ORIGINAL SCIENTIFIC PAPER RECEIVED: FEBRUARY 2018 REVISED: FEBRUARY 2018 ACCEPTED: FEBRUARY 2018 DOI: 10.2478/ngoe-2018-0003 UDK: 330.35:005.412 JEL: O31,047, C23 Citation: Nekrep, A., Strasek, S., & Borsic, D. (2018). Productivity and Economic Growth in the European Union: Impact of Investment in Research and Development. Nase gospodarstvo/Our Economy, 64(1), 18-27. DOI: 10.2478/ngoe-2018-0003 Productivity and Economic Growth in the European Union: Impact of Investment in Research and Development Andreja Nekrep PhD student at the Faculty of Economics and Business, University of Maribor, Slovenia andreja.nekrep@um.si Sebastjan Strasek University of Maribor, Faculty of Economics and Business, Slovenia sebastjan.strasek@um.si Darja Borsic University of Maribor, Faculty of Economics and Business, Slovenia darja.borsic@um.si Abstract This paper focuses on investment in research and development as a factor of Labour productivity and economic growth. Our analysis confirms the Link between expenditure for research and development (expressed in % of GDP) and labour productivity (expressed in the number of hours worked) based on selected data for EU Member States in the period 1995-2013. A causal link between variables of the concave parabola was confirmed, and the value of expenditure for research and development (2.85% of EU GDP) maximising productivity (per hour of work) was determined based on the examined data. In accordance with these findings, EU's target of reaching 3% of GDP spent on research and development to be achieved by 2020 seems in support of reaching maximum productivity in the EU. Key words: investment in research and development, productivity, economic growth, correlation, panel analysis NG NASE GOSPODARSTVO OUR ECONOMY Vol. . 64 No. 1 2018 pp . 18-27 Introduction How to increase the level of productivity and consequently economic growth in comparison to other leading economies in the world such as the USA and Japan remains of the main topics of economic and political discussions in the European Union. Such discussions quickly come across the determinants of growth and productivity. That is why the preceding paper focuses on investment in research and development and explains its role in determining productivity and economic growth. Theory and empirical literature provide a wide variety of authors, who analyse the relationship among investment in research and development, and productivity and economic growth. There have been several views in the modern theory of economic growth since the middle of the twentieth century. The first, neoclassical 18 Andreja Nekrep, Sebastjan Strasek, Darja Borsic: Productivity and Economic Growth in the European Union: Impact of Investment in Research and Development growth theory, formalized by Solow (1956, 1957) and Swan (1956), is based on the assumption of exogenous technological progress, while explaining the increasing relation among production factors capital and labour as a main source of economic growth. On the other hand, endogenous growth theories emphasize production factors such as new knowledge (Romer, 1990; Grossman & Helpman, 1991); research, development and innovations (Aghion & Howitt, 1992); and human capital (Lucas, 1988) as main sources of productivity and economic growth. Arrow (1962) is one of the authors who introduced the concept of learning by doing and defined the technological change as an unplanned outcome of new knowledge, which is generated in the process of learning by doing. Grossman and Helpman (1991) added the notion that modern technological progress requires intentional investment of the private sector in research and development, while the state should neutralize the spill-over effect of the new knowledge. They applied the spill-over effect of knowledge also to the cross-country level as an important source of productivity growth in individual countries and differences among them. Aghion and Howitt (1992) are founders of a group of models, in which research activities are crucial for creating new knowledge, and where new improvements of products and processes generate growth of productivity and economic growth. There are also empirical papers by Coccia (2009) and Zachariadis (2004), who confirmed the positive impact of expenditure for research and development on productivity. Contrary, Pack (1994) found out that in some OECD countries, productivity declined despite increased expenditure for research and development. The author explained his findings by the impact of production organization and social and institutional characteristics of the economies. A similar approach can be noted in the third standpoint related to the causes of economic growth, which place more interest on noneconomic factors such as: new institutional economics (North, 2003) or the concept of national innovation systems (Lundvall, 1992; Nelson, 1993). Being aware of the findings of economic theory about the role of investment in research and development for enhancing productivity and economic growth, the EU pays special attention to the expenditure level for research and development. Already, since the 1950s, when the economic and political integration in Europe began, a need for an effective, common research and development policy has been present. The aim is to gain synergy effects of research activities by overcoming the partial national research policies, to avoid the duplication of research and to reach common directions in research and innovations for solving key challenges of European society and to increase effectiveness of investment in research and development. In 2000, the EU introduced the Lisbon Strategy with special attention to establishing European Research Area (ERA), common internal market for research with free mobility of researchers, scientific discoveries and technologies. The EU maintained ERA as a central element also in the present strategy of Europe 2020 and its leading incentive Innovation Union, which were presented in 2010. Since 1984, the EU has been stimulating research and development activities through five-year framework programs, which are key EU financial instruments for supporting research and development. These framework programs are supplemented by several structural funds on the national and regional levels. By implementing such support for research activities, the EU strives to become the leading research area in the world, to enhance competitiveness of the European economy and to find solutions for the EU's modern social challenges (such as demographic changes and population aging, healthy food, scarce energy sources, etc.). In the current program period (2014-2020), the framework program Horizon 2020 takes place with the biggest budget in EU history, which is an additional indicator of the importance that the EU places on research and innovations for enhancing the productivity and competitiveness of the European economy. The paper analyses the impact of expenditure for research and development on labour productivity in EU-28 for the period from year 1995 to year 2013. The original contribution of the paper to the observed economic phenomena is empirically testing the relationship between investment in research and development by taking into account a different set of countries and different time frame, as compared to other similar empirical works (such as Coccia, 2009; Zachariadis, 2004; Hall & Mairesse, 1995; Amendola et al., 1993; Lichtenberg & Siegel, 1991). In addition, we empirically tested the link among the size of investment in research and development and potential maximal productivity, which was done by only a few authors (Coccia, 2009). Furthermore, Pokrivcak and Zahorsky (2016) found empirical evidence of statistically significant impact of investment in research and development in the Czech Republic, Poland, Romania, and Slovenia among all CEE countries. Meanwhile, Gocer at al. (2016) and Gehringer et al. (2016) estimate the effect of investment of research and development on income and economic growth, respectively. The paper proceeds with a review of the level of investment in research and development in EU member states. The third part explains the data used and methodology applied, which is followed by the presentation of empirical results in section four. The fifth and last section provides the conclusions. 19 NAŠE GOSPODARSTVO / OUR ECONOMY Vol. 64 No. 1 / March 2018 Investment in Research and Development in EU Member States The indicator for the size of expenditure for research and development is gross domestic expenditure on research and development (GERD) as a % of gross domestic product (GDP). The share of expenditure for research and development in GDP is also defined as R&D Intensity (Eurostat, 2016). Since investment in research and development presents one of the key determinants of productivity and enhancing competitiveness, the EU Lisbon Strategy had set a goal of devoting 3% of GDP for research and development in year 2010, which was not achieved. According to Eurostat, the share of investment in research and development in GDP in EU reached 1.93% in 2010 (Eurostat, 2016). EU kept the goal of 3 % also in its Europe 2020 Strategy for smart sustainable and inclusive growth with its leading incentive, Innovation Union, which is supposed to be realized by 2020. Individual EU member states set different national goals by 2020 (Table 1). Among them, six states (Belgium, Denmark, Germany, Estonia, France and Slovenia) set the same goal as the EU (3%) while three states (Austria, Finland and Sweden) set a higher goal (Eurostat, 2016). The size of expenditure for research and development in EUR per capita by EU member states is presented in Figure 1. Taking a look at individual EU member states (Table 1), one can notice the highest R&D Intensity in 2014 in Finland (3.17%), Sweden (3.16%), Denmark (3.04%) and in Austria (2.99%). Nine member states devoted less than 1% of GDP to research and development. These are, besides Greece, many of the members who joined the EU in 2004 or later. However, Slovenia is above EU average with 2.39%, while Czech Republic (2.00%), Estonia (1.46%), Hungary (1.38%) are below EU average but above 1 % of GDP (Eurostat, 2016). Figure 2 presents comparison of expenditures for research and development in EU-28 and other selected economies: USA, Japan and South Korea. According to Eurostat, EU-28 member states, on average, devoted 1.80% of GDP for research and development in year 2003, although this amount decreased to 1.76% in 2005, it has grown since 2006, with slight fall in 2010, to 2.03% in year 2014. Despite the growing trend in the observed period, the share of GDP devoted for research and development in EU-28 in 2012 was lower than in other selected economies, particularly Japan (3.34%), USA (2.81%) and South Korea (4.03%) (Eurostat, 2016). Data and Methodology Data about the size of expenditure for research and development in % of GDP, and data about the labour productivity in EUR per hour, were obtained from the Eurostat database for individual EU-28 member states for the period of 1995-2013. Our empirical analysis of productivity is limited to only one determinant (the expenditure for research and development), even though there are other factors influencing the productivity. The expenditure for research and development are considered as total and not divided to several sectors (government, private, higher education. etc.). Our database consists of EU-28 member states (N=28) for the period Table 1. Expenditure for research and development in EU-28 and target values for 2020 % of GDP 2000 2005 2010 2014 2020 Target EU-28 1.79 1.76 1.93 2.03 3 BE 1.93 1.78 2.05 2.46 3 BG 0.49 0.45 0.59 0.80 1.5 CZ 1.12 1.17 1.34 2.00 1 DK 2.19 2.39 2.94 3.08 3 DE 2.39 2.42 2.71 2.84 3 EE 0.60 0.92 1.58 1.46 3 IE 1.09 1.20 1.62 1.55 2 EL n.a. 0.58 0.60 0.83 1.21 ES 0.89 1.10 1.35 1.20 2 FR 2.08 2.04 2.18 2.26 3 HR n.a. 0.86 0.74 0.79 1.4 IT 1.01 1.05 1.22 1.29 1.53 CY 0.23 0.37 0.45 0.47 0.5 LV 0.44 0.53 0.60 0.68 1.5 LT n.a. 0.75 0.78 1.02 1.9 LU 1.57 1.59 1.53 1.24 2.3 HU 0.79 0.93 1.15 1.38 1.8 MT n.a. 0.53 0.64 0.85 2 NL 1.81 1.79 1.72 1.97 2.5 AT 1.89 2.38 2.74 2.99 3.76 PL 0.64 0.57 0.72 0.94 1.7 PT 0.72 0.76 1.53 1.29 2.7 RO 0.36 0.41 0.45 0.38 2 SI 1.36 1.41 2.06 2.39 3 SK 0.64 0.49 0.62 0.89 1.2 FI 3.25 3.33 3.73 3.17 4 SE n.a. 3.39 3.22 3.16 4 UK 1.73 1.63 1.69 1.72 n.a. Vir: Eurostat (2016). 20 Andreja Nekrep, Sebastjan Strasek, Darja Borsic: Productivity and Economic Growth in the European Union: Impact of Investment in Research and Development Figure 1. Public expenditure for research and development (in EUR per capita) for EU member states in year 2014 700,0 - 600,0 500,0 400,0 - - - 300,0 ------ - 200,0 -------- - - 100,0 -----------0,0 —ILUC^^ooluI— _JQl/1Q-LU LU U 1/1 O —I Oi —I LU ^ m CL >| O U —1 CO Source of data: Eurostat (2016). Figure 2. Expenditure for research and development, in % of GDP, in EU-28, USA (ZDA), Japan (JP) and South Korea (JK), 2000-2012 4,5- 35 .................. 3,5 3----- 2,5--- 2-^- 1,5- 1- 0,5- 0- 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 — EU-28 ZDA JP —- JK Notes: ZDA - USA, JK - South Korea Source of data: Eurostat (2016). 1993-2013 (T=19), resulting in a panel dataset of dimension NxT (532). Considering the missing data for some observations, we applied the empirical analysis to the panel data with 454 observations. The empirical analysis consists of four parts. First, by applying time series data for individual EU member states, we tested what kind of correlation among R&D intensity (expenditure for research and development as a share of GDP in %) and productivity existed in the period of 1995-2013. Second, we explored the effect of time lags in the size of expenditure for research and development in their correlation to productivity. In the third part, we explored the functional relationship among expenditure for research and development and productivity by utilizing a panel data set. Fourth, based on the results from the previous part, the size of expenditure for research and development, which maximises the productivity in the panel of EU member states, was calculated. 21 NAŠE GOSPODARSTVO / OUR ECONOMY Vol. 64 No. 1 / March 2018 Results of the Empirical Analysis Correlation among expenditure for research and development, and labour productivity in EU-28 The Pearson correlation coefficient (rxy) defines the direction and strength of correlation among two variables, y and x.. It can be calculated by (Artenjak, 2003, p. 154): '■xy rxy =-" Jjfl^iC*i-x)2 jiUM-y)2 1 __ _^T.i=iXjyi-xy_ J^eUxM2) -^lAyf-y2) ' (1) where c is covariance of y and x, a standard deviation xy ' x of variable x, ay standard deviation of variable y, N is the number of observations, y is arithmetic mean of y, and x is arithmetic mean of x. The value of the correlation coefficient can be in the interval of -1 < rxy < 1, where the absolute values of the coefficient present different strength of the correlation among the observed variables (Artenjak, 2003, p. 154): \rxy\ = 0, no correlation, 0 < lrxy\ < 0.50, weak correlation, 0.51 < \rxy\ < 0.79, moderate correlation, 0.80 < \rxy\ < 0.99, strong correlation, \rxy\ = 1, perfect correlation. Based on the data for the size of expenditure for research and development, and labour productivity, we calculated the correlation coefficients (rxy) for individual EU-28 member states in the period of 1995-2013. Table 2 presents results obtained in SPSS. In 17 out of 28 EU member states, there is positive and statistically significant correlation among expenditure for research and development and labour productivity. In one case, there is statistically significant negative correlation (p < 0.05), while other countries exhibited statistically insignificant correlation among observed variables. Table 3 presents the number of EU-28 member states regarding the direction and strength of the correlation for the significance level of 5%. Table 2. Correlation coefficients (r ) among expenditure for research and development, and labour productivity in EU-28 rxy P BE 0.529" 0.020 BG 0.350" 0.042 CZ 0.793"" 0.000 DK 0.850"" 0.000 DE 0.907"" 0.000 EE 0.865"" 0.000 IE 0.777"" 0.000 EL 0.205 0.523 ES 0.790"" 0.000 FR -0.084 0.734 HR n.a. n.a. IT 0.655"" 0.002 CY 0.958"" 0.000 LV 0.796"" 0.001 LT 0.893"" 0.000 LU 0.436 0.178 HU 0.866"" 0.000 MT 0.037 0.914 NL -0.252 0.298 AT 0.988"" 0.000 PL 0.451 0.053 PT 0.888"" 0.000 RO 0.005 0.983 SI 0.794"" 0.001 SK -0.455 0.050 FI 0.883"" 0.000 SE -0.104 0.711 UK -0.462" 0.046 Notes: *Correlation coefficient is statistically significant at 5%. ** Correlation coefficient is statistically significant at 1%. n.a. - Due to missing data for Croatia, the correlation coefficients were not calculated. Table 3. Number of EU-28 member states regarding the direction and strength of correlation Positive and weak Positive and moderate Positive and strong Negative correlation correlation correlation correlation Statistically significant at 5% 1 7 9 1 Statistically insignificant 9 22 Andreja Nekrep, Sebastjan Strasek, Darja Borsic: Productivity and Economic Growth in the European Union: Impact of Investment in Research and Development Correlation among expenditure for research and development and labour productivity in EU-28 with time lags Besides the basic correlation coefficient among the observed variables, we have checked also the effects of time lags in expenditure for research and development on labour productivity by applying Pearson correlation coefficients for periods (t-1), (t-2), (t-3). For labour productivity, the period of 1998-2013 was applied, while for expenditure for research and development, we employed time periods (t) 1998-2013, (t-1) 1997-2012, (t-2) 1996-2011 and (t-3) 1995-2010. We calculated Pearson correlation coefficients (r) in SPSS for individual EU member states and presented them in Table 4. Considering the time period t (without time lags in expenditure for research and development), there are 16 EU member states with statistically significant positive moderate or strong correlation coefficients. Regarding one, two and three-year lags in expenditure for research and development, there are 14, 13 and 13 EU member states with positive moderate or strong correlation coefficients, respectively. When compared to the correlation without the time lags, one can note that the correlation is stronger with 1-year lag for 10 EU member states, with 2-year lag in 11 states and with 3-year lag in 10 EU member states (out of 16 EU member states with statistically significant positive moderate or strong correlation without time lags). Additionally, it can be noted that 8 out of 16 member states have the highest Table 4. Correlation coefficients (rxy ) in EU-28 with time lags in expenditure for research and development rxy rxy r xy r xy t (t-1) (t-2) (t-3) BE 0.308 (p=0.245) 0.265 (p=0.322) 0.316 (p=0.234) 0.510" (p=0.044) BG 0.495 (p=0.051) 0.436 (p=0.091) 0.186 (p=0.491) -0.186 (p=0.490) CZ 0.713"" (p=0.002) 0.729""(p=0.001) 0.826"" (p=0.000) 0.919"" (p=0.000) DK 0.757"" (p=0.001) 0.839"" (p=0.000) 0.905"" (p=0.000) 0.926"" (p=0.000) DE 0.846"" (p=0.000) 0.855"" (p=0.000) 0.895"" (p=0.000) 0.916"" (p=0.000) EE 0.865"" (p=0.000) 0.798"" (p=0.000) 0.829"" (p=0.001) 0.909"" (p=0.000) IE 0.777"" (p=0.000) 0.705"" (p=0.002) 0.595" (p=0.015) 0.458 (p=0.074) EL 0.205 (p=0.523) 0.469 (p=0.067) -0.086 (p=0.801) 0.383 (p=0.245) ES 0.764"" (p=0.001) 0.854"" (p=0.000) 0.922"" (p=0.000) 0.962"" (p=0.000) FR 0.329 (p=0.214) 0.189 (p=0.483) -0.146 (p=0.589) -0.427 (p=0.099) HR n.a. n.a. n.a. n.a. IT 0.442 (p=0.087) 0.413 (p=0.112) 0.467 (p=0.068) 0.545" (p=0.029) CY 0.958"" (p=0.000) 0.949"" (p=0.000) 0.981"" (p=0.000) 0.985"" (p=0.000) LV 0.796"" (p=0.001) 0.871"" (p=0.000) 0.803"" (p=0.001) 0.682"" (p=0.007) LT 0.893"" (p=0.000) 0.904"" (p=0.000) 0.700 (p=0.053) 0.519 (p=0.233) LU 0.463 (p=0.178) 0.410 (p=0.211) 0.147 (p=0.706) -0.248 (p=0.554) HU 0.809"" (p=0.000) 0.853"" (p=0.000) 0.886"" (p=0.000) 0.898"" (p=0.000) MT 0.037 (p=0.914) 0.400 (p=0.175) -0.573 (p=0.107) -0.524 (p=0.183) NL -0.073 (p=0.789) -0.294 (p=0.269) -0.516" (p=0.041) -0.685"" (p=0.003) AT 0.981"" (p=0.000) 0.984"" (p=0.000) 0.987"" (p=0.000) 0.984"" (p=0.000) PL 0.519"" (p=0.039) 0.398 (p=0.127) 0.195 (p=0.469) 0.055 (p=0.841) PT 0.874"" (p=0.000) 0.910"" (p=0.000) 0.923"" (p=0.000) 0.906"" (p=0.000) RO 0.531"" (p=0.034) 0.248 (p=0.355) -0.212 (p=0.431) -0.526" (p=0.036) SI 0.794"" (p=0.001) 0.730"" (p=0.003) 0.672"" (p=0.009) 0.659"" (p=0.010) SK 0.005 (p=0.985) -0.445 (p=0.084) -0.723"" (p=0.002) -0.866"" (p=0.000) FI 0.741"" (p=0.001) 0.829"" (p=0.000) 0.883"" (p=0.000) 0.903"" (p=0.000) SE -0.663"" (p=0.013) 0.469 (p=0.067) -0.188 (p=0.558) 0.242 (p=0.448) UK -0.371 (p=0.157) -0.291 (p=0.275) -0.336 (p=0.203) -0.528" (p=0.036) Notes: *Correlation coefficient is statistically significant at 5%. ** Correlation coefficient is statistically significant at 1%. n.a. - Due to missing data for Croatia, the correlation coefficients were not calculated. 123 NAŠE GOSPODARSTVO / OUR ECONOMY Vol. 64 No. 1 / March 2018 correlation coefficient among expenditure for research and development and labour productivity with 3-year lag (t-3) for expenditure for research and development. While two member states exhibit the strongest correlation among the observed variables with 2-year lag in expenditure for research and development, there are two member states with 1-year lag and four member states without time lags. Regarding all EU member states included into the analysis, it can be concluded that 15 countries (out of 27) exhibit positive and statistically significant correlation among labour productivity in the current period and expenditure for research and development with 3-year lag. This is followed by 14 countries with positive and statistical significant correlation in the case of 2-year lag in expenditure for research and development, and by 13 countries with positive and statistical significant correlation in the case of 1- year lag in expenditure for research and development. Nonlinear relation among expenditure for research and development and labour productivity The scatter plot in Figure 3 displays nonlinear relation among expenditure for research and development in % of GDP (independent variable) and labour productivity per hour of work (dependent variable). Distribution of observations in the diagram illustrates that the best fit would be a parabola (polynomial of degree 2). The estimation of quadratic function was conducted on the panel of EU-28 member states. Figure 3. Expenditure for research and development in % of GDP (IZDAT_X) and labour productivity (PROD) in EU-28 in period 1995-2013 70 -,- 60 50 Quadratic regression model is in general expressed as (Pfajfar, 2014, p. 167): Q O ai CL 40 30 20 10 oo oo 0 JOO Oo= k t o>»°° °° r 0