1# RECORD eeedb48b06f3660101615bf3de6b27e1951880ee2# TITLE: Impact of public water supply unreliability on residential property prices in Marion County, West Virginia3# YEAR: 2021.0 VENUE: Agricultural and Resource Economics Review DOI: 10.1017/age.2021.244# AUTHORS: Fahad Alzahrani; A. Collins5# PAGES: 25 CANDIDATE TABLE PAGES (1-based): [8, 11, 12, 15, 21, 22, 23, 24]6# ABSTRACT: Abstract Water supply unreliability in many public water systems stems from aging infrastructure. We measure unreliability by the issuance of boil water notices (BWNs) within one year prior to single-family residential sale observations. Using a spatial quantile regression framework on transactions between 2012 and 2017, we find statistically significant, negative relationships between BWNs and residential properties. The estimated impacts of unreliability on residential housing prices, however, are not uniform across the distribution of prices. Specifically, we find that BWNs have a larger impact on medium- to low-priced houses (at or below the 60 percent quantile) compared with high-priced houses. An aggregate marginal willingness-to-pay value of $4.2 million was computed for a one-day reduction in annual BWN throughout Marion County.789===== PAGE 1 (score 8, numeric-paren 0, table-words 1, var-words 6) =====1011RESEARCH ARTICLE12Impact of public water supply unreliability on13residential property prices in Marion County,14West Virginia15Fahad Alzahrani116and Alan R. Collins2*171King Faisal University, Al-Ahsa, Eastern Province, Saudi Arabia and 2West Virginia University,18Morgantown, WV, USA19*Corresponding author. Email: alan.collins@mail.wvu.edu20(Received 1 February 2021; revised 28 September 2021; accepted 14 October 2021; first published online 21Abstract22Water supply unreliability in many public water systems stems from aging infrastructure.23We measure unreliability by the issuance of boil water notices (BWNs) within one year24prior to single-family residential sale observations. Using a spatial quantile regression25framework on transactions between 2012 and 2017, we find statistically significant, nega-26tive relationships between BWNs and residential properties. The estimated impacts of27unreliability on residential housing prices, however, are not uniform across the distribu-28tion of prices. Specifically, we find that BWNs have a larger impact on medium- to29low-priced houses (at or below the 60 percent quantile) compared with high-priced30houses. An aggregate marginal willingness-to-pay value of $4.2 million was computed31for a one-day reduction in annual BWN throughout Marion County.32Keywords: boil water notices; hedonic property price model; spatial quantile regression; water33infrastructure34Introduction35Studies have shown that even developed countries, like the U.S., face water challenges due to36problems associated with affordability, demographic changes, high environmental quality37expectations, and aging infrastructure (Wescoat, Headington, and Theobald 2007; OECD382016; Allaire, Wu, and Lall 2018; Allen et al. 2018). Examining affordability concerns,39Mack and Wrase (2017) project increasing water affordability issues with raising water40rates, particularly in the state of West Virginia, where the highest percentage of at-risk cen-41sus tracts (46 percent) for households unable to afford water bills of any state in the nation.42Aging water infrastructure, coupled with major funding shortfalls, is threatening the43nation’s water security (Alfredo et al. 2016).1 Many of the pipes and main lines that44© The Author(s), 2021. Published by Cambridge University Press on behalf of the Northeastern Agricultural and Resource45Economics Association. This is an Open Access article, distributed under the terms of the Creative Commons Attribution46licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in47any medium, provided the original work is properly cited.481Water infrastructure involves what is constructed to pump, divert, transport, treat, store, and deliver49water, as well as to collect, treat, and discharge storm and wastewater.50doi:10.1017/age.2021.2451Agricultural and Resource Economics Review (2022), 51, 105–1295215 November 2021)53https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press545556===== PAGE 8 (score 27, numeric-paren 9, table-words 6, var-words 4) =====5758Unlike OLS, quantile regression specifies changes in the conditional quantile, which59allows us to observe how housing characteristics are valued differently by consumers of60housing at different price ranges. Similar to equation 8, the spatial quantile lag model is61expressed as62ln(P) = r(t)W ln(P) + ZB(t) + 1(t)63(9)64which will result in estimates based on the conditional quantile, taking account of the65spatial effect at the same time.66To estimate the spatial quantile model, the two-stage quantile regression approach67proposed by Kim and Muller (2004) is used to account for the endogeneity in the spa-68tially lagged variable. In the first stage, the spatially lagged exogenous variables WZ and69Z are used to predict the spatially lagged endogenous variable W ln(P) at an individual70quantile. The predicted7172W ln(P) is substituted for W ln(P) in the spatial lag model to73eliminate the correlation between the spatially lagged endogenous variable and the74error term. Then, the second-stage regression for that quantile is performed to obtain75ρ(τ) and β(τ). This two-stage procedure is then repeated for each quantile.76Data77Our analysis is based on two main sources of data: (1) West Virginia Office of78Environmental Health Services (OEHS) and (2) Marion County Assessor. The next79two subsections provide descriptions of these data.80Boil water notices81To measure water supplier unreliability, we use BWNs issued by either public water sys-82tems or the Marion County Health Department. The OEHS provides information on all83BWNs in West Virginia starting from 2012. Between January 2012 and December 2017,84there were a total of 440 BWNs in Marion County, varying annually between 64 and 8085with no discernible trend over time. For each notice, the provided information includes86the following: when the notice is issued and lifted, the name of the public water system87plus its PWSID, the public health sanitation district where the notice is issued, the88Table 2. Boil Water Notices (BWNs) means by price quantile89Price quantile90Mean number of91BWNs days per year92Mean of BWN experienced93one year prior to sale94$31,928 (0.1)954.694960.81197$59,864 (0.2)981.907990.471100$81,286 (0.3)1010.8901020.313103$102,267 (0.4)1041.3141050.319106$124,881 (0.5)1070.9951080.301109$144,611 (0.6)1101.1851110.291112$167,371 (0.7)1130.5401140.268115$210,602 (0.8)1160.5491170.228118$301,113 (0.9)1190.5051200.162121Alzahrani and Collins122112123https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press124125126===== PAGE 9 (score 25, numeric-paren 2, table-words 1, var-words 21) =====127128reason for the notice, and details about the affected areas or street names.4 Using infor-129mation from the SDWIS/Fed database and details from each notice, we obtained city130and county affected by each notice.5131Housing data132The Marion County Assessor provides data for all residential property sales between133January 2012 and December 2017. This data set contains 7,972 residential property134transactions. By dropping all property sales without structures, the number of transac-135tions declined to 5,525. To ensure only arm’s length sales in our analysis, a process fol-136lowing Herriges, Secchi, and Babcock (2005) is used to drop all properties that were less137than 50 percent of their assessed values and/or sold for less than $5,000. Therefore, after138eliminating these sales, the total number of transactions included in our analysis is1391,985 single-family residential properties.6140To capture housing attributes previously found in the literature to influence housing141prices, we include variables for housing characteristics in the hedonic model such as142size, story height, age, number of bedrooms, total number of bathrooms, and existence143of house amenities (e.g., air-conditioning and fireplace). In addition, we control for two144variables that have been consistently neglected in hedonic property price models: (1)145house physical condition and (2) material and workmanship quality. Based upon asses-146sor data for physical condition, we rank houses from Unsound = 1 to Excellent = 6. For147quality, we use the quality grade factor provided by the county’s assessor ranging from148E = Poor (0.5) to X = Excellent (2.5). According to the assessor, these two variables are149unrelated, as the first one measures only the physical condition, but the second mea-150sures the quality of the building regardless of the physical condition. We compute a151simple correlation coefficient, and we find that it is equal to 0.60, indicating a moderate152correlation between the physical condition and the quality of the building.153All property sales are geocoded using house’s street address with ArcGIS to measure154neighborhood and proximity characteristics such as crime, school quality, distance to155Fairmont State University, and to the nearest highway interchange.7 For crime, we156use the total crime index that compares the average local crime level to that of the entire157U.S. (an index of 100 is average). This index includes both property and violent crimes158for 2010 at the block group level, and it is obtained from the ArcGIS database. For159school quality, we follow the literature and use proficiency tests as a proxy160(Brasington 1999). Specifically, we use the percentage of students in elementary school161who scored at or above proficiency levels in math tests. We obtain these shapefiles for162attendance zones in the county from the National Center for Education Statistics163(NCES) and the data for math tests from the GreatSchools website. For distance1644PWSID is an identification code that starts with the state’s initials and followed by seven numbers.165These codes can be used to find more information about the water system (e.g., city and county served166or the source of the primary water) in the Safe Drinking Water Information System/Federal Warehouse167(SDWIS/Fed).1685Some water systems were not available on SDWIS/Fed, so we used the WV Drinking Water Watch to169obtain the necessary information. Available at: http://129.71.204.189:1977/DWWpublic/index.jsp.1706County Assessor data did not include information about the source of potable water for each property.171Given the high percentage of the county population being served by public water systems, all properties in172this data set are assumed to be supplied by public water for potable water needs as opposed to private wells.1737This variable was included since Fairmont serves a “bedroom” community for the more prosperous174communities of Morgantown to the north and Clarksburg to the south.175Agricultural and Resource Economics Review176113177https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press178179180===== PAGE 11 (score 27, numeric-paren 19, table-words 2, var-words 4) =====181182Results183First, a single market assumption for residential properties in Marion County is con-184firmed from our data using a Chow test on models comparing two data subsamples:185(1) the city of Fairmont and (2) the rest of Marion County. To examine the changing186impact of water supply unreliability over the distribution of housing prices, the spatial187quantile regression model results are reported in Appendix Tables A1 and A2. All esti-188mates for the spatial autoregressive coefficient (ρ) are positive and statistically signifi-189cant, meaning that a positive spatial similarity exists between residential property sale190prices. Focusing on the BWN variables with coefficients statistically different from191Table 3. Property transactions and BWNs by tax districts in Marion County, West Virginia, between 2012192and 2017193Number194District195Property196transactions197BWNs198Properties affected by199BWNs (%)2001201Barrackville20253203420430 (56.6)2052206Fairmont District207782082020923 (29.5)2103211Fairmont Citya21248521351214138 (28.5)2154216Grant Annexa2171362182021945 (33.1)2205221Union Citya2221772231122434 (19.2)2256226Winfield Citya22728228422921 (75)2307231Fairview232923332341 (11.1)2358236Farmington23714238523910 (71.4)2409241Grant District2422362432024432 (13.6)24510246Grant Town24719248224912 (63.2)25011251Lincoln District2521332533725469 (51.9)25512256Mannington District25717258325911 (64.7)26013261Mannington City26275263726459 (78.7)26514266Monongah-Grantb26724268026914 (58.3)27015271Monongah-Lincolnb27219273027412 (63.2)27516276Paw Paw District277592783127941 (69.5)28017281Rivesville282502832428437 (74)28518286Union District2871132881228929 (25.7)29019291Winfield District292862931129422 (25.6)29520296Worthington297529812994 (80)30021301White Hall30225303230411 (44)30522306Pleasant Valley3071443081730945 (31.3)310Multiple Districts311—31265313—314Total3151,985316350317700 (35.3)318aPart of Fairmont City.319bBWNs that affect districts 14 and 15 usually affect other districts as well.320Agricultural and Resource Economics Review321115322https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press323324325===== PAGE 12 (score 22, numeric-paren 0, table-words 2, var-words 16) =====326327Table 4. Summary statistics and variable descriptions328Variable329Description330Unit331Average332St. Dev.333Min.334Max.335Dependent variable336House price337House’s sale price338$339135,33834081,0823416,000342650,000343Independent variables344BWN variables345Days BWNs_year346Number of days the house was under BWNs a year347before the sale348Days3491.403505.523510352116353BWN3541 = if the house was affected by a BWN, 0 = otherwise3550/13560.353570.4835803591360Structural variables361SQFT362Building area in square feet363ft23641,645.45365658.893664803675,488368Story height369Story height370Number3711.273720.4237313743375Construction material376Material used in construction, 1 = if the material used is377Alum/Vinyl, 0 = otherwise3780/13790.633800.4838103821383Physical condition384Physical condition of the house (1–6, 1 = unsound, 6 = excellent)385Number3864.753870.8238813896390Quality grade factor391Quality of the material and workmanship (0.5–2.5, 0.5 = E, 2.5 = X+)392Number3931.053940.243950.53962.5397Age398Age of the house399Years40063401364020403167404Bedrooms405Number of bedrooms406Number4072.804080.76409141012411Total bathrooms412Total number of bathrooms (full + half)413Number4141.884150.8141614176418AC419Heating system with AC, 1 = central with A/C, 0 = otherwise4200/14210.774220.4242304241425Fireplace4261 = if there is a wood-burning fireplace, 0 = otherwise4270/14280.314290.4643004311432Basement garage4331 = if there is basement garage, 0 = otherwise4340/14350.294360.4543704381439Alzahrani and Collins440116441https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press442443444===== PAGE 13 (score 13, numeric-paren 0, table-words 1, var-words 9) =====445446Property variables447Lot size448Lot size449Acres4500.964515.304520.01453130.28454Sidewalk4551 = if there is a sidewalk, 0 = otherwise4560/14570.064580.2445904601461Fence4621 = if there is a fence around the property, 0 = otherwise4630/14640.494650.5046604671468Neighborhood and proximity variables469Crime470Total crime index471Number47254.0547324.2847419475101476School quality477Percentage of students scoring at or above proficiency levels in478the math test479%48048.9148114.264822948371484Distance to highway exit485Distance to the nearest highway interchange (I-79)486Miles4872.634882.684890.0949017.47491Distance to FSU492Distance to Fairmont State University (FSU)493Miles4943.194952.574960.1449717.30498Note: Observations = 1,985.499Agricultural and Resource Economics Review500117501https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press502503504===== PAGE 15 (score 47, numeric-paren 33, table-words 3, var-words 4) =====505506Table 5. Two stage least square (2SLS) estimates for the spatial quantile models (dependent variable is ln (P))5072SLS5080.15090.25100.35110.45120.55130.65140.75150.85160.9517Days518BWNs_year519−0.0065***520(0.0016)521−0.0664***522(0.0173)523−0.0271**524(0.0121)525−0.0155526(0.0095)527−0.0075528(0.0048)529−0.0064**530(0.0033)531−0.0049**532(0.0025)533−0.0048534(0.0034)535−0.0015536(0.0033)5370.0009538(0.0030)539BWN540−0.1853***541(0.0186)542−0.5392***543(0.0653)544−0.2971***545(0.0483)546−0.1750***547(0.0265)548−0.1017***549(0.0209)550−0.0676***551(0.0171)552−0.0509***553(0.0185)554−0.0449**555−(0.0449)556−0.0074557(0.0205)5580.0077559(0.0172)560Note: Bootstrapped standard errors in parentheses.561***p<0.01, **p<0.05.562Table 6. MWTP ($) for a change in water supply reliability from the estimated spatial models5632SLS5640.15650.25660.35670.45680.55690.65700.75710.85720.9573Days BWNs_year5741,1605752,6795762,093577982578868579BWN58035,11658128,69058224,74658318,30458412,87558510,4215869,0905879,129588Notes: The values represent coefficients that are statistically significant at 5 percent; empty cells are for insignificant coefficients. Since BWN is a binary variable, its marginal impact is equal to:589Mean P in quantile * (exp(b) −1) * (1/1 −ρ).590Agricultural and Resource Economics Review591119592https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press593594595===== PAGE 16 (score 8, numeric-paren 0, table-words 3, var-words 2) =====596597spatial correlation impacts between houses impacted by the BWN that are embedded in598the MWTP estimates, (b) enhanced risk perceptions of households impacted by the599BWN as explained below, and (c) the capitalized defensive expenditure value may rep-600resent the cost of leaving the home temporarily while the BWN is in effect rather than601the economic cost of in-house responses.602Furthermore, we see that the MWTP values for consumers of low-priced houses are603greater than for consumers of high-priced houses. That is, a reduction of one day under604BWN per year will increase housing prices at the medium to lower end of the housing605price distribution by $2,100 to $2,700, whereas houses at the mid-range of the distribu-606tion will increase by $850 to $1,000 (Table 6). These impacts range from 0.60 percent to6078.4 percent of average sale prices across these quantiles. Houses at the 0.7 quantile and608above of the price distribution show no statistically significant impacts from the days of609the BWN variable. Finally, given the mean distribution of BWNs by quantile in Table 2,610a threshold value (π) of approximately 1.0 BWNs annually is indicated by the statisti-611cally insignificant results at the 0.7 quantile and above.612One interpretation of these results is that the greater prevalence of BWN for613lower-priced houses leads to greater risk perceptions and awareness about public water614supply problems among these housing consumers. Previous studies have shown that615consumers who experience more problems with their water supply perceive the associ-616ated health risks differently and were willing to pay more to fix water supply problems617than those consumers who experience water supply problems infrequently (Anadu and618Harding 2000; Genius and Tsagarakis 2006). As shown in Table 2, experience with and619duration of BWNs consistently declines with housing price quantiles.620Overall, water supply unreliability has a substantial impact on the value of residential621properties in Marion County. To calculate the aggregate MWTP across the entire622county for a one-day reduction of BWN, three factors are multiplied together: (1)623the average value of statistically significant coefficients from the spatial quantile regres-624sion ($1,694), (2) the estimated annual average number of owner-occupied housing625units that would be expected to be affected by the BWN in Marion County between6262013 and 2017 (6,150), and (3) finally, from the quantile regression results, a factor627of 0.4 as the percentage of houses would incur a price impact from these notices.628Following this procedure, we obtain an aggregate MWTP value of $4.2 million for a629one-day reduction of the annual BWN throughout Marion County.630Conclusions631Aging water infrastructure, combined with funding shortfalls, poses serious challenges632to local governments and water systems across the U.S. The number of water main633breaks in the U.S. has been increasing every year resulting in more interruptions in pub-634lic water service (Folkman 2018). Since achieving full reliability in water service is most635likely not an optimal policy, decision makers at local water utilities must attempt to bal-636ance between cost and risk associated with their policies (Howe et al. 1994). However,637water customers’ perceptions of the risk associated with supply disruptions are usually638unknown to decision makers. Therefore, studies have attempted to examine the WTP639for640water641supply642reliability643using644stated645preference646approaches647(Genius648and649Tsagarakis 2006; Vásquez and Espaillat 2016).650In this research, we examine the impact of water supply unreliability on single-651family, residential property sales in Marion County, West Virginia, using spatial quan-652tile hedonic property price models. To measure the impact of water supply unreliability,653Alzahrani and Collins654120655https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press656657658===== PAGE 21 (score 198, numeric-paren 163, table-words 2, var-words 13) =====659660Appendix661Spatial quantile regression models—dependent variable is ln (P)11662Table A1. Spatial quantile regression results for model (1)—days BWNs_year6630.16640.26650.36660.46670.56680.66690.76700.86710.9672Days BWNs_year673−0.0664***674(0.0173)675−0.0271**676(0.0121)677−0.0155678(0.0095)679−0.0075680(0.0048)681−0.0064**682(0.0033)683−0.0049**684(0.0025)685−0.0048686(0.0034)687−0.0015688(0.0033)6890.0009690(0.0030)691SQFT6920.0002***693(0.0000)6940.0002***695(0.0000)6960.0002***697(0.0000)6980.0002***699(0.0000)7000.0002***701(0.0000)7020.0003***703(0.0000)7040.0002***705(0.0000)7060.0002***707(0.0000)7080.0002***709(0.0000)710Story height711−0.0575*712(0.0306)713−0.0303714(0.0283)715−0.046*716(0.0244)717−0.0413**718(0.0192)719−0.0323*720(0.0173)721−0.0266722(0.0198)723−0.0092724(0.0243)7250.0052726(0.0225)727−0.0063728(0.0222)729Construction material730−0.0672**731(0.0286)732−0.0668***733(0.0231)734−0.0636***735(0.0185)736−0.0525***737(0.0154)738−0.0297**739(0.0141)740−0.0298*741(0.0160)742−0.0285*743(0.0168)744−0.0256745(0.0174)746−0.0311*747(0.0159)748Physical condition7490.2683***750(0.0281)7510.2637***752(0.0217)7530.2266***754(0.0187)7550.1998***756(0.0146)7570.1758***758(0.0156)7590.1634***760(0.0158)7610.1472***762(0.0181)7630.1373***764(0.0167)7650.1473***766(0.0159)767Quality grade factor7680.4243***769(0.1245)7700.3806***771(0.0792)7720.3937***773(0.0677)7740.3875***775(0.0544)7760.3909***777(0.0488)7780.3472***779(0.0507)7800.364***781(0.0668)7820.4398***783(0.0532)7840.4374***785(0.0657)786Age787−0.0025***788(0.0007)789−0.0025***790(0.0005)791−0.0026***792(0.0005)793−0.0024***794(0.0003)795−0.0025***796(0.0004)797−0.0026***798(0.0004)799−0.0025***800(0.0004)801−0.0021***802(0.0004)803−0.0016***804(0.0003)805Bedrooms8060.0235807(0.0264)8080.0198809(0.0212)8100.0411**811(0.0164)8120.0487***813(0.0150)8140.0518***815(0.0126)8160.0463***817(0.0158)8180.0319**819(0.0152)8200.0448***821(0.0140)8220.0474***823(0.0133)824Total bathrooms8250.0059826(0.0320)8270.0258828(0.0192)8290.0372**830(0.0153)8310.0572***832(0.0134)8330.034***834(0.0123)8350.028**836(0.0142)8370.0279*838(0.0145)8390.0152840(0.0149)8410.0152842(0.0178)843AC8440.2074***845(0.0540)8460.2328***847(0.0335)8480.2202***849(0.0309)8500.2217***851(0.0250)8520.2169***853(0.0244)8540.2099***855(0.0259)8560.1711***857(0.0233)8580.1615***859(0.0271)8600.0909**861(0.0371)862(Continued)863Agricultural and Resource Economics Review864125865https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press866867868===== PAGE 22 (score 185, numeric-paren 154, table-words 1, var-words 9) =====869870Table A1. (Continued.)8710.18720.28730.38740.48750.58760.68770.78780.88790.9880Fire place8810.1527***882(0.0362)8830.1059***884(0.0220)8850.113***886(0.0191)8870.1052***888(0.0138)8890.0905***890(0.0158)8910.0867***892(0.0150)8930.0799***894(0.0163)8950.0667***896(0.0176)8970.0553***898(0.0153)899Basement garage9000.0573**901(0.0263)9020.0169903(0.0203)9040.0298*905(0.0168)9060.0159907(0.0140)9080.0113909(0.0144)9100.0142911(0.0144)9120.0085913(0.0185)9140.0032915(0.0172)916−0.0058917(0.0179)918Lot size919−0.0009920(0.0075)9210.0088922(0.0065)9230.0112**924(0.0044)9250.0105**926(0.0050)9270.0117*928(0.0065)9290.0108930(0.0073)9310.0244**932(0.0102)9330.0212**934(0.0100)9350.0381***936(0.0070)937Sidewalk938−0.0753939(0.0730)940−0.0562941(0.0529)942−0.0216943(0.0462)944−0.035945(0.0325)946−0.0415947(0.0291)948−0.0376949(0.0298)950−0.0559*951(0.0315)952−0.0885***953(0.0298)954−0.0933**955(0.0410)956Fence9570.1351***958(0.0308)9590.1127***960(0.0249)9610.0967***962(0.0160)9630.0944***964(0.0155)9650.0903***966(0.0130)9670.0912***968(0.0163)9690.0764***970(0.0161)9710.0659***972(0.0156)9730.0434***974(0.0161)975Crime976−0.0004977(0.0010)978−0.0004979(0.0005)980−0.0007981(0.0005)982−0.0005983(0.0005)984−0.0002985(0.0004)986−0.0005987(0.0004)988−0.0008989(0.0005)990−0.0005991(0.0004)992−0.0006993(0.0005)994School quality9950.0014996(0.0014)9970.0001998(0.0011)999−0.00061000(0.0008)1001−0.0014**1002(0.0007)1003−0.0011004(0.0007)1005−0.0011006(0.0008)1007−0.00091008(0.0010)1009−0.00051010(0.0007)1011−0.00071012(0.0007)1013Distance to highway exit1014−0.00461015(0.0116)1016−0.0071017(0.0085)1018−0.0156**1019(0.0072)1020−0.0204***1021(0.0061)1022−0.0139**1023(0.0058)1024−0.0103*1025(0.0062)1026−0.0154**1027(0.0064)1028−0.0128**1029(0.0064)1030−0.014**1031(0.0067)1032Distance to FSU1033−0.00931034(0.0106)1035−0.01151036(0.0078)1037−0.00721038(0.0060)10390.00051040(0.0054)1041−0.00161042(0.0052)1043−0.00441044(0.0052)1045−0.00241046(0.0056)10470.00151048(0.0061)10490.00551050(0.0067)1051ρ10520.2085***1053(0.0439)10540.2248***1055(0.0341)10560.1709***1057(0.0302)10580.1705***1059(0.0245)10600.1862***1061(0.0248)10620.1841***1063(0.0266)10640.1633***1065(0.0288)10660.1628***1067(0.0267)10680.1694***1069(0.0264)1070Constant10716.7475***1072(0.5188)10736.7145***1074(0.4221)10757.569***1076(0.3615)10777.7436***1078(0.2913)10797.6613***1080(0.2727)10817.8642***1082(0.3232)10838.3129***1084(0.3527)10858.2941***1086(0.3355)10878.2796***1088(0.3264)1089Notes: Bootstrapped standard errors in parentheses. The statistical software used for the estimation of spatial quantile regression, and the qregspiv command of the McSpatial package in the1090software R. This command does not return goodness-of-fit measures for the regression results. Year-fixed effects are included in this model.1091***p<0.01, **p<0.05, *p<0.1.1092Alzahrani and Collins10931261094https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press109510961097===== PAGE 23 (score 231, numeric-paren 192, table-words 1, var-words 16) =====10981099Table A2. Spatial quantile regression results for model (2)—BWN11000.111010.211020.311030.411040.511050.611060.711070.811080.91109BWN1110−0.5392***1111(0.0653)1112−0.2971***1113(0.0483)1114−0.175***1115(0.0265)1116−0.1017***1117(0.0209)1118−0.0676***1119(0.0171)1120−0.0509***1121(0.0185)1122−0.0449**1123−(0.0449)1124−0.00741125(0.0205)11260.00771127(0.0172)1128SQFT11290.0003***1130(0.0000)11310.0002***1132(0.0000)11330.0002***1134(0.0000)11350.0002***1136(0.0000)11370.0002***1138(0.0000)11390.0003***1140(0.0000)11410.0002***1142(0.0002)11430.0002***1144(0.0000)11450.0002***1146(0.0000)1147Story height1148−0.03571149(0.0372)1150−0.03171151(0.0285)1152−0.0466**1153(0.0205)1154−0.0374*1155(0.0197)1156−0.02381157(0.0217)1158−0.0326*1159(0.0194)1160−0.00651161−(0.0065)11620.00891163(0.0211)1164−0.00441165(0.0208)1166Construction1167material1168−0.078***1169(0.0262)1170−0.0762***1171(0.0235)1172−0.0519**1173(0.0202)1174−0.0567***1175(0.0159)1176−0.0407***1177(0.0152)1178−0.0311**1179(0.0146)1180−0.0275*1181−(0.0275)1182−0.01971183(0.0170)1184−0.0308**1185(0.0151)1186Physical1187condition11880.2086***1189(0.0238)11900.2265***1191(0.0227)11920.2058***1193(0.0173)11940.2014***1195(0.0173)11960.1733***1197(0.0176)11980.1622***1199(0.0153)12000.1478***1201(0.1478)12020.1407***1203(0.0166)12040.1474***1205(0.0181)1206Quality grade1207factor12080.5079***1209(0.0963)12100.4762***1211(0.0734)12120.4752***1213(0.0509)12140.4117***1215(0.0578)12160.4025***1217(0.0609)12180.3566***1219(0.0585)12200.3477***1221(0.3477)12220.4225***1223(0.0632)12240.4505***1225(0.0692)1226Age1227−0.0021***1228(0.0006)1229−0.0025***1230(0.0005)1231−0.0024***1232(0.0004)1233−0.0025***1234(0.0005)1235−0.0026***1236(0.0004)1237−0.0026***1238(0.0003)1239−0.0023***1240−(0.0023)1241−0.0023***1242(0.0004)1243−0.0017***1244(0.0003)1245Bedrooms12460.01651247(0.0161)12480.0241249(0.0192)12500.0406**1251(0.0166)12520.0463***1253(0.0165)12540.0447***1255(0.0154)12560.0481***1257(0.0130)12580.0366**1259(0.0366)12600.0433***1261(0.0128)12620.047***1263(0.0141)1264Total bathrooms1265−0.01781266(0.0211)12670.0211268(0.0223)12690.0406***1270(0.0148)12710.0454***1272(0.0168)12730.0335***1274(0.0118)12750.0273*1276(0.0141)12770.037***1278(0.0370)12790.01481280(0.0172)12810.01431282(0.0153)1283AC12840.1847***1285(0.0347)12860.2161***1287(0.0372)12880.2305***1289(0.0272)12900.2244***1291(0.0271)12920.2281***1293(0.0268)12940.2081***1295(0.0295)12960.1684***1297(0.1684)12980.1468***1299(0.0287)13000.0936***1301(0.0334)1302Fire place13030.1368***1304(0.0260)13050.1306***1306(0.0234)13070.1319***1308(0.0204)13090.1094***1310(0.0159)13110.0972***1312(0.0149)13130.0896***1314(0.0159)13150.0836***1316(0.0836)13170.077***1318(0.0165)13190.0604***1320(0.0171)1321Basement garage13220.02091323(0.0253)13240.03261325(0.0214)13260.02321327(0.0184)13280.0171329(0.0179)13300.01531331(0.0151)13320.00811333(0.0153)1334−0.00061335−(0.0006)13360.00191337(0.0165)1338−0.01041339(0.0181)1340(Continued)1341Agricultural and Resource Economics Review13421271343https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press134413451346===== PAGE 24 (score 150, numeric-paren 126, table-words 1, var-words 6) =====13471348Table A2. (Continued.)13490.113500.213510.313520.413530.513540.613550.713560.813570.91358Lot size13590.00221360(0.0049)13610.00881362(0.0066)13630.0122**1364(0.0050)13650.0111**1366(0.0047)13670.01*1368(0.0057)13690.01121370(0.0078)13710.022***1372(0.0220)13730.0218**1374(0.0099)13750.0372***1376(0.0074)1377Sidewalk1378−0.01411379(0.0515)1380−0.05411381(0.0345)1382−0.05271383(0.0443)1384−0.03851385(0.0334)1386−0.05231387(0.0400)1388−0.03741389(0.0312)1390−0.0647**1391−(0.0647)1392−0.0876**1393(0.0357)1394−0.0889**1395(0.0445)1396Fence13970.0783***1398(0.0277)13990.104***1400(0.0226)14010.1003***1402(0.0176)14030.0936***1404(0.0182)14050.0993***1406(0.0156)14070.0938***1408(0.0170)14090.0729***1410(0.0729)14110.066***1412(0.0144)14130.0514***1414(0.0155)1415Crime1416−0.00011417(0.0008)1418−0.00091419(0.0007)1420−0.00071421(0.0005)1422−0.00041423(0.0004)1424−0.00011425(0.0005)1426−0.00041427(0.0004)1428−0.0007*1429−(0.0007)1430−0.00061431(0.0005)1432−0.00071433(0.0005)1434School quality14350.00181436(0.0012)14370.000021438(0.0011)1439−0.00081440(0.0009)1441−0.00121442(0.0008)1443−0.00051444(0.0007)1445−0.00091446(0.0007)1447−0.00081448−(0.0008)1449−0.00051450(0.0007)1451−0.00051452(0.0007)1453Distance to1454highway exit1455−0.01581456(0.0104)1457−0.0151*1458(0.0089)1459−0.02***1460(0.0067)1461−0.022***1462(0.0065)1463−0.0123*1464(0.0071)1465−0.0124**1466(0.0058)1467−0.0146***1468−(0.0146)1469−0.0163***1470(0.0061)1471−0.0141*1472(0.0076)1473Distance to FSU1474−0.00241475(0.0094)1476−0.00721477(0.0079)1478−0.00321479(0.0064)14800.00141481(0.0061)1482−0.00331483(0.0061)1484−0.00061485(0.0048)1486−0.00171487−(0.0017)14880.00121489(0.0056)14900.00431491(0.0065)1492ρ14930.2047***1494(0.0326)14950.1631***1496(0.0298)14970.1507***1498(0.0284)14990.1497***1500(0.0236)15010.1619***1502(0.0281)15030.1693***1504(0.0274)15050.158***1506(0.1580)15070.1442***1508(0.0255)15090.1539***1510(0.0263)1511Constant15127.0339***1513(0.3612)15147.5819***1515(0.3441)15167.8347***1517(0.3136)15187.9704***1519(0.2879)15207.9343***1521(0.3276)15228.0273***1523(0.3113)15248.3479***1525(8.3479)15268.5324***1527(0.3105)15288.4602***1529(0.3152)1530Notes: Bootstrapped standard errors in parentheses. The statistical software used for the estimation of spatial quantile regression, and the qregspiv command of the McSpatial package in the1531software R. This command does not return goodness-of-fit measures for the regression results. Year-fixed effects are included in this model.1532***p<0.01, **p<0.05, *p<0.1.1533Alzahrani and Collins15341281535https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press153615371538===== PAGE 25 (score 3, numeric-paren 0, table-words 0, var-words 3) =====15391540Figure A1. Distance Histogram Between Neighbors in the Weight Matrix.1541Cite this article: Alzahrani F, Collins AR (2022). Impact of public water supply unreliability on residential1542property prices in Marion County, West Virginia. Agricultural and Resource Economics Review 51, 105–129.1543https://doi.org/10.1017/age.2021.241544Agricultural and Resource Economics Review15451291546https://doi.org/10.1017/age.2021.24 Published online by Cambridge University Press1547