The Broadband Penetration in Europe

. The Broadband Penetration in Europe


Introduction
In this article we investigate the determinants of broadband penetration in 37 European countries in the period 2010 -2019.Broadband penetration is essential to promote the effective digitalization of the economy.In effect broadband penetration can sustain Internet of Things, telemedicine, smart cities and the promises of artificial intelligence and machine learning.
The ability to promote innovation technology especially in ICT sector is deeply associated with the efficiency and capacity of broadband penetration.Innovation technology is a driving force of economic growth such as in the Solow's model (Solow 1956, 65-94) in the theory of endogenous growth (Romer 1994, 3-22) and in the case of Schumpeterian economics (Schumpeter 1934).Furthermore, innovation technology has positive effects on human resources (Leogrande and Costantiello 2021, 240-259) and on the ability of firms to improve their sales either at an international level (Costantiello, Laureti, De Cristoforo and Leogrande 2021, 19).The investment in innovation technology is also positively associated to deeper expenditures in Research and Development either at a country level (Leogrande, Massaro and Galiano 2020, 72-101) either in the corporate dimension (Leogrande, Massaro and Galiano 2020, 186-201).Finally, innovation technology, in the sense of digitalization, also has positive effects on employment (Costantiello and Leogrande 2020, 166-187).For all these motivations it is important to analyze the question of broadband penetration for its role in promoting digitalization and innovation technology.Broadband penetration requires investments in fiber optics (Agboje, Adedoyin and Ndujiuba 2017, [1][2][3][4][5][6][7][8][9][10][11][12].There is a positive relationship between broadband penetration and the economic growth in Asia either in high income countries either 1 Strada Statale 100 km 18, 70010 Casamassima, Bari, Puglia, Italy, EU. in medium income countries (Kurniawati 2021), (Alam, Sultana and Rayhan 2019).The development of broadband penetration infrastructures improves total factor productivity and GDP growth (Kabaklarli and Atasoy 2019, 105-114).The investments in internet penetration is positively associated to economic development and growth in Africa (Kabaklarli and Atasoy 2019, 105-114).The creation of political economies that are specifically oriented to the creation of digital infrastructures in developing countries, such as in the case of Indonesia, shows that the effect on trade and ecommerce is structurally higher than the effect of human capital in terms of HDI (David and Grobler 2020, 1394-1418).
The development of telecommunication infrastructure also has a positive role in promoting digitalization and economic growth especially in high performing EU-OECD economies (Oztunc, Demirbas and Orhan 2019, [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16].The investment in broadband penetration in Asian countries is positively associated to GDP growth especially in the case of low inflation as a tool to promote job creation in the digital economy (Shodiev, Turayey and Shodiyev 2021).While on one side, families and individual customers could prefer mobile internet connection to broadband penetration, since these groups use internet for communication and entertainment, on the other side firms and business activities express a demand of broadband penetration to improve the profitability of their activities (Dwiardi 2020, 41-58).Broadband penetration has negative impact on social capital in terms of relationships and interactions since users tend to prefer virtual life to real life (Geraci, Nardotto, Reggiani and Sabatini 2018).
The diffusion of broadband penetration is positively associated to population density (Benseny, J., Töyli, J., Hämmäinen, and Arcia-Moret 2019, 139-155) Broadband penetration can have a positive effect on productivity and regional convergence such as in the case of Brazil (Jung and López-Bazo 2020).The investment in broadband penetration has positive effect on economic growth at regional level in Russia even in the theoretical and applicative framework of the Solow's model (Imasheva and Kramin 2020, 26-28).The mechanism of pricing can have a role in shaping the degree of broadband penetration among countries i.e. tariff diversity is positively associated to a widespread broadband penetration (Lange 2017, 285-312).The diffusion of broadband penetration improves the ability of citizens to actively participate in political debate in Europe (Hendriks Vettehen, Troost, Boerboom, Steijaert and Scheepers 2020, 967-987).
Since the diffusion of internet at a country level is positively associated to the number of customers and the quantity of digital contents then countries that are interested in promoting broadband penetration should also improve internet services (Na, Hwang and Kim 2020, 97-111).The prediction of broadband penetration in developing countries such as India can also offer an outlook on the ability to promote economic growth and an improvement in Human Development Index-HDI (Chattopadhyay and Majumdar 2019, 110-119).There also controversial relationship among broadband penetration, income inequality and digital divide in the sense that either income inequality can reduce broadband penetration and either that digital divide can weaken the relation between digitalization and economic growth (Aissaoui 2017, 799-808).
The impact of telecommunication infrastructure on economic growth is greater for developing countries in respect to developed ones (Maneejuk and Yamaka 2020).Positive relationships between broadband penetration and economic growth are present also in Latin America (Alderete 2017, 549-569).Similar relationships also characterize the Nigerian economy (Akinwale, Sanusi and Surujlal 2018, 129-142) and Middle East countries (Habibi and Zabardast 2020).Broadband penetration in South Korea has reached highest levels as a consequence of a co-opetition between public sector and the market and the presence of firms able to offer high qualitative services to customers that show a significative orientation towards product and service innovativeness (Shin and Koh 2017, 31-38).Empirical studies have shown that an increase in 10% of mobile broadband penetration can rise GDP per capita of an amount in the range 0,59-0,76% at a country level if the economy also have a developed industrial sector and the human capital necessary to create value added through digitalization (Bahia, Castells and Pedrós 2019).The article continues as follows: the second paragraph presents the econometric model, the third paragraph contains the cluster analysis, the fourth paragraph shows the results of the machine learning algorithms used to predict the value of broadband penetration, the fifth paragraph concludes.

The Econometric Model
We use data from European Innovation Scoreboard in the period 2010-2019 for 37 countries to estimate the following equation: (1) where i = 37 and t = [2010,2019].
The level of broadband penetration is defined as a relationship between a numerator that is the number of enterprises with a maximum contracted download speed of the fastest fixed internet connection of at least 100 Mb and a denominator that is based on the total number of enterprises.The level of broadband penetration is a measure that capture the ability of European countries to develop a digital economy based on e-commerce and internet of thing.The level of broadband penetration is negatively associated to: • Government procurement of advanced technology products: is a measure that consider the ability of a government to promote technological innovation through the public demand.There is a negative relationship between "Broadband penetration" and the level of "Government Procurement of Advanced Technology Products".This negative relationship can be interpreted in the sense that the public demand of technological innovation is not sufficient to promote a widespread diffusion of internet among European countries.• Sales impacts: is an indicator that is created as the sum of three different measures that are "Exports of medium and high-tech products", "Exports of knowledge intensive services", "Sales due to innovation activities".There is a negative relationship between Broadband penetration and sales impacts.This negative relationship can be interpreted considering that European countries, even when export, do not use digitalization and e-commerce and prefer traditional and classical methodologies.• Firm investments: is based on three different indicators that are R&D investments, non-R&D investments, and the investment of R&D to improve the skills of human resources.There is a negative relationship between firm investment and broadband penetration.The negative relationship means that the investment of the firms does not depend on the ability of a country to promote a widespread and efficient diffusion of internet network.• Opportunity-driven entrepreneurship: is a ratio that measures the relationship between persons that create enterprises to improve their income and persons that creates enterprises since they have no other opportunity to work.The fact that there is a negative relationship between broadband penetration and opportunity driven entrepreneurship means that the level of digitalization in the European Countries is not sufficient to promote the creation of new firms that operate for profits maximizing technological improvements in the digital economy.• Most-cited publications: is ratio in which at the denominator there is the number of scientific publication in the top 10 most cited publication worldwide and the denominator is the total number of publication.The fact that there is a negative relationship between broadband penetration and most-cited publication means that the impact of digitalization of the efficiency of research system is not sufficient to promote an increase in the quality of publications at an international level.• Rule of law: is a measure that capture the ability of economic agent to trust public institutions and their ability of governments to promote the quality of contract enforcement, property 4 rights, police services, and courts and act in the reduction of crime and violence.The negative relationship between broadband penetration and rule of law means that public institutions are not able to apply the digital services to justice and public order.Probably an improvement in the ability of government to efficiently use digital services to promote the role of law could have a positive effect in strengthening the of rule of law.The level of broadband penetration is positively associated with: § Enterprises providing ICT training: is ratio between the number of firms that provide training in ICT sector for their personnel and the total number of enterprises.The idea is that the greater the number of firms that train their human resources in ICT skills the greater the ability of firms to perform in the digital economy with positive effects either in the sense of profits either in the sense of GDP.The presence of a positive relationship between the broadband penetration and the number of firms that among European countries promote ICT skills for their personnel shows that the digitalization through infrastructure improve the level of human capital.Furthermore, this positive relationship can also better understood considering the role of MOOCs platforms and remote learning that are used to professional training in the digital economy.§ Innovative sales share: is a ration between the total turnover of new products and total turnover for all enterprises.The indicator shows the level of turnover associated either with new products either with the diffusion of new technologies.The existence of a positive relationship between innovative sales share and the level of broadband penetration shows that the investment in broadband penetration has effective a positive role in promoting new products and services and in the diffusion of new technologies in the markets.Innovative sales share can also be considered as a tool that shows the presence of a higher degree of customer innovativeness that can effectively sustain the marketization of new products and services.§ Intellectual assets: is a measure that capture three different forms of intellectual propriety rights that are PCT patent applications, trademark application and design application.The fact that there is a positive relationship between broadband penetration and the value of intellectual assets means that the digitalization of the economies among European Countries is positively associated to the promotion of a knowledge economy based on intangibles assets.The strengthening of the knowledge economy is either the condition either the output of the digitalization of the economy and can be efficiently realized in connection with an increase in the value of human resources at a country and international level.§ Knowledge-intensive services export: is ratio between the export of knowledge intensive services and the total export of services.The indicator can be considered as a measure of the ability of a country to export knowledge intensive services for example in the sector of transportation, finance, research and developments, ICT, media, and creative industries.The ability of a country to export knowledge intensive services is a consequence of the general degree of innovativeness of an economy.Furthermore, a higher degree of knowledge intensive services is also a signal of the ability of a country to actively participate in knowledge intensive global value chains.The positive relationship between broadband penetration and know-ledge intensive services exports means that effectively the production of high quality services in the knowledge economy requires a higher degree of digitalization.§ Turnover share SMEs: is a ratio between the turnover in SMEs and the total turnover of the business economy except financial activities.There is a positive relationship between broadband penetration and turnover share in SMEs.This means that the SMEs have greater benefits from broadband penetration in respect to big corporations.The economic opportunities for SMEs improve in the presence of higher level of broadband penetration.This result can be better understood considering the network effects and positive externalities offered by internet and the relative low cost of the ICT in respect to other innovations that are costlier such as in the case of the pharmaceutical and automotive sectors.The digital technologies are more sustainable for SMEs, either in the sense of financial resources either in the sense of know-how.Specifically, broadband penetration can promote SMEs' competitiveness and productivity creating the conditions for better results in terms of turnover at a national level among European countries.§ Innovation-friendly environment: is measure that capture the ability of firms to operate in an environment that can sustain innovation and digital firms.There is a positive relationship between broadband innovation and innovation friendly environment.The positive relationship can be in part explained considering that the presence of a digital business culture, the development of human resources that can work in ICT sectors with STEM skills, and even the possibility to have access to digital markets are effectively related to the presence of high level of broadband penetration.Furthermore, there is a recursive effect between the innovation-friendly environment and the broadband penetration: at the first stage the broadband penetration promotes a more developed innovation friendly environment and in a second stage the innovation-friendly environment expresses a demand for improvements in broadband penetration.

Clusterization
The clusterization model is realized with the application of the k-Means algorithm optimized with the silhouette coefficient.We found three different clusters that are: We can create an order of clusters based on the value of the median.Specifically, the value of the median of "Broadband Penetration" for the cluster 2 is equal to 197,48, the value of the median for 6 the cluster 1 is equal to 132,08, and the value of the median for the cluster 3 is equal to 118,25.Based on the value of the median we can build the same order: C2>C1>C3.

Machine Learning and Predictions
We use eight different typologies of machine learning algorithms to choose the best predictor of the level of "Broadband Penetration".We evaluate the algorithms based on their ability to maximize R-squared and to minimize "Mean Absolute Error", "Mean Squared Error", "Root Mean Squared Error", "Mean Signed Difference".To test the algorithms, we use data from the European Innovation Scoreboard for the period 2014-2021 for 37 countries.70% of the dataset has been used for the learning rate and the remaining 30% has been used to effectively predict the degree of "Broadband Penetration" in Europe.We found obtain the following ranking of algorithms: § Polynomial Regression with a payoff equal to 10; § Probabilistic Neural Network-PNN with a payoff equal to 11; § Artificial Neural Network-ANN with a payoff equal to 19; § Gradient Boosted Tree with a payoff equal to 20; § Tree Ensemble with a payoff equal to 24; § Linear Regression with a payoff equal to 27; § Simple Regression Tree with a payoff equal to 32; § Random Forest with a payoff equal to 37. Generally, the level of broadband penetration is expected to growth in Europe even if an heterogenous way among different countries.

Conclusions
In this article we estimate the determinants of broadband penetration in Europe.We use data from the European Innovation Scoreboard of the European Commission for 37 countries in the period 2010-2019.In the first paragraph we present a theoretical introduction considering the role of broadband penetration as a tool to promote economic growth.Empirical studies have shown the presence of a positive relationship between broadband penetration and economic growth either in developed either in developing countries.
In particular, the effect of broadband penetration on economic growth is greater in countries with less income inequalities and a higher level of human capital and investments in intangible assets such as R&D.In the second paragraph we estimate the level of broadband penetration in European countries using different econometric models i.e.We apply Panel Data with Fixed Effects, Panel Data with Random Effects, WLS, OLS and Dynamic Panel.The results obtained are consistent among different models and statistically significant.We found that the level of "Broadband Penetration" in Europe is positively associated to "Enterprises Providing ICT Training", "Innovative Sales Share", "Intellectual Assets", "Knowledge-Intensive Service Exports", "Turnover Share SMEs", "Innovation Friendly Environment" and negatively associated with "Government procurement of advanced technology products", "Sales Impact", "Firm Investments", "Opportunity-Driven Entrepreneurship", "Most Cited Publications", "Rule of Law".
The cluster analysis with k-Means algorithm optimized with the Silhouette Coefficient shows the presence of three clusters that substantially coincide, with some exceptions, with Northern, Central and Southern Europe.Finally, we apply eight machine learning algorithms to predict the level of "Broadband Penetration" in Europe.We find that the Polynomial Regression algorithm is the best predictor and that the level of the broadband penetration is expected to increase of 10,4%.Our results show that if policy makers are interested in promoting the economic growth then they should invest more in broadband penetration at a European level.

Figure 1 .
Figure 1.Synthesis of the main results of the econometric estimation realized with the usage of Panel Data with Fixed Effects, Panel Data with Random Effects, Pooled OLS, Dynamic Panel Data, and WLS.

Figure 2 .
Figure 2. The Silhouette Coefficient of the Three Clusters Analyzed

Figure 3 .
Figure 3. Ranking of algorithms in terms of performance for the maximization of R-squared and the minimization of errors.

Figure 4 .
Figure 4. Percentage variation of the value of Broadband penetration among different countries in Europe as predicted using the machine learning algorithm of Polynomial Regression.