New Housing Starts, While Better, Are Still Depressed

US housing starts, private single family homes, January 1980 to August 2012The US Census Bureau released this morning its regular monthly report on US housing starts.  News reports were positive, noting that housing starts are rising and are now well above where they were.  Starts on private single family homes in August grew by 27% over what they were a year ago to a pace of 535,000 (at a seasonally adjusted annual rate), while starts on all private housing units, including multi-family units such as apartments, grew by 29% over the year ago figure to a pace of 750,000.

Such growth rates are substantial.  But looking at the figures over a longer period than just a year shows that the increases, while welcome, are not as strong as they would appear.  The housing market remains depressed, with housing construction still far below what a more normal level might be, and even further below where it was during the 2002 to 2006 housing bubble.  The graph above puts the recent figures in the longer term context.

The graph shows how new private housing starts (monthly, but at seasonally adjusted annual rates) have moved since 1980.  Housing starts can be volatile, but they have never been so volatile (going back to 1980) as the recent boom and collapse.  The housing bubble started to build in early 2002, and new starts reached an annualized (and seasonally adjusted) peak of over 1.8 million new units in January 2006.  They then fell steadily, and the collapse in the housing market was the major underlying cause of the overall economic collapse in 2008, in the last year of the Bush Administration.  They reached a trough of 358,000 in January 2009, the month Obama was inaugurated, a fall of 80% from the peak.  Since then they have increased, to 535,000 last month, but remain far below what they had been.

A recovery to the previous bubble peak would be unwarranted (on a sustained basis), as it was the build-up of an excess supply of housing which led to the bubble collapsing.  But the American population continues to grow and needs housing, and it is clear that the current pace of construction is insufficient (based on historical patterns).  Prior to 2002, new housing starts was on an upward trend, but at a moderate pace.  But to keep things simple and conservative, one can take as a reasonable floor of where housing starts need to be as the average in 2001, when 1.27 million units were started.

Based on this conservative benchmark, the new housing starts of 535,000 single family homes in August 2012 would need to increase by a factor of  almost 2 1/2 to return to a more normal level.  While this is better than where it was last year in August (when it would have had to triple to reach the benchmark), it still has a long way to go.

But as has been noted previously in this blog (see the posts here and here), the shortfall in home construction since the collapse of the bubble indicates suggests a substantial potential, once housing begins to recover.  (Note that these earlier blog posts focused on new home sales, while the current post focuses on the broader concept of new home starts.  The starts figure includes starts of home units that would not only be sold, but also those which would eventually be rented, whether by original intention or because the new home could not be sold, plus homes which were built by or for a specific owner.)  The need for new homes remains, as the population continues to grow.  In the short-run, families double up, or adult children continue to live with their parents, as was discussed in the blog posts cited above.  But as soon as they are able, these people want to buy their own homes.

Based on a 1.27 million units per year norm, the graph above shows the excess of new homes (shaded in blue) between 2002 and late 2006, and then the deficiency (shaded in red) since then to now.  Based on this norm, the excess of housing started during the bubble totaled 1.3 million units over the full period.  This excess has now been more than worked off.  The cumulative shortfall (shaded in red) comes to 3.9 million units, or triple the previous excess.  Stated another way, there is now a shortfall of a net 2.6 million single family housing units.  There will be pressure to catch up on this once the economy, and the housing market, begins to recover.

Such a catch-up on the accumulated short-fall in new home construction of recent years could serve as a significant stimulus to the economy, as was discussed in the blog posts cited above.  Other commentators, such as Paul Krugman recently, have noted this as well.  But while such a stimulus to demand would be welcome, one needs also to recognize that fiscal drag has been quantitatively more important than the collapse in residential construction in explaining the lack of a strong recovery from the 2008 collapse.  This was discussed in a posting on this blog from last March.  Residential construction is only 2.4% of GDP currently, down from over 6% of GDP at the peak of the bubble, and about 4% of GDP in more normal times.  Government consumption and investment (as in the GDP accounts) is about 20% of GDP, and total government spending (including transfer payments, such as for Social Security or Medicare) is 36% of GDP.  Government is a much larger share of the economy than is residential construction.  Because of this, reversing the fiscal drag resulting from the scaling back of government expenditures in recent years (particularly at the state and local level) and allowing it to grow as it had during the Reagan years, would add more to the economy than a recovery in housing, welcome as a recovery in housing would be.  Numbers are provided in the March post cited above.

In summary, while there have been recent positive signs, housing construction remains depressed.  However, because housing construction has been so depressed for so long, there is now a shortfall in housing units relative to what is needed for a growing population.  Hence a recovery in new home construction should be expected as the economy begins to recover, and could lead to a doubling or tripling of new home construction from where it is now.  This would be a welcome stimulus to the economy.  But welcome as this would be, allowing government expenditures to recover would make an even larger contribution.

Recent Data on Home Prices and New Home Sales: Still Far To Go

Case - Shiller Home Price Index, 10-city composite, January 1987 to April 2012

US new home sales, 1980 to May 2012, annual data

A pair of new reports on housing released yesterday and today have sparked positive reports on conditions in the housing market.  Both indicate that conditions have improved.  But comparisons to the recent past can be misleading as conditions have been so miserable.  It is important to look at the data also in a long-term context.  This blog post updates two which were posted on this site last December (here and here).

Yes, compared to the recent past, conditions have improved.  But viewed over a longer term context, one cannot yet say that the changes are significant.  The recent data might ultimately turn out to have marked a turning point.  But it is too early to say that.  Plus there is far to go before one can say there has been a meaningful recovery from the downturn in US housing that started in early 2006 when the housing bubble burst.

The top graph shows the Case-Shiller 10-City Composite home price index for the period from 1980 to April 2012, where the April figures were released this morning.  The Case-Shiller numbers are three month moving averages (so the “April” numbers represent an average over February, March, and April in their raw data).  The index is calculated by looking at changes over time of individual home prices, comparing the price of the home when it was sold to the price when it was purchased.  There is also a broader 20-City Composite Index, but this index only goes back to 2000.

The Case-Shiller numbers indicate an uptick in prices in recent months.  But the upticks are small, with monthly increases of just 0.7% in April and also in March, no change in February, and negative before.  Compared to a year ago, the index was 2.2% lower.

But all these changes are small compared to the fall of one-third in prices from the peak of the housing bubble in early 2006 to now.  Prices were plummeting in 2007 and 2008, and then finally stabilized within a few months of Obama taking office.  But there has not been a significant change since then.  Nor is it necessarily likely that there will be a significant change anytime soon.  As one can see in the diagram, average home prices were fairly flat for almost a decade, from late 1988 to late 1997.

The New Home Sales figures, released yesterday by the Census Bureau, were somewhat more positive.  Estimated new home sales in May reached 369,000 at an annualized rate.  This was almost 20% higher than the 308,000 figure for May 2011.  The January to May, 2012, average pace of new home sales was 352,800, which was 17% above the 300,400 pace of new home sales over January to May 2011 (with all figures at annual rates).

These increases are more encouraging.  But they are still small compared to the pace of new home sales that reached close to 1.3 million in 2005 at the peak of the housing bubble, as seen in the graph above.  The high rate of new home construction and sales during the bubble was clearly excessive.  As was discussed in the earlier blog post, annual sales of about 900,000 a year in the US right now might be considered roughly what is needed, on average, given the US population and its growth.  Sales at a pace of 369,000 units a year is still far below this.  An increase of 17% over the pace of 300,400 in the January to May 2011 period is good, but sales would need to almost triple (an increase of 200%) to reach 900,000 a year.

Over time, one should expect home building to recover to this roughly 900,000 level.  When this happens, it will serve as a significant spur to the economy.  And it might well start soon.  Taking the 900,000 figure as a rough benchmark, there was excess home construction during the bubble years of the Bush administration from 2002 to 2006.  This is shown as the area in blue in the figure above.  The excess during this period (i.e. the excess over the 900,000 benchmark) totaled 1.1 million housing units between 2002 and 2006.  Construction and sales then plummeted as the bubble burst.  With the continued depressed state of the economy, new home demand has remained low, and the cumulative shortfall from 2007 to now (calculated again relative to a 900,000 home unit per year pace as the “normal” demand) has come to 2.5 million units as of May 2012.  There is therefore now a net shortfall in housing units of 2.5 million minus the 1.1 million previous excess, for a net of 1.4 million units.  And at the current rate of 369,000 units per year (at annualized rates), the net shortfall is growing at a rate of 900,000 – 369,000 = 531,000 units per year, or about 44,000 units per month.

All this is consistent with recent published reports (see this Census Bureau report, or this Washington Post article based on it, from June 20) on how households are “doubling up” in record numbers, particularly with adult children in their 20s continuing to live with their parents.  The Census Bureau estimates that the number of doubled up households increased by 2.0 million between 2007 and 2010, with the number of “additional” adults (over and above the household head and his or her spouse or partner, and excluding students) in such households increasing by 3.8 million over this period.

Many of these additional adults will seek their own homes as soon as they can.  This will happen when the economy improves, and the home purchases will then in turn serve to spur further improvement in the economy.  When this happens, the impact on growth will be significant.  A rough calculation in the previous blog post suggested that new home construction and sales returning to a pace of 900,000 per year would add about 1% of GDP, or 2% of GDP assuming a multiplier of two.  The Congressional Budget Office estimates that GDP is about 5% below potential, so such growth in new housing construction could act to make up a significant share of the gap.

This pent up housing demand could therefore act as a significant spur to the economy once the process starts.  This serves to underscore again how important it is to end the fiscal drag that is holding back the economy, and instead allow fiscal growth such as that which acted as a significant spur to the economy during the Reagan years (as was discussed in this earlier post on this blog).  Once growth starts, the recovery of housing construction and sales to a more normal level will act to reinforce the recovery.

It is, however, premature to claim that the recent housing data provides an indication that this recovery is underway.  While positive, the changes are still too small, when seen in the longer term context, to bear much weight in drawing such a conclusion.

Home Prices Stagnate, at Levels Similar to Those of 2003

US home prices, Case-Shiller 10-City index, 1987 to 2011

US home prices continue to stagnate, at levels well below the peak reached in 2006 during the housing bubble. They fell sharply in 2007 and 2008 during the last two years of the Bush Administration and then stabilized under Obama, first rising a bit and then falling back a bit, but with no overall trend so far.  Based on the 10-city composite home price index of S&P / Case-Shiller, the prices of single-family homes in September 2011 were 33% below the peak reached in April 2006.

It is useful to view this in the longer term context, as presented in the graph above.  While home prices are fully a third off their peak, they are still higher than they ever were prior to 2003.  The sharp fall in prices is a reflection of the sharp rise in the middle of the decade, as a bubble built up and policy makers decided not to try to do anything to moderate it.  Indeed, many politicians, as well as many existing homeowners, felt quite good about the rapidly rising prices.

Then the bubble burst, and the consequences for the economy have been clear, as the economy collapsed in the sharpest downturn since the Great Depression.  This was then followed by an anemic recovery, with still high unemployment.  Recovery from such “balance sheet recessions” are normally slow, as the entities with the over-extended balance sheets (mortgage holders in the US; the corporate sector in Japan in the 1990s) seek to hunker down and save their way out of their predicament.  Asset prices recover only slowly at best.

If nothing is done, US home prices are likely to continue to stagnate, and may well fall further.  As indicated above, while prices after the bubble burst fell by a third, they are still only at the level seen in 2003.  Yet between 1997 and 2003 they had already doubled.  That home prices have not now fallen further than simply to 2003 levels is therefore even a bit of a surprise.  They could fall more.  And as seen prior to 1997, there can be long periods when prices are basically just flat.

Those households with negative equity in their homes (commonly referred to as “underwater”) face major difficulties, even if they can afford to make continued payments on their homes.  They cannot refinance at the current low rates for mortgages, unless they can come up with extra cash to bring the mortgage down to 80% (generally) of their current lower home value.  And they cannot sell their house to someone else, perhaps to move elsewhere for a new job, without bringing extra cash to the closing to pay off the remaining mortgage balance.  The housing market remains frozen, and with that, the economy remains in the doldrums.

Unless something major is done, this weak housing market will likely keep the economy in the doldrums.  And there is no reason to believe that there will be a jump in housing prices to levels similar to those at the peak of the bubble, with this then curing the problem of the underwater mortgages.   Rather, a comprehensive program, led by government, will be necessary to restructure these mortgages, to unfreeze this market and allow the economy to recover.

New Home Sales Remain Abysmal, But The Potential Should Be Recognized

New home sale figures for October, recently released, remain abysmally low at just 307,000 sales at an annual rate.  They have fluctuated within only a narrow range over the past year, with no clear trend and no signs to be encouraged by.  What is more interesting is the longer term context.

New home sales in the US fluctuate a good deal, but around a rising trend of 1.0% a year (calculated from a regression over the 1980 to 2010 data).  They reached a fairly stable level of approximately 900,000 housing units a year between 1998 and 2001.  But there was then a bubble between 2002 and 2006, reaching a peak of almost 1,300,000 in 2005, and then crashing to a bit over 300,000 in 2010 and 2011, where it now seems to have leveled off.

Only 300,000 new home sales a year is abysmal and a major drag on the economy.  It is only one-third of the 900,000 level of the early part of the last decade, and less than one-quarter of the levels reached in the bubble years.  But it should also be noted that people need homes, and that that need grows over time due to population growth.  The trend 1.0% growth found in the data is consistent with this.  And with new home construction now far below trend, it will be a significant spur to the economy if and when housing construction recovers.  Brad DeLong has made a similar point in his blog a number of times.  Here is one such post.

Keeping things simple and conservative, let’s assume that new home sales will, for the purposes here and over periods of time, on average equal the number of new homes constructed.  All are eventually sold, with very few exceptions.  And while there are figures available for new housing starts, these are consistently higher than sales for reasons that are not clear, but may reflect homes never completed.  Using the 900,000 units sold a year between 1998 and 2001 as a reasonable norm, and for simplicity not even allow for a rising trend in this figure, one can calculate the excess over the bubble years 2002 to 2006 (the area in blue in the graph) and then the shortfall that has developed over 2007 to 2011 and continues (the area in red).  Based on the figures underlying the graph, the excess during the bubble was 1.1 million units, while the shortfall in the subsequent bust through 2011 was 2.2 million units.  There is therefore now a net shortfall in housing supply of 1.1 million, as the excess over 2002 to 2006 has been more than worked off.  And as long as new home sales and construction remain at only 300,000 a year, that shortfall (relative to a 900,000 per year norm) will be growing by 600,000 housing units a year.

This suggests that a major pent-up demand for homes has built up.  It has not been resolved due to continued problems in the housing market and especially in the housing finance market, with many mortgages underwater and hence do not move, mortgages that remain hard to get by potential new buyers, an overhang of houses not sold, and a foreclosure process that has not worked well.  The market remains close to frozen.

But if these housing market problems could be resolved, so that construction could revert to its previous norms, the stimulus to the economy would be huge.  Assuming simply a return to construction of 900,000 units a year (and hence ignoring what would likely be an overshooting of this for a few years to make up for the net shortfall of recent years), implies construction of an additional 600,000 units a year.  At the average new home price of $221,800 in 2010, the aggregate value of such homes would be $133 billion per year.  This is just short of 1% of GDP.  Most of this would add to GDP as there is some, but only limited, “leakage” through some inputs that are imported (housing lumber, fuel for trucks, etc.).

But there will also be indirect multiplier effects, especially in the current depressed economy.  A reasonable estimate of the multiplier on housing construction would be 2 to 3.  A return of new home construction to previous levels would therefore add 2 to 3% of GDP. With actual GDP currently about 5% below potential GDP (estimate for end 2011 by the Congressional Budget Office, page 12), such an addition from housing construction would bring the economy about half-way back to where it needs to be.  But this will not happen until the market for homes and for home mortgage financing begins to function reasonably well.