We’re Better Off Without Seasonally Adjusted Housing Stats
Posted by Jonathan J. Miller -Monday, January 9, 2012, 9:58 AM
Comments Off
[click to expand]
I’ve been outspoken about the misuse of seasonal adjustments in housing statistics. While they are not all bad and are favored by economists for their ability to smooth out a year’s worth of information so one can see where greater than normal changes stand out, there cause more harm than good when it comes to understanding housing.
The basic problem with seasonally adjusting the numbers are as follows:
- there is no standardized methodology or period for making the adjustments
- the adjustments are rarely disclosed
- they often adjust already adjusted numbers (i.e. annualizing NAR existing home sales)
- the reader often doesn’t understand what it is
- while it may be useful for analysis, the results often contradict current conditions
and most importantly…
- the adjustments become skewed shortly after a significant market change.
As discussed in this Fed research piece on “…seasonality gone awry”:
While seasonally adjusted data can be extremely helpful, they should be used with care. In particular, the statistical methods used for seasonal adjustment may generate misleading results when applied to data with structural breaks, where the underlying properties of the data change significantly during the period studied.
Case in point is the National Association of Realtor’s Pending Home Sales Index that has basically been a mess for the past 18 months after the expiration of the federal home buyer’s tax credit. The disparity between the indices. The NAR press releases sharply contradict the feeling on the street with agents, buyers and sellers, marginalizing the meaning of the report results.
- Cash Assets of Foreign Banks: An Example of Seasonal Adjustment Gone Awry [FRB NY]
- Pending Home Sales Index [NAR]
