Seasonally Adjusting the Google Real Estate Index
Barry did an interesting and short post on the Google Real Estate Index. It measures the amount of search going on over real estate. My question was: okay, are we over or under the trend at present, on a seasonally adjusted basis?
I decided to run a regression where each month would have a similar effect across years, and each year would have its own effect. December 2009 was the baseline. Here are the results:
Wow. Very significant results. As Barry said, “Go figure: Even the search pattern for Real Estate is highly seasonal;” It’s not that surprising. People don’t search in the fourth quarter, because they know the inflexibility derived from children and schools. (Only 2% of the population homeschools and can act like turtles, taking their homes with them as they walk. That said, homeschoolers don’t typically use that flexibility.) But at the start of each calendar year, people look forward to the new year, and make new plans on real estate.
From the annual coefficients, there is also no surprise — 2005-2007 were great, 2008 was worse, and 2009 was horrible.
It should not them be surprising that with a 94% R-squared, that the following graph would be tight, actual versus expected:
But looking at the bottom, the purple line indicates when people have been more willing than normal to search for housing. This is such a time — on the low end, from what I am seeing, many people are more interested in housing given the current lower prices. Should we jump up and down about this? Not sure, but it does point out what I have said recently, that housing on the low end has reached equilibrium with foreclosures.
Don’t get too excited by this, 2009 is still a bad year for real estate, but maybe a few things are starting to turn up.
PS — all of this assumes that search on Google has some correlation with actual intent to buy or sell real estate. I think that is a reasonable assumption.
