1# RECORD 819ec939f0241cca8513d3a72e45a56254da737d2# TITLE: House Prices and Flood Risk Exposure: An Integration of Hedonic Property Model and Spatial Econometric Analysis3# YEAR: 2022.0 VENUE: Journal of real estate finance and economics DOI: 10.1007/s11146-022-09930-z4# AUTHORS: Nam Bui; Le Wen; B. Sharp5# PAGES: 32 CANDIDATE TABLE PAGES (1-based): [14, 15, 16, 17, 18, 19, 22, 24, 25, 26]6# ABSTRACT: The impact of climate change is devastating in developing countries where flood protection and insurance schemes are limited. Certain parts of Ho Chi Minh City, Vietnam are under the constant threat of inundation due to sea-level rise. We integrate the hedonic property model in a difference-in-differences framework and spatial econometric analysis into a single analytical framework to estimate the economic effect of pluvial flooding. We find prices for affected houses were discounted by 9% after a large flood event on 30 September 2017. This research contributes to the existing literature as follows. First, we study the economic impact of pluvial floods, which has received less attention in existing studies where large and irregular floods are their focus. Second, the inclusion of legal status as a control variable accounts for the unique character of the Vietnamese housing market. Third, we also identify the recovery of house prices after the flood event.789===== PAGE 1 (score 9, numeric-paren 0, table-words 0, var-words 9) =====1011Vol:.(1234567890)12The Journal of Real Estate Finance and Economics (2024) 69:100–13113https://doi.org/10.1007/s11146-022-09930-z141 315House Prices and Flood Risk Exposure: An Integration 16of Hedonic Property Model and Spatial Econometric 17Analysis18Nam Bui1 · Le Wen2 · Basil Sharp219Accepted: 31 October 2022 / Published online: 15 November 2022 20© The Author(s) 202221Abstract22The impact of climate change is devastating in developing countries where flood 23protection and insurance schemes are limited. Certain parts of Ho Chi Minh City, 24Vietnam are under the constant threat of inundation due to sea-level rise. We inte-25grate the hedonic property model in a difference-in-differences framework and 26spatial econometric analysis into a single analytical framework to estimate the eco-27nomic effect of pluvial flooding. We find prices for affected houses were discounted 28by 9% after a large flood event on 30 September 2017. This research contributes 29to the existing literature as follows. First, we study the economic impact of pluvial 30floods, which has received less attention in existing studies where large and irregu-31lar floods are their focus. Second, the inclusion of legal status as a control variable 32accounts for the unique character of the Vietnamese housing market. Third, we also 33identify the recovery of house prices after the flood event.34Keywords House price · Flood risk · Hedonic property model · Spatial analysis · 35Vietnam36Introduction37Sea level rise is a constant threat to many shorelines around the world, especially 38in dense urban areas. Vojinović (2015) differentiates pluvial floods from fluvial 39and coastal ones. Fluvial floods break or overflow levees and enter the city while 40coastal floods follow storm surges and enter the city by sea or delta. Pluvial floods 41happen in large towns where the amount of rainfall is greater than the ability of 42 * Le Wen 43 44l.wen@auckland.ac.nz451 46Department of Economics, University of Auckland, 1142 Auckland, New Zealand472 48Energy Centre, Department of Economics, University of Auckland, 1142 Auckland, 49New Zealand505152===== PAGE 14 (score 18, numeric-paren 0, table-words 2, var-words 2) =====5354113551 356House Prices and Flood Risk Exposure: An Integration of Hedonic…57^ standard deviation58Table 4 (continued)59Pre-flood listings60Post-flood listings61Control group62Treatment group63Post-treatment group64Mean65sd^66Mean67sd^68Mean69sd^70Mean71sd^72Mean73sd^74 3 (pink book)750.387760.487770.369780.483790.367800.482810.419820.494830.363840.48285Direction86 0 (unknown)870.628880.483890.725900.447910.719920.449930.510940.500950.558960.49897 1 (north)980.064990.2441000.0301010.1711020.0341030.1821040.0961050.2961060.0681070.253108 2 (northeast)1090.0381100.1901110.0221120.1461130.0281140.1661150.0371160.1891170.0111180.102119 3 (east)1200.0391210.1941220.0401230.1951240.0391250.1931260.0411270.1981280.0261290.160130 4 (southeast)1310.0771320.2661330.0511340.2201350.0621360.2421370.0741380.2621390.0791400.270141 5 (south)1420.0861430.2801440.0671450.2501460.0671470.2511480.1111490.3151500.1161510.321152 6 (southwest)1530.0201540.1401550.0111560.1061570.0131580.1111590.0281600.1651610.0211620.144163 7 (west)1640.0301650.1701660.0241670.1521680.0211690.1431700.0481710.2141720.0371730.189174 8 (northwest)1750.0191760.1371770.0311780.1741790.0161800.1251810.0541820.2261830.0841840.278185 LONG_X186106.7161870.015188106.7171890.015190106.7171910.017192106.7151930.008194106.7171950.008196LAT_Y19710.7351980.01319910.7362000.01320110.7352020.01520310.7352040.00320510.7342060.003207 Observations2081,3052091,0622101,828211539212190213214215===== PAGE 15 (score 31, numeric-paren 6, table-words 9, var-words 6) =====216217114218 219N. Bui et al.2201 3221prices among the treatment group from before to after the event. Because these 222houses lie within the radius of the affected roads, their prices were reduced 1.6% 223(= 0.075—0.091).224Last, based on the coefficient of the DD term, house prices among the treatment 225group experienced a discount by 9.1% compared to the control group from before to 226after the event. A difference-in-differences plot is used to show the parallel trends 227(Di Maggio et al., 2019). Thus, Fig. 3 is provided to illustrate the results of the dif-228ference-in-differences estimation for the flood event on 30 September 2017.229Results in Table 6 reveal that house prices are determined by house characteris-230tics. For example, house prices are expected to increase with proximity to the city 231centre. It is desirable to stay far away from traffic congestion spots, nearer to large 232parks and schools, and inside secured areas. The directions views to west and south-233west boost house prices. Listings in the red book, which indicates houses with land 234use rights and ownership, are expected to command a higher price.235The current interpretation of direct impacts can be improved further. To identify 236the total effect of flood risk on house prices, we consider the spillover effect which 237assumes that house prices and/or flood risk of a certain property may have an impact 238on nearby ones. As discussed previously, the SAC model is used to interpret the 239impact. Thus, we continue to use the SAC model to adjust the price discount inter-240pretation under the spillover effects.241The decomposition estimation results are reported in Table 7 where the spillover 242and total effects of all variables are significant and concur with the fact that sig-243nificant coefficients of the spatial weight matrix W in the case of the SAC model as 244shown in Table 5. Our empirical evidence in Table 7 is consistent with the conclu-245sion drawn by Delgado and Florax (2015) in which an appropriate spatial estimator 246should be used when both spatial sorting and spatial interaction exist in treatment 247responses.248Direct and indirect effects indicate the rate of price discount that houses in the 249flooded areas were facing. Separating indirect effects from total effects helps to dif-250ferentiate between how much the reduction is from being affected by the floods and 251how much it is caused by staying near inundated areas. Understanding this can aid 252when formulating a tax policy. The amount accounted for a direct effect could be 253Table 5 “Move in tandem” test254Dependent Variable: Natural logarithm of house prices.t statistics in 255parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. This table illustrates 256the “move in tandem” test suggested by Gertler et al. (2016). Results 257show that house prices in both the control and treatment group were 258increasing before the event. The variable Time is measured by using 259the listing date of properties260Treatment Group261Control Group262Time2630.002**2640.005***265(2.12)266(5.55)267Constant268-40.722*269-87.901***270(-1.76)271(-5.09)272Observations2736122741,104275276277===== PAGE 16 (score 151, numeric-paren 122, table-words 1, var-words 11) =====2782791152801 3281House Prices and Flood Risk Exposure: An Integration of Hedonic…282Table 6 Spatial estimation results from alternative spatial models (July 2017—January 2018)283Models284OLS285SEM286SAR287SLX288SDM289SDEM290SAC291Equation (2)292Equation (3)293Equation (4)294Equation (5)295Equation (6)296Equation (7)297Equation (8)298Direct effect299 D_POST3000.089***3010.082***3020.087***3030.092***3040.099***3050.067***3060.075***307(3.66)308(3.90)309(3.57)310(3.79)311(4.01)312(3.34)313(3.70)314 D_TR3150.214***3160.190***3170.213***3180.082*3190.104**3200.0293210.184***322(5.01)323(4.51)324(5.02)325(1.83)326(2.21)327(0.66)328(4.52)329 D_POST X D_TR330-0.139***331-0.155***332-0.119**333-0.116**334-0.123**335-0.079*336-0.091**337(-2.62)338(-3.43)339(-2.23)340(-2.19)341(-2.31)342(-1.84)343(-2.08)344 CBD (km)345-0.129***346-0.195***347-0.139***348-0.112***349-0.127***350-0.138***351-0.191***352(-6.31)353(-5.95)354(-6.79)355(-5.56)356(-5.76)357(-4.75)358(-6.54)359 Traffic_congestion (km)3600.067**3610.070*3620.094***3630.052**3640.076**3650.0523660.163***367(2.52)368(1.73)369(3.42)370(1.97)371(2.50)372(1.44)373(4.42)374 Nearest_river (km)375-0.041376-0.274***3770.013378-0.030379-0.044380-0.0313810.067382(-0.81)383(-3.79)384(0.24)385(-0.56)386(-0.81)387(-0.41)388(0.91)389 Large_parks (km)390-0.144***391-0.372***392-0.177***393-0.105***394-0.128***395-0.322***396-0.483***397(-4.58)398(-8.40)399(-5.46)400(-3.25)401(-3.65)402(-7.48)403(-11.72)404 Hospital (1 km)4050.0204060.0324070.026408-0.160***409-0.152***410-0.0454110.117***412(0.49)413(0.70)414(0.62)415(-3.21)416(-3.03)417(-0.73)418(2.70)419 School (1 km)4200.314***4210.370***4220.332***4230.355***4240.365***4250.372***4260.355***427(11.73)428(10.74)429(12.24)430(12.45)431(12.49)432(9.62)433(11.03)434 Area (m2)4350.006***4360.005***4370.006***4380.006***4390.006***4400.004***4410.004***442(33.09)443(29.35)444(32.93)445(32.98)446(32.96)447(28.22)448(28.67)449 Number_of_bedrooms4500.0174510.039***4520.0184530.0224540.0224550.057***4560.057***457(1.05)458(2.77)459(1.09)460(1.32)461(1.38)462(4.27)463(4.23)464 Number_of_bathrooms4650.091***4660.049***4670.084***4680.074***4690.072***4700.0074710.014472(5.76)473(3.63)474(5.30)475(4.68)476(4.56)477(0.56)478(1.06)479480481===== PAGE 17 (score 123, numeric-paren 98, table-words 1, var-words 7) =====482483116484 485N. Bui et al.4861 3487Table 6 (continued)488Models489OLS490SEM491SAR492SLX493SDM494SDEM495SAC496Equation (2)497Equation (3)498Equation (4)499Equation (5)500Equation (6)501Equation (7)502Equation (8)503 Floor_level504-0.010***505-0.006***506-0.010***507-0.009***508-0.009***509-0.005***510-0.005***511(-7.01)512(-4.86)513(-6.63)514(-6.52)515(-6.34)516(-3.83)517(-3.97)518 FRW (m)5190.003***5200.002***5210.002***5220.002**5230.002**5240.0015250.001*526(3.13)527(3.02)528(2.95)529(2.52)530(2.53)531(1.38)532(1.70)533 Secured_area5340.172***5350.143***5360.166***5370.156***5380.156***5390.104***5400.116***541(5.26)542(5.24)543(5.08)544(4.83)545(4.84)546(3.98)547(4.37)548 Balcony_or_Terrace549-0.0205500.022551-0.023552-0.046553-0.046554-0.0045550.007556(-0.47)557(0.62)558(-0.56)559(-1.12)560(-1.10)561(-0.12)562(0.21)563 Fiber_Internet564-0.147***565-0.112***566-0.146***567-0.127***568-0.126***569-0.099***570-0.109***571(-4.12)572(-3.79)573(-4.12)574(-3.62)575(-3.58)576(-3.50)577(-3.80)578 Fully_furnished579-0.014580-0.012581-0.008582-0.038583-0.034584-0.0115850.006586(-0.38)587(-0.37)588(-0.21)589(-1.01)590(-0.92)591(-0.35)592(0.19)593 Garden5940.0505950.0375960.0535970.0545980.0525990.0326000.033601(1.20)602(1.06)603(1.27)604(1.32)605(1.28)606(0.98)607(0.98)608 Swimming_pool609-0.314***610-0.286***611-0.294***612-0.226***613-0.228***614-0.135***615-0.185***616(-7.88)617(-8.38)618(-7.33)619(-5.57)620(-5.61)621(-3.98)622(-5.49)623 Washing_machine6240.084**6250.094***6260.081**6270.074**6280.073*6290.074**6300.080***631(2.25)632(3.00)633(2.17)634(1.99)635(1.96)636(2.48)637(2.62)638Legal status639 1(building permit)6400.001641-0.012642-0.009643-0.015644-0.013645-0.056646-0.042647(0.01)648(-0.24)649(-0.16)650(-0.27)651(-0.22)652(-1.20)653(-0.90)654 2(red book)6550.176***6560.162***6570.136***6580.0676590.064660-0.0186610.042662(3.61)663(3.95)664(2.74)665(1.33)666(1.28)667(-0.45)668(1.04)669 3 (pink book)6700.186***6710.114***6720.174***6730.155***6740.155***6750.051**6760.075***677(6.42)678(4.65)679(6.01)680(5.38)681(5.36)682(2.17)683(3.16)684685686===== PAGE 18 (score 123, numeric-paren 106, table-words 1, var-words 2) =====6876881176891 3690House Prices and Flood Risk Exposure: An Integration of Hedonic…691Table 6 (continued)692Models693OLS694SEM695SAR696SLX697SDM698SDEM699SAC700Equation (2)701Equation (3)702Equation (4)703Equation (5)704Equation (6)705Equation (7)706Equation (8)707Direction708 1 (north)7090.165***7100.084*7110.142***7120.127**7130.123**7140.0147150.028716(3.10)717(1.88)718(2.66)719(2.40)720(2.34)721(0.34)722(0.64)723 2 (northeast)7240.0947250.111**7260.0737270.0877280.0897290.0537300.053731(1.44)732(2.05)733(1.13)734(1.35)735(1.38)736(1.03)737(1.01)738 3 (east)739-0.0847400.032741-0.120**742-0.133**743-0.133**744-0.069745-0.049746(-1.45)747(0.67)748(-2.04)749(-2.27)750(-2.28)751(-1.46)752(-1.04)753 4 (southeast)7540.0767550.086**7560.0627570.0527580.0527590.0557600.056761(1.64)762(2.20)763(1.33)764(1.13)765(1.13)766(1.47)767(1.47)768 5 (south)7690.108**7700.0397710.100**7720.086**7730.084*7740.0037750.016776(2.50)777(1.07)778(2.32)779(2.00)780(1.95)781(0.10)782(0.46)783 6 (southwest)7840.281***7850.195***7860.260***7870.230***7880.230***7890.116*7900.126*791(3.25)792(2.71)793(3.01)794(2.69)795(2.69)796(1.70)797(1.80)798 7 (west)7990.255***8000.241***8010.211***8020.196***8030.193***8040.108**8050.130**806(3.70)807(4.21)808(3.02)809(2.83)810(2.80)811(1.96)812(2.32)813 8 (northwest)8140.119*8150.140**8160.0898170.0418180.0398190.0618200.069821(1.68)822(2.36)823(1.26)824(0.58)825(0.56)826(1.07)827(1.20)828Constant8297.911***8308.894***8317.906***8327.848***8337.906***8348.435***8358.568***836(53.84)837(42.76)838(53.95)839(53.91)840(53.13)841(44.65)842(45.53)843WY844W*845 e.Ln_Price8462.415***8472.581***8482.581***849(95.11)850(63.62)851(63.47)852 Ln_Price8530.022***8540.0268550.079***856(3.84)857(1.58)858(13.76)859860861===== PAGE 19 (score 45, numeric-paren 25, table-words 7, var-words 2) =====862863118864 865N. Bui et al.8661 3867Table 6 (continued)868Models869OLS870SEM871SAR872SLX873SDM874SDEM875SAC876Equation (2)877Equation (3)878Equation (4)879Equation (5)880Equation (6)881Equation (7)882Equation (8)883WX884W*885 D_POST886-0.431***887-0.715***888-0.039889(-3.52)890(-3.29)891(-0.24)892 D_TR8930.932***8940.614**8951.708***896(5.65)897(2.34)898(11.99)899 D_POST X D_TR900-0.1809010.313902-1.863***903(-0.45)904(0.61)905(-4.98)906 / var(e.lnprice)9070.255***9080.177***9090.254***9100.247***9110.247***9120.160***9130.164***914(15.54)915(12.51)916(15.61)917(15.37)918(15.45)919(12.01)920(12.53)921 Pseudo R29220.6479230.5989240.6479250.6589260.6569270.5719280.599929 AIC9303552.29312692.29323539.89333484.49343484.19352522.89362461.1937 BIC9383748.39392894.19403741.89413697.99423703.39432730.59442680.4945 Log likelihood946-1742.1947-1311.1948-1734.9949-1705.2950-1704.0951-1225.4952-1192.6953 Chi-squared9544337.19553981.59564368.39574549.29584547.99595139.09605379.5961 Wald test of spatial terms962096309640965096609670968 Observations9692,3679702,3679712,3679722,3679732,3679742,3679752,367976Dependent Variable: Natural logarithm of house prices. t statistics in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. This table presents the direct effect of flood risk 977on house prices. The first column is the standard linear model (OLS) based on Eq. (2). The rest of spatial models are established based on Eqs. (3) to (8). We used the 978Moran’s I test to show that this model is not suitable for interpreting the effect of flood risk on house prices. The model SAC has the lowest AIC/BIC values and is used 979for further consideration. The part of the table regarding the spatial weight matrix W shows the available coefficients of the spatial weight matrices depending on which 980type of models. The formulae in Section 2.2 help understand why W coefficients differ across the models. See Appendix Table A1 for explanations regarding these spatial 981estimates982983984===== PAGE 20 (score 25, numeric-paren 0, table-words 9, var-words 7) =====9859861199871 3988House Prices and Flood Risk Exposure: An Integration of Hedonic…989attributed to real estate companies that built properties on exiting waterways of the 990city. This is compensation for harming the environment and affecting people’s lives. 991On the other hand, tax revenue equal to the indirect effect might be used for inform-992ing the residents of flood risk areas, thus raising the awareness of flood risk.993Therefore, house prices increase 7.7% from before to after the event among the 994control group relative to a 1.7% (= 0.077—0.094) decrease in house prices among 995the treatment group during the same time span. Converting from the logarithm form, 996the effect of the pluvial flood on house prices was reflected by a discount by 9% of 997the treatment group compared to the control group from before to after the event. 998This discount rate could be used as a foundation for proposing an environmental 999tax (more details are in conclusion). Our results (9% price discount) are consist-1000ent with results obtained from other studies (e.g. 5.7%—Bin and Polasky (2004), 10018.6%—Posey and Rogers (2010), 10.3%—Rabassa and Zoloa (2016), 6.7%—Ram-1002baldi et al. (2013), and 6.2%—Samarasinghe and Sharp (2010)).1003Price Recovery After the Flood1004More analyses are carried out to examine whether houses regain their value. 1005The original dataset used in Table 5 is added with more observations for several 1006months after the flood. Specifically, compared to the original D_POST in Table 5, 1007the new variable DPOST_5m in Table 8 keeps the same three months before the 1008event (value = 0 for the date of listing from 01-Jul-2017 to 30-Sep-2017) and adds 1009two more months after the event (value = 1 for 01-Nov-2017 and 01-Apr-2018). 1010Similarly, DPOST_6m, DPOST_7m, DPOST_8m, DPOST_9m, and DPOST_10m 1011include data 6 months, 7 months, 8 months, 9 months, and 10 months respectively 1012after the flood. The housing market was inactive as Vietnamese people were taking a 1013break for Lunar New Year; hence the disappearance of DPOST_4m.1014Table 8 shows the recovery of house prices after the flood. Table 8-Panel A repre-1015sents the direct effect in the SAC model while Table 8-Panel B reports the spillover 1016and total effects. The positive and significant spillover effects indicate house prices 1017are not only affected locally but also by neighbouring inundated areas. Results in 1018Table 8-Panel B help to conclude that the negative effect from the flood event on 1019house prices slowly diminishes as the houses regain their value from month to 1020month. Based on the total effects, houses in the treatment group increased their 1021prices gradually from 4.6% to 5.1% compared to those in the control group, if the 1022duration of after-effect months is extended from 3 to 7–10 months.1023Figure 4 shows house price changes for the control and treatment groups after 1024the flood event on 30 September 2017. First, we find that house prices (red line) in 1025the treatment group are higher than those (black line) in the control group. This is 1026because houses in the treatment group are close to the city centre, with easy access 1027to the hospital, inside secured areas and are more likely to list with pink book-the 1028certification for both land use rights and house ownership based on the summary 1029statistics by sub-groups in Table 4. Second, in the first four months after the flood 1030event, house prices were reduced in the treatment group and increased in the control 103110321033===== PAGE 22 (score 37, numeric-paren 15, table-words 3, var-words 14) =====1034103512110361 31037House Prices and Flood Risk Exposure: An Integration of Hedonic…1038regarding these properties have potential risks of future disputes. Moreover, obtain-1039ing only a building permit is not enough to significantly increase the price.1040The views to the west, and southwest were believed to be the best direction to 1041have for a home. The north view looks towards the energetic city centre, while the 1042west and southwest views show the peaceful green reserve of HCMC—the value-of-1043a-view perspective (Luttik, 2000; Samarasinghe & Sharp, 2008).1044Possible explanations for the negative sign of the swimming pool variable include 1045the concerns for either hygiene (in shared pools of apartments) and high mainte-1046nance costs (in villas or townhouses). Vietnamese house buyers realised the incon-1047venience of a swimming pool and later preferred a playground for their children or 1048an in-house garden for their parents.1049The sign of the floor level variable also differs from expectation. In contrast to 1050a Westerner’s perspective, Vietnamese like to reside in lower floors for the con-1051venience of mobility. The structure of Vietnamese households, in general, includes 1052children and elderly parents who would feel safer when they can stay closer to the 1053ground.1054Unlike other countries, installed fibre connection surprisingly has a negative 1055impact on house prices. Most building contractors in Vietnam work with an internet 1056provider to set up the fibre system. House owners need to connect to a predestined 1057provider; therefore, it restricts their freedom of choice of providers. In the recent 1058years, the freedom of choice has been improved for other new projects. Further 1059research need to be done to validate the outcome of this situation.1060Although this study adopts the basic DD model by Atreya and Ferreira (2015), 1061we have further developed the DD model by integrating spatial models. In doing so, 1062we can examine the spillover effect, i.e. if residential property prices are affected 1063by neighbouring house prices and if flood event impact on a certain zone might 1064influence house prices of bordering zones. Since this paper studies the impact of 1065a flood event on house prices and the model does not control for the inundated 1066property directly as done in Atreya and Ferreira (2015); thus, results are different 1067from those of Atreya and Ferreira (2015) in which inundated floodplain properties 1068were discounted by 48% immediately after the flood. We find that the effect of the 1069Table 7 SAC decomposition 1070estimation results (July 2017—1071January 2018)1072Dependent Variable: Natural logarithm of house prices. t statistics 1073in parentheses.* p < 0.10, ** p < 0.05, *** p < 0.01. This table focuses 1074on the direct, indirect (spillover) and total effects of the variable D_1075POST and the interaction term D_POST × D_TR in the SAC model. 1076These values are used for our final interpretation of flood risk impact 1077on house prices1078SAC model1079Direct effect1080Spillover effect1081Total effect1082D_POST10830.075***10840.003***10850.077***1086(3.70)1087(3.59)1088(3.70)1089D_TR10900.184***10910.007***10920.191***1093(4.52)1094(4.35)1095(4.52)1096D_POST × D_TR1097-0.091**1098-0.003**1099-0.094**1100(-2.08)1101(-2.09)1102(-2.08)110311041105===== PAGE 23 (score 7, numeric-paren 0, table-words 0, var-words 7) =====110611071221108 1109N. Bui et al.11101 31111pluvial flood on house prices among the treatment group experienced a discount by 11129% compared to the control group from before to after the event. We also study if 1113houses retain their value after a temporary reduction caused by the flood, which has 1114not been included in Atreya and Ferreira (2015).1115Residents in HCMC usually experience a typical six-month rainy season, and 1116pluvial flooding events tend to happen in the last three months. Therefore, we select 1117the four months after the flood event as the post period. However, it may raise the 1118concern that the short period – four months after the event- may not be sufficient to 1119isolate the impact of flood risk on house prices due to several plausible alternative 1120explanations from the demand and supply sides of the housing market. For instance, 1121potential buyers might want to postpone purchasing homes in an area affected by 1122flooding due to other reasons such as access, unavailability of labour to do neces-1123sary repairs, and sellers needing time to repair the affected homes. Moreover, sell-1124ers who sold homes straight after the flood event could be the liquidity-constrained 1125sellers who needed to sell as quickly as possible. For example, these sellers could be 1126the ones who had committed to purchasing another house and were therefore forced 1127to sell at low prices in a bad market. In our future research, the propensity score 1128matching technique (Papadogeorgou et al., 2019) can be employed to account for 1129selection bias in affected houses in the housing market by using houses with similar 1130characteristics.1131Conclusion1132This study sets out to determine the impact of a large pluvial flood event on house 1133prices within the southern area of HCMC, Vietnam. The results of this investigation 1134show that houses exposed to the flood event experienced a significant price discount 1135by 9% after a large flood event on 30 September 2017.1136The paper extends our knowledge by estimating the impact of a pluvial flood 1137event on house prices in Vietnam. It contributes to the literature by shifting the focus 1138towards the context of developing countries and challenging the issue of data availa-1139bility. In the past, pluvial floods have received little attention, even though the cumu-1140lative costs from constantly disrupting the people’s daily travel are relatively high 1141compared to a single large flood. This is the reason why we choose pluvial floods to 1142be the main focus of the present study. The methodology of the integrated hedonic 1143model and spatial economic approach into a single analytical framework is the first 1144attempt to be applied to estimate the economic impact of pluvial floods, and it can 1145be applied extensively. The findings of the study provide insights to design effective 1146insurance policies.1147The evidence from this study can be used as a reference for policymakers to 1148design a flood insurance scheme. Without sufficient capital to properly repair the 1149houses after the flood, owners can become even more vulnerable. The scheme can 1150help financially-difficult people in high-risk zones. The premium price could also 1151act as a source of risk information for both owners and sellers.1152Blocking the existing channels with new residential areas led to frequent flooding. 1153Now being constantly under the threat of inundation, even when it is not raining, the 115411551156===== PAGE 24 (score 97, numeric-paren 82, table-words 2, var-words 2) =====1157115812311591 31160House Prices and Flood Risk Exposure: An Integration of Hedonic…1161Table 8 Price recovery after the flood (SAC estimation results)11625 months11636 months11647 months11658 months11669 months116710 months1168Panel A – Direct effect1169 D_TR = 111700.324***11710.306***11720.300***11730.323***11740.320***11750.319***1176(12.90)1177(12.54)1178(12.18)1179(13.87)1180(13.96)1181(13.98)1182 D_POST_5m = 111830.031*1184(1.68)1185 D_TR = 1 X D_POST_5m = 111860.0301187(1.28)1188 D_POST_6m = 111890.031*1190(1.68)1191 D_TR = 1 X D_POST_6m = 111920.0361193(1.59)1194 D_POST_7m = 111950.035*1196(1.89)1197 D_TR = 1 X D_POST_7m = 111980.045**1199(1.96)1200 D_POST_8m = 112010.042**1202(2.30)1203 D_TR = 1 X D_POST_8m = 112040.051**1205(2.25)1206 D_POST_9m = 112070.045**1208(2.46)1209 D_TR = 1 X D_POST_9m = 112100.050**1211(2.25)1212 D_POST_10m = 112130.045**1214(2.50)1215 D_TR = 1X D_POST_10m = 112160.050**1217(2.23)1218W*1219 Ln_Price12200.045***12210.049***12220.051***12230.056***12240.057***12250.057***1226(15.43)1227(18.82)1228(19.95)1229(23.88)1230(24.75)1231(24.85)1232 e.Ln_Price12333.170***12343.434***12353.616***12364.940***12375.336***12385.376***1239(55.00)1240(96.01)1241(79.90)1242(52.30)1243(58.82)1244(60.17)1245 / var(e.Ln_Price)12460.084***12470.085***12480.087***12490.086***12500.086***12510.086***1252(43.47)1253(46.84)1254(49.24)1255(53.10)1256(55.29)1257(55.59)1258 Pseudo R212590.74712600.75112610.75212620.76412630.76812640.7701265 Log likelihood1266-688.11267-823.91268-965.21269-1082.71270-1171.91271-1173.91272 Chi-squared127330,498.7127443,989.7127548,711.41276105,231.31277130,008.81278137,067.81279 Wald test of spatial terms1280012810128201283012840128501286Panel B – Spillover and total effects1287Spillover effect1288 D_TR12890.007***12900.008***12910.007***12920.008***12930.008***12940.008***1295(10.37)1296(10.98)1297(11.14)1298(12.69)1299(12.83)1300(12.81)130113021303===== PAGE 25 (score 73, numeric-paren 57, table-words 3, var-words 4) =====130413051241306 1307N. Bui et al.13081 31309city is considering a four-billion-USD flood prevention project (Vietnamnet, 2018). 1310Housing projects built on top the channels should be taxed based on the converted 1311land area to fund flood protection infrastructure.1312Another important practical implication is that estimates of price discount due 1313to flood risk could be used to justify flood prevention policies by the government. 1314Many flood protection measures in HCMC are criticised as to their cost-effective-1315ness. For example, the city hired an expensive pump for removing surplus water on 1316Nguyen Huu Canh Street (Huu, 2018). Therefore, further research around the costs 1317and benefits of different flood protection measures is worthy of being investigated.1318Dependent Variable: Natural logarithm of house prices. t statistics in parentheses. * p < 0.10, ** p < 0.05, 1319*** p < 0.01. ~ Nm denotes the corresponding month, here, N = [5, 6, 7, 8, and 9]. The model includes 1320other control variables, such as, location, structure, neighbourhood, facilities, direction and legal status1321Table 8 (continued)13225 months13236 months13247 months13258 months13269 months132710 months1328 D_POST_Nm~13290.001*13300.001*13310.001*13320.001**13330.001**13340.001**1335(1.67)1336(1.67)1337(1.88)1338(2.29)1339(2.45)1340(2.49)1341 D_TR × D_POST_Nm13420.00113430.00113440.001*13450.001**13460.001**13470.001**1348(1.27)1349(1.59)1350(1.96)1351(2.24)1352(2.24)1353(2.23)1354Total effect1355 D_TR13560.331***13570.314***13580.308***13590.331***13600.328***13610.327***1362(12.93)1363(12.56)1364(12.21)1365(13.90)1366(13.99)1367(14.01)1368 D_POST_Nm13690.032*13700.032*13710.036*13720.043**13730.046**13740.046**1375(1.68)1376(1.68)1377(1.89)1378(2.30)1379(2.46)1380(2.50)1381 D_TR × D_POST_Nm13820.03013830.03713840.046*13850.052**13860.051**13870.051**1388(1.28)1389(1.59)1390(1.96)1391(2.25)1392(2.25)1393(2.23)1394 Observations13953,78013964,38913974,85013985,64413996,11714006,18314018.014028.114038.214048.314058.414068.514078.614088.714098.81410**1411**1412**1413**1414**1415**1416*1417**14184 months141910 months14209 months14218 months142230 September 20171423Flood event1424s1425e1426cir1427p1428e1429s1430u1431o1432h1433f1434o1435m1436h1437tir1438a1439g1440olla1441r1442u1443t1444a1445n1446 control group1447 treatment group14487 months1449**1450*1451Fig. 4 House prices recovery after the flood event on 30 September 2017. Notes: **5% level of signifi-1452cance; *10% level of significance145314541455===== PAGE 26 (score 40, numeric-paren 1, table-words 17, var-words 2) =====1456145712514581 31459House Prices and Flood Risk Exposure: An Integration of Hedonic…1460Appendix1461Table 9 Auxiliary tests1462We perform further test for the hypotheses that the SAR or SEM model can be nested in the SAC model 1463(see Fig. 1). First, if λ = 0, the SAC model can be reduced to SAR. Second, if ρ = 0, the SAC model can 1464be replaced with SEM. Wald tests of spatial terms together with these LR tests show that the hypotheses 1465of the simplification of SAC into either SAR or SEM must be rejected1466Panel A: Results of Moran’s I test for house prices1467 Ho: error is i.i.d1468 Moran’s I14690.1111470 Chi2(1)14716404.821472 P-value14730.001474Panel B – Spatial Autocorrelation Tests1475 Ho: Error has no spatial autocorrelation1476 GLOBAL Moran MI = 0.65901477Z(226.008)1478 P-Value14790.001480 GLOBAL Geary GC = 0.41101481Z(-18.983)1482 P-Value14830.001484 GLOBAL Getis-Ords GO = -0.29251485Z(-6.3e + 04)1486 P-Value14870.001488 Moran MI Error Test = 475.97661489Z(1.6e + 05)1490 P-Value14910.001492 LM Error (Burridge):Chi2(1)14932.07e + 041494 P-Value14950.001496 LM Error (Robust):Chi2(1)14972.01e + 041498 P-Value14990.001500 Ho: Spatial lagged dependent variable has no spatial autocorrelation1501 LM Lag (Anselin):Chi2(1)15022937.96781503 P-Value15040.001505 LM Lag (Robust): Chi2(1)15062341.00111507 P-Value15080.001509 Ho: No General Spatial Autocorrelation1510 LM SAC (LMErr + LMLag_R): Chi2(2)15112.31e + 041512 P-Value15130.001514 LM SAC (LMLag + LMErr_R): Chi2(2)15152.31e + 041516 P-Value15170.001518Notes: To enhance the validity of the Moran’s I test, we used LM tests and robust LM tests. Both tests 1519indicate that the hypotheses of no spatially lagged dependent variable and no spatially auto-correlated 1520error term must be rejected with the significance level of 1%1521Panel C– Postestimation for the SAC model—LR tests1522 LR test for SAR (λ = 0): Chi2(1)152340.1671524 P-Value15250.001526 LR test for SEM (ρ = 0): Chi2(1)152772.4101528 P-Value15290.001530 LR Test SAC vs. OLS (ρ + λ = 0): Chi2(2)1531102.0261532 P-Value15330.00153415351536===== PAGE 27 (score 5, numeric-paren 0, table-words 1, var-words 3) =====153715381261539 1540N. Bui et al.15411 31542Table 9, Figs. 5, 61543Fig. 5 Map of houses on arterial road. Notes: This study uses only the houses located on arterial roads 1544within a radius of one kilometre. Each of these roads is at least 1 km different from the others. If the arte-1545rial road is flooded, there is no chance that the small roads connect to it can stay dry since arterial roads 1546are built higher than smaller roads connected to them1547