1# RECORD 90decb41b454a38dee63486b408022b766c2dc062# TITLE: Effects of New Grocery Store Development on Inner-City Neighborhood Residential Prices3# YEAR: 2017.0 VENUE: Agricultural and Resource Economics Review DOI: 10.1017/age.2016.294# AUTHORS: Belkis Cerrato Caceres; J. Geoghegan5# PAGES: 16 CANDIDATE TABLE PAGES (1-based): [9, 10, 11, 12]6# ABSTRACT: A difference-in-differences approach is used to measure the impact of new inner-city grocery store developments on residential housing values in Worcester, Massachusetts. Using geocoded housing sales from 1988–2011, we develop a hedonic model, exploiting temporal and spatial discontinuities, to identify the effect of 12 new grocery stores on neighborhood housing prices. Results suggest these new stores were associated with an increase in sale prices of nearby homes, and these results could help inform current policies related to urban food deserts, in that new grocery stores have the potential to improve neighborhood wealth as well as health.789===== PAGE 1 (score 8, numeric-paren 0, table-words 0, var-words 8) =====1011Effects of New Grocery Store12Development on Inner-City13Neighborhood Residential Prices14Belkis Cerrato Caceres and Jacqueline Geoghegan15A difference-in-differences approach is used to measure the impact of new inner-16city grocery store developments on residential housing values in Worcester,17Massachusetts. Using geocoded housing sales from 1988–2011, we develop a18hedonic model, exploiting temporal and spatial discontinuities, to identify the19effect of 12 new grocery stores on neighborhood housing prices. Results suggest20these new stores were associated with an increase in sale prices of nearby21homes, and these results could help inform current policies related to urban food22deserts, in that new grocery stores have the potential to improve neighborhood23wealth as well as health.24Key words: difference-in-differences, grocery stores, hedonic models25The burgeoning obesity epidemic in the United States (Nguyen and El-Serag262010, Flegal et al. 2012), with large disparities among different population27groups along ethnicity and socioeconomic lines (Wang and Beydoun 2007),28has highlighted the need for access to affordable healthy food throughout the29United States. The presence of strong residential segregation in housing30markets among these socioeconomic groups may affect the local food31environment, which potentially can be a factor in leading to these negative32health outcomes for disadvantaged groups. Geographic areas lacking in33affordable healthy food have been termed food deserts, and a growing34research literature from many academic disciplines has demonstrated that35these food deserts are more likely to occur in poorer neighborhoods (Powell36et al. 2007, Chen et al. 2010, Walker, Keane, and Burke 2010). These37geographic regions are more likely to have more convenience stores and fast38food outlets than grocery stores, resulting in less healthy food consumption39and worse health outcomes for their residents (Brown et al. 2008, Galvez40Belkis Cerrato Caceres is an Assistant Professor of Economics at Edgewood College. Jacqueline41Geoghegan is a Professor of Economics at Clark University. Correspondence: Belkis Cerrato42Caceres ▪Department of Social Science, Edgewood College ▪1000 Edgewood College Dr ▪Madison,43WI 53711 ▪Phone 508.793.7709 ▪Email bcerrato@edgewood.edu44The authors would like to thank John Brown, Wayne Gray and Wang Jin for advice and45suggestions, as well as two anonymous reviewers for very helpful comments.46In addition,47research support for this project from the Mosakowski Institute for Public Enterprise at Clark48University is gratefully acknowledged.49Agricultural and Resource Economics Review 46/1 (April 2017) 87–10250© The Author(s) 2017. This is an Open Access article, distributed under the terms of the Creative51Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits52unrestricted re-use, distribution, and reproduction in any medium, provided the original work is53properly cited.54https://doi.org/10.1017/age.2016.29 Published online by Cambridge University Press555657===== PAGE 9 (score 58, numeric-paren 31, table-words 4, var-words 13) =====5859post-opening grocery store indicator are shown in Table 2. Column 1 includes60all the properties in the sample, column 2 excludes properties that had a sales61price in the top 1 percent of sales price observations, and column 3 excludes62properties that had a price in the top 5 percent of sales price observations.63Standard errors were clustered at the blockgroup level. The estimates of the64coefficients of the spatial variables suggest that properties located within65each of the two spatial buffers were associated with lower prices compared66to properties located more than 800 meters away. Each of the indicator67variables D0399:968ij69and D40080070ij71has a negative and statistically significant72coefficient indicating that properties located within 800 meters from a73grocery store site sold at a price between 7 and 8 percent below the value of74comparable properties located more than 800 meters away from the store.75Table 1. Summary Statistics of Housing Data76All Residential77Properties780–399.979Meters80400–80081Meters82800 Meters83or More84Price 1988–201185(Average)286$175,759.387$173,661.788$171,136.989$177,67190(76,329.54)91(77234.4)92(75530.63)93(76461.58)94lot size (sq ft)951542.93961166.54971373.8981641.1899(5893.29)100(3127.04)101(4267.75)102(6570.30)103Bathrooms1041.861052.051061.921071.82108(0.87)109(0.94)110(0.95)111(0.83)112Bedrooms1134.051144.841154.581163.78117(2.25)118(2.51)119(2.62)120(2.01)121Age12265.4912382.0112475.8312560.13126(37.05)127(34.31)128(36.84)129(36.11)130Annual median131household income132(2010 inflation-133adjusted $)134$58,196.42135$44,492.96136$49,115.13137$62,020.17138(25,373.01)139(15,694.31)140(18,382.09)141(26,702.95)142% of households that143own 1 or more cars14487.7414587.7514686.7714788.03148(12.28)149(9.61)150(10.9)151(12.83)152% White15370.7115456.4715563.9715673.87157(21.31)158(24.54)159(23.17)160(19.43)161Population density162per square mile1636186.431647650.121657512.051665671.35167(4825.95)168(5236.47)169(5324.29)170(4525.45)171# Property Sales1723394917322111748576175231621762 This average excluded properties in the sample that had a sales price in the top 1 percent.177Cerrato Caceres and Geoghegan178Effects of New Grocery Store Development 95179https://doi.org/10.1017/age.2016.29 Published online by Cambridge University Press180181182===== PAGE 10 (score 45, numeric-paren 24, table-words 5, var-words 8) =====183184Overall, this suggests that grocery stores are located in lower priced185neighborhoods. The first row of Table 1 shows the average sales price of all186the properties sold between 1988 and 2011. Average sales price for187properties within the 800-meter buffers were below the city’s average sales188price and below the prices of properties located more than 800 meters away189from a grocery store. This could reflect the differences in housing structure190and neighborhood characteristics. Houses within the 800-meter buffers are191on average older and have a smaller lot size, and these neighborhoods are192more racially diverse and densely populated.193In terms of the impact of a new grocery store development on real estate194prices, the estimated coefficient of the interaction term Dd195ijPost iy identifies196the effect on sales price of a house located within a specific distance band197from the grocery store. The estimated coefficients of D0399:9198ij199Post iy and200D400800201ij202Post iy have positive magnitudes. In addition, the coefficient is larger203for properties located closest to the grocery store. For example, as shown in204columns 1–3 of Table 2, properties located within 399.9 meters from a205Table 2. Impact of Grocery Store Openings in Housing Prices206Sample207All208Observations209Excludes Properties210with Top 1% Prices211Excludes Properties212with Top 5% Prices213(1)214(2)215(3)216Variables217logprice218logprice219logprice220Grocery store distance2210–399.9 meters2220.076***2230.073***2240.072***225(0.028)226(0.027)227(0.027)228Grocery store distance229400–800 meters2300.081***2310.080***2320.076***233(0.018)234(0.018)235(0.017)236Grocery store distance2370–399.9 m*Post238Grocery store2390.073***2400.070***2410.071***242(0.024)243(0.024)244(0.024)245Grocery store distance246400–800 m*Post247Grocery store2480.039**2490.039**2500.037**251(0.019)252(0.019)253(0.018)254Housing Characteristics255Yes256Yes257Yes258Neighborhood259Characteristics260Yes261Yes262Yes263Clustered SE264Yes265Yes266Yes267N26833,84926933,58327032,619271R-squared2720.5892730.5992740.584275Robust standard errors in parentheses276*** p < 0.01, ** p < 0.05, þ p < 0.1277Note: All regressions control for year fixed effects278Agricultural and Resource Economics Review27996 April 2017280https://doi.org/10.1017/age.2016.29 Published online by Cambridge University Press281282283===== PAGE 11 (score 53, numeric-paren 39, table-words 3, var-words 3) =====284285grocery store experienced a 7 percent increase in price after the store opened,286while properties located within 400–800 meters from the grocery store287experienced approximately a 4 percent increase in value. As a comparison,288Pope and Pope (2015) found that a new Walmart store increases housing289prices between 2 and 3 percent for houses located within 0.5 miles of the290Table 3. Impact of Grocery Store Openings in Housing Prices for More291Disaggregated Buffers292Sample293All294Observations295Excludes Properties296with Top 1% Prices297Excludes Properties298with Top 5% Prices299(1)300(2)301(3)302Variables303logprice304logprice305logprice306Grocery store distance3070–199.9 meters3080.103663090.100623100.09851311(0.06349)312(0.06409)313(0.06452)314Grocery store distance315200–399.9 meters3160.07078**3170.06747**3180.06674**319(0.02840)320(0.02844)321(0.02763)322Grocery store distance323400–599.9 meters3240.0749***3250.0705***3260.0713***327(0.02149)328(0.02099)329(0.02115)330Grocery store distance331600–800 meters3320.0845***3330.0845***3340.0788***335(0.02107)336(0.02056)337(0.02026)338Grocery store distance3390–199.9 m*Post340Grocery store3410.09786þ3420.096853430.09464344(0.05840)345(0.05965)346(0.05956)347Grocery store distance348200–399.9 m*Post349Grocery store3500.06805**3510.06516**3520.06703**353(0.02634)354(0.02617)355(0.02598)356Grocery store distance357400–599.9 m*Post358Grocery store3590.016953600.015613610.01632362(0.02448)363(0.02377)364(0.02409)365Grocery store distance366600–800 m*Post367Grocery store3680.05274**3690.05241**3700.04979**371(0.02271)372(0.02231)373(0.02200)374Housing Characteristics375Yes376Yes377Yes378Neighborhood379Characteristics380Yes381Yes382Yes383Clustered SE384Yes385Yes386Yes387N38833,84938933,58339032,619391R-squared3920.589253930.599443940.58413395Robust standard errors in parentheses396*** p < 0.01, ** p < 0.05, þ p < 0.1397Note: All regressions control for year fixed effects398Cerrato Caceres and Geoghegan399Effects of New Grocery Store Development 97400https://doi.org/10.1017/age.2016.29 Published online by Cambridge University Press401402403===== PAGE 12 (score 40, numeric-paren 18, table-words 3, var-words 13) =====404405store, while similar positive results for a new Walmart within a two-mile radius406of a residential prices can be found in Johnson et al. (2015).407The estimated coefficients on the control variables have the expected signs408and are statistically significant: homes with more bedrooms and bathrooms409sell at a higher price, as do properties with larger lots. Homes in410neighborhoods with lower income, lower education levels and higher411percentages of minorities sell at lower prices.412In order to test the robustness of these results, we further disaggregated the413800-meter distance into four mutually exclusive rings (0–199.9 meters, 200–414399.9 meters, 400–599.9 meters, and 600–800 meters) centered around each415of the grocery stores. One of our hypotheses was that a new grocery store416would have a nonmonotonic effect on neighboring housing prices, where417properties in close proximity to a grocery store experienced a negative impact,418and properties located further away experienced a positive distance-decaying419effect. However, as shown in Table 3, this relation was not observed across the420coefficients of the interaction terms. In fact, the coefficients of the interaction421terms D0199:9422ij423Post iy and D400599:9424ij425Post iy are not statistically significant.426Table 4. Falsification Tests of Grocery Store Opening Year and Impact on427Housing Prices428Number of Years Grocery Store Opening429Date is Shifted4303 years431earlier4324 years433earlier4345 years435earlier436(1)437(2)438(3)439Variables440logprice441logprice442Logprice443Grocery store distance 0–399.9 meters4440.124***4450.121**4460.111**447(0.030)448(0.034)449(0.035)450Grocery store distance 400–800 meters4510.097***4520.102***4530.103***454(0.019)455(0.020)456(0.021)457Grocery store distance 0–399.9 m*Post458Grocery store4590.0444600.0364610.020462(0.031)463(0.037)464(0.040)465Grocery store distance 400–800 m*Post466Grocery store4670.0184680.0244690.024470(0.016)471(0.017)472(0.018)473Housing Characteristics474Yes475Yes476Yes477Neighborhood Characteristics478Yes479Yes480Yes481Clustered SE482Yes483Yes484Yes485R-squared4860.5924870.5924880.592489N49026,72449126,72449226,724493Robust standard errors in parentheses494*** p < 0.01, ** p < 0.05, þ p < 0.1495Note: All regressions control for year fixed effects496Agricultural and Resource Economics Review49798 April 2017498https://doi.org/10.1017/age.2016.29 Published online by Cambridge University Press499500501===== PAGE 13 (score 9, numeric-paren 0, table-words 1, var-words 7) =====502503Finally, falsification tests were conducted to identify whether the observed504increase in housing prices was due to a grocery store opening a short505distance from the property. One of the concerns is that there may be506confounding factors affecting the price of houses close to the grocery store’s507site before its announcement and opening date. Hence, similar to Pope and508Pope (2015), we constructed false opening dates for the grocery stores and509shifted the opening years of grocery stores three, four, and five years prior to510the actual opening years3. If the grocery store had an impact on property511values, we should observe a positive effect when the grocery store is512announced and opened, and not in prior years. The results of these tests are513shown in Table 4. None of the coefficients of the interaction terms provide514evidence of property values near a grocery store’s site increasing in value515three, four, and five years before the store’s actual opening year. Thus, these516results suggest that the opening of a new grocery store does have a positive517effect on nearby residential property values.518Conclusions519In this study, we investigated the effect of the opening of a new grocery store on520housing prices in Worcester, Massachusetts for the period 1988–2011. Using521geocoded housing sales data of single-family homes, different mutually522exclusive neighborhoods were created around each new store opening. A523hedonic model with a difference-in-differences econometric specification was524developed to test the impact of the opening of each new grocery stores on its525neighboring property values. Our results indicate that properties located526between527400–800 meters528and 0–399.9 meters529from530a531grocery store532experienced between 4 and 7 percent increase, respectively, in the level of533prices relative to similar properties located across the city and more than534800 meters away from a grocery store.535Our results also suggest that grocery stores in Worcester were in lower536property-valued neighborhoods. In particular, our results suggest that prices537of properties located within 800 meters from a grocery store were between 7538and 8 percent below the value of properties located more than 800 meters539away. Although our analysis does not incorporate an empirical analysis540specific to food desert areas due to data limitations, many of the grocery541stores were established in the city’s low-income census blockgroups. In fact,542most grocery stores were in neighborhoods where household income was543below the city’s average median income and where minorities made up a5443 Pope and Pope (2015) found that the average Walmart in their sample took about 1.5 years to545be built after its construction and opening were announced. As our dataset includes only the year546in which the grocery store opened, we shifted the grocery store opening by three, four, and five547years. This allowed us to reduce possible changes in housing prices due to announcement548rather than the opening of the grocery store.549Cerrato Caceres and Geoghegan550Effects of New Grocery Store Development 99551https://doi.org/10.1017/age.2016.29 Published online by Cambridge University Press552