Tag: statistician

Posted on

TICs, San Francisco’s Involuntary Reflex: Part 2 — The Data

There are weeks when I look through the new listings on the MLS (Multiple Listing Service) and it seems like there are more TICs for sale than condominiums.  Turns out, this just isn’t true.  Here’s a chart showing relative sales volumes since 2003 (click to enlarge).

Units Sold By Month
Look at that!  Excluding those wonderfully regular dips every Xmas, condo sales are generally at around 200 units per month.  TICs rarely break 40.

Here’s how TIC and condo median prices stack up against each other on a monthly basis.

Condo vs TICs Median Prices By Month
Dueling spaghetti you say?  That was my reaction, too.  The huge variability in prices from month to month on the TIC line is a direct result of the paucity of sales.  And this chart certainly doesn’t help get at the key question, which is this:

Given that TICs are riskier and less flexible than condos, what’s the premium that you pay for buying a condo vs.  a TIC?

In fact many TIC buyers do so with the hope of being able to realize this “premium” by converting their TICs into condos down the road.  Fat chance unless you’re buying a TIC in a two unit building which — for now at least — remain exempt from San Francisco’s byzantine annual lottery system.

Luckily, I have a bona fide statistician mathematical genius phd for a wife, and she always lends a hand on methodology when I need it.  She suggested that where one set of data (condos) is so much larger than another, using averages provides a more reliable “apples to apples” comparison than medians.   Also, with so few monthly TIC sales, I decided to look at annual rather than monthly trends.

Here’s attempt number two.

Condos vs. Tics Annual Average Sales Prices

Much more useful!  (By the way, the fact that TICs were more expensive than condos in 2003 and 2004 can be explained by a few massively (in excess of $8 million) expensive TIC sales in those years.  This is a great example of how using medians or averages can really affect the results.)

So, can we drill down further and come up with a condo premium per square foot? Stay tuned….

Posted on

Supply/Demand: Does it predict price? Maybe not.

Now hold on there, matie!  Basic economic theory  says more supply than demand, prices will fall, right?  Well take a look at this graph. It shows the absorption rate of single family home listings from January 2006 through December 2008 plotted against median prices (click to make it bigger):

absorption-price-chart1

“Absorption” is basically the number of weeks it would take to sell all the homes available on the market based on the number of homes that are selling at that time.  (I’ve tweaked the formula to diminish the spikes caused by the huge seasonal dropoff in new listings each December/January.)   There are many ways to calculate absorption, but the basic idea is simply to capture how quickly demand is eating supply.  Less time to absorb the supply should reflect a “hotter” market where sellers can demand top dollar. A higher absorption rate, on the other hand, means that there’s relatively more listings on the market than demand for them.  That would tend to suggest a buyer’s market and softer prices.

In the chart above, we’d expect to see  median prices rising when the Absorption Rate line falls and median prices falling as the Absorption Rate line rises.  ie. an inverse correlation.

Well, I’ve looked at this chart long and hard and I just don’t see those lines moving that way at all.  In fact I’ve looked at similar data as far back as 2002 and the only thing that’s clear is that people forget about buying or selling a home at the end of the year. Look at 2006:  the market got tighter but prices stayed pretty flat. In 2007, stuff was being absorbed more slowly (the red line goes up), but prices went up anyway.  In 2008, you’d think that with only two weeks of supply available, home prices would be skyrocketing.  Obviously that aint happenin’.

I’m not an economist or a statistician, but I did get my dear wife, who eats statistics for breakfast, to check my methodology.  I think these results are quite counter-intuitive.  Here are the explanations I can think of.  Please chime in with your own:

  • If you cut off the peaks and troughs, the Absorption Rate  mostly stays within a band of around 4 to 7 weeks.   That suggests that supply/demand in San Francisco is actually pretty stable.  And that in turn suggests that something else must be driving prices.  Note, for example, that my chart  doesn’t reflect the number of offers that are made on any particular house.  There might be 10 offers on a house, but at the end of the day just one house gets sold.  Anyone who was playing during the frenzies of 2005 – 2007 doesn’t need to be told how multiple offers affect price, but that sort of demand isn’t reflected in an absoprtion rate.
  • Relatively speaking, San Francisco is not a stressed market.  Supply/demand is not hugely out of whack.  Foreclosures are not piling up (yet).  Under these circumstances, prices are “sticky.”  They don’t react quickly to changes in demand.  If people don’t get the price they want, what do they do?  They don’t sell unless they really have to.  And SF home-owners tend to be people who don’t have to sell.  More on this in another post.
  • The price increases of the last few years and their recent tumble may are probably most directly attributable to one thing, pure and simple:  easy money.  That aint gonna show up on this graph either.

Conclusion:  “Absorption”  isn’t a good measure of supply/demand.

So is there any other metric that correlates more closely with changes in price?  How about the famous “DOM” — Days on Market.  This is how long a property takes to go from being listed to being sold.  The theory goes that when properties sell quickly the market is “hot.”  Why do properties sell quickly?  Probably because … there’s more demand than supply.  ie.  More people making offers, more people getting the loans they need, more people willing to waive inspection contingencies and buy “as is” just to get the deal done.  So maybe DOM actually does capture those muliple offers where the Absorption Rate just doesn’t.

So will DOM tell us how “hot” the market is and where prices are headed?  Or is DOM dumb.  Stay tuned….