Nonparametric approach to evaluation of economic and social development in the EU28 member states by DEA efficiency

Journal of Risk and Financial Management Article Nonparametric Approach to Evaluation of Economic and Social Development in the EU28 Member States by DEA Efficiency Lukáš Melecký1, Michaela Staníˇcková1,* and Jana Hanˇclová2 1Department of European Integration, Faculty of Economics, VŠB—Technical University of Ostrava, Sokolskátˇrída 33, 702 00 Ostrava 1, Czech Republic; [email protected] 2Department of System Engeneering, Faculty of Economics, VŠB—Technical University of Ostrava, Sokolskátˇrída 33, 702 00 Ostrava 1, Czech Republic; [email protected] *Correspondence: [email protected]; Tel.: +420-597-322-237 Received: 27 March 2019; Accepted: 22 April 2019; Published: 24 April 2019   Abstract: Data envelopment analysis (DEA) methodology is used in this study for a comparison of the dynamic efficiency of European countries over the last decade. Moreover, efficiency analysis is used to determine where resources are distributed efficiently and/or were used efficiently/inefficiently under factors of competitiveness extracted from factor analysis. DEA measures numerical grades of the efficiency of economic processes within evaluated countries and, therefore, it becomes a suitable tool for setting an efficient/inefficient position of each country. Most importantly, the DEA technique is applied to all (28) European Union (EU) countries to evaluate their technical and technological efficiency within the selected factors of competitiveness based on country competitiveness index in the 2000–2017 reference period. The main aim of the paper is to measure efficiency changes over the reference period and to analyze the level of productivity in individual countries based on the Malmquist productivity index (MPI). Empirical results confirm significant disparities among European countries and selected periods 2000–2007, 2008–2011, and 2012–2017. Finally, the study offers a comprehensive comparison and discussion of results obtained by MPI that indicate the EU countries in which policy-making authorities should aim to stimulate national development and provide more quality of life to the EU citizens. Keywords: competitiveness; country competitiveness index; DEA; efficiency; European Union; factors; indicators; Malmquist productivity index 1. Introduction It is generally accepted that the level of economic development is not uniform across territories. On the contrary, it substantially differs. This plays an essential role in many research studies that sought to assign an appropriate evaluation of economic and social development in the European area (e.g., Balcerowicz et al. 2013;Easterly and Levine 2012;Watt and Botsch 2010;Ghosh et al. 2009). As human activities are related to economic development and affected by territorial development, the way of measurement of the conditions of national development is essential in the determination of a country’s socio-economic policies (Halkos and Tzeremes 2005). The issue of socio-economic advancement, as well as disparities of territories, is closely linked to the setting and evaluation of competitiveness (Gardiner et al. 2004;Lukovics 2009;Ocubo 2012). The pursuit and the promotion of competitiveness increasingly shape the dynamics of economic, social, political, and cultural change in the contemporary world. The economy’s entry into the globalization phase radically altered the nature of competition. Numerous new actors from every market in the world are simultaneously in competition on every market. This new competition J. Risk Financial Manag. 2019,12, 72; doi:10.3390/jrfm12020072 www.mdpi.com/journal/jrfm J. Risk Financial Manag. 2019,12, 72 2 of 34 accentuated the interdependence of the different levels of globalization. Globalization obliged all countries to raise their standards of economic efficiency, resulting in a growing interest in and concern about competitiveness; nations, regions, and cities have no option but to strive to be competitive in order to survive in the new global marketplace and the “new competition” being forged by the further information or knowledge-driven economy (Gardiner et al. 2004). Policy-makers at all levels are being swept up in this competitiveness fever. This growing interest may perhaps be partly attributable to their awareness of the fact that all countries have to contend with raised standards of economic efficiency as a result of the globalization of goods and factor markets. The economy may be competitive, but if the society and the environment suffer too much, the country will face significant difficulties and vice versa. Therefore, governments, in the long run, cannot focus alone on the economic competitiveness of their country; instead, they need an integrated approach to govern the country. The complexity of competitiveness, decomposed by Esser et al. (1995), in the view of efficiency analysis is used in this paper—every country has standard features which affect and drive the competitiveness of all the entities located there, even if the variability of competitiveness level of the entities within the country may be very high. In the European Union (EU), the process of achieving an increasing trend and a higher level of competitiveness is significantly complicated by the heterogeneity of countries and regions in many areas. Although the EU is one of the most developed parts of the world with high living standards, there exist significant and substantial economic, social, and territorial disparities influencing a level of worldwide production and efficiency; so far, the EU competitiveness stands as a global player in the world economy. Considering the increasing importance of economic growth in the society and competitive world, evaluation of territorial performance is remarkably considered, and various measures are brought up as criteria in the assessment of territorial performance. The EU competitiveness depends on a multiplicity of actions that can optimize the potentials within its countries. All EU member states possess development opportunities; however, enough use of these options will increase the competitiveness of the EU countries and, thus, they must be efficient enough. From this point of view, the purpose of the paper is to achieve a more detailed productivity analysis and assessment of EU28 countries based on the concept of country competitiveness index (Annoni and Kozovska 2010;Annoni and Dijkstra 2013;Annoni et al. 2017) using a multivariate method of factor analysis (FA), identifying the main factors of socioeconomic development determining the competitiveness level of European countries. These factors of competitiveness are used for further productivity score evaluation performed using an advanced data envelopment analysis (DEA) approach—Malmquist productivity index (MPI) (Färe et al. 1994a,1994b) in the reference period 2000–2017. The application of MPI allows providing an efficiency analysis of EU member states in three selected periods, 2000–2007 (pre-crisis period), 2008–2011 (crisis period), and 2012–2017 (post-crisis period), concerning the internal and external assumption for their economic growth and competitive position. 2. Theoretical Background At a time when the EU member states have to deal with increased pressure on public balances, stemming from demographic trends and globalization, the improvement of the efficiency and effectiveness of public spending features high on the political agenda. The current economic situation determined by persisting effects of the crisis is causing the governments of countries worldwide to streamline their processes in terms of collecting revenue from the state budget and then redistributing it on the principle of performance and economic efficiency. Therefore, this resulted in the fact that markets provided by developed countries will be more critical for developing countries and their trade practices, as well as commercial practices of national or/and private companies (MacGregor Pelik á nov á 2017). Comparative analysis of efficiency in the public sector is, thus, a starting point for studying the role of efficiency, effectiveness, and total performance regarding economic governance of resource utilization by general management for achieving medium/long-term objectives of economic recovery and sustainable development of national economies (Mihaiu et al. 2010). J. Risk Financial Manag. 2019,12, 72 3 of 34 The analysis of efficiency and effectiveness is about the relationships between inputs (entries), outputs (results), and outcomes (effects). Farrell (1957) already investigated the question of how to measureefficiencyandhighlighteditsrelevanceforeconomicpolicy-makers. Since thattime, techniques to measure efficiency improved, and investigations of efficiency are more frequent. Nevertheless, the measurement of efficiency and effectiveness of countries remains a conceptual challenge. Problems arise because public spending has multiple objectives and because public sector outputs are often not sold on the market, which implies that price data are not available and that the output cannot be quantified (Mandl et al. 2008). Efficiency is, thus, a central issue in analyses of economic growth, the effects of fiscal policies, the pricing of capital assets, the level of investments, the technology changes and production technology, and other economic topics and indicators. Efficiency can be achieved under the conditions of maximizing the results of action about the resources used, and it is calculated by comparing the effects obtained by their efforts. In a competitive economy, therefore, the issue of efficiency, particularly dynamic efficiency, can be resolved by comparing these economic issues. The ratio of inputs to outputs gives the efficiency, but there is a difference between the technical efficiency and the allocative efficiency. The technical efficiency implies a relationship between inputs and outputs on the frontier production curve; however, not any form of technical efficiency makes sense in economic terms, and this efficiency is captured through the allocative efficiency that requires a cost/benefit ratio. The effectiveness, in terms of this meaning, implies a relationship between outputs and outcomes. In this sense, the distinction between the output and the outcome must be made. The outcome is often linked to welfare or growth objectives and, therefore, may be influenced by multiple factors (including outputs, as well as exogenous “environment” factors). The effectiveness is, thus, more challenging to assess than efficiency, since the outcome is influenced by political choice. There are thus three key topics for the article concept: competitiveness–productivity–stage of development, and their interdependence is as follows, resp. for the logical interconnection of theoretical and empirical part see Figure 1. J. Risk Financial Manag. 2019, 6, x FOR PEER REVIEW 3 of 35 The analysis of efficiency and effectiveness is about the relationships between inputs (entries), outputs (results), and outcomes (effects). Farrell (1957) already investigated the question of how to measure efficiency and highlighted its relevance for economic policy-makers. Since that time, techniques to measure efficiency improved, and investigations of efficiency are more frequent. Nevertheless, the measurement of efficiency and effectiveness of countries remains a conceptual challenge. Problems arise because public spending has multiple objectives and because public sector outputs are often not sold on the market, which implies that price data are not available and that the output cannot be quantified (Mandl et al. 2008). Efficiency is, thus, a central issue in analyses of economic growth, the effects of fiscal policies, the pricing of capital assets, the level of investments, the technology changes and production technology, and other economic topics and indicators. Efficiency can be achieved under the conditions of maximizing the results of action about the resources used, and it is calculated by comparing the effects obtained by their efforts. In a competitive economy, therefore, the issue of efficiency, particularly dynamic efficiency, can be resolved by comparing these economic issues. The ratio of inputs to outputs gives the efficiency, but there is a difference between the technical efficiency and the allocative efficiency. The technical efficiency implies a relationship between inputs and outputs on the frontier production curve; however, not any form of technical efficiency makes sense in economic terms, and this efficiency is captured through the allocative efficiency that requires a cost/benefit ratio. The effectiveness, in terms of this meaning, implies a relationship between outputs and outcomes. In this sense, the distinction between the output and the outcome must be made. The outcome is often linked to welfare or growth objectives and, therefore, may be influenced by multiple factors (including outputs, as well as exogenous “environment” factors). The effectiveness is, thus, more challenging to assess than efficiency, since the outcome is influenced by political choice. There are thus three key topics for the article concept: competitiveness–productivity– stage of development, and their interdependence is as follows, resp. for the logical interconnection of theoretical and empirical part see Figure 1. Figure 1. The relationship between the efficiency and the effectiveness impacting competitiveness (source: Mandl et al. 2008; own extension and elaboration). Figure 1. The relationship between the efficiency and the effectiveness impacting competitiveness (source: Mandl et al. 2008; own extension and elaboration). J. Risk Financial Manag. 2019,12, 72 4 of 34 Drucker (2001) believes that there is no efficiency without effectiveness because it is more important to do well what you proposed (the effectiveness) than do well something else that was not necessarily concerned. The relationship between efficiency and effectiveness is that of a part to the whole; the effectiveness is a necessary condition for achieving efficiency. This implies that efficiency and effectiveness are not always easy to isolate. 3. Materials and Methods The most common quantitative methods convenient for a high number of multivariate measured variablescanbeidentifiedasmultivariatestatisticalmethods. Multivariateanalysisisanever-expanding set of techniques for data analysis, encompassing a wide range of possible research situations (Hair et al. 2009). Between collections of multivariate statistical methods, we can include, e.g., principal component analysis, factor analysis, cluster analysis, or data envelopment analysis. 3.1. Factor Analysis Many scientific studies feature the fact that “numerous variables are used to characterize objects”. Because of these big numbers of variables that are in play, the study can become rather complicated. Moreover, it could well be that some of the variables measure different aspects of the same underlying variable. For situations such as these, factor analysis (FA) was invented. FA is the statistical approach that can be used to analyze interrelationships among a large number of variables and to explain these variables in terms of their standard underlying dimensions, i.e., factors. The main applications of FA techniques are, thus, to reduce the number of variables and to detect structure in the relationships among variables, so as to classify variables. The objective of FA is to reduce the number of variables by grouping them into a smaller set of factors; for this purpose, FA is applied in the paper. FA is a collection of methods for investigating whether some variables of interest (Y 1 ,Y 2 , . . . ,Y n ) are linearly related to a smaller number of unobservable factors (F 1 ,F 2 , . . . ,F k ). If we suggest that one measured variable, Y 1 , is a function of two underlying factors, F 1 and F 2 , then it is assumed that Y variable is linearly related to the two factors F, as follows (Hair et al. 2009): Y1=β10 +β11F1+β12F2+e1. (1) The error terms e 1 serves to indicate that the hypothesized relationships are not exact. In the specialized vocabulary of FA, the parameters βi,j are referred to as loadings, e.g., β12 is called the loading of variable Y1on factor F2. Why carry out factor analyses? If we can summarize a multitude of measurements with a smaller number of factors without losing too much information, we achieve some economy of description, which is one of the goals of scientific investigation. It is also possible that FA will allow us to test theories involving variables, which are hard to measure directly. Finally, at a more prosaic level, FA can help us establish that sets of questionnaire items (observed variables) are in fact all measuring the same underlying factor (perhaps with varying reliability) and, hence, can be combined to form a more reliable measure of that factor. There are some different varieties of FA (Stevens 1986). For an elaboration of FA, the software IBM SPSS Statistics 25 is used in the paper. 3.2. DEA-Based Malmquist Productivity Index Charnes et al. (1978) first proposed data envelopment analysis (DEA). Since DEA was first introduced, researchers in some fields quickly recognized that it is an excellent and easily used methodology for modeling operational processes for efficiency evaluations, accompanied by other developments. There are several researchers which also employed the DEA method in the context of studies about a country’s macroeconomy and Knowledge-based economies (KBE) (see Appendix A); Meleck ý (2018) and Stan í ˇckov á (2017) also consider DEA as a convenient tool for measuring efficiency as a mirror of national and regional competitiveness. Several studies using the DEA approach also focused J. Risk Financial Manag. 2019,12, 72 5 of 34 its attention on efficiency analysis in the context of EU member states in research and development (Conte et al. 2009). DEA is based on the simple Farrell model (Farrell 1957) for measuring the efficiency of decision-making units (DMUs) with one input and one output. This method was initially expanded in 1978 by Charnes, Cooper, and Rhodes (CCR model) assuming constant returns to scale (CRS), and it was later modified in 1984 by Banker, Charnes, and Cooper (BCC model) assuming variable returns to scale (VRS). DEA methods also include advanced additive models, such as the slack-based model (SBM) performed by Tone in 2002 or free disposal hull (FDH) and free replicability hull (FRH) models that were firstly formulated in 1984 by Deprins, Simar, and Tulkens. In recent years, research efforts focused on the investigation of the causes of productivity change and its decomposition. Malmquist productivity index (MPI) became the standard approach in productivity measurement over time within nonparametric research. MPI was introduced firstly by Caves et al. (1982). Färe et al. (1994a,1994b) defined and applied an input-oriented productivity index as the geometric mean of the two MPIs developed by Caves et al. Although it was developed in a consumer context, MPI recently enjoyed widespread use in a production context. MPI can be used to construct indexes of input, output, or productivity, as ratios of input or output distance functions. There are various methods for measuring distance functions, and the most famous one is the linear programming method. MPI allows measuring of total productivity using distance-function calculation, which can be estimated from a solution of mathematical programming problems of the DEA kind. With respect to the nonparametric approach, it is worth mentioning differences between parametric and nonparametric methods in statistics, especially concerning the fact that we use several descriptive statistics in the paper. Methods are classified by what we know about the population we are studying. Parametric methods are typically the first methods studied in an introductory statistics course. The basic idea is that there is a set of fixed parameters that determine a probability model. Parametric methods are often those for which we know that the population is approximately normal, or we can approximate using a normal distribution after we invoke the central limit theorem. There are two parameters for a normal distribution: mean and standard deviation. To contrast with parametric methods, we define nonparametric methods. These are statistical techniques for which we do not have to make any assumption of parameters for the population we are studying. Indeed, the methods do not have any dependence on the population of interest. The set of parameters is no longer fixed, and neither is the distribution that we use. It is for this reason that nonparametric methods are also referred to as distribution-free methods. Nonparametric methods are growing in popularity and influence for some reasons. The main reason is that we are not constrained as much as when we use a parametric approach. We do not need to make as many assumptions about the population that we are working with as what we have to make with a parametric approach. Many of these nonparametric methods are easy to apply and to understand. It is safe to say that most people who use statistics are more familiar with parametric analyses than nonparametric analyses. What is the comparison of both methods? There are multiple ways to use statistics to find a confidence interval about a mean. The parametric method would involve the calculation of a margin of error with a formula, and the estimation of the population mean with a sample mean. The nonparametric method to calculate confidence mean would involve the use of bootstrapping. Why do we need both parametric and nonparametric methods for this type of problem? Many times, parametric methods are more efficient than the corresponding nonparametric methods. Although this difference in efficiency is typically not that much of an issue, there are instances where we do need to consider which method is more efficient. Concerning statistical error, the difference lies in the fact that nonparametric methods (data envelopment analysis, DEA) use optimization to solve statistical errors, and parametric methods (stochastic frontier analysis, SFA) use econometrics to resolve statistical errors. As mentioned above, empirical analysis is based on a frontier nonparametric approach and aims to study productivity growth and efficiency. This part of the analysis is based on MPI for measuring the change of technical efficiency and the movement of the frontier in terms of individual DMUs J. Risk Financial Manag. 2019,12, 72 6 of 34 (Färe et al. 1994a,1994b). Suppose we have a production function in period tas well as period t+1. MPI calculation requires two single-period and two mixed-period measures. The two single-period measures can be obtained using the CCR CRS model. For simplicity of MPI calculation, it is presented as a basic DEA model based on the assumption of single input/output. With regard to the selected type of DEA model and its assumptions, it is still appropriate to mention here the general way of model selection. The classic input-oriented DEA model can be specified under the condition that the production function has constant returns to scale (CRS). If the production function has variable returns to scale (VRS), there may be returns to scale (RTS) described as increasing (IRTS) or decreasing (DRTS). The selection of the DEA model based on RTS can be assessed according to three methods, as specified Seiford and Zhu (1999), i.e., CCR RTS method, BCC RTS method, and scale efficiency index method. For evaluation of territorial efficiency, DEA in the form of the CRS model is often used, e.g., Lacko and Hajduová(2018), Makridou et al. (2014), Otsuka (2014), or Malhotra and Malhotra (2006). Suppose each DMU j (j=1, 2, . . . n) produces a vector of output yt j=yt 1j,. . . ,yt sj by using a vector of inputs xt j=xt 1j,. . . ,xt mj at each period t,t=1, . . . ,T. From time tto time t+1, DMU 00 s efficiency may change and/or the frontier may shift. MPI is calculated via Equation (2) comparing xt 0 to the frontier at time t, i.e., calculating θt 0xt 0,yt 0 in the following input-oriented IO CCR CRS envelopment model (Zhu 2011): θt 0xt 0,yt 0=minθ0, (2) subject to n P j=1λjxt j≤θ0xt 0, n P j=1λjyt j≥yt 0, λj≥0, j=1,. . . ,n, where θ0 indicatestheefficiencyscoreofobservedDMU 0 , and xt 0=xt 10,. . . ,xt m0 and yt 0=yt 10,. . . ,yt s0 are the input and the output vectors of DMU0among others. MPI is further calculated via Equation (3) comparing xt+1 0 to the frontier at time t+1, i.e., calculating θt+1 0xt+1 0,yt+1 0 in the following input-oriented CCR CRS envelopment model (Zhu 2011): θt+1 0xt+1 0,yt+1 0=minθ0, (3) subject to n P j=1λjxt+1 j≤θ0xt+1 0, n P j=1λjyt+1 j≥yt+1 0, λj≥0, j=1,. . . ,n. MPI is further calculated via Equation (4) comparing xt 0 to the frontier at time t+1, i.e., calculating θt+1 0xt 0,yt 0via the following linear program (Zhu 2011): θt+1 0xt 0,yt 0=minθ0, (4) subject to n P j=1λjxt+1 j≤θ0xt 0, n P j=1λjxt+1 j≥yt 0, λj≥0, j=1,. . . ,n. J. Risk Financial Manag. 2019,12, 72 7 of 34 MPI is further calculated via Equation (5) comparing xt+1 0 to the frontier at time t, i.e., calculating θt 0xt+1 0,yt+1 0via the following linear program (Zhu 2011): θt 0xt+1 0,yt+1 0=minθ0, (5) subject to n P j=1λjxt j≤θ0xt+1 0, n P j=1λjyt j≥yt+1 0, λj≥0, j=1,. . . ,n. MPI measuring the efficiency change of production units between successive periods tand t+1, is formulated via Equation (6). M0(xt+1,yt+1,xt,yt) = ECH0·FS0, (6) where ECH 0 is the change in the relative efficiency of DMU 0 about other units (i.e., due to the production possibility frontier) between periods tand t+1. FS 0 describes the change in the production possibility frontier as a result of the technology development between periods tand t+1. The formulation of MPI in Equation (7) makes it possible to measure the change of technical efficiency and the movement of the frontier in terms of a specific DMU0(Zhu 2011). M0=θt 0 θt+1 0xt+1 0,yt+1 0        θt+1 0xt+1 0,yt+1 0 θt 0xt+1 0,yt+1 0· θt+1 0xt 0,yt 0 θt 0xt 0,yt 0        1 2 . (7) The first component on the right-hand side measures the magnitude of technical efficiency change between periods tand t+1, indicating that technical efficiency improves, remains, or declines. The second term measures the shift in the possibility frontier, i.e., technology frontier shift, between periods tand t+1. Trends in MPI, ECH, and FS are illustrated in Table 1. Table 1. Trends in Malmquist productivity index (MPI) and its components (source: Zhu 2011). MPI Productivity ECH FS MPI >1 Improving Change >1, improving Change >1, improving MPI =1 Unchanging Change =1, unchanging Change =1, unchanging MPI <1 Declining Change <1, declining Change <1, declining MPI—malmquist productivity index; ECH—change in relative efficiency; FS—change in production possibility frontier. DEA is a popular method for general business management because it has a number of advantages: (1) it can evaluate a DMU’s performance with multiple inputs and multiple outputs (fulfilling the criteria of our dataset, i.e., many input and output factors based on the number of numerous initial indicators); (2) it allows the units of input and output variables to be different (again, this criterion meets the paper outline, where the dataset represents various aspects of competitiveness on both side of input and output indicators); and (3) it is not necessary to know the type of production function in advance. However, DEA also has several limitations: (1) the DMUs must be homogeneous (in our case, the criterion of homogeneity represent 28 countries of the EU); (2) to obtain the best results, the number of DMUs must be at least twice the total number of input and output variables (this condition is fulfilled as the following paragraph and equations explain); and (3) isotonicity must exist, that is, the output must not decrease while the input increases (this condition is met as confirmed in the following paragraph). J. Risk Financial Manag. 2019,12, 72 8 of 34 If a performance measure (input/output) is added or deleted from consideration, it will influence the relative efficiencies. Empirically, when the number of performance measures is high in comparison with the number of DMUs, then most of the DMUs are evaluated efficiently. Hence, the obtained results are not reliable. There is a rule of thumb proposed by Cooper et al. (2007) which expresses the relationship between the number of DMUs and the number of performance measures. Toloo et al. (2015) checked more than 40 papers that contain practical applications and, statistically, they found out that, in nearly all of the cases, the number of inputs and outputs did not exceed six. Suppose there are nDMUs which consume minputs to produce soutputs. A simple calculation shows that when m ≤ 6 and s≤6, then 3 (m+s)≥m×s. As a result, in this paper, the following formula is applied: n≥3(m+s). (8) In the article, this rule is met, i.e., the number of DMUs is three times higher than the sum of inputs and outputs, i.e., 28 ≥3(6 +3). In this section, we check the validity of the model in terms of the model specification and the existence of potential outliers in the sample. Firstly, we introduce the isotonicity test for checking the validity of the model specification. Specifically, we checked whether an increase in input indicators brought growth in outputs rather than a decrease in outputs (see Avkiran 2006;Adusei 2016; Hwang et al. 2018;Jiang and He 2018 ). Input data for the DEA model must meet the isotonicity criteria, i.e., the level of outputs is at least the same, and does not fall when inputs increase. More specifically, the requirement is that the relationship between inputs and outputs is not erratic. Increasing the value of any input while keeping other factors constant should not decrease any output but should instead lead to an increase in the value of at least one output. By calculating the correlations between the input and output variables, we found that, if the pairwise correlation is statistically significant at 5% level of significance, the correlation moves in the range from 0.49 to 0.69 with one exception. Only in the case of input factor 2 (level of infrastructure) and output factor 3 (labor market) was there a correlation (−0.13). These results of the isotonicity test justify the selection of variables. Secondly, we conducted outlier detection with the idea of a scatter matrix, especially using boxplots and the number of extreme outliers. Based on the assumptions for principal component analysis (PCA), we use standardized variables for the extraction of rotated factors using SPSS Statistics, which recommends determining extreme outliers as component scores with values out of interval (quartile 1 – 3 × IQR; quartile 3 +3 × IQR). According to Chandola et al. (2009) or Jiang and He (2018), this methodology is simple and widely used. We performed the procedure and calculated the accumulated times that the data of a country are considered to be an outlier with the value of component scores out of interval (quartile 1 − 1.5 × IQR; quartile 3 +1.5 × IQR). The results show that there are outliers, but they are more or less exceptions. In the case of input factors, we found that Germany and the United Kingdom in factor 2 (infrastructure), Malta in factor 5 (participation in education), and Bulgaria in factor 6 (expenditure on education and civilization diseases) outperformed the other countries during some but not all years of the reference period 2000–2017. In the case of output factors, we found that only Spain in factor 2 (knowledge-based economy) outperformed the other countries during some but not all years of the reference period 2000–2017. In the case of outliers, the DEA method might be inconsistent. Because these countries were outliers and not extreme outliers, present only in some input or output factors and not in all years of the reference period, they were left within the framework of the evaluation. For the solution of the DEA method, a software tool based on solving linear programming problems is used in the paper—Solver in MS Excel 2016, similar to DEA Frontier. 4. Results The empirical analysis starts by building a database of indicators that are part of the country competitiveness index (CCI) approach created by Annoni and Kozovska (2010) in 2010, and then J. Risk Financial Manag. 2019,12, 72 9 of 34 updated by Annoni and Dijkstra (2013) and Annoni et al. (2017). CCI also has its dimension in the regional competitiveness index (RCI). The roots of CCI/RCI lay in the most known competitiveness indicator, the global competitiveness index reported by the World Economic Forum. Pillars of CCI/RCI are grouped according to the different dimensions (input versus output aspects) of national competitiveness they describe. The terms “inputs” and “outputs” are meant to classify pillars into those which describe driving forces of competitiveness, in terms of long-term potentiality, and those which are direct or indirect outcomes of a competitive society and economy Annoni and Kozovska (2010). The CCI/RCI data file consisted of 66 indicators in 2010, 73 indicators in 2013, and 74 indicators in 2016; however, not all indicators are used in the paper because of a lack of data for every country within EU28—15 countries are classified as old EU member states (origin countries from 1957 and countries joining the European community in 1973, 1981, 1986, and 1995), and 13 countries belong to the group of new EU member states (joining the EU in 2004, 2007, and 2013). Some indicators are excluded from analysis because of a lack of data for many of countries and periods; from this point of view, only 61 indicators are used in the paper—37 represent inputs, and 24 represent outputs (see Table 2). Related tothe issueof thenatureof thedataset andindividual indicators, theused databaseincludes quantitative (numerical) indicators with the exact measured values, and not qualitative (categorical) indicators. The data source for downloading these indicators was the European Statistical Office (Eurostat). Table 2. Country competitiveness index (CCI) indicators in input and output dimensions * (source: own elaboration). Dimension Pillar Indicator of Input Input Institution Political stability (PS), voice and accountability (VA), government effectiveness (GE), regulatory quality (RQ), rule of law (RL), control of corruption (CC) Macroeconomic stability Harmonized index of consumer prices (HICP), gross fixed capital formation (GFCF), income, saving, and net lending/net borrowing (ISLB), total intramural research and development expenditure (GERD), labor productivity per person employed (LPPE) Infrastructure Railway transport—length of tracks (RTLT), air transport of passengers (ATP), volume of passenger transport (VPT), volume of freight transport (VFT), motorway transport—length of motorways (MTLM), air transport of freight (ATF) Health Healthy life expectancy (HLE), infant mortality rate (IMR), cancer disease death rate (CDDR), heart disease death rate (HDDR), suicide death rate (SDR), hospital beds (HB), road fatalities (RF) Primary, secondary and tertiary education; training and lifelong learning Mathematics, science, and technology enrolments and graduates (MSTEG), pupils to teachers ratio (PTR), financial aid to students (FAS), total public expenditure at primary level of education (TPEPLE), total public expenditure at secondary level of education (TPESLE), total public expenditure at tertiary level of education (TPETLE), participants in early education (PEE), participation in higher education (PHE), early leavers from education and training (ELET), accessibility to universities (AU), lifelong learning—participation in education and training (LLPET) Indicators for technological readiness Level of internet access (LIA), E-government availability (EA) J. Risk Financial Manag. 2019,12, 72 16 of 34 Table 7. MPI descriptive statistics (source: own calculation and elaboration). Statistics Period 2000–2007 2008–2011 2012–2017 MPI ECH FS MPI ECH FS MPI ECH FS NValid 28 28 28 28 28 28 28 28 28 0 Missing 0 0 0 0 0 0 0 0 0 Mean 0.97235 0.99416 0.97789 0.99377 1.00394 0.98987 0.99961 0.99965 0.99996 SD 0.159261 0.027815 0.155983 0.135679 0.013288 0.135008 0.027900 0.005487 0.027283 Variance 0.025 0.001 0.024 0.018 0.000 0.018 0.001 0.000 0.001 Range 0.867 0.156 0.867 0.844 0.070 0.844 0.165 0.028 0.165 Minimum 0.785 0.885 0.785 0.396 0.991 0.396 0.869 0.986 0.869 Maximum 1.653 1.041 1.653 1.240 1.061 1.240 1.033 1.014 1.033 In Tables 8–10, the MPI results for periods 2000–2007, 2008–2011, and 2012–2017 are outlined, including information about the number of evaluated DMUs (the first column), codes of EU28 member states (the second column), efficiency scores of MPI (the third column), scores of efficiency change (the fourth column), scores of frontier shift (the fifth column), rank of EU28 member states based on MPI (the sixth, seventh, and eighth columns), and groups of countries (the ninth column). In Tables 8–10, results of the MPI and its dimensions are highlighted using the traffic light method. The range of colors of this method changes from dark to light shades of gray. Countries with the highest values of the MPI, catch-up, and frontier shift suggest a better level of efficiency and, thus, competitiveness; they are highlighted by dark shades of gray—the higher the value is, the darker the shade of gray is. On the contrary, countries with the lowest values of the MPI and its two dimensions (catch-up and frontier shift) suggest a worse level of efficiency; they are highlighted by light shades of gray—the lower the value is, the lighter the shade of gray is. Countries with values of the MPI falling between efficient (dark shades of gray) and inefficient (light shades of grey color) are highlighted by medium shades of gray. Table 8. MPI results for period 2000–2007 (source: own calculation and elaboration). No. DMUs IO CRS MPI Efficiency Change Frontier Shift Rank Group of Countries 1 BE 0.972 1.000 0.972 1. MT 1.653 1st (3 EU15, 5 EU13) 2 BG 0.785 1.000 0.785 2. HR 1.138 3 CZ 0.903 0.994 0.908 3. PT 1.109 4 DK 0.939 1.000 0.939 4. RO 1.097 5 DE 0.880 1.000 0.880 5. IE 1.055 6 EE 0.928 1.013 0.915 6. LT 1.049 7 IE 1.055 1.000 1.055 7. FI 1.036 8 EL 0.948 0.978 0.970 8. SK 1.010 9 ES 0.858 1.000 0.858 9. AT 0.979 2nd (7 EU15, 6 EU13) 10 FR 0.888 1.000 0.888 10. BE 0.972 11 IT 0.892 1.000 0.892 11. LU 0.970 12 CY 0.907 1.000 0.907 12. EL 0.948 13 LV 0.904 0.996 0.907 13. DK 0.939 14 LT 1.049 1.000 1.049 14. EE 0.928 15 LU 0.970 1.000 0.970 15. HU 0.927 16 HU 0.927 1.000 0.927 16. PL 0.920 17 MT 1.653 1.000 1.653 17. SE 0.919 18 NL 0.906 1.000 0.906 18. CY 0.907 19 AT 0.979 1.041 0.940 19. NL 0.906 20 PL 0.920 0.885 1.040 20. LV 0.904 21 PT 1.109 1.030 1.077 21. CZ 0.903 J. Risk Financial Manag. 2019,12, 72 17 of 34 Table 8. Cont. No. DMUs IO CRS MPI Efficiency Change Frontier Shift Rank Group of Countries 22 RO 1.097 1.000 1.097 22. IT 0.892 3rd (5 EU15, 1 EU13) 23 SI 0.826 0.940 0.879 23 FR 0.888 24 SK 1.010 1.000 1.010 24. DE 0.880 25 FI 1.036 1.000 1.036 25. ES 0.858 26 SE 0.919 1.000 0.919 26. UK 0.827 27 UK 0.827 0.959 0.863 27. SI 0.826 28 HR 1.138 1.000 1.138 28. BG 0.785 4th (1 EU13) Note: Belgium (BE), Bulgaria (BG), Czech Republic (CZ), Denmark (DK), Germany (DE), Estonia (EE), Ireland (IE), Greece (EL), Spain (ES), France (FR), Italy (IT), Cyprus (CY), Latvia (LV), Lithuania (LT), Luxembourg (LU), Hungary (HU), Malta (MT), Netherlands (NL), Austria (AT), Poland (PL), Portugal (PT), Romania (RO), Slovenia (SI), Slovakia (SK), Finland (FI), Sweden (SE), United Kingdom (UK), Croatia (CR). Table 9. MPI results for period 2008–2011 (source: own calculation and elaboration). No. DMUs IO CRS MPI Efficiency Change Frontier Shift Rank Group of Countries 1 BE 0.989 1.000 0.989 1. MT 1.240 1st (1 EU13) 2 BG 0.396 1.000 0.396 2. CY 1.120 2nd (1 EU13) 3 CZ 1.034 1.000 1.034 3. PT 1.075 3rd (8 EU15, 8 EU13) 4 DK 0.987 1.000 0.987 4. NL 1.065 5 DE 1.021 1.000 1.021 5. LU 1.065 6 EE 1.013 1.061 0.954 6. AT 1.062 7 IE 0.905 1.000 0.905 7. LT 1.042 8 EL 0.942 1.000 0.942 8. SI 1.039 9 ES 1.035 1.000 1.035 9. ES 1.035 10 FR 0.975 1.000 0.975 10. CZ 1.034 11 IT 1.002 1.000 1.002 11. RO 1.029 12 CY 1.120 1.000 1.120 12. PL 1.021 13 LV 0.948 1.000 0.948 13. DE 1.021 14 LT 1.042 1.000 1.042 14. HR 1.014 15 LU 1.065 1.000 1.065 15. EE 1.013 16 HU 0.875 1.000 0.875 16. SK 1.010 17 MT 1.240 1.000 1.240 17. FI 1.003 18 NL 1.065 1.000 1.065 18. IT 1.002 19 AT 1.062 1.000 1.062 19. BE 0.989 4th (7 EU15, 1 EU13) 20 PL 1.021 1.008 1.013 20. DK 0.987 21 PT 1.075 1.030 1.044 21. SE 0.982 22 RO 1.029 1.000 1.029 22. FR 0.975 23 SI 1.039 1.000 1.039 23 LV 0.948 24 SK 1.010 1.000 1.010 24. EL 0.942 25 FI 1.003 1.000 1.003 25. UK 0.938 26 SE 0.982 1.000 0.982 26. IE 0.905 27 UK 0.938 0.991 0.947 27. HU 0.875 5th (1 EU13) 28 HR 1.014 1.020 0.993 28. BG 0.396 6th (1 EU13) J. Risk Financial Manag. 2019,12, 72 18 of 34 Table 10. MPI results for period 2012–2017 (source: own calculation and elaboration). No. DMUs IO CRS MPI Efficiency Change Frontier Shift Rank Group of Countries 1 BE 1.005 1.000 1.005 1. EL 1.033 1st (12 EU15, 7 EU13) 2 BG 0.869 1.000 0.869 2. EE 1.020 3 CZ 0.993 1.000 0.993 3. RO 1.019 4 DK 1.004 1.000 1.004 4. UK 1.016 5 DE 0.996 1.000 0.996 5. IE 1.015 6 EE 1.020 1.014 1.007 6. PL 1.015 7 IE 1.015 1.000 1.015 7. ES 1.012 8 EL 1.033 1.000 1.033 8. LU 1.011 9 ES 1.012 1.000 1.012 9. FI 1.009 10 FR 1.002 1.000 1.002 10. HU 1.008 11 IT 1.002 1.000 1.002 11. BE 1.005 12 CY 0.988 0.986 1.002 12. SK 1.005 13 LV 0.989 1.000 0.989 13. SE 1.004 14 LT 1.001 1.000 1.001 14. DK 1.004 15 LU 1.011 1.000 1.011 15. MT 1.004 16 HU 1.008 1.000 1.008 16. NL 1.003 17 MT 1.004 1.000 1.004 17. FR 1.002 18 NL 1.003 1.000 1.003 18. IT 1.002 19 AT 0.997 1.000 0.997 19. LT 1.001 20 PL 1.015 1.014 1.001 20. AT 0.997 2nd (3 EU15, 5 EU13) 21 PT 0.985 0.988 0.997 21. DE 0.996 22 RO 1.019 1.000 1.019 22. HR 0.996 23 SI 0.991 1.000 0.991 23 CZ 0.993 24 SK 1.005 1.000 1.005 24. SI 0.991 25 FI 1.009 1.000 1.009 25. LV 0.989 26 SE 1.004 0.999 1.005 26. CY 0.988 27 UK 1.016 1.000 1.016 27. PT 0.985 28 HR 0.996 0.991 1.006 28. BG 0.869 3rd (1 EU13) Broader aspects enter the overall evaluation of economics, and these aspects are unnoticeable for DEA, i.e., parts of the qualitative assessment in line with the evaluation of overall performance. Performance is linked concerning competitiveness; a good performance in the innovation group is expected to also be a good performance in the efficiency and the basic groups as they are instrumental in increasing levels of competitiveness. As countries move along the path of development, their socio-economic conditions change, and different determinants become more important for the national level of competitiveness. As a result, the best way to improve the competitiveness of more developed countries will not necessarily coincide with the way to improve less developed countries. Consistent with the theory of economic growth and economic development, CCI results confirm that the most competitive countries are those with the highest level of economic development (for more information, see Annoni and Kozovska 2010;Annoni and Dijkstra 2013; or Annoni et al. 2017). It is striking that several of the top competitors are traditionally economically strong countries. At the end of the competitiveness scale, it is possible to find some countries which are unfortunately steadily the worst performers. These differences in CCI editions indicate that the EU moved far from a homogeneous entity in terms of competitiveness, but CCI results show a more polycentric pattern. Therefore, part of the explanation of inequalities among the EU member states has to do with differences in competitiveness. An economic entity with a low level of competitiveness may not have similar opportunities as a highly competitive economic entity. This fact remains and is confirmed. However, what does it mean for efficiency in competitiveness? In the case of efficiency analysis of competitiveness and in the time comparison analysis of change, the results are just a little bit different. Why? The concept of competitiveness may then be necessary not only to evaluate why some countries grow faster than J. Risk Financial Manag. 2019,12, 72 19 of 34 others, but also why some countries have a better and more efficient distribution of competitiveness over time than others. Is a high level of competitiveness necessarily associated with a high level of efficiency, and vice versa? It may not always be the case because evaluated countries have a lower level of input; these countries were able to achieve competitiveness at the level of CCI. While the CCI value may not be high in the less competitive countries, it is necessary to compare the values of inputs and outputs. In DEA efficiency analysis, although the IO CRS MPI value is not so high, overall, it is possible to state that the country operates more efficiently at the end than at the beginning of the reference period. Such a conclusion is relevant by comparing values of inputs and outputs, and the fact that outputs are achieved with given inputs. More specifically, based on MPI results in periods 2000–2007, 2008–2011, and 2012–2017, it is important to notice that many European countries achieved a value of MPI higher than 1.000 and, thus, productivity is increasing. As mentioned above, part of the explanation of the large inequalities within EU countries is linked with the differences in competitiveness. Finally, Tables 8–10 show reordered countries, from best to worst, their MPI score, and the corresponding rank. The results state positive trends within the community of EU member states. Based on the MPI results, it is clear that the best efficiency changes in competitiveness comparing reference years were achieved by countries belonging to the group of EU13 countries, i.e., new EU member states, than in the case of countries belonging to the group of EU15 countries, i.e., the old EU member states. This fact is not surprising, because it has the following key political implications with several reasons/factors: • The new EU member states constantly fall into the category of less developed and competitive states based on gross domestic product (GDP) per head in Purchasing Parity Standard (PPS), which is the reason for their inclusion in the appropriate categorization stage of development (see Annoni and Kozovska 2010;Annoni and Dijkstra 2013;orAnnoni et al. 2017); • The association of each country with the relevant stage of development testifies to its competitive advantages and disadvantages and determines its weaknesses. A medium stage of development is associated with economies primarily driven by factors such as lower skilled labor and basic infrastructures. Aspects related to good governance and quality of public health are considered basic inputs in this framework. An intermediate stage of development is characterized by labor market efficiency, quality of higher education, and market size, factors which contribute to a more sophisticated economy and more significant potential for competitiveness. In the high stage of development, factors related to innovation, business sophistication, and technological readiness are necessary inputs for innovation-driven economies (Annoni and Dijkstra 2013); • The threshold defining the level of GDP as a percentage of EU average was taken as a reference as it is the criterion for identifying countries and their regions eligible for funding under the established criteria of the EU regional policy framework. European funds are an essential tool for regional development and reducing economic, social, and territorial disparities among European countries and their regions. Reducing disparities have a significant impact on competitiveness, and these two concepts are, thus, the EU complementary objectives. Of the total budget allocated to regional policy, a substantial part goes just to the NUTS 2 regions of EU13 countries (i.e., the basic regions for the application of regional policies classify based on the EU Nomenclature of Territorial Units for Statistics), where development is significantly supported; • New EU member states are often considerably dependent on exports into the old EU member states and on the flow of money for this exchange shift. Theabove facts can raisethe questionofwhether the resultsautomaticallyprovidethe prerequisites for improving the development of the new EU member states. This is the question of the convergence process de jure and de facto. For the Baltic, Balkan, and central and eastern European countries, joining the EU held the implicit promise of economic convergence to Western European standards of living represented by the old EU member states. As officially stated by the European Commission (2019), this was true for both the first wave of eastern enlargement in 2004 and the subsequent accession of Bulgaria J. Risk Financial Manag. 2019,12, 72 20 of 34 and Romania in 2007 and finally Croatia in 2013. As of the 15th anniversary of the 2004 accession, this expectation was largely met; access to the European single market (i.e., internal market) created new business opportunities, triggered vast capital flowed to the new EU member states, and facilitated their integration into global supply chains. The catch-up process, thus, gained additional impetus during the accession talks and negotiation and again upon joining the EU. Although a significant gap remains today, it is shrinking at a rapid pace, highlighting central improvements among the new EU member states, as well as convergence of the group of EU13 countries to the group of EU15 countries in the following areas: income convergence; convergence in labor productivity; convergence in workforce; convergence in participation rates; convergence in educational attainment; convergence in competitiveness; convergence in quality of governance; convergence in research, development, and innovation; convergence in digital connectivity; convergence in openness to trade and integration into European supply chains; and convergence in openness to foreign direct investment. Figure 2constitute the box plots of all parts of MPI, i.e., MPI, ECH, and FS. Box plots of each MPI part show data skewness and kurtosis to mean values, reflected by the equal location of the median (X 50 ) between the upper (X 75 ) and lower (X 25 ) quartiles. In the cases of MPI, ECH, and FS, data are skewed to the upper levels—the median is shifted to the upper quartile (X 75 ). The shapes of box plots also indicate the symmetrical layout. Each box plot represents data from the normal distribution, not only due to its symmetry but also due to the position of the median, which lies almost in the middle of the rectangle. In the context of efficiency analysis, the outliers and extreme values are interesting, i.e., the highest or lowest values of MPI in comparison to values of MPI of other countries in the evaluated sample within the reference period. The current version of the EU has 28 member states and, therefore, we aim to have relevant analysis for all EU countries and not only the selected sample. Therefore, DMUs present the EU countries in the form of outliers, and extreme values are not excluded from our empirical analysis. Our analysis aims to have comprehensible results for the entire sample of countries and not a partial sample; we are concerned about the diversity that the EU is characterized by, as highlighted by its motto “unity in diversity”. The classification of EU15 and EU13 member states concerning the nature of technical and technological change is illustrated in Figure 3. In all reference periods, the location of all European countries is recorded concerning results of ECH and FS. Evaluated countries are divided into two groups (the EU15 member states and the EU13 member states) for a better comparison of common features and differences. It is convenient to remind the reader that ECH and FS values of 1.000 mean no productivity change, values higher than 1.000 mean that productivity is improving, and values lower than 1.000 mean that productivity is deteriorating. From this point of view, it is possible to divide European countries into four categories or quadrants. Via the illustration of Figure 2, information about differences in efficiency recorded by MPI among three reference periods is confirmed. Across the reference periods, most European countries are located in quadrants with a low level of FS, and a higher or lower level of ECH. It means that efficiency change is especially caused by the difference in the production possibility frontier because of the technology development between reference years, i.e., technology frontier shift. This fact is positive information concerning factors of competitiveness; it signifies that countries can utilize their internal factor endowment efficiently and can apply technological progress for boosting their competitive advantages, i.e., they contribute to qualitative-based economic growth, allowing raising the steady state. On the other side, some European countries are located in quadrants with a high level of FS, and a more upper or lower level of ECH. It means that efficiency change is due to a change in the relative efficiency of the evaluated country with respect to other countries, due to the production possibility frontier between reference years, i.e., technical efficiency change. This fact is not such positive information because it means that countries extract their efficiency based on shifts in sources of competitiveness, i.e., they make changes in composition and quantity of sources based on their exchange business with other countries. The characteristic of technical efficiency change, thus, contributes only to quantitative-based economic J. Risk Financial Manag. 2019,12, 72 21 of 34 growth, which has its limits; this is disconcerting concerning limited sources, utilization of sources, and possibility/impossibility of their recovery. J. Risk Financial Manag. 2019, 6, x FOR PEER REVIEW 20 of 35 comprehensible results for the entire sample of countries and not a partial sample; we are concerned about the diversity that the EU is characterized by, as highlighted by its motto “unity in diversity”. Figure 2. Box plots of Malmquist productivity index (MPI), change in relative efficiency (ECH), and change in production possibility frontier (FS)—outliers (source: own calculation and elaboration). 2000–2007 2008–2011 2012–2017 Figure 2. Box plots of Malmquist productivity index (MPI), change in relative efficiency (ECH), and change in production possibility frontier (FS)—outliers (source: own calculation and elaboration). J. Risk Financial Manag. 2019,12, 72 22 of 34 J. Risk Financial Manag. 2019, 6, x FOR PEER REVIEW 22 of 35 Figure 3. Comparison of EU15 and EU13 distances in ECH and FCH (source: own calculation and elaboration). 2000–2007 2008–2011 2012-2017 I. Low FS – Hi g h EFCH II. High FS – Hi g h EFCH III. High FS – Low EFCH IV. Low FS – Low EFCH Figure 3. Comparison of EU15 and EU13 distances in ECH and FCH (source: own calculation and elaboration). J. Risk Financial Manag. 2019,12, 72 23 of 34 The practicality or applicability of these results in terms of economic policy is, however, limiting because the results only refer to relative efficiency. What does it mean? In the framework of the evaluation, it is necessary to move from efficiency to effectiveness, i.e., instead of conducting economic policy activities based on their setting and objectives; however, this cannot be done using the DEA method. For future research, it is necessary to rely on the evaluation of the relationship between output and outcome (effectiveness) and not input and output (efficiency), which the DEA method evaluates. The reconstructed or newly built technical and transport infrastructure, the reconstruction of buildings and companies, and buying new technical tools, i.e., factual or physical re-modernization should be taken in account, as well as the possibilities of proper use in activities generating added value for the economy, i.e., qualitative, competitive advantage, which is key for the knowledge economy. This should be the topic of future research, i.e., how the factor endowment of the given economy contributes to its growth and how the economy can use not only its quantitative but also its qualitative competitive advantages. To this end, however, it is necessary to find suitable methods used in the evaluation of effectiveness. The quality and utility of assessment could be improved further by developing a more integrated and ongoing approach to evaluation. All these factors affect the convergence trend of the new EU member states and their regions to the old EU member states, and the growth in the old EU member states has an implicative impact on growth in the new EU member states. This growth may have the same degree in EU13 countries as in EU15 countries or may be higher. Many of the differences in economic growth and quality of life within a country may be explained by the differences in competitiveness. Countries with more paved roads, with better institutions, with better business environment, and with better human capital, for example, may experience faster economic growth and a clearer reduction in poverty levels (Charles and Zegarra 2014). All these trends and facts have very significant effects on the competitiveness of all EU member states, changing the efficiency/inefficiency development. The internal variation and heterogeneity also underline the inevitable steps needed at the national level. Policies oriented to solve the main economic and social problems of citizens may then not only focus on the improvement of the aggregate or average indicators of competitiveness, but also on the reduction of the regional differences in competitiveness. Effective thematic policies and efficient use of public spending on the established aims will help the overall efficiency of the whole system, ensuring desired outcomes—effectiveness that has a significant impact on reducing disparities and improving competitiveness. The White Paper on the Future of Europe makes a powerful statement about the current precarious state of European integration and its uncertain future. The continuing effects of the financial, economic, and migration crises are associated with reduced confidence and trust in democratic institutions and politicians, and a rise in populism, threatening the unity of the EU. A significant cause is the unequal impact of globalization and technological change on different parts of the EU. Thus, the EU not only needs to accelerate sustainable growth but also to resume convergence so that all parts of the EU can exploit the opportunities from the globalization of trade and technological change. The past three decades were characterized by trade liberalization, the rise of global value chains, and global production networks. The integration of emerging countries challenged the EU’s attractiveness as a production location, because of import competition and off-shoring. Furthermore, technological change and digital transformation (the fourth production revolution) is associated with jobless growth and concerns that the EU is falling behind technologically. Europe generally has a strong position concerning advances in technology, value added, productivity, profitability, and profits, but there are significant questions about its technological leadership. There are significant opportunities from the structural change that the EU is well placed to exploit. The cost advantages of some emerging economies are eroding, labor costs are becoming a less critical factor in location decisions, and some supply chains are being shortened to ensure greater control. These trends do not guarantee the renewed competitiveness of developed economies but depend on the ability of developed economies to effect the necessary structural transformation. Structural change across the EU requires a different policy and institutional focus on “ecosystems” of open, interconnected J. Risk Financial Manag. 2019,12, 72 24 of 34 networks of stakeholders, cooperating through strategic partnerships able to respond rapidly and flexibly to technological, market, and social changes. Disruptive innovation and creativity require multidisciplinaryandopenmodels of collaboration. The support of anenvironmentfor suchecosystems will unavoidablyneed tobe tailoredtospecific national, regional, oreven localcontexts. Policy packages need to be integrated and coordinated, delivered at a national, regional, and local level, while being adapted to the needs of different territories (Bachtler et al. 2017). Many observers believe that Europe is at the beginning of a new industrial revolution, considered to be the fourth such leap forward labeled Industry 4.0. The ubiquitous use of sensors, the expansion of wireless communication and networks, the deployment of increasingly intelligent robots and machines, as well as increased computing power at a lower cost and the development of “big data” analytics, have the potential to transform the way goods are manufactured in Europe. This new digital industrial revolution holds the promise of increased flexibility in manufacturing, mass customization, increased speed, better quality, and improved productivity. However, to capture these benefits, enterprises will need to invest in equipment, information and communication technologies (ICT), and data analysis, as well as the integration of data flow throughout the global value chain. The EU supports industrial change through its industrial policy and research and infrastructure funding. Member states are also sponsoring national initiatives such as Industrie 4.0 in Germany, the Factory of the Future in France and Italy, and Catapult centers in the United Kingdom (UK). However, challenges remain. The need for investment, changing business models, data issues, legal questions of liability and intellectual property, standards, and skill mismatches are among the challenges that must be met if benefits are to be gained from new manufacturing and industrial technologies. If these obstacles can be overcome, Industry 4.0 may help reverse the past decline in industrialization and increase total value added from manufacturing to a targeted 20% of all value added by strategy Europe 2020. Based on the facts mentioned in the two paragraphs above, the issue of reducing disparities among the EU member states and improving internal and external competitiveness can be solved by the current technical and digital revolution (Industry 4.0), especially via the EU cohesion policy instruments, e.g., in the form of Cohesion Policy 4.0 and through the European Structural and Investment Funds for current programming period 2014–2020. The challenge for the EU as a whole and the individual member state policy-makers is to develop or adopt policy frameworks and strategies that will stimulate sustainable growth, in a manner that ensures greater inclusiveness, especially in access to employment and capacity for entrepreneurship. This demands a more granular approach to structural policy, tailored better to the specific conditions of the different types of regions and communities across the EU. Different strategies are needed for frontier regions, intermediate regions (some catching up but others only keeping pace), and lagging regions. Existing EU strategies—from Lisbon strategy for period 2000–2010 to current strategy Europe 2020 for period 2010–2020—are only partially successful, with limited results about the scale of the challenge. Notwithstanding specific achievements, strategies were over-ambitious about the resources available, the deficits in governance (especially on coherence and the coordination of policies), and the performance of interventions. Importantly, policy responses gave inadequate recognition of the spatial unevenness of current and development needs and challenges for economic growth and development in the EU. Looking forward, any new EU strategic approach needs to recognize the lessons from the past and be realistic about what can be achieved. With relatively limited budgetary resources at the EU level, the EU will need to establish some principles for a new EU strategy. The critical requirement is a coherent, consistent, and mutually enforcing policy framework. Sectoral policies cannot deliver on a new EU agenda without integrated territorial policy packages. Equally, integrated territorial policy approaches cannot achieve prosperity and inclusive growth in the EU without well-designed sectoral and structural policies and reforms. The EU model of integration delivered unmatched long-term growth and economic and social convergence. However, the model is threatened by the effects of the financial and economic crises on employment opportunities and living standards. The EU needs both to accelerate sustainable growth J. Risk Financial Manag. 2019,12, 72 25 of 34 and ensure that all parts of the EU can exploit the growing globalization of trade and technological change. Structural transformation should be central to renewed policy priorities, requiring a new balance between policies for competitiveness and cohesion. The pursuit of economic and social cohesion is a collective task of both national and EU policies. Member states have the primary responsibility for the conduct and coordination of their economic policies to meet cohesion objectives. The same obligation applies to all EU policies and actions, including the implementation of the internal market. The agenda for Cohesion 4.0 is, thus, a much broader task than for cohesion policy alone. It requires the EU member states to demonstrate that they implemented structural reforms to support growth and cohesion before uploading domestic interests to the European level. It also underscores the necessity of an integrated approach to structural transformation and cohesion under all EU regulatory and investment policies (Bachtler et al. 2017). The informative ability of the results depends on the methods used; the results, as such, are dependent on the selected measurement methods that affect their limits and usage assumptions. Generally, the results of each analysis and method depend on the data used, i.e., they depend on data quality. There exist different assumptions that your data must meet for the method used to give a valid result. In the case of our analysis, the limitations are lined primarily with using principal component analysis (PCA). When we chose to analyze our data using PCA, part of the process involved checking to make sure that the data we wanted to examine could be analyzed using PCA. In practice, checking for these assumptions required using SPSS Statistics to carry out a few more tests, as well as to think a little bit more about our data. When analyzing our data using SPSS Statistics, one or more of these assumptions may be violated (i.e., not met). This is not uncommon when working with real-world data rather than textbook examples. However, even when our data fail certain assumptions, there is often a solution to try and overcome this. The particulars that we had to deal with in our analysis mainly concerned that data should be suitable for data reduction and there should be no significant outliers. Involving DEA assumptions, we had to deal with the homogeneity of units, sampling adequacy, i.e., comparison of the number of units and the number of input and output variables, and last but not least isotonicity. All of these limitations were addressed, tested, and explained in the article. 6. Conclusions Currently, the EU consists of 28 member states and is continually expanding to include new countries. The considerable geographic, demographic, and cultural diversity of the EU also brings differences in the socio-economic position of the EU member states. Different results in economic performance and living standards of the population indicate the status of the competitiveness of every country. Each country should know its competitive advantages and disadvantages and aim to strengthen advantages and reduce disadvantages, i.e., key factors of competitiveness. One of the main aims of the paper was to define the main factors of socio-economic development that determine the competitiveness level of EU member states. Based on FA results, it is possible to state that, in most of the cases, the old EU member states reflect the best results in driven forces of competitiveness (inputs aspects) as an assumption for better outcomes of economic activities and functioning of society (outputs aspects). The competitiveness of territory resides not only in the competitiveness of its constituent firms and their interactions, but also in the broader assets and social, economic, institutional, and public attributes of the country itself. The notion of competitiveness is as much about qualitative factors and conditions (such as untraded networks of informal knowledge, trust, social capital, and the like) as it is about quantifiable attributes and processes (such as inter-firm trading, patenting rates, labor supply, and so on). Furthermore, the causes of competitiveness are usually attributed to the effects of an aggregate of factors rather than the impact of any individual factor. The sources of competitiveness may also originate at a variety of geographical scales, from the local through to the regional, national, and even international. Therefore, the possibility of isolating the precise effects of any individual factor is limited, as mentioned by Martin (2003). The emergence of new perspectives in creating competitive advantages at the national level clearly emphasizes the role of local factors and economic initiatives in J. Risk Financial Manag. 2019,12, 72 32 of 34 References Adusei, Michael. 2016. Modelling the efficiency of universal banks in Ghana. Quantitative Finance Letters 4: 60–70. [CrossRef] Afzal, Munshi Naser Ibne, and Roger Lawrey. 2012. Evaluating the Comparative Performance of Technical and Scale Efficiencies in Knowledge-Based Economies (KBEs) in ASEAN: A Data Envelopment Analysis (DEA) Application. European Journal of Economics, Finance and Administrative Sciences 51: 81–95. Annoni, Paola, and Lewis Dijkstra. 2013. EU Regional Competitiveness Index 2013. Luxembourg: Publication Office of the European Union. Annoni, Paola, and Kornelia Kozovska. 2010. EU Regional Competitiveness Index 2010. Luxembourg: Publication Office of the European Union. Annoni, Paola, Lewis Dijkstra, and Nadia Gargano. 2017. EU Regional Competitiveness Index 2016. Working Paper WP 02/2017. Brussels: European Commission. Avkiran, Necmi K. 2006. Productivity Analysis in the Service Sector with Data Envelopment Analysis. SSRN Working Paper 2006. Brisbane: The University of Queensland. Bachtler, John, Joaquim Oliveira Martins, Peter Wostner, and Piotr Zuber. 2017. Towards Cohesion Policy 4.0: Structural Transformation and Inclusive Growth. Brussels: Regional Studies Association. Balcerowicz, Leszek, Andrzej Rz ó nca, Lech Kalina, and Aleksander Łaszek. 2013. Economic Growth in the European Union. Brussels: Lisbon Council asbl. Bansal, Pooja, and Aparna Mehra. 2018. Multi-period additive efficiency measurement in data envelopment analysis with non-positive and undesirable data. OPSEARCH 55: 642–61. [CrossRef] Barnum, Darold, Jason Coupet, John Gleason, Abagail McWilliams, and Annaleena Parhankangas. 2017. Impact of input substitution and output transformation on data envelopment analysis decisions. Applied Economics 49: 1543–56. [CrossRef] Breuss, Fritz, Mikul á s Lupt á cik, and Bernhard Mahlberg. 2000. How far away are the CEECs from the EU economic standards? A data envelopment analysis of the economic performance of the CEECs. In EI Working Papers/Europainstitut, 35. Vienna: Vienna University of Economics and Business. Caves, Douglas W., Laurits R. Christensen, and W. Erwin Diewert. 1982. The Economic Theory of Index Numbers and the Measurement of Input, Output, and Productivity. Econometrica 50: 1393–414. [CrossRef] Chandola, Varun, Banerjee Arindam, and Vipin Kumar. 2009. Anomaly detection: A survey. ACM Computing Surveys 41: 1–58. [CrossRef] Charles, Vincent, and Luis Felipe Zegarra. 2014. Measuring regional competitiveness through Data Envelopment Analysis: A Peruvian case. Expert Systems with Applications 41: 5371–81. [CrossRef] Charnes, Abraham, William W. Cooper, and Edwardo L. Rhodes. 1978. Measuring the efficiency of decision making units. European Journal of Operational Research 2: 429–44. [CrossRef] Cheng Chen, Chih. 2017. Measuring departmental and overall regional performance: Applying the multi-activity DEA model to Taiwan’s cities/counties. Omega 67: 60–80. [CrossRef] Christopoulos, Dimitris K. 2007. Explaining country ' s efficiency performance. Economic Modelling 24: 224–35. [CrossRef] Chortirat, Thunyaporn, Boonorm Chomtee, and Juthaphorn Sinsomboonthong. 2011. Comparison of four data transformation methods for weibull distributed data. Kasetsart Journal-Natural Science 45: 366–83. Conte, Andrea, Philip Schweizer, Adriaan Dierx, and Fabienne Ilzkovitz. 2009. An Analysis of the Efficiency of Public Spending and National Policies in the Area of R&D. European Economy—Occassional Papers 54. Brussels: European Commission. Cooper, William W., Lawrence M. Seiford, and Kaoru Tone. 2007. Data Envelopment Analysis: A Comprehensive Text with Models, Applications, References and DEA-Solver Software. New York: Springer. Deliktas, Ertugrul, and Mehmet Balcilar. 2005. A Comparative Analysis of Productivity Growth, Catch-Up, and Convergence in Transition Economies. Emerging Markets Finance and Trade 41: 6–28. [CrossRef] Drucker, Peter. 2001. The Efficiency of the Decision Makers. Bucharest: Editura Destin. Easterly, William, and Ross Levine. 2012. The European Origins of Economic Development. NBER Working Paper Series 18162; Cambridge: National Bureau of Economic Research. Esser, Klaus, Wolfgang Hillebrand, Dirk Messner, and Jörg Meyer-Stamer. 1995. Systemic Competitiveness. New Governance Patterns for Industrial Development. London: Frank Cass. J. Risk Financial Manag. 2019,12, 72 33 of 34 European Commission. 2019. 11 Trends for 11 Countries on EU Convergence: The EU Enlargement Countries in the Baltics, Balkans and Central and Eastern Europe. Available online: https://ec.europa.eu/info/conference-15thanniversary-2004-eu-enlargement-looking-back-looking-forward/edited-volume_en (accessed on 1 April 2019). Färe, Rolf, Shawna Grosskopf, and C. A. Knox Lovell. 1994a. Production Frontiers. Cambridge: Cambridge University Press. Färe, Rolf, Shawna Grosskopf, Marry Norris, and Zhongyang Zhang. 1994b. Productivity Growth, Technical Progress and Efficiency Change in Industrialized Countries. The American Economic Review 84: 66–83. Farrell, Michael James. 1957. The measurement of productivity efficiency. Journal of the Royal Statistical Society 120: 253–90. [CrossRef] Foddi, Marta, and Stefano Usai. 2013. Technological catching up among European regions. Lessons from Data Envelopment Analysis. WP4/02 Search Working Paper. Brussels: European Commission. Gardiner, Ben, Ron Martin, and Peter Tyler. 2004. Competitiveness, Productivity and Economic Growth across the European Regions. Journal of Regional Studies 38: 1045–67. [CrossRef] Ghosh, Jayati, Peter Havlik, Marcos Poplawski-Ribeiro, and Waltraut Urban. 2009. Models of BRICs’ Economic Development and Challenges for EU Competitiveness. Vienna: The Vienna Institute for International Economics Studies. Golany, Boaz, and Sten Thore. 1997. Restricted best practice selection in DEA: An overview with a case study evaluating the socio-economic performance of nations. Annals of Operations Research 73: 117–40. [CrossRef] Goryushina, Evgenija, and Karine Mesropyan. 2013. Economic Inequality and Political Instability Measuring by DEA and Alternative Indices: State of the Art and Research Perspectives for Cross-Regional Studies. Der Donauraum 52: 445–64. [CrossRef] Hair, Joseph F., William C. Black, Barry J. Babin, and Roplh E. Anderson. 2009. Multivariate Data Analysis. Upper Saddle River: Prentice Hall. Halkos, George, and Nickolaos Tzeremes. 2005. A DEA Approach to Regional Development. MPRA Paper 3992. Available online: https://mpra.ub.uni-muenchen.de/id/eprint/3992 (accessed on 1 July 2007). Hseu, Jiing-Shyang, and Jui-Kou Shang. 2005. Productivity Changes of Pulp and Paper Industry in OECD Countries, 1991–2000: A Non-Parametric Malmquist Approach. Forest Policy and Economics 7: 411–22. [CrossRef] Hsu, Maxwell, Xueming Luo, and Gary H. Chao. 2008. The Fog of OECD and Non-OECD Country Efficiency: A Data Envelopment Analysis Approach. The Journal of Developing Areas 42: 81–93. Hwang, Yun-Gi, Soohyun Park, and Daecheol Kim. 2018. Efficiency Analysis of Official Development Assistance Provided by Korea. Sustainability 10: 2697. [CrossRef] Izadikhah, Mohammad, Reza Farzipoor Saen, and Razieh Roostaee. 2018. How to assess sustainability of suppliers in the presence of volume discount and negative data in data envelopment analysis? Annals of Operations Research 269: 241–67. [CrossRef] Jiang, Huichen, and Yifan He. 2018. Applying Data Envelopment Analysis in Measuring the Efficiency of Chinese Listed Banks in the Context of Macroprudential Framework. Mathematics 6: 184. [CrossRef] Lacko, Roman, and Zuzana Hajduov á . 2018. Determinants of Environmental Efficiency of the EU Countries Using Two-Step DEA Approach. Sustainability 10: 3525. [CrossRef] Lambooy, Jan G., and Ron A. Boschma. 2001. Evolutionary economics and regional policy. The Annals of Regional Science 35: 113–31. [CrossRef] Lukovics, Miklos. 2009. Measuring Regional Disparities on Competitiveness Basis. In Regional Competitiveness, Innovation and Environment. Edited by Zoltán Bajmócy and Imre Lengyel. Szeged: JATE Press, pp. 39–53. MacGregor Pelik á nov á , Radka. 2017. European myriad of approaches to parasitic commercial practices. Oeconomia Copernicana 8: 167–80. [CrossRef] Makridou, Georgia, Kostas Andriosopoulos, Michael Doumpos, and Constantin Zopounidis. 2014. An Integrated Approach for Energy Efficiency Analysis in European Union Countries. Working Paper 2014.02. Chania: Technical University of Crete. Malhotra, Rashmi, and Davinder K. Malhotra. 2006. Evaluating the efficiency of European Union integration. International Journal of Commerce and Management 19: 233–52. [CrossRef] Mandl, Ulrike, Adriaan Dierx, and Fabienne Ilzkovitz. 2008. The Effectiveness and Efficiency of Public Spending. Brussels: European Commission-Directorate General for Economic and Financial Affairs. J. Risk Financial Manag. 2019,12, 72 34 of 34 Martin, Ron. 2003. A Study on the Factors of Regional Competitiveness. Available online: http://ec.europa.eu/ regional_policy/sources/docgener/studies/pdf/3cr/competitiveness.pdf (accessed on 1 September 2003). Meleck ý , Luk á š. 2018. The main achievements of the EU structural funds 2007–2013 in the EU member states: efficiency analysis of transport sector. Equilibrium. Quarterly Journal of Economics and Economic Policy 13: 285–306. [CrossRef] Mihaiu, Diana Marieta, Alin Opreana, and Marian Pompiliu Cristescu. 2010. Efficiency, effectiveness and performance of the public sector. Romanian Journal of Economic Forecasting 1: 132–47. Mohammad, Nordin. 2007. A Linear Programming Formulation of Macroeconomic Performance: The Case of Asia Pacific. Matematika 23: 29–40. Nurboja, Bashkim, and Marko Košak. 2017. Banking efficiency in South East Europe: Evidence for financial crises and the gap between new EU members and candidate countries. Economic Systems 41: 122–38. [CrossRef] Ocubo, Toshihiro. 2012. Antiagglomeration subsidies with heterogeneous firms. Journal of Regional Science 52: 285–87. [CrossRef] Otsuka, Akihiro. 2014. Analysis of Productive Efficiency in Japanese Regional Economies. Studies in Regional Science 44: 453–65. [CrossRef] Rabar, Danijela. 2013. Assessment of Regional Efficiency in Croatia using Data Envelopment Analysis. Croatian Operational Research Review 4: 76–88. Ramanathan, Ramakrishnan. 2006. Evaluating the comparative performance of countries of the Middle East and North Africa: A DEA Application. Socio-Economic Planning Sciences 40: 156–67. [CrossRef] Seiford, Lawrence M., and Joe Zhu. 1999. An investigation of returns to scale in data envelopment analysis. Omega 27: 1–11. [CrossRef] Shu, Guoping, Beiyan Zeng, Deanne Wright, and Oscar Smith. 2002. Impact of Data Transformation on the Performance of Different Clustering Methods and Cluster Number Determination Statistics for Analyzing Gene Expression Profile Data. Paper presented at 14th Annual Conference on Applied Statistics in Agriculture, Manhattan, Kansas, April 28–30; pp. 94–110. Stan í ˇckov á , Michaela. 2017. Can the implementation of the Europe 2020 Strategy goals be efficient? The challenge for achieving social equality in the European Union. Equilibrium. Quarterly Journal of Economics and Economic Policy 12: 383–98. Stevens, James P. 1986. Applied Multivariate Statistics for the Social Sciences. Mahwah: Lawrence Erlbaum Associates. Tan, Hui-Boon, Chee-Wooi Hooy, Sardar M.N. Islam, and Alex Manzoni. 2008. Relative efficiency measures for the knowledge economies in the Asia Pacific region. Journal of Modelling in Management 3: 111–24. Toloo, Mehdi, Mona Barat, and Atefeh Masoumzadeh. 2015. Selective measures in data envelopment analysis. Annals of Operations Research 226: 523–642. [CrossRef] Tung, Shiue-Jen, Guo-Ya Gan, and Wen-Li Chyr. 2018. Efficiency Measures for VRM Models Dealing with Negative Data in DEA. Journal of Marine Science and Technology-Taiwan 26: 180–84. Watt, Andrew, and Andreas Botsch. 2010. After the Crisis: Towards a Sustainable Growth Model. Brussels: European Trade Union Institute. Wu, Po-Chin, Tzu-Hsien Huang, and Sheng-Chieh Pan. 2014. Country Performance Evaluation: The DEA Model Approach. Social Indicators Research 118: 835–49. [CrossRef] Zhu, Joe. 2011. Manual DEA Frontier—DEA Add-In for Microsoft Excel. Available online: http://www.deafrontier. net (accessed on 1 May 2011). © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).