1# RECORD d5fa1a30f0b446809e7d5129c56b30f298ed13b52# TITLE: Land conservation policy and rental markets: evaluating unintended housing price effects3# YEAR: 2026.0 VENUE: The annals of regional science DOI: 10.1007/s00168-026-01528-04# AUTHORS: Corey Lang; Jarron VanCeylon5# PAGES: 19 CANDIDATE TABLE PAGES (1-based): [11, 13, 14, 15]6# ABSTRACT: Environmental policies can generate distributional impacts through the housing market, impacting homeowners and renters in unequal ways. While environmental improvements often benefit disadvantaged groups, one concern is that these improvements can increase the desirability of neighborhoods, which in turn can increase housing costs for renters and leaves the overall welfare effect ambiguous. In contrast, homeowners benefit from the additional amenities and the equity gain. This paper seeks to evaluate how rental prices respond to gains in nearby land conservation relative to owner-occupied home values using a panel of census and conservation data for the whole USA. We employ a first-difference regression model and use propensity score matching as a pre-processing method to improve the comparability of areas with and without new conservation. Our results suggest that local gains in conservation do not capitalize into rental prices but do capitalize into home values. Overall, our findings indicate that concerns for renters’ welfare and unintended consequences of land conservation are unsupported.789===== PAGE 1 (score 6, numeric-paren 0, table-words 0, var-words 6) =====1011Vol.:(0123456789)12The Annals of Regional Science (2026) 75:11013https://doi.org/10.1007/s00168-026-01528-014ORIGINAL PAPER15Land conservation policy and rental markets: evaluating 16unintended housing price effects17Corey Lang1 · Jarron VanCeylon218Received: 7 January 2026 / Accepted: 15 July 2026 / Published online: 18 August 2026 19© The Author(s) 202620Abstract21Environmental policies can generate distributional impacts through the housing 22market, impacting homeowners and renters in unequal ways. While environmen-23tal improvements often benefit disadvantaged groups, one concern is that these 24improvements can increase the desirability of neighborhoods, which in turn can 25increase housing costs for renters and leaves the overall welfare effect ambiguous. In 26contrast, homeowners benefit from the additional amenities and the equity gain. This 27paper seeks to evaluate how rental prices respond to gains in nearby land conserva-28tion relative to owner-occupied home values using a panel of census and conserva-29tion data for the whole USA. We employ a first-difference regression model and use 30propensity score matching as a pre-processing method to improve the comparability 31of areas with and without new conservation. Our results suggest that local gains in 32conservation do not capitalize into rental prices but do capitalize into home values. 33Overall, our findings indicate that concerns for renters’ welfare and unintended con-34sequences of land conservation are unsupported.35JEL classification Q24 · Q51 · Q52361 Introduction37Environmental public goods and hazards are unequally distributed, with minority 38and low-income communities facing far worse conditions (Brooks and Sethi 1997; 39Christensen et al. 2022; Colmer et al. 2020; Depro et al. 2015; Godwyll and Buzinde 402023; Jbaily et al. 2022). Due to unequal exposure, environmental improvements, 41 * Corey Lang 42 43clang@uri.edu44 * Jarron VanCeylon 45 46jvanceylon@uri.edu471 48Department of Environmental and Natural Resource Economics, University of Rhode Island, 49Kingston, USA502 51Department of Economics, St. John Fisher University, Rochester, USA525354===== PAGE 11 (score 67, numeric-paren 44, table-words 1, var-words 10) =====5556Land conservation policy and rental markets: evaluating…57Page 11 of 19 5811059Table 1 Summary statistics & sample comparison60Full sample means61Propensity score matched sample means62Variables: (Year 2000)63Conservation gain64No conservation gain65Standardized66difference67Conservation gain68No conservation gain69Standardized70difference71Median home value ($000 s)72174.5473161.93740.11575178.4276180.0277− 0.01478(106.86)79(111.72)80(108.91)81(120.72)82Median monthly rent ($)83801.9684835.0485− 0.10186812.5787812.15880.00189(328.04)90(328.33)91(338.76)92(342.70)93Median household income ($000 s)9463.659555.81960.3359763.679863.6999− 0.001100(24.16)101(22.72)102(24.72)103(26.01)104Population density (Pop./Acres)1051.401068.83107− 0.6781081.551091.74110− 0.045111(3.86)112(15.01)113(4.11)114(4.05)115Renter population (%)11620.6611733.75118− 0.68211921.5612022.01121− 0.027122(15.51)123(22.28)124(16.23)125(16.46)126High school educated (%)12753.6312851.321290.20513053.1013153.001320.010133(10.73)134(11.85)135(10.86)136(11.22)137College educated (%)13827.7813926.031400.10914127.9714228.04143− 0.004144(15.52)145(16.48)146(16.01)147(16.42)148Minority population (%)14914.8315032.33151− 0.64015216.6215317.38154− 0.033155(21.91)156(31.88)157(23.19)158(23.56)159Median rooms1605.861615.311620.6191635.791645.781650.023166(0.81)167(0.96)168(0.82)169(0.85)170Median year built1711970.231721963.621730.4561741971.401751971.361760.003177(13.36)178(15.56)179(13.07)180(13.36)181Distance to CBSA (mi.)18216.311839.781840.55618516.3818616.40187− 0.002188(12.63)189(10.76)190(13.19)191(14.74)192Conserved acres (%)19311.691944.981950.45119612.0619712.19198− 0.006199200201===== PAGE 12 (score 9, numeric-paren 4, table-words 2, var-words 1) =====202203 204C. Lang, J. VanCeylon 205110 206Page 12 of 19207Standard deviations in parenthesis. We split the sample into block groups that experienced a gain in conservation between 2001 and 2009 and block groups that did not. 208All variables are in 2000 levels. A standardized difference greater than 0.1 in absolute value indicates imbalance between treatment and control209Table 1 (continued)210Full sample means211Propensity score matched sample means212Variables: (Year 2000)213Conservation gain214No conservation gain215Standardized216difference217Conservation gain218No conservation gain219Standardized220difference221(17.53)222(11.61)223(18.02)224(20.22)225Observations22614,223227148,03622811,84122947,091230231232===== PAGE 13 (score 37, numeric-paren 6, table-words 8, var-words 13) =====233234Land conservation policy and rental markets: evaluating…235Page 13 of 19 2361102372018; Lang et al. 2023). Further, the owner findings indicate that the null results for 238renters are not a consequence of the research design nor data limitations.2394.1 Heterogeneous price impacts for renters240One concern in our study is that new conservation may financially burden vulner-241able households. While we see little evidence of price responses for the median 242monthly rents, there may be heterogeneous price impacts at different parts of the 243price distribution or for renters in low income or racially segregated neighborhoods. 244In Table 3, we estimate our preferred first-difference model (Table 2 Column 3) 245for median monthly rent and interact conservation with a series of socioeconomic 246indicator variables: below median rent, below median income, and above median 247minority population. Across all columns, none of the interaction coefficients are sta-248tistically significant, and all are small in magnitude, suggesting that a lack of rental 249response is common across these different types of neighborhoods.250We propose that conservation may impact rents at different points along the dis-251tribution of prices. Since our main analysis only uses median monthly rents, we 252may miss part of the story. Following the approach of Gamper-Rabindran and Tim-253mins (2013), we construct rent price quintiles for each block group and estimate 254Table 2 Homeowner & renter capitalization models255Observations are at the block group level and are weighted by total households. All specifications include 256full set of differenced housing & neighborhood characteristics including median household income, 257median rooms, median year built, housing units, % minority population, % high school educated, % col-258lege educated, % renter population, % over age 65, % under age 18, % employed population, vacancy 259rate, housing units, population density, % population employed in real estate, finance, insurance, scien-260tific management, or administrative fields, as well as base-year 2000 controls: % acres in 2000, block 261group neighbor acres in 2000, housing units in 2000, distance to nearest CBSA and distance to nearest 262CBSA squared. For region and state interactions, all housing, neighborhood, and base-year character-263istics are interacted with indicators for a specific region or state. All models include county-trend fixed 264effects. Robust standard errors are clustered at the tract level and shown in parentheses: *** p < 0.01, ** 265p 0.05, * p < 0.1266Dependent variable: Δ Log 267median rent268Dependent variable: Δ Log Median value269Independent variables270(1)271(2)272(3)273(4)274(5)275(6)276ΔConserved acres (%)2770.0006278–0.0002279–0.00052800.00352810.00312820.0027283(0.0009)284(0.0008)285(0.0008)286(0.0008)***287(0.0007)***288(0.0007)***289Controls:290Housing & neighbor-291hood characteristics292X293X294X295X296X297X298County-Trend FE299X300X301X302X303X304X305Region interactions306X307X308X309State interactions310X311X312Observations31358,93231458,93231558,93231658,93231758,93231858,932319R-squared3200.28003210.45203220.48443230.45673240.49103250.5304326327328===== PAGE 14 (score 40, numeric-paren 10, table-words 11, var-words 6) =====329330 331C. Lang, J. VanCeylon 332110 333Page 14 of 19334the effects of conservation changes on the 20th, 40th, 60th, and 80th price percen-335tiles. In Table 4, our results indicate that rental prices across the distribution do not 336respond to gains in conservation. The coefficient on conservation change in Column 3373 for the 60th percentile of rent prices has the largest magnitude of all percentiles 338at 0.0009, but there is no clear pattern across quintiles. All recovered estimates are 339not statistically different from zero and do not indicate a divergence from the main 340findings. We see this as a positive sign that vulnerable tenants at the bottom of the 341income distribution are not negatively impacted by local gains in conserved lands.3424.2 Alternative hypotheses343In the online appendix, we test several hypotheses that would undermine our conclu-344sion. First, if proximity is the primary avenue for valuation of conservation, rental 345units may be less likely to be proximate than owner-occupied homes. We subset 346block groups that are small, so all housing units would be relatively proximate, and 347still find no rent impact (Table A4). Second, it is possible that changes in the rental 348market lag the owner market (Lang 2015), in which case we may not have enough 349data to observe the eventual uptick in monthly rent. We test for lagged effects in 350rental prices and perform a placebo test using conservation from 1990–1999, simi-351larly finding no price response for either model (Table A5 and Table A6).352Table 3 Heterogeneous price impacts for renters353Observations are at the block group level and are weighted by total households. Binary interaction vari-354ables are defined as above or below the median for that variable. All specifications include the full set of 355controls as in Table 2 Column 3. Robust standard errors clustered at the tract level and shown in paren-356theses: *** p < 0.01, ** p < 0.05, * p < 0.1357Independent variables358Dependent variable: Δlog median rent359(1)360(2)361(3)362(4)363ΔConserved acres (%)364− 0.0007365− 0.0005366− 0.0008367− 0.0010368(0.0009)369(0.0012)370(0.0010)371(0.0013)372ΔConserved acres * Low rental prices (0/1)3730.00043740.0001375(0.0016)376(0.0020)377ΔConserved acres * High minority population (0/1)378− 0.00013790.0005380(0.0014)381(0.0015)382ΔConserved acres * Low-income area (0/1)3830.00063840.0005385(0.0014)386(0.0017)387Controls:388Housing & neighborhood characteristics389X390X391X392X393County-trend FE394X395X396X397X398State interactions399X400X401X402X403Observations40458,93240558,93240658,93240758,932408R-squared4090.48634100.48484110.49184120.4947413414415===== PAGE 15 (score 24, numeric-paren 4, table-words 6, var-words 7) =====416417Land conservation policy and rental markets: evaluating…418Page 15 of 19 4191104205 Conclusion421This study establishes a novel relationship between rental prices and the conserva-422tion of open space at a national level. While land conservation can enhance the value 423of properties by offering valuable spatial amenities, there are documented environ-424mental justice and equitable distribution concerns like gentrification and access 425issues (Lang et al. 2023; Sims et al. 2022). We examine the housing market impacts 426of conservation efforts between 2001 and 2009 using national-scale census data and 427find positive effects on owner-occupied home values but no impact on monthly rent. 428These results differ from previous findings focused on different amenities where rent 429prices capitalize on amenity changes, just at a lower rate compared to owner values 430(Banzhaf and Farooque 2013; Bento et al. 2015; Davis 2011; Grainger 2012).431One concern about environmental improvements is that, while they provide 432benefits for many, some longstanding or vulnerable residents may be displaced 433due to rising housing prices, a process referred to as environmental gentrification 434or green gentrification (Black and Richards 2020; Immergluck and Balan 2018). 435Indeed, there are times when people may advocate against environmental improve-436ments, such as land conservation, but also cycling infrastructure and river restora-437tion, for fear of green gentrification (Pearsall and Anguelovski 2016; Harris et al. 4382020). Additionally, even when concerns about green gentrification or inequitable 439access to conserved land are present among practitioners, action can be limited due 440to resources, lack of necessary data, or being beyond their scope of mission (Nesbitt 441et al. 2023; Keller et al. 2024). Our findings may help relieve consternation among 442policymakers, NGOs, and community groups concerned about the negative impacts 443of local conservation on vulnerable communities. One possible explanation of our 444findings is that our sample primarily comprises small-scale investments instead of 445Table 4 Heterogeneous impacts 446on the distribution of rent prices447Observations are at the block group level and weighted by total 448households. All specifications include the full set of controls as in 449Table 2 Column 3. All models include county-trend fixed effects. 450Robust standard errors clustered at the tract level in parentheses: *** 451p < 0.01, ** p < 0.05, * p < 0.1452Dependent variable: Δ log median Rent 453(Percentile)45420th45540th45660th45780th458Independent variables459(1)460(2)461(3)462(4)463Δ Conserved acres (%)4640.0004465–0.00024660.00094670.0003468(0.0017)469(0.0011)470(0.0009)471(0.0009)472Controls:473Housing & neighbor-474hood characteristics475X476X477X478X479County-trend FE480X481X482X483X484State interactions485X486X487X488X489Observations49058,25349158,25349258,25349358,253494R-squared4950.25584960.35484970.38634980.3419499500501===== PAGE 16 (score 12, numeric-paren 0, table-words 0, var-words 12) =====502503 504C. Lang, J. VanCeylon 505110 506Page 16 of 19507transformative investments, such as Atlanta’s BeltLine or New York City’s High 508Line, which is somewhat consistent with “just green enough” practices that seek 509to improve environmental quality without displacing residents (Black and Richards 5102020; Curran and Hamilton 2012). Kim and Wu (2022) additionally find evidence 511that the characteristics of green space matter and not all types foster gentrification. 512While certainly more support is needed, these data points suggest that some of the 513concern about environmental gentrification is perceived and not actual.514A couple directions of future research would help bolster our findings or con-515textualize them. Clearly, utilizing individual, geocoded rental transactions instead 516of aggregated median monthly rent would improve the analysis, though we know 517of no publicly available source. Second, while we assume that preferences for con-518served land are the same for renters and owners, we are unable to verify that. Stated 519preference research reports mixed results comparing homeowner and renter prefer-520ences. Tu et al. (2016) find that renters (N = 180) are willing to pay more than own-521ers to live closer to parks and that homeowners with substitutes like private yards 522are willing to pay even less. In contrast, more general contingent valuation studies 523find that homeowners have a higher WTP than renters for neighborhood amenities 524(restaurants, public transportation options) (Caplan et al. 2021; Lindsey and Knaap 5251999). However, it is possible that homeowners exhibit a higher WTP due to cur-526rent scarcity; renters likely have more of these specific neighborhood amenities due 527to proximity to downtown. Thus, a large-scale survey that directly compares home-528owner and renter preferences for conserved open space would be valuable for future 529research.530Supplementary Information The online version contains supplementary material available at https://doi.531org/10.1007/s00168-026-01528-0.532Acknowledgements For useful feedback, we thank Pengfei Liu and Ben Blachly.533Author contribution C.L. and J.V. conceived the study, developed the theoretical framework, and co-534wrote the manuscript. J.V. conducted the primary data collection and analysis. C.L. secured funding and 535supervised the project. Together, J.V. and C.L. refined the analyses and finalized the manuscript. All 536authors reviewed and approved the final version.537Funding National Institute of Food and Agriculture, USA, 2018-67024-27695538Data availability Anonymous link for data replication package available at: [https://osf.io/fmwun/overv539iew?view_only=f882f0338d8b43478168b878355cfe14].540Declarations 541Conflict of interest The authors declare no conflict of interests.542Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, 543which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as 544you give appropriate credit to the original author(s) and the source, provide a link to the Creative Com-545mons licence, and indicate if changes were made. The images or other third party material in this article 546are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the 547material. If material is not included in the article’s Creative Commons licence and your intended use is 548not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission 549