1# RECORD 4326e33d5aa8be80e2f1d1ae91adb42331da513d2# TITLE: E-Grocery and Housing Markets: Evidence from the Amenity Effect of Amazon Fresh as a Digital Amenity3# YEAR: 2026.0 VENUE: Journal of real estate finance and economics DOI: 10.1007/s11146-026-10055-w4# AUTHORS: Jamie Chung; Liuming Yang5# PAGES: 41 CANDIDATE TABLE PAGES (1-based): [12, 14, 19, 22, 23, 24, 25, 27, 30, 38, 39]6# ABSTRACT: 7 8 This paper examines the impact of e-grocery entry on the housing market, focusing on Amazon Fresh, which provides grocery items and delivery services without a physical store. We find that areas receiving Amazon Fresh services experience a 2.5% increase in housing prices compared to areas within the same metropolitan region but do not receive such services. Local households show increased online consumption following Amazon Fresh’s entry, highlighting its amenity effect. Furthermore, the price effect is stronger in areas where households are more inclined to use online platforms and in socio-economically disadvantaged areas, including those with lower income, less education, higher rental occupancy, higher minority populations, lower vehicle ownership, and limited retail access. These patterns have policy implications for alleviating the "9 Food Desert10 " issue in urban areas through technological advancements in grocery markets by leveraging digital grocery services.11121314===== PAGE 1 (score 8, numeric-paren 0, table-words 0, var-words 8) =====1516The Journal of Real Estate Finance and Economics17https://doi.org/10.1007/s11146-026-10055-w18Abstract19This paper examines the impact of e-grocery entry on the housing market, focus20ing on Amazon Fresh, which provides grocery items and delivery services without 21a physical store. We find that areas receiving Amazon Fresh services experience a 222.5% increase in housing prices compared to areas within the same metropolitan 23region but do not receive such services. Local households show increased online 24consumption following Amazon Fresh’s entry, highlighting its amenity effect. Fur25thermore, the price effect is stronger in areas where households are more inclined 26to use online platforms and in socio-economically disadvantaged areas, including 27those with lower income, less education, higher rental occupancy, higher minor28ity populations, lower vehicle ownership, and limited retail access. These patterns 29have policy implications for alleviating the "Food Desert" issue in urban areas 30through technological advancements in grocery markets by leveraging digital gro31cery services.32Keywords E-grocery · Digital amenity · Housing market · Retail desert33JEL Classification D12 · D63 · J15 · L81 · R12 · R3234Received: 18 June 2025 / Accepted: 24 March 202635© The Author(s) 202636E-Grocery and Housing Markets: Evidence from the 37Amenity Effect of Amazon Fresh as a Digital Amenity38Jamie Chung139 · Liuming Yang240Jamie Chung and Liuming Yang contributed equally to this work.41 42 Jamie Chung43jeonghyunchung@unomaha.edu44Liuming Yang45liumingyang@cuhk.edu.hk461 47University of Nebraska Omaha, Omaha, USA482 49The Chinese University of Hong Kong, Hong Kong, China501 3515253===== PAGE 12 (score 17, numeric-paren 0, table-words 2, var-words 10) =====5455J. Chung, L. Yang56value of online consumption is $123,355, which reveals that online shopping con57stitutes a minor proportion of overall consumption throughout the observed time58frame. By computing the share, it's evident that online expenditure contributes to 59around 2.25% of the total expenditure. Despite falling below, the national average, 60which stands at 5.6% over 2004–2019 according to the Census Bureau's 2022 data 61(Lee & Zheng, 2025), NHCP dataset holds potential for identifying the changes in 62patterns of online and offline consumptions.14 As mentioned, the majority of total 63consumption comes from offline consumption in NHCP, which accounts for around 6497.57% ($4,960,819/$5,084,173) on average. Panel C illustrates the socioeconom65ics characteristics at the ZIP Code level, which are included as control variables in 66the model. On average, each ZIP Code comprises a population of approximately 6733,299 individuals, with 37.32% of them holding bachelor’s degree or higher edu68cational qualifications. Furthermore, 37.30% of households opt to rent rather than 6914 Literature about using Nielsen Household Panel Data to explore the online consumption pattern could be 70found in Baker, et al. (2021); He (2023); Lee and Zheng (2025).71Table 1 Summary statistics72Panel A. Property Values73Obs74Mean75Std. Dev76Min77Max78Home Value ($)79118,51780364,89181241,8038228,063831,074,54884Home Value ($)—1 bedroom8554,60886223,01987150,8478816,97589698,86590Home Value ($)—2 bedrooms91107,18692254,34193170,5929421,06895766,42596Home Value ($)—3 bedrooms97115,61998343,94899219,07510036,1461011,004,458102Home Value ($)—4 bedrooms103106,735104447,683105288,19510643,9631071,421,147108Home Value ($)—5 bedrooms10987,603110623,715111463,45711254,7681132,467,025114Panel B. Household Consumptions115Obs116Mean117Std. Dev118Min119Max120Total Consumption ($)121183,0111225,084,1731234,845,73212467012563,700,000126Online Consumption ($)127183,011128123,355129400,733130013122,600,000132Offline Consumption ($)133183,0111344,960,8191354,727,535136013760,400,000138Share of Online Consumption (%)139183,0111400.02251410.06031420.00143100.00144Panel C. Socioeconomics145Obs146Mean147Std. Dev148Min149Max150Population151118,51715233,29915319,688154615155111,893156High Education (%)157118,51715837.3215919.661602.5616186.26162Renter Occupied (%)163118,51716437.3016518.611661.1416788.96168Non-White Population (%)169118,51717038.6017123.861721.5217399.25174Median Age175118,51717636.911775.1517821.1017976.20180Median Household Income ($)181118,51718269,62818328,48318412,994185190,995186Non-Vehicle Ownership (%)187118,51718813.9618916.741900.0019185.50192Household Size193118,5171942.771950.481961.421975.18198Vacancy Rate199118,5172000.072010.4720202030.54204Competing Retail Est. per households205118,5172060.092070.0720802090.82210This table presents the summary statistics for the data used in the analysis. Panel A provides summary 211statistics for home values, both overall and by property type. Panel B offers summary statistics for 212household consumption, with the Share of Online Consumption calculated by dividing online 213consumption by total consumption. Lastly, Panel C outlines the socio-economic factors at the ZIP Code 214level.2151 3216217218===== PAGE 13 (score 16, numeric-paren 0, table-words 2, var-words 12) =====219220E-Grocery and Housing Markets: Evidence from the Amenity Effect of…221own a home. About 38.6% of the population identifies as non-White. The median 222age and median household income are 37 years old and $69,628, respectively. The 223ownership of automobiles can potentially impact the utilization of online services, 224and it’s noteworthy that around 14% of households do not possess a car. The aver225age household size is 2.77, and the housing vacancy rate is approximately 7%. 226Local retail market dynamics may influence the adoption of e-grocery services 227and consequently affect our results. Following An & Chung (2025), we identify 228competing retail stores relevant to e-grocers, including supermarkets, convenience 229stores, and retailers of daily necessity goods.15 On average, there are 0.09 compet230ing retail stores per household within each ZIP Code.231Methodology232Our main research question is whether e-grocery service is treated as an amenity and 233increases housing values. To answer this question, we run the difference-in-difference 234regression described as:235 Yl,t = α + β1EGroceryl + β2Postt + β3EGroceryl × Postt + βiXl,t + γl + δt + ϵl,t236where the dependent variable is the logarithm of home value in ZIP Code l at year-237month t. EGrocery is an indicator variable which equals to 1 if Amazon Fresh service 238is available in ZIP Code l and 0 if Amazon Fresh service is not available although 239it is located within the same MSA. This approach is used in a recent paper by An & 240Chung (2025) that investigates the impact of Amazon Fresh on the local competing 241retail businesses. Post is a time indicator variable which equals to 0 for 36 months 242prior to the entry of Amazon Fresh and 1 for 36 months after the entry of Amazon 243Fresh. The interaction term of EGrocery and Post is the main variable of our interest, 244which captures the impact of Amazon Fresh service on the local home values. If β3 245is positive and statistically significant, it indicates that Amazon Fresh service has a 246positive amenity effect on local home values. Conversely, if β3 is negative or statisti247cally insignificant, it describes that the e-grocery service either results in a decrease 248in home values or has no discernible impact.249As our dependent variable is the average home values at the ZIP Code level, we 250include local socio-economic factors that could affect the housing market. Specifi251cally, Xl,t includes a vector of local factors such as the logged population, the logged 252median age, the logged number of individuals in the labor force, the percentage of 253the non-White population, the percentage of individuals with a bachelor’s degree or 254higher, the logged median household income, the vacant unit rate, unemployment 255rate, and the logged median household size at the ZIP Code level. We include a ZIP 256Code fixed effect to account for time-invariant location-related factors and a year-257month fixed effect to mitigate the influence of seasonal effects exist in the housing 258market. Standard errors are clustered at the ZIP Code level.25915 See An and Chung (2025) for the complete list of NAICS codes for competing retail stores.2601 3261262263===== PAGE 14 (score 32, numeric-paren 10, table-words 7, var-words 7) =====264265J. Chung, L. Yang266Empirical Findings267Baseline: Impact of Amazon Fresh on Housing Prices268Table 2 represents the results when we run the regression following Eq. 1. In column 2691, where ZIP Code level control variables are excluded, we observe a 2.5% increase 270in home values following the entry of Amazon Fresh. When local control variables 271are introduced in column 2, the estimated coefficient remains positive and statisti272cally significant at the 1% level. The magnitude of the effect is unchanged, indicating 273an average increase in home values of approximately 2.5%, which corresponds to 274about $9,122 over a three-year period within each ZIP Code.16275This positive price response suggests that Amazon Fresh is perceived as a valuable 276local amenity in the housing market. Notably, the magnitude of the effect is compara277ble to the estimated amenity value of physical grocery stores documented in the prior 278literature (e.g., Pope & Pope, 2015).17 Increases in housing prices are typically ben279eficial for homeowners, reflecting a positive amenity effect. While some may raise 280concerns that higher assessed values could lead to increased property tax burdens, 281particularly for homeowners who do not sell, higher house values nonetheless rep282resent an improvement in housing wealth. They could also expand opportunities for 283extracting home equity through mechanisms such as refinancing, home-equity loans, 284or reverse mortgages, even in the absence of a transaction.28516 It indicates the variation of ZIP Code level controls are largely absorbed in ZIP Code fixed effects.28617 For example, Pope & Pope (2015) find that the opening of a Walmart leads to an increase in nearby 287housing prices by 2% to 3%.288Table 2 The impact of e-grocery on housing prices289(1)290(2)291(3)292(4)293(5)294Dep. Variable: ln(ZHVI)295Baseline296Anticipation Effects297Announcement Within Market298Across Market 299Amazon Fresh x Post3000.025***3010.025***3020.029***3030.020***3040.028***305(0.001)306(0.001)307(0.003)308(0.002)309(0.002)310Observations311118,517312118,51731380,30431493,15431597,413316R-squared3170.9893180.9903190. 9293200.9903210.989322Controls323no324yes325yes326yes327yes328ZIP Code Fixed Effect329yes330yes331yes332yes333yes334Year x Month Fixed Effect335yes336yes337yes338yes339yes340This table presents the influence of Amazon Fresh on home values. In column 1, we exclude the ZIP 341Code level control variables while in column 2, we include the control variables in the model. ZIP 342Code fixed effect and year by month fixed effect are included in columns 1 and 2. In column 3, we 343exclude two markets, San Francisco and New York, where Amazon Fresh announced entry prior to 344the commencement of actual operations. Column 4 addresses within-market anticipation effects by 345excluding ZIP Codes that are immediately adjacent to treated ZIP Codes. Column (5) addresses across-346market anticipation effects by excluding three markets, San Francisco, San Diego, and Philadelphia, 347where nearby markets received the service prior to their own openings. Standard errors are clustered at 348the ZIP Code level. Robust standard errors, clustered at the ZIP Code level, are reported in parentheses. 349***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.3501 3351352353===== PAGE 15 (score 7, numeric-paren 0, table-words 2, var-words 3) =====354355E-Grocery and Housing Markets: Evidence from the Amenity Effect of…356Anticipation Effects357We next examine whether anticipation effects influence our estimates. Although entry 358dates generally align with the timing of actual operations and are verifiable for most 359markets, we identify two exceptions in which early media coverage preceded the 360official service launch. In New York, an August 2014 report documented Amazon 361Fresh’s anticipated entry ahead of its official October launch, while in San Francisco, 362an NPR article published in June 2013 described the service’s forthcoming arrival 363several months before its official December launch.18 Such early coverage may 364induce announcement effects in local housing markets. To address this concern, we 365exclude these two markets from the sample and re-estimate the model in column 3 of 366Table 2. The estimated effect remains robust, with home values increasing by 2.9%.367We further consider two types of potential anticipation spillovers. The first 368is within-market anticipation effects, whereby residents in ZIP Codes adjacent to 369treated ZIP Codes may anticipate imminent service expansion. To mitigate this con370cern, we exclude all geographically adjacent ZIP Codes from the control group and 371re-estimate the model using only non-adjacent ZIP Codes as the comparison group 372in column 4. This restriction substantially reduces the scope for within-market spill373overs. The estimated effect declines modestly to 2.0%, suggesting some anticipation 374among immediate neighbors. Importantly, however, the estimate remains positive 375and statistically significant, and its magnitude is consistent with the baseline results, 376indicating that within-market anticipation does not materially affect our conclusions.377The second concern is across-market anticipation effects, which may arise when 378nearby markets experience closely timed openings. We identify three such markets, 379San Francisco, San Diego, and Philadelphia, that launched Amazon Fresh within 380three years of larger neighboring markets (Los Angeles or New York) and are located 381within 150 miles or within the same state. To test for cross-market spillovers, we 382exclude these later-opening markets in column 5. After removing these markets, 383the estimated effect increases by 0.3% (2.8% compared to the baseline estimate of 3842.5%). This pattern suggests that within-market anticipation is more pronounced than 385across-market anticipation, likely reflecting greater exposure to and familiarity with 386the service at close geographic proximity. Overall, the results across all specifications 387remain robust, indicating that anticipation effects are limited and do not drive our 388main findings.389Parallel Trend390To examine the timing of the price effects, we conduct dynamic analyses of Amazon 391Fresh’s entry. Figure 3 presents coefficient plots from these event-study specifica392tions, illustrating the annual effects of Amazon Fresh’s entry on home values rela393tive to the year prior to entry. In Panel A, which uses a balanced three-year pre- and 394post-entry window, the estimated coefficients in the pre-entry period are statistically 39518 https://www.bizjournals.com/newyork/blog/techflash/2014/08/spotted-in-nyc-amazonfresh-trucks.htm396l; https://www.npr.org/sections/thesalt/2013/06/05/188959969/amazon-s-grocery-delivery-a-trojan-hors397e-to-get-into-your-door3981 3399400401===== PAGE 19 (score 35, numeric-paren 14, table-words 7, var-words 5) =====402403E-Grocery and Housing Markets: Evidence from the Amenity Effect of…404Table 3 Robustness tests405(1)406(2)407(3)408(4)409(5)410(6)411(7)412Dep. Variable: ln(ZHVI)413FHFA HPI414Weighted 415Sample416Treated vs 417Not-yet-Treated418Focusing on Neighboring 419ZIP Codes420Excluding Alternative 421E-grocery Service422Atlanta Case423Initial Service 424ZIP Codes425Continued Treated 426vs. Never Treated427Amazon Fresh x Post4280.026***4290.028***4300.015***4310.019***4320.045***4330.073***4340.089***435(0.002)436(0.001)437(0.001)438(0.006)439(0.001)440(0.010)441(0.027)442Observations44310,669444118,517445191,82744650,73644799,02944819,4654494,088450R-squared4510.9904520.9904530.9914540.9914550.9904560.9854570.984458Controls459yes460yes461yes462yes463yes464yes465yes466ZIP Code Fixed Effect467yes468yes469yes470yes471yes472yes473yes474Year x Month Fixed Effect475yes476yes477yes478yes479yes480yes481yes482This table presents the results of robustness tests. Column 1 uses the Federal Housing Finance Agency (FHFA) home price index as the dependent variable. Column (2) 483re-estimate the baseline specification by weighting observations by the share of treated ZIP Codes within each market. Column 2 re-estimates the baseline model using 484weights equal to each market’s share of treated ZIP Codes. Column 3 compares the impacts between treated areas and not-yet-treated areas, where the not-yet-treated areas 485are the cities that received Amazon Fresh services in later years, beyond the sample period. Column 4 incorporates control ZIP Codes that are geographically adjacent to 486the treated ZIP Codes; treated ZIP Codes without adjacent control ZIP Codes are excluded from this analysis. Column 5 excludes ZIP Codes within the service areas of 487alternative e-grocery providers to ensure the results are not confounded by overlapping service regions. Standard errors are clustered at the ZIP Code level. The last two 488columns focus on the Atlanta metropolitan market: column 6 utilizes the initial Amazon Fresh service area data from 2016, while column 7 uses the continued service 489area from 2016 to 2022, comparing these two areas where Amazon Fresh was never introduced during the sample period. Robust standard errors, clustered at the ZIP 490Code level, are reported in parentheses. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.4911 3492493494===== PAGE 20 (score 13, numeric-paren 0, table-words 3, var-words 7) =====495496J. Chung, L. Yang497Subsample Analysis: Neighboring ZIP Codes498Although the coefficient plots (Fig. 3) indicate that home values are almost iden499tical within the metropolitan statistical areas prior to Amazon Fresh’s entry, we 500may still be concerned that ZIP Codes that receive e-grocery services are system501atically different from those without such services. To address this concern, we 502use an alternative approach by exclusively comparing geographically adjacent ZIP 503Codes because geographically contiguous regions often share similar unobservable 504characteristics, such as local economic conditions and labor market dynamics. Fol505lowing the methodology of Dube, et al. (2010), who leverage spatial proximity to 506estimate the effects of minimum wage policies on employment and earnings, we 507restrict our analysis to only adjacent ZIP Codes, rather than including all ZIP Codes 508in the market.509Specifically, we focus on treated ZIP Codes located along the edge of the Ama510zon Fresh service area and use their neighboring ZIP Codes (e.g., shaded in green 511in Fig. 2) as the control group.21 This approach allows us to compare nearby ZIP 512Codes that are likely similar in economic and demographic factors, differing pri513marily in access to Amazon Fresh services. The results are presented in Table 3 514column 4. We find that areas with Amazon Fresh services experience a robust 1.9% 515increase in home values compared to areas without Amazon Fresh services. These 516findings suggest that the positive pricing effects observed in our analysis are pri517marily driven by the entry of Amazon Fresh, indicating that e-grocery services can 518function as a positive amenity for local households, similar to the effect of a physi519cal big-box store.520Alternative e-grocer: Instacart521While Amazon Fresh emerged as a dominant player, accounting for approximately 522one-third of the U.S. online grocery market during the study period, other e-grocers 523may have simultaneously entered focal markets, potentially introducing estimation 524bias. To address this concern, we design a regression model that excludes overlap525ping periods between Amazon Fresh and Instacart. We focus on Instacart as the pri526mary alternative for two reasons. First, Instacart experienced rapid expansion during 527the sample period and became the second-largest e-grocer nationally.22 Second, like 528Amazon Fresh, Instacart publicly delineates its service areas at the ZIP Code level 529on its website, allowing for precise spatial identification of market entry. Our analy530sis uses a pre-period spanning three years prior to the first opening of either service 531(Amazon Fresh or Instacart) and a post-period covering three years after Amazon 532Fresh’s entry.23 The results, presented in Table 3 column 5, indicate that the posi533tive pricing impact remains significant when using this clean pre- and post- period 53421 We exclude treated ZIP Codes that are not geographically adjacent to any control ZIP Codes.53522 According to eMarketer, Instacart held a 3.7% market share of the U.S. digital grocery sales market in 5362017, which increased to 5.9% in 2018, 10.2% in 2019, and 21.5% in 2020. https://www.oberlo.com/stati537stics/instacart-market-share Accessed in June 2025.53823 This empirical setting is illustrated graphically in Figure A3.5391 3540541542===== PAGE 22 (score 19, numeric-paren 4, table-words 5, var-words 5) =====543544J. Chung, L. Yang545appeal of online grocery services (CBRE, 2022). Lastly, the Atlanta metropolitan 546area features a highly diverse and youthful population, including a large proportion 547of professionals and university students, who are among the most active users of 548digital platforms. These combined factors likely amplify the positive housing market 549effects observed in the Atlanta case, relative to other urban markets. However, it is 550more crucial to emphasize that, even after accounting for potential dynamic changes 551in service areas, we can still conclude that the e-grocery service amenity is capital552ized into housing prices.553Heterogeneous Treatment Effects554Although our analysis focuses on 11 markets that adopted Amazon Fresh over a rela555tively short period (2013–2016), the two-way fixed effects (TWFE) estimator may still 556be limited in its ability to accommodate heterogeneous treatment effects or staggered 557adoption. To address this concern, we follow De Chaisemartin and d’Haultfoeuille 558(2023) and implement two alternative estimators: (i) the did_multiplegt estimator 559and (ii) the did_imputation estimator. Both approaches recover group–time average 560treatment effects and avoid the negative-weight problem associated with TWFE esti561mators. Using these methods, we continue to find positive and statistically significant 562effects of Amazon Fresh access on housing prices, with estimated increases of 2.6% 563and 4.4%, respectively, as reported in columns 1 and 2 of Table 4. These robustness 564checks indicate that our main conclusions are not driven by the TWFE specification 565or by heterogeneous treatment effects arising from staggered adoption.566Mechanism567Household Consumption Patterns568We find that the entry of Amazon Fresh services positively impacts local home 569values. We hypothesize that this effect is attributable to the perception of e-gro570(1)571(2)572Dep. Variable: ln(ZHVI)573did_multiplegt574did_imputation575Amazon Fresh x Post5760.026***5770.044***578(0.008)579(0.0001)580Observations581118,517582118,517583Controls584yes585yes586ZIP Code Fixed Effect587yes588Yes589Year x Month Fixed Effect590yes591yes592This table shows the results when we use did_multiplegt and 593did_imputation estimators suggested by De Chaisemartin and 594d’Haultfoeuille (2023). All regressions include ZIP Code fixed 595effects, year-by-month fixed effects, and the full set of baseline 596controls. Robust standard errors, clustered at the ZIP Code level, 597are reported in parentheses. ***, **, and * indicate statistical 598significance at the 1%, 5%, and 10% levels, respectively.599Table 4 Heterogeneous treat600ment effects601 6021 3603604605===== PAGE 23 (score 32, numeric-paren 6, table-words 11, var-words 3) =====606607E-Grocery and Housing Markets: Evidence from the Amenity Effect of…608cery services as an amenity for local households. To understand the underlying 609mechanism behind this positive pricing effect, we examine whether local house610holds alter their consumption behavior subsequent to Amazon Fresh’s entry. Using 611the Nielsen Household Consumer Panel (NHCP) data, we run the regressions 612described in Eq. 1 on the household consumption patterns. The results are pre613sented in Table 5.614First, we conduct a regression on the logarithm of total consumption to gauge 615whether there are changes in overall household consumption subsequent to Ama616zon Fresh’s introduction. In column 1, we find that total consumption experiences 617a 3.7% increase, amounting to $188,114 over the three-year period within each ZIP 618Code after Amazon Fresh’s arrival. While the economic impact of Amazon Fresh’s 619entry on the total consumption is notable, it is statistically insignificant. Secondly, 620we confirm that the surge in total consumption primarily stems from heightened 621online consumption following Amazon Fresh’ entry. When we use the logarithm 622of online consumption as the dependent variable in column 2, we ascertain that 623local households enhance their online consumption by 29.1% (approximately 624$35,896) in comparison to households residing in ZIP Codes where Amazon Fresh 625service is not available. However, local households exhibit no significant altera626tions in their offline consumption, as demonstrated in column 3. Lastly, we adopt 627the online consumption share as the dependent variable in column 4 and consis628tently find that households’ proportion of online consumption elevates by 0.3% 629following the entry of Amazon Fresh. Relative to the mean value, it is about 13.3% 630(0.3%/2.25%) increases of the proportion of online consumption over the total 631consumption.632These results suggest that the operation of Amazon Fresh prompts an upswing in 633local households’ consumption via the online channel. This augmented online con634sumption appears to be an additional expenditure driven by the convenience and 635variety brought by Amazon fresh. These findings support our main findings that the 636entry of Amazon Fresh services increases home values, which also proposes a poten637Table 5 Mechanism: household consumption patterns638(1)639(2)640(3)641(4)642Dep. Variables:643ln(Total Con644sumption + 1)645ln(Online 646Consumption + 1)647ln(Offline 648Consumption + 1)649ln(Online 650Share + 1)651Amazon Fresh x Post6520.0376530.291***6540.0336550.003**656(0.023)657(0.099)658(0.023)659(0.001)660Observations661183,011662183,011663183,011664183,011665R-squared6660.6926670.3556680.6896690.293670Controls671yes672yes673yes674yes675ZIP Code Fixed Effect676yes677yes678yes679yes680Year x Month Fixed Effect681yes682yes683yes684yes685This table presents the influence of Amazon Fresh on consumption at the ZIP Code level. The dependent 686variables are the natural logarithm of the amount of consumption (plus 1), the natural logarithm of the 687amount of online and offline consumption (plus 1) as well as the online consumption share in column 6881–4, respectively. Standard errors are clustered at the ZIP Code level. Robust standard errors, clustered 689at the ZIP Code level, are reported in parentheses. ***, **, and * indicate statistical significance at the 6901%, 5%, and 10% levels, respectively.6911 3692693694===== PAGE 24 (score 31, numeric-paren 10, table-words 5, var-words 9) =====695696J. Chung, L. Yang697tial avenue whereby Amazon Fresh services function as a positive amenity within the 698housing market.699Local Market Changes700Another mechanism contributing to the increase in home values is the potential 701improvement in local markets following the entry of Amazon Fresh. The availability 702of this new service often drives significant investments in infrastructure and may bol703ster the local labor market. For instance, Chung (2023) demonstrates that the open704ing on Amazon fulfillment centers leads to increased retail property values, through 705improvements in the labor market within the transportation and warehousing sector. 706If Amazon Fresh exerts similar impacts on local markets, we would expect that its 707entry may lead to significant changes in local economic and demographic factors. 708To explore this possibility, we analyze how these factors evolve in response to the 709introduction of Amazon Fresh services.710As shown in Table 6, Amazon Fresh services appear to marginally improve local 711market conditions. Following its entry, the total population in treated areas increases 712by 0.3%, and the labor force grows by 0.4%. Additionally, the median age rises by 7130.3%. However, the economic impacts of these changes are relatively small, making 714it difficult to argue that Amazon Fresh significantly alters local market demographics. 715Other factors, such as median household income, household size, and unemployment 716rates, show no statistically significant changes. These findings provide evidence that 717online platforms like Amazon Fresh may not fundamentally drive local market trans718formations but instead serve as an added amenity for local households. Moreover, the 719lack of significant changes in economic factors suggests that the observed increase in 720home values is unlikely to be driven by broader local market shifts, such as gentrifi721cation or demographic changes.722Table 6 Local market changes723(1)724(2)725(3)726(4)727(5)728(6)729(7)730Dep. Variables:731Population Laborforce Income732Median 733Age734House735hold Size736Unemploy737ment Rate738Employ739ment Size740Amazon Fresh 741x Post7420.003**7430.004***744−0.0027450.003***746−0.0027470.002748−0.001749(0.001)750(0.001)751(0.001)752(0.001)753(0.001)754(0.004)755(0.0005)756Observations757118,517758118,517759118,517760118,517761118,517762118,517763118,517764R-squared7650.9987660.9987670.9907680.9727690.9847700.9047710.917772ZIP Code Fixed 773Effect774yes775yes776yes777yes778yes779yes780yes781Year x Month 782Fixed Effect783yes784yes785yes786yes787yes788yes789yes790This table presents the results when using the economic and demographic factors as dependent variables. 791The dependent variables are logged. Standard errors are clustered at the ZIP Code level. Robust standard 792errors, clustered at the ZIP Code level, are reported in parentheses. ***, **, and * indicate statistical 793significance at the 1%, 5%, and 10% levels, respectively.7941 3795796797===== PAGE 25 (score 38, numeric-paren 10, table-words 6, var-words 15) =====798799E-Grocery and Housing Markets: Evidence from the Amenity Effect of…800Heterogeneity801Property Characteristics802E-commerce is frequently favored by younger generations and considering that 803younger customers often occupy smaller units with fewer bedrooms, while fami804lies with children tend to reside in larger units with more bedrooms, we posit that 805the amenity effect of Amazon Fresh could differ based on the number of bedrooms. 806Using the ZHVI data by the number of bedrooms, we run the regressions on home 807values of each property type in column 1–5 of Table 7. We find that home values of 808smaller units are sold at higher prices when compared to similar size homes in areas 809without Amazon Fresh service following the entry of Amazon Fresh. Specifically, 810homes with 1 bedroom to 3 bedrooms are transacted at higher prices following the 811entry of Amazon Fresh. The magnitude is larger when a home has 1 bedroom at 2.3% 812and 2 bedrooms at 2.9%, and the effect decreases to 2.0% when a home has 3 bed813rooms. Homes with 4 bedrooms and 5 bedrooms also have positive amenity effect 814from Amazon Fresh entry, but with smaller impacts of 1.5% and 1.3%, respectively. 815Our findings suggest the positive pricing effects of the Amazon Fresh’s entry are 816more pronounced on homes where young people are likely to reside, i.e. units with 817smaller the number of bedrooms, which is consistent with the previous literature that 818e-grocery services are advantageous for the younger demographic (e.g., Couture, et 819al., 2021).820Socioeconomic Status821Urban areas with low socioeconomic status are often referred to as “Food Deserts” 822due to their lack of accessible grocery stores and nutritious food options (e.g., All823cott, et al., 2019). Furthermore, the distribution of retail outlets varies depending on 824the income of the neighborhood; for instance, areas with high poverty rates tend to 825exhibit lower business density, including fewer retail stores or chain supermarkets 826Table 7 Heterogeneity tests—property types827(1)828(2)829(3)830(4)831(5)832Dep. Variable: ln(ZHVI)8331-bedroom8342-bedroom8353-bedroom8364-bedroom8375-bedroom838Amazon Fresh x Post8390.023***8400.029***8410.020***8420.015***8430.013***844(0.004)845(0.002)846(0.002)847(0.002)848(0.002)849Observations85054,535851106,987852115,420853106,47385487,477855R-squared8560.9878570.9908580.9928590.9928600.993861Controls862yes863yes864yes865yes866yes867ZIP Code Fixed Effect868yes869yes870yes871yes872yes873Year x Month Fixed Effect874yes875yes876yes877yes878yes879This table illustrates the impacts of Amazon Fresh on home values based on property type. Column 8801 employs ZHVI for 1-bedroom units, column 2 for 2-bedroom units, column 3 for 3-bedroom units, 881column 4 for 4-bedroom units, and column 5 for 5-bedroom units. Both ZIP Code and year-month 882fixed effects are included. Standard errors are clustered at the ZIP Code level. Robust standard errors, 883clustered at the ZIP Code level, are reported in parentheses. ***, **, and * indicate statistical significance 884at the 1%, 5%, and 10% levels, respectively.8851 3886887888===== PAGE 26 (score 14, numeric-paren 0, table-words 2, var-words 10) =====889890J. Chung, L. Yang891(e.g., Alwitt & Donley, 1997; Schuetz, et al., 2012). Our study aims to investigate 892the potential mitigating effect of e-grocery services on these retail disparities in eco893nomically disadvantaged urban neighborhoods. Specifically, we explore whether the 894impacts of Amazon Fresh’s entry on property values differ across regions with vary895ing socioeconomic statuses. To address this question, we utilize household demo896graphic and economic data at the ZIP Code level from the American community 897Survey. We create dummy variables based on the median values of each demographic 898and economic factor and intersect these dummy variables with the interaction term 899of the treatment and post variables. Specifically, we explore whether the results vary 900according to local socioeconomic factors and if so, which attributes have significant 901effects. The findings from our heterogeneity analysis are presented in Table 8.902In column 1, where we use median age as the criterion, residential properties 903located in younger-population areas see a greater rise in home values. Focusing on 904the median household income in column 2 of Table 8, we find that areas with higher 905income experience a lower home value increase following the entry of Amazon Fresh. 906While higher income households are more inclined to shop online (e.g., Dolfen, et al., 9072023), we find the opposite impact that the entry of e-grocery services, which provide 908daily necessities, may be more beneficial to lower income households through ame909nity channel. Analyzing the proportion of renter-occupied households in column 3, 910we find that areas with high renter occupancy witness a larger boost in home values 911following the entry of Amazon Fresh. When examining the ratio of the highly edu912cated population (bachelor’s degree and above) in column 4, we find that homes in 913less educated areas encounter a greater increase in home values following the entry of 914Amazon Fresh. Considering household race in column 5, we discover that home val915ues increase more in areas with minority households (non-White). These results are 916in line with the previous literature that e-commerce offers greater benefits to younger 917population and minorities in rural households (e.g., Couture, et al., 2021). Further918more, since e-grocery services can be useful when local households are not able to 919utilize automobiles, we investigate whether local automobile ownership rates affect 920the results. Notably, we find that areas with lower automobile ownership exhibit a 921stronger and positive pricing effect following the entry of Amazon Fresh in column 6. 922These findings suggest that areas with underdeveloped socioeconomic factors benefit 923more from the availability of e-grocery services, experiencing a greater appreciation 924in home values.925Additionally, we investigate whether local interest in Amazon Fresh affects the 926results in column 7. Unlike physical stores, online platforms are not directly visible to 927buyers, though they provide significant convenience to local customers. In our study, 928we assume that local households are aware of e-grocery services or use them signifi929cantly enough to impact the local housing market. To connect local household inter930est in Amazon Fresh to housing market outcomes, we use Google Trends’ keyword 931search outcomes for ‘Amazon Fresh’ as a proxy for local interest in e-grocery ser932vices. We find that the pricing effect is more pronounced when local household inter933est in e-grocery services is higher. This result supports our argument that e-grocery 934services serve as an additional amenity to the local housing market and are priced 935in home values. The presence of such services may act as a signal of neighborhood 936desirability and investment potential.9371 3938939940===== PAGE 27 (score 62, numeric-paren 45, table-words 4, var-words 4) =====941942E-Grocery and Housing Markets: Evidence from the Amenity Effect of…943(1)944(2)945(3)946(4)947(5)948(6)949(7)950(8)951(9)952Dep. Variables: ln(ZHVI)953Age954Income955Renter956Education957Minority958No Vehicle959Google960Trends961Competing962Retail Est963All964Amazon Fresh x Post9650.060***9660.055***967−0.019***9680.046***969−0.006*9700.0049710.00339720.033***9730.009974(0.001)975(0.001)976(0.001)977(0.001)978(0.001)979(0.001)980(0.025)981(0.001)982(0.039)983x Age984−0.072***985−0.004***986(0.007)987(0.001)988x Income989−0.059***9900.00005**991(0.007)992(0.001)993x Renter Occupied9940.086***9950.122***996(0.006)997(0.031)998x Education999−0.041***1000−0.00031001(0.006)1002(0.001)1003x Minority10040.064***10050.110***1006(0.008)1007(0.018)1008x Non-Vehicle10090.043***10100.001**1011(0.007)1012(0.001)1013x Interest10140.0363***10150.021***1016(0.005)1017(0.009)1018x Competing Retail Est1019−0.016***10200.000061021(0.007)1022(0.001)1023Observations1024118,5171025118,5171026118,5171027118,5171028118,5171029118,5171030118,5171031118,5171032118,5171033R-squared10340.99110350.99010360.99110370.99010380.99010390.99010400.99010410.99010420.9911043Controls1044yes1045yes1046yes1047yes1048yes1049yes1050yes1051yes1052yes1053Table 8 Heterogeneity tests – socio-economic factors10541 3105510561057===== PAGE 28 (score 15, numeric-paren 0, table-words 3, var-words 9) =====10581059J. Chung, L. Yang1060(1)1061(2)1062(3)1063(4)1064(5)1065(6)1066(7)1067(8)1068(9)1069Dep. Variables: ln(ZHVI)1070Age1071Income1072Renter1073Education1074Minority1075No Vehicle1076Google1077Trends1078Competing1079Retail Est1080All1081ZIP Code Fixed Effect1082yes1083yes1084yes1085yes1086yes1087yes1088yes1089yes1090yes1091Year x Month Fixed Effect1092yes1093yes1094yes1095yes1096yes1097yes1098yes1099yes1100yes1101This table provides the results of the heterogeneity test analyzing the impact of the entry of e-grocery services on housing prices. We use the median value of each 1102indicator to create dummy variables, which are then intersected with the interaction term of Amazon Fresh (Treat) and Post. Demographic and economic factors at the 1103ZIP Code level are from the 2010 American Community Survey. Income is median household income, age is median household age, and education is the ratio of educated 1104individuals with a bachelor's degree and above. Minority is the ratio of non-White population. No Vehicle indicates the share of household without vehicle. Google Trends 1105indicates the search engine outcomes for the keyword “Amazon Fresh” at the market level. Competing retail establishments (Competing Retail Est.) indicates the share of 1106retail establishments that compete with e-grocer compared to the total retail establishment. Standard errors are clustered at the ZIP Code level. Robust standard errors, 1107clustered at the ZIP Code level, are reported in parentheses. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.1108Table 8 (continued)1109 11101 3111111121113===== PAGE 30 (score 34, numeric-paren 12, table-words 6, var-words 9) =====11141115J. Chung, L. Yang1116tically significant, with estimated effects of 16.6%, 14.9%, and 19.3%, respectively. 1117This pattern indicates that online consumption increases more in ZIP Codes experi1118encing larger housing-price effects and in areas with housing structures and tenure 1119patterns, specifically smaller housing units and higher owner-occupancy, that are 1120more likely to benefit from Amazon Fresh. Overall, these findings provide additional 1121support for the proposed consumption channel. As the Nielsen data do not identify 1122whether online purchases are specifically attributable to Amazon Fresh, the results 1123should be interpreted as suggestive evidence consistent with the proposed mecha1124nism, rather than as a direct measure of the treatment mechanism. 1125Conclusion1126The e-grocery industry has experienced rapid expansion in recent decades, yet its 1127economic impact remains underexplored. This study contributes to the literature by 1128exploring the impact of e-grocery services on the housing market. We use Amazon 1129Fresh, which provides both products and delivery services to local households as a 1130representative case of e-grocers as Amazon Fresh is a leading e-grocer during our 1131Table 9 Heterogeneity tests—online consumptions1132(1)1133(2)1134(3)1135Dep. Variable: ln(Online Consumption)1136Amazon Fresh x Post11370.221*11380.213**11390.252**1140(0.119)1141(0.102)1142(0.125)1143Amazon Fresh x Post x D(Strong Price Effects)11440.166**1145(0.071)1146Amazon Fresh x Post x D(Higher Share Units with 1–3 Bedrooms)11470.149***1148(0.049)1149Amazon Fresh x Post x D(Higher Share Owner-Occupied Units)11500.193**1151(0.080)1152Observations1153183,0111154183,0111155183,0111156R-squared11570.35611580.35411590.3551160Controls1161yes1162yes1163yes1164ZIP Code Fixed Effect1165yes1166yes1167yes1168Year x Month Fixed Effect1169yes1170yes1171yes1172This table examines heterogeneity in the effect of Amazon Fresh entry on online consumption using 1173household-level Nielsen data. The dependent variable is Online Consumption, an index of online 1174grocery spending at the ZIP Code–month level (exact construction described in the main text). Amazon 1175Fresh × Post equals one for ZIP Codes in metros where Amazon Fresh is available in month t and zero 1176otherwise. In column (1), D(Strong Price Effects) equals one for ZIP Codes whose estimated house-1177price response to Amazon Fresh entry is above the sample median and zero otherwise. In column (2), 1178D(Higher Share Units with 1–3 Bedrooms) equals one for ZIP Codes with an above-median share of 1–3 1179bedroom units. In column (3), D(Higher Share Owner-Occupied Units) equals one for ZIP Codes with 1180an above-median owner-occupancy rate. Interaction terms capture how online consumption responds 1181in ZIP Codes with stronger price effects and housing structures that are more likely to benefit from 1182e-grocery. All specifications include ZIP Code fixed effects, year-by-month fixed effects, and the full set 1183of baseline controls. Robust standard errors, clustered at the ZIP Code level, are reported in parentheses. 1184***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.11851 3118611871188===== PAGE 31 (score 10, numeric-paren 0, table-words 1, var-words 8) =====11891190E-Grocery and Housing Markets: Evidence from the Amenity Effect of…1191study’s sample period. Furthermore, we accurately define the exact servicing areas 1192of e-grocer by using detailed data on e-grocer servicing areas at the ZIP Code level.1193Our study reveals that e-grocery services are perceived as a valuable amenity by 1194local residents. Specifically, we discover that the entry of Amazon Fresh positively 1195influences the housing market, resulting in a 2.5% increase in local housing prices. 1196Our findings also remain robust when we redefine treated and control groups to focus 1197on neighboring ZIP Codes and consider alternative e-grocer impacts on local home 1198values. Additionally, our mechanism analysis shows that the presence of Amazon 1199Fresh results in increased local household consumption through online shopping 1200channels. This supports our argument that e-grocery services provide amenity effects 1201comparable to those of physical retail stores in the housing market.1202Our heterogeneity analysis indicates that the positive impact on home values is 1203more pronounced in areas with lower socioeconomic status, including those with 1204lower income, less education, higher renter occupancy, and lower automobile own1205ership. Moreover, residents in "Retail Desert" areas, regions with limited access to 1206physical stores, benefit particularly from e-grocery services compared to those in 1207areas with better retail access. Because e-grocery services do not require physical 1208storefronts and can cover broader geographies efficiently (for example, an Amazon 1209Fresh fulfillment center can serve a larger area than a single retail store), increased 1210accessibility to daily necessities can help address disparities in the housing market.1211This study provides new insights into the complex interactions among technologi1212cal innovation, consumer behavior, and housing markets. The expansion of e-grocery 1213services, exemplified by Amazon Fresh, has the potential to generate meaningful ben1214efits for local housing markets, particularly in areas facing socioeconomic challenges. 1215However, this study does not directly examine whether the observed increases in 1216home values translate into improved housing affordability for low-income or under1217served households. While greater access to online grocery services may enhance 1218convenience and welfare in underserved areas, rising home values could also pose 1219affordability challenges, especially for renters. Future research could address this 1220issue by incorporating household-level measures of housing affordability.1221Moreover, an important concern relates to the potential predictability of Ama1222zon Fresh entry through facility site selection, development approvals, or permitting 1223processes. Such activities could render entry partially anticipated and thus weaken 1224a strict “surprise” interpretation of entry announcements. Because we are unable 1225to obtain comprehensive permitting or zoning data, we rely instead on systematic 1226media coverage to identify entry timing. This approach is reasonable insofar as most 1227households are more likely to learn about Amazon Fresh’s entry through news reports 1228than by directly observing development or permitting activity. Future research could 1229extend our analysis by explicitly examining the developmental and pre-operational 1230stages of e-grocery entry.1231As digital services continue to reshape urban life, understanding these nuanced 1232dynamics remains essential for informing policy discussions and promoting equi1233table access to digital amenities. In particular, the rapid expansion of other e-grocery 1234providers beyond our sample period (post-2019) both across markets and over time 1235may moderate or reshape amenity capitalization effects. Future research can inves1236tigate the dynamic effects of e-grocers on local economies to fully assess how digi12371 3123812391240===== PAGE 38 (score 43, numeric-paren 16, table-words 7, var-words 7) =====12411242J. Chung, L. Yang1243(1)1244(2)1245(3)1246All Sample Matched 1247Sample1248Considering1249Alternative 1250E-grocery1251Dep. Variable: ln(Home 1252Value)1253Instacart x Post12540.003***1255-0.0001256-0.004***1257(0.00)1258(0.00)1259(0.00)1260Observations1261784,9581262458,0381263284,7521264R-squared12650.99112660.99112670.9911268Controls1269yes1270yes1271yes1272ZIP Code Fixed Effect1273yes1274yes1275yes1276Year x Month Fixed Effect1277yes1278yes1279yes1280This table presents the influence of Instacart on home values. We 1281use all the markets that received Instacart services between 2013 1282and 2016 in the first column while we use the matched markets to the 1283Nielsen data in the next column. Column 3 excludes areas serviced 1284by alternative e-grocery provider’s servicing areas from the sample. 1285ZIP Code fixed effect and year by month fixed effect are included. 1286Standard errors are clustered at the ZIP Code level. Robust standard 1287errors, clustered at the ZIP Code level, are reported in parentheses. 1288***, **, and * indicate statistical significance at the 1%, 5%, and 128910% levels, respectively.1290Table A12 The impact of insta1291cart on housing prices1292 1293Treated Group1294Control Group1295T-test1296Mean1297Std. 1298dev1299Mean1300Std. 1301dev1302Difference1303Home Value 1304($)1305367,181130631,1781307330,874130823,227130936,307 1310(0.92)1311Population131230,39213132,257131426,23013152,21413164,161 (1.31)1317Education 1318Rate13190.66413200.04213210.56813220.04013230.097 1324(1.66)*1325Renter Oc1326cupied Rate13270.23013280.01813290.20913300.01513310.021 (0.86)1332Minority 1333Rate13340.30413350.02813360.23413370.02313380.07 (1.89)*1339Median Age134039.68513410.564134240.87813430.5911344-1.19 (1.46)1345Median HH 1346Income ($)134787,94513484,211134984,13713503,84513513,807.98 1352(0.66)1353Non-Vehi1354cle (%)13556.87213561.40813574.86713580.56713592.004 (1.27)1360Household 1361Size13622.87613630.05713642.84313650.04813660.033 (0.44)1367Vacancy 1368Rate13690.05713700.00413710.07213720.0081373-0.015 1374(-1.84)*1375Competing 1376Retail Est. 1377Rate13780.26213790.05713800.36413810.0651382-0.10 1383(-1.17)1384Table A11 Pre-event summary 1385statistics1386This table shows the summary 1387statistics of pre-event period 1388and t-test difference in the 1389last column. ***, **, and * 1390indicate statistical significance 1391at the 1%, 5%, and 10% levels, 1392respectively.1393 13941 3139513961397===== PAGE 39 (score 27, numeric-paren 6, table-words 8, var-words 4) =====13981399E-Grocery and Housing Markets: Evidence from the Amenity Effect of…1400Funding No funding was received for conducting this study.1401Data Availability The datasets contain confidential information and cannot be shared publicly.1402Declarations1403Ethics Approval This manuscript is original, has not been published, and is not currently under consider1404ation for publication elsewhere.1405Consent to Participate Not applicable.1406Consent to Publish Not applicable.1407Competing Interests The authors have no financial or non-financial interests to disclose.1408Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, 1409which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long 1410as you give appropriate credit to the original author(s) and the source, provide a link to the Creative 1411Commons licence, and indicate if changes were made. The images or other third party material in this 1412article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line 1413to the material. If material is not included in the article’s Creative Commons licence and your intended use 1414is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission 1415directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licen1416ses/by/4.0/.1417(1)1418(2)1419(3)1420(4)1421Dep. Variables:1422ln(Total 1423Consump1424tion + 1)1425ln(Online 1426Consump1427tion + 1)1428ln(Offline 1429Consump1430tion + 1)1431ln(Online 1432Share + 1)1433Amazon Fresh 1434x Post1435-0.01014360.129**1437-0.01314380.002***1439(0.012)1440(0.053)1441(0.012)1442(0.001)1443Observations1444302,0651445302,0651446302,0651447302,0651448R-squared14490.69314500.34314510.69014520.2581453Controls1454yes1455yes1456yes1457yes1458ZIP Code Fixed 1459Effect1460yes1461yes1462yes1463yes1464Year x Month 1465Fixed Effect1466yes1467yes1468yes1469yes1470This table presents the influence of Instacart on consumption at the 1471ZIP Code level. The dependent variables are the natural logarithm 1472of the amount of consumption (plus 1), the natural logarithm of the 1473amount of online and offline consumption (plus 1) as well as the 1474online consumption share in column 1–4, respectively. ZIP Code 1475fixed effect and year by month fixed effect are included. Standard 1476errors are clustered at the ZIP Code level. Robust standard errors, 1477clustered at the ZIP Code level, are reported in parentheses. ***, **, 1478and * indicate statistical significance at the 1%, 5%, and 10% levels, 1479respectively.1480Table A13 Instacart: household 1481consumption patterns1482 14831 3148414851486===== PAGE 40 (score 18, numeric-paren 0, table-words 0, var-words 18) =====14871488J. Chung, L. Yang1489References1490Agarwal, S., Ding, Y., Kuang, W., & Zhu, X. (2023). Are environmental punishments good news or bad 1491news? Evidence from China. Journal of Environmental Economics and Management, 120, Article 1492102847.1493Allcott, H., Diamond, R., Dubé, J. P., Handbury, J., Rahkovsky, I., & Schnell, M. (2019). Food deserts and 1494the causes of nutritional inequality. Quarterly Journal of Economics, 134(4), 1793–1844.1495Alwitt, L. F., & Donley, T. D. (1997). Retail stores in poor urban neighborhoods. Journal of Consumer 1496Affairs, 31(1), 139–164.1497An, B. Y., & Chung, J. (2025). Who bears the brunt of disruptive innovation? The effect of grocery e‐com1498merce on local retail competitors. Journal of Regional Science, 65(3), 843–865.1499Baker, S. R., Johnson, S., & Kueng, L. (2021). Shopping for lower sales tax rates. American Economic 1500Journal, Macroeconomics, 13(3), 209–250.1501Barwick, P. J., Li, S., Lin, L., & Zou, E. Y. (2024). From fog to smog: The value of pollution information. 1502American Economic Review, 114(5), 1338–1381.1503Beracha, E., & Hardin, W. G., III. (2021). The housing price premium associated with charter schools. 1504Real Estate Economics, 49(4), 1267–1289.1505Bogin, A., Doerner, W., & Larson, W. (2019). Local house price dynamics: New indices and stylized facts. 1506Real Estate Economics, 47(2), 365–398.1507Caceres, B. C., & Geoghegan, J. (2017). Effects of new grocery store development on inner-city neighbor1508hood residential prices. Agricultural and Resource Economics Review, 46(1), 87–102.1509Capital One Shopping. (2024). Online Grocery Shopping Statistics: Trends, demographics, and market 1510share. Retrieved from https://capitaloneshopping.com/research/online-grocery-shopping-statistics/. 1511Accessed in June 2025.1512CBRE. (2022). Bigger and Bolder: Large Warehouse Leases Fueled US Industrial & Logistics Market in 15131H 2022. Retrieved from https://www.cbre.com/press-releases/bigger-and-bolder-large-warehouse-l1514eases-fueled-us-industrial-logistics-market-in-1h-2022. Accessed in June 2025.1515Chava, S., Oettl, A., Singh, M., & Zeng, L. (2024). Creative destruction? Impact of e-commerce on the 1516retail sector. Management Science, 70(4), 2168–2187.1517Chung, J. (2023). The spillover effect of e-commerce on local retail real estate markets. Regional Science 1518and Urban Economics. https://doi.org/10.1016/j.regsciurbeco.2023.1039191519Conroy, S. J., & Milosch, J. L. (2011). An estimation of the coastal premium for residential housing prices 1520in San Diego County. Journal of Real Estate Finance and Economics, 42, 211–228.1521Couture, V., Faber, B., Gu, Y., & Liu, L. (2021). Connecting the countryside via e-commerce: Evidence 1522from China. American Economic Review: Insights, 3(1), 35–50.1523Damianov, D. S., & Escobari, D. (2016). Long-run equilibrium shift and short-run dynamics of US home 1524price tiers during the housing bubble. Journal of Real Estate Finance and Economics, 53, 1–28.1525Danielsen, B. R., Harrison, D. M., & Zhao, J. (2014). It makes a village: Residential relocation after char1526ter school admission. Real Estate Economics, 42(4), 1008–1041.1527Daunfeldt, S. O., Mihaescu, O., Öner, Ö., & Rudholm, N. (2021). Retail and place attractiveness: The 1528effects of big‐box entry on property values. Geographical Analysis, 53(3), 467–498.1529De Chaisemartin, C., & d’Haultfoeuille, X. (2023). Two-way fixed effects and differences-in-differences 1530with heterogeneous treatment effects: A survey. Econometrics Journal, 26(3), C1–C30.1531Diao, M., Qin, Y., & Sing, T. F. (2016). Negative externalities of rail noise and housing values: Evidence 1532from the cessation of railway operations in Singapore. Real Estate Economics, 44(4), 878–917.1533Ding, X., Bollinger, C., Clark, M., & Hoyt, W. (2024). Too late to buy a home? School redistricting and the 1534timing and extent of capitalization. Journal of Regional Science, 64(1), 207–237.1535Dolfen, P., Einav, L., Klenow, P. J., Klopack, B., Levin, J. D., Levin, L., & Best, W. (2023). Assessing the 1536gains from e-commerce. American Economic Journal: Macroeconomics, 15(1), 342–370.1537Dube, A., Lester, T. W., & Reich, M. (2010). Minimum wage effects across state borders: Estimates using 1538contiguous counties. Review of Economics and Statistics, 92(4), 945–964.1539Goldmanis, M., Hortaçsu, A., Syverson, C., & Emre, Ö. (2010). E-commerce and the market structure of 1540retail industries. Economic Journal, 120(545), 651–682.1541Goolsbee, A. D., & Klenow, P. J. (2018). Internet rising, prices falling: Measuring inflation in a world of 1542e-commerce. AEA Papers and Proceedings, 108, 488–492.1543Gupta, A., Van Nieuwerburgh, S., & Kontokosta, C. (2022). Take the Q train: Value capture of public 1544infrastructure projects. Journal of Urban Economics, 129, Article 103422.15451 31546