# RECORD 4d2cf36586abe5a8dcb4cd171a95a31095559520 # TITLE: The influence of urban green and recreational areas on the price of housing in Zagreb # YEAR: 2024.0 VENUE: Hungarian Geographical Bulletin DOI: 10.15201/hungeobull.73.3.3 # AUTHORS: Mirela Turk Cerovečki; Zoran Stiperski # PAGES: 20 CANDIDATE TABLE PAGES (1-based): [8, 10] # ABSTRACT: Urban green and recreational spaces are important for residents and one of the foundations for the functioningof cities. They offer numerous benefits, including the influence on the price of real estate as an economic benefit.In this paper, the influence of green and recreational areas in Zagreb on the price of apartments in residential buildings is analysed using the hedonic pricing model. The data for the study was processed and analysed using a geographic information system. A local hedonic price model was developed. The results show that some of the observed green and recreational areas increase the price of housing due to their proximity, while others decrease it. Some areas have no influence on the price of housing. The influence of the observed areas on housing prices was heterogeneous. Spatial differences were also found in the influence of the observed areas on housing prices. The most attractive factors when choosing a place to live are Jarun and the parks in the city centre. The price of apartments that near the Jarun increases by 41 EUR/m2 if the apartment is 100 m closer to the Jarun. The price of the apartment increases by 91 EUR/m2 the closer the apartment is to the parks in the city centre (per 100 m). Apartments near Jarun and the parks in the city centre are not available for residents with lower socio-economic status. The forest areas, Maksimir Park and the banks of the Sava river are not favoured when choosing a place to live. Neighbourhoods near these areas are available to the lower socioeconomic status population. Bundek raises the price of one part of the apartments and lowers the price of the other. The results of this study can contribute to the study of green gentrification in Zagreb, but further research is needed on socio-economic indicators and other changes in the region. ===== PAGE 1 (score 15, numeric-paren 0, table-words 0, var-words 15) ===== 249 Turk Cerovečki, M. and Stiperski, Z. Hungarian Geographical Bulletin 73 (2024) (3) 249–268. DOI: 10.15201/hungeobull.73.3.3 Hungarian Geographical Bulletin 73 2024 (3) 249–268. Introduction A city is a space that is characterized by a compact structure and whose elements are in constant interaction. One of the elements of the functional-spatial structure of the city are recreational areas. Recreational areas are open spaces with built infrastructure in­ tended for recreational activities. Their basic characteristics are the predominant natural environment, which includes vegetation, i.e. greenery (forest, grass, flowers), and often also water areas (river, canal, lake) and a certain number of built objects (paths, roads, buildings, playgrounds, sports halls, sports facilities). These areas are used by users for recreational activities. The largest recrea­ tional areas are generally located on the out­ skirts of the city, as this is where the largest open spaces are available. Green areas in the city consist of parks, gardens, urban forests, promenades and other smaller areas that are covered with vegetation but have fewer or no buildings (they may include walking paths, benches and other rest areas). In addition to other functions, the green areas integrated into the urban structure also serve active and passive recreation. 1 Postgraduate Studies in Geography. Department of Geography, Faculty of Science, University of Zagreb. Knezovec 148, Knezovec, 40000 Čakovec, Croatia. E-mail: mirela.turk@gmail.com 2 Department of Geography; Faculty of Science, University of Zagreb, Marko Marulić square 19/II. 10 000 Zagreb; Croatia. E-mail: zstiper@geog.pmf.hr The influence of urban green and recreational areas on the price of housing in Zagreb Mirela TURK CEROVEČKI 1 and Zoran STIPERSKI 2 Abstract Urban green and recreational spaces are important for residents and one of the foundations for the functioning of cities. They offer numerous benefits, including the influence on the price of real estate as an economic benefit. In this paper, the influence of green and recreational areas in Zagreb on the price of apartments in residential buildings is analysed using the hedonic pricing model. The data for the study was processed and analysed using a geographic information system. A local hedonic price model was developed. The results show that some of the observed green and recreational areas increase the price of housing due to their proximity, while others decrease it. Some areas have no influence on the price of housing. The influence of the observed areas on housing prices was heterogeneous. Spatial differences were also found in the influence of the observed areas on housing prices. The most attractive factors when choosing a place to live are Jarun and the parks in the city centre. The price of apartments that near the Jarun increases by 41 EUR/m2 if the apartment is 100 m closer to the Jarun. The price of the apartment increases by 91 EUR/m2 the closer the apartment is to the parks in the city centre (per 100 m). Apartments near Jarun and the parks in the city centre are not available for residents with lower socio-economic status. The forest areas, Maksimir Park and the banks of the Sava river are not favoured when choosing a place to live. Neighbourhoods near these areas are available to the lower socioeconomic status population. Bundek raises the price of one part of the apartments and lowers the price of the other. The results of this study can contribute to the study of green gentrification in Zagreb, but further research is needed on socio-economic indicators and other changes in the region. Keywords: urban green areas, recreational areas, hedonic model, geographically weighted regression, real estate Received April 2024, accepted August 2024. ===== PAGE 8 (score 36, numeric-paren 0, table-words 13, var-words 4) ===== Turk Cerovečki, M. and Stiperski, Z. Hungarian Geographical Bulletin 73 (2024) (3) 249–268. 256 example when mutually correlated variables are of great interest for research. Since the environmental variables are of interest, none of the highly correlated variables were omit­ ted. Due to the use of georeferenced data, only the results of the local hedonic model are presented in this paper. The geographically weighted regression was performed with 120 neighbours. The Gaussian weighting method was used, ac­ cording to which the weight decreases ex­ ponentially the further the point is from the observed point. Results The spatial autocorrelation in the local hedonic model was tested with the Mo­ ran I index, which is 0.185698 (p-value = 0.00000, z-score = 12.780390) and shows a statistically significant clustering of the analysis results. The coefficient of determi­ nation is R2 = 0.4090 and the adjusted coef­ ficient of determination is adj. R2 = 0.3616 (Table 2). There are differences in the coefficient of determination between the apartments (Figure 2). The highest coefficients of deter­ mination are observed west of Maksimir Park and in the western part of Novi Zagreb (Kajzerica). The lowest coefficient of determi­ nation is observed in the wider surroundings of the city centre (Donji grad, Trnje). The statistical significance of the model was tested at the 10 percent level (p-value < 0.1, t-value > 1.660). Since the attractiveness of certain factors in space is determined by their proximity (distance) to the apartment, a geographically weighted regression was used to account for apartments that are up to Table 2. Local (GWR) model coefficients Variables Coefficients min mean max Intercept -36,531.74 -1808.923 18,759.15 Structural variables (S) area floor -7.01288 -7.27306 -3.596185 54.1917 -0.424051 221.9165 Neighbourhood variables (N) dist_public dist_kinder dist_elem num_fax dist_shop dist_ind dist_health dist_hosp num_kult num_sportobj num_sportigr -0.490508 -0.91591 -2.562863 -1404.26 -1,533588 -1.073735 -1.153004 -0.606382 -111.232 -114.2322 -277.8436 1.27809 0.558948 -0.014774 6.95362 -0,084314 0.209952 0.250067 0.154938 7.15274 -11.7921 -3.03338 4.15467 2.48765 3.054714 206.193 0,841975 2.349503 2.759074 2.718514 176.7045 173.7824 558.92 Environmental variables (E) dist_zona dist_forest dist_jarun dist_bundek dist_bank dist_maksimir dist_park -5.15309 -3.542074 -2.05610 -14.39090 -14.45573 -2.89827 -4.37027 0.079955 0.075858 0.972864 0.983093 -1.035173 0.507891 -0.664219 3.65268 5.27866 8.43639 12.36982 17.62028 5.95143 0.91697 N R2 Adjusted R2 2314 0.4090 0.3616 Source: Authors’ own elaboration. ===== PAGE 9 (score 20, numeric-paren 0, table-words 4, var-words 12) ===== 257 Turk Cerovečki, M. and Stiperski, Z. Hungarian Geographical Bulletin 73 (2024) (3) 249–268. 2000 m away from a green or recreational area (Table 3). The distance given is the upper limit for a walk (about 25 minutes). Some studies have shown that the influence of the observed environmental elements on the property price extends up to a maximum distance of 2000 m (Melichar, J. and Kaprová, K. 2013), in some cases also over a shorter distance (Tyrväinen, L. and Miettinen, A. 2000; Biao, Z. et al. 2012; Daamsa, M.N. et al. 2019). Although the neighbourhood variables are not the direct object of interest of this re­ search, they should be briefly commented on. According to their mean values of the coef­ ficients, four variables increase the value of real estate with their attractiveness (when all other variables in the regression are constant) (see Table 2). These are the distance from the elementary school, the distance from the shopping center, the number of higher edu­ cation institutions and the number of cultural institutions within 500m of the apartment. The average of the coefficients shows a high value for the variables number of higher edu­ cation and cultural institutions. The proxim­ ity of the shopping center is an attractive fac­ tor despite the crowds, noise and large built- up areas (buildings, parking lots, roads). Other neighbourhood variables show unattractive factors for housing. Increasing the distance from health institutions (health center, hospital) increases the price of real estate, which is contrary to expectations. The same was expected for the variables distance from the nearest public transport station and the nearest kindergarten. The coefficients of the variables number of sports facilities and number of sports fields within 500 m of the property show the opposite of what was ex­ pected. Increasing the number of such facili­ ties reduces the price of real estate. Possible causes are crowding and noise. The results of the local hedonic model (GWR) show that there are large ranges of variation in Fig. 2. Coefficient of determination R2 in the local hedonic model. 1 = SRC Bundek, 2 = RSC Jarun. Urban park-forests: 3 = Čulinečina, 4 = Šestinski dol, 5 = Susedgrad, 6 = Lisičina, 15 = Mirogoj, 16 = Remetinski ka­ menjak, 17 = Granešina (Urban of Youth), 18 = Remete, 19 = Oporovec, 20 = Novoselčina, 21 = Miroševečina, 22 = Dankovečina, 23 = Zamorski Breg, 24 = Dotršćina, 26 = Grmošćica. Park-forests: 7 = Zelengaj, 8 = Jelenovac, 9 = Vrhovec, 10 = Tuškanac-Dubravkin put-Cmrok, 12 = Kraljevec, 13 = Pantovčak, 14 = Prekrižje. 11 = Maksimir Park, 25 = Sava dam. Source: Compiled by the authors. ===== PAGE 10 (score 18, numeric-paren 0, table-words 5, var-words 6) ===== Turk Cerovečki, M. and Stiperski, Z. Hungarian Geographical Bulletin 73 (2024) (3) 249–268. 258 Table 3. Results of the local hedonic model for apartments up to 2000 m away from a green or recreational area Variables Count Percentage Statistic for significant properties (coefficients) Part of the apartment price* in EUR Non sign Sign** Sign - Sign + Sign Sign - Sign + min max mean min max mean dist_zona dist_forest dist_jarun dist_bundek dist_bank dist_maksimir dist_park 930 852 188 66 486 180 620 1202 723 213 89 38 190 487 609 143 212 50 3 0 457 593 580 1 39 35 190 30 56 46 53 57 7 51 44 51 20 100 56 8 0 94 49 80 0 44 92 100 6 -5.153090 -3.542070 -0.621160 -1.512264 -0.432529 0.167898 -1.945374 3.652680 5.278660 0.296880 0.631705 1.285763 2.580130 0.908357 0.083059 0.337271 -0.406531 -0.231906 0.705826 0.623056 -0.907034 -515 -354 -62 -151 -43 17 195 365 528 30 63 129 258 91 8 37 -41 -23 71 62 -91 *Part of the apartment price for statistically significant apartments by individual variable, the change was estimated for every 100 m of distance from a green or recreational area. The Sign (-) shows a decrease in price with increasing distance from the observed green or recreational area, and the Sign (+) shows the inverse relationship. The unit of measurement is EUR/m2. **Statistically significant (p < 0.1). Source: Authors’ own elaboration. the coefficients of the individual environmental variables (see Table 2). These results should be viewed with caution as they include apartments for which a single environmental variable is not statistically significant. For this reason, we will not interpret them here. We will interpret the results for the apartments for which the envi­ ronmental variables are statistically significant (p < 0.1, t > 1.660) and which are located at a dis­ tance of up to 2000 m from the observed green or recreational area (see Table 3). The variable distance to the nearest green or recreational area (dist_zona) is statistically significant for 56 percent of the apartments. It has a negative sign for 51 percent of the apart­ ments, which means that proximity to this area increases the price of the apartment. Analogous to the coefficients, the value range of the price share associated with this variable ranges from -515 to 365 EUR/m2, with an average of 8 EUR/m2 (see Table 3). The geographical distri­ bution of the coefficients (Figure 3) shows that the coefficients with a negative sign are concen­ trated in the western and south-western part of the city, while the coefficients with a positive sign predominate in the rest of the city, with the exception of the area around the forests of Mirogoj, Remetinski kamenjak and Remete. Proximity to an urban park-forest or park-for­ est (dist_forest) has a statistically signifi­ cant effect on 46 percent of apartments up to 2000 m away. The coefficients of the variables have a positive sign for most apartments (80%). According to the coefficients, the share of hous­ ing price attributable to proximity to the near­ est forest ranges between -354 and 528 EUR/m2, i.e., the extreme values, and the mean value is 37 EUR/m2 (see Table 3). The coefficients with a negative sign (the variable increases the price of the apartment with decreasing distance from the forest) are located near the seven park-for­ ests in the area of the historic city centre (Park- forests Zelengaj, Jelenovac, Vrhovec, Tuškanac- Dubravkin put-Cmrok, Kraljevec, Pantovčak and Prekrižje) and two urban park-forests (Zamorski Breg, Šestinski dol). It can be seen that other forest areas lower the price of housing the closer they are to the forest (coefficients with a positive sign) (Figure 4). ===== PAGE 11 (score 3, numeric-paren 0, table-words 0, var-words 3) ===== 259 Turk Cerovečki, M. and Stiperski, Z. Hungarian Geographical Bulletin 73 (2024) (3) 249–268. Fig. 3. Coefficients of the variable distance to the nearest green or recreational area. Numbers 1–25: see Fig. 2. Source: Compiled by the authors. Fig. 4. Coefficients of the variable distance to the nearest urban park-forest or park-forest. Numbers 1–25: see Fig. 2. Source: Compiled by the authors.