The New Home Sale stats were released yesterday and provided a “pick and choose” resource for how the housing market actually performed. Every month, the report provides a plethora of reasons why you shouldn’t rely on data that is adjusted, re-adjusted and shows a wide range of results.

Here’s what I mean, as shown in this text taken from the very top of the released New Residential Sales report [pdf]

Sales of new one-family houses in May 2006 were at a seasonally adjusted annual rate of 1,234,000, according to estimates released jointly today by the U.S. Census Bureau and the Department of Housing and Urban Development. This is 4.6 percent (±13.1%)* above the revised April rate of 1,180,000, but is 5.9 percent (±10.8%)* below the May 2005 estimate of 1,311,000.

In other words, the question is whether May 2006 New Home Sales were:

A. up 4.6% from April 2006, but down 5.9% from May 2005
B. up 17.7% from April 2006, but down 16.7% from May 2005
C. down 8.5% from April 2006, but up 4.9% from May 2005
D. down 8.5% from April 2006, but down 16.7% from May 2005
E. up 17.7% from April 2006, but up 4.9% from May 2005
F. somewhere in between
G. All of the above

The consensus of news coverage opted to use the nominal figures provided suggesting new home sale gains.

UPDATE: As a reader just pointed out, the confidence level for the monthly change in the Northeast of -7.9% had a confidence interval of 30.9%. The West was nearly as bad.

Thats crazy inaccurate. In other words, don’t use their month over month housing stats. Imagine doing month over month real estate stats in a local market with a much smaller data set?

Tags: ,

One Response to “New Home Sales: To Good To Be True, Literally”

  1. techy lname says:

    You probably noticed the confidence interval for the Northeast : +/- 30.9%. With that kind of interval, there really isn’t a lot of useful information contained in the report (at least for the Northeast).