Inner Peripheries: Beyond the Demographic Drift

David Peónab, Edelmiro López-Iglesiasc, Xosé Martínezd

a Corresponding author. [email protected] b Department of Financial Economics and Accounting, University of A Coruña c Department of Applied Economics, University of de Compostela d Department of Applied Economics II, University of A Coruña (UDC)

Abstract

Demographic ageing and depopulation of rural areas foster an economic decline that marginalize these territories and puts the access to basic services at risk. However, this situation is often a heritage from large migration processes in past decades and the demographic imbalances that left in consequence. Following a previous paper by two of us (Martínez and Peón, 2015), we use a statistical method to remove the demographic drift due to past migration. We apply it to the case of , to identify some inner peripheries that were able to moderate the process of ageing and depopulation dragged from the past. We characterize inter- and intra- county dynamics, and explore possible explanations for a better performance. Results show that most councils in recovery are head of counties of an intermediate size in terms of population. This would be in line with observed results about rurbanization across most of Europe (Eliasson et al., 2015).

Keywords: Rural population drift, growth poles, spread and backwash effects, regional development, Galicia, demography

JEL Classification: J11, R58, R12.

1

1. Introduction

Economic authorities pursue to ensure social and territorial development and cohesion, such that the benefits of economic growth are widely shared. Nevertheless, the literature and statistical data continue to show a significant lag of most rural areas in terms of economic development and social well-being (Spoor, 2013; Akgün et al, 2015), which is reflected in their demographic trends (European Commission, 2013). There is, however, an increasing heterogeneity, with some rural areas performing much better than others and, in some cases, better than urban areas (Bryden and Munro, 2000). This view of the diversity of rural dynamics has replaced the traditional urban-rural dichotomy. Focusing on intra-rural divides is an interesting field of study, since it would help to design public policies by learning from experiences of successful rural areas (Rizzo, 2016).

Here we consider the specific case of Galicia (), though the case study might easily apply to other European regions with a similar background. Galicia represents an example of an aged region in demographic decline, resulting from an unbalanced demographic structure inherited from strong migration processes since the nineteenth century and especially in the period 1950-1975. In these regions, history may represent a heavy burden, particularly for rural areas, when they experienced large migration processes in the past. The consequences of population decline are often self-reinforcing, bringing about more population decline (Elshoff et al., 2014).

Following a previous paper by two of us (Martínez and Peón, 2015), we use a statistical method to remove the demographic drift due to past migration. We apply it to the case of Galicia, to identify some municipalities that were able to moderate the process of ageing and depopulation dragged from the past. Thus, we observe that the generalized negative performance in terms of population growth hide some regions that are actually able to mitigate the drift. These councils, despite they are showing a bleak picture of an ongoing depopulation process inherited from the past, they hide an, at least partial, truly recovery.

The main contribution of this paper, beyond the analysis at county level performed by Martinez and Peón (2015), is to describe the results within the Galician counties, while we explore for some potential explanations for out and underperformance of the different rural areas. Following the literature of regional development, we study the role

2 of growth poles for regional development (Perroux, 1955; Parr, 1999). We characterize inter- and intra- county dynamics, and explore possible explanations for a better performance. The results, for the case of Galicia, show that most municipalities in recovery are head of counties of an intermediate size in terms of population. This would be in line with observed results about rurbanization across most of Europe (Eliasson et al., 2015), a polarized development of the countryside in which large districts are depopulating, while certain areas, mainly around big- and mid-sized cities, exhibit a transformation of rural communities to communities with urban values and lifestyles.

The structure of the article is as follows. Section 2 introduces the case of study: the region of Galicia, a paradigmatic example in Europe of a demographic decline in rural areas in the last decades. In Section 3 we analyse the historical causes behind this process: a demographic drift due to past migration. In Section 4 we use Martinez and Peón (2015)’s approach to remove the drift, and explore the results obtained. Section 5 is devoted to a characterization of the inter- and intra- county dynamics, following the literature of the role of growth poles for regional development. Finally, Section 6 concludes.

2. Recent demographic dynamics in Galicia: a depopulation process in rural councils continues in the last two decades (1991-2011)

Galicia, in the North-West of the Iberian Peninsula, is one of the seventeen Autonomous Communities in Spain, with a population of 2.75 million inhabitants in about thirty thousand squared kilometers (an extension similar to that of Belgium). The lack of good communications to central Europe is probably one of the key reasons behind its historical lag in economic terms. In the last 100 years, especially since the mid- twentieth century, Galicia has lagged Spain in terms of population and GDP growth, consistently losing relative weight. The GDP per capita was by 2014 80% of the EU-28 average in PPS terms, down from 92.3% in 2009 (, 2014).

What makes Galicia an interesting case of study is that it represents a region in demographic decline. The average age of population is four years higher than that of the Spaniards and Europeans, as a result of a more unbalanced demographic structure, with 4 percentage points less of young people and 5 points more of population over 65 years

3 than Europe. The fertility rate is among the lowest in the world (1.07 children per woman), contributing to a strongly negative vegetative balance (-3.03 per thousand).

The overall population stagnation was accompanied by a growing imbalance in its geographical distribution. Since the mid-twentieth century Galicia has experienced a late and abrupt agricultural sector decline, reducing its share in total employment from 70% to less than 5%. This intense sectoral restructuring resulted in a reduction of total employment, leading to rural-urban migration flows within the region (López Iglesias, 1995). Today, almost 70% of the population lives in 15% of the territory, known as the Eixo Atlántico –a line in the West that goes from North to the Portuguese border in the South, and includes the biggest cities of A Coruña and (about 300,000 inhabitants each) as well as the administrative capital, Santiago de Compostela –see Figure 1.

Figure 1. Territorial distribución of the Galician population. Density by municipalities 2011

Source: Own elaboration. Data: INE, Census 2011

Figure 2 shows the population change of each municipality between 1991 to 2011 – what we denote DESPOB.9111. We may see a clear picture of an inner Galicia experiencing a strong depopulation process, with the more acute cases often related to municipalities in the mountain side –to the East and South. The population dynamics continue to favour the concentration of the Galician population in the Eixo Atlántico, with the exceptions being the Northern coastal area, the two inner capitals of province – and - and a few villages of an intermediate size. These would include the

4 most notable exceptions of O Barco and Verin, in the inner , as well as some municipalities in the surroundings of the capital of that province – and O Carballiño, among others).

Figure 2. Population change of Galician municipalities, 1991-2011

Source: Own elaboration. Data: INE, Census 1991, 2011

In this situation, projections estimate a worst case scenario where Galicia might lose more than a million inhabitants, 38% of its current population, by year 2050 (Xunta de Galicia, 2013), what would totally empty rural areas. This performance is difficult to reverse, as it would not obey to continued negative migration flows, but to the negative vegetative balance these rural areas experience due to a demographic drift consequence of past migrations (López Iglesias, 2013). Such drift we intend to analyze in the next Section.

5

3. A ‘demographic drift’: migration during the 1950-1991 period as a conditioning of the recent population dynamics of rural municipalities

Following Martinez and Peón (2015), we set the following hypothesis of study: the depopulation of rural municipalities in recent decades is due to the demographic structure set in early 1990s and this, on its turn, is a consequence of past migration in the period 1950-1991 –a variable we will denote EMIG50-91.

Step 1 – Defining and characterizing EMIG50-91.

From 1950 to 1975, most Galician municipalities experienced a massive emigration. Though this process ends after the 1970s crisis, due to a sudden cut of the number of migrants to Europe, part of that migration does not appear in the official statistics until the 1991 Census –reflected in a significant net migration officially recognized for the 1981-1991 decade (Fernández Leiceaga and López Iglesias, 2000). For such reason, we define the variable EMIG50-91 as the annualized percentage change between 1950 and 1991, proxy of the strong emigration observed in these four decades and, particularly, in the 1950-1975 period. Figure 3 shows the performance of such variable.

Figure 3. Population change from 1950 to 1991, at a municipal level

6

Source: Own elaboration. Data: INE, Census 1950, 1991

We may appreciate a widespread population decline in these decades, where three quarters of the Galician territory saw its human potential dinimished. In the Eastern provinces, all councils lost population with the only exceptions of the capitals, , , and in the North, due to the economic momentum of Burela and the founding of Alúmina-Aluminio, O Carballiño thanks to migrants’ transfers, A Rúa and O Barco because of the slate industry, Verín and Xinzo. In the Atlantic provinces the situation was also negative in the inland regions and in the . The positive performance concentrated in the Eixo Atlántico, mostly in the metropolitan areas of A Coruña, Santiago and Vigo, the Ferrolterra, and the Rias Baixas. Out of this area, Aldrey (2013) points out the positive evolution of Camariñas and Corcubion, As Pontes, Barbanza, Lalín and the Baixo Miño.

If we compare EMIG50-91 to the population density in 1950 we obtain the results provided in Figure 4.1 This relationship shows similarities to the features noted in the analysis at the county level (Martínez e Peón, 2015). Thus, a dual pattern is clearly observed: councils with a population density higher than 200 inhab/km2 were able to increase or barely lose population, while there are only a few exceptions among municipalities below 150 inhab/km2 by 1950 that were able to avoid a population loss.

Figure 4. Population density in 1950 and population change 1950-1991 compared

1 For better interpretation of Figure 4 we do not represent the four cities of higher density in 1950, which all experienced positive population changes of 1.5% to 2.0% from 1950‐1991 except (+0,49%).

7

Source: Own elaboration. Data: INE, Census 1950, 1991

The positive exceptions are cases of new industrialization in the North (As Pontes, Burela, Cervo, Narón), with the counterpoint of negative cases in the same area of Neda, Mugardos and, to a lower extent, Ferrol –which exhibits the lowest population increase of all Galician cities (+0,49%). Martínez and Peón (2015) already pointed out the exceptional negative behavior of , Terra de and (counties with high population density in 1950 that experienced a strong emigration in the 1960s and 1970s). Now, in this analysis at the municipality level, we consider this exception deserves further analysis beyond this paper: most extremely negative cases correspond to municipalities in Ribeiro (Beade, , , ) and Celanova (, , , , Ramirás). Beyond these regularities, there is a large variability in the population dynamics of councils with similar densities. This variability is particularly large in municipalities from 50 to 150 inhab/km2, suggesting different paces in the depopulation process of rural areas, while the development of urban and coastal areas tended to be more homogeneous.

Step 2 – Historic migration, reflected in the population change 1950-1991 (EMIG50-91) as determinant of the demographic structure of the municipalities in 1991.

We test the hypothesis that the variable EMIG50-91 explains the differences in the demographic structure of the Galician municipalities in 1991, and particularly the heavily skewed structure we find (to different degrees) in rural municipalities. To that

8 purpose we analyze the correlation of EMIG50-91 with the demographic variables mean age (d1), the percentage of population under 20 (d2), the percentage of population over 65 (d3), the elderly dependency ratio (d4), the 85+ over 65+ ratio (d5), the age- dependency ratio (d6), the labour force structure (d7) and the labour force replacement (d8).

Overall, the results confirm the hypothesis: correlations of 75% to 90%, significantly different from zero, and signs of consistent interpretation. The only exception is a correlation of -0.16 with variable d5 (the percentage of population 85 years or more on the group of 65 or more years). It makes sense, since this ratio would explain demographic phenomena before 1950. We may see in Table 1 how the higher correlation appears with age structure (d1 and d3, a correlation of -0.89, and d2 with 0 86). Then, with correlation coefficients higher than 0.80, variables that reflect the level of aging population at the municipal level (d4 and d8), and with coefficients above 0.75 those relative to the structure of active population (d6 and d7).

Table 1 – Correlations between population change 1950-1991, demographic structure in 1991, and population change 1991-2011

Source: Own elaboration. Data: INE, Census 1950, 1991, 2011

Hence, these results support the intuition to use of EMIG50-91 as a synthetic indicator or proxy of the "demographic drift” effect –that is, the inertia of the past population dynamics. However, we need to show whether it is more appropriate, in order to reflect that drift, using EMIG50-91 or any of the indicator(s) above regarding the demographic structure in 1991. We carry out that analysis in the next step.

9

Step 3 – Testing the validity of EMIG50-91 as a proxy of the demographic drift

We may see in Table 1 that the correlation between EMIG50-91 and DESPOB91-11 is strong as well, close to 65%, significant and of a consistent interpretation. However, the relevant issue is not such correlation, but testing whether the goodness of fit of a regression that uses EMIG50-91 as the explanatory variable is similar to that using the variables referred to the demographic structure in 1991. We show next it does, what justifies using EMIG50-91 as a the explanatory variable.

We perform a selection of variables for the regression DESPOB91-11 to EMIG50-91 and the demographic variables d1 to d8. As selection criteria we use BID and Mallows Cp, for regressions with one or two regressors, and taking the explanatory power of the model in consideration. As in Martínez and Peón (2015), DESPOB91-11 = f(EMIG50-91) is the regression that better fits the recent demographic performance, together with the regressions against d2 e d3 (percentage of younger thatn 20 and older than 65) –see Table A.1 in the SM.

The values obtained by the regression DESPOB91-11 = f(EMIG50-91 are almost identical in terms of R2 to the regression with best results –the one that uses d2 as a regressor- but introducing EMIG50-91 and d2 together does not add much. It would perhaps make sense to use EMIG50-91 and d5, since these two variables are not correlated. However, the regression barely gives additional information, d5 appears to be non-significant, and the 2 adj-R worsens. Therefore, we choose EMIG50-91 as main regressor, given the above analysis in terms of correlations and regressions.

4. The population performance beyond the demographic drift; continuity and changes in the population dynamics of rural municipalities in the last two decades

We noticed above, in Table 1, there is a strong correlation between EMIG50-91 and

DESPOB91-11, significant and of consistent interpretation. This result indicates that the dynamics of the population by municipalities in the period 1991-2011 is strongly affected by emigration in the previous decades. We point out two facts. First, data confirms that the population dynamics at the municipal level in recent decades exhibits a strong relationship with the trends in the period 1950-1991. However, the value of the correlation coefficient (0.65) is far from one, which indicates the existence of significant variations in demographic behavior of municipalities with regard to previous trends.

10

Second, such coefficient is clearly lower than the one obtained using data at the county level (0.83). This indicates the existence of significant new developments with regard to past trends when we move down to the municipal level. In addition, though it confirms a quite predictable fact, it deserves to be pointed out: counties hide relevant differences in the behavior of the municipalities within. Although the correlation between migration and demographic performance in the last two decades is evident, it is interesting to identify the extent to which the dynamics exhibit new developments with regard to previous trends. This is what the study of the regression residuals allows to do.

We define RESID91-11 as the residuals of the regression DESPOB91-11 = f (EMIG50-91). They provide the demographic performance of the Galician municipalities once the drift from past migrations is discounted. Figure 5 shows the residuals RESID91-11.

Figure 5 – Residuals of the demographic drift (councils and counties, compared)

Source: Own elaboration. Data: INE, Census 1950, 1991, 2011

We show as well the results by councils in Martinez and Peon (2015) to compare results. Bear in mind for visual comparison of the two maps that the scales are very different. This makes, for instance, that regions like or Verin, with very positive residuals at the county level (dark blue), does not have any counties of that tone.

Figure 5 provides an alternative view of the depopulation process of Galicia, with similar results to those observed at the county level, but also suggesting some differences and nuances when we go down to the municipal level. For instance, we

11 verify that the depopulation process not inherited from the past is now not widespread. The areas with a better performance, clearly improving their relative dynamics in the last two decades, are primarily concentrated in the vicinity of the cities and along the Eixo Atlántico. The effect of the urban area of A Coruña extends to Carballo and Ares stuary, and the urban area of Santiago (mainly the municipalities of Ames, Teo and Oroso). The councils in Baixo Miño and , and others to the Southern urban area of Vigo, and the surroundings of Ourense, with spread effects that reach much of the northwest quarter of the province.

Martínez and Peón (2015) observed a central ring of positive performance. Now, rather than that, we may talk of a central Galicia that remains stable, once the drift from the past is removed. This territory would include the urban areas of Coruña-, Santiago, Lugo and Ourense, and the county capitals of Arzúa, , Monforte, O Carballiño, Lalín e A Estrada, among others. Negative cases in this territory are the councils of Dozón –where the worst residual is observed- Agolada and Rodeiro in the county of Deza, the neighboring municipalities of Antas de Ulla, Paradela and in Lugo, and Cea in Ourense. Finally, the bad performance of Ferrolterra observed at the county level (Ferrol, Eume and ) now is reduced to the councils of Ferrol, Ortigueira, Fene and As Pontes, in contrast with the expansion of other municipalities these counties.

The negative demographic performance (i.e., an accelerated depopulation beyond the drift effect from the past) appears to be linked to an adverse orography. Thus, much of the Galician municipalities with negative residuals are placed in the continuous formed by Xistral, Ancares, Courel, Macizo Central, and Trevinca, which extends along the Eastern third of Galicia. Similar results are observed in the Dorsal Central, where the councils with worst performance form a continuum from Agolada, Rode and Dozón in the region of Deza, to Cerdedo, Forcarei, A Paradanta, and the Western side of Ourense (, Avión). This relationship to the orography is more clearly observed now than in the county analysis. In the same group we may include most municipalities of Celanova, Baixa Limia and , in the border with .

Notwithstanding, the worst behavior at the municipal level, in terms of relative decline compared to previous trends, is not a mountainous area. We may find it in the Costa da Morte and all the Western side of the A Coruña province, from Bergantiños to Barbanza and getting inland to the rural municipalities in the North of Santiago de Compostela.

12

The only exceptions are Carballo and Laracha in the neighbouring of Coruña, and the metropolitan area of Compostela to Negreira.

5. Study in detail by county typology: inter and intra county dynamics

Following the general analysis above provided, a first possible approach to explore more in detail the different demographic performance beyond the drift of the different councils is to examine the dynamics within the different Galician counties. This allows us to distinguish regions with a positive performance that is originated exclusively in the local head (e.g., Valdeorras) from others where the result is just the opposite (e.g., the metropolitan area of Santiago). Overall, this approach that starts from the counties to go down to councils, allows us to “see what happens inside them” and trace differences.

To this purpose, we produce several scatter plots that compares the residuals of each municipality, in the X-axis, to their historic migration, EMIG50-91, listed in Figure 6, what allows us to compare the improvement or worsening of recent demographic dynamics (1991-2011) with the historical migration process (1950-1991). Councils are classified, by columns, in four types of regions depending on whether the county capital had more than 50,000, 20,000, 5,000 or less than 5,000 inahabitants in 1991, respectively –a few corrections were introduced for broad municipalities of a very low density. For each of these types, we established a triple division of municipalities: county capital (first row), municipalities with a population greater than or equal to 5,000 inhabitants (second row), and those with less than 5,000 inhabitants (third row). Only for type 4 regions, the division is limited to only two groups (capitals, and all the other municipalities) since, by definition, none of these municipalities were above 5,000 inhabitants in 1991.

13

Figure 6 – Scatter plots Residuals (X-axis) vs. EMIG50-91 Municipalities of a different size (rows) in each type of county (columns) Type 1 counties Type 2 counties Type 3 counties Type 4 counties

RESID.91‐11 ‐ EMIG.50‐91 (C1‐cabeceiras) RESID.91‐11 ‐ EMIG.50‐91 (C2‐cabeceiras) RESID.91‐11 ‐ EMIG.50‐91 (C3‐cabeceiras)

4.00% 4.00% 4.00%

2.00% 2.00% 2.00%

0.00% 0.00% 0.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00%

‐2.00% ‐2.00% ‐2.00%

‐4.00% ‐4.00% ‐4.00%

RESID.91‐11 ‐ EMIG.50‐91 (C1‐sup.5000) RESID.91‐11 ‐ EMIG.50‐91 (C2‐sup.5000) RESID.91‐11 ‐ EMIG.50‐91 (C3‐sup.5000) RESID.91‐11 ‐ EMIG.50‐91 (C4‐cabeceiras)

4.00% 4.00% 4.00% 4.00%

2.00% 2.00% 2.00% 2.00% Fene

0.00% 0.00% 0.00% 0.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00%

‐2.00% ‐2.00% ‐2.00% ‐2.00%

‐4.00% ‐4.00% ‐4.00% ‐4.00%

RESID.91‐11 ‐ EMIG.50‐91 (C1‐inf.5000) RESID.91‐11 ‐ EMIG.50‐91 (C2‐inf.5000) RESID.91‐11 ‐ EMIG.50‐91 (C3‐inf.5000) RESID.91‐11 ‐ EMIG.50‐91 (C4‐non.cabeceira)

4.00% 4.00% 4.00% 4.00%

2.00% 2.00% 2.00% 2.00%

0.00% 0.00% 0.00% 0.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% ‐4.00% ‐2.00% 0.00% 2.00% 4.00% Pereiro de Aguiar ‐2.00% ‐2.00% ‐2.00% ‐2.00% Nogueira de Ramuín ‐4.00% ‐4.00% ‐4.00% ‐4.00%

14

The scatter plots allow us to appreciate all 315 councils at a glance. Notwithstanding, results are easiest to interpret using Table 2, which summarizes the data in Figure 6 to the average values.

Table 2 – RESID91-11 vs. EMIG50-91 by counties (mean values)

Type 1 counties Type 2 counties Type 3 counties Type 4 counties num RESID.91‐11 EMIG.50‐91 num RESID.91‐11 EMIG.50‐91 num RESID.91‐11 EMIG.50‐91 num RESID.91‐11 EMIG.50‐91

County capital 717‐0,65% 1,25% 0,43% 0,25% 20 0,20% ‐0,39%

Municips > 5.000 inhab. 31 0,84% 0,24% 34 ‐0,18% ‐0,16% 22‐0,69% ‐0,40% 9 ‐0,30% ‐1,32%

Municips < 5.000 inhab. 28 0,87% ‐1,13% 48 ‐0,11% ‐1,36% 72‐0,24% ‐1,40% 27 ‐0,39% ‐2,00%

We must be aware that any conclusion based on an analysis that uses the average values of the different clusters should be treated with caution. In the table we appreciate how the population growth in the past (1950 to 1991) was concentrated in the cities and, to a lesser extent, in the municipalities of more than 5,000 inhabitants of the type 1 and type

2 counties (see values EMIG50-91).

In regards to current performance beyond the drift, we obtain some conclusions. Firstly focusing on county capitals, a slow down is observed in the growth of the seven cities. Today, the population growth is concentrated in the periphery of the cities and, to a lesser extent, a moderate increase in the county capitals of a medium size (20,000 to 50,000 and 5,000 to 20,000 inhabitants). This accelerates the growth of the former, while it stabilizes the number of inhabitants of the latter. Contrariwise, the depopulation process continues in the capitals of smaller counties (<5,000 inhabitants), while in type 2 and type 3 counties, the positive performance of the capital does not spread to municipalities in their surroundings, which experience an accelerated depopulation process.

Focusing now on municipalities that are not a county capital, we observe some sort of ‘county effect’: the greater the size of the county capital, residuals are higher – suggesting a spreading effect that depends on the population of the node. this would be in line with the literature of nodes of regional development. We do not appreciate relevant differences in the performance of municipalities above or below 5,000 inhabitants.

All the above results in three patterns within counties. First, seven cities that slow down but the population growth spreads to their neighbouring, regardless of their size –a peri- urbanisation where what would be relevant is perhaps proximity and the quality of

15 communications to the node. Second, in counties organized around towns of certain size (types 2 and 3), there is an invigoration of the capitals that operates fully or partially at the expense of emptying the rural hinterland –evidenced in the negative residuals of those municipalities). Third, in counties without a capital above 5.000 inhabs (type 4) the depopulation is high and accelerating.

Further statistical research is required, but the scatter plots in Figure 6 help us to observe different individual performances compared to the broad regularities above mentioned. For instance, in type 2 regions we observed an invigoration of the capitals at the cost of an accelerated demographic contraction of the rural hinterland. However, the scatter plots evidence, on one hand, that the historic emigration (EMIG50-91) of the capitals was very different. Given that starting point, almost all of them saw their momentum reinforced in the period 1991-2011 (residues positive), but the differences are even greater. The extreme positive residuals are Ponteareas (2.24%), O Carballiño (1.11%) and O Barco (1.10%). The negative ones, (-0.12%), (-0.59%) and Padrón (-0.70%).

For councils above and below 5,000 inhabitants in type 2 and type 3 counties, dispersion is very wide. We may see examples of a positive historical drift (a positive

EMIG50-91) but a negative residual today, such as Burela and Cervo –though both might reflect a census bias- and the opposite (e.g., Oza dos Ríos). Finally, smaller municipalities (<5,000 inhabs) are almost all of them in the lower left corner of the scatter plot (suggesting a demographic regression from 1950 that accelerated in recent decades). However, some exceptions that deserve mention are (+0.29%) and (+0.80%).

6. Provisional conclusions and future lines of research

Following Martinez and Peón (2015) methodology to remove the drift of past migrations, the regression residuals we obtain for the municipalities of Galicia are useful to identify some rural areas that were able to moderate the process of ageing and depopulation inherited from the past. We characterized inter- and intra- county dynamics, to observe some clear patterns, such as a peri-urbanisation process around the seven cities, or a invigorating process of county capitals of a medium size (5,000 to 20,000 inhabitants) that operates fully or partially at the expense of emptying the rural

16 hinterland. Notwithstanding, there is an obvious dispersion of results in all groups, so we need to explore possible explanations for a better performance.

In the research in progress we aim to conclude the exploratory analysis in this article with an interpretation of possible explanatory factors of the differential performance of rural municipalities in Galicia after the demographic drift. Variables to be considered include geographical and territorial (see Salvatti and Carluci, 2016 for a list), population density (Smailes et al., 2002), the role of amenities (Figueiredo, 2009) and agritourism (Phelan and Sharpley, 2011), employment and economic diversification (e.g., Marsden and Sonnino, 2008) and access to public services and better institutional governance (Sánchez-Zamora et al., 2014).

REFERENCES

Akgün, A.A., T. Baycan and P. Nijkamp (2015), Rethinking on sustainable rural development, European Planning Studies 23(4), 678-692.

Aldrey Vázquez, J.A. (2013), O reto demográfico para o rural galego, In R. Rodríguez González (coord.): Galicia, un mundo rural Vivo. UIMP – Concello de Lalín, . Pp. 38-59. 2013.

Bryden, J. and G. Munro (2000), New approaches to economic development in peripheral rural regions, Scottish Geographical Journal 116(2), 111-124.

Eliasson, K., H. Westlund and M. Johansson (2015), Determinants of net migration to rural areas, and the impacts of migration on rural labour markets and self-employment in rural Sweden, European Planning Studies 23(4), 693–709.

Elshoff, H., L. van Wissen and C.H. Mulder (2014), The self-reinforcing effects of population decline: An analysis of differences in moving behaviour between rural neighbourhoods with declining and stable populations, Journal of Rural Studies 36, 285-299.

European Commission (2013). Rural Development in the European Union. Statistical and Economic Information-Report 2013, Brussels: European Commission. Directorate-General for Agriculture and Rural Development.

Fernández Leiceaga, X., and E. López Iglesias (2000), Estrutura Económica de Galiza, Edicións Laiovento.

Figueiredo, E. (2009), One rural, two visions: Environmental issues and images on rural areas in Portugal, Journal of European Countryside 1(1), 9-21.

17

Martínez, X.M. and D. Peón (2015), Patróns de despoboamento do rural galego: Unha análise por comarcas, Revista Galega de Economía 24(1), 63-80.

Parr, J.B. (1999). Growth-pole strategies in regional economic planning: A retrospective view, Urban Studies 36(7), 1195-1215.

Perroux, F. (1955). Note sur la notion de poles croissance, Economic Appliquee 1&2, 307-320.

Rizzo, A. (2016), Declining, transition and slow rural territories in southern Italy: Characterizing the intra-rural divides, European Planning Studies 24(2), 231–253.

Spoor, M. (2013), Multidimensional social exclusion and the ‘rural-urban divide’ in Eastern Europe and Central Asia, Sociologia Ruralis 53(2), 139-157.

López Iglesias, E. (1995). Demografía e estruturas agrarias. Análise da dinámica demográfica e das mudanças nas estruturas fundiárias da agricultura galega, 1950-1993. Universidade de Santiago de Compostela.

López Iglesias, E. (2013). A gobernanza e xestión do medio rural galego a comezos do século XXI: Reflexións e propostas para o debate, En Román Rodríguez González (coord.): Galicia, un mundo rural Vivo. UIMP – Concello de Lalín, Pontevedra. Pp. 130-147.

Marsden, T. and R. Sonnino (2008), Rural development and the regional state: Denying multifunctional agriculture in the UK, Journal of Rural Studies 24, 422–431.

Phelan, C. and R. Sharpley (2011), Exploring agritourism entrepreneurship in the UK, Tourism Planning & Development 8(2), 121-136.

Salvatti, L. and M. Carluci (2016), Patterns of sprawl: The socioeconomic and territorial profile of dispersed urban areas in Italy, Regional Studies 50:8, 1346-1359, DOI: 10.1080/00343404.2015.1009435

Sánchez-Zamora, P., R. Gallardo-Cobos and F. Ceña-Delgado (2014), Rural areas face the economic crisis: Analyzing the determinants of successful territorial dynamics, Journal of Rural Studies 35, 11–25.

Smailes, P.J., N. Argent and L.C. Griffin (2002), Rural population density: Its impact on social and demographic aspects of rural communities, Journal of Rural Studies 18, 385–404.

Xunta de Galicia (2013). Plan para a Dinamización Demográfica de Galicia 2013 – 2016 Horizonte 2020, Santiago de Compostela: Consellería de Presidencia, AA.PP. e Xustiza, Xunta de Galicia. Accesible en http://www.xunta.es/c/document_library/get_file?folderId=577713&name=DLFE- 19604.pdf

Xunta de Galicia (2014). Diagnóstico de situación socioeconómica e territorial de Galicia, Santiago de Compostela: Consellería de Facenda, Dirección Xeral de Planificación e Orzamentos. Accesible en http://www.conselleriadefacenda.es/documents/10433/1347376/Diagnostico-Socioeconomico- Galicia-gal.pdf/a2b3ac61-d536-4058-9346-d3d0e18fb752

18

IBM SPSS Statistics version 21 and R Project (Packages rriskDistributions and zipfR) were used for statistical analysis.

R Core Team (2012). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/.

19