LogoESDP.gif (262 bytes)  d Spatial Indicators
for European Regions
d c
SPESP

 



 

.


emblem1_BBR.gif (1058 bytes)

adler_y_Germany.gif (1087 bytes)

Federal Office for Building and Regional Planning
..

The database of regional indicators has been integrated by BBR (Germany) based on the set of indicators produced by SPESP Working Groups during 1999.

BBR plans to complete and maintain the database in the future, please visit BBR website to download future versions clicking on www.bbr.bund.de/english/europe.htm

..
.. This page has access to:

 

..
Final Report
.
..  

Table of contents:

I Mission and proceeding of the Concerted Action Group

II Indicators

III Analyses of effects

IV Factor analysis

V Spatial differentiation of the EU territory and regional typologies

VI Summary

VII References

Annex

 

 

 

I Mission and proceeding of the Concerted Action Group

In Nijmegen, after the presentation and discussion of the working groups on strand one and the reflection on how to proceed and feed the various single results into the overall framework of the Study Programme, it was agreed that some kind of synthetic view is needed.

The German National Focal Point offered to start a concerted action. Concerted action meaning that first, we would collect indicators and data from the respective working groups and compile a common file. And second, we would conduct a quantitative analysis with bivariate and multivariate operations.

The objective was to try to feed some results of the working groups into the overall framework and integrate them into some kind of synthetic view. It was aimed at establishing a first and cautious idea on how the seven criteria for spatial differentiation of the EU territory as identified in the ESDP, interrelate and on the picture they portray of the territory of the EU. Doing this, an overview of data availability according to presently existing data sources would be arrived at, too.

However, coming back to doubts that at least partly stem from remaining questions regarding the purpose and the intended application of the indicators and a certain dissatisfaction with the conceptual framework, it is to be pointed out, that the objective of this initiative was and is NOT to summarise the results of the seven working groups nor to establish a synthetic higher order analysis.

In Nijmwegen, it was agreed that working groups I.1 to I.7 would send a set of about five indicators each - including data – to the BBR. The selection was supposed to follow two criteria: the significance of the indicator in depicting the respective spatial criterion and data availability. Regarding scale and the level of spatial differentiation we arrived at the pragmatic and feasible compromise to restrict ourselves to the NUTS 2 level, with the obvious and known short-comings. To ensure compatibility, we distributed the respective NUTS 2 classification (version 6) to all National Focal Points.

A total of 49 indicators were sent to us. Irrespective of data quality they were all integrated into one data-file consisting of 54 indicators (including 5 additional indicators from the BBR). The complete data set is available as at the website of the BBR (www.bbr.bund.de/english/europe.htm), as is its description. Both will be updated regularly. However, for the calculations presented below, because of insufficient data quality, a limited set of 23 only (plus three computed typology indicators) was used.

With this data file we conducted various bi- and multivariate operations. In order to ensure a balanced input of the working groups not only the single indicators but also aggregated synthetic indicators (one per spatial criterion) entered the analyses.

The approach and the objectives of the analyses were the following:

  • Correlation analyses were computed in order to identify interrelationships of the provided indicators and variables and describe their strengths (chapter III).
  • A factor analysis served to reduce the number of indicators to six factors describing and explaining the spatial diversity of the EU territory independently of each other (chapter IV).
  • A third approach is based on classifications of the EU territory according to three region typologies representing different dimensions: while the first typology expresses the geographic location of a region the second is based on its dominant settlement structure (measured as a combination of population density and existence/size of centre, see Schmidt-Seiwert 1997). The third typology was developed by the authors by means of a cluster analysis. Here, regions similar to each other with regards to the characteristics depicted by the provided indicators were summarised into groups, totalling the number of ten.

Taking these three typologies as independent variables and the contributions of the six working groups as dependent variables, comparisons of means showed whether and how the distributions of the respective attributes vary significantly over the regions and, furthermore, put to a test, which of the three typologies is suited best for differentiating the EU territory according to the indicators provided (chapter V).

In the following, the approach and the results are outlined in detail.

 

II Indicators

II.1 Contributions from the working groups

Until April 2000 49 indicators were forwarded to the BBR by the working groups. An integrated data-file was composed consisting of all data and indicators forwarded to us plus two recoded indicators (see below) and three regional typology indicators added by the BBR. However, only those 26 indicators (including the three typology indicators) for which data quality respectively availability was sufficient for (most of) the 202 NUTS-2 regions of the EU were used for the analyses presented below.

In the following, the received contributions are described with special focus on the selection of indicators and, where applicable, restrictions of data quality. A separate Excel file documents their formats, measurement levels, sources, coverage etc. in a more formalised way. This description as well as the data-file itself can be downloaded from the homepage of the BBR at www.bbr.bund.de/english/europe.htm.

Working Group 1 – Geographic position

The working group on geographic position supplied nine indicators: geographic, physical, cultural and accessibility indicators. The two geographical indicators are geographical latitude and geographical longitude, both taken from a world atlas. Above all, these indicators are important for identifying the location of a place and/or the Euclidean distance between two places, here: regions. As physical indicators, elevation above sea-level (in meters), length of seashores (in percentage of region’s perimeter) and annual sunshine radiation (in kWh/m²) were selected. They convey information on aspects of climate and remoteness respectively an important natural resource. The cultural indicator language shows the major and secondary languages in the European regions. Finally, three accessibility indicators based on calculations of German IRPUD were selected: accessibility by road to population, accessibility by rail to population and accessibility by air to GDP. They are potential indicators, measuring the potential of an area as the total of destinations in other areas that can be reached from the area discounted by a negative function of the effort to reach them. Values for Ceuta y Melilla, the Canarias as well as the Portuguese islands partly had to be estimated. For the analyses only the accessibility indicators (as only they can actively be influenced by policies) were used.

Working group 2 – Economic strength

The working group on economic strength sent five classic economic indicators: GDP per employee, measuring output and productivity, GDP per capita, measuring wealth, share of employment in agriculture and share of employment in research and development (each by total employment/100 employees) as indicators of sectoral structure respectively future orientation, and, lastly, the unemployment rate as a figure of exclusion from the labour market. The data for these five indicators are taken from the REGIO database from EUROSTAT. Because of missing values, data for the R+D indicator had to be filled up with data from preceding years going back to 1993 and/or higher NUTS levels. In case of the R+D indicator, values for the two regions of Corse and Luxembourg had to be estimated. For the GDP (mean values from 1993 to 1995) and the unemployment indicators an index was chosen, with the EU mean as reference resembling 100%.

Working group 3 – Social integration

The working group on social integration identified seven sets of indicators: demographic structure, economic structure, labour market, income, education, housing and living conditions, social cohesion and political resources. However, data availability on regional level was sufficient for labour market only. The working group supplied three indicators in two formats: unemployment, long-term unemployment and female activity rate in percent and as indices. Reference for the indices is the national average, thus the indicators point to regional differences within the Member States. As the situation on the labour market influences a lot of other spheres, these indicators do not only quantify the access to the labour market but also convey information on financial and social living conditions as well as the participation and integration of women. Data are from 1996 and are, as above, taken from the REGIO database.

Working Group 4 – Spatial integration

The working group on spatial integration originally named five indicators: GDP per capita, unemployment, inter-regional migration, net-freight flow between regions and "a co-operation indicator".

The two former indicators were selected by other working groups, too (1.2/1.3). Regarding the latter, two indicators were actually developed: a twinning indicator, representing the ratio of EU-funded host municipalities per region, based on a data-base of the project "Town-twinning" of DGRegio and an interreg indicator measuring national financing of INTERREG II A projects by GDP. The latter, by definition, is restricted to NUTS 2 regions on national borders and therefore of limited use for the analysis at this stage.

Inter-regional migration and net-freight flow between regions as available from EUROSTAT measure inner-country migration and freight flows only and were therefore not used at this stage. Furthermore, data scale partly is non-regionalised at all or only available on NUTS 1. However, the working group further developed these indicators for UK, the Netherlands and Portugal (see the final report of the working group). Because of the different format it was not possible to integrate these data into the common data-file.

Working Group 5 – Land-use pressure

Three ordinal indicators were provided by the working group on land-use pressure: land abandonment, agricultural intensification and land use pressure due to urbanisation and economical growth. The latter integrates data on urban agglomerations and transport networks from CORINE and the variation of GDP between 89 and 96 taken from the REGIO database. In addition, these effects are weighted for three land-use categories established by the working group: natural areas, semi-natural areas and wetlands or water surfaces. The aggregated indicators agricultural intensification and land abandonment rely on information such as the share of agricultural holdings, extensive and intensive land uses, the change of agricultural accounts and the geographic proximity to urban settlements. As data for all the indicators show substantial gaps, they were not used at this stage. Detailed information on how the indicators were computed can be found in the final report of the working group.

Working Group 6 – Natural assets

The natural assets working group provided six indicators. Following the pressure state response approach they are comprised of two pressure indicators - pressures on the environment and emissions of acidifying gases - and four state indicators – ecosystemic diversity, natural hazards, coastal value and natural protected areas. The data stem from different sources (IUCN, UNEP, USGS, EEA etc.) and were calculated on ordinal scale on NUTS 2 by the working group (for detailed information on the computation see final report of the working group). Unfortunately, despite of close collaboration with EEA, for ecosystemic diversity values for Finland, Sweden and the UK are still missing.

Working Group 7 – Cultural Assets

The working group on cultural assets is split into two parts – cultural landscapes on the one and cultural sites and monuments on the other hand. For cultural landscapes seven single indicators were provided that are combined in the complex indicators significance and endangering degree. For the former this is agricultural production by utilised agricultural areas in ECU, share of farms with less than 20 ha by total utilised agricultural area and yearly tourist stays. For the latter it is population growth by total area, dissection (length of transportation network by total area), use of energy and lubricants and standard gross margin, the last two by utilised agricultural area. The single indicators were provided in five classes whereas the composite indicators consist of three. For most of these indicators, data coverage unfortunately is not complete. Considering that both, extreme population growth and extreme population decline pose a serious threat to rural areas and in line with the provided aggregation model a modified formula of the variable on population growth was added by the BBR, describing the absolute variation from the mean growth and labelled population change accordingly. The same is true for the standard gross margin respectively the additional variable standard gross margin change.

The central concept the built heritage part of the working group adhered to is tourism. The data provided consisted of four indicators reflecting this focus: presence of cultural sites, concentration of cultural sites, tourist pressure and touristicity, only the latter one with major gaps.

 

 

Image29_p.jpg (4050 bytes)
Synthetic indicator Geographic position

Image30_p.jpg (4313 bytes)
Synthetic indicator Economic strength

Image31_p.jpg (4299 bytes)
Synthetic indicator Social integration

Image32_p.jpg (4058 bytes)
Indicator Spatial integration (here: twinning)

Image33_p.jpg (4216 bytes)
Indicator Land use pressure (here: agric. intens.)

Image34_p.jpg (4293 bytes)
Synthetic indicator Natural assets

Image35_p.jpg (4451 bytes)
Synthetic indicator Cultural landscapes (C.A.1)

Image36_p.jpg (4321 bytes)
Synthetic indicator Built heritage (C.A.2)

II.2 Synthetic indicators

The indicators described above represent the seven criteria of spatial differentiation as outlined in the Nordwijk draft of the ESDP. Depending on the topic quantity as well as quality of scientific theoretical and empirical references vary significantly. Accordingly, the contributions of the working groups in terms of numbers of indicators but also in terms of their theoretical and explanatory strength vary extensively, too. In order to avoid yet another example of conceptual domination of well established topics such as e.g. economic strength, it was aimed at establishing a greater balance between the criteria by constructing one dimension per working group that would enter into the analyses with the same weight each.

With this objective one synthetic indicator per working group was computed. The proceeding was that all indicators were brought into the same direction (high value – positive impact) and standardised by means of a z-transformation. The mean of the z-values of all indicators per working group thus represent the respective synthetic indicator.

Table II.3 lists the variables that were used in this way with "(-)" indicating where the direction has been changed. The table is followed by maps (II.3.1 to II.3.7b) depicting the regional distribution of the synthetic indicators in the EU.

Table II.3: Synthetic indicators: å z-values/no. of variables per working group

Synthetic Indicators Variables
1.1 Geographic Position Accessibility of population by road

Accessibility of population by rail

Accessibility of GDP by air

1.2 Economic Strength GDP per employee

GDP per capita

Unemployment (index, EU=100) (-)

Share of employment in agriculture (-)

Share of employment in R+D

1.3 Social Integration I/II Unemployment (I: in %, II: NUTS 0=100) (-)

Long-term unemployment (I: in %, II: NUTS 0=100) (-)

Female activity rate (I: in %, II: NUTS 0=100)

1.4 Spatial Integration Twinning
1.5 Land use pressure Agricultural intensification (-)
1.6 Natural Assets Emissions (-)

Natural hazards (-)

Environmental pressures (-)

Natural protected areas

1.7a Cultural Assets 1: Cultural Landscapes Yearly tourist stays (for significance)

Share of farms with < 20 ha utilised agricultural area (for significance)

Population change (for endangering degree) (-)

Standard gross margin change1 (for endangering degree) (-)

1.7b Cultural Assets 2: Built heritage Presence of cultural sites

Concentration of cultural sites

Tourist pressure on site (-)

 

III Analyses of effects

Correlation analyses offer a first insight into the picture the selected indicators portray of the territory of the EU. In the following, the bivariate relationships among and between the indicators of the different working groups are described. While in a first step, zero-order correlations are presented for the single indicators (III.1) the attention is turned to the synthetic indicators in a second step (III.2). In addition to the zero-order correlations partial correlations for both the single and the synthetic indicators are listed in the annex. As can be observed by the interested reader the picture changes substantially when the effects of intervening variables are controlled. However, as in each case all other variables are treated as third variables changes in the zero-order effects cannot be attributed to single variables. An elaborate interpretation of the partial correlations therefore is renounced at this stage.

 

Table III.1:
Bivariate correlations of single indicators (Pearson’s and Spearman coefficients)

III.1 Correlation analysis with single indicators

Concerning bivariate effects within the working groups the selected indicators highly correlate with the other indicators of the same working group with only few exceptions (unemployment in economic strength, natural protected areas in natural assets and population change in cultural assets, for details see table III.1). This is especially true for the criteria geographic position, economic strength and social integration, less so for natural and cultural assets. Particularly strong correlations (>.600) exist between the three accessibility indicators, the two GDP indicators, female activity rate and long-term unemployment (negative) as well as emissions and environmental pressures.

However, regarding the criterion economic strength and the indicator unemployment rate the only effect is with GDP per capita chosen as a measure for wealth. Yet, regions with a higher share of employment in agriculture tend to be "less wealthy"(-.522), "less productive"(-.359 for GDP per employee) and "less future oriented" regarding the economy (-.377 for share of employment in R+D). To what concerns the two GDP indicators, the opposite is true for share of employment in R+D (.519 and .340 respectively).

To what concerns natural assets it has to be pointed out that natural hazards correlates negatively with both emissions and environmental pressures.

Regarding cultural assets it can be stated that the two variables chosen as descriptors for the significance and the endangering degree of cultural landscapes correlate only weakly with each other. The indicators for the built cultural heritage, that are based on the concept of tourism, show stronger correlations.

Looking at the relationships between the working groups the well established indicators of the first three working groups – and here the accessibility indicators especially - prove to have the most and the strongest effects.

Regions with high accessibility tend to be economically strong as well as socially and spatially integrated but at the same time they are exposed to a high level of emissions and environmental pressures (p<.60) and have less tourist pressure albeit a higher concentration of cultural sites. Particularly high correlations emerge between the accessibility indicators and environmental pressures (.707 to .796) and emissions (.613 to .635) on the one and accessibility to GDP by air and GDP per capita (.661) and share of employment in agriculture (-.605) on the other hand.

Regarding economic strength the sectoral indicator is the strongest: a high rate of employment in agriculture goes in line with high long-term unemployment and a low rate of female employment, few environmental pressures and few emissions, but a strong exposure to natural hazards and a high share of small farms. Without exception, the correlations for the R+D indicator chosen as representation for the future orientation of the economy are an asymmetrical reflection hereof. To what concerns the GDP indicators not much unexpected is to be said: regions with a high GDP (both per employee and per capita) tend to be exposed to a lot of environmental stress (emissions and pressures), have a stable population and a high concentration of cultural sites. Concerning women’s integration into the labour market the effects of the GDP indicators go in opposite directions: women are more likely to be employed in a regions with a more wealthy population, and less likely in regions with highly productive economies. Turning the attention to the spatial distribution of the indicators for social integration there is some evidence that long-term unemployment is particularly high in regions that are not very attractive in other respects neither: higher emissions, environmental pressures and natural hazards and only little tourist pressure prevail.

Spatial integration is measured as the ratio of municipalities receiving funding from DGX for hosting twinning activities. The bivariate correlations with the indicators of the other working groups can only very tentatively be described as pointing at a trend that spatial integration in terms of twinning is stronger in highly accessible, wealthy and future oriented areas with low unemployment and little exposure to natural hazards but high exposure to other environmental stresses.

Lastly, it should be pointed out, that the indicator natural hazards in addition to the above mentioned correlations also highly correlates with share of farms < 20 ha (.599) and female activity (-.535). These figures confirm the relationship that was pointed at before: Many of the regions exposed to a particularly high risk of e.g. earthquakes and vulcanos are situated in the Mediterranean and are characterised by relatively traditional social and economic structures. It can be assumed that the most adequate explaining variable thus is a regional one.

 

Table III.2:
Bivariate correlations of synthetic indicators

III.2 Correlation analyses with synthetic indicators

As described in II.3, for each criterion of spatial differentiation a synthetic indicator was computed with the aim of reducing the number of indicators in the analysis and balancing the working groups so that each would enter the analysis with the same weight.

The zero order correlations in table III.2 show interesting features. As can be seen clearly, all synthetic indicators correlate significantly with at least two of the other synthetic indicators. However, with one exception, the correlations are only weak.

Confirming the results described above, geographic position (here: accessibility) and economic strength by far show the strongest correlations with the other criteria. The zero-order correlation between the two is a convincing .685, indicating that most often accessibility and economic strength go hand in hand. The relationships of the two with spatial integration, social integration and cultural assets 2 (built heritage) is less strong (between .217 and .412). Still they point at other features of the afore mentioned accessible and strong regions: socially as well as spatially, they also tend to be better integrated.

As to be expected there also is a negative correlation between geographic position and natural assets as well as cultural landscapes (cultural assets 1).

 

IV Factor analysis

A factor analysis is an operation that helps to reduce the number of influencing variables and to identify the central dimensions in a model. Whereas the correlation analyses in chapter III investigated the relationships between all the indicators - single and synthetic - this operation tests whether effects of single indicators can be grouped around factors that account for the spatial diversity of the EU territory independently of each other. The explanatory strength of these factors is investigated, too. For the analysis, 19 of the 23 indicators of the data-file are entered into the computation. By means of the principal components analysis and with respect to the Kaiser criterion a total of six factors is extracted. The amount of total variance these factors account for rates at 74%. Table IV shows the factor loadings of the grouped variables resulting from a varimax rotation. The factors were labelled accordingly:

 

Table IV: Rotated factor matrix for single indicators5
.

Factor 1

Centrality

Factor 2

Traditionality

Factor 3

Economic strength

Factor 4

Social inclusion

Factor 5

Protected areas

Factor 6

Population change

Accessibility to population by rail

.880

. . . . .
Accessibility to population by road

.855

. . . . .
Environmental pressures

.796

. . . . .
Emissions

.780

. . . . .
Accessibility to GDP by air

.714

.

.499

. . .
Share of farms <20ha .

.812

. . . .
Female activity rate .

-.780

. . . .
Natural Hazards .

.773

. . . .
Long-term unemployment .

.617

.

-.567

. .
Share of employment in agriculture

-.454

.590

. . . .
GDP per employee . .

.875

. . .
GDP per capita . .

.816

. . .
Concentration of cultural sites . .

.726

. . .
Share of employment in R+D .

-.403

.578

. . .
Unemployment . . .

-.829

. .
Tourist pressure . . .

.565

. .
Protected Areas . . . .

.784

.
Population change . . . . .

.830

Twinning . . . . .

.418

5 only factor loadings >.400 are displayed factor loadings >.600 are printed in bold

 

Image37_p.jpg (4397 bytes)
Factor 1 Centrality



Factor 1 consists of the three accessibility indicators as well as two environmental indicators: emissions and environmental pressures and shows that high accessibility still tends to be achieved at the expense of the environment. The correlations identified in chapter III already showed the strong interrelations of these indicators. Factor loadings are very high and range between .714 and .880. The regional picture (see map IV.1) shows that this factor can still be described as centrality factor depicting an however diffuse centre-periphery polarisation. The Benelux countries and Western Germany are attributed the highest and the Nordic countries, parts of Scotland and the Mediterranean countries Spain and Greece the lowest values. Apart from the Nordic countries, the metropolises are clearly set apart from their surrounding regions. Factor 1 alone explains 22% of the variation

 

Image38_p.jpg (4753 bytes)
Factor 2 Traditionality

The indicators grouped around factor 2 represent traditional economic and social structures: share of farms < 20 ha and the sectoral indicator (share of employment in agriculture) with strong positive loadings, employment in R&D with a somewhat lower negative loading on the one and two indicators on inclusion respectively exclusion from the labour market (female activity rate and long-term unemployment) on the other hand. Furthermore, the loading for natural hazards quantifying pressure on the environment for example in terms of flooding and earthquakes is very high, too. Factor 2 thus expresses proximity to traditional economic and social structures combined with a high exposure to natural hazards – an interdependency that was accounted for by the correlation analysis, too. Factor 2 explains 16% of the variance and was labelled traditionality. It shows a strong north-south contrast most obviously between the Nordic countries including Denmark on the one and Greece and southern Italy on the other hand (see map IV.2). However, there are some high values for several Benelux and western German regions, too.

 

Image39_p.jpg (4809 bytes)
Factor 3 Economic strength

Factor 3 has been labelled economic strength as it combines the major economic indicators: the GDP indicators for economic output and for wealth, as well as the modernity indicator share of employment in R+D, and accessibility to GDP by air. Concentration of cultural sites is included here too, as due to its operationalisation (number by total area), agglomerations have higher densities of registered monuments. Factor 3 clearly sets apart the economically dominant metropolises and regions of Hamburg, Berlin, Brussels, Vienna, London, Luxembourg, the Ile de France and the Rhine-Main Area against the new German Länder, parts of Greece, Portugal, and the Austrian Burgenland at the border of the European Union and single regions in the UK (see map IV.3). With 15% of explained variance factor 3 is almost as strong as the traditionality factor 2.

While the first three factors account for more than half of the spatial variation of the EU territory (53%) the other three factors, factors 4 to 6, explain another 21% of the total variance.

 

Image40_p.jpg (4766 bytes)
Factor 4 Social inclusion

 

Factor 4 combines two of the indicators of the working group social integration quantifying the access to the labour market: unemployment and long-term unemployment, both with a negative loading. The factor is thus labelled social inclusion. Although tourist pressure is included here, too, the regional picture confirms that unemployment is the strongest indicator in this factor, as large parts of Spanish and Italian tourist areas still have lowest values. Their counterparts are Algarve and Centro in Portugal and Austrian and Italian regions in the Alps like Tirol, Kärnten, Trentino. Factor 4 accounts for 9% of the regional variation.

Image41_p.jpg (4694 bytes)
Factor 5 Protected areas

 

Factor 5 represents natural protected areas. The regional distribution of this factor shows lowest values for western metropolises like Amsterdam, Berlin and Vienna, the Nordic countries and parts of England, Portugal and Greece (see map IV.5). With only 6% of explained variance factor 5 which is labelled protected areas is of minor significance in the model.

Image42_p.jpg (4698 bytes)
Factor 6 Population change

The same is true for factor 6, that, too, explains 6% of total variance. Labelled population change it is composed of population change and twinning, the former clearly dominating with a loading of .830 Portugal, north-western Spain and parts of Greece in the South as well as many regions in Germany display the highest instability to what concerns population, whereas for the Republic of Ireland and parts of England and Scotland, Atlantic regions of France as well as many parts of Italy the opposite is true.

 

V Spatial differentiation of the EU territory and regional typologies

While the analyses in chapters III and IV focused on the interrelations of the selected indicators, this chapter introduces another aspect. The spatial differentiation of the EU-territory is further investigated by summarising the EU regions into groups with similar characteristics. The applied criteria are:

  • geographic location
  • settlement structure
  • the results of a cluster analysis

The resulting typologies enter the analyses as independent variables.

In the following the presentation first focuses on the description of the three typologies (V.1 - V.3). Special attention is paid to the results of the cluster analysis. The profiles of the clusters are characterised and described according to the picture that emerges from a comparison of means for the indicators (single and synthetic) and the factors identified in chapter IV.

In a last step, the three typologies are put to a test comparing their descriptive and differentiating power by comparisons of means and investigating further the added value of the cluster typology as opposed to the two other less complex ones (V.4).

 

Image43_p.jpg (3961 bytes)
Typology Geographic location

V.1 Geographic location

The typology geographic location (see map V.1) was developed by the authors for the purpose of this report. The NUTS 2 regions of the EU regions were classified manually into the following five groups:

  • Centre,
  • North,
  • Atlantic,
  • Mediterranean and
  • Eastern EU-border.

 

Image44_p.jpg (4816 bytes)
Typology
Settlement structure

V.2 Settlement structure

The second typology (settlement structure) was established by Schmidt-Seiwert in 1997. Based on the two criteria population density and size of centres he distinguishes agglomerations, urbanised areas and rural areas and defines the following six types of regions (see map V.2):

I Agglomerated regions with a centre > 300,000 - and a population density > 300/km² (I.1)
. . - or a population density 150 - <300/km² (I.2)
. . . .
II Urbanised regions with a centre 150 - <300.000 - and a population density 150 - <300/km² [or a smaller population density (100 - <150/km²) with a bigger centre (> 300.000)] (II.1)
. . - or a population density 100 - <150/km² (II.2)
. . . .
III Rural regions with a population density < 100/km² - and a centre > 125.000 (III.1)
. . - or a population density < 100/km² with a centre < 125.000 (III.2)

 

Image45_p.jpg (4826 bytes)
Typology
Region clusters

 

V.3 Region clusters

For the third typology a cluster analysis was computed. 201 Nuts 2 regions were entered while one, Ceuta y Melilla, had to be dropped due to missing values. The chosen method was the hierarchical cluster analysis applying the Ward algorithm where the linkage of the groups is based on the minimum increase in variance thus creating groups with the highest homogeneity possible. Euclidean distance was chosen as proximity measure.

When selecting the indicators for the cluster analysis it was aimed at maintaining a balance between the different working groups. In case of the working group on cultural assets, some indicators were dropped accordingly. For cultural landscapes one significance (share of farms < 20ha) and one endangering degree indicator (population change) were selected, for the built heritage part concentration of cultural sites and tourist pressure were chosen. In order to arrive at a model incorporating most EU regions (i.e. all but the Portuguese islands, the Canarias and Ceuta y Melilla), agricultural intensification (n=190) had to be dropped, leaving the working group on land-use pressure with no indicator in the equation.

The cluster analysis is an explorative statistical operation arriving at rather sensitive results. Even minor variations of the entering data-set may cause major changes in the cluster allocation and the resulting model reflects the selected input in terms of method and variables. The model considered most appropriate identified ten clusters. Table V.3 describes their regional composition and their characteristics resulting from comparisons of means for the single and the synthetic indicators (for the actual values see table V.4).

 

Table V.3: Region clusters and their characteristics
.. Label Regions Characteristics

1

Peripheral core Most parts of France, northern and western UK, Friesland, Germany: northern regions, Trier, Koblenz, Eastern Bavaria; Eastern Austria, Ahvenanmaa, Good integration into the labour market, including the female active rate, most other indicators, especially accessibility and GDP, close to EU mean.

2

Central and central metropolitan regions Metropolitan regions: London, Ile de France, Berlin, Hamburg, Vienna; Belgium except Brussels, Luxembourg, most parts of Netherlands, most western German regions, Piemonte, Lombardia Central and economically strong regions: accessibility, GDP and employment in R+D above and employment in agriculture well below average; at the same time highest environmental pressures in the EU albeit high rate of natural protected areas.

3

Tourist regions Algarve, Baleares, Notio Aigaio, Salzburg, Tirol, Trentino, Voralberg, Valle d’Aosta Regions at the Mediterranean and in the Alps with extreme pressure due to tourism. Attractive despite lower accessibility also because of low (long-term) unemployment and little environmental pressures.

4

Brussels and Bremen Brussels and Bremen Highly central and economically strong regions. Urban disparities manifest themselves in highest values for GDP per capita and per employee and share of employment in R+D on the one and long-term unemployment (56%) on the other hand.

5

German New Länder New German Länder, Alentejo, Flevoland Economically weak regions (lowest value for GDP per employee) undergoing extreme population change (four times the EU mean); high unemployment (15%) and long-term unemployment (50%) while female activity rate is high. Together with low share of small farms a "survival" from the former GDR.

6

Central and Eastern UK Central and Eastern UK plus Alsace, Hanover, Noord-Holland, Nord-Pas de Calais Good inclusion into the labour market (including female activity) and most stable population, share of employment in agriculture well below average but high environmental pressures and few cultural sites.

7

Nordic countries Nordic countries except Ahvenanmaa Peripheral regions with highly modern economic and social structure: share of employment in R+D well above average, lowest long term unemployment, highest female activity rate; few environmental pressures.

8

Peripheral Southern Europe Greece (without Attiki and Notio Agaio) and Norte and Centro in Portugal Peripheral regions with weak and agrarian economies (lowest value for GDP per capita, share of employment in agriculture five times the EU mean, high rate of small farms, high long-term unemployment and low female activity rate), high pressure due to natural hazards but few other environmental pressures.

9

Mediterranean plus Ireland The isle of Ireland, Spain (without Galicia/Baleares), southern Italy, Marche, Corse and Sardegna Relatively poorly accessible regions with high exclusion from the labour market: highest unemployment rate at 20%, high long term unemployment and lowest female activity rate; high presence of cultural sites.

10

Northern Italy Most regions in the northern half of Italy as well as Lisboa and Attiki, Provence-Alpes-Cote d’Azur Strong economy with both GDP indicators above average; long-term unemployment and female activity rate slightly poorer than the EU mean; high values for cultural sites; high exposure to natural hazards.

 

Table V.4.1:
Single indicators and the three typologies: comparisons of means

Table V.4.2: Synthetic indicators and the three typologies: ETA²

V.4 The three typologies in comparison

The comparison of the three region typologies reveals a hierarchy regarding their descriptive and differentiating power. The tables below show the EU means and the means for the sub-groups of the three typologies. ETA² as the measure for the amount of variance of the dependent variable that is explained by the variation of the independent variable is listed too. Table V.4.1 displays the values for the single indicators, table V.4.2 for the synthetic ones. The presentation below is a summary of the most significant results.

Although the typology based on the settlement structure has significant values for all the geographic and economic indicators and for some of the other indicators too, the amount of explained variance is relatively low, only once exceeding 50% and accounting for more than 20% only in case of the accessibility, the wealth, the sectoral structure and "the modernity" indicators. With the exception of type II.2 all these effects are linear differentiating between highly urbanised agglomerations on the one and rural areas on the other hand.

Looking at the synthetic indicators, these findings can be summarised pointedly: Their distribution varies significantly with p<= .001 for geographic position, economic strength, natural assets and cultural assets 2, accounting for 17 up to 47% of the total variance. Whereas agglomerations of type 1 are characterised by good accessibility, economic strength and few natural assets the opposite is true for the rural areas of type 5 and 6.

With only few exceptions (esp. accessibility to GDP by air and share of employment in agriculture) the typology based on geographic location has a higher discriminating power than the more differentiated typology based on settlement structure. The accessibility indicators are particularly significant. The centre has the highest accessibility and is followed by the Eastern EU-border regions, the Atlantic and the Mediterranean with the North falling back significantly. The core area also has the highest values for both GDP indicators as well as the lowest unemployment rate in the EU. Another rather common feature is confirmed: The Mediterranean turns out the most traditional area with highest values for share of employment in agriculture, unemployment and long-term unemployment and the lowest for female activity rate, GDP per capita and share of employment in R+D. On the other end of the scale is the North with low long-term unemployment, few environmental pressures also in terms of emissions or natural hazards on the one and the highest rates for female activity and share of employment in R+D on the other hand.

The ETA² values for the comparison of means for the synthetic indicators confirm the described tendencies again. With .64 ETA² is highest for geographic position, polarising as mentioned above between the poorly accessible Nordic countries and peripheral Mediterranean regions and the Centre. Explained variances for the other synthetic indicators range from .14 (spatial integration) to .38 (social integration).

The typology based on the region clusters clearly is the most significant of the three. With the exception of land use pressure all ETA² values are significant at .001 level and explain up to eighty percent of the variance of the spatial indicators. Without exception, the amount of explained variance is substantially higher for the cluster typology than it is for either of the other two typologies. As before, ETA² for geographic position measured as accessibility to population is still highest with now .78 respectively .75, only exceeded by concentration of cultural sites.

 

VI Summary

Based on the contributions of the working groups of strand 1, a data-file with 54 indicators was compiled.

The approach applied was an experimental and a comparative one: the interrelationships of the selected indicators were tested against the background of different operations. The indicators entered the analyses as separate variables but also summarised (and standardised) for each working group respectively spatial criterion, giving each dimension the same weight.

Zero-order correlations tested the effects among single and synthetic indicators.

A factor analysis grouped the indicators into six independent dimensions explaining three quarters of the overall variation. According to the loading of the indicators, theses factors were labeled centrality, traditionality, economic strength, social inclusion, protected areas and population change.

Finally, three typologies of regions were developed and applied to further investigate the spatial differentiation of the EU: a typology based on geographic location, on settlement structures and on the results of a cluster analysis. The cluster analysis identified ten groups of regions showing both, a north-south and a centre-periphery divide. Comparisons of means for the subgroups of the three typologies revealed that the typology of the region clusters clearly had the highest discriminating and explanatory power. The cluster profiles were further elaborated by computing comparisons of means for both the single and the synthetic indicators.

Due to the framework this report is based on and due to the poor data quality, the results described on the pages above are clearly of a preliminary character.

However, depending on the feedback from the other working groups, advancements concerning the selection of indicators, improvements of data quality and last but not least, the further development of ESPON, the started experiment could well be elaborated further to arrive at punchy policy messages in the future.

The data-file can be downloaded from the web page of the BBR (www.bbr.bund.de/english/europe.htm). If there are any amendments, modifications or actualisations regarding the data from your side or should you want to provide additional indicators please do not hesitate to contact the authors.

 

VII References

This report is solely based on the contributions of the seven working groups of Strand 1 – Criteria for the spatial differentiation of the EU territory in terms of indicator descriptions, data, reports and emails.

For the typology based on settlement structure the following additional reference was used:

Schmidt-Seiwert, Volker (1997): Landkarten zum Vergleich westlicher Regionen. In: Hradil S.; Immerfall S. (Eds.). Die westeuropäischen Regionen im Vergleich. Opladen. PP. 603-628.

 

Annex 1:
Partial correlations of single indicators

Annex 2:
Partial correlations of synthetic indicators

Annex

Annex 1 Partial correlations of single indicators

Annex 2 Partial correlations of synthetic indicators

 


We welcome your Feedback

: