Wednesday, April 25, 2012

Housing overview part 2: prices



- by New Deal democrat

  In the first part of this overview, we looked at housing sales and construction, with its attendant effects on employment and GDP. It appears that 6 years after the housing bust started, we are finally seeing a modest uptrend from a very low level.

 Now let's look at housing prices. If sales and construction are 3 years after their worst levels, the debate about prices is whether we are bottoming now or there is further to go. There are a number of indexes, all relying on different methods. There are asking prices indexes, median and mean sales price indexes, and repeat sales indexes. Within each type there are seasonally adjusted and non-seasonally adjusted metrics. I'll look at each of the three groups in turn.  If you want to avoid the verbiage and the graphs,  skip to the conclusion for a list describing the results.

 List prices 

 The theory here is that sellers have to price to meet the market. If they are doing so, then the trend in list prices will lead the trend in sales prices, since sales are consummated months after the property is listed. There are three of these, two non-seasonally adjusted and one seasonally adjusted.

 Every week I report the Housing Tracker YoY change. This is a non-seasonally adjusted index of 54 markets with data back to the peak of the housing bubble. This index began telegraphing a bottom by last summer, and turned positive on a YoY basis in December, as shown in this index below: 

Month2007 2008 2009 2010 2011 2012
January ----7.5%-11.5%-5.8%-8.7%+2.9%
February ----7.8%-12.0% -5.2%-8.4%+4.1%
March ----8.3% -10.9%-5.0%-7.3%+3.7%
April -2.7% -8.6%-9.6%-5.0%-6.8%---
May -3.5% -9.1% -8.1%-5.0%-5.6%---
June -5.0%-9.8%-7.0%-5.0%-4.4%---
July -5.4% -10.4%-6.1% -5.1%-4.2%---
August -6.0% -10.6%-5.5%-6.1%-2.8%---
September -6.2% -11.1%-5.1%-6.6%-1.7%---
October -6.7% -11.4% -4.5%-7.0%-0.9%---
November -6.6%-11.7%-4.5%-6.7%-0.7%---
December -7.2% -11.4%-5.6% -7.8%+1.1%---


 The NAR just entered this fray, by a non-seasonally adjusted index made up of all of its listings. The NAR's index showed that asking prices were still negative, down 2.5% in December 2011, but turned positive YoY by February, up 6.82%. As of its last report for March, asking prices were up 5.56% YoY. 

Most recently the Trulia asking price index debuted. This is a seasonally-adjusted index. In their inaugural report for March 2012, they indicated that asking prices had bottomed in January:
Nationally, asking prices on for-sale homes – which lead sales prices by approximately two or more months – were 1.4 percent higher in March than one quarter ago. Prices increased month over month 0.9 percent in March and 0.6 percent in February. The Trulia Price Monitor is seasonally adjusted, so these monthly and quarterly increases are on top of typical springtime price jumps. . . . According to the Monitor, asking prices had been declining prior to February and reached a low in January. Throughout 2011, asking prices rose slightly in several months of the year, but never more than 0.2 percent in a month. Asking prices in March were 0.7 percent below their level one year earlier.
Here's their graph of seasonally adjusted monthly changes in asking prices for the last year:

 Median and mean sales prices

There are three of these indexes.  The first only takes into account new home prices.  None are seasonally adjusted, so we have to look at YoY trends.

Each month the Census Bureau reports on new home sales, and yesterday reported for March. The report includes both mean and median home prices on a non-seasonally adjusted basis. Both the YoY median and mean prices turned positive in February and remained positive in March. Median sales prices are now up 6.3% YoY, as shown in red, left scale in the graph below, along with the actual monthly median price, blue, right scale:

 

 Mean sales prices are also positive by 11.7% YoY.

The NAR's existing home sales report also includes sale prices. According to the NAR, sales prices for existing homes on a non-seasonally adjusted basis turned positive YoY in March, up 2.5%, as shown in the graph below:






Radar Logic is another non-seasonally adjusted index of sales, based on prices paid per square foot. They reported on Monday that:
Radar Logic 25 MSA Composite data reported on residential real estate transactions (condos, multi and single family homes) that settled as late as February 20 and averaged for the month indicates that with increasing spring transactions has come an increase of prices (the typical trend) with the national index increasing 0.71% since January but falling 3.76% below the level seen in February 2011.
This is still the best YoY comparison in the Radar Logic series in over a year, as shown by the red shaded area in the graph below:

 


 Repeat sales indexes


There are five sources of repeat sales and other indexes which control for price or quality of house.  The first is not seasonally adjusted, so we must look YoY.  The other 4 are seasonally adjusted, so a turn in trend can be spotted without waiting for the YoY metric to turn.



CoreLogic makes use of repeat sales transactions to create a House Price Index which is not seasonally adjusted. Their last report:
shows national home prices, including distressed sales, declined on a year-over-year basis by 2.0 percent in February 2012 and by 0.8 percent compared to January 2012, the seventh consecutive monthly decline. Excluding distressed sales, month-over-month prices increased 0.7 percent in February from January. The CoreLogic HPI® also showed that year-over-year prices declined by 0.8 percent in February 2012 compared to February 2011.
Here is the graph of their most recent data:

 

Yesterday the FHFA, which gathers seasonally adjusted house prices,  reported that
U.S. house prices rose 0.3 percent on a seasonally adjusted basis from January to February ...  While prices in January were unchanged according to initial estimates reported in the last [House Price Index] release, the January result has been revised downward to reflect a 0.5 percent decrease. For the 12 months ending in February, U.S. prices rose 0.4 percent, the first 12-month increase since the July 2006 - July 2007 interval. 
That this repeat sales index has bottomed can be easily seen in their accompanying graph:

  FNC is not actually a repeat sales index but FNC "has developed a hedonic index based on the data collected from public records and blended with data from appraisals." It is reported for 10, 30, and 100 metropolitan areas. It's last report covers February 2012, and indicated that "prices on non-distressed home sales (excluding foreclosure auction sales, REO sales, and short sales) continue to slide, down 0.8% from February or 3.0% from a year ago."  As shown in the graph below, this is the least negative comparison in two years:

 



Lender Price Services (LPS) has a House Price Index which controls for houses for sale in a variety of price brackets in order to avoid problems with the median result based on the mix of houses for sale.  Its last report was for January, their updated LPS HPI national average home price index declined 3.2% YoY, as indicated in their graph below:

This was still not as great as the typical 5.2% YoY decline of the past 2 years.

  Finally, we come to the Case-Shiller home price indexes. Yesterday almost every report focused on the data that "the 10- and 20-city composites were each down 0.8% in February from a month earlier, and fell 3.6% and 3.5% respectively from the year-ago period." In so doing they completely missed what was perhaps the most significant installment of this report in 5 years.

Beginning in 2007 the seasonally adjusted Case Shiller reports fell every single month until the implementation of the $8000 housing credit. During the period of that credit, prices rose, only to resume their relentless fall the moment the credit ended. In fact, through January 2012, only one month - April 2011 - had risen from the month prior. Until yesterday. In February, both the 10 and 20 city composite price indexes rose from January, as shown in the graph below:

 

 Except for the fact that there seems to be a statistical glitch in the Case Shiller report for March of the last 3 years, causing each March to break trend to the downside by about 1%, it looks very much like the Case Shiller home price indexes have made a bottom with yesterday's report.

When we collate all of the indexes above, a compelling picture emerges:

Seasonally adjusted indexes:

  List price - Trulia - bottomed in January

  Repeat price - FHFA - bottomed in February 2011
                         FNC - 3.2% decline YoY (but least in two years)
                         LPC - 3.0% decline YoY
                         Case Shiller - bottomed in January?

Non seasonally adjusted indexes:

  List price - Housing Tracker - YoY positive in December 2011
                    Realtor.com - YoY positive in February

  Median and mean sales price indexes:
                -  Census Bureau - positive in February, off and on for 2 years
                   NAR - turned positive in March
                   Radar Logic 3.76% decline YoY (but least in one year)

  Repeat sales indexes:
                - Core Logic 2.0% decline YoY in February (least in two years)

All of the list prices indexes and all but one of the median/mean sales price indexes appear to have bottomed.  One of the repeat sales indexes has bottomed, and the most famous - the Case Shiller index - looks very close to a bottom if it didn't just happen.  Three other repeat sales indexes continue to show declines, but in two of them the rate of decline is rapidly abating.