1# RECORD fb704039a7cb48f523fc8649b0477ed96bedc7f72# TITLE: Improved Methods for Predicting Property Prices in Hazard Prone Dynamic Markets3# YEAR: 2016.0 VENUE: Environmental and Resource Economics DOI: 10.1007/s10640-016-0076-54# AUTHORS: K. de Koning; T. Filatova; Okmyung Bin5# PAGES: 17 CANDIDATE TABLE PAGES (1-based): [5]6# ABSTRACT: Property prices are affected by changing market conditions, incomes and preferences of people. Price trends in natural hazard zones may shift significantly and abruptly after a disaster signalling structural systemic changes in property markets. It challenges accurate market assessments of property prices and capital at risk after major disasters. A rigorous prediction of property prices in this case should ideally be done based only on the most recent sales, which are likely to form a rather small dataset. Hedonic analysis has been long used to understand how various factors contribute to the housing price formation. Yet, the robustness of its assessment is undermined when the analysis needs to be performed on relatively small samples. The purpose of this study is to suggest a model that can be widely applicable and quickly calibrated in a changing environment. We systematically study four statistical models: starting from a typical standard hedonic function and gradually changing its functional specification by reducing the hedonic analysis to some basic property characteristics and applying kriging to control for neighbourhood effects. Across different sample sizes we find that the latter performs consistently better in the out-of-sample predictions than other traditional price prediction methods. We present the specific improvements to the traditional spatial hedonic model that enhance the model’s prediction accuracy. The improved model can be used to monitor price changes in risk-prone areas, accounting for changes in flood risk and at the same time controlling for autonomous market responses to flood risk.789===== PAGE 1 (score 5, numeric-paren 0, table-words 0, var-words 5) =====1011Environ Resource Econ (2018) 69:247–26312https://doi.org/10.1007/s10640-016-0076-513Improved Methods for Predicting Property Prices in14Hazard Prone Dynamic Markets15Koen de Koning116· Tatiana Filatova1 ·17Okmyung Bin218Accepted: 15 October 2016 / Published online: 26 October 201619© The Author(s) 2016. This article is published with open access at Springerlink.com20Abstract Property prices are affected by changing market conditions, incomes and prefer-21ences of people. Price trends in natural hazard zones may shift significantly and abruptly after22a disaster signalling structural systemic changes in property markets. It challenges accurate23market assessments of property prices and capital at risk after major disasters. A rigorous24prediction of property prices in this case should ideally be done based only on the most recent25sales, which are likely to form a rather small dataset. Hedonic analysis has been long used26to understand how various factors contribute to the housing price formation. Yet, the robust-27ness of its assessment is undermined when the analysis needs to be performed on relatively28small samples. The purpose of this study is to suggest a model that can be widely applicable29and quickly calibrated in a changing environment. We systematically study four statistical30models: starting from a typical standard hedonic function and gradually changing its func-31tional specification by reducing the hedonic analysis to some basic property characteristics32and applying kriging to control for neighbourhood effects. Across different sample sizes we33find that the latter performs consistently better in the out-of-sample predictions than other34traditional price prediction methods. We present the specific improvements to the traditional35spatial hedonic model that enhance the model’s prediction accuracy. The improved model36can be used to monitor price changes in risk-prone areas, accounting for changes in flood37risk and at the same time controlling for autonomous market responses to flood risk.38B Koen de Koning39k.dekoning@utwente.nl40Tatiana Filatova41t.filatova@utwente.nl42Okmyung Bin43BINO@ecu.edu44145Department of Governance and Technology for Sustainability (CSTM), University of Twente,46P.O. Box 217, 7500AE Enschede, The Netherlands47248DepartmentofEconomics,CenterforNaturalHazardsResearch,EastCarolinaUniversity,Greenville,49NC 27858, USA50123515253===== PAGE 5 (score 27, numeric-paren 0, table-words 3, var-words 19) =====5455Improved Methods for Predicting Property Prices in Hazard…5625157Table 1 Summary statistics of58the property attributes59Variables60Summary (N = 4779)61Mean62SD63Sales price64USD 156,6126587,35466Age of the house6722.4years681969Number of bedrooms703.2710.5972Total structure square feet7323917499375Lot size in acres760.64772.478Gas heating (= 1)790.35800.4881Fireplace (= 1)820.77830.4284Face brick (= 1)850.48860.5087Hard wood flood (= 1)880.25890.4390Good quality (= 1)910.031920.1793Vacant home (= 1)940.0048950.06996Distance to creek97854 feet9859699Distance to airport10033,966 feet10117,859102Distance to major road103135 feet10499105Distance to business centre1064632 feet1072452108Distance to railroad1095498 feet1106378111Distance to Tar River11220,999 feet11317,587114Distance to park1157490 feet1167051117Sold between Fran and Floyd (= 1)1180.341190.48120Sold after Floyd (= 1)1210.371220.47123Floodplain (= 1)1240.0641250.24126The area provides an excellent natural experiment setting for this study in that it had enjoyed127a period of relative calm, not experiencing major hurricane flooding since Hurricane Hazel in1281954, followed by two major hurricanes. Hurricane Fran (1996) produced millions of dollars129in property damages resulting from profuse rainfall, flash floods, and severe storm surge.130Three years later Hurricane Floyd (1999) brought torrential rains and record flooding which131resulted in one of the largest peacetime evacuations in U.S. history (Bin and Polasky 2004).132The data include information on sales price, property characteristics such as age and size,133dummy variables that represent the presence or absence of extra facilities of the property,134spatial information on the distance to amenities and disamenities, flood zoning, and time135of sales. The summary statistics of all the relevant property characteristics can be found in136Table 1.1372.2 Model Specifications138The hedonic price function of a property is given by139Footnote 1 continued140in spatial error models are constructed based on the assumptions of the user, whereas in kriging they are based141on the spatial structure of the error, which is defined in the construction of the semivariogram. The spatial142error model is particularly relevant for proper estimation of the coefficients in the hedonic price estimation,143whereas kriging focuses on the prediction of the dependent variable.144123145146147===== PAGE 6 (score 31, numeric-paren 0, table-words 5, var-words 21) =====148149252150K. de Koning et al.151Table 2 Input variables and their functional forms for the hedonic models152Variables153M1154M2155M3156M4157Age of the house158X + X2159X +160√161X162X163X164Number of bedrooms165X + X2166X +167√168X169X170X171Lot size in acres172X + X2173X +174√175X176√177X178√179X180Total structure square feet181X + X2182X + ln(X)183ln(X)184ln(X)185Gas heating (=1)186X187X188Fireplace (=1)189X190X191Face brick (=1)192X193X194Hard wood flood (=1)195X196X197Good quality (=1)198X199X200Vacant home (=1)201X202X203Log of distance to creek204X205X206Kriging207Log of distance to airport208X209X210Kriging211Log of distance to major road212X213X214Kriging215Log of distance to business centre216X217X218Kriging219Log of distance to railroad220X221X222Kriging223Log of distance to Tar River224X225X226Kriging227Log of distance to park228X229X230Kriging231Sold between Fran and Floyd (=1)232X233X234Sold after Floyd (=1)235X236X237Floodplain (=1)238X239X240X241X242Floodplain × sold btw Fran and Floyd243X244X245Floodplain × sold after Floyd246X247X248A variable ‘Schools’ has not been included in the analysis since the school quality in this particular areas249is rather homogeneous. Furthermore, school rating does not have statistically significant effect on property250prices251ln Pi = β0 +252K253254k=1255βkxi256k + Ei257(2)258where ln Pi is the natural log of the sales price of property i, β0 is the intercept, βk is the259coefficient for each property characteristic k, xi260k is the value of characteristic k of property i,261and Ei is the residual of the predicted property price. Our objective is to identify the model that262provides the smallest prediction errors in the out-of-sample predictions for various sample263sizes. The models to be compared include:264– A spatial hedonic model from Bin and Landry (2013) (M1)265– An adjusted version of M1 with different functional forms (M2)266– M2 with a reduced number of input variables (M3)267– M3 whereby spatial variability in property prices is predicted with kriging (M4)268The hedonic analysis is used to estimate the coefficients of the input variables βk (Eq.2692). In M4 hedonic analysis is used to understand the influence of the core spatial variable270of interest (flood risk) on property prices while the rest of the spatial variability in prices271is captured by interpolating the residuals (Ei, Eq. 2) using kriging. A list of all the input272variables can be found in Table 2. M1 is same model with the same specifications as used273123274