patrick.net
An Antidote to Corporate Media
1,198,119 comments by 14,126 users - Al_Sharpton_for_President, just_passing_through, komputodo, socal2 online now
How About a Better Metric of Sales & Pricing
invite response
2012 Jul 23, 11:50am 2,677 views 3 comments
Comments 1 - 3 of 3 Search these comments
Patrick
2012 Jul 23, 11:52am
I'd do it if I could get access to the sales data. But it's really hard to get access. Counties all have different formats and different access policies.
bmwman91
2012 Jul 23, 11:55am
Hmm, bummer. Well, I guess that people arguing the point in [insert local RE market] can always look up past average sales data & volumes and do the math themselves. Roberto & the bears that follow him should have a field day with that for the AZ market lol.
I'm not sure it's that straightforward to just compare mean or median home prices month-to-month or even year-over-year. Since prices rise during the spring and drop in the fall, some rise at this time of the year is normal. The question is whether we are seeing greater rise than usual.
And as for the yoy numbers, we had a lot more distressed inventory a year ago. Many of these had condition issues. So homes selling now may be worth more on a $/sf basis, on average, than those sold last year at this time.
It takes time to drill down to the house-by-house details for a specific area. You have to lump the sales into different categories like: 1) foreclosure, 2) short sale, 3) regular sale in need of updates, 4) recently renovated, 5) fixer, etc. There are also add-ons like pool, prime lot, culdesac, schools.
And finally, you have to have enough similar recent sales in each area to be statistically significant. One similar home sold 6 months ago, and one dissimilar home sold last month, don't tell you much about the value of current listings.
Ideally, a buyer's agent would be able to shed some light on value in their local market. The reality is that they just cherry pick recent sales to justify the seller's asking price to quickly extract a high offer (my experience).
I currently use Redfin to review the zip code for past sales, and drill down to the neighborhood I'm interested in. By looking at the pictures of past sales, sqft, $/sf, short sale/reo/standard sale, you can get an idea of what prices are doing. For example, one area had four distinct price ranges for 2011:
189-197 $/sf [short sale] (3 sales)
214-216 $/sf [fixer] (2 sales)
226-232 $/sf [good condition] (3 sales)
243-245 $/sf [recent reno, prime lot/location] (2 sales)
The short sales seem well priced, but they also took 6 months or more to close, and invariably had condition issues. It just depends on what you are willing to take before and after you close.
You can see by the mix of these sales that if short sales and fixers aren't being sold, then suddenly the average sale prices rise from $219/sf to $235/sf. The buyer's agent will exploit this knowledge to a buyer's disadvantage, claiming that prices have risen by 7% during the last year! (Buy now before you are priced out of the market!) If you've done your research you can quickly call BS and fire that lying sack.
There is always the possibility that the prices have actually risen in that neighborhood, but if you do your own comparative market analysis, you will know: a) whether prices have risen, and b) by how much. With mortgage rates dropping by 1% over the last year, it wouldn't be surprising for prices to rise. If you think about it, when rates drop from 4.5% to 3.5%, your borrowing costs drop by (3.5-4.5)/4.5 = -22.2%! This is a massive increase in buyer purchasing power, and does have an effect on housing prices. But the converse is true as well, as rates rise, borrowing costs rise even more quickly: (4.5-3.5)/3.5 = +28.6%. This will destroy buyer purchasing power should rates rise (which they won't -- see Japan). Beware of realtors who use this line to extract an offer.
It seems like there is constant debate in here about whether prices are going up or down, and sales volume usually works its way in somehow or another depending on who is arguing the point.
As of now, there is a fair amount of evidence showing that inventories are down, and prices on what is available is up. That's not really a surprise. Despite this, some people on here seem to be in bear-denial-land and insist that prices cannot possibly be rising. Well, they are, on the smaller-than-usual sales volumes. Constant demand + lower supply = duh.
How about a marriage of average or median pricing and sales volumes? Multiply the two, in a calculated and meaningful fashion, to arrive at a RE transaction spending figure that might represent something useful to someone. Is a market selling 1000 houses per month at an average of $300k healthier than when it is selling 500 houses per month at an average of $400k a year later? "A 33% YoY increase in price", declares the NAr! In one case, $300M is being spent per month, whereas in the other case $200M is being spent per month on housing. I think that it might be a useful number for us to get a better understanding or feel for what is going on.
Patrick, can your data mining operation come up with something here? You definitely cannot multiply the median by the sales volume. At the same time, it would probably be good to leave out "outlier" sales of mega-mansions since the people buying those generally buy those properties independent of what the larger market is doing. Come up with a way to eliminate the outlier sales data and then, add up all the sales prices of units that sold in a given month in whatever area you are focusing on. That will give the RE spending number I am talking about. If you don't have that sort of granularity, then you could always just take monthly average sales prices and monthly volume and multiply them. Yeah, the outliers are stuck in there, but being outliers they probably don't mess up things with that much frequency.
Ideally, if all sales were available as data points, you could do some stochastic outlier removal and then sum up all sales that are within +/- 3 sigma of the mean to get a reading of the "core" market sales total. That would remove abnormally high priced and abnormally low priced sales from things. The data I have seen is usually pretty heavily skewed, so a regular Gaussian distribution function probably isn't going to work too well.
#housing