A hedonic model comparison for residential land value analysis

Abstract

Distantial attenuation is a significant characteristic of the urban environment. In this paper we explore the relative contributions of different land uses to the urban residential environment according to distances with an analysis on the most significant scope of these factors. Two types of models are developed. One employs the distance to city center, main road, public facilities and environment factors as the variables for macro environment analysis, and the another one includes all factors to analyze the influences of the micro environment. In the models of the latter type, the contributions of the neighboring land uses are magnified with a specially designed variable measurement scheme in which variables are evaluated specially for each model so that the contributions of the variables become comparable from model to model. With an application to the case study of Danyang, China, we measure the distantial functions of the urban environment (e.g. the surrounding land uses) including both the most influential scopes and the relative contributions by means of model summary and regression coefficient comparisons. Finally, case comparison on the samples is introduced to figure out the difference of the functions under micro environment.

Introduction

Distantial attenuation is a significant feature of the location. Since every urban land use activity should consider the best location to acquire the most utilities, the distance dependences to each of the environment factors are sure to be of interest to researchers. In most of these studies land value always evolved as a substitute for the location utility. Perhaps the most fundamental one among these studies is the work done by Wingo (1963) who firstly tried to systematically analyse the relationship between the distance function of location and the urban land value by modeling the value of consumption in transport. He developed his famous location equilibrium theory which declared that transportation costs can be substituted by the space costs or land values. Then there is Alonso's profound trade-off theory (Evans, 1987, Kivell, 1993) in which Alonso (1964) pointed out that land values are generally determined in the urban area, more specifically speaking, the value (or the utility) of the urban land should decline with the increasing the distance to the CBD. Based on the fundamental researches of Wingo and Alonso, more considerate investigations were performed. For decades, scholars have spent a lot of endeavor in investigating various commercial places and retail supplies other than being solely focused on the CBD (Ragas, 1987, Song and Sohn, 2007) accompanied with generalized transport including the accessibility changes (Smersh and Smith, 2000) and transporting improvements (Yiu and Wong, 2005) were considered. Location of all kinds of public services such as schools (Chin and Foong, 2006), outdoor recreations (Wilman, 1988), etc. or a synthetic supply (Randall, 2005) also provides many prostheses for such discussion. As a traditional public issue, the accessibility to the environment amenities such as open space (Jim and Chen, 2006, Jim and Chen, 2007, Kong et al., 2007) or urban forest (Tyrväinen, 1997, Tyrväinen and Miettinen, 2000) arouses a lot of interest too. Compared to the pioneers, these researches are no more rigidly macroscopic and theoretic and the results work more confidently. Recently some scholars began paying their attention to the neighborhood effects (Galster et al., 2004, Hui et al., 2007, Tse, 2002). Some founded that in the classical smooth attenuated models for urban environment influence analysis, the contiguous influences are always underestimated. Although there are no systemic and specific investigations about it yet, some of the researches have surely proved that the influence of the distantial attenuation of the urban environment exists in both the “unique spatial fixity in the urban area” (i.e. macro environment) and the “neighborhood characteristics” (i.e. micro environment) (Jim and Chen, 2007, Wendt, 1957).

Urban land is a multiattribute utility comprising of heterogeneous characteristics (Colwell and Munneke, 1997, Qadeer, 1981, Wendt, 1957), hedonic regression method (HRM) makes its numeral measuring possible. In the HRM, land value can be specified into a bundle of the locational characteristics of the plot and expressed as some kinds of the form of $U(z_1,\rightleftarrows z_2,\rightleftarrows \dots ,\rightleftarrows z_n)$, where U means the accumulation of a bunch of utilities $u_i(z_i)$, and z_i is the normalized distance functions. Statistics and regression procession figures out each value of a particular locational distance, which is described as the “pure willingness to pay” in Bockstael and McConnel's (2007) manuscript. HRM has significantly promoted the progress of the locational environment studies, especially when it concerns the land value.

However, in most of the HRM applications literature researches rely on mono-distance functions, in despite of the difference between macro and micro environment having been mentioned a lot. When Yeates (1965) having noticed that the land value declines only being functional in the area within 1.5 miles to the CBD, for decades the inter-activities between micro and macro environment are seldom discussed qualitatively. Occasionally relative conclusions appeared as some sort of by-products, for example Wu et al. (2004) found that the distance to the nearest park, river or lake being 1000 feet can make sufficient influence on housing price, and Mahan et al. (2000) declared that 1000 feet can be the margin of the influential distance of the wetland in contributing to the house price, etc. Most of these findings are scattered and not specific. Yes, it is easy to tell that a plot nearing to a main street is less profitable in the city fringe than in the city center, and being near to a park is more attractive than to the main street in the city fringe. Though, how to measure them? The answer is still under the seal.

This paper is intended to figure out such differences. Methodologically the paper sketches a model-analysis of the cooperation of the macro environment and the micro environment. Variables contributing to them are strictly distinguished and are selectively added to different HRM models. Firstly the influences of the macro environment are specifically figured out by employing the residential benchmark price (RBP) which is designed and adopted as a kind of general mass appraisal for the physical homogenous environment. Then micro environment is involved in, with introducing the market values of the residential land (MVRL) as the result of complex interactions between the characteristics contributing to both the macro environment and micro environment. The analysis and results cover both the influential scopes and the relative contributions. Different hedonic regressions are designed to estimate the equations for the land values under different environments (Eqs. (1), (2)).

$$\text{RBP}=f(\text{macro}\text{environment})$$

$$\text{MVRL}=f(\text{macro}\text{environment},\text{micro}\text{environment})$$

With the application of the geographic information systems (GIS) in handling and processing large amounts of spatial data, it is possible to associate and process the land value data and locational data we needed and to undertake the analysis with multi-original data (Nzau, 2003, Turkstra, 1998). At last, explicit case comparisons are taken for deducing the results.