4.1. Evolution of Technology Infrastructure in Terms of Speed and Coverage
In 2010, DAE was created as one of the flagship initiatives of the Europe 2020 strategy and included specific broadband coverage targets stretching to 2020: Universal broadband coverage of speed of at least 30 Mbps by 2020 and 50% of households should have broadband subscriptions of 100 Mbps or more by 2020 [
9]. Examining coverage levels by the individual speed categories, at the end of 2021, it can be observed in
Table 2 and
Figure 1 that four countries from the eastern EU member states, namely Bulgaria, Czechia, Hungary, and Romania, exceeded the EU-27 average percentage of 89.9% for the fixed broadband networks, which are capable of providing them with actual download speed of at least 30 Mbps.
For the other two countries, the values of the BC30 indicator are significantly below the European average, respectively 82.3% for Slovakia and 77% for Poland. This evolution was driven by the recorded growth in NGA coverage and the technological advancement provided by a higher number of very high-speed digital subscriber line (VDSL) networks capable of supporting a 30 Mbps download speed.
Coverage of networks supporting at least 100 Mbps at the EU-27 level was 82.1% at the end of 2021. In
Figure 2, it can be observed that Romania, Czechia, Bulgaria, and Hungary exceeded this percentage, and Poland (69.2%) and Slovakia (75.4%) were below this value. This is a result of the growth of two broadband types: VDSL2 vectoring coverage (connection to a VDSL2-enabled cabinet or exchange, and the vectoring solution is applied to receive at least 100 Mbps of download speed) and fiber to the premises (FTTP) coverage (connection to a fiber service without requiring the construction of new fiber infrastructure, which is available for connection within reasonable time and cost limits).
The overall fixed broadband coverage category has been designed to provide a measure of progress in the deployment of fixed broadband access technologies, which are capable of providing households with broadband services of at least 2 Mbps of download speed. Four technologies make up the overall fixed broadband coverage: Digital subscriber line (DSL), cable, FTTP, and fixed wireless access (FWA).
In the year 2021, the European average of households with an FBC connection was 97.9%. Referring to this average, it can be seen in
Table 3 that only two countries exceed this value, namely, the Czech Republic (99.9%) and Hungary (98.4%). Slovakia and Bulgaria are close to the European average with a percentage of over 97% of households with a fixed connection, while Romania (94.1%) and Poland (89.7%) are making considerable efforts to reach the targets proposed within DAE.
The NGA category comprises technologies capable of delivering a service speed of at least 30 Mbps. This category must be improved considering the main objective of DAE to have complete coverage of European households at this speed by 2020. In
Figure 2, it can be observed that this coverage registered the lowest values in the analyzed countries. Therefore, the analysis of the combination category constitutes an evaluation of the rollout of the relevant technologies and progress toward this goal. The European average for NGA is 90.1%. Of the six countries included in this research, only two are below this value (Slovakia—84.3% and Poland—78.2%).
The average LTE coverage metric is an important indicator since it has also been included as one of the components of the Connectivity dimension of the Digital Economy and Society Index (DESI) [
48,
49]. The European average of mobile broadband technologies coverage exceeds the percentage of 99.8% of all households. Of the six countries included in the analysis, only Slovakia (98.4%) is below this value and Hungary which is only 0.01 percentage points away from reaching the European average.
The statistical data analysis regarding the level of broadband coverage and the download speed confirms the first hypothesis formulated, H1, regarding the fact that in at least two of the analyzed states (Poland and Slovakia), the technological infrastructure requires improvements to reach the targets established within the DAE.
4.2. Contribution of Broadband Technologies to Tax Collection
Figure 3 points out the relationship between total receipts from taxes and social contributions (REV) on the horizontal axis and broadband technologies indicators on the vertical axis. It indicates that the relationship is positive: More coverage and broadband speed lead to more revenues from taxes and contributions to the central government. Furthermore, the figure proves that the relationship is not linear. From the graph, we can observe two areas of concentration, one denser in the value range from EUR 10,000 to 80,000 million and another less dense in the value range from EUR 120,000 to 180,000 million.
Figure 4 shows the presence of outliers in the datasets of our variables. As we mentioned in the previous section, it is necessary to identify these outliers, and if they are many, it is recommended to use robust regression for a more robust analysis [
50]. As it can be seen, all the variables in the model present one or more outliers. Even if there are remedial measures for influential outliers and non-normal distributions, they are not always effective for large amounts of contamination and are not easy to automate. Occasionally, it is essential to keep outliers in the data and not remove them completely. Removing data will reduce the sample size, which is not good from an estimation point-of-view.
To test the stationarity of the data we used LLC unit root test. As we notice from
Table 4, all data are stationary at level or at first difference according to the
p-value associated to t-statistics.
After examining the stationary properties of the data, the study employs a correlation matrix to confirm that the data for the current investigation is free of the problem of collinearity.
Table 5 presents the results of the correlation coefficients, which highlight that there is no multicollinearity problem in the data since the coefficient of correlation among any two variables is less than 1.00.
In proceeding to apply robust least squares regression on the proposed variables and using the S-estimator as the estimation method, we obtained the results in
Table 6.
The table reveals the significant impact of broadband speed from a statistical point-of-view, in order that the use of connections with low download speed leads to a reduction in tax revenues, while the use of technological networks with high download speed positively influences the volume of revenues from taxes and fees to the central budgets. The importance of broadband coverage speed can also be explained through the lens of users’ behavior; the majority of people are reluctant to use difficult applications which process commands slowly and often encounter blockages during the execution of some commands. This confirms the second hypothesis, H2, regarding the influence of broadband speed coverage on the volume of revenues from taxes and contributions to the central government.
On the one hand, regarding the coverage infrastructure, among the three variables, the FBC variable is statistically insignificant. On the other hand, a strong influence of NGA coverage on the collection of taxes and fees can be observed, as well as the use of mobile broadband which has a positive impact on the payment of taxes to the central institutions. Taking into account that there is at least one mobile connection in every household in the analyzed countries [
9], we can observe the significant positive impact of LTE on the growth of revenues from taxes and fees collection. The results in
Table 6 confirm hypothesis, H3, regarding the significant positive influence of the degree of broadband coverage on the degree of tax collection.
The bottom portion of the output displays the R-squared and adjusted R-squared and indicates that the model accounts for roughly 50% of the variation in the constant-only model. The Rn-squared statistic of 1710.806 and the corresponding p-value of 0.00 indicate a strong rejection of the null hypothesis that all non-intercept coefficients are equal to zero.
To strengthen the results obtained by applying the S-estimator to the proposed model, we additionally performed the Granger causality test [
45,
50], the results of which are specified in
Figure 5.
We notice that regarding the direction of causality between the variables, the results of the Granger test indicate a direct causality on the performance of the fiscal policy from the higher speed broadband (BC100), as well as from the modern technological infrastructure (NGA), which can also support higher data transmission speeds.
Causality relationships were also identified between the independent variables: From NGA and LTE to BC100, an absolutely normal relationship, since the development of a modern infrastructure that supports high transmission speeds automatically leads to an increase in speed; from NGA to FBC, also a normal relationship as an increase in new generation infrastructure will lead to the reduction in rudimentary types of infrastructure.