Broadband
Broadband Speeds COVID-19

How global broadband speeds changed during COVID-19 lockdown periods

Cable.co.uk analysed data from the Oxford Coronavirus Government Response Tracker (OxCGRT), and over 364 million broadband speed tests courtesy of M-Lab to compare average internet speeds in 114 countries during and outside of their most stringent COVID-19 lockdown periods

Download data (.xlsx)

Interactive map

Using the map

The map above is interactive. You can use the +/- symbols to zoom in and out. Hovering over an individual country will reveal the statistics uncovered in the study as they apply to that specific country.

Data available in the hover-over includes the start and end dates for the country's lockdown period. Note that this is where the OxCGRT tracker records a lockdown stringency index score exceeding 50, rather than the first or last day any form of lockdown measures came into force. Likewise, the end date is either the last day lockdown stringency exceeded 50, or the final day of the study period – 30 June 2020 – whichever comes first.

You will also find the mean average lockdown stringency (see the downloadable methodology on this page for an in-depth explanation of the OxCGRT stringency index), the total number of speed tests across the six-month study period, the average download speed during lockdown, and the download speed outside of lockdown.

Why some countries are missing data

There are 114 countries that qualified for inclusion in this study, which leave some 100+ countries and territories that did not qualify. Non-qualifiers fall into three separate categories, which are again represented with the hover-over explanations in the map. The three types of non-qualifier are as follows:

'Included in OxCGRT tracker, but stringency never exceeded 50'. These six countries did institute lockdown measures but these measures were not life-limiting enough to cause behaviours, such as staying at home, which might correlate with changes in network speed. Since there are only six countries in this category, we can list them here. They are: Belarus, Burundi, Japan, Nicaragua, Sweden and Taiwan.

Then there are countries 'Included in OxCGRT tracker, but less than 2,000 total IPs tested in M-Lab data (small sample size)'. These are countries that were tracked in the OxCGRT study and did indeed institute lockdown measures where stringency exceeded 50. However, with fewer that 2,000 speed tests taken across the six-month fieldwork period, there were not enough tests to produce reliable results. There are 53 countries in this category and they are listed under the 'Excluded countries' tab in the downloadable spreadsheet.

Finally, there are countries 'Not included in OxCGRT lockdown tracker (no lockdown in most cases)'. And these are, as suggested, countries not tracked by the OxCGRT lockdown study. In most cases this means there were no lockdowns in these countries during the fieldwork period. However, there may be other reasons for their exclusion. If you wish to find out why a specific country was excluded from the Oxford tracker, we recommend contacting them.

Our other projects

Previously, Cable.co.uk has produced internationally cited reports on global broadband speeds more generally, global mobile data pricing, and global fixed-line broadband package pricing.

Broadband comparison services

Cable.co.uk helps you quickly and easily compare broadband packages from the UK's leading broadband providers.

Highlights

Here is a quick look at the countries that recorded the highest percentage speed loss during their lockdown periods

#1. China
Fixed broadband penetration in China has expanded rapidly across the last half-decade. Fibre broadband grows the fastest, as DSL and cable modem access declines.
#2. Panama
Internet penetration is growing rapidly in Panama as consumers respond to government fixed-line projects in part, but mainly due to a high uptake of mobile broadband connectivity.
#3. Madagascar
National fibre backbone in Madagascar continues to roll out apace, with incumbent provider Telma connecting major cities and expanding the nation's fibre backbone.
#4. Chile
Fixed broadband penetration is relatively high for the country, with fibre broadband services among the fastest and least expensive in Latin America.

Drops in broadband speed during lockdown periods go against typical speed trend data

Our study uncovers that, overwhelmingly, speeds dropped in most countries included in the study during their most stringent lockdown periods. However, to recognise the true significance of these findings, one must acknowledge the typical way broadband speeds change globally over time.

The average global broadband speed measured by Cable.co.uk and M-Lab during the period from 11 May 2016 to 10 May 2017 was 7.40Mbps. The average global broadband speed measured during the period from 30 May 2017 to 29 May 2018 was 9.10Mbps – a rise of 23.35%. The average speed measured in the period from 9 May 2018 to 8 May 2019 was 11.03Mbps – a further rise of 20.65% over the previous year. Each year, typically, we see a rise in global broadband speeds in excess of 20%.

To see so many countries heading in the opposite direction of global broadband speed inflation is, to put it mildly, rather startling. Though we steadfastly cannot attribute direct causation to the measured effects thanks to myriad confounding variables, the strong correlation between stringent lockdown periods and the reversing of inertia in global speed trends points to some rather obvious conclusions.

The effects uncovered in the study could have been even more extreme

In terms of network capacity, the most dominant culprits when it comes to bandwidth consumption are video streaming services. Altogether, downstream video accounted of 60.6% of total network volume during 2019. For this reason, in March of this year, almost all market-leading streaming services (including Netflix, Amazon, YouTube and more) reduced the quality of their video offering to offset the potential network impacts caused by populations spending so much more time at home during lockdowns.

The correlations uncovered by Cable.co.uk with the help of its data partners, then, without service impact intervention could feasibly have been far more pronounced, provided the observed effects were causal in nature – something that cannot be committed to due to confounding variables (this being one).

Whether or not lockdowns would have brought national networks to a grinding halt were it not for such interventions is a matter for conjecture. But also an interesting factor to consider nonetheless.

The picture is highly regionalised

Though we typically include a regional column with our annual speed reports, we noticed early on in the project that there were pronounced regional correlations in speed drops (and rises) to consider. Countries close to one another tended to perform similarly poorly or similarly well.

For this reason, on this occasion, we have taken a more regional approach to examining the results, carving the regionally grouped data into a second tab in our downloadable spreadsheet.

Regions

Here we take a closer look at how internet speeds changed overall in each of 13 global regions, combining the individual country changes during their lockdown periods to create the average.

Asia (excl. Near East)
In Asia (excluding Near East), internet speeds dropped by -16.33% on average for the region. 12 of the 19 qualifying countries experienced slower internet speeds during their lockdown periods, with China (-50.97%) showing the largest drop both in the region and the world.
South Korea (-30.45%), Malaysia (-29.51%), Sri Lanka (-23.29%), Bangladesh (-21.76%), India (-21.05%), and Nepal (-20.77%) also experienced significantly slower speeds during their lockdowns. Afghanistan (+23.95%), and Vietnam (+12.84%), bucked the trend in this region, showing statistically significant increases in speed during their lockdowns.
Baltics
The Baltics, comprising three qualifying countries, saw the closest to no change of any region in this study, with a positive change of just +0.41%. Lithuania (-8.87%), and Estonia (-5.97%) both saw small drops in speed during their lockdowns, while Latvia (+17.91%) saw instead a fairly pronounced rise.
Caribbean
The Caribbean appears to have fared best overall as a region, though only four countries qualified for measurement. Dominical Republic (-26.52%) was the only country in the region to experience a drop in speed during its lockdown. Puerto Rico (+0.75%), Jamaica (+7.68%), and Trinidad and Tobago (+19.16%) all saw rises in speed during their lockdown periods.

Central America
The region where speeds dropped by the greatest percentage overall was Central America, with an average drop of -26.03% during its combined lockdown periods. Only one of the six countries qualifying in this region experienced a rise (Costa Rica at +0.82%). Meanwhile, Panama (-48.99%), Guatemala (-14.30%), Honduras (-3.69%), Mexico (-2.35%), and El Salvador (-0.01%) all experienced drops in speed of varying severity.
CIS (former USSR)
CIS (Former USSR) countries experienced an average drop in measured speeds during lockdown periods of -6.12%, a number very close to the global average. Four out of six qualifying countries in the region were measured to have experienced drops in internet speeds. Kyrgyzstan (-36.48%) and Russia (-12.60%) were measured to have the most pronounced drop in internet speeds during their lockdowns. Meanwhile, Ukraine (+14.86%), and Uzbekistan (+7.55%) saw small rises.
Eastern Europe
Eastern Europe measured an average drop in measured internet speeds during lockdowns of just -1.17%. A majority of the countries in the region showed almost no change. However, Hungary saw a drop in speeds of -13.26%, while at the other end of the scale significant rises were measured during lockdown periods in Moldova (+25.54%), and Bosnia and Herzegovina (+45.33%).

Near East
On average, the Near East region, comprising 14 qualifying countries, saw almost no change during its combined lockdown periods, with internet speeds down just -1.49% on average. Six countries recorded drops in average speed, seven gains in speed both small and large, and one (Qatar) was one of only two countries in the study to show no change at all (0.00%). The countries which recorded the largest drop is speed were United Arab Emirates (-18.42%), Israel (-14.59%), Saudi Arabia (-12.62%), and Yemen (-11.69%).
Northern Africa
Northern Africa recorded the second-highest overall drop in internet speeds during lockdown periods, with all five qualifying countries in minus figures. Libya (-27.38%) recorded the largest drop, followed by Tunisia (-23.23%), Morocco (-18.72%), and Algeria (-13.76%). Egypt's internet speeds experienced the smallest drop at -5.96%.
Northern America
Northern America – a region with only two qualifying countries – averaged a drop in lockdown internet speeds of -2.91%. Speeds in Canada fell by -10.82% during its stringent lockdown period. Meanwhile the United States actually saw a small rise in speeds during its lockdown period (+3.32%).

Oceania
The two qualifying countries in Oceania combined exhibited an average drop in measured internet speeds -12.05%. However, this is due to New Zealand alone, which experienced a drop of -17.88%. Australia actually showed a small rise in measured speed of +5.38% during its lockdown period.
South America
South America showed an average drop in internet speeds of -16.11%, with seven out of the 11 qualifying countries measuring drops in speed during their lockdown periods. Chile (-36.77%), Peru (-36.76%), Paraguay (-29.91%) and Bolivia (-21.45%) experienced the most precipitous drops in measured speeds. Conversely, small increases in speed were measured in Guyana (+8.51%), Brazil (+3.55%), and Uruguay (+2.37%) during their lockdown periods.
Sub-Saharan Africa
In Sub-Saharan Africa, 13 of the 14 qualifying countries recorded drops in measured internet speeds during their lockdown periods, with an average decrease of -14.24%. Angola bucked the trend in the region, showing a surprising increase of +117.19% during its lockdown period. Meanwhile, Madagascar (-37.71%), Cote d'Ivoire (-30.77%), Ghana (-24.58%), and Nigeria (-20.84%) experienced the largest drops in measured speeds during their respective lockdowns.

Western Europe
Western Europe (-4.66% overall) shows an interesting picture, with 12 of its 15 qualifying countries exhibiting measured drops in internet speed during their lockdown periods. However, Ireland (-2.23%), United Kingdom (-1.70%), Portugal (-0.19%), and Switzerland (+2.90%) all exhibit changes in speed that should be considered somewhat negligible. The countries showing the largest drops in average speed during their lockdown periods were Finland (-24.81%), Netherlands (-13.01%), Austria (-10.47%), and Italy (-10.44%).

Our comments

We set out to identify changes in network speeds in countries around the world which had instituted lockdown policies that put restrictions on populations that may cause more time to be spent at home, and thereby more time spent using the internet. Thanks to the Oxford COVID Government Response Tracker (OxCGRT) and its highly useful stringency index, we were able to pinpoint the precise period in each country where lockdowns were at their most life-limiting, and match them up with over 364 million (M-Lab) speed tests.
The results are startling. Although an overall drop of just -6.31% across all countries doesn't sound like an awful lot, this figure moves very much against the tide. Our annual global broadband speed tracker has demonstrated global increases of around 20% year-on-year since 2017. For the majority of countries in this study to be moving in the opposite direction during their COVID-19 lockdowns, then, is all the more significant.
While it is impossible to attribute direct causality, our study shows a high correlation between lockdown periods around the world and dips in measured internet speeds, with some regions of the world measuring the most substantial drops, and others more or less unchanged. The picture is an interesting one indeed, and I hope our work will be of interest to anyone who seeks to further understand the influence of stringent lockdown measures on network capacity and capability.

Resources

Downloadable versions of the data set (.xls), the original press release and the research methodology (.pdf)

The data
The full data can be downloaded here, including tabbed splits for regions and excluded/non-qualifying countries.
DOWNLOAD
Press release
If you wish to see the original press release for this research, you can download it here as a PDF.
DOWNLOAD
Methodology
Our research methodology and notes on how to interpret speed data can be downloaded here as a PDF.
DOWNLOAD

Contact us

Dan Howdle
Consumer telecoms analyst
Email: [email protected]
Mark Ashton
Head of Research and Development
Email: [email protected]

How global broadband speeds changed during COVID-19 lockdown periods

Interactive map

Using the map

Why some countries are missing data

Our other projects

Broadband comparison services

Highlights

#1. China

#2. Panama

#3. Madagascar

#4. Chile

Drops in broadband speed during lockdown periods go against typical speed trend data

The effects uncovered in the study could have been even more extreme

The picture is highly regionalised

Regions

Asia (excl. Near East)

Baltics

Caribbean

Central America

CIS (former USSR)

Eastern Europe

Near East

Northern Africa

Northern America

Oceania

South America

Sub-Saharan Africa

Western Europe

Our comments

Resources

The data

Press release

Methodology

Contact us

Dan Howdle

Mark Ashton

Entering your postcode