A Study on the Contribution of 12 Key-Factors to the Growth Rates of the Region of the East Macedonia-Thrace (EMTH) by Using a Neural Network Model

A Study on the Contribution of 12 Key-Factors to the Growth Rates of the Region of the East Macedonia-Thrace (EMTH) by Using a Neural Network Model

E. Stathakis (Kavala Institute of Technology, Greece), M. Hanias (University of Athens, Greece), P. Antoniades (Kavala Institute of Technology, Greece), L. Magafas (Kavala Institute of Technology, Greece) and D. Bandekas (Kavala Institute of Technology, Greece)
DOI: 10.4018/ijpmat.2012010102
OnDemand PDF Download:
No Current Special Offers


This study gives a new methodological framework regarding the measuring of the contribution of some key-factors on the regional growth rate and forecasting the future development rates, based on Neural Network Models (NN Models). It’s a serious attempt to study the contribution of twelve key-factors to the change of the Regional Gross Domestic Product of the Region of East Macedonia -Thrace during a long-term of growth process, by creating and using a suitable Neural Network Model. Specifically, twelve key-factors, time functioned in the period 1991-2008, are studied for the first time, in order to be investigated, scientifically, firstly their % contribution to growth of the regional economy and secondly, to be predicted how much the (Regional Growth Domestic Product) RGDP-under certain conditions-will be changed. It’s a NN Model with inputs the twelve key-factors in order to be evaluated and measured, at the best precise, their percentage contribution to the RGDP. The model and results can be found further into the article.
Article Preview

Introduction, Brief Literature, And Some Explanations

Considering, in theoretical basis the regional problem, we can say that the two predominant economic growth approaches, we mean the neoclassical model and the model of endogenous growth, are in force even to the regional growth. Romer (1986) and Rebelo (1991), argue that the absence of convergence and the presence of disparities and inequalities across economies of various spatial levels throughout the world represent strong evidence against the neoclassical model and in favor of the theories of endogenous growth. But, some newer economists, fans of the neoclassical economic school, argued that the neoclassical growth model amended with partial capital mobility and reinforced by technological diffusion, explain better the issue of convergence/divergence in region level (Baro, 1991). But undoubtedly, the overexploitation of some resources in certain areas and the inactivation of some resources in other areas, in a status of widened open and transacting economy, lead toward to be formed developmental disparities and inequalities into and among areas or regions and, the result is what interprets well the growth gap between north-south.

In Greece, the regional problem is enough significant and complicated, since the inequalities and disparities among the thirteen regions are greater than to other euro zone countries. An analysis of data derived from databanks of Hellenic Statistical Service (HSS) shows that up to end of ‘80s the polar pattern was predominant in Greece (Petrakos & Saratsis, 2000). Also, from the analysis of such data, we conclude that the dynamic developing sectors are concentrated in three areas, greater area of Athens, Thessalonica, and urban-industrial complexes along the growth axis between Athens-Thessalonica. Also, local development advantages were present in regions around the above complexes. Despite the state policies for a new regional strategy aiming to reduce the disparities among the thirteen regions, even today, the different developmental levels continue to be existed, since five regions from the thirteen ones have, at least, a 20% lower GDP/capita than the eight others in average terms, a growth divergence greater than anyone or a European society can accept.

In order anyone can satisfactory explain when, why and how the convergent or divergent growth among regions of a country take place, he will have to collect, process and present multiannual statistical series -without significant time lags pertaining to regional socioeconomic conditions. Of course, it is much more significant to collect data which are more spatially specific than the statistical series already available at the national level. Take in account all the previous, we do remark the difficulties anyone meets to collect statistical data at regional level in Greece, since to this country enough trustworthy and uniform statistical data at regional level have been collected by the NSSG for the last fifteen years only.

So, our attempt to collect analyze, and present such time series of eighteen years, 1991-2008, was very great deal. Even, the data used by us were adapted and formed in a uniform type in order to be able to be compared for the first time. Another key part of this work is to look behind the factors of success for explanation of EMTH region growth and spatial performance, using very specialized method from Econo(mo)Physics.

Econo(mo)physics is an interdisciplinary research field, applying theories and methods originally developed by physicists in order to solve very complex problems in economics. Usually, such models have used in financial economic and more specifically are used in the prediction of future behaviour of shares in the largest Stock Exchange Markets. In this work, another particularity is that we have used a NN Model able to produce results using even small time-series- only twenty years, although this model is suitable for the case of many inputs. So, it was needed to do many adaptations in order the model to become functioning and reliable.

A neural network is a computational model that is loosely based on the neuron cell structure of the biological nervous system. Given a training set of data, the neural network can learn the data with a learning algorithm. In this research, the most common algorithm, back propagation, is used. Through back propagation, the neural network forms a mapping between inputs and desired outputs from the training set by altering weighted connections within the network.

Complete Article List

Search this Journal:
Open Access Articles
Volume 7: 2 Issues (2019)
Volume 6: 2 Issues (2018)
Volume 5: 2 Issues (2017)
Volume 4: 2 Issues (2016)
Volume 3: 2 Issues (2015)
Volume 2: 2 Issues (2014)
Volume 1: 4 Issues (2012)
View Complete Journal Contents Listing