1# RECORD bcc88aa6a74894b8030f7071e7fd929a9fa086682# TITLE: Urban versus rural disparities in amenity proximity and housing price: the case of integrated urban–rural city, Sejong, South Korea3# YEAR: 2024.0 VENUE: Journal of Housing and the Built Environment DOI: 10.1007/s10901-023-10098-y4# AUTHORS: Jiyeon Shin; Galen D. Newman; Y. Park5# PAGES: 21 CANDIDATE TABLE PAGES (1-based): [8, 12, 13, 14]6# ABSTRACT: Due to the growing importance of accessibility to urban amenities on quality of life and neighborhood satisfaction in South Korea, various efforts have been made to improve accessibility to urban amenities by establishing a minimum standard, an approach called Living Social Overhead Capital (Living SOC); for example, one such standards states that an elementary school should be reachable within a 15-min walk. However, these endeavors have not yet resulted in an equal distribution of Living SOC in urban and rural areas. Many residents feel the disparity of Living SOC is greatest in areas where the rural city was incorporated into an adjacent urban city—often called an integrated urban–rural city—or when new towns were developed in rural cities. This situation can lead to the creation of dissimilar landscapes and infrastructure within a single city. As such, this research observes how the resultant disparity in proximity between urban and rural areas impacts neighborhood satisfaction, as reflected in housing prices. The outcomes indicate that, with the exception of senior care and education, all observed amenities were found to be located closer to residences in urban areas. Additionally, the closer the Living SOC, the higher the housing prices. The findings of this study posit an urgent need for a comprehensive Living SOC plan for entire cities, not only for newly developed areas, to avoid widening the existing gap between urban and rural Living SOC.789===== PAGE 1 (score 9, numeric-paren 0, table-words 0, var-words 9) =====1011Vol.:(0123456789)12Journal of Housing and the Built Environment (2024) 39:727–74713https://doi.org/10.1007/s10901-023-10098-y141 315ARTICLE16Urban versus rural disparities in amenity proximity 17and housing price: the case of integrated urban–rural city, 18Sejong, South Korea19Jiyeon Shin1 · Galen D. Newman2 · Yunmi Park3 20Received: 27 May 2022 / Accepted: 11 December 2023 / Published online: 6 January 2024 21© The Author(s) 202422Abstract23Due to the growing importance of accessibility to urban amenities on quality of life and 24neighborhood satisfaction in South Korea, various efforts have been made to improve 25accessibility to urban amenities by establishing a minimum standard, an approach called 26Living Social Overhead Capital (Living SOC); for example, one such standards states that 27an elementary school should be reachable within a 15-min walk. However, these endeav-28ors have not yet resulted in an equal distribution of Living SOC in urban and rural areas. 29Many residents feel the disparity of Living SOC is greatest in areas where the rural city 30was incorporated into an adjacent urban city—often called an integrated urban–rural 31city—or when new towns were developed in rural cities. This situation can lead to the crea-32tion of dissimilar landscapes and infrastructure within a single city. As such, this research 33observes how the resultant disparity in proximity between urban and rural areas impacts 34neighborhood satisfaction, as reflected in housing prices. The outcomes indicate that, with 35the exception of senior care and education, all observed amenities were found to be located 36closer to residences in urban areas. Additionally, the closer the Living SOC, the higher the 37housing prices. The findings of this study posit an urgent need for a comprehensive Living 38SOC plan for entire cities, not only for newly developed areas, to avoid widening the exist-39ing gap between urban and rural Living SOC.40Keywords Urban amenities · Living social overhead capitals · Proximity · Urban–rural 41integrated city · Housing price · Urban–rural disparities42 * Yunmi Park 43 44ym.park@snu.ac.kr45 46Jiyeon Shin 47 48jys.urban@ewhain.net49 50Galen D. Newman 51 52gnewman@arch.tamu.edu531 54Architectural and Urban Systems Engineering, College of Engineering, Ewha Womans University, 55Seoul 03760, South Korea562 57Department Landscape Architecture and Urban Planning, Texas A&M University, College Station, 58TX 77843, USA593 60Department of Civil and Environmental Engineering and Institute of Construction 61and Environmental Engineering, Seoul National University, Seoul 08826, South Korea626364===== PAGE 8 (score 15, numeric-paren 0, table-words 3, var-words 6) =====656673467 68J. Shin et al.691 370• Senior (Factor 3): Senior facilities (senior care facilities, senior education facilities)71Traditionally, housing prices have been analyzed using hedonic price models, but sev-72eral recent studies have also considered the spatial dependence of housing prices. Due 73to the spatial dependency of each neighborhood (Moran’s I score = 0.614, p = 0.001), we 74assumed that there would be a spatial dependency. A Lagrange Multiplier (LM) test was 75conducted to determine the proper model between two spatial regression models. The 76results indicated the spatial lag model (SLM) was favorable. In this model, the influence 77of the surrounding area on the dependent variable is added to the regression model as a 78new explanatory variable. The result of ordinary least square regression (OLS) was also 79reported for simple comparison. The basic formula of SLM is:80where W is the spatial weight matrix, and ρ is the spatial autocorrelation coefficient.814 Results824.1 Descriptive statistics83The average housing price was about 4.3 million Korean won per square meter (≈ $3,400/84m2). The gap between the highest and lowest prices was about 7 million Korean won/m2 85won, with the most expensive houses in the urban area. The mean distance from apartment 86complexes to each facility was under 1 km, except for the senior education facility where 87the mean distance was larger than for other facilities, spanning up to 23 km. The mini-88mum distances to Daycare, Senior care, and Clinics appeared to be zero, suggesting that 89these facilities are located within the apartment complexes themselves. As mentioned, the 90Housing Construction Standard Regulation mandates that large apartment complexes must 91have daycare and senior care facilities within their boundaries. Additionally, clinics and 92other small retail shops are often located within large apartment complexes in Korea. The 93(1)94y = 휌Wy + X훽+ 휖95Table 2 Factor Analysis Results96Variables97Factor198Learning and Care99Factor2100Education 101and Leisure102Factor3103Senior104Pharmacy1050.89411060.37421070.1175108Clinic1090.88151100.43461110.0542112Daycare Center1130.81051140.44651150.0796116Public Library1170.78021180.34961190.3973120Kindergarten1210.42741220.86961230.3973124Elementary School1250.43721260.86611270.1282128Urban Park1290.71731300.72211310.1161132Sports Center1330.50311340.65291350.4624136Senior Care1370.13251380.24511390.6137140Senior Education1410.17611420.09931430.4953144145146===== PAGE 9 (score 22, numeric-paren 0, table-words 7, var-words 7) =====147148735149Urban versus rural disparities in amenity proximity and housing…1501 3151average age of each apartment was relatively low due to new construction in a new town, 152and, on average, the bus stop can be reached within 1 km (Table 3).1534.2 Welch’s t‑test154Since the urban (n = 129) and rural (n = 39) had unequal sample sizes and variance (log-155transformed housing price variance: 휎2156Urban = 0.057, 휎2157rural = 0.112), a Welch’s t-test was 158performed to assess the difference in proximity to Living SOC (Table 4.). The analysis 159revealed a statistically significant difference in network distance to Living SOC between 160urban and rural areas, with the exception of sports centers. In rural locations, the network 161distance to most Living SOC was found to be 0.67 km longer on average than in urban 162areas. Conversely, the distance to senior care (difference = − 0.14 km) and senior education 163(difference = − 4.65 km) facilities in rural areas was closer than in urban areas. For senior 164education facilities, the mean network distance in rural areas was approximately half that 165of urban areas. Besides daycare centers and senior care centers, which have regulations that 166are essential to be installed in large complexes of more than 200 households, parks were 167the closest facilities to apartments in urban areas. In rural areas, due to the few apartment 168complexes with over 200 households, the distance to daycare centers was a lot farther than 169in urban areas, and the closest facility was followed by senior care centers and sports cent-170ers. The facility that had the shortest distance gap between both regions was the senior care 171center. In addition to the Living SOC variables, housing prices and other variables, except 172for bus stop distance, exhibited a statistically significant difference between the two areas. 173Housing prices were significantly higher, while the average age of houses was lower in 174urban areas.1754.3 Results of the spatial lag model176As shown in Table 5, the spatial autoregressive parameter ( 휌 = 0.248, p < 0.01), which 177reflects the spatial dependency of housing price, was highly significant. Compared with the 178OLS model, the explanatory power of SLM was higher (R2179SLM = 0.899, R2180OLS = 0.895) and 181the AIC score, in which a lower score means a better model, was lower (AICSLM = − 72.222, 182AICOLS = − 67.644). Therefore, the results of SLM are interpreted, and OLS is presented 183for simple reference (Table 5). Learning & Care facilities (b = − 0.039) and Education 184& Leisure facilities (b = − 0.057) were negatively associated. Meanwhile, Senior facili-185ties (b = 0.057) were positively associated. This suggests that the closer the houses are to 186Learning & Care facilities and Education & Leisure facilities, the higher the value, while 187the opposite is true for elderly facilities. Building age (b = − 0.157) and the dummy vari-188able of total units (b = − 0.094) were negatively associated with housing prices, while other 189control variables were positively associated. This refers to that newly built and larger apart-190ment complexes and the higher apartments tend to have higher prices. Also, the closer to 191bus stops and Geum River the higher the price. As regional characteristics, apartments in 192urban areas tend to have higher prices than rural areas. The sign of each variable remained 193the same in both the SLM and OLS models, but the CBD was the only significant variable 194in the OLS model.195Table 6 presents the results of the SLM model separated by urban and rural areas. Since 196the sample size of the rural area was small (39 apartment complexes with 832 sales cases), 197the interpretation of the rural model should be done cautiously. The spatial autoregressive 198199200===== PAGE 12 (score 31, numeric-paren 15, table-words 1, var-words 4) =====201202738203 204J. Shin et al.2051 3206Table 4 Welch’s t-test of each variable207Urban (n = 129)208Rural (n = 39)209Variables210Mean (km)211SD212Mean (km)213SD214Urban–Rural (km)215t216Housing Price217507.911218122.506219157.95122066.683221349.96222− 23.058 ***223Public Library2240.7482250.6542261.2442271.782228− 0.4962291.703 *230Daycare Center2310.1902320.1892331.0962341.880235− 0.9062363.007 ***237Kindergarten2380.5042390.3402401.3422411.512242− 0.8382433.433 ***244Elementary School2450.5382460.3522471.3902481.534249− 0.8522503.442 ***251Senior Care2520.2692530.3192540.1322550.2642560.137257− 2.691 ***258Senior Education2599.3652601.8762614.7172626.4272634.648264− 4.459 ***265Urban Park2660.3822670.2062680.8852691.391270− 0.5032712.525 **272Clinic2730.4992740.3942751.3092761.882277− 0.8102782.669 **279Pharmacy2800.5512810.4072821.2112831.757284− 0.6602852.324 **286Sports Center2870.5722880.4992890.8692901.617291− 0.2972921.130293Age2944.3332951.95429621.2052977.001298− 16.87229914.876 ***300Max Floor30121.4733028.1613035.1793045.01530516.294306− 8.366 ***307Bus stop3080.2053090.1423100.2223110.111312− 0.0173130.769314CBD3153.0293161.41931714.5443186.446319− 11.51532011.075 ***321Freq322Percent323Freq324Percent325t326Total units 327(dummy)328032910733082.9503312433261.5403332.500 **33413352233617.0503371533838.460339Geum River 340(dummy)341034211534389.1503443834597.440346− 2.205 *34713481434910.85035013512.560352353354===== PAGE 13 (score 53, numeric-paren 25, table-words 10, var-words 2) =====355356739357Urban versus rural disparities in amenity proximity and housing…3581 3359parameter ( 휌Urban = 0.877, 휌Rural = 0.436) was still highly statistically significant in both 360areas. The explanatory power of the SLM in urban areas was about 57%, and around 80% 361in rural areas. Comparing the R-squared value (R2362SLM.Urban = 0.567, R2363SLM.Rural = 0.802) 364and AIC score (AICSLM.Urban = − 88.899, AICSLM.Rural = − 14.990), the SLM showed better 365explanatory power in both areas. The signs of Living SOC variables remained the same, 366but Senior facilities were not statistically significant in rural areas. In urban areas, Learn-367ing & Care facilities (bUrban = − 0.085) and Education & Leisure facilities (bUrban = − 0.087) 368had negative effects on housing prices, while Senior facilities (bUrban = 0.103) had a posi-369tive effect. In rural areas, Senior facilities did not have a statistically significant effect 370on housing prices, while the other two Living SOCs had a significant negative effect 371(bRural.Learning&Care = − 0.034, bRural.Education&Leisure = − 0.050). This implies that the apart-372ments located closer to both Learning & Care and Education & Leisure facilities had 373higher prices in both urban and rural areas, while, in the case of Senior facilities, it showed 374higher prices when the distance got farther only in urban areas.3755 Discussion376This study observed the disparities in Living SOC in urban and rural areas and their 377effects on neighborhood satisfaction that were reflected in housing prices in one sample of 378urban–rural integrated city, Sejong, Korea. Although this is exploratory work done with a 379single sample, the outcomes of this study illustrate a statistically significant difference in 380Table 5 SLM & OLS results381***p < .01, ** < p < .05,*p < .1382SLM383OLS384Coef. (b)385z-value386Coef387t-value388Learning & Care389− 0.039**390− 2.504391− 0.040**392− 2.397393Education & Leisure394− 0.057***395− 3.304396− 0.053***397− 2.950398Senior3990.057**4002.3784010.062**4022.479403ln(Age)404− 0.157***405− 4.774406− 0.163***407− 4.687408Total Units (dummy)409− 0.094**410− 2.319411− 0.099**412− 2.320413ln(Max Floor)4140.117***4153.2344160.120***4173.139418Bus stop4190.323***4202.9884210.333***4222.918423CBD424− 0.007425− 1.348426− 0.012**427− 2.476428Urban (dummy)4290.383***4303.4964310.587***4327.054433Geum River (dummy)4340.090*4351.7304360.092*4371.674438Rho ( 휌)4390.248**4402.489441− 442CONSTANT4434.075***4447.1054455.421***44630.234447R-squared4480.8994490.895450Log-Likelihood/Adj.R-squared45148.111/–45244.822/0.888453Akaike info criterion (AIC)454− 72.222455− 67.644456Jarque–Bera test45771.539***45846.944***459Breusch-Pagan test460–46165.928***462463464===== PAGE 14 (score 52, numeric-paren 25, table-words 8, var-words 1) =====465466740467 468J. Shin et al.4691 3470Table 6 SLM & OLS results for each region471***p < .01, ** < p < .05,*p < .1472Urban(N = 129)473Rural(N = 39)474SLM475OLS476SLM477OLS478Coef. (b)479z-value480Coef481t-value482Coef. (b)483z-value484Coef485t-value486Rho ( 휌)4870.877***48810.228489–4900.436***4913.117492–493Learning & Care494− 0.085*495− 1.871496− 0.089*497− 1.776498− 0.034*499− 1.907500− 0.040*501− 1.731502Education & Leisure503− 0.087*504− 1.943505− 0.078506− 1.592507− 0.050**508− 2.527509− 0.056**510− 2.183511Senior5120.103***5134.0485140.122***5154.3625160.0575171.3855180.0295190.560520ln(Age)521− 0.081**522− 2.169523− 0.076*524− 1.865525− 0.586***526− 8.638527− 0.564***528− 6.390529Total Units (dummy)530− 0.076531− 1.612532− 0.062533− 1.189534− 0.074535− 1.330536− 0.081537− 1.109538ln(Max Floor)5390.184***5404.5265410.191***5424.273543− 0.041544− 0616545− 0.011546− 0.121547Bus stop5480.284***5492.8225500.306***5512.767552− 0.164553− 0.6445540.0185550.055556CBD557− 0.035***558− 2.916559− 0.022560− 1.647561− 0.003562− 0.631563− 0.006564− 1.113565Geum River (dummy)5660.107**5672.1255680.117**5692.100570− 0.098571− 0.480572− 0.119573− 0.448574CONSTANT5750.3195760.5805775.686***57833.1975794.853***5806.5635816.900***58216.524583R-squared5840.5675850.5165860.8025870.748588Log-Likelihood/Adj. R-squared58955.449/–59049.511/0.48059118.495/–59214.686/0.670593Akaike info criterion (AIC)594− 88.899595− 79.022596− 14.990597− 9.371598Jarque–Bera test599–6001.664601–6020.501603Breusch-Pagan test6046.7026057.9356068.2336077.406608609610===== PAGE 15 (score 6, numeric-paren 0, table-words 0, var-words 6) =====611612741613Urban versus rural disparities in amenity proximity and housing…6141 3615the proximity of each Living SOC between urban and rural areas and how they affect hous-616ing prices. As expected, most of Living SOC except senior care and education facilities 617were much closer to the apartments in urban areas, while only senior care and senior edu-618cation facilities were closer to the apartments in rural areas. All three categories of Living 619SOC have significant effects on housing prices. The long-distance Learning & Care and 620Education & Leisure facilities tended to decrease housing prices in both urban and rural 621areas. In other words, the closer the Living SOC, the better the neighborhood satisfaction. 622These findings confirmed the previous findings that observed the positive impacts of urban 623amenities on neighborhood satisfaction (Choi et al., 2021; Wen et al., 2017; Yuan et al., 6242020). In the case of Senior facilities, housing prices increased when the distance to facili-625ties became farther, and there was no significant effect of proximity to facilities in rural 626areas.627The average distance disparities to Learning & Care and Education & Leisure facilities 628was 670 m. Based on this distance, adults in rural areas would need to walk eight minutes 629more, and the elderly would require an additional two minutes than that, based on the aver-630age walking speed provided by Bohannon and Andrews (2011); mean walking speed of the 631healthy elderly (60–69 years old) is approximately from 1.24 to 1.33 m/sec which could 632decline to 1 m/sec when gets older or sick. The average distance to senior care facilities 633was similar in urban and rural areas, and the distance to senior education facilities was 634much closer in rural areas. Like this, even with the expectation of positive spill-over devel-635opment and infrastructure to rural areas through balanced development, natural spillover, 636on the other hand, was not detected in existing areas. Rather, it was reported that neighbor-637hood satisfaction decreased and people claimed that relative deprivation occurred in nearby 638areas (Lee et al., 2018). Thus, new town plans should have consolidated urban amenity 639plans that cover adjacent existing areas.640As urban areas held a relatively low proportion of individuals over 60 in urban areas 641(5.8%) compared to rural areas (23.3%) (Ministry of Interior & Safety, 2022), Senior facili-642ties for the young may not be necessary or could be located far away. Even in rural areas 643with many elderly persons, Senior facilities would not be considered necessary facilities. 644This situation in which Senior facilities become useless would be a unique Korean phe-645nomenon. According to Kim (2019), the participation rate of the elderly in education was 646low compared to other age groups. Also, the elderly mostly use their community centers 647for interaction and relaxation in rural areas (Lee et al., 2012). So, there seems to be lit-648tle need for official Senior facilities as community centers can work as Senior facilities. 649It is, therefore, necessary to update the procedures for providing official senior care and 650education facilities, as well as the division of responsibilities. Moreover, as urban areas 651also become older as time passes, a strategic approach supporting the lifestyle of the urban 652elderly should be implemented as senior facilities work as a core for social integration of 653the elderly and significantly contribute to their Living SOC (Aday et al., 2006; Yoo, 2015; 654Zarghami et al., 2019).655Senior education facilities observed in this study are facilities that are specifically 656designed to aid the elderly in their social adaptation and enhance their physical and men-657tal well-being through various programs, including club activities and educational services 658such as self-development and social participation through computers (Han et al., 2011). In 659Sejong, a total of seven senior education facilities exist, each built before 2015, and all of 660them are located in rural areas. This is because the proportion of the elderly population is 661higher in rural areas, and also, more than half of the elderly in rural areas require literacy 662and other technological education for their daily lives (Ma et al., 2018). However, good 663accessibility to Senior facilities in rural areas where the elderly population continues to 664