September Existing Home Sales: Highly Questionable Data
by Dave Kranzler, Investment Research Dynamics In the context of today’s new home sales report catastrophe – and I’ll have a detailed analysis on that later this week – I continue to question the validity of the National Association of Realtor’s existing home sales report. Last Thursday, the NAR released its September exiting home sales report, which purported that existing home sales occurred in September at a “seasonally adjusted” annualized rate of 5.55 million. Notwithstanding the absurdity of taking a data measurement for a single month and expressing it as an annualized rate, the report itself is highly questionable. Part of my problem with the existing home sales report – aside from the overall methodology of measurement – is that fact that over the last two years there has been a complete disconnect in the correlation between the existing home sales and new home sales reports: A Pearson correlation coefficient analysis on the 24 months (9/2013 – 9/2015) of data showed just a 47% correlation between new and existing home sales. Interestingly, I ran the correlation coefficient on the data from May 1999 – Sept 2012 and it showed an 87% correlation. In my opinion, the low correlation between existing and new home sales which has surfaced in the last two years suggests a high degree of estimation error in the sampling and seasonal adjustment methodology used by both the NAR and the Census Bureau. To the extent that errors are introduced into the monthly calculation, this error is compounded by a factor of 12 when the monthly data is converted into an annualized rate metric. I wrote an article for Seeking Alpha in which I discuss the reasons why the NAR’s report is highly unreliable in terms of measuring the true level of existing homes sales. You can read my analysis here: September’s Existing Home Sales Report Is Highly Questionable.