Broadband, effective labor, and economic growth during the COVID-19 pandemic period: evidence from a cross-country study

2.1 The diffusions of ICTs and the digital divide in the world

The role of mass media in social and national development is one of the core research questions in communication studies. In the 1960s and 1970s, several seminal studies such as Lerner (1958), Schramm (1964), and Rogers (1962) contributed significantly to this research question. These initial attempts attracted many followers who conducted much more research on this topic. Their accumulated scholarship formed a significant paradigm—modernization paradigm—which is regarded as one of the research traditions of development communication studies (Servaes 2016). A central proposition of modernization paradigm is that mass media have positive influence on social and national development. This paradigm has been more focused on the contribution of information communication technologies to national development (ICT4D) during the past decades. As the Internet and broadband have become the fundamental telecommunication platforms, researchers nowadays are more interested in exploring the role of the Internet and broadband in national development.

The majority of research in ICT4D provided supporting evidence for the central proposition of modernization paradigm. That is, ICTs play an essential role in the national development. There are numerous studies that demonstrated the positive correlations between the penetrations of ICTs, especially the Internet and broadband, and national development variables such as the income, economic growth, human capital, labor productivity (e.g., Bagchi 2005; Baker et al. 2020; Billón et al. 2009; Bohlin et al. 2010; Choi and Yi 2009; Gruber and Koutroumpis 2011; Hargittai 2002; Vu 2019). On the other hand, researchers also noted that ICTs diffused unevenly across the world and explored the patterns and determinants of these uneven diffusions. These uneven ICT diffusions across the world result in the gaps in the adoptions and usage of ICTs among countries, societies, and communities, which is called digital divide. Many people believe the existence of digital divide impedes the positive functions of ICTs in the underdeveloped countries and societies, and hinders the development potentials of underprivileged people.

The concept of digital divide was coined in the mid-1990s and refined by the US Department of Commerce (Gunkel 2003). It denotes the social gap between the people who have access to ICTs and those who have not. The binary definition (has and has not) of digital divide is reductive as it does not reveal the multifaceted differences among people’s various ICT adoption and use behaviors. To better conceptualize this complex phenomenon, Organization for Economic Co-operation and Development (OECD) (2001) developed a refined concept: “digital divide refers to the gap between individuals, households, businesses and geographic areas at different socio-economic levels with regard both to their opportunities to access ICT and to their use of the Internet for a wide variety of activities” (p. 32). This refined definition not only refers to the first level of digital divide (the gap between “has” and “has not”), but also involves the second level of digital divide (the inequality in terms of resources and capacities in using ICTs) (Albarran 2013).

Broadband (high-speed Internet) enables advanced applications and spurs a new wave of innovations and market opportunities in the telecommunication industries. It also provides transformative possibilities for firms in other industries (Mack 2014). International Telecommunications Union (2003) argued the diffusion of broadband promotes economic growth as it encourages innovation, increases productivity, and attracts foreign investment. There are fundamentally two types of broadband: fixed broadband and mobile broadband. Srinuan et al. (2012) defined fixed broadband as the platform that is developed or upgraded from dial-up telephone modems to an Asymmetric Digital Subscriber Line (ADSL), or the platform that is connected through cable or fiber. Cheong and Park (2005) defined mobile broadband as the platform using wireless networks with the feature of “any place” and “anytime” accessibility. Although mobile broadband has diffused much faster than fixed broadband in many countries (International Telecommunication Union 2010; Molleryd et al. 2009), mobile broadband and fixed broadband are not only competitive but also complementary (OECD 2012). There is accumulated evidence that rural people with lower levels of income and education are more likely to adopt mobile broadband than urban people (e.g., Horrigan and Duggan 2015; Puschita et al. 2014).

2.2 ICTs, labor productivity, and economic growth

ICTs have been regarded as the driving force of economic productivity for a long time. Jipp (1963), for example, tested the relationship between income per capita and telephone density using a dataset of 48 countries. This seminar study was followed by many other studies that investigated the impacts of other ICTs (e.g., computer, Internet, mobile phone) on economic growth (e.g., Choi and Yi 2009; Gruber and Koutroumpis 2011). As the Internet becomes a major telecommunication backbone in the contemporary society, the contribution of the Internet to economic growth has become a major component of the ICT4D literature. Researchers conducted multiple empirical studies to estimate the impacts of Internet on economic productivity at the country level (e.g., Badran 2012; Ghosh 2017; Koutroumpis 2009), the regional level (e.g., Crandall et al. 2006; Jung and López-Bazo 2020), and the firm level (e.g., Farooqui and van Leeuwen 2008; Haller and Lyons 2019). Although many of these empirical studies demonstrated that the Internet has positive influences on productivity at various levels, several studies did not find any statistically significant results (e.g., Haller and Lyons 2015).

There are a number of studies that demonstrate broadband penetration or network size has significant impact on economic growth (Czernich et al. 2011; Gruber et al. 2014; Koutroumpis 2009; Vu 2019). There are a few studies that examined the impact of mobile broadband on economic growth. For example, Katz and Jung (2021) found that mobile broadband has higher impacts on economic growth than fixed broadband based on the data of 139 countries. Gbahabo and Ajuwon (2019) found that mobile broadband penetration impacts economic growth positively in the Nigerian economy.

The Internet and other ICTs affect economic productivity via multiple mechanisms. Prior research explored these mechanisms from three different perspectives (Zhang 2021). The first perspective argues ICTs affect both capital productivity and labor productivity, and therefore should contribute to total factor productivity (TFP) (e.g., Haller and Lyons 2019). The second perspective posits that ICTs work as a specific capital that increases the amount of capital per labor, a process called capital deepening (e.g., Farooqui and van Leeuwen 2008). The third perspective holds ICTs work as the tools that help people learn skills and knowledge (e.g., Akerman et al. 2015).

Human capital is an economic concept that represents the sets of skills and knowledge people have that can improve labor productivity (Becker 1964). Education and health are believed as two major components of human capital (Becker 2007). ICTs, especially broadband, are nowadays the essential tools that help people accumulate knowledge and skills. Scholars also conducted empirical research to investigate the specific effects of ICTs on human capital accumulation. Akerman et al. (2015), for example, found the Internet improved the productivity of skilled workers, but had a negative impact on unskilled workers. Meanwhile, Haller and Lyons (2015) found these impacts were nonexistent or non-significant.

Although the literature shows inconsistent findings regarding the impacts of ICTs on economic productivity and human capital, broadband plays a unique role in improving labor participation in the COVID-19 pandemic. Specifically, broadband plays a crucial role in helping people engage in productive work during the pandemic period. Millions of people adopted a working-from-home mode to accommodate the restrictions of the containment measures. This telework mode is heavily relied on broadband. A survey of 8672 adults in the United States during the COVID-19 pandemic period showed more people in higher income quintiles (second, middle, forth, and top) worked at home rather than were not able to work. In the top quintile, 71% of respondents worked at home, and only 19% respondents were unable to work (Reeves and Rothwell 2020). This new telework mode has also been adopted in many other countries. This new working mode would not have been viable if broadband had not diffused fast across the world. Without this working mode, many countries would have more economic losses during the pandemic.

Autor et al. (2003) argued ICTs have different impacts on different types of tasks. Specifically, they argued that ICTs substitute for workers in performing routine tasks, which are repetitive and predictable tasks such as picking or sorting, repetitive assembly, record-keeping, and calculation; and complement workers in executing nonroutine tasks, which require for problem-solving and communications skills such as responding to discrepancies, improving production processes, and coordinating and managing the activities of others. Based on this study, it is reasonable to argue that broadband helps improve the productivity of white-collar workers more than blue-collar workers. Specifically, during the COVID-19 pandemic period, when people have to work at home due to the restrictions of the containment measures, white-collar works benefit more than blue-collar workers from broadband connection as many of the nonroutine tasks can be performed via broadband networks while the assembly line work cannot be performed at home.

Another important issue in examining the impacts of broadband on economic growth is the network effect, i.e., the value of the network, specifically the broadband network, increases when more people use this network. In other words, the larger size the broadband network, the more contribution it has to economic growth. This argument was supported by several studies in the literature which demonstrated that the size of broadband network has a significant impact on economic growth rate (e.g., Koutroumpis 2009; Vu 2019).

2.3 Epidemic/pandemic, COVID-19, and economic growth

The public health crises, especially the epidemic and pandemic, have negatively impacted economic growth as these infectious diseases are detrimental to public health, deteriorate human capital, and impede productive activities. There are numerous studies in the literature that explored the impacts of epidemic or pandemic on economic growth. Prior to the COVID-19 pandemic, researchers investigated how several epidemics affected the economic growth in some countries. For example, Gallup and Sachs (2000) investigated the relationship between poverty and Malaria for tropical and subtropical countries, and found that the countries with severer Malaria suffered from lower economic growth and experienced deeper poverty. Holtkamp et al. (2013) estimated the negative impact of the swine fever epidemic on the US economy from 2005 to 2010, and found the average annual economic loss was about 664 million US dollars. Katz et al. (2020) studied the impacts of the 2003 SARS pandemic on economic growth in 178 countries and found countries with higher broadband adoption were able to counteract, to some degree, the effects of the outbreak.

Compared to these epidemics, the COVID-19 pandemic spread over much wider geographical areas, infected a much greater number of people, caused much higher numbers of deaths, and created much more social disruptions and economic losses. This pandemic exceeds the 1918 influenza (Spanish flu) in terms of the affected areas, the number of infected people, and the losses of life in many countries. Due to its tremendous impacts on the global economy, many economists conduct research to explore how this pandemic negatively impacted economic growth from either theoretical or empirical perspectives.

Several published studies explored the mechanisms of how COVID-19 affects economic growth in theory. Bischi et al. (2022), for example, combined epidemic SIS (Susceptible, Infectious, Susceptible) population model and Solow’s (1956) growth model to construct an economic growth model. This new model exhibits multiple steady state equilibria. One scenario is the economic growth ceases and the economy falls into a poverty trap when the propensity to save is low and the containment measures do not effectively stop the spread of pandemic. The opposite scenario is the economic growth returns to a steady state characterized by high capital accumulation when the containment measures work effectively. Likewise, Xiang et al. (2021) combined the epidemiological model and economic growth model and constructed a new interdisciplinary economic growth model. The equilibrium of this model shows the outbreaks of infectious diseases reduce labor supply and negatively affect economic output, and suggests that government’s effective public health policy can promote economic growth during the pandemic.

Many institutions and economists assessed the impacts of COVID-19 on the economic growth of the world and many economies. For example, The World Bank (2020) estimated a 5.2 percent contraction in global economy in 2020 due to the negative impacts of the pandemic in spite of the fiscal and monetary stimulate policies of governments. The International Monetary Fund (IMF) (2021) assessed the global annualized economic growth rate fell to around −3.2% in 2020. OECD (2020) estimated that the containment policies of governments resulted in a 20%–25% reduction in GDP in several OECD countries in 2020. Deb et al. (2022) used the data of 84 economies to assess the impact of containment measures on economic activities and found that these measures resulted in about a 10 percent loss in industrial production over 30 days following their implementation. Coccia (2021) found that a longer period (two months) of lockdown resulted in a deeper contraction of GDP (about −21%) than a shorter period (15 days) of lockdown did (about −13%).

Several studies were specifically focused on the impacts of COVID-19 on labor market. Anderton et al. (2021), for example, estimated that COVID-19 caused a sharp decrease (approximately 44%) in the number of people employed in the euro area. The negative impacts of COVID-19 on the labor markets in many economies were obviously observed and exhibited in the employment data. Interestingly, Bischi et al. (2022) pinpointed a significant finding that the total hours worked decreased by 16.8% and the average hours worked decreased by 14.3% in annual terms in 2020, which are much higher than the decrease of employment in euro area. This pattern is obvious when many people cannot fully engage in productive work during the COVID-19 pandemic due to various containment measures employed by the governments.

Nevertheless, broadband plays a positive role in helping people engage in productive work. Millions of people adopted a working-from-home mode to accommodate the restrictions of the containment measures. This telework mode is heavily relied on broadband. The adoption rate of this telework account for 20%–25% in developed economies and 20% in developing economies, and these people worked from home three to five times per week on average (Congressional Research Service 2021). Zhang (2021) explored this unique role of broadband Internet in helping people engage in production activities and found that broadband alleviated China’s economic losses during the first quarter of 2020 when the coronavirus spread across this country. The contribution of broadband to economic growth during the COVID-19 was also found by other studies. Katz and Jung (2021), for example, investigated the broadband’s contribution to mitigating the economic disruption of COVID-19 using the panel data of 121 countries, and found that the higher broadband penetration an economy has, the higher level of economic mitigation it was able to achieve. Another study of Katz and Jung (2022) found the similar result using the data of 50 states in the United States.

3.1 The model

Neoclassical growth models, including both exogenous and endogenous growth models, built a consensus on the three major sources of economic growth: capital growth, labor growth, and technological progress. This consensus became the foundation for growth accounting and laid the basis for many economic growth models. The economic model of this study is also developed from this consensus. It identifies the special function of broadband in labor growth during the pandemic. Specifically, broadband enables millions of workers to work from home and help economies survive under these tough restriction circumstances. This study captures this unique contribution of broadband by using a concept of effective labor.

Effective labor is the labor that effectively engages in work-related activities. In normal time, the effective labor is equal to employed labor and measured by the number of people employed. During the pandemic period, effective labor is different from the employed labor as many employed people cannot work due to the containment measures such as lockdown, quarantine, etc. Effective labor reflects the extent to which employed labor participates in productive activities. It is negatively influenced by the severity of the pandemic as well as the extents of the containment measures or suppression policies. As discussed above, this study argues that broadband helps people engage in productive activities. Therefore, the penetration levels of broadband should be positively associated with the effective labor during the COVID-19 pandemic period. Based on these arguments, this study proposes a new economic growth model for the pandemic period. The production function of this model is a revised Cobb-Douglas function:

where Y is total output of the production of an economy, L e is effective labor, K is capital. A is technology (total factor productivity).

The effective labor is defined by the following function:

where L is labor. δ is the coefficient of effective labor. In normal time, this coefficient is 1 ( δ = 1 ) as effective labor is not different from labor. That is: L e = L . During the pandemic period, this coefficient is less than 1 ( δ < 1 ) .

The definition of effective labor is based on the restrictions of COVID-19 containment measures on the labor participation and productivity. When there is no containment measure, the production participation and productivity of workers is normal, the effective labor is equal to the normal labor ( δ = 1 ) , which can be measured by the number of employed workers. During the COVID-19 pandemic period, many employed workers cannot go to workplace and participate in production activities as normal. This was what actually happened in many countries. Anderton et al. (2021) provided further evidence that the total hours worked decreased to a larger extent than the unemployment rates in European countries in 2020. Under such a special circumstance, the normal labor concept cannot accurately reflect labor participation. Effective labor concept captures this unique characteristic in labor participation during the pandemic period. Specifically, it reflects the loss of labor participation ( δ < 1 ) due to the various containment measures.

Broadband enables many people to work from home and helps sustain productive activities in many economies. This telework mode cannot be performed without broadband. This is how broadband affects effective labor. Specifically, for the scenario that there is no broadband connection, when people have to stay at home but cannot work from home, they cannot participate in production activities at all, and their productivity is zero. In this scenario, the effective labor is zero ( δ = 0 ) even people are employed. When there is broadband connection, people are able to participate in productive activities to some extent ( δ < 1 ) . It should be noted that the extent to which people participate in production and their productivity depends on many factors such as the nature and style of work, broadband speeds, devices and applications used, automation of production, etc. Specifically, broadband speeds determine what kinds of jobs can be performed at home. For example, if the remote work only requires standard definition videoconferencing, the minimum broadband (either fixed or mobile) speed is 1 Mbps. Other applications require higher speeds. It is recommended that a minimum of 50–100 Mbps download speeds and at least 10 Mbps upload speeds for working from home (McNally 2021).

The fundamental argument of this study is labor in the production function should be replaced by effective labor during the COVID-19 pandemic period, and effective labor is affected by broadband connections. Therefore, economic growth rate is dependent on effective labor, as well as broadband variables. This mechanism is explained by the following equations:

Based on the discussion of the unique contribution of broadband to effective labor, as well as the negative impacts of COVID-19 on effective labor, the coefficient of effective labor is defined by the following function:

where f ( B ) represents the influence of broadband on effective labor. p ( C ) represents the influence of the COVID-pandemic on effective labor.

The effective labor equation is:

The growth rate of effective labor is derived from Eq. (4):

where the dot above the variable represents the time derivative of the variable. For example, L ˙ e is the time derivative of effective labor: L e ˙ = d L e d t .

The economic growth function is derived from Eq. (1) and Eq. (5):

f ( B ) ˙ f ( B ) is the impact of broadband on economic growth rate. If f ( B ) is defined as the broadband penetration, f ( B ) ˙ f ( B ) is the growth rate of broadband penetration. On the other hand, many previous studies found that the broadband penetration rate has a significant correlation with economic growth rate during the normal time and pandemic time (e.g., Gruber et al. 2014; Katz and Jung 2021). Meanwhile, there are also several studies that demonstrated the size of broadband network has an impact on economic growth (e.g., Koutroumpis 2009; Vu 2019). p ( C ) ˙ p ( C ) is the impact of COVID-19 on economic growth rate. As the specifics of p ( C ) is not determined, p ( C ) ˙ p ( C ) is also not determined. Following previous studies (Katz and Jung 2021; Zhang 2021), this study uses the number of infected or death cases to measure this variable.

Based on Eq. (6) and previous studies, this study tests the following hypotheses:

4.1 Data

This study uses the secondary data from multiple authoritative sources. Specifically, the economic growth rate (GDP annual % growth), capital growth rate (gross fixed capital formation annual % growth), total labor force, total population, and population density data come from the World Bank Development Indicator database (https://databank.worldbank.org/source/world-development-indicators). Fixed broadband subscriptions (per 100 people) and mobile broadband subscription (per 100 people) data come from International Telecommunication Union database (https://www.itu.int/en/ITU-D/Statistics/Pages/stat/default.aspx). The number of COVID-19 infections come from World Health Organization database (https://covid19.who.int/data). Because this study examines the influence of broadband on economic growth during the COVID-19 pandemic period and the outbreak of this pandemic happened in the end of 2019, only two years’ data (2020, 2021) are available in the World Bank (2022) database and the World Health Organization database to test the hypotheses of this study.

4.2 Data analysis methods

The literature suggests that economic growth is also influenced by demographic structure. Specifically, the population size and population density have been investigated as the determinants of economic growth (e.g., Alders and Peter 2004; Kelley and Schmidt 2005; Pham and Hong Vo 2021). These two variables are added into the basic economic growth model, and the following regression model is used for testing the hypotheses:

where Y G i , t is economic growth rate, K G i , t is capital growth rate, LG i , t is labor growth rate, B B i , t is broadband variables (i.e., broadband penetration growth rate, broadband penetration rate, and the size of broadband network), P C i , t is the number of infected cases, P O P i , t is population size, θ is the coefficient of population size, P O D i , t is population density, ϑ is the coefficient of population density, ξ i , t is error term, i denotes an economy, t denotes time.

Panel-data (cross sectional and longitudinal data) regression method is used in the empirical analysis. There are two techniques to analyze panel data: fixed effects and random effects. The fixed effects model based on Eq. (7) is:

where σ i ( i = 1 , … , n ) is the unknown intercept of each economy (economy-specific intercepts), u i , t is the error. Other variables are the same as those in Eq. (7).

The random effects model based on Eq. (7) is:

where δ i , t is between-economy error, ε i , t is within-economy error. Other variables are the same as those in Eq. (7).

In the fixed effects model, the unobserved time-invariant differences between the economies (such as culture, political systems, religions) are controlled for, and these variables are absorbed by the intercepts. In the random effects model, the variation across economies is assumed to be random and uncorrelated with the predictor, and thus these time-invariant variables can be included in the model (Greene 2008). Hausman test is usually used to determine whether the fixed effects model or random effects model is more appropriate. When the p-value (prob > chi2) of the test is significant (<0.05), fixed effects model is a more appropriate model than random effects model (Torres-Reyna 2007).

Besides the panel-data regression, Pearson’s correlation test is performed to assess the correlations among the penetration rate, penetration growth rate, and network size of fixed broadband and mobile broadband, and their correlations with economic growth rate.