Start Date
31-10-2017 1:00 PM
End Date
31-10-2017 2:45 PM
Description
Due to physical, legal, and other barriers, as well as cost prohibitive reasons associated with data collection and storage, sparse data can be a common hurdle in the effectiveness of governments who depend on or are considering the implementation of a property tax regime. The ability to estimate–with some degree of confidence–property values for certain geographic areas is oftentimes a highly difficult task, particularly in areas with little or no sales transactions. In developing countries with limited, inaccurate or no cadastre or multiple listing service, the data needed to create reliable estimates of value is simply not available. Estimating the existing housing stock (a potential tax base) in such a country for research or implementation purposes would result in costly "boots on the ground" efforts, as well as heightened technical and data storage requirements. This research will be the first of its kind (to the authors' knowledge) in that it develops a model to estimate house price determinants across an entire country (Malawi, Africa) using data collected in the Malawi Integrated Household Survey. Several questions in this large-scale survey ask respondents about their perceived selling value of their homes, as well as physical characteristics of their dwelling (size, distance to road, wall materials, etc.). Further more, this research evaluates methods to pockets or sparse data within a city, state, or country. Similarly behaving patterns of data often emerge over geographic space. What has come to be referred to as Waldo Tobler’s First Law of Geography (“everything is related to everything else, but near things are more related than distant things”) has led to the development of statistical methodologies whose accuracy benefits from the inclusion of various GIS data. Spatial interpolation is one such methodology, that estimates data for a respective location based on its surroundings. Because land markets behave so differently over geographic space, and location plays such a large roll in value formation, conventional modeling techniques may not be able to accurately estimate value (Fotheringham et al. 2002; McMillen 2010). Spatial interpolation of variables from known data locations provides response surfaces from which additional non-measured properties can be assigned respective location-based predictions. Depicted visually on a map, a thematic view of potential geographic data patterns can be conveyed. This technique, generally referred to as “response surface analysis” (RSA), has been demonstrated with a high degree of accuracy to estimate sparse pockets of property markets. The use of RSA in mass appraisal AVMs has shown to increase predictability power, equity, and uniformity in property tax valuations (O’Connor 1982; O’Connor & Eichenbaum 1988; Ward et al. 1999; McCluskey et al. 2000; González et al. 2005; d’Amato 2010). This paper will add to the existing literature base by examining feasibility of estimating housing stock using various spatial interpolation techniques used in the mass appraisal community, side-by-side, and will compare them based on IAAO ratio study standards.
Publication Date
October 2017
Recommended Citation
McCluskey, William PhD; Davis, Peadar PhD, MRICS; McCord, Michael PhD; and Bidanset, Paul, "Estimating value in the absense of transactions: Spatial AVMs using census data from Malawi Africa" (2017). International Research Symposium. 14.
https://researchexchange.iaao.org/irs/irs17/sessions/14
Estimating value in the absense of transactions: Spatial AVMs using census data from Malawi Africa
Due to physical, legal, and other barriers, as well as cost prohibitive reasons associated with data collection and storage, sparse data can be a common hurdle in the effectiveness of governments who depend on or are considering the implementation of a property tax regime. The ability to estimate–with some degree of confidence–property values for certain geographic areas is oftentimes a highly difficult task, particularly in areas with little or no sales transactions. In developing countries with limited, inaccurate or no cadastre or multiple listing service, the data needed to create reliable estimates of value is simply not available. Estimating the existing housing stock (a potential tax base) in such a country for research or implementation purposes would result in costly "boots on the ground" efforts, as well as heightened technical and data storage requirements. This research will be the first of its kind (to the authors' knowledge) in that it develops a model to estimate house price determinants across an entire country (Malawi, Africa) using data collected in the Malawi Integrated Household Survey. Several questions in this large-scale survey ask respondents about their perceived selling value of their homes, as well as physical characteristics of their dwelling (size, distance to road, wall materials, etc.). Further more, this research evaluates methods to pockets or sparse data within a city, state, or country. Similarly behaving patterns of data often emerge over geographic space. What has come to be referred to as Waldo Tobler’s First Law of Geography (“everything is related to everything else, but near things are more related than distant things”) has led to the development of statistical methodologies whose accuracy benefits from the inclusion of various GIS data. Spatial interpolation is one such methodology, that estimates data for a respective location based on its surroundings. Because land markets behave so differently over geographic space, and location plays such a large roll in value formation, conventional modeling techniques may not be able to accurately estimate value (Fotheringham et al. 2002; McMillen 2010). Spatial interpolation of variables from known data locations provides response surfaces from which additional non-measured properties can be assigned respective location-based predictions. Depicted visually on a map, a thematic view of potential geographic data patterns can be conveyed. This technique, generally referred to as “response surface analysis” (RSA), has been demonstrated with a high degree of accuracy to estimate sparse pockets of property markets. The use of RSA in mass appraisal AVMs has shown to increase predictability power, equity, and uniformity in property tax valuations (O’Connor 1982; O’Connor & Eichenbaum 1988; Ward et al. 1999; McCluskey et al. 2000; González et al. 2005; d’Amato 2010). This paper will add to the existing literature base by examining feasibility of estimating housing stock using various spatial interpolation techniques used in the mass appraisal community, side-by-side, and will compare them based on IAAO ratio study standards.