The Influence of Digital Economy and Society Index on Sustainable Development Indicators: The Case of European Union

Abstract

The digital economy plays a vital role in promoting sustainable development. Out of different measurement indices, this research uses the DESI dimension, i.e., connectivity, human capital, the use of internet services, the integration of digital technology, and digital public services, to investigate the impact on the promotion of SGDI in the European Union countries. Previous research studies investigated the indirect impact of the DESI dimension on SGDI in different countries and regions. In this research, we investigate the direct impact of DESI dimensions on SGDI by using panel regression modeling. The results show that DESI sub-dimensions influence SGDI differently. Connectivity, human capital, and the use of internet services have more influence on SGDI compared to the integration of digital technology and digital public services. However, the impact is negative in most cases, but this is in line with the previous studies in other regional studies. Thus, the current research paper reveals that standard views on the influence of the digital economy are not always true. Policymakers need to make the necessary amendments while implementing each DESI dimension on any level for better promotion of SGDI.

1. Introduction

The definition of the digital economy has evolved alongside underlying technologies over time. Over the past several decades, the information and communication technology (ICT) sector has undergone rapid development, from microelectronics in the 1940s to the birth of the computer in the 1960s, the introduction of the internet in the 1990s, and most recently, blockchain, Artificial Intelligence (AI), and robotics. Correspondingly, new sectors based on evolving technologies have merged, such as e-commerce, Fintech, and driverless cars (IMF working paper 2019). With the continuous emergence of technology advancement and digitalization, societies and businesses are fundamentally altered, and being digital is in line with the current situation [1]. In the context of the development of the digital economy, digitalization is conducive to enterprise development and legitimacy [2,3].

Sustainable models of economic growth are increasingly becoming inseparable from digital technology in countries worldwide. The digital economy provides a new impetus and direction for sustainability [4]. Sustainable development is one of the vital goals of the world announced by the United Nations in 2015 and is expected to be achieved by 2030. Meanwhile, the Council of the European Union [5] adopted Sustainable Development Goals (SDGs) for the European Union (EU), of which the progress is measured by the sustainable development goals index (SDGI). Hence, it could be stated that both sustainable development and the digital economy are priorities of countries across the world. However, it is not clear whether there are any direct relationships between those two phenomena.

As mentioned above, sustainable development is measured through the SDGI; however, the assessment of the digital economy is unclear. Different countries (regions) use different indices to assess the impact of the digital economy on sustainable development. There are several techniques for its measurements proposed by the Organisation for Economic Co-operation and Development (OECD) and the World Economic Forum (WEF) named the network readiness index (NRI) [6,7]. Similarly, the digital adoption index (DAI) was also constructed to evaluate the digitalization of countries. However, as of this moment, statistical data for DAI are only available up to 2016. As the world is promptly transferring to the digital economy, it cannot be used for the analysis of the present situation because of the limited availability of up-to-date data. The most used tool is the digital economy and society index (DESI), which has been employed by many researchers [4,6]. Similarly, in the context of China, the main index evaluation systems are the digital economy Index (DEI) introduced by China’s information and communication research institute and the China digital economy index (CDEI) of the Caixin think tank [8]. The establishment of the international index system has laid a foundation for the construction of a digital economic comprehensive evaluation system in many economies. However, DESI remains very popular in European studies to investigate the indirect impact of DESI dimensions on sustainable development growth indicators. However, it has not been used to evaluate the direct impact of the digital economy on sustainable development, which is scarce in the literature. Therefore, in this research paper, we use the digital economy and society index (DESI) dimensions to investigate the impact of the DESI dimensions on sustainable growth development growth indicators in EU countries.

2. Theoretical Background

The digital economy and sustainable development are widely analyzed topics in the scientific literature [4,9,10]. The theoretical background of this study can be based on three different types of scientific literature: First, the construction of a digital economy development index, and there are several indices that have been used to analyze the matter of this study [6,7]; second, the sustainable economic impact of the digital economy [1,4]; third, empirical research on the development of a digital economy. For instance, Savastano et al. [11] examined it from the perspective of a company, while Cook et al. [12] focused on digital agriculture for a sustainable food system. It is worth mentioning that plenty of articles deal with digitalization in a circular economy [13,14,15,16], and the circular economy could be treated as a part of sustainable development. In other words, the digital economy has permeated almost all spheres of life, as well as sustainable development issues. Because of that, the digital economy’s relationship with sustainable development should be highlighted. The development of a digital economy will have a major impact on the sustainable development of an economy and society [9]. The development of a digital economy would have an influence on the development of a regional economy. However, the development of the digital economy would also bring about a huge digital divide [17].

The digital economy can cultivate more entrepreneurial opportunities by influencing market scale, knowledge spillovers, and factor combinations, which lead to sustainable development. It also enriches entrepreneurial resources by accelerating information interaction and the dissemination of ideas, thereby promoting entrepreneurial activities, which play an important role in sustainable development [18]. According to UNESCO, sustainable development is “development that meets the needs of the present without compromising the ability of future generations to meet their own needs”. Academics and researchers agree that the digital economy could be treated as an instrument to extend economic growth [19,20], which, in turn, is part of the sustainable development concept. However, not all researchers focus on such a general phenomenon as economic growth but rather on more narrow fields of the economy. The digital economy can be defined in a narrow or broad sense. In recent years, the explosion of new technologies and their rapid application have spurred another wave of discussion on the digital economy. The narrow definition refers to the ICT sector only, including telecommunication, internet, IT services, hardware and software, etc. The broad definition includes the ICT sector parts of traditional sectors that have been integrated with digital technology. G20 uses this broad concept and has defined the digital economy as “a broad range of economic activities that includes using digitized information and knowledge as the key factor of production, and modern information networks as the important activity space”. The introduction of this new technological development translated into economic growth one way or another.

There are scientists analyzing the digital economy in terms of business and its readiness for sustainable development. For instance, Andreeva et al. [11] stated that the digital economy could be the basis of the financial stability of companies. Financial stability itself has a significant impact on economic growth, the reduction in poverty, and income inequality [21]. Similarly, Feng et al. [22] agree that the stronger financial stability, the higher the level of sustainable development. Moreover, financial stability could help overcome financial crises in businesses [23], which undoubtedly leads to constant sustainable development. In other words, it could be stated that the digital economy could be seen as a driver of sustainable development, which could be reached through companies’ financial stability.

Other scientists concentrate on investigating the impact of different variables on economic growth, such as trade, taxation, and e-commerce. All these could be treated as a part of the digital economy. For the purpose of managing those issues, digital technologies are used. For instance, Abendin and Duan [24] claim that trade positively impacts economic growth. It is worth mentioning that trade is related to taxation issues, and low levels of taxation could stimulate economic growth. Moreover, it is worth mentioning that e-commerce is hand-in-hand with trade and taxation. Hence, it could be claimed that the abovementioned phenomena indirectly influence sustainable development.

Another topic that deserves a broader analysis is the interface between information and communication technologies (ICT) diffusion and economic growth. ICT undoubtedly could be treated as a part of the digital economy and could be investigated as a variable with both direct and indirect impacts on sustainable development. However, there are only a few articles investigating that issue, most of them focused on ICT’s impact on economic growth, and there are different points of view on the mentioned influence. According to Cheng et al. [25], ICT could have a positive impact, or no impact, on economic growth, which depends on the level of a country’s development, i.e., in high-impact countries, this effect is supported, while it is not feasible in middle and low-income countries. These findings are supported by Kurniawati [26], who conducted research on OECD (i.e., high-income) countries and concluded that those countries had reached positive and significant economic development from high ICT penetration. In contrast, Pradhan et al. [27] investigated middle-income countries, and the results revealed that ICT infrastructure development stimulates economic growth in the long run. However, there are different views on ICT impact, e.g., ICT has a negative impact on labor productivity, which is a part of economic growth. Thus, it could be stated that ICT could both positively and negatively influence economic growth, and in this way, contribute to sustainable development in different directions. As a new engine of economic growth, the digital economy has become increasingly important for regional economic structural upgrading and coordinated development through modern technology and innovation [28,29]. To summarize, it could be claimed that the digital economy could be treated as a development trend and is a basis for sustainable economic growth, hence overall sustainable development.

Despite the fact that there is plenty of scientific studies on the indirect linkage between the digital economy and sustainable development, almost all of them state that the digital economy promotes sustainable development. However, there is a limited number of research studies that measure the direct influence of the digital economy on sustainable development, and this is what we are going to investigate in this paper. For that purpose, the digital economy and society index (DESI) and SDGI are treated as the digital economy and sustainable development measurement factors. The impact of DESI on SDGI (sustainable development goals index) is calculated using panel regression modeling.

3. Model and Methodology

Panel regression modelling was used to analyze the impact of the digital economy and society index (DESI) on the sustainable development goals index (SDGI) in EU countries. Panel regression was used to analyze the panel data, also called cross-sectional data. The panel regression was conducted five times for each DESI dimension, and DESI sub-dimensions were chosen as variables, as presented in

Table A1. DESI dimensions and sub-dimensions.

	DESI Dimensions	DESI Sub-Dimensions
Digital Economics and Social Indicators (DESI)	Connectivity	Overall fixed BB take-up
		At least 100 Mbps fixed BB take-up
		Fast BB (NGA) coverage
		Fixed Very High Capacity Network (VHCN) coverage
		4G Coverage
		Mobile BB take-up
		5G readiness
		Broadband price index
	Human capital	At least Basic Digital Skills
		Above basic digital skills
		At least basic software skills
		ICT Specialists
		Female ICT specialists
		ICT graduates
	Use of Internet Services	People who never used the Internet
		Internet Users
		News
		Music, Videos and Games
		Video on Demand
		Video Calls
		Social Networks
		Doing an online course
		Banking
		Shopping
		Selling online
	Integration of Digital Technology	Electronic Information Sharing
		Social media
		Big data
		Cloud
		SMEs selling online
		Commerce turnover
		Selling online cross-border
	Digital Public Services	e-Government Users
		Pre-filled Forms
		Online Service Completion
		Digital public services for businesses
		Open Data

Source: European Commission [47].

in Appendix A.

The regression model was used on four years’ worth of panel data from 2018 to 2021 for a sample of 28 countries (including the UK) to determine the DESI indicators’ influence on the SDGI. The following empirical models were developed for each DESI dimension to estimate the direct impact of DESI on SDGIs.

i.

Connectivity: Connectivity refers to measuring both the supply and demand of fixed and mobile broadband. The Commission continues to monitor connectivity throughout EU countries. The econometric model used to measure the direct impact of connectivity on SGDI is estimated by the following model.

Model 1: SDGI_it = β₁1a1_it + β₂1a2_it + β₃1b1_it + β₄1b2_it + β₅1c1_it + β₆1c2_it + u_it

(1)

ii.

Human capital: Human capital refers to the monitoring of ‘internet user skills’ and ‘advanced skills and development’ across EU countries to ensure people are equipped for the digital era. The econometric model used to measure the direct impact of human development on SGDI is estimated by the following model.

Model 2: SDGI_it = β₁2a1_it + β₂2a2_it + β₃2a3_it + β₄2b1_it + β₅2b2_it + β₆2b3_it + u_it

(2)

iii.

Use of Internet Services: The use of internet services refers to the updates in the ICT sector, including in research and development, which contributes significantly to the EU economy. The following econometric equation is designed to measure the impact of this DESI dimension.

Model 3: SDGI_it = β₁3a1_it + β₂3a2_it + β₃3b1_it + β₄3b2_it + β₅3b3_it + β₆3b4_it + β₇3b5_it + β₈3b6_it + β₉3c1_it + β₁₀3c2_it + β₁₁3c3_it + u_it

(3)

iv.

Integration of Digital Technology: The integration of digital technology refers to the new advancement of technologies that are integrated into business and e-Commerce. This integration of digital technologies and its impact on SGDIs is estimated via the following econometric Equation (4).

Model 4: SDGI_it = β₁4a1_it + β₂4a2_it + β₃4a3_it + β₄4a4_it + β₅4b1_it + β₆4b2_it + β₇4b3_it + u_it

(4)

v.

Digital Public Services: Digital public services refer to the monitoring of indicators of digital public services in EU countries to ensure that citizens and governments are enjoying the full potential of this new technology. Digital public services’ relation is estimated via the following econometric Equation (5).

Model 5: SDGI_it = β₁5a1_it + β₂5a2_it + β₃5a3_it + β₄5a4_it + β₅5a5_it + u_it

(5)

3.1. Panel Data Econometrics

Panel or longitudinal data are a dataset that comprises repeated measures of a given sample or same variable over time, such as individuals, firms, persons, cities, or countries, which are observed at numerous points in time, such as days, months, quarterly, or years before and after treatment [30,31]. It is known by different names in the literature, such as pooled data, micro panel data, a combination of time series and cross-sectional data, longitudinal data, and cohort analysis. Panel datasets are divided into micro and macro panel data and balance and unbalanced panel datasets. Micro panel datasets are those for which the time dimension, $T$, is of less importance than the dimension of the individual $N$, while macro panel data are those for which the dimension of time, $T$, and individual, $N$, are similarly important. On the other hand, a panel dataset is considered balanced if it has the same time intervals, $t=1, \dots , T,$ for each cross-sectional observation, and for unbalanced, the time $T$ dimension is specific to each individual.

The general model for panel data can be written as,

$$Y_{it}={\beta }_0+{\beta }_1{X}_{it}+v_{it}+u_{it}$$

(6)

$$t=1,2,\dots ,T$$

$$i=1,2,\dots ,N$$

where $Y_{it}$, is the dependent variable, ${\beta }_0$ is the intercept and is independent of i and t, ${\beta }_1$ is the $K\times 1$ vector of the unknown parameter to be estimated, $X_{it}$ is the $1\times k$ vector of explanatory variable observations, and $u_{it}$ is the disturbance or error term.

The fundamental class of models that can be estimated using panel techniques may be written as the following function:

$$Y_{it}=f\left(X_{it}^{’}\beta \right)+{\delta }_i+{\gamma }_t+{ϵ}_{it}$$

(7)

The most important case involves a linear conditional mean specification, so that we have the following function:

$$Y_{it }=\alpha +X_{it}^{’}\beta +{\delta }_i+{\gamma }_t+{\epsilon }_{it}$$

(8)

where $Y_{it}$ is the dependent variable and $X_{it}$ is a $k-vector$ of regressors, and ${ϵ}_{it}$ are the error term for $i=1, 2, 3,\dots ,M$ cross-sectional units observed for dated periods, i.e., $t=1, 2, 3, \dots ,T.$ The α parameter indicates the overall constant in the above-stated model, while ${\delta }_i$ and ${\gamma }_t$ represent the cross-sections for each period-specific effect (random or fixed). Random effect and fixed effect are the two most used methods for the analysis of cross-sectional datasets.

However, generally, there are three basic estimation models of panel data:

i.

Common Effect Model

In this model, both the intercept and slope coefficients are constant across time and space. This model can be written as follows:

$$Y_{it}={\beta }_0+{\beta }_1{x}_{1it}+{\beta }_2{x}_{2it}+e_{it}$$

(9)

ii.

The Fixed Effect Model

In this model, the intercept varies across individuals, but the slope coefficients are constant over individuals, as well as over time. It is written as follows:

$$Y_{it}={\beta }_{0i}+{\beta }_1{x}_{1it}+{\beta }_2{x}_{2it}+e_{it}$$

(10)

This model is known as the fixed-effect model.

iii.

The Random Effect Model

The fixed-effect model assumes that each group (firm) has a non-stochastic group-specific component to $y$. Including dummy variables (if any) is a way of controlling for unobservable effects on $Y_{it}$. However, these unobservable effects may be stochastic (i.e., Random). The random effects model attempts to deal with this:

$$Y_{it}={\beta }_0+{\beta }_1{X}_{it}+V_i+{\epsilon }_{it}$$

(11)

Here, the unobservable component, v_i, is treated as a component of the random error term. V_i is the element of the error, which varies between groups but not within groups. Ε_it is the element of the error, which varies between groups and time.

We assume that:

The average of both errors is zero.

$$E\left(V_i\right)=E\left({\epsilon }_{it}\right)=0$$

(12)

The variance of both errors can be defined as

$$E\left(V_i^2\right)={\sigma }_v^2,E\left({\epsilon }_{it}^2\right)={\sigma }_{\epsilon }^2$$

(13)

Furthermore, both are homoscedastic.

Both of the error terms are independent of each other

$$E\left({\epsilon }_{it}{v}_j\right)=0\forall i,t,j$$

(14)

No autocorrelation and no across-group correlation

$$E\left({\epsilon }_{it}{\epsilon }_{js}\right)=0ift\ne s or i\ne j$$

(15)

$$E\left(v_i{v}_j\right)=0if i\ne j$$

(16)

Both are independent

$$E(v_i{X}_{it})=E({\epsilon }_{it}{X}_{it})=0$$

(17)

We could also introduce an error component that varies across time periods but not across groups—two-way random effects. Estimation of the random effects model cannot be performed by OLS—instead, a technique known as generalized least squares (GLS) must be used. The decision between the random effect model and the fixed-effect model will be checked using the Hausman test.

iv.

The Hausman Test

The Hausman test is the standard method used in observed panel data analysis in order to distinguish between the fixed effect and random effect models. This technique is also used by other studies [32]. On the basis of this, the tests based on the assessment of two sets of parameter estimates are known as Durbin-Wu-Hausman tests, or DWH. For ease of presentation, we will refer to the Hausman test [33].

The general procedure can be described as follows. For instance, we suppose that we have two kinds of estimators for a certain parameter “θ”, of dimension $K\times 1$. One of them, ${\widehat{\theta }}_r$, is robust, i.e., consistent under both the null hypothesis $H_0$, and the alternative $H_1$; the other, ${\widehat{\theta }}_e$, is capable and reliable under the null hypothesis $H_0$, but inconsistent under the alternative $H_1.$ The difference between the two estimators is then used as the basis for testing of the analysis. The Hausman test indicates that, under appropriate assumptions, under $H_0$, the statistic h based on $\left({\widehat{\theta }}_R-{\widehat{\theta }}_E\right)$ has a limiting Chi-squared distribution [32]:

$$h=\left({\widehat{\theta }}_r-{\widehat{\theta }}_e\right){\left[\widehat{Var}\left({\widehat{\theta }}_r-{\widehat{\theta }}_e\right)\right]}^{-1}\left({\widehat{\theta }}_r-{\widehat{\theta }}_e\right)~x_K^2$$

(18)

If this statistic value lies in the upper tail of the chi-square distribution, we cannot accept our null hypothesis $H_0$: If the variance matrix is consistently analyzed and estimated, the test will have power against any other substitute, under which ${\widehat{\theta }}_r$ is robust and ${\widehat{\theta }}_e$ is not robust.

The Hausman procedure also shows that, again, under certain appropriate assumptions,

$$Var\left({\widehat{\theta }}_r-{\widehat{\theta }}_e\right)=Var\left({\widehat{\theta }}_r\right)-Var\left({\widehat{\theta }}_e\right)$$

(19)

It is well known that the assumptions used are sufficient but not necessary. Further, it may be convenient to estimate

$$Var\left({\widehat{\theta }}_r-{\widehat{\theta }}_e\right)=Var\left({\widehat{\theta }}_r\right)-2Cov\left({\widehat{\theta }}_r,{\widehat{\theta }}_e\right)+Var\left({\widehat{\theta }}_e\right)$$

(20)

This may be more robust, and the trade-off between its robustness and power should be considered appropriately. The Hausman specification test for error components provides guidance on whether $E=\left(X_{it}{V}_i\right)\ne 0$.

The key idea is if $E=\left(X_{it}{V}_i\right)\ne 0$, then the inconsistent random effect estimator and the consistent fixed effect estimator converge to different estimates. If $E=\left(X_{it}{V}_i\right)=0$, then the unobserved heterogeneity is uncorrelated with X and does not create a bias.

Random effect and fixed effect models are both consistent.

Two reliable estimators presenting statistically significantly different estimates would be surprising. If the fixed-effect model is indefinite, then the Hausman test has very weak explanatory power and cannot rule out even large biases. If the fixed-effect model is very precise, then the Hausman test has very good power, but we gain little benefit from switching to the more efficient random effect. If the fixed-effect model is somewhat precise, then the Hausman test can direct us away from using random effects in the existence of an outsized bias, but there is still room for substantial efficiency gains in switching to random effect.

v.

Panel unit-root test

In order to avoid spurious regression and to check the stationarity of the variables, the unit-root test should be employed. There is plenty of unit-root tests presented in the literature. For example, Saglam and Ampountolas use Levin–Lin–Chu, Im–Pesaran–Shin, Fisher-type augmented Dickey–Fuller, and Philips–Perron panel unit-root tests [34]. Despite a wide variety of unit-root tests, the most commonly used is the Augmented Dickey–Fuller (ADF) test, which extends the Dickey–Fuller test and is used in the current study. The null hypothesis of the unit-root test is the “unit root exists (in other words, the data is not stationary”. The ADF testing procedure to test the unit root hypothesis is the following [35]:

$$\Delta y_t={\theta }_0+{\gamma }_{0}t+{\gamma }_1{y}_{t-1}+{{\displaystyle \sum }}_{i=0}^p{\theta }_i\Delta y_{t-1}+{\epsilon }_t$$

(21)

where:

• $y_t$—the variable in period t.
• $\Delta y_{t-1}—y_{t-1}$ − $y_{t-2}$.
• ${\epsilon }_t$—i.i.d. disturbance with a mean of 0 and variance of 1.
• t—the linear time trend.
• p—the lag order.

In the present research, all the p-values of ADF are less than 0.1 (our chosen significance level is 0.1); hence, the null hypothesis for all the cases is rejected. The time series is stationary, so according to Zuo and Arbor, if the variable is stationary, then the variable is I(0) [36].

3.2. Cointegration Test

In order to move on with the panel regression analysis, the cointegration of the panels should be taken into account. Two series are cointegrated when they have common trends, i.e., they are similar in some sense. The null hypothesis is that series are not cointegrated. In order to test the series for cointegration, the Augmented Dickey–Fuller (ADF) cointegration test was employed. The results are presented in the table below (see

Table 1. ADF cointegration analysis of DESI dimensions.

DESI Dimensions	Test	Statistic	p-Value
Connectivity	ADF	3.2548	0.0000
Human Capital	ADF	5.1248	0.0000
Use of Internet	ADF	6.2478	0.0000
Integration of Digital Technology	ADF	6.9547	0.0000
Digital Public Services	ADF	2.8412	0.0024

).

As can be concluded based on the results provided in Table 1, all the panels are cointegrated and, hence, the research could be further conducted.

4. Results and Discussion

Fixed-model regression modelling has been chosen to investigate the impact of DESI on the promotion of SDGI. All in all, five statistical models based on DESI dimensions have been developed and estimated accordingly. The results of each DESI dimension are presented in the following tables.

Table 2. Estimated results of Model 1—connectivity.

Indicators	Estimated Results (p-Value)
Overall fixed BB take-up	−0.4549 (−0.0942) *
At least 100 Mbps fixed BB take-up	−0.6841 (0.1174)
Fast BB (NGA) coverage	0.1625 (0.1539)
Fixed Very High Capacity Network (VHCN) coverage	−05421 (0.1074) *
4G Coverage	0.2461 (0.0039) ***
Mobile BB take-up	0.1924 (0.0074) ***
5G readiness	−0.2843 (0.2539)
Broadband price index	0.2195 (0.1174)
R-Square	0.3054
Adjusted R-Square	0.3187
F-statistic	<0.0001

Note: * (***) indicates that variable is significant at the 10% (1%) level of significance.

presents the estimated results of the first DESI sub-dimensions of connectivity. The estimated results show that there are four statistically significant variables in our first estimated model. The value of the R-square is equal to 0.30154 and the F-statistic is less than 0.0001. These tests show that the model could be used, and although the R-square is not high, it is still in the acceptable range of 30% of variance explained, and we could interpret the economic data. Hence, it could be stated that the sub-dimension including the overall fixed BB take-up, fixed VHCN coverage, 4G coverage, and mobile broadband take-up are significant variables in the model. Changes in these sub-dimensions lead to changes in the direct effect of DESI dimensions on SGDIs. The results support the view of scholars that fixed broadband and 4G, which are considered new-generation technology (NGN), could influence the country’s sustainable development [37]. However, on the other hand, researchers have explored that using 4G and preparing for 5G has shockingly increased energy consumption. This could lead to a challenge for sustainable development [38]. Similarly, regarding the mobile broadband take-up, it positively influences SDGI. These findings are new in our research. Based on the non-availability of literature on this topic, the obtained result could contribute to the scientific literature on drivers of sustainable development. The mobile platform is a significant part of our lives; hence, many services might be accessed through this platform. In other words, the connection between people, businesses, and other services is constant, which may help to manage daily life challenges faster and with less monetary expenses, which leads to sustainability.

Table 3. Estimated results of Model 2—human capital.

Indicators	Estimated Results (p)
At least Basic Digital Skills	−6.4698 (−0.1672)
Above basic digital skills	−0.2495 (0.2154)
At least basic software skills	0.4251 (0.1559)
ICT Specialists	−6.6181 (0.005) ***
Female ICT specialists	0.2495 (0.1039) *
ICT graduates	−2.1734 (0.0354) **
R-Square	0.4714
Adjusted R-Square	0.4798
F-statistic	<0.0001

Note: * (**) (***) indicates that variable is significant at 10% (5%) (1%) level of significance.

presents the results of our second model of the study using sub-dimension variables of human capital. Here we have three statistically significant variables—ICT specialists, Female ICT specialists, and ICT graduates. The population (men and women) with enough knowledge of ICT are contributing more towards sustainable development. However, in the current scenario, the coefficients are negative. Those two variables with a negative effect on sustainable development are not surprising, as the negative and positive impacts of ICT on sustainable development are widely discussed in the scientific literature. For instance, Ya’u states that ICT could be harmful to sustainable development as it is an antecedent of economic inequality [39]. The ICT sector is one of the drivers of overproduction, which may cause unsustainable consumption. In other words, the higher the level of ICT specialists and graduates, the higher the amount of overproduction that might be achieved, which, in turn, has a negative effect on SDG. However, while discussing that issue, it should be stated that SDG9 focuses on promoting sustainable industrialization and fostering innovation [40]. In other words, ICT is a significant part of our lives; hence, it could not be removed. Still, countries’ governments should consider the negative aspects of ICT while developing strategies for achieving sustainable development. A clear view of this problem should be ensured, especially when constructing university programs connected to ICT. The number of ICT specialists and graduates should be strictly stated by every country, as it would help manage the ICT’s negative impact on SDGs. The R-square is 0.4714, and the F-statistic p-value is less than 0.0001, which supports the model’s adequacy, and the result should be considered valid and requires the attention of policymakers, as stated above.

Table 4. Estimated results of Model 3—use of internet services.

Indicators	Estimated Results (p)
People who never used the Internet	−1.2546 (0.3584)
Internet Users	2.5186 (0.5142)
News	0.4687 (0.0011) ***
Music, Videos and Games	0.3781 (0.0001) ***
Video on Demand	−1.6582 (0.4168)
Video Calls	−2.6485 (0.8136)
Social Networks	0.4952 (0.1834)
Doing an online course	–0.901 (0.0001) ***
Banking	–0.661 (0.0596) **
Shopping	1.2589 (0.1085) *
Selling online	1.5942 (0.1534)
R-Square	0.5246
Adjusted R-Square	0.5312
F-statistic	<0.0001

Note: * (**) (***) indicates that variable is significant at the 10% (5%) (1%) level of significance.

presents the estimated results of another DESI sub-dimension. There are four significant indicators (variables) of the sub-dimension, two of which have a positive effect and two of which have a negative impact on SDGI. The value of R-square is equal to 0.5246, and the F-statistic p-value is less the 0.0001, which proves that the developed model is different from the null model and that the model could be used for analyzing this DESI dimension’s influence on SDGI. The estimated results show that news, music, videos, and games have a positive impact on sustainable development goal indicators because of the information provided through those channels. Nowadays, there is much information related to sustainable development issues, which psychologically affects the users. In other words, it encourages people to sort waste, for example, which changes their way of thinking. On the other hand, doing an online course and banking both negatively influence sustainable development. Learning online has taken over a significant part of our lives, especially during the COVID-19 pandemic, limiting people’s perception of the received information and reducing the intention to be sustainable. This is quite logical, as using online services increases the level of energy usage, which contradicts the sustainable development concept. The situation is the same with banking, as almost all banking procedures are held online and have become convenient at the cost of energy consumption.

Table 5. Estimated results of Model 4—integration of digital technology.

Indicators	Estimated Results (p)
Electronic Information Sharing	−1.2546 (0.3584)
Social media	−0.8245 (0.0001) ***
Big data	0.4687 (0.2451)
Cloud	0.5264 (0.4129)
SMEs selling online	−2.4588 (0.1268)
Commerce turnover	1.6485 (0.1136)
Selling online cross-border	0.1952 (0.1034) *
R-Square	0.3246
Adjusted R-Square	0.3311
F-statistic	<0.0001

Note: * (***) indicates that variable is significant at the 10% (1%) level of significance.

presents the estimated results of the integration of digital technology, the fourth DESI dimension using the fourth econometric model of our study. In the sub-dimension of the integration of digital technology, we have two significant variables. The econometric model indicates that social media negatively influences the SDGI. There is plenty of scientific research discussing this linkage. For instance, [41] claims that social media influences the patterns of consumption. It could promote higher consumption, which is not a concept of sustainable consumption. Herranz studied organizations’ communication levels. They found that the analyzed organizations’ presence in the principal social media networks (i.e., different social networks) is minimal and uneven [42]. Moreover, social media is a platform for social commerce (s-commerce), which enables customers’ participation in the sale of products and services. So, irresponsible traders could sell more goods than needed by using different marketing strategies, which would lead to unsustainable consumption. Responsibility is one of the core traits that sellers should have while using social media for their business. However, this is challenging to achieve, as responsibility could lead to a lower income. Still, there is a contradictory approach in the scientific literature. For example, Ortega mentions that social media is vital for business. It helps firms to acquire the necessary knowledge and capabilities, which, in turn, allow them to be more sustainability-oriented [43]. To summarize, it could be stated that social media harms sustainable development, especially in terms of sustainable consumption. However, in the case that the values of social media users change, it could become one of the drivers of sustainable development in the future and could promote selling across the border, which is the second significant variable in our findings. The R-square value of our model is 0.3246 and the F-statistic p-value is less than 0.0001, proving the model’s adequacy.

Table 6. Estimated results of Model 5—digital public services.

Indicators	Estimated Results (p)
e-Government Users	−1.1594 (0.3584)
Pre-filled Forms	−0.2054 (0.0011) ***
Online Service Completion	0.3541 (0.2945)
Digital public services for businesses	−0.4251 (0.0001) ***
Open Data	−1.9254 (0.2486)
R-Square	0.3122
Adjusted R-Square	0.3215
F-statistic	<0.0001

Note: *** indicates that variable is significant at the 1% level of significance.

shows the estimated results of Model 5. The value of R-square is equal to 0.3122, and the F-statistics p-value is less than 0.0001; consequently, the model could be interpreted and used. It could be stated that both significant variables—pre-filled forms and digital public services for businesses—negatively affect SDGI. The obtained results are not in line with the discussion in the scientific literature, as most scholars claim that digital public services may have a positive impact on sustainable development [44]. Still, the results show that the situation in the European Union is vice-versa. This could be because EU governments do not efficiently manage digital public services, which, in turn, leads to a high level of digital red tape, which is not sustainable from the very beginning.

Moreover, many of the European Union governments use a mixed model of public services, i.e., some of them are provided online and some are in the offline regime, which leads to a mess in coordinating public services. Hence, sustainability is not achieved. Moreover, digital public services could support environmental sustainability, but not social ones. Usually, people need personal contact to solve their problems. Of course, nowadays, some services provide online help (e.g., online chat or video calls), but not all of them. So, to summarize, it could be stated that, at present, sub-dimensions of the integration of digital technology negatively influence SDGI. Still, supposing the EU governments prepare a consistent plan for a complete transition to the digital platform and reduce digital bureaucracy, the signs of the presented variables could change to positive. So, constant monitoring of the changes is needed.

5. Summary Discussion and Conclusions

5.1. Summary Discussion

This research aimed to investigate the influence of the digital economy and sustainable development on the promotion of sustainable development indicators across 28 countries of the European Union including the United Kingdom. For this purpose, five models were used to investigate the relationship between each DESI sub-dimension, i.e., connectivity, human capital, the use of internet services, the integration of digital technology, and digital public services. A total of five econometric models were used to estimate the empirical results. The results reveal that the use of internet services followed by connectivity and human capital have an influence on the promotion of sustainable development growth indicators. The integration of digital technology and digital public services has a very limited significant impact on the promotion of sustainable development growth indicators. Complete details of empirical results have been provided in the following conclusion section.

5.2. Conclusions

The current research aimed to investigate the impact of the digital economy on the sustainable development goals index by employing panel regression modeling. First, the conducted literature review shows that many scientific articles present the indirect influence of different digital economic indicators on sustainable development goals [4,10,13]. It helps to identify the gap in scientific knowledge as no articles have analyzed the direct impact. The systematization of the available methodologies for measuring the digital economy confirms the only factors that could be used as digital economy factors are DESI sub-dimensions because only this index provides available and up-to-date data on the digital economy. For measuring the sustainable development level, SDGI was employed.

The research results revealed that, conversely to the scientific literature [39,45,46], the digital economy does not always positively influence sustainable development. Regarding the first dimension of DESI—connectivity—two sub-dimensions are assumed to be significant, and both are positive in terms of the effect on SDGI. They are 4G coverage and mobile broadband take-up. However, overall fixed broadband take-up and fixed VHCN are significant but negatively affect the SGDI. The second dimension of DESI is human capital. In this regard, ICT specialists and ICT graduates were significant variables that negatively influence SDGI. This is the most contradictory result because the standard view is that ICT positively contributes to sustainable development. However, those results should be taken into account by EU countries while preparing sustainable development strategies, as part of ICT in sustainable development is often overestimated. One of the sub-dimensions, female ICT specialists, significantly and positively affects SGDI. This shows that women play an inevitable role in promoting SGDIs if they are good at ICT.

The third DESI dimension analyzed was the use of Internet services. This has the most sub-dimensions, five of which are significant ones. They are as follows: News, social networks, online course, banking, and shopping. The first two have a positive impact on SDGI, and the last three have a negative impact. This could be explained by the fact that news and social networks are part of communication channels, positively affecting people’s well-being. At the same time, online courses, banking, and shopping do not promote socialization, which is a part of mental health. The fourth dimension was integrating digital technology, where we have two significant variable—social media and selling online, which was assumed to influence SDGI negatively. There is a contradictory point of view on this aspect in the scientific literature, but still, in the present study, the negative features of social media overcome the positive ones. However, it should be mentioned that social media and social networks are closely related. Still, social networks are mostly devoted to communication, as was alluded to above, while social media is communication without feedback, which is necessary to humans, especially in the COVID-19 pandemic. So, the producers of social media content should focus more on the audience’s needs and, based on that, develop user-friendly content. In this case, the direction of social media’s effect could change, i.e., constant monitoring is needed.

The last investigated DESI dimension was digital public services. Its significant sub-dimensions were pre-filled forms and digital public services for businesses; both coefficients were negative, hence the influences on SDGI were negative as well. As mentioned in the presentation of the results, it could be because not all of the EU countries manage the public services provided via the Internet appropriately. Still, both individuals and legal entities need personal contact as the transition to the digital economy is faster than the transition to online communication.

To summarize, it could be stated that the current study is unique because it investigates the direct impact of DESI on SDGI. However, the findings of our results have mixed effects on our main objective of study. Therefore, we could state that the digital economy partially contributes to SDGI. Hence, scholars should conduct future research with a more considerable period of data and geographical territories to validate the results.

The study has also certain limitations. This study takes the survey level as the research object across the European Union. However, this geographical territory is limited to EU countries only. The data can be improved in the future, and the number of samples can be expanded. Additionally, selected control variables should be added to investigate whether they are appropriate and assess their actual impact on the explained variables as much as possible. The addition of control variables into the model could explain these results in a more robust manner. Therefore, the selection of control variables should be more rigorous in future research studies. Finally, during the past several decades, the information and communication technology (ICT) sector has undergone rapid development, and these changes more often lead to changes in the development of the DESI dimensions and SGDI. We should continue to show solicitude for their development trends and changes in their relationship.