Miller, Robert
(2024)
GeoPrice: The development of an efficient,
rapidly-updating, mix-adjusted median
property price index model using stratified
geospatial matching.
PhD thesis, National University of Ireland Maynooth.
Abstract
The topic of house price index modelling is one which is central to a significant
number of market stakeholders; governments, central banks, homeowners and businesses,
among others. The impact which property price indices have on inflation,
economic growth and policy-making are profound, yet the methodology and processes
behind the generation of these statistics tools are rather opaque.
National statistical agencies will typically use one of two de-facto standard methods
for modelling the housing market, those being hedonic regression and repeatsales.
While these methods bring with them distinct advantages, they also suffer
from significant drawbacks. One of the most problematic of the these is the volume
of rich property data required by the model. These data requirements often necessitate
the use of non-public data sources, usually acquired through privileges as a
government agency. As such, it is difficult for end-users of these statistics to verify
the veracity, reliability and accuracy of the results.
Furthermore, these intensive data requirements induce a typical lag to publication
in excess of two months. As a result, not only homeowners and businesses, but
even policy-makers are operating on stale information, which is a substantial limitation
given the critical influence exerted by the housing market on so many facets of
the economy.
Our proposal is a novel, geospatially stratified house price index model which
can be computed automatically on publicly available datasets. The algorithm does
not require additional, privately-held attribute data for each property, nor does it
necessitate a great deal of statistical expertise to implement, maintain and interpret,
as the existing standards do. In this thesis, we will outline our methodology and
demonstrate the performance of the index, initially on the Irish property market.
Following an initial study on Irish sale transactions, the model is extended to
a database of asking prices for homes online, thus demonstrating the flexibility of
the approach. This illustrates the accessibility of the model to operate on a variety
of data sources. Finally, our algorithm will be employed to create a property price
index for the United Kingdom, where the public dataset of sale transactions is significantly
more plentiful. The results of this demonstrate that our index is not only
as good as the official hedonic regression model produced by the ONS in the UK,
but far exceeds the smoothness and noise reduction achieved by said model, while
maintaining a month-to-month correlation in excess of 85%. Moreover, our proposal
achieves this with a lag time from data publication in the order of hours, rather than
weeks, as per the ONS house price index.
| Item Type: |
Thesis
(PhD)
|
| Keywords: |
GeoPrice; development; efficient; rapidly-updating; mix-adjusted median; property price index model; stratified
geospatial matching; |
| Academic Unit: |
Faculty of Science and Engineering > Computer Science |
| Item ID: |
19043 |
| Depositing User: |
IR eTheses
|
| Date Deposited: |
15 Oct 2024 11:23 |
| URI: |
|
| Use Licence: |
This item is available under a Creative Commons Attribution Non Commercial Share Alike Licence (CC BY-NC-SA). Details of this licence are available
here |
Repository Staff Only(login required)
 |
Item control page |
Downloads per month over past year
Origin of downloads
Altmetric
Altmetric