R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 155 www.r-economy.ru Online ISSN 2412-0731 Original Paper © Ju. G. Lavrikova, A. V. Suvorova, 2019 doi 10.15826/recon.2019.5.4.016 Spatial aspects of regional infrastructure distribution (the case of Sverdlovsk region) Ju. G. Lavrikova1 , A. V. Suvorova1, 2 1 Institute of Economics, Ural Branch of the Russian Academy of Sciences, Ekaterinburg, Russia; e-mail: lavrikova_ug@mail.ru 2 Ural State University of Economics, Ekaterinburg, Russia ABSTRACT The article discusses the correlation between the localization of specific infra- structure objects within a region and characteristics of this region’s territorial development. Conceptually the study is grounded in the theory of regional economics, spatial analysis and modelling and uses the tools of spatial autocor- relation analysis, such as the global and local Moran’s I, and map-based spatial analysis. The settlement system of Sverdlovsk region (Russia) is considered as a key characteristic of its territorial development and the analysis shows the cor- relation between settlement patterns and the distribution of certain objects of social infrastructure (places of attraction) across the region’s territory. Access to infrastructure is an important factor which attracts people to this or that municipality. However, the key parameter that determines the spatial aspects of infrastructure distribution in the region is the emergence and development of the factors underlying this process. The article demonstrates that the local- ization of infrastructure objects built to generate economic effects and bring profit to their developers to a greater extent correlates with the prospective transformations of the settlement system (primarily agglomeration processes) rather than with its current characteristics (such correlation is more typical of the infrastructure objects specifically intended to address social issues). These research findings can be used by policy-makers for setting priorities of region- al development, which would shape the spatial transformations of the territory. KEYWORDS space, settlement system, distribution, social infrastructure, region, spatial autocorrelation, Moran’s I, map-based spatial analysis, Sverdlovsk region ACKNOWLEDGEMENTS The research was supported by the Institute of Economics, Ural Branch of the Russian Academy of Sciences in accordance with the plan for 2019–2021. FOR CITATION Lavrikova Ju. G., Suvorova A. V. (2019) Spatial aspects of regional infrastructure distribution (the case of Sverdlovsk region). R-economy, 5(4), 155–167. doi: 10.15826/recon.2019.5.4.016 Пространственные аспекты инфраструктурного обустройства региона: пример Свердловской области Ю. Г. Лаврикова1 , А. В. Суворова1, 2 1 Институт экономики Уральского отделения Российской академии наук, г. Екатеринбург, Россия; e-mail: lavrikova_ug@mail.ru 2 Уральский государственный экономический университет, г. Екатеринбург, Россия АННОТАЦИЯ Статья посвящена оценке степени соответствия характера размещения в  пространстве региона элементов инфраструктуры особенностям его территориального развития. Теоретическую и методологическую основу исследования составляет совокупность научных представлений в  обла- сти региональной экономики, пространственного анализа и моделиро- вания. На основе оценки пространственной автокорреляции (с помощью определения величин как глобального, так и локального индекса Морана) и осуществления картографического анализа выделены и сопоставлены друг с другом особенности сложившейся в Свердловской области систе- мы расселения (как одной из ключевых характеристик ее территориаль- ного развития) и взаиморасположения в регионе элементов инфраструк- туры мест проживания и мест притяжения. Показано, что размещение объектов социальной инфраструктуры в целом соответствует характеру расположения на территории ее основных потребителей – жителей реги- она. Однако ключевым параметром, определяющим пространственные аспекты инфраструктурного обустройства территории, является гене- зис факторов, лежащих в основе данного процесса. Доказано, что лока- лизация инфраструктурных объектов, главной целью создания которых КЛЮЧЕВЫЕ СЛОВА пространство, система расселения, размещение, социальная инфраструктура, регион, пространственная автокорреляция, индекс Морана, картографический анализ, Свердловская область БЛАГОДАРНОСТИ Работа выполнена при поддержке Института экономики Уральского отделения Российской академии наук в соответствии с планом на 2019–2021 гг. http://doi.org/10.15826/recon.2019.5.4.016 http://doi.org/10.15826/recon.2019.5.4.016 mailto:lavrikova_ug@mail.ru mailto:lavrikova_ug@mail.ru 156 www.r-economy.ru R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 Online ISSN 2412-0731 Introduction The distribution of infrastructure elements across space is one of the main topics not only in regional economics and economic geography but also in policy-making on different levels of the territorial hierarchy. The principal difficulty is to determine the parameters for the optimal location of such infrastructure objects. It seems obvious that the key criterion should be the ability of infrastructure objects to meet the needs of the main stakeholders, regardless of the level of the territory – cities, regions or the coun- try as a whole. It means that the distribution of infrastructure should correlate with the concen- tration of its users, that is, the latter should be pro- vided with a convenient access to these objects. Developed infrastructure, in its turn, attracts more residents to the area, which is conducive to socio-economic growth and turns infrastructur- al development into a powerful tool of regional policy-making. Irrespective of whether the infra- structure is going to be developed in accordance with the already existing settlement patterns and distribution of productive forces across the terri- tory or with the view to future transformations of the socio-economic space, decision-making in this sphere is based primarily on the analysis of the current situation: before building new objects of infrastructure, it is necessary to assess different parameters of the region’s development, in partic- ular the already existing infrastructure, and iden- tify the gaps and disproportions that need to be addressed. This study is aimed at analyzing the charac- teristics of territorial development of a region and revealing their correlation with the localization of infrastructure objects in the given area. It should be noted that such analysis should take into ac- count different types of infrastructure. Our study focuses on the discrepancies between settlement patterns of a region (characteristics of its territo- rial development) and localization of some ele- ments of social infrastructure in the same region (infrastructure necessary for maintaining and im- proving the living conditions). Theoretical framework Spatial aspects of economic development now attract considerable scholarly attention in Russia, especially after the adoption of the federal law ‘On Strategic Planning in the Russian Federation’1 in 2014. This law identifies the strategy of spatial development as one of the key strategic planning documents. However, it should be noted that the research on the relationship between the distribution of economic entities across space and specific param- eters of territorial development goes back to the nineteenth century. The classical location theory developed by J. H. von Thünen [1], A. Weber [2], A. Lösch [3], W. Christaller [4], and C. W. F. Laun- hardt [5] described the factors that determine the localization of industries in space. Spatial aspects of territorial development were also considered by the growth poles theory and the theory of polar- ized development, theories and concepts of urban development [8; 9], and so on. Questions related to distribution of productive forces were also discussed by Soviet economists, such as N. N. Nekrasov [10], I.  G.  Alexandrov [11], A. E. Probst [12] and others. Interestingly enough, as A. I. Tatarkin and E. G. Animitsa point out in their article on the paradigm theory of re- gional economy, seminal works written by West- ern authors had little impact on the theoretical views of Soviet scholars in what concerned the distribution of industrial enterprises and region- al development. Nevertheless, the development of territorial studies in the USSR, which dealt primarily with the radical shifts in the location of productive forces, theory and practice of eco- nomic zoning, factors that determine the location of industries, to some extent coincided with the international trends. The research of the role played by spatial fac- tors in the development of socio-economic sys- tems requires a methodological approach that would not rely exclusively on evaluating the dy- 1 Federal Law No. 172-FZ of 06.26.2014 ‘On Strategic Planning in the Russian Federation’. Retrieved from: http:// www.consultant.ru/document/cons_doc_LAW_164841 выступает генерация экономических эффектов и получение прибыли, в большей степени коррелирует не с текущими особенностями системы расселения (что характерно для объектов, создание которых призва- но способствовать решению социальных проблем), а  с  перспективами ее преобразования, проявляющимися тенденциями (в  первую очередь, с агломерационными процессами). Полученные результаты могут найти применение при определении приоритетов осуществления региональ- ной политики, пространственных преобразований территорий. ДЛЯ ЦИТИРОВАНИЯ Lavrikova Ju. G., Suvorova A. V. (2019) Spatial aspects of regional infrastructure distribution (the case of Sverdlovsk region). R-economy, 5(4), 155–167. doi: 10.15826/recon.2019.5.4.016 http://doi.org/10.15826/recon.2019.5.4.016 http://www.consultant.ru/document/cons_doc_LAW_164841 http://www.consultant.ru/document/cons_doc_LAW_164841 R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 157 www.r-economy.ru Online ISSN 2412-0731 namics of certain objects in time but consider the specific parameters of these objects’ distribution across space, that is, the proximity of objects to each other, their concentration within one area and the scale of the systems they form. Therefore, such studies prioritize methods of spatial analysis and modelling. Without going into a detailed discussion of the history of spatial analysis, we need to mention that this methodology goes back to the 1940s and 1950s, when the first papers on spatial modelling were published [14; 15]. At the subsequent stag- es [16; 17], more new methods for estimating the spatial effects produced by the transformations on different levels were proposed. These methods provided sufficient foundation for a vast number of empirical studies, including the studies based on Russian data. Spatial autocorrelation analysis has been gain- ing popularity among Russian scholars [18–22]. Spatial autocorrelation can be defined the follow- ing way: for set S containing n geographical units, spatial autocorrelation is a correlation between the variable observed in each of the  n  localities and a measure of geographical proximity defined for all n (n − 1) pairs chosen from S [23]. In other words, spatial autocorrelation analysis shows the strength of correlation between the parameters characterizing the development of territories lo- cated in close proximity to each other. One of the most widely applied (and relatively easy to use) parameters is Moran’s I. The test for spatial autocorrelation proposed by Patrick Mo- ran is used in most Russian studies of patterns of spatial dependence between neighbouring terri- tories. Various indicators can be used to describe the situation in the given territories: for example, Y. V. Pavlov and E. N. Koroleva analyzed territori- al clusters in Samara region by looking at the pop- ulation data of its municipalities [18]. A. A. Grig- oriev estimated the scale of spatial autocorrelation in Russian regions by using such parameters as ed- ucation, crime rates, birth rates, infant mortality rates, urbanization, migration, urbanization and household income [19]. O. A. Demidova focused on the level of unemployment [20]; O. S.  Balash, on the GRP per capita [21]; E. S. Inozemtsev and O. V.  Kochetygova, on birth rates and life expec- tancy [22]. If we look at the theoretical and method- ological foundations of Russian and international studies of economic space, we can see that spa- tial analysis methods hold enormous potential as they help us search for correlations between vari- ous parameters of territorial development and the localization of infrastructure within this territory. Methodology and data This study focuses on the case of Sverdlovsk region in Russia, which comprises 73 municipali- ties – 68 urban districts and 5 municipal districts. The choice of indicators was determined by the fact that any area can be seen from the per- spective of its potential users as a place to live and work in and as a place of attraction, that is, as a source of opportunities for leisure and recreation. Elements of social infrastructure can be classified the same way: amenities and benefits for living and work; infrastructure for sport and leisure. In this study, we decided to focus on the social infrastructure used by people in their daily lives (we use the supply of new housing as an indicator) and the infrastructure that turns certain spots into places of attraction (for example, the number of stadiums with terraces). We did not consider in- frastructure objects that are necessary for creating a comfortable working environment, although the proposed methodology would make it possible to consider those as well. Moreover, this method- ological approach can be applied to analyze the spatial distribution of infrastructure elements of other types, for instance, those unrelated to the social sphere or linked to other indicators such as cultural facilities, public improvements and so on. As an indicator characterizing settlement pat- terns, we took the number of permanent residents in the municipalities of the region. In order to obtain the necessary data on the population, new housing supply and the number of stadiums with terraces for specific municipalities, we used the database2 of the Federal State Statistics Service. The study period was one year – 2017. The study comprised several stages: at the first stage, we focused on the settlement patterns in the region and searched for correlations between the population size of neighbouring municipalities. Thus, we were able to identify clusters within the regional settlement system. At the second stage, we investigated the distribution of specific ele- ments of infrastructure across the region and its correlation with the settlement patterns. Methodologically, this study relies on calcula- tions for Moran’s I and map-based analysis. 2 Official website of the Federal State Statistics Service. Database of municipal indicators. Retrieved from: http://www. gks.ru/dbscripts/munst/ http://doi.org/10.15826/recon.2019.5.4.016 http://www.gks.ru/dbscripts/munst/ http://www.gks.ru/dbscripts/munst/ 158 www.r-economy.ru R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 Online ISSN 2412-0731 We calculated the parameters of spatial auto- correlations (based on Moran’s test) by following the procedure described below. First, a distance matrix was generated. The matrix shows the distances between all the given territorial units. Entries for the matrix can be de- termined in different ways: for example, an entry may be equal to 0 (if the territories do not share a border) or 1 (if they do). Entries can be also de- termined by using aerial distance data, the length of the roads or railways between the territories in question. We built the distance matrix by using the data on the length of the roads connecting administra- tive centres of the municipalities. The region has three municipalities whose administrative cen- tres are located outside their borders and, there- fore, coincide with the administrative centres of the neighbouring municipalities (Kamensky and Krasnoufimsk urban districts, Kamyshlovsky mu- nicipal district). The distance between these mu- nicipalities and the neighbours which they share their ‘capital’ with was taken as 0. Second, we calculated the global Moran’s I and looked for the spatial autocorrelation or its absence. The formula for the global Moran’s I (1) looks the following way: = = = − − = − ∑ ∑ ∑ 1 1 2 0 1 ( )( ) , ( ) n n ij i j i j n i i n w x x x x I S x x (1) where I is the global Moran’s I, x is the given param- eter, S0 is the sum of spatial weights ( = = = ∑ ∑0 1 1 ij i j S w ), and n is the number of territories. The index values may vary between –1 and 1. We need to compare the actual value with the ex- pected value (2) to make a conclusion about the presence or absence of spatial autocorrelation and its character. − = − 1 ( ) , 1 E I n (2) where E(I) is the expected value and n is the num- ber of territories. These values can be interpreted the following way. If the calculated value of Moran’s I exceeds the expected value, we observe a positive spatial auto- correlation (the values of the given indicator for neighbouring areas are similar or close to each oth- er); if the expected value exceeds the value of Mo- ran’s I, it means that there is a negative spatial au- tocorrelation (the values of the given indicator for neighbouring areas are different). If the expected value of Moran’s I coincides with the actual value, it means the absence of spatial autocorrelation [21]. To test for significance of Moran’s I, we use a z-test – a traditional procedure for hypothesis testing in econometrics. The z-score for Moran’s global I is calculated by applying the following formula: − = −2 2 ( ) - , ( ) ( ) I E I z score E I E I (3) where I is the global Moran’s I and E(I) is the ex- pected value. The z-score thus obtained is the measure of how many standard deviations above or below the expected value the actual value of Moran’s I is. If the above value is sufficiently high, it means that the actual distribution did not occur by chance. Third, we calculate the local Moran’s I and find the strength of correlation between the ter- ritories. The local Moran’s I shows the interdepen- dence between the territories and its strength [25, p. 147]. The local Moran’s I can be calculated by applying formula (4): = ∑ ,iL i ij jI z w z (4) where ILi is the local Moran’s I for the ith territo- ry, wij is the standardized distance between the ith and jth territories, zi and zj are the standardized values of the given indicator for the ith and jth ter- ritories. The values we obtain may be negative (min- imum –1) or positive (maximum 1) and can be interpreted by following the same logic as for the global Moran’s I. It is also interesting to look at the separate components of local index (5), whose values char- acterize the strength of interdependence between the two territories [18]: = ,ij i j ijLISA z z w (5) where LISAij is the strength of interdependence between the ith and the jth areas, wij is the stan- dardized distance between the ith and jth areas, zi and zj are the standardized values of the given indicator for the ith and jth areas. Fourth, the territories are grouped according to the correlation between the standardized val- ues of the given indicators and the values of the spatial factor. If we combine the standardized values of the given indicator (z) with its spatial centred weights (wz) for each given territory within one system of http://doi.org/10.15826/recon.2019.5.4.016 R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 159 www.r-economy.ru Online ISSN 2412-0731 axes, we can notice that the points (corresponding to the territorial units) concentrate in one of the four quadrants [24, p. 50]. If values z and wz are positive (quadrant HH), it means that territories characterized by high val- ues in the given indicator are clustered with adja- cent territories, which also demonstrate high val- ues. If values z and wz are negative (quadrant LL), it means that the territories are located near other areas with similar values in the given parameter, but in both cases the territories demonstrate a low level of performance in the given indicator. If value z is positive while value wz is negative (quadrant HL), the territory is different from its neighbours – it is ahead of the adjacent territories in this indicator. If value z is, on the contrary, neg- ative, while value wz is positive (quadrant LH), the territory lags behind its neighbours. Thus, terri- tories with a positive autocorrelation fall within the quadrants HH and LL, with negative autocor- relation – quadrants HL and LH. Such grouping demonstrates the place of each territorial unit in this spatial system, shows its leaders (extrema) and peripheral areas and allows us to make spatial clustering. The map helps us display the results of spa- tial data analysis and complements other research methods. Maps can be used as spatial models of real-life situations, illustrating the already existing or planned structures and relationships in a so- cio-economic space. If we add new information to the map (symbols and pictograms characterizing the localization of the objects, lines in different thicknesses to show the strength of interdepen- dence between the specific territories, different colours to highlight some parts of the map, and so on), we can show subtle trends, relationships and correlations. Results and discussion The population density in Sverdlovsk region (the map of the region with its municipalities is shown in Figure 1) is uneven, with 34.7% of the population living in the region’s administrative centre – Ekaterinburg. The population of the ur- ban agglomeration of Ekaterinburg (its boundar- ies are defined by the Territorial Planning Scheme of Sverdlovsk region3) is over 2,242 thousand peo- ple or 51.8% of the total population of the region. 3 Decree of the Government of Sverdlovsk Region No. 1000-PP of August 31, 2009 ‘On the Approval of the Territorial Planning Scheme of Sverdlovsk Region’. Retrieved from: http:// docs.cntd.ru/document/895218020 It should be noted that the size of the area of the region’s constituent municipalities is only 6.8% of the total area of Sverdlovsk region4. In order to estimate spatial autocorrelation, we analyzed the data on the population of mu- nicipalities in Sverdlovsk region and found that there is an inverse relationship between the val- ues of this indicator for nearby localities: the actual value of the global Moran’s I (–0.021) is smaller than the expected value, which means that there is a negative autocorrelation. The sig- nificance of this result is confirmed by the z-test. This means that the population size varies sig- nificantly from municipality to municipality. It should be noted, however, that negative values of Moran’s I can be explained by the sheer size of the largest municipality – Ekaterinburg: it dif- fers considerably not only from the region’s av- erage but also from its nearest neighbours, even though many nearby territories have quite large populations. The local Moran’s I for Ekaterin- burg is –0.010, which means that if we exclude this municipality from our calculations, the val- ue of the global Moran’s I (for the whole region) will exceed the expected value. The values of local indices calculated with the help of formula (4) show that large munici- palities, such as Ekaterinburg, Nizhny Tagil, and Kamensk-Uralsky, differ significantly from their neighbours. The same applies to the municipali- ties located in closest proximity to these cities (see Table 1). Thus, we can suppose that Ekaterinburg, Nizhny Tagil and Kamensk-Uralsky concentrate most population in the region (extrema) and that they are the leaders of their respective territorial clusters. Table 1 Municipalities characterized by negative values of the local Moran’s I Municipality ILi Municipality ILi Ekaterinburg –0.010 Degtyarsk –0.001 Kamensky –0.003 Closed settlement ‘Uralsky’ –0.001 Kamensk-Uralsky –0.002 Nizhny Tagil –0.002 Verkhnee Dubrovo –0.001 Gornouralsky –0.002 Verkh-Neyvinsky –0.001 Sredneuralsk –0.001 Aramilsky –0.001 The table does not include the data on those municipalities whose values of the local Moran’s I are negative but are closer to 0 than to –0.001. 4 Official website of the Federal State Statistics Service. Database of municipal indicators. Retrieved from: http://www. gks.ru/dbscripts/munst/ http://doi.org/10.15826/recon.2019.5.4.016 http://docs.cntd.ru/document/895218020 http://docs.cntd.ru/document/895218020 http://www.gks.ru/dbscripts/munst/ http://www.gks.ru/dbscripts/munst/ 160 www.r-economy.ru R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 Online ISSN 2412-0731 In order to make a more solid conclusion about the spatial characteristics of the settlement system in Sverdlovsk region, we need to group the territories according to the correlation between the standardized values of the indicator and the values of the spatial factor. Moran’s diagram of spatial dispersion (Figure 2) illustrates the distri- bution of z points in the system of axes z and w. Each point corresponds to one of the municipal- ities. The three points located to the right of the vast majority of points are the obvious leaders we have already identified above. Nevertheless, along with the easily identified extrema, there are other municipalities in the Ivdelsky UD Pelym UD Garinsky UD Severo- uralsky UD 1 2 Krasnoturyinsk UD Karpinsk UD Novolyalinsky UD Sosvinsky UD Verkhotursky UD Alapaevskoye MD 3 4 5 6 7 Kush- vinsky UD Gornouralsky UD Nozhneturinsky UD Kachkanarsky UD Lesnoy UD Verkhnyaya Tura UD Nizhny Tagil Taborinsky MD Turinsky UD Tavdinsky UD Slobodo-Turinsky MD 18 Rezhe- vskoy UD 19 20 Arte- movsky UD Irbitskoye MD 8 39 Krasnou�msk UD Shakinsky UD 11 21 9 10 1312 14 1516 17 2425 27 2622 23 30 31 32 34 Revda UD 35 33 28 36 Kamensk-Uralsky UD Kamyshlovsky UD Kamyshlovsky MD 37 38 40 41 42 29 1 – Volchansky urban district; 2 – Serovsky urban district; 3 – Krasnouralsk urban district; 4 – Verkhnesaldinsky urban district; 5 – closed settlement Svobodny; 6 – NizhnyayaSalda urban district; 7 –Alapaevsk municipal district; 8 – Irbit municipal district; 9 – Kirovgradsky urban district; 10 – Nevyansky urban district; 11 – Staroutkinsk urban district; 12 – Nizhny Tagil urban district; 13 – Verkh-Neyvinsky urban district; 14 – Novouralsky urban district; 15 – Verkhnyaya Pyshma urban district; 16 – Sredneuralsk urban district; 17 – Berezovsky urban district; 18 – Malyshevsky urban dis- trict; 19 – Reftinsky urban district; 20 – Asbestovsky urban district; 21 – Bisertsky urban district; 22 – Degtyarsk urban district; 23 – Ekaterinburg urban district; 24 – Verkhnee Dubrovo urban district; 25 – Zarechny urban district; 26 – Be- loyarsky urban district; 27 – closed settlement Uralsky; 28 – Aramilsky urban district; 29 – Baikalovsky municipal dis- trict; 30 – Аchitsky urban district; 31 – Krasnoufimsky municipal district; 32 – Artinsky urban district; 33 – Pervouralsk urban district; 34 – Nizhneserginsky municipal district; 35 – Polevskoy urban district; 36 – Sysertsky urban district; 37 – Kamensky urban district; 38 – Bogdanovich urban district; 39 – Sukhoy Log urban district; 40 – Pyshminsky urban district; 41 – Talitsky urban district; 42 – Tugulymsky urban district Figure 1. Municipalities of Sverdlovsk region http://doi.org/10.15826/recon.2019.5.4.016 R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 161 www.r-economy.ru Online ISSN 2412-0731 HL group (areas with a higher population con- centration than their neighbours) such as Serov, Novouralsk and Krasnoturyinsk (Table 2). Their values of spatial autocorrelation are too close to zero, which means that their impact on the sur- rounding territories is insignificant. One more group of municipalities with rel- atively large populations (and positive autocor- relation values) includes seven territories (group HH). These municipalities do not qualify as cen- tres of the settlement system and, therefore, they do not dominate the surrounding territories. At –0.010 –0.005 0.000 0.005 0.010 0.015 0.020 –1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 W z Z Figure 2. Moran’s diagram of spatial dispersion (parameter – ‘resident population size’) Table 2 Groups of municipalities with different positions in the regional settlement system Municipality ILi Municipality ILi LH HH Kamensky urban district –0.0029 Pervouralsk urban district 0.0015 Gornouralsky urban district –0.0016 Verkhnyaya Pyshma urban district 0.0013 Sredneuralsk urban district –0.0014 Berezovsky urban district 0.0006 Closed settlement ‘Uralsky’ –0.0012 Polevskoy urban district 0.0002 VerkhneeDubrovo urban district –0.0011 Sysertsky urban district 0.0001 Aramilsky urban district –0.0011 Revda urban district 0.0001 Degtyarsk urban district –0.0007 Asbestovsky urban district 0.0000 Verkh-Neyvinsky urban district –0.0006 Zarechny urban district –0.0003   Staroutkinsk urban district –0.0003   Bysertsky urban district –0.0003   Beloyarsky urban district –0.0003   Malyshevsky urban district –0.0002   Closed settlement ‘Svobodny’ –0.0002   VerkhnyTagil urban district –0.0002   Reftinsky urban district –0.0002   Kirovgradsky urban district –0.0001   Nizhneserginsky municipal district –0.0001   Nevyansk urban district –0.0001   Shalinsky urban district –0.0001   Rezhevskoy urban district –0.0001   Bogdanovich urban district –0.0001   Artinsky urban district –0.0001   NizhnyayaSalda urban district –0.0001   Sukhoy Log urban district 0.0000   Achitsky urban district 0.0000   Alapaevskoye municipal district 0.0000   Makhnevskoye municipal district 0.0000   Alapaevsk urban district 0.0000   Artemovsky urban district 0.0000   Verkhnesaldinsky urban district 0.0000   LL HL Other municipalities Ekaterinburg –0.0104 http://doi.org/10.15826/recon.2019.5.4.016 162 www.r-economy.ru R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 Online ISSN 2412-0731 the same time, they have large populations, which means that we cannot consider them simply as pe- ripheral areas. It would be more appropriate to re- fer to them as constituent parts of agglomerations, elements of the area of population concentration. These also include those urban districts (Per- vouralsk, Verkhnyaya Pyshma, and Berezovsky) which have closer relations with their neighbours than other municipalities in this group (their val- ues of the local Moran’s I are the highest). The LH group includes municipalities with comparatively low values in the given indicator, located in proximity with densely populated ter- ritories and thus inevitably influenced by these neighbours. Eight of these municipalities (placed at the top of the corresponding part of the table) are more closely connected with the neighbouring municipalities (these municipalities are the most influential ones) than with others. All other municipalities in the region (not included in any of the groups) have a positive autocorrelation (which means a certain similar- ity to their neighbours) and have relatively low values of the population size. These are included into the LL group: they are neither influenced by their neighbours nor influence their neighbours themselves. The map in Figure 3 illustrates these results. The four groups of municipalities are highlighted by different colours and the saturation of the co- lour depends on how closely these municipalities interact with their neighbours: red, dark green or dark yellow are used for municipalities with the highest values of the local Moran’s I in their re- spective groups. The territories with strongest interdependence are connected by lines (we used formula 5 to assess the strength of influence be- tween the two possible pairs of territories). Territories signi�cantly in�uenced by the leaders Territories insigni�cantly in�uenced by the leaders Territories unin�uenced by the leaders Kamensk-Uralsky Krasnoturyinsk Serov Nizhny Tagil Novouralsk Ekaterinburg Centres of the regional settlement system Centres of local settlement systems Densely populated areas with a positive spatial autocorrelation with theirneighbours Densely populated areas with a negative spatial autocorrelation with their neighbours Strongest interterritorial relations Figure 3. Spatial autocorrelation between municipalities in Sverdlovsk region (parameter – ‘resident population size’) http://doi.org/10.15826/recon.2019.5.4.016 R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 163 www.r-economy.ru Online ISSN 2412-0731 These results confirm our previous conclusion that the population distribution across Sverdlovsk region is uneven: the region has three extrema with large populations and these municipalities are sur- rounded by other territories, which are also quite densely populated (areas of population concentra- tion). The centres of the regional settlement system (and the surrounding areas of influence) are located in the south-western part while the rest of the region looks like a ‘desert’, comprising scarcely populated municipalities. Although some researchers expect the city of Serov in the north of the region to evolve into a full-fledged urban agglomeration [26; 27], it is still too early to speak of it as a newly emerged centre in the regional settlement system. Serov and Krasnoturyinsk have much larger populations than the surrounding territories, which turns them into local leaders, although their resources are not suffi- cient for scaling up their activities and for creating an agglomeration effect. Parameters of spatial autocorrelation identi- fied through the analysis of infrastructure local- ization are slightly different from the previously identified strength of correlations between the resident populations of the given municipalities. The value of the global Moran’s I (0.025) ex- ceeds its expected value: we observe a positive spatial autocorrelation, which means that in gen- eral there are no significant disparities between the development of the neighbouring territories. What we see is a gradual change in the given indicators. An undisputed leader in terms of new hous- ing supply is Ekaterinburg. The neighbouring ter- ritories are behind Ekaterinburg but they still tend to perform above the average level in the region. Therefore, Ekaterinburg together with the adja- cent territories (group HH) form an area charac- terized by intensive construction of new housing (see Table 3). As Table 3 illustrates, the strongest correlations between the values of this indica- tor for this area are observed for Ekaterinburg, Berezovsky, Sysert and Verkhnyaya Pyshma. Table 3 Leaders in new housing supply Group Municipality ILi HH Verkhnyaya Pyshma 0.0065 Berezovsky 0.0061 Ekaterinburg 0.0059 Sysert 0.0044 Beloyarsky 0.0010 Pervouralsk 0.0009 Sredneuralsk 0.0006 Kamensk-Uralsky 0.0000 HL Nizhny Tagil –0.0001 In the HL group (territories whose rates of new housing supply are considerably higher than in the neighbouring municipalities), only one municipality – Nizhny Tagil – can be considered to be a local leader, able to compete (though not very successfully) with Ekaterinburg and its sur- roundings. The majority of municipalities in these groups are characterized by a negative autocorrelation (group LH) since their performance in this indi- cator is not very high while their proximity to the top municipalities means that they are influenced by these leaders. The LL group again includes those municipalities which account for over a half of the region’s total area, primarily, its northern and eastern parts (Figure 4). If we compare the results shown in Figure 3 and Figure 4, we shall see that in general munici- palities with positive spatial autocorrelation in the two given parameters demonstrate the following trend: areas with a high concentration of popula- tion and objects of infrastructure (including the zones of influence surrounding these objects) are located in the south-western part of the region while its northern and eastern parts are maxi- mally remote (not only geographically but also regarding the specific aspects of territorial devel- opment) from the regional leaders. At the same time our analysis of spatial au- tocorrelation has brought to light a significant difference between the spheres in question. The adjacent municipalities may differ considerably in terms of the population size while the difference between their rates of new housing supply is usu- ally not that substantial, which can be explained by the differences inherent in the nature of the phenomena in question. The population size re- sults from the impact of a whole set of complex socio-economic processes while the data on new housing (characterizing the process as such) cor- relate with the economic parameters of territorial development and are driven by market factors. The demand in the housing market is to a great extent determined by the number of poten- tial buyers – local residents. Nevertheless, housing developers’ decision-making depends even more on the trends in the sphere of land use planning and development. Those who build infrastructure for this or that residential space seek to maximize their profits and occupy new market niches. In doing so, they try to predict in what direction the transformation of the settlement system in this territory will be heading and at the same time http://doi.org/10.15826/recon.2019.5.4.016 164 www.r-economy.ru R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 Online ISSN 2412-0731 adjust this transformation to their needs. Trans- formations of agglomerations mostly involve the development of the territories surrounding the centre, which means that new living spaces tend to emerge within the boundaries of these territo- ries rather than beyond them. In their turn, the territories which do not play a significant role in the settlement system and hold little potential in this respect continue to rank low in the regional system of living spaces. Analysis of the data on places of attraction (for example, stadiums with terraces) built in the region shows a negative spatial autocorrelation (there are differences in the given indicators be- tween the adjacent territories): the global Moran’s I is 0.058. We have thus arrived at some interest- ing results (see Figure 5). First, the distribution of the given infrastruc- ture objects across the region cannot be called even, although fewer municipalities are uninflu- enced by the regional leaders (in comparison with the distribution of population and new housing considered above). Second, the number of centres (mostly local) where stadiums are built (HL group) is quite large (23). The factor that influenced the results of this study is that the number of stadiums in the region (or equivalents thereof ) is insignificant. Centres of the local system of distribution of infrastructure objects Areas of infrastructure concentration with a positive spatial autocorrelation with their neighbours Areas of infrastructure concentration with a negative spatial autocorrelation with their neighbours Territories signi�cantly in�uenced by the leaders Territories insigni�cantly in�uenced by the leaders Territories unin�uenced by the leaders Territories with no open statistical data on the amount of new housing supply Nizhny Tagil Ekaterinburg Kamensk-Uralsky Figure 4. Spatial autocorrelation between municipalities in Sverdlovsk region (parameter – ‘new housing supply’) http://doi.org/10.15826/recon.2019.5.4.016 R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 165 www.r-economy.ru Online ISSN 2412-0731 Third, the proximity of certain municipalities without stadiums of their own to the areas with stadiums enabled them to join the zone of influ- ence created by the leaders (that is, municipalities which have at least one stadium), which means that inhabitants of the former can enjoy access to the infrastructure of the latter. If we look at the maps in Figure 3 and Fig- ure 5, we can notice that, despite the perceived differences in the distribution of infrastructure across municipalities (location of stadiums), there are certain correlations in terms of infrastructure concentration (concentration areas are located in the south-western part of the region), location of hubs in urban districts, such as Nizhny Tagil and Kamensk-Uralsky, and the settlement system. The difference between the spatial character- istics of the infrastructure in residential areas and places of attraction (Figure 4 and Figure 5) is even more significant. This can be explained by the fact that it is usually the local authorities who initi- ate the building of such objects as stadiums and their further development, because these projects are not considered profitable by local business- es (except for large stadiums in big cities) and, therefore, do not attract much private investment. Thus, the distribution of such objects in space is determined not so much by the economic factors Centres of the local systems of distribution of infrastructure objects Areas of infrastructure concentration with a positive spatial autocorrelation with their neighbours Areas of infrastructure concentration with a negative spatial autocorrelation with their neighbours Territories signi�cantly in�uenced by the leaders Territories insigni�cantly in�uenced by the leaders Territories unin�uenced by the leaders Territories with no open statistical data on the number of stadiums with terraces Centres of the regional system of distribution of infrastructure objects Nizhny Tagil Ekaterinburg Kamensk-Uralsky – 7 – 5 – 3 – 2 – 1 Number of sports facilities (stadiums with terraces), units Figure 5. Spatial autocorrelation between municipalities in Sverdlovsk region (parameter – ‘number of stadiums with terraces’) http://doi.org/10.15826/recon.2019.5.4.016 166 www.r-economy.ru R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 Online ISSN 2412-0731 but by social factors such as the standards of in- frastructure provision (determined by the current demographic characteristics of the area), resi- dents’ needs and expectations. Conclusion Scholarly interest in spatial socio-economic systems of different levels and their dynamics as well as the need for efficient regional policy-mak- ing has led to the development of a comprehensive system of analytical methods. These methods are applied for analysis of the localization of objects and its characteristics, spatial aspects of territorial transformations, and problems of spatial develop- ment. Characteristics of regional settlement sys- tems, infrastructure distribution and the relation- ship between them can be studied with the help of spatial autocorrelation analysis combined with map analysis. In our study we revealed a correlation be- tween the patterns of distribution of different social infrastructure elements in Sverdlovsk re- gion and the region’s settlement patterns, which can be explained by the fact that these objects of infrastructure attract their potential users, thus increasing the population concentration in these areas. Distribution and concentration of infra- structure of different types is determined by var- ious factors, and, therefore, the infrastructural systems can meet the needs of local residents to a greater or lesser extent. For example, the spa- tial organization of the regional infrastructure, its emergence and further transformations stem from the need to generate economic effects and, therefore, correlate to a greater extent with the prospective transformations of the settlement sys- tem rather than with its current characteristics. Since it is regional and local authorities who are in charge of building places of attraction, the lo- calization of such infrastructure correlates more with the current settlement system. Formation and transformation of the region’s infrastructural framework can contribute to lev- elling the differences between the territories and thus enhance the shrinkage of space and its de- fragmentation (provided that the key factor of such transformation is the agglomeration pro- cesses and the changes they cause). Territorial infrastructure should be able to respond prompt- ly to the region’s needs in spatial development, which makes monitoring of the qualitative and quantitative characteristics of the infrastructure vitally important. References 1. Thünen, J. (1826). Der isolirte Staat in Beziehung auf Landwirtschaft und Nationalökonomie. Hamburg: Wirtschaft & Finan. 2. Weber, A. (1922). Standort der Industrien. Tubingen. 3. Losch, А. (1954). The Economics of Location. New Haven: Yale University Press. 4.  Christaller, W. (1980). Die zentralen Orte in Süddeutschland. Eineökonomisch-geographische Untersuchungüber die Gesetzmäßigkeit der Verbreitung und Entwicklung der Siedlungen mit städ- tischer Funktionen. Wissenschaftliche Buchgesellschaft, Darmstadt. 5. Launhardt, W. (1882). Die Bestimmung des zweckmässigsten Standortes einer gewerblichen Anlage. Zeitschrift des Vereinesdeutscher Ingenieure, 26, 106–115. 6. Boudeville, J. (1968). L’espace et les Pôles de Croissance. Paris: PUF. 7. Perroux, F. (1954). L’Europe sans Rivages. Grenoble: Presses universitaires de Grenoble. 8. Friedmann, J. (1986). The World city hypothesis. Development and Change, 4, 12–50. 9.  Fujita, M., Krugman, P., &Venables, A. J. (1999). The Spatial Economy: Cities, Regions, and International Trade. Cambridge: The MIT Press. 10. Nekrasov, N. N. (1978). Regional Economy: Theory, Problems, Methods. Moscow: Ekonomi- ka. (In Russ.) 11. Aleksandrov, I. G. (1921). Economic Zoning of Russia. Moscow. (In Russ.) 12.  Probst A.E. (1965). Efficiency of Territorial Organization of Production: Methodological Es- says. Moscow: Mysl. (In Russ.) 13. Tatarkin, А. I., & Animitsa, E. G. (2012). Formation of the paradigm theory of the regional economy. Economy of Region, 3, 11–21. (In Russ.) 14. Moran, P. (1948). The interpretation of statistical maps. Journal of the Royal Statistical Soci- ety, 10(2), 243–251. http://doi.org/10.15826/recon.2019.5.4.016 https://books.google.com/books?id=K-M2AQAAMAAJ R-ECONOMY, 2019, 5(4), 155–167 doi: 10.15826/recon.2019.5.4.016 167 www.r-economy.ru Online ISSN 2412-0731 15.  Geary, R. (1954). The continiguity ratio and statistical mapping. The Incorporated Statisti- cian, 5, 115–145. 16. Cliff, A., & Ord, J. K. (1981). Spatial Processes: Model, and Application. London: Pion. 17. Anselin, L. (1988). Spatial Econometrics: Methods and Models. Dordrecht: Kluwer Academic. 18. Pavlov, Ju. V., & Koroleva E. N. (2014). Spatial interactions: estimation based on global and local Moran indexes. Spatial Economics, 3, 95–110. (In Russ.) 19.  Grigoriev, A. A. (2018). Spatial autocorrelation of educational attainment in the Russian Federation. Psychology. Journal of Higher School of Economics, 15(1), 164–173. (In Russ.) DOI: 10.17323/1813-8918-2018-1-164-173 20.  Demidova, O., & Signorelli, M. (2012). Determinants of youth unemployment in Russian regions. Post-Communist Economies, 24(2), 191–218. 21. Balash, O. S. (2012). Statistical research of the spatial clustering of regions of Russia. News of Tula State University. Economic and Legal Sciences, 2-1, 56–65. (In Russ.) 22.  Inozemcev, E. S., & Kochetygova, O. V. (2018). Spatial panel analysis of fertility and life expectancy in Russia. Izv. Saratov Univ. (N. S.), Ser. Economics. Management. Law, 18 (3), 314–321. (In Russ.) DOI: 10.18500/1994-25402018-18-3-314-321 23. Hubert, L. J., Golledge R. G., & Costanza C. M. (1981). Generalized procedures for evaluating spatial autocorrelation. Geographical Analysis, 13, 224–233. DOI: 10.1111/j.1538-4632.1981.tb00731.x 24.  Geography, Institutions and Regional Economic Performance. (2012). In R. Crescenzi, & M. Percoco (Eds.). Berlin: Springer Science & Business Media. 25. Rusanovskiy, V. A., & Markov, V. A. (2016). The effect of a spatial factor on regional differ- entiation of unemployment in the Russian economy. Studies on Russian Economic Development, 5, 144–157. (In Russ.) 26. Averkieva, K. V., Antonov, E. V., Denisov, E. A., & Faddeev, A. M. (2015). Spatial structure of the urban system of the North of Sverdlovsk oblast. Bulletin of the Russia Academy of Sciences. Geographical series, 4, 24–38. 27. Izhguzina, N. R. (2017). The calculation of synergistic effect of urban agglomerations (exem- plified by Sverdlovsk oblast). Journal of the Ural State University of Economics, 2(70), 75–89. (In Russ.) Information about the authors Julia G. Lavrikova – Doctor of Economics, Associate Professor, Director of the Institute of Eco- nomics, Ural Branch of Russian Academy of Sciences (29 Moskovskaya St., 620014, Ekaterinburg, Russia); e-mail: lavrikova_ug@mail.ru Arina V. Suvorova – Candidate of Economics, Deputy Director for Research of the Institute of Economics, Ural Branch of Russian Academy of Sciences, Associate Professor, Department of Re- gional, Municipal Economy and Management, Ural State University of Economics (29 Moskovskaya St., 620014, Ekaterinburg, Russia); e-mail: av_suvorova_av@mail.ru ARTICLE INFO: received July 2, 2019; accepted September 10, 2019 Информация об авторах Лаврикова Юлия Георгиевна – доктор экономических наук, доцент, директор, Институт экономики Уральского отделения Российской академии наук (620014, Россия, г. Екатеринбург, ул. Московская, 29); e-mail: lavrikova_ug@mail.ru Суворова Арина Валерьевна – кандидат экономических наук, врио зам. директора по научной работе, Институт экономики Уральского отделения Российской академии наук, до- цент, кафедра Региональной, муниципальной экономики и управления, Уральский государ- ственный экономический университет (620014, Россия, г. Екатеринбург, ул. Московская, 29); e-mail: av_suvorova_av@mail.ru ИНФОРМАЦИЯ О СТАТЬЕ: дата поступления 2 июля 2019 г.; дата принятия к печати 10 сентября 2019 г. This work is licensed under a Creative Commons Attribution 4.0 International License Эта работа лицензируется в соответствии с Creative Commons Attribution 4.0 International License http://doi.org/10.15826/recon.2019.5.4.016 http://doi.org/10.17323/1813-8918-2018-1-164-173 http://doi.org/10.18500/1994-25402018-18-3-314-321 http://doi.org/10.1111/j.1538-4632.1981.tb00731.x mailto:av_suvorova_av@mail.ru mailto:av_suvorova_av@mail.ru