Main Variables Influencing Residential Property Values Using the. Entropy Method the Case of Auckland

Transcription

1 Main Variables Influencing Residential Property Values Using the Entropy Method the Case of Auckland 1 Dr. Ge, Xin Janet and 2 Du, Yue 1 School of the Built Environment, UTS, Sydney 2 Beijing Widnell Engineering Consultancy Co. Ltd ABSTRACT Property value is an essential aspect of property markets world wide. They are determined by a variety of factors and determining those factors is a significant part of valuing properties. The Entropy method is one of the ranking approaches used for property valuation. The benefits of the entropy method include that it requires less information than do other methods. This paper investigates the main variables that influence residential property values in the Auckland, New Zealand property market and ranks the variables using the Entropy method. The paper starts by reviewing the literature and identifies the possible determinants for residential property value. Secondly, a total of 6791 data sets were collected from the Auckland property market and reasonable assumption made to quantify the data. The application of entropy based on weight of importance, and the priority ranking is than obtained. It is found that number of Bedrooms, CPI, land area, and location are the most significant variables contributing to property values. Keywords: Property value, Valuation, Entropy, Auckland INTRODUCTION Real estate property valuation is a topic of interest for stakeholders for different purpose. Governments value properties for tax purposes. Investors like to know the purchase price of the property in which they are going to invest. Developers seek to find out the Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 2

2 feasibility of selling price for their decision-making. Other purposes are for bank lending, accounting reports and insurance. There are different concepts of market value and price. Market value is defined by FNMA (2007) as: The most probable price which a property should bring in a competitive and open market under all conditions requisite to a fair sale, the buyer and seller, each acting prudently, knowledgeably and assuming the price is not affected by undue stimulus. The definition of the International Valuation Standards Committee (IVSC) (2007) has been adopted by many national property organizations as follows: Market value is the estimated amount for which an asset should exchange on the date of appraisal between a willing buyer and a willing seller in an arm s length transaction after proper marketing wherein the parties had each acted knowledgeably, prudently and without compulsion. Both definitions focus on property transactions in a fair market. Price can be the same, under or above the market value depending on specific market conditions as a result of the interaction of the forces of supply and demand. Real estate property valuation is thus the study of market value of properties. There are risks or uncertainty in property valuation as it could be subjective (Joslin, 2005). To be able to accurately estimate a property value, we must study the main attributes and it is also important how the attributes impact on the value. This paper investigates the main attributes that influence residential property values applying the Entropy method, using data from the Auckland property market. The paper starts by identifying the possible determinants for residential property value and methods used for the study by reviewing the literature. Secondly, the method and basic steps to apply Entropy will be described. Thirdly, a total of 6791 data sets were Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 3

3 collected and applied to the Entropy method, followed by a discussion of results and conclusions. REVIEW OF RELATED WORK Previous research has found that the value of residential properties depend a number of characteristics because housing is a heterogeneous commodity. Kauko (2003) listed a set of attributes that has been commonly used in property valuation research including accessibility factors, neighbourhood level factors, specific negative externalities, public services, taxes, and density factors. McCluskey, et al. (2000) measured the effect of location on residential house prices using the Ordnance survey of Northern Ireland data. The attributes they considered in their multiple regression model, calibrated by location adjustment factors, is sale price, date of sale, age of property, size, number of bedrooms and bathrooms, number of garages, type of central heating, condition, neighbourhood and group cluster. The founding was that location and structural characteristics are the key determinants of residential property values. Tse and Love (2000) identified four categories of attributes, namely structural, physical, neighbourhood and environmental, to measure residential property values using a hedonic equation in Hong Kong. Their findings showed that estate-type dwelling units are valued by the homebuyers as attributes reflecting the preferred quality of a living environment. In addition, the availability of car parks has a strong effect, and a view of a cemetery has a negative influence on property value. Similarly, Chau, et al. (2004) studied the effect of balconies on the residential properties value in Hong Kong and found a positive effect on the value of a property, irrespective of the quality of the view. Table 1 shows parts of the literatures that identified attributes influencing property value. Table 1: Main attributes influencing property value Authors Location (Joslin, 2005) Age Size Determinant Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 4

4 (Kauko, 2003) (Paz, 2003) (Wong, Hui, & Seabrooke, 2003) (Case & Shiller, 2003) (Han, Yu, Malone-Lee, & Basuki, 2002) (McCluskey, Deddis, Lamont, & Borst, 2000) (Blackley, 1999) (Meen & Andrew, 1998) (Cheshire & Sheppard, 1998) (Lenk, Worzala, & Silva, 1997) Location Shopping centers Highway Parks Metro Neighborhood characteristics GDP Level of income Migration Construction Activity Economics Activity Purchasing Power Interest rate Number of Employment Land area Parks CBD Schools Location Changes in tax policy Age composition of the population Rate of household formation Income, real interest rates Nominal interest rates General level of prices Household wealth Demographic variables Tax structure Financial liberalization Housing stock Income Interest rates Demographic Structure Housing stock Location Level of income Population Transport Policy Neighbourhood characteristics Number of bathrooms Number of bedrooms Age of house Lot size Basement area Contact: Dr. Ge, Xin Janet on 5

5 Total area of house Number of fire place Number of car garages The hedonic approach with regression based techniques, introduced by Rosen (1974), has been a traditional appraisal approaches, widely used in the residential property valuation (Rosiers, Theriault, & Villeneuve, 2000; Rossini, 1997). Kauko (2003) argued that there are issues involved such as nonlinearity, data quality, model flexibility and functional discontinuity. Spatial analysis is an alternative approach that uses a set of data based on discrete points for sub-areas applying GIS framework. A best function that represents the whole surface is then selected to predict values at other sub-areas. The dynamics of the spatial distribution of landed residential property values were studied by Han, et al. (2002) in Singapore. They found that the distribution of transaction prices among the various geographic units was quite unique, and concluded that the value distribution of housing in Singapore was by and large reflecting a dichotomous pattern. Location was an important factor, and government intervention had an impact on the value distribution of landed housing. Uncertainty is a real and common phenomenon in valuation. French and Gabrielli (2004) stated that uncertainty is due to the lack of knowledge and poor or imperfect information about all the inputs that can be used in the valuation. Valuers would better consider all possible purchasers in the market in order to make a best price in the future (Mallinson & French, 2000). They pointed out that the lack of a standardised approach to identify uncertainty is an ongoing difficulty within the industry. Therefore, seeking a technique that minimizes uncertainty is a concerned of many researchers. Probability theory is used in measuring uncertainty. It can help the user to identify a range of outcomes for the most important variables and to allot probabilities for these variables. Since Simulation was developed to analyse probability (French & Gabrielli, 2004), Monte Carlo simulations for real estate valuation has helped to corporate uncertainty of valuation parameters (Hoesli, Jani, & Bender, 2006). Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 6

6 The validity and accuracy of valuation procedures has increasingly been questioned by concerns relating to data quality and the need for more objective appraisals based upon quantitative methods. Dotzour (2001) researched the reliability and precision of the cost approach, testing the hypothesis that the cost approach yields unbiased estimates of market value, and the factor of age of the subject properties to the precision of the cost approach, as well as comparing with the sales comparison approach to determine the better reliability. He found that the cost approach is a less reliable method of estimating value when compared to the sales comparison approach. His result also revealed that both appraisal methods systematically overvalued the properties in the sample. However, it can help to identify the top end of the expected range of value for which the subject property will sell, and can be used as a check on the estimate of value from the sales comparison approach. An artificial Neural Network method was recently introduced to residential property valuation (McCluskey & Anand, 1999; Rossini, 1997). The method has the advantages of being relatively inexpensive and easy to apply. Larger data-sets are necessary for testing and evaluation of neural networks and the black box is its main limitations. McGreal, et al. (1998) evaluated the ability of a neural network model to predict the value of properties using Belfast data as a case study. They concluded that the use of neural networks for appraisal purposes remains problematic. Though some vary close predictions were possible, others deviated appreciably from the sale price. To sum up, the fundamental theory of housing economics shows that product value is based on the interaction between demand and supply in the market. The frequent change of market equilibrium leads to the property value moving up and down. The problem of valuation is the uncertainty, which may occur for various reasons. It is thus necessary to identify the main attributes and how significant they are to the property value. Entropy was introduced to property valuation for determining which attributes contribute and Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 7

7 how much in Hong Kong (Ge, Lam, & Lam, 2005). In this work relationships between property value and attributes, are explored in the Auckland market. ENTROPY METHOD AND APPLICATIONAL PROCEDURE Property valuation has some uncertainty associated with it (French, 2007). Entropy measures the randomness or uncertainty of a random process (Pandurangan & Upfal, 2007). It is one of the methods for rating (Shannon & Weaver, 1963) and provides a quantitative measurement of this uncertainty (Kapur, 1989). Chan, et al. (1999) suggested that the method required less and more easily available information for ranking and was easy to apply. The method has been used in areas such as multiattribute decision-making (Zanakis, Solomon, Wishart, & Dublish, 1998) and under uncertainty (Dyer & Sarin, 1979). Entropy employs a mathematical algorithm to estimate priority of attributes. The steps of estimation are: a) to identify attributes through literature review; b) to collect and scale the relevant data; c) to calculate the expected value of each individual variable; and d) to rank attributes by Shannon (1963) entropy (S), i.e., a measure of uncertainty of a probability distribution; and d) to rank priority of each attributes. Consider a property value impacted by number of attributes, expression vector x=(x 1, x 2,, x n ) for N attributes and an observation x j for attribute j. The Shannon entropy of the attribute is calculated as S = N j= 1 P Log j 5 P j ) ( (1) Where P j is the probability of observing an intensity value of x j for attribute j defined as P j = x j N j = 1 / x (2) j S ranges from zero to log 5 (N), with the value 0 for property value expressed in not influencing and log 5 (N) for property value uniformly in all the interrogated attributes. Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 8

8 The research methodology for the application of entropy is shown in Figure 1. Figure 1: Research methodology of this research EMPIRICAL STUDY There were 6791 residential data sets collected from Real Estate Institute New Zealand (REINZ) and the Department of Statistics in New Zealand. Procedure of data collection Contact: Dr. Ge, Xin Janet on 9

9 is listed in Figure 2. Each data set includes one dependent variable, property price, and 18 independent variables which are illustrated in Table 2. The independent variables consist of both macro and micro economic factors. The macro economic factors include GDP, CGDP, CPI, CCPI, population, employment rate and unemployment rate; and the 11 microeconomic variables include land size, location, garden, bedroom, primary and secondary school net, access to shopping centre, coast and motor way. Data Collection Independent Variable Dependent Variable Macro-economics Factor Micro economics Factor Property Price (REINZ) Statistics New Zealand Real Estate Company (REINZ) Apply Council Property Rate Search Figure 2 Procedure of Data Collection Table 2 Dependent and Independent Variables Variable Types Abbreviation Variable Names Dependent variables PRICE Price Independent Variables AREA Land size (square metre) (Micro Economic Factors) LOCATION location GN Garden BED Number of bedrooms PSN Primary School Net HRPS High Reputation Primary School Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 10

10 Independent Variables (Macro Economic Factors) SSN HRSS DNC MWA DM GDP CGDP CPI CCPI POP EM UEM Secondary School Net High Reputation Secondary School District Nearby Coast Motor Way Access Shopping Mall available Gross Domestic Product (GDP) Construction GDP Consumer Price Index Construction CPI Population Number of Employment Number of Unemployment Quantitative data is required for applying Entropy method. The transformation for the qualitative data is listed in Table 3. Table 3: Quantification Methods of Each Variable Variables Quantification Methods GN 1 for houses with gardens and 0 if without. HRPS 1 for houses within district of high reputation primary school and 0 if not. HRSS 1 for houses within district of high reputation secondary school and 0 if not. DM 1 for houses close to shopping mall and 0 if not. MWA 1 for houses easily access to Motor way, 0 if not. DNC 1 for houses is nearby coast and 0 otherwise. The role of weights is to assist in the expression of the importance of each attribute in relation with the others, scaling of data is required. In order to apply the entropy method, all data should be scaled. Sekaran (2003) suggested that different cultures reacted differently to scaling issues. In this research, all variables are scaled by using the 5-point scale since many studies have indicated that the 5-point scale was able to efficiently capture the meaning of the subjective judgements (Chen, Jung, & Peacock, 1994; Jung & Jung, 2001; Luczak & Ge, 1989). Figure 3 defines the mode with respect to Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 11

11 significance of scaling. Figure 3: Definition of mode with respect to significance Some appropriate assumptions are made in order to scale the variables. Table 4 and 5 present the assumptions as well as scaling methods for each variable. Table 4- Assumptions for Scaling of Each Variable Variables Assumptions AREA It is assumed that the bigger land size, the higher the value. LOCATION It is assumed that the value is higher with closer to the CBD. GN It is assumed that the value is higher if there is a garden. BED It is assumed that the value is higher for more bedrooms. PSN It is assumed that the closer the primary school, the higher the value. HRPS It is assumed that the closer the high reputation primary school, the higher the value. SSN It is assumed that the closer the secondary school, the higher the value. HRSS It is assumed that the closer the high reputation second school, the higher the value. DNC It is assumed that the value is higher with nearby the coast. MWA It is assumed that the motor way entry nearby, the higher the value. DM It is assumed that the value is higher with closer to the shopping mall. GDP It is assumed that the higher the GDP, the higher the property value. CGDP It is assumed that the higher the CGNP, the higher the value. CPI It is assumed that the higher the CPI, the higher the value. CCPI It is assumed that the higher the CCPI, the higher the value. POP It is assumed that the great the population, the higher the value. EM It is assumed that the higher employment rate, the higher the value. UEM It is assumed that the higher the unemployment, the lower the value. Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 12

12 Table 5: Scaling of Variables m Variables Scale AREA < > DIST GN Without With 4 BED >6 5 PSN HRPS Without With 7 SSN HRSS Without With 9 DM Without With 10 MWA Without With 11 DNC Without With 12 GDP or below > CGDP 1423 or below > CCPI 1106 or below > CCPI 1176 or below > POP 4015 or below > EM or > below UEM or below > The scaled attributes are then applied to prioritize in the entropy method. The property value is affected by a variety of attributes. The ranking results measure whether those attributes are more significant than others shown in Table 6. The result shows that the number of bedrooms, CPI, location, Land size, and CPI (construction) are the top 5 most important attributes in relation with the residential property value in Auckland. The highest weight is bedroom with , and the lowest is garden with It implies that property value is affected by a combination of many attributes because each Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 13

13 of the attributes possesses a considerable weight. Table 6: Priority Ranking of Attributes Priority Ranking Attributes Normalization 1 BED CPI LOCATION LAND SIZE CCPI SSN PSN HRPS HRSS GDP POP EMP MWA DNC CGDP DM UEMP GN DISCUSSION AND CONCLUSION Many attributes impact on property value as shown by the priority ranking built in the Entropy method where we can find the contribution of each attribute. The findings of this study suggest that both macroeconomic and micro-attribute variables contribute to property value. The priority ranking of importance of macroeconomics variables are listed in the Table 7, which shows that CPI is the most important factor in terms of the macroeconomics factors while unemployment is the least important factor in terms of contribution to Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 14

14 property value. The overall weight of macroeconomics factors are Table 7: Priority of Ranking Macroeconomics Variables Macroeconomics Variables No Priority variables Ranking 1 CPI CCPI GDP POP EMP CGDP UEMP Total This result is similar to(paz, 2003). New Zealand S economy has been growing rapidly since Figure 4 shows consumer price index of Auckland with inflation reaching 4 percent from the June 2005 quarter to the June The most significant upward contributions came from the transportation group (up 10%) and the housing sector (up 4.8%). The annual increase in the CPI was influenced by higher prices for petrol (up 32.2%), as well as the purchase and construction of new dwellings (up 5.0%) (Bascand, 2006). Figure 4: Consumer Price Index (Resource: Statistics New Zealand) Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 15

15 The higher CPI leads to higher property values. The CPI construction has been increasing around 25% in the last 4 years (Figure 5). The higher construction cost and the higher energy consumption are the major contribution to increasing property values. It demonstrates that the CPI has a strong relationship with property value in Auckland. Figure 5: Consumer Price Index for Construction (Resource: Statistics New Zealand ) A considerable proportion of the GDP is related to property value and GDP is the most widely used measure of economic performance. The growth in the GDP is also underlining the viability of the housing market and lends support to the rising aspiration of home ownership. The construction activity, indicates the injection by housing into the GDP and Paz (2003) suggests that a high level of construction can raise the competition in the housing market and may lower the price of existing houses. Auckland attracts international immigration. The population in Auckland has been increasing rapidly since the economy started booming. The increase in population stimulates the construction activity and housing market both regarding new and existing prices (Paz, 2003) showing how a high demand for property induces growth of property value. Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 16

16 Employment and unemployment rate also contributes to the property market indirectly y affecting the demand for investment in the property market. The data shows the higher the employment, the higher the property value (Case & Shiller, 2003). The priority ranking of house related attributes is depicted in Table 8. The number of bedrooms, location and land size are the most important attributes determining property value in Auckland. Motorway access, distance to Coast, shopping mall and garden are the least important indicator. The total of weights of Micro-attributes is Table 8: Priority of Ranking Housing Related Factors Microeconomics Variables No Priority Variables Ranking 1 BEDROOMS LOCATION LAND SIZE SECONDARY SCHOOL NET PRIMARY SCHOOL NET REPUTABLE PRIMARY SCHOOL REPUTABLE SECONDARY SCHOOL MOTOR WAY ACCESS DISTANCE TO COAST DISTANCE TO SHOPPING GARDEN Total The research findings agree with those of Joslin (2005), Kauko (2003), Han, et al. (2002) and Lenk (1997). The number of bedrooms has the highest weight in this study. It is not difficult to Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 17

17 understand that residences with more space and living room fit families and help create a good quality of life command higher prices. More rooms mean higher construction costs, thus the higher the value of the property. Location is a very important factor for property value according to the literature. McCluskey et al (2000) contended that location is a basic attribute of a dwelling, which is a direct determinant of housing quality and market value. Significant differences in value can occur over short distances. In the Auckland housing market, location plays a significant role for determining property value. The closer to the CBD, the higher the value of property. People like to live close to CBD for convenience and to save time and costs for transport. The scarcity pushes up the price of property. The bigger the land size, the higher the value of a property. Households normally prefer a bigger leisure area. Furthermore, the bigger land size implies a potential for further improvement and land subdivision. According to the weigh the access to schools is an important determinant as suggested by previous studies (Han, Yu, Malone-Lee, & Basuki, 2002). It is obvious that households are willing to pay more for a property close to a school, particularly one with a high reputation. Kauko (2003) has researched externalities, and evaluated, in monetary terms, the effect of externalities. This results indicated that locational externalities includes not only very localized negative externalities (social, physical or visual ones), but also positive ones (services, greenery, status etc.). The result shows a similar weight for accessing to motor way, shopping center, and coast. Motor way provides a better link between a residential area and business activities without affecting of quality of living. The living near the coast could be a contribution to a high quality life style. Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 18

18 This paper has identified, using the Entropy method, the key attributes to property value, and provided a ranking and the weights of those attributes in the case of the Auckland property market of 18 attributes. It is concluded that the number of bedrooms, CPI, land size and location are the most important attributes in relation with property value. The Entropy method embodies an effective and convenient tool for property valuation incorporating local economic and demographic condition. REFERENCE Bascand, G. (2006). Consumers price index. Retrieved 20 September, 2006, from C869A3EAD5C9/0/consumerspriceindexjun06qtrhotp.pdf Blackley, D. M. (1999). The long-run elasticity of new housing supply in the United State: empirical evidence for 1950 to journal of Real Estate Finance and Economics, 18(1), Case, K. E., & Shiller, R. J. (2003). Is there a bubble in the housing market? Brookings Papers on Economic Activity, 2, Chan, L. K., Kao, H. P., Ng, A., & Wu, M. L. (1999). Rating the importance of customer needs in quality function deployment by fuzzy and entropy methods. International Journal of Production Research, 37(11), Chau, K. W., Wong, S. K., & Yiu, C. Y. (2004). The value of the provision of a balcony in apartments in Hong Kong. Property Management, 22(3), Chen, J. G., Jung, H. S., & Peacock, B. J. (1994). A Fuzzy Sets Modelling Approach for Ergonomic Workload Stress Analysis. Internaitonal Journal of Industrial Ergonomics, 13(3), pp Cheshire, P., & Sheppard, S. (1998). Estimating the demand for housing, land, and neighbourhood characteristics. Oxford Bulletin of Economics and Statistics, 60(3), Dotzour, M. G. (2001). An empirical analysis of the reliability of precision of the cost approach in residential appraisal. Journal of Real Estate Research, 5(1), Dyer, J. A., & Sarin, R. K. (1979). Measurable mylti-attribute value functions. Operations Research, 27, Federal National Mortgage Association (FNMA). (2007). Real estate appraisal [Electronic Version]. Wikipedia, the free encyclopedia, Retrieved 17 May 2007 from French, N. (2007). Valuation uncertainty - common professional standards and methods. Paper presented at the 13th Pacific-Rim Real Estate Society Conference, Fremantle, WA, Australia. French, N., & Gabrielli, L. (2004). The uncertainty of valuation. Journal of Property Investment & Finance, 22(6), Ge, X. J., Lam, K. Y., & Lam, K. C. (2005, September). Entropy method in the rank of attributes for property value. Paper presented at the Proceedings of the 3rd International Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 19

19 Structural Engineering and Construction Conference, Shunan, Japan Han, S. S., Yu, S. M., Malone-Lee, L. C., & Basuki, A. (2002). Dynamics of property value distribution in an Asian metropolis - the case of landed housing in Singapore, Journal of Property Investment & Finance, 20(3), Hoesli, M., Jani, E., & Bender, A. (2006). Monte Carlo simulations for real estate valuation. Journal of Property Investment & Finance, 24(2), Joslin, A. (2005). An investigation into the expression of uncertainty in property valuation. Journal of Property Investment & Finance, 23(3), Jung, H. S., & Jung, H. S. (2001). Establishment of Overall Workload Assessment Technique for Various Tasks and Workplaces. Internaitonal Journal of Industrial Ergonomics, 28, pp Kapur, J. N. (1989). Maximum-Entropy models in science and engineering. New York: Wiley. Kauko, T. (2003). Residential property value and locational externalities - on the complementarity and substitutability of approaches. Journal of Property Investment & Finance, 21(3), Lenk, M. M., Worzala, E. M., & Silva, A. (1997). High-tech valuation: should artificial neural networks bypass the human valuer? Journal of Property Valuation and Investment, 15(1), Luczak, H., & Ge, S. (1989). Fuzzy Modelling of Relations Between Physical Weight and Perceived Heaviness: The Effect of Size-Weight Illusion in Industrial Lifting Task. Ergonomics, 32, pp Mallinson, M., & French, N. (2000). Uncertainty in property valuation - the nature and relevance of uncertainty and how it might be measured and reported. Journal of Property Investment & Finance, 18(1), McCluskey, W. J., & Anand, S. (1999). The applicatin of intelligent hybrid techniques for the mass appraisal of residential properties. Journal of Property Investment & Finance, 17(3), McCluskey, W. J., Deddis, W. G., Lamont, I. G., & Borst, R. A. (2000). The application of surface generated interpolation models for the prediction of residential property values. Journal of Property Investment & Finance, 18(2), McGreal, S., Adair, A., McBurney, D., & Patterson, D. (1998). Neural networks: the prediction of residential values. Journal of Property Valuation & Investment, 16(1), Meen, G., & Andrew, M. (1998). Modelling regional house prices: a review of the literature. Reading: University of Reading. Pandurangan, G., & Upfal, E. (2007). Entropy-based bounds for online algorithms [Electronic Version]. Retrieved 05 May from Paz, P. T. (2003). Determinants of housing prices in Spanish cities. Journal of Property Investment & Finance, 21(2), Rosen, S. (1974). Hedonic prices and implicit markets: product differentiation in pure competition. Journal of Political Economy, 82(1), Rosiers, F. D., Theriault, M., & Villeneuve, P. Y. (2000). Sorting out access and neighbourhood factors in hedonic price modelling. Journal of Property Investment & Finance, 18(3), Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 20

20 Rossini, P. (1997). Artificial neural networks versus multiple regression in the valuation of residential property. Australian Land Economics Review, 3(2), Sekaran, U. (2003). Research methods for business: a skill building approach. New York: Wiley. Shannon, C. E., & Weaver, W. (1963). The mathematical theory of communication. Urbana: University of Illinois Press. Tse, R. Y. C., & Love, P. E. D. (2000). Measuring residential property values in Hong Kong. Property Management, 18(5), Valuation, I., & (IVSC), S. C. (2007, 12 Jan. 2006). International Valuation Standards Retrieved 05 June, 2007, from Wong, T. Y. J., Hui, C. M. E., & Seabrooke, W. (2003). The impact of interest rates upon housing prices: an empirical study of Hong Kong's market. Journal of Property Investment & Finance, 21(2), Zanakis, S. H., Solomon, A., Wishart, N., & Dublish, S. (1998). Multi-attribute decision making: a simulation comparison of select methods. European Journal of Operational Research, 107, Contact: Dr. Ge, Xin Janet on xinjanet.ge@uts.edu.au. 21