, , ,

Academic research can seem rather odd to people not at a university.  Most of it seems obscure, until somebody needs to find the expert on a particularly troublesome topic.

This podcast explores some research done about 15 years ago by one of the members of GREG.  The discussion at the top of the page by the interviewer shows how this work shifts thinking on the connections amongst list prices, selling prices and how long it takes to sell, and is followed by the podcast.  (Click on the green arrow at the bottom to listen to the podcast itself.)

Understanding this very old problem has implications for understanding some current dilemmas.  For example, if a house is sold in one week, is that a good thing or not?  (Usually not since, in such a short time, some high bidders were probably excluded.)  Or, if an absurdly low list price is used as an advertising gimmick, what does that mean for the expected selling price?  What if (as often happens) that house does not sell as quickly as expected?  (This unresolved problem is hard to state with enough precision to analyse.)

On a technical level, this research studies time on market in a way that may be unfamiliar.  Most people think that if a house sells after three weeks, then 3 is the useful fact in the same sense that a selling price of 300000 is a useful fact.  But, because times moves forward, “3” reveals more.  For example, if 100 houses are offered for sale on the same day and 25 sell during the first week, then the probability of sale during that week is 25 percent.  If 25 more houses sell during the second week, then the weekly probability of sale increases (to 33 percent).  If 25 sellers give up at the end of the second week (i.e. do not sell, which happens surprisingly often) and only 10 more houses sell during the third week, then the probability of sale again increases (i.e. to 40 percent= 10/(100- 25- 25- 25)).

Estimating these effects accurately requires some specialized techniques, and access to high quality data (which is hard to find in Canada).

This pattern of probability over time affects the job of a real estate agent and the bargaining positions of a seller.  It changes with market conditions and, increasingly, information technology.  Distinguishing these effects from random variation accurately and precisely is worth studying.