Closing Broadband Connectivity Divides for All

From Evidence to Practice

Report

10 July 2025

2. State of connectivity between and within countries

Copy link to 2. State of connectivity between and within countries

Abstract

This chapter provides a comparative analysis of broadband availability and performance across and within OECD Member countries and partner economies. It examines fixed and mobile broadband performance and coverage using harmonised subnational spatial classifications and third-party data through various indicators that capture user experience across multiple dimensions. The findings reveal significant territorial gaps that persist despite network rollouts, particularly in low- and middle-income countries. Finally, the chapter discusses how various factors such as income, education, age, gender, and trust in digital services influence digital divides across the OECD and beyond.

Broadband connectivity is the backbone of digital transformation. Access to affordable high-quality broadband services provides economic opportunities in OECD Member countries and regions (OECD, 2023[1]). Policy makers across the OECD are increasingly concerned with ensuring that citizens are not only connected but connected well in terms of access to high-quality broadband services at affordable prices.

There are three main components to assess the state of connectivity divides between and within countries: availability, affordability and high quality.

Availability: Broadband connectivity needs to be available everywhere geographically in a country, whether in cities, towns or rural areas.1 Having choice is even better. Namely, all citizens should ideally have more than one option of broadband provider and offer to subscribe. In this way, they can choose the option that best fits their needs.
Affordability: Broadband connectivity should be affordable for all citizens. Prices that are unaffordable for certain segments of the population reinforce digital divides, regardless of whether broadband offers are available in their geographic area.
High quality: Broadband connectivity should be high quality to allow users to fully experience a full range of use cases, especially those requiring high bandwidth, high reliability and low latency, which are increasingly prevalent. Having high-quality, reliable broadband connectivity is also affected by the state of other critical infrastructure, such as access to a reliable electricity grid.

While affordability is a key aspect when considering whether a user is connected well, this report focuses on broadband availability and performance. For further discussion on the affordability of broadband communication offers across the OECD, see the chapter “Trends in access and connectivity” in OECD (2024[2]).

This chapter first considers the availability of broadband communication offers from the perspective of access. Can a user contract a broadband service anywhere in the territory? It presents available data for fixed and mobile networks. For fixed networks, it relates to metrics whereby individuals or households living in different areas of a given country can have access to high-quality broadband offers. For mobile networks, it portrays indicators on the share of the population covered. Moreover, the percentage of time with a signal by technology is presented to provide further insight into how often a user is on a 3G, 4G or 5G network in the places they frequent.

While complementary metrics on the geographical footprint or coverage of fixed and mobile networks would be ideal, there are several hurdles. On a technical level, it is challenging to collect data to create and maintain an up-to-date broadband map. This prompted the OECD to convene the Expert Group on Broadband Mapping to discuss and share national experiences in these areas. In addition, countries use different classifications of geographical areas. This makes it difficult to harmonise geographical broadband coverage data collected by communication regulators.

Following the discussion on availability, the chapter considers broadband performance. What is the quality of broadband connectivity in the country, across rural and remote areas, towns and metropolitan areas?

This discussion covers several indicators to assess broadband quality. Download and upload speeds of the network connection are assessed over time, and by geography where data exist, for fixed and mobile networks. It also presents latency over time and by geography. Latency, also called the delay or ping rate, is the round-trip time for information to travel between two devices across the network (OECD, 2019[3]). Higher latency translates into delayed network response times (OECD, 2022[4]). Therefore, the lower the latency, the better the user experience.

To consider additional aspects of mobile broadband quality, this chapter employs a composite “consistent quality” metric that examines several indicators, including “download throughput, upload throughput, latency, jitter, packet discard [loss] and time to first byte” (Opensignal, 2025[5]). It also presents data on the percentage of time that users have no mobile signal. Such “user centric” approaches to understand and assess network experience seek to go beyond traditional quality of service to measure quality of experience (OECD, 2022[4]).

This report considers these questions – of being connected and connected well – from the perspective of OECD Member countries, as well as of partner economies for which data are available.

Third-party source data on availability and quality of broadband for 61 countries (38 OECD Member countries and 23 selected partner economies)2 are used in this publication. The partner economies selected include eight accession countries (Argentina, Brazil, Bulgaria, Croatia, Indonesia, Peru, Romania and Thailand) and 15 additional countries of relevance for the study (Bangladesh, Cambodia, India, Kenya, Nigeria, the People’s Republic of China [hereafter ‘China’], Philippines, Rwanda, Senegal, Singapore, the Solomon Islands, South Africa, the United Republic of Tanzania [hereafter ‘Tanzania’], Togo and Viet Nam). These selected countries include key partners and countries of particular interest for the study, for instance in the Southeast Asia region, and selected countries for development co-operation based on official development assistance (ODA) priorities.

The underlying data sources for indicators on the quality of broadband connectivity in this chapter are third-party providers, Ookla and Opensignal. The different methodologies of these two providers are complementary, offering different perspectives of the Internet to provide a more comprehensive view of broadband performance. At a high level, Ookla compiles its results from speed tests launched by users towards their “test servers” to measure broadband performance. It uses quality of service metrics such as speeds and latency, as well as quality of experience metrics. Opensignal collects end-to-end data from real-world users, focusing on quality of experience. The data are collected from the device, such as mobile phones, to common Internet endpoints, including content delivery networks (CDNs) – measuring the entire connectivity chain. For more detailed discussion on each methodology, please see Box ‎2.1.

Box ‎2.1. Methodologies of third-party measurements of actual broadband speeds

Copy link to Box ‎2.1. Methodologies of third-party measurements of actual broadband speeds

The two main third-party sources used in this report to measure actual broadband speeds, Ookla and Opensignal, have differing methodologies that should be considered when assessing their data.

Ookla has two primary datasets shared for this report. One is generated from the consumer-initiated Speedtest quality of service (QoS) tests. The other is generated from samples taken from a range of host applications known as Ookla’s consumer quality of experience (QoE) dataset. The consumer- initiated tests are directed to Ookla’s “dedicated test servers”, which are distributed all over the world and hosted in numerous different Internet points. This is a multi-server QoS test designed to run in the foreground (via applications of websites), maximising the full bandwidth of four test servers. As such, it is a QoS test measuring the sustained peak throughput achievable on the access network by users.

In practice, when a user launches a Speedtest, the device pings Ookla’s nearby dedicated testing servers, saturates the network connection and measures the sustained peak speed achieved by the device during the test window (OECD, 2021[6]).Therefore, the measured indicator does not reflect everyday speeds experienced by users. Rather, it measures the actual maximum speeds attainable by the network connection when a user’s device sends the maximum amount of data to four of Ookla’s over 15 000 test servers (OECD, 2022[4]; Ookla, 2024[7]). The consumer QoE datasets are focused on coverage, video capabilities and file transfer to content delivery networks and cloud hosting servers. These are more representative of a user’s day-to-day network experience.

Opensignal’s methodology for broadband speeds measures the “end-to-end” consumer network experience. In other words, Opensignal tests are conducted from the device to common Internet endpoints, such as CDNs like Google, Akamai and Amazon (OECD, 2022[4]). This mimics the way connections are made each day to typical websites and content, providing a view on end-to-end network experience.

Opensignal’s measurements provide insights on how consumers are using their digital devices (e.g. time using applications like video and games) to support network improvements. The measurements are designed to prevent operators from optimising test traffic; they are not allowed to treat the tested traffic differently, which would affect results without improving their networks (Opensignal, 2024[8]).

Source: Adapted from Box 6 of OECD (2022[4]), “Broadband networks of the future”, OECD Digital Economy Papers, No. 327, https://doi.org/10.1787/755e2d0c-en.

While informative, these metrics are influenced by market circumstances. For instance, the data on fixed network performance reflect offers on the market, as well as subscriptions (e.g. whether a user has a 1 Gbps subscription versus a 100 Mbps one). Consequently, the metrics are not a pure measurement of a network’s theoretical capacities. Similarly, mobile users could only connect to an available 5G network with a 5G-capable mobile device and a 5G subscription; such connections may be influenced by variables such as income levels and the price of 5G services and devices. Other factors may also influence these metrics. For example, the topology of the access network, as well as the physical condition of network equipment and infrastructure, could affect performance.3

With a view to support the eventual goal of enabling cross-country comparison across the OECD, this report applies established geographical classifications to the data from Ookla and Opensignal. The OECD’s territorial definitions and classification of regions based on accessibility to cities – small regions (Territorial Level 3, or TL3) – is applied where possible. Certain datasets go beyond OECD Member countries where the TL3 regional spatial classification cannot be applied. These datasets use harmonised geographical classification based on population settlement grids and classified by the degree of urbanisation (DEGURBA). Box ‎2.2 provides further information on both classifications.

All graphs will list which one of the two methodologies applies. For this report, TL3 classification is used for graphs based on Ookla data. This classification is not available for partner economies. Therefore, the DEGURBA territorial classification for subnational analysis of the third-party broadband performance data is used for graphs containing data for non-OECD countries. Likewise, all graphs based on Opensignal data present results where the DEGURBA territorial classification has been applied.

Box ‎2.2. OECD territorial definitions and classification of regions

Copy link to Box ‎2.2. OECD territorial definitions and classification of regions

The OECD has developed territorial classifications that aim to provide an analytical tool to better understand regional developments, both within and across countries. These classifications may be based on administrative units (e.g. small regions (TL3)) or grids (spatial units of the same shape and size, e.g. degree of urbanisation).

Small regions (TL3) Classification

Regions within the 38 OECD Member countries are classified into two territorial levels reflecting the administrative organisation of countries. The 433 OECD large (TL2) regions represent the first administrative tier of subnational government (OECD, 2023[9]). The 2 414 OECD small (TL3) regions correspond to administrative regions, except for Australia, Canada, Latvia and the United States (OECD, 2023[9]).

TL3 regions can be classified based on the presence/absence of metropolitan areas and the level of accessibility of the latter by the population in each region. The concept of metropolitan area corresponds to a functional urban area (FUA) of at least 250 000 inhabitants (Fadic et al., 2019[10]). FUAs are identified based on relatively high-resolution (1 km²) population grids. The classification further distinguishes between regions, as follows (Fadic et al., 2019[10]):

TL3 classification	Description	Group
Large metropolitan region	Region with a very large city >1.5M inhabitants	Metropolitan region
Metropolitan region	Region with a large city >250k inhabitants	Metropolitan region
Region near a metropolitan area	Region near a city >250k inhabitants	Regions near a metropolitan area
Region with/near a small-medium city	Region near a city between 50k and 250k inhabitants	Regions far from a metropolitan area
Remote region	Remote region	Regions far from a metropolitan area

Degree of Urbanisation (DEGURBA) classification

The DEGURBA classification is based on the same delineation criteria for all regions/countries. It produces a harmonised and universal mapping along the urban-rural continuum. DEGURBA combines population size and population density thresholds to a 1 km² population grid (OECD et al., 2021[11]). As a benefit of using the grid, all spatial units (cells) have the same shape and size, and their borders are stable over time. In this way, it captures three mutually exclusive classes: “Cities”, “Towns and semi-dense areas”, and “Rural areas” (level 1 of the DEGURBA classification).

Cities consist of contiguous grid cells with a density of at least 1 500 inhabitants per km² or are at least 50% built up. They must have a population of at least 50 000.
Towns and semi-dense areas consist of contiguous grid cells with a density of at least 300 inhabitants per km² and are at least 3% built up. They must have a total population of at least 5 000.
Rural areas are cells that do not belong to a city or a town and semi-dense area. Most of these have a density below 300 inhabitants per km².

The second level of DEGURBA captures the full settlement hierarchy of cities, towns and villages, which are often quite heterogeneous and do not identify specific types of settlement. “Towns and semi-dense areas” are split into three subclasses: “Dense towns”; “Semi-dense towns”; and “Suburban or peri-urban” (OECD et al., 2021[11]). Rural areas are split into three subclasses: “Villages”; “Dispersed rural areas”; and “Mostly uninhabited areas” (OECD et al., 2021[11]).

Being connected well in all places: Broadband coverage

Copy link to Being connected <em>well</em> in all places: Broadband coverage

The first criterion for being connected well is broadband availability or access. This means that citizens can contract broadband services from anywhere in their geographic territory. Thereby, at least one broadband service offer is available, although more than one is preferable for consumer choice and competition. This availability is usually expressed in coverage metrics; however, it does not reveal whether a user has actually contracted a service (only that the user could contract a service).

OECD Member countries see fixed and mobile broadband as complementary services rather than as substitutes (OECD, 2022[4]). Fixed access at households is indispensable for ensuring the full benefits of digital transformation and should not be overlooked where there is already mobile coverage (OECD, 2023[1]; 2022[12]). In a “remote economy”, more and more business processes move on line and people increasingly work and learn from home. Given the trend towards a remote economy, and the continuous growth of data-intensive applications, the demand for high-quality networks is only expected to increase. Fibre needs to be deployed deeper into networks to increase broadband performance across all access technologies, including mobile networks. Network densification inherent to 5G will require greater fibre rollout to support backhaul capacity. In addition, more fibre is needed to bring frontier technologies, such as the Internet of Things (IoT) and artificial intelligence (AI), to their full potential (OECD, 2022[4]).

Knowing the availability of broadband services capable of providing given speeds can help policy makers better understand coverage. Many countries assess and publish their own statistics on broadband coverage on both national and subnational bases, including for rural households.

High-speed broadband coverage in rural areas remains a major challenge for many OECD Member countries. The availability of fixed broadband networks, in terms of geographical coverage, that are capable of providing speeds of at least 30 Mbps and of at least 100 Mbps reveals significant gaps between rural and urban households. Available data for OECD Member countries show that around half (14 of 27) had fixed broadband coverage of at least 30 Mbps download speeds for 95% or more of total households in 2023. However, the share of rural households with access to fixed broadband networks capable of delivering such speeds was frequently lower. Overall, 92.3% of total households across the OECD had access to a fixed broadband offer capable of delivering at least 30 Mbps compared to 78.5% for rural households in 2023 (Figure ‎2.1, Panel A). In the European Union, for comparison, 92.95% of households overall lived in areas where fixed broadband networks offered services capable of providing speeds of at least 30 Mbps. However, only 78.65% of rural households had similar coverage in 2023 (Figure ‎2.1, Panel A). With respect to fixed broadband coverage at higher speeds, while 81.3% of households overall were located in areas with fixed broadband networks capable of delivering 100 Mbps download speeds, only 58.7% of rural households were covered with such speeds across the OECD (Figure ‎2.1, Panel B).

With respect to mobile networks, 4G, and increasingly 5G networks are increasingly prevalent in OECD Member countries, as measured by the percentage of the population covered. As of Q4 2024, 4G mobile networks around the OECD reached 99.1% of the population, 3G networks covered 93.8% and 5G networks had 83.6% population coverage (GSMA Intelligence, 2025[13]).

Measuring coverage of a mobile network by the share of the population covered has a caveat: it does not represent the network’s geographical footprint in the same way as the percentage of a country’s territory covered by a mobile network. Some countries with population highly concentrated in urban areas may have great disparity between these two metrics.4

Beyond the OECD, 4G networks reached at least 95% of the national population in Q4 2024 among most of this group. Exceptions occurred in Nigeria, the Solomon Islands and the United Republic of Tanzania (hereafter “Tanzania”) where 3G networks were more prevalent (Figure ‎2.2) (GSMA Intelligence, 2025[13]). 5G network coverage is lower than the OECD average in most partner economies included in the report. However, Bulgaria, Croatia, People’s Republic of China (hereafter “China”), Singapore and Thailand reported 5G population coverage above the OECD average (GSMA Intelligence, 2025[13]).

The 5G network coverage shown in Figure ‎2.2 comprises both standalone (SA) and non- standalone (NSA) 5G networks. However, 5G SA deployments have the potential for enhanced network performance compared to NSA networks, the latter of which leverage 4G core network infrastructure and use NSA-5G radio interface standards. Most 5G networks available commercially to date are based on NSA-5G (OECD, 2024[2]). Globally, 154 operators in 63 countries have invested in public 5G-SA deployments, either through trials, or planned and actual deployments. Meanwhile, at least 67 operators in 35 countries had launched public 5G-SA networks as of January 2025 (GSA, 2025[14]).

Figure ‎2.1. Fixed broadband coverage is not yet ubiquitous within OECD Member countries
Copy link to Figure ‎2.1. Fixed broadband coverage is not yet ubiquitous within OECD Member countries

Figure ‎2.2. 3G and 4G networks reach more people, although 5G population coverage is growing
Copy link to Figure ‎2.2. 3G and 4G networks reach more people, although 5G population coverage is growing

3G, 4G and 5G network coverage as a percentage of the population in select partner economies, Q4 2024

Mobile user experience metrics assessing the proportion of time users had a 2G, 3G, 4G or 5G signal may shed further insights on network availability (Figure ‎2.3). Opensignal’s mobile signal availability metric aims to assess users’ mobile experience in the places they frequent, although it does not measure the geographical coverage of a mobile network (Opensignal, 2025[5]). According to Opensignal data in Q4 2024, users across the OECD, on average, spent most of their time on 4G networks (86.1%), followed by 5G networks (8.8%), 3G networks (3.1%) and 2G networks (1.1%). They spent less than 1% of time with no mobile network signal in Q4 2024 (Figure ‎2.3).

While the time without a mobile network signal is relatively small, when it occurs it can range from a minor inconvenience to undertake daily tasks to more critical consequences, depending upon the application and circumstance. For instance, not having mobile signal on a remote roadway may be dire in the case of autonomous driving. In addition, as shown later in Figure ‎2.19, geographic differences exist; users in rural areas have a higher percentage of time without a mobile signal than those in cities across the OECD.

Australia, Korea, and the United States led the OECD with the highest percentage of time spent on 5G networks in Q4 2024, with the United States reporting 26.7%, followed by Australia (19.1%) and Korea (19.0%). Several factors play a role in this metric, including spectrum policy decisions. Moreover, only users with a 5G device and subscription can connect to a 5G network. Take-up of 5G services depends on users’ interest, as well as the affordability of the services and devices. This may help explain the difference between the percentage of time that users experienced a 5G signal (OECD average 8.8%, according to Opensignal data for Q4 2024) and the percentage of the population covered by a 5G network (OECD average of 83.6%, according to GSMA Intelligence data for Q4 2024) (Figure ‎2.2). Additionally, business users of 5G may report different results, given the growing emergence of private networks in different sectors and 5G B2B use cases.

Figure ‎2.3. Users across the OECD spent most time connected through 4G networks
Copy link to Figure ‎2.3. Users across the OECD spent most time connected through 4G networks

Percentage of time with a signal by mobile technology, Q4 2024

From a more granular perspective, Opensignal data as of Q4 2024 show that users in cities across most OECD Member countries report a higher percentage of time spent on 5G networks than users in rural areas. This may reveal the nature of 5G deployments that often require network densification (OECD, 2019[20]), where initial rollouts are often made in densely populated areas. It could also indicate users’ choice to delay upgrading mobile devices in rural areas, which are often correlated with lower income levels. Regardless, investments by network operators to upgrade their networks to the latest mobile technology are an encouraging development to ensure networks will be fit for purpose. These upgrades may privilege urban areas in the short term. Consequently, further expansion into all areas should be supported through an enabling policy environment to promote investment and ease infrastructure deployment, as expounded in “Chapter 3. Policies to bridge connectivity divides”.

Broadband maps play a key role in providing end-user transparency and increasing the effectiveness of broadband policy measures. Improving the accuracy of broadband data at a granular level is crucial. Broadband maps can influence the allocation of funds to close connectivity gaps in unserved and underserved areas. For example, harmonising subnational coverage indicators across the OECD through a “cross-border” broadband map would greatly facilitate cross-country comparison. This, in turn, would allow better understanding of broadband availability across regions. Given the complexities of collecting data to create an up-to-date cross-country broadband map, the Expert Group on Broadband Mapping, convened by the OECD, provided a forum of exchange on broadband mapping experiences (Box ‎2.3).

Box ‎2.3. The OECD Expert Group on Broadband Mapping and Digital Divides

Copy link to Box ‎2.3. The OECD Expert Group on Broadband Mapping and Digital Divides

The Expert Group held seven meetings between Q3 2023 and Q3 2024 with 30-35 experts from ten countries attending regularly (i.e. Australia, Brazil, Canada, Colombia, France, Japan, Mexico, Portugal, Sweden and the United States). The Expert Group also included participation of the International Telecommunication Union and invited experts in broadband measurement from GSM Association (GSMA), the Internet Society and Packet Clearing House.

The Expert Group identified several challenges for building a broadband map. The foundational level of locating households, buildings and roads on maps was cited as a key difficulty for two reasons.1 First, the available datasets are low resolution, i.e. they are not plotting households as individual geographic datapoints. Second, operators use different datasets to plot or report their own networks, which are challenging to integrate and reconcile. Disputes over the accuracy of the information on a map is another obstacle. For example, homeowners may refute a map that indicates their households are covered by technology capable of certain speeds. Technology-specific challenges for broadband access technologies relying on wireless last-mile connections, such as fixed-wireless access and direct-to-home satellite, are also important. Establishing common parameters for models that estimate possible coverage of different systems, while theoretically possible, is complex, technical and contentious.

1. As another challenge, addresses do not always reflect the geographic location of houses.

Being connected well in all places: Broadband performance

Copy link to Being connected <em>well</em> in all places: Broadband performance

Being connected well also means that broadband services are both available and of high quality. There are several aspects of quality, including download and upload speeds, latency, reliability and quality of experience (OECD, 2022[4]). However, “high quality” is an ever-changing target, and one highly dependent on the evolving needs of users. Broadly, the experienced quality of broadband services should be good enough to support modern applications, even at peak periods of usage across a network. Considering download and upload speeds as an example, broadband services should be sufficient to support common applications such as videoconferencing and streaming. For example, Zoom recommends 3.8 Mbps upload speeds and 3 Mbps download speeds for high definition (1080p) group video calling (Zoom, 2024[21]). For its part, Netflix recommends 15 Mbps or higher for ultra-high definition (4K) streaming (Netflix, 2024[22]; Zoom, 2024[21]). However, other factors can come into play, such as user equipment and network congestion (e.g. two users video calling on the same fixed home connection), which may affect experienced quality.

This section elaborates on trends in the quality of fixed and mobile communication networks across geographical regions using harmonised spatial classifications. Broadband performance metrics explored include download and upload speeds, latency and network reliability (e.g. through metrics such as consistent quality and percentage of time without a mobile signal).

Relative to national performance, there is still a large geographical divide in fixed and mobile broadband download speeds

Beyond differences between countries, connectivity divides in the urban-rural continuum within countries is a substantial challenge in terms of broadband speeds. Territorial differences in connectivity also translate into user experiences that vary substantially depending on where people live or work. This is evidenced by the differences in actual download speeds experienced by individuals in cities compared to those living in regions far from metropolitan areas. According to data from self- administered connection speed tests by Ookla, people living in metropolitan regions in OECD Member countries on average experience faster mean fixed and mobile download speeds than those living in regions far from a metropolitan area (Figure ‎2.4).5 Across the OECD, mean download speeds over fixed networks in regions far from metropolitan areas were on average 23.5 percentage points below those experienced in metropolitan areas in Q4 2024.

However, this average masks substantial variations within OECD Member countries in the differences in fixed download speeds experienced across regions. For example, in Colombia, Greece and Türkiye, the gap between regions far from metropolitan areas and those in metropolitan areas was over 40 percentage points. On the other hand, countries like Korea, the Netherlands and Norway show small differences in fixed broadband download speed across regions (five percentage points or less between the fastest and slowest regions) over the same period as measured by Ookla.

The territorial gap in mobile download speeds is wider than in fixed download speeds. On average, in OECD Member countries, Ookla data show a gap of 35.0 percentage points in mobile speeds between metropolitan regions and regions far from metropolitan areas. This worked out to 10.0% faster speeds in metropolitan regions than the national mean, and 25.0% slower speeds than the national mean in non-metropolitan regions far from metropolitan areas in Q4 2024.

The variation between fastest and slowest regions in terms of mean mobile download was also larger than the variation in fixed download speeds. For example, gaps between region types were over 40 percentage points in seven countries (Australia, Belgium, Colombia, Ireland, Poland, Portugal and the United Kingdom). Of these, four had a gap of 50 percentage points or more (Belgium, Colombia, Poland and the United Kingdom), as of Q4 2024 in Ookla data.

The trend in absolute median fixed broadband speeds (Figure ‎2.5) echoes the one in percentage deviation (Figure ‎2.4). People in cities experienced median fixed broadband download speeds 43.8% higher than people living in regions far from metropolitan areas in Q4 2024, according to Ookla data (Figure ‎2.5). The countries with the largest absolute gaps in the levels of median speeds between metropolitan regions and regions far from metropolitan areas were Colombia (106.1 Mbps), Switzerland (94.3 Mbps), Canada (71.2 Mbps), Poland (62.8 Mbps), the United Kingdom (58.3 Mbps), the United States (56.8 Mbps) and Hungary (53.9 Mbps), in Q4 2024 based on Ookla data. For Canada, Hungary, Switzerland and the United States, despite the magnitude of the absolute domestic gap, the regions far from metropolitan areas showed median speeds higher than the OECD average for that type of region (i.e. 132.7 Mbps).

Figure ‎2.4. Across the OECD, most remote regions have fixed and mobile download speeds slower than national means, with considerable differences in magnitude
Copy link to Figure ‎2.4. Across the OECD, most remote regions have fixed and mobile download speeds slower than national means, with considerable differences in magnitude

Percentage deviation in mean fixed and mobile download speeds from national means, Q4 2024, small regions (TL3) classification

Figure ‎2.5. Country-by-country, there is significant variation in levels of fixed broadband speeds
Copy link to Figure ‎2.5. Country-by-country, there is significant variation in levels of fixed broadband speeds

Median fixed broadband download speeds in the OECD, Q4 2024, small regions (TL3) classification

Geographical gaps persist over time in the OECD, despite increases in fixed download speeds

In terms of regional differences in absolute median levels of fixed broadband download speeds (Figure ‎2.6), geographical connectivity divides persist and are growing in time. This is observed despite a reduction in relative difference in terms of percentage deviation observed previously (see Figure 1.2 in Chapter 1).6 When measuring median speed levels, median fixed download speeds for OECD Member countries increased, on average, from 52.9 Mbps in Q4 2019 to 177.8 Mbps in Q4 2024 (Figure ‎2.6). While median fixed broadband speeds across all regions improved overall, the gap between the regions with the fastest and the slowest speeds widened over this period. In the fourth quarter of 2019, the gap between metropolitan areas and regions far from a metropolitan area across the OECD was 22.3 Mbps. However, by the fourth quarter of 2024, the gap grew to 58.4 Mbps, based on analysis of Ookla data.

According to data from Ookla, median fixed download speeds for OECD Member countries more than tripled in five years, on average. This represented 236.3% growth from Q4 2019 to Q4 2024, or a compound quarterly growth rate of 5.9% (OECD average, Table ‎2.1). Aggregate growth in upload speeds in OECD Member countries also saw strong growth at 318.7% in the same period. Meanwhile, network response times (i.e. latency) improved with a 23.0% reduction.

Figure ‎2.6. Fixed broadband download speeds increased in all regions, but gaps in speeds for users living outside metropolitan areas also increased compared to those living within them
Copy link to Figure ‎2.6. Fixed broadband download speeds increased in all regions, but gaps in speeds for users living outside metropolitan areas also increased compared to those living within them

Median fixed broadband download speeds in the OECD, small regions (TL3) classification

Growth rates	Metropolitan regions	Regions near a metropolitan area	Regions far from a metropolitan area	OECD average
Compound quarterly (%)
Download	5.9	6.1	6.5	5.9
Upload	7.1	6.6	7.2	7.1
Latency	-1.2	-1.4	-1.4	-1.2
Aggregate (%)
Download	232.08	247.67	276.29	236.26
Upload	322.08	283.40	329.66	318.72
Latency	-22.24	-25.50	-25.20	-22.98

Notes: Based on data for 34 OECD Member countries where data are available (complete data for Costa Rica, Iceland, Israel and Luxembourg were not available). Measurements are based on tests performed by users around the globe via the Speedtest platform. For a more comprehensive picture on broadband performance metrics, see OECD (2022[4]). Within small regions (TL3 classification), the OECD has three main classifications: “Metropolitan regions”, “Regions near a metropolitan area”, and “Regions far from a metropolitan area”. The last category has two further subcategories: “Regions close to small/medium city” and “Remote regions” (see https://doi.org/10.1787/20737009).

Source: Based on OECD analysis of Ookla’s Speedtest Intelligence data.

Within countries, remote regions – coming from lower initial levels of broadband performance – saw stronger improvements than metropolitan regions in fixed download and upload speeds, as well as in latency, in terms of percentage aggregate growth. For example, in the Q4 2019-Q4 2024 period, metropolitan areas saw a growth in fixed download median speeds of 232.1% compared to 276.3% in regions far from metropolitan areas. Fixed upload speeds in metropolitan areas grew by 322.1%, with regions far from metropolitan areas exhibiting the largest growth of 329.7%. Improvements in latency were largest in regions near a metropolitan area (25.5% reduction), with a similar improvement in regions far from a metropolitan area (25.2% reduction), followed by metropolitan regions (22.2% reduction) from Q4 2019 to Q4 2024 (Table ‎2.1).

Meanwhile, the territorial gap in levels (i.e. absolute values) between metropolitan areas and regions far from a metropolitan area in latency is narrowing slightly. However, the territorial gaps in terms of levels of fixed download and upload speeds between metropolitan regions and regions far from a metropolitan area are increasing (Figure ‎2.7). These trends are likely explained by network upgrades, which began in metropolitan areas to meet demand. At the same time, they highlight how the gaps in broadband performance are persistent over time and the need for continued efforts to bridge connectivity divides.

Figure ‎2.7. Absolute gaps across regions in upload and download speeds have been growing over the past five years, while gaps in latency have narrowed slightly
Copy link to Figure ‎2.7. Absolute gaps across regions in upload and download speeds have been growing over the past five years, while gaps in latency have narrowed slightly

Index of the gap between OECD average metropolitan regions and regions far from metropolitan areas for fixed download and upload speeds (Mbps) and latency (milliseconds), small regions (TL3) classification

How are national average speeds associated with territorial gaps?

An increase in average national median fixed download speeds is positively associated with larger gaps in the median fixed download speed levels between metropolitan and non-metropolitan regions (Figure ‎2.8). A one unit increase in national average fixed download speeds is associated with a 0.13 Mbps larger gap in absolute fixed download speeds in Q4 2024, according to Ookla data. This can be expected, as gaps in absolute terms tend to be greater the higher the level of absolute speeds. It also reflects the natural progression of investment in communication networks.

Network operators must continue to invest to upgrade networks to ensure they are equipped to meet ever-increasing demands stemming from digital transformation. Urban and metropolitan regions have a larger pool of subscribers to recoup investments. They also have potentially more initial demand for advanced services requiring higher quality networks. Consequently, the business case in these areas makes them the natural choice for expansion or upgrades. More investment in urban and metropolitan regions could, in turn, increase connectivity gaps unless operators are incentivised to invest in all territories.

The role of policy makers, therefore, is to incentivise and ease the deployment of high-quality networks into all areas and regions. These networks should enable everyone to access a network connection sufficient to allow their meaningful participation in society (especially in key areas like the economy, public services and education). The deployment of such networks should, where possible, avoid increasing persistent gaps. This will be discussed in more detail in Chapter 3 of the report.

Figure ‎2.8. Territorial gaps in fixed download speeds slightly widen as national speeds increase
Copy link to Figure ‎2.8. Territorial gaps in fixed download speeds slightly widen as national speeds increase

Absolute gaps in median fixed download speeds in the OECD by national average fixed download speeds, Q4 2024, small regions (TL3) classification

Territorial gaps beyond OECD Member countries

Territorial gaps between countries also exist beyond the OECD. According to data from user-initiated tests, select partner economies in this report (i.e. 14 of 20 countries where data were available) experienced lower mean fixed broadband download speeds compared to G20 countries (with an average of 184.7 Mbps) in Q4 2024 (Ookla) (Figure ‎2.9). The countries with higher national mean fixed download speeds, as reported by user-initiated tests, were Singapore (402.8 Mbps), China (327.2 Mbps), Thailand (288.8 Mbps), Romania (257.2 Mbps), Peru (224.9 Mbps) and Bulgaria (223.2 Mbps).7

Figure ‎2.9. Attaining high fixed download speeds is a challenge for some partner economies
Copy link to Figure ‎2.9. Attaining high fixed download speeds is a challenge for some partner economies

Mean fixed download speeds in select partner economies, Q4 2024, degree of urbanisation classification

There is substantial variation in the mean fixed download speeds (absolute values) between the partner economies included in the report. Among OECD accession countries, the mean speeds ranged from 51.4 Mbps (cities) and 31.9 Mbps (rural areas) in Indonesia to 304.2 Mbps (cities) and 257.5 Mbps (rural areas) in Thailand in Q4 2024, according to analysis of Ookla data. For the remaining partner economies in the report, mean fixed speeds in cities ranged between 28-40 Mbps (Kenya, Nigeria and Togo) to over 400 Mbps (Singapore). Rural areas similarly saw wide ranges in speeds, from between 30-40 Mbps (Kenya, Nigeria and Togo) to over 300 Mbps (Singapore) (Figure ‎2.9).

Within this group of partner economies, most countries have a geographical disparity, with higher fixed download speeds in cities compared to rural areas. Kenya, Nigeria and Togo were the exceptions, where speeds in rural areas slightly beat fixed broadband download speeds in cities. This may be driven in part by the low level of tests for these countries.

The absolute gap of mean fixed download speeds between urban and rural areas for G20 countries was 64.9 Mbps, on average. This can also be expressed as mean speeds that were 42.8% higher in cities than in rural areas. The OECD accession countries with the largest absolute geographical gaps were Peru at 173.7 Mbps, Croatia at 108.1 Mbps, Bulgaria at 104.3 Mbps, Brazil at 97.8 Mbps and Argentina at 93.8 Mbps. Speeds in cities in Peru were 246.4% higher than in rural areas, followed by Argentina at 139.8%, Brazil at 123.1% and Croatia at 85.4% in Q4 2024, according to analysis of Ookla data.

These partner economy results highlight common challenges to tackle disparities between quality of connectivity both across and within countries. The high speeds in cities in some OECD accession countries demonstrate progress towards upgrading networks and improving quality in these areas. However, the degree of inequity in the quality of connectivity within certain of these countries also represents an opportunity for policy makers. They have a role to ensure their policies incentivise and ease the deployment of high-quality networks to support the objective of all areas being well connected.

Geographical gaps in mobile speeds increasing with technology advancement

The past five years have seen improvements in median mobile download speeds, across all regions. According to Ookla data, the median mobile download speed for metropolitan areas around the OECD was 31.9 Mbps on average and 27.1 Mbps in regions far from metropolitan areas in the fourth quarter of 2019. This increased to 126.4 Mbps in metropolitan areas and 81.5 Mbps in regions far from metropolitan areas in Q4 2024 (Figure ‎2.10). However, median speeds in metropolitan areas across the OECD increased more than in regions far from metropolitan areas. In the fourth quarter of 2019, the gap between regions was 4.7 Mbps. However, by the fourth quarter of 2024, the gap grew to 44.9 Mbps, based on OECD analysis of Ookla data.

Figure ‎2.10. Territorial gaps in mobile download speeds are growing across the OECD
Copy link to Figure ‎2.10. Territorial gaps in mobile download speeds are growing across the OECD

Median mobile download speeds, small regions (TL3) classification

There was an uptick in speeds experienced by metropolitan users towards the end of 2020 and the beginning of 2021 in countries across the OECD (Figure ‎2.10). This increase coincides with the commercial launch of 5G networks in several OECD Member countries and around the world. From 2020 to 2021, 5G connections as a percentage of total mobile connections in the OECD grew from 2.1% to 8.4%, rising to 38.3% at the end of 2024 (Figure ‎2.11) (GSMA Intelligence, 2025[13]).

The rollout of 5G networks around the OECD and the accompanying growth in 5G connections may explain the jump in median mobile download speeds across the OECD seen in Figure ‎2.10. Successful upgrades to the network raise the quality of mobile networks. These should be encouraged to make sure networks across the OECD are fit to meet current and future demands. As noted above, in the short term this may increase spatial connectivity gaps due to the initial business case to deploy networks in densely populated areas to recover build-out costs. Raising the bar across the board should – and has so far – improved the quality of mobile broadband speeds across all regions. Nevertheless, closing the gap between regions should still be a focus for policy makers so that citizens everywhere can fully benefit from the opportunities afforded by high-quality connectivity.

Figure ‎2.11. 5G connections have grown globally since 2019
Copy link to Figure ‎2.11. 5G connections have grown globally since 2019

Percentage of 5G connections in total mobile connections

Data from Opensignal (see methodological explanation in Box ‎2.1) similarly show variations in mobile download speeds between and within OECD Member countries (Figure ‎2.12). For the OECD, overall mobile download speeds experienced in Q4 2024, on average, were 74.5 Mbps in cities, 66.2 Mbps in towns and semi-dense areas, and 54.3 Mbps in rural areas, as measured by Opensignal (Figure ‎2.12). This means that, in the OECD, users in cities experienced mobile download speeds 37.2% higher than those in rural areas.

Korea stands apart from other countries in this dataset. Mobile download speeds in rural areas of Korea were 156.1 Mbps in Q4 2024, faster than those in cities in all other OECD Member countries. Furthermore, the gap in the level of mobile speeds (i.e. absolute values rather than relative gaps) between all three regions in Korea were minimal (less than 10 Mbps). Other countries with low absolute gaps between cities and rural areas include Greece, Israel and Türkiye, with gaps of 2 Mbps or less, and Chile, Colombia, Italy and Mexico (gaps of 5 Mbps) in Q4 2024 as reported by Opensignal (Figure ‎2.12). However, these countries, apart from Korea, reported speeds across all three geographic classifications lower than OECD averages. Gaps tend to be narrower when the level of speeds across all regions is lower.

The largest gaps in overall mobile download speeds experienced by urban users compared to those in rural areas in Q4 2024 were reported in Sweden and Denmark (absolute value gap of 58.0 Mbps), Norway (44.6 Mbps) and Australia (44.2 Mbps). However, some of these countries have among the highest mobile download speeds experienced in cities and were above OECD averages across all three geographic classifications (Figure ‎2.12). As one possible cause of these gaps, operators in these countries have upgraded mobile networks starting in urban areas due to demand and the quickest recoup of their investment. This underscores the persistence of connectivity divides; service in rural areas may improve while networks in cities are being upgraded, moving the target for rural areas once again.

Looking specifically at 5G, urban users of 5G in most OECD Member countries reported higher mobile download speeds than users in rural areas, according to Opensignal data in Q4 2024 (Figure ‎2.13). In that period, 5G mobile download speeds experienced in the OECD by users in cities (222.6 Mbps) was, on average, 28.3% higher than in rural areas (173.5 Mbps). Korea was the leader in 5G speeds, for all regions. As seen in overall mobile download speeds (Figure ‎2.12), users in rural areas in Korea similarly experienced 5G speeds faster than those in cities in all other OECD Member countries (Figure ‎2.13). Across the OECD, on average, download speeds experienced by users in 5G for each of the three regions were close to, or more than, three times higher than the OECD average overall mobile download speeds for the respective region (Figure ‎2.12 and Figure ‎2.13).

Figure ‎2.12. Users in cities across the OECD experienced mobile download speeds 37% higher than those in rural areas
Copy link to Figure ‎2.12. Users in cities across the OECD experienced mobile download speeds 37% higher than those in rural areas

Overall mobile download speeds experienced in cities, towns and rural areas, Q4 2024, degree of urbanisation classification

Figure ‎2.13. In all OECD Member countries where 5G has been deployed, 5G download speeds surpass 85 Mbps in all regions
Copy link to Figure ‎2.13. In all OECD Member countries where 5G has been deployed, 5G download speeds surpass 85 Mbps in all regions

Gaps in 5G mobile download speeds experienced in cities, towns and rural areas, Q4 2024, degree of urbanisation classification

Beyond the OECD, there is a wide variation in mobile download speeds experienced in the partner economies included in the report, according to Opensignal data. Nevertheless, a common theme is evident: users in cities experienced higher speeds than those in rural areas in Q4 2024. In G20 countries, users in urban areas experienced mobile download speeds 31.7% higher than their rural counterparts, on average, in Q4 2024 (Figure ‎2.14). In all partner economies included in the report, where data exist, the level of mobile speeds across all regions was lower compared to G20 averages, except for Singapore, Bulgaria, Croatia and India. In those countries, speeds across all regions were higher than G20 averages.

In terms of urban-rural gaps, Bulgaria (gap of 45.3 Mbps), India (26.0 Mbps), Croatia (25.2 Mbps), Thailand (16 Mbps) and Romania (15.8 Mbps) reported the largest gaps in levels of speeds (absolute values) in Q4 2024, according to Opensignal. On the other end of the spectrum, South Africa reported less than 1 Mbps difference between regions. Other countries with small absolute gaps between cities and rural areas include Singapore (1.6 Mbps), Indonesia (3.2 Mbps), and Cambodia and Peru (3.6 Mbps) in Q4 2024.

As in OECD Member countries, partner economies with narrow absolute gaps tended to have lower speeds across all regions, such as in Cambodia, Indonesia, Peru and South Africa. Singapore is a notable exception in this regard, with high speeds and a narrow gap between regions. However, this may be explained in part by the small geographic area of the city-state.

Gaps in terms of absolute values were wider across regions in Bulgaria, India and in Croatia in Q4 2024, although this was coupled with higher overall speeds. As seen across the OECD (Figure ‎2.12), these gaps imply a positive development as they indicate investment in mobile network upgrades. For example, users in rural areas in Bulgaria and Croatia experienced mobile download speeds of 63.5 Mbps and 59.7 Mbps, respectively, in Q4 2024 as measured by Opensignal. This was higher than the speeds in cities for most partner economies included in Figure ‎2.14 and higher than the G20 average for rural areas of 48.5 Mbps.

Figure ‎2.14. Speeds are higher in cities than in rural areas, although with a wide disparity between partner economies
Copy link to Figure ‎2.14. Speeds are higher in cities than in rural areas, although with a wide disparity between partner economies

Overall mobile download speeds experiences in select partner economies, Q4 2024, degree of urbanisation classification

Latency over fixed networks has improved since 2019, although territorial gaps remain

Latency is highly influential on the user’s experience. Ookla estimates that humans can perceive a delay as small as 15 milliseconds (ms), with applications such as voice and video calling requiring a latency of no greater than 300 ms for basic functioning (Ookla, 2023[25]). However, for some advanced applications such as remote surgery or autonomous driving, connections must provide near instantaneous feedback (i.e. low latency) for optimal performance. For example, within a latency of 250 ms, a car travelling at 96 kilometres per hour (60 miles per hour) would travel 6.8 metres (22.5 feet), a distance that could cause accidents in the case of an autonomous vehicle (Ookla, 2023[25]). Within this context, improvements in latency can have a high impact on quality of experience.

Over the past five years, latency over fixed networks in the OECD has steadily improved across all regions and the territorial gap is narrowing. Between the fourth quarters of 2019 and 2024, median fixed latency across the OECD fell from 16.6 ms to 12.8 ms on average, demonstrating a 23% decline over this period, based on OECD calculations of Ookla data. Similarly, the gap between metropolitan areas and regions far from metropolitan areas dropped from 5.5 ms to 3.6 ms over the same period. Improvements in latency were more prominent in regions far from (25.2% reduction), and in regions near to metropolitan areas (25.5% reduction) over the five-year period. This compared to a 22.2% decrease for metropolitan areas (Ookla).

Nevertheless, territorial gaps persist (Figure ‎2.15). People in cities across the OECD experienced, on average, 23% lower median latency over fixed networks compared to people living in regions far from metropolitan areas in Q4 2024, based on OECD calculations of Ookla data. This equated to a median latency of 12.1 ms for metropolitan areas in the OECD, and 15.8 ms for regions far from a metropolitan area. Gaps between metropolitan areas and regions far from metropolitan areas were 1 ms or less in Czech Republic (hereafter “Czechia”), Denmark, Finland, Ireland, Norway and Slovenia. However, Australia and Mexico reported territorial gaps of more than 10 ms (absolute values), followed by Japan (9 ms gap) in Q4 2024 according to analysis of Ookla data.

Figure ‎2.15. Internet users in cities across the OECD experienced 23% lower fixed broadband median latency on average, compared to people in regions far from metropolitan areas
Copy link to Figure ‎2.15. Internet users in cities across the OECD experienced 23% lower fixed broadband median latency on average, compared to people in regions far from metropolitan areas

Median latency for fixed networks, Q4 2024, small regions (TL3) classification

Outside of the OECD, there is a wide disparity between partner economies in terms of both mean latency over fixed networks and territorial gap between regions. Moreover, the territorial gap in mean latency between rural areas and cities is substantial in some partner economies (Figure ‎2.16). For instance, Kenya, South Africa and Tanzania reported territorial gaps of more than 35 ms in mean latency between cities and rural areas in Q4 2024. Meanwhile, Argentina, Indonesia, Peru, the Philippines and Togo reported gaps of more than 20 ms, according to Ookla data.

Overall, in 13 of the partner economies in the study, Ookla data showed rural areas having connections with mean latency at least 10 ms higher than cities in Q4 2024 (Figure ‎2.16). Across the G20, on average, users in rural areas experienced connections with mean latency 15 ms higher than users in cities in Q4 2024 (28.6 ms compared to 13.3 ms), according to Ookla data.

Figure ‎2.16. Territorial gap in latency between rural areas and cities is substantial in some partner economies
Copy link to Figure ‎2.16. Territorial gap in latency between rural areas and cities is substantial in some partner economies

Mean latency over fixed networks in select partner economies, Q4 2024, degree of urbanisation classification

Rural areas lag behind urban areas on other mobile quality indicators measuring network response times and reliability

Latency across mobile networks improved over the past five years, as measured by Ookla, but a territorial gap remains. From Q4 2019 to Q4 2024, the OECD average of median latency over mobile networks decreased from 34.2 ms to 28.1 ms. These decreases were seen in OECD averages across all geographic regions (i.e. metropolitan areas, as well as regions both near and far from metropolitan areas) over the same period. The absolute gap between metropolitan areas and regions far from metropolitan areas also decreased slightly, from 6.4 ms in Q4 2019 to 6.1 ms in Q4 2024 as measured by Ookla. By geography, the median latency decreased at similar rates over the five years (e.g. 20.6% in regions near, and 19.5% in regions far, from metropolitan areas), but with metropolitan areas decreasing at a slightly faster rate (23%). Thus, Ookla data suggest that median mobile latency across regions improved over the past five years. However, a territorial gap remains as regions far from metropolitan areas did not “catch up” to their counterparts in metropolitan regions.

Opensignal uses the “consistent quality” metric to measure additional indicators of broadband quality beyond latency, which comprises “download throughput, upload throughput, latency, jitter, packet discard and time to first byte” (Opensignal, 2025[5]). Opensignal sets thresholds for each component indicator to support requirements for common applications such as video calling, e-commerce and mobile banking without disruptions. These thresholds are based on real-world performance requirements from widely accepted industry standards, service provider recommendations, and user experience expectations. Opensignal calculates the percentage of tests that meet or exceed each threshold across the six components of the consistent quality metric.8

In Q4 2024, Opensignal data show a territorial gap between cities and rural areas in the OECD, although there is a wide variation across countries. On average, across the OECD, 78.3% of tests in cities benefited from consistent quality compared to 73.1% in rural areas in Q4 2024 (Figure ‎2.17). In some countries, the gap between cities and rural areas was small. Denmark and Finland lead the pack with consistent quality scores above 85% for all regions. There was a negligible difference recorded in Czechia, Switzerland and Spain (less than a percentage point in favour of cities). Austria, Japan and the Netherlands also had relatively similar levels of consistent quality across regions, with differences between rural areas and cities of less than two percentage points in Q4 2024.

However, users in cities in Colombia, Costa Rica and Mexico had more consistent quality than did rural users, with gaps greater than ten percentage points in Q4 2024, as measured by Opensignal. Even a difference of ten percentage points may greatly affect the rural users’ mobile experience, frustrating their attempts to do common tasks on their devices. It means that rural users will be unable to engage in listed use cases (i.e. video calling, uploading images on social media, or smart home applications) an additional 10% of the time compared to urban users.

Figure ‎2.17. Consistent quality is higher in cities than in rural areas on average, across the OECD
Copy link to Figure ‎2.17. Consistent quality is higher in cities than in rural areas on average, across the OECD

Consistent quality, Q4 2024, degree of urbanisation classification

Nearly all partner economies had territorial gaps in consistent quality akin to OECD Member countries. Opensignal results in Q4 2024 showed territorial gaps in consistent quality in all but one partner economies outside the OECD included in the study for which data exist. South Africa was the exception, with just over a percentage point gap, with rural areas slightly beating cities (Figure ‎2.18).

The G20 average saw a three percentage-point gap between cities and rural areas. Bangladesh, Bulgaria, Croatia and Singapore also reported gaps between regions of less than four percentage points of difference between cities and rural areas in Q4 2024. Six partner countries (Argentina, Brazil, Indonesia, the Philippines, Peru and Thailand) reported gaps of greater than ten percentage points, with users in cities having a higher percentage of consistent quality than rural users in Q4 2024 (Figure ‎2.18).

Bulgaria, Croatia, Romania and Singapore reported high levels of consistent quality in Q4 2024 across regions, beating G20 averages (Figure ‎2.18). Other partner economies had lower levels of consistent quality (absolute value) across regions. Network improvements could therefore increase users’ quality of experience not only in rural areas but across the board.

Figure ‎2.18. Consistent quality varies in partner economies, and is higher in cities than rural areas
Copy link to Figure ‎2.18. Consistent quality varies in partner economies, and is higher in cities than rural areas

Consistent quality in select partner economies, Q4 2024, degree of urbanisation classification

Perhaps more telling from a perspective of quality is the amount of time spent with no mobile signal. As seen in Figure ‎2.3, lack of mobile signal represents a small proportion of time overall across the OECD according to Opensignal data in Q4 2024 (less than 1% on average across regions). Nevertheless, the data reveal variations between OECD Member countries and between geographies (Figure ‎2.19). In the OECD, on average, rural areas had a slightly higher percentage of time without a mobile signal compared to their urban counterparts in Q4 2024 (0.9% in rural areas compared to 0.7% in cities). This was more pronounced in some OECD Member countries. Over the same period, rural areas in Colombia, Costa Rica and Mexico reported more than 2% of time without mobile signal. Meanwhile, urban areas in these countries reported around 1% or less of time without a mobile signal.

Beyond the OECD, partner economies in the study varied widely in the percentage of time without a mobile network. Some partner economies reported levels similar to, or even better than, OECD averages in Q4 2024, with less than 1% of time without a mobile signal based on Opensignal data. For instance, Bulgaria, Cambodia, Croatia, India, Singapore, South Africa, Thailand and Viet Nam reported less than 1% of time without a mobile signal across all regions. The territorial gap between cities and rural areas was particularly high (above two percentage points) in Argentina, Brazil, Nigeria, Peru and the Philippines, according to Opensignal data for Q4 2024.

Figure ‎2.19. Users in rural areas across the OECD have a higher percentage of time without a mobile signal than those in cities
Copy link to Figure ‎2.19. Users in rural areas across the OECD have a higher percentage of time without a mobile signal than those in cities

Time without mobile signal, Q4 2024, degree of urbanisation classification

Various factors influence digital divides across the OECD and beyond

Copy link to Various factors influence digital divides across the OECD and beyond

Several factors can affect the use and adoption of communication services, which then influence broader digital divides. The previous chapter presented comprehensive data showing persistent disparities across places in the availability and quality of fixed and mobile communication services, across the OECD and beyond. However, divides along other dimensions such as between men and women, or by age, educational attainment and income level, also influence the use and adoption of communication services. The users’ level of trust in digital services and applications, as well as the relevance of digital content and availability in local languages, may also affect adoption of these services. This, in turn, may limit access to essential online services or employment opportunities (OECD, 2022[26]). While these trends are apparent in countries both in and outside the OECD, they are often amplified in low- and middle-income countries.

Beyond expanding access to broadband, there is still work to be done to close digital divides

Across the OECD, there remains a divide in adoption of basic broadband services (fixed or mobile, of at least an advertised speed of 256 Kbps) between urban and rural areas. In 2024, according to data from 28 OECD Member countries, urban households were, on average, 4 percentage points more likely than rural households to have subscribed to basic broadband services (OECD, 2025[27]). This gap has narrowed in recent years, falling from a high of 11 percentage points in 2008 (OECD, 2025[27]). Nevertheless, the gap has remained relatively stable since 2020 at between four and five percentage points (OECD, 2025[27]).

Income level often plays a role in the use of digital technologies across the OECD. In 2023, more individuals in the fourth income quartile across the OECD reported using the Internet daily or almost every day within the last three months compared to their counterparts in the first income quartiles. On average across the OECD, 67.5% of individuals aged 16-74 living in households in the first income quartile used the Internet daily or almost every day. Conversely, 93.4% of individuals aged 16-74 living in households in the fourth income quartile in 2023 used the Internet every day or almost daily. This represents a 25.9 percentage-point gap between the two income quartiles (OECD, 2024[28]).

Internet usage gaps between income groups have improved over the last decade in OECD Member countries. In 2013, 45.7% of individuals aged 16-74 living in households in the lowest income quartile used the Internet daily or almost every day, compared to 82.9% of individuals in the highest income quartile (37.2 percentage-point gap) (OECD, 2024[28]).

Affordability is one of the main barriers to broadband uptake in many OECD Member countries. While analysis of affordability is outside the scope of this report, OECD (2024[2]) examines affordability with respect to fixed and mobile communication services. It finds a significant decrease in mobile broadband prices over 2013‑23 in OECD Member countries. However, prices of bundled communication services across the OECD over the more recent past (2020-23) remained relatively stable. The exception was for the medium-usage profile, triple-play bundle (i.e. fixed broadband, fixed voice and television), which saw a 19.5% decrease (OECD, 2024[2]). Improving the affordability of communication services, as well as devices, leads to greater adoption and a more inclusive participation in the digital economy and society across all income levels.

Other dimensions, such as age and educational level, similarly play a role in use of the Internet. In 2024, 97.0% of individuals aged 16-24 across the OECD, on average, reported using the Internet daily or almost every day in the past three months compared to 76.4% of those aged 55-74 (OECD, 2025[27]). Across the OECD, individuals with lower levels of educational attainment also tend to use the Internet less than those with higher educational levels. In 2024, 79.4% of individuals with no or a low level of educational attainment used the Internet daily or almost every day, while 97.0% of individuals with a high educational attainment did (OECD, 2025[27]). Additionally, the level of ICT skills affects use of the Internet, and digital tools and services more broadly. A higher level of ICT skills, as well as other foundational and complementary skills, enable a more diverse and complex use of the Internet and online tools.

Digital inclusion of women and girls is also unequal in some cases around the OECD, not only in terms of use and adoption of digital technologies, but also in their role to develop them. Previous OECD work has noted that women hold fewer senior-level positions, receive lower pay and participate at a lower rate in science, technology, engineering and mathematics (STEM) fields in labour markets around the world (UNESCO/OECD/IDB, 2022[29]). Moreover, data from 30 OECD Member countries reported in 2024 that young men (16-24 years old) are more than twice as likely as young women to be able to programme, on average (OECD, 2025[27]). This divide between men and women can have an especially important influence on the ability to mitigate potential bias in general purpose technologies in the future.

Digital divides can be major challenges in low- and lower middle-income countries

Other partner economies face similar disparities in the use and adoption of digital technologies. However, while all countries experience these trends, they are amplified in low- and middle-income countries and may be exacerbated by other challenges, such as access to basic infrastructure, as described below. Digital divides can thus compound existing challenges such as lower returns to work, poorer access to public services and social exclusion. A comparison of basic indicators on digital access and use show that low- and lower middle-income countries experience the greatest digital divides, on average (Table ‎2.2).

	Fixed broadband subscriptions per 100 inhabitants	Mobile cellular telephone subscriptions per 100 inhabitants	Active mobile broadband subscriptions per 100 inhabitants	Percentage of individuals using the Internet	Ratio of youth (15-24) to rest of population using Internet	Population covered by at least a 3G mobile cellular network, rural (%)
LIC	0.5	71.0	40.1	26.5	1.88	69.5
LMIC	4.8	94.5	66.3	54.0	1.41	92.1
UMIC	32.0	129.5	111.7	80.7	1.24	93.9
HIC	38.1	138.2	152.7	93.4	1.06	97.1

Notes: LIC=Low-income country; LMIC=Lower middle-income country; UMIC=Upper middle-income country; HIC=High-income country.

Source: OECD based on data from ITU (2024[30]), ITU World Telecommunication/ICT Indicators database, www.itu.int/en/ITU-D/Statistics/Pages/stat/default.aspx.

Supply- and demand-side challenges drive digital divides in low- and middle-income countries

Low- and middle-income countries face several major barriers to reduce digital divides, although they equally have much to gain from closing them. This is especially true considering economic hardships in recent years marked by rising debts, inflation, shifts in trading patterns and wage contraction. However, several significant barriers hamper efforts to expand digital access and use.

Basic infrastructure gaps are the foundation of digital divides in low- and middle-income countries

Access to electricity, a prerequisite for digital access and use, remains a significant issue particularly in sub-Saharan Africa. According to projections, 645 million people will remain without access to electricity in 2030 (IEA, 2024[31]). Of these, 85%, or about 545 million people, will be in sub-Saharan Africa (IEA, 2024[31]). The investment required to reach universal access to electricity by 2030 is projected to be USD 50 million each year, on average (IEA, 2024[31]).

Within low- and middle-income countries, low-population-density areas are often characterised by scarce power and transportation infrastructure, as well as challenging geography. This, together with the lower household incomes commonly found in such areas, further reduces financial incentives for private sector investment in communication infrastructure (Williams and Bachiri, 2021[32]).

Demand-side barriers drive digital divides in low- and middle-income countries

Demand-side barriers, such as affordability, low skill levels, lack of relevant content, and safety and security, are also driving digital divides in low- and middle-income countries. Most people who remain unconnected to mobile Internet – most of them in low- and middle-income countries – live in an area already covered by mobile broadband but do not use mobile Internet services (Cruz and Tiel Groenestege, 2021[33]).9 A number of demand-side barriers account for this disparity. Affordability of handsets and services are critical obstacles. In 2020, almost 2.5 billion people lived in countries where the most affordable smartphone cost more than a quarter of the average monthly income (Sarpong, 2021[34]). In the same year, more than 1 billion people lived in countries where 1 gigabyte (GB) of mobile broadband data was available at a cost of more than 2% of per capita gross national income (GNI); in the Central African Republic, for example, 1 GB of data cost 24.4% of GNI per capita in that year (Sarpong, 2021[34]). Some low- and middle-income countries apply taxes on communication devices or services (Matheson and Petit, 2021[35]), which could exacerbate affordability concerns. Other barriers include low levels of skills and lack of relevant content, including content in local languages. These can lead to lower demand for communication services. Additional demand-side barriers arise due to safety, security and data protection concerns. These may include risks associated with misinformation, online harassment, scams and theft.

Low trust in government authorities to protect rights online may also hinder adoption and usage of digital services. At one end of the spectrum lies issues related to poor oversight capacity, such as concerns about consumer protection in e‑commerce markets. Such concerns, for example, were identified as a key driver for lower-than-expected online purchases in Thailand (Attrey, 2021[36]). Fear that digital identification systems would fail to protect personal data is another concern (Masse et al., 2021[37]). At the other end are issues related to the use of digital tools for surveillance, disinformation and reduction of rights (Khan, 2021[38]; Roberts and Bosch, 2021[39]). Research in ten African countries also found evidence of Internet shutdowns, SMS and app blocking, targeting of specific groups and arrests for online speech (Roberts and Bosch, 2021[39]).

Demand-side barriers tend to affect women and girls disproportionately, contributing to digital divides between men and women. In 2024, women were 14% less likely to use mobile Internet compared to men in low- and middle-income countries (LMIC) (GSMA, 2025[40]). Of the 885 million women who were not using mobile Internet in LMICs in 2024, around 60% lived in South Asia and sub-Saharan Africa, where gaps between men and women in mobile Internet use are widest (GSMA, 2025[40]). Moreover, women are also more likely to be disadvantaged by social norms. Restricted movement, for example, may prevent them from purchasing devices or opening accounts in their names. Women are also less likely to have the necessary foundational identity documents for registration (Cruz and Tiel Groenestege, 2021[33]). Safety and security risks, such as online harassment or bullying, may also disproportionately affect women and girls compared to male counterparts. When factoring in other parameters beyond ability to connect, research shows that women often have poorer quality of access or inadequate devices relative to connected men (Sarpong, 2021[34]). In addition, women in LMICs are 14% less likely to own a smartphone compared to men in 2024 (GSMA, 2025[40]).

References

[36] Attrey, A. (2021), “Agile, comprehensive and principles-based: Policy making for a digital age”, in Development Co-operation Report 2021: Shaping a Just Digital Transformation, OECD Publishing, Paris, https://doi.org/10.1787/ce08832f-en.

[16] CRTC (2024), Communications Monitoring Report 2024, (database), https://crtc.gc.ca/eng/publications/reports/PolicyMonitoring/ban.htm (accessed on 4 June 2025).

[33] Cruz, G. and M. Tiel Groenestege (2021), “Tackling digital disadvantage with people-centred policies”, in Development Co-operation Report 2021: Shaping a Just Digital Transformation, OECD Publishing, Paris, https://doi.org/10.1787/ce08832f-en.

[17] European Commission (2024), “Digital decade 2024: Broadband coverage in Europe 2023”, report commissioned by European Commission, Luxembourg, OMDIA and Point House, https://digital-strategy.ec.europa.eu/en/library/digital-decade-2024-broadband-coverage-europe-2023.

[10] Fadic, M. et al. (2019), “Classifying small (TL3) regions based on metropolitan population, low density and remoteness”, OECD Regional Development Working Papers, No. 2019/06, OECD Publishing, Paris, https://doi.org/10.1787/b902cc00-en.

[18] FCC (2025), “Area Summary”, FCC National Broadband Map, (database), https://broadbandmap.fcc.gov/area-summary (accessed on 4 June 2025).

[14] GSA (2025), “5G- Standalone January 2025 Summary”, GSA Report, Global mobile Suppliers Association, Surrey, United Kingdom, https://gsacom.com/paper/5g-standalone-january-2025/ (accessed on 14 February 2025).

[40] GSMA (2025), Mobile Gender Gap Report 2025, Global System for Mobile Communications Association, London, http://www.gsma.com/r/wp-content/uploads/2025/05/The-Mobile-Gender-Gap-Report-2025.pdf.

[24] GSMA (2023), The State of Mobile Internet Connectivity Report 2023, Global System for Mobile Communications Association, London, https://www.gsma.com/r/somic/.

[13] GSMA Intelligence (2025), Database, http://www.gsmaintelligence.com/data/ (accessed on 27 September 2024).

[31] IEA (2024), SDG7: Data and Projections, IEA, Paris, https://www.iea.org/reports/sdg7-data-and-projections.

[30] ITU (2024), “Individuals Using the Internet”, ITU World Telecommunication/ICT Indicators, No. Version November 2024, for Facts and Figures 2024, (database), https://www.itu.int/en/ITU-D/Statistics/Pages/stat/default.aspx (accessed on 31 January 2025).

[38] Khan, I. (2021), “In my view: To tackle disinformation, we must uphold freedom of opinion and expression”, in Development Co-operation Report 2021: Shaping a Just Digital Transformation, OECD Publishing, Paris, https://doi.org/10.1787/ce08832f-en.

[37] Masse, E. et al. (2021), “Case study: Combatting cyber threats, disinformation, and Internet shutdowns”, in Development Co-operation Report 2021: Shaping a Just Digital Transformation, OECD Publishing, Paris, https://doi.org/10.1787/ce08832f-en.

[35] Matheson, T. and P. Petit (2021), “Taxing telecommunications in developing countries”, International Tax and Public Finance, Vol. 28, pp. 248-280, https://doi.org/10.1007/s10797-020-09621-6.

[22] Netflix (2024), “Internet Connection Speed Recommendations”, webpage, https://help.netflix.com/en/node/306 (accessed on 18 October 2024).

[27] OECD (2025), Going Digital Toolkit, (database), https://goingdigital.oecd.org/ (accessed on 30 September 2024).

[28] OECD (2024), “ICT Access and Usage by Households and Individuals”, OECD Data Explorer, (database), http://oe.cd/hhind (accessed on 30 September 2024).

[2] OECD (2024), OECD Digital Economy Outlook 2024 (Volume 2): Strengthening Connectivity, Innovation and Trust, OECD Publishing, Paris, https://doi.org/10.1787/3adf705b-en.

[1] OECD (2023), Enhancing Rural Innovation in the United States, OECD Rural Studies, OECD Publishing, Paris, https://doi.org/10.1787/22a8261b-en.

[9] OECD (2023), OECD Territorial Grids, OECD, Paris, http://www.oecd.org/content/dam/oecd/en/data/datasets/oecd-geographical-definitions/territorial-grid.pdf.

[4] OECD (2022), “Broadband networks of the future”, OECD Digital Economy Papers, No. 327, OECD Publishing, Paris, https://doi.org/10.1787/755e2d0c-en.

[26] OECD (2022), “Building better societies through digital policy: Background paper for the CDEP Ministerial meeting”, OECD Digital Economy Papers, No. 338, OECD Publishing, Paris, https://doi.org/10.1787/07c3eb90-en.

[12] OECD (2022), Rural Policy Review of Colombia 2022, OECD Rural Studies, OECD Publishing, Paris, https://doi.org/10.1787/c26abeb4-en.

[6] OECD (2021), Bridging digital divides in G20 countries, OECD Publishing, Paris, https://doi.org/10.1787/35c1d850-en.

[3] OECD (2019), “The operators and their future: The state of play and emerging business models”, OECD Digital Economy Papers, No. 287, OECD Publishing, Paris, https://doi.org/10.1787/60c93aa7-en.

[20] OECD (2019), “The road to 5G networks: Experience to date and future developments”, OECD Digital Economy Papers, No. 284, OECD Publishing, Paris, https://doi.org/10.1787/2f880843-en.

[11] OECD et al. (2021), Applying the Degree of Urbanisation: A Methodological Manual to Define Cities, Towns and Rural Areas for International Comparisons, OECD Regional Development Studies, OECD Publishing, Paris/European Union, Brussels, https://doi.org/10.1787/4bc1c502-en.

[7] Ookla (2024), “Ookla’s Speedtest® Methodology”, webpage, https://www.ookla.com/resources/guides/speedtest-methodology (accessed on 26 September 2024).

[25] Ookla (2023), “Latency is the next frontier of consumer experience. Are you ready?”, 8 February, Ookla, https://www.ookla.com/articles/latency-the-next-frontier-of-consumer-experience.

[19] Opensignal (2025), Insights, http://www.opensignal.com.

[5] Opensignal (2025), “Mobile Network Experience Metrics”, http://www.opensignal.com/our-approach/mobile-metrics (accessed on 5 May 2025).

[8] Opensignal (2024), “Insights Built on Real-world Network Experiences”, webpage, https://www.opensignal.com/our-approach (accessed on 26 September 2024).

[39] Roberts, T. and T. Bosch (2021), “Case study: Digital Citizenship or Digital Authoritarianism?”, in Development Co-operation Report 2021: Shaping a Just Digital Transformation, OECD Publishing, Paris, https://doi.org/10.1787/5jrqdl7rvns3-en.

[34] Sarpong, E. (2021), “Beyond the basics: Quality, speed, affordability, relevance”, in Development Co-operation Report 2021: Shaping a Just Digital Transformation, OECD Publishing, Paris, https://doi.org/10.1787/5jrqdl7rvns3-en.

[23] Speedtest by Ookla (2025), Global Fixed and Mobile Network Performance Maps, http://www.ookla.com.

[29] UNESCO/OECD/IDB (2022), The Effects of AI on the Working Lives of Women, UNESCO, Paris, https://doi.org/10.1787/14e9b92c-en.

[15] United States Census Bureau (2023), “Urban and Rural”, webpage, https://www.census.gov/programs-surveys/geography/guidance/geo-areas/urban-rural.html (accessed on 6 December 2024).

[32] Williams, M. and O. Bachiri (2021), “Financing options for the future of digital”, in Development Co-operation Report 2021: Shaping a Just Digital Transformation, OECD Publishing, Paris, https://doi.org/10.1787/ce08832f-en.

[21] Zoom (2024), “Zoom system requirements: Windows, macOS, Linux”, 11 February, Zoom, https://support.zoom.com/hc/en/article?id=zm_kb&sysparm_article=KB0060748.

Notes

Copy link to Notes

← 1. With respect to the territorial classifications used in this report, the analysis based on the Ookla data for OECD Member countries uses the TL3 regions: “Metropolitan regions” (i.e. cities), “Regions near a metropolitan area” and “Regions far from a metropolitan area”. For the analysis of Ookla data for partner economies and for Opensignal data, the degrees of urbanisation levels used are “Cities”, “Towns and semi-dense areas” and “Rural areas”.

← 2. The 61 countries that were the focus of this report comprise:

38 OECD Member countries,
8 OECD accession countries: Argentina, Brazil (key partner country), Bulgaria, Croatia, Indonesia (key partner country), Peru, Romania and Thailand,
3 OECD key partner countries: China (People’s Republic of), India and South Africa,
3 priority Southeast Asian countries: Singapore, Cambodia, and Viet Nam,
9 countries from the list of lower middle-income and Least Developed Countries suggested by the OECD’s Development Co-operation Directorate based on official development assistance (ODA) priorities across regions: Bangladesh, Kenya, Nigeria, Philippines, Rwanda, Senegal, Solomon Islands, Tanzania and Togo.

← 3. In the case of user-initiated tests, these may introduce a selection bias as users may be more likely to test their speeds while experiencing connection issues (i.e. abnormal network conditions) or may be more “tech savvy” than the general population.

← 4. For instance, a country where most people live in a few large metropolitan areas may have a large mobile footprint as a percentage of the population, but a small footprint with respect to national territory. Nevertheless, coverage in terms of percentage of the population data is often more widely available and can still be helpful to consider.

← 5. Figure ‎2.4 expands upon the final period (Q4 2024) in Figure ‎1.2 by showing for each country how its geographical regions differ from the national mean.

← 6. As explained previously, this can occur when absolute values analysed are relatively high, such that despite a reduction in relative terms of percentage difference gaps (average dispersion from the mean), the actual gap in level values in terms of medians is higher. Nevertheless, it reflects two important trends: the relative performance of each observational unit (regions) and the absolute performance of each observational unit (regions).

← 7. Ookla data depend on the existence and frequency of user-initiated tests for speed. As such, where no tests were initiated, no speed results can be reported. Significant efforts are undertaken to reduce the impact of any bias on results that may occur in some cases. For this report, the analysis excluded countries where any subnational area had fewer than 1 000 tests reported in Q4 2024, as described in the note below Figure ‎2.9. However, further caution may be warranted when interpreting the analysis of these data from large countries with diverse geographies. In addition, comparisons of fixed broadband download speeds between OECD member countries (Figure ‎2.5) and non-OECD countries (Figure ‎2.9) are not possible. This is due to differences in geospatial classification and the unavailability of median fixed download speed data for non-OECD countries.

← 8. According to Opensignal’s methodology, to calculate this value, “the proportion of tests that pass the requirements of Consistent Quality is multiplied by the test success ratio, which is the proportion of completed tests to all tests conducted. Tests that pass indicate these activities will be possible without noticeable lag or slowdown (Opensignal, 2025[5]).”

← 9. This finding is according to GSMA as cited in Cruz, G. and M. Tiel Groenestege (2021[33]), “Tackling digital disadvantage with people-centred policies”, in Development Co‑operation Report 2021: Shaping a Just Digital Transformation, OECD Publishing, Paris, https://doi.org/10.1787/ce08832f-en.

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Countries

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Publications

Publications

Featured publications

Data

Data

Featured data

News & events

News & events

Featured events

About OECD

About

Engage with us

Work with us

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment