is actually more dire, because it consists of borrowers who are having trouble making payments now, as opposed to borrowers who are anticipated to default at some undetermined point in the future.

That being said, except for the lower housing unit loan base Goodman’s analysis seems rock-solid, though it would mushroom if used with my higher base housing unit figures and more pessimistic view of servicer’s ability to mitigate defaults. Together they would paint a devastating picture of the future, so I won’t try to reconcile them .. at least not yet.

Using the assumptions above, and applying the LPS data state-by-state, there are 9,800,000 houses in shadow inventory.

If these loans were taken out for the median value of a state-by-state home price, using data from the FHFA, for Q2, 2006, there is $2.3 trillion of home values at near the market peak. The mortgage balances are going to be lower than that, but given how widespread equity extraction came to be (and it is probably that the most levered homes are hitting the wall), it is not unreasonable to assume LTV ratios relative to peak values of 80%. Loss severities on prime mortgages are running at roughly 50% and are 70% on subprime (note that with more borrowers fighting foreclosures, and given that loss severities on a contested foreclosure can come in at 200% or even higher, so using these assumption is certain to understate actual results). $2.3 trillion x 80% x 50% = $900 billion.

These losses will be distributed across the GSEs (meaning taxpayers), banks that have second liens (with the biggest losers being Bank of America, Citibank, JP Morgan, and Wells Fargo), investors in private label (non GSE) mortgage securities, and other US and foreign banks. Balanced against this liability is some amount figure for the underlying asset, the house. Given that servicer advances, foreclosure costs and servicer fees come close to and even exceed the value of the property, comparatively little of this $2.3 trillion will be recovered in property liquidations.

It is unclear where the money from these write-offs will come from, or whether they losses have been adequately budgeted. Obvious sources are Fannie Mae, Freddie Mac, European and US banks, none of which have reported anywhere near this level of reserves. We know that the Federal Reserve has been buying up MBS and related instruments in bulk; maybe the central bank plans to print more money to cover the losses and enable the foreclosures. Printing this much money, for this purpose, in this political environment, in secret, seems unlikely.

In support of the conclusion that banks cannot afford to recognize this shadow liability is the sharp decrease of foreclosure filings in 2011 and the seeming unwillingness of banks to move foreclosures through the system. They file foreclosures, then let them linger, not taking homes even when every possible borrower defense is exhausted. Some of this slowdown may be due to more scrutiny of foreclosure documentation, particularly in judicial foreclosure states, but there is clearly more at work. In the most obvious example, servicers are reluctance of banks to take title to the homes after obtaining a judgment; even after the judgment is a year old and cannot be challenged.

For example, filing volume in Palm Beach County, FL, started to increase towards the end of 2010 but judgments remained flat and certificates of title — where a bank actually takes title to a house, recognizing the underlying financial loss and evicting the family — actually slowed down despite an enormous backlog of judgments. This contrasts to the banks incessant complaints of a broken court system, because a judgment more than one year old in FL cannot be challenged for fraud. This leads to the conclusion that it is the banks — who are unwilling or unable to absorb the losses — rather than the courts or homeowners that are actually slowing down liquidations.

Let’s walk through these figures. In Palm Beach County, the number of Certificates of Title issued for Q1-4, 2011, was 1,594, 1,886, 1,413, and 1,299 respectively; the number of judgments was 289, 480, 281, and 367 respectively. Let’s compare that to 2010, when there were 3,105, 9,704, 7,259, 1,033 judgments in Q1-4 respectively and 1,534, 2,207, 3,065, and 2,738 titles transferred.

Many of these cases are uncontested; yet it is not uncommon in foreclosure court to see bank lawyers arguing vehemently for delays with nobody on the other side.

Let’s review more figures: in Palm Beach County there are 10,794 more final judgments of foreclosure that are at least a year old than there are certificates of title issued. Again, there is nothing anybody can do to challenge a judgment after one year. Servicers appear to be milking ongoing costs and fees from investors. Cross-referencing that to a softer data point I’m reminded of a worker, in my home state of FL, sent by a company I hired to perform a home repair. He’s a young man who said he purchased a condo, lost a prior job that paid better, and stopped paying for his condo for which, he noted, similar models were selling for at a 80% discount to what he owed. He filed no defense to his foreclosure whatsoever — he was positively clueless about the judicial system and did not hire a lawyer — but he ran to his truck to show me a Notice of Voluntary Dismissal of his foreclosure, asking what it meant. It’s clear that while some homeowners do their best to avoid the auction block, even those who do nothing all have a statistically good chance of staying put.

There is other anecdotal evidence suggesting banks do not want these houses or, more accurately, do not want the write-offs that actually taking the houses would force:

• Foreclosure defense lawyers have clients who have not paid their mortgage in years, but face neither a foreclosure nor even a negative mark on their credit report. I recently received a call from a man who said he had not paid his $1.6 million mortgage in two years but his servicer has not foreclosed, and he faces no derogatory information on his credit report; he was frustrated because he is retired and just wants to move to a cottage. This phenomenon, which apparently isn’t rare, might explain why shadow inventory reports that rely on credit reports to extrapolate shadow inventory are often dramatically lower than these calculations.
• Every year the Republican dominated Florida legislature introduces legislation to speed along foreclosures, and every year the legislation fails. I personally believe this legislation to be both immoral and arguably illegal. However, it is impossible to believe this bank beholden governmental body is willing to repeatedly bite the hand that feeds them .. unless their master makes it quietly clear that they do not actually wish to accelerate liquidations but cannot publicly admit as much.
• It is common for foreclosure mill lawyers to argue for delays in selling a home when nobody is representing a borrower. Judges, who want to clear their dockets, will rail at bank lawyers about the age of the case even while bank lawyers argue for yet another delay, while the other table — where the borrower, the defendant, is supposed to sit — is empty.
• Bank-instituted delay tactics are not limited to Florida. Not long ago I spent the day with Sean O’Toole, CEO of foreclosureradar.com. Sean knows the foreclosure world and his data is, literally, the best in the Western states he covers. He noted the same effect in CA; lender-initiated delay after delay after delay selling a home. In CA, after three delays both parties must approve a further delay but Sean said banks routinely file stipulated delays when, in fact, borrowers just want to literally move on.
• There is the well-known tendency of servicers to “lose” paperwork, where borrowers beg for mortgage modifications, short-sales, or deeds-in-lieu. These delay tactics — rather than just answering “no” to a request — make sense in this context because leaving a house in foreclosure limbo, forever, is the only solution that delays the inevitable balance sheet busting write-offs.
• Lastly is the unwillingness of banks to agree to principal reductions, or even modifications with principal balloon payments, which would yield more long-term money than a foreclosure. Servicers appear to want these homes in the higher-yielding default status, even if they are reluctant to actually push the homes to liquidation, to take title on behalf of investors.

We’ve written relentlessly about servicer abuses, but we’ve almost always contextualized these abuses through their effect on borrowers. Staring through data, especially data at this scale, complexity, and with strong economic ramifications, is like looking through a dirty window. But as we wipe away layer after layer of schmutz the picture is becoming clearer. Yes, servicers continue to prey upon ordinary Americans. But evidence suggests that they’re also preying on investors. Individual American families do not deserve to suffer these behaviors, that increase the losses while delaying the uncertainty, and neither do pension funds, European villages, municipalities, or other unsuspecting entities who actually funded these loans.

Few people are going to complain when they’re not paying their mortgage that there is no mark on their credit-report nor a foreclosure; a few of the more perplexed ones — or those that want to bring a bad mortgage to resolution — may speak out, but most remain silent.

Similarly, many investors, and surely the banks themselves, know about these figures. But as both sides spin their wheels, the problem continues to spiral out of control.

Finally, there is government behavior that makes no sense, especially from the Obama Administration. We have repeatedly seen federal intervention when it is inappropriate and unwelcome, and we’ve seen no intervention when it is warranted. For example, the Administration has actively intervened in the multistate Attorney General settlement talks even though this is, by definition, a state issue. However, they have done nothing to prosecute overt and clearly proven interstate crimes surrounding document forgery.

There is a strong argument that campaign donations are at work, but given the lopsided donations from the financial services industry to Republicans one would think Obama would send a message by taking firm control over the FHFA, the FDIC, the SEC, the OCC, the Treasury, the Justice Department, and strong-arming the Federal Reserve into offering substantive help to borrowers and investors. Yet, at every level, the President has failed ordinary Americans. Even the most egregious behavior results in dead silence .. we don’t even get a yawn. Every program has been an unmitigated disaster, especially HAMP. When Administration figures do intervene their influence is overtly skewed in favor of the banks.

Surely Obama and his advisers realize these problems. It seems inevitable that we will soon face either widespread bank failures and a staggering loss in home values (although arguably an increase in middle-class liquidity), or another much larger bailout; a fraud bailout. Either option is likely to sink President Obama’s popularity rating in much the same way it is likely to sink individual home values. Despite this, the president continues to play Kick the Can, presumably hoping these problems won’t be widely recognized prior to the election in November, while the banks continue to kick everybody else.

Market manipulation used to be illegal, especially in cases where there was asymmetrical information or unequal bargaining power. Pundits use the term “heads we win, tails you lose,” but that actually understates the problem because it implies that there still exists individual parties and counter-parties. Our more modern arrangement looks more like an aristocracy, where there isn’t a genuine market at all but rather a pseudo-market operating like a private ancient tax collector, demanding the increasingly poor peasants feed the monarchs and his cronies rather than feeding their own children.

I’m often told that people don’t care about deadbeats who haven’t paid their mortgages. But people fail to realize that this affects everybody. Ordinary Americans see the effects of this manipulation every day; it affects them profoundly, even if they don’t understand it. All but the most irresponsible aristocrats throughout history realized there were boundaries. Their motivations may have differed — some cared about the well-being of the peasantry while others feared the guillotine — but for millennia all but the stupidest acknowledged and avoided pushing the populace too far. If we’re going to live under an American Nuevo-Feudal system, the least we deserve are overlords at least as smart as the despots they’re trying to imitate.