Unveiling core-periphery disparities through multidimensional spatial resilience maps

Abstract

Resilience has become a crucial concept in understanding the ability of complex systems to withstand shocks and adapt to future challenges. This concept has recently gained much attention in various disciplines, including geography and regional science. For example, the European Union has recognized the importance of resilience-based policies in the face of crises such as the COVID-19 pandemic. This study focuses on core-periphery disparities within European regions and aims to examine the impact of economic peripherality on regional resilience. Economic peripheries often face challenges such as adverse sectoral structures, low activity rates, and lower levels of innovation, making them more vulnerable to various shocks. The main hypothesis is that peripherality increases vulnerabilities and is correlated with lower resilience. Nevertheless, peripheral regions are more motivated to enhance resilience compared to regions already considered resilient. We argue that these peripheries need to renew existing structures or create new paths to enhance their resilience. However, historical and path dependency factors make it difficult to bring about such changes in a timely manner. The study explores different conceptual and methodological approaches to core-periphery models and emphasizes the importance of assessing socio-economic disparities from a spatial resilience perspective. Various factors, including trade integration, GDP, industrialization, human capital, and institutional efficiency, appear to contribute to core-periphery differentiation. Moreover, the paper highlights the long-run impact of the recent 2007–2012 economic crisis on peripheral regions, particularly in Southern areas of Greece, Italy, and Spain. To understand the spatial patterns of resilience comprehensively, the study presents a range of European multidimensional resilience maps, including consistent spatial indicators, building on previous resilience atlases that illustrated resilience at various territorial levels. These maps provide evidence-based insights into regional resilience performance and capacity in European regions, facilitating a better understanding of their potential to bounce back from major shocks and disruptions.

Appendices

Appendix 1: Types of peripheralization

See Fig. 9a–c.

Fig. 9

Multimodal accessibility. b Distance to political heartland. c Economic peripherality

Appendix 2: Data sources

See Tables 2 and 3.

Table 2 Core-Periphery index

Table 3 The European resilience atlas

Appendix 3: Statistical Approach (resilience indexes) (details in Bănică et al. 2021).

3.1 Resilience performance

3.1.1 Analysis of indicators and delimitation of resistance and recovery periods

For each of the selected indicators, the delimitation of the periods of resistance (resistance) and recovery (recovery) was carried out by analysing the aggregated indicators at the EU level (if available) or their average value (the average of the units at the NUTS0 or NUTS2 level). In this sense, analysing the evolution of the indicators during the period of the shocks, the following milestones were identified: the year in which the value preceding the shock (peak) was reached, the year in which the minimum value during the shock was reached (trough); the year of full recovery, if applicable, or the most recent value. Taking into account these benchmarks, the periods corresponding to the two dimensions that are generally considered in the analysis of resilience, respectively the resistance period and recovery periods, were established:

• (a) Resistance (Impact of the crisis)—is the period between the maximum value preceding the shock (peak) and the minimum value reached during the period corresponding to the shock (trough). It should be noted that the minimum value can also be reached after short recovery periods.

• (b) Recovery (Bounce back)—corresponds to the period between the minimum value reached during the shock period (trough) and the year in which the indicator value indicates a complete recovery (reached or exceeded the pre-shock value). If the indicator shows a full recovery and exceeds the value corresponding to the pre-shock period, the limit of the interval becomes the first year in which this happens. If the indicator is still lower than its pre-shock value, the latest available data was used.

  The following models were used to extract resistance and recovery trends:

  $$x_{r,t} = c + \beta_{{{\text{resistance}}\_{\text{period}}_{{t,{\text{EU}}}} }} ,$$

  (3)

  where the slope corresponding to the resistance period is β, x refers to the variable in the region or country r, in the resistance period t, and resistance_period_t,EU represents the years of resistance defined by the analysis of indicators at the EU27 level.

  $$x_{{r,t^{\prime } }} = c + \delta_{{{\text{recovery}}\_{\text{period}}_{{t,{\text{EU}}}} }}$$

  (4)

  where the slope corresponding to the recovery period is δ, x refers to the variable in the region or country r, during the recovery period t, recovery_period_t,EU represents the recovery years defined from the analysis of indicators at the EU27 level.

  In order to normalize the data, we applied a min–max formula to transform all variables of region or country i to a scale between 0 and 1, which is often used in calculating indices:

  $${\text{Normalized}}\;{\text{indicator}}_{i} = \left( {\frac{{\left. {x_{i} - {\text{min}}(x} \right)}}{{\left. {\max \left( x \right) - {\text{min}}(x} \right)}}} \right),$$

  (5)

  where x_i is the value of the indicator corresponding to region or country i, and min(x) and max(x) are the minimum and maximum values of the indicator. The inversely measured variables (for which the minimum value indicates the best performance) were normalized by reversing the minimum and maximum values.

• (c) Resilience performance index—The composite index was calculated as the weighted average of the values of the intermediate components. The weights were given by the squared factor

  $$I_{i} = 1/v\mathop \sum \limits_{j = 1}^{v} \omega_{j} x_{i,j}$$

  (6)

  loading of all dimensions assigned to that factor, later normalized. In this sense, the stages indicated in the OECD guide for constructing indices were followed (Nardo et al. 2008)

  :

  $$\omega_{j} = \mathop \sum \limits_{m = 1}^{M} \left[ {\frac{{{\text{Explained}}\;{\text{variance}}_{m} \left( {{\text{loading}}_{j,m} } \right)^{2} }}{{\mathop \sum \nolimits_{l = 1}^{M} {\text{Explained}}\;{\text{variance}}_{l} \mathop \sum \nolimits_{n = 1}^{v} \left( {{\text{loading}}_{n,m} } \right)^{2} }}} \right],$$

  (7)

  where M is the number of retained components and where the loading of dimension j on component m is zero when it has not been assigned to this component. Also, as recommended by Nardo et al. (2008), the following criteria were taken into account when extracting factors: (a) choosing factors with eigenvalue > 1; (b) factors that individually contribute at least 10% to the overall variation; (c) the number of retained factors must cumulatively explain at least 60% of the global variance.

3.2 Resilience drivers and resilience capacity

To select the main determinants of resilience, an econometric model was used in which the dependent variables were the resilience index, respectively its two components, the index that measures the performance in terms of shock resistance, respectively the one that measures the recovery after the shock. The following econometric models were estimated:

$${\text{Resilience}}_{i} = \alpha_{i} + \beta X_{i} + C_{i} + \varepsilon_{i} ,$$

(8)

where Resilience_i represents the value of the resilience index for region i, X represents a set of potential determinants that were included in the model to explain resilience, C_i are the dummy variables corresponding to the countries taken into account for state-specific effects but not observed in the model, and ε_i is the residual term. The same potential determinants were also tested in alternative models that separately explain the resistance and return components.