Monday, 16 October 2006
I’ve been playing with sales data for houses in the Bay area for a while, and have always wanted to come up with an index of same-home sales — reputed to be one of the more accurate ways to do an index, because you’re not having to compensate for differences in intrinsic value between different houses.
I’m kind of stuck on that*, but in the meantime, the Flipperdex is a list of same-house sales in the Bay area, sorted by annualised gain.
I’ve filtered out sales where the square footage or build year of the house changes (figuring it to be the result of a teardown or an improvement, and therefore not suitable), as well as sales that happen very close together (to avoid capturing corrections in the data). I’ve also tried to get rid of apartments, as much of the data doesn’t include a unit number.
It’s not perfect, of course, but it is very interesting. There are lots of sales where a substantial loss is made; I imagine that these are mortgagee sales and the like. OTOH, there are some with stupendous gains; probably private transactions within a family (or just very astute/lucky investors).
The most interesting, though, is the median. If you scroll to the middle of the page, it looks like the median annualised return for a short-term housing investment in the Bay area over the last few years has been about 22%, not including expenses like transaction costs, interest, upkeep, or property tax (which is substantial in California).
That’s a pretty good return on investment, but not stellar; most self-respecting (if delusional) mutual funds seemed to be claiming that kind of return in the boom, and there was far less hassle in buying one of those.
- …because I’m not sure what the appropriate math for turning a bunch of discrete sales events into an index over time is. For example, if house A is bought for $100,000 on Jan 1, 2004, and sold for $150,000 on Jan 1, 2006, while house B is bought for $100,000 on Jan 1, 2005, and sold on Jan 1, 2006 for $150,000, how should each affect the index? In particular, how should the gain of each be distributed over the time period between when they were bought and sold? There are a few different housing indices based on same-house sales, but I haven’t been able to find out how they handle this. Any help?