Alabama, Birmingham, and Huntsville Real Estate Market in 2012-Forecast

While the Real Estate market adage is “Location, Location, Location” I submit that based on this study that maybe it should be “Jobs, Jobs, Jobs”

Over the years, I have often been asked to predict what will happen in the future in the residential real estate market. I have generally resisted providing answers to this, pointing out that the current data which I provide gives some pretty strong hints.

However, in doing some consulting work I have come across some useful correlative data that holds promise for quantifiable, accurate, projections of future real estate sales.

While it is  important to keep the adage “correlation is not causation” in mind, correlations can be useful and dependable if they pass the “common sense” test.

I have found that for a given area, the Bureau of Labor Statistics Unemployment Statistics can reasonably and reliably predict future sales. This makes sense in that buying homes is not a spur of the moment decision, but rather influenced by consumer future expectation, which is nicely captured and influenced by the unemployment rate.

To date, I have tested this concept in multiple markets with similar results. I expect that additional implementation will show similar predictive capabilities.

Since both the real estate and labor market have large seasonal swings, I chose to use the January unemployment rate to project the entire year’s transaction volume. I then spread the volume over the individual months, based on the historical percentage of sales that a given month has in a year. The percentage sales in a month holds steady in “normal” years. I have currently used 2004-2011 to arrive at the percentage sales per month. I suspect that some inaccuracy comes from including 2010 when we had the really abnormal melt-down followed by the extraordinary tax credit. Rather than continue adjusting numbers, I have chosen to publish these preliminary findings in the hope that others can help with the methodology.

Summarizing the findings:

How Unemployment correlates to Real Estate Sales from 2004-2010:
Correlation rate:
Statewide         80%
Birmingham     83%
Huntsville         73%

I have also tested a few other areas, outside Alabama, with similar results.

Interestingly, when I included years before 2004, correlation rates went down. I suspect that this is due to the way financing changed in 2004. I also suspect that years before 2004 will show strong correlation, and that years post 2004 will continue to show even more correlation just not across the 2004 “divide”. This remains for more research. By the way I attempted to find some correlation based on mortgage interest rates. In short, there was none. There may be some very short term impact from interest rates, but nothing close to the jobs market.

Based on the initial correlations a linear regression can predict sales for 2011, and preliminarily for 2012 (based on the estimated January 2012 unemployment rate) for the same three markets.

Market	2010 (actual)	2011 (proj)	2012 (proj)
Statewide	36,234	37,407	39,429
Birmingham	12,235	12,736	13,430
Huntsville	8,543	8,321	8,507

The statewide totals are via the Alabama Center for Real Estate

The monthly projected results compared to the actuals for 2011for Birmingham are as follows:

2011	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov
proj	755	877	1135	1137	1239	1314	1180	1191	1090	933	914
actual	683	773	1072	1022	1106	1255	1237	1224	1145	983	944
%error	-10.5	-14.9	-5.9	-11	-12	-4	4	2.7	4.8	5	3
Cumulative error											2.9

As you can see this is quite encouraging. The cumulative error for 2011 projections is under 3% for 2011. I believe that excluding 2010, to calculate the monthly changes, due to the tax credit impact will improve the monthly accuracy.

Why might this be useful?

If the results during the year do not unfold as projected one should be looking for the reasons such as the tax stimulus or similar government action. It is also likely that other impacts and shocks to consumer expectations will cause the market to react in ways that are not accounted for in the projections. One can think of major disasters such as the coastal oil spill and multiple weather events. I have not yet tried to see how the coastal markets can be projected.