– the case of Slovakia
EVA VÝROSTOVÁ 1 , TOMÁŠ VÝROST 2
2. Analysis of Cohesion policy implementation at district level in Slovakia
2.2 Data description
As can be seen from the previous section on the distribution of funds, any analysis on the district level of Slovakia is necessarily heavily influenced by a single outlier representing the capital city of Bratislava. It is only natural for Bratislava to behave differently, for several reasons – on one hand, Bratislava has a separate operations programme with constrained legibility of applications, but on the other hand many applicants for funding represent institutions based in Bratislava, even though the specific projects to be implemented may be related to the operations of these institutions in other Slovak regions. Excluding Bratislava region from the set of applicants under analysis thus may reduce the bias inherent in the analysis.
The dataset thus consists of 70 Slovak districts, plus the city of Košice.
The approach to the quantification of the relationship of regional development and funding during the 2007–2013 period is based on publicly available data on the district level, compiled primarily by the Statistical Office of the Slovak republic, National bank of Slovakia and Central Office of Labour, Social Affairs and Family.
The analysis is exploratory in nature – thus, instead of testing an underlying assumption on the relationship, the objective is to determine the local variables
with observable relationship to funding. Our approach was thus based on com-piling the most complete dataset on the district level, which would be used for the selection of the most significant variables.
The dependent variable in our analysis is in all cases represented by the overall funding of projects in a specific district. The information on funding was compiled from the list of all projects within the programming period in Slovakia. The funding was attributed to district based on the official seat of the applicant and then aggregated. This approach has obvious shortcomings, in that the seat of an institution or a company does not necessarily imply spending in the same region, but as more detailed information is not publicly available, this choice makes for the best proxy under such constraints (we are presently working with relevant institutions to obtain more detailed data on the spending within the financed projects). We analyze the total funding, consisting of both EU funds and national co-financing, as we are interested in the overall financial effect attracted by the projects.
The objective of the constructed models is to determine, whether the funding allocation may be explained by specific attributes and properties at the district level. To characterize the regions, we use 25 variables measured on the district level.
First, National strategic reference framework 2007–2013 in Slovakia is based on the principle of territorial concentration. According to document of Ministry of construction and regional development (2007) certain municipalities (33.65%
of total number of municipalities) are identified as “innovation and cohesion growth poles”. These poles should be the accelerators of development, which were constructed in accordance with the polycentric concept of the settlement development in Slovakia (ROP, 2013). Variable POLE represents the share of the district’s population living in such poles. Other variables include the level of unemployment (UNEMP), average wage (WAGE), population density as a proxy for overall attractiveness and intensity of agglomeration (DENS), the amount of foreign direct investments within the district (INVEST), inward and outward migration flows (MIG_IN and MIG_OUT), the number of economically active population (ACTIVE), number of personal enterprises as a proxy for entrepre-neurship (ENT), the number of small and medium-sized enterprises (SMALL) and the proportion of employees in mentally challenging occupations (MENT). Health and socially oriented indicators include the average life expectancy (AGE), infant mortality rate (INFA), number of medical facilities for adults (DOCT_ADLT) and children (DOCT_CHILD), number of classes in pre-school (S_CHILD), basic schools (S_BASIC) and secondary schools (S_SEC) and the number of recipients of social benefits in need (BENEF). Housing development is represented by the number of completed new flats (FLATS). Environmental aspects is taken into account by measuring solid particulates (EM_PART), sulfur oxides (EM_SULF),
nitrous oxides (EM_NITR) and CO2 (EM_CO2) and number of households connected to the sewer system (SEWER).
All variables have been expressed in terms of intensity per inhabitant, with the exception of those already in relative terms (e.g. UNEMP) and the pollution variables, which are expressed in terms per square kilometer. To avoid artifacts in the data and reduce the effect of possible singular swings, all variables represent five-year averages spanning the years 2003–2007 directly preceding the program period (we included the year 2007, as there was practically no funding granted in 2007 and first project were financed in 2008 in Slovakia). This choice was made to allow for the interpretation, on whether the state of regional development on the district level was consistent with the policy objectives in the programming period.
Figure 2. Correlation matrix Source: Authors’ calculations.
Table 2. Simple (univariate) and full regression models for Cohesion policy funding
Simple regressions Full model
Estimate t-stat p-value R2 Estimate t-stat p-value
(Intercept) –3 844.00 –0.52 0.60
POLE 2 175.11 2.48 0.02 0.08* 517.50 0.36 0.72
UNEMP –13.43 –0.85 0.40 0.01 25.12 0.55 0.59
WAGE 2.92 1.99 0.05 0.05 1.41 0.65 0.52
DENS 2.54 3.09 0.00 0.12 –3.90 –1.71 0.09
AGE –30.33 –0.42 0.67 0.00** 95.21 1.16 0.25
INFA –36.89 –1.01 0.32 0.01* –20.79 –0.42 0.68
EM_PART 96.64 2.92 0.00 0.11 –312.20 –1.75 0.09
EM_SULF 23.88 1.84 0.07 0.05* –12.35 –0.68 0.50
EM_NITR 50.36 3.02 0.00 0.12 195.50 1.28 0.21
EM_CO2 5.85 3.22 0.00 0.13 3.66 0.35 0.73
SEWER 1 200.35 0.25 0.80 0.00** –2 973.00 –0.73 0.47
INVEST 74.33 1.68 0.10 0.04* 48.30 0.77 0.44
FLATS –15 665.43 –0.27 0.79 0.00** 46 140.00 0.49 0.63
MENT 1 482.79 0.53 0.60 0.00** –589.70 –0.23 0.82
DOCT_ADLT 1 270 568.00 0.91 0.37 0.01* –1 306 000.00 –0.93 0.36 DOCT_CHILD –5 013 961.00 –1.70 0.09 0.04 –8 619 000.00 –3.48 0.00**
MIG_OUT 122 914.60 3.37 0.00 0.14 125 900.00 1.95 0.06
MIG_IN 16 322.17 0.64 0.52 0.01** –49 880.00 –0.96 0.34 S_BASIC –62 903.03 –0.36 0.72 0.00** 189 000.00 0.86 0.40 S_CHILD 71 826.77 0.09 0.93 0.00*** –773 800.00 –0.89 0.38
S_SEC 669 017.70 7.41 0.00 0.44 544 100.00 5.07 0.00 ***
BENEF –4 364.43 –0.85 0.40 0.01 5 852.00 0.44 0.67
ACTIVE –3 044.38 –0.88 0.38 0.01* –6 593.00 –1.76 0.09
ENT 15 282.56 2.16 0.03 0.06 8 476.00 0.85 0.40
SMALL 84 168.26 4.29 0.00 0.21 61 450.00 2.17 0.04*
Note: Individual univariate regressions (left) and the full model (right). For the full model, the R2 was 0.7517 and the adjusted R2 was 0.6137. For the simple regressions, the
Bonferroni-adjusted p-value is 0.002 at 5% significance level.
Source: Authors’ calculations.
Figure 2 shows the correlation between the dependent variable (FUND) and all the explanatory variables used in the study. There are several notable clusters showing stronger dependence between variables – in the top left corner, we see strong positive correlation between the variables for various form of pollution, but also population density and outward migration. On the right, personal enterprises (ENT) are negatively correlated with the number of people on benefits, infant mortality and unemployment. These three variables also form a strongly corre-lated cluster in the lower right. As for the dependent variable (FUND), the only noticeable correlation is with S_SEC, the number of classes in secondary schools.
Even though somewhat surprising at first glance, the relationship is easily explained by the fact that the secondary schools, notably grammar schools aiming to prepare the students for universities are located mostly in the regional capitals (or capitals of higher regions), and thus are indicative of regions receiving larger portion of the funding.
Table 2 (left) presents the univariate relationships of the explanatory variables to the amount of funding (the dependent variable is explained solely by a constant and a single regressor). As we aggregate the funding for the whole programming period for each district (the effects of the projects are long-term, and it is thus not reasonable to analyze annual changes), we face a problem with a large number of regressors with respect to the sample size of 71 districts.
We take into account the problem in the following manner. First, we estimate individual regressions of single explanatory variable versus the funding. We see several significant variables. However, when we apply the Bonferroni correction for multiple comparisons, conservatively accounting for the larger number of comparisons made, the only significant results are for emissions EM_CO2, outward migration (MIG_OUT), number of classes in secondary schools (S_SEC) and number of small and medium sized enterprises (SMALL).
The individual regressions however might hide a lot of information, as they provide only a very narrow view on the relationship and interaction with respect to the dependent variable. There is therefore a need to use all the variables in a fully specified model, which is shown on the right side of the table 2. Here, only number of medical facilities for children, number of classes in secondary schools and number of small and medium-sized enterprises remains significant. This model is however also not satisfactory, as it contains rather large number of re-gressors given the sample size. We have therefore performed a stepwise regres-sion to construct a parsimonious model, keeping only the relevant variables.
The results of the stepwise regressions are depicted in Table 3. The regressors remaining in the model are not highly correlated (see Figure 2), providing for a valid specification. The only remaining and statistically significant variables in this model include number of classes in secondary schools (S_SEC), outward migration (MIG_OUT), number of small and medium-sized enterprises (SMALL)
and the number of classes in basic schools (S_BASIC). The only variable negatively related to the funding is number of medical facilities for children (DOCT_CHILD).
Table 3. Reduced and spatial regression analysis for EU cohesion policy funding Reduced OLS model
Bootstrap conf. int. (95%)
Corresponding spatial lag model
Estimate p-value Estimate p-value
(Intercept) –959.80 0.18 –2 128.0 761.2 –913.60 0.19
S_SEC 539 377.10 0.00*** 92 114.1 846 194.5 538 809.37 0.00 DOCT_CHILD –8 514 889.40 0.00*** –14 590 949.9 –1 506 957.0 –8 540 782.31 0.00 MIG_OUT 90 834.60 0.00*** 38 230.0 196 508.2 90 762.40 0.00
SMALL 55 950.00 0.00** –6 703.0 96 604.6 55 977.95 0.00
S_BASIC 333 344.40 0.01** 107 872.0 550 294.9 334 206.99 0.00
rho –0.04 0.76
Note: The reduced OLS model (left), bootstrap confidence intervals for the regression coefficients (middle) and corresponding spatial lag model (right). For the reduced model, the R2 was 0.6595
and the adjusted R2 was 0.6333.
Source: Authors’ calculations.
The presented rsults might be subject to some skepticism, as any use of stepwise regressions might be suspect to the presence of data-snooping bias.
Therefore, we have also constructed confidence intervals for the regression coefficients based on 10,000 bootstrap resamples from the original data, yielding the results in the middle portion of the table. As can be seen, all variables except the number of small and medium sized enterprises remain significant. To test for possible spatial effect, a spatial lag model has also been estimated, leading to very similar estimates (right), with the results of the LR test unsupportive of the presence of spatial effects on the district level.
3. Conclusion
The paper analyzes spatial allocation of the overall EU funding with co-financing from state budget in the Slovak Republic. We were able to identify several determinants of Cohesion policy funding allocation at district level, namely number of classes in secondary and basic schools, outward migration, number of small and medium-sized enterprises and number of medical facilities for children.
Other variables, especially unemployment rate was not confirmed to be the determinants for spatial allocation of EU funds at the district level.
The presented analysis has several weaknesses, such as the fact that the spatial distribution of applicants may be influenced by other factors, such as political situation, ability to secure co-financing to the EU, capacity to attract and use the funds, administrative capacities, corruption and others, which were not analysed within this paper. Further analysis ought to be based on the allocation of EU funds on the place of the implementation of the project. Only this more precise information may lead to the conclusion, whether the resources spent are truly focused on the less developed districts. As higher EU fund spending per se does not guarantee an improvement in the situation of a region, the analysis of the effectiveness of these funds is also necessary.
Acknowledgements
This work represents part of the project VEGA no. 1/0153/18.
References
BODENSTEIN, T., KEMMERLING, A. 2012. Ripples in a Rising Tide: Why Some EU Regions Receive More Structural Funds than Others. In: European Integration online Papers (EIoP), vol. 16, article 1. Available at: http://eiop.or.at/eiop/texte/2012-001a.htm. DOI:
10.1695/2011007.
BOUVET, F., DALL'ERBA, S. 2010. European regional structural funds: How large is the influence of politics on the allocation process? In: JCMS: Journal of Common Market Studies. Vol. 48, No. 3, pp. 501–528.
CAMAIONI, B. et al., 2013. How rural is the EU RDP? An analysis through spatial fund allocation.
In: Bio-based and Applied Economics. Vol. 2, No. 3, pp. 277–300.
CENTRAL OFFICE OF LABOUR, SOCIAL AFFAIRS AND FAMILY. Sociálne dávky. Available at: https://www.upsvr.gov.sk/statistiky/socialne-veci-statistiky/2011.html? page_id=312199 CHARRON, N. 2016. Explaining the allocation of regional Structural Funds: The conditional effect
of governance and self-rule. In: European Union Politics. Vol. 17, No. 4, pp. 638–659.
Consolidated version of the Treaty on the functioning of the European Union. Official Journal of the European Union. 9.5.2008.
COUNCIL REGULATION (ES) 1083/2006 of 11 July 2006 laying down general provisions on the European Regional Development Fund, the European Social Fund and the Cohesion Fund and repealing Regulation (EC) No 1260/1999. Official Journal of the European Union.
CRESCENZI, R. 2009. Undermining the principle of concentration? EU development policies and the socio-economic disadvantage of European regions. In: Regional Studies. Vol. 43, No. 1, 111–133. https://doi.org/10.1080/00343400801932276.
DALL’ERBA, S. 2005. Distribution of regional income and regional funds in Europe 1989–1999:
an exploratory spatial data analysis. In: The Annals of Regional Science. Vol. 39, No. 1, pp. 121–
148.
DELLMUTH, L. M., STOFFEL, M. F. 2012. Distributive politics and intergovernmental transfers:
The local allocation of European Union structural funds. European Union In: Politics. Vol. 13, No. 3, pp. 413–433.
EUROPEAN COMMISSION, 2014. Investment for jobs and growth. Promoting development and good governance in EU regions and cities. Sixth Report on economic, social and territorial cohesion. Brussels.
EUROPEAN COMMISSION, 2016. Ex Post Evaluation of Cohesion Policy Programmes 2007–
2013, focusing on the European Regional Development Fund (ERDF) and the Cohesion Fund (CF). August 2016. Brussels. Available at: https://ec.europa.eu/regional_policy/sources/
docgener/ evaluation/pdf/expost2013/wp1_synth_report_en.pdf
EUROPEAN COURT OF AUDITORS, 2019. Rapid case review Allocation of Cohesion policy funding to Member States for 2021–2027.March 2019. https://www.eca.europa.eu/Lists/
ECADocuments/RCR_COHESION/RCR_COHESION_EN.pdf
HÁJEK, O., et al. 2012. Regionální disparity a financování regionální politiky – některé poznatky z České republiky. In: Politická ekonomie. Vol. 60. No. 3, pp. 330–348.
KPMG, 2015. Available at: http://www.nsrr.sk/sk/hodnotenie/programove-obdobie-2007---2013/
MICHÁLEK, A. 2014. Disparity v alokácii a čerpaní zdrojov s dôrazom na marginálne regióny Slovenska. In: Geografický časopis. Vol. 66, No. 3, pp. 269–286.
MILIO, S. 2007. Can administrative capacity explain differences in regional performances?
Evidence from structural funds implementation in southern Italy. In: Regional Studies. Vol. 41, No. 4, pp. 429–442.
MINISTRY OF CONSTRUCTION AND REGIONAL DEVELOPMENT, 2007. Zoznam pólov rastu pre Národný strategický referenčný rámec na roky 2007–2013. Rozhodnutie ministra výstavby a regionálneho rozvoja SR č. 3/2007 zo 6. júna 2007.
MINISTRY OF FINANCE OF SR, 2016. Informácia o čerpaní EÚ fondov k 31.08.2016. 14.9.2016.
[cit. 2016-09-21].
Available at: https://www.finance.gov.sk/Default. aspx?CatId=84&NewsID=982
MOHL, P., HAGEN, T. 2010. Do EU structural funds promote regional growth? New evidence from various panel data approaches. In: Regional Science and Urban Economics. Vol. 40, No.
5, pp. 353–365.
NATIONAL BANK OF SLOVAKIA. Priame zahraničné investície. Available at:
https://www.nbs.sk/sk/statisticke-udaje/statistika-platobnej-bilancie/priame-zahranicne-investicie
NATIONAL STRATEGIC REFERENCE PROGRAMME, 2007. Available at:
http://www.nsrr.sk/download.php?FNAME=1210870605.upl&ANAME=NSRR290607.zip NOVOSÁK, J. et al. 2017. Structural funding and intrastate regional disparities in post-communist
countries. In: Transylvanian Review of Administrative Sciences, No. 51 E/2017, 53–69.
ROP, 2013. Regional operational programme. Available at: www.nsrr.sk
STATISTICAL OFFICE OF THE SLOVAK REPUBLIC, 2019. Datacube. Available at:
https://slovak.statistics.sk/
STERNBERG, R. 2012. Do EU regional policies favour regional entrepreneurship? Empirical evidence from Spain and Germany. In: European Planning Studies. Vol. 20, No. 4, pp. 583–
608.
SVOBODA, D. 2007. Kam smerujú eurofondy na Slovensku? Bratislava: Konzervatívny inštitút M.R. Štefánika. Available at: http://www.monitoringfondov.eu/article.php?cal-2007-08-21-kids
SVOBODA, T. 2016. Efektivnost využívání strukturálních fondů EU. Vybrané právní aspekty. Spisy Právnické fakulty Masarykovy Univerzity. Edice Scientia, svazek č. 577. Brno: Masarykova univerzita.
THE GOVERNMENT OFFICE OF THE SLOVAK REPUBLIC, 2016. Zoznam prijímateľov finančných prostriedkov z fondov EÚ v PO 2007–2013. Stva k 30.6.2016. Zdroj: ITMS.
Available at: http://www.nsrr.sk/download.php?FNAME=1476084825.upl&ANAME=
zoznam-16_06_30.xls
VANOLO, A. 2010. European Spatial Planning Between Competitiveness and Territorial Cohesion:
Shadows of Neo-liberalism. In: European Planning Studies, Vol. 18, No. 8, pp. 1301–1315.
VEIGA, L. G. 2012. Determinants of the assignment of EU funds to Portuguese municipalities. In:
Public Choice. Vol. 153, No. 1–2, pp. 215–233.
VÝROSTOVÁ, E. 2016. Spatial allocation of EU cohesion policy funding in Slovakia for 2007–
2013. In: KLÍMOVÁ, V. a ŽÍTEK, V. (eds.) XIX. mezinárodní kolokvium o regionálních vědách. Sborník příspěvků. (Čejkovice 15.-17.6.2016). Brno: Masarykova univerzita, s. pp.
130–138.
ZUBEK, N., HENNING, C. 2016. Local government, spatial spillovers and the absorption of EU structural funds. In: Journal of Agricultural Economics. Vol. 67, No. 2, pp. 368–397.