Home | OECD iLibrary

Introduction

Information and communication technologies (ICTs) are the backbone of the digital economy and society. This chapter reviews trends and recent developments in core sectors, including ICT supplying sectors and communication services, which provide the foundations for access to and connectivity within digital environments. It focuses on the ICT sector, communication markets, broadband networks and the Internet of Things (IoT).

The ICT sector is a key driver of innovation, accounting for the largest share (23%) of business expenditure on research and development (BERD) in the OECD. About 37% of all patent applications are in ICT technologies. In 2015, the ICT sector accounted for 4.5% of total value added in OECD countries, largely concentrated in services (80%). At the end of 2016, over 70% of the United States’ venture capital (VC) investments went to the ICT industry.

Employment in the ICT sector has proved resilient to the 2007 crisis and has been growing since 2013. This trend is mainly driven by sustained job creation in information technology (IT) services and software. These trends are expected to continue in the coming years, as the share of VC investment in ICTs is back to the peak it reached in 2000.

Communication networks are critical for the development of digital economies. They underpin the broader use of ICTs for economic and social development and assist in achieving policy goals. In recent years, communication infrastructures and services have continued to develop apace, driven by increasing demand, tremendous innovation and growing competitiveness. More than ever, OECD countries welcome these developments, recognising their potential to strengthen and sustain their economies and to improve social welfare.

In terms of infrastructure, communication operators have deployed fibre optics further into their networks to support the evolving “last mile” technologies that are designed to make copper, wireless and coaxial cable able to deliver higher speeds or, in the case of some, taken fibre all the way to the premises of their customers. While the devices people use in their daily lives are increasingly wireless, whether over cellular mobile services or Wi-Fi connectivity, this is only possible if fixed networks are available with sufficient capacity to meet the growing demands for data that are generated in digital economies.

Backbone facilities have for many years been made up almost entirely of fibre networks. The lines used to connect to these backbones provide the backhaul necessary to connect wireless towers or end users directly. For fixed network access this is necessary to support the increasing capacity being offered to users. The first 10 Gigabits per second (Gbps) commercial broadband offers have started to be deployed and, while still few in number, point to the future. Not so long ago offers from 100 Megabits per second (Mbps) to 1 Gbps were the outliers but today they are increasingly commonplace in OECD countries.

Fixed-line gigabit services will require major investment in backhaul networks to meet demand and, for the same reason, so will wireless networks. Many believe commercial 5G services will be brought into play around 2020 with an increasing number of trials already underway. Just as each previous generation of mobile technology needed higher capacity backhaul networks, so too will 5G. Moreover, the cells for 5G are expected to be smaller than for previous generations requiring many more locations. Some will be the traditional towers with a greater array of transmitters, to make more efficient use of available spectrum, but others will be located on urban infrastructures, from lampposts to street signs and rooftops.

The development of gigabit fixed and 5G wireless networks will necessitate ever-closer attention, as more infrastructures will need to be deployed and the use of IoT devices and machine to machine (M2M) keeps growing, for example for autonomous vehicles, which are likely to spur large increases in the amount of data they generate. Some of the demands of these services will be well met by technologies such as Long-term Evolution for Machines (LTE-M),2 with the first such networks now being deployed.

Key findings for the ICT sector are that since the global economic crisis, value added in the ICT sector as a whole has decreased in the OECD in line with total value added. Within the ICT sector, however, value added in telecommunication services and in computer and electronics manufacturing has decreased while it has increased in IT services and remained constant in software publishing. These contrasting trends, which are being reflected in OECD ICT employment, are expected to continue in the coming years as the share of venture capital investment in ICTs – an indicator of business expectations - is back to its 2000 peak. The ICT sector remains a key driver of innovation, accounting for the largest share of OECD BERD and for over one third of total patent applications worldwide.

Key findings for infrastructures, services and the IoT are that demand and innovation are driving positive developments in communication infrastructures and services. Fixed broadband subscriptions continue to increase, while average prices for both fixed and mobile broadband access are decreasing, and mobile broadband subscriptions are at a new high, with mobile data usage growing exponentially in some countries and Wi-Fi helping to offload some of the traffic. The IoT continues to evolve with M2M subscriptions on the rise and different wireless options promising improved connectivity.

Trends in the ICT sector

Growth in the ICT sector is increasingly driven by software production and services, with the latter accounting for more than 80% of total ICT value added. Slower growth in the ICT sector seems due to the sluggish performance of the semiconductor industry, which was previously a key branch of the industry. However, despite an overall decline in value, the share of ICT goods and services in total trade continues to increase. Production and export of ICT goods and services are increasingly concentrated in a few OECD countries, with six of them accounting for about 80% of world exports of ICT goods. The ICT sector remains a key driver of innovation, with over 30% of all patent applications in the OECD being related to ICT.

The ICT sector has not fully recovered from the crisis, but computer and data-related services drive a positive outlook

Recent trends in value added and employment

Since the global economic crisis, value added in the OECD ICT sector has remained constant, in line with the total value added (Figure 3.1). Several factors are at play and there are undoubtedly shifts between the sectors that make up this category. Between 2008 and 2015, value added in telecommunication services (-10%) and computer and electronics manufacturing (-7%) decreased as a result of a combination of factors, including increased use of production in OECD partner economies and the value added being recorded in different areas. While demand for devices and services are increasing, this is to some extent offset by global and local competition reducing prices. In addition, once recorded in telecommunication services, value added faces increased competition due to the greater use of software defined services. On the other hand, value added increased by 16% in IT services and by 12% in software.

In 2015, the ICT sector accounted for 5.4% of total value added for selected OECD countries (Figure 3.2). This share shows large variations across countries, ranging from over 10% of total value added in Korea to less than 3% in Mexico and Turkey. Sweden has the second-largest share (over 7%), followed by Finland (close to 7%).

In the majority of OECD countries, the value added tends to concentrate in ICT services, which accounts for three-quarters of total ICT sector value added (4% of total value added), reflecting a broader trend of specialisation in services rather than manufacturing. Within ICT services, IT and other information services industries are prominent in most OECD countries. Exceptions are Greece, Luxembourg and Mexico, where the value added is concentrated in the telecommunications industries.

Figure 3.3 shows the evolution in the years following the crisis of the share of ICT goods and services in total value added by country. The picture is somewhat mixed. In certain countries, notably Finland, Ireland, Japan and Luxembourg, this share decreased between 2008 and 2015. In others, however, the share increased – notably in Estonia, Iceland, Latvia, Norway and Sweden.

From 2008 to 2015, employment in the ICT sector proved resilient and it has grown faster than total employment (Figure 3.4). This is mainly caused by the continued growth in the number of people employed in specific sub-sectors, such as the IT and other information services industries and the software publishing industries. On the other hand, the two sub-sectors that have not shown any signs of recovery following the crisis in terms of employment are the ICT manufacturing and the telecommunication industries, which continue to decrease.

In 2015, the ICT sector accounted for 3% of total employment for selected OECD countries. Estonia, Korea and Luxembourg had the largest shares of ICT employment in total employment, at 4% and over. The smallest shares were in Greece, Lithuania, Mexico and Portugal (less than 2% of total employment). ICT services (software publishing, together with the telecommunications industry and IT and other information services), accounted for almost 80% of ICT employment on average (Figure 3.5).

Figure 3.6 shows changes in the share of ICT in total employment in the years following the crisis. In most countries, with the exception of Finland, Germany, Japan and Mexico, the ICT sector’s share of total employment has held steady or increased since 2008.

A significant part of ICT value added and employment in OECD countries is accounted for by foreign affiliates (i.e. local firms owned or controlled by a foreign company). In 2015, the share of ICT value added produced by foreign affiliates was above 75% in Estonia and Hungary, 62% in Poland and above 50% in Austria and the Czech Republic. ICT employment matches these figures, although the employment shares tend to be lower (except in Estonia and Finland) due to higher productivity of foreign affiliates relative to domestic firms (Figure 3.7).

Outlook for the ICT sector

Statistics on value added and employment are only available until 2015. However, some short-term indicators can provide an outlook for the ICT sector in more recent times. In 2016, production in the ICT sector had not yet fully recovered from the global economic crises that occurred in 2007 and 2009. Output growth in ICT manufacturing industries was sluggish from late 2010 onwards in most countries, especially for those hit more severely by the crisis. The same trend can be observed in ICT services, although the effects of the crisis were milder (OECD, 2015).

Over 2015-16, output growth in manufacturing slowed down in most economies (Figure 3.8), with a few noteworthy exceptions:

The People’s Republic of China (hereafter “China”) continued to grow at a sustained rate of 10% annually.
Growth in the United States has remained relatively stable, around 5% per year.
In the European Union (EU), ICT manufacturing output grew significantly (15%) in 2015 but growth seems to has come to a halt more recently.
Growth in Korea has has largely been positive since late 2015 while Japan has witnessed mainly negative growth rates over that period.
Output growth was negative in Chinese Taipei from June 2015 to June 2016 but started to increase afterwards, reaching 10% by the end of 2016.

Figure 3.9 shows trends in ICT services over 2013-16. Turnover in telecommunication industries (panel A) has remained stable in most economies, except China where it has been growing at a spectacular rate since 2014, reaching over 60% in 2016.

Trends in IT and other information services industries (panels B and C) are more positive. Overall, turnover3 in computer and other related services industries increased in 2016, ranging between 7% in the EU15 and about 2% in Japan. Turnover growth was also positive in data-processing industries in 2016. The United States and the EU15 have shown increasing growth rates since mid-2014, up to 10% in the third quarter of 2016, slowing down afterwards. Korea registered the highest growth rate (15%) in 2016, after a dip in 2015.

Semiconductor production remains a leading indicator for the ICT sector. Semiconductors are fundamental for growth and innovation in the digital economy – e.g. mobile technologies, IoT, smart technologies (sensors, visual recognition, etc.). Industry sales have been growing very modestly in the past two years, only 1.1% in 2016, and are not expected to regain major traction in the near future (Figure 3.10). The main reasons seem to be declining average sales prices for semiconductors coupled with high research and development (R&D) and investment costs for the increasingly complex fabrication of semiconductors (KPMG, 2016). The Asia-Pacific region and Japan account for 71% of total annual sales. This is also where growth remains the highest: semiconductor sales in China and Japan grew by 9.2% and 3.8% respectively.

VC investment, a market indicator of upcoming business opportunities, shows a global slowdown. In 2016, global VC investment was about USD 101 billion, a 23% decrease over the previous year. While VC investment in Asia and North America continued to fall in the last quarter of 2016, Europe saw an increase in funding (PwC, 2017).

In the United States, despite the overall slowdown, the ICT industries remain a key area of focus for VC investment, accounting for 71% of total VC investment in Q4 2016 (Figure 3.11). This share has remained stable since 2014, and is back to the level before the dot.com bubble.

The share of ICT goods and services in total trade continues to increase despite an overall decline in value terms

This section presents developments in gross trade patterns of ICT goods and services over time. These sectors are core building blocks of the digital economy and gross trade patterns help illustrate how international demand for and transactions of ICT goods and services have evolved. Chapter 5 includes a description of how the digital transformation is reshaping the broader trade landscape, particularly for services, and includes an analysis of trade in ICT goods and services in value-added terms, as well as data on trade restrictions on services in certain ICT services.

Trade in ICT goods

Over 2008-15, the value of world trade in ICT goods increased by 12%, exports from China increased by 49%, while OECD exports decreased by 13% (Figure 3.12). Over the same period, imports of ICT goods in value terms from China increased by 60% while OECD imports remained stable (1%).

In 2015, the value of world exports of ICT goods decreased by 3.4%, down to USD 1.9 trillion4 while the share of ICT goods in total goods exports increased by 11%. However, the decrease was smaller for ICT goods than for total trade in goods. As a result, the share of ICT in total trade in goods increased (Figure 3.13). Imports of ICT goods followed the same pattern. In 2015, the share of ICT goods in total imports worldwide increased (from 11.8% to 13.1%) but the value of global imports of ICT goods declined by 3.3% to just over USD 2.1 trillion.5

Exports of ICT goods are increasingly concentrated in a few economies. In 2016, the top ten exporters, which include six OECD countries, accounted for 85% of world exports of ICT goods, up from 70% in 2001 (Figure 3.14). Partly due to offshoring of production, Japan’s share in world exports of ICT goods decreased from 10% in 2001 to 4% in 2016, while China’s share grew from 6% to 32%, with a tenfold increase in current US dollars. Korea is the only OECD country whose share continues to grow (5.5% in 2001, 6.8% in 2007 and 7.6% in 2016).

The trend towards a re-composition of exports from computers and peripherals to communication equipment continued (Figure 3.15). In 2015, the share of ICT exports in communications equipment has reached the share of computers and peripherals exports (26%), while electronic components exports continue to account for the largest share of ICT exports (33%).

Trade in ICT services

Over 2010-16, the value of OECD export of ICT services increased by 40%, just below the growth of world trade of ICT services but faster than total trade in services (Figure 3.16). In 2016, world exports of ICT services increased by 5%, from USD 470 billion to up to USD 493 billion. As a result, the share of global exports of ICT services in total services increased by 2 percentage points, reaching over 10% in 2016.

As for trade in ICT goods, a few economies account for a significant share in global exports of ICT services (Figures 3.17 and 3.18). Ireland, which benefits from the presence of a high concentration of transnational corporations relative to the size of its domestic market, continues to be the leading exporter of ICT services (over 14% of global services), followed by India (11%) and the Netherlands and the United States (both with 8%). China is also among the top ten exporters of ICT services, along with France, Germany, Sweden, Switzerland and the United Kingdom. Together, these ten countries account for two-thirds of total exports of global services.

ICTs play a key role in today’s innovation activities

Enterprises in the ICT sector are leading across all types of innovation activities, while innovators are often intensive users of ICTs. In most OECD countries, the ICT sector accounts for the largest share of BERD, representing about 24% of total BERD and 0.4% of the gross domestic product (GDP). In 2015, ICT BERD relative to GDP was highest in Chinese Taipei (1.77%), Korea (1.73%), Israel (1.61) and Finland (1.04), followed by the United States, Sweden and Japan (about 0.6%), (Figure 3.19).

Figure 3.20 shows detailed information on the breakdown of the business R&D expenditure in the ICT sector and provides information on the weight of the ICT sector BERD in total BERD. In 2014-15, Chinese Taipei and Korea devoted 71% and 49% of their total BERD to ICT manufacturing. Despite the drop in Nokia’s activities, Finland continues to spend over 41% of its total BERD on ICT manufacturing, which is the same as Singapore, followed by Japan, Sweden and the United States, which all spent above 15% of total BERD.

IT and other information services represent more than 50% of total ICT business R&D expenditure in a majority of countries. The highest shares of R&D expenditure on software publishing in total ICT BERD were observed in the United States and Norway, accounting for 33% and 23% respectively. Telecommunication services account for a lower share of ICT BERD in most countries, except for Australia, Portugal and the United Kingdom, where it represents about 25% of total ICT BERD.

While R&D provides one measure of innovation input, patents, registered designs and trademarks capture elements of innovation output. In 2012-15, more than 0.9 million patent families were filed within the Five Intellectual Property (IP5) offices (i.e. the European Patent Office [EPO], the Japan Patent Office [JPO], the Korean Intellectual Property Office [KIPO], the State Intellectual Property Office of the People’s Republic of China [SIPO] and the United States Patent and Trademark Office [USPTO]). Patent applications in ICT technologies accounted for almost 37% of total applications, against 35% over the 2002-05 period. In OECD countries, ICT-related patents accounted for almost 34% of all applications, a slight decrease compared to the 2002-05 level, while applications by Brazil, the Russian Federation, India, Indonesia, China and South Africa (BRIICS) almost doubled, reaching 58%, largely as a result of increased patenting by China (Figure 3.21).

Registered ICT and audiovisual-related designs can be used to proxy innovation in relation to the aesthetic feature of products and provide information about product differentiation and customisation and, more generally, about the role played by design to shape competition in the marketplace. In 2011-14, registered designs in ICT and audiovisual devices accounted for 9.6% of European Registered Community Designs (RCD), representing a 2 percentage point increase over 2006-09. Across all economies, about 60% of registered ICT and audiovisual-related designs refer to data-processing and recording equipment, followed by communication and audiovisual devices (Figure 3.22).

The United States and Korea are the most active economies in ICT and audiovisual-related RCD (both gaining shares with respect to 2006-09), followed by Germany and Japan (both losing shares), with the other large European economies tailing behind. China more than doubled its share but remains a minor player with regard to designs registered in Europe. The United States scores high in data-processing equipment and Korea in communication equipment, while France and Japan lead in the design of audiovisual devices.

Korea shows the strongest specialisation in ICT and audiovisual-related designs, which represent almost 65% of Korean total RCD. Other economies specialising in this field are Canada, Japan, Chinese Taipei and the United States.

The distribution of trademarks offers a distinctive perspective on the competitive position of economies concerning ICT products. Indeed, national trademark shares do not align with R&D, patents or export shares. The United States appears to be the largest overall player, accounting for 76% of total ICT-related trademark applications at the United States Patent and Trademark Office (USPTO) and more than 12% at the European Union Intellectual Property Office (EUIPO) (Figure 3.23). ICT-related trademarks on the European market are conversely led by applicants in Germany, followed by the United States, the United Kingdom, Spain, France and Italy. In the last five years, a number of large trademark players, such as Japan and the United States, lost shares in EU branding to the benefit of China, Korea and smaller EU countries, while Germany and Spain were able to hold their positions.

Communication markets

Communication networks are critical for the development of Digital Economies. They underpin the broader use of all ICTs for economic and social development as well as in assisting to achieve the many goals set by policy makers. Indicators on network dimensions and development, as well as the take-up of services over these infrastructures, are at the forefront of any assessment of the ability of a country to seize the potential benefits of ICTs.

While the number of telecommunication subscriptions continues to grow, industry revenue fell slightly over the period 2013-15. This may be explained by the evolving market players and the changing nature of subscriptions as well as increased competition. Network operators continue to provide the access paths and connections, but new players such as over-the-top (OTT) providers are increasingly offering applications, which may influence reported sector revenue. In addition, mobile, fixed broadband and M2M subscriptions are increasing, while traditional fixed lines are decreasing. However, these subscriptions are offered at different price points; for example M2M subscriptions are often at a lower price than traditional mobile services (i.e. lower average revenue per subscription unit), which may contribute to current revenue trends relative to ongoing subscription growth.

Trends in the number of subscriptions and industry revenue appear to decouple

The long-term relationship between increased communication subscriptions and growth in industry revenue, which was consistent for more than a century, appears to have somewhat decoupled in recent years. After reaching peaks in 2008 and 2011, total industry revenues has been flat or declined over 2011-15. Between 2013 and 2015, telecommunication revenue decreased by 6%, from USD 1.312 trillion to USD 1.235 trillion (Figure 3.24). Notwithstanding the decrease in industry revenue, the number of subscriptions to telecommunication services continued its extraordinary growth as witnessed already over the past two decades.

By 2015, there were more than 2.3 billion telecommunication access paths in OECD countries (Figure 3.25). This was up more than 150 million access paths from 2013 for an overall increase of 7%. The access paths that continue to grow are fixed and mobile broadband subscriptions, as well as subscriptions for M2M communication services. In contrast, the number of fixed lines with traditional telephony subscriptions continued its longer term decline. This raises the question of why there is a divergence between subscription growth and overall telecommunication revenue.

Changes to telecommunication earnings and expenses are due to a number of factors. Some are relatively neutral in terms of incoming receipts and outgoing payments. Others reflect a longer term decoupling of the historically close relationship between service revenue growth and that for access paths. A decrease in termination charges results in lower revenues, but also lower costs. At the same time, if consumers purchase devices independently from subscription plans, that revenue remains in the broader sector but is not counted in terms of service revenue by network operators nor as costs, which overall results in a relatively neutral outcome. Conversely, the changes that result in decreasing revenues recorded by access providers at a time of increased subscriptions more fundamentally reflect a partial substitution by new players of the entities that used to provide services and devices over those networks.

Historically, network operators offering telephony or cable television provided what might today be called the complete ecosystem of access and services. Now, they continue to provide the access paths and garner revenue from connections and usage but do not necessarily provide the applications. Sometimes these services are provided by entities that some call OTT providers. The income for OTT services, such as Voice over Internet Protocol (VoIP) or video on demand, is not wholly captured in statistics on telecommunication revenue unless provided by network access operators. In other words, the overall revenue that may be attributed to the sector with increased access is not necessarily declining, but could be shifting or expanding in new directions given the large increase in OTT services.

A further factor explaining why telecommunication revenues are not proceeding at the same pace as the growth of access subscriptions is the changing nature of subscriptions. Between 2013 and 2015, the number of traditional fixed telecommunication lines decreased by 12.5%. Over the same period, mobile subscriptions increased by 8.5%, fixed broadband by 7.9% and M2M by 50.5%. The pricing of some of these services is, however, often substantially different from traditional approaches or the greater use of bundles (i.e. the inclusion of services once priced separately). While the price for unlimited access to the Internet from a dedicated SIM card in an automobile may be similar to one for a smartphone, this is likely not the case for many other M2M services (e.g. in an area such as environmental monitoring by sensors). That said, this market is expected to grow substantially in the coming years, providing tremendous opportunities for wireless networks in the enterprise sector.

In 2015, telecommunication investment was higher in proportion to revenue at 15.7% though at USD 194 billion some 3% lower than in 2013 in absolute terms. In terms of individual countries, New Zealand devoted the largest proportional share of revenue to telecommunication investment (Figure 3.26). This high investment share is in association with the development of a national fixed broadband network and an expansion of mobile broadband coverage. It continues to be reflected in higher demand for “fibre to the residence” subscriptions and an increase in the country’s fixed penetration ranking among peers. Nevertheless, expanding rural coverage for mobile broadband remains a priority in New Zealand. Meanwhile countries such as Korea, Latvia and Japan, which have the highest penetration of fibre in fixed networks and well-developed mobile broadband coverage, are devoting a lower relative proportion of revenue to investment. In those countries the next increase to overall investment is likely to be the result of forthcoming 5G mobile networks.

Broadband networks

Fixed and mobile broadband subscriptions have continued to increase in the OECD, reflecting their ongoing complementarity. Substitution certainly occurs, such as when mobile telephones are used instead of fixed telephones for voice services, but the most intensive use of wireless devices, such as smartphones, is over Wi-Fi supplied by fixed networks. Prices in both fixed and mobile broadband have decreased, with mobile plans increasingly being priced based on data usage rather than telephony, mirroring the rapid increase in the demand for mobile data in the market. In terms of fixed broadband technology, digital subscriber line (DSL) still represents the largest category, though it is gradually being replaced by fibre as network operators invest in faster networks. Given the increased importance of mobile broadband, this edition of the OECD Digital Economy Outlook measures for the first time the actual usage of mobile data volume and finds a steep increase in the use of mobile data per mobile broadband subscription across the OECD.

Fixed broadband subscriptions continue to increase across the OECD

The number of fixed broadband subscriptions continues to increase across the OECD. Data for fixed-line broadband show that subscriptions in OECD countries reached 387 million as of December 2016, up from 372 million a year earlier and leading to an average penetration rate of 30.1%. Switzerland, Denmark, the Netherlands and France topped the list with 50.1%, 42.4%, 41.9% and 41.4% respectively (Figure 3.27).

For many countries growth is slower than in previous years, reflecting their higher penetration rates. Turkey and Mexico defied this trend adding 9.3% and 9.2% respectively between December 2015 and December 2016, but generally the countries growing at the highest rates have penetration rates below the OECD average. Notable increases were experienced in Portugal (7.6%), Australia (7.5%) and Greece (5.5%) (Figure 3.28).

Overall, the data indicate that the market continues to view fixed and mobile broadband technologies as complimentary. That being said, in five countries the number of fixed broadband subscriptions declined between December 2015 and December 2016. These were Estonia, Switzerland, Luxembourg, Finland and Poland.

Wi-Fi access is undoubtedly the technology that best illustrates the complimentary nature of fixed and wireless networks. OpenSignal, a tool that uses crowdsourced data provided by smartphone users on a voluntary basis, illustrates this phenomenon well. In August 2016, OpenSignal reported the amount of time their users were connected to Wi-Fi. Across OECD countries this ranged from 40% in Turkey to 70% in the Netherlands. This likely reflects the Netherlands’ high penetration of fixed broadband networks – among the highest in the OECD – together with a high population density and as a result a greater proximity to Wi-Fi coverage. As OpenSignal points out, however, these figures are the amount of time connected to Wi-Fi and not the amount of data downloaded. Nonetheless, all available indicators show that users download the bulk of smartphone traffic when connected to Wi-Fi networks. This can be over 80% in some OECD countries and even higher in countries with lower Internet access penetration. In India, for example, users of Wi-Fi provided by Google in partnership with RailTel, a telecommunication operator with fibre along railway tracks, download 15 times more data on their smartphones than on days where they only rely on cellular networks (Rajan, 2016). Key factors include the availability of reliable power and fibre backhaul at Indian railway stations as well as the fact that this service is offered free to users. In other words, Indian users regard the uncharged access in the same way users in OECD countries regard access to Wi-Fi as being at a lower cost than cellular networks.

In a sense, all wireless technologies are essentially extensions of fixed networks. Wi-Fi extends fixed networks over a short range and cellular networks over a much larger area while both allow nomadic and mobile usage. Which one is viewed as substituting for the other is only moot in the sense that a user will take decisions on types of subscriptions (e.g. higher or lower amounts of data included in a plan; not personally subscribing to a fixed service if a combination of their needs are met by Wi-Fi and cellular or even giving up a conventional cellular subscription if their needs can be met by services that primarily rely on Wi-Fi or a service such as FreedomPop).

At present all available data indicate that substitution occurs largely in a user’s choice of access technology at any point in time rather than between subscriptions. In other words, the bulk of users substitute Wi-Fi for cellular when at home or at work. They nonetheless maintain both fixed and mobile subscriptions due to their complimentary nature and the offloading of traffic benefits both cellular providers and users. What may change that relationship over time would be if cellular networks increased speeds and data allowances to the point where they met the needs of enough users to give up, for example, their fixed residential connection. The first signs of this may have occurred in Finland and Latvia but other factors may countervail such developments.

The key constraints on cellular networks for substituting for fixed broadband is capacity in most countries, whether defined as the amount of available spectrum or the type of backhaul technology connecting any cellular tower. Smartphones accessing data make far greater demands on this capacity than mobile telephony did. As a result, the number of simultaneous users accessing data is more constrained than telephony and this is reflected in how cellular networks are priced, the actual speeds available compared to fixed networks, and how much data users download on both. In the final quarter of 2015, for example, the average mobile user in Australia downloaded 1.4 Gigabytes (GB) per month (ABS, 2016). The average download for Australia’s National Broadband Network, across a mix of fixed technologies, was around 80 times that amount in the same period.

Speed and technology

Since the first commercial fixed network broadband services were introduced in the second half of the 1990s there have always been some outliers in respect to the fastest speeds offered to consumers. Business services are a different segment in this respect. This is because individual offers to business users, educational institutions and the public sector can be tailored to their requirements through products such as leased lines between specific locations. Highlighting the leading offers to consumers is useful because it enables all stakeholders to look forward in terms of their own trajectories.

In the period under review here, the leading advertised download speed in the OECD area is 10 Gbps, with only a small number of consumer offers available at that level, such as in Japan, though even there it is not yet on a nationwide basis. Experience shows, however, that it might take a decade or more before such speeds are widely available in all countries. In 2002, for example, operators in Korea introduced broadband at 10 Megabits per second (Mbps) and, at the time, this was a pacesetter. Today, the baselines for some purposes, such as defining high-speed services or delivering actual service levels, for many countries exceeds this threshold. Notwithstanding such developments, even in these countries delivering such speeds to all geographical locations remains a challenge. This is one reason why average speeds vary substantially across OECD countries (Figure 3.29) and why it is preferable to look at speeds in tiers related to penetration (Figure 3.30).

Consumer offers marketed at 1 Gbps are increasingly common across OECD countries, particularly where there is fibre to the premises or upgraded cable broadband networks. This is the case in countries with high population densities, such as Japan and Korea, as well as in an increasing number of cities in New Zealand, Sweden and the United States. Residential offers at 1 Gbps are most common where there is either strong infrastructure competition between operators or competition between retail providers using wholesale networks. In Korea, for example, there is widespread infrastructure competition with residential apartments commonly being able to access three “fibre to the basement” providers. This means the building residents, who own the inside wiring, are in a strong position to jointly negotiate very competitive prices for connections to all residences. As a result, 1 Gbps services with unlimited data usage are available in Korea at around USD 25 per month.

In countries with cities that have a greater mix of apartments and stand-alone residential dwellings, 1 Gbps services are also increasingly common. For the most part, in the United States, 1 Gbps services are offered through end-to-end infrastructure competition rather than functional or structural separation of wholesale network providers. There are a few cases where retail Internet service providers (ISPs) have used unbundling associated with telecommunication lines as a launch pad for installing their own gigabit fibre networks, such as Sonic Internet in San Francisco. Sonic Internet provides a bundle of Internet access up to 1 Gbps with unlimited data usage and voice telephony for USD 40 per month.

While Sonic Internet has benefited from regulation that unbundled a legacy telecommunication network, and followed a “ladder of investment” strategy for developing its own fibre network, a different approach has been adopted by Layer3 TV. In September 2016, the start-up company entered the Chicago market as a retailer of commercially negotiated cable broadband access. This is not, however, in the form of what might generally be described as wholesale access or even as an OTT service. In other words, Layer3 is not seeking to act as its customer’s ISP using a cable network for physical broadband access. Rather, Layer3 is in some ways more like a content delivery network that injects its video content directly into the cable operator’s broadband network for final delivery to its customers by the provider that acts as the customer’s ISP. Customers of Layer3 TV would still need to purchase a separate broadband access service. A stand-alone Internet broadband access connection at, for example, USD 49 for 25 Mbps per month over a two-year contract would therefore be required in addition to Layer3 TV’s service. A user would also need to consider any additional charges that may apply if traffic was metered in the instance of a data cap or charge for a cable modem. At a time when some suggest that “cord cutting” will increase for traditional cable television, the company aims to attract customers with what it says is a superior set-top box for navigation and other features; a no contract period; and a combination of traditional cable, broadcast and online channels.

If Layer3 TV is successful, cable broadband networks in other countries may also begin to consider similar commercial arrangements for retailers, especially as “cord cutting” gains momentum against traditional approaches. At a time when regulators are closely examining set-top boxes to see the extent to which “walled gardens” may hinder competitive choice, the cable broadband industry faces as much change as the telecommunication industry has faced for many years. Aside from changing patterns of consumer demand, stimulated by the availability of OTT content, there has been an increase in the array of set-top boxes from companies such as Amazon, Apple, Google, Roku and many others. The capabilities of these devices go far beyond programme navigation to areas such as digital assistants. They also have applications (apps) that can carry content provided by traditional players or OTTs. In France, for example, Apple TV carries the Molotov.tv app, which offers content from free to air and pay TV suppliers. This may assist cable networks in meeting new forms of competition by players, for example, such as Twitter, which has launched an app for devices such as Apple TV, Amazon Fire TV and the Xbox One for viewers to watch free sports events while browsing curated content from apps such as Periscope. Layer3 TV set-top box also offers access to other OTT services such as Amazon and Netflix as well as integrating social media. In that sense it aims to provide a service that goes beyond the traditional cable television set-top box. Traditional ISPs have launched competitive responses to provide video content delivery. In the United States, for example, cable operator Comcast launched its X1 system, which aggregates video content from Comcast and other video providers and performs functions similar to third-party set-top boxes.

What such changes mean for infrastructure providers is an open question. Some with many years of commercial experience in providing end-to-end infrastructure and services will undoubtedly be able to compete. On the other hand, municipal or other publicly owned networks may be severely challenged to meet rapidly changing demand, given their strength is likely to be in providing utility-like infrastructure rather than services, unless they offer a degree of openness that enables retailers to innovate. An end-to-end network, for example, can offer symmetrical services if there is market demand. A retail network, however, can only offer such a service if the wholesaler enables that product to be sold. This is why the most successful publicly owned wholesale networks tend to be those that provide the greatest amount of latitude to retailers, such as Stokab in Sweden. In other words, it provides retailers with the same capabilities as end-to-end networks to meet customer demand, though the experience of UTOPIA (an open access municipal network in Utah, United States) demonstrates that this model does not guarantee success.

One of the most striking examples of a structurally separated network is that of Singapore. Here the wholesale infrastructure company provides dark fibre to ISPs who are free to provide any layer of service above that level. As a result it is among the first countries with commercial 10 Gbps services for consumers but also with the ability for ISPs to configure broadband access in ways they assess will most drive the take-up of services. One example is the offer for customers of two 1 Gbps fibre connections to a single household. While there are many countries that have more than 100% mobile penetration due to users having multiple SIM cards, Singapore has become the first and only country in the world to have more fixed-line subscriptions than households. This does not, of course, mean that more than 100% of households are connected, but rather that there are more fixed-line subscriptions in Singapore than the number of household premises. In other words, ISPs in a very competitive market have been tremendously successful in assessing demand in ways that may not have been obvious to a wholesale provider. By way of example, MyRepublic, an ISP in Singapore, sells 1 Gbps Internet access at USD 36 but two such connections for USD 43. Clearly, both the retail providers and consumers are attracted by the marginal cost and the wholesale arrangements enabled such an innovative approach.

In addition to offering multiple 1 Gbps connections, MyRepublic’s method for assuring service quality is one based on discriminating between different traffic types in the assignment of transmission priority. Some suggest this offering would be against net neutrality non-discrimination rules in some other countries. A further notable aspect of developments in Singapore is subscription plans aimed at users wishing to have 1 Gbps connections prioritised for playing games. If a user believes latency is critical to their online gaming experience, they can opt for MyRepublic’s “GAMER” plan, which enables them to request custom routing with the aim of optimising an individual game’s performance. Features such as custom routing can generally be found only in business plans with specific service-level agreements, not in residential plans.

The issue for policy makers and regulators is not to ask why users would need a 1 Gbps connection, why they would need two such connections in different rooms of a residence, or why some would pay more for one of those connections to be optimised for what they see as an edge in game playing. Market demand in the way broadband connections are used and what stimulates more adoption and utility evolves rapidly. The challenge for policy makers and regulators is to ensure the market is responsive to such demand by ensuring competition between end-to-end infrastructure providers or that wholesale providers maximise the ability of retailers to respond to such demand in the same manner as end-to-end providers in a competitive market. This is in an environment subject to ongoing change in access technologies.

DSL now makes up 43% of fixed broadband subscriptions as it continues to be gradually replaced by fibre, now accounting for 21.2% of subscriptions, up from 12.3% in December 2010 (Figure 3.31). Cable (32.7%) made up most of the rest. Japan, Korea, Latvia and Sweden have the highest shares of fibre in fixed-line broadband at 74.9%, 74.2%, 62.7% and 55% respectively.

All fixed broadband connections have upgrade paths available to them to provide the final connection to users. For cable broadband networks the technology is DOCSIS 3.1, a suite of specifications that supports up to 10 Gbps downstream and 1 Gbps upstream with the first commercial offers becoming available at 1 Gbps via companies such as Comcast in the United States for selected cities. For the historical copper lines, technologies such as XG.Fast have also demonstrated speeds up to 10 Gbps in laboratories, though commercial offers of the various DSL technologies (e.g. VDSL2) are generally commercially offered up to 100 Mbps, such as in Australia and Germany.

Copper lines face two main constraints when delivering broadband access. The first is that speeds decrease with distance due to signal attenuation; the second is the interference between different copper lines in a bundle. What a technology called vectoring does is cancel the noise between these different lines, enabling higher speeds. Nonetheless, what all these technologies rely on – whether DOCSIS or DSL – is taking fibre backhaul closer to premises. At the maximum extent, this approach takes fibre all the way to the premises, whether these are business locations or residential dwellings. This is often called fibre to the home, but there are many points along a network, such as fibre to the node or fibre to the cabinet. While different operators are following different network architectures depending on a range of factors and may not agree on which point to take fibre to in any given network, they are uniform in deploying fibre deeper into their network. This is why any fibre deployment is regarded as being “future proof”, because, however, the final connections evolve, and fibre is needed to ensure effective backhaul. This includes fixed wireless and mobile networks. The key issue for policy makers is not to be prescriptive in the choice of technology but to ensure that any technological choice involves sufficient competition and a path for demand-driven innovation. In countries with end-to-end network competition this implies having sufficient fixed-line infrastructure competition, while for networks that rely on regulated access it involves wholesalers not being able to stifle competition or innovation among retail providers.

Prices

Between 2013 and 2016, average prices across the OECD decreased for both fixed and mobile broadband access (Figure 3.32). This is drawn from a comparison over time of the averages for specific OECD price comparison baskets for telecommunication services. The baskets are designed to provide a snapshot of prices at any given time rather than as a series. Accordingly, the lowest cost plan is selected at any point in time and may have different characteristics from earlier plans (e.g. higher speed or increased amount of data). That caveat aside, it is nonetheless worth noting an average for all OECD countries as an indicator of likely trends, though all baskets are available online and provide a more accurate set of indicators for any country in relation to its peers.

The trend common between fixed and mobile broadband services is lower prices for data, with the highest gains being made by plans with larger volumes. This is reflected in the relatively constant price for a fixed broadband plan with a low use of 20 GB per month. By way of contrast, the average for plans with 200 GB declined 15.4%, from USD 43.25 to USD 36.57, in purchasing power parity, between June 2013 and June 2016. Mobile broadband prices have also declined, with the largest decreases associated with higher volumes. A mobile user with a 2 GB plan spent USD 70.88 in May 2013 but this had been reduced to USD 39.28, in purchasing power parity, in May 2016. Across all patterns of usage, during this period, there were reductions: some 44.5% for 2 GB plans, 32.6% for 1 GB and 23.9% for 500 Megabytes (MB).

While unit prices are declining, of course not all users are paying less, because they may prefer to pay the same amount as before for plans with higher included amounts of data, higher speeds and so forth. In mobile markets, where prices have decreased the most, it involves more competition in some countries but also the fact that data allowances are changing in response to greater demand. These factors are considered in the next section for mobile markets in relation to technology, speed and prices.

Mobile broadband subscriptions are at a new high

By December 2016, mobile broadband penetration in the OECD area had risen to 99.3%, meaning that there was nearly one high-speed mobile broadband subscription for every inhabitant (Figure 3.33). This is up from a penetration rate of 91% in December 2015. In December 2016, the addition of 113 million new mobile broadband subscriptions in OECD countries resulted in a year-on-year rise of 10%, driven by continued growth in the use of smartphones and tablets, lifting the OECD total to 1.275 billion subscriptions for a population of 1.28 billion people.

Data for the 35 OECD countries show that in between December 2014 and December 2016 Japan has overtaken Finland as the mobile broadband penetration leader, with a penetration rate of 152% versus 147% in Finland. The United States has moved up to fourth from eighth place, reflecting a growing demand for mobile video and data in general and increasingly competitive offers in that segment.

Speed and technology

For convenience, different generations of mobile networks are often referred to as 2G, 3G and 4G, even though there has been a range of different technologies associated with their evolution. All three generations are in use today, though after more than two decades of service the first 2G GSM networks are starting to be switched off. Telstra in Australia and AT&T in the United States switched off their 2G networks in 2016. Meanwhile, Singapore shut down all 2G networks at the same time, in April 2017. A range of GSM operators have announced this will occur in their 2G networks from 2018 to 2021, e.g. in countries such as Canada and Switzerland.

Consumers have long progressed to 3G and 4G services thanks to the popularity of smartphones, but 2G networks are still widely used by M2M communication. This is for a number of reasons, including the lower cost of 2G equipment (i.e. modems) and the lower need for data usage and speed for some M2M applications as well as the longevity of devices (e.g. consumers may switch handsets every two years or so whereas M2M equipment may be used for a decade or more). Meanwhile, the number of mobile towers that provide 4G coverage continues to increase in OECD countries relative to those providing 3G coverage only. At the same time, the first trials of so-called 5G networks commenced in 2016, though a standard has yet to be agreed.

In many ways 4G, or more precisely Long-term Evolution (LTE) networks, represent a major change in technologies because they were the first mobile networks designed for an IP-based system with significantly reduced transfer latency compared to 3G architecture. Plans for 5G networks aim to further optimise capabilities for data transfer, and while standards are not yet agreed, past experience demonstrates that some operators will launch services ahead of such agreements being reached to seek a competitive edge and meet growing demand. A likely characteristic of 5G networks is the use of smaller cells and, similar to 4G services, the need to improve backhaul capabilities over fixed networks for offloading traffic.

Prices

While tariff reductions are sometimes described in the media as a “price war”, in the mobile sector they can also represent more fundamental changes in a market characterised by technological and commercial shifts, as well as evolving consumer demand. The entry of a new mobile network operator (MNO) or a change in strategy by an existing player determined to win market shares almost always involves a change in pricing designed to attract a larger share of customers. In mobile markets today, which are characterised by forces observed earlier in fixed markets, as they converged with the Internet, this was reflected in a shift away from pricing primarily based on telephony to one based on the use of data.

The pricing of 4G services often differs greatly from that of 3G, taking advantage of architecture designed for IP-based traffic. In France, for example, since 2015 Iliad Free Mobile has offered 3 GB of 3G data per month but 50 GB of 4G data in a single subscription. In other words, as 4G coverage expands, so too does the opportunity to use a greater amount of data available for the same price. In other countries the same elements are evident in different aspects of tariffs but can be summarised as a trend away from charging separately for voice and text (i.e. they are just included as an unmetered part of a bundle) and relying on prices that reflect data usage. In other words, if 2G and 3G were optimised for voice and 4G for data, the shifts witnessed in an increasing number of countries would not be so much “price wars” as more reflecting a combination of changes in the market that will not return to the previous status quo.

While it has often been said that there was a trend away from unlimited data offers for mobile service following the sharp increase in the use of smartphones, they were never very prevalent across OECD countries. The United States was somewhat of an exception with some operators retaining unlimited usage plans while others grandfathered their use in the 3G era. The introduction of 4G services, however, has been characterised by unlimited data offers with price discrimination undertaken by speeds. In the United States, for example, this has been achieved by companies such as Sprint and T-Mobile, which have sometimes offer speeds for video quality at standard definition or for an additional payment at high definition. Among a number of operators in other countries, such as in Finland and Switzerland, users select the speed for all services but have no cap on data usage.

In Finland, Elisa offers tiered advertised speeds for 4G services at 50 Mbps, 100 Mbps and 300 Mbps and 3G services at 120 kilobits per second (2017). All the Elisa offers and those for other operators in Finland include unlimited data usage. Meanwhile, in Switzerland, Swisscom introduced unlimited data plans in 2012 when it launched 4G. Swisscom offers speeds starting at 1 Mbps, 10 Mbps and 50 Mbps after which offers are advertised at “highest speed”. Meanwhile, other operators in the Swiss market offer a combination of unlimited and tiered data offers, at the same 4G speed, as do operators in Latvia. A further differentiation in the Latvian market is the use of a combination of unlimited offers, such as one by Bite, an operator in that country, for USD 18.57 and a bundled subscription to the Deezer music service. Tiers with capped levels of usage on the other hand incorporate “zero rated” data usage for services such as Facebook and WhatsApp.

While tiered data pricing is still more prevalent than tiered speed pricing, there is generally an increase in the size of data caps across the OECD. This is contributing to higher usage of data, with Finland and Latvia leading the way (Figure 3.34). The average use of data per subscription in Finland was 11 GB per month in 2016, up from 7 GB the year before (Figure 3.35). Meanwhile the introduction of unlimited offers in Latvia was associated with an increase of 8.2 GB on average per month in 2016 from 5.8 GB in 2015. Across all OECD available countries the amount of mobile data grew from 18 000 petabytes (PB) to 27 500 PB, a 52% increase between 2015 and 2016. This indicator does not include the use of Wi-Fi by devices such as smartphones, which makes up the predominant data use for many users.

Substitution: Are we there yet?

While increased mobile data usage is common to all countries, the fact that Finland had the highest levels of usage and has also witnessed decreases in fixed broadband in 2016 is notable. This raises the question of whether mobile networks have reached the tipping point where some users will give up their fixed broadband lines. Competition between fixed and mobile does not require the two services to be perfect substitutes for all customers. Nonetheless, while there is certainly substitution for services between mobile and fixed subscriptions, such as telephony, the capacity constraints in terms of spectrum and backhaul have to date meant that they are largely viewed as complementary for Internet access by many users. Over time this may change for some users, as seems to be the case in Finland, if unlimited data offers are increased in other countries. On the other hand, the fact that Switzerland is still experiencing an increase in fixed broadband connections also suggests that fixed networks can leverage the higher speeds possible to retain and increase subscriptions.

It is important to point out that as far as networks are concerned, fixed and mobile are definitely complimentary. The spread of Wi-Fi means that most users in OECD countries are connected to this technology for more than half their day and download far more data over Wi-Fi than on cellular networks. Moreover, the offloading of this traffic improves the performance of cellular access for other users because fixed networks are doing the “heavy lifting”. That being said, the substitution effect can be greater in countries without a high penetration of fixed broadband. In India, according to OpenSignal, smartphone users are connected to Wi-Fi for less than a fifth of the time (18.4%) and in Myanmar it is less than a sixth (14.6%) (2017). This is one reason average speeds are much lower in these countries than in OECD countries.

In emerging countries substitution between mobile and fixed networks has taken a different form than in OECD countries. Where there are less-developed fixed networks, users have, of course, opted for mobile connections rather than fixed ones. In a 2G and even a 3G era, typified by voice and SMS rather than data, this was less of a constraint than with growing demand for data over 4G networks. Today, however, when a company such as Reliance launches a 4G network in India it underlines the establishment of Wi-Fi hotspots as an integrated part of its planning. It ensures Wi-Fi use is part of its tariff plans and that users can seamlessly transfer from cellular to Wi-Fi to ensure the maximum amount of traffic is offloaded onto fixed networks (Box 3.1).

Box 3.1. Tele2 and Reliance Jio

Most mobile operators started with 2G networks, some with 3G and recently a few with 4G as their first deployment. In 2015, the first new entrant that commenced with 4G, without a legacy network, was Tele2 in the Netherlands. Some 92% of the population was covered at the time of Tele2’s Dutch launch. Tele2 did, however, use the T-Mobile NLs 2G/3G network to increase coverage and to handle 2G/3G data in situations where there was no coverage or where devices did not support 4G data. Tele2 has faced some challenges in moving to a 4G only network. Support of Voice over Long-term Evolution (VoLTE) has proved to be device and manufacturer specific, with several 4G devices not capable of supporting VoLTE on Tele2’s network. In addition, many devices will revert back to the 2G/3G mode in the case of voice calls or emergency calls. In September 2016, Reliance Industries launched a 4G network called Jio in India, after investing over USD 20 billion, with a goal to cover 90% of India’s population during 2017.

What both of these 4G networks have in common is that they entered the market with offers available that only charge for data and included unlimited voice and text services. In addition, they both chose to offer data at lower prices and to allow users to subscribe to plans with greater amounts of data than had previously been available. Consumers in both countries are avid users of Wi-Fi, though on average Dutch users are connected a far greater amount of time each day. To address availability, Jio plans Wi-Fi hotspots that will leverage Reliance’s extensive fibre backbone, as will its proposal to provide a fibre to the home service in 100 cities at 1 Gigabit per second.

The Internet of Things

There has been an increase in the number of M2M subscriptions, reflecting the take-up of one part of the IoT. Mostly, IoT connected devices will generate lower amounts of data than traditional use, though there are expected to be many more such connected devices. New types of network capabilities (e.g. low-power, wide-area [LPWA]) are being rapidly implemented across OECD countries to meet this demand. That being said, some predict that the use of autonomous vehicles will generate much larger amounts of data, though it is yet unknown how much will need to be transmitted in real time. Irrespective of the balance of demand between immediate transmissions on a highway and a vehicle being garaged, this development could have major implications for infrastructure requirements in the future, along with the development of fixed and wireless networks.

Machine-to-machine subscriptions are increasing, marking the uptake of the Internet of Things

The year 2016 saw an increase in the take-up of M2M communications, with 149 million M2M SIM cards in use by year end versus 108 million at the end of 2014. Sweden, New Zealand, Norway, Finland and Italy are the leaders in M2M SIM cards per 100 inhabitants, with the caveat that data are not yet fully comparable for all countries. Sweden counts 87 M2M SIM cards per 100 inhabitants – a much higher level than for most other OECD countries that provided data, though not all of these devices are located in Sweden.

There are many uses for SIM cards in M2M communications. By way of example, the following paragraphs focus on the automotive industry. A connected vehicle may have two or more SIM cards inserted by the manufacturer – one for telemetry and the other for entertainment services. Some, like Tesla Motors, have chosen to sell vehicles with the use of connectivity provided by these SIM cards incorporated in the price of a vehicle. Users can also purchase stand-alone devices for any vehicle with an On-Board Diagnostics II (OBD-II) port such as that sold by the company Automatic.6 In the United States, the “Automatic pro” plugs into the OBD of any vehicle and has a 3G service included in the purchase price with unlimited data connectivity for five years. Other devices using the OBD port and an embedded SIM card seek to not only offer monitoring for diagnostics, but a broader range of services.

Vinli, a company offering a dongle that uses the OBD port, offers a variety of apps, from safety to entertainment to on-board Wi-Fi.7 In some cities in the United States, Uber uses Vinli to provide Wi-Fi for its users. Vinli’s dongle connects a users’ vehicle to their smartphone or computer and, in the United States, offers 4G connectivity via the embedded SIM card from T-Mobile, with users reporting actual speeds of 30 Mbps to 40 Mbps. The pricing for both the devices and data usage is dependent on Vinli’s partners who provide and distribute the product and service. The Vinli Sync sending data to the cloud is incorporated in the price of the device for the first two years, after which it has a yearly fee. In 2016, Vinli’s developer platform had over 2 000 partners and developers using Vinli’s cloud platform and app marketplace. Developers list their products in the Vinli app catalogue with the apps being available from the Apple and Google stores. In 2016, Vinli began offering the service outside the United States in partnership with MNOs in those countries.

Some MNOs have begun to sell dongles for use in the OBD with their SIM cards embedded. AT&T, for example, sells the ZTE Mobley unit with a two-year contract to their DataConnect plans starting at USD 20 for 1 GB or USD 30 for 3 GB (AT&T, 2017).8 Alternatively, the device can be added to some shared AT&T plans for an access charge of USD 10 per month and purchased without the contract for USD 100. Not all devices use the OBD to provide Wi-Fi in vehicles. In the United Kingdom, Three sells SIM card dongles with 2 GB of data per month for USD 13.23 (GBP 10), which connect via a USB port or 12v sockets. Such devices aim to provide connectivity but not vehicle diagnostics. In some vehicles users also have the option to connect their own smartphone and use their existing mobile subscriptions. These services can benefit from the subscriptions a user already has such as for music, as well as potentially improved signal strength using the vehicle as an aerial or for making hands-free calls. Like the dongles, however, such services are not integrated in the same manner as via an OBD device or factory embedded SIMs.

Automobile manufacturers have developed options for connectivity through SIM cards embedded in their vehicles. One of the first was General Motors (GM), which in partnership with AT&T offers its “OnStar” service.9 GM connects via AT&T’s 4G LTE service, with a variety of periods included for new or pre-owned vehicles for basic and premium services. Following the end of such periods, users need to extend their subscriptions via OnStar or, if they are an AT&T customer, add their vehicle to their mobile data plan for USD 10 per month. Other manufactures such as BMW and Audi have also embedded SIM cards in vehicles and offer services in an increasing number of countries with local MNOs.

In October 2015, BMW introduced a Wi-Fi hotspot connection to provide up to ten devices with Internet access. In Germany the company offers services via Deutsche Telekom and in the United States via AT&T. BMW’s “ConnectedDrive” menu includes access to current location-based information, such as weather and news, as well as an online search enabled via Google.10 Services and features such as parking information as well as travel and hotel guides can be accessed directly in the vehicle’s SIM card without a smartphone. The proprietary apps from the BMW store have unlimited access to selected services or through the user selecting a subscription.

In Germany the first BMW vehicles with virtual eSIMs were offered in mid-2016. In the future, such SIMs may enable users to switch providers, once standards are finalised and agreed. For the moment, in vehicles from most manufacturers, the approach is to use hardware with SIM cards being soldered into the mobile radio platform in the vehicle’s head unit. This means that users do not have a choice of SIM provider when they purchase a vehicle nor can they change provider.

In the United States, Audi’s built-in 3G/4G mobile SIM card enables the vehicle to access data services such as navigation via Google Earth and Google Street View as well as information regarding routing, road and traffic conditions, and parking. Notably, drivers get access to their Twitter accounts, e-mail and smartphone calendar. These are unlimited data volumes with regards to these services, incorporated in the price of a vehicle for three years, with separate plans for Wi-Fi. Users can also use their own smartphone and data plan, though this does not benefit from Audi’s selected unmetered services. Across most European countries, when users are travelling, the system connects automatically to Audi’s chosen MNO, avoiding roaming fees. Along with other manufacturers, such as BMW and Toyota, and in co-operation with Deutsche Telekom Audi is holding trials to assess the capabilities of Long-term Evolution for Vehicles (LTE-V), the vehicle version of the 4G cellular radio technology LTE (Hammerschmidt, 2016; see also Allevin, 2016).

Connected autonomous vehicles are expected to generate large amounts of wireless data

Connected automobiles are generating an increasing amount of data. Some of these data are simply users accessing entertainment services through embedded SIMs, but these also include IoT communication between devices. The Vinli dongle, for example, can communicate with Amazon’s Alexa device to enable home automation functionality, as can the services of automobile manufacturers or MNOs.

Autonomous vehicles are expected to generate large amounts of data. Some of these data can be offloaded to a fixed connection, such as using Wi-Fi when a vehicle is garaged, while other data must be transmitted in real time. The data generated from sensors on board of modern vehicles, for example, can be used to warn other cars on the road about possible dangers.11 HERE, the Open Location Cloud company, for example, aims to provide location-relevant data to verify and enhance maps and attributes, detect road incidents in advance, as well as warn about poor road conditions (e.g. potholes and construction). Such information they point out will be essential in vehicles that are given greater amounts of control. HERE, originally an American company that is now co-owned by Audi, BMW and Daimler, has developed a design for a universal data format that will allow for standardised vehicle data exchange, including for self-driving vehicles (Tipan, 2016). This enables the exchange of real-time traffic, weather and parking spaces across the vehicles of different manufacturers.

By October 2015, Google’s website for its “self-driving car” said the project had recorded data for 1.5 million miles.12 The company has collected more than 1.3 billion miles of data from autopilot-equipped vehicles operating under diverse road and weather conditions around the world (Hull, 2016). All Tesla cars after the first 60 000 have autopilot hardware and are providing autopilot data to Tesla Motors. Ford Motors says that its older models generated 500 MB of data an hour but current models may exceed 25 GB per hour.13 Only part of these data is, of course, being transmitted in real time. According to Chevrolet, its customers consumed more than 5 600 terabytes of data from December 2014 to December 2016 (Figure 3.36). Over time the overall volume of data can be expected to increase as more connected vehicles are sold and more applications developed, but also due to falling prices. In June 2016, for example, Chevrolet reduced the monthly price of 1 GB from USD 20 to USD 10 and of 20 GB from USD 80 to USD 40.

For the future, Intel says that the volume of data starting to be produced by semi-autonomous vehicles suggests fully autonomous cars will produce 4 000 GB per day by 2020, or the equivalent of the then average daily use of 3 000 people with smartphones (Waring, 2016a). Once again, it is necessary to point out that such volumes of data would not necessarily all be transmitted in real time over cellular networks, but it still does underline the potential need for further developments in areas such as 5G, fibre backhaul, vehicle-to-vehicle short-range communication, and other technologies and data pricing for the IoT. In addition, the further development of the Internet Protocol version 6 (IPv6) is to be welcomed given the exhaustion of IPv4.

Adoption of the Internet Protocol version 6 is progressing

Measuring an evolving process such as the adoption of IPv6 worldwide requires the use of different methodologies assessing different parts of the Internet (OECD, 2014). Usage has been growing significantly over recent years, although from a very low base. Some differences can still be observed depending on the measurement and vantage point used:

Data from the Asia Pacific Network Information Centre measuring the capability and preference of networks to use IPv6 show that global penetration increased from 2.5% to 6.5% between October 2014 and mid-September 2016.
Google IPv6 statistics, which track the percentage of users that access its services over, show that 13.6% of these users connected via IPv6 in mid-September 2016, compared to 3.9% at the beginning of October 2014 (Figure 3.37).
The percentage of IPv6-enabled networks was 26.3% as of July 2016 based on estimations by RIPE NCC using the global the Border Gateway Protocol table, showing an increase from the 18.0% in July 2014.14

The difference observed between the Google and the Asia Pacific Network Information Center (APNIC) data is likely due to the type of measurement. APNIC data show the capability of networks to use IPv6 whereas Google data show the percentage of end systems able to make a connection using IPv6. As more networks become IPv6-capable, network effects will be visible in Google’s measurement due to an increase in the total number of end-to-end IPv6 connections made by users.15

When analysing data from Internet exchanges, adoption trends vary greatly based on the data utilised. IPv6 traffic at the Amsterdam Internet Exchange (AMS-IX), the second-largest Internet Exchange Point, represents only 1.5% of the total combined IPv4/IPv6 traffic exchanged among the nearly 800 connected networks.16 The London Internet Exchange (LINX), another major European hub, accounts for seven times less active IPv6 prefixes in one of its route servers.17 However, when looking into active sessions, the number of IPv6 sessions represents about 38% of the combined IPv4/IPv6, a much more positive outlook.18

Comparing the adoption of IPv6 per country provides a useful benchmark to policy makers. As of October 2016, Belgium was the OECD leader in IPv6 adoption with 45.4%, largely ahead of the United States at 28.8%, Greece at 26.1% and Switzerland at 26.1% according to Google’s metrics (Figure 3.38).19 Efforts to accelerate the adoption carried out by governments, non-governmental institutions and the technical community seem to have been only partially successful: only six OECD countries had a user penetration higher than 20% and ten OECD countries still had less than 1% penetration as of October 2016.

The exhaustion of the Internet Protocol version 4 address space continues to be an issue

The depletion of the IP address space remains a relevant topic as regional Internet registries continue to run out of IPv4 blocks. The exhaustion of the American Registry for Information Numbers’ address pool for general use at the end of September 2015 followed the exhaustion of addressing resources in the other registries: APNIC in April 2011, RIPE NCC in September 2012 and LACNIC in June 2014. The African Network Information Center is the only registry with general use addresses left and its remaining pool is projected to run out in July 2018 assuming that the levels of demand in that region continue at their current levels.

With a nearly exhausted IPv4 address space, other areas of interest emerge for the industry and the research community. One may be a better understanding of how the IPv4 address space is being used and its implications for operational practices and governance decisions.

A recent measurement study argues that a simple address count does not capture the increasingly complex situation of usage of the IPv4 address space (Richter et al., 2016). The study counted 1.2 billion active, globally unique IPv4 addresses, more than any other previous estimation. Data show that the set of active IP addresses varies as much as 25% over the course of a year. The study presented implications for several stakeholders. For the measurement community, results show that remote active measurements are insufficient for an IP census, particularly at the IP-level granularity. For Internet governance purposes, the authors suggest that the use of metrics providing insight into the actual utilisation of the IPv4 address space can support governance bodies, such as Regional Internet Registries, in determining the compliance to their respective transfer policies. Network management professionals could gain better insights into their IPv4 assignment practices and achieve more efficient results. Finally, security professionals could take more informed decisions and better adjust host-based access controls and host reputation mechanisms.20

Connectivity for the Internet of Things can be provided through different wireless options

The IoT has a number of existing and emerging wireless options to provide connectivity. One is the use of LPWA communication technologies using unlicensed spectrum. Alternatively, standardised LPWA technologies for mobile operators using licensed spectrum are under development by the 3rd Generation Partnership Project and are expected to be commercially available in 2017. This technology is designed for M2M networking, aimed at interconnecting devices with low-bandwidth connectivity, while improving range and power efficiency.

Proponents say LPWA network technologies can effectively eliminate significant barriers for the development of IoT applications that do not require low latency networks, in particular related to device costs, power consumption and network deployment costs. Within this approach, the ability to use unlicensed bands, such as the ISM 868-902 Megahertz (MHz) in Europe and North America, and an increasing demand for low-power applications for the IoT have driven the development of two main competing LPWA technology systems: Sigfox and LoRa.

Sigfox, a company headquartered in France, was founded in 2009 and pioneered the use of cellular-type ultra-narrow band technology. As Sigfox’s infrastructure is independent of existing telecommunication operators, expanding its network requires the development of partnerships with local technology providers. By March 2017, Sigfox operated in 32 countries, with plans to expand its reach to 60 countries by 2018 (Sigfox, 2017). Total invested in the start-up at the end of 2016, and the company announced a partnership with Telefónica in March 2017 (Sigfox, 2017). The company, together with its partners, had nationwide coverage in countries such as France, Ireland, Luxembourg, the Netherlands, Portugal and Spain.

The LoRa Alliance was established to promote the LoRa protocol (LoRaWAN) as an open global standard for secure and carrier grade IoT connectivity. A certification programme for device manufacturers aims to guarantee compliance and interoperability between operators, one of the main challenges for establishing a global IoT. The Netherlands, Switzerland and Korea were the first three countries with nationwide LoRaWAN coverage, as announced by KPN, Swisscom and SK Telecom respectively between March and July 2016. In the Netherlands, at the time of the national launch, KPN had already contracted 1.5 million devices to be connected to its network stemming from the popularity of the service in its initial locations in Rotterdam and The Hague (KPN, 2016). Meanwhile, in Korea, SK Telecom announced it would invest USD 90 million in LoRa infrastructure and is expecting to connect 4 million IoT devices by the end of 2017 (Waring, 2016b). For its part, Swisscom aims for its LoRa network to feature 80% outdoor coverage as well as light indoor coverage in selected cities, such as Zurich, Geneva, Lausanne and many others.

Proponents say LoRa networks are a very cost-effective connectivity solution, particularly for public MNOs that seek to complement their current M2M product offering using 2G, 3G and 4G mobile networks. Existing transmission towers can be upgraded with certified LoRa equipment (antenna and gateway), creating a new connectivity solution for sensor-based applications. The long range and penetration features of the 900 MHz band allow ranges of 2 kilometres (km) to 5 km per antenna in dense urban environments and up to 15 km in suburban outdoor areas. According to the LoRa Alliance, its protocol offers many benefits over competing technologies, such as bi-directionality, security, mobility for asset tracking and accurate localisation (Lora Alliance, 2017).

Initial low-power, wide-area pricing approaches

As the market for IoT connectivity grows, network operators are developing new approaches for tariffs more suitable to the demands of the market. In many ways, the experimentation being seen in the first deployments mirrors that of any new network. Commercial offers and price plans for connected devices using LPWA networks vary significantly, even when operators use the same underlying technology. In the case of Sigfox, a cost indication provided in specialised media pointed at USD 1 per device per year for contracts with 50 000 or more devices (Shankland, 2016). Meanwhile, other approaches are being followed by operators in Korea and Switzerland (Table 3.1).

Table 3.1. Commercial offerings and price plans for low-power, wide-area networks
SK Telecom (Korea)			Swisscom (Switzerland)
Price plan	Data allowance1	Monthly flat rate	LPN bundle service per device	Number of messages per day2
Band IoT 35	100 kB	USD 0.30	XS	2/1
Band IoT 50	500 kB	USD 0.43	S	4/1
Band IoT 70	3 MB	USD 0.61	M	24/2
Band IoT 100	10 MB	USD 0.87	L	48/4
Band IoT 150	50 MB	USD 1.31	XL	96/9
Band IoT 200	100 MB	USD 1.75	XXL	144/14
← 1. Data usage exceeding the data allotment provided will be charged at KRW 0.005 per 0.5 kB.
← 2. Uplink/downlink.
IoT = Internet of Things; kB = kilobyte; MB = Megabyte.
Sources: SK Telecom (2016), “SK Telecom commercializes nationwide LoRa network for IoT”, www.sktelecom.com/en/press/detail.do?idx=1172; Swisscom (2017), “Low power network product and service overview”, http://lpn.swisscom.ch/e/our-offering (accessed 22 March 2017).

In Korea, SK Telecom offers six different price plans that include a data allowance at a monthly flat rate. The lowest cost plan, Band IoT 35, offers 100 kilobytes of data allowance at about USD 0.30 per month. For applications with greater data use, plans include Band IoT 100 for 10 MB at USD 0.87 per month and Band IoT 200 for 100 MB at approximately USD 1.75 per month (SK Telecom, 2016). SK Telecom’s LoRa services cost merely one-tenth of their LTE-based IoT services and extensive discount rates are offered to business customers depending on their contract duration and number of lines contracted.

In Switzerland, Swisscom’s LPN connectivity plans are designed as a bundle service per device. Instead of a data allowance, the bundle includes a number of uplink and downlink messages per day. The smallest package (XS) allows for 2 uplink and 1 downlink messages, the M package comes with 24/2 messages, and the largest plan (XXL) includes 144 uplink and 14 downlink messages (Swisscom, 2017).

Global roaming for the Internet of Things

Before LPWA specifications are included in the 3rd Generation Partnership Project standards used by the mobile industry, several connectivity players have announced their interest in establishing a LoRa-based global roaming system (Yoon, 2016). One such system would allow LoRaWAN end devices to be deployed in multiple networks and roam from one network to another irrespective of network infrastructure or operator. For such an international roaming system to be implemented, operators of LoRa networks would need to negotiate roaming agreements as the mobile industry has done for two decades.

References

ABS (Australian Bureau of Statistics) (2016), Australian Bureau of Statistics website, www.abs.gov.au (accessed July 2017).

Akamai (2017), “Akamai state of the Internet IPv6 adoption”, Akamai, Cambridge, Massachusetts, https://www.akamai.com/uk/en/our-thinking/state-of-the-internet-report/state-of-the-internet-ipv6-adoption-visualization.jsp.

Akamai (2016), “Akamai’s state of the Internet report: Q1 2016 report”, Akamai, Cambridge, Massachusetts, www.akamai.com/us/en/multimedia/documents/state-of-the-internet/akamai-state-of-the-internet-report-q1-2016.pdf.

Allevin, M. (2016), “Deutsche Telekom, Huawei among those testing LTZ-V for next-gen auto tech”, FierceWireless, 1 July, www.fiercewireless.com/tech/deutsche-telekom-huawei-among-those-testing-lte-v-for-next-gen-auto-tech.

APNIC (Asia Pacific Network Information Center) (2017), “IPv6 Measurement Maps”, webpage, http://stats.labs.apnic.net/ipv6 (accessed July 2017).

AT&T Inc. (2017), “ZTE Mobley”, webpage, https://www.att.com/devices/zte/mobley.html#sku=sku7700323 (accessed July 2017)

Chevrolet (2016), “Chevrolet lowers 4G LTE data pricing up to 50 percent”, press release, 29 June, http://media.chevrolet.com/media/us/en/chevrolet/home.detail.html/content/Pages/news/us/en/2016/jun/0629-onstarData.html (accessed 19 October 2016).

Google (2016), “Per-country IPv6 adoption”, webpage, https://www.google.com/intl/en/ipv6 (accessed July 2017).

Hammerschmidt, C. (2016), “Audi vehicles get their own IoT identity”, EE Times, 31 May, www.automotive-eetimes.com/news/audi-vehicles-get-their-own-iot-identity.

Hull, D. (2016), “The Tesla advantage: 1.3 billion miles of data: Silicon Valley and Detroit can’t keep up with Elon Musk’s trove of real-world metrics”, Bloomberg, 20 December, https://www.bloomberg.com/news/articles/2016-12-20/the-tesla-advantage-1-3-billion-miles-of-data.

KPMG (2016), “KPMG global semiconductor outlook 2016: Seismic shifts underway”, KPMG International Cooperative, https://assets.kpmg.com/content/dam/kpmg/pdf/2016/02/kpmg-global-semiconductor-outlook.pdf.

KPN (2016), “The Netherlands has first nationwide LoRa network for Internet of Things”, press release, 30 June, KPN, The Hague, https://www.kpn.com (accessed 22 September 2016).

Lora Alliance (2017), “LoRa Alliance™ Technology”, webpage, https://www.lora-alliance.org/What-Is-LoRa/Technology (accessed 22 March 2017).

OECD (2015), OECD Digital Economy Outlook 2015, OECD Publishing, Paris, https://doi.org/10.1787/9789264232440-en.

OECD (2014), “The Internet in transition: The state of the transition to IPv6 in today’s Internet and measures to support the continued use of IPv4”, OECD Digital Economy Papers, No. 234, OECD Publishing, Paris, https://doi.org/10.1787/5jz5sq5d7cq2-en.

OECD (2011), OECD Guide to Measuring the Information Society 2011, OECD Publishing, Paris, https://doi.org/10.1787/9789264113541-en.

PwC (PricewaterhouseCoopers) (2017), “PwC/CB Insights MoneyTree™ report Q4 and full-year 2016”, www.pwc.com/us/en/moneytree-report/assets/PwC%20&%20CB%20Insights%20MoneyTree%20Report%20-%20Q4%2716_Final%20V1.pdf.

Rajan, N. (2016), “Google’s free Wi-Fi: This is why it chose railway stations to connect India”, Indian Express, 5 August, http://indianexpress.com/article/technology/google-sundar-pichai-free-wifi-railtel-stations-2943720 (accessed 21 September 2016).).

Richter, P. et al. (2016), “Beyond counting: New perspectives on the active IPv4 address space”, ICM 2016 Proceedings, 14-16 November, Santa Monica, California, https://net.t-labs.tu-berlin.de/~prichter/imc174-richterA.pdf.

Shankland, S (2016), “’Sigfox’s Internet of Things network heads to Denmark, too”, CNET, 9 June, https://www.cnet.com/news/sigfox-internet-of-things-network-heads-to-denmark-too (accessed 22 September 2016).

Sigfox (2017), “Sigfox and Telefónica strike global deal to offer IoT services worldwide”, press release, 22 March, www.sigfox.com/en/news/sigfox-and-telefonica-strike-global-deal-offer-iot-services-worldwide (accessed 6 April 2017).

SK Telecom (2016), “SK Telecom commercializes nationwide LoRa network for IoT”, press release, 7 April, www.sktelecom.com/en/press/detail.do?idx=1172.

Swisscom (2017), “Low power network product and service overview”, webpage, http://lpn.swisscom.ch/e/our-offering (accessed 22 March 2017).

Tipan, E. (2016), “SENSORIS to fast-track development of self-driving cars”, Autoindustriya, 1 July, www.autoindustriya.com/auto-industry-news/sensoris-to-fast-track-development-of-self-driving-cars.html (accessed 22 September 2016).

Waring, J. (2016a), “Intel CEO: 5G crucial to manage coming M2M data flood”, Mobile World Live, 2 September, www.mobileworldlive.com/asia/asia-news/intel-ceo-says-coming-m2m-data-flood-requires-5g.

Waring, J. (2016b), “SK Telecom plans nationwide LPWA network based on LoRa”, Mobile World Live, 16 March, https://www.mobileworldlive.com/asia/asia-news/skt-plans-nationwide-lpwa-network-this-year.

Yoon, S.W. (2016), “SKT pushing for IoT global roaming”, The Korea Times, 14 July, www.koreatimes.co.kr/www/news/tech/2016/07/133_209420.html (accessed 22 September 2016).

Chapter 3. Access and connectivity1

Introduction

Trends in the ICT sector

The ICT sector has not fully recovered from the crisis, but computer and data-related services drive a positive outlook

Recent trends in value added and employment

Outlook for the ICT sector

The share of ICT goods and services in total trade continues to increase despite an overall decline in value terms

Trade in ICT goods

Trade in ICT services

ICTs play a key role in today’s innovation activities

Communication markets

Trends in the number of subscriptions and industry revenue appear to decouple

Broadband networks

Fixed broadband subscriptions continue to increase across the OECD

Speed and technology

Prices

Mobile broadband subscriptions are at a new high

Speed and technology

Prices

Substitution: Are we there yet?

The Internet of Things

Machine-to-machine subscriptions are increasing, marking the uptake of the Internet of Things

Connected autonomous vehicles are expected to generate large amounts of wireless data

Adoption of the Internet Protocol version 6 is progressing

The exhaustion of the Internet Protocol version 4 address space continues to be an issue

Connectivity for the Internet of Things can be provided through different wireless options

Initial low-power, wide-area pricing approaches

Global roaming for the Internet of Things

References

Notes