SOCIO-SPATIAL DIFFERENTIATION IN TRANSITION: A PRELIMINARY COMPARATIVE ANALYSIS OF POST-SOVIET SAINT PETERSBURG AND RIGA

Research into the socio-spatial dynamics in Central-Eastern European cities is an important area of contemporary transition studies. Open issues in this domain range from defining a theoretical framework to data availability and methodological approaches. As to the former aspect, recent literature focuses on the hybrid nature of the post-socialist urban space, which underwent transformation in the conditions of globalization and economic liberalization; the earlier model of spatial development changed dramatically as a result. The multi-scalar and comparative approaches may shed new light on the complex patterns of urban socio-spatial differentiation and its post-Soviet dynamics. Growing regional socio-economic imbalances observed in the former socialist states are lending new urgency to this area of research. This study employs a comparative approach to investigate post-1991 socio-spatial transformations in St Petersburg and Riga — the two largest post-Soviet urban centres in the Baltic Sea region. An important result of the research is a methodology for multi-leva el analysis of changes in the urban environment of post-socialist cities. Data from post-1991 national censuses and population registers are used to calculate measures of social well-being in urban districts as well as to identify territorial imbalances. Comparative analysis makes it possible to trace the spatial patterns of post-Soviet differentiation and set out guidelines for further research in the area.


Introduction
It is widely accepted that growing socioeconomic inequality is a major trend in most postsocialist countries of Central and Eastern Europe. It is ob served at all levels -from interregional to intracity ones [1; 2]. Spatial differ ences emphasise the depth of this inequality. On the one hand, there is a growing contrast between the socioeconomic and demographic performance of capital regions and provinces, especially, the periphery, which was most strongly af fected by the economic transformation of the 1990s [3]. On the other hand, the sociospatial dynamics of large cities give rise to various forms of spatial seg mentation at the district level.
Exploring the spatial dynamics of social transformation is an important area of urban studies. Such research contributed enormously to the understanding of what causes variations in the forms of the increasing spatial segmentation of cities. These forms include segregation, gentrification, and sub-and de-urban isation. Most studies in the field are empirical. They focus on various aspects of spatial development in postsocialist cities. What is often overlooked is the multiscalar dimension of changes.A multiscalar perspective is decisive for the formation of a theoreticalconceptual framework for the analysis of urban envi ronment transformations [4; 5]. Recent works have considered postSoviet cit ies as hybrid spaces, the transformation of which is determined by the logic of neoliberalism and market rather than by changes occurring within the inherited spatial development paradigm [5].
Most historiographical works of the past three decades, despite all the dissim ilarities, view the Soviet model as an alternative project of modernity, different from the Western capitalistic one (see, for instance, [6]). That project predeter mined particular architectural and urban planning of infrastructure [7]. Although Soviet urban planning had much in common with the international modernist tendencies in the West, socialist urban planning differed dramatically from cap italist planning in terms of functions, property structure, and the organisation of everyday life [8]. It seems relevant to study the nonlinear results of interactions between the postsocialist market economy and the organisation of urban spac es inherited from the Soviet era [9]. According to many economic geographers and specialists in regional economics, such research requires multilevel analysis that treats space as a more complex dimension than urban territory [10][11][12]. For example, Golubchikov [9] emphasises the fact that most analyses of social transformation in postsocialist cities are restricted to empirical accounts of so cioterritorial change. However, many of these analyses fail to give a full picture of the actual dynamics in the economic and societal sphere.
This study attempts to use available statistical data to examine, in a compara tive perspective, spatial aspects social differentiation in the two largest postSo viet cities of the Baltic region as a starting point for a multiscalar analysis of postsocialist urban transformations.

Data sources and methodology
We compare interconnected indicators calculated using official statistical data to understand the patterns of sociospatial differentiation at a city level in St Petersburg and Riga. Our exploration of residential conditions differenti ation in both cities, which became more pronounced as social differentiation started to grow, aims to give a new perspective on the postsocialist transition period. The study area is limited by the administrative borders of St Petersburg and Riga. Their internal divisions into statistical areas/municipalities are tak en into account.
Riga is divided into fifty-eight statistical units (micro-districts). This divi sion developed in three stages, within which distinctive city zones and their constituent micro-districts emerged. The historical parts of the 'inner' and 'out er' city formed before World War II. Today, they are home to approximately 25% of the capital's residents. City zones developed based on the principles of the market economy. Having escaped the interest of Soviet urban planners, both areas became fields of physical desolation and social degradation. Rapid residential development began in Riga after World War II and continued until the end of the Soviet period in 1991. Flats in large blocks were centrally dis tributed among skilled workers (the middle class of the Soviet era) and the nomenklatura (the upper class in the Soviet social structure). Today, 75% of the residents of Riga still live in buildings constructed in the Soviet time. Although new houses have been built in the city centre after the Soviets, most of them are located in fringe areas. The districts of the Sovietbuilt tower blocks have been little affected by new residential development. As a rule, new housing built after 2000 is the most expensive and thus available only to the most welloff residents of Riga.
St Petersburg has a similar territorial structure. The city is divided into eight een large districts, which comprise 111 municipalities. 1 Similarly to Riga, the centre of St Petersburg consists of districts built primarily before the twentieth century (twenty municipalities). They are home to 11% of the residents of the city. St Petersburg is a monocentric city, a 'nut in a shell'. It did not deeply change until the 1950s when it grew northward and southward [13]. Sovietera residential areas built in the 1950-80s account for most of St Petersburg's area. They are home to three fourths of all the city's residents. This new city, which emerged in the late 1980s, has a greater area and population than the his torical centre. Moreover, the former is a conglomerate of residential zones that are isolated both from each other and from the historical centre [14]. Only a few transport corridors running between industrial premises connect the centre of St Petersburg with its dormitory districts. The suburbs of the city are large modern dacha communities (Repino, Komarovo, Lisiy Nos, Pargolovo, Strelna, etc.) as well as the towns that grew around 18 th century palaces (Pavlovsk, Pushkin, Lo monosov, Petrodvorets, etc.) and 19 th /early 20 th century factories (Sestroretsk, Kolpino, Pontonny, Metallostroy). These territories became administrative con stituents of Leningrad (St Petersburg) in the Soviet period. These once small towns have become sites of largescale residential development [15]. Suburbs account for 16% of the St Petersburg population.
Our approach to the study is based on comparing two groups of indicators of the social differentiation of urban space. The first group comprises the indi cators of the average concentration of residents with a higher socioeconomic status. The second group assesses housing prices as an indicator of the attrac tiveness of a certain residential area. The data for Riga come from the 2011 cen sus. For St Petersburg, we used various information sources: data on the social parameters of the population (in particular, the level of education) are from the 2010 census, whereas information on housing prices comes from 2016-2019 evaluations.
This is a two-stage study. At first, we considered spatial differences in the in dicators of socioeconomic status. Most statistics come from the 2010/2011 na tional censuses and population registers. The price indicators of housing attrac tiveness in St. Petersburg are calculated based on 2016-2019 evaluations. At stage two, we produced a taxonomy of municipalities and microdistricts in St Petersburg and Riga respectively, according to their social affluence. Table 1 Socio-economic indicators used in the study Russian and Latvian censuses do not contain any information on household and individual incomes. The average concentration of welloff residents was thus calculated based on employment/occupational status and education de gree data. To obtain the employment data for Latvia, we used the International Standard Classification of Occupations (ISCO) as a major socioeconomic sta tus indicator. At the same time, managers and skilled professionals comprise the category of the population with the highest social status. There is general con sensus that the ISCO gives a comprehensive picture of social differentiation in rapidly transforming postSocialist countries [16]. Since Russian censuses do not report occupations and areas of employment, we analysed the percentage of entrepreneurs with employees to assess the social status of the residents of St Petersburg. In the case of Riga, we considered all entrepreneurs. The level of education, which is believed to be an indicator of the socioeconomic status (see [17]) was also used in the spatial analysis of social differentiation. Particularly, we took into account the percentages of people with a) a university degree; and b) a doctoral degree. 2 As to the housing price indicator, it became evident when gathering informa tion that there were significant differences in the availability of these data for the two cities. In Riga, just as any other Latvian city, there are no systematised data on land and housing prices in municipalities. Latvian censuses, however, contain detailed and accurate information on the years when each residential building was constructed. In Russia, housing data are difficult to obtain. Therefore, we used different indicators to assess the comfort of the urban environments of St Petersburg and Riga. In the case of St Petersburg, these were price per one sq. m and the average property tax per person.
Alongside the indicators of the socioeconomic differentiation of urban areas, we employed data on intracensus population change in municipalities. We de liberately did not consider ethnic differentiation in the city: unlike multiethnic Riga, St Petersburg is a predominately monoethnic city. Signs of ethnic segrega tion have appeared only recently in St Petersburg in response to a mass migration to the city from CIS countries [18].
To illustrate spatial differences, we use sketch maps showing the municipal divisions of cities. Percent deviation from the city average is given for the indi cator in question.

Results. St Petersburg
At the end of the Soviet period, the population of St Petersburg (then, Len ingrad) exceeded five million people. That happened at the end of 1989. In 1991, just like any other large city in Russia, St Petersburg witnessed a rapid population decline caused by a natural decrease and negative net migration [19]. There are very different evaluations of the rate at depopulation in St Pe tersburg in the last decade of the 20 th century/the first years of the 21 st centu ry. According to statistics, the population of St Petersburg had to be below 4,600 thousand people at the beginning of 2002. The census carried out in Oc tober 2002 reported a population of 100 thousand people above the estimate. The underestimation of migrations has undermined the reliability of population data in Russian regions.
In 2003, the population of the Northern Capital started to grow. At first, the growth was sustained by immigration; in 2012, natural increase became a sig nificant factor (Fig. 1). Of course, a positive difference between the birth and death rates has accounted for a mere 15-20% of population growth in recent years ( Fig. 2).  Since 2000, the population of St Petersburg has increased by 15.6%. 3 The rate of increase, however, varied from district to district.
In the past three decades, the city has gone through significant changes in population distribution. The city experienced the most rapid growth in the second half of the 1960s when largescale residential development started behind the factory belt 4 that surrounded the 20 th century St Petersburg. This resulted in the fragmentation of the city -its division into the centre and peripheral dormitory districts. Each of the latter is connected to the city centre by onetwo transport corridors. Connections between contiguous dormitory districts are often less sta ble than those with the centre are. Over the sixteen years from 2003 to 2018, thirtyfour of St Petersburg's 111 municipalities saw their population increase by 10-30%, eleven by 30-50%, ten by 50-100%. The population of three municipalities (the villages of Pargolovo (Vyborg district), Shushary, Aleksandrovskaya (Pushkin district) more than doubled. 5 In the same period, ten municipal units (MU) in the city saw their population decrease by 10-30% and one (the 'Palace District' in the city centre) by 34%. In fortytwo municipalities, the number of residents has changed within 10% from the last census (2002) (Fig. 3).
There are distinct spatial patterns of the population dynamics in the districts of St Petersburg. Most municipalities in the historical centre are losing popula tion, whereas suburbs are turning into new dormitory districts, and the number of their residents is growing. The lack of residential development lands within the administrative boundaries of the city and the spatial configuration of St Pe tersburg contribute to rapid population growth in the municipalities located in the north (Primorsky and Vyborgsky) and south (Pushkinsky and Krasnoselsky) of St Petersburg. The population is redistributed from the centre of the city to its periphery. All large cities (St Petersburg is no exception) are socially heterogeneous. Although in the Soviet period, inequality was effectively concealed, the resi dential quarters of Leningrad differed dramatically in comfort and thus had dif ferent social compositions. The Stalinist buildings of the 1930-50s with better layouts had a higher consumer value that the 'khrushchyovkas' and 'brezh nevkas' of the 1960-70s. During the 1990s' transition to market relationships, the spatial division of urban societies started to resemble differentiation, pri marily, in terms of income. For most people, the flat they owned was the main and often only financial asset and an indicator of their living standards [20].
The commercial value of residential property is an important, yet not the only, indicator of social differentiation. There are many indicators, which can be divided into several groups measuring the level of income, the level of education, occupation, and behavioural rule compliance.
Unfortunately, the available statistics reporting the standards of living across all 111 municipalities of St Petersburg has a limited number of indicators. Many of these measures register the situation at the time of the census only. The most recent national survey was held in October 2010.
Nevertheless, we will consider the indicators of quantitative differences in the social composition of St Petersburg population at a municipal level and try to identify the least welloff districts.
We selected the following available indicators of social differentiation in the city: 1) the commercial value of a sq. m of residential housing; 2) individual property tax per person in a given municipality; 3) percentage of entrepreneurs employing any number of people; 4) the percentage of people with a university degree; 5) the percentage of people with a doctoral/postdoctoral degree.
Housing prices. The source of data on housing prices in St Petersburg mu nicipalities is the database of CIAN 6 -Russia's largest real estate agency. According to CIAN, at the beginning of 2019, housing prices ranged between 62 to 247 roubles per sq. m across St Petersburg municipalities (Fig. 4).
The most expensive residential properties are in the historical centre of the city -the Central, Admiralteyski, and Perogradsky districts. The most luxuri ous area is the 'Palace District' municipal unit, which lies between the Palace Embankment of the Neva and the beginning of Nevsky Prospect. The average price of one sq. m of residential property in the area is almost 250 thousand roubles (around 3.5 thousand euros). Above 200 thousand roubles per sq. m are the prices for residential property in the Chkalovskoe municipal unit (Pe trogradsky district), which include Krestovsky Island -a favoured openair spot of St Petersburgians. This area is being built over with posh residential properties, which already cost almost as much as those lining the main street of the city -Nevsky Prospect.
The cheapest residential property is in the industrial outskirts of St Petersburg, primarily, its southern districts. In the villages of Pesochny (Kurortny district), Pontonny, and Saperny (Kolpino district both), and the 'town of Krasnoe Selo' (Krasnoselksy district), the price per sq. m did not exceed 70 thousand roubles (below one thousand euros). Fig. 4 demonstrates the centreperiphery distribution of housing prices in St Petersburg with a northward axis of expensive properties.   9 In eleven MUs, most of which are situated in the south of St Petersburg, this proportion does not exceed 1.3%. Overall, there is a correlation between the con centration of entrepreneurs and the amount of property tax paid (a correlation coefficient of 0.544) (Fig. 6). Percentage of people with a university degree. An important indicator of so cial structure is the level of education. A major Russian centre of culture and ed ucation dubbed the 'cultural capital', St Petersburg has a high percentage of peo ple with a university degree. According to the 2010 census, 38% 10 of the city's population aged twenty and older had a university degree. In this respect, Saint Petersburg ranks second in Russia after Moscow (42.4%). 11 The distribution of people with a university degree is very uneven in St Pe tersburg. The highest concentration of this category of St Peterburgians is asso ciated with the central districts. In the Admiralteysky district, their proportion among all residents aged 20 and older is almost 39%; in the Petrogradsky and Central districts, it is above 46%. Moreover, in some municipal units, such as Aptekarsky Island in the Perogradsky district and the Palace District in the Central district, over half of the adult population had a university degree. 12 A high level of education is characteristic of the population of some remote municipal ities: Komarovo (Kurortny district) (51.8% of people with a university degree) and Tyarlevp (Pushkinsky district) (49.4%) 13 (Fig. 7).
The Moskovsky and Puskinsky districts also stand out for their high percentage of people with a university degree (40% on average).
The largest dormitory districts of St Petersburg, Primorsky and Nevsky, as well as the town of Kronstadt and the remote industrial Kolpinsky district have the lowest proportion of people with a university degree. In 2010, this indicator ranged between 26.4 to 29.8%, which is 13-23% below the city average. The lowest level of education is associated with small municipal units of the Kurortny and Kolpinsky districts -the villages of Smolyachkovo, Molodezhnoe, and Pontonny located at the farthest distance from the centre of St Petersburg. The percentage of people with a university degree in those municipalities was one third below the city average.
When analysing spatial differences in the level of education of St Petersburg residents, it is important to keep in mind that the transition from districts (munic ipalities) performing better and worse on this indicator is rather abrupt.. This may be attributed to the incipient spatialization trends of social segregation. 10 Of respondents who answered the question about the education background. 11 According to the 2010 census, the national percentage of people aged 20+ with a university degree was 23.8%. Source: Russian census 2010. Volume 3. Education. The population of Russian regions by age, sex, and level of education. URL: http://www.gks.ru/free_doc/new_ site/perepis2010/croc/perepis_itogi1612.htm (access date: 02.09.2019). 12 Here and below, the percentage of people with a university degree is calculated as a proportion of people aged twenty and older. 13   Nevertheless, the spatial unevenness of the level of education in St Petersburg should have a demographic explanation since the characteristic in question de pends strongly on the age structure of the population. Whereas the average city percentage of people with a university degree among the population aged 20 and older is 38%, in the 25-29 age cohort, this indicator reached 45.7% in 2010. In the 50-54 age group, it was 33.4% and, among the population aged seventy and older, only 24.5%. Thus, the older the age group, the lower the percentage of people with a university degree. For St Peterburgians, as well as residents of other Russian regions, higher education is becoming obligatory, turning from an indicator of one's professional status to that of social status. This category is more numerous in the areas where there are many young (and technically more educated) people.
Percentage of people with a (post-doctoral) degree. Today, when higher edu cation is becoming universal, not to say obligatory, the percentage of people with a university degree cannot any longer be considered a reliable indicator of social differentiation. It is necessary to consider the spatial distribution of both the pop ulation with a university degree and those with a doctoral/postdoctoral degree (Fig. 8). These data, similarly to other information on the level of education, are reported in censuses. 14 According to the most recent census (2010), St Petersburg was home to 56.1 thousand people with a doctoral and 12.9 thousand with a postdoctoral de gree. In the 25+ age group, 15 16 In the municipalities of the Kolpinksy, Krasnoselsky, Nevsky, and Frunzensky districts, the percentage of people with a doctoral/postdoctoral degree is rather low.
A comparison of the above indicators makes it possible to rank St Petersburg municipalities by the level of social affluence. To this end, we will rank munici palities by each indictor in descending order. For our indicators (property tax; the percentage of entrepreneurs employing any number of people; the percentage of people with a university degree; the percentage of people with a doctoral/postdoctoral degree), the rank will be from 1 to 111, and for one indicator (housing prices), it will be from 1 to 99: data on the commercial value of one sq. m of housing is unavailable for twelve St Petersburg municipalities. The obtained ranks summed up, and the average rank value calculated. 17 14 The population of St Petersburg by age and level of education. Russian census 2010 in St Petersburg. A book of statistics. Part 1. Saint Petersburg, 2012. 15 In view of the years of study at secondary school and university in Russia, it is virtual ly impossible for a person under twenty-five to obtain a doctoral degree. The 2010 census did not report in St Petersburg any people under twenty-five with a doctoral degree. 16 The 'town of Pushkin' municipality. 17 For ninety-nine St Petersburg MUs, the sum of ranks is divided by five and for twelve MUs by four. The lowest indicator score (let us call it an average rank by social affluence, ARSW) is associated with the most prosperous municipalities. Vice versa, the highest indicator score is characteristic of areas with lower living standards. Al though, theoretically, the ARSW of St Petersburg municipalities can range be tween 1 and 111, its actual range is narrower -from 2.6 (the 'Palace District' MU in the Central district) to 107.3 (Smolyachkovo MU in the Kurortny district) (Fig. 9). The worst-performing municipalities in terms of social affluence are those in the southern part of the city (Kolpinsky, Nevsky, Krasnoselsky, and Petrodvortsovy districts and Kronstadt). Fig. 9 shows that the incipient social differentiation of St Petersburg has clear spatial localisation: 'poor' and 'rich' municipalities group in different parts of the city.

Results. Riga
At the beginning of 2019, Riga -the capital and the largest city of Latviahad a population of 632.6 thousand people and was home to almost half of the country's urban residents. Since the early 1990s, the population of Riga has been gradually decreasing (Fig. 10). Both natural decline and negative net migration contributed to that process. Immigration became a decisive factor in the 2000s when Latvia acceded to the EU (Fig. 11). In the past three decades, the population of Riga has been rapidly decreas ing: in 1989-2008, it declined by almost onethird (in 1989, Riga was home to 915.1 thousand people). 18 The postSocialist transition brought about dramatic changes in population distribution across different parts of the city. This process accelerated after 2000. Large residential areas built in the Soviet period (dormito ry districts) were also becoming less populous, albeit at a higher rate. Only in the suburbs of Riga where new lowrise residential properties appeared, the popula tion was growing in that period. The above applies to both new detached house districts and luxurious historical suburbs such as Mežaparks and Vecāķi. Out of the districts in the centre of the city, only one, Skanste, saw its population in crease. The growth is explained by the area's advantageous location and large scale residential development (Fig. 12). The average percentage of people with a university degree among the popula tion aged 15+ is 32.1% across the city. The highest indicator score is observed in the Old Town (49.1%) and the lowest in Spilve (4.1%), a fringe area of Riga. The concentration of people with a university degree is above average in the most luxurious microdistricts of the historical cities and some multistorey residential areas of the late Soviet period (the second half of the 1980s), sich as Zolitūde and Purvciems. The percentage of people with a university degree is the lowest in the Sovietera microdistricts built in the 1960-70s (Ķengarags, Daugavgrīva) and some remote areas that have a 'semi-rural' type of housing (Fig. 13). The average percentage of people with a doctoral degree among the popu lation aged twenty-five and older is 8‰; the highest is in the suburban district of Kleisti (32‰). In some less populous microdistricts there are no residents with a doctoral degree. An aboveaverage percentage of people in this category is observed in other affluent districts of Riga, whereas the Soviet-built residential areas, especially remote ones, underperform in this respect (Fig. 14). Among the city's population aged fifteen and older, the average percentage of managers and skilled professionals is 29.7%. Mežaparks, the most luxurious historical district outside the city centre, has the highest indicator score among Riga's districts (50.6%). The percentage of people in the category is the lowest in Spilve (3/9%) (Fig. 15).
Higher values are observed in the city centre, the late Soviet microdistricts, and some of the more affluent suburbs where the population has been increasing after the Soviet period. The average percentage of entrepreneurs among the economically active pop ulation of Riga aged fifteen and older is 4.0%. This proportion is the highest in the micro-district of Mūkupurvs (13.5%), which is part of the 'outer city'. The distribution of entrepreneurs across Riga districts largely coincides with that of managers and skilled professionals. A high percentage of people in the category concentrating in some 'outer city' districts is explained by the patterns of residen tial development in the late Soviet period (Fig. 16).
Despite some differences, all the sketch maps illustrate similar trends: resi dents with higher social status tend to concentrate in the 'inner city', the affluent historical districts of the 'outer city', and the new suburbs that have developed rapidly in the past two decades. Only a few Sovietera multistorey residential areas, the attractiveness of which is explained by their transport accessibility and developed infrastructure, boast a significant concentration of people with high social status. The ranking of Riga micro-districts by the level of social affluence, calculated using the above indicators, shows that the most luxurious districts of Latvia's capital are the quarters of the historical centre (the 'inner city') and some pre-So viet quarters of the 'outer city', such as Mežaparks and Vecāķi. Among socially affluent areas of Riga are some 'dormitory districts' built at the end of the Soviet period and located in the eastern and western parts of the city. The population of these districts has been growing since 2000. Less affluent areas are primarily located in the southern and northern parts of the city. They consist of quarters built in the 1960-70s, which have lowquality housing and poorly developed in frastructure. One of such areas is the microdistrict of Maskavas, which receives zero investment in reconstruction and development. It has buildings of both So viet and pre-Soviet period; its level of social affluence is the lowest among the central districts of Riga (Fig. 17).

Discussion and conclusions
Our study sought to compare spatial trends in demographic development and social segmentation at a municipal level in the postSoviet cities of St Petersburg and Riga. To this end, we analysed available quantitative data reflecting the social status of residents and the attractiveness of housing.
Overall, the models of postSoviet spatial segmentation are very similar for the two cities, though St Petersburg and Riga have different geography, demog raphy, and historical traditions of urban development. In both cases, the obliv ion of the Soviet period was followed by the socioeconomic modernisation of city territories with a high architectural value (preSoviet city centres) and/or luxurious suburbs. There are, however, differences in demand for Sovietera housing. Stalinist buildings are valued in both cities for their high construction quality and bigger apartments. Abundant in St Petersburg, these buildings are quite rare in Riga where the erection of Stalinist houses began only in 1946. Attitudes to the 'economy-class' multi-storey housing built after 1954, which has plenty of maintenance problems, are different in the two cities. In St Pe tersburg, almost none of the dormitory districts built in the 1960-80s are con sidered attractive, whereas the situation in Riga is more complicated. Some of Riga's microdistricts built in the late Soviet era (the second half of the 1980s) have good layouts, spacious flats, and high transport accessibility. All this trans lates into considerable attractiveness to the city's residents [21]. The demand for postSoviet housing is also different in St Petersburg and Riga. The former city is building multistorey residential property, whereas the latter is erecting housing aimed at the middle class. Riga's new housing boasts low density and abundant green space. In St Petersburg, this type of new housing is classified as 'premium'. In Riga, new suburbs beyond the bounds of dormitory areas consist almost exclusively of private houses or lowrise buildings. They have greater social homogeneity than their counterparts in St Petersburg do.
In other words, social differentiation manifests itself mainly in the spatial segregation of residents by income and socioeconomic status. Welloff people tend to concentrate in the historical cores and the most exclusive suburbs. Yet the process of social polarization is more complicated. It involves all areas of the two cities, including Sovietera dormitory districts (Riga) and coastal dacha communities (St Petersburg). This conclusion may be illustrated by the rapid residential development and a high percentage of entrepreneurs in Riga's dis tricts whose level of social affluence is average, as well as by the considera ble increase in the population of St Petersburg's northern districts, whose res idents are far from being welloff. An earlier spatial analysis of the population composition in Riga shows that the socioeconomic polarisation outside the central part of the city was insignificant in 1991 and quite noticeable in 2011 [22]. The emergence of more socially affluent micro-districts can be explained by both the growing fortunes of their residents and the migration of wealthi er residents to those areas. To understand the actual dynamics and features of this process, it is necessary to carry out a more detailed microlevel analysis of the population composition. In this respect, data on private investment and the development of the housing market are as important as the information on the occupational structure of labour resources. Limited attraction of EU fund ing and an overall lack of investment in the reconstruction of Riga's Sovietera buildings prompt middleclass residents to contribute money to the construction of new housing rather than to the renovation of the existing ones. New construc tion projects aimed at more affluent residents lead to the polarization of urban space, segregation at microdistrict level, and changes in everyone's quality of life (see [23]). A comparison of the five indicators of social affluence of St Petersburg munic ipalities suggests that these measures are interconnected. Our analysis pointed to a high degree of correlation between most of the studied indicators. The Pearson coefficient of correlation between housing prices and the percentage of people with a doctoral/postdoctoral degree, calculated for 111 municipalities, is 0.813. That between housing prices and the percentage of entrepreneurs employing any number of people is 0.797. The coefficient of correlation between housing prices and the amount of property tax per person is 0.739 ( Table 2).
The situation is quite similar in Riga. There is a high degree of correlation between the distribution of people with a university degree across the city and the areas of preferred residence of managers and skilled professionals. The Pearson coefficient of correlation between these measures is 0.947. There is a significant correlation between the areas of preferred residences of managers and skilled professions, on the one hand, and entrepreneurs, on the other (0.727); between people with a university degree and entrepreneurs (0.626); between people with a doctoral degree and managers (0.636) ( Table 3). Table 3 The Pearson correlation coefficient for social differentiation indicators in the urban space of Riga

Indicator
Correlation coefficient Our findings point to two conclusions. Firstly, there are similarities between the spatial trends of social segmentation in St Petersburg and Riga at munici pal/microdistrict level. The differences in the demographic structure and histor ical urban settlement patterns do not have a significant effect on spatial differen tiation, which is shaped by similar parameters as well as the dynamics and logic of socioeconomic and spatial development. The two cities have historical simi larities, namely, they are experiencing a strong effect of preSoviet urbanisation, which is absent in most industrial cities that developed in the Soviet period [24]. It would be helpful to carry out a comparative study covering different types of cities in the postSoviet space to investigate their differences in historical devel opment and urban planning. The similarities identified in this analysis suggest that there are transitional processes common to all postSoviet cities.
Moreover, our analysis of social differences demonstrated that, despite being an important research tool, it is not sufficient to give a comprehensive picture of the spatial dimension of transformations in the urban environment. On the one hand, there is a need to compare urban planning strategies and national spatial de velopment documents to understand how much ideology and political priorities affect the observed processes of social differentiation in postSoviet cities. On the other hand, the results obtained lay the groundwork for further research into urban space segmentation and social segregation at an intradistrict level. This way, our work may contribute to the discussion on the prospects of multiscale studies of postSocialist cities.