by Calculated Risk on 10/28/2008 10:00:00 AM

This morning the Census Bureau reported the homeownership and vacancy rates for Q3 2008. Here are a few graphs and some analysis ...

Click on graph for larger image in new window.

The homeownership rate decreased slightly to 67.9% and is now back to the levels of the summer of 2001. Note: graph starts at 60% to better show the change.

Here is an excerpt from a piece I wrote earlier this year on the impact of the change in homeownership rate (with a hat tip to Jan Hatzius):

As the graph shows, the homeownership rate increased from 64% in 1994 to 69% in 2004, or about 0.5% per year. With 110 million total households in the U.S., this change in the homeownership rate would mean an increase of about 550 thousand new homeowners per year during that period – even with no population growth.

The U.S. population has been growing just under 3 million people per year on average, and there are about 2.4 people per household. Assuming no change in these numbers, there would be close to 1.25 million new households formed per year in the U.S.

Since about two thirds of all households are owner occupied, an increase of 1.25 million households per year would imply an increase in homes owned of about 800K+ per year.

So we could add the 550K from the increasing homeownership rate, to the 800K due to the increase in households (due to population growth), and the U.S. would have needed 1.35 million additional owner occupied homes per year during the period from 1995 to 2004. If the homeownership rate now stabilized, the U.S. would only need 800K additional units per year.

And if the homeownership declined – as it has been for the last 2+ years – at a rate of around 0.5% per year, the U.S. would need 800K minus 550K new units per year, or only 350K additional owner occupied units per year!

This number can't be compared directly to the Census Bureau housing starts and new home sales. There are many other factors that must be accounted for, but this does show the homebuilders had a tailwind behind them for a decade, and are now flying into a headwind.

Even when the homeownership rate stabilizes, the U.S. would only need 800K new owner occupied homes per year – far below the level of 1995 to 2004.

This means the builders have two problems over the next few years: 1) There is too much inventory, and 2) demand will be significantly lower over the next few years than from 1995 to 2004.

Why did the homeownership rate increase?

A 1007 research paper by Matthew Chambers, Carlos Garriga, and Don E. Schlagenhauf (Sep 2007), "Accounting for Changes in the Homeownership Rate", Federal Reserve Bank of Atlanta, suggests that there were two main factors for the increase in homeownership rate between 1994 and 2004: 1) mortgage innovation, and 2) demographic factors (a larger percentage of older people own homes, and America is aging). The authors found that mortgage innovation accounted for between 56 and 70 percent of the recent increase in homeownership rate, and that demographic factors accounted for 16 to 31 percent. Even as we unwind some of the excesses of recent years, not all innovation is going away (securitization and some smaller down payment programs will stay). And the population is still aging, so the homeownership rate will probably only decline to 66% or 67%, not all the way to 64%.

In summary: For as long as the homeownership rate declines – probably for at least another couple of years - this means the need for new owner occupied units will stay depressed, and even when the homeownership rate stabilizes and the inventory is reduced, demand will only be about 2/3 of the 1995-2004 period.

The second graph shows the homeowner vacancy rate since 1956. The homeownership vacancy rate was steady at 2.8% (down from a record 2.9% in Q1).

A normal rate for recent years appears to be about 1.7%. There is some noise in the series, quarter to quarter, so perhaps the vacancy rate has stabilized in the 2.7% to 2.9% range.

This leaves the homeowner vacancy rate almost 1.1% above normal, and with approximately 75 million homeowner occupied homes; this gives about 825 thousand excess vacant homes.

The rental vacancy rate decreased slightly to 9.9% in Q3 2008, from 10.0% in Q2. The rental vacancy rate had been flat or trending down slightly for almost 3 years (with some noise).

It's hard to define a "normal" rental vacancy rate based on the historical series, but we can probably expect the rate to trend back towards 8%. According to the Census Bureau there are close to 40 million rental units in the U.S. If the rental vacancy rate declined from 9.9% to 8%, there would be 1.9% X 40 million units or about 760,000 units absorbed.

This would suggest there are about 760 thousand excess rental units in the U.S.

There are also approximately 200 thousand excess new homes above the normal inventory level (for home builders) - plus some uncounted condos.

If we add this up, 760 thousand excess rental units, 825 thousand excess vacant homes, and 200 thousand excess new home inventory, this gives about 1.8 million excess housing units in the U.S. that need to be absorbed over the next few years. (Note: this data is noisy, so it's hard to compare numbers quarter to quarter, but this is probably a reasonable approximation).

These excess units will keep pressure on housing starts and prices for some time.

by Calculated Risk on 10/28/2008 09:12:00 AM

S&P/Case-Shiller released their monthly Home Price Indices for August this morning. This includes prices for 20 individual cities, and two composite indices (10 cities and 20 cities). Note: This is not the quarterly national house price index.

Click on graph for larger image in new window.

The first graph shows the nominal Composite 10 and Composite 20 indices (the Composite 20 was started in January 2000).

The Composite 10 index is off 22.0% from the peak.

The Composite 20 index is off 20.3% from the peak.

Prices are still falling, and will probably continue to fall for some time.

The second graph shows the Year over year change in both indices.

The Composite 10 is off 17.7% over the last year.

The Composite 20 is off 16.6% over the last year.

The following graph shows the price declines from the peak for each city included in S&P/Case-Shiller indices.

In Phoenix and Las Vegas, home prices have declined about 36% from the peak. At the other end of the spectrum, prices in Charlotte and Dallas are only off about 3% from the peak.

This shows the difference between the bubble areas (Krugman in 2005 called the "Zoned Zones") and the "Flatland" areas.

There was a bubble in Flatland too caused by the rapid migration from renting to buying - facilitated by loose lending - that pushed up Flatland prices. But those bubbles were small compared to the bubbles in the Zoned Zones.

Now that the bubble has burst, prices in the more bubbly Zoned Zones are falling much more than in Flatland.

Detroit is an exception with prices off 27% from the peak, even though Detroit never had a price bubble. The reason is Detroit has a weak economy and a declining population. Since housing is very durable, there is excess supply in Detroit, and prices for existing homes are below replacement costs.

Another exception is New York. Prices in New York are only off 10.6% even though New York is part of the Zoned Zone. New York had a price bubble, but until recently prices had held up pretty well.

by Calculated Risk on 10/28/2008 09:10:00 AM

From Rex Nutting at MarketWatch: Home prices off record 16.6% in past year, Case-Shiller says

Home prices in 20 major U.S. cities dropped 1% in August compared with July and had fallen a record 16.6% from the previous year ... Prices have fallen in all 20 cities compared with a year ago.

More - plus graphs - as soon as the data is available online.

by Calculated Risk on 10/28/2008 12:02:00 AM

The 8-K filed by First Fed with the SEC today has some interesting information on current LTVs. (hat tip Brian)

Click on table for larger image in new window.

This table shows the original LTV of First Fed's $4.5 billion loan portfolio, and the current LTV using OFHEO House Price Index for price declines.

This shows that about 22.3% of First Fed loans (by dollar) are underwater. It would be a larger percentage using the Case-Shiller price index.

Approximately $1.0 billion in loans are underwater out of $4.5 billion in loans.

Using the above table, and the delinquent loans by LTV on page 8, we can determine the percent delinquent by LTV category.

As expected, the higher the current LTV, the larger the percentage of delinquent loans. Probably most of the loans listed as 90% to 100% LTV are also currently underwater too since First Fed uses OFHEO (and Case-Shiller is probably worse and I believe more representative of actual price changes).

Also see the bottom of page 7 for delinquencies by borrower documentation type. For Verified Income/Verified Assets loans, 5.7% of loans are delinquent. For Stated Income (and no income) loans, around 20% of loans are delinquent. Not exactly a surprise ...

This is important for the entire industry: the higher the LTV, the higher the delinquency rate. As house prices continue to fall, and more and more households have negative equity - Moody's Economy.com estimates 12 million households currently are underwater, and this will probably increase to 20+ million by the time housing prices bottom - the delinquency rate (and foreclosures) will continue to increase.

by Calculated Risk on 10/27/2008 08:22:00 PM

From The Times: Banks may need further support from taxpayers as recession bites

Britain's banks may need to raise capital above and beyond the £50 billion of taxpayer-underwritten money already earmarked for them.

The Bank of England's report into financial stability today suggests that a recession as severe as that of the early 1990s would lead to credit losses of £130 billion for Britain's six biggest financial institutions and possibly wipe out the entire government-backed funding package.

The U.S. GDP is about five times the U.K, and that would suggest the eventual cost of the U.S. bailout might be over $1 trillion.

Note: £130 billion X 5 / exchange rate 0.646329 = $1 trillion.

That is less than the number Professor Krugman speculated about over the weekend:

Do the math ... these numbers seem to suggest that an eventual outlay of $2 trillion is in the realm of possibility.

But more might be required.

by Calculated Risk on 10/27/2008 05:31:00 PM

From the WSJ: GM May Get Loan for Chrysler Deal

The Department of Energy is working to release $5 billion in loans to General Motors Corp. ... The funds would come from a pool of $25 billion in low-interest loans approved by Congress to help Detroit retool its plants to meet new fuel-efficiency standards. It's not clear how quickly the money could be made available or whether it would come with strings attached.

And from Reuters: US Treasury working on aid for GM, Chrysler merger

The U.S. government is considering direct financial assistance to facilitate a possible merger between General Motors Corp and Chrysler ... The Treasury Department is weighing aid of at least $5 billion, which could include capital injections and government purchases of bad auto loans ...

It looks more and more likely that GM will acquire Chrysler.

Also of interest, just last week Daimler wrote the value of their Chrysler holdings down to zero according to a report in the Free Press: Daimler: Chrysler worth nothing

The German automaker has depreciated its stake in Chrysler to zero from $268 million at the end of June, the company said Thursday. A little over a year ago, the company valued its 19.9% stake in Chrysler at $2.2 billion.

by Calculated Risk on 10/27/2008 03:44:00 PM

From MarketWatch: Wal-Mart U.S. to add remodels, trim new store growth

Capital spending for Wal-Mart U.S. is expected to decline to $5.8 billion to $6.4 billion for fiscal 2009 from $9.1 billion last year. For fiscal 2010, capital spending is pegged at $6.3 billion to $6.8 billion ...

Just more evidence of the imminent non-residential construction downturn.

by Calculated Risk on 10/27/2008 01:17:00 PM

New home sales in 2008 might be at the lowest level since 1982. However adjusted for changes in owner occupied units, the current year is the worst on record.

The following graph shows both annual new home sales (from the Census Bureau) and sales through September.

Click on graph for larger image in new window.

In 2008, sales through September (before revisions) have totaled 402 thousand. This is slightly ahead of the pace in 1991 (391 thousand sales through September).

However sales have slowed in the 2nd half of 2008, and it appears that annual sales might be below the 509 thousand in 1991. It will probably be close, but if sales are below the 1991 level, this would mean sales would be the lowest since 1982 (412 thousand).

Of course the U.S. population and the number of households were much lower in 1982. In 1982 there were 54.2 million owner occupied units in the U.S., in 1991 there were 61.0 million, and there are approximately 76 million today.

If we use a ratio of owner occupied units to compare periods, the low in 1982 was 412 thousand X (76/54.2) = 578 thousand units (based on the number of owner occupied units today).

The calculation for 1991 gives 634 thousand units (to compare to today).

By this measure, 2008 is the worst year for new home sales since the Census Bureau started tracking new home sales (starting in 1963).

by Calculated Risk on 10/27/2008 11:37:00 AM

I spoke with a senior manager at a public company this morning, and his company has just received loan commitments from two major lenders for an acquisition. Both lenders are on the list of banks receiving capital from the Treasury. This is a marginal credit risk deal, so I consider this a positive sign. The deal isn't done, but this is definite progress. (Note: the company is publicly traded so I can't reveal any details).

LIBOR: From the WSJ: Credit Gauges Little Changed

[LIBOR was] 3.5075%, according to Monday's daily Libor fixing by the British Bankers Association. That's down from 3.51625% Friday ...

The yield on 3 month treasuries: 0.82% up slightly from 0.80% (unchanged)

The Fed is expected to lower rates this week by anywhere from 25 bps to even 75 bps, but I'd still like to see the three month treasury closer to 1.0% (or whatever the Fed Funds rate is this week). The effective Fed Funds rate is about 0.93%, so the three month yield is still a little low.

The TED spread: 2.76 up slightly from 2.70 (slightly worse) This is still way too high. I'd like to see the spread move back down to 1.0 or lower - at least below 2.0.

The two year swap spread from Bloomberg: 117.75 down from 125.02 (better). I'd like to see this under 100.

Activity in the Treasury's Supplementary Financing Program (SFP). This is the Treasury program to raise cash for the Fed's liquidity initiatives. If this program slows down borrowing, I think that would be a good sign.

Here is a list of SFP sales. The Treasury announced another $40 Billion for the Fed this morning - no progress.

The A2P2 spread is 4.32, down from 4.48. better.

During a recession, this spread usually increases because the risk of default for lower quality paper increases. However the recent values (over 400 bps) are far in excess of normal. If the credit crisis eases, I'd expect a significant decline in this spread.

The progress is slow, but this is a positive day in the credit markets.

by Calculated Risk on 10/27/2008 10:00:00 AM

According to the Census Bureau report, New Home Sales in September were at a seasonally adjusted annual rate of 464 thousand. Sales for August were revised down slightly to 452 thousand.

Note that the most recent wave of the credit crisis started in mid-September. Since New Home sales are reported when the contract is signed, September sales were only partially impacted by the credit crisis.

Click on graph for larger image in new window.

The first graph shows monthly new home sales (NSA - Not Seasonally Adjusted).

Notice the Red columns for 2008. This is the lowest sales for September since 1981. (NSA, 36 thousand new homes were sold in September 2008, 28 thousand were sold in September 1981).

As the graph indicates, sales in 2008 are substantially worse than the previous years.

The second graph shows New Home Sales vs. recessions for the last 45 years. New Home sales have fallen off a cliff.

Sales of new one-family houses in September 2008 were at a seasonally adjusted annual rate of 464,000, according to estimates released jointly today by the U.S. Census Bureau and the Department of Housing and Urban Development.

This is 2.7 percent above the revised August rate of 452,000, but is 33.1 percent below the September 2007 estimate of 694,000.

And one more long term graph - this one for New Home Months of Supply.

"Months of supply" is at 10.4 months.

Note that this doesn't include cancellations, but that was true for the earlier periods too. Sales are falling quickly, but inventory is declining too, so the months of supply is slightly lower than the peak of 11.4 months in August 2008.

The all time high for Months of Supply was 11.6 months in April 1980.

And on inventory:

The seasonally adjusted estimate of new houses for sale at the end of September was 394,000. This represents a supply of 10.4 months at the current sales rate

Inventory numbers from the Census Bureau do not include cancellations and cancellations are falling, but are still near record levels. Note that new home inventory does not include many condos (especially high rise condos), and areas with significant condo construction will have much higher inventory levels.

As I noted a couple of months ago, I now expect that 2008 will be the peak of the inventory cycle for new homes, and could be the bottom of the sales cycle for new home sales. But the news is still grim for the home builders. Usually new home sales rebound fairly quickly following a bottom (see the 2nd graph above), but this time I expect a slow recovery because of the overhang of existing homes for sales (especially distressed properties).

This is a another very weak report.