ABSTRACT
Poverty reduction and the tackling of social exclusion are overarching goals of development and welfare policies. This paper explores the extent to which decentralization contributes to poverty and social exclusion alleviation in European countries and regions. We find evidence that increases in central government transfers of political, administrative and fiscal authority to subnational tiers of government reduce poverty and address social exclusion at the national level. This, however, mainly happens in countries with a high degree of governance quality and, fundamentally, in urban areas. The link between decentralization and poverty and social exclusion alleviation is more uniform at the regional level, as greater regional autonomy is connected to lower poverty and social exclusion, regardless of the quality of regional government. Hence, when regional governments have the capacity to design their own independent policies, a reduction of poverty and social exclusion and improvements in well-being generally ensue.
ACKNOWLEDGEMENTS
We are grateful to Klaus Dodds, the Editor in Chief of Territory, Politics, Governance and to the anonymous reviewers for their challenging but helpful comments in the different reviewing rounds. We also thank Foteini-Antonia Papadioti (Panteion University of Social and Political Sciences) for her excellent research assistance in the compilation of the regional dataset.
DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.
Notes
1 There is also some blurring and overlap between the concepts of fiscal, political and administrative decentralization, deconcentration, delegation and devolution (Pike et al., Citation2012).
2 Generally, the:
effectiveness of public policies can be defined as the extent to which the policies are achieving the benefits they are supposed to achieve plus any unanticipated side effects [and] efficiency of public policies can be defined as the extent to which they are keeping costs down, especially monetary costs, as indicated by either total costs or a ratio that involves both benefits and costs. (Nagel, Citation1986, p. 99)
3 This indicator is part of the United Nations Sustainable Development Goals (SDG) indicator and the EU 2020 strategy indicators.
4 The choice of meso-level generally coincides with the territorial units with the greatest degree of autonomy in each country. It thus varies from one country to another: Länder in Germany, Regioni in Italy, Comunidades Autónomas in Spain or Cantons in Switzerland.
5 A limitation of a simple aggregative index is that the components with a large value have greater influence on the index. The value of each component for the quality of governance index that we use ranges from –2.5 to 2.5.
6 Austria, Belgium, Bulgaria, Croatia, Czechia, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, the Netherlands, North Macedonia, Norway, Poland, Portugal, Romania, Serbia, Slovenia, Spain, Sweden, Switzerland, Turkey and the UK.
7 Austria, Bulgaria, Croatia, Czechia, Denmark, Finland, Germany, Greece, Hungary, Ireland, Italy, the Netherlands, Romania, Slovenia, Spain and Sweden.
8 For a mapping of the regional poverty and social exclusion level, see Appendix 1 in the supplemental data online.
9 For a mapping of the regional RAI, see Appendix 2 in the supplemental data online.
10 For a mapping of the regional quality of governance, see Appendix 3 in the supplemental data online.
11 For example, Austria has two regional governments/tiers: ordinary Länder and the capital region, Vienna. Hence, r = 1, 2. It implies that we add two different dummies for Austria.
12 The size of a coefficient describes the size of the effect that an independent variable is having on the dependent variable (PovSocEx). The sign on the coefficient (positive or negative) shows the direction of the effect.
13 The controls included in the analysis are not the only ones capable of affecting poverty and social exclusion. Others, such as educational attainment, economic development, inactivity, tax, innovation, urbanization economies and physical geography, among others, may also have a non-negligible influence on poverty and social exclusion rates. However, the independent variables included in a regression model should be independent from one another. That is, they should not be correlated. Highly correlated independent variables produce problems to fit the model and to interpret the results. This results in multicollinearity, making coefficients highly sensitive to small changes, reducing the precision of the estimated coefficients and weakening the statistical power of the regression model. Due to these multicollinearity problems, we do not control for several variables originally considered as controls. These include (1) education, proxied by a human capital index based on the average years of schooling (Barro & Lee, Citation2013) and returns to education (Psacharopoulos, Citation1994) (source: Penn World Table – PWT); (2) per capita GDP (ln) (source: World Bank), measuring the economic development of the country; (3) the percentage of inactive adults (source: Eurostat); (4) tax revenue (source: World Bank); (5) patent applications to the European Patent Office (EPO) or the intramural (research and development – R&D) expenditure as the percentage of GDP (source: Eurostat), as proxies for innovation; (6) urban population as the percentage of the total population (source: World Bank), and the population in the largest city as the percentage of the urban population (source: World Bank); and (7) the distance from a country’s centroid to the nearest coastline or navigable river or the percentage of land area within 100 km of the nearest coastline or navigable river (source: Center for International Development, Harvard University; Gallup et al., Citation2010). These variables are highly correlated with Dec, Gov and/or some of the Controls.
14 For instance, the RAI of an Austrian NUTS-II region (i.e., Burgenland, Lower Austria, Vienna, Carinthia, Styria, Upper Austria, Salzburg, Tyrol and Vorarlberg) is the maximum RAI score of the differentiated regions r which refer to this specific NUTS-II region.
15 Due to multicollinearity problems, we do not control for (1) infant mortality rate, (2) the per capita GDP (ln), and (3) the per capita number of patent applications to the EPO or the intramural (R&D) expenditure as the percentage of GDP. These variables are highly correlated with Dec and/or Controls.