As some of you know, I started keeping my own property listings of Redondo Beach homes for sale starting in late September. I wanted to get better measures of how long it really takes to sell a house here.
I'm also keeping tabs on sales. Unfortunately, there seems to be a number of problems with sales records. It makes me wonder how the data reporting agencies like Trendgraphix or DataQuick manage to report anything reliably. The biggest problem I see is that not all sales, as reported from what I find in Melissa data, make it into any form in Domania or Zillow. A bunch of sales records just aren't there, or they just aren't easy to pull up. Another major problem is that the sales record does not always accurately reflect the square footage and rooms of a property, making it darn difficult to match with an inventory listing. This happens in the case of a remodel or new construction that has been sold. It's not unusual for a sale record to show 9 square feet! This makes it a problem to compare square feet versus sale price, or to do good frequency distributions by square feet. It makes it too difficult to compare the square footage of what has just sold with the square footage of what is up for sale. The sales records will understate the square footage. The MLS listings look far more accurate when listing square footage and rooms, though such listings are not perfect. I've tried to use the county tax assessor as a third source of data, and I honestly wonder if the county collects all the taxes it's supposed to get, because there are properties that appear in sale records that don't ever seem to show up in the tax assessor records, many months after the sale. But I'm not about to go down that direction!
As an example, Melissa data tells me that in December 2006 there were 43 sales apiece in 90277 and 90278 for a total of 86. These zip codes are Redondo Beach, though little tiny pieces of Torrance do spill over into these zip codes. I don't think there is enough Torrance spillover, though, to explain that I can only find a total of 61 sale records through Domania and Zillow. What happened to the remaining 25? To add to the problem, Zillow has great data but a terrible user interface for getting to it. On the other hand, I can semi-automate data extraction out of Domania, but Domania does not tell me the exact sale date, only month and year.
And of those sale records, I don't have a listing record to match each sale, which means that for that sale, I cannot calculate days on market, or keep track of what the original asking price was and how much the price has been reduced. I sure hope realtors have all that data at their fingertips AND the tools to access it, because I would think they would need to know this kind of information to factor in whenever they list a property for sale.
Getting back to my data collection problems - what I've ended up doing is just updating the sales fields on my listing records when I stumble across sales records that match. In my December example, from 86 hypothetical sales, I can find only 61 sale records, and of those, I've got only 49 matching inventory records. Then maybe out of those I have to throw away a few because no square footage was recorded either in the listing or the sale.
Then there is the problem of what to present. At this point I can scrape up only 47 sale records for December. If I expand that to Q4 2006, I've got 98 records, but this collection is heavily weighted to December. In a market that's changing pretty quickly, presenting data on sale price and what square footage sells at that price, spread out over an entire quarter, can be rendered useless equally quickly. Probably better to just present December data, to juxtapose against January, even though there are so many fewer records.
So knowing how limited this data is, I tried to compare December 2006 sales (47 records) against my unresolved inventory (566 records, properties that had been listed at some point since late 2005, but had not sold by the end of December 2006).
The inventory statistics are summarized as follows:
Ask Price SQFT
MEDIAN 799900 1824
AVERAGE 907681 1871
MIN 379000 410
MAX 3200000 4700
The December sales statistics are summarized as follows:
Orig Ask Sale Price SQFT
MEDIAN 809000 735000 1751
AVERAGE 821785 776850 1771
MIN 529000 475000 812
MAX 1649000 1400000 3324
This first chart shows the two series compared to each other. The dark blue diamonds are inventory and the red squares are sales. Linear trend lines are drawn in. Each point plot (square feet, price). For inventory it is current asking price, and for sales it is sale price.
This second chart blows up the area where the median values fall. The large yellow square is the closet value to the median sale "point." The large yellow diamond is the closest value to the median inventory point.
It's nice to see how inventory and sales scatter next to each other, but these charts don't really tell us anything about which categories of square feet or price sold the most frequently. Where is the bulk of this market?
For December 2006, it appears the bulk of the market was in the $750K-$800K range. In this third chart, I've taken frequency distributions of the inventory and of the sales, then I divided the number in each category into its total for that category to get percentages. So 11% of the inventory falls in the $450K-$500K range but there were few buyers for that range. Why? Are these properties located in less desirable areas? Are they little 1 bedroom 1 bathroom boxes? Are they major fixer uppers that nobody wants to spend half a megabuck to fix? Notice how 45% of sales were covered by three ranges: $550K-$600K, $600K-$650K, and $750K-$800K. That's difficult to ignore - in those ranges, buyers feel like they are getting something.
I guess to complete this exercise, I should do the same percentage distribution by square feet. I've been on this computer all day so I have to stop here, but will include it next time I try this.
UPDATE: See this post for a correction to that last chart.