Date: Thu, 21 Mar 2002 11:23:20 -0300
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Hector Maletta <firstname.lastname@example.org>
Subject: Re: extrapolation help required!
Content-Type: text/plain; charset=us-ascii
interpolating for the intervening years, however it is done, assumes a
continuous adjustment of rent. Is that actually so? Unless some
adjustment or indexation clause is included, rental contracts usually
remain fixed for the duration. I do not know whether contracts of this
sort are annual, or for two years, or what. Even when you have a new
contract with a new rental price (contract T+1 in your parlance), you
don't know whether the previous tenant ended up paying the same amount
agreed upon at the beginning of the contract, or got his rent adjusted
to reflect the changing market price.
All this regards rent paid, in nominal terms. You may have other
purposes in mind. For instance, you may ask: "This house was rented for
$1000 a month in the year 1998; how much would it be rented for in
1999?" If the house remained occupied by the same tenant from 1998 to
later years, with an unchanged contract, you won't have an actual
contract made in 1999 to answer your question with precision, so you'd
have to estimate an answer. This answer you may find by looking at the
general evolution of rental prices in the relevant area (and type of
housing) from 1998 to 1999. For this you would need a price index of
housing, that you wopuld apply to the original rent agreed for that
house in past years and thence estimate the "price" in 1999. That would
be only an estimate of course, based on other dwellings and not the one
you're interested in.
Estimating price indexes for housing rent is actually one of the most
difficult problems in the field of price indexes. All statistical boards
in the world (including the Australian one I'm sure) have come up with
some solution, as has the BLS in the US.- In the BLS Web site there are
some references to papers written by BLS staff on this matter.
Universidad del Salvador
Buenos Aires, Argentina
Lisel O'Dwyer wrote:
> Hi all
> does anyone have any bright ideas on how to do the following...?!
> I have a dataset of N records of rental properties' rental histories over
> the last 10 years - the dates the property was let to a new tenant, how
> long the lease lasted and the value of rent paid at the *outset* of the
> tenancy. I want to calculate/estimate what the value of the rent was at the
> *end* of the tenancy. Since rents (in Australia!) have not increased at the
> same rate as inflation, I cannot simply multiply by inflation.
> Instead, I have assumed that the rent at the end of a tenancy T is the same
> as the rent recorded for the beginning of tenancy T +1. Then I can work out
> what the rent was during tenancy T after one, two or X years of a tenancy of
> length Y years.
> This works out fine for records where a property was let more than once over
> the 10 year period.
> The trouble is for properties which were let only once, say, in 1992 for 5
> years (ie the lease was completed in 1997) but then the property was not let
> again. How can I estimate what the rent in 1993-1997 was? I think it must
> be something to do with interpolating/extrapolating based on the end rent
> values of dwellings with similar characteristics, (eg location, dwelling
> structure, number of bedrooms, let in the same year) but which were let more
> than once.
> But how do I do that?
> Thanks in advance!!!!
> Dr Lisel O'Dwyer
> School of Geography, Population and Environmental Management
> Flinders University
> GPO Box 2100
> Adelaide, South Australia 5001
> IAG home page: http://www.ssn.flinders.edu.au/geog/iag/
> Telephone 61 8 8201 2969
> Fax 61 8 8201 3521