(a) To establish how many dwellings exist in an area they use the census figures from 2006. Additionally they estimate the dwellings in 2005, 2007 and 2008 (not sure how) and from all this you come up with a figure for how many dwellings an area has.
(b) They somehow establish how many houses are rented. They say they go back to census figures for this. This tells you how many houses are up for rental at some time or other. Sure it will change, houses are sold to investors or investors sell to owner occupiers but the aim is to say "this area has x amount of houses occupied by renters or available for occupancy by renters", ie rental stock is established.
(c) Now to ascertain how many of those rental houses are vacant they monitor adverstising of dwellings for rent. They track new listings, listings withdrawn and then come up with how many are still listed at the end of the month.
The previous month had x still listed at the end of it, this month had y still listed and then they "conduct a reconciliation" which is not clear.
They seem to take steps to weed out possible fake ads (I didn't know such things existed) and they only use ads that are advertised for two weeks or more and are still advertised at the end of the month.
-------------------------------------------
I'm impressed with the work that goes into the methodology and would be interested to know how others do it since the figures are so different.
The problems I see with SQM's method (problems which are unavoidable) is that the number of dwellings in 2006 (adjusted with estimates from other years) may not be as representative as they think. If a new apartment block has gone up this would increase the number of advertised dwellings but if you haven't updated the number of available dwellings (accurately) you would yield a higher apparent vacancy rate.
If houses are having townhouses built in the backyard (or granny flats) and then advertised for rent the collaters of data probably would not be aware that this is new stock which should be added the available rental stock figure.
It would be hard to monitor constantly shifting systems. Like I said I would like to learn how other bodies monitor this and how they come up with figures.
