President, SMR Research Corp.
A decade has passed since the housing crisis began in 2007. And, thankfully, foreclosures have fallen dramatically.
But does this mean that housing is inured to future crises? Is another problem period coming, and if so, when and why?
This paper will make two key points:
● Another round of home price depreciations now looks likely in metro areas comprising about one-third of the U.S. population.
This would be a lesser problem for credit than we saw in the 2007-2011 period – but still a potentially serious dilemma.
● Credit risk in the pool of mortgage borrowers is lower than it was in the “bad old days.” But perhaps not as much lower as some people think.
The second point about credit risk is easier to address, so let’s look at that first.
Credit Risk In The Mortgage Pool
Positive Changes In Credit Quality
In many ways, the degree of risk in the pool of all borrowers has declined from the period in 2004-2006. Then, risk was high and getting higher.
But since 2008, very few loans have been made with negative amortization features – or with interest-only payments that later reset and require large, sudden increases in monthly payment amounts.
At the same time, stated-income loans (with unverified borrower incomes) have decreased. And, loans made to borrowers with extremely low credit scores are largely gone.
Moreover, risk-layering is no longer taking place much. This practice, halted by regulators in late 2006, referred to loans produced to borrowers who had multiple high-risk features, such as low credit scores and high leverage.
All these changes have been good for credit quality, and nearly everyone is aware of them.
But maybe, folks are too aware of all this. A feeling of comfort can make you stop thinking about whether any alternative problems exist.
Have subprime-credit borrowers really disappeared? Or have they simply morphed into a new style of borrowing?
To be sure, there are no pure-play subprime lenders left. The old ones are all bankrupt or otherwise gone: Ameriquest, New Century Financial, Household Finance, Accredited Home Lenders, IndyMac, and dozens more.
But, government-sponsored loans through the FHA and VA have dramatically increased. The vast majority are FHA.
The typical FHA borrower has a higher credit score today than the typical subprime borrower did yesterday. On the other hand, FHA specializes in handling borrowers with lower-than-normal incomes – and lower-than-normal cash on hand.
Using our own property database, we can see that the average FHA customer borrows, on average, 96.6% of a home’s purchase price. VA customers are even higher, at 98.9% loan-to-value on a home purchase. Both are a far cry from the traditional 80% leverage that conventional borrowers are encouraged to have.
And, the numbers of these borrowers do seem to suggest that the subprime borrower of older times may have simply shifted to FHA loans.
Back in 2003, as the housing boom was heating up, there were 763,638 FHA/VA loans produced to finance home purchases, according to the government’s annual HMDA database. These were only 13.7% of all home purchase loans produced in that year.
Fast-forward to 2009, in the depths of the housing bust and with subprime lending supposedly curtailed.
FHA/VA loans originated for home purchase in 2009 were 1,323,746, or 47.5% of all home purchase loans made that year. So, in a very bad year for home buying, government-insured loans increased by 73%, and their share of the purchase market more than tripled.
Today, we are well past the crisis period – yet FHA continues to roll. In 2014, FHA/VA insured 1,018,631 purchase loans, or 31.5% of all purchase loans. In 2015, FHA/VA again insured 1,251,839 home purchase loans, or 34.2% of all of them.
So, it looks like FHA’s share of market has stabilized around one-third of all home purchase loans. And that share turns out to be an important number.
In the bad old days, subprime conventional loans used to run around 25% of all loans. If you summed the old subprime and FHA/VA loans back then, you’d end out at about the same 33% share that FHA/VA has right now.
Has FHA become the new subprime? Not exactly, but sort of. The FHA credit scores are often higher. But the incomes and cash assets of the FHA borrowers are very weak, and borrower leverage is very high.
To fight the Great Recession, the Fed dropped short-term interest rates to near-zero. And, in an unprecedented move, the Fed has kept rates incredibly low for eight subsequent years.
If you were a net saver with money to invest, what did you do over all these years? Bank deposits and CDs have yielded virtually no interest income.
So, you put cash in the stock market. Anything else? If stocks made you uneasy, you may have considered investing in a house that you could rent out. After all, house prices were very low from 2009 through 2012.
One result of this has been substantial purchasing of rental homes.
The credit problem with these is that they are at least somewhat more likely to default if and when home prices decline. Lenders have always known this and charge higher interest on non-owner-occupied homes. There was a good paper on this subject published in May of 2010 by the Federal Reserve Banks of Richmond and Indianapolis.
In 2009, home lenders financed 284,526 purchases of non-owner-occupied homes. In 2014, they financed 375,111 of these, and in 2015, 400,399. These numbers are gleaned from the HMDA annual databases, which clearly distinguish purchases of primary homes from those intended as either vacation homes or rentals (most are rentals).
At the same time, we’ve seen another recent phenomenon: corporations buying up single-family homes and condos to rent out. One outfit, Invitation Homes LP, now says it owns nearly 50,000 houses for rental purposes. And there are competitors nearly as large.
The debt that these business home buyers take on is not residential mortgage debt, but rather commercial debt. Still, should a downturn take place, one wonders about the inherent risk on these commercial loans.
In all, we recently counted in our own database 21.7 million non-owner-occupied 1-4 unit homes owned by individuals – and another 5.13 million owned by corporations or partnerships.
Virtually all the corporate-owned homes are rentals, and most of the individually owned homes are, too. Of the 21.7 million owned by individuals, 13.3 million were located in the same county that the owners lived in – probably not vacation homes unless you vacation right down the street from where you live.
Inflated Again In Many Places
So, the pool of mortgage debtors today includes a lot of FHA borrowers – and a lot of investors who always pose (on average) higher-than-normal credit risk.
Any other problems? Yes. We are seeing another round of house price bubbles.
The foreclosure debacle of 2007-2011 was by far most explosive in places where home prices crashed the most.
And, the best predictor of foreclosure was not always the borrower’s credit score, but rather his underwater status (mortgage debt exceeding the newly fallen home value). In any new price crash, the highly leveraged FHA/VA borrowers would be at risk. And the increased number of non-owner-occupied home borrowers would not help.
With this as a backdrop, the near-future news, we think, will be: Home prices will indeed decline – not nationally, but in selected markets spanning perhaps one-third of the USA.
In short, this will not be the mega-disaster that we saw last time – but will be a substantial change with foreclosures rising again.
For many months, the National Association of Realtors (NAR) has warned that in many markets, home affordability is far too low.
They are looking at local average incomes as a percent of local average home prices. We tried that method long ago and found the predictive power of that single ratio was poor.
People in California, for instance, have spent a larger-than-normal share of their income on housing for decades, so a simple ratio always made California appear to be on the brink of a price depreciation.
The more predictive way to look at affordability, we found, was to study the degree of change over time in house prices against the degree of change in incomes.
When house prices rise faster than incomes for a long time and to a great degree, our data (going back to the 1970s) shows that a depreciation is near-certain to occur.
To implement our method, you need to index both average home prices and average incomes and then look at the difference in recent indexed values. The Federal Housing Finance Agency (FHFA) already does publish a home price index by metro area, with prices set to Q1 of 1995 = 100. We index our data on localized average incomes to the same 1995=100.
Let’s walk through an example to illustrate how this works: the Miami, FL, metro area.
From 1976 all the way through Q2 of 2001, the two indexes are closely aligned. Basically, every 1% increase in incomes caused a roughly 1% increase in home prices. It made sense. But then, we entered the bubble period.
Here are the numbers:
At the end of year 2000, the Miami FHFA home price index was 128.15 and the income index was a near-identical 128.58. By the end of 2002, the price index was 154.73 and the income index was 136.28.
A gap of 18 points had opened, but that was just a warm-up.
This gap widened to a maximum of 163.54 points by Q1 of 2007. At that time, the HPI was 342.27 while the income index stood at only 178.73.
The bubble had grown larger for five years, and then came the crash.
By Q2 of 2011, the Miami HPI had dropped nearly in half to 180.51, with thousands of foreclosures resulting as homeowners walked away from underwater properties. In that quarter, the income index stood at 184.81, once again near parity with home prices.
But, starting in 2013, the same sort of bubble started building up again. Home prices began rising faster than incomes, leaving a gap in HPI over the income index of 25.84 points by the end of that year.
By the end of 2016, the gap stood at 76.53 points – a level that historically has always resulted in an eventual price depreciation. So, Miami was one of the epicenters of the last mortgage crisis, and it looks like it will be again at some near-future moment.
Here is a graphical representation of the two indexes in Miami:
This method of price depreciation forecasting seems to work almost flawlessly across time and across nearly 400 U.S. metro areas.
In Buffalo, NY, there was no price bubble in the 2000s, and there was no crash. In Denver, the 2000s bubble was modest, and the crash was modest.
Back in the late 1980s, there was a bubble gap that formed in most of California using our method. A price crash in that state followed about three years later. Other historical events of the 1980s and 1990s followed the same pattern.
This all makes sense because of one underlying truism: When home prices exceed what the local population can afford, prices must come down.
In some markets today, prices are up partly because demand exceeds supply (limited homes for sale). But that sort of imbalance is usually temporary.
Even with limited supply, prices cannot keep rising ahead of incomes forever, unless there is a strong and continuing influx of wealthy buyers from outside the metro area. Still, ultimately, their incomes get mixed in with those of the longtime residents, so the tracking of price vs. income changes returns to having predictive power.
Triggers And Timing
Why have we seen a resumption in house price bubbles in markets like Miami?
One reason is the long-extended period of low mortgage interest rates. They have allowed people to buy much more expensive homes than they otherwise could have afforded.
Consider: In late 2006, before the crisis began, the 30-year fixed-rate mortgage was at around 6.3%. Someone able to afford a $1,500 monthly P&I payment would qualify for a mortgage of around $242,000. At 80% LTV, the house price would have been $302,500.
At a more recent 30-year rate of 4%, the same person able to make a $1,500 monthly payment would qualify for a mortgage of $313,000, enabling him at 80% LTV to buy a house for $391,250.
So, super-low rates by themselves have enabled a roughly 29% increase in home prices.
But of course, this also works in reverse. The Fed began raising rates in 2016 and has continued in 2017. So far, as of July 2017, the higher short-term rates have not impacted mortgage rates by much, but eventually they will. Then, we will likely see the downward pressure on home prices get underway.
Yet, until this happens, the game of musical chairs that causes home prices to rise will continue. Perceiving steady price increases, more people buy homes to flip them – and more people rush to buy before prices grow even higher.
This tends to play out the most in expensive housing markets. There, house flipping can yield more cash. And if you just want a place to live, the fear of rising prices is most warranted because they can rise by such large sums.
A 5% rise in prices in Akron, OH, may cause prices to rise by $8,000 or so. A 5% rise in prices in San Jose may equate to $25,000 or more. For this reason, the vast majority of metro areas today with new house price bubbles are the more expensive real estate markets.
Note for a moment how utterly ironic all of this is.
The greatest economic catastrophe in the last 50 years was caused by a giant house price bubble coupled with massive subprime lending. The Federal Reserve is supposed to prevent economic disasters.
But the Fed’s rate policy has now helped created a new set of bubbles (not yet nationally, but regionally). And, Fed officials seemingly haven’t noticed that the growth in FHA lending has created another large group of highly leveraged mortgage borrowers. Their homes will quickly move underwater as soon as any price decline begins.
Where Price Declines
Are Most Likely To Happen
Once you have a predictive set of numbers, you can apply them to all local markets and simply rank-order them.
We do so based on the gap between the HPI and the income indexes. The gap has been predictive in the past, and we think it will be as well in the future.
Click here to see a report covering all major U.S. metro areas as of year-end 2016.
As you will see in that report, a large majority of U.S. cities look OK. Incomes have kept up with home prices, or have even exceeded home price growth. Major cities like Chicago and Atlanta look fine, as do literally all cities in Ohio and many other places.
But most of California looks poised for another price collapse. So does South Florida (but not North Florida), Denver, Portland (OR), Charleston, Seattle, and the counties of Long Island, NY.
The upcoming problem will be serious not so much because of the number of at-risk places, but because they are so populous. They are roughly one-third of the entire U.S. population.
The big question is: When might the bubbles begin to collapse?
No one knows. We do know that the last big house price bubble inflated for about five years before bursting. This one got underway in 2013, so five years would put us at 2018. The regional price bubbles of the late 1980s and early 1990s took only three years to burst, so we are already beyond that timing.
Our best guess on timing is that as soon as mortgage rates begin to rise substantially, the bubble-popping will begin. It could well be that this will start in 2018.
And let us restate, finally, one thing: The foreclosure problem likely to develop when the bubbles burst will not be as bad as it was from 2007-2011. But it probably will be bad enough to make some serious economic waves.