Over at voiceofsandiego.org I wrote a bit about the bifurcation of our market, by which I mean the fact that the high-end markets have held up so much better than low-end markets. My idea is to start tracking the price and volume of two representative sets of zip codes: the strong markets and the weak ones. So for instance, I might do a chart of the median price/square foot of homes in the three strongest zip codes (above a certain size, to prevent noise) and of the three weakest. To determine strongest or weakest I would look at pricing and volume, mostly pricing since that’s what we’re really most interested in.
I thought about separating out things by price range, but that’s kind of weird because price is one of the things we will be measuring and changing prices could cause houses to jump categories. By using zips instead, I at least have a more constant set of houses… no house is going to change zip codes regardless of its price movements.
Adam (aka SD Realtor) has already given me some good pointers that I want to look into regarding which zip codes to use, how to handicap them, etc. If anyone else has suggestions please feel free to post them.
I’d be very interested in
I’d be very interested in what you find out about Bay Park, 92110.
It might be cumbersome to
It might be cumbersome to do, but breaking the different zip areas into their 3 or 4 primary market segments would give you a detailed view of what they were doing.
Generally speaking, most of the development occurs during the boom periods with only some clean up construction occuring during the busts. Each successive boom period introduced larger homes and more features. By breaking all the data into the 3 or 4 different cycles during which they were built you’d be comparing roughly similar sizes, similar designs and floorplans, and similar overall appeal.
It would be a hassle to set up but once you got it going maintaining it would be a snap. It’s the kind of analysis that could be extended indefinitely, and would be real helpful in identifying emerging trends and future opportunities. You’d be able to watch a trend start on the west side and sweep eastward or vice versa, and pick an area for investment prior to the trend getting there.
Bugs, that is a good idea
Bugs, that is a good idea but I think it is “out of scope” for the amount of time I am able to put into this. I am hoping to be able to key off of what is available in the MLS reports, without creating a separate database or pulling together multiple info sources. That would be really cool, though. Maybe someday. 🙂
ktcat, I’m sorry, but I imagine Bay Park is somewhere in the middle, relative strength wise, so I don’t imagine it will figure into any of this. Maybe one of the realtors on the board can give you some insights into this submarket.
rich
Rich, thanks for the reply.
Rich, thanks for the reply. I can understand not doing 92110. I recently had my house in Bay Park appraised and the appraiser’s computer program claimed the darn thing had appreciated 30% in the last 12 months! That seems like madness to me.
Anyone know more about
Anyone know more about this?
As of 11-15-2007, there will be a new tool for figuring out how much toxic waste is in investment banks’ balance sheets. The new US accounting rule SFAS157 requires banks to divide their tradable assets into three “levels” according to how easy it is to get a market price for them.
This kind of full disclosure will reveal just how bad things really are “/
It’s my understanding
It’s my understanding there calling it “level three accounting” and it will be bad news for financial stocks. Most of them use the “mark to whatever we say it’s worth model” not the market price.
Gold well into four digits is coming……………..
Statistics to the Rescue
Two
Statistics to the Rescue
Two suggestions. Try running a regression on medians or median/sf by zip code, something like:
MedianPrice2007 = a + b*MedianPrice2006 + c*MedianPrice2006^2
Or, if you think there is a specific cutoff, you can assume is is BLAH (maybe $417k), and write:
MedianPrice2007 = a + b*MedianPrice2006 + c*LessThan417 + d*LessThan417*MedianPrice2006
Or you can ask a computer to find that cutoff using a method called maximum likelihood.
Free software that does all of this is at http://cran.r-project.org/
Or I can run it for you.
SANDAG has lots of detailed
SANDAG has lots of detailed current information on a ZIP code level including factors like median household income, population, available housing units (broken up between single and multi-family), occupied housing units (single and multi seperated), and # of people per household. This can be found at http://datawarehouse.sandag.org/
In addition to this, if you want even more detailed data on specific ZIP codes, http://profilewarehouse.sandag.org/ has data on median age, ethnicity, and more detailed household income.
It might be interesting to see if you could identify 2 ZIPs with similar median values but different incomes and see if this makes a difference as prices change (higher income ZIP holds up better?) You could also test other variables this way like vacancy rate or persons per household. This could essentially end up as a regression like trex mentioned but with more variables and greater predictive ability.
One last thing that’s nice about SANDAG’s data is they update it annually. The bad thing is it’s an estimate based on the census, which can lead to errors in smaller areas like ZIPs.
SANDAG BLAAAAAAAAAAAAAAA
SANDAG BLAAAAAAAAAAAAAAA
SANDAG BLAAAAAAAAAAAAAAA
SANDAG BLAAAAAAAAAAAAAAA
Hi Rich,
I wonder if it
Hi Rich,
I wonder if it would be useful to have historical charts for appreciation on a price per square foot basis to compare with the down cycle data for each of the zip codes you select? It seems like you are saying the higher end markets and lower end markets appreciated for different reasons? The lower end because of easy credit. Why did the higher end appreciate then? Maybe it will take a different catalysts to cause weakness in the higher end? A lagging catalyst. Perhaps more economy driven than consumer credit driven? I do know one theory says appreciation allowed for a chain of move ups shifting demand up the scale. You also imply an economy where money is really made instead or borrowed, or borrowed from one’s other assets, being a larger contributing factor in the higher price range. How and when does must sell inventory hit this “second market”, which is supported by wealth? What would make prospective buyers unwilling or unable to pay current prices for it?
I am not sure if this will
I am not sure if this will help or add to the confusion, but evaluating data in a spatial context can really help you get to the lowest common denominator. The two main analysis levels that you will want to focus on, at least as I see it, are either zip code or parcel. Most of the broader economic analyses produced by other parties appear to be done on a zip code level.
The shortcoming of this is that there are often sub-areas within zip codes that perform better or worse than others (e.g., 92008 zip code is vastly different depending on whether the property is west-of or east-of Interstate 5). Focusing on the parcel level might initially appear to be more work than it's worth, but once the base structure is set, it's just a matter of updating and table-joining APN's. The best analogy I can think of for the parcel level analysis would be Zillow's Heat Maps. I also think using this methodology would be more consistent with a measurement such as Case-Shiller's Home Price Index, which tracks existing home sale price changes.
[img_assist|nid=5392|title=|desc=|link=node|align=middle|width=466|height=337]
Rather than tracking a zip code number, you are tracking geographical areas (individual parcels). What's even more interesting is that using spatial analysis you can overlay other layers of data, such as sales volume changes, and compare your results. I could go on and on. My main point (finally) is that you might be better off using maps to help get your ideas across to the public. Run this by Vlad and he might help give you some more insight on how to do this.
Hi Rus, no, I wasn’t really
Hi Rus, no, I wasn’t really saying that different areas appreciated for different reasons; just that the tightening has had more effect in the lower priced areas because A) subprime has tightened more than prime and B) people in higher end areas may be able to tap other wealth sources if financing gets tighter. I think the whole market rose by and large on EZ credit and speculative enthusiasm; it’s just unwinding differently in different areas. Also, it would be excellent to chart the historical data you cite (time permitting) but I myself don’t actually have it — do you know where that could be attained?
Anyway, thanks everyone for all the suggestions. There have been some excellent ones here and in email. The problem as I mentioned to Bugs is that to do this in a really kickass manner would probably take more time than really am able to spend, so there’s a good chance I won’t be able to implement many of the suggestions.
A guy who works with Gary London (and who has put a lot of thought into this topic) read the article at the Voice and contacted me about partnering with USD to create some better indices. That would be ideal because it would obviously be more sophisticated than anything I’d have time to do on my own. I am going to see how that plays out — hopefully some new and more useful indicators will result, but if not I will probably try to do something very simplistic on my own.
rich
Hello Rich,
The historical
Hello Rich,
The historical picture for price per square foot by zip codes can be derived from the MLS excepting those transactions that didn’t go through that system. It would be tedious project. I also don’t know the protocol for using that data to derive and publicly post graphs.I am sure you know that though.
I would think there is another source wherein price, size, zip code, and close of escrow date are accumulated in a more handy format. Perhaps there is another subscription data service or a publicly available compilation from the tax assessor which would have it?