**4. Results**

This study began with the selection of variables with an emphasis on the vital sector of rural areas. The results of the KMO test as a measure of sample adequacy (0.730) were moderately good according to the Kaiser classification. In addition, Bartlett's test of sphericity was statistically significant (Table 1). The results of these two tests indicated the adequacy of the use of Factor Analysis in this study. Subsequently, a Correlation Analysis was performed, followed by a Factor Analysis.

### **Table 1.** KMO and Bartlett's Test.


Source: the authors' calculations.

This factor explained 71.511% of total variance, with eigenvalues higher than 1 (3.576) (Table 2). The correlation matrix indicated that GDP per capita was positively correlated with labour productivity (total economy and primary sector), while it was negatively correlated with the share of employees in the primary sector as well as the share of the primary sector in total GVA (Table 2), thus indicating that high dependence on the primary sector is a feature of regions that are in a less favourable economic situation and are thus less competitive regions. Factor loadings for this dimension are also presented in Table 2. The positive sign in front of the factor loadings of the variables GDP per capita, total labour productivity of all sectors, and labour productivity in the primary sector indicate overall socioeconomic development in the region, while the negative sign in front of the factor loadings of the variables share of employees in the primary sector as well as the share of the primary sector in the creation of GVA indicate that the primary sector is of less importance in more economically developed regions. The dominant variable within this factor, and with the highest correlation with the factor, was the GDP per capita (0.872). The calculated factor scores for this factor indicated the level of economic development, or wellbeing, across regions in the EU and Serbia, with the best rated observation units showing the best socioeconomic performance.

Factor scores, i.e., Index of Socioeconomic Performance, were ranked within a range of −3 to 3 and divided into quintiles. The averages for the five groups identified in Table 3 were drawn according to the level of socioeconomic development. Group 1, which included most of the intermediate and predominantly rural regions in Serbia, had an average of 27.6% of employees working in the primary sector; the primary sector had an 11.2% share of GVA creation, and the lowest levels of GDP per capita, and labour productivity both in total and in the primary sector. These results are disturbing and point to the grea<sup>t</sup> importance of the primary sector in the overall regional economies of NUTS 3 regions. The share of the primary sector in employment and GVA of the region declines and GDP per capita and labour productivity increases were highest in Group 1 and then decline for each subsequent group. In Group 5, the average share of employment in the primary sector was 3% and the average share of GVA was 2%, which indicates other sectors contribute much more to the economy. There has been a decline in the share of employees in agriculture in the EU-15 since 1990, with

an average reduction of 2–3% per year, which has resulted in an absolute reduction in the agricultural workforce by about 340,000 workers, or 190,000 annual work units (AWU) [52]. According to the same source, the only exceptions in the EU-15 that do not show a declining trend in the agricultural workforce are in those regions with a high proportion of part-time workers and a larger share of farms engaged in other profitable activities.

**Table 2.** Results of Factor Analysis: Socioeconomic performance of intermediate and predominantly rural regions.




Factor extraction method: Principal Components Analysis. Source: the authors' calculations.



Source: the authors' calculations.

In Figure 1, the darkest colour indicates the group with the best socioeconomic performance, and the group with the lowest socioeconomic performance is marked with the lightest colour. Regional inequalities are noticeable both among different countries and within one country. In this study, the focus was on several significant regional inequalities within the observation units.

**Figure 1.** Index of Socioeconomic Performance of regions in EU plus Serbia as candidate for membership. Source: Author processed based on results of FA. Note: The specific status of Kosovo and Metohija excluded it from the analysis. Adobe Photoshop CC 2015 and NUTS 3 maps of the European Commission were used.
