Baltic Sea Region Territorial Monitoring System

Territorial Cohesion - 10 indicators for BSR territorial cohesion

Distribution indicators (1-3)

The three first indicators measure overall cohesion in a distributive manner, each from its own specific point of view.

(1.) The Gini Concentration Ratio (GCR) is one of the most widely utilised inequality indicators. It measures the dispersion of a phenomenon and it operates within the range 0-1, where a value of 0 would indicate perfect equality (i.e. in our case that all regions would be exactly the same) and a value of 1 in turn maximum inequality (i.e. that all that is measured would be concentrated into a single region alone). A GCR value of e.g. 0.45 could be interpreted as the amount (45 %) required to be shifted for perfect equality to take place. Apart from being non-spatial, the GCR has the analytic limitation that it reacts in relative terms equally on changes within the middle band of regions as it does to changes in the extremes, which is troublesome, for it is most often occurrences at the extreme ends of the scale that are of interest to policy.

(2.) The Atkinson index seeks to address this shortcoming of the GCR by introducing a sensitivity parameter (ε value) that enables giving greater emphasis to, in our case, small or low performing regions. It operates on a similar scale as the GCR, i.e. 0 would indicate perfect equality and a 1 maximum inequality. For the purpose of this analysis the sensitivity parameter (ε value) is always set at 0.8, which implies that greater weight is given to changes among the lower performers. By comparing the results of the Atkinson index to those of the GCR, we are able to draw conclusions whether the changes in inequality stem from the changes in the lowest performers or not.

(3.) The 80/20 ratio (also known as the Kuznets ratio) is a simple bivariate analytic technique that concerns the relationship between the highest (top 20 %) and the lowest (bottom 20 %) performers. It is calculated as the ratio between these two and does as such not concern itself at all with what happens in the three middlemost quintiles. The higher the value, the larger is the discrepancy between the two extreme groups, and vice versa. A value of e.g. 8.0 indicates that the best performing group (i.e. the top quintile or the highest 20 % of regions) has eight times more of what is measured than the corresponding lowest performing group.

Convergence indicators (4-5)

The following two indicators measure the process of convergence by means of two commonly used standard techniques. By applying both methods in parallel, one can obtain a picture whether the process of convergence – or lack thereof – is of a sigma type (i.e. reduction of disparities in general) or of a beta type (i.e. convergence through a catch up of the low performers).

(4.) Sigma-convergence occurs when disparities in general are reduced. It is commonly measured simply by the coefficient of variation, which is calculated as standard deviation divided by the mean of all regions. The higher the value, the larger are the overall differences between all regions, and vice versa. This indicator is very sensitive to extreme outliers and can be used as a supplement to e.g. the GCR. A catch-up process of the poorest performers affects the value as much as would similar reductions among the best performers.

(5.) Beta-convergence concerns itself primarily with disparity reduction via a catch-up process by the poorest performers. It is (in this paper) measured by means of a linear regression model where the dependent variable is the level of the region at beginning of a year and the independent variable the change that has occurred during this particular year. By looking at the unstandardised "b" regression coefficient from each model, one can obtain a picture of how much the growth rate is affected by the initial level. A negative rate implies increasing convergence, as it de facto (on average) implies that the lower a region’s performance is, the higher has been its growth rate. A positive value indicates the opposite, i.e. a pull-off by the best performers.

Targeted BSR territorial cohesion indicators (6-10)

The remaining five indicators are targeting five specific aspects of territorial cohesion with particular relevance in a BSR context. Simple though they are from a methodical point of view, they nonetheless are able to provide a more diversified picture of different aspects of territorial cohesion in the BSR with a clear focus on regional specifities, and may be used in addition to the more traditional indicators described above. One aim of these is to capture the three principal divides of the BSR. Each indicator is bivariate meaning that it compares two groups of regions against each other. The last four of these indicators are based on four different DG Regio territorial typologies (supplemented by information on Belarus and NW Russia) and as such can only be applied on data available at NUTS level 3. Each indicator is calculated as a straightforward ratio, and for example a value of 1.3 would indicate that the numerator (e.g. “east” in the “east/west ratio” or “south” in the “south/north ratio”) has 30 % more of the measured entity than has the corresponding denominator.

(6.) The east/west ratio compares the amount of a phenomenon in eastern BSR to that in western ditto. Eastern BSR is comprised of the new German Länder, the Baltic States, Poland, Belarus and NW Russia. The Nordic countries and former West Germany including the NUTS 3 region of Berlin are in turn classified as Western BSR.

(7.) The south/north ratio is based on the DG Regio typology of sparsely populated areas (supplemented by information on NW Russia and Belarus). All regions classified as sparse in the typology (i.e. less than 12.5 inhabitants/km² at NUTS 3 level or less than 8 inhabitants/km² at SNUTS level 2 in NW Russia and Belarus) are classified as “north, the remaining areas as “south”.

(8.) The urban/rural ratio is based on the DG Regio Typology on urban-rural regions supplemented by information on NW Russia and Belarus. The indicator compares the class “predominantly urban regions” with the class “predominantly rural regions”. The latter class includes both regions “close to a city” as well as “remote” regions. This indicator hence excludes the middlemost category of the typology (“Intermediate regions”) and is able to provide a crude picture on relative changes between the top and bottom section of the urban-rural hierarchy.

(9.) The non-border/border ratio is based on the DG Regio typology “Border regions - internal and external” supplemented by information on Belarus and NW Russia. It compares the external border regions of the BSR to all the remaining regions. Based on this typology, there are no external border regions identified in Denmark and BSR Germany. Please note that for reasons related to easier interpretation, we have throughout calculated the ratio as “non-border regions” divided by “border regions” instead of the opposite.

(10.) The coast/inland ratio is based on the DG Regio “Typology on coastal regions”, where coastal regions are classified on basis of the (low, medium, high or very high) share of population living within the coastal zone. Our indicator compares the entire group of coastal NUTS 3 regions to all other regions.