Natural population growth and urban management in metropolitan regions: Insights from pre-crisis and post-crisis Athens,Greece

Abstract

Chinese

Between the 1970s and the 1990s, cities in Southern Europe experienced a progressive delocalisation of population, settlements and activities over larger regions. Economic downturns have increasingly influenced more recent waves of metropolitan growth, shaping differentiated patterns of urban change. While some cities evolved towards accelerated population dynamics in central districts responding to re-urbanisation impulses, other agglomerations were intrinsically bounded in a sort of ‘late suburbanisation’, with demographic shrinkage of both inner districts and rural areas, and uneven expansion of suburban population. By providing a comprehensive interpretation of the socioeconomic mechanisms underlying recent urban expansion, this study illustrates a diachronic analysis of population dynamics over multiple spatial scales and time frames in a metropolitan region of Southern Europe (Athens, Greece) between 1999 and 2019. Natural population balance was investigated vis à vis selected territorial indicators using descriptive, inferential and multivariate statistics. Results of the analysis identify different social forces underlying suburban population growth during economic expansion (2000s) and recession (2010s), evidencing a distinctive response of local communities to economic downturns that depends mostly on the background context (affluent versus disadvantaged neighbourhoods). Given the multiplicity of territorial dimensions involved in urban growth, our findings highlight how economic downturns distinctively shape metropolitan development based on locally differentiated demographic dynamics.

Keywords

multivariate statistics natural population balance Southern Europe suburbanisation urban governance

Introduction

Sequential waves of urbanisation, suburbanisation and re-urbanisation were a basic feature of metropolitan systems worldwide (Arribas-Bel et al., 2011; Carbonaro et al., 2018; Fernández Maldonado et al., 2014). Assessing the spatial evolution of population growth rates over recent stages of metropolitan expansion contributes to a refined understanding of urban change (Angel et al., 2011), allowing a more precise estimation of spatial direction and intensity of local development processes over broader regions in the near future (Buzar et al., 2007; De Rosa and Salvati, 2016; Dijkstra et al., 2015). This evidence will provide a solid knowledge of the latent drivers of metropolitan transition (Seto et al., 2011), informing appropriate and timely strategies of urban management adapted to varying socioeconomic contexts (Ciommi et al., 2018; Couch et al., 2007; Garcia-López, 2010). Multiple approaches have been proposed to evaluate the contribution of different components of population growth in total urban expansion (Ciommi et al., 2019; Kroll and Kabisch, 2012; Turok and Mykhnenko, 2007). While migration balance was sometimes considered the main engine of urban growth in advanced economies – especially when the local effects of the first demographic transition slowed down – this component was demonstrated to be, in most cases, heterogeneous over time (Bocquier and Bree, 2018; Lerch, 2013, 2019). Even more uncertain outcomes were observed at the spatial level of neighbourhoods (or urban districts), making the analysis of urban change a very hard task and short-term predictions of city growth quite imprecise (Salvati and Serra, 2016).

Conversely, urban scenarios were increasingly based on information derived from natural population balance, intended as a more stable and local-based attribute of urban and rural communities (e.g. Salvati et al., 2016). By considering ‘stabilised’ rates of natural population growth (e.g. averaging annual rates of births and deaths over longer time frames) was a consensus strategy for analysis of the latent expansion of metropolises because of different reasons: (i) unlike migration rates, natural balance reflects the intrinsic population age structure – considered a basic attribute of every local community – giving direct information on demographic dynamics at the same time (Kabisch and Haase, 2011; Serra et al., 2014; Zambon et al., 2018); (ii) especially in the economic systems more affected by the recent crisis, and even more intensively by the COVID-19 pandemic – the contribution of natural balance to total population growth was demonstrated to increase systematically, being predicted to be the most relevant engine of metropolitan expansion, for example, in some peripheral contexts of Europe (Carlucci et al., 2017; Lerch, 2013; Morelli et al., 2014); (iii) although responding less rapidly than migration balance to external shocks, natural balance is a more stable property of local demographic systems, since it considers together the impact of temporary fertility recovery (or decline) and progressive ageing, whose changes are reflective of socioeconomic transitions directly affecting the overall path of metropolitan growth (Duvernoy et al., 2018; Kroll and Kabisch, 2012; Nijkamp and Kourtit, 2013); and, finally, (iv) natural balance can be estimated more precisely than migration balance on a very detailed time frame (e.g. years) and spatial scales (e.g. municipalities/neighbourhoods/local districts) using widely accessible (and mostly free) statistical sources, revealing latent (and subtle) urban changes likely more rapidly than any other socioeconomic indicator (Buzar et al., 2007; Hatz, 2009; Pili et al., 2017).

For these motivations, a specific analysis of the endogenous growth of metropolitan populations, especially in recent times, allows a better comprehension of demographic dynamics underlying distinctive stages of the city life cycle, with specific regards to suburbanisation and re-urbanisation impulses (Nijkamp and Kourtit, 2013; Nüssli and Schmid, 2016; Salvati, 2016). Mediterranean cities are frequently regarded as the most advanced districts in Southern Europe (Zambon et al., 2017). A complex interplay of demographic forces, influencing metropolitan growth at varying spatial scales – from local to regional – has determined social stratification and economic development, especially in the largest agglomerations of the area (Arapoglou and Sayas, 2009; Kandylis et al., 2012; Rontos et al., 2016). As in other European regions, demographic transitions in Mediterranean Europe contributed to urban change and metropolitan transformations at large, promoting a progressive shift from compact settlements and radio-centric growth to a spatially discontinuous and low-density urban fabric (e.g. Salvati et al., 2018). In these contexts, the contribution of vital dynamics and migration to urban growth was demonstrated to be heterogeneous over both time and space, being intrinsically oriented along multiple geographical gradients especially – but not exclusively – associated with elevation, accessibility or proximity to the sea coast (Arribas-Bel et al., 2011; Cuadrado-Ciuraneta et al., 2017; Garcia-López, 2010). However, immigration rates declined rapidly with the 2007 recession and emigration at working age – especially younger people, university students and high-tech professionals – rose significantly, minimising the contribution of migration in total population growth (Kandylis et al., 2012; Panori et al., 2019; Rontos et al., 2016). The COVID-19 pandemic accelerated this process, further reducing internal migrations (a typical trait of Mediterranean demographic systems since the early 1950s) and in turn enhancing counter-urbanisation impulses, already observed – at least in nuce– with the earlier recession stage (Gkartzios, 2013).

Based on these premises, our study moves in the (post-crisis) debate on urban cycles and population dynamics in expanding metropolitan regions (e.g. Scott and Storper, 2015), integrating a comparative analysis of (endogenous) population growth with a socioeconomic vision of metropolitan development. Assuming demographic dynamics as spatially asymmetric – since they are (directly or indirectly) influenced by the dominant socioeconomic structure – we verify if natural population growth in Athens (Attica, Greece) over the last two decades was determinant in the consolidation of a late suburbanisation stage or in a more latent recovery of central locations – following re-urbanisation impulses. Based on exploratory analysis of official statistics, the study period encompassed an intense economic expansion followed by a drastic crisis. In these regards, the relationship between natural population balance and a large set of territorial variables delineating selected characteristics of local contexts has been investigated to test if the socioeconomic structure of the Athens’ metropolitan region – contrasting with the stereotypical, radio-centric and compact Mediterranean cities – definitely shifted towards a fully ‘suburbanised’ model. We assumed the Athens’ metropolitan region as a paradigmatic example of demographic dynamics typical of many urban regions in Southern Europe. To identify scale-dependent demographic processes, the analysis was run at different spatio-temporal levels, from regional to local spatial scales, and from medium-term to short-term time intervals, using administrative analysis’ units and indicators derived from official statistics. The implications of this study for urban management are finally discussed, envisaging metropolitan scenarios and policy guidelines facing with a more cohesive city growth that can be extended to other contexts with similar patterns of change in both advanced countries and emerging economies.

Methodology

Study area

The area investigated in this article encompasses the majority of the administrative region of Attica in Central Greece (corresponding with the NUTS-2 (Nomenclature of Territorial Statistical Units) level of the European classification of territorial statistical units) and coincides with the Athens’ metropolitan region. Extending more than 3000 km², the areal boundaries reflect the operational definition of the European Urban Atlas, an international initiative covering more than 700 cities in the continent (https://land.copernicus.eu/local/urban-atlas/urban-atlas-2018), and is administered by seven regional units (Central, Western, Northern and Southern Athens, Piraeus, Western and Eastern Attica) and 115 municipal authorities enforced with the ‘Kallikratis’ reform of the local governance structure in Greece (Rontos et al., 2016). The area (hosting nearly 3.8 million inhabitants in 2011) alternates coastal and inland plains – the largest being the ‘Lekanopedio Attikis’ corresponding with the Greater Athens’ area, the core agglomeration constituted of 56 municipalities hosting more than 3 million inhabitants in 2011 – and mountain ranges frequently exceeding 1000 m of elevation. Compact urban expansion occurred mainly in the first decades after World War II (late-1940s to late-1970s). Suburbanisation processes with expansion of low-density residential settlements have systematically observed since the early 1980s (Morelli et al., 2014; Pili et al., 2017; Salvati, 2016). A comprehensive outlook of recent processes of urban growth in the study area has been provided by Arapoglou and Sayas (2009), Chorianopoulos et al. (2010), Gospodini (2009), and Salvati and Serra (2016), among others.

Elementary data, indicators and spatial analysis’ units

A comparative investigation of recent population dynamics in Athens and the latent relationship with the local context was carried out using demographic indicators and background socioeconomic variables derived from multiple statistical sources. More specifically, we considered the vital dynamics of births and deaths per year and municipality collected and disseminated by ELSTAT (National Statistical Authority of Greece, Piraeus) at the same spatial and temporal scale. Based on these estimates, we calculated the natural population growth over two time scales assumed to depict different demographic responses to economic downturns: (i) three time windows of seven years each (1999–2005, 2006–2012, 2013–2019) and (ii) seven time windows of three years each (1999–2001, 2002–2004, 2005–2007, 2008–2010, 2011–2013, 2014–2016, 2017–2019). By adopting the first time schedule, it was possible to estimate (direct and indirect) impacts of an intense economic expansion culminated with the 2004 Olympics in Athens, together with more latent (and delayed) effects of the subsequent recession covering (more or less homogeneously) the large part of the 2010s (Salvati, 2016). The second time schedule was considered with the aim of verifying the internal stability of the demographic indicators calculated following the first time schedule, and refining the empirical results of such analysis with more precise evidence at a reduced temporal lag. Natural population growth rates at each municipality of the study area were calculated subtracting 1 to the absolute ratio between the cumulated number of births and deaths over the desired time frame (three or seven years long, see below). By reflecting the local outcomes of natural balance, this index assumes positive values when births exceed deaths (population growth) and negative values when deaths exceed births (population decline). Indexes calculated over a sufficiently long time interval provide a more reliable, stable and statistically robust estimation of demographic rates at the local scale, as demonstrated in similar studies (e.g. Muniz, 2009).

Natural balance was also aggregated over few spatial partitions distinguishing (i–ii) central districts (downtown Athens and Piraeus, corresponding with the same municipalities) and (iii) the rest of the Greater Athens’ area (54 municipalities) from the suburban districts of (iv) Messoghia (located at the Eastern fringe of Athens and encompassing 29 municipalities), and (v) Thriasio (located at the Western fringe of Athens, encompassing a total of 15 municipalities), as well as the rural district of Northern Attica (encompassing 15 municipalities that include the main sub-centre of Oropos). This spatial aggregation provided an appropriate classification in urban, suburban and rural districts aimed at verifying the specific assumptions of our study.

A total of 65 background variables were made available from official statistics (mainly derived from ELSTAT) at the municipal scale with the aim at delineating a socioeconomic profile of local communities in the study area (Table 1). Variables were sub-divided in six research dimensions: topography and accessibility (8 variables), territorial features including natural amenities (7 variables), land-use (19 variables), settlement morphology (9 variables), productive structure (12 variables) and social conditions including wealth and a summary evaluation of local job markets (10 variables). This dataset provides a comprehensive description of municipalities, taken as representative of urban, suburban and rural communities in the study area (Morelli et al., 2014). The variables adopted in our study were already described in earlier studies on Athens’ development (Pili et al., 2017; Rontos et al., 2016; Salvati and Serra, 2016) and used extensively in previous studies on urban geography (Ciommi et al., 2019; Di Feliciantonio and Salvati, 2015; Di Feliciantonio et al., 2018).

Table 1.

Results of non-parametric Spearman rank correlation analysis between natural growth rate and contextual variables in the Athens’ metropolitan region by time interval (only significant coefficients at p < 0.05 after Bonferroni’s correction for multiple comparisons were shown*).

Variable	Seven-year interval			Three-year interval
	1999–2005	2006–2012	2013–2019	1999–2001	2002–2004	2005–2007	2008–2010	2011–2013	2014–2016	2017–2019
Elevation		0.25	0.24		0.23		0.31		0.26
Distance from Maroussi	−0.36	−0.32	−0.32	−0.39	−0.28	−0.28	−0.23	−0.36	−0.32	−0.31
Distance from Oropos	−0.27	−0.23			−0.28	−0.24	−0.26	−0.24	−0.23
Self-contained growth	−0.22	−0.28	−0.30	−0.23		−0.30	−0.26	−0.32	−0.31	−0.24
Residential building					0.26	0.23	0.24	0.23
Multiple usage buildings		−0.25	−0.23			−0.24	−0.28	−0.25	−0.25
Town master plan				0.23
Per-capita urban area		0.29	0.29			0.24	0.36	0.24	0.29	0.24
Discontinuous dense urban fabric				0.28						0.25
Discontinuous medium density settlements	0.27	0.34	0.26	0.27	0.23	0.31	0.31	0.36	0.28
Discontinuous low density settlements		0.28				0.26	0.29	0.30
Transport infrastructure	0.25	0.29	0.45	0.23	0.23	0.26	0.31	0.32	0.41	0.43
Land without current use			0.24						0.24	0.25
Mining activities								0.24		0.25
Hospitality activities	−0.33	−0.32	−0.29	−0.37	−0.32	−0.29	−0.30	−0.31	−0.29	−0.29
Financial activities								0.25	0.24
Technology/advanced services	0.25	0.25	0.25	0.29				0.31	0.25
Per-capita income	0.33	0.33	0.33	0.36	0.23	0.28	0.25	0.39	0.34	0.29
Income growth during recession	0.28	0.33			0.33	0.27	0.34	0.32	0.24
Participation rate	0.31	0.30	0.25	0.36		0.27	0.23	0.34	0.23	0.23
Living/working in the same municipality				−0.27

Proximity to the sea coast, distance from downtown Athens, distance from downtown Piraeus, distance from Markopoulo Messoghias, distance from Aspropyrgos (Thriasio), soil quality index, climate quality index, population density, per cent share of population residing in sparse settlements in total population, protected areas, inhabitants per building, average building height (vertical profile of settlements, diversity in urban land-use, industrial buildings, hotel-use buildings, service-commercial buildings, compact and continuous residential urban fabric, discontinuous very low (<10%) residential urban fabric, isolated structures, industrial/commercial areas, other roads and associated land, railways, port areas, airports, construction sites, sites, construction sites, green urban areas, sport and leisure facilities, cropland, forests, manufacturing, energy activities, constructions, publishing businesses, commerce, transports, telecommunications, industry-to-service businesses ratio, income growth rate during economic expansion (2001–2008), percentage of workers in Class 1 of the European SocioEconomic Classification (ESEC 1) in total workers, percentage of population living in the same municipality within the last five years in total population, percentage of non-European foreign-born residents in total population, percentage of native Greeks in total population, percentage of European citizens in total population.

Statistical analysis

We adopted a multi-step exploratory approach considering together descriptive statistics, mapping, inferential statistics based on non-parametric pair-wise correlations, multivariate statistics (Principal Component Analysis) and econometric models (step-wise multiple regressions). Integration of different analytical tools was considered appropriate to delineate (apparent and latent) relationships among data structures intrinsically complex over time and space. This objective was pertinent to a refined analysis of (apparent and latent) mechanisms of long-term metropolitan growth based on a reliable demographic indicator (natural population balance) and the socioeconomic context underlying such changes. Descriptive and spatial analysis were used to define the general trends characteristic of natural population growth in the spatial partitions adopted here.

To identify (linear and non-linear) relationships between population dynamics and local socioeconomic contexts, a non-parametric Spearman rank analysis was run on the whole sample of 115 municipalities testing pair-wise correlations between natural population balance and each of the 65 background variables described above. The statistical significance of each correlation coefficient was tested at p < 0.05 applying the Bonferroni’s correction for multiple comparisons. Spearman correlation analysis allowed a first selection of variables with a supposedly important relationship with the natural population balance. The variables selected in this first step constituted the base of the subsequent analysis, which was functionally divided in two operational steps as a function of the time scale adopted: (i) natural population balance calculated on individual time frames of seven years was compared with the socioeconomic context characteristic of each municipality using a Principal Component Analysis; (ii) the same demographic rate, calculated on individual time frames of three years, was the dependent variable of step-wise multiple regressions modelling the spatial variability of natural population growth with the background local context (predictors).

More specifically, a Principal Component Analysis (PCA) was run on a data matrix including three demographic rates (seven year time frame) at each of the 115 municipalities of the study area and the background variables found as significant in the exploratory Spearman correlation analysis. PCA was aimed at decomposing different dimensions of metropolitan expansion in Athens along few independent components explaining the largest part of the variability in the input data matrix. By selecting principal components with eigenvalue >1, the analysis provided an indirect assessment of urbanisation patterns and processes and the underlying socioeconomic forces (De Rosa and Salvati, 2016). Correlation of background variables to each extracted component was investigated using loadings (reflecting the pair-wise correlation rate between variables and components). Municipalities with characteristic population dynamics and socioeconomic profiles were discriminated inspecting (and mapping) component scores.

Multiple linear regression models were finally run separately for each three-year time frame adopting a forward stepwise strategy for selecting significant predictors (starting from the same PCA inputs, see above) of natural population growth in each municipality, considered as the dependent variable. Predictors were included in each model with a value of the associated Fisher-Snedecor F test (p-level) below 0.01. All variables were standardised prior to analysis. Empirical results include slope coefficient estimates and the associated Student t statistic testing for the null-hypothesis of non-significant regression coefficient at p < 0.05. The goodness-of-fit of each regression model was measured using adjusted R² and tested for significance (against the null hypothesis of a non-significant model) through a Fisher-Snedecor F test with p < 0.001.

Results

Population dynamics in Athens

Considering the entire study period, which includes economic expansion (1999–2005), a rapid economic decline (2006–2012) and a subsequent period of stagnation (2013–2019) due to the effects of the global recession (Figure 1), a differential trend in the natural population balance was observed in urban areas (downtown Athens, downtown Piraeus, the rest of the Greater Athens’ area) and fringe districts (suburban Western and Eastern Attica, rural Attica). Fringe districts were characterised by an intense natural population increase reaching the highest value (2009) in Western Attica (Thriasio) and Eastern Attica (Messoghia). In both districts, despite a drastic demographic decline corresponding to the 2010s recession, natural population growth was always positive. While maintaining a fairly similar trend over time, urban and rural areas displayed less intense demographic dynamics, with the highest positive natural population balance observed in 2008 or 2009. Downtown Athens and Piraeus were districts with the lowest natural population growth in the study area, showing systematically negative values throughout the period. The rural area, while displaying moderate demographic performances with economic expansion, faced a drastic reduction in the natural population growth rate since 2012, stabilising on systematically negative values afterwards. Greater Athens (with the exclusion of downtown Athens and Piraeus) showed a comparable trend with what was observed for suburban locations, although with a less marked growth peak at the end of the 2000s and a more marked decrease with recession. Out of 115 municipalities, the number of municipalities with a systematically negative natural population growth rate increased sharply in the three study intervals (1999–2005: 27 municipalities; 2006–2012: 33 municipalities; 2013–2019: 67 municipalities). Considering shorter time frames, the increase was even more evident, after an initial decline likely associated with the early 2000s economic expansion, in line with a moderate recovery of fertility levels (1999–2001: 33; 2002–2004: 28; 2005–2007: 25; 2008–2010: 25; 2011–2013: 46; 2014–2016: 60; 2017–2019: 82 municipalities).

Figure 1.

Trends over time in the natural balance of population in selected geographical partitions of the study area.

The spatial distribution of the natural population growth rate was mapped considering three intervals of seven years each (Figure 2). Results of the spatial analysis reveal a trend in the natural population balance that is consistent in the study area. The most dynamic areas include the suburban districts West (Thriasio) and East of Athens (Messoghia). This trend was observed throughout the study period, although the highest values of natural population growth varied largely in the first, second and third time intervals. Natural growth rates above 1 were observed in suburban municipalities mainly between 2006 and 2012, reflecting a temporary increase in fertility and rising life expectancy at older ages typical of economic expansions.

Figure 2.

Spatial distribution of natural growth rate in the Athens’ metropolitan region by time interval: 1999–2005 (left); 2006–2012 (middle); 2013–2019 (right) (dot in the left map indicates downtown Athens; 0.5 value belongs to soft grey class).

Correlation analysis

A pair-wise correlation analysis carried out using non-parametric Spearman coefficients identified the variables associated with natural population growth in Athens. Variables displaying a significant correlation coefficient with the natural rate of population growth at least in one time interval of the study period were shown in Table 1. Correlation analysis provided homogeneous results for time intervals of seven and three years, allowing identification of the socioeconomic variables more closely associated with the natural rate of population growth in Athens. The specific impact of these variables on the natural population balance was studied through multivariate statistical techniques, the results of which are presented in the following sections. Out of a total of 65 contextual variables tested in Spearman’s analysis, only 21 variables were significantly correlated with natural population balance in one (or more) time windows. For these 21 variables, both intensity and sign of the correlation are comparable for the two time scales considered (three and seven years).

The most intense (and stable over time) positive pair-wise correlation coefficients were observed between the natural population growth rate and (i) the average municipal per capita income, (ii) the market participation rate of work, (iii) the percentage of the surface area destined for transport infrastructures in total landscape, (iv) the percentage of surface area destined to discontinuous medium-density (30%–50%) residential urban fabric in total landscape, as well as (v) the amount of urban settlements (surface area) per head. The most intense and stable negative correlation coefficients over time were observed between natural population balance and the distance from two target locations (Maroussi and Oropos) in the study area, the percent volume of self-contained urban expansion, as well as the percent share of economic activities (hotels and restaurants) in the stock of economic activities at the municipal level.

Although less intense, other variables were positively correlated with natural population balance: municipal average elevation, the percentage of residential buildings in total building stock, the presence of an approved municipal master plan, the percent shares of both discontinuous dense (50%–80%) and discontinuous low-density (10%–30%) urban fabric in total landscape, the percent share of new construction sites in total landscape, the percent share of mining, financial and high-tech activities in total economic activities at the municipal level, as well as the average municipal income growth rate during recession. Significant variables negatively correlated with the natural balance of the population included the percentage of buildings with multiple, mixed use (e.g. residential and commercial, typical of urban contexts) and the percentage of population living and working in the same municipality in total resident population.

Finally, a summary analysis evaluating trends over time in the number of significant pair-wise correlations between the natural population growth rate (dependent variable) and the background variables illustrated in Table 1 indicated that the dependent variable was associated to a smaller number of socioeconomic predictors in the first time interval (10 variables: 1999–2005), increasing substantially in the second time interval that reflects the effects of economic expansion (14 variables: 2006–2012) and decreasing again in the third time interval reflecting the impact of recession on demographic dynamics in Athens (12 variables: 2013–2019). Considering shorter time intervals, the number of significant correlations followed a similar trend over time, with relatively low values at the beginning of the study period (11 variables: 1999–2001 and 9 variables: 2002–2004), increasing substantially with economic expansion and early recession (13, 14, 16 and 15 significant variables respectively for 2005–2007, 2008–2010, 2011–2013 and 2014–2016) and declining again in the last time interval (10 variables: 2017–2019).

Principal Component Analysis

Results of a PCA highlighting the multivariate relationship between the contextual variables selected through Spearman’s non-parametric correlation analysis were illustrated in Table 2. The analysis delineated territorial profiles that underlie a specific (spatial) relationship between natural population growth and the socioeconomic context, considering separately three time intervals of seven years (1999–2005, 2006–2012, 2013–2019). Analysing a total of three dependent variables and 21 contextual variables, three components were extracted which together accounted for about 50% of the overall variability. The first and third components were associated with natural population balance, while the second component was associated exclusively with contextual variables. Component 1 (21.5% of the overall variability) showed a positive relationship between natural population growth and socioeconomic conditions (e.g. per-capita income). Loadings of the selected variables on Component 1 have also highlighted how natural population growth rates increased with the percent share of technologically intensive activities in total economy, the participation rate in the labour market, and the surface area destined for discontinuous urban fabric. Conversely, natural population growth rates decreased with the distance from the central location of Maroussi, with the percentage of accommodation activities (hotels and restaurants) in the business stock at the municipal scale, and with the percent share of population living and working in the same municipality in total population. Based on the spatial distribution of component scores (Figure 3, left), Component 1 identified the municipalities along the suburban fringe of Athens, both West (districts of Thriasio and Megara) and East (districts of Messoghia, Marathona and Lavrio) of the Greek capital. Component 2 (18.2% of the overall variability) definitely outlined the urban-rural gradient in Athens (Figure 3, middle), as the polarisation between discontinuous dense residential urban areas (negative loading) and discontinuous low-density ones (positive loading) clearly demonstrates. Component 3 (10.3% of the overall variability) delineated a more subtle relationship between natural population growth and the local context, with the first variable positively associated with municipalities in disadvantaged socioeconomic conditions (per-capita income below the average) and with land-use dominated by transport infrastructures (Figure 3, right). These municipalities prevail in the Northern fringe of Athens (Aspropyrgos, Magoula, Ano Liosia, Filis, Acharnes), while they are more scattered in Eastern Attica (Messoghia), a suburban district characterised by less disadvantaged economic conditions.

Table 2.

Results of a principal component (PC) analysis run on natural growth rates (seven-year interval) and selected background variables in the Athens’ metropolitan region.

Variable	PC 1	PC 2	PC 3
Natural population growth
1999–2005	0.64		0.53
2006–2012	0.64		0.64
2013–2019	0.59		0.66
Contextual predictors
Elevation		0.43
Distance from Maroussi	−0.73	0.47
Distance from Oropos
Self-contained urban growth
Residential buildings		0.67
Multiple usage buildings		−0.86
Town master plan		−0.54
Per-capita urban area		0.75
Discontinuous dense urban fabric	0.48	−0.50
Discontinuous medium density settlements	0.63
Discontinuous low density settlements	0.44	0.54
Transport infrastructures			0.46
Land without current use
Mining
Hospitality activities	−0.52	0.52
Financial activities
Technology/advanced services	0.71
Per-capita income	0.81		−0.47
Income growth during recession		0.48
Participation rate	0.44	−0.46
Living/working in the same municipality	−0.65
Explained variance (%)	21.5	18.2	10.3

Figure 3.

Classification of municipalities in the Athens’ metropolitan region based on scores (see Table 2) of principal Component 1 (left), 2 (middle) and 3 (right); 0 value belongs to soft grey class.

Step-wise multiple regression

A step-wise multiple regression analysis was finally run to identify the background variables that explain the greater spatial variability of the natural population growth rate (dependent variable) at the municipal level in the study area. Results of the regression analysis (Table 3) for each of the seven time intervals highlight a substantial heterogeneity in the population growth drivers, as emerged from Spearman’s non-parametric correlation analysis. The variables selected by step-wise regressions (from 2 to 4, as a function of time intervals) produced models with highly variable precision. The adjusted R² gradually increased over time, starting from a rather modest value for 1999–2001 (0.15), reaching the highest value in 2011–2013 (0.28) and then decreasing again. In the first period, the natural growth rate decreased with the distance from the central location of Maroussi and with the stock of mixed-use buildings. In the second period, population grew in residential areas with land-use destined for transport infrastructures, decreasing in more rural locations where economic activities were oriented towards hospitality (hotels and restaurants). A similar socioeconomic profile characterises the spatial distribution of the natural population balance between 2005 and 2007 and between 2008 and 2010, highlighting how accelerated demographic dynamics exquisitely characterised the Athens’ fringe, to the detriment of both urban areas and rural districts. A regression model focusing on the distance from the central location of Maroussi was significant for 2014–2016, highlighting the importance of suburban areas with medium-low settlement density and a modest rate of self-contained urban expansion. Finally, the last three years delineated a rather different model of population growth, where (residual) demographic increase was mainly linked to contexts with land-use dominated by transport infrastructures and mining activities. These contexts included municipalities in the northern fringe of Athens, with medium-low economic conditions, in line with the results of Component 3 (see earlier section).

Table 3.

Results of a forward step-wise multiple regression with natural growth rate (three-year interval) as dependent variable and selected contextual variables as predictors in the Athens’ metropolitan region by time interval.

Predictor	1999–2001	2002–2004	2005–2007	2008–2010	2011–2013	2014–2016	2017–2019
Distance from Maroussi	−0.365 (0.089)**					−0.386 (0.088)**
Multiple usage buildings	−0.277 (0.089)*		−0.416 (0.089)**	−0.404 (0.087)**
Hospitality activities		−0.319 (0.087)**	−0.334 (0.089)**	−0.360 (0.088)**	−0.358 (0.086)**
Residential buildings		0.344 (0.086)**
Transport infrastructure		0.206 (0.085)*		0.209 (0.083)*	0.204 (0.081)*		0.337 (0.084)**
Land without current use			0.191 (0.084)*
Per-head urban a.					0.327 (0.086)**	0.382 (0.090)**
Self-contained exp.					−0.234 (0.082)*	−0.198 (0.084)*
Mining							0.286 (0.085)*
Adjusted R²	0.15	0.20	0.21	0.26	0.28	0.24	0.19
F-test	10.9 (2.112)**	10.7 (3.111)**	10.9 (3.111)**	14.4 (3.111)**	12.2 (4.110)**	12.8 (3.111)**	14.1 (2.112)**
Durbin-Watson test	2.18	1.88	2.08	2.00	2.42	1.90	2.42

*0.001 < p < 0.05; **p < 0.001

Discussion

Southern European metropolises underwent intense demographic changes, experiencing population shrinkage as a joint result of fertility decline, mortality at older ages and an unexpected reduction of international immigration with recession. The related mismatch in settlement expansion and population growth – a direct consequence of more heterogeneous patterns and processes of urban development – makes understanding of the mechanisms governing metropolitan development a particularly hard task. This is particularly relevant in the case of Mediterranean cities featuring long-term compact expansion (Cuadrado-Ciuraneta et al., 2017; De Muro et al., 2011; Gospodini, 2009). Going beyond scale and agglomeration economies, Athens’ development encompasses sequential waves of economic expansion and crisis, revealing a complex form-function relationship at the base of recent urbanisation (Morelli et al., 2014). The empirical results of this study outline coherent trends at both regional and local scales, considering longer (seven year) and shorter (three year) time frames. These findings document the appropriateness of a demographically stable rate, such as the natural population balance, to delineate (apparent and latent) mechanisms of metropolitan growth in recent decades. As a result of economic change, heterogeneous socio-demographic dynamics in recent times were associated with new family relationships (Billari and Kohler, 2004), resulting in a decline of total fertility and ageing (Coleman, 2006). In this perspective, demographic dynamics have had an influential effect on the evolution of central cities, suburbs and rural districts re-densifying and, in some ways, diversifying those areas (Bayona-Carrasco and Gil-Alonso, 2012). Understanding fertility and mortality dynamics (i.e. the components of natural balance) allows identification of the socioeconomic factors oriented along density gradients, evidencing the latest trends in urban/suburban growth likely better than other indicators (Bocquier and Bree, 2018; Buzar et al., 2007; Champion, 2001). Although urban–suburban–rural divides in fertility decreased over time in the study area, significant differences among various types of settlements still persist (Pili et al., 2017). In this context, economic downturns led to even more heterogeneous demographic processes over space (e.g. Kulu, 2013; Kulu and Boyle, 2009; Lesthaeghe, 2010). For instance, while contributing significantly to such dynamics, fertility recovery typical of the Mediterranean countries in the 2000s has been little explored on a local scale. At the same time, economic downturns have led to a more latent change in housing preferences, since today many younger people prefer to live in smaller households than in the past (Delladetsima, 2006; Pérez, 2010; Rosti and Chelli, 2009). This reveals a preference for urban lifestyles, with an indirect effect on demography and household composition, and more variable economic prospects (Helbich, 2012; Hoekveld, 2015; Lerch, 2013). These conditions impact local fertility rates consolidating marriage and childbearing postponement (Hatz, 2009; Kulu, 2013; Liu, 2005). Flexible work arrangements with temporary jobs and lower income may exalt such processes (Carlucci et al., 2018; Chelli and Rosti, 2002; Gkartzios, 2013; Rosti and Chelli, 2009). At the same time, the arrival of labour-related foreign immigrants, especially in low-quality neighbourhoods, contributed to re-shape local fertility trends impacting labour markets (Gil-Alonso et al., 2016).

Apart from the intuitive divergence observed between urban and rural municipalities, the existence of an intermediate group of (suburban) municipalities featuring accelerated population dynamics is in line with the mainstream literature evidencing suburbs as the most dynamic places in metropolitan regions thanks to the positive joint impact of higher fertility (‘suburban fertility hypothesis’, for example see sensu Kulu, 2013) and younger age structures, with intrinsically lower death rates. This context is coherent with the ‘suburbanisation’ stage of the city life cycle, rejecting the hypothesis of a new re-urbanisation wave characterising central cities (Champion, 2001; Kabisch and Haase, 2011; Morelli et al., 2014).

In a truly exploratory perspective, results of the regression analysis – considering both the explained and unexplained proportion of variance, together with the selected predictors – open up new perspectives of investigation searching for non-economic (and non-demographic) determinants of population growth (Salvati et al., 2018, 2019). This perspective is typical of non-confirmatory, exploratory approaches referring to the mainstream ecological, geographical and territorial literature (e.g. Bayona-Carrasco and Gil-Alonso, 2012; Chorianopoulos et al., 2010; Pili et al., 2017). Such dynamics should be more effectively interpreted and governed in a holistic perspective, focusing on the intrinsic relationship between location factors, land-use and demographic trends (Lerch, 2019; Serra et al., 2014; Vobetká and Piguet, 2012).

Conclusions

Demographic dynamics indicate – likely better than other socioeconomic variables – how urban spill-over processes, which have shaped Athens’ expansion over the last decades, are producing highly differentiated suburban spaces (Arapoglou and Sayas, 2009). The results of the multivariate analysis in this study have clearly demonstrated such exclusive patterns of growth (Salvati and Serra, 2016), discriminating affluent from disadvantaged (suburban) municipalities with high birth rates in respect with urban and rural districts (Kandylis et al., 2012). The analysis has also displayed a different fertility response to economic downturns, increasing more rapidly in structurally affluent communities during economic expansion and declining less rapidly with recession in disadvantaged municipalities. These mechanisms – that should be more intensively investigated comparing local contexts in Europe – suggest how local fertility levels outline a different ability of suburban communities with distinctive socioeconomic profiles to respond to external shocks.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Luca Salvati

References

Angel

Parent

Civco

, et al. (2011) The dimensions of global urban expansion: Estimates and projections for all countries, 2000–2050. Progress in Planning 75(2): 53–107.

Arapoglou

Sayas

(2009) New facets of urban segregation in southern Europe. European Urban and Regional Studies 16(4): 345–362.

Arribas-Bel

Nijkamp

Scholten

(2011) Multidimensional urban sprawl in Europe: A self-organizing map approach. Computers, Environment and Urban Systems 35(4): 263–275.

Bayona-Carrasco

Gil-Alonso

(2012) Suburbanisation and international immigration: The case of the Barcelona Metropolitan Region (1998–2009). Tijdschrift voor Economische en Sociale Geografie 103(3): 312–329.

Billari

Kohler

H-P

(2004) Patterns of low and lowest-low fertility in Europe. Population Studies 58(2): 161–176.

Bocquier

Bree

(2018) A regional perspective on the economic determinants of urban transition in 19th-century France. Demographic Research 38: 1535–1576.

Buzar

Ogden

Hall

, et al. (2007) Splintering urban populations: Emergent landscapes of reurbanisation in four European cities. Urban Studies 44(4): 651–677.

Carbonaro

Leanza

McCann

, et al. (2018) Demographic decline, population aging, and modern financial approaches to urban policy. International Regional Science Review 41(2): 210–232.

Carlucci

Chelli

Salvati

(2018) Toward a new cycle: Short-term population dynamics, gentrification, and re-urbanization of Milan (Italy). Sustainability 10(9): 3014.

10.

Carlucci

Grigoriadis

Rontos

, et al. (2017) Revisiting a hegemonic concept: Long-term ‘Mediterranean urbanization’ in between city re-polarization and metropolitan decline. Applied Spatial Analysis and Policy 10(3): 347–362.

11.

Champion

(2001) A changing demographic regime and evolving polycentric urban regions: Consequences for the size, composition and distribution of city populations. Urban Studies 38(4): 657–677.

12.

Chelli

Rosti

(2002) Age and gender differences in Italian workers’ mobility. International Journal of Manpower 23(4): 313–325.

13.

Chorianopoulos

Pagonis

Koukoulas

, et al. (2010) Planning, competitiveness and sprawl in the Mediterranean city: The case of Athens. Cities 27(4): 249–259.

14.

Ciommi

Chelli

Carlucci

, et al. (2018) Urban growth and demographic dynamics in southern Europe: Toward a new statistical approach to regional science. Sustainability 10(8): 2765.

15.

Ciommi

Chelli

Salvati

(2019) Integrating parametric and non-parametric multivariate analysis of urban growth and commuting patterns in a European metropolitan area. Quality and Quantity 53(2): 957–979.

16.

Coleman

(2006) Immigration and ethnic change in low-fertility countries: A third demographic transition. Population and Development Review 32(3): 401–446.

17.

Couch

Petschel-held

Leontidou

(2007) Urban Sprawl in Europe: Landscapes, Land-use Change and Policy. London: Blackwell.

18.

Cuadrado-Ciuraneta

Durà-Guimerà

Salvati

(2017) Not only tourism: Unravelling suburbanization, second-home expansion and “rural” sprawl in Catalonia, Spain. Urban Geography 38(1): 66–89.

19.

Delladetsima

(2006) The emerging property development pattern in Greece and its impact on spatial development. European Urban and Regional Studies 13(3): 245–278.

20.

De Muro

Monni

Tridico

(2011) Knowledge-based economy and social exclusion: Shadow and light in the roman socio-economic model. International Journal of Urban and Regional Research 35(6): 1212–1238.

21.

De Rosa

Salvati

(2016) Beyond a ‘side street story’? Naples from spontaneous centrality to entropic polycentricism, towards a ‘crisis city’. Cities 51: 74–83.

22.

Di Feliciantonio

Salvati

(2015) ‘Southern’ alternatives of urban diffusion: Investigating settlement characteristics and socio-economic patterns in three Mediterranean regions. Tijdschrift voor Economische en Sociale Geografie 106(4): 453–470.

23.

Di Feliciantonio

Salvati

Sarantakou

, et al. (2018) Class diversification, economic growth and urban sprawl: Evidences from a pre-crisis European city. Quality & Quantity 52(4): 1501–1522.

24.

Dijkstra

Garcilazo

McCann

(2015) The effects of the global financial crisis on European regions and cities. Journal of Economic Geography 15(5): 935–949.

25.

Duvernoy

Zambon

Sateriano

, et al. (2018) Pictures from the other side of the fringe: Urban growth and peri-urban agriculture in a post-industrial city (Toulouse, France). Journal of Rural Studies 57: 25–35.

26.

Fernández-Maldonado

Romein

Verkoren

Parente Paula Pessoa

(2014) Polycentric structures in Latin American metropolitan areas: Identifying employment sub-centres. Regional Studies 48(12): 1954–1971.

27.

Garcia-López

(2010) Population suburbanization in Barcelona, 1991–2005: Is its spatial structure changing? Journal of Housing Economics 19(2): 119–132.

28.

Gil-Alonso

Bayona-i-Carrasco

Pujadas-i-Rúbies

(2016) From boom to crash: Spanish urban areas in a decade of change (2001–2011). European Urban and Regional Studies 23(2): 198–216.

29.

Gkartzios

(2013) ‘Leaving athens’: Narratives of counterurbanisation in times of crisis. Journal of Rural Studies 32: 158–167.

30.

Gospodini

(2009) Post-industrial trajectories of Mediterranean European cities: The case of post-Olympics Athens. Urban Studies 46(5–6): 1157–1186.

31.

Hatz

(2009) Features and dynamics of socio-spatial differentiation in Vienna and the Vienna metropolitan region. Tijdschrift voor Economische en Sociale Geographie 100(4): 485–501.

32.

Helbich

(2012) Beyond postsuburbia? Multifunctional service agglomeration in Vienna’s urban fringe. Tijdschrift voor Economische en Sociale Geographie 103(1): 39–52.

33.

Hoekveld

(2015) Spatial differentiation of population development in a declining region: The case of Saarland. Geografiska Annaler: Series B, Human Geography 97(1): 47–68.

34.

Kabisch

Haase

(2011) Diversifying European agglomerations: Evidence of urban population trends for the 21st century. Population, Space and Place 17(3): 236–253.

35.

Kandylis

Maloutas

Sayas

(2012) Immigration, inequality and diversity: Socio-ethnic hierarchy and spatial organization in Athens, Greece. European Urban and Regional Studies 19(3): 267–286.

36.

Kroll

Kabisch

(2012) The relation of diverging urban growth processes and demographic change along an urban–rural gradient. Population, Space and Place 18(3): 260–276.

37.

Kulu

(2013) Why do fertility levels vary between urban and rural areas? Regional Studies 47(6): 895–912.

38.

Kulu

Boyle

(2009) High fertility in city suburbs: Compositional or contextual effects? European Journal of Population 25(2): 157–174.

39.

Lerch

(2013) Fertility decline during Albania’s societal crisis and its subsequent consolidation. European Journal of Population 29(2): 195–220.

40.

Lerch

(2019) Regional variations in the rural-urban fertility gradient in the global South. PLoS One 14(7): e0219624.

41.

Lesthaeghe

(2010) The unfolding story of the second demographic transition. Population and Development Review 36(2): 211–251.

42.

Liu

(2005) Fertility trends in China’s more developed urban districts: The case of four cities. Population, Space and Place 11(5): 411–424.

43.

Morelli

Rontos

Salvati

(2014) Between suburbanisation and re-urbanisation: Revisiting the urban life cycle in a Mediterranean compact city. Urban Research & Practice 7(1): 74–88.

44.

Muniz

(2009) Spatial dependence and heterogeneity in ten years of fertility decline in Brazil. Population Review 48(2): 32–65.

45.

Nijkamp

Kourtit

(2013) The “new urban Europe”: Global challenges and local responses in the urban century. European Planning Studies 21(3): 291–315.

46.

Nüssli

Schmid

(2016) Beyond the urban–suburban divide: Urbanization and the production of the urban in Zurich North. International Journal of Urban and Regional Research 40(3): 679–701.

47.

Panori

Psycharis

Ballas

(2019) Spatial segregation and migration in the city of Athens: Investigating the evolution of urban socio-spatial immigrant structures. Population, Space and Place 25(5): e2209.

48.

Pérez

JMG

(2010) The real estate and economic crisis: An opportunity for urban return and rehabilitation policies in Spain. Sustainability 2(6): 1571–1601.

49.

Pili

Grigoriadis

Carlucci

, et al. (2017) Towards sustainable growth? A multi-criteria assessment of (changing) urban forms. Ecological Indicators 76: 71–80.

50.

Rontos

Grigoriadis

Sateriano

, et al. (2016) Lost in protest, found in segregation: Divided cities in the light of the 2015 ‘Oχι’ referendum in Greece. City, Culture and Society 7(3): 139–148

51.

Rosti

Chelli

(2009) Self-employment among Italian female graduates. Education and Training 51(7): 526–540.

52.

Salvati

(2016) The dark side of the crisis: Disparities in per capita income (2000–12) and the urban-rural gradient in Greece. Tijdschrift voor Economische en Sociale Geografie 107(5): 628–641.

53.

Salvati

Ciommi

Serra

, et al. (2019) Exploring the spatial structure of housing prices under economic expansion and stagnation: The role of socio-demographic factors in metropolitan Rome, Italy. Land Use Policy 81: 143–152.

54.

Salvati

Ferrara

Chelli

(2018) Long-term growth and metropolitan spatial structures: An analysis of factors influencing urban patch size under different economic cycles. Geografisk Tidsskrift/Danish Journal of Geography 118(1): 56–71.

55.

Salvati

Sateriano

Grigoriadis

(2016) Crisis and the city: Profiling urban growth under economic expansion and stagnation. Letters in Spatial and Resource Sciences 9(3): 329–342.

56.

Salvati

Serra

(2016) Estimating rapidity of change in complex urban systems: A multidimensional, local-scale approach. Geographical Analysis 48(2): 132–156.

57.

Scott

Storper

(2015) The nature of cities: The scope and limits of urban theory. International Journal of Urban and Regional Research 39(1): 1–15.

58.

Serra

Vera

Tulla

, et al. (2014) Beyond urban–rural dichotomy: Exploring socioeconomic and land-use processes of change in Spain (1991–2011). Applied Geography 55: 71–81.

59.

Seto

Fragkias

Güneralp

, et al. (2011) A meta-analysis of global urban land expansion. PLoS One 6(8): e23777.

60.

Turok

Mykhnenko

(2007) The trajectories of European cities, 1960–2005. Cities 24: 165–182.

61.

Vobetká

Piguet

(2012) Fertility, natural growth, and migration in the Czech Republic: An urban–suburban–rural gradient analysis of long-term trends and recent reversals. Population, Space and Place 18(3): 225–240.

62.

Zambon

Benedetti

Ferrara

, et al. (2018) Soil matters? A multivariate analysis of socioeconomic constraints to urban expansion in Mediterranean Europe. Ecological Economics 146: 173–183.

63.

Zambon

Serra

Sauri

, et al. (2017) Beyond the ‘Mediterranean city’: Socioeconomic disparities and urban sprawl in three Southern European cities. Geografiska Annaler: Series B, Human Geography 99(3): 319–337.