1# RECORD 38bf59597966e99f31288b3b3bfbe1d0b22ac0b52# TITLE: Spatial Discontinuities in Water Rights: Estimating Household Willingness-to-Pay for Unmetered Irrigation Water3# YEAR: 2024.0 VENUE: The Review of Regional Studies DOI: 10.52324/001c.1279624# AUTHORS: Matthew K. Gnagey5# PAGES: 18 CANDIDATE TABLE PAGES (1-based): [10, 11, 12, 13, 14, 15, 16]6# ABSTRACT: House prices capture the value households place on environmental attributes and amenities. Access to an unlimited water right for irrigating lawns in an arid region is a prized environmental amenity. The distribution of this amenity varies across space, and often houses with these water rights are located in neighborhoods that differ substantially from other neighborhoods. This study aims to disentangle these neighborhood differences to identify the value that households place on unmetered water rights for lawn irrigation. Identification relies on a regression discontinuity design exploiting the variation in unmetered water access that occurs based on historical land use and geographic features. By examining houses on the border of the water rights zones, the analysis isolates houses that only differ by their access to unmetered irrigation water, thereby eliminating potentially correlated variation with this access, such as differences in socioeconomic attributes, local tax rates, school quality, and environmental amenities. The findings indicate that households are willing to pay a premium of 2% for access to unmetered irrigation water. This premium is not driven by factors such as lawn greenness or the curb appeal from more heavily irrigated landscapes.789===== PAGE 1 (score 10, numeric-paren 0, table-words 0, var-words 10) =====1011(2024) 54, 241–25812Spatial Discontinuities in Water Rights: Estimating13Household Willingness-to-Pay for Unmetered Irrigation14Water ∗15Matthew K. Gnageya16aDepartment of Economics, Weber State University, USA17Abstract: House prices capture the value households place on environmental attributes and amenities.18Access to an unlimited water right for irrigating lawns in an arid region is a prized environmental amenity.19The distribution of this amenity varies across space, and often houses with these water rights are located20in neighborhoods that differ substantially from other neighborhoods. This study aims to disentangle these21neighborhood differences to identify the value that households place on unmetered water rights for lawn22irrigation. Identification relies on a regression discontinuity design exploiting the variation in unmetered23water access that occurs based on historical land use and geographic features. By examining houses on the24border of the water rights zones, the analysis isolates houses that only differ by their access to unmetered25irrigation water, thereby eliminating potentially correlated variation with this access, such as differences in26socioeconomic attributes, local tax rates, school quality, and environmental amenities. The findings indicate27that households are willing to pay a premium of 2% for access to unmetered irrigation water. This premium28is not driven by factors such as lawn greenness or the curb appeal from more heavily irrigated landscapes.29Keywords: Regression discontinuity design, hedonic analysis, water conservation, irrigation30JEL Codes: D12, L95, Q21, 25311.32INTRODUCTION33Environmental attributes are frequently capitalized into house prices. These amenities can34include external attributes at the regional or neighborhood level, but can also include the35actual environmental attributes on the property, such as low air pollution or an appealingly36landscaped lawn. The provision of these environmental attributes can result from local or37regional policy decisions, such as emission testing for air quality, or water rights to facilitate38green landscaping. In the arid American West, access to an unlimited water right for irri-39gating lawns is anecdotally a highly valued attribute. Because these water rights vary across40∗Gnagey is an Associate Professor of Economics at Weber State University, Ogden, UT 84403. Corresponding41Author: M. K. Gnagey, E-mail: mattgnagey@weber.edu42(c) Southern Regional Science Association 202443ISSN 1553-0892, 0048-49X (online)44www.srsa.org/rrs454647===== PAGE 10 (score 168, numeric-paren 130, table-words 4, var-words 16) =====484925050The Review of Regional Studies 54(3)51Table 2: Regression Results Using Boundary Fixed Effects.52Dependent Variable is Natural Log of the Real House Price53Variable54All Houses55<1000ft56<500ft57<100ft58Secondary Water590.0368***600.0395***610.0304***620.0198**63(0.003)64(0.004)65(0.005)66(0.010)67House Size (’000 sq ft)680.1510***690.1060***700.1007***710.1576***72(0.007)73(0.011)74(0.015)75(0.027)76House Size Squared770.0078***780.0143***790.0146***800.009181(0.001)82(0.002)83(0.003)84(0.006)85Acreage860.0011*870.0275880.0877***890.1766***90(0.001)91(0.018)92(0.027)93(0.036)94House Age95-0.0039***96-0.0037***97-0.0039***98-0.0038***99(0.000)100(0.000)101(0.000)102(0.000)103Bathrooms1040.0287***1050.0238***1060.0256***1070.0263***108(0.001)109(0.002)110(0.003)111(0.008)112Bedrooms1130.0697***1140.0603***1150.0570***1160.0492***117(0.002)118(0.003)119(0.004)120(0.009)121Downtown Ogden (min)1220.0074***1230.0087***1240.0114***1250.0095***126(0.000)127(0.001)128(0.001)129(0.002)130Air Conditioning1310.0796***1320.0703***1330.0813***1340.0763***135(0.002)136(0.004)137(0.005)138(0.012)139Deck1400.0214***1410.0160***1420.00671430.0208**144(0.002)145(0.003)146(0.005)147(0.009)148Swimming Pool149-0.1137***150-0.0552***151-0.0477***1520.0418153(0.005)154(0.008)155(0.013)156(0.047)157Fireplace1580.0741***1590.0625***1600.0499***1610.0145162(0.002)163(0.004)164(0.005)165(0.013)166Elevation (’000ft)1670.0421***1680.1027***1690.1011***1700.0764*171(0.006)172(0.012)173(0.015)174(0.044)175View1760.0590***1770.0577***1780.0498***1790.0290**180(0.002)181(0.004)182(0.005)183(0.012)184Cul-de-sac1850.0152***1860.0180***1870.0246***1880.0487***189(0.003)190(0.005)191(0.006)192(0.014)193Corner Lot1940.0187***1950.0232***1960.0163***197-0.0100198(0.003)199(0.004)200(0.006)201(0.013)202Constant20311.3196***20411.0503***20511.0085***20610.8637***207(0.033)208(0.089)209(0.102)210(0.217)211Observations21274,37521322,45421411,3852151,906216R-squared2170.74172180.75022190.74622200.7733221Boundary FE222Yes223Yes224Yes225Yes226Month/Year FE227Yes228Yes229Yes230Yes231Notes: Standard errors clustered at each boundary segment. *** p<0.01, ** p<0.05, * p<0.1.232©Southern Regional Science Association 2024.233234235===== PAGE 11 (score 167, numeric-paren 130, table-words 4, var-words 15) =====236237GNAGEY: SPATIAL DISCONTINUITIES IN WATER RIGHTS238251239Table 3: Regression Results Using Neighborhood Fixed Effects.240Dependent Variable is Natural Log of the Real House Price241Variable242All Houses243<1000ft244<500ft245<100ft246Secondary Water2470.0078***2480.0248***2490.0235***2500.0201**251(0.003)252(0.004)253(0.005)254(0.011)255House Size (’000 sq ft)2560.1632***2570.1200***2580.1090***2590.1617***260(0.007)261(0.010)262(0.015)263(0.025)264House Size Squared2650.0068***2660.0132***2670.0144***2680.0094269(0.001)270(0.002)271(0.003)272(0.006)273Acreage2740.0017**2750.03432760.0975***2770.1895***278(0.001)279(0.021)280(0.027)281(0.033)282Age283-0.0038***284-0.0038***285-0.0039***286-0.0041***287(0.000)288(0.000)289(0.000)290(0.000)291Bathrooms2920.0246***2930.0206***2940.0226***2950.0231***296(0.001)297(0.002)298(0.003)299(0.008)300Bedrooms3010.0677***3020.0598***3030.0572***3040.0471***305(0.002)306(0.003)307(0.004)308(0.008)309Downtown Ogden (min)3100.0114***3110.0113***3120.0115***3130.0096***314(0.000)315(0.001)316(0.001)317(0.002)318Air Conditioning3190.0681***3200.0629***3210.0796***3220.0771***323(0.002)324(0.004)325(0.005)326(0.012)327Deck3280.0233***3290.0174***3300.0084*3310.0211**332(0.002)333(0.003)334(0.005)335(0.009)336Swimming Pool337-0.1057***338-0.0700***339-0.0723***3400.0417341(0.005)342(0.008)343(0.012)344(0.042)345Fireplace3460.0750***3470.0642***3480.0493***3490.0118350(0.002)351(0.004)352(0.006)353(0.012)354Elevation (’000ft)3550.0379***3560.0545***3570.0565***3580.0539359(0.007)360(0.014)361(0.018)362(0.046)363View3640.0615***3650.0555***3660.0496***3670.0355***368(0.002)369(0.004)370(0.005)371(0.011)372Cul-de-sac3730.0108***3740.0160***3750.0229***3760.0360**377(0.003)378(0.005)379(0.006)380(0.014)381Corner Lot3820.0159***3830.0239***3840.0186***385-0.0136386(0.003)387(0.004)388(0.006)389(0.014)390Constant39111.1422***39211.1068***39311.0899***39411.0645***395(0.033)396(0.062)397(0.082)398(0.202)399Observations40074,37540122,45440211,3854031,906404R-squared4050.73434060.74314070.74194080.7741409Boundary FE410Yes411Yes412Yes413Yes414Month/Year FE415Yes416Yes417Yes418Yes419Notes: Standard errors clustered at each neighborhood. *** p<0.01, ** p<0.05, * p<0.1.420©Southern Regional Science Association 2024.421422423===== PAGE 12 (score 34, numeric-paren 8, table-words 9, var-words 7) =====424425252426The Review of Regional Studies 54(3)427Table 4: Regression Results Using Non-Summer Month Sales.428Dependent Variable is Natural Log of the Real House Price429Variable430All Houses431<1000ft432<500ft433<100ft434Secondary Water4350.0363***4360.0404***4370.0343***4380.0220*439(0.003)440(0.004)441(0.006)442(0.011)443Observations44453,02744516,0414468,1034471,379448R-squared4490.74304500.74914510.74894520.7985453Control Variables454Yes455Yes456Yes457Yes458Boundary FE459Yes460Yes461Yes462Yes463Month/Year FE464Yes465Yes466Yes467Yes468Notes: Control variables include the complete set of variables presented in Table 2. Standard errors469clustered at each boundary segment. *** p<0.01, ** p<0.05, * p<0.1.470in curb appeal of the properties.471The regression models from Table 2 are re-run using this restricted dataset. If the price472premium observed earlier in the main analysis is primarily driven by greener lawns in summer473months, there would be a substantially reduced or non-existent premium in this model.474However, if the premium persists, this is evidence that the access to unmetered secondary475water is valuable, not merely the visual reflection of that access manifested through greener476landscapes.477The results are statistically and numerically similar using this restricted sample (Table 4).478Despite losing slightly more than 25% of the observations (unsurprisingly, a disproportion-479ately large number of sales happen in the summer months compared to the other seasons),480the point estimates and statistical significance of the estimates are similar across the entire481geographic region, and each of the three boundary restrictions, and the point estimates are482actually slightly larger in magnitude on this restricted sample, suggesting that the major-483ity of the premium is not driven by seasonal variation in landscape quality. Additionally,484restricting the sample to each individual season (spring, summer, autumn, winter) yields a485statistically significant and premium between 2-3% for each specification (results available486from the author upon request).487A second approach to model this phenomenon is through interacting the secondary water488variable with a summer dummy variable that indicates whether the property sale occurred489during a summer month. Estimates from this interaction in the RDD are similar to the490previous estimated premium for secondary water (Table 2). Houses sold during the summer491months on average sell for higher prices than houses sold during other times of year. This492reflects the well-known seasonality of the property market.493The estimated premium for494secondary water remains sizeable and statistically significant. Notably, there is no evidence495of an additional premium for secondary water during the summer months.496Even when497transactions are made outside of the peak irrigation months, buyers still are willing to pay498a substantial premium for properties that offer unmetered water access. Both sets of results499in this section support the main findings in this paper, that households place a substantial500©Southern Regional Science Association 2024.501502503===== PAGE 13 (score 43, numeric-paren 20, table-words 6, var-words 8) =====504505GNAGEY: SPATIAL DISCONTINUITIES IN WATER RIGHTS506253507Table 5: Regression Results Testing for Higher Premiums for508Secondary Water During Summer Month Sales. Dependent509Variable is Natural Log of the Real House Price510Variable511All Houses512<1000ft513<500ft514<100ft515Secondary Water5160.0371***5170.0403***5180.0348***5190.0210**520(0.003)521(0.004)522(0.006)523(0.010)524Summer5250.0199***5260.0257***5270.0352***5280.0699***529(0.004)530(0.006)531(0.008)532(0.020)533Secondary*Summer534-0.0013535-0.0030536-0.0136537-0.0226538(0.005)539(0.007)540(0.010)541(0.025)542Observations54374,37554422,45454511,3855461,906547R-squared5480.74065490.74835500.74415510.7688552Control Variables553Yes554Yes555Yes556Yes557Boundary FE558Yes559Yes560Yes561Yes562Month/Year FE563Yes564Yes565Yes566Yes567Notes: Control variables include the complete set of variables presented in Table 2. Standard errors568clustered at each boundary segment. *** p<0.01, ** p<0.05, * p<0.1.569value on access to unmetered secondary water, and this value is not merely due to immediate570aesthetic benefits such as differences in curb appeal that would be more prominent in the571summer than during other seasons.5726.2.573Heterogenous Effects574While the estimated premium of having access to secondary water is relatively consistent575across the various specifications, the valuation of secondary access may vary across prop-576erty types. Interaction effects are estimated between secondary water access and property577characteristics to identify differences in the premiums across different properties. Table 6578presents the results of interactions between secondary water access and lot size, as well as579secondary water access and house size. While findings are consistent for the full sample,580for conciseness Table 6 shows results using properties within 500 feet and 100 feet of the581boundary.582There is no statistically significant difference in premium for larger lots, indicating that583purchasers of larger properties do not value the premium differently than smaller properties,584despite the theoretical prediction that it would be less costly to irrigate smaller properties585using primary culinary water. This finding may reflect that larger lots are more expensive,586and thus these consumers may be less price sensitive to relatively small expenditures like587water bills. Consistent with this hypothesis, buyers of larger houses exhibit a diminished588premium for secondary access than those of smaller houses. These findings suggest that589further investigation into the heterogenous effects of secondary water access could yield590important insights into the relationship between individuals or property characteristics and591the valuation of water infrastructure.592©Southern Regional Science Association 2024.593594595===== PAGE 14 (score 32, numeric-paren 14, table-words 4, var-words 8) =====596597254598The Review of Regional Studies 54(3)599Table 6: Regression Results Examining Whether the Premium600for Secondary Water Varies Across Property Types. Dependent601Variable is Natural Log of the Real House Price602Impact of lot size603Impact of house size604Variable605<500ft606<100ft607<500ft608<100ft609Secondary Water6100.0275*6110.0223**6120.0750**6130.0756*614(0.012)615(0.011)616(0.016)617(0.032)618Secondary*Acres6190.0115620-0.0314621(0.047)622(0.065)623Secondary*Sqft624-0.0231**625-0.0327*626(0.008)627(0.015)628Observations62911,3856301,90663111,3856321,906633R-squared6340.74626350.77346360.74666370.7741638Control Variables639Yes640Yes641Yes642Yes643Boundary FE644Yes645Yes646Yes647Yes648Month/Year FE649Yes650Yes651Yes652Yes653Notes: Control variables include the complete set of variables presented in Table 2. Standard errors654clustered at each boundary segment. *** p<0.01, ** p<0.05, * p<0.1.6556.3.656False Boundary Analysis657A key identifying assumption in this model is the absence of spatially varying unobservable658factors that are correlated with the boundary between areas with and without secondary659water access. These unobservables, which could range from historical land use patterns,660street layouts, distance to amenities, or socioeconomic sorting across space, could bias the661estimates of the premium for secondary water.662To test for the presence and potential impact of these spatially varying unobservables,663I employ a falsification test using artificial boundaries following a well-developed strategy664in analyses that rely on a geographical discontinuity RDD (Becker et al., 2016; Buggle,6652016; Conklin et al., 2020; Oto-Peral´ıas, 2020). If the main results are driven by the water666boundary itself, rather than an underlying spatial trend, then there should not be any effect667when false secondary water boundaries are used for the analysis.668False boundaries represent moving the actual boundary 750 and 2000 feet into the non-669secondary areas and also by moving the boundary 750 and 2000 feet into the area with670secondary water, respectively. These new boundaries are likely to be correlated with unob-671servable factors that vary across space in similar ways to the secondary water access points,672but properties across these false boundaries all have the same access (or non-access) to673secondary water.674For each of these false boundaries, the RDD analysis is replicated, treating houses on one675side of the boundary as if they have access to unmetered irrigation water and houses on the676other side of the boundary as if they do not have access. If the original results are robust,677there should not be any price discontinuities across these false boundaries.678©Southern Regional Science Association 2024.679680681===== PAGE 15 (score 22, numeric-paren 4, table-words 7, var-words 3) =====682683GNAGEY: SPATIAL DISCONTINUITIES IN WATER RIGHTS684255685Table 7: Regression Results Using Artificial Boundaries Moved686Into the Secondary Water Areas. Dependent Variable is Natural687Log of the Real House Price688False Boundary 750ft689False Boundary 2000ft690Variable691<500ft692<100ft693<500ft694<100ft695Secondary Water - Artificial Boundary6960.00046970.00816980.00196990.0076700(0.004)701(0.008)702(0.005)703(0.011)704Observations70513,6697062,8247078,0197081,616709R-squared7100.74737110.77937120.73667130.7585714Control Variables715Yes716Yes717Yes718Yes719Boundary FE720Yes721Yes722Yes723Yes724Month/Year FE725Yes726Yes727Yes728Yes729Notes: Control variables include the complete set of variables presented in Table 2. Standard errors730clustered at each boundary segment. *** p<0.01, ** p<0.05, * p<0.1.731Results from all four artificial boundaries can be found in Tables 7 and 8. Columns 1732and 3 in both tables restrict the analysis to buffers within 500 feet on either side of the733artificial boundary.734Columns 2 and 4 present the analyses with 100-foot buffers of the735artificial boundaries. Table 7 presents results where the artificial boundary was moved 750736feet and 2000 feet into the secondary area, while Table 8 presents the results where the737artificial boundary was moved 750ft and 2000ft respectively into the non-secondary areas.738All regression discontinuities fail to identify a statistically significant premium associated739with differences in the artificial boundary.7407.741CONCLUSION742Environmental attributes are capitalized into the sale price of houses, allowing policymakers743the ability to understand the premium households are willing to pay for a specific environ-744mental good or service. Access to an unlimited water right for irrigating lawns is a highly745valued attribute in the arid American West.746I estimate the value of access to unmetered irrigation water through a regression disconti-747nuity design which isolates differences in house prices on either side of a boundary separating748those with access and those without. The remaining difference in mean house value across749this boundary after controlling for property attributes is the valuation of this access.750Households are willing to pay a premium of approximately 2% for access to secondary751irrigation water. This result indicates that access to secondary water is highly valued, ap-752proaching the value of an additional bathroom for houses in the sample. This finding is753indicative of the value of secondary water access in Northern Utah. While the context may754be similar to other regions in the arid American West, direct extrapolation outside the study755area must be done cautiously absent further research on a broader geographic region.756Several promising avenues exist for future research on this topic. These include, but757©Southern Regional Science Association 2024.758759760===== PAGE 16 (score 18, numeric-paren 4, table-words 5, var-words 3) =====761762256763The Review of Regional Studies 54(3)764Table 8: Regression Results Using Artificial Boundaries Moved765Into the Non-Secondary Water Areas. Dependent Variable is766Natural Log of the Real House Price767False Boundary 750ft768False Boundary 2000ft769Variable770<500ft771<100ft772<500ft773<100ft774Secondary Water - Artificial Boundary7750.00237760.00247770.0040778-0.0218779(0.006)780(0.013)781(0.007)782(0.015)783Observations7847,8307851,5957866,0267871,415788R-squared7890.76047900.78927910.78427920.8007793Control Variables794Yes795Yes796Yes797Yes798Boundary FE799Yes800Yes801Yes802Yes803Month/Year FE804Yes805Yes806Yes807Yes808Notes: Control variables include the complete set of variables presented in Table 2. Standard errors809clustered at each boundary segment. *** p<0.01, ** p<0.05, * p<0.1.810are not limited to extending the analysis to other geographic regions, identifying temporal811changes in the premium through variation in drought periods, investigating the influence812of different demographic characteristics on the size of the premium, and exploring whether813access to secondary water influences landscaping decisions, such as xeriscaping vs traditional814lawns.815Such research directions are especially relevant given the ongoing policy debate about816effective measures to conserve water in dynamic regions that are experiencing demand pres-817sures from population growth, and supply pressures due to climate change reducing the818supply of water. As new technology facilitates better measurement of irrigation water using819the secondary water lines, households may expect to pay the full price of this previously820highly subsidized benefit to homeowners in certain neighborhoods. Furthermore, knowledge821of these values is critical to understanding potential pushback from households toward water822pricing policies based on use, as well as understanding the value that residents place on823irrigation water availability.824©Southern Regional Science Association 2024.825826827===== PAGE 17 (score 10, numeric-paren 0, table-words 0, var-words 9) =====828829GNAGEY: SPATIAL DISCONTINUITIES IN WATER RIGHTS830257831REFERENCES832Asabere, Paul K and Forrest E Huffman. (2009) “The relative impacts of trails and greenbelts833on home price,” The Journal of Real Estate Finance and Economics, 38, 408–419. http:834//doi.org/10.1007/s11146-007-9089-8.835Becker, Sascha O, Katrin Boeckh, Christa Hainz, and Ludger Woessmann. (2016) “The836empire is dead, long live the empire! Long-run persistence of trust and corruption in837the bureaucracy,” The Economic Journal, 126(590), 40–74. http://doi.org/10.1111/ecoj.83812220.839Black, Sandra E. (1999) “Do better schools matter? Parental valuation of elementary ed-840ucation,” The quarterly journal of economics, 114(2), 577–599. http://doi.org/10.1162/841003355399556070.842Bourassa, Steven, Martin Hoesli, and Jian Sun. (2005) “The price of aesthetic externalities,”843Journal of real estate literature, 13(2), 165–188. http://doi.org/10.1080/10835547.2005.84412090160.845Buggle, Johannes C. (2016) “Law and social capital: Evidence from the Code Napoleon in846Germany,” European Economic Review, 87, 148–175. http://doi.org/10.1016/j.euroecorev.8472016.05.003.848Chidsey Jr, Thomas C, Vicky Dyson, John Rogers, LaVonne J Garrison, Jim Davis, Cody849Stewart, and Samantha Mary Julian. (2018) “Utah Geological Survey,” Energy Policy,850801, 538–8700.851Conklin, James, Moussa Diop, and Herman Li. (2020) “Contact high: The external effects of852retail marijuana establishments on house prices,” Real Estate Economics, 48(1), 135–173.853http://doi.org/10.1111/1540-6229.12220.854Des Rosiers, Fran¸cois, Marius Th´eriault, Yan Kestens, and Paul Villeneuve. (2002) “Land-855scaping and house values: an empirical investigation,” Journal of real estate research,85623(1-2), 139–162. http://doi.org/10.1080/10835547.2002.12091072.857Gonzalez, Robert M. (2021) “Cell phone access and election fraud: evidence from a spatial858regression discontinuity design in Afghanistan,” American Economic Journal: Applied859Economics, 13(2), 1–51. http://doi.org/10.1257/app.20190443.860Gourley, Patrick. (2021) “Curb appeal: how temporary weather patterns affect house861prices,” The Annals of Regional Science, 67(1), 107–129.862http://doi.org/10.1007/863s00168-020-01042-x.864Hahn, Jinyong, Petra Todd, and Wilbert Van der Klaauw. (2001) “Identification and esti-865mation of treatment effects with a regression-discontinuity design,” Econometrica, 69(1),866201–209. http://doi.org/10.1111/1468-0262.00183.867Harrison Jr, David and Daniel L Rubinfeld. (1978) “Hedonic housing prices and the demand868for clean air,” Journal of environmental economics and management, 5(1), 81–102. http:869//doi.org/10.1016/0095-0696(78)90006-2.870Henry, Mark S. (1999) “Landscape quality and the price of single family houses: further871evidence from home sales in Greenville, South Carolina,” Journal of Environmental Hor-872ticulture, 17(1), 25–30. http://doi.org/10.24266/0738-2898-17.1.25.873Holmes, Thomas J. (1998) “The effect of state policies on the location of manufacturing:874Evidence from state borders,” Journal of political Economy, 106(4), 667–705. http://doi.875org/10.1086/250026.876©Southern Regional Science Association 2024.877