“LUCIAN BLAGA” UNIVERSITY OF SIBIU THE FACULTY OF ECONOMICS DOCTORAL STUDIES DOCTORAL THESIS Scientific coordinator: Prof. PhD. ILEANA TACHE PhD… [303389]
“LUCIAN BLAGA” UNIVERSITY OF SIBIU
THE FACULTY OF ECONOMICS
DOCTORAL STUDIES
DOCTORAL THESIS
Scientific coordinator:
Prof. PhD. ILEANA TACHE
PhD Candidate:
FLORIN TEODOR BOLDEANU
Sibiu, 2017
“LUCIAN BLAGA” UNIVERSITY OF SIBIU
FACULTY OF ECONOMICS
DOCTORAL STUDIES
THE INFLUENCE OF PRIVATE AND PUBLIC SECTORS ON ECONOMIC GROWTH IN THE EUROPEAN UNION
Scientific coordinator:
Prof. PhD. ILEANA TACHE
PhD Candidate:
FLORIN TEODOR BOLDEANU
Sibiu, 2017
ABBREVIATIONS
ASEAN = Association of Southeast Asian Nations
COFOG = Classification of the Functions of Government
EU = European Union
EUROSTAT = European Statistical Service
FDI = Foreign direct investment
FE = Fixed Effects Model
FGLS = Feasible Generalized Least Squares
FMOLS = Fully Modified Ordinary Least Square
GDP = Gross Domestic Product
GMM = Generalised Method of Moments Estimator
GNP = Gross National Product
GVA = Gross Value Added
ICT = Information and communications technology
IMF = International Monetary Fund
LM = Lagrange multiplier
N.e.c = not elsewhere classified
NUTS = Nomenclature of Territorial Units for Statistics
OECD = The Organization for Economic Cooperation and Development
OLS = Ordinary Least Square
PPS = Purchasing Power Standard
QML = Quasi-maximum likelihood
R&D = Research and Development
REM = Random Effects Model
UN = United Nations
LIST OF TABLES
Table 1: The UN Classification of the Functions of Government……………………………….. 37
Table 2: The similarities between Harrod’s and Domar’s model………………………………… 70
Table 3: Summary statistics of the variables used…………………………………………………….. 98
Table 4: The correlation matrix of the variables used in the model……………………………… 99
Table 5: Unit-root tests results for the variables used………………………………………………. 101
Table 6: Hausman test…………………………………………………………………………………………. 102
Table 7: Breusch-Pagan / Cook-Weisberg test for heteroskedasticity………………………… 103
Table 8: [anonimizat]………………………………………………….. 104
Table 9: Wooldridge test……………………………………………………………………………………… 104
Table 10: Pesaran test………………………………………………………………………………………….. 104
Table 11: Frees' test of cross sectional independence………………………………………………. 105
Table 12: Friedman's test of cross sectional independence……………………………………….. 105
Table 13: Breusch-Pagan Lagrange multiplier (LM) ………………………………………………. 105
Table 14: The results of the pooled OLS and REM regressions………………………………… 106
Table 15: The results of the FGLS and FMOLS regressions…………………………………….. 109
Table 16: The results of the GMM and system GMM……………………………………………… 113
Table 17: Ramsey RESET test using powers of the fitted values of Ly……………………… 116
Table 18: Shapiro-Wilk W test for normal data………………………………………………………. 119
Table 19: Summary statistics of the variables used in Eq 1………………………………………. 141
Table 20: Summary statistics of the variables used in Eq 2………………………………………. 142
Table 21: The correlation matrix of the variables used in equation 1…………………………. 143
Table 22: The correlation matrix of the variables used in equation 2…………………………. 144
Table 23: Unit-root test results for the variables used………………………………………………. 144
Table 24: Hausman test for the QML estimation…………………………………………………….. 147
Table 25: Parm test for the QML-FE estimation……………………………………………………… 147
Table 26: The results of the GMM method for the NUTS 1 regions………………………….. 148
Table 27: The results of the system GMM method for the NUTS 1 regions……………….. 151
Table 28: The results of the GMM method for the NUTS 2 regions………………………….. 153
Table 29: The results of the system GMM method for the NUTS 2 regions……………….. 155
Table 30: The results of the QML estimation for the NUTS 1 regions……………………….. 157
Table 31: The results of the QML estimation for the NUTS 2 regions……………………….. 159
Table 32: Top ten metropolitan areas in the EU by population in 2013……………………… 174
Table 33: Top ten and bottom ten metropolitan areas by GDP/capita and GDP PPS/inhab in 2012………………………………………………………………………………………………………………….. 177
Table 34: Top ten and bottom ten metropolitan areas crude rate of net migration plus adjustment in 2012……………………………………………………………………………………………… 178
Table 35: Top five and bottom five metropolitan areas by gross value added from agriculture, forestry and fishing in 2012………………………………………………………………… 179
Table 36: Top five and bottom five metropolitan areas by gross value added from industry in 2012………………………………………………………………………………………………………………. 180
Table 37: Top five and bottom five metropolitan areas by gross value added from manufacturing in 2012…………………………………………………………………………………………. 180
Table 38: Top five and bottom five metropolitan areas by gross value added from construction in 2012……………………………………………………………………………………………. 181
Table 39: Top five and bottom five metropolitan areas by gross value added from wholesale and retail trade, transport, accommodation and food service activities in 2012………………………………………………………………………………………………………………….. 182
Table 40: Top five and bottom five metropolitan areas by gross value added from Information and communication in 2012……………………………………………………………….. 182
Table 41: Summary statistics of the variables used…………………………………………………. 183
Table 42: The correlation matrix of the variables used…………………………………………….. 184
Table 43: Unit-root test results for the variables used………………………………………………. 185
Table 44: Hausman test for the QML method…………………………………………………………. 186
Table 45: The Parm test……………………………………………………………………………………….. 187
Table 46: The results of the GMM estimator………………………………………………………….. 187
Table 47: The results of the system GMM estimator……………………………………………….. 189
Table 48: The results of the QML-FE method………………………………………………………… 192
Table 49: The results of the QML-FE estimation with the time period divided in ante and post economic crisis……………………………………………………………………………………………. 194
Table 50: The results of the QML-FE estimation for Western European and Central-Eastern Europe…………………………………………………………………………………………………… 196
LIST OF FIGURES
Figure 1: The evolution of GDP for the 28 EU countries from 1995 to 2014…………………. 2
Figure 2: Stationary state in the neoclassical Solow growth model……………………………… 74
Figure 3: The increase in savings in the Solow growth model……………………………………. 75
Figure 4: Voice and accountability indicator in 1996 and 2014 in the EU28………………… 94
Figure 5: Political stability and absence of violence/terrorism indicator in 1996 and 2014 in the EU28……………………………………………………………………………………………………………… 94
Figure 6: Government effectiveness indicator in 1996 and 2014 in the EU28………………. 95
Figure 7: Regulatory quality indicator in 1996 and 2014 in the EU28…………………………. 96
Figure 8: Rule of law indicator in 1996 and 2014 in the EU28…………………………………… 96
Figure 9: Control of corruption indicator in 1996 and 2014 in the EU28……………………… 97
Figure 10: The evolution of neglog GDP/capita in the EU 28 between 1990 and 2014… 100
Figure 11: Residual plot………………………………………………………………………………………. 103
Figure 12: Kernel density estimate………………………………………………………………………… 116
Figure 13: Histogram of residuals…………………………………………………………………………. 117
Figure 14: Standardize normal probability plot………………………………………………………. 118
Figure 15: Quintile normal plots…………………………………………………………………………… 118
Figure 16: Heterogeneity across countries for the panel data……………………………………. 119
Figure 17: Heterogeneity across years for the panel data…………………………………………. 120
Figure 18: Population on 1 January by NUTS2 regions in 2013……………………………….. 134
Figure 19: Regional gross domestic product (PPS per inhabitant) by NUTS2 regions in 2013………………………………………………………………………………………………………………….. 135
Figure 20: Motorways network (kilometres) by NUTS 2 regions in 2013………………….. 136
Figure 21: Total intramural R&D expenditure (GERD) % GDP by NUTS2 regions in 2013………………………………………………………………………………………………………………….. 137
Figure 22: Tertiary education attainment, age group 25-64 % total by sex and NUTS2 regions in 2013…………………………………………………………………………………………………… 138
Figure 23: Nights spent by non-residents at tourist accommodation establishments by NUTS2 regions in 2013……………………………………………………………………………………….. 139
Figure 24: Nights spent by residents at tourist accommodation establishments by NUTS2 regions in 2013…………………………………………………………………………………………………… 140
Figure 25: EU regional average GDP growth vs. initial GDP – 2001-2008………………… 162
Figure 26: EU regional average GDP growth vs. initial GDP – 2008-2013………………… 162
Figure 27: Metropolitan GDP per capita and in PPS/inhabitant in 2013…………………….. 176
Figure 28: Retail sales per capita, annual % change, 2009-2012……………………………….. 177
Figure 29: Metropolitan crude rate of net migration plus statistical adjustment in 2012. 178
INTRODUCTION
The motivation for choosing the theme and the importance of the subject
Researchers, Nobel Prize winners and public institutions tried to find the proper definition for the concept of economic growth. Why should we focus on this dry statistical issue? This is because economic growth is a key factor in the well-being of billions. From the advantages brought by the industrial revolution, advanced countries that experience constant growth help their citizens to live well and longer. The recent economic crisis of 2008 showed that certain events can also pay a significant role in determining the variation of gross domestic product. Better understanding the mechanism behind what influences the economy will help us in mitigating or eliminating the negative outcomes that affect economic development.
Economic growth is the pinnacle of the twentieth century. Entire nations continue to see it as an extremely important objective economically and politically, the only factor that ensures the economic success of a nation in the long term.
The theme proposed for this scientific research aims at showing how private and public variables have had an influence on economic growth in the European Union at different territorial levels, more specifically at country, regional – NUTS areas and metropolitan level.
The link between government investment and economic development is a widely explored topic. Research studies that targeted the public sector are important for policy-makers from different countries, who are interested in allocating government funds more efficiently. The analysis of the influence of the private sector on economic growth is a less investigated theme in the literature. Very often in research papers that focused on the public sector, there were also some private variables that were included (gross capital formation, public investment, FDI, exports, etc.).
Gross domestic product in the EU has risen considerably. Figure 1 shows that between 1995 and 2014 many states saw improvements regarding economic development. For most of the Eastern European countries, the European integration was an advantage because of the new capital investments and the benefits of open trade. GDP is an aggregate
Figure 1: The evolution of GDP for the 28 EU countries from 1995 to 2014
Source: own contribution
indicator and it is important to quantify the exact factors that determined the rise in EU economic growth. This thesis investigates what factors have determined economic growth in the EU for different territorial levels and tries to quantify and to make a comparison with other studies. This study will be important for policy makers in better determining the exact factors that foster economic development.
Placing the thesis in the scientific context
The shifts that are taking place in the economy in the recent years have seen many developing states play a more important role in the world. Migration, globalization and the opening of new trade markets helped states like China, India or Brazil to have each year a sustained economic growth rate.
Emerging markets account for more than 50% of the world’s total output and China has already outpaced the US as the top economy. What were the other factors that determined this economic advancement? Regarding the European Union, the 2008 economic crisis impacted negatively many nations. Greece lost more than a third of its GDP since the onset of the economic crisis and Western European countries have each year a below 1% GDP growth rate. Are these outcomes a direct consequence of the austerity measures? What were the determinants of economic growth for the European Union states? These factors can be measured using economic variables, but some of them like trust, uncertainty, panic, political instability are non-economic factors.
The literature makes a clear distinction between economic and non-economic factors. For example “proximate” or economic sources refer to factors like capital accumulation, technological progress and labour and “ultimate” or non-economic sources refer to factors like government efficiency, institutions, terrorism, political and administrative systems, cultural and social factors, geography and demography (Rodrik 2003; Acemoglu et al 2005; Arvanitidis 2007; Acemoglu 2009).
Europe is in the middle of a changing economic and political landscape. The developing nations of the EU are seeing improved economic growth with the industrialized countries facing more political and social problems than economic ones. The 2016 vote for the Brexit may impact Europe in a negative way if policy makers will not come with concise measures. The huge wave of migrants and terrorism will have serious consequences on the economy and on human trust.
Are the economic growth models still viable in this ever changing world economy? More and more people are involved in creating virtual goods which are produced with smaller costs and distributed much easier. All you need for this is access to the internet and an innovative idea. We should not ignore the Space industry, which generates 300 billion each year. This industry could have a decisive effect for the economy in the future.
Knowledge stage
Economic growth theories and econometric models highlight the various ways in which the present economic activity can influence the future and identify sources that may lead to continuous growth. These theories have evolved over time, depending on the dynamics of the economic reality and the evolution of economic analysis tools. The interest in this subject was and is very high. From the classical economists to the present new economic growth scholars, this topic is very debated and researched.
There are a large number of scientific investigations in this field proved by the considerable number of articles, books, journals and other such works. Many theoretical and empirical works helped improve the knowledge regarding the determinants of economic growth. There are a large number of economists that have devoted an important part of their life to study the concept of economic growth and what influences this difficult concept. I will only name a few of them, such as: A. Smith, D. Ricardo, T. Malthus, J. M. Keynes, R.J. Barrio, R. Solow, Sala-I-Martin.
Research studies investigated the impact on economic growth of such determinants like investment, human capital, economic and fiscal policies, trade openness, foreign direct investment, research and development, institutional and political framework, socio-cultural factors, geography and demography. These studies were conducted mostly on country samples, but in the recent decades, there is a surge of empirical analysis done for regional, metropolitan or city samples.
Many authors have dealt with the relationship between public expenditure, foreign direct investment, openness, public or private investment, non-economic variables, among others and economic growth at country level (Khan and Reinhart 1990; Barro 1990, Barro and Sala-I-Martin 1995; Devarajan et al. 1996; Brasoveanu et al. 2008; Arpaia and Turrini 2008; Acemoglu 2009; Bagli and Adhikary 2014; Shera et al. 2014). Some of them focused on a single field of study like for example the role of health and education on economic growth, or the role of public and private investment.
The empirical research in the field of regional economic growth has tried to determine what variables have an influence on growth and to come to a consensus on the relevant sign of the variation. There are a number of research studies that determined a significant link between innovation (R&D expenditures, patent application, population employed in research), transportation (airport infrastructure, roads, highways), population growth, capital formation, energy consumption, public investments and economic growth at EU regional level (Bottazzi and Peri, 2002; Parent and LeSage 2012; Rodriguez-Pose et al. 2012, 2015). Like in the case of economic growth at country level, there is still not a consensus on the effects of some variable. Also, contradictions in results may appear from studies done for different regions like South America, China, North America or Russia (Golubchikov 2007; Spiezia and Weiler 2007; Hartono et al. 2007).
The notion that cities and metropolitan regions are a source of economic growth is gaining more and more focus in the recent period. Cities and urban zones are considered to be the fundamental sites for the concentration of economic activity. This is in part because of the new research done by many scholars in the field of new economic geography (agglomeration economies) or the ones involved in the “new growth theory” (Glaeser et al. 1992; Combes 2000; Melo et al. 2009).
Urban areas are human centres that allow for the exchange of goods, ideas and people and in turn the society reaps the benefits from trade and specialization (Christiaensen and Todo 2013; Glaser et al. 1992; Combes 2000). They facilitate all these factors to come together to allow for more production and labour specialization. Towns and cities rose to become market places in which goods and services are transferred faster and more efficiently.
These concepts and findings will represent the theoretical and methodological framework for this thesis. Also, this investigation will use the latest research regarding the concept of economic growth, published in top journals. The study will identify the possibilities to extend the investigation in this field and to provide comprehensive comparisons with the findings captured by previous studies. All of these literature works will be presented at the end of the thesis in a separate section entitled References.
The main objectives of the thesis
The most important objective of this thesis is to determine the main factors that influence economic growth in the European Union. This objective will be researched in the three empirical chapters of this study.
The first empirical chapter will have as its main objective to establish the most important factors that impacted economic growth for the 28 European Union countries. The other goals of this chapter, that stem from this main objective, will be to provide comprehensive knowledge regarding what public or private variables have a more important role on economic growth. Furthermore, the division of education into primary, secondary and tertiary levels will show which type of schooling is more significant. By using a dynamic panel data model, the lag dependent variable will also highlight meaningful knowledge related to the economic convergence hypothesis.
The second empirical chapter has as a main objective to provide conclusive information regarding the most relevant determinants at regional/territorial level in the European Union for 98 NUTS and 273 NUTS 2 areas. Another objective will be to establish if the convergence hypothesis holds for the above mentioned regions. By graphically illustrating the changes that took place between 2000 and 2013, this hypothesis will be confirmed or denied.
The objective of the last empirical chapter is to determine the most important factors that influenced economic growth at EU metropolitan level. The secondary objective that stems from this initial one is to find out which economic sectors are significant in fostering economic development. Another goal is to see if population measured by density, size and growth and net migration had a relevant effect for the variation of per capita gross domestic product at metropolitan level. Furthermore, an essential goal of this chapter is to present conclusive information regarding the difference between Western and Central and Eastern European metropolitan regions.
Thesis methodology and the expected results
The methodology of this thesis is an empirical one in the sense that it is using econometric models by which the influence of the main important factors of economic growth in the European Union (at country, regional and metropolitan level) was evaluated. The data for this empirical investigation is collected from renowned international organizations. It is collected from credible sources like the World Bank’s Statistical Database, the European Commission’s statistical database (Eurostat), the Annual Macroeconomic database of the European Commission (AMECO), which process the information gathered from state and private institution.
The main goal of this thesis is to determine the factors that influenced economic growth in the EU after the 2000’s. This investigation involves the use of certain research methods and techniques, as follows:
The documentation and literature review involves the use of references, of theoretical documentation by consulting journals, books, national or international papers. Also, this documentation comprises of further processing and a complex interpretation of the findings.
The mathematical and the statistical methods requires the use of classification, static and dynamic analysis, the correlation between variables, econometric modelling and the use of panel data techniques suited for the models created, graphical representations to show the trend of the variables used in the models, the representation of the minimum, maximum, mean and standard deviation.
The interdisciplinary methods are based on economic (use of economic variables like GDP, FDI, etc. or of economic ratios), econometrics (using certain specification tests for determining the proper models to be used like the Hausman, Fisher, Parm, In-Pesaran-Shin,Breusch-Pagan), mathematics and informatics (the use of the STATA program).
To accomplish the aim of the thesis and to empirically investigate it, the study will want to demonstrate which determinants are the most important in fostering economic growth in the European Union at different territorial levels, namely at country, regional and metropolitan division. A secondary objective of this thesis is to determine if the convergence hypothesis still holds for the EU. The investigation of the thesis will be solved by three empirical investigations in Chapter Three, Four and Five and these chapters have the following methodology:
Chapter 3 will want to provide conclusive information regarding the variables that determine economic growth for the 28 European Union countries from 1990 to 2014. It will empirically investigate the relationship between the dependent variable, real gross domestic product per capita and the independent variables life expectancy, final energy consumption, financial sector leverage, general government debt, total general government expenditure, government deficit, employment rate, exports, imports, trade openness, private sector debt, real labour productivity per hour worked, gross fixed capital formation, foreign direct investment, inflation, population size, primary, secondary and tertiary education. The investigation will use several dummy variables to measure if the governance indicators (control of corruption, absence of violence of terrorism, government effectiveness, rule of law, etc.) used by the World Bank have an influence on economic growth and also if there is any difference between Western, Eastern, Northern, Southern or Western Asian regions. The study will use a dynamic panel data model and the variables will be legitimated using the neglog transformation which doesn’t drop observation form the panel. The chapter will highlight a summary statistics for the variables (mean, std.dev, observations, etc.) and the correlation matrix. Some preliminary tests will be conducted to determine what kind of econometric model will be properly used for this investigation, like the Fisher and Im-Pesaran-Shin unit root tests, Hausman, Breush- Pagan/Cook-Weisberg, LR for panel-level heteroskedasticity, Wooldridge tests. The Pesaran, Frees and Friedman test will be also computed to determine the cross section independence. To offer more robustness of the results, several panel date techniques will be used, namely the pooled OLS, REM, FGLS and FMOLS regressions. The investigation will use also the GMM and system GMM estimations. These two methods are popular for dynamic panel investigation. They increase efficiency of estimation, are suited for autocorrelation and heteroskedasticity and try to fix endogeneity biases. The estimated results of this chapter will demonstrate if government expenditure, trade openness, productivity, foreign direct investment, employment, energy consumption, gross fixed capital formation and life expectancy have a positive influence on economic growth. The negative variables that should hinder growth should be public deficit, imports, public and private debt. The governance indicators should show that control of corruption, rule of law, absence of terrorism have a positive influence on growth and that because of the lag between the investment in primary and secondary education and the outcome on GDP, tertiary education should be the most important schooling variable. Finally the methodology will demonstrate if the model is properly fitted by conducting some normality tests – Ramsey, Shapiro Wilk W and by plotting the residual.
Chapter 4 will continue the investigation by empirically analysing the factors that determine economic growth for 98 NUTS 1 and 273 NUTS 2 regions from 2000 to 2013, regions that are in the same 28 European Union countries as in the previous chapter. The study will empirically investigate the relationship between two dependent variables used to measure economic growth, namely real GDP/capita and real GDP per inhabitant in purchasing power standard. The independent variables are population size, fertility rate, life expectancy, early leavers from education and training, persons with tertiary education, average hours of usual weekly hours of work at main job (male and female), employment rate (total, male and female), R&D expenditures, infrastructure (motorways and other roads), total nights spend by residents and non-residents in tourist accommodations, the stock of vehicles, population density and net migration. This study will also use a dynamic panel data model and neglog transformation. Some preliminary tests will be conducted to determine what kind of econometric model will be properly fitted, like the Fisher unit root test, Hausman (to decide between a fixed or random effects model) and Parm tests. To offer robustness the GMM, system GMM and QML methods will be used. The quasi-maximum likelihood method does not use any instruments like the GMM or system GMM methods. Also the weak instruments that may be used in the GMM and SysGMM are avoided in QML estimation. The expected results of this chapter should demonstrate that fertility rate, life expectancy, persons with tertiary education, employment, R&D expenditure and infrastructure endowment are positively correlated with regional economic growth. Also, agglomeration should yield positive results as stated by the findings of the agglomeration economies theory (Dogaru 2015). Regional economic growth would be positively influenced by domestic and international growth and that domestic tourism plays a more important role than international tourism at regional level (Paci and Marrocu 2014). By graphically illustrating the variation of GDP/ capita between 2000 and 2013 the chapter will also highlight if the convergence hypothesis holds. Because of the 2008 crisis, many regions saw a drop in output and this in turn affected GDP.
Chapter 5 will investigate the variables which played an important role at the metropolitan level in the European Union. The investigation is carried out for 14 years (2000-2013). The chapter will empirically investigate the relationship between two dependent variables used to measure growth, namely metropolitan real GDP/capita and real GDP per inhabitant in purchasing power standard. The explanatory variables that will measure the impact of certain economic branches on economic growth are the share of metropolitan gross value added of agriculture, forestry and fishery, industry, manufacturing, construction, wholesale and retail trade, transport, accommodation and food service activities and finally, information and communication in total metropolitan gross value added. Other independent variables are the number of employees, population size, density and growth, economically active population, net migration and a dummy variable that controls for the effects of European enlargement. This study will also use a dynamic panel data model and neglog transformation. The Hausman test will be used to determine if the model is a random or fixed effects one. The Parm test will be used to show if the model needs time fixed effects. Like in the previous chapter the GMM, system GMM and QML methods are used. To further improve efficiency and offer more robustness the study will opt to split the time period in two (2000-2008 and 2008-2013) and also divide the panel sample so as to measure the difference between Western and Central and Eastern metropolitan areas. Because of the high level of industrialization, the results should demonstrate the determinant role of industry in shaping economic growth. Because the chapter uses population size, growth and density the empirical outcomes should show if agglomeration plays a significant part in advancing metropolitan economic growth. Finally, with the considerable migration wave which the EU was confronted since 2015, it is important to see if this variable positively impacts economic growth
Research limitations
The concept of economic growth has some inherent limitations and we should not avoid underlining these facts because they may have serious economic and social repercussions. For this thesis a first limitation can be the problem of measurement or the occurrence of systematic errors which can have a negative effect on the outcomes of any empirical analysis and can bias the results.
Another problem regarding economic growth analysis is the fact that missigness has been always a common occurrence, especially for large panel data. For regional and metropolitan data missing values can affect the empirical results of the analysis. The study tries to overcome this problem by using several panel data techniques and by applying the quasi-maximum likelihood which is better suited to overcome this bias.
An important topic for future analysis of the models constructed in this thesis is the necessity to include spillovers which measure positive or negative externalities. For example, knowledge spillovers which are created by companies or institutions can affect other firms or institutions and can lead to more economic growth. In the category of spillovers we can find other types like industry, environmental or spatial spillovers.
CHAPTER 1 THE LITERATURE REVIEW CONCERNING ECONOMIC GROWTH THEORY. THE MAIN DETERMINANTS
1.1. Introduction
In general, there is consensus on the need for sustainable economic development. Growth theories and econometric models draw attention to the various ways in which the present economic activity can influence the future development of a nation/region and identify sources that may lead to continuous growth. These theories have developed over time, relying on the dynamics of the economic reality and the evolution of economic analysis tools (Boldeanu and Constantinescu 2015).
In this chapter the study will focus on highlighting the main growth theories and the stages in developing new ways of measuring and refining the models to suit the actual period. This will be achieved by understanding exactly what economic growth means and reviewing the literature so that we can determine the main factors that help foster or not the growth of a state.
1.2. Economic Growth
The concept of economic growth had over time a multitude of interpretations. Researchers, Nobel Prize winners and public institutions tried to find the proper definition of economic growth. Why should we focus on this dry statistical issue? This is because economic growth is a key factor in the well-being of billions. From the advantages brought by the industrial revolution, advanced countries that experience constant growth help their citizens to live well and longer. We cannot say the same for the poorer countries that have now in the 21st century a GDP/capita lower than the one Europe had in the nineteenth-century.
Economic growth is the pinnacle of the twentieth century. Entire nations continue to see it as an extremely important objective economically and politically, the only factor that ensures the economic success of a nation in the long term.
At first glance, we are tempted to limit the definition of growth in gross national product per capita growth, but we are better characterizing growth by improving living standards.
The World Bank defines economic growth as an expansion or a quantitative change of a state’s economy. Economic growth is generally calculated as the percentage change of gross domestic product or gross national product during one year (World Bank Publications 2004). Denison (1962) considers that economic growth is an expansion of real GDP or GDP per capita, an increase of the national product that is calculated in constant prices.
Economic growth can have two different forms: if the growth is "extensive" then the economy uses more capital (resources) in the form of human, physical or natural capital or the economy can grow "intensively" by utilizing the above mentioned resources productively (an increase in efficiency). If the expansion of the economy is due to the utilization of more human capital (labour), then there is not per capita growth. However, if this economic growth is a result of employing all these resources the outcome is an increase in per capita income growth. This leads to a development of living standards (World Bank Publications 2004).
We have to consider that in the first case, the economic growth approach is made in terms of production and not through the well-being of citizens (the consumers). From this standpoint, the use of gross national product (GNP) as a whole or per capita seems lacking. A series of elements considered as consumer expenditures would be better to be passed as investments, such as health and education spending. Also, some public expenditure – expenditure for arming, national defence – should be excluded. On the contrary, we should take into account household services expenditures and "do-it-yourself" activities (Toffler 1981). The amount of free time should be also included, as a value of externalities – positive or negative.
William Nordhaus and James Tobin have tried to make these correlations, creating a new indicator – the measure of economic welfare – which is estimated to be modified in the same way as GNP but with a lower rate (Nordhaus and Tobin 1972).
Robinson defined economic growth as an increase in aggregate product, total or per inhabitant, without any modifications in the structure of the economy or in social or cultural values (Robinson 1972).
In my opinion economic growth should be considered a long-term process because the production capacity can be increased only in the long-run. Sustainable long-term growth in operational terms is an increased in environmentally net domestic product, keeping in mind some specific conditions (Bartelmus 1994).
Economic growth is determined by direct factors such as human resources (the increase of the active population, education – investing in human capital), natural resources (underground resources, soil, climate conditions), the increase in capital employed or technological changes/advancements and indirect factors such as institutions (private administrations, financial institutions, etc.), the size of the aggregate demand (the absorption capacity of the internal market), the efficiency of the banking system, investment rates and saving rates, the migration of labour and capital, fiscal and budgetary policies of the state and the efficiency of the government (Boldeanu and Constantinescu 2015).
Economic growth in the long term has two major sources:
Quantitative growth of production factors (the number of people, the amount of fixed or working capital used). It is also called extensive growth;
Qualitative growth factors, i.e. the factors of production efficiency (productivity thereof). It is the result of intensive economic growth.
Thus, economic growth is derived from existing quantity and quality of human capital (labour force). Consequently, the quantity of human capital may lead to a sustainable growth in economic terms only if the stock of capital increases. Otherwise, increasing the amount of work force used when the capital stock remains the same can lead to a less efficient utilization of the production factors. This in turn can have a negative consequence on the production per capita (because of decreasing returns).
The human capital quality refers both to the qualification, degree of culture, and also to the health of the population, its longevity.
Increasing the amount of capital used increases the amount of goods and services that are made in a national economy, but in the same manner as the increasing amount of labour. Technical progress, described as a ”residual factor” in the established economic models (that of Robert Solow and those that followed him), is one of the most significant determinant of economic growth, an important source of growth and productivity (Solow 1957).
Economic growth brings economic benefits. These include the improvement in living standards. Increasing the volume of final goods and services in a country is usually equivalent to a change in consumption. Long-term economic growth brings an increase not only in the quantity of the good and services consumed but also in their quality. Furthermore, if the income distribution is correct (no corruption and other inequalities), economic growth can alleviate poverty. Increasing production capacities can generate more jobs and therefore more numerous sources of income for households.
Another benefit is the change in the consumption structure. In recent decades, in countries with developed economies, growth has led citizens to focus more and more on other needs like education, culture, leisure, communication, etc. because the physiological and safety needs like food, shelter, water etc. (Maslow 1943) are ensured by a part of their income.
However, economic growth has its costs. Pollution is one of the most significant costs of economic growth in the nineteenth century and especially the twentieth century. A lot of funds are allocated for environmental protection (water treatment, waste management, pollution abatement) that the state has to allocate each year to treat the hazardous problems of pollution.
The allocation of resources for growth means the allocation of opportunity cost. The opportunity cost of economic growth consists of sacrificed current consumption. In a world dominated by scarcity almost nothing is free. Such growth requires a massive investment of resources in capital goods, which do not always cause immediate benefits (i.e. the investment in education), thus the present generation is assuming some sacrifices.
The social costs and personal costs of economic growth are also important to be highlighted. An economy that traverses a process of growth is a changing economy. Innovation makes some cars to wear-and-tear (moral) and also people must constantly adapt to changes.
The new technologies require new skills and intensive training and those who will not be able to adapt will suffer (will lose their jobs). The progress will be more suited for the younger generation.
Also a downside can be considered the distribution of costs and benefits of economic growth over time. The costs of technological changes tend to be generated approximately immediately, unlike the benefits that are felt in the future.
1.3. The theories of economic growth
There are four main determinants influencing economic growth and these are: human resources (including labour, education), natural resources, capital formation and technological progress, but the influence that economists had conferred to each factor was always different (Boldeanu and Constantinescu 2015). Prominent economists (Smith 1776; Ramsey 1928; Young 1928; Schumpeter 1934; Knight 1944) presented, over time, the most essential ingredients which appear in modern economic growth theories. Some of their ideas cover competitive behaviour approach and dynamic equilibrium, the role of decreasing yields and their relation to labour force and physical capital accumulation, the population growth rate and the rate of per capita income, the effect of technological progress in deepening specialization of human capital and the introduction of new products and methods of production and the role of the monopolies – as an incentive to technological progress, etc.
Understanding the concepts and factors involving growth implies the investigation of the theory of economic growth, imposed by the demands of economic life, in the struggle with the restrictions of natural and man-made environment. It took all these growth problems so as the concept of economic growth to be addressed and studied.
a. The classical theory of economic growth
British classical authors studied the first elements of a theory of economic growth. Classical economic models have described the evolution of the economy in terms of growing population and limited land (Smith 1776; Malthus 1798; Ricardo 1817). The model first elaborated by Adam Smith (1776) and developed by Malthus (1798) had an agrarian under layer. While the lands were free the population growth was unlimited. All individuals in this manner could obtain through their work enough production to survive and support their families.
Classical theoreticians were confronted with the rapid changes in their societies. The feudal system started to be replaced by industrial capitalism. By developing the field of political economics, they tried to explain the main forces that govern the new economic system. Also, there was a philosophical concern regarding the notion of progress and what exactly are the social forces that can help to foster economic growth. The forces that hinders economic growth was also a big concern. The highest level of thought for the classical theory was reached by David Ricardo (1817) in his works.
In “Wealth of Nations”, Adam Smith (1776) introduced the basic concepts of economic development. He began by describing a time when the earth was free for all people and capital accumulation did not matter. Since no capital, with production doubling the population doubles. Salaries monopolized the entire national income, as there was no rent for lands and there was no interest on capital. Production increases in the same pace as the population, so that real wages per worker remain constant over time. In the absence of technical progress, population growth leads to a total usage of the land that is available. With an increasing population density, the law of decreasing returns means that there was no more land and income for workers. Economic growth leads to higher land rent and, in parallel, lowering wages.
Ricardo (1817) was researching the main laws that govern distribution. He thought that:
„Political economy is an inquiry into the nature and causes of wealth. I think it should rather be called an inquiry into the laws which determine the division of the produce of industry among the classes which concur in its formation (Hedlund 1983)”.
David Ricardo analysed profit as a residual component of the surplus. Furthermore, he eliminated the rent component that was calculated as a difference between the product on marginal land and the one on intra-marginal units (Harris 1978). Ricardo found a link between the total rate of profit and the wage rate. This outcome was determined only for the sector of corn production. Corn was a commodity that could be both as a wage to be paid for the work force and also a capital good that enters into the production cycle.
When all the land was cultivated, the further increase in population meant that the number of workers on the same surface was supplemented. Each new worker produces an amount of supply that decreases over time. The reduced marginal production of labour implies that the real wage decreases. How long can such a situation worsen? Malthus believed that population continues to grow as long as wages are above the level of subsistence (Malthus 1798). The pressure of excess population can have big implications on the economy in which wages can fall below the subsistence level, which leads to increased mortality and the reduction of population.
According to Malthus:
” It is, undoubtedly, a most disheartening reflection that the great obstacle in the way to any extraordinary improvement in society is of a nature that we can never hope to overcome. The perpetual tendency in the race of man to increase beyond the means of subsistence is one of the general laws of animated nature which we can have no reason to expect will change” (Malthus 1798).
Malthusian equilibrium is reached when the salary falls at subsistence level, below which a job offer is no longer reproduced at the same level and the economy remains in a stationary state. In the opinion of Malthus (1798), in the absence of a real increase of per capita production, most of the population was condemned to live at a subsistence level achieved by adjusting the mechanism by demographic dynamics. According to Ricardo's (1817) theory, the earth was not the object of accumulation, but a source of profits for the owners. Capital was considered rather a substitute for labour and not a source of growth in productivity. Classics underestimated or omitted the contribution of technical progress in increasing productivity.
The Reverend Thomas R. Malthus was an Anglican minister, a man of deep moral and religious convictions. He considered the existence of three forms of controlling the growth of the population:
Poverty, i.e. hunger, disease and war;
Sin, i.e. the release of human passions through sexual practices that do not lead to procreation
Moral self-restriction, i.e. sexual abstinence.
The latter was the solution that the Reverend Malthus (1798) proposes for solving the fast growing population.
Marx refuted this theory implying that the population can be ”recycled” through its replacement mechanism from existing employment due to structural shifts in production and mechanization (Harris 1978).
An important contribution made by classical researchers for the economic growth theory was the understanding that the productive and accumulation of investment is an important part behind economic growth. In the capitalism system this type of investment is mainly the profit reinvested.
b. Keynesian theory, post and neo-Keynesian
In the early twentieth century amid serious economic and social distortions that culminated with the economic depression of 1929-1933, the research was intensified towards new theoretical concepts for the dynamic analysis of macroeconomic processes. The first and most significant reaction to adapt to the requirements of these economic realities belongs to the English economist J.M. Keynes. The Keynesian growth model is a macroeconomic model according to which national income grows in response to increasing aggregate demand. Keynes used new concepts relating to economic growth (Keynes 1936).
J.M. Keynes had very good knowledge in both economic science and mathematics, constructing a descriptive economic and mathematical model so as to facilitate the understanding of economic activity and the relationship between its different variables.
The model consists essentially of three types of elements:
Variables that consist of either a series of macroeconomic indicators (national income, demand, supply, consumption, savings, investment, etc.) or their rates;
Relationships between variables constructed using equations (definition and behavioural) and inequalities, and the interdependency of them constructed using functions (employment function, supply function, the actual function of global demand, investment function, etc.);
The "investment multiplier" (K) with which you can express the intensity of the influence of a variable.
J.M. Keynes considered two types of variables: endogenous variables and exogenous variables (Keynes 1936).
endogenous variables are global or aggregate indicators (macro categories), which characterize the economic activity, at the national scale, namely:
global supply price of obtained production (Z);
effective demand for goods (D) or the total proceeds from the sale of goods obtained by the entrepreneurs. Effective demand in turn consists of two parts: the demand for final consumer goods or individual goods (D1) and the demand for goods necessary for investment (D2);
global income (Y);
global final consumption (C);
savings (S);
global investment (I);
the amount of labour (N).
Keynes's (1936) opinion is that effective demand for goods (D) is the most important of the endogenous variables for the economy to function. This is because the other variables depend on the level and change of demand, their mutual interdependence stands under the influence of exogenous variables.
exogenous variables (independent) consist of a series of rates (proportions), data on the behaviour of economic agents, both as consumers (propensity to consume) and as entrepreneurs (marginal efficiency of capital and interest rates). The propensity to consume is regarded as the ratio between consumption and income (the average expression C / Y) or between the consumption growth and growth of income (marginal propensity to consume ΔC / ΔY) with its reverse – inclination towards savings, in average expression S / Y and marginal expression ΔS / ΔY; marginal efficiency of capital or the percentage of profit at the last investment; interest rate or the percentage of interest paid on borrowed capital (Keynes 1936).
When the state of technology, resources and factorial cost of a unit of labour employed are given, the volume of employment depends in every enterprise and branch, and on the whole, on the volume of revenues that entrepreneurs expected to obtain on the respective volume of the production. Entrepreneurs are interested to set the volume level of employment at the level that they would get a maximum profit. Revenues depend on demand (level and structure).
The global supply function is, if the global price of supply is Z for an output obtained when using N people:
Z = Φ (N)
The global demand function is D = f (N), where D represents the total income that entrepreneurs are expected to obtain using the N persons.
The relationships between the variables mentioned above are constructed using a system of equations and inequalities, with very different roles, i.e. behavioural equations (functions), the fundamental equation of the model, the balance equation, mathematical expressions of various forms of imbalances, etc.
In chapter 8 of the „General Theory”, Keynes (1936) states that the main theme of his investigations and the ultimate goal of the analysis is “to discover what determines the volume of labour employed”.
Keynes (1936) formulated the opinion that the number of workers who find work (N) depends on the effective demand for goods (D) or entrepreneur’s proceeds from the sale of production, i.e. the suitable demand function N = f (D). Given the structure of demand for goods and the link between the volume of effective demand and the supply price (Z) and the size of the global income (Y), Keynes concludes that the sum of the global final consumption (C) and global investments (I) is equivalent to the total income (Y) when the economy is in equilibrium, situation synthetically expressed by the fundamental equation of his model, namely C + I = Y. Whereas in reality, where the economy is imbalanced, Keynes expresses this relationship as C + I <Y which shows that revenues are lower than the production offered, the global demand for final goods (C) and goods for investment (I) is less than the supply Y and Z, that a part of the production could not be sold and thus part of the workers could not find work, involuntary unemployment persists. In search of explanations and solutions, Keynes examines the two components of demand, taking into account the decisions of economic agents, which delay the sale of the entire production owing to the fundamental equation of his model. Thus, he demonstrated, if you deduct the overall final consumption (C) from total income (Y), there remains a part of that income, for example the savings (S), obtaining the equation S = Y – C (Keynes 1936).
If the savings part of the total income would be equal to the investments we could achieve economic equilibrium and may lead to increased employment. S = I can be considered the equilibrium equation. In fact, Keynes observed that not everything that is saved or is accumulated is automatically invested, thus S>I expresses the persistent imbalance in the contemporary economy, the difficulties in selling a proportion of the goods offered on the market and thus maintaining or even increasing involuntary unemployment. One of the main causes of imbalances in the contemporary economy is the large share of savings (accumulation) in global revenue/income. Other causes may stem from the role of money in the market economy, in particular, of the consequences the quantity of money on other variables and the combination of these consequences with psychological inclinations of people or certain groups of economic agents (Keynes 1936).
A big outcome of Keynes’s theory is the investment multiplier. Investment multiplier or the multiplier effect is one of the major achievements of modern macroeconomics. The investment multiplier (K) is the ratio between the increasing levels of production or income (ΔY) and the initial growth of investments (ΔI):
K = ΔY / ΔI
Keynes writes regarding the multiplier effect:
”It is, however, to the general principle of the multiplier to which we have to look for an explanation of how fluctuations in the amount of investment, which are a comparatively small proportion of the national income, are capable of generating fluctuations in aggregate employment and income so much greater in amplitude than themselves (Keynes 1936, P 81).”
The essence of the general theory developed by J.M. Keynes can be synthesized as follows:
If at a certain size of the number of employed labour (N), the expected volume of receipts (revenues) is higher than the global demand price (D>Z), entrepreneurs will be encouraged to increase the volume using labour beyond (D) and if needed, to climb costs and the competition for production factors to the size of (N) to which (Z) equals (D). As such, the use of labour is determined by the point of intersection between global demand and global supply curves, because this is the point at which entrepreneurs expect the profits to be maximized. The size of (D) at the point on the curve where global demand and global supply intersect will be labelled as effective demand, because this is the essence of the General Theory of Employment (Keynes 1936).
In accordance with the Say Law, the global supply and global demand functions will equalize themselves, and the global demand price of the whole production is equal to its global supply price at any volume of production, which would equate to the observation that full employment does not meet any obstacles (Say 1934).
In the early 1940s, R.F. Harrod and Domar developed models in line with the post Keynesian theory, presented later together because of background similarities, characterized by concerns for economic stability and unemployment, as well as short-term analysis (Harrod 1939; Domar 1946).
The dynamic analysis undertaken by Harrod (1939) and Domar (1946) followed two theoretical aspects:
Explaining the quantitative relationship between different technical-economic variables of the growth model used by them;
Identifying different types of growth rates, according to the objectives of individual businesses and also the society as a whole.
The Harrod-Domar growth model was constructed using endogenous variables: income (Y), investments (I), savings (S), capital (K) and the income growth rate (G) and with exogenous variables: population growth, technical progress and labour productivity levels, referred as growth fundamentals. The authors use two new parameters, i.e. capital coefficient and investment efficiency, which illustrates the functional relationship between investment growth and income growth, taking into account both the volume of possible savings and the type of existing technology, therefore, its cost in relation to an increase by one unit of income (Harrod 1939; Domar 1946).
Capital coefficient in average expression (c) and marginal (Cr) was used by Harrod to determine how the additional capital necessary to obtain increased output or income (Y) by one unit, according to the relationships (Harrod 1939):
c = K / Y; Cr = I / ΔY
The marginal efficiency of investment (δ) is a parameter used by Domar to determine how much product or income can be obtained with an additional capital unit invested δ = ΔY / I.
The two parameters in different ways and models express the same relationship between investment growth and income growth, but starting at different points. Harrod’s model has as a starting point the income growth, trying to determine how much capital must be increased to achieve the objective (economic growth) and Domar’s model has as its starting point the capital increase, trying to identify the results that may be obtained (how much is the increase in income?) per one unit of additional capital.
According to Harrod-Domar’s model, the income growth (G = ΔY / Y) is conditioned mainly by the marginal propensity to save (s) that depends on available funds to be invested and the existing technology and its cost, characterized by a marginal coefficient of capital in case the other variables in the model are given (Harrod 1939; Domar 1946).
The fundamental equation of the Harrod-Domar model is: G = s / Cr. This equation highlights the direct link between economic growth and the accumulation share in national income.
Harrod-Domar’s model of economic growth and the corresponding theory tries to explain the contemporary economic instability in the market over time. Harrod distinguished between two diametrically opposed types of growth rates, respectively, warranted rate (Gw) and the natural rate (Gn) (Harrod 1939; Domar 1946).
The warranted rate of economic growth (Gw) is the one that meets the interests of entrepreneurs, even if it doesn’t ensure the full use of all factors of production (specifically the labour force), assuming the existence of unemployment.
The natural rate of economic growth (Gn) is the one that ensures the use of all available production factors, but it may not meet the interests of entrepreneurs.
Actual Growth Rate (G) does not usually coincide with either of the two possible rates. In principle it may be greater or less than the warranted rate and it is usually lower than the natural rate. The oscillation of the real growth rate around the warranted rate and the natural rate express the instability of the contemporary market economy in the opinion of Harrod. When the actual growth rate is lower than the warranted rate (G <Gw), there is recession; when the actual growth rate is higher than the warranted rate (G>Gw), business is good, the economy is expanding; when the warranted rate is lesser than the natural growth rate (Gw < Gn), there is unemployment (Harrod 1939).
To mitigate these oscillations, to correct some insufficiencies or excesses in the effective demand for goods on the market, in the size of the economy, the interest rates and the amount of money in the market, Harrod proposed economic state intervention using the stop and go policy. This is the use of monetary and fiscal policies in order to influence the behaviour of individual economic agents so as to balance the divergent trends. When the economy is too "heated" (increase demand, increase investment, increase income and wages) it is necessary for the state to stem this expansion by raising taxes and increasing interest rates, and when there is stagnation of economic activity (demand for goods on the market is low, investments are reduced) then it is necessary to stimulate economic activity by reducing taxes, increasing the amount of money circulating in the market and decreasing interest rates.
The progress made by Harrod and Domar in macroeconomic analysis is obvious and remarkable. They highlighted a number of contradictions and conflicts that raise serious problems for private companies and state bodies. Harrod-Domar’s model highlights three major problems: the possibility of sustained growth; the likelihood of sustained growth in condition of full employment and the warranted rates of growth. The framework of the model is large enough to incorporate technical progress (Harrod 1939; Domar 1946).
c. Neo-classical growth theory
In the context of an ideologically political economy, the period between 1870 and 1914 was marked by large-scale changes in the history of economic thought. While the classical theory seems to have reached its limits and that Marxism was ignored by most economists, a new theory arose in the 1870s with the works of Marshall, Pareto, Jevons, Menger and Walras. Later called "neoclassical" or "marginal" economists, they generated streams of thought, representations, pedagogical traditions that as a whole were very similar to what Kuhn called a „new paradigm” (Marshall 1890; Pareto 1971; Jevons 1871; Menger 1871; Walras 1954).
Although these theorists were contested, their paradigm that followed will dominate economic thinking in the twentieth century.
The issue of the neoclassical economists was how to deal with the spread of power between the workers/labour and the industrialists. In the 19th century the labour force strength throughout Europe grew. In 1871 England adopted the legalisation of unions.
Neoclassical economists tried to change and improve the theories of Ricardo (1817) and to demonstrate that he was wrong when he stated that profit is a residual. What determines the profit is the level of marginal productivity of capital and the salaries of the work force is determined similarly by the marginal productivity of labour.
Neoclassical economists argued that if unions accomplish to raise the salaries of workers the outcome will be unemployment. Following the same principles of his contemporaries, Pareto dismissed the concept of Utilitarianism that calls for income redistribution and formulated a new economic efficiency definition. This definition says that the higher union salaries grow the bigger is the economic inefficiency.
After the Second World War there was a theoretical synthesis of tackling macroeconomic and dynamic approach regarding economic growth. Thus, there was an outline of the modern economic growth theory, as a component of contemporary economics. For this neo-classical growth theory have contributed many post war economists which had revised and improved on Keynesian and classical theory (Harrod 1939; Domar 1946; Solow 1956; Swan 1956; Samuelson 1948; Kaldor 1957; Dornbusch 1975).
The economic context was far different from the one in which classical theories were formed. Thus, with the entry of capitalist economies in the twentieth century, they have developed important new industries related to electricity, telephone and car. The accumulation of capital and the new technologies have become the dominant forces that determined growth. Growth theory that incorporates capital accumulation forms the core of modern analysis. The neoclassical growth model explains how capital accumulation and technological changes affect the economy.
A new approach was designed by Robert Solow, winner of the Nobel Prize for his contributions to the economic growth theory, which drew in the mid-1950s a stable growth model, inspired by the models of Harrod and Domar (Solow 1956).
In the same period, Kaldor presents a new theory, in which he suggested that there is a possibility to achieve a stable secular growth rate for an economy if the savings rate changes through income redistribution (Kaldor 1957).
Solow's neoclassical growth model is a fundamental landmark in the analysis of economic growth. Through this model, neoclassical economics shows how the increase in the savings rate, population growth and technological progress and economic growth affect production levels over a given period (Solow 1956).
In his study, Solow starts from the following assumptions:
• Perfectly competitive economy;
• The perfect mobility of factors of production;
• Full employment in terms of resources used;
• Capital (production factor) is subject to decreasing returns;
• The returns of scale are constant.
According to Solow’s growth model:
– How much a country invests and saves is a key factor for the living standards of individuals in that state;
– Developing countries which start from a level of capital accumulation less than the optimum steady state of capital, must obtain a higher savings rate. This can be achieved by increasing government savings (reducing public spending and increasing public revenues) and increasing incentives to save for households by lowering taxes.
– Developing countries must allocate more investment in infrastructure, human capital, education etc.;
– To create technological spillovers /externalities requires an effective industrial policy;
– Technical progress should be encouraged and states and companies must invest in new technologies.
The results of the Solow’s growth model are:
The most important contribution to the growth in output is the increase in labour.
In steady-state equilibrium the growth rate of output is the same as the growth rate of population or the labour force and is exogenous of the saving rate. This means that it doesn’t depend on the savings rate.
The rate of saving determines an increase of per capita income in the steady-state equilibrium (determining therefore the total income) by raising capital per inhabitant (head).
The neoclassical theory of economic growth demonstrated that the output is a function of growth in factor inputs, like primarily the capital and labour, and also the technological progress.
The long-run economic growth rate, equal to the steady state rate of income growth per capita is influenced by technological progress.
The total output per head will converge to steady state equilibrium if technological progress doesn’t exist.
An important conclusion of the neoclassical growth theory of Solow (1956) is that if we have two countries that have the same savings rates and the same growth rates of population and they have the same access to technology, their levels of per capita income will eventually converge, that is they will ultimately become equal.
Assessing Solow’s growth model we can find that the model was very important to future empirical work, with a neoclassical model that was simple to understand. Also the utilization of the Solow residual was important in determining the relative impact of capital accumulation and technological progress on income growth. The problem of convergence is also a theme that is discussed in the model.
Some problems with the neoclassical growth model constructed by Solow were that it highlights only a simple picture when it comes to convergence and technological progress is viewed as exogenous. He only explained non-convergence in the model and the vast differences across countries regarding income and technology. Also, technological progress is just a residual and under neoclassical view the accumulation of physical capital cannot explain the long-run growth.
Based on the works of Solow (1956) and Kaldor (1957) there were developed a large number of econometric models and their variants, in which technical progress was included in its various forms. The technological progress was considered for the most part to be exogenous.
The accumulation of capital and labour is the essence of modern analysis of the neoclassical economic growth model. It uses an instrument known as aggregate production function, which relates the labour and capital factors with potential GDP size. In the absence of technological progress and innovation, increasing the availability of capital per worker is not followed by a proportional increase in output per worker because of diminishing return to capital. Therefore, the sustained growth of capital contributes to a decline in its rate of return.
d. The new growth theory
In the mid-1980s, the new theory of economic growth emerged, promoted by Paul Romer and Robert Lucas on the lines of the traditional neoclassical growth model (Romer 1986 and 1990; Lucas 1988). The premise of this theory is that capital accumulation is usually associated with an accumulation of knowledge.
The new growth theory treats the variation of technology as an endogenous variable that responds to market signals. Investment in human capital, education, research and development, etc. gives positive externalities. The sequential increase in investment could yield growing returns, increasing the volume of total production.
The new theory of growth brings significant changes:
Technological progress is considered a product of economic activity, contrary to previous theories that stated that technology is produced by outside market forces. It was treated exogenously. The new theory is called the theory of endogenous growth because internalizes technology into a working model of the economy;
It is considered that, unlike objects of physical nature, knowledge and technology is characterized by increasing returns. Ideas can be shared and reused indefinitely and they can increase and propel economic growth without a limit. The new theory reflects the transition from a resource-based economy to a knowledge-based one.
Assuming that the emergence of new knowledge is an essential source of economic growth, it is permissible for small events, but at the right time, to change the trajectory of growth. The new theory contradicts the concept of unique and optimal equilibrium, which implies a reduced ability to forecast future results.
In 1987 Paul Romer (1987) designed the AK model. It implies a proportional output with the capital stock in an economy. The capital is accumulated according to the equation of the Solow model, namely: Δ K = s Y – δ K, considering the rate of population growth to be zero. The economy is in a situation of steady and balance growth whatever is the use of capital stock; its increase does not stop, because the total investment is always higher than the depreciation (s Y > δ K). This is a model in which capital has constant returns, so that the marginal productivity of each additional unit of capital employed is always A.
The income growth rate is also an increasing function relative to the rate of investments therefore public policies that increase this rate have a positive influence on the process of economic growth (Romer 1987).
There are several models of endogenous growth. Lucas's model from 1988 utilizes a production function like the following:
Y = K α * (hL) 1-α, where h represents the human capital per inhabitant (capita).
The Lucas model assumes that human capital derives according to the following equation: ∂ h = (1-u) *h, where u is the time spent working and (1-u) is the time allocated to training. Consequently, the latter increases in a permanent way the rate of growth of human capital ∂ h / h = (1-u) and also of income.
In other endogenous growth models, technical progress is the product of a specific sector of the economy, like the sector of research and development, which is using part of the resources of an economy. A sustained economic growth exists only if the number of new ideas created and applied in each period is increased. This result is similar to that achieved in the Solow (1956) model with technical progress. But the mechanism is not the same, this being an endogenous creation of new ideas: the larger the number of people who generate these ideas, as workers in research and development, the bigger will be their impact, because using ideas is unrivalled and non-exclusive.
Contributing to the increased production potential, technical progress allows a higher output with the same inputs of labour and capital. The theory confirms what economic history shows us, that technical progress brings higher productivity of factors of production, wages and living standards. Using quantitative techniques, economists have demonstrated that innovation and education have a greater impact on labour productivity or GDP growth than the growth of capital.
e. The theories concerning the role of institutions
Other approaches to the economic growth process highlights the strengths that are beyond the abstract formulas of econometric models. In this category we mention concerns about growth forecast through the historical evolution of national economies, like the stages of economic growth of Rostow (Rostow 1960) or empirical statistical studies that highlight, in a descriptive manner, sociological factors, demographic and institutional issues of growth.
Empirical research on the phenomena of growth and transformation has begun in the 1960s by Simon Kuznets, Nobel laureate. He has dealt with long-term processes in a comparative analysis of the growth experiences of nations. Kuznets introduced the concept of modern economic growth, which is more broadly defined than other specialists, because he included also institutional changes that favour structural modifications. Comparative analysis allows the establishment of common traits and the construction of a model and identification of model biases. According to Kuznets's assessment, there are joint trans-national factors and an interaction mechanism through which growth occurs in the modern world (Kuznets 1973). The main highlighted trans-national factors, which are common in present national economies, are:
Industrial system or production system based on the potential of the modern technological production capacity. Some of the system’s requirements are: a minimum level of intellectual preparation, a type of non-family organization and a high degree of urbanization;
Human community desires and aspirations. This is manifested by a relatively low resistance to the dissemination of modern technology and the widespread desire to achieve greater standards of economic performance and higher living standards;
Organizing the world in nation states. Specific national factors common to nation states are the size and location of natural resources, religion and historical heritage.
1.4. The main determinants of economic growth
Modelling economic growth refers to the development and use of economic and mathematical models, either for theoretical purposes, for description and explanation, either for pragmatic purposes, for forecasting and directing the growth process.
Economic growth models represent the formalization, in mathematical expression, by functions and specific parameters, of the relationships between factors and outcomes of economic growth, highlighting the mechanisms, intensity and its trends.
Characteristics for the economic growth model are the following basic elements:
a) These are macroeconomic models, covering all the national economic landscape. In this regard, it is noted that this class of models operates with macroeconomic indicators of effort and results and also reflects the overall structure and functioning of the economic system;
b) Models of economic dynamics that reflect the changes in time of the values of various parameters specific to economic growth and the correlation between them. Essential to the dynamic nature of growth models is the concept of economic growth trajectory, signifying a succession of moods, actually achieved or expected in the development process of growth on a specific time horizon. Due to its outstanding complexity, the process of growth is difficult to be represented by a simple model.
Growth models and theories can be classified according to several criteria, but I detail the two most important in terms of teaching, namely: the level at which they address the problems and the theoretical framework at which they start (Keynesianism, neoclassicism, Marxism, radicalism).
In terms of the degree of aggregation, growth models can be divided in one sector analysis or multi sector analysis.
Mono sectorial models do not differentiate the inputs and outcomes on economic sectors, the abstraction being higher. In turn, the indicators of effort and results take, in this case, the highest aggregate expression and the combination of factors is considered to be of the same nature throughout the economy.
Multi sectorial models make differentiation between sectors. They differ in the transformation functions of growth factors in the results and the specific contribution of each branch to obtain synthetic macroeconomic indicators. In general, such models are operating with variables like: per capita gross national product and its dynamics, the capital accumulation rate, the volume of capital and labour, the coefficients of substitution of factors of production, investment volume, import and export etc.
Like I mentioned in the beginning of this chapter, economic growth measured as GDP is the increase in the value of goods and services that are produced by a country from one period to another.
There are different ways to calculate GDP. Using the expenditure approach yields the following formula:
Gross domestic product calculated by the expenditure approach takes into account four main components. The consumption (C) represents the total personal consumption broken into services, durable and non-durable goods.
Investment (I) signifies the gross private investment that can be broken in fixed capital investments and changes in inventories.
Government (G) represents the government expenditures on consumption of goods that are used in the current period (current expenditures) and capital goods. Government spending doesn’t take into account the transfers, because they are not part of the current production of goods.
Net Exports (X-M) are calculated as a difference between national exports and total imports. Exports are goods and service that are produced in the home country and sold to other nations and imports are good and services that the home country buys from foreign states. This difference may be positive or negative.
The determinants of growth are inter-related factors that influence the rate of growth of an economy, in a direct way by increasing the real GDP of an economy. Six major factors determine economic growth with four of them being grouped under supply determinants and the other two are efficiency and demand (Boldeanu and Constantinescu 2015).
The four supply variables that influence economic growth are natural resources, capital goods, human resources and technological change. These determinants have a direct effect on the value of goods and services supplied.
Natural Resources represent anything with economic value that can be exploited in nature and they can have a direct effect on the quantity or quality of products. Major resources that can determine growth are for example energy resources (gas, fuel or coal), rare and valuable metals or wildlife.
Capital Goods represent fixed assets such as machinery and plants that can be put in use to produce goods and services. For tangible assets we need an initial investment and the increase in productivity means, if the capital spending was efficient and well prospected, an increase in growth rates in the future.
Human Resources are skilled and unskilled labour force. A rise in the quantity and quality of the labour force means an increased rate of economic growth. When the workforce is raised they produce more goods and services and when the workforce increases in its work skills it produces goods and services with high value.
Technology represents methods and procedures that are used to produce certain goods and services. By changing the current technology with a better one or by improving the one we already have, production can be increased or by doing it in a more efficient way. This in the end means a future growth of the economy.
Efficiency means achieving greater output to input ratio. For a maximum economic growth rate, the economic system must use the available resources with a minimum cost to produce an optimal mix of products and services. The resources must be used at the maximum extent possible.
The demand factor means that the increase in the supply of products and services has to suit the needs of consumers by an increased in the demand for the goods and services that the economy produces.
Economic growth measured by gross domestic product signifies the increase of the growth rate of GDP, but what influences the rise of each component is very different. Public and private factors have different outcomes on economic growth. Public spending, inflation, capital formation, private investment, employment rates, etc. have different consequences on economic growth and we should take into account that these factors have different implications if the countries are developed or not. There are also socio-political factors and events that have a major influence on the economic advancement of a nation (Boldeanu and Constantinescu 2015).
The difference between economic and non-economic factors is made also by the research literature. “Proximate” or economic sources are referring to variables like capital accumulation, technological progress, labour and “ultimate” or non-economic sources are referring to determinants like government efficiency, cultural and social factors, institutions, political and administrative systems, geography and demography (Rodrik 2003; Acemoglu et al 2005; Arvanitidis 2007; Acemoglu 2009).
Acemoglu formulated this question in his 2009 research:
”if technology, physical capital and human capital are so important in understanding differences in the wealth of nations and if they can account for differences in income per capita across countries, then why is it that societies do not improve their technologies, invest more in physical capital, and accumulate more human capital?” (Acemoglu 2009)
Because the economic growth theory is not a unified one, the empirical research is multi-theoretical based and there are contradictions in the results and findings of different authors regarding the determinants of growth.
Research studies analysed the implications on economic growth of such determinants like investment, human capital, R&D, economic and fiscal policies, trade openness, FDI, institutional and political framework, socio-cultural factors, geography and demography. In the next part of this chapter the thesis will highlight the main findings and ideas of the research papers that studied the main determinants of economic growth.
1.4.1. Public expenditure
Countries apply different strategies for sizing public spending depending on the national, but also the international context. Therefore, in times of uncertainty, like for example economic crisis, governments have to dispense of certain investments so as to fund social and economic activities that are needed to support its development and to help protect its vulnerable citizens. The level of development of a country has a serious influence on the structure and volume of public expenditure, since developing countries need a bigger increase in public spending to minimize the gap between them and developed countries.
In the economic growth literature there are several conflicting views about the outcomes of public spending. In addition to providing social protection and transfers to maintain an optimal level of social welfare, the government invests in the economy both in the public sector (infrastructure, budgetary resources) and in the private sector to increase productivity.
Short-term unproductive expenditure in education and health may facilitate long-term growth of labour productivity. In addition, the government can provide information to the economic environment may reduce financial risks and change the incentives received by different companies. But some economists believe that public goods provided by the state can be ineffective.
Also, there are negative effects on economic growth by changing tax (if the tax increases or cuts are not well understood, are not implemented at the right time or the effects are not properly measured) and by the transfer mechanism between ministries. Increased taxes can cause a misallocation of surplus funds and cause a constraint to businesses.
Recent empirical research (Devarajan et al. 1996; Brasoveanu et al. 2008; Arpaia and Turrini 2008; Holzner 2011; Afonso and Alegre 2011) shows that for economic growth it is more important the composition of public spending than their overall level and in turn giving policy-makers in the government a clear picture how to better intervene in the economy and to obtain sustainable long-term economic growth (Boldeanu and Tache 2015).
In the process of accomplishing its functions and tasks the state has to ensure to satisfy the general public needs through financial relations. Mobilizing financial resources is realized by public expenditures.
Public spending refers to all expenditure used by all public institutions, which cover either the state budget, local budgets or extra-budgetary funds or the budgets of institutions (that finance themselves by their own revenues).
Unlike all public spending, budgetary expenditure is only part of the public spending and applies only to those expenses that are covered by the state budget, local budgets or from social security funds.
The public expenditure system includes all expenditure that is made from the funds of the state for economic activities, social – cultural, for the maintenance of state bodies, national defence, etc.
State expenditures materialize in payments from resources mobilized in various ways, for purchases of goods or services necessary to meet the various objectives of state policy: general public services, social-cultural, defence, economic activities etc.
State public expenditures are classified in several ways. If using the administrative classification the criteria is based on institutions that allocates expenses, if using economic classification the expenditures are current or capital, if financial classification is utilized the expenses are definitive, temporary or virtual. Regarding the role of public spending in the economic reproduction the expenses are real expenditures (negative) or economic expenditures (positive). There is another classification used by the UN for classifying public expenditures.
In 1999 the OECD has designed the latest version of the functional classification of government expenditure and that was afterwards published by the UN’s Statistics Division. Their method of classification is used by many states and by international originations. The European Commission’s statistical database, Eurostat, uses also this classification for EU public spending
In accordance with the functional classification (COFOG – Classification of the Functions of Government), public expenditure is grouped into 10 classes (defence, health, social protection, etc.), and each class has up to 9 subclasses. This classification is made according to the national accounts system, which groups together public spending by nature and its utilization.
Using the 10 groups of public expenditure, Tsouhlou and Mylonakis (2011) analysed the evolution and structure of public expenditure in each EU country and also the change in GDP during 1996-2007. The results showed that all of the EU countries reduced their public expenditure event if a majority of them had a medium size public sector. He concluded that:
”The average rate of growth in EU countries with large public sectors ranged between 1.4 and 3%. In countries with medium-sized public sectors the rate of growth ranged from 1.5 to 4.6%, while in countries with small public sectors it ranged between 3 to 7% (Tsouhlou and Mylonakis 2011)” (the conclusions were made by comparison between available data for the 27 EU countries without any econometric analysis).
The classes and subclasses of public expenditures for each type of expenses are shown in the following table (OECD 2011):
Table 1: The UN Classification of the Functions of Government
Source: unstats.un.org
The theoretical and empirical research on the impact of public expenditure on economic growth is still an important topic for economists. After the 1990s the analysis of public expenditure was intensified with contribution to the field made by Robert Barro and Sala-i-Martin (Barro 1990, Barro and Sala-i-Martin 1995).
Endogenous growth theory has introduced government spending into the production function of firms or into the utility consumer function. It has demonstrated that public expenditure can have permanent, positive and significant influence on long-run economic growth. Also the decomposition into components of public spending (health, education, defence, public order and safety, etc.) was analysed by many empirical papers.
Like it was mentioned earlier in this subchapter, there are conflicting views regarding the influences of government spending components on economic growth. For example Ghosh and Gregoriou (2008) and Benos (2009) had different outcomes of their research using the same method of analysis, respectively the GMM technique. The first authors showed that the current component of public expenditure had a positive influence on growth for their sample group of 15 developing countries. Meanwhile, the other researcher concluded that infrastructure and human capital had a significant effect on long-run growth for a group of 14 EU states. This will be achieved only by a reallocation of the components of public spending.
Another important research paper that focused on the different impact of current and capital components of public expenses on economic growth was the one of Devarajan, Swaroop and Zou, published in 1996 (Devarajan et al. 1996). Devarajan investigated how changing the structure of government spending would affect GDP. The article was considered innovative and the information they produced was used in 2010 by three professors from Japan to build a mathematical model for maximizing economic growth (Miyakoshi et al. 2010).
Devarajan et al.'s (1996) research study was based on a panel of forty three developing countries. The authors analysed the period between 1970 and 1990. They used as government variables: defence, education, health, transport and communications in both current and capital investments measured as percentage in total public expenditure in GDP.
To cancel the random fluctuations of GDP, the authors use a moving average mean of 5 and the public expenditures were constructed with a lag of five to per capita GDP as spending needs time to affect the economy.
Devarajan et al. (1996) considered that the influence of public spending depends not only on the nature of the investments–productive or unproductive expenditures – but also theirs percentage size in GDP. In conclusion the authors found that the actual growth of public current expenditure was positively determining the variation of economic growth. Capital expenditures have a negative influence on GDP per capita and therefore these expenses even if productive, used in excess, can become unproductive for the economy. Their empirical results have shown that developing countries wrongly allocated investments relying more on capital expenditures at the expense of current expenditures.
Pollard et al. (2012) analysed 150 countries using the Penn World Table 7.0 and showed that there is an inverse relationship between government expenditure and economic growth. Government expenditure was either used inefficiently by the state or not for promoting economic growth. They stated that there was also a difference between regions (convergence club hypothesis) and that slow growing nations of a region are catching up to the richer nations that are in the same region.
Brașoveanu et al. (2008) carried out an empirical investigation to capture the link between government spending (as percentage of gross domestic product) and economic growth (measured as the growth rate of real GDP and per capita GDP) for Romania between 1990 and 2011. The authors employed the COFOG classification and split government expenditure into three classes: productive expenditure (that have a positive effect on economic growth), unproductive expenditure (hindering effect of economic growth) and other expenses.
First off, Brasoveanu et al. (2008) assumed that economic growth is negatively linked with government expenditure. The authors also consider that this is explained due to a decreasing relationship between government size and the level of development of a country (Boldeanu and Tache 2015).
In accordance with the economic growth model formulated by Barro and Sala-i-Martin (1995), the Romanian researchers chose to divide the ten government expenditure in three categories. The productive expenditures were: general public services, defence, public order and safety, education, health, housing and community amenities, environment and water protection, transport and communications. The unproductive once were: culture, recreation and religion, social protection and economic affairs. The last category was formed by other expenses. Their results showed that all categories of expenditure have negative effects on economic growth in Romania. For example a 1 pp increase in government expenditure determines the real GDP growth rate to fall by 0.45 pp for the category of productive expenditure, with 1.57 pp for unproductive expenditure and 1.92 pp for other expenses (Boldeanu and Tache 2015).
Yu et al. (2009) studied the significance of expenditure on economic growth for forty four developing countries on 3 separate continents (Africa, Asia and Latin America) between 1980 and 2004 (totalling 80% of the GDP for the states in that category). To analyse the significance of public expenditure on economic growth, the authors employed the GMM method. The categories of public spending investigated by Yu et al. (2009) are: defence, health care, education, agriculture, telecommunication and social protection.
This investigation found that there is an important link between government spending and the growth rate of GDP, but each category of expenditure affects differently the dependent variable depending on the region. Thus, in Africa, human capital expenditure had a positive effect on economic development. In Asia capital expenditures, agriculture and education have a positive influence, while in Latin America it was found that no category of expenditure promotes economic growth.
This study demonstrates the importance of downsizing public expenditure structure by increasing the most effective of them and reducing the less productive, such as those for the defence. Also, developing countries should undertake research and development spending in areas such as education, agriculture and health, to raise the standard of living of citizens.
Mario Holzner (2011) examined the effects of the relationship between income inequality and public spending on economic growth in developing countries of Central and Eastern Europe between 1989 and 2006.
The first model reflects the influence of two types of explanatory variables (public spending and variables that summarize the changes in the transition of a country from a socialist to a market economy).
The results suggest that there is a negative correlation between government spending and inequality. Large-scale privatization, trade liberalization and implementing foreign exchange systems have had a positive effect, leading to a decrease in inequality.
The second model explains the evolution of the rate of growth of GDP by analysing public expenditure broken down into nine categories according to the functional classification. Such expenses pertaining to economic affairs, housing and community amenities and education are negatively correlated with GDP growth.
Payne et al. (2006), Lamartina and Zaghini (2008), Arpaia and Turini (2008), Szarowská (2012), Menyah and Wolde-Rufael (2013) and other several studies investigated the correlation between government expenditure and economic growth utilizing the concepts of the Wagner’s law (1911). The outcomes of the investigation made by Lamartina and Zaghini (2008) proved Wagner's theory, due to the fact that the coefficient of public expenditure elasticity in relation to GDP took a value above par (at a 1% increase in GDP, general government expenditure increased by 1,028%). The study also affirmed that the values of the expected long-term elasticity coefficient are higher in countries with lower GDP per capita, suggesting an attempt to realize economic development funded by the state (Boldeanu and Constantinescu 2015). In 2004 Wahab (2004) also confirm the Wagner Law for a group of OECD countries.
Arpaia and Turrini (2008) have investigated the influence of government spending on per capita gross domestic product growth. Their study was conducted for both long-term and short-term analysis, for the EU 15 countries between 1970 and 2003. The cointegration tests that were employed by the authors demonstrated the correlation between long-term economic growth and government expenditure. Also the elasticity coefficient between the two variables are below par, but there is no long-term stability (the coefficient of elasticity decreases considerably).
In less developed countries, less indebted, where the degree of population aging is high, public spending grew faster than GDP per capita. The analysis also showed that it takes approximately three years for the change of government expenditure to cancel their long-term deviation compared with potential GDP. Furthermore the Nordic states and also the Anglo Saxon ones achieve long-run economic growth equilibrium in a shorter time, unlike the countries of southern Europe, where the period of adjustment is greater. An exception to this result is Greece that has a smaller period and Germany that has a longer time period.
Szarowská (2012) has analysed the direct link between government expenditure (% of GDP) and gross domestic product in short and long-term for 5 EU states, namely Bulgaria, Czech Republic, Hungary, Romania and Slovakia. She also studies to see if public spending is countercyclical. Szarowská (2012) used the 10 government expenditures detailed in the COFOG classification of the United Nation. The results of the paper invalidated the countercyclical effect of the two variables.
Several recent papers that investigated OECD and developing countries or Latin American nations showed that, contrary to the theory, government expenditure is procyclical (Gavin 1997; Lane 2003; Hercowitzand Strawczynski2004; Alesina et al. 2008; Ganelli 2010; Abbott and Jones 2011). Szarowská (2012) pointed out that there is a long-run statistical correlation between public spending and economic growth, but for the short-run effect the statistical significance is too low.
Menyah and Wolde-Rufael (2013) conducted an investigation to find the long-run and casual link between government spending and national income for Ethiopia between 1950 and 2007. Using four different estimators, they obtain a link starting from national income to government expenditure. Public expenditure increases (1.73 to 1.79) more than national income and the Keynesian hypothesis in which government expenditure is an instrument for promoting economic growth doesn’t hold for Ethiopia.
Chang et al. (2004) determined a unidirectional Granger causality that starts from income to public expenditure in South Korea, Taiwan, Japan, U.S.A and United Kingdom. In Australia, New Zealand, South Africa, Thailand and Canada there was no statistical relationship between public spending and economic growth.
Liliana Cioban (2014) carried on an investigation regarding the education and health care sectors among other variables. Between 2000 and 2007 health and education expenses were higher in the developed countries (Germany, Norway, USA, Canada etc.). In Romania and other Eastern countries, these sectors are under financed. Also the author noted that R&D is a sector with big financial incentives from the state in the developed countries and poorly financed in the developing ones. The author found a significant link between economic growth and health, R&D and education (Cioban 2014).
A significant, but a negative link between health and education government expenditure and gross capital formation and economic growth was confirmed by the study of Dao for 28 developing economies (Dao 2012). Some of his variables do not have the expected sign because there was possible collinearity between some of them.
The empirical literature also emphasized the importance of education expenditure on economic growth for different periods and samples of countries (developing, in transition and developed). I consider that a great contribution to this subject was made by researchers like Barro (1991), Levine and Renelt (1992), Sala-i-Martin et al. (2004), among others. Education is also a key measurement tool and proxy for the quality of human capital in the sense that educated and skilled workers can have an important contribution to production and economic growth (Boldeanu and Constantinescu 2015).
Robert Barro’s study of the importance of primary and secondary enrolment for 98 developed and developing countries, analysed between 1965 and 1985, he found a significant influence of these variables on economic growth (GDP/capita). Also education reduces fertility rates and increases private investment productivity. Education has an impact on these two phenomena (Barro 1991). Mankiw et al. (1992) analysed the impact of human capital investment (secondary school net enrolment rate * fraction of population of secondary school age in 1960-85 average) and his study also yielding a positive effect on economic growth. The two papers are similar in results and suggested that increasing primary and secondary enrolment rates can increase economic growth, but the analysis was limited because of the small number of regressors (can suffer from omitted variable bias).
Levine and Renelt (1992) used 40 variables and many regressions to check for robustness of the effect of his estimators on economic growth. Compared to the above studies, he demonstrated that primary and secondary enrolment rates have no conclusive effect on growth. Barro and Lee (1994) and Barro and Sala-I-Martin (2004) analysed also the role of male and female enrolment rates on economic growth. The research papers concluded that male secondary education has a positive and significant influence on economic growth with female secondary education being negative or being insignificant.
Bils and Klenow (2000) showed that human capital used in the production function demonstrates that it has a significant and positive influence on economic growth. In their paper they already assumed at the starting point that the influence between schooling and education exists and wanted to find the correlation between the two.
A problem with the early research was the measurement error and the omitted variable bias, which after the 2000s was addressed and tried to be solved. Hanushek and Kimko (2000), Sala-I-Martin et al. (2004), Hanushek and Woessmann (2008) indicated that reducing measurement errors yields better results for the impact of education on growth. By calculating the school quality (measured by internationally comparable tests) impact on economic growth had also an influence on the variable bias and measurement error. Glewwe et al. (2014) reiterated the importance of variable bias and that education quality plays an important role in Sub-Saharan African countries.
Benoit (1973, 1978), Pieroni (2009), Dunne and Tian (2013), Ho and Chen (2014) among others, investigated the influence of military spending on growth. Most of the research papers concluded that in general defence expenditure has a negative effect on economic growth. Benoit (1798) was the pioneer in this line of research and found that in less developed countries military expenditure had a positive effect on economic growth. The assumption that this component of government spending can have a positive effect depends on the samples, the different theoretical specifications and periods and on the methodology used. Yildirim et al. (2005), McDonald and Eger (2010), indicated that defence expenditure had a small or rather insignificant effect on economic growth. On the other hand Galvin (2003), Pieroni (2009), Ho and Chen (2014) concluded that military spending has a negative influence on growth.
Ho and Chen (2014) stated that:
“Military expenditure could influence the economy through a number of channels, broadly grouped into demand, supply and security effects. In the demand side, an increase in military expenditure could have the Keynesian multiplier effect on the economy by increasing total demand and resource utilization and reducing unemployment, if there is spare capacity. It can have opportunity costs and crowd out human and capital investment. In the supply side, factors of production used by the military are not available for civilian use and may cause negative effects on growth. On the other hand, the impact could be positive through externalities (such as military training and technology spin-offs). In the security side, military expenditure promotes growth by increasing security which is essential to enhance the incentives to accumulate capital and innovation. “
1.4.2. FDI and trade determinants
There are an important number of research papers that analysed and explained the link between FDI and trade components (exports, imports, net export, trade openness, trade restrictions, etc.) and economic growth with some different views depending on the period, sample and regions. A big part of the literature has shown that states that have economies open to trade have higher per capita GDP and grow much faster (Romer 1990; Grossman and Helpman 1991; Edwards 1992; Sachs and Warner 1995; Dollar and Kraay 2000, Barro 2003).
Tekin (2012) found that a rise in exports has a positive effect on economic growth. He concluded that this outcome is explained due to the multiplication effect of foreign trade. Using the Granger causality test, the Johansen Cointegration Methodology, Sultan and Haque (2011), Simuț and Meșter (2014) determined a long-term and direct influence between some trade determinants of economic growth. Simuț and Meșter (2014) identified a direct correlation and causality between exports, openness and economic growth for 10 East European countries using quarterly data from 2000 and 2013 and Sultan and Haque (2011) found that there is a long-run relationship between exports and growth for India.
In 2012 Simuț (2012) showed that exports and openness have a positive statistical impact on economic growth for the 10 Eastern European countries, those that were integrated into the economic community between 2004 and 2007. Her findings were based on a time sample between 2000 and 2010.
The impact of trade components on economic growth in the Middle East has been debated by many researchers. AL- Raimony (2011) empirically analysed the relationship between real export and real import growth, among other variables, and economic growth in Jordan. He concludes in his paper that real export growth positively affects growth, while real import growth negatively affects economic growth and that Jordan has to be more involved in stimulating exports. In 2014 Abu-Eideh also analysed real domestic exports and imports of goods and services and the way these variables affect real gross domestic product in Palestine for the period 1994 to 2013 (Abu-Eideh 2014). In his research, he summarized that real domestic exports have a positive impact on growth in Palestine while real domestic imports a negative one.
Abu-Eideh (2013) determined in another research of Palestine for the period 1994-2011 that exports have a positive impact on economic growth measured by GDP. He concluded that for Palestine there is a need of export policies so that the country could maximize the benefits of exports and for the goods and services to be more competitive on the foreign market.
Openness can have an impact on growth through a multitude of different channels like for example, technological transfers, the increase in economies of scale, competitive advantage (Chang et al. 2009). Edward (1992) analysed for 30 developing countries the impact of openness on economic growth (GDP). Using a linear regression with a period between 1970 and1982 he showed that trade openness has a favourable effect on real GDP and that trade liberalization will accelerate economic growth and countries will be capable to enter more easily foreign markets.
Ynikkaya (2003) also analysed the impact of trade openness on growth for 120 countries between 1970 and 1997. He used several variables to measure trade openness like the volume of exports, the volume of imports, the sum of exports and imports and the volume of trade with developed countries in total GDP. He also used trade policy variables for measuring restrictions or openness of trade. The result of his study showed that for developed and developing countries the indicators that measure the volume of trade are positively impacting GDP per capita. An interesting outcome is that trade restrictions have the effect of accelerating growth of GDP in developing countries.
Using a panel of 32 European countries (15 old members, 12 new members and Croatia, Bosnia and Herzegovina, the Former Yugoslav Republic of Macedonia, Serbia and Albania), Malešević-Perović et al. (2014) investigated the relationship between trade openness and financial openness on economic growth. The results confirm that trade openness and financial openness (FDI) have a significant impact on growth and also that institutional openness is affecting indirectly the economy via trade and FDI. For the new EU countries (in their panel EU 12) that are less developed compared with the old EU members, higher productivity and export specialisation have a big impact on economic growth.
Mihuț and Luțaș (2014) also found that for Central and Eastern European countries, respectively, for the 12 new EU member states, the degree of openness and human capital are positively correlated with economic growth. The data set had a time horizon between 1992 and 2011.
In the research literature, there are also views that contradict the results of the above mentioned authors regarding openness (Levine and Renelt 1992; Rodriguez and Rodrik 1999). Singh (2011) concluded that for Australia he obtained a negative impact of imports on economic growth and Ahmad and Kwan (1991) found that for 47 African states there is no link between trade and growth.
Li and Liu (2005) investigated the role of FDI on economic growth for a large sample of countries, both developing and developed. Their study showed that FDI directly and positively influences growth. The findings of other researchers at the beginning of the 2000s demonstrated that FDI may have a positive link between it and economic growth (Hermes and Lensink 2000; Lensink and Morrissey 2006).
In the case of some European Union countries, like Romania, Poland, Hungary, Czech Republic, Latvia and Lithuania, some research papers showed that FDI had a positive impact on growth (Marinaș 2007; Bhandari et al. 2007). Also investment and exports are the main two transmission channels of this relationship. Using a VAR Multivariate Autoregressive model, Pereira and Xu (2000) found that increasing exports has a positive outcome on GDP growth. This result was confirmed also by another study of Marinaș (2008).
FDI inflows have a positive impact on the economy and can accelerate the rhythm of economic growth especially in developing countries. This fact was demonstrated by Johnson (2006) for a panel of 90 developing countries during the period 1980-2002. He concluded that in developed states the FDI inflows do not have the same accelerating impact like in developing ones. By enabling positive externalities like the diffusion of know-how and new technologies, FDI can have a direct impact in the sectors in which these funds were allocated, but also an indirect impact on the whole productivity in the economy (de Vita and Kyaw 2009).
Bagli and Adhikary (2014) investigated the impact of the growth of FDI and the growth of openness on economic growth in India during the periods of moderate liberalization (1980-1990) and strong liberalization (1991-2010). Their assessment of these factors has shown that there is no statistical effect between FDI growth and openness growth and economic growth. After the 1991 much of the foreign direct investment in India is a derivative investment in the service sector. In this sense, Bagli and Adhikary (2014) make the following remark:
“A part of FDI in India introduces inappropriate technology and retards the development of domestic capital goods industries. A part of profit of FDI outflows from the host country. In this sense FDI is more dangerous than external borrowings. While borrowings create repayment compulsion for a certain period of time, FDI may generate an unending commitment. If every year a constant amount of FDI inflows in a country and a constant rate of profit on it, outflows, it definitely generates a negative net inflow for this country after a certain lag. It may be the probable explanation of the very recent falling tendency of growth in India.“
1.4.3. Public and private investment
Public and private investments are fundamental components of economic growth theory used by neoclassical and endogenous growth models. There are many empirical researchers that investigated the correlation between investment and economic growth with contradictory outcomes (Khan and Reinhart 1990; Levine and Renelt 1992; Barro and Sala-I-Martin 1995; Nazmi and Ramirez 1997; Bond et al, 2001).
Public investment in recent studies is viewed as expenditure made by the state for goods or services that the private sector is not involved in producing at suitable amounts because of the cost being too high at the start up level or the benefits (profits) are not high enough compared with the production costs. But this public investment is necessary for both the well-being of the citizens (communication, social expenses, transportation, energy facilities, water supply, ports, etc.) and also for the private investment by raising total factor productivity and labour productivity.
A problem with public investments is the possibility of creating and subsidizing state-own companies that can become market inefficient and can have an impact on future economic growth. These can create a crowding-out effect of private investment in some sectors that can otherwise be profitable (Pereira 2001). Public investment can use scarce natural resources inefficiently, that can otherwise be used by private investment.
Most policymakers and research economists consider that private investment is more efficient than public investment and contributes more to economic growth (Coutinho and Gallo 1991, Mankiw et al. 1992).
For example Khan and Reinhart (1990) stated that research studies dealing with the effects of economic growth do not differentiate between private and public investments. Through their research study, they separated the effects of private and public sector investment, analysing 24 developing countries (Asia and Latin America) during 1970-1979. The results determined that private investment has a bigger influence than public investment. Khan and Kumar’s (1997) results for 95 developing countries (which were situated in 4 different regions, namely in Europe and Middle East, Africa, Asia and finally Latin America) during 1970–90 showed that there is a substantial difference between private and public investment, with the first one being more important.
Zou (2006) analysed the correlation between public and private investment and economic growth for Japan and the USA using GMM and OLS empirical methods and a time period from 1958 to 1997. The results show that there are differences between the two countries. Private investment being more important in the USA compared with public investment. For Japan, both public and private investment greatly contributes to the increase in GDP growth.
Nazmi and Ramirez (1997) investigated the contribution of public and private investment in determining economic development for Mexico between 1950-1990. Their analysis concluded that private and public investment had the same statistical positive effect on growth. Also the direct contribution of public investment in fostering growth has an impact on private investment by depressing it. Odedokun (1997) investigating a sample of forty eight developing states during 1970-90 affirmed that there isn’t any material difference between private and public investment. Also public investment partly crowds-out private investment in the long-run.
Ghali (1998) investigated the dynamics between private and public investment and economic growth. He applied this research for Tunisia and determined that in the long-run public investment has a negative contribution on growth and on private investment. Also, in the short-run public investment negatively affects private investment, but doesn’t have an effect on growth.
Analysing the ASEAN 5 countries (Singapore, Thailand, Indonesia, Malaysia and Philippines) Prasetyia (2013) concluded that private investment is more important for growth in the long-run, but public investment contributes more to economic growth in the short-run.
1.4.4. Energy consumption
As world energy consumption and demand is still growing on a fast rate and prices for petroleum tend to fluctuate in the future, the influence of the energy sector on economic growth is very important. Researchers in this area are still debating if there is a causality that runs from energy consumption (coal, petroleum, electricity) to economic growth or if the link is the other way around. Others consider that there is a bidirectional link between energy consumption and growth.
The research papers that favour the unidirectional correlation from energy consumption to economic growth suggest that energy has a significant role in shaping growth and that the state has to use energy policies wisely as not to harm the economy. Yu and Choi (1985), Masih and Masih (1996), Yang (2000), Soytas and Sari (2003), Lee (2005), Narayan and Prasad (2008), Apergis and Payne (2011), Bhattacharya and Bhattacharya (2014), Mahalik and Mallick (2014) showed that for developing countries (India, China, Pakistan, Turkey, Brazil, Indonesia, etc.) and also for developed countries (France, Australia, Italy, Korea, Japan, etc.) energy consumption influences economic growth.
Research papers that highlight the unidirectional link from economic growth to energy consumption suggest that the economic output and its growth influence how much energy is consumed. Pioneers in this branch of economic growth theory were Kraft and Kraft (1978) who analysed the impact of energy consumption on GNP in the USA during 1947-1974. They found a unidirectional link that runs from economic growth to energy consumption. Abosedra and Baghestani (1989), Masih and Masih (1996), Soytas and Sari (2003), Ghosh (2002), Yoo (2006), Narayan and Prasad (2008), Zhang and Xu (2012), Bhattacharya and Bhattacharya (2014) also found a unidirectional link from economic growth to energy consumption for countries like Pakistan, USA, Korea, India, Indonesia, Thailand, etc.
Finally, there are research studies that support the bidirectional relation between energy consumption and growth (Masih and Masih 1997; Paul and Bhattacharya 2004; Odhiambo 2009; Solarin and Shahbaz 2013; Bhattacharya and Bhattacharya 2014).
Like we can see from the literature, there are authors that in the same study obtained different outcomes for their country sample. This demonstrates that in certain states energy consumption influences economic growth, whereas in others economic growth influences energy consumption.
Mallick (2009) used the Granger causality test and the variance decomposition analysis to investigate whether energy consumption influences economic growth in India during 1970-71 to 2004-05. The Granger causality showed that there is a unidirectional relation from economic growth to crude oil and electricity consumption and that only coal consumption drives economic growth. The variance decomposition analysis demonstrated that there is a bidirectional link between electricity consumption and economic growth, with the other result remaining unchanged (Boldeanu and Tache 2015).
Bhattacharya and Bhattacharya (2014) tested the correlation between energy consumption and economic growth for India and China between 1980 and 2010. They disaggregated energy consumption using electricity consumption, coal consumption and petroleum. The results show that in the case of India there is a bidirectional relation between coal consumption and economic growth, a unidirectional causality from petroleum consumption to growth and finally there is no relation with energy consumption. For China, the authors concluded that there is a unidirectional causality from economic growth to coal consumption, a unidirectional causality from petroleum consumption to growth and like in the case of India, there is no causality between electricity consumption and economic growth.
Odhiambo (2014) studied the Granger-causality between electricity consumption and economic growth in the Democratic Republic of Congo during the period 1980-2011. Using the ARDL-bounds test procedure he demonstrated that in the short run there is a bidirectional causality between electricity consumption and economic growth and in the long run economic growth influences electricity consumption.
Bhattacharya and Bhattacharya (2014) affirmed that the study of energy consumption can have significant implications for economic growth, but also global warming has to be taken into account. They stated that:
“Energy is needed for economic growth, for improving the quality of life and for increasing opportunities for development along with serious environmental consequences. There is a growing concern that the present pattern of energy production and consumption are unsustainable in the long run.”
1.4.5. “Ultimate” or non-economic determinants
I have already stated in the beginning of the subchapter that “ultimate” or non-economic sources refer to factors like government efficiency, institutions, political and administrative systems, cultural and social factors, geography and demography (Rodrik 2003; Acemoglu et al 2005; Acemoglu 2009; Arvanitidis 2007).
In his economic growth analysis, Arusha (2009) investigated the effects that the size of government expenditure has on economic growth. Also, he used a qualitative factor, namely the political governance indicator.
The impact of governance and public expenditure on economic growth was investigated for 71 states with a time frame between 1996 and 2003. There were three types of countries, namely developed, developing and in transition, which were also classified related to the income distribution. The model explained the evolution of GDP / capita using the main categories of expenditures (health, education and defence), investment, technology development level or the impact that the financial sector has on the economy. The author also used several dummy variables to measure the significance of the governance indicator.
The author believed that the government has a considerable role in the allocation and distribution of resources within a country. The state is very important for social organization, the enactment of laws and political stability.
Using the governance dummy variables, Arusha (2009) demonstrated that compared with states with “very weak” governance the countries with "very good" governance have an increase of GDP/capita of approximately 8.33%, those with "good" governance an increase of 8.11% and those where the governance was "weak" grew with 5.65%.
To see if poorer countries have an economic growth rate higher than the richest ones for his time sample, Arusha (2009) divided the panel into three categories respectively states with low, medium and high per capita income. The author confirms that for the period under review poorer countries have a higher economic growth than those with a high average income. Developed states are in a process of economic stagnation. The research paper points out the need for good governance and along with increased public consumption will result in sustainable economic growth.
A significant factor in shaping growth in the research literature is the state institutional framework. At the middle of the 1950s and beginning of the 1960s the role of institutions was starting to be acknowledged with the seminal work of Lewis (1955) and afterwards by Ayres (1962) and after the beginning of the 1990s, it really started to be more and more researched by many works in the field of economic growth (Mauro 1995; Rodrik 1999; Acemoglu et al. 2002).
North (1990) argues that institutions are used to structure the uncertain interactions between humans. They are used by the state to constrain certain relations between people. In his opinion institutions can be created to be a benefit to economic growth or can lead to stagnation. The role of institutions is to create a framework in the society in which people and organizations can take advantage of certain opportunities. The skills and knowledge fostered by the institutional framework and the benefits created by them leads to developments in the economy and in time create new institutions and incentives for the society.
Voigt (2009) defined institutions as:
“commonly known rules used to structure recurrent interaction situations that are endowed with a sanctioning mechanism”
Institutions can have a direct effect on economic growth, but can also affect other determinants like human and physical capital, technological change or investment. Rodrik (2000) argued that five kinds of institutional frameworks (property rights, regulatory institutions, institutions for macroeconomic stabilization, institutions for social insurance and institutions of conflict management) can have a direct outcome on growth and on other determinants of economic growth (Boldeanu and Constantinescu 2015).
The key measurements of quantifying the role of institutions on economic growth are corruption, rule of law, bureaucracy, property rights and the repudiation of contracts by the government (Knack and Keefer 1995).
Corruption is a non-economic variable that can have a significant influence on economic growth in many countries. This is an important problem for the society and the economy. Many developing states have a significant problem with corruption because of the level of poverty. It can have an adverse effect on investment. It can have a negative impact on public productivity by making government spending inefficient. Many researchers are now of the opinion that we have to mitigate and control the level of corruption in the society.
Murphy et al. (1993), Mauro (1995) and other researchers have shown that corruption tends to have a negative effect one growth by affecting innovation and other start up activities and as a consequence may reduce productivity. In the case of innovation, corruption can limit the new entrepreneurs to enter the market with their start-up products and services. The enterprises that have to pay a big amount of money for bribes tend to reduce the production and also distort their figures (De Soto 1989; Svensson 2003).
Mo (2001) and Mauro (1995) empirically tested the direct impact of corruption on growth and concluded that decreasing corruption may lead to an increase of 0.8 % of annual per capita GDP growth rate or an increase of only one unit of the corruption index may lead to a fall of 0.5455 p.p. in GDP growth rate.
Shera et al. (2014) analysed the consequence of corruption on economic growth for twenty two developing countries for the period 2001-2012 in conjunction with other economic variables like trade, secondary school enrolment, inflation, public spending and capital formation. The panel sample was represented by former socialist states and satellites of the USSR in the Balkans, East and Central Europe and Asia and includes also Mongolia. The results of their research showed that corruption had an important and a negative outcome for the 22 developing countries.
In contrast to studies in which corruption is viewed as an inhibitor of economic growth, there are researchers that consider that corruption can be beneficial because it can make the economy more efficient and facilitate for investors a way to pass more restrictive and established rules. By offering bribes, business people can facilitate the influenced they need to make the bureaucratic process go much faster and to circumvent certain obstacles (Leff 1964; Huntington 1968; Bardhan 1997; Acemoglu and Verdier 2000). Lui (1985) demonstrated in his model that bypassing red tape, corruption can facilitate market performance. He constructed a queue concept where the decision of paying bribes is decentralized to customers.
Rock and Bonnett (2004) also implied that for the major industrialized countries in East Asia (Japan, Korea, Indonesia, China and Thailand) corruption has a beneficiary effect on economic growth. Braguinsky (1996), Kaufman and Wei (2000) stated the fact that in certain circumstances corruption can have a lubrication effect on growth.
Graziano (1980) and Huntington (2002) suggested that bribes, especially among government official can be considered the important glue that keeps countries together. Corruption may tend to keep violence and conflicts at a low and in so doing can foster economic growth.
Political variables like political regimes (democracy, totalitarianism), political stability/instability, civil freedom, the perception of politics play also an important role in fostering economic growth or can inhibit development. They can also impact the relationship between the state and its citizens (Lipset 1959; Scully 1988; Lensink et al. 1999). Political instability has a negative consequence on companies and their willingness to invest, can create violence and anarchy in the society and in the end can have serious consequences on economic growth (Boldeanu and Constantinescu 2015). Alesina and Rodrick (1994) studied the relationship between politics and economic growth. Policies that maximize economic growth are optimal only for “capitalistic” states. If inequality is high, the higher are the rates of taxations and the economic growth is smaller.
Using a panel of 24 states in the Middle East and North African from 2001 to 2009, Tang and Abosedra (2014) examined the importance of tourism, energy consumption and political instability on economic growth. Their results show that energy consumption and tourism have a significant contribution to growth for their panel sample and demonstrates that political instability hinders the process of economic growth. They conclude that the analysed regions need to overcome the political instability so that they can attract more tourists and foster economic growth.
Aisen and Veiga (2013) determined a significant negative effect of higher degrees of political instability related to the growth rate of GDP for 169 states in the period 1960-2004. The channels of transmission through which political instability affects economic growth are productivity, physical and human capital accumulation. Also democracy may have a small negative effect on economic growth. The empirical literature is still ambiguous in relating the significance of democracy in fostering economic growth. It may have a negative or not a significant influence on growth (Boldeanu and Constantinescu 2015). Raufhon (2015) conducted an investigation to determine the effects of democracy and intelligence on economic growth for ninety three countries between 1970 and 2013. He concluded that the link between democracy and real gross domestic product growth varies with a country's level of cognitive abilities.
Socio-cultural variables also play a significant role in the economy. Ethnic diversity and fragmentation, language, religion, superstitions, social conflicts, civic norms, beliefs are among the sociocultural determinants that may have an effect on economic growth (Putnam et al. 1993; Helliwell and Puttman 1995; Knack and Keefer 1997; Inglehart and Baker 2000; Acemoglu 2009). The link between them and growth is not direct. Ethnic and cultural diversity has an indirect impact on economic growth by stimulating uncertainty in the society and propagating ethnic conflicts and in turn affecting growth. Or ethnic diversity may have a positive impact on growth by creating an environment in which new concepts and ideas from other cultures can be propagated and the receiving society can benefit.
Knack and Keefer (1997) were of the opinion that trust and civic norms are important incentives for innovation, the accumulation of physical capital and to influence human resources and in turn to foster economic growth. Also in their research they showed that trust and civil norms have a bigger influence in countries with high income and more equality, with better education and ethnically homogeneous population.
Ethnic diversity may have a negative impact on economic growth by reducing trust, having a negative effect on education (low schooling), political instability, underdeveloped financial systems, high public deficit and underdeveloped infrastructure. These negative effects of ethnic diversity were analysed by Easterly and Levine (1997) for Sub-Saharan Africa for 30 years starting from 1960. Granato et al. (1996) affirmed that both cultural and economic determinants matter for economic development. For his panel of 25 countries, motivation and post materialism are significant for growth, with the first being positive and the second having a negative effect on growth.
Acemoglu (2009) affirmed that the difference in religion is the clearest sign that cultural differences may impact economic behaviour. In his opinion cultural differences are propagated by two important channels, specifically the willingness of people to trade-off certain activities or consumption patterns and the degree of cooperation and trust. The first channel impacts economic growth by influencing market structure, the accumulation of human and physical capital, saving rates, occupational choices. The second channel has an impact on productivity in the sense that trust and cooperation can have a serious effect on the production capacity of a state. Acemoglu (2009) stated that:
“Culture is much harder to influence and any advice to a society that it should change its culture is almost vacuous”.
The importance of geography on economic growth has been well established. After World War II there was a surge in the empirical analysis of geography. Braudel (1981-1984), Mcneill (1963), Jones (1981) and Crosby (1986), Diamond (1997) analysed the impact of geography and climate change in Europe and its dominance over the colonies. North-Atlantic and Mediterranean Europe were the creative centres of the world after the middle Ages ended. Crosby (1986) showed the importance of temperate zones for climate and productivity in agriculture.
The most used proxies in the research literature for geography are latitude and longitude in absolute values, the relative distance from the equator, distance from the coast, the temperature and the level of rain, the nature and quality of the soil, natural resources and diseases (Gallup et al. 1999; Rodrik et al. 2002; Easterly and Levine 2003).
Acemoglu (2009) stated that geography can affect in different ways the economy. Soil quality can have an influence on agricultural productivity. Natural resources directly contribute to the industrialization of a country by providing essential components for production. Climate has a direct impact on productivity and the attitudes regarding consumption. The topography of a region or state can have a positive or negative impact on transportation costs and on communication. And not least, diseases can affect health care, production and the accumulation of human and physical capital.
There are also contradicting views regarding the effects of geography on economic growth. For example, Sachs is an avid supporter of the idea that geography is beneficiary for the economy. Gallup and Sachs (2001) and Bloom and Sachs (1998) stated that malaria is credited with reducing the annual growth rate for sub-Saharan African countries by 2.6 percent per year. If malaria had been eradicated since 1950 the income per capita in that part of the world would have been double today.
Climate, natural resources or topography have also a significant influence on the economy. These variables are responsible in increasing agricultural production, lowering transportation costs and increasing competitiveness (Bloom and Sachs 1998; Armstrong and Read 2004).
Other authors like Rodrik, Easterly and Levine showed in their research that geography doesn’t have any conclusive and significant effect on economic growth (Rodrik et al. 2002; Easterly and Levine 2003).
Demographic trends play also an important role in shaping economic growth. Among the variables that have a significant role in the dynamics of growth I can mention population density, population growth, migration trends and age distribution (Becker and Barro 1988; Kormendi and Meguire 1985; Kelley and Schmidt 1995; Barro 1996).
Population distribution has a significant role in shaping growth. A high number of working-age citizens can have a positive impact on the economy compared with a large number of young and old age citizens that need social protection and other public expenditures.
A positive outcome on economic growth may be obtained with a certain population density, by increased specialization and knowledge. Also migration plays an important role in the receiving and also in the sending countries. For the receiving state, there can be a benefit if the migrating subjects have acquired knowledge and skills that can boost production. Also the expenditures for education and training were supported by the sending country without any compensation and added value of these costs if people go and migrate to another state.
The rise of population size could have a significant negative influence on economic growth by impacting the investment and savings behaviours of citizens, the dependency ratio and the quality of human capital. Shera et al. (2014) concluded in their research of former communistic states that the rise in population size with a unit will decrease GDP by 0.044 %. Tolo (2011) demonstrated that in the case of Philippine, high population growth can have negative implications.
Van Der Gaag and Beer (2015) stated that in the EU the decline in the working-age population percentage in total population will decline even more for all countries and that this will have a negative consequence on economic growth. Furthermore, they investigated if there is a considerable difference between rural and urban regions when it comes to the outcomes on growth. Their results concluded that raising employment will solve only partially the problem of economic growth and moreover there is no difference between population growth outcomes on growth in rural or urban regions.
Barro (1996) affirmed these about fertility and population growth:
“If the population is growing, then a portion of the economy’s investment is used to provide capital for new workers, rather than to raise capital per worker. A higher rate of population growth has a negative effect growth. Another, reinforcing effect is that a higher fertility rate means that increased resources must be devoted to childrearing, rather than to production of goods.”
Summarizing the main ideas of the impact of demography on economic growth, I can state that in general demography has a negative influence on growth. I have to mention also that some authors have a contradictory view of my above affirmation. For example Grier and Tullock (1989), Pritchett (2001) found that there is no association between economic growth and demography. Bloom and Williamson (1998) for East Asia’s economies, Braun et al. (2009) for Japan, Kelley and Smith (2005) for the world and Sanchez-Romero (2013) for Taiwan, demonstrated that demography can play a significant role in fostering economic growth. For example, Sanchez-Romero using the computable general equilibrium found that demography accounts for 22 % of Taiwan’s per capita output growth during 1965–2005. In his opinion fertility is the most important demographic factor.
1.5. Conclusion
The study has shown in the above sections that economic growth theory is a very complex process that involved many researchers and decades and centuries to refine. From the beginning of the classical theory of Adam Smith in the 18th century to the present days of the new growth theory, models have evolved constantly to take into account the changes in the economy.
The chapter has highlighted the main determinants of economic growth raging from public expenditure, FDI, openness, export, imports, private and public investment or “ultimate” (non-economic) causes. There are many more determinants that are being refined and disaggregated to be used into new and advanced models of economic growth. Also, as new mathematical and statistical models and tests are being produced, the old assumptions have to be retested and if differences occur the theory has to be modified.
For a well-balanced economic growth model we have to take into account the proximate causes (economic determinants) and also the ultimate causes or the fundamental causes (Acemoglu 2009). Also, as new statistical data are being published, the research has to focus not only on country analysis, but on regional analysis as well, like for example in the European Union the Nomenclature of Territorial Units for Statistics.
Moral-Benito (2007) affirmed that: “in the search for a satisfactory statistical model of growth, the main area of effort has been the selection of appropriate variables to include in linear growth regressions. The cross-country regression literature concerned with this task is enormous: a huge number of papers have claimed to have found one or more variables correlated with the growth rate, resulting in a total of more than 140 variables proposed as growth determinants.”
CHAPTER 2 ECONOMIC GROWTH MODELS
In the above chapter the study has highlighted the main determinants of economic growth. In the following chapter of the thesis the investigation will focus on the main economic growth models and how they were used for determining the main influences on growth. Economic growth models represent an important example of the dynamics of the economies of different countries during certain periods of time.
Economic growth models are an important part of the growth theory. The evolution in time of these models was used for capturing the main characteristics of the most important macroeconomic indicators that influence the development of the contemporary economy.
Economic growth models have been a key interest for researchers since the classical period. Like it was mentioned earlier, economists like Adam Smith (1776) or David Ricardo (1817) tackled with the problems of determining the appropriate factors that influence growth. Keynesian models and the ones that followed (the Neo-Keynesians) argued that to have a stable economy requires the use of macroeconomic policies and direct state intervention in reaching equilibrium and stimulating economic growth. At the other extreme we have the neoclassical models who claim that the economy is stable and that it will return to a steady-state if different shocks will occur.
All the determinants that were highlighted in the first chapter have to be included in economic growth models and we must emphasize this because validating the variables included in the models is very important. With a thorough empirical analysis we can say for sure that our determinants have a significant importance.
There are many economic growth models that have been used to determine certain results depending on the specifications of the researchers. In the following part of the chapter the study will highlight the most important models that have been used and implemented throughout the decades by different researchers.
2.1. Harrod model
The Harrod (1939) model has at its centre the correlation between economic growth rates and the accumulation rate, which depends on the investment rate. The basic elements of the model are three equations, by which you can calculate three possible rates of economic growth: the actual rate of economic growth, the warranted growth rate and natural growth rate.
Harrod (1939) model starts from the following assumptions:
The state of technology is presumed. It requires all the inputs to be employed in fix proportions. The production function with its fixed coefficient takes the following form:
Y(t) = , with α>0, β>0 (1)
The coefficients α, β represent the labour output ratio and the capital output ratio. The two ratios are required for the production of one unit of output.
The economy is a closed system that produces only one type of goods that is partly invested and consumed.
The labour force grows determined by an exogenous constant symbolised with λ. This constant is an exponential rate.
(2)
If t=0 then for L(0):
(3)
From equation (3) we can obtain the labour size at time t.
The investments are illustrated by the acceleration principle.
, with ν >0 (4)
“ν” is the accelerator coefficient. The model considers that there is no technological progress and that there is no depreciation of capital stock.
The inclination to save is at a constant proportion of the output at any time.
, where 0<s<1 (5)
“s” represents the propensity to save of the society at a given time measured by “t”.
The entrepreneurs are interested in their profit maximization.
The Harrod (1939) model represents one of the first post-Keynesian models. From the beginning I must emphasize that the model uses three fundamental concepts, namely:
a) the actual rate of economic growth “G”, namely that which is obtained in a period of time, usually one year;
b) the justified growth rate “GW”, which Harrod calls the warranted growth rate;
c) the natural rate of growth “Gn”.
To highlight the first concept, Harrod (1939) denotes with “G” the growth rate of total product, namely of the national income (Y), expressed as a ratio between the absolute growth ΔY and total product of the previous period Y, such as:
G =
Next, Harrod (1939) represents the coefficient “C” of private capital in marginal shape, which is explained by the ratio between the investment rate and the increase in income I / ΔY. “S” represents the propensity to save, represented by the part of income that is saved, I / Y. The first fundamental equation of the Harrod model is as follows:
(6)
If we use the “C” and “S” ratio in the above equation we obtain:
or then I = S
Once Harrod (1939) establishes the methodology and terms to be used he resorts to the following two hypotheses. First, he assumes that we are dealing with a neutral technical progress. This means that investments are neutral and that this does not affect the capital ratio. In fact, the concept of neutrality is regarded as an average, assuming that, in the considered time period, the effect of investments that require more capital (labour saving) per unit of product is offset by the opposite effect of investments that reduce capital ratio (capital saving). Secondly, he considered that the interest rate is constant, which is difficult to be envisaged in a market economy.
Equation (6) is defined by Harrod (1939) as a truism rate (de facto – actual growth rate). It can be considered as an ex-post rate (at the end of the period). Since it is not at all certain that investments will be equal to savings, this rate may be satisfactory or not for entrepreneurs. If the growth rate turns out to be too low, it might displease a good part of investors, seeing their profits diminished.
I continue my analysis of the model with the second concept proposed by Harrod, namely, that of the warranted rate of growth. Entrepreneurs do not need any growth rate, but a rate that can ensure their profits. If in the equation (1), “C” represents a coefficient de facto of capital, made ex post, this time to achieve a desired rate “Gw”, which guarantees the expected profits, it requires a substantial inflow of capital “Cr”. This way, we get to the second fundamental equation:
(7)
Afterwards Harrod (1939) compared equation one and two. If G ≠ Gw means that in the economy act centrifugal forces that cause the economic system to not be in line with the necessary growth rate.
In steady state equilibrium, the actual growth rate should be equal to the desired economic growth rate of entrepreneurs:
If this equality is confirmed, then the actual savings coincide with planned savings. The same equality will be found between the actual capital ratio and the expected capital. In this case, savers, on the one hand, and entrepreneurs, on the other hand, will see their own projects realized and the economy will be in a steady state growth. Of course, the fulfilment of such equilibrium would mean an ideal situation. But it is far from being achieved, because, as noted above, always will arise certain forces that will not permit this to happen.
Consequently, population growth and technical progress impose limits on the possibility of increasing the capital invested. Beyond these limits, the new investments could not be used, even if there was an availability of necessary capital. So the maximum rate of growth of national income that is permitted by increasing population and technological progress is the natural economic growth rate, which Harrod (1939) noted it as “Gn”. This rate is the third fundamental concept used in his model. In other words, it is considered to be a natural rate because it will always be determined by the natural growth of the labour force and natural growth – in the sense of exogenous variable – of technical progress, namely labour productivity. From this point of view, the natural economic growth rate would coincide with the pace of economic expansion needed to meet population growth and technical progress, which, this time, no longer has a neutral character.
In these circumstances, it becomes absolutely clear that, by definition, the actual growth rate “G” may never exceed the natural economic growth rate “Gn” (the maximum rate allowed by the economic system).
Therefore, it can be noted that:
= S (8)
Obtaining this natural growth rate, regarded as a maximum rate and also the optimal growth of national income would also satisfy the interests of the society, because it would exclude unemployment.
That is why the equality between the natural growth rate, the warranted growth rate and the actual growth rate (G = Gw = Gn) are considered as ideal, the economy being in a situation of perfect balance. Thus, the growth rates achieved are the maximum permitted by the availability of production factors and the scientific knowledge.
Unfortunately, the economic facts do not evolve as we would like them to do. Usually the actual growth rate “G” tends to be inferior to the warranted one “Gw”. In this case, the savings will not be used in their entirety. Contrary, if the natural growth rate is higher than the warranted growth rate (Gn>Gw), then we are in a situation whereas population growth and technical progress offers opportunities for investments, but they cannot be exploited fully because of lack of spontaneous savings. If, however, the volume of investments would exceed the savings, the economy would be on an inflationary spiral, because they would be made on account of the increasing budget deficit.
In conclusion, the balance shown is an unstable equilibrium. The considerations regarding the discrepancies between the three growth rates embody an attempt to explain the business cycle fluctuations. We will have an ideal growth process only when:
G = Gw = Gn
If there would be such a situation, the actual economic growth rate would not only satisfy all the expectations of the entrepreneurs, but would completely absorb any available manpower.
2.2. Domar model
Basically, Domar’s (1946) model of economic growth doesn’t differ much from that of Harrod. However, he highlighted some rather interesting particular issues. Domar starts from the observation that the Keynesian model, while containing a detailed analysis of demand and the impact of investments on it, completely ignores the effects that the same investments have on the supply component. For Domar (1946) the investments that appear in Keynes's model have a multiplying effect on demand, but no multiplying effect on the productive capacity, considered to be constant in time.
A fundamental point of economic growth for Domar is the fact that the act of investment always produces a double effect: on one hand, it increases global demand and, on the other hand, it leads to increasing production capacity, of real supply. A balance growth is only possible when the two effects are quantitatively equal, so only when demand growth is equal with the real supply.
For developing his model Domar (1946) starts from certain assumptions, with the most important being:
a) the savings and investments refers to the income at the same period;
b) he takes into account only the net investments, which means that the assumption are aimed only for those investments that increase the existing production capacity and not those that replace the used fixed capital.
Domar’s model regarding the supply side:
The capacity of a society to invest is calculated using the change of the rate of potential output that an economy can produce. Domar (1946) affirmed that the output produced in the economy is proportional to the amount of the stock of capital. This is symbolised by:
(1)
The above relation is the linear production function that has only the capital component. “Yp” is the potential output at a given stock of capital and “σ” is the output capital ratio (the coefficient of capital) that is considered to be constant. Finally “k” represents the stock of capital. Differencing the above relation with “t” we obtain the change in potential level of output:
(2)
Domar’s (1946) model regarding the demand side:
At any point in time the actual level of income is defined by a simple multiplier process.
Like in the Harrod (1939) model “s” represents the propensity to save of the society at a given time measured by “t”.
Taking into consideration the modifying rate of income over time the following equation can be formulated:
(3)
Taking into account these specifications mentioned above, Domar's (1946) model consists essentially of three fundamental equations. They are:
(4)
(5)
(6)
The first equation shows us the change of the productive capacity, namely the total supply, which depends on the investments made in the past and their marginal productivity. The second equation does not present anything other than the Keynesian multiplier relationship, which measures the increasing total demand due to increased investment. The equilibrium condition is given by the balance between demand and supply in the third equation. This allows the following equation:
(7)
From (7) it can be deduced that:
(8)
In formal terms, this means that we can have economic growth based on the full utilization of the workforce only if investments grow at a rate equal to the product between the marginal inclination towards savings and the marginal productivity of investments.
The Domar and Harrod models have many similarities even if there is a time gap of 7 years between them. This is exemplified in Table 2.
Table 2: The similarities between Harrod’s and Domar’s model
Source: Own contribution
The two models have parameters that are the same. For example Gw (the warranted rate of growth) in the Harrod model is the same with the full employment growth rate (ασ) in the Domar model.
Even if the two models use similar concepts and similar equations, there are also important differences. Domar focuses its assumption on the importance of investment (the demand and supply of investment) in determining economic growth, whereas Harrod considers that income is the key factor in the growth process (the importance of demand and supply of savings).
As I have already shown Harrod uses three separate economic growth rates (“G”, “Gw” and “Gn”) in his model compared with Domar that has at its centre only the full employment growth rate (ασ).
Some limitations of the models:
The capital output ratio and the propensity to save are presumed to be constant in Domar’s model, but in the long run they are likely to modify and can have an influence on the steady state of growth.
Harrod considers that capital and also labour should be utilized in fixed proportions. This is not justified because labour can be substituted by capital and the economy will not be confronted with high inflation or high unemployment if the actual rate of growth doesn’t coincide with the warranted rate of growth.
Both models didn’t assess the importance of price change. Changes in price occur over time and can have a stabilizing effect on the economy.
Interest rates change over time. The assumption that the interest rate does not change is not realistic. They have a direct contribution on investment rates.
The models do not take into consideration the role of the government in stimulating growth and economic development.
There is no real distinction in the models between consumer goods and capital goods.
2.3. Solow model
Solow’s (1956) model shows us how increasing savings rate, population growth and technological progress affect economic growth and the production level over a certain period. Before this model the most used one was the model developed by Harrod and Domar (Harrod 1939, Domar 1946) which the study has already presented it in the beginning of this chapter.
Solow's (1956) neoclassical model represents a fundamental landmark in the analysis of the process of economic growth. Aghion and Howitt (2009) said about the Solow model that it shows how economic policy can stimulate economic growth rate by stimulating citizens to save. Also the model predicts this kind of increase in economic growth cannot last indefinitely. In the long-run, the country’s growth rate will return to the rate of technological progress. Acemoglu (2009) stated that: “Solow’s model has shaped the way we approach not only economic growth but the entire field of macroeconomics.”
In his study, Solow starts from the following assumptions:
the economy is perfectly competitive;
perfect mobility of factors of production;
full employment in terms of resources used;
capital (the production factor) is subject to decreasing returns;
returns of scale production are constant.
I will study Solow’s representative model of economic growth in two steps:
a) the analysis of how supply and demand of goods determines the accumulation of capital (initially he considered labour force and technological progress as constants);
b) reducing the constraints of the model by introducing changes in labour force and, later, for the state of technology.
Let’s examine a production function, with two inputs (physical capital K and labour L) which assumes diminishing marginal returns of production factors, constant returns to scale and that takes into account the four Inada (1963) conditions:
Y – the production function.
Inada (1963) conditions:
, for λ > 0.
The condition for constant returns to scale implies a presentation of the production function (income) as follows:
, where
(capital stock/capita) (income (production) per capita)
So the following production function is: y = f (k).
The model takes into account a closed economy with only one sector, where the production (Y) is homogeneous, intended for consumption (C) or investment (I) and to create new units of capital (K). The savings are equal to investments (I = S). If “s” is the part of income that is saved (s is constant and positive), then (1 – s) represents the fraction that is consumed. The capital is subject to depreciation with a constant and positive rate.
(1)
, where K is the derivative of K with respect to time t
Stationary state and capital accumulation
Aghion and Howitt (2009) stated that:
„Because we are assuming away population growth and technological change, the only remaining force that can drive growth is capital accumulation. Output will grow if and only if the capital stock increases.”
An economy's capital stock increases as a result of accumulation (I) and reduces by the depreciation of existing capital (δ × K). If the equation ΔK is dividend in relation to labour force (L), the following can be obtained:
(2)
y,i, δ δ*k (depreciation or required investment)
45o
i*= δ*k* s*f(k) (investment made)
k* k
Figure 2: Stationary state in the neoclassical Solow growth model
Source: own contribution
The level of capital per worker in stationary state is k*, level at which the capital stock remains constant. This means that ΔK = 0 ⇒ s * f (k) = k*δ (depreciation is equal to investments).
The examination of the stationary state leads to the following conclusions:
An economy that is in the stationary state will remain in that state;
An economy that is not in a stationary state, in time will converge to the stationary state;
If the economy starts from a level of capital accumulation k1< k* meaning that investments exceed depreciation, the stock of capital per worker, k will continue to grow until k *;
If the economy starts from a level of capital accumulation k2> k* meaning that investments are lower than the depreciation, the capital stock per work k will fall to k *.
In fact, the optimal level of capital corresponding to the stationary state represents the long-term equilibrium of the economy.
Increasing savings rate (s) and growth
Solow believes that the success of developed economies is due to a high rate of saving. Let's try to highlight this conclusion using graphical representation.
y, i , δ
k*δ
i2* = δ* k2* f(k) *s2
f(k) * s1
i1* = δ* k1*
k1* k2*
Figure 3: The increase in savings in the Solow growth model
Source: own contribution
It can be observed from the above figure (Figure 3) that an increase of the saving rate leads to an increase in capital stock per work so that the economy reaches a stationary state at a higher level of capital (k) and investment (i). Thus k2> k1 and i2> i1.
From the above investigation. we can now understand why large budget deficits have a negative influence on economic growth. Thus, an unstable financial situation reduces savings and causes the crowding-off effect of investment (reducing private investments). We obtain lower levels of savings and of physical capital.
In Solow's (1956) model only temporarily high savings rates lead to a rapid increase in economic growth, because the economy will grow until a new stationary state, with higher levels of savings “s” and capital “k”. Even if we still have a high level of savings rate, it will maintain capital “k” and production “y” at a high level, but cannot maintain a high economic growth rate in the long term.
Solow's (1956) model has demonstrated that only capital accumulation cannot explain sustained economic growth: high rates of savings lead to economic growth only temporarily.
To explain the sustained economic growth the model has to be extended and to incorporate two other factors that have a significant influence on economic growth, these being population growth and technical progress.
Increasing population and economic growth process
Suppose that the population and the labour force grow at the same constant rate represented by “n”. To explain the stationary state with a growing population, we must consider how population growth influences the accumulation of capital per worker. Investments increase the capital stock and depreciation of investments decreases it. If the number of workers grows and K is constant, then the ratio between capital and labour falls.
The investments lead to higher physical capital “k”, whereas depreciation and population growth decrease “k”.
In the stationary state, population growth (n2 > n1) reduces the capital per worker from K1 * to K2 *.
This situation explains why countries with high population growth have a lower level of capital per worker and therefore lower incomes.
Technological progress and economic growth
The preceding analysis has shown that accumulation of factors only contributes to short term economic growth. Only technological progress, which is considered exogenous in the neoclassical analysis, can have an influence on long-term economic growth. All economies will benefit from technological progress (A), which will grow with the same constant rate: A (t) = A (0) eat. I will use a Cobb-Douglass production function, where technological progress is included in the model represented by an intensive utilization of labours.
The Solow production function with technological progress is:
Y = f (K, L * E), where E represents the work efficiency.
If technical progress increases also labour productivity increases. The term L × E is the actual number of workers. Labour productivity E grows at a constant rate g. Because workforce grows at a rate “n” and labour productivity at a rate “g”, the actual number of workers grows at a rate ”n + g”.
The stationary state expressed by equation Δk = s * f(k) – δ*k becomes:
Technological progress favours the process of economic growth, because with the same capital stock we can obtain a higher level of income. In the Solow (1956) model, only technological progress ensures a higher growth rate of y. In the short term, economic growth is driven by capital and technological progress, but in the long-term economic growth is determined solely by technological advancements.
An obvious limitation of the Solow (1956) model is its inability to explain technological progress. Although the model shows that technological progress contributes to economic growth, it does not explain why technological progress occurs. The rate of technological progress is set to “g” without any theoretical relationship to other variables within the model (i.e. the rate is considered exogenous). The normal justification offered is that technological change is rooted in knowledge produced by the public research (e.g. universities, public research institutions) outside the economic system (Brânză 2007).
2.4. Romer’s model of endogenous growth
Acemoglu (2009) and Christopher Carroll (2014) affirmed that Romer’s model initiated the endogenous growth literature and reinitiated the interest in economic growth theory within the community of researchers and economists. The important breakthrough for the existing literature was that his paper “presented a model of increasing returns in which there was a stable positive equilibrium growth rate that resulted from endogenous accumulation of knowledge” (Carroll 2014).
Kim and Heshmati (2014) affirmed that Romer’s model of endogenous economic growth confirmed that technology has a significant importance in the economic growth process. He used technology as an endogenous variable. Romer (1986) stated that technological progress can lead to continuous growth. “Many economists believed that the phenomenal economic growth of the New Economy in the US had been built on ICT technologies (Kim and Heshmati 2014)”.
Romer (1986) has formulated his model of endogenous growth taking into account the knowledge externalities. The higher the average knowledge stock of other companies the higher is the production of a given company. His first model of endogenous growth was improved over the following years (it has to be mentioned the important contribution of the 1990 model (Romer 1990)).
The Romer production function in which we do not have any population growth, the production side is represented by a set of firms in the interval [0,1], with labour augmenting knowledge takes the following form:
, (1)
Where and represent the labour and capital that are rented by the company i. represents the technology which is the same for all firms.
Then we will have to normalize the measure of final goods producers to one and have the following specifications:
and
The constant level of labour is symbolized by “L”. Let’s presume that companies are competitive on every market, employing the same capital to labour ratio and the factor prices will be influenced by the marginal products. The following differentials can be written as:
Even if companies consider knowledge “A (t)” as granted given, the stock of technology moves endogenously for the whole economy. In his model, Romer (1986) considers that this process takes place because of spillovers across firms. He accredits the spillovers to physical capital (Acemoglu 2009).
He assumed that there will be sufficient externalities for continuous growth of knowledge in the economy. He formulated that the stock of knowledge is proportional to the stock of capital, thus:
(2)
If the knowledge A(t) will be substituted from equation (2) into the initial equation (1) the following production function will be obtained:
(3)
Dividing equation (3) by K (t):
The output per capita is:
where represents the capital to labour ratio.
Also E(t) and P(t) can be normalized with f(L) and the following relations can be represented:
and , where P(t) is constant
The market rate of return is also constant (Acemoglu 2009) and it is defined by the following assumption:
, where δ is the depreciation rate
The usual Euler equation implies that consumption is growing at a constant rate
, (4)
where represents the growth rate of consumption;
is the intertemporal elasticity of substitution;
ρ is the discount rate.
The rates of growth of capital, consumption and output are given by the equation (4). Also the rate of growth of capital is the same as the rate of growth of consumption.
Assuming that , we will have positive economic growth. Also the above assumption will no violate the following condition:
(5)
Because of the relation that occurs between the equilibrium rate of growth and population, the latter must be constant. The population growth is represented by “n”. Because f(L) –f’(L) is always growing in L, the higher the population growth the higher the economic growth is. If the labour force (L) is rising, then the economy will not permit a steady state and economic growth rate will rise over time.
According to Acemoglu (2009), output will reach infinity in a defined time and this will violate the assumption of equation (5).
Romer’s (1986) model demonstrates that in certain situations, constant returns to economy-wide knowledge can contribute to endogenous growth. The spillovers are also crucial for long-run economic growth because of the declining yields to private knowledge capital (Klenow and Rodriguez-Clare 2004).
2.5. The Schumpeterian Model
The Schumpeterian model (generally referred to as the Schumpeterian growth theory) centres on quality improving innovations that makes old products obsolete and thus involves the so called “creative destruction” force (Aghion and Howitt 1992).
Schumpeter realized the fact that the financial sector plays a very important part in the process of creative destruction. This contradicts the neoclassical economic view that ignores transaction costs and presumed that the financial sector freely channels savings to the innovative activities and to the most productive investments (Van den Berg 2013).
The production function in a Schumpeterian model at the industry level is as follows (Aghion and Howitt 1992, 1998):
, with 0<α<1 (1)
where Yt represents the output of the final production good in a certain period t;
At represents a parameter of productivity of the intermediate input in period t;
L represents the labour force with identical and constant number of individuals in each period that have a life span of only one period. Each individual offers one unit of labour per one unit of time at zero cost;
Kt represents the quantity used of the intermediate product;
The index t represents time – the period. The time period is a discrete sequence.
Labour receives a unit of labour services supplied inelastic. The utility is dependent on consumption and also is risk-neutral. All the labour in the economy is used for the production of final-goods. The goal of the model is to maximize the expected consumption. The individual consumes a single product (final production goods) that is obtained using labour and a single intermediate product. Also the firms are in perfect competition between each other. At multiplied by L represents the effective labour supply.
Aghion and Howitt (2009 ) declare that in every period ”t” the intermediate product is the outcome of a monopoly, using as all input the final good one-for-one (every unit of final good used as input for every unit of intermediate product). The final output not used for intermediate production is used for research and consumption. This part of the final output expresses the GDP of the economy. The gross domestic product can be written as:
(2)
The gross domestic product is the sum of consumption and investment in intellectual capital (research). Aghion and Howitt’s (2009) model has an economy with no government expenditure, no foreign sector or investments in human or physical capital. Economic growth is a result of technological innovation that has a positive impact on the parameter At. This is done by the improvement in the quality of the intermediate product. In state equilibrium gross domestic product and the intermediate profit of monopolies are direct proportional with the effective labour supply.
At period t the maximization of consumption for a monopolist is obtained by maximizing the profit Φt calculated in units of the final good:
where pt represents the intermediate product price relative to the final good produced;
pt*Kt is the revenue calculated as price multiplied by the quantity of the intermediate product;
Kt measures the quantity of final good, the input cost that has to be the same as the obtained output.
In an economy with perfect competition the equilibrium price of a production factor used is the same as its marginal product value. From equation (1) it can be deduced the intermediate product price as (Aghion and Howitt 2009):
(3)
The profit maximization using the quantity Kt can be described as:
(4)
The equilibrium quantity is:
(5)
The equilibrium profit is:
, (6)
The equilibrium profit and quantity are proportional to the effective labour supply. Also, if we will substitute into equation (1) and (2) the quantity used of the intermediate product Kt with the values from equation (5) we will observe that gross domestic product and final output are also proportional with effective labour supply AtL (Aghion and Howitt 2009).
and (7)
In each period one entrepreneur has the opportunity to create a new version of the intermediate product. So he will be involved in an innovative process. If the outcome of the innovation is positive, the new product will be more productive that in the past period. If the outcome is negative, the product during this period will be the same as the one in the previous period (Aghion and Howitt 2009). Firms that are involved in innovation can have a monopoly on the production of the new intermediate product. But research is a costly process. The probability βt that innovation occurs in a certain t period is defined as:
, representing the productivity-adjusted level of expenditure related to research (Aghion and Howitt 2009).
where stands for the productivity of the new product obtained if we have successful research;
Rt is the amount of final good used for research.
As innovation advances the complexity to research new products becomes much harder. This is why the probability of innovation is inverse proportional to . A Cobb-Douglas innovation function can be written as:
, 0<σ<1 (8)
where π stands for a parameter that signifies the productivity of research in the sector.
According to (Aghion and Howitt 2009) the marginal product of research in generating innovation can be written as:
and
The marginal product of research is positive and is decreasing.
In period t the entrepreneur that obtains a new intermediate product by innovating will be the new monopolistic firm in that period. The return obtained by the firm is the profit earned thanks to the new research multiplied by the probability of successful innovation and minus the cost of research Rt. The return of the firm can be formulated as:
The firm will want to maximize his return taking into account the expenditure for research Rt, so we must derivate the above condition according to Rt:
Using equation (6) you obtain:
(9)
From equation (9) Aghion and Howitt (2009) deduced that productivity-adjusted level of research nt is a constant n and from this resulting that probability of innovation βt is a constant of β = φ(n). From equation (8) the following can be written:
and (10)
Finally, it can be stated that the economic growth rate grt in a Schumpeterian model can be written as gross domestic product per capita. In line with equation (7) the economic growth rate is proportional to the growth rate of the parameter of productivity of the intermediate input.
Depending if a firm is successful in innovating or not in a certain period t, the economic growth will be directly influenced by this event, specifically the probability of innovation.
If there is innovation in period t, the growth rate will be . The probability of this event will be β.
If there is no innovation in period t, the growth rate will be . The probability of this event will be 1-β.
The long-run average economic growth rate can be written as:
G = E (gt) = β*(γ-1)
The average economic growth rate in the long run is equal to the frequency of innovations multiplied with the size of innovations. From equation (10) we have the following:
According to Van den Berg (2013):
“Schumpeter predicted that the improvements in technology and knowledge that entrepreneurs and corporations brought about would ultimately undermine capitalism: the modern corporation, although the product of the capitalist process, socializes the bourgeois mind; it relentlessly narrows the scope of the capitalist motivations; not only that, it will eventually kill its roots.”
CHAPTER 3 INVESTIGATING THE MOST IMPORTANT FACTORS THAT DETERMINE ECONOMIC GROWTH IN THE EUROPEAN UNION: AN ANALYIS OF EU 28 COUNTRIES BETWEEN 1990 AND 2014
3.1. Introduction
Many authors have dealt with the relationship between public expenditure, foreign direct investment, openness, public or private investment, non-economic variables, among others and economic growth (Khan and Reinhart 1990; Barro 1990, Barro and Sala-i-Martin 1995; Devarajan et al. 1996; Brasoveanu et al. 2008; Arpaia and Turrini 2008; Acemoglu 2009; Bagli and Adhikary2014; Shera et al. 2014). There are not too many researchers that tried to analyse in a single model the two components (the private and public sector). Some of them focused on a single field of study like for example the role of education, health care on economic growth, or the role of public and private investment. This is the reason why this study tries to incorporate into the analysis both the public and private influences and determine their link with economic growth in the EU.
As mentioned in the beginning of the first chapter, economic growth theories highlight the different ways in which the present economic activity can influence the future and identify sources that may lead to continuous growth (Boldeanu and Constantinescu 2015). Economic growth theories have evolved over time from the early 18th century (Adam Smith and David Ricardo) to the present new-growth theory and depend on the dynamics of the economic reality and the evolution of economic analysis tools.
The rest of the chapter has the following sections: section 3.2 consists of a short overview of the existing literature, concentrating on the most important economic (“proximate”) and non-economic (“ultimate”) factors that have an influence on growth. Section 3.3 describes the methodology used in the study and what data was used to construct the empirical model. Section 3.4 highlights the results of the study and finally section 3.5 is the conclusion.
3.2. Overview of existing literature
As far as detailing the determinants of economic growth, the fourth sub-chapter of the first chapter tried to underline the main influences that affect growth at country level. Researchers tested the impact on economic growth of such determinants like investment, human capital, research and development, economic and fiscal policies, trade openness, FDI, institutional and political framework, socio-cultural factors, geography and demography. Also, they separated the “proximate” or economic sources from the “ultimate” or non-economic sources (Rodrik 2003; Acemoglu et al 2005; Acemoglu 2009).
Regarding the influence of public expenditure on economic growth there are many conflicting views in the research literature. Some public expenditures do not facilitate growth and are counterproductive. Endogenous growth theory has introduced government spending into the production function of firms or into the utility consumer function. Recent empirical research (Devarajan et al. 1996; Brasoveanu et al. 2008; Arpaia and Turrini 2008; Holzner 2011; Afonso and Alegre2011) shows that for growth it is more important the composition of public expenditure than their overall level.
After the 1990s there are many research studies that tried to disaggregate public expenditure into its components (defence, education, health, infrastructure, etc.) and to evaluate the impact of one or more determinant of economic growth. Holzner (2011) showed that economic affairs, housing and community amenities and education are negatively correlated with economic growth. A negative link between growth of per capita public health expenditure in GDP, growth of per capita public expenditure on education in GDP, growth of the share of total health expenditure in GDP and economic growth was confirmed by the study of Dao (2012) for 28 developing economies.
An important researcher in the 1990s that studied the effects of public expenditure on economic growth was Robert Barro. Barro’s (1991) study of the importance of primary and secondary enrolment for developed and developing countries found a significant effect of these factors on economic growth (GDP/capita).
Furthermore a significant number of researchers analysed and explained the link between FDI and trade components (exports, imports, net export, trade openness, trade restrictions, etc.) and economic growth with some different views depending on the period, sample and regions. A big part of the literature has shown that states with open economies have higher per capita GDP and grow much faster (Romer 1990; Grossman and Helpman 1991; Edwards 1992; Sachs and Warner 1995; Dollar and Kraay 2000; Barro 2003). FDI inflows have a positive impact on the economy and can accelerate the rhythm of economic growth especially in developing countries.
Public and private investments are fundamental components of economic growth theory used by neoclassical and endogenous growth models. Most policymakers and research economists consider that private investment is more efficient than public investment and contributes more to economic growth (Coutinho and Gallo 1991; Mankiw et al. 1992; Khan and Reinhart 1990; Levine and Renelt 1992; Barro and Sala-I-Martin 1995; Nazmi and Ramirez 1997; Bond et al. 2001).
Khan and Kumar’s (1997) showed that for developing countries there is a substantial difference between private and public investment, with the first one being more influential. Prasetyia (2013) concluded that for ASEAN 5 countries (Singapore, Thailand, Indonesia, Malaysia and Philippines) private investment is more relevant for growth in the long-run, but public investment contributes more to economic growth in the short-run.
“Ultimate” or non-economic sources have a very considerable influence on economic growth. The relation is mostly indirect. The transmission operates from the non-economic determinants to the economic ones. For example, institutions can have a direct effect on growth, but can also act upon other determinants like human and physical capital, technological change or investment. Corruption can have an adverse effect on investment. It can have a negative impact on public productivity by making government spending inefficient and creating the distortion of allocated resources (Murphy et al. 1993; Mauro 1995).
Political determinants like political regimes (democracy, totalitarianism), political stability/instability, civil freedom, the perception of politics have also a considerable role in fostering economic growth and can alter the relationship between the state and its citizens (Lipset 1959; Scully 1988; Lensink et al. 1999). Political instability has a negative effect on companies and their willingness to invest, can create violence and anarchy in the society and in the end can have serious consequences on economic growth (Boldeanu and Constantinescu 2015).
Socio-cultural variables also play a great role in economic development. Ethnic diversity and fragmentation, language, religion, superstitions, social conflicts, civic norms, beliefs are among the sociocultural determinants that may have an effect on economic growth. Trust in individuals or in a group of citizens is the most important outcome of sociocultural factors (Putnam et al. 1993; Helliwell and Puttman 1995; Knack and Keefer 1997; Inglehart and Baker 2000; Acemoglu 2009).
Demography and geography are also two significant factors determining economic development. Geography can impact in different ways the economy. Soil quality can have an influence on agricultural productivity. Natural resources directly contribute to the industrialization of a country by providing essential components for production. Climate has a direct impact on productivity and the attitudes regarding consumption. The topography of a region or state can have a positive or negative impact on transportation costs and on communication. And not least diseases can affect health care, production and the accumulation of human and physical capital (Boldeanu and Constantinescu 2015).
The analysis in this chapter of the thesis tries to re-examine the above mentioned findings for an up to date dynamic panel consisting of all the 28 UE countries with a time frame between 1990 and 2014. The model constructed in this study will try to investigate what economic (“proximate” factors) and non-economic (“ultimate” factors) determinants are significant in fostering economic growth in the EU.
As there is not yet a unified growth model the analysis will offer some key insight regarding the factors that are most relevant in explaining growth variation using a number of 30 variables and 6 different panel data techniques. The usage of different panel data techniques is meaningful for this type of study in order to obtain robust and consistent results. The findings of this investigation can shed new light on what is really a very complex and debated subject.
3.3. Methodology and Data
The purpose of this chapter is to establish the main determinants of economic growth in the European Union (EU28) from 1990 to 2014 taking into account private and public influences. In order to investigate what determinants are important for economic growth, this chapter utilizes a statistical model (a growth equation) in which there are a total of 30 variables. It uses as dependent variable the real annual GDP/capita of the 28 countries that are member states of the European Union.
The study also employs time series data in a dynamic panel data model. Some of the variables contain gaps (omitted values) because of the database from which they were collected.
Panel data models are widely used by many researchers in the field of economic growth. Shera et al. (2014) highlighted in their paper the main advantages of using panel data for economic growth models, like controlling for unobserved heterogeneity, improving efficiency of estimation compared with cross-section models. Panel data also control for omitted variables and measurement errors.
The variables that were employed in this chapter will be transformed using logarithm. The investigation does not use the standard logarithm or natural logarithm like most of the research articles. This is because in the data panel there are positive and negative values for some variables (government deficit, foreign direct investment or inflation). Using standard logarithm will reduce the number of observations. In this context, it is appropriate to choose a transformation that behaves like ln (z) when z is positive and like – ln (-z) when z is negative. So the analysis will use a logarithm called “L” = sign(z)*ln(|z| + 1, where z is the value of the variable. It has been called the neglog transformation (Whittaker et al. 2005).
The economic growth equation used in this chapter has the following formula:
, j= (1)
where:
LY: the neglog of real GDP per capita; this variable represents the negative logarithm of per capita real gross domestic product, expressed in euros.
Lyt-1: the neglog of one lag real GDP per capita;
LLE: the neglog of total life expectancy (years); the literature considers that life expectancy has a primary effect on population growth. Improving life expectancy can slow down population growth and can also encourage human capital accumulation. These improvements in life expectancy can also have an important effect on income per capita (Bloom and Sachs 1998; Cervelatti and Sunde 2009; Acemoglu 2009).
LFEC: the neglog of Final Energy Consumption (1000 tonnes of oil equivalent); this variable represents the total sum of the energy which is provided to the final. This sum represents the total final energy that is consumed in agriculture, industry, transportation, households, services, etc.
LFSL: the neglog of financial sector leverage (debt to equity), non-consolidated (%); this ratio of debt-to-equity illustrates the relative proportion of debt used to finance assets to shareholders' equity. This determinant will measure if financial over-indebtedness has a negative outcome on economic development.
LGDEBT: the neglog of General government gross debt (EDP concept), consolidated – annual data (% GDP); this indicator measures the total gross debt at nominal value outstanding at the end of the year and consolidated between and within the sectors of general government.
LEXP: the neglog of Total general government expenditure (% GDP); according to the COFOG classification, total government expenditure is comprised of the total sum of the 10 categories of public spending, namely general public services, defence, public order and safety, economic affairs, environmental protection, housing and community amenities, health, recreation, culture and religion, education and social protection.
LDEFICIT: the neglog of Deficit – Net lending (+) / net borrowing (-) (% GDP); the difference between general government total revenue and total expenses is known as general government net lending (+) / net borrowing (-) and is usually referred to as government deficit (or surplus). This figure is an important indicator of the overall situation of government finances. It is usually expressed as a percentage of GDP.
LEMPL: the neglog of Employment rates by sex, age and degree of urbanisation (% Total); this indicator represents the number of employed persons as a percentage of the working age population between 15 and 64 of years.
LEXPORT: the neglog of Exports of goods and services (% GDP); this indicator represents the exports of goods and services from residents to non-residents.
LIMP: the neglog of Imports of goods and services (% GDP); this variable represents the imports of goods and services from non-residents to residents.
LOPEN: the neglog of Trade Openness (% GDP); the sum between exports and imports proportional to GDP. Because the data for exports and imports is already divided by GDP, openness is L(Exports + Imports).
LPDEBT: the neglog of Private sector debt, consolidated (% of GDP); this indicator represents the total sum of liabilities in the hands of non-financial corporations, non-profit institutions and households. This variable does not consider the transactions within the same sector.
LPROD: the neglog of Real labour productivity per hour worked (Euro/hour/worked);
LGFCF: the neglog of Gross fixed capital formation (Direct investment) (% GDP); this indicator represents the resident producers’ investments, deducting disposals, in fixed assets during a given period.
LFDI: the neglog of FDI – Direct investment in the reporting economy (flows) – annual data (% GDP); this indicator is the international foreign investment of a resident entity that acquires at least 10% of the equity of an enterprise in another country than of the investment.
LINF: the neglog of Inflation, consumer prices (annual %); this indicator is conventionally measured as the variation of the consumer price index in one year.
LPOP: the neglog of population (inhabitants); the total number of persons inhabiting a country measured in a year.
LEDUC1: the neglog of Less than primary, primary and lower secondary education (levels 0-2) (% total);
LEDUC2: the neglog of Upper secondary and post-secondary non-tertiary education (levels 3 and 4) (% total);
LEDUC3: the neglog of Tertiary education (levels 5-8) (% total);
D: is a vector of 10 dummy variables. It contains six dummy variables with which the analysis will want to measure the governance impact on economic growth. It employs the 6 governance indicators established by Kuafmann, Kraay and Mastruzzi (2010) – Voice and Accountability, Political Stability and Absence of Violence/Terrorism, Government Effectiveness, Regulatory Quality, Rule of Law and Control of Corruption. The dummy variables have two values, one and zero. One is given if the rank of a specific governance indicator for a certain country is above 50 and zero if the rank is below 50. The authors rank the indicator from 0 (lowest) to 100 (highest). Also, to observe if there are differences between countries regarding their location in Europe, the chapter uses regional dummies. The World Bank’s “composition of macro geographical (continental) regions” divides each country in separate regions. The 28 EU countries will be separated into 5 regions: Eastern Europe (Bulgaria, Czech Republic, Hungary, Poland, Romania and Slovakia), Northern Europe (Denmark, Estonia, Finland, Ireland, Latvia, Lithuania, Sweden and UK), Southern Europe (Croatia, Greece, Italy, Malta, Portugal, Slovenia and Spain), Western Europe (Austria, Belgium, France, Germany, Luxembourg and Netherlands) and Western Asia (Cyprus). The dummy variable will take the value 1 if the country is in the correct region and 0 if it is not in that specific region. It is expected that a positive or a big influence on economic growth will be observed for the countries in the North or Western Europe and negative or smaller influence for countries in the South and Eastern Europe or Western Asia. The analysis will use only 4 regional dummies, Western Asia being excluded to avoid multicollinearity.
η: is the unobserved country-specific effect;
ε: is the disturbance term;
i is the individual country dimension and t is the time period dimension.
Data are taken from the Annual Macroeconomic database of the European Commission (AMECO), from the World Bank’s Statistical Database and from Eurostat database. All monetary data are expressed in constant prices and denominated in a common currency (ECU). Nominal GDP is deflated using the Eurostat country deflator, with the base year being 2010.
The study will first highlight some key findings related to the evolution of the governance indicators in the European Union.
The indicator for voice and accountability has improved for some EU countries since 1996. This fact can be observed from the below figure for UK, Estonia, Lithuania, Slovakia and Hungary. Other countries have fallen in rank like for example Greece and Portugal.
Figure 4: Voice and accountability indicator in 1996 and 2014 in the EU28
Source: World Bank governance indicators
Figure 5: Political stability and absence of violence/terrorism indicator in 1996 and 2014 in the EU28
Source: World Bank governance indicators
The figure (Figure 5) for political stability and the absence of violence and terrorism highlights a different picture for the 28 EU states. Countries like Sweden, Germany, the United Kingdom, Portugal or Italy were confronted with political instability. This can be attributed to the economic crisis.
Figure 6: Government effectiveness indicator in 1996 and 2014 in the EU28
Source: World Bank governance indicators
Government effectiveness has been improving in the EU in the past years. This is illustrated very well in Figure 6. Countries like Romania, Bulgaria, the Baltic States, Czech Republic and Slovakia saw an upgrade in government effectiveness.
The same amelioration for the Eastern European countries can be seen also for the rule of law and regulatory quality (Figures 7 and 8). But a negative outcome is observed for the Southern EU states like Italy, Greece and Spain and also for Hungary and France.
Figure 7: Regulatory quality indicator in 1996 and 2014 in the EU28
Source: World Bank governance indicators
Figure 8: Rule of law indicator in 1996 and 2014 in the EU28
Source: World Bank governance indicators
Regarding the control of corruption in the EU it can be concluded from the below figure (Figure 9) that Romania, the Czech Republic and Latvia have made some significant advances.
Figure 9: Control of corruption indicator in 1996 and 2014 in the EU28
Source: World Bank governance indicators
Like stated before, the aim in this chapter is to establish the main determinants of economic growth in the European Union (EU 28) from 1990 to 2014 taking into account private and public influences. To investigate what determinants are important for economic growth, the analysis will use a statistical model (a growth equation) in which there are a total of 30 variables and uses as dependent variable the annual real GDP/capita of the 28 countries that are member states of the European Union. The investigation also employs time series data in a dynamic panel data model.
As mentioned in the beginning of the sub chapter the empirical methodology is based on 28 developed and developing countries from the European Union over the period from 1990 to 2014. The choice of the 28 countries and the time period is determined by the availability of data. The list of the states included in the sample, the variables used and the data sources are presented in Appendix I.
Table 3 of the analysis highlights the summary statistics of the variables used, starting with the mean, the standard deviation, the minimum and maximum of the variables and the number of observations. Because there are a total of 28 countries and a time period of 25 years the maximum number of observations is 700.
Table 4 presents the correlation matrix for all the independent variables and the dependent variable – real GDP/capita. This correlation offers a first crude glance of the relationship between these variables. Table 4 shows that real GDP/capita has a positive correlation with life expectancy, final energy consumption, government debt and expenditure, deficit, employment rate, private debt, real labour productivity per hour worked, population, less than primary, primary and lower secondary education and finally with tertiary education. Real GDP/capita has a negative correlation with financial sector leverage, exports, imports and trade openness, gross fixed capital formation, foreign direct investment, inflation and upper secondary and post-secondary non-tertiary education. The graphical illustrations of the correlation (graph correlation matrix) for the variables used in Table 4 are presented in Appendix I.
Table 3: Summary statistics of the variables used
Source: Stata v12
Table 4: The correlation matrix of the variables used in the model
Source: Stata v12.
Figure 10 shows the evolution of neglog of GDP/capita for the 28 countries analysed. From this graph it is obvious that Eastern European states have made a significant progress from the fall of communism in 1989. Countries like Romania, Latvia, Lithuania, Estonia, Bulgaria or Czech Republic have seen improvements in annual growth.
Pollard et al. (2012) analysed 150 countries using the Penn World Table 7.0 and found that there was a difference between regions (convergence club hypothesis) and that slow growing nations of a region are catching up to the richer nations that are in the same area.
Figure 10: The evolution of neglog GDP/capita in the EU 28 between 1990 and 2014
Source: Stata v12
3.4. Empirical results
To empirically estimate the relation between the independent variables and the neglog of real GDP/capita this chapter of the thesis will use several panel data estimation techniques. It will employ the pooled ordinary leased square, random effects model, the feasible generalized least squares estimator, the fully modified OLS, the first difference GMM estimator and the system GMM estimator. This will also offer robustness of the estimation results.
Some of the variables may be nonstationary. The regressions that involve independent nonstationary variables can generate “spurious” results (Ghosh 2012).
To test if the variables contain unit-roots, it is necessary to perform some specific tests. Because the panel is balanced, but having gaps in values, the empirical analysis will utilise two different tests to investigate the stationary hypothesis. The two tests are the Fisher-type unit-root test and the Im–Pesaran–Shin test, which are both suited for unbalanced panel data. These two tests were developed by Fisher (1925) and Im-Pesaran-Shin (2003) for determining the lag and if the variables contain or not unit roots.
Table 5: Unit-root tests results for the variables used
Legend: *, **, *** denote significance at 1%, 5%, and 10%, respectively
Source: Stata v12
Table 5 reports the unit root results for the Fisher and Im–Pesaran–Shin tests. The results confirm the presence of a unit root for several variables. The test is also repeated for the first order differenced variables. By doing so the nonstationary variables became stationary in first difference. The model will be rewritten with all the variables in first order difference, except for the regional dummies. The first difference of variables that do not depend on time can change into a series of only one constant. For the regional dummy variables the differentiation will transform the series into a constant row of zeros. The unobserved country specific effects are removed when the first-differences are taken.
The economic growth equation with the differenced variables is as follows:
, j= k= (2)
The above equation uses the first difference for the governance dummy variables and did not use first difference for the regional dummies. This is because the regional dummies are not time variant and if differenced the values will be a row of constant zeros.
To determine if the empirical investigation should use a FE or a RE model the Hausman test will be computed for the differenced equation (2). If the p-value is less than 5%, then the null hypothesis is rejected and the study has to use fixed effects. Because the p-value is larger than 5%, the null hypothesis can be confirmed. In this case the model is better suited for random effects (Greene 2008).
Table 6: Hausman test
chi2(25) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 28.38
Prob>chi2 = 0.2908
(Ho: difference in coefficients not systematic)
Source: Stata v12
After the above test it is necessary to investigate the following conditions: serial correlation (autocorrelation), homoskedasticity and spatial autocorrelation.
Because the Hausman test confirmed that the model is a random effects one, the Wald test is not appropriate. Therefore, it is required to investigate the hypothesis of homoskedasticity by applying the Breusch-Pagan / Cook-Weisberg test. According to the results of Table 7 the investigation can reject the null hypothesis and conclude that the model is a heteroskedastic one. Furthermore the above heteroskedaticity findings are also tested by applying the LR test for panel-level heteroskedasticity. This test was developed by Poi and Wiggins (2001).
Table 7: Breusch-Pagan / Cook-Weisberg test for heteroskedasticity
chi2(29) = 227.38
Prob>chi2 = 0.0000
(Ho: Constant variance)
Source: Stata v12
The LR test also confirms the presence of heteroskedasticity in the model. The results are presented in Table 8. The issue of heteroscedasticity can be observed also in the below figure (Figure 11). The residuals are concentrated and not well distributed.
Figure 11: Residual plot
Source: Stata v12
Table 8: LR test for panel-level heteroskedasticity
Likelihood-ratio test LR chi2(24) = 291.05
(Assumption: homosk nested in hetero) Prob> chi2 = 0.0000
Source: Stata v12
Using the Wooldridge (2002) test the analysis will examine for serial correlation in the model. The results of the test in Table 9 confirm the presence of serial correlation. According to Torres-Reyna (2007):
“Serial correlation causes the standard errors of the coefficients to be smaller than they actually are and higher R- squared.”
Table 9: Wooldridge test
F(1, 25) = 48.807
Prob>chi2 = 0.0000
(Ho: no first-order autocorrelation)
Source: Stata v12
Baltagi (2008) has stated that cross-sectional dependence can lead to bias results especially in macro panels. The null hypothesis of the Pesaran (2004) test is that the residuals are not correlated across entities. Because the p-value is higher than 5% it can be stated that there is no cross-sectional dependence in the model.
Table 10: Pesaran test
Pesaran's test of cross sectional independence = 1.134, Pr = 0.2566
Average absolute value of the off-diagonal elements = 0.293
Source: Stata v12
The findings of the Pesaran test can be also corroborated with the ones developed by Frees (1995) and Friedman (1937). The two tests show that there is cross-section independence, which is in line with the results of the Pesaran test.
Table 11: Frees' test of cross sectional independence
Frees' test of cross sectional independence = 1.768
Critical values from Frees' Q distribution
alpha = 0.10 : 0.4127
alpha = 0.05 : 0.5676
alpha = 0.01 : 1.9027
Source: Stata v12
Table 12: Friedman's test of cross sectional independence
Friedman's test of cross sectional independence = 8.281, Pr = 0.9993
Source: Stata v12
To opt between a random effects regression and a simple OLS regression the analysis will use the Breusch-Pagan LM test. Because the p-value is higher than 5% the null hypothesis can be confirmed. There are no significant differences across countries. The model is suited for a simple OLS regression.
Table 13: Breusch-Pagan Lagrange multiplier (LM)
Test: Var(u) = 0
chibar2(01) = 0.00
Prob>chibar2 = 1.0000
(Ho: variances across entities is zero)
Source: Stata v12
After applying the above tests the study can continue by empirically estimating the relation between the independent variables and the neglog of real GDP/capita and by this it will use several panel data estimation techniques. The investigation will use the pooled ordinary leased square and the random effects model, even if the Breusch-Pagan LM suggested only using OLS. It will employ also the FGLS estimator, the fully modified OLS, the first difference GMM estimator and the system GMM estimator. This will offer also robustness of the empirical results.
For the OLS regression the study will require to control in Stata for heteroskedasticity and serial correlation by using cluster-robust standard errors because of the above validation tests. Therefore, the study will compute the pooled OLS regression. The investigation will also have cluster-robust standard errors for the random effects model and for the FGLS estimator to control for heteroskedasticity and serial correlation.
Table 14: The results of the pooled OLS and REM regressions
Notes:
Standard errors in parentheses
*** p < 0.10, ** p < 0.05, * p < 0.01
1. Voice and Accountability was omitted because of collinearity. By regressing the depended variable and the dummy variables we could determine that for Voice and Accountability the coef. is 0.29 with a p-value of 0.295. Government Effectiveness and Eastern Europe were significant at 1% and Political Stability and Absence of Violence/Terrorism and Northern Europe at 5%, respectively.
2. Voice and Accountability was omitted because of collinearity. By regressing the depended variable and the dummy variables we could determine that for Voice and Accountability the coef. is 0.29 with a p-value of 0.278. Government Effectiveness and Eastern Europe were significant at 1% and Political Stability and Absence of Violence/Terrorism, Control of Corruption and Northern Europe at 5%, respectively.
Source: Stata v12
Table 14 presents the results of the pooled OLS and the random effects model. Columns 1 and 2 highlight the results for the OLS and REM. The models have the same coefficients and standard errors. Only the confidence intervals and the p-values change. Some of the variables were significant at 1%, 5% and 10% and others were not statistically significant. Life expectancy, financial sector leverage, total government expenditure, deficit, exports, imports and openness, private debt, gross fixed capital formation and foreign direct investment, less than primary, primary and lower secondary education and upper secondary and post-secondary non-tertiary education were not significant. Also, some of the dummy variables had p-values higher than the threshold, like voice and accountability, political stability and absence of violence/terrorism, the control of corruption and some regional dummies like Eastern Europe.
The results of the pooled OLS and REM models affirm that final energy consumption, the employment rate, the real labour productivity per hour worked, tertiary education and inflation had a positive effect on real GDP/capita in the European Union. Real labour productivity had the biggest influence on real GDP/capita growth (0.89%). Labour productivity is one of the main determinants of economic growth and prosperity (ADAMCZYK-ŁOJEWSKA 2013). The explanatory variables which had a negative impact were general government gross debt and population. For example a rise in population by 1% determines a drop in real GDP/capita growth of -1.15%. The interesting results are from the dummy variables. The rule of law has an important effect on real GDP/capita in EU 28 (raising real GDP by 4%), with a negative effect from government effectiveness and regulatory quality. This can be attributed to the specification in this analysis of the governance dummy variables.
The regional dummies also offer interesting results. Because the study opted to exclude from the regional dummies the Western Asia (Cyprus) region, the results of the analysis have to be compared with this one. Countries from Western Europe grew slower than the ones in Western Asia. The same is applied to Southern Europe and Northern Europe. In conclusion, from the regional dummy results, it can be affirmed that Western European countries grew the slowest compared with the ones in the rest of the EU 28.
The positive value of the lag of initial real GDP per capita is a sign of divergence between the 28 countries analysed. Regional convergence has ended in the EU since 2008, at the start of the economic crisis. Divergence between countries and especially EU regions has accentuated since 2008 (Timbeau 2014; Boldeanu and Ion 2015).
Table 15: The results of the FGLS and FMOLS regressions
Notes:
Standard errors in parentheses
*** p < 0.10, ** p < 0.05, * p < 0.01
1. Voice and Accountability was omitted because of collinearity. By regressing the depended variable and the dummy variables we could determine that for Voice and Accountability the coef. is 0.16 with a p-value of 0.027. Eastern Europe is significant at 1%.
Source: Stata v12
Table 15 presents the results for the FGLS and the fully modified OLS estimations. The results from column (1) have almost the same outcomes as the other two methods (OLS and REM) with some exceptions. Like the previous methods life expectancy, financial sector leverage, total government expenditure, deficit, exports, imports and openness, private debt, foreign direct investment and less than primary, primary and lower secondary education were not significant. Almost all of the dummy variables had p-values higher than the threshold except for voice and accountability and the regional dummy for Western Europe.
The FGLS estimation states that final energy consumption, the employment rate, real labour productivity per hour worked, gross fixed capital formation, upper secondary and post-secondary non-tertiary education, tertiary education and inflation had a positive effect on real GDP/capita in the European Union.
Like in the OLS and REM estimations, real labour productivity had the biggest influence on real GDP/capita growth (0.81%). The explanatory variables which had a negative impact were general government gross debt and population. For example a rise in population by 1% determines a drop in real GDP/capita growth of -0.67%.
Regarding the dummy variables, voice and accountability had a positive impact on growth. The countries from Western Europe grew much slower compared with the others in the EU.
The second column of Table 15 presents the results of the FMOLS regression. These have the same sign as the FGLS result, but also some other variables are statistically significant. Life expectancy, FDI and rule of law have a positive effect on economic growth. Interesting is that exports, government effectiveness and regulatory quality did not offer the expected outcome.
The GMM estimator and system GMM estimator techniques are based on the contributions of Arellano and Bond (1991), Arellanoand Bover (1995), Blundell and Bond (1998) and Roodman (2009). The GMM methodology is increasingly popular for estimating panel dynamics. It is suited for “small T and large N”, heteroskedasticity and autocorrelation among other characteristics. It is also employed to correct any endogeneity biases.
By applying the GMM estimator (also called the “difference GMM estimator”) the analysis needs to use the first difference of equation (1). This will be in this case equation (2) because to obtain stationary dependent and independent variables the analysis had to difference equation (1). The right hand side variables of the differenced equation could be used as instruments in the regression. The GMM method uses the lagged dependent variable which is an additional explanatory variable. It was already included in the first equation (eq. 1).
The error component is also differenced and it is assumed that it is not correlated with the independent variables. Also the error term is not serial correlated with the lagged differenced dependent variable.
The study will also compute the system GMM estimator which consists of the first difference and the level equations. Using only lagged level variables as in the difference GMM estimator can bias the parameters (Baum 2006).
According to Arellano and Bover (1995) and Blundell and Bond (1998) the system GMM estimator consists of two equations. One is the standard first-difference equation (eq. 2) and an additional set of level equations. When adding the second equation, we obtain additional instruments. The independent variables in levels in the second equation are instrumented with their own first differences. This process usually increases efficiency.
System GMM uses more instruments compared with the difference GMM. It may not be appropriate to use system GMM with a dataset where there are not too many countries. When the number of countries is smaller than the number of instruments the Sargan test may be weak.
Table 16: The results of the GMM and system GMM
Notes:
Standard errors in parentheses
*** p < 0.10, ** p < 0.05, * p < 0.01
1. X61 was omitted because of collinearity.
2. “GMM estimator” is the first difference GMM “system GMM estimator” is the system GMM.
3. The Hansen J-test and the Diff-in-Hansen test are the p-values for the null hypothesis of instrument validity and the p-values for the validity of the additional moment restriction necessary for system GMM, respectively. AR(1), AR(2) and AR(3) are the p-values for first, second and third order auto-correlated disturbances in the first difference equations.
Source: Stata v12
According to Roodman (2009):
“The Sargan/Hansen test should not be relied upon too faithfully, because it is prone to weakness. Intuitively speaking, when we apply it after GMM, we are first trying to drive (1/N) Z__E close to 0, then testing whether it is close to 0. Counter intuitively, however, the test actually grows weaker the more moment conditions there are and, seemingly, the harder it should be to come close to satisfying them all.”
The regional dummies were dropped from the GMM and system GMM regressions. This is because the two methods compute the first difference of all the variables. As mentioned earlier, the first difference of variables that do not vary in time will yield constant values of 0, 1 or -1.
Table 16 presents the results of the GMM estimator and the system GMM estimator. Column 1 highlights the results for the first-difference GMM estimator. Life expectancy, financial sector leverage, total general government expenditure, the deficit, imports, private sector debt, less than primary, primary and lower secondary education, upper secondary and post-secondary non-tertiary education and gross fixed capital formation were not statistically significant. The explanatory variables that had a significant and positive effect on real GDP/capita were final energy consumption, employment rate, trade openness, real labour productivity per hour worked, foreign direct investment, tertiary education and finally inflation. Trade openness and real labour productivity are the factors that had the most influence on real GDP/capita in the EU28.
The variables that negatively influenced real GDP/capita in the EU 28 according to the GMM estimator were government debt, exports and population. The dummy variables that were significant were government effectiveness, regulatory quality,the rule of law and control of corruption. The last rows of the results highlight the validation tests. The Arellano-Bond tests indicate that there exists a second order serial correlation and the results of the Hansen J-test indicate that the instruments are valid.
The results of the system GMM confirm that life expectancy, general government gross debt, employment rate, imports, foreign direct investment, inflation and all the variables related to education were not statistically significant.
The explanatory variables that had a significant and positive effect on real GDP/capita were final energy consumption, deficit, trade openness, real labour productivity per hour worked. Trade openness is the factor that had the most influence on real GDP/capita in the EU28. The variables that had a negative and significant influence on growth were financial sector leverage, total general government expenditure, exports, private debt, gross fixed capital formation and population. Regulatory quality, control of corruption and rule of law are among the dummy variables that have an effect on real GDP/capita in the EU. Compared with the other methods, the sysGMM showed that financial sector leverage is negatively influencing economic growth. The over-indebtedness of this sector has a negative consequence on all of the other branches of the economy. The results of the Arellano-Bond tests indicate that there is no second order serial correlation. There is third order serial correlation. The null hypothesis of the Hansen J-test and the Diff-in-Hansen tests cannot be rejected.
After estimating the results of the empirical model the study will continue by running a regression diagnostic to check for the normality of the residuals, omitted and unnecessary variables. It has been shown that the model is hetereoskedatic, but this outcome was controlled by using cluster-robust standard errors. The study will also graphically illustrate the heterogeneity across years and countries.
Testing for omitted variable bias is important because it relates to the assumption that the independent variables and the error term are not correlated. Omitted variables have a significant influence on the dependent variable and should be in the model but are excluded. The study will conduct a Ramsey (1969) test to see if the model needs more variables. Because the probability obtained in Table 17 is greater than 10% the null hypothesis can be accepted and it can be concluded that there is no need for more variables in the specified model.
Table 17: Ramsey RESET test using powers of the fitted values of Ly
F(3, 313) = 3.82
Prob > F = 0.0104
(Ho: model has no omitted variables)
Source: Stata v12
Regression models should have residual that behave normality, are normally distributed. The normality hypothesis investigation is needed for testing the validity of the hypothesis. This insures that the p-values for the t-tests and F-test are valid.
Figure 12: Kernel density estimate
Source: Stata v12
For the normality hypothesis the analysis in this section will compute a kernel density estimated graph. It can be observed from Figure 12 that the residual do not follow a normal pattern. There is an upward tail compared to the normal distribution. In practice normality does not represent a big issue when we have a big sample. The fact that the residual do not follow a normal Gaussian distribution can be observed also from the below histogram.
Figure 13: Histogram of residuals
Source: Stata v12
The standardized normal probability plot (Figure 14) checks the non-normality hypothesis in the middle of the range in the residual. The residual follows a pattern slightly off the line, but it can be stated that there is no specific reason to consider this a major problem.
Figure 14: Standardize normal probability plot
Source: Stata v12
Figure 15: Quintile normal plots
Source: Stata v12
The quintile-normal plot is checking for non-normality in the extremes of the data. It plots quintiles of residuals against quintiles of a normal distribution. From Figure 15 it can be observed that there are two tails a bit off the normal.
The study also utilizes a non-graphical test for the normality hypothesis. It will compute the Shapiro-Wilk (1965, p.593) test for normal distribution. The null hypothesis for this test is that the distribution is normal. Because in the below table the probability is < 5%, the null hypothesis can be accepted (at 95%).
Table 18: Shapiro-Wilk W test for normal data
(Ho: the residual are normally distributed)
Source: Stata v12
Figures 16 and 17 present the heterogeneity graphs across countries and years.
Figure 16: Heterogeneity across countries for the panel data
Source: own contribution
Figure 17: Heterogeneity across years for the panel data
Source: own contribution
3.5. Conclusions
This study examined the most important determinants of economic growth in the European Union (EU 28) for the period 1990-2014 using private and public variables
Durlauf, Johnson and Temple (2005) stated that there are approximately 145 different regressors that were found to have a statistically importance in shaping economic growth in at least one paper. Because we have data for almost 150 countries, it is hard to construct a model in which we can calculate their influence on growth. Even if we can include all the variables in a model the probability of gaps in the data (missing values) is quite high.
This study employed different statistical methods to find the significant influence of the regressors on economic growth. It used the pooled OLS regression, the random effects model, the FMOLS regressions, the FGLS estimator and the GMM and system GMM methods. This offered also robustness for the investigation.
From the first two regression methods (pooled OLS and random effect model) it was shown that life expectancy, financial sector leverage, total government expenditure, deficit, exports, imports and openness, private debt, gross fixed capital formation and foreign direct investment, less than primary, primary and lower secondary education and upper secondary and post-secondary non-tertiary education were not statically significant to influence economic growth in the EU 28.
The results of the pooled OLS and REM methods affirm that final energy consumption, the employment rate, real labour productivity per hour worked, tertiary education and inflation had a positive effect on real GDP/capita in the European Union. Real labour productivity had the biggest influence on real GDP/capita. This means that in the EU, the state and private companies should concentrate on stimulating employment and productivity. The unidirectional link from energy consumption to economic growth suggests that energy has a meaningful role in shaping growth and that the state has to use energy policies wisely as not to harm the economy. This concept was hardly debated and confirmed by many research papers. Yu and Choi (1985), Masih and Masih (1996), Lee (2005), Narayan and Prasad (2008), Bhattacharya and Bhattacharya (2014), Mahalik and Mallick (2014) showed that for developing countries (India, China, Pakistan, Turkey, Brazil, Indonesia, etc.) and also for developed countries (France, Australia, Italy, Korea, Japan, etc.) energy consumption plays an important role in shaping economic growth.
The explanatory variables which had a negative impact on growth are general government gross debt and population. For example a rise in population by 1% determines a drop in real GDP/capita growth of -1.15%. Population growth could have a significant negative influence on economic development by impacting the investment and savings behaviours of citizens, the dependency ratio and the quality of human capital. There are also many research papers that suggest that population growth can have negative consequences on economic growth (Shera et al. 2014; Tolo 2011).
Some of the dummy variables that were used to measure governance had a significant influence on growth. Rule of law has an important effect on real GDP/capita in EU28 (raising real GDP by 4%), with a negative effect from government effectiveness and regulatory quality. This can be attributed to the specification of the governance dummy variables in this study.
The regional dummies also offer interesting results. Western European countries grew the slowest compared with the ones in the rest of the EU 28. Also Southern and Northern Europe did not perform as expected.
The FGLS results showed that life expectancy, financial sector leverage, total government expenditure, deficit, exports, imports and openness, private debt, foreign direct investment and less than primary, primary and lower secondary education were not significant.
The FGLS estimator states that final energy consumption, the employment rate, real labour productivity per hour worked, gross fixed capital formation, upper secondary and post-secondary non-tertiary education, tertiary education and inflation had a positive effect on real GDP/capita in the European Union. As with the other methods (OLS and REM), real labour productivity had the biggest influence on real GDP/capita growth (0.81%). The explanatory variables which had a negative impact were general government gross debt and population. The FMOLS regression confirms the above results and also states that life expectancy, FDI and rule of law have a positive effect on economic growth. Interesting is that exports, government effectiveness and regulatory quality did not offer the expected outcomes.
The results of the GMM estimator show that life expectancy, financial sector leverage, total general government expenditure, the deficit, imports, private sector debt, less than primary, primary and lower secondary education, upper secondary and post-secondary non-tertiary education and gross fixed capital formation were not statistically significant. Final energy consumption, employment rate, trade openness, real labour productivity per hour worked, foreign direct investment, tertiary education and inflation had a positive influence on growth. Trade openness and real labour productivity are the factors that had the most influence on real GDP/capita in the EU28.
The variables that negatively influenced real GDP/capita in the EU 28 according to the GMM estimator, were government debt, exports and population. The dummy variables that were significant were government effectiveness, regulatory quality, control of corruption and the rule of law.
The results of the system GMM confirmed that life expectancy, general government gross debt, employment rate, imports, foreign direct investment, inflation and all the variables related to education were not statistically significant.
Final energy consumption, deficit, trade openness, real labour productivity per hour worked had a positive and significant influence on economic growth. Trade openness is the factor that had the most influence on real GDP/capita in the EU28. Trade openness can have an influence on economic growth through a multitude of different channels like technological transfers, the increase in economies of scale, competitive advantage (Chang et al. 2009).
The variables that had a negative and significant influence on growth were financial sector leverage, total general government expenditure, exports, private debt, gross fixed capital formation and population. It is interesting to note that exports and general government expenditure have had a negative effect on growth.
In conclusion, it can be stated from the investigation in this chapter that the EU countries should try to concentrate on stimulating employment and real labour productivity. Tertiary education is the most important type of schooling education that had a significant and positive effect on growth. Education policies should concentrate on stimulating higher education and innovative research.
Furthermore stimulating energy consumption can improve growth, but also we have to take into account climate change. Renewable energy can be considered an alternative to classical energy production. Inflation also had a small positive effect on economic growth. Because many EU countries have small or even negative rates of inflation, this positive influence obtained in this chapter is thus explained. High levels of inflation hinder economic growth, but smaller rates can be helpful to growth. (Barro 1996; Sarel 1996; Mallik and Chowdhury 2001). The negative impact of government debt has to determine EU states to lower or better manage their borrowing and debt service. Also policy makers should take into account that the over-indebtedness of the financial sector has a negative consequence on all of the other branches of the economy. The results of this chapter must be interpreted with caution because of the inherent endogeneity and the omitted variable biases. Furthermore, certain economic and political shocks could have had significant implication for this empirical framework. Further investigation of these inherent shocks could affect the estimation coefficients.
CHAPTER 4 TERRITORIAL ECONOMIC GROWTH IN THE EU: AN ANALYSIS OF NUTS 1 AND NUTS 2 REGIONS BETWEE 2000-2013
4.1. Introduction
Economic growth analysis at country level is well established and even if there is not a unified model, researchers have pinpointed some significant factors that affect the economy. Durlauf, Johnson and Temple (2005) confirmed in their investigation of the economic growth literature that there are approximately 145 different variables that affect growth at country level and that the growth theory at the moment is an open-ended one. Recently there has been a surge among researchers regarding regional economic growth as different models can be constructed and applied based on more data availability.
The empirical research in the field of regional economic growth has tried to determine what variables have an influence on growth and to come to a consensus on the relevant sign of the variation. There are a number of articles that determined a significant link between innovation (R&D expenditures, patent application, population employed in research), transportation (airport infrastructure, roads, highways), population growth, capital formation, energy consumption, public investments and economic growth at EU regional level (Bottazzi and Peri, 2002; Parent and LeSage 2012; Rodriguez-Pose et al. 2012, 2015).
Like in the case of economic growth at country level, there is still not a consensus on the effects of some variable. Also contradictions in results may appear from studies done for different regions like South America, China, Indonesia, North America or Russia (Golubchikov 2007; Spiezia and Weiler 2007; Hartono et al. 2007).
The aim of this chapter is to contribute to the regional growth literature by testing and updating the importance of several determinants (variables). The study will also use a number of different methods to quantify and statistically demonstrate the link between these different variables and economic growth. The growth analysis will be measured for two different territorial levels in the EU 28. Firstly, the investigation will test an economic model on the 98 NUTS 1 regions between 2000 and 2013. NUTS1 areas represent the major socioeconomic regions in the European Union with administrative functions. After that the study will go in depth and analyse a growth model for 273 NUTS 2 regions in the EU also between 2000 and 2013. NUTS 2 regions represent medium-sized regions with a population that varies from 100 000 to 10 million inhabitants.
Gross fixed capital formation, population, fertility rate, population density (the agglomeration factor), life expectancy, employment, tertiary education, infrastructure, tourism and migration are among the variables that will be tested to see if they have an influence on regional growth.
This part of the thesis will also investigate if the regions of the EU 28 are converging or not, by analysing the time frame between 2000 and 2013. Because the investigation is also using in the econometric models the initial level of the dependent variable, we can determine if the coefficient is negative (representing the convergence between regions) for the NUTS 1 and NUTS 2 regions.
At this moment there are contradictions in the literature regarding the fact that there is regional divergence or that there is convergence in the EU. A big part of the researchers consider that, EU regions are diverging, especially after 2008 (Pellegrini et al. 2013; Rodriguez-Pose, Psycharis and Tselios 2012; Timbeau 2014).
In order to achieve the results of the empirical investigation the rest of this chapter is structured around five sections. First, this short introduction is followed by the literature review on regional economic growth. Section 4.3 highlights the methodology used and the data sources with some graphical illustrations of the variables included in the analysis. Section 4.4 presents the findings of the empirical methods applied in this case study. The chapter ends with the conclusions.
4.2. A summary of the existing literature
Economic growth analysis at territorial/regional level is starting to be more and more important for many researchers. This type of study can shed new light on what kind of influences can facilitate economic development at regional level. Different territories have certain characteristics and levels of development starting from infrastructure, industry, the spread of services, tourist facilities or regional taxes. Better understanding how certain regions are influenced by socio, cultural and economic determinants will facilitate us in creating specific policies for fostering regional economic growth.
Pellegrini et al. (2013) analysed the role of European Regional Policy in fostering economic growth in the European Union (EU 15). They wanted to measure if the Structural and Cohesion Funds are efficient tools for mitigating the differences between European regions. The sample analysed consists of 190 NUTS 2 regions, 57 that received Objective 1 structural funds between 1994 and 2006, and 133 that didn’t receive a considerable amount of Structural and Cohesion Funds. Their results showed that Regional Policy in the EU 15 had an impact of 0.6–0.9 percentage points per year in annual per capita GDP growth for regions in Objective 1. Between 1994 and 2006 the total growth of per capita GDP rose more than a quarter for those regions. The results were not so favourable for regional convergence. If we consider only the impact of Structural and Cohesion Funds, regional convergence will be achieved in more than 50 years. The aim of reducing disparities between levels of development of the various EU regions will be a hard process and it will take a long time to be achieved.
Guclu, M. (2013) investigated Turkey’s regional economic growth process between 1990 and 2000 in the context of the Kaldor’s laws. He found that the growth of neighbouring regions has an effect on the growth of a region. Also the manufacturing sector has a very important role in regional economic growth for the 71 provinces analysed by Guclu (2013). The results of his research have confirmed that all the relations proposed by Kaldor’s laws are valid for Turkey.
There are contradicting views regarding the impact of public investment at regional level. Some view public investment (especially infrastructure investment) as an important factor for growth and productivity (Aschauer 1990; Munnell 1992) and others are sceptic on the exact returns and the implications of public investment on economic growth (Garcia-Mila and McGuire 1992; Crescenzi and Rodríguez-Pose 2012; Rodriguez-Pose, Psycharis and Tselios 2012).
Rodriguez-Pose, Psycharis and Tselios (2012) showed that public investment has a significant impact on the economy. This link is stronger in the long-run than in the short-run. Their results also indicate that growth is affected differently by different types of per capita public investment expenditure and that the spillovers of some types of public investment (especially investments in transport infrastructure) are essential for Greek regional economic growth.
Many authors demonstrated the importance of public investment spillovers in the diffusion of externalities across regions (Ottaviano 2008). Their analysis of 51 regions (NUTS 3 level) in Greece also showed that both in the short-run and in the long-run, research and education, infrastructure investment and housing are the most important public investments that the Greek state has made.
Concerning convergence between Greek prefectures Rodriguez-Pose, Psycharis and Tselios (2012) implied that there is no clear indication of this happening. They stated that “Greek development policy was either not territorially progressive enough or that public investment may have been more efficient in more developed regions.”
There are many views in the literature that consider political factors to be very important in allocating public investment at regional level. Usually politicians can be bias and allocate resources to already developed regions, because they want to please their voters. Building roads, ports or bridges is also a very public and visible statement for politicians in showing that they are implicated in regional development.
Infrastructure investment can bring significant external benefits. It can generate an investment multiplier effect (job creation, increase in productivity) creating an increase in personal wealth and shaping the environment (Kessides 1993). Investment in infrastructure can decrease transportation costs and lower the waiting time in production. These effects have a beneficiary outcome on trade and lower the prices of goods (Pol 2003).
Crescenzi and Rodríguez-Pose (2012) also analysed the importance of public investment, particularly transport infrastructure (kilometres of motorways) in determining economic growth at European territorial level (NUTS regions) between 1990 and 2004. The correlation between infrastructure and economic growth was put in relation also with innovation, a social filter and migration. Contrary to the established thought that infrastructure is positively related to growth, their results showed that infrastructure endowment is poorly linked with economic growth. Also the regions that were surrounded by those with good infrastructure were not significantly influenced. Innovation and the social filter were more important for regional growth in the EU and also the regions that attracted migrants were influenced positively.
The positive link between innovation (investment in science and technology and R&D) and territorial growth has been demonstrated also by recent scientific works (Crescenzi et al. 2007; Usai, 2011; Rodríguez-Pose and Villarreal Peralta 2015).
The recent research of Rodríguez-Pose and Villarreal Peralta (2015) showed that between 2000 and 2010 Mexican states which invested more in research and development had a significant increase in regional GDP growth compared with the states that invested less or that did not invest at all. Also human capital endowment had important consequences on regional economic growth in Mexico. The states that were neighbours to those with high investment in research and development and high GDP/capita tend to grow faster compared with the ones that have poor state neighbours.
Population density can play an important role in regional economic development. High agglomeration in capital cities and large urban areas can have an influence on growth, increasing labour specialization and productivity (Puga 2002). van Oort, de Geus and Dogaru (2015) showed that agglomeration plays an important role for 15 EU countries at regional level, specifically for 205 EU NUTS2 regions. Regional heterogeneity is influencing employment growth and that different levels of specialization are related to productivity growth. Many authors have found that there is an important difference between small and medium size urban areas and large size urban areas and that this has an impact on economic development (Theissen et al. 2013).
Ali and Ahmad (2010) studied the link between FDI and regional economic development in Malaysia. The authors showed that foreign direct investment is significantly and positively influencing economic growth, but because FDI was concentrated more in the most developed regions (Selangor, Penang, Malacc and Johore) it created a divergence shift between them and the poorer regions in the country. In conclusion they found that FDI contributed to regional disparities in Malaysia between 1980 and 2006.
Another study that investigated the economic performance at regional level is the one of Ghosh (2012). He analysed the importance of some economic factors in determining growth for 15 Indian states during 1960 and 2007 (ante and post reform periods) and also if there was regional disparities between them. Using the principal component analysis the author showed that physical and social infrastructure has a significant and positive effect on long term economic growth for Indian states. Investing more funds in these two areas can improve economic growth and reduce regional imbalances for the states that have lower steady-state levels. In regards to regional disparities Ghosh (2012) has shown that in India, regional divergence has accentuated since 1991 even if the post-reform period saw an increase in GDP/capita. He also determined that there are different club convergence regions, one converging to the steady-state path and two non-convergent. The regional divergence in per capita income has originated from the divergence in income in the industry and services sectors.
At country level, there are comprehensive and well established papers that investigated the role of tourism on economic growth, but not too many studies focused on analysing the regional component. Paci and Marrocu (2014) investigated the impact of tourism (domestic and international) on economic growth for 179 regions (Western European regions) between 1999 and 2009. Their results showed that regional economic growth is positively influenced by domestic and international growth and that domestic tourism plays a more important role than international tourism at regional level.
The chapter of the thesis will continue by presenting the methodology used and the data selected for this investigation.
4.3. Methodology and the data used
The primary objective of this chapter is to evaluate the most important determinants of regional economic growth for NUTS 1 and NUTS 2 regions in the European Union between 2000 and 2013. For each level of territorial division in the EU (NUTS 1 or NUTS 2) the investigation will employ a separate growth equation and it will use as dependent variable the regional real GDP per capita and regional real GDP in purchasing power standard per inhabitant.
The determinants that will be measured by the growth equations are population, fertility rate, population density (the agglomeration factor), life expectancy, employment, R&D expenditure, tertiary education, infrastructure, tourism, migration, employment rate among other. All the values are expressed at constant market prices and denominated in euros for the monetary variables. Nominal GDP is deflated using the Eurostat country deflator, with the base year being 2010. The models will be applied on dynamic panel data for a number of 98 NUTS 1 regions and 273 NUTS 2 regions. The NUTS classification refers to the Nomenclature of Territorial Units for Statistics of Eurostat, the Statistical Office of the European Union.
All the variables that are used will be transformed using the neglog transformation. This is because there are also negative values for some variables. The neglog transformation behaves like ln (z) when z is positive and like – ln (-z) when z is negative (Whittaker et al. 2005). Therefore the study will use a logarithm called “L” = sign(z)*ln(|z| + 1, where z is the value of the variable.
Because the case study of this chapter of the thesis will want to investigate two different territorial levels, it will have to employ two separate growth equations.
The regional economic growth equation for the NUTS 1 level has the following formula:
(1)
The regional economic growth equation for the NUTS 2 level has the following formula:
(2)
where:
LY: the neglog of regional real GDP per capita. This variable will be expressed also as the regional real GDP in PPS standard per inhabitant to see it there are differences between the two indicators of growth. According to Eurostat expressing gross domestic product in purchasing power standards cancels the differences in price levels between countries. By calculating GDP per inhabitant makes it easier to compare different countries and regions in comparison with a calculation in absolute size. Also the eligibility of the structural programs of the European Union for the NUTS 2 are offered by determining and comparing the GDP per inhabitant in purchasing power standards.
Lyi,t-1: represents the neglog of one lag regional real GDP per capita or one lag regional real GDP in PPS standard per inhabitant. It is usually introduced in the growth equation to measure the convergence or divergence hypothesis. If the coefficient of this variable is negative then we can state that the EU regions are converging or the less developed ones are catching-up to the most developed ones. This indicator is very important for this type of regional analysis.
LPOP: the neglog of regional population (inhabitants);
LFERT: the neglog of regional fertility rate. It is the average number of children that would be born to a woman over her lifetime. Micheli and Zuanna (Micheli and Zuanna 2005, p.80) see fertility rate as a proxy for the spread of a full motherhood experience;
LLIFE: the neglog of regional life expectancy measured in years. In the research literature, it is an important indicator and proxy for measuring the health of the inhabitants.
LELET: the neglog of early leavers from education and training. Flisi et al. (Flisi, Goglio, Meroni, and Vera-Toscano 2015) consider it to be a proxy of the size of the group of individuals most at risk on the labour market;
LTERT: the neglog of regional persons with tertiary education (percentage of total). It is a measure for human capital and for skilled labour. Some debate if persons which have finished tertiary educations are more skilled and find jobs faster than less educated persons.
LWHOURf: the neglog of regional average number of usual weekly hours of work in main job for female. The LWHOURm is the neglog of regional average number of usual weekly hours of work in main job for male. With this variable, this case study will want to determine if the number of hours worked has an impact on growth. Because of regulations and other socioeconomic factors, the average number of hours worked has declined in the developed world.
LEMPL: the neglog of regional employment rate. This indicator represents the number of employed persons as a percentage of the working age population between 15 and 64 of years. This indicator will be also divided into male and female employment to investigate if there are differences between genders.
LR&Dexp: the neglog of regional total intramural research and expenditure for all sectors (% of GDP). Intramural expenditures are expenditures for research and development during a specific period, whatever the source of funds. Both current and capital expenditures are included.
LMOTORWAY and LROADS: the neglog of regional motorway and roads (other roads besides highways) measured in kilometres. These two indicators are proxies for regional infrastructure development. Infrastructure is seen as a key investment for regional development and also for the convergence hypothesis.
LTOURISMint and LTOURISMext: the neglog of regional total nights spent by residents and non-residents in tourist accommodations (% of total). According to the United Nations World Tourism Organization, a ‘tourist’ is defined as a person who spends at least one night in official tourist accommodation establishments. These indicators are measured as percentage of total population.
LVEHICLES: the neglog of regional vehicles (except trailers and motorcycles). It is a proxy for stock of vehicles.
LDENSITY: the neglog of regional population density (persons per km2). Population density is the ratio between the annual average population and the land area of the region. It is a proxy for regional agglomeration. Usually large and densely populated regions should have a positive effect on regional economic growth.
LMIGRATION: the neglog of regional net migration (%). The rate of net migration is the ratio of net migration plus adjustment during the year to the average population in that year.
η: is the unobserved regional-specific effect;
ε: is the disturbance term;
i is the individual regional dimension and t is the time period dimension.
Data are taken from the Eurostat database. All monetary data are expressed at constant market prices and denominated in a common currency (ECU). Regional nominal GDP is deflated using the Eurostat country deflator, with the base year being 2010. Below, the study will continue with some graphical illustrations of regional disparities for some of the variables analysed.
The chapter will use several panel data estimation techniques to obtain the economic growth results and to offer some robustness. It will employ the first difference GMM estimator and the system GMM estimator and the cross-section time-series dynamic panel data estimation by quasi-maximum likelihood. The last estimation has been developed by Kripfganz (2016). The ML (maximum likelihood) approach was pioneered by Bhargava and Sargan (1983), further developed by Hsiao, Pesaran and Tahmiscioglu (2002) and is suited also for panel data with missing values. Missingness can be solved by implementing a ML estimation or a multiple imputation technique.
According to Hsiao, Pesaran and Tahmiscioglu (2002) maximum likelihood estimation “approach appears to dominate the GMM approach both in terms of the bias or root mean squares error of the estimators and the size and power of the test statistics.”
Quasi-maximum likelihood method does not use any instruments like the GMM or system GMM methods. Also the weak instruments that may be used in the GMM and SysGMM are avoided in QML estimation.
The below picture illustrates the regional population structure in the EU. The current demographic situation of the 28 EU countries is characterised by increasing population (1.3-1.7 million each year). The population of the 28 EU countries as a whole has increased in 2013, but there are 12 EU countries that had a decrease in population. From Figure 18 we can see that the regions with the smallest population are the ones in the Scandinavian countries, Greece and parts of Central Europe.
Figure 18: Population on 1 January by NUTS2 regions in 2013
Source: Eurostat
In 2013 regional gross domestic product varied across EU regions. As we can see from the below figure (Figure 19) most of the Eastern European regions, the South of Italy, Greece, Portugal and South of Spain and the Baltic states are below 75% of the EU 28 average. The leading regions in 2013 were Inner London (United Kingdom) with 325% of the average, the Grand Duchy of Luxembourg (258%), Bruxelles/Brussel in Belgium (207%), Hamburg in Germany (195%) and Groningen in the Netherlands (187%). Mayotte in France had the lowest ranking (27%), followed by regions in Bulgaria and Romania: Severozapaden (30%), Severen tsentralen (31%) and Yuzhen tsentralen (32%) in Bulgaria and Nord-Est in Romania (34%).
Figure 19: Regional gross domestic product (PPS per inhabitant)
by NUTS2 regions in 2013
Source: Eurostat
Figure 20: Motorways network (kilometres) by NUTS 2 regions in 2013
Source: Eurostat
Between 2007 and 2013 (the EU programming period) more than 28% of Regional Development Funds (ERDF) and Cohesion Fund were allocated for infrastructural investments. EU politicians consider infrastructure and especially highway development a key mechanism in order to achieve regional convergence and economic development.
Between 2000 and 2012 the most significant infrastructure development has occurred in Ireland, Spain, France and Hungary. The picture looks different when absolute changes over the same period are considered. In absolute figures, the longest highways were built in Ireland – Southern and Eastern Irish regions (+556 km) and in Spain – (Extremadura (+456 km), Galicia (+378 km), Aragón (+364 km), Región de Murcia (+291 km) and Comunidad de Madrid (+276 km)). In relative terms, the most significant development took place in the Romanian (Sud-Est region), the Border, Midland and Western region of Ireland and Lódzkie in Poland.
Figure 21: Total intramural R&D expenditure (GERD) % GDP
by NUTS2 regions in 2013
Source: Eurostat
Expenditure for research and development is one of the most widely used measures of innovation inputs. Figure 21 shows the differences between the allocation of research and development expenditure at regional level in the EU. In 2013, R&D expenditure ranged from 0.48% (Cyprus) to 3.32% across the EU. The most involved regions in research and development are the Nordic regions (Finland and Sweden) and Austria. Europe’s 2020 strategy set a target in relation to R&D expenditure of at least 3.00 % of the EU’s GDP. The regions with the lowest R&D expenditure levels were mostly in Eastern and Southern Europe (Romania, Bulgaria, Cyprus, Malta and Greece).
The second target of the Europe 2020 strategy is raising the share of the population that finishes tertiary or equivalent education to at least 40 %. Low levels of tertiary education can have a consequence on competitiveness and limit innovation.
Figure 22: Tertiary education attainment, age group 25-64 % total
by sex and NUTS2 regions in 2013
Source: Eurostat
The overwhelming majority of regions in the EU-28 have tried to follow the EU 2020 strategy regarding education attainment. As an outcome tertiary levels of education have risen in 222 regions between 2002 and 2013. A negative fact was that tertiary level education declined in 41 regions, for example in Ceuta (Spain), in the northern Bulgarian region of Severen tsentralen, in Dresden and Chemnitz (Germany), in Basse-Normandie and Languedoc-Roussillon (France).
Europe was the most visited continent by tourists in 2013, with more than a half of all international tourist arrivals in the world.
Figure 23: Nights spent by non-residents at tourist accommodation establishments
by NUTS2 regions in 2013
Source: Eurostat
Figure 23 and 24 provides an image of the total number of overnight stays (residents and non-residents) in all types of tourist accommodation in 2013. From the two figures it is clear that the most nights spend by tourists were concentrated in coastal regions (principally in the Mediterranean), the Alpine regions and some of the EU’s major cities.
According to the World Travel and Tourism Council domestic tourism flows is the most generalized method of spending leisure time for European citizens (more than 70%) (WTTC 2012).
In 2013 Canarias (Spain) had the highest number of nights spend by residents and non-residents with more than 89 million nights. In top five we can also find two other Spanish regions, Cataluña and the Illes Balears, the Île de France region and Jadranska Hrvatska island region of Croatia. An important fact was also that 28 NUTS 2 regions achieved more than 20 million nights spend by tourists. These were located in Italy, Spain, France, Germany, Greece, Austria, Ireland, Croatia, Netherlands and United Kingdom. Also among these 28 regions, 6 were capital city regions in Germany, France, Ireland, Italy, Netherlands and UK.
Figure 24: Nights spent by residents at tourist accommodation establishments
by NUTS2 regions in 2013
Source: Eurostat
From figure 23 and 24 we can see a clear difference between nights spend by residents and non-residents. For Poland and Romania the residents account for more than 80%, while in Cyprus, Malta and Croatia the non-residents account for more than 90% of the nights spend in tourist accommodation. In Germany the northern regions were predominantly visited by residents of Mecklenburg-Vorpommern. Nights spend by nationals also peaked in Lincolnshire (UK), in Basilicata (Italy), in Principado de Asturias (Spain) and in Auvergne (France).
As mentioned in the beginning of the chapter the aim of this section is to determine what variables influence regional economic growth in the EU 28 between 2000 and 2013. The growth analysis will be measured for two different territorial levels in the EU 28 (98 NUTS 1 regions and 273 NUTS 2 regions).
The list of the EU 28 regions included in the sample, the variables used and the data source is presented in Appendix II.
Table 19 illustrates the summary statistics of the variables used in equation 1 for the NUTS 1 regions. Because there are a total of 98 regions and a time period of 14 years the maximum number of observations is 1372.
Table 19: Summary statistics of the variables used in Eq 1
Source: Stata v12
Table 20 highlights the summary statistics of the variables used in equation 2 for the NUTS 2 regions. Because there are a total of 273 regions and a time period of 14 years the maximum number of observations is 3822.
Table 20: Summary statistics of the variables used in Eq 2
Source: Stata v12
Table 21 presents the correlation matrix for all the independent variables and the two dependent variables in equation 1 (NUTS 1). Real GDP/capita and real GDP in PPS/inhabitant are positively correlated with fertility rate, life expectancy, early leavers from education and training, persons with tertiary education, employment rate (total employment, male and female employment), R&D expenditure, motorway infrastructure, nights spent by non-residents (% of total) and the number of vehicles in a region. Real GDP/capita and real GDP in PPS/inhabitant are negatively correlated with population, the average number of usual weekly hours of work in main job (for male and female), other road infrastructure and total nights spent by residents in tourist accommodations.
Table 21: The correlation matrix of the variables used in equation 1
Source: Stata v12
Table 22 presents the correlation matrix for all the independent variables and the two dependent variables in equation 2 (NUTS 2). Real GDP/capita and real GDP in PPS/inhabitant (NUTS 2 regions) are positively correlated with fertility rate, life expectancy, early leavers from education and training, persons with tertiary education, employment rate (total employment, male and female employment), R&D expenditure, motorway infrastructure, nights spent by non-residents (% of total), population density, net migration and the number of vehicles in a region. Real GDP/capita and real GDP in PPS/inhabitant are negatively correlated with population, the average number of usual weekly hours of work in main job (for male and female) and total nights spent by residents in tourist accommodations (% of total). Also GDP/cap is positively correlated with other road infrastructure, but GDP PSS/inhabitant is negatively correlated with this variable.
Table 22: The correlation matrix of the variables used in equation 2
Source: Stata v12
The graphical illustrations of the correlation (graph correlation matrix) for the variables used in table 21 and 22 are presented in Appendix II.
4.4. Empirical results
This part of the study will highlight the variables that had a significant influence on regional economic growth in the EU 28.
Before applying the regression models it is important to do some preliminary investigation on the equations and variables used. To not have “spurious” results the variables need to be stationary (Ghosh 2012). For testing the stationary hypothesis the study applied the Fisher-type (1925) unit-root test which is suited for this panel data. The results are presented in Table 23.
Table 23: Unit-root test results for the variables used
Legend: *, **, *** denote significance at 1%, 5%, and 10%, respectively
Source: Stata v12
Table 23 describes the unit root results for the Fisher test. Some variables are not stationary. The issue was solved by applying the first order difference for the variables that were not-stationary. The two regional economic growth equations will be rewritten with all the variables in first difference. The unobserved regional specific effects are removed when the first-differences are taken.
The regional economic growth equations with the differenced variables are as follows:
NUTS 1 regions:
(1)
NUTS 2 regions:
(2)
Before implementing the GMM, system GMM and QML estimation, the Hausman test has to be computed for the Quasi-maximum likelihood estimation to see if fixed effects or random effects are needed. For the NUTS 2 regions the QML-FE estimation will be computed for the models with real GDP/capita and real GDP in PPS/inhab with the total employment variable and only for the model with real GDP in PPS/inhab with the employment rate split into male and female rates. A QML-RE estimation will be applied for the model with real GDP/capita with the employment rate split into male and female rates.
Table 24: Hausman test for the QML estimation
Source: Stata v12
To eliminate the most common sources of cross-sectional dependence, the panel estimation techniques will include time dummies for the QML-FE and for the GMM methods. The Parm test was utilized to see if time fixed effects are needed for the QML-FE regressions and it confirmed the null hypothesis of the importance of time dummy inclusion.
Table 25: Parm test for the QML-FE estimation
Source: Stata v12
For the GMM and system GMM techniques the lagged values of the dependent variable (real GDP/capita and real GDP in PPS/inhab.) and the variables that are weakly exogenous are used for GMM style instruments. Average number of usual weekly hours of work in main job for male and female, the employment rate (total, female and male), total intramural R&D expenditure and tertiary education were used as GMM style instruments with lag two and also with all available lags. The rest of the regressors were used for the IV style instruments. The study will also introduce time dummy variables in the models. To limit the number of instruments (as a rule the number of panel data units should be higher than the number of instruments), the collapse option is used in Stata. The third to the fifth lag of the dependent variables are used in the GMM style instruments. The analysis will include also the “robust“ command to control for heteroskedasticity and autocorrelation within panels.
The results of the GMM and system GMM estimations for the NUTS 1 and NUTS 2 regions are presented below. In table 26 columns 1-4 are for the GMM estimator with real GDP/capita as dependent variable and 5-8 for the real GDP in PPS/inhabitant. Columns (1), (2), (5) and (6) have the GMM style instruments in second order lag only. Columns (3), (4), (7) and (8) have the GMM style instruments for all suitable lags.
Table 26: The results of the GMM method for the NUTS 1 regions
Notes: t statistics in parentheses. * p < 0:10, ** p < 0:05, *** p < 0:01. All regressions include time dummies
Source: Stata v14
From Table 26 it can be observed that all the coefficients of the lagged dependent variables (real GDP/capita and in PPS/inhabitant) were positive, with a significance level smaller than 1% in almost all of the regressions. This suggests a divergence trend among the regions in the European Union. This is a confirmation of the findings in the third chapter where all of the 28 countries in the EU were investigated.
Life expectancy is not statistically representative in the regressions, but the coefficients were mostly positive. Regarding the other regressors it can be stated that population had from all the 8 columns a negative effect on economic growth. Fertility rate appears to be significant and negative only in the equation with real GDP in PPS/inhabitant when taking into account the second order lag of the GMM style instruments. As expected, early leavers from education and training is negatively linked with growth at regional level. If the percentage of the population aged 18-24 in lower secondary education and who were not in further education or training rises by 1%, real GDP drops between 0.10% and 0.55%. This is an effect that has to be taken into consideration at European level.
Analysing tertiary education and average number of weekly hours at main job offers interesting hypothesis. Tertiary education is negatively determining economic growth and its coefficient was significant only for the regional real GDP/capital regression with all suitable lags as instruments. The average number of weekly hours worked by male has a very important negative effect on growth. This is in contrast to the positive effect of male employment rate, which is positively determining regional economic growth. The employment rate for female is negatively influencing growth only in the sample for all suitable lags as instruments in the real GDP/capital model.
Another obvious result from Table 26 is the fact that expenditure in research and development is fostering economic growth. It is true that the variable is significant only in regression (7) and (8), but the level of significance is at 5%. These results were confirmed also by Crescenzi and Rodríguez-Pose (2012) for EU 15 regions.
Regarding infrastructure development it can be said that motorway development has a small but negative effect on growth and other types of roads a positive effect when the real GDP/capita equation is taken into consideration.
Total nights spend by residents has a significant and negative impact on growth when we take into consideration the real GDP measured in power purchasing standard per inhabitant. Total nights spend by non-residents have a positive effect on regional economic growth. The last variable to be highlighted in Table 26 is the stock of vehicles. It has a positive outcome on regional growth, but is statistically weak.
In the case of the Sargan and Hansen tests, the p-value for the Sargan test is significant for the results in columns (1,) (2), (5) and (6). The p-value for the Hansen test confirms the validity of selected instruments. According to Roodman (2009) the Hansen test is more realistic if the number of instruments is low. The Arellano-Bond test for serial correlation indicates that there is no first order serial correlation, but second order serial correlation is detected.
Table 27: The results of the system GMM method for the NUTS 1 regions
Notes: t statistics in parentheses. * p < 0:10, ** p < 0:05, *** p < 0:01. All regressions include time dummies
Source: Stata v14
Regarding the system GMM results for the NUTS 1 regions, the evidence of regional divergence is still present. The negative outcome of population size is only statistically significant in columns (3) and (4), with smaller coefficients than previously illustrated. In Table 27 the fertility rate appears to be significant and positively influencing economic growth. Life expectancy appears to be a very important factor in stimulating growth at the regional level. The coefficients for early leavers from education and training are still negatively linked with growth at the regional level, but with values not so high like in the GMM estimator and only significant in 5 out of the 8 columns.
Compared with the GMM estimator in Table 26, the system GMM highlights the importance of having tertiary education. Average number of usual weekly hours of work in main job for male is still negative, but statistically significant only in column (3).
The results for employment are comparatively the same with the ones obtained for the GMM method. Total employment has a positive effect on growth. Male employment is also significant and positive and female employment is negatively correlated with growth.
A contradictory result is the fact that in the system GMM expenditure in research and development is negatively affecting regional economic growth. This outcome has to be evaluated in the future regressions of this chapter. Another reversal of the results is obtained for infrastructure development. It can be observed that motorway development has a small but significant effect on growth and other types of roads a small but negative effect.
The total nights spend by residents and non-residents are important for regional growth. The coefficient for the first variable was significant only in regression (3), but the one for non-residents was significant in half of the regressions. Also the regional stock of vehicles is important for growth and this has to be taken into consideration by policy makers.
The p-values for the Sargan and Hansen tests were smaller compared with the results obtained by the GMM regressions. The Arellano-Bond test for serial correlation indicates that there is no first order serial correlation, but second order serial correlation is detected only for the estimation with real GDP in PPS/inhab.
The next step in investigating the panel data using GMM and system GMM is to analyse what variables influence regional economic growth for the 273 NUTS 2 regions.
Table 28: The results of the GMM method for the NUTS 2 regions
Notes: t statistics in parentheses. * p < 0:10, ** p < 0:05, *** p < 0:01. All regressions include time dummies
Source: Stata v14
In table 28 columns 1-4 are for the GMM estimator with real GDP/capita as dependent variable and 5-8 for the real GDP in PPS/inhabitant. Columns (1), (2), (5) and (6) have the GMM style instruments in second order lag only. Columns (3), (4), (7) and (8) have the GMM style instruments for all suitable lags.
This section of the analysis follows the same methodology for constructing the GMM and system GMM regressions, like for the NUTS 1 regions. The two new independent variables, population density and migration, were used for the IV style instruments.
The same trend of regional divergence among the NUTS 2 regions is still observed from the results in Table 28. Regarding the other independent variables in the regressions, population size had an important effect on growth in columns (3) and (4). Fertility rate appears to be statistically significant and negative in six out of the eight columns and life expectancy is positive at 10% only in column (2). The variable for early leavers from education and training is negatively linked with growth only in the regressions with all the suitable lags of the instruments.
Tertiary education negatively determines growth and its coefficient was significant only for the regional real GDP/capital regression with all suitable lags as instruments. The average number of weekly hours worked by male has a very important negative effect on economic growth in column (1) when only the second order lag of the instruments is used. In columns (3) and (4) weekly hours worked by male are significant and positive. Also the average number of weekly hours worked by female is negative in the real GDP in PPS/inhabitant equation. Total employment rate and male employment have an important outcome on regional economic growth, with female employment rate being negatively correlated with growth.
Research and development expenditure had a statistically significant coefficient only in the regression of real GDP/capita with all suitable lags of the instruments. Regarding infrastructure development the results confirm the importance of only other types of roads besides highways. The motorway infrastructure variable had a negative coefficient, but it was not in the accepted threshold (1% to 10%).
Concerning nights spend by tourists at regional level, it can be stated that only non-residents had a weak but negative effect on growth. This was highlighted only in column (7) for real GDP in PPS/inhabitant and for the instruments in all suitable lag order. The stock of vehicles is positively linked with growth at NUTS 2 level. Usually the increase in the number of vehicles (especially cars) it’s a sign of economic development.
An interesting result is the fact that population density (proxy for agglomeration) has a big and negative impact on regional economic growth. This is in contrast with the agglomeration economies theory that sees the increase in urban population as a stimulus of economic growth (Rosenthal and Strange 2004; van Oort, de Geus and Dogaru 2015). Net migration was not statistically significant.
The Sargan and Hansen tests were significant for the results in columns (1,) (2), (5) and (6). For the other columns there might be a problem with the relevance of the instruments used. There is no first order serial correlation detected, but second order serial correlation is detected only for the estimation with real GDP in PPS/inhab.
Table 29: The results of the system GMM method for the NUTS 2 regions
Notes: t statistics in parentheses. * p < 0:10, ** p < 0:05, *** p < 0:01. All regressions include time dummies
Source: Stata v14
There are similarities between the results of the GMM estimator for the NUTS 2 regions and the results for the system GMM. The coefficients for the lagged dependent variables are very significant and positive, meaning that there is regional divergence in the European Union. The variable for population is still positive, but significant in only column (8) and with the coefficient much smaller than the one for the GMM estimator. For the variable that measures fertility rate the coefficients are also small. The proxy for the persons that are the most at risk on the labour market is still statistically representative and negative, meaning that EU policy makers have to create measures for this social group.
Tertiary education in the system GMM regression is statistically positive, but only for the equation with real GDP/capita. Furthermore the results for the employment rates are in line with the ones for the GMM estimator. Total employment and male employment are significant and positive and female employment is negatively correlated with growth.
Total intramural research and development expenditure is still negative and also significant in almost all regressions. Motorway development had a small positive outcome on growth, with other type of roads being positive in regression (2) and small and negative in regressions (7)-(8). Nights spend by non-residents are statistically representative and this determinant improves regional economic growth. An interesting result is the fact that the stock of vehicles is in the system GMM negatively linked with regional growth. Population density and net migration are not statistically significant.
An important concern is the fact that the p-values for the Sargan and Hansen tests were small, which can reject the null hypothesis of valid instrument selection. There is no first order serial correlation detected, but second order serial correlation is detected only for the estimation with real GDP in PPS/inhab.
The next step is to compute the QML estimations for the NUTS 1 and NUTS 2, taking into consideration the results provided by the Hausman and Parm tests. Table 30 provides the results for the QML for the NUTS 1 regions. Column (1) and (2) is for the real GDP/capita estimations and columns (3) and (4) for the real GDP in PPS/inhabit. All the regressions use fixed effects with time dummies.
Table 30: The results of the QML estimation for the NUTS 1 regions
Notes: t statistics in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. All regressions include time dummies
Source: Stata v14
The lagged dependent variable is positive, confirming yet again the presence of regional divergence in the EU. Only in column (1) fertility rate has a weak statistically positive effect on regional growth. Life expectancy has an important outcome on growth. This means that healthier citizens contribute to a prosperous society.
Once again, it appears that early leavers from education and training have a negative impact on growth. The same conclusion was found for the GMM and system GMM estimations. Furthermore, average weekly hours worked by male are an important negative determinant of regional growth in the EU.
From Table 30, total employment rate and male employment rate contribute to regional economic growth. Female employment rate was not statistically significant in the quasi-maximum likelihood estimation. In regards to infrastructure development, the conclusion is that motorways measured by km do not have a statistical significance on economic growth. Other road development appears to have a small but statistically significant coefficient. The impact of other roads is negatively related to economic growth.
Regarding the variables for tourism, total nights spent by residents do not have a significant coefficient. Total nights spent by non-residents are positively correlated with regional economic growth in the real GDP/capita equation and are negative in the real GDP in PPS/inhab estimation.
Population, tertiary education, average weekly hours worked by female and the stock of vehicles were not statistically significant in determining regional economic growth in any of the QML estimations. The same can be said for research and development expenditure, even if the coefficients are negative in every column.
The following table provides the outcomes of the QML test for the NUTS 2 regions. Columns (1) and (2) present the results for the real GDP/capita estimations and columns (3) and (4) for the realGDP in PPS/inhab. Column (2) is a random effect estimation and the rest of the columns being fixed effect methods.
Table 31: The results of the QML estimation for the NUTS 2 regions
Notes: t statistics in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. All regressions include time dummies.
Source: Stata v14
The Quasi-maximum likelihood estimation still provides conclusive results for the divergence hypothesis between EU regions. The coefficients are positive and statistically significant. Population appears to be influencing regional growth. Fertility rate increases economic growth when the dependent variable is real GDP/capita and has a negative influence when real GDP in PPS/inhab is used.
Like in the GMM and system GMM, life expectancy has a concrete outcome on regional growth. Life expectancy is used as a proxy for the health level of the population. It makes sense that a healthier and longer life positively impacts the economy.
Like it was stated before in the other regression, early leavers from education and training have a negative influence on growth. This social category is at risk economically and socially. Persons with tertiary education help in fostering economic growth, but the coefficients in Table 31 are small.
Average weekly hours worked by male appear to be negative in the QML estimations with fixed effects and positive in the QML estimation with random effects. In the same QML estimation average weekly hours worked by female is statistically significant and negative.
The analysis of employment rates offers the same conclusion as before: total employments and male employment are beneficiary for the economy and female employment decreases economic growth.
Research and development had a weak statistically significance on economic growth only in the QML-FE for real GDP/capita. This can mean that innovation is not contributing too much to EU regional growth as it was believed to do. Also infrastructure development appears to have small coefficients and none of them are statistically smaller than 10%.
Regarding total nights spent by residents and non-residents in tourist accommodations the results are not conclusive to say that these indicators have a major impact on regional growth. In Table 31 nights spend by residents were not significant to be validated and nights spent by non-residents contributed to growth in the QML-RE estimation and are negatively correlated with growth in the QML-FE estimation for real GDP in PPS/inhab (column 3).
The stock of vehicles at regional level is a variable that is useful for economic growth. From the results obtained for population density it seems that agglomeration is not an important factor at regional level. Finally, net migration is statistically significant in the estimation with real GDP/capita as dependent variable, but the coefficients were very small, implying that migration is not contributing very much to regional development.
From all the empirical estimations the coefficient of the lagged dependent variable of real GDP/capita was positive, implying that there is regional divergence in the EU. The divergence or convergence hypothesis will be graphically illustrated below. To see if there is a difference between the post crisis and after crisis period, the time frame was divided in two. Figure 25 shows that between 2001 and 2008 regional convergence was present, even if the correlation was small and relatively weak.
Figure 25: EU regional average GDP growth vs. initial GDP – 2001-2008
Source: own calculations
From 2007 onwards the divergence between EU regions has accelerated. The financial crisis may have had a big influence, even if it is not well proven by empirical research. The coefficient of determination in Figure 26 is higher, meaning a stronger correlation between the two variables.
Figure 26: EU regional average GDP growth vs. initial GDP – 2008-2013
Source: own calculations
4.5. Conclusions
The purpose of this chapter was to determine what factors influence economic growth at NUTS 1 and NUTS 2 levels in the European Union between 2000 and 2013. To find the role of each determinant of regional economic growth the study used two separate growth equations and as dependent variable the regional real GDP per capita and regional real GDP in purchasing power standard per inhabitant.
The models were applied on dynamic panel data for a number of 98 NUTS 1 regions and 273 NUTS 2 regions for all the EU country states (EU 28). Several panel data estimation techniques were utilised (the first difference GMM estimator and the system GMM estimator and cross-section time-series dynamic panel data estimation by quasi-maximum likelihood), which improved also efficiency. The GMM methodology is very common in panel data analysis for economic growth variation, but the QML estimation is a relatively new variation of the ML (maximum likelihood) approach pioneered by Bhargava and Sargan (1983). This is why the results have to be considered with caution.
Because the QML (quasi-maximum likelihood estimation for dynamic panel data models) is more efficient for this type of analysis, it can be concluded that the results offered by this method should be the most reliable, but we should not ignore the ones produced by the GMM and system GMM methods.
It was found that EU regions are not converging. There was a steady state of convergence between EU regions before 2008. This ended most probably because of the financial crisis of 2008 (Pellegrini et al. 2013; Timbeau 2014).
From the results of the QML estimation for NUTS 2 regions, population appears to be influencing regional growth. The ones for NUTS 1 were not significant at 10%. The outcome for fertility rate offered mixed results. It increases economic growth when the dependent variable is real GDP/capita and has a negative influence when real GDP in PPS/inhab is used.
The results confirm that life expectancy has a concrete impact on regional growth. Life expectancy is used as a proxy for the health level of the population. It makes sense that a healthier and longer life positively impacts the economy.
Early leavers from education and training are a negative influence on regional economic growth. This social category is at risk economically and policy makers have to adopt measures for the better integration of this group in the society and on the labour market.
Persons with tertiary education appear to contribute to regional economic growth, but the coefficients were small and not statistically significant in most of the results.
Regarding the average weekly hours worked by male this chapter comes to the conclusion that it hinders economic development. Also, the variable for average weekly hours worked by female is negative, but mostly not statistically significant.
The investigation into the effects of employment rates offers the following conclusion: total employments and male employment are beneficiary for the economy and female employment decreases economic growth.
Research and development had a negative impact on regional development in almost all of the regressions, even if some of the coefficients were not significant. Also infrastructure development appears to not have a defining role in shaping regional economic growth. Infrastructure endowment is poorly linked to economic growth and the exact returns and implications of this type of investment is not so clear (Crescenzi and Rodríguez-Pose 2012; Rodriguez-Pose, Psycharis and Tselios 2012).
Concerning total nights spent by residents and non-residents in tourist accommodations the results are not conclusive to say that these indicators have a major impact on regional growth.
In general, from this case study’s regressions, the stock of vehicles at regional level is a variable that was positively correlated with growth. Furthermore the results obtained for population density contradict the agglomeration economies theory. It seems that regional agglomeration is not an important factor. This outcome can be attributed to Europe’s high number of small and medium size cities and the negative externalities of living in a big city like congestion cost, labour competitiveness, pollution and high rental costs (Dijkstra et al. 2013).
Finally, net migration is statistically significant only for the QML estimation and for real GDP/capita as the dependent variable. The coefficients were very small, implying that migration is not contributing very much to regional development.
Furthermore the claims of this chapter require further analysis to empirically test the assumptions made. As the QML estimation technique is being improved further analyses have to be conducted. This investigation has considerable policy implications for policymakers. Furthermore, certain economic and political shocks could have had significant implication for this empirical framework, like for example the 2008 economic crisis. Further investigation of these inherent shocks could affect the estimation coefficients and might offer different results.
CHAPTER 5 ARE EUROPEAN METROPOLITAN REGIONS STILL RELEVANT AND WHAT ARE THE DRIVING FORCES OF URBAN ECONOMIC GROWTH?
5.1. Introduction
From the beginning of human recorded history cities were centres of culture, economic wealth, artistic innovation and magnets for the most talented people of the era. From the Phoenician city states (Tyr, Sidon and Byblos), the Greek poleis, Rome, the city was considered the cradle of civilization and sometimes its destruction meant the end of that culture (Carthage was razed to the ground in 146 BC after the Roman defeat).
The notion that cities are a source of economic growth is gaining more and more focus in the recent period. Cities and urban zones are considered to be the fundamental sites for the clustering of economic activity. This is in part because of the new research done by many scholars in the field of new economic geography (agglomeration economies) or the ones involved in the “new growth theory” (Glaeser et al. 1992; Combes 2000; Melo et al. 2009).
Cities are human centres that allow for the exchange of goods, ideas and people and in turn the society reaps the benefits from trade and specialization (Christiaensen and Todo 2013; Glaser et al. 1992; Combes 2000). Cities facilitate all these factors to come together to allow for more production and labour specialization. Towns and cities rose to become market places in which goods and services are transferred faster and more efficiently.
When focusing on Europe, it’s important to state that more that 75% of its citizens are living in urban areas. From this number we can affirm that Europe has the highest density of urban zones in the world. Urbanization is a fast growing trend in the EU even if population growth is small compared with many other regions (Asia, Africa or Latin America). Half of European cities are small with between 50 000 and 100 000 inhabitants and only two can be considered global cities – London and Paris. Smaller cities have more than 40 % of the EU population (OECD 2012).
The focus of this chapter is to contribute to the metropolitan economic growth literature by implementing an analysis for 271 areas located in the European Union. For this endeavour the study uses several empirical methods to quantify and statistically demonstrate the link between the independent variables and real GDP measured in per capita and in PPS per inhabitant.
The study will use several independent variables to check which of them have an influence on metropolitan growth. These are the percentage of agriculture, industry, construction, retail, transportation and accommodation services, manufacturing and information and communication in total metropolitan gross value added, the number of employees, the population size, growth and density and finally net migration.
To investigate the robustness of the results, the empirical model is also estimated by dividing the time period in two parts (post and ante economic crisis) and by splitting the sample of metropolitan regions in two components – the Western more developed regions and the Central and Eastern (the formal communist states, except for Cyprus) metropolitan areas.
The key questions that this study will want to answer are:
What are the most important economic sectors for metropolitan growth?
Does population size, population density or population growth have an effect on metropolitan regions?
Is migration a positive influence on development?
Are metropolitan regions diverging and did the European enlargement substantially influenced growth in these areas?
In order to achieve the results of the empirical investigation the rest of this chapter is structured around six sections. First, this short introduction is followed by the literature review on urban economic growth. Section 5.3 highlights the methodology used and the data sources with some graphical illustrations of some of the variables utilised for the analysis. Section 5.4 presents the findings of the empirical methods used. Section 5.5 conducts some robustness checks. The chapter ends with the conclusions.
5.2. Literature review
At the end of the 19th century Alfred Marshall (1890) argued that urban agglomeration has many benefits for regional and state development. These benefits can be summed up as follows: providing easier goods and services to firms and also to consumers, knowledge spillovers and labour market pooling. The early 1990s saw a revitalization of the urban economics and economic geography literature (Porter 1990; Krugman 1991) and continued into the new millennium, especially by North American researchers (Henderson 2010; Glaeser 2011). These seminal works mostly concentrated on North American cities and some on the ones in the developing countries. Because of the important shifts that are now taking place regarding the importance of cities in driving economic growth this study will try to offer future insight and answer some important questions.
The research literature has found some important aspects regarding the role of cities in shaping economic growth and the different ways by which they can affect the development of countries. It is well known that urban centres are the engines of regional economic growth. States with more dispersed urban centres with medium size population have reduced poverty compared with countries that have a big concentration of population and large cities (Christiaensen and Todo 2013).
Rapid urbanization has occurred extensively after the second half of the 20th century. This process is unprecedented in human history and has manifested more in countries with low per capita income (Cohen 2004). We can attribute the fast pace of urbanization to changes in the economic system and mostly to globalization. In the case of India, Sridhar (2010) pointed out that at the beginning of the 19th century, only Calcutta had a population of more than 1 million inhabitants. This process was intensified after the half of the 20th century in India. In 1991 in India were twenty three cities with a population of over 1 million inhabitants and by 2001 the number rose to 35 cities (38% of the total urban Indian population).
Urban economists have shown that larger cities have high population density because of the increasing competition for capital gains (returns) and labour. Larger cities also have higher productivity and per capita income compared with the smaller ones, but this statement is highly dependent on the political and economic system in the country (Polese 2005). According to Combes (2000) large cities growth more if the infrastructure endowment is better (better schools, roads, hospitals). Au and Henderson (2006) found that because of migratory restriction a big number of Chinese cities are not growing as fast as they should, in turn affecting urban economic growth and income.
The size of a city can be also detrimental to its growth. There are negative outcomes of becoming too large as a city. These are being defined by the literature as “agglomeration diseconomies” (Henderson and Becker 2000). These diseconomies can range from increased crime rates, air pollution (some examples can be the Chinese mega cities), higher costs of living, social inequality or traffic congestion due to too many cars and an infrastructure that is lagging behind.
According to the agglomeration economies theory, there are productivity gains for companies and citizens by the fact that they are clustered in an urban community. For example, companies which are located in an urban zone benefit from the economy of scale (a bigger market size), lower transaction, information and infrastructure costs, a bigger sampling pool for recruitment and more skilled workers or more suppliers to choose from. The human capital accumulation of skilled workers determined the fast growth of Indian cities by making them more attractive for companies (Sridhar 2010).
As stated above agglomeration economies are mostly beneficiary for companies, but urbanization, and especially the formation of large cities brings more competition. It is still argued if higher competition is going to lead to future economic growth for urban areas (Glaser et al. 1992; Usai and Paci 2003; Combes 2000).
The rapid growth of urban centres in India and China in the past years can be attributed to a mix of economic reforms and the extent of the manufacturing sector vis-à-vis to the service sector. The fast city growth in the south of India is linked to the employment surge in the service sector (Paul and Sridhar 2015).
There are many research contributions that focused on the urban economic growth in China and what determines city growth. Between 1991 and 1998 urban economic growth was influence by foreign direct investment, infrastructure endowment and investment in human capital. Population growth and domestic investment had a negative impact on GDP per capita (Lin and Song 2002). Population growth was found to influence real urban GDP growth in 220 Chinese cities, but to negatively influence GDP/capita (Anderson and Ge 2004). Compared to the government sector, the private sector contributed the most to city growth in China. Au and Henderson (2006) stated that in China agglomeration economies (diversified industries and population), the accumulation of capital and foreign direct investment were significant sources of growth.
Urban centres are very important for rural inhabitants in many ways. First of all they provide a market for their products and in turn cities provide for rural inhabitants specialised goods and services. Secondly commuting from rural area to urban areas for employment reasons is a common fact in the modern era. Many people in the developing countries of Eastern Europe, Latin Africa, Africa and Asia increase their rural income by working in the medium and large urban centres (Reardon et al. 2001). Cities can be considered also hubs for fostering cultural, economic and social communication between citizens of the same country or from different corners of the work. This is because usually the infrastructure is more developed in urban areas than in rural ones, offering more connectivity between people.
In accordance with the endogenous growth theory, it is important to state that urban centres are essential for knowledge formation and diffusion. They promote the flow of new ideas and facilitate innovation (McCann 2007). The knowledge diffusion of cities is beneficial for creating spatial externalities and spillovers that can contribute more to regional and state development. Knowledge spillovers are increased if companies in the same industry are geographically proximate.
According to McCann and Acs (2011) productivity rises with the size of the city in the US, Korea and Japan, but in general productivity is more related to growth if the city is better connected with other cities/regions.
Cities are also socially diverse, with inhabitants from different backgrounds, with different religions, norms or habits. This is more common in large urban areas like London, Paris, New York or Beijing. Audretsch et al. (2010) have found that urban social diversity has an important effect on regional economic growth.
Berdegue et.al (2015) confirmed that the presence of a city in a rural-urban region has a positive outcome on economic growth in Columbia and Chile and that it reduces poverty. They found that cities favour territorial development by the diffusion of ideas, the flow of information and knowledge. Furthermore, they provide access to specialized services.
In the case of Brazil, the rise in rural population, the development of inter-regional infrastructure and higher levels of education for the work force has a considerable influence on the growth of a city (Da Mata et al. 2005). The increase in criminality rates has a negative impact on the growth rate of a Brazilian urban area.
Climate also plays an important factor in urban population growth and can affect the economic growth of a city. Urban areas that are less favourable for human inhabitants tend to grow slower that the ones with climate endowed (Haurin 1980).
There are also scholars that, contrary to the literature, consider that cities do not influence growth and that the evidence so far is not conclusive. Polese (2005) considers that cities do not cause income to rise in the long run, but the rise in income is a result of an adjustment process of national economic growth.
5.3. Methodology and data
The aim of this chapter is to evaluate the factors determining urban economic growth at metropolitan level between 2000 and 2013 in the European Union for 271 metropolitan regions. The Directorate General for Regional Policy of the European Commission defines metropolitan regions as NUTS 3 regions or a combination of NUTS 3 regions which represent all agglomerations of at least 250.000 inhabitants. These agglomerations were described using the Urban Audit's Functional Urban Area (FUA). Each metropolitan agglomeration is represented by at least one NUTS 3 region.
The analysis at metropolitan level is important in the context of increasing urbanization in the EU. The study will investigate the role of some important economic sectors like agriculture, industry, manufacturing, construction, service activities and information and communication services in facilitating urban development. Other variables that are examined are the number of employees, population density, population size and growth and net migration. The study will also use a dummy variable to control for the importance of EU enlargement on metropolitan areas.
The study is based on a growth equation with the dependent variable being the metropolitan real GDP per capita or metropolitan real GDP in purchasing power standard per inhabitant. All the monetary values are expressed at consnta market prices and denominated in euros. Metropolitan nominal GDP is deflated using the Eurostat country deflator, with the base year being 2010.
The variables will be transformed using the neglog transformation. Some of the variables in the study are negative (net migration, population growth) and the utilization of normal logarithm will result in data loss. The neglog transformation behaves like ln (z) when z is positive and like – ln (-z) when z is negative (Whittaker et al. 2005). So the investigation will use a logarithm called “L” = sign(z)*ln(|z| + 1, where z is the value of the variable.
The economic growth equation has the following form:
(1)
where:
LY: the neglog of metropolitan real GDP per capita or GDP in PPS standard per inhabitant to see if there are differences between the two indicators of growth. The metropolitan gross domestic product is defined as the market value of all final goods and services produced within a metropolitan area in a given period of time. According to Eurostat expressing gross domestic product in purchasing power standards cancels the differences in price levels between countries. By calculating GDP per inhabitant makes it easier to compare different regions in comparison with a calculation in absolute size.
Lyi,t-1: represents the neglog of one lag metropolitan real GDP per capita or one lag metropolitan real GDP in PPS standard per inhabitant. It is usually introduced in the growth equation to measure the convergence or divergence hypothesis. If the coefficient of this variable is negative then we can state that the EU metropolitan regions are converging or the less developed ones are catching-up to the most developed ones. This indicator is very important for this type of urban analysis.
LGVAagr: represents the neglog of the share of metropolitan gross value added of agriculture, forestry and fishing in total metropolitan gross value added. It represents the contribution that this specific economic activity/sector has on metropolitan economic output.
LGVAind: represents the neglog of the share of metropolitan gross value added of industry in total metropolitan gross value added. It represents the contribution that this specific economic activity/sector has on metropolitan economic output.
LGVAmanuf: represents the neglog of the share of metropolitan gross value added of manufacturing in total metropolitan gross value added. It represents the contribution that this specific economic activity/sector has on metropolitan economic output.
LGVAconst: represents the neglog of the share of metropolitan gross value added of construction in total metropolitan gross value added. It represents the contribution that this specific economic activity/sector has on metropolitan economic output.
LGVAserv: represents the neglog of the share of metropolitan gross value added of wholesale and retail trade, transport, accommodation and food service activities in total metropolitan gross value added. It represents the contribution that this specific economic activity/sector has on metropolitan economic output.
LGVAitc: represents the neglog of the share of metropolitan gross value added of information and communication in total metropolitan gross value added. It represents the contribution that this specific economic activity/sector has on metropolitan economic output.
LEMPL: the neglog of the total number of employees at metropolitan level. This indicator will measure the impact of employed persons on metropolitan economic growth.
LDENSITY: the neglog of metropolitan population density (persons per km2). Population density is the ratio between the annual average population and the land area of the region. This variable is a proxy for regional agglomeration. Usually large and densely populated urban area should have a positive effect on regional economic growth.
LEAP: the neglog of economically active population (inhabitants).
LPOP: the neglog of metropolitan population (inhabitants). It measures the impact of population size on metropolitan economic output;
LPOPgr: the neglog of metropolitan population (inhabitants) growth. It measures the impact of population growth on metropolitan economic output. The study uses the crude rate of population change.
LMIGRATION: the neglog of metropolitan net migration (%). The study uses the crude rate of net migration plus statistical adjustment.
D: represents the dummy variable for European enlargement. This dummy variable will assess if EU enlargement had an impact on the economic growth of metropolitan areas. Because the study analyses all the 28 EU metropolitan areas between 2000 and 2013, some of them were not part of the EU before 2004, 2007 or 2013. The variable will take the value 1 if the metropolitan area was part of the EU and 0 if the metropolitan area was not part of the EU.
η: is the unobserved regional-specific effect;
ε: is the disturbance term;
i is the individual regional dimension and t is the time period dimension.
Data are taken from the Eurostat database, more specifically from the metropolitan regions database. All monetary data are expressed at constant market prices and denominated in a common currency (ECU).
Before starting the empirical investigation it is essential to present some key facts about the metropolitan areas in the European Union. The biggest metropolitan areas by population are Paris and London. In 2013 the population of London was approximately 13.6 million, of which 24% were living in Inner London. Paris had approximately 11.9 million people. The third and fourth places are held by two major metropolitan areas from Spain. Madrid had in 2013 a population of 6.4 million and Barcelona a population of 5.4 million.
Germany has the following places in the list of the biggest metropolitan areas with more than 5 million residents (Rhine area and Berlin). Both metropolitan areas have a population of 5.1 million. Table 32 presents the top ten metropolitan areas in the EU by population size.
Table 32: Top ten metropolitan areas in the EU by population in 2013
Source: Eurostat database
The metropolitan areas in the countries that joined the EU after 2004 have higher population density compared with the old EU countries. This difference occurred because of the strong central planning in the metropolitan areas of the former socialist countries in Eastern Europe.
Compared with the US, Europe has almost twice as many major metropolitan areas, with a total population of 207 million. In comparison the US has 173 million inhabitants living in major metropolitan areas. But large metropolitan areas represent almost 55% of the US population and in Europe they represent only 40%.
Migration from the new members of the EU and also from the other parts of the developing world and the continuing move from rural to urban regions are driving the fast growth of metropolitan zones. For example the metropolitan region of Madrid is the fastest growing in the EU with 1.5% growth annually. Rome is the second highest, fastest growing metropolitan area, followed by Brussels, London, Prague and Valencia, each increasing with more than 1% per year. But there is also stagnation or even a reduction in population for some important metropolitan areas like Essen and Katowice (Upper Silesian area of Poland) and Naples.
Figure 27: Metropolitan GDP per capita and in PPS/inhabitant in 2013
Source: Eurostat database
From Figure 27 it is obvious that there is variation between metropolitan areas in the EU regarding economic development. There is divergence in GDP/capita (GDP in PPS/inhab.) between Western European metropolitan regions and Eastern European ones. Table 33 also presents a top ten and bottom ten list related to urban area GDP/capita and GDP in PPS/inhab. Luxembourg, Oslo and Groningen are in the top 3 urban regions by GDP/capita and GDP in PPS/inhabitant. The only metropolitan region from Central and Eastern Europe that is in the top ten is Bratislava. Furthermore, only one of the two global metropolitan areas (Paris and London) is in the top ten. Paris has a GDP/capita in 2012 of more than 52 210 euro and ranks the ninth in the list.
Regarding the regions in the bottom ten Plovdiv (Bulgaria) is the last one, with a GDP/capita 18 times smaller than that of Luxembourg. The other underperforming urban areas are located in Romania, Bulgaria, Hungary, Poland and Croatia.
Table 33: Top ten and bottom ten metropolitan areas by GDP/capita and GDP PPS/inhab in 2012
Source: Eurostat database
Figure 28: Retail sales per capita, annual % change, 2009-2012
Source: Eurostat database, Moody’s analytics
Consumer spending has fallen in Spanish, Portuguese, Irish, Greek and many Italian metropolitan areas (Figure 28). Also Czech and many French metropolitan areas saw a very small rise in retail sales between 2009 and 2012. Consumer spending was highest in Romanian, Polish, Bulgarian and the Baltic metropolitan regions.
According to Figure 29 and Table 34 the highest crude rate of net migration was in Luxembourg, followed by two cities in Italy, specifically Florence and Bologna. The list is completed with other metropolitan areas from Germany, Italy and France. The urban areas that registered the highest negative migration in 2012 were Thessaloniki, Barcelona and Coimbra.
Figure 29: Metropolitan crude rate of net migration plus statistical adjustment in 2012
Source: Eurostat database
Table 34: Top ten and bottom ten metropolitan areas crude rate of net migration plus adjustment in 2012
Source: Eurostat database
Gross value added from the sector of agriculture, forestry and fishing contributes differently to economic growth across metropolitan regions in the EU. The most GVA produced by this sector was in Reims followed by another region of France, Bordeaux. The contribution of GVA from agriculture, forestry and fishing was insignificant for four metropolitan regions of south and central England and for the German region of Wuppertal.
Table 35: Top five and bottom five metropolitan areas by gross value added from agriculture, forestry and fishing in 2012
Source: Eurostat database
Table 36 highlights the top 5 regions with the most GVA generated from industry (except construction). The total value of goods and services produced by the industry sector in the metropolitan area of Paris adds up to almost 50.000 million euro (approximately 50 billion), but the share of this sector in total GVA is roughly 9%. The urban areas of Stuttgart and Ruhrgebiet are also high producers of industrial goods and services, with the share in total GVA being higher. The top five list is completed by London and Milan. The smallest GVA generated from industry is obtained in the metropolitan region of Slit (Croatia), followed by Pecs (Hungary).
Table 36: Top five and bottom five metropolitan areas by gross value added from industry in 2012
Source: Eurostat database
The metropolitan region of Paris is again the most important region regarding GVA obtained by the manufacturing sector. In the top five we also find 3 German metropolitan areas – Stuttgart, Munchen and Ruhrgebiet and the Italian region of Milan. The smallest value of goods and services produced by the manufacturing sector is attained by the metropolitan region of Brighton and Hove. A smaller value of manufacturing GVA is obtained also in Split (Croatia), Varna (Bulgaria), Pecs (Hungary) and Perpignan (France).
Table 37: Top five and bottom five metropolitan areas by gross value added from manufacturing in 2012
Source: Eurostat database
An important sector of the economy is construction. Because of the financial crisis, many construction projects were closed or abandoned and many companies had to go bankrupt. The highest GVA from the construction sector is obtained by the metropolitan region of London. Paris comes second with more than 25 billion euro. Madrid, Milan and Barcelona occupy the third, fourth and fifth places, but at considerable distance from London. The smallest GVA from construction is attained by four metropolitan areas of Hungary and by Plovdiv (Bulgaria).
Table 38: Top five and bottom five metropolitan areas by gross value added from construction in 2012
Source: Eurostat database
Yet again the Paris metropolitan region is first in Europe regarding GVA obtained from wholesale and retail trade, transport, accommodation and food service activities. The share of this sector in total GVA is more than 18%. London is second in the list with more than 92 billion euro. The ranking is completed by Madrid, Barcelona and Milan. The urban regions that are underachieving in regards to GVA from wholesale and retail trade, transport, accommodation and food service activities are from Hungary and Romania.
Table 39: Top five and bottom five metropolitan areas by gross value added from wholesale and retail trade, transport, accommodation and food service activities in 2012
Source: Eurostat database
Another essential sector of the economy is information and communication. The Paris metropolitan area had the highest value of goods and services produced by the ITC sector in the EU adding up to more than 56 billion euro. At a small distance we find the metropolitan area of London. Milan, Rome and Dublin are in the top five ranking in regards to ICT, but at a major distance between them and Paris. The underperforming urban regions are found in Romania, Hungary and Bulgaria. For example the total gross value added produced by the ICT sector in Varna adds to almost 35 million euro.
Table 40: Top five and bottom five metropolitan areas by gross value added from Information and communication in 2012
Source: Eurostat database
As stated in the beginning of the chapter, the aim of this investigation is to determine what variables influence metropolitan economic growth. The analysis is focused on 271 metropolitan regions of the European Union within a time frame of 14 years (2000-2013).
The tables containing all the names of the 271 metropolitan regions, the variables used in the investigation and the data source are presented in Appendix III.
The next part of the analysis offers a summary statistics of the variables used and the correlation matrix.
Table 41 of this chapter highlights the summary statistics of the variables used, starting with the number of observations, the mean, standard deviation and the minimum and maximum. Because the time range is between 2000 and 2013 and there are 271 metropolitan regions used in this investigation, the maximum number of observations is 3794.
Table 41: Summary statistics of the variables used
Source: Stata v14
Table 42 presents the correlation matrix between the dependent variables – real GDP per capita and real GDP in PPS per inhabitant and the independent variables. The correlation matrix constitutes a first crude glance of the relationship between these variables. Metropolitan real GDP per capita and real GDP in PPS per inhabitant are negatively linked with GVA from agriculture, industry, manufacturing, construction and wholesale. Metropolitan real GDP per capita and real GDP in PPS per inhabitant are positively linked with GVA from ITC, the number of employees, with population density, size and growth, with economically active population and net migration.
Table 42: The correlation matrix of the variables used
Source: Stata v14
The correlation matrix graphs are illustrated in Appendix III for the variables used in table 41.
5.4. Empirical results
For analysing the influence of the independent variables presented in the methodology, the study will use several panel data estimation techniques. The panel data techniques used are the first difference GMM estimator and the system GMM estimator and the quasi-maximum likelihood estimation. The study uses a linear dynamic panel data. The QML estimation was developed by Kripfganz (2016). The ML (maximum likelihood) approach was pioneered by Bhargava and Sargan (1983), further developed by Hsiao, Pesaran and Tahmiscioglu (2002) and is suited also for panel data with missing values. Missingness can be solved by implementing a ML estimation or a multiple imputation technique.
Quasi-maximum likelihood estimation does not use any instruments like the GMM or system GMM methods. Also the weak instruments that may be used in the GMM and SysGMM are avoided in the QML estimation. The estimators in a QML technique are extended to accommodate for unbalance panel data, like in the present investigation related to metropolitan economic growth.
Before applying the regression models it is important to make some preliminary investigations. Some of the variables may be nonstationary. The regressions that involve independent nonstationary variables can generate “spurious” results (Ghosh 2012). For testing the stationary hypothesis the investigation applies the Fisher-type (1925) unit-root test which is suited for this panel data. The results are presented in Table 43.
Table 43: Unit-root test results for the variables used
Legend: *, **, *** denote significance at 1%, 5%, and 10%, respectively
Source: Stata v14
Table 43 details the results of the Fisher test. It confirms the presence of a unit root for several variables. The test is also conducted for the first differenced variables. By doing so, the nonstationary variables become stationary in first difference. The model will be rewritten with all the variables in first difference except for the dummy one. The metropolitan specific effects are removed when the first difference is implemented.
The economic growth equation with the differenced variables is as follows:
(2)
The study will compute the Hausman test to determine if the quasi-maximum likelihood will be a fixed effects or a random effects method. The results confirm that the study should use a quasi-maximum likelihood method with fixed effects.
Table 44: Hausman test for the QML method
Source: Stata v14
To eliminate the most common sources of cross-sectional dependence, the investigation will include also time dummies. To see if time fixed effects are needed the Parm test will be computed. The results of the Parm test from Table 45 confirm the null hypothesis of the importance of time fixed effects.
Table 45: The Parm test
Source: Stata v14
According to the economic growth literature, for the GMM and system GMM the lagged values of the dependent variable (real GDP/capita and real GDP in PPS/inhab.) and the variables that are weakly exogenous are used for GMM style instruments. The six variables that measure the shares of different economic sectors in the total metropolitan gross value added, population density and population size were used as GMM style instruments with lag two and also with all available lags. The rest of the regressors were used for the IV style instruments. The investigation will also introduce time dummy variables in the models. To limit the number of instruments (as a rule the number of panel data units should be higher than the number of instruments), the collapse option is used in Stata. The first to the fifth lag of the dependent variables are used in the GMM style instruments. The analysis will also include the “robust“ command to control for heteroskedasticity and autocorrelation within panels.
The two below tables (Table 46 and 47) present the results for the GMM and system GMM estimations. In Table 46, columns (1) and (2) highlight the results for the model with real GDP/capita as dependent variable, whereas columns (3) and (4) focus on the results for the model with real GDP in PPS/inhab as dependent variable.
Table 46: The results of the GMM estimator
Notes: t statistics in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. All regressions include time dummies
Source: Stata v14
As seen from the above table the coefficients of the lagged dependent variable (real per capita and real GDP in per inhabitant) are positive, with a significance level of 1%. This suggests that metropolitan regions are not converging to the steady state of growth. This is true when we consider the different levels of development among metropolitan areas and the gap between Western regions and Eastern regions. For example, according to Eurostat, the only metropolitan region from Central-Eastern Europe that is in the top ten list regarding in purchasing power per inhabitant in the year 2012 is Bratislava. In this regard underperforming urban areas are located in Romania, Bulgaria, Hungary, Poland and Croatia. The Plovidv metropolitan area of Bulgaria has a /capita 18 times smaller than that of Luxembourg. This paints a negative picture regarding the measures taken by the EU to limit the gaps between regions and it seems that the process of integration is difficult.
Concentrating now on determining what economic sectors are important for EU metropolitan growth, we can see different results from Table 46. The sector of agriculture, forestry and fishing appears to have a small but negative effect on metropolitan growth. This result was obtained when the GMM methodology used all the available lags of the IV instruments. Two important sectors that are driving metropolitan growth in the EU are industry and construction. The fact that industry has such an important contribution to metropolitan growth is not startling when we consider that most of the countries in the EU are very industrialized. A 1 % rise in the share of industry in total GVA will determine the metropolitan real GDP to rise between 0.2% and 2%.
Even if the construction sector was severely hit by the crisis that started in 2008 this sector is still an important and one that contributes to metropolitan development. The same statement cannot be said about the manufacturing sector. From the results obtained by the GMM estimation the share of manufacturing has a big negative effect on metropolitan growth. An interesting outcome was the fact that neither the wholesale, retail trade, transport, accommodation and food service activities sector nor the ITC sector had a statistically significant coefficient even if they were positive. It means that these two sectors are not contributing to metropolitan economic growth.
As expected, the number of employees positively influences economic growth. If the number of employed persons rises by 1%, metropolitan real GDP rises by almost 0.25-0.29%. Economically active population size has a small negative influence on growth, but is statistically significant only in column (1).
European enlargement appears to have contributed to metropolitan development, but the coefficients are not too considerable (0.0413 and 0.0313). Population density, population size, population growth and net migration did not have any statistical significance in Table 46.
The Arellano-Bond test has detected first order serial correlation for the estimation in the first column. Second order serial correlation has also been detected for the GMM estimation. The p-values for the Sargan and Hansen tests validate the use of the instruments only for the results in columns (1) and (3). The use of instruments with all the lags included appears to be weak.
Table 47: The results of the system GMM estimator
Notes: t statistics in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. All regressions include time dummies
Source: Stata v14
In Table 47 columns (1) and (2) highlight the results for the model with real GDP/capita as dependent variable, whereas columns (3) and (4) focus on the results for the model with real GDP in PPS/inhab as dependent variable. The coefficients of the lag dependent variables are positive, implying that there is a gap between metropolitan regions regarding development.
Compared with the results of the GMM estimator, the only sectors that had statistically significant coefficients were agriculture, forestry and fishing and information and communication. Both of them had small but negative coefficients which imply that these two sectors do not contribute to metropolitan economic growth.
The number of employees is positively influencing metropolitan growth. If the number of employed persons rises by 1%, metropolitan real GDP rises by almost 0.17-0.33%. Economically active population size had statistically insignificant coefficients. It is interesting to see that population density has a small (less than 0.05%) negative influence on metropolitan development. Population density is used as a proxy for agglomeration. According to Puga (2002) high agglomeration in capital cities and large urban areas can have an influence on growth increasing labour specialization and productivity. van Oort, de Geus and Dogaru (2015) showed that agglomeration plays an important role for 15 EU countries at regional level, specifically for 205 EU NUTS2 regions. The results obtained by the system GMM estimator are in contrast with the agglomeration economies theory that sees the increase in urban population as a stimulus of economic growth (Rosenthal and Strange 2004; van Oort, de Geus and Dogaru 2015).
Population size measured by the number of inhabitants has a significant negative effect on metropolitan growth, which is in correlation with the results obtained for population density. An interesting fact is that the coefficients for population growth were positive, but the overall impact is very small which implies that density and size is more important than the growth of the population.
Net migration appears to be negatively influencing metropolitan economic growth. These results are in correlation with the ones obtained in the fourth chapter of the thesis where the coefficients for net migration were also very small and most of them not statistically significant.
European enlargement appears to have contributed to metropolitan development, but only one coefficient was accepted since it was in the confidence interval.
The Arellano-Bond test has detected second order serial correlation for the estimation in the first column. Second order serial correlation has also been detected for the GMM estimation. The p-values for the Sargan and Hansen tests imply that the instruments used might be weak.
The next method applied in this study is the Quasi-maximum likelihood (QML). Compared with the GMM methods, the QML estimation does not use instruments which can bypass many problems identified by Roodman (2009) like for example instrumental selection. The QML estimators can also raise efficiency. The paper will use quasi-maximum likelihood with fixed effects and time dummies. The results are presented in Table 48.
Table 48: The results of the QML-FE method
Notes: Standard errors in parentheses, * p < 0:10, ** p < 0:05, *** p < 0:01. All regressions include time dummies
Source: Stata v14
The coefficients of the lagged dependent variables are again positive, implying that there is divergence between metropolitan regions in the EU. The share of agriculture, forestry and fishing has a small negative impact on growth at a confidence interval of 95%. It seems that raising the share of this sector in the EU would not be so beneficial for metropolitan development. The industrial and construction sectors are two domains that add value to the EU economy, but compared with the results of the GMM estimator, the coefficients are smaller. By increasing the share of industry, for example by 1% the metropolitan real GDP will rise by approximately 0.1%. The manufacturing sector was not statistically significant which is in line with the results obtained so far from the other two methods.
Compared with the GMM and system GMM results for the sector of wholesale and retail trade, transport, accommodation and food service activities, the coefficients of the QML-FE are in the confidence intervals of 95% and 99%. The results suggest that this sector has a bigger impact on metropolitan growth than the other sectors of the economy.
Even if the EU has put a lot of emphasis on the importance of investing in the field of information technology, it appears that this sector has not a big impact especially on metropolitan regions.
Yet again we see that the number of employees is positively influencing metropolitan growth. If the number of employed persons rises by 1%, metropolitan real GDP rises by almost 0.18-0.22%.
Economically active population size had statistically insignificant coefficients, with only one being in the confidence interval. Population density, the proxy for agglomeration, has in the case of the QML-FE estimation a positive coefficient of 0.08, but still the effect is not so considerable. Yet again population size has a negative coefficient. A considerable increase in population size can be associated with a rise in public expenditure (child care and other contributions) that can put a strain on the economy. Population growth has a very small effect on metropolitan economic growth and the significance level is only at 5%.
Net migration is not statistically significant. European enlargement appears to have contributed to metropolitan development, but the coefficients are small. This can be a concern for the EU authorities in light of rising euroscepticism and the 2016 British referendum for the Brexit.
Robustness check
In the final part of the investigation, to gain some robustness, the time period is split in two parts and also the list of metropolitan areas is divided so as to measure the difference between the Western part of the continent and the Central and Eastern part of Europe.
The first robustness check will analyse the econometric model by dividing the period in two samples, one being the period from 2000 to 2007 and the other the period after the economic crisis, from 2008 to 2013. By doing so, the study will determine if the period before the crisis was totally different compared with the period after the crisis. This methodology will be conducted only for the QML-FE estimation.
In Table 49 columns (1) and (2) highlight the results for the QML-FE estimation with real GDP/capita as dependent variable, whereas columns (3) and (4) present the results for the estimation with real GDP in PPS/inhab as dependent variable.
Table 49: The results of the QML-FE estimation
with the time period divided in ante and post economic crisis
Notes: Standard errors in parentheses, * p < 0:10, ** p < 0:05, *** p < 0:01. All regressions include time dummies
Source: Stata v14
The results in Table 49 suggest that there is still divergence between the 271 metropolitan areas used in this study, but the interesting point is that the coefficients for the 2000-2007 sample period were much bigger than the ones for 2008-2013. This implies that the economic crisis has smoothened the gap between the metropolitan regions analysed in this study.
Regarding the impact of different economic sectors on metropolitan growth, from Table 49 we can see that agriculture, forestry and fishing had a significant coefficient only in the after crisis period, with a negative value. The coefficients for the industrial sector were negative before the crisis, but after 2008 it seems that this sector has a positive correlation with metropolitan economic growth. Concerning the role of manufacturing in boosting metropolitan development, the results show that only one coefficient was statistically significant and quite small. This suggests that this sector is not detrimental in influencing regional development.
As to the importance of construction, the results of Table 49 imply that this sector is among the driving factors that play a role in metropolitan development, but its impact is not so substantial because of the small coefficients.
According to the results in the above mentioned table, the sector of wholesale, retail trade, transport, accommodation and food service activities and the one for information and communication were influencing metropolitan economic growth only in the ante crisis period. Their statistical coefficients were small and the significance level was only at 5% or 10%.
Same as in section 5.4 the number of employees is an influential determinant of metropolitan economic growth. Economically active population size had negative coefficients before the economic crisis, but after the crisis the values are positive. This implies that the work force is more important in stimulating growth in periods of turmoil. Population density (proxy for agglomeration) has a considerable effect on metropolitan real GDP/capita after the 2008 and slightly smaller coefficients before the economic crisis. Population size seems to put pressure on metropolitan development if we consider the big impact it had after the 2008. Population growth and net migration were not statistically significant. Also the dummy variable that measures the influence of EU enlargement has statistically significant coefficients, but quite small. The ones for the subsample period 2008-2013 are blank because the data for Croatian regions is missing. The real GDP for the year 2013 for Croatian metropolitan regions was not available on Eurostat, which was the year when Croatia entered the EU.
The second robustness check conducted in this study involves the division of the sample data into Western metropolitan areas and Central-Eastern metropolitan area. What do Western and Central-Eastern mean in this case? Western metropolitan regions are the areas of the sample data from the following countries: Belgium, Denmark, Germany, Ireland, Greece, Spain, France, Italy, Luxembourg, Malta, Netherlands, Austria, Portugal, Finland, Sweden and United Kingdom. Central-Eastern metropolitan areas are the regions from the following countries: Bulgaria, Czech Republic, Estonia, Croatia, Cyprus, Latvia, Lithuania, Hungary, Poland, Romania, Slovenia and Slovakia.
In the below table, columns (1) and (2) highlight the results for the QML-FE estimation for the Western metropolitan areas, whereas columns (3) and (4) present the results for the Central-Eastern ones. Columns (1) and (3) have real GDP/capita as the dependent variable. Columns (2) and (4) have real GDP in PPS/inhabitant as the dependent variable.
Table 50: The results of the QML-FE estimation
for Western European and Central-Eastern Europe
Notes: Standard errors in parentheses, * p < 0:10, ** p < 0:05, *** p < 0:01. All regressions include time dummies
Source: Stata v14
The results in Table 50 show that there is divergence between metropolitan regions in Western and also in the Central-Eastern Europe. The coefficients were statistically significant for both the samples. Regarding the impact that certain economic sectors have on metropolitan growth, the study demonstrates that for Western regions, agriculture, industry, manufacturing and construction are the most important determinants. Wholesale and retail trade, transport, accommodation and food service activities and information and communication sectors were not statistically significant. Agriculture and manufacturing had a negative impact on metropolitan growth, whereas the industrial and construction sectors had a positive one.
The only economic sector that influences metropolitan development in Central-Eastern Europe is construction. The other branches of the economy were not statistically significant.
The number of employees has a determinant impact on metropolitan growth. Moreover the coefficients for the Central and Eastern regions were much higher. Population density and the number of economically active population appear to be statistically significant only for the Central-Eastern metropolitan regions. The coefficients for these variables were negative. Net migration, population size and growth were not statistically significant. The results for the dummy variable show that enlargement did not have a positive impact on metropolitan growth.
Conclusions
The principal goal of this chapter is to contribute to the metropolitan economic growth literature by implementing an analysis for 271 areas located in the European Union. For this endeavour the study uses several empirical methods to quantify and statistically demonstrate the link between the independent variables and real GDP measured in per capita and in PPS per inhabitant and to answer some important questions.
The key questions that this study will want to answer are:
What are the most important economic sectors for metropolitan growth?
Does population size, population density or population growth have an effect on metropolitan regions?
Is migration a positive influence on development?
Are metropolitan regions diverging and did the European enlargement substantially influenced growth in these areas?
For investigating the robustness of the results, the empirical model is also estimated by dividing the time period in two parts (post and ante economic crisis) and by splitting the sample of metropolitan regions in two components – the Western more developed regions and the Central and Eastern (the formal communist states, except for Cyprus) metropolitan areas.
The results of this chapter clearly show that metropolitan regions are not converging to the steady state of growth. There are considerable differences in development among metropolitan areas and there is a visible gap between Western regions and Eastern regions. For example the only metropolitan region from Central-Eastern Europe that is in the top ten list regarding in purchasing power per inhabitant in the year 2012 is Bratislava. In this regard underperforming urban areas are located in Romania, Bulgaria, Hungary, Poland and Croatia. The Plovidv metropolitan area of Bulgaria has a /capita 18 times smaller than that of Luxembourg. This paints a negative picture regarding the measures taken by the EU to limit the gaps between regions and it seems that the process of integration is difficult.
The main findings of this chapter regarding the influences of economic sectors on metropolitan growth are that agriculture, forestry and fishing can have a negative impact on economic growth. A considerable portion of EU funds is employed for stimulating investment in agricultural production and the big countries are also subsidizing this sector so as to be more competitive. The results of this chapter suggest that these allocations appear to not be efficient for metropolitan growth.
Industry, construction and wholesale and retail trade, transport, accommodation and food service activities are positively related to metropolitan growth. It is true that in the system GMM the coefficients were not significant, but for the GMM estimator and QML-FE estimation the values are significant. The manufacturing and information and communication sectors were, in general, statistically insignificant. These findings have considerable policy implications for policymakers in the sense that EU funds must stimulate mostly the economic branches with the most value added to the economy.
The number of employees positively influences EU metropolitan economic growth. If the number of employed persons rises by 1%, metropolitan real GDP rises by almost 0.25-0.29%. European enlargement appears to have contributed to metropolitan development, but the coefficients are not too considerable.
The results also show that population density has a small influence on metropolitan development. Population density is used as a proxy for agglomeration. According to Puga (2002) high agglomeration in capital cities and large urban areas can have an influence on growth increasing labour specialization and productivity. van Oort, de Geus and Dogaru (2015) showed that agglomeration plays an important role for 15 EU countries at regional level, specifically for 205 EU NUTS2 regions. The results obtained by the system GMM estimator are in contrast with the agglomeration economies theory that sees the increase in urban population as a stimulus of economic growth (Rosenthal and Strange 2004; van Oort, de Geus and Dogaru 2015). Population size measured by the number of inhabitants has a significant negative effect on metropolitan growth and the coefficients for population growth were positive, but the overall impact is very small which implies that density and size is more important that the growth of the population.
Net migration appears to be negatively influencing metropolitan economic growth when using the System GMM method and is not statistically significant for the other two techniques.
The robustness check also offered considerable outcomes. First of all it showed that the agriculture, forestry and fishing sector had a significant negative coefficient only in the after crisis period (2008-2013). The coefficients for the industrial sector were negative before the crisis, but after 2008 this sector has a positive correlation with metropolitan economic growth. Concerning the role of manufacturing in boosting metropolitan development, the results show that only one coefficient was statistically significant and quite small. This suggests that this sector is not detrimental in influencing regional development.
The construction sector is among the driving factors that play a role in metropolitan development, but its impact is not so substantial because of the small coefficients. The sector of wholesale, retail trade, transport, accommodation and food service activities and the one for information and communication were influencing metropolitan economic growth only in the ante crisis period.
Secondly the robustness check showed that for Western European regions agriculture, industry, manufacturing and construction are the most important determinants. Wholesale and retail trade, transport, accommodation and food service activities and information and communication sectors were not statistically significant. Agriculture and manufacturing had a negative impact on metropolitan growth, whereas the industrial and construction sectors had a positive one. The only economic sector that influences metropolitan development in Central-Eastern Europe is construction. Population density and the number of economically active population appear to be statistically significant only for the Central-Eastern metropolitan regions. European enlargement did not have a substantial positive impact on metropolitan growth for the Central-Eastern regions.
The results of this chapter must be interpreted with caution because of the inherent endogeneity and the omitted variable biases. However, the investigation undertook some important steps to limit these particularities within this empirical framework. The study used three different panel data estimation techniques which offered more robustness of the results. This offered conclusive results regarding the influences of agriculture, forestry and fishery, industry, construction and wholesale and retail trade, transport, accommodation and food service activities on metropolitan economic growth in the European Union. By splitting the time period and the panel sample the results were more conclusive and showed that certain economic shocks (like the 2008 crisis) have significant impact on the empirical outcomes. On the other hand further investigation has to be done regarding the QML methodology like is with all new panel data estimation techniques. This study has considerable implications for policy makers so as to limit the gaps at metropolitan level.
CONCLUSION
The importance of determining what factors are influencing economic growth is very important for the society as a whole. A lot of emphasis was put on the subject by many researchers starting with the classical economists (Smith, Ricardo, etc.) so as to determine the proper ways in which the economic system may be enhanced.
The determinants of growth are interrelated factors that influence the growth rate of an economy, in a direct way by increasing the real GDP of an economy. Public and private factors have different outcomes on economic growth. Public expenditure, inflation, capital formation, private investment, employment rates, FDI etc. have different consequences on economic growth. There are also socio-political factors and events that have a major influence on the economic advancement of a nation.
The literature made a differentiation between economic and non-economic factors and their implications on state or regional/territorial development. “Proximate” or economic sources are referring to variables like capital accumulation, technological progress, labour and “ultimate” or non-economic sources are referring to determinants like government efficiency, institutions, political and administrative systems, cultural and social factors, geography and demography (Boldeanu and Constantinescu 2015). Research studies analysed the impact on economic growth of such determinants like investment, human capital, research and development, economic and fiscal policies, trade openness, FDI, institutional and political framework, socio-cultural factors, geography and demography.
Because the theory of economic growth is not a unified one, the empirical research is multi-theoretical based and there are contradictions in the results and findings of different authors regarding the determinants of growth. The thesis pursued to fill the gaps in the literature by utilizing up to date empirical tools. A major purpose of the thesis was to analysis the economic and non-economic factors at three distinct levels, specifically at country, regional and metropolitan level in the European Union. This offered more comprehensive results and better empirical robustness. Utilising different econometric techniques and by splitting the panel sample, the investigation also tried to determine if certain economic and political events (the European enlargement, the 2008 economic crisis) had any effect on the results.
The thesis used empirical investigations by which the main results were obtained showing what the positive or negative determinants of economic growth are. The study employed econometric models to evaluate the effects of the public and private factors on economic growth in the EU for different territorial levels (country, regional – NUTS and metropolitan). New panel data estimation techniques were used to cope with the problem of missingness. The data was collected from credible sources like the World Bank’s Statistical Database, the European Commission’s statistical database Eurostat, the Annual Macroeconomic database of the European Commission AMECO, which process the information gathered from state and private institution.
The purpose of the thesis was to determine which factors can improve economic development in the European Union and which have a negative or even an insignificant effect on economic growth. Utilizing dynamic panel data models the study also investigated the convergence hypothesis. For this endeavour the study conducted a thorough investigation divided into three separate empirical chapters as follows: The aim of the first empirical chapter (chapter 3) was to establish the main determinants of economic growth for the 28 European Union countries for the period from 1990 to 2014. In order to investigate which factors are relevant for the country sample this chapter utilized a dynamic panel data model with a number of 30 independent variables. Panel data models are widely used by many researchers in the field of economic growth. The main advantages of using panel data for economic growth models are that it controls for unobserved heterogeneity, it improves efficiency of estimation compared with cross-section models. Panel data also control for omitted variables and measurement errors. To offer robustness several panel data techniques were used, specifically the pooled OLS, the random effects model, FGLS and FMOLS estimation. Finally the Generalized method of moments (GMM) and system GMM estimations were used because it increased efficiency over the other methods. By utilizing the World Bank’s governance indicators, the study also measured the influence of certain non-economic variables like political stability, the control of corruption, absence of violence/terrorism, the rule of law, etc. Based on the findings of the third chapter, the following question was if some of these determinants have the same impact also at territorial level. This was addressed in Chapter 4 of the thesis for 2 separate territorial levels, namely for the NUTS 1 and NUTS 2 regions of the European Union for a time period from 2000 to 2013. This chapter also applied panel data techniques, specifically the GMM and system GMM estimations and the quasi-maximum likelihood (QML) estimation, using two separate growth equations and as dependent variable the regional real GDP per capita and regional real GDP in purchasing power standard per inhabitant.. The QML estimation does not use any instruments like the GMM or system GMM methods. Also the weak instruments that may be used in the GMM and SysGMM are avoided in QML estimations. This chapter also graphically illustrated the fact that regional divergence is a common occurrence after the 2008 economic crisis. Chapter 5 continued the investigation by applying an analysis at metropolitan level and tried to answer some key question. It showed that industry, construction and wholesale and retail trade are the economic sectors that are positively related to metropolitan growth and that agriculture, forestry and fishery hinders it. It showed that population density and size is more important that the growth of the population. Net migration and European enlargement did not have a big impact on metropolitan growth. By dividing the panel sample the chapter demonstrated the impact that a certain economic event like the 2008 crisis has on growth and also that there are still big differences between Western European and Central and Eastern European regions. This chapter applied several methods (GMM, system GMM and QML estimation) on an unbalanced and dynamic panel data model for 271 metropolitan regions in the EU over the period from 2000 to 2013.
The three empirical chapters mentioned above were preceded by an introduction of the thesis and two theoretical chapters. The first empirical chapter analysed the main determinants of economic growth for the 28 European Union countries and if the convergence hypothesis if confirmed. The second empirical chapter highlighted the most important factors that determine growth for two separate territorial levels in the EU, namely the NUTS 1 and NUTS 2 areas and graphically illustrated the much debated convergence hypothesis. Finally, Chapter 5 investigated which economic sectors have an influence on metropolitan growth and if population, employment, migration and the EU enlargement had a definitive impact on development for the panel sample.
Summary of the findings
The summary of the findings of the thesis is presented below:
1 – The results of the third chapter, based on panel data analysis, have shown that among EU countries there was still regional divergence. The results of the pooled OLS and REM methods provided evidence that final energy consumption, employment rate, real labour productivity per hour worked, tertiary education and inflation had a positive effect on economic growth in the EU 28. Real labour productivity had the biggest influence on real GDP/capita. The variables that had a negative impact on growth were general government gross debt and population. For example a rise in population by 1% determines a drop in real GDP/capita growth of -1.15%. Also rule of law had an important effect on economic growth in EU28, with a negative effect from government effectiveness and regulatory quality. The regional dummies also offered interesting results. Western European countries grew the slowest compared with the ones in the rest of the EU 28. Also Southern and Northern Europe did not perform as expected.
The FGLS estimation determined that final energy consumption, the employment rate, real labour productivity per hour worked, gross fixed capital formation, upper secondary and post-secondary non-tertiary education, tertiary education and inflation had a positive effect on growth. The explanatory variables which had a negative impact were general government gross debt and population size. The FMOLS regression confirmed the above results and also showed that life expectancy, FDI and rule of law had a positive effect on economic growth. The interesting fact was that exports, government effectiveness and regulatory quality did not offer the expected outcomes.
The results of the GMM estimator provided evidence of the positive influence of final energy consumption, employment rate, trade openness, real labour productivity per hour worked, FDI, tertiary education and inflation. Trade openness and real labour productivity were the factors that had the most influence on real GDP/capita in the EU28. The negative coefficients were for government debt, exports and population. The dummy variables that were significant were government effectiveness, regulatory quality, control of corruption and the rule of law.
The results of the system GMM confirmed that final energy consumption, deficit, trade openness, real labour productivity per hour worked had a positive and significant influence on economic growth. Trade openness was the factor that had the most influence on real GDP/capita in the EU28. Trade openness can have an influence on economic growth through a multitude of different channels like technological transfers, the increase in economies of scale, competitive advantage (Chang et al. 2009). The variables that had a negative and significant influence on growth were financial sector leverage, total general government expenditure, exports, private debt, gross fixed capital formation and population.
2 – The results of Chapter Four provided compelling information regarding the main determinants at regional level in the EU. The chapter proved that for the 98 NUTS 1 and 273 NUTS 2 areas analysed there was regional divergence. There was a steady state of convergence between EU regions before 2008. From the results of the QML estimation for NUTS 2 regions, population size appeared to be influencing regional growth. The ones for NUTS 1 were not significant at 10%. The outcome for fertility rate offered mixed results. It increased economic growth when the dependent variable was real GDP/capita and had a negative influence when real GDP in PPS/inhab was used. The results provided evidence of the importance of life expectancy. It is used as a proxy for the health level of the population. It makes sense that a healthier and longer life positively impacts the economy. Early leavers from education and training had a negative influence on regional economic growth. Tertiary education appeared to contribute to regional economic growth, but the coefficients were small and not statistically significant in most of the results. Average weekly hours worked by male hindered economic development and the variable for average weekly hours worked by female was negative but mostly not statistically significant. The investigation into the effects of employment rates offered the following conclusion: total employments and male employment were beneficiary for the economy and female employment decreased economic growth. Research and development had a negative impact on regional development in almost all of the regressions, even if some of the coefficients were not significant. Also infrastructure development appeared to not have a defining role in shaping regional economic growth. For total nights spent by residents and non-residents in tourist accommodations the results were not conclusive to say that these indicators had a major impact on regional growth. In general, from this case study, the stock of vehicles at the regional level was a variable that was positively correlated with growth. Furthermore the results obtained for population density contradict the agglomeration economies theory. It seems that regional agglomeration is not an important factor. This outcome can be attributed to Europe’s high number of small and medium size cities and the negative externalities of living in a big city like congestion cost, labour competitiveness, pollution and high rental costs (Dijkstra et al. 2013). Finally the chapter suggests that migration was not contributing very much too regional development.
3 – The main results of Chapter Five demonstrated that agriculture, forestry and fishing had a negative impact for the 271 EU metropolitan areas studied. Industry, construction and wholesale and retail trade, transport, accommodation and food service activities were positively related to metropolitan growth. The manufacturing and information and communication sectors were, in general, statistically insignificant and didn’t contribute to economic growth. The number of employees was positively linked with EU metropolitan growth. European enlargement appeared to have contributed to metropolitan development, but the coefficients were small. The results of this chapter demonstrated that population density had a small influence on metropolitan development. The results obtained were in contrast with the agglomeration economies theory that sees the increase in urban population as a stimulus of economic growth (Rosenthal and Strange 2004; van Oort, de Geus and Dogaru 2015). Population size measured by the number of inhabitants had a significant negative effect on metropolitan growth and the coefficients for population growth were positive, but the overall impact was very small which implied that density and size was more important that the growth of the population. Net migration was negatively influencing metropolitan economic growth for the System GMM method and was not statistically significant for the other two techniques.
Academic contributions
The empirical findings of this thesis can provide important contributions to the economic growth theory, which has seen many conflicting views regarding the main determinants that can foster economic development. Because it analysed the growth factors at distinct levels (country, regional – NUTS and metropolitan), the outcomes were more comprehensive and offered a better comparison. The major academic contributions of this thesis can be summarised as follows:
1 – Chapter Three has shown that economic growth theory is very complex and a unified model is difficult to be constructed because of the inherent estimation and data collection problems. This chapter used several panel data techniques which offered more comprehensive information regarding the main determinants of economic growth in the European Union at country level and also increased efficiency in estimation. The investigation also included public and private economic variables so as to demonstrate if one is more beneficial than the other. The results showed that debt in general, private or public, has a negative effect on economic growth. Public investment measured as total government investment hinders growth, but FDI is positively linked with it. The study opted also for decomposing education into 3 components. This has shown the positive role of tertiary education in fostering economic growth compared with primary and secondary education. As mentioned in the literature, non-economic variables can play an important role, albeit an indirect one, in the economic growth process. The results showed that the rule of law is an important determinant of economic development. Control of corruption is also an important non-economic variable. As there are more than 145 determinants that were demonstrated by the literature to have contributed to economic growth in at least one article, creating the framework to include all of them into a single model will yield more comprehensive knowledge on a much debated subject.
2 – Chapter Four continued the previous investigation by applying panel data techniques for two separate growth equations for a panel of 98 NUTS 1 and 273 NUTS 2 regions. This chapter contributed to the existing regional economic literature by helping to better understand the role of the main factors that determine growth. This investigation used the GMM and system GMM techniques which are widely used in growth analysis and also utilized a recently new method, namely the quasi-maximum likelihood. It was used to improve the estimations for the panel data which had missing values for some specific regions. Many empirical studies that dealt with regional investigation used variables like population, infrastructure, innovation, employment, migration. This study also used these variables simultaneously and disaggregated some of them. For example, it used average weekly hours at the main work place for male and female. Also employment was used as female and male employment and as a total. The results showed that the main factors that improve regional economic growth are employment, specifically male employment, life expectancy, tertiary education. Furthermore this chapter, like the other two empirical ones of this thesis, used the lag dependent variable to demonstrate or infirm the convergence hypothesis. In this chapter the regional divergence was furthermore confirmed by graphically illustrating the fact that after the 2008 crisis the EU areas are drifting more apart from each other, especially the ones from the former communist states.
3 – Chapter Five provides a comprehensive glance regarding the main determinants of economic growth at metropolitan level. Compared with studies done at country or regional level, there are not too many empirical investigations for metropolitan areas. This chapter will try to fill in the gaps in the literature. An interesting fact was that it used all the economic metropolitan sectors (industry, agriculture, construction, wholesale and retail trade, transport, accommodation and food services, manufacturing and ITC) to investigate which branch has a positive or negative influence. To offer robustness it used three panel data techniques, namely GMM, system GMM and QML. The main results showed that agriculture, forestry and fishing had a negative impact for the 271 EU metropolitan areas studied. Industry, construction and wholesale and retail trade, transport, accommodation and food service activities were positively related to metropolitan growth. The manufacturing and information and communication sectors were, in general, statistically insignificant and didn’t contribute to economic growth. The number of employees was positively linked with EU metropolitan growth. Population size had a significant negative effect on metropolitan growth and the coefficients for population growth were positive, but the overall impact was very small which implied that density and size was more important that the growth of the population. To further improve efficiency and offer robustness, the investigation opted to split the time period in two (2000-2008 and 2008-2013) and also dividend the panel sample to measure the difference between Western and Central and Eastern metropolitan areas.
Policy implications
There are several policy recommendations that can be drawn from this thesis.
1 – The finding of Chapter Three showed that employment and productivity have a big contribution to economic growth. This means that in the EU, the state and private companies should concentrate in stimulating employment and productivity. The unidirectional link from energy consumption to economic growth suggests that energy has a meaningful role in shaping growth and that the state has to use energy policies wisely as not to harm the economy. This concept was hardly debated and confirmed by many research papers. Yu and Choi (1985), Masih and Masih (1996), Lee (2005), Narayan and Prasad (2008), Bhattacharya and Bhattacharya (2014), Mahalik and Mallick (2014) showed that for developing countries (India, China, Pakistan, Turkey, Brazil, Indonesia, etc.) and also for developed countries (France, Australia, Italy, Korea, Japan, etc.) energy consumption plays an important role in shaping economic growth. The chapter suggests that a rise in population by 1% determines a drop in real GDP/capita growth of -1.15%. Population growth could have a negative influence on economic development by impacting the investment and savings behaviours of citizens, the dependency ratio and the quality of human capital.
Rule of law had a significant contribution to economic growth in Chapter Three. The EU authorities have to concentrate on safeguarding the justice system as to not hinder economic development. Better measures in controlling corruption will have a beneficial effect on the rule of law. Control of corruption is also a variable that had a positive effect on real GDP/capita. Trade openness was a determinant with a significant positive effect on economic growth. Better access to markets for developing countries in the EU can facilitate economic development. Trade openness can have an influence on economic growth through a multitude of different channels like technological transfers, the increase in economies of scale, competitive advantage (Chang et al. 2009). Tertiary education is the most important type of schooling education that had a significant and positive effect on growth. Education policies should concentrate on stimulating higher education and innovative research. The negative impact of government debt has to determine EU states to lower or better manage their borrowing and debt service. Also policy makers should take into account that the over-indebtedness of the financial sector has a negative consequence for the economy.
2 – Chapter Four suggests that early leavers from education and training are a negative influence on regional economic growth. This social category is at risk economically and policy makers have to adopt measures for the better integration of this group in the society and on the labour market. Infrastructure development appears to not have a defining role in shaping regional economic growth. Infrastructure endowment is poorly linked to economic growth and the exact returns and implications of this type of investment is not so clear (Crescenzi and Rodríguez-Pose 2012; Rodriguez-Pose, Psycharis and Tselios 2012). The results obtained for population density contradict the agglomeration economies theory. It seems that regional agglomeration is not an important factor. This outcome can be attributed to Europe’s high number of small and medium size cities and the negative externalities of living in a big city like congestion cost, labour competitiveness, pollution and high rental costs (Dijkstra et al. 2013).
3 – The results of Chapter Five clearly show that metropolitan regions are not converging to the steady state of growth. There are considerable differences in development among metropolitan areas and there is a visible gap between Western regions and Eastern regions. For example the only metropolitan region from Central-Eastern Europe that is in the top ten list regarding in purchasing power per inhabitant in the year 2012 is Bratislava. In this regard underperforming urban areas are located in Romania, Bulgaria, Hungary, Poland and Croatia. Regional policy makers have to take this into account and try to limit the gaps between these regions by providing better cohesion funds. The main findings of this chapter regarding the influences of economic sectors on metropolitan growth are that agriculture, forestry and fishing can have a negative impact on economic growth. A considerable portion of EU funds is employed for stimulating investment in agricultural production and the big countries are also subsidizing this sector so as to be more competitive. The results of this chapter suggest that these allocations appear to not be efficient for metropolitan growth. Industry, construction and wholesale and retail trade, transport, accommodation and food service activities are positively related to metropolitan growth. The manufacturing and information and communication sectors were, in general, statistically insignificant. These findings have considerable policy implications for policymakers in the sense that EU funds must stimulate mostly the economic branches with the most value added for the economy.
The results also show that population density has a small influence on metropolitan development. Population density is used as a proxy for agglomeration. According to Puga (2002) high agglomeration in capital cities and large urban areas can have an influence on growth increasing labour specialization and productivity. van Oort, de Geus and Dogaru (2015) showed that agglomeration plays an important role for 15 EU countries at regional level, specifically for 205 EU NUTS2 regions. The results obtained by the system GMM estimator are in contrast with the agglomeration economies theory that sees the increase in urban population as a stimulus of economic growth (Rosenthal and Strange 2004; van Oort, de Geus and Dogaru 2015). Population size measured by the number of inhabitants has a significant negative effect on metropolitan growth and the coefficients for population growth were positive, but the overall impact is very small which implies that density and size is more important that the growth of the population.
APPENDICES
APPENDIX I
The variables and data sources
List of countries included in the empirical analysis
Correlation matrix graph between Ly Lyt-1 LLE LFEC LFSL LGDEBT LEXP LDEFICIT
Correlation matrix graph between Ly LEMPL LEXPORT LIMP LOPEN LPDEBT LPROD
Correlation matrix graph between Ly LGFCF LFDI LINF LPOP LEDUC1 LEDUC2 LEDUC3
APPENDIX II
The variables and data sources
List of regions included in the empirical analysis
NUTS 1 regional map
NUTS 2 regional map
Correlation matrix graph for the NUTS 1 variables
Correlation matrix graph for the NUTS 2 variables
APPENDIX III
The variables and data sources
List of metropolitan regions included in the empirical analysis
EU metropolitan regions map
Correlation matrix graph
REFFERENCES
Abbott, A. and Jones, P. (2011), Procyclical government spending: Patterns of pressure and prudence in the OECD, Economics Letters, 111: 230–232.
Abosedra, S. and Baghestani, H. (1989), New evidence on the causal relationship between United States energy consumption and gross national product, Journal of Energy Development, 14: 285-292.
Abu-Eideh, O.M. (2013), Palestinian Exports Performance and Its Impact on Economic Growth: An Econometric Study During The Period (1994 – 2011), Alazhar University Journal , Humanities Series, 15(1): 347-376.
Abu-Eideh, O.M. (2014), Factors of economic growth in Palestine: an empirical Analysis during the period of (1994-2013), International Journal of Business and Economic Development, 2(2): 70-84.
Acemoglu D., Johnson S. and Robinson J. (2002), Reversal of fortune: geography and institutions in the making of the modern world income distribution, Quarterly Journal of Economics, 117(4): 1231–1294.
Acemoglu, D. (2009), Introduction to modern economic growth, Princeton NJ, Princeton Univiversity Press.
Acemoglu, D. and Verdier, T.A. (2000), The Choice between Market Failures and Corruption, American Economic Review, 90(1): 194-211.
Acemoglu, D., Johnson, S. and Robinson, J. (2005), Institutions as a Fundamental Cause of Long-Run Growth, in P. AGHION and S. DURLAUF (eds.), Handbook of Economic Growth, 1A (pp , 385-472), Amsterdam: Elsevier.
ADAMCZYK-ŁOJEWSKA, GRAŻYNA. (2013), Work productivity as an economic growth and prosperity factor in Polan from 2004-2010. Studia i Materialy Polskiego Stowarzyszenia Zarzadzania Wiedza / Studies & Proceedings Polish Association for Knowledge Management, Issue 67: pp.38-53.
Afonso A. and Alegre, J.G. (2011), Economic growth and budgetary components: A panel assessment for the EU, Empirical Economics., (41)3: 703–723.
Aghion, P and Howitt, P. (2009), The economics of growth, MIT Press, MIT Press Books.
Aghion, P. and Howitt, P. (1992), A Model of Growth Through Creative Destruction, Econometrica, 60(2): 323-351.
Aghion, P. and Howitt, P. (1998), Endogenous Growth Theory, Cambridge: MIT Press.
Ahmad, J. and Kwan, A.C.C. (1991), Causality between exports and economic growth: Empirical evidence from Africa, Economics Letters, 37(3): 243-248.
Aisen, A. and Veiga, F.J. (2013), How does political instability affect economic growth?, European Journal of Political Economy, 29(March): 151-167.
Alesina, A. and Rodrik, D. (1994), Distributive politics and economic growth, Quarterly Journal of Economics, 109: 465-490.
Alesina, A., Campante, F. and Tabellini, G. (2008) Why is fiscal policy often procyclical?, Journal of the European Economic Association, 6(5): 1006–1030.
Ali, H., and Ahmad, S. (2010), Foreign direct investment, economic growth and regional disparities: The malaysian experience, International Journal of Interdisciplinary Social Sciences, 5(8): 405-416.
AL-Raimony, A.D. (2011), The Determinants of Economic Growth in Jordan. Abhath Al-Yarmouk, Humanities and Social Sciences Series, 27(3): 2297-2305.
Anderson, G. and Ge, Y. (2004), Do Economic Reforms Accelerate Urban Growth? The Case of China, Urban Studies, 11: 2197–2210.
Apergis, N. and Payne, J.E. (2011), A dynamic panel study of economic development and the electricity consumption-growth nexus, Energy Economics, 33(5): 770-781.
Arellano, M. and Bond, S. (1991), Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations, Review of Economic Studies, 58: 277–297.
Arellano, M. and Bover, O. (1995), Another look at the instrumental variable estimation of error-components models, Journal of Econometrics, 68: 29–51.
Armstrong, H. and Read, R. (2004), The Economic Performance of Small States and Islands: The Importance of Geography, Islands of the World VIII International Conference, Taiwan.
Arpaia A. and Turrini A. (2008), Government Expenditure and Economic Growth in the EU: Long-Run Tendencies and Short-Term Adjustment,” SSRN Working Paper Series., pp. 800–844.
Arusha, C. (2009), Government Expenditure, Governance and Economic Growth, Comparative Economic Studies, 51(3)- 401-418.
Arvanitidis, P., Petrakos, G. and Pavleas, S. (2007), Determinants of economic growth: the experts’ view, DYNREG20, Economic and Social Research Institute (ESRI).
Aschauer, D.A. (1990), Why is infrastructure Important? Boston, MA: Munnell AH (ed) Is there a shortfall in public capital investment? Federal Reserve Bank of Boston.
Au, C-C. and Henderson, V. (2006), Are China’s Cities Too Small?, Review of Economic Studies, 73: 549–576.
Audretsch, D., Dohse, D., and Niebuhr, A. (2010), Cultural diversity and entrepreneurship: a regional analysis for Germany, Annals of Regional Science, 45: pp.55–85.
Ayres, C. (1962), “The theory of economic progress. A study of the Fundamental Economic Development and Cultural Change”, New York: Schocken.
Bagli, S. and Adhikary, M. (2014), FDI inflow and Economic Growth in India An Empirical Analysis, Economic Affairs, 59(1): 23-33.
Baltagi, B. H. (2008), Econometric Analysis of Panel Data, John Wiley & Sons Ltd.
Bardhan, P. (1997), Corruption and Development: A Review of Issues, Journal of Economic Literature, 35(3): 1320-1346.
Barro, R.J. (1990), Government spending in a simple model of endogenous growth, Journal of Political Economy, 98(5):S103–S125.
Barro, R.J. (1991), Economic growth in a cross-section of countries, Quarterly Journal of Economics, 106(2): 407–443.
Barro, R.J. (1996), Determinants of Economic Growth: A Cross-Country Empirical Study, NBER Working Paper, 5698.
Barro, R.J. (2003), Determinants of economic Growth in a Panel of Countries, Annals of Economics and Finance, 4(2): 231-274.
Barro, R.J. and Lee, J.-W. (1994), Sources of economic growth, Carnegie-Rochester Conference Series on Public Policy, 40: 1–46.
Barro, R.J. and Sala-I-Martin, X. (1995), Economic Growth, (D. R. Henderson, Ed.) European psychiatry the journal of the Association of European Psychiatrists, 25: 159–163.
Barro, R.J., and Sala-I-Martin, X. (2004), Economic growth (2nd ed.), Cambridge, MA: MIT Press.
Bartelmus, P. (1994). Environment, Growth, and Development: The Concepts and Strategies of Sustainability, Routledge, pp:163.
Baum, F. (2006), An introduction to modern econometrics using Stata, Texas: Stata Press.
Becker, G.S. and Barro, R.J. (1988), A Reformtiation of the Economic Theory of Fertility, Quatierly Journal of Economics, 103( 1): 1–25.
Benoit, E. (1973), Defense and Economic Growth in Developing Countries, Lexington, MA: Lexington Books.
Benoit, E. (1978), Growth and Defense in Developing Countries, Economic Development and Cultural Change, 26 (2): 271-280.
Benos, N. (2009), Fiscal policy and economic growth: empirical evidence from EU countries, University of Ioannina.
Berdegue, J., Carriazo, F., Jara, B., Modrego, F. and Soloaga, I., (2015), Cities, Territories, and Inclusive Growth: Unraveling Urban–Rural Linkages in Chile, Colombia, and Mexico, World Development, 73(C): pp. 56-71.
Bhandari, R., Dharmendra, D., Gyan, P. and Kamal, U. (2007), Foreign Aid, FDI and Economic Growth in East European Countries, Economics Bulletin, 6(13): 1-9.
Bhargava, A. and J. D. Sargan (1983), Estimating Dynamic Random Effects Models from Panel Data Covering Short Time Periods, Econometrica, 51(6): 1635-1659.
Bhattacharya, M. and Bhattacharya, S.N. (2014), Economic growth and Energy consumption nexus in Developing, World: The case of China and India, Journal of Applied Economics and Business Research, 4(3): 150-167.
Bils, M., and Klenow, P. (2000), Does schooling cause growth?, American Economic Review, 90(5): 1160–1183.
Bloom, D.E. and Sachs, J.D., (1998) Geography, Demography, and Economic Growth in Africa, Brookings Papers on Economic Activity, 29(2): 207-296.
Bloom, D.E. and Williamson, J.G. (1998), Demographic transitions and economic miracles in emerging Asia, World Bank Economic Review, 12(3):419–455.
Blundell, R. and Bond, S. (1998), Initial conditions and moment restrictions in dynamic panel data models, Journal of Econometric,s 87: 115–143.
Boldeanu, F. T. and Constantinescu, L. (2015), The main determinants affecting economic growth. Bulletin of the Transylvania University of Brasov. Economic Sciences, 8(2), pp.329-338.
Boldeanu, F.T. and Tache, I. (2015), Sub-division expenditures and economic growth in Europe based on United Nation’s Classification of the Functions of Government, International Journal of Economic Practices and Theories, 5(5): 435-446.
Boldeanu, F.T. and Tache, I. (2016), A regional approach to the metropolitan economic growth: Evidence from the European Union, Journal of Smart Economic Growth, 1(1): 29-72.
Boleanu, F.T. and Ion, M.S. (2015), The Macro-Regional Framework And Diversity In Europe, EURINT proceedings, 2(2): 52-63.
Bond, S.R., Hoeffler, A. and Temple, J. (2001), GMM Estimation of Empirical Growth Models, Economics Papers, 2001-W21, Economics Group, Nuffield College, University of Oxford.
Bottazzi, L. and Peri, G. (2002) Innovation and Spillovers: Evidence from European Regions, European Economic Review, 47: 687–710.
Braguinsky, S. (1996), Corruption and schumpeterian growth in different economic environments, Contemporary Economic Policy, 14(3): 14–25.
Brasoveanu, I., Brasoveanu L.O. and Paun, C. (2008), Correlations Between Fiscal Policy And Macroeconomic Indicators In Romania, Theoretical and Applied Economics, Asociatia Generala a Economistilor din Romania – AGER, 11(528): 51-59.
Braudel, F. (1981-84), Civilization and Capitalaism, 15th-18th Century, London, Collins (3 volumes).
Braun, A.R. Ikeda, D. and Joines, D.H. (2009), The saving rate in Japan: why it has fallen and why it will remain low, International Economic Review, 50(1):291–321.
Brânză, R. (2007), ROLUL CAPITALULUI UMAN ÎN MODELE DE CRESTERE ECONOMICA, The Romanian Economic Journal, 24(Year X): 43-53.
Carroll, C.D. (2014), The Romer (1986) Model of Growth, John Hopkins University, http://www.econ2.jhu.edu/people/ccarroll/public/lecturenotes/Growth/Romer86.pdf
Cervellati, M. and Sunde, U. (2009), Life Expectancy and Economic Growth: The Role of the Demographic Transition, Institute for the Study of Labor (IZA), IZA Discussion Papers No. 4160.
Chang, R., Kaltani, L. and Loayza, N. (2009), Openness is Good for Growth: The Role of Policy Complementarities, Journal of Development Economics, 90(1): 33-49.
Chang, T., Liu, W. and Caudill, S.B. (2004), A re-examination of Wagner’s law for ten countries based on cointegration and error correction modelling techniques, Applied Financial Economics, 14: 577–589.
Christiaensen, L. and Todo, Y. (2013), Poverty reduction during the ruralurban transformation: the role of the missing middle, Policy Research Working Paper Series 6445, World Bank.
Cioban, G.L. (2014), The tertiary civilization – concept of the economic growth, The USV Annals of Economics and Public Administration, 1(19): 48-55.
Cohen, B. (2004), Urban Growth in Developing Countries: A Review of Current Trends and a Caution Regarding Existing Forecasts, World Development, 32(1): 23–51.
Combes, P. (2000), Economic Structure and Local Growth: France, 1984–1993, Journal of Urban Economics, 47: pp. 329–355.
Coutinho, R. and Gallo, G. (1991), Do Public and Private Investment Stand in Each Other’s Way, WDR Background Paper, World Bank.
Crescenzi, R. and Rodríguez-Pose, A. (2012), Infrastructure and regional growth in the European Union, Regional Science, 91(3): 487-513.
Crescenzi, R., Rodriguez-Pose, A. and Storper, M. (2007), The territorial dynamics of innovation: a Europe-United States comparative analysis, Journal of Economic Geography, 7(6):673-709.
Crosby, A.W. (1986), Ecological Imperialism: The Biological Expansion of Europe, 900-1900, Cambridge, Cambridge University Press.
Da Mata, D., Deichmann, U., Henderson, J.V., Lall, S.V. and Wang, H.G. (2005), Determinants of City Growth in Brazil. National Bureau of Economic Research (NBER) Working Paper 11585. Available from URL: http://www.nber.org/papers/w11585. August, National Bureau of Economic Research, Cambridge, MA.
Dao, M.Q. (2012), Government expenditure and growth in developing countries, Progress in Development Studies, 12(1): 77-82.
De Soto, H. (1989), The Other Path: The Invisible Revolution in the Third World. Harpercollins.
de Vita, G. and Kyaw, K. (2009), Growth effects of FDI and portfolio investment flows to developing countries: a disaggregated analysis by income levels, Applied Economics Letters, 16: 277–283.
Denison, E.F. (1962), The Sources of Economic Growth in the United States and Alternatives Before Us, CED Supplementary Paper, No 13.
Devarajan, S., Swaroop, V., and Zou, H. (1996), The composition of public expenditure and economic growth, Journal of Monetary Economics, 37(2): 313–344.
Diamond, J. M. (1997), Guns, Germs and Steel: The Fate of Human Societies, New York NY, W.W. Norton & Co.
Dijkstra, L., Garcilazo, E., and McCann, P. (2013), The economic performance of European cities and city regions: myths and realities, European Planning Studies, 21 (3): 334-354.
Dollar, D. and Kraay, A. (2000), Trade, Growth and Poverty, The World Bank Development Research Group, Washington.
Domar, E. (1946), Capital Expansion, Rate of Growth and Employment, Econometrica, 14(2):137-147.
Dornbusch, R. (1975), Exchange Rates and Fiscal Policy in a Popular Model of International Trade, American Economic Review, 65(5): 859-71.
Dunne, J.P. and Tian, N., (2013), Military Expenditure and Economic Growth: A Survey, The Economics of Peace and Security Journal, (8)1: 5–11.
Durlauf, S., Johnson, P. A., and Temple, J. (2005), Growth econometrics, Amsterdam: North Holland, in P. Aghion, & S. Durlauf (Eds.), Handbook of economic growth.
Easterly W. and Levine R. (2003) “Tropics, germs and crops: how endowments influence economic development”, Journal of Monetary Economics, 50(1), 3-39.
Easterly, W. and Levine, R. (1997), Africa's Growth Tragedy: Policies and Ethnic Divisions, Quarterly Journal of Economics 112(4): 1203-1250.
Edwards, S., (1992), Trade orientation, distortions and growth in developing countries. Journal of Development Economics, 39: 31– 57.
Edwards, S., (1998), Openness, Productivity and Growth: What Do We Really Know?. Economic Journal,108: 383-398.
Fisher, R.A. (1925), Statistical Methods for Research Workers, Edinburgh: Oliver and Boyd.
Flisi, S., Goglio, V., Meroni, E.C. and Vera-Toscano, E. (2015), School-to-work transition of young individuals: what can the ELET and NEET indicators tell us?, Luxembourg: Publications Office of the European Union, EUR – Scientific and Technical Research Reports.
Frees, E. W. (1995), Assessing cross-sectional correlation in panel data, Journal of Econometrics, 69: 393–414.
Friedman, M. (1937), The use of ranks to avoid the assumption of normality implicit in the analysis of variance, Journal of the American Statistical Association, 32: 675–701.
Gallup J., Sachs, J. and Mellinger, A. (1999), Geography and Economic Development. International Regional Science Review, 22(2): 179-232.
Gallup, J.L. and Sachs, J.D. (2001), The economic burden of malaria, American Journal of Tropical Medicine and Hygiene, 64(1-2 Suppl):85-96.
Galvin, H., (2003), The Impact of Defence Spending on the Economic Growth of Developing Countries: A Cross-Section Study, Defence and Peace Economics, 14(1): 51–59.
Ganelli, G. (2010), The international effects of government spending composition, Economic Modelling, 27(3): 631-40.
García–Milá, T., McGuire, T.J. and Porter, R.H. (1996), The effect of public capital in state-level production functions Reconsidered, The Review of Economics and Statistics, 78: 177–180
Gavin, M. and Perptti, R. (1997), Fiscal policy in Latin America, Macroeconomics Annual, 12: 11–70.
Ghali, K.H. (1998), Public investment and private capital formation in a vector error-correction model of growth, Applied Economics, 30 : 837-844.
Ghosh, M. (2012), Regional Economic Growth and Inequality in India during the Pre- and Post-reform Periods, Oxford Development Studies, 40(2): 190-212.
Ghosh, R. A. (2012), Revisiting the Relationship between Economic Growth and Government Size, Economics Research International, Volume 2012:1-8.
Ghosh, S, and Gregoriou, A, (2008), The composition of government spending and growth: Is current or capital spending better?, Oxford Economic Papers, 60(3): 484-516.
Ghosh, S. (2002), Electricity consumption and economic growth in India, Energy Policy, 30: 125-129.
Glaeser, E. L. (2011), Triumph of the City: How Our Greatest Invention Makes Us Richer, Smarter, Greener, Healthier, and Happier, New York: The Penguin Press.
Glaeser, E.L., Kallal, H.D., Scheinkman, J.A. and Shleifer, A. (1992), Growth in Cities, Journal of Political Economy, 100: pp. 1126–1152.
Glewwe, P., Maiga, E. and Zheng, H. (2014), The Contribution of Education to Economic Growth: A Review of the Evidence, with Special Attention and an Application to Sub-Saharan Africa, World Development, 59: 379-393.
Golubchikov, O. (2007), Re-Scaling the Debate on Russian Economic Growth: Regional Restructuring and Development Asynchronies, Europe-Asia Studies, 59(2): 191-215.
Granato J., Inglehart, R. and Leblang, D. (1996), The effect of cultural values on economic development: theory, hypotheses, and some empirical tests, American Journal of Political Science 40(3): 607-631.
Graziano, L. (1980), Clientelismo e Sistema Politico, Il Caso dell Italia, F. Angeli, Milan.
Greene, H.W. (2008), Econometric analysis, 6th ed., Upper Saddle River, N.J.: Prentice Hall.
Grier, K. and Tullock, G. (1989), An empirical analysis of cross-national economic growth, 1951-1980, Journal of Monetary Economics, 24(1): 259-276.
Grossman, G. and Helpman, E. (1991), Innovation and growth in the global economy, Cambridge, MIT Press.
Guclu, M. (2013), Manufacturing and Regional Economic Growth in Turkey: A Spatial Econometric View of Kaldor’s Laws, European Planning Studies, 21(6): 854–866.
Hanushek, E. and Kimko, D. (2000), Schooling, labor force quality and the growth of nations, American Economic Review, 90(5): 1184–1208.
Hanushek, E. and Woessmann, L. (2008), The role of cognitive skills in economic development, Journal of Economic Literature, 46(3): 607–668.
Harris, D.J. (1978), Capital Accumulation and Income Distribution, Redwood City, Stanford University Press.
Harrod, R.F. (1939), An Essay in Dynamic Theory, The Economic Journal, 49(193):14-33.
Hartono, D. et al. (2007), REGIONAL ECONOMIC INTEGRATION AND ITS IMPACTS ON GROWTH, POVERTY, AND INCOME DISTRIBUTION: THE CASE OF INDONESIA, Review of Urban & Regional Development Studies, 19(2): 138–153.
Haurin, D. (1980), The Regional Distribution of Population, Migration and Climate. The Quarterly Journal of Economics, 95(2): 293-308.
Hedlund, J.D. (1983), Distribution Theory Revisited: An Empirical Examination of the Weintraub Synthesis, Journal of Post Keynesian Economics, 6(1): 73-81.
Helliwell, J. and Putnam, R. (1995), Economic Growth and Social Capital in Italy, Eastern Economic Journal, XXI: 295-307.
Henderson, J. V. (2010), Cities and development, Journal of Regional Science, 50(1): pp. 515–540.
Henderson, J.V. and R. Becker (2000), Political Economy and City Sizes and Formation, Journal of Urban Economics, 48: pp. 453–484.
Hercowitz, Z. and Strawczynski, M. (2004), Cyclical ratcheting in government spending: Evidence from the OECD, Review of Economics and Statistics, 86(1): 353-61.
Hermes, N. and Lensink, R. (2000), Foreign direct investment, financial development and economic growth, Journal of development studies, 40(1): 142-163.
Holzner, M. (2011), Inequality, growth and public spending in Central, East and Southeast Europe, ECINEQ WP 2011, pp. 1–25.
Hou, N. and Chen, B. (2014), Military Spending and Economic Growth in An Augmented Solow Model: A Panel Data Investigation for OECD Countries, Peace Economics, Peace Science, and Public Policy, 20(3): 395-409.
Hsiao, C., Pesaran, M. H. and Tahmiscioglu, A. K. (2002), Maximum Likelihood Estimation of Fixed Effects Dynamic Panel Data Models Covering Short Time Periods, Journal of Econometrics, 109: 107—150.
Huntington, S.P. (1968), Political Order in Changing Societies, New Haven, Yale University Press.
Huntington, S.P. (2002) Modernization and corruption, Political Corruption: Concepts and Contexts (Eds) A. Heidenheimer and M. Johnston, Transaction, New Brunswick, NJ, pp: 253–263.
Im, K.S., Pesaran, M.H. and Shin, Y. (2003), Testing for unit roots in heterogeneous panels, Journal of Econometrics, 115: 53–74.
Inada, K-I. (1963), On a Two-Sector Model of Economic Growth: Comments and a Generalization, The Review of Economic Studies, 30 (2): 119–127.
Inglehart, R and Baker, W. (2000), Modernization, cultural change and the persistence of traditional values, American sociological review, 65: 19-51.
Jevons, W.S. (1871), The Theory of Political Economy, London, Macmillan.
Johnson, A. 2006. The Effects of FDI Inflows on Host Country Economic Growth, CESIS – Centre of Excellence for Science and Innovation Studies, Royal Institute of Technology Working Paper Series, 58: 1-58.
Jones, E.L. (1981), The European Miracle: Environments, Economies, and Geopolitics in the History of Europe and Asia, Cambridge, Cambridge University Press.
Kaldor, N. (1957), A Model of Economic Growth, The Economic Journal, 67 (268): 591–624.
Kaufmann, D. and Wei, S. J. (2000), Does ‘grease money’ speed up the wheels of commerce?, International Monetary Fund, Working Paper No. WP/00/64, International Monetary Fund, Washington, DC.
Kaufmann, D., Kraay, A. And Mastruzzi, M. (2010), The worldwide governance indicators : methodology and analytical issues, Policy Research Working Paper Series 5430, The World Bank.
Kelley, A. and Schmidt, R. (1995), Aggregate Population and Economic Growth Correlations: The Role of the Components of Demographic Change, Demography, 32: 543-55.
Kelley, A.C. and Smith, R.M. (2005), Evolution of recent economic-demographic modeling: a synthesis, Journal of Population Economics ,18(2): 275–300.
Kessides, C. (1993), The contributions of infrastructure to economic development: A review of experience and policy implications, Discussion Paper, 214, The World Bank, Washington DC.
Keynes, J.M. (1936), The General Theory of Employment, Interest and Money, London, Macmillan.
Khan, M.S. and Kumar, M.S. (1997), PUBLIC AND PRIVATE INVESTMENT AND THE GROWTH PROCESS IN DEVELOPING COUNTRIES, OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 59(1): 69-88.
Khan, M.S. and Reinhart, C.M. (1990), Private Investment and Economic Growth in Developing Countries, World Development, 18: 19-27.
Kim, T-Y. and Heshmati, A. (2014), Economic Growth The New Perspectives for Theory and Policy, New York: Springer.
Klenow, P.J. and Rodriguez-Clare, A. (2004), Externalities and Growth, NBER Working Papers 11009, National Bureau of Economic Research, Inc.
Knack, S. and Keefer, P. (1995), Institutions and Economic Performance: Cross-Country Tests Using Alternative Institutional Measures, Economics and Politics, 7(3): 207-227.
Knack, S. and Keefer, P. (1997), Does social capital have an economic impact? A cross-country investigation, Quarterly Journal of Economics, 112(4): 1252-88.
Knight, F.H. (1944), Diminishing returns from investment. Journal of Political Economy, 52 (March): 26-47.
Kormendi, R. and Meguire, P. (1985), Macroeconomic determinants of growth: cross-country evidence, Journal of Monetary Economics, 16(4): 141-63.
Kraft, J. and Kraft, A. (1978), On the relationship between energy and GNP, Journal of Energy Development, 3(2): 401-403.
Kripfganz, S. (2016), xtdpdqml: Quasi-maximum likelihood estimation of linear dynamic panel data models in Stata, Manuscript, University of Exeter.
Krugman, P. (1991), Geography and Trade, Cambridge, MA: MIT Press.
Kuznets, S. (1973), Modern Economic Growth: Findings and Reflections, American Economic Review, 63(3): 247-258.
Lamartina, S. and Zaghini A. (2008), Increasing public expenditure: Wagner’s Law in OECD countries, German Economic Review, 12(2): 149–164.
Lane, P. R. (2003), The cyclical behaviour of fiscal policy: Evidence from the OECD, Journal of Public Economics, 87(12): 2661-75.
Lee, C. C. (2005), Energy consumption and GDP in developing countries: A cointegrated panel analysis'', Energy Economics, 27: 415–427.
Leff, N. H. (1964), Economic Development Through Bureaucratic Corruption, The American Behavior Scientist, 8(3): 8-14.
Lensink, R., Bo, H. and Sterken, E. (1999), “Does Uncertainty Affect Economic Growth? An Empirical Analysis, Weltwirtschaftliches Archiv, 135: 379-396
Lensink, W. and Morrissey O. (2006), 'Foreign Direct Investment: Flows, Volatility and the Impact on Growth, Review of International Economics, 14(3): 478-493.
Levine, R. and Renelt, D. (1992), A sensitivity analysis of cross-country growth regressions, American Economic Review, 82(4): 942–963.
Lewis, A. (1955), The Theory of Economic Growth, London ,George Allen and Unwin.
Li, X. and Liu, X. (2005), Foreign Direct Investment And Economic Growth: An Increasingly Endogenous Relationship, Worm Development, 33(3): 393-407.
Lin, S. and Song, S. (2002), Urban Economic Growth in China: Theory and Evidence, Urban Studies, 39: 2251–2266.
Lipset, S.M. (1959), Some Social requisites of Democracy: Economic Development and Political Legitimacy, American Political Science Review, 53(March): 69-105.
Lucas, R.E. (1988), On the Mechanics of Economic Development, Journal of Monetary Economics, 22 (1): 3–42.
Lui, F. (1985), An equilibrium Queuing model of bribery, Journal of Political Economy, 93(4): 760–781.
Mahalik, M.K. and Mallick, H. (2014), ENERGY CONSUMPTION, ECONOMIC GROWTH AND FINANCIAL DEVELOPMENT: EXPLORING THE EMPIRICAL LINKAGES FOR INDIA, The Journal of Developing Areas, 48(4): 139-159.
Malešević Perović, L., Simic, V. and Muštra, V. (2014), Investigating the Influence of Economic and Socio-Political Openness on Growth, International Journal of Economic Sciences and Applied Research, 6 (3): 35-59.
Mallick, H. (2009), EXAMINING THE LINKAGE BETWEEN ENERGY CONSUMPTION AND ECONOMIC GROWTH IN INDIA, The Journal of Developing Areas, 43(1): 249-280.
Mallik, G and Chowdhury, A. (2001), Inflation and economic growth: Evidence from four South Asian countries, Asia-Pacific Development Journal, 8(1): 123-135.
Malthus, T.R. (1798), Essay on the Principle of Population (1 ed.), London: Macmillan, 1926(1st ed.).
Mankiw, N. G., Romer, D., and Weil, D. (1992), A contribution to the empirics of economic growth, Quarterly Journal of Economics, 107(2): 407–437.
Marinaș, C-M. (2007), The Estimation of the Cointegration Relationship between the Economic Growth, Investments and Exports. The Romanian Case, Theoretical and Applied Economics, 7(512): 11-16.
Marinaș, C-M. (2008), Covergența economică. Bucharest: Economic Publishing House.
Marshall, A. (1890), Principles of economics, London, Macmillan.
Masih, A .M. M. and Masih, R. (1996), Electricity consumption, real income and temporal causality: Results from a multi-country study based on cointegration and error–correction modeling techniques, Energy Economics , 1996, 18: 165–183.
Masih, A. M. M. and Masih, R. (1997), On the causal relationship between energy consumption, real income prices: Some new evidence from Asian NICs based on multivariate cointegration / vector error correction approach, Journal of Policy Modeling, 19: 417-440.
Maslow, A.H. (1943), A theory of human motivation, Psychological Review, 50 (4): 370–96.
Mauro, P. (1995), Corruption and Growth, Quarterly Journal of Economics, 110: 681-712.
McCann, P. (2007), Sketching out a model of innovation, face-to-face interaction and economic geography, Spatial Economic Analysis, 2(2): pp.117–134.
McCann, P. and Acs, Z. J. (2011), Globalisation: Countries, cities and multinationals, Regional Studies, 45(1): pp. 17–32.
McDonald, B.D., Eger, R.J., (2010), The Defense-Growth Relationship: An Economic Investigation into Post-Soviet States, Peace Economics, Peace Science and Public Policy, 12(1): 1–26.
McNeill, W.H. (1963), The Rise of the West: A History of the Human Community, Chicago, University of Chicago Press.
Melo, P.C., Graham, D.J. and Noland, R.B. (2009), A Meta-analysis of Estimates of Urban Agglomeration Economies, Regional Science and Urban Economics, 39: pp. 332–342.
Menger, C. (1871), Principles of Economics, Vienna, Braumüller.
Menyah, K. and Wolde-Rufael, Y. (2013), Government Expenditure and Economic Growth: The Ethiopian Experience, 1950-2007, The Journal of Developing Areas, 47( 1, Spring 2013): 263-280.
Mihuț, I.S. and Luțaș, M. (2014), SUSTAINABLE GROWTH: RECENT TRENDS ACROSS CENTRAL AND EASTERN EUROPEAN ECONOMIES, Annals of the University of Oradea, Economic Science Series, 23(1): 175-186.
Miyakoshi, T., Kono, T., Terasawai, K., Koyanagi M. (2010), Economic growth and public expenditure compozition: Optimal adjustment using the gradient method, Japanese Economic Review, 61(3): 320-340.
Mo, P.H. (2001), Corruption and Economic Growth, Journal of Comparative Economics, 29(1): 66-79.
Moral-Benito, E. (2007), Determinants of Economic Growth: A Bayesian Panel Data Approach, CEMFI Working Papers Paper, 0719: 1-32.
Munnell, A.H. (1992), Infrastructure investment and productivity growth, Journal of Economic Perspectives, 6: 189–198.
Murphy, K.M., Shleifer, A. And Vishny R.W. (1993), Why Is Rent-Seeking So Costly to Growth?, American Economic Review, 84(2): 409-414.
Narayan, P. K.. and Prasad, A. (2008), Electricity consumption-real GDP causality nexus: Evidence from a bootstrapped causality test for 30 OECD countries, Energy Policy, 36: 910-918.
NAZMI, N. and RAMIREZ, M. D. (1997), PUBLIC AND PRIVATE INVESTMENT AND ECONOMIC GROWTH IN MEXICO, Contemporary Economic Policy, 15(1): 65–75.
Nordhaus, W.D. and Tobin, J. (1972), Is Growth Obsolete?, NBER Chapters: Economic Research: Retrospect and Prospect.
North, D. (1990), Institutions, Institutional Change and Economic Performance, Cambridge: Cambridge University Press.
Odedokun, M.O. (1997), Relative effects of public versus private investment spending on economic efficiency and growth in developing countries, Applied Economics, 29 : 1325-1336.
Odhiambo, N.M. (2009), Electricity consumption and economic growth in South Africa: A trivariate causality test, Energy Economics, 31(5): 635– 640.
Odhiambo, N.M. (2014), Electricity Consumption, Exports, and Economic Growth in the Democratic Republic of Congo: An ARDL-Bounds Testing Approach, Journal of Developing Areas, 48(4): 189-207.
OECD. (2012), Redefining "Urban": A New Way to Measure Metropolitan Areas, Paris: OECD Publishing.
OECD. (2011), Government at a Glange 2011, Paris, OECD Publishing.
Ottaviano, G.I.P. (2008), Infrastructure and economic geography: An overview of theory and evidence, EIB Papers, 13: 8–35.
Paci, R. and Marrocu, E. (2014), Tourism and regional growth in Europe, Papers in Regional Science, Wiley Blackwell, (93):S25-S50, November.
Parent, O. and LeSage, J.P.(2012), Determinants of Knowledge Production and Their Effects on Regional Economic Growth, Journal of Regional Science, 52(2): 256-284.
Pareto, V. (1971), Manual of political economy. New York, Scholars Book Shelf.
Paul, S. and Bhattachrya, R.B. (2004), Causality between energy consumption and economic growth in India: A note on conflicting results, Energy Economics, 26: 977-983.
Paul, S. and Sridhar, K.S. (2015), The Paradox of India’s North-South Divide in Lessons from the States and Regions, New Delhi: Sage Publications India.
Payne, J.E., Ewing B.T. and Mohammadi, H. (2006), Wagner’s Hypothesis: New Evidence from the US Using the Bounds Testing Approach, Ott, A.F. and Cebula, R.T. (Eds.), The Elgar Companion to Public Economics, Empirical Public Economics, Edward Elgar, UK, 2006, pp. 30-40.
Pellegrini, G. (2013), Measuring the effects of European Regional Policy on economic growth: A regression discontinuity approach, Papers in Regional Science, 92 (1): 217-233.
Pereira, A. M. (2001), On the effects of public investment on private investment: what crowds in what, Public Finance Review, 29: 3–25.
Pereira, A.M. and Xu, Z. (2000), Export Growth and Domestic Performance, The Review of Economic Studies, 67(1): 60-73.
Pesaran, M. (2004), General diagnostic tests for cross section dependence in panels, CESifo Working Paper Series No. 1229; IZA Discussion Paper No. 1240.
Pieroni, L. (2009), Military expenditure and economic growth, Defence and Peace Economics, 20 (4): 327-339.
Poi, B. and Wiggins, V. (2001), Testing for panel-level Heteroskedasticity and Autocorrelation, Stata Corp LP. Retrieved from http://www.stata.com/support/faqs/stat/panel.html.
Pol, P. (2003), The economic impact of the high-speed train on urban regions, European Regional Science Association ERSA, conference papers 03 397.
Polèse, M. (2005), Cities and national economic growth: A reappraisal, Urban Studies, 42(8): pp.1429-1451.
Pollard, S.K., Piffaut, P.V. and Shackman, J. (2012), Government Expenditure, Economic Growth and Conditional Convergence: What Does the Penn World Table 7.0 Tell Us?, Academy of Business Research Journal , Vol II.
Porter, M. E. (1990), The Competitive Advantage of Nations, New York: The Free Press.
Prasetyia, F. (2013), The role of government and private sector on economic development in ASEAN 5, Journal of global business and economics, 7(1): 54-67.
Pritchett, L. (2001), Where has all the education gone?, World Bank Economic Review, 15: 367-91.
Puga, D. (2002), European regional policies in light of recent location theories, Journal of Economic Geography, 2: 373-406.
Putnam, R., Leonardi, R. and Y Nanetti, R. (1993), Making Democracy Work: Civic Traditions in Modern Italy , Princeton, NJ: Princeton University Press.
Ramsey, F.P. (1928), A Mathematical Theory of Saving, The Economic Journal, 38(152): 543-559.
Ramsey, J. B. (1969), Tests for Specification Errors in Classical Linear Least Squares Regression Analysis, Journal of the Royal Statistical Society, Series B 31 (2): 350–371.
Raufhon, S. (2015), Democracy and economic growth: The role of intelligence in cross-country regressions, Intelligence, 50: 228-234.
Reardon, T., Berdegue´, J. A., and Escobar, G. (2001), Rural non-farm employment and incomes in Latin America: Overview & policy implications, World Development, 29(3): pp.395–409.
Ricardo, D. (1817), On the Principles of Political Economy and Taxation (1 ed.), London: John Murray.
Robinson, S. (1972), Theories of Economic Growth and Development: Methodology and Content, Economic development and Cultural Change, 21(1): 54-67), Chicago: The University of Chicago Press.
Rock, M.T. and Bonnett, H. (2004), The Comparative Politics of Corruption: Accounting for the East Asian Paradox in Empirical Studies of Corruption, Growth and Investment, World Development, 32 (6): 999– 1017.
Rodriguez, F. and Rodrik D. (1999), Trade Policy and Economic Growth: a Skeptic’s Guide to the Cross-national Evidence, NBER Working Paper 7081, Cambridge MA, National Bureau of Economic Research.
Rodríguez-Pose, A. and Villarreal Peralta, E.M. (2015), Innovation and Regional Growth in Mexico: 2000–2010, Growth and Change, Wiley Blackwell, 46(2): 172-195.
Rodriguez-Pose, A., Psycharis, Y., and Tselios, V. (2012), Public investment and regional growth and convergence: Evidence from Greece. Papers in Regional Science, 91(3), 543-568.
Rodrik, D (ed.) (2003), In Search of Prosperity: Analytic Narratives on Economic Growth, Princeton, NJ, Princeton University Press.
Rodrik, D. (1999), Where did all the growth go? External shocks, social conflict and Growth collapses, Journal of Economic Growth 4(4); 385–412.
Rodrik, D. (2000), Institutions for High-quality Growth: What they are and How to Acquire them, Studies in Comparative International Development ,35: 3–31.
Rodrik, D., Subramanian, A. and Trebbi, F. (2002) “Institutions rule: the primacy of institutions over geography and integration in economic development”, NBER Working Paper, no. 9305.
Romer, P.M. (1986), Increasing Returns and Long-Run Growth, Journal of Political Economy, 94:1002-1037.
Romer, P.M. (1987), Growth Based on Increasing Returns Due to Specialization, American Economic Review, 77(2): 56-62.
Romer, P.M. (1990), Endogenous Technological Change, Journal of Political Economy, 98(I):S71-S102.
Roodman, D. (2009), How to do xtabon2: An introduction to difference and system GMM in Stata, The Stata Journal, 9(1):86-136.
Rosenthal, S.S. and Strange, W.C. (2004), Evidence on the nature and sources of agglomeration economies, J.V. Henderson & J.F. Thisse (eds.), Handbook of Regional and Urban Economics, Amsterdam: Elsevier: 2119–2171.
Rostow, W.W. (1960), The Stages of Economic Growth: A Non-Communist Manifesto, Cambridge, Cambridge University Press.
Sachs, J. and Warner, A. (1995), Economic Reform and the Process of Global Integration, Brooking Papers on Economic Activity, 1: 1-118.
Sala-I-Martin, X., Doppelhofer, G., and Miller, R. (2004), Determinants of long-term growth: A Bayesian averaging of classical estimates (BACE) approach, American Economic Review, 94(4): 813–835.
Samuelson, P.A. (1948), Economics: An Introductory Analysis, First Edition, New York, McGraw-Hill.
Sánchez-Romero, M. (2013), The role of demography on per capita output growth and saving rates, Journal of Population Economics, 26(4): 1347-1377.
Sarel, M. (1996), Nonlinear effects of inflation on economic growth, IMF Staff Papers, 43(1): 199-215.
Say, J-B. (1834), A Treatise on Political Economy (sixth American ed.), Philadelphia, Grigg & Elliott.
Schumpeter, J. (1934), The Theory of Economic Development, New York: Oxford University Press.
Scully, G. (1988), The Institutional Framework and Economic Development, Journal of Political Economy, 96 (3): 652–62.
Shapiro, S. S. and Wilk, M. B. (1965), An analysis of variance test for normality (complete samples), Biometrika, 52 (3–4): 591–611 (p. 593).
Shera, A., Dosti, B. and Grabova, P. (2014), Corruption impact on Economic Growth: An empirical analysis, Journal of Economic Development, Management, IT, Finance and Marketing, 6(2): 57-77.
Simuț, R. (2012), The Impact Of Investments, Exports, And Openness On Economic Growth. A Comparative Study On The East European Countries, Annals of the University of Oradea, Economic Science Series, 1(2): 420-425.
Simuț, R. and Meșter, I. (2014), AN INVESTIGATION OF COINTEGRATION AND CAUSALITY BETWEEN INVESTMENTS, EXPORTS, OPENNESS, INDUSTRIAL PRODUCTION AND ECONOMIC GROWTH: A COMPARATIVE STUDY FOR THE EAST EUROPEAN COUNTRIES, Annals of the University of Oradea, Economic Science Series, 23(1): 369-378.
Singh, T. (2011), International trade and economic growth nexus in Australia: a robust evidence from time-series estimator, The World Economy, 34(8): 1348-1394.
Smith, A. (1776), An Inquiry into the Nature and Causes of the Wealth of Nations, New York: Modern Library, 1937.
Solarin, S.A. and Shahbaz, M. (2013), Trivariate causality between economic growth, urbanisation and electricity consumption in Angola: Cointegration and causality analysis, Energy Policy, 60: 876-884.
Solow, R. (1956), A contribution to the theory of economic growth, The Quarterly Journal of Economics, 70(1): 65-94.
Solow, R. (1957), Technical change and the aggregate production function, Review of Economics and Statistics, 39 (3): 312–320.
Soytas, U. and Sari, R. (2003), Energy consumption and GDP: Causality relationship in G-7 countries and emerging markets, Energy Economics, 25: 33–37.
Spiezia, V. and Weiler, S. (2007), Understanding Regional Growth, The Review of Regional Studies, 37(3): 344 – 366.
Sridhar, K.S. (2010), Determinants of city growth and output in India, Review of urban and regional development studies, 22(1): 22-38.
Sultan, Z.A. and Haque, M.I. (2011), The Estimation of the Cointegration Relationship between Growth, Domestic Investment and Exports: The Indian Economy, International Journal of Economics and Finance, 3(4): 226-232.
Svensson, J. (2003), Who Must Pay Bribes and How Much?, Quarterly Journal of Economics, 118(1): 207–30.
Swan, T.W. (1956), Economic Growth and Capital Accumulation, Economic Record, 32: 334-361.
Szarowská, I. (2012), THE CYCLICALITY OF GOVERNMENT EXPENDITURE AND WAGNER'S LAW- CASE OF CZECH REPUBLIC, SLOVAKIA, HUNGARY, BULGARIA AND ROMANIA, Scientific Papers of the University of Pardubice. Series D, Faculty of Economics & Administration, 18(24):188-198.
Tang, C.F. and Abosedra, S. (2014), The impacts of tourism, energy consumption and political instability on economic growth in the MENA countries, Energy Policy, 68: 458-464.
Tekin, R.B. (2012), Economic growth, exports and foreign direct investment in Least Developed Countries: A panel Granger causality analysis, Economic Modelling, 29: 868–878.
Theissen, M.F. et al. (2013), Regional competitiveness and smart specialization in Europe. Place-based development in international economic networks, Cheltenham: Edward Elgar.
Timbeau, X. et al. (2014), iAGS 2015- Third Report, Independent Annual Growth Survey.
Toffler, A. (1981), The Third Wave, London: Pan Books Ltd.
Tolo, W.B.J. (2011), The Determinants of Economic Growth in the Philippines: A New Look, IMF Working Papers, WP/11/288.
Torres-Reyna, O. (2007), Panel data analysis fixed and random effects using Stata (v.4.2), Princeton: Princeton University.
Tsouhlou, A. and Mylonakis, J. (2011), Public Expenditure, Public Sector Size and Growth: The European Union Marked Structural Differences, Review of European Studies, 3(2): 33-51.
Usai, S. (2011), The geography of inventive activities in OECD regions, Regional Studies, 45(6):711-731.
Usai, S. and Paci, R. (2003), Externalities and Local Economic Growth in Manufacturing Industries. In: B. Fingleton, eds., European Regional Growth, pp. 293–323. Berlin: Springer.
Van den Berg, H. (2013), Growth theory after Keynes, part I: the unfortunate suppression of the Harrod-Domar model, The Journal of Philosophical Economics, VII(1): 1-23.
Van Der Gaag, N. and Beer, J. (2015), From Demographic Dividend to Demographic Burden: The Impact of Population Ageing on Economic Growth in Europe, Journal of Economic & Social Geography, 106(1): 94-109.
van Oort, F., de Geus, S. and Dogaru, T. (2015), Related Variety and Regional Economic Growth in a Cross-Section of European Urban Regions, European Planning Studies, vol. 23(6): 1110-1127.
Voigt, S. (2009), How (Not) to Measure Institutions, MAGKS Papers on Economics, Philipps-Universität Marburg, Faculty of Business Administration and Economics, Department of Economics (Volkswirtschaftliche Abteilung).
Wagner, A. (1991), Staat in nationalokonomischer Hinsicht (The state in Economic Perspective), Handworterbuch der Staatswissenschaften. 3rd ed. Jena: Verlag Gustav Fischer, pp. 727–739.
Wahab, M. (2004), Economic growth and government expenditure: evidence from a new test specification, Applied Economics, 36: 2125–2135.
Walras, L. (1954), Elements of pure economics; or, The theory of social wealth, London, Published for the American Economic Association and the Royal Economic Society by Allen and Unwin.
Whittaker, J., Whitehead. C. and Somers, M. (2005), The neglog transformation and quantile regression for the analysis of a large credit scoring database, Applied Statistics, 54(5): 863 – 878.
Wooldridge, J. (2002), Econometric Analysis of Cross Section and Panel Data, The MIT press.
World Bank Publications. (2004), Beyond Economic Growth: An Introduction to Sustainable Development, Second Edition, Washington, no. 14865, pp: 205.
World Scientific. (1992), Economic Sciences, 1969-1980: The Sveriges Riksbank (Bank of Sweden Prize in Economic Sciences in Memory of Alfred Nobel), Sweden, pp: 87-100/ 442.
World Travel & Tourism Council. (2012), World Travel and Tourism Council report 2012.
Yang, H.Y. (2000), A note on the causal relationship between energy and GDP in Taiwan, Energy Economics, Vol: 309-317.
Yildirim, J., Sezgin, S., Öcal, N., (2005), Military Expenditure and Economic Growth in Middle Eastern Countries: A Dynamic Panel Data Analysis, Defence and Peace Economics, 16(4): 283–295.
Ynikkaya, H. (2003), Trade Openness and Economic Growth: a cross country empirical investigation, Journal of Development Economics, 72: 57-89.
Yoo, S. (2006), The causal relationship between electricity consumption and economic growth in ASEAN countries, Energy Policy, 34: 3573–3582.
Young, A.A. (1928), Increasing Returns and Economic Progress, The Economic Journal, 38(152): 527-542.
Yu, B., Fan, S. and Saurkar, A. (2009), Does composition of government spending matter to economic growth?, Proceedings of the International Association of Agricultural Economists Conference, Beijing.
Yu, E. S. H. and Choi, J.Y. (1985), The causal relationship between electricity and GNP: An international comparison, Journal of Energy and Development , 10: 249–272.
Zhang, C. and Xu, J. (2012), Retesting the causality between energy consumption and GDP in China: Evidence from sectoral and regional analyses using dynamic panel data, Energy Economics, 34(6): 1782-1789.
Zou, Y. (2006), Empirical studies on the relationship between public and private investment and GDP growth, Applied Economics, 38 : 1259-1270.
Zuanna, G.D. and Micheli, G.A. (eds.), Strong Family and. Low Fertility: A Paradox? New Perspectives in Interpreting Contemporary Family and Reproductive behaviour, Netherlands: Springer Netherlands.
Copyright Notice
© Licențiada.org respectă drepturile de proprietate intelectuală și așteaptă ca toți utilizatorii să facă același lucru. Dacă consideri că un conținut de pe site încalcă drepturile tale de autor, te rugăm să trimiți o notificare DMCA.
Acest articol: “LUCIAN BLAGA” UNIVERSITY OF SIBIU THE FACULTY OF ECONOMICS DOCTORAL STUDIES DOCTORAL THESIS Scientific coordinator: Prof. PhD. ILEANA TACHE PhD… [303389] (ID: 303389)
Dacă considerați că acest conținut vă încalcă drepturile de autor, vă rugăm să depuneți o cerere pe pagina noastră Copyright Takedown.