COVID-19: Measurement issues and assessments

This virus is shrewd in its camouflage and unabashed in its crueltyAysha Taryam

Timeline of a pandemic

On 31 May 2020, the WHO reported that more than 5.9 million people had been confirmed infected with COVID-19. That same day, 367 166 deaths globally were attributed to the virus.1

Five months earlier, on 31 December 2019, the WHO country office in China was notified that a new strain of pneumonia of unknown cause had been detected in the Hubei Province. On 7 January 2020, the Chinese authorities identified this pneumonia as a new strain of coronavirus. By mid-January, ministries of health in both Thailand and Japan confirmed imported cases of the novel coronavirus. The Republic of Korea reported their first case on 20 January. The following day, the WHO began issuing daily situation reports2 and confirmed 282 cases across the four affected countries, with six deaths in China.

Thereafter, events unfolded quickly (see figure 1) and, by the end of January, the day after the WHO designated “2019-nCoV acute respiratory disease” as the interim name of the disease, their Emergency 2019-nCoV Committee declared a PHEIC under the 2005 International Health Regulations (WHO, 2005). That day, the WHO reported 9 826 confirmed cases across 20 countries and 213 deaths (all in China).3 The first confirmed cases in Italy were also reported that day.

Figure 1. The first five months: Some key events, until end-May 2020
loading
  • December 31

    First report to WHO of new strain of pneumonia of unknown cause in Hubei Province, China

  • January 7

    New strain of coronavirus isolated

  • January 12

    China shares genetic sequence of the novel coronavirus

  • January 13

    Confirmed case of novel coronavirus in Thailand

  • January 15

    Confirmed case of 2019-nCoV in Japan

  • January 20

    Confirmed case of 2019-nCoV in Republic of Korea/WHO start daily Situation Reports/282 cases and 6 deaths

  • January 21

    1st confirmed case in South-East Asia region (Thailand)

  • January 23

    1st confirmed case in region of the Americas (USA)

  • January 25

    1st confirmed case in Europe region (France)

  • January 31

    WHO Declares PHEIC / Italy reports 1st cases of 2019-nCoV

  • February 1

    UK reports 1st cases of 2019-nCoV

  • February 2

    1st 2019-nCoV related death outside China (Philippines)

  • February 6

    1st cases reported on a cruise ship (Japanese waters)

  • February 11

    WHO names the disease

  • February 26

    1st case in Africa region (Algeria)

  • February 28

    WHO increases assessment of Global risk of spread and impact from HIGH to VERY HIGH / 83 652 confirmed cases/2 791 deaths/54 countries

  • March 6

    Number of cases globally passes 100 000 (101 254)

  • March 7

    Number of countries reporting confirmed cases passes 100

  • March 11

    WHO declares COVID-19 a Pandemic / 118 319 cases / 4 292 deaths / 114 countries / Italy passes 10 000 cases (10 149)

  • March 20

    Global death toll passess 10 000 (11 186)

  • March 26

    Number of countries reporting confirmed cases passes 200 (201)

  • March 29

    Italy passes 10 000 deaths (10 023)

  • April 3

    Spain passes 10 000 deaths (10 003)

  • April 4

    Number of cases globally passes 1 000 000 (1 051 697)

  • April 8

    France and United States of America pass 10 000 deaths (France 10 313) / (United States of America 10 845)

  • April 12

    Global death toll passes 100 000 (105 952)

  • April 13

    United Kingdom passes 10 000 deaths (10 612)

  • April 17

    Number of cases globally passes 2 000 000 (2 074 529)

  • May 6

    UK passes out Italy to become country with most deaths in Europe region (29 427)

  • May 11

    Brazil passes 10 000 deaths (10 627)

  • May 23

    Number of cases globally exceed 5 000 000

Source: UNCTAD derived from WHO (2020a).

On 26 February, the first cases of COVID-194 were reported on the African continent (all in Algeria), at which point COVID-19 was present in 45 countries or territories across all six WHO regions5 (see figure 2). Two days later, the WHO (2020a) increased their assessment of the global risk of spread and impact from high to very high. At this point, there were 83 652 confirmed cases spread across 54 countries.6

Figure 2. Number of countries, territories or areas reporting confirmed cases of COVID-19
Source: UNCTAD calculations based on WHO (2020a).

On 6 March, the number of global confirmed cases attributed to COVID-19 passed the 100 000 mark (see figure 3). The following day, the number of countries reporting confirmed cases exceeded 100. Four days later, the WHO declared COVID-19 a pandemic. In doing so, the Director General of WHO expressed concern at both the alarming levels of spread and severity, and the alarming levels of inaction. He explained that the WHO had assessed that COVID-19 could now be characterized as a pandemic, clarifying that this did not change the threat level (WHO, 2020b).

Figure 3. Number of global confirmed cases per day
Source: UNCTAD calculations based on WHO (2020c).

By the end of May 2020, the aggregate cumulative number of confirmed cases and deaths reported by countries to the WHO was 5.9 million and 367 thousand, respectively. As of 31 May 2020, Europe and the United States of America combined accounted for 65 per cent of all confirmed cases and 77 per cent of all COVID-19 deaths, as shown in figure 4 (readers should be aware that there are particular measurement problems with COVID-19 statistics as currently reported by all sources (see section Measurement issues below)).

Figure 4. Ten most affected countries, cumulative deaths, 31 May 2020
Source: UNCTAD calculations based on WHO (2020c) and United Nations (2019).

In the first three or four months of the pandemic, the global cumulative total deaths was led by European countries (notably Belgium, France, Germany, Italy, Spain and the United Kingdom) as well as by the United States of America. But since then, as shown in figure 5, it has been the Americas that have accounted for most of the growth (most notably Brazil and Mexico, in addition to the United States of America).

Figure 5. Global COVID-19 deaths by WHO region per day, until 31 May 2020
Source: UNCTAD calculations based on WHO (2020c).

At a country level, the spread and prevalence of COVID-19, as well as the measures taken to contain its spread, have varied considerably. For a variety of reasons, a number of countries showed much higher prevalence rates than others. The trajectory of the number of confirmed cases in a selection of hardest hit countries is compared in figure 6.

Figure 6. The first 100 days: Number of confirmed cases per million people, selected countries
Source: UNCTAD calculations based on WHO (2020c) and United Nations (2019).

A curiosity of COVID-19, in the early months at least, is that it has hit developed countries much harder than developing countries, in terms of prevalence, with the notable exceptions of the Islamic Republic of Iran and, more recently, Brazil and Mexico. In figure 6, the time axis is normalized to the start date (i.e. the date when a confirmed case was first reported to the WHO by a country) so that the trajectory of COVID-19 spread in the first 100 days can be compared.

Comparing the worst hit countries in Europe with badly hit countries elsewhere in the world, the patterns are immediately different in both timing and scale. Although Italy rose to prominence in the media, Spain and Belgium have been the worst affected countries to date on a per capita basis. The spread accelerated early and rapidly in Italy, Spain and Belgium, peaking in Italy and Spain around day 62 (i.e. approximately two months after the first confirmed cases were reported). The number of new cases peaked about two weeks later in Belgium (day 74).

In France and Germany, acceleration started about 10 days later than in Italy and Spain. Both countries experienced similar trajectories and prevalence to each other, with the spread of COVID-19 peaking around days 69 and 70. Initially, the United Kingdom had an almost identical trajectory to France, albeit lagged by a few days; however, new cases peaked on day 70 in France, whereas the spread continued accelerating in the United Kingdom and did not reach its maximum for another week. Furthermore, unlike France and Germany, the peak was not followed by a rapid decline. Rather, the number of new cases continued at a slightly reduced rate, until around day 91, when the number of cases began rising again, and then finally began to decline around day 100.

Some of the worst hit countries elsewhere in the world have had a markedly different experience to that in Europe. Acceleration was much more varied, beginning around day 13 in the Islamic Republic of Iran, day 40 in Brazil but not until around day 60 in the United States of America. To date, the per capita number of cases per day in the United States of America and Brazil are a little over half of what was experienced in Spain at its peak.

The Islamic Republic of Iran was hit by COVID-19 relatively early and rapidly. When most countries were just beginning to experience an acceleration in spread (around day 45), the spread in the Islamic Republic of Iran was already at its maximum. Unlike most European countries, however, this country did not experience a sharp downturn, but rather a gradual deceleration which troughed around day 76 and then began slowly rising again.

The United States of America, once acceleration began, experienced quite a steep trajectory similar to that observed in Italy. By day 83, cases in the United States of America peaked at 95 cases per million people (higher than Italy’s peak of 91 per million people) and then, similar to the United Kingdom, the number of new cases did not reduce significantly, but fell back to a slightly reduced rate of spread. Again, like the United Kingdom, the trajectory began increasing again around day 97.

The number of cases in both the Russian Federation and Brazil have increased steadily along a similar curve, albeit with acceleration in the Russian Federation lagging by about 20 days. By day 100, the number of cases per capita in Brazil, at 110 per million, had far surpassed the peak in the United States of America. Cases in Mexico have been rising slowly but inexorably.

An important aspect to note about the outbreaks within each of these countries is their highly heterogenous and, at least initially, concentrated nature, both in terms of geography and demography. Most countries initially experienced severe outbreaks in one or several geographic areas, for example Lombardy in Italy or New York in the United States of America, rather than a uniform development across the country. Specific communities or groups have also been affected differently by the virus, with many countries experiencing outbreaks in care-homes, meat-processing plants, or low-income communities. This has also led to second-wave developments in some countries as at-risk communities experience outbreaks amidst an otherwise “under control” situation, such as it has been the case with meat-processing plant workers in Germany or migrant workers in Singapore, or as previously spared geographic areas succumb to the pandemic, as with the southern United States of America.

Measurement issues

One of the challenges of analysing COVID-19 statistics is that their quality is unproven and considerable methodological differences exist across countries. They likely suffer from problems considering that organizing a new global data collection during a pandemic, at both national and international level, on a disease about which relatively little is known, is not going to be without teething problems. There is also always the risk that some countries may inaccurately or not report COVID-19 related statistics at all (BBC, 2020). The WHO notes "Differences are to be expected between information products published by WHO, national public health authorities, and other sources using different inclusion criteria and different data cut-off times. While steps are taken to ensure accuracy and reliability, all data are subject to continuous verification and change. Case detection, definitions, testing strategies, reporting practice, and lag times differ between countries/territories/areas. These factors, amongst others, influence the counts presented, with variable underestimation of true case and death counts, and variable delays to reflecting these data at global level."7 Furthermore, when making international comparisons, one should also be cognisant that a range of factors not directly related with the state of a country’s health system likely impact infection rates. These include: the age structure of populations, the density or urbanization of populations, prevalence rate of chronic diseases and perhaps also ethnicity.

The two principal variables, ‘confirmed cases’ and ‘deaths’, are to some extent problematic, and this impacts on the veracity and general quality of derived variables, such as mortality rates. The number of ‘confirmed cases’ is based on the number of laboratory-confirmed cases, which rely on the quantity and consistency of testing in countries. This varies enormously, as experience has shown (see figure 7). Some countries undertake large-scale population testing, whereas others have adopted less comprehensive approaches. As countries have learned more about COVID-19, some have changed their testing methods and schemes, causing methodological breaks and discontinuities in time series. For example, several countries have changed their reporting ‘day’, i.e. the 24-hour period that comprise the reference period, as did the WHO themselves on 18 March.8 Furthermore, uncertainty regarding the numbers of asymptomatic and undiagnosed cases, and of course misdiagnosed cases, means that the actual number of cases may be quite different to the number of officially confirmed cases. An early study from China suggests that almost 80 per cent of cases of infection were classified as mild or asymptomatic (Day, 2020). In a pre-print study published in April 2020, Lu et al. (2020) estimate the proportion of asymptomatic cases to be lower, ranging from 18 to 50 per cent. Therefore, it is important when using the ‘confirmed cases’ metric to understand that this statistic is the number of cases reported by each country, and that the reference date may not always accurately reflect the date of the event.

Figure 7 illustrates the variation in the numbers of tests undertaken by countries, which as noted above, will immediately impact the number of confirmed cases reported. From a surveillance and control perspective, it should also be noted there is an important distinction between the number of tests performed and the number of tests analysed and reported. The time delay between the two is also of critical importance.

Figure 7. Total COVID-19 tests per thousand people
Source: Our World in Data (2020).
Notes: Counts refer to 26 May 2020 or nearest available. For some countries, the statistics are not updated regularly: India (24 April); Brazil (20 April); France (5 May); China, Hong Kong SAR (19 May); Spain (21 May). No data were available for China.

Almost certainly, both ‘confirmed cases’ and ‘deaths’ are undercounted, probably to different degrees, which no doubt will, with time, explain some of the apparently high mortality rates. Given the problems with the reported statistics, the actual prevalence of COVID-19 in populations remains for the time being unknown. In a February 2020 interview, Neil Ferguson, Professor of epidemiology at Imperial College London, estimated that China had only detected around 10 per cent or less of its coronavirus cases (Ferguson, 2020). In France, a recent study by Salje et al. (2020) estimated that on 11 May9 about 2.8 million10 people (or 4.4 per cent of the population) had been infected by COVID-19 – some 20 times more than the official estimate of 137 073 reported to the WHO for that day.11 A serological antibody test conducted in the canton of Geneva in Switzerland (Hôpitaux Universitaires de Genève, 2020) found seroprevalence in the population to be 5.5 per cent (or 27 000 people) on 17 April 2020, or some five times higher than official estimates. Although using different approaches, these studies yield similar results, which suggest that the proportion of the target populations infected in April/May was between four and six per cent.

From a policy perspective, these studies suggest that countries are a long way from developing herd immunity, which in turn implies that population immunity is probably insufficient to avoid a second wave. On 25 April, the WHO (2020d) warned there was no evidence that people with COVID-19 are immunised. They noted, on 12 May, that the concept of herd immunity is generally used for calculating how many people will need to be vaccinated in a population to protect others, not for calculating the occurrence of immunity through infections (Independent, 2020). Many studies support the conclusion that a relatively small proportion of the population has been infected to date. A study using three different approaches (Lu et al., 2020) estimated that as much as 10 per cent of the population in the United States of America may have been infected by mid-April 2020. On 20 April, the WHO noted that early studies suggest that only two to three per cent of the global population had been inflected (WHO, 2020b). The ONS in the United Kingdom reported on 5 June that, as of 24 May, 6.8 per cent of people who provide blood samples tested positive for antibodies to COVID-19 (ONS, 2020).

At first glance, ‘deaths’ statistics appear to be less problematic, but on closer examination, a number of problems are also evident. In several countries it has emerged that deaths (initially at least) only included deaths in hospitals, and that deaths in other institutional or private households had not been included. There have been several revisions to official reports, as causes of death have been re-evaluated as more is learned about the disease. This too has led to reporting lags, and problems matching events to dates properly. Furthermore, analyses of ‘excess deaths’, i.e. the deviation in mortality from the expected level, suggests that deaths attributed to COVID-19 are being undercounted.

EuroMOMO (2020) monitors mortality for several countries in Europe.12 Their data suggest that between weeks 12 and 20, i.e. between the weeks beginning 16 March and finishing 26 April, there were 142 577 excess deaths in Europe (see figure 8). The excess mortality in 2020 is notable, both in scale and in seasonal pattern. The weeks in which excess mortality was unusually high during the first quarter of 2020 were quite distinct from the typical seasonal flu patterns associated with the winter months. For the same countries and during the same period, the WHO reported that deaths attributed to COVID-19 rose from 13 786 on 16 March to 116 029 on 26 April, an increase of 102 243. This number is 40 000 lower than the number of excess deaths reported by EuroMOMO (2020). While most of these excess deaths can, in all probability, be attributed to COVID-19, caution must again be exercised: it is likely that other medical treatments were postponed or cancelled, as people avoided doctors and hospitals. This in and of itself may have had led to a spike in excess mortality.

Figure 8. Excess deaths by week (all ages), weeks 1 to 20 of 2020 (1 Jan to 9 May)
Source: UNCTAD calculations based on EuroMOMO (2020).
Notes: These figures include the 24 countries or territories participating in the EuroMOMO network.

A false sense of security?

In October 2019, a new GHS Index was launched jointly by Johns Hopkins University and the Nuclear Threat Initiative, with the purpose of conducting a first comprehensive assessment and benchmarking of health security and related capabilities across the 195 countries that are signatories to the WHO International Health Regulations. The index was constructed by the Economist Intelligence Unit, in consultation with Nuclear Threat Initiative and the Johns Hopkins University Center for Health Security and advised by an international panel of experts (Johns Hopkins University et al., 2019).

The GHS Index assesses not only countries’ health security capacities, but also the existence of functional, tested and proven capabilities for stopping outbreaks at the source. It also tests whether that capacity is regularly tested and shown to be functional in exercises or real-world events. It was not designed to warn specifically against COVID-19, but to assess the readiness of countries to deal with a biological event or pandemic, such as COVID-19, in general.

In their 2019 inaugural report, the authors issued some stark warnings, reporting that countries were not prepared for a globally catastrophic biological event, nor were they fully prepared for epidemics or pandemics. Collectively, they note, international preparedness was weak. Many countries did not show evidence of the health security capacities and capabilities that needed to prevent, detect, and respond to significant infectious disease outbreaks. Prophetically, they warned: "knowing the risks, however, is not enough. Political will is needed to protect people from the consequences of epidemics, to take action to save lives, and to build a safer and more secure world". They also noted that "unfortunately, political will for accelerating health security is caught in a perpetual cycle of panic and neglect”.

The GHS Index is described as a multidimensional analytical framework, commonly known as a benchmarking index. It is essentially a composite, comprising six categories: (1) prevention; (2) detection and reporting; (3) rapid response; (4) health systems; (5) compliance with international norms; and (6) risk environment. Those categories are populated with 34 indicators and 85 sub-indicators. The overall index for each country is the weighted sum of the category scores, where the weights are agreed by an expert panel. In constructing the index, three other weighting types were tested: neutral weights; equal weights; and weights derived from a principal component analysis.

One would hope that the GHS index never need be tested in a live situation. But it has been, and like many metrics, it has been confounded by COVID-19. In retrospect, the GHS (as is often the case with composite indices) may have hidden as much as it has revealed. It highlights again the question of whether country rankings have any real utility or simply distract readers from important underlying messages. Although the report issued many stark warnings, the indices themselves may have conveyed a different message; at least for countries ranked near the top, with scores in excess of 70, the indices may have given a false sense of security. Developments in the first half of 2020 have made some of the GHS country rankings appear incongruous. It is too early to conduct any definitive analyses of COVID-19, thus any assessment is necessarily premature. Perhaps in the longer term, the index rankings may correlate better with events. That said, the first six months have generated some noteworthy comparisons.

The index ranked the United States of America as the best prepared country in the world, followed by the United Kingdom. Additionally, included in the top 20 best prepared countries were Belgium, France, Netherlands, and Spain. It is striking that these are some of the hardest hit countries by the COVID-19 pandemic in both absolute and per capita terms. It also ranked Brazil and Mexico in the top 30 and placed New Zealand only 35th.

Table 1. GHS top 20 best prepared countries compared with 20 worst affected countries by COVID-19 (as of 31 May 2020)
(Ranks)
CountriesGHS overallWorst affected countries
Best prepared rankingConfirmed COVID-19 casesConfirmed COVID-19 cases per millionCOVID-19 deathsCOVID-19 deaths per million
United States of America11818
United Kingdom251732
Netherlands32630157
Australia46911478133
Canada517321211
Thailand69117994158
Sweden72511165
Denmark858434223
Republic of Korea95613058120
Finland1070575230
France11153456
Slovenia12112767436
Switzerland1336272614
Germany1411371122
Spain1561263
Norway1666516142
Latvia171218511086
Malaysia186712170134
Belgium19221391
Portugal2032262817

Four measures of ‘worst affected’ are presented: confirmed cases; confirmed cases per million of population; deaths; and deaths per million of population. The United States of America, the United Kingdom and Spain are in the top 20 hardest hit, no matter which measure is used. Sweden, France and Belgium are in three of the four measures. Canada and Germany feature in two (see table 1).

It is of course easy to be wise with hindsight. But the particular importance of indicators for Political and Security Risk and Public Health Vulnerabilities are striking. In their commentary, the authors portentously noted the importance of ‘political will’ – this seems to have been the critical factor in how well countries have dealt with COVID-19 to date. Unfortunately, it is extremely difficult to measure political will. Furthermore, in light of developments, this dimension might warrant a higher weight in the overall index. Perhaps the overall risk environment needs to be supplemented as well, as the index doesn’t seem to adequately address potential transmission vectors. For example, some connectivity and globalization indices would arguably strengthen the robustness of the index.

The importance of public health systems is also now clear. COVID-19 has graphically illustrated the importance of government and public infrastructure and services more generally and the critical role they play during a time of crisis. Thus, a wider reflection of public services generally, including the strength and investment in national statistical systems, but in particular investment in public health systems, might also improve the index.

Another composite index, compiled using an AI approach, has been published by the Deep Knowledge Group (2020). This index is more bespoke, targeting COVID-19 directly. The latest version, from June 2020, also provides country rankings, which in light of events appear more credible. However, it should be stressed that this index is updated contemporaneously and is specific to COVID-19, so this should not be surprising. Unlike the GHS, the purpose of which was to conduct an ex-ante assessment of countries preparedness for a biological event, such as, a pandemic, the purpose of the index from the Deep Learning Group is to inform government decisions during the current pandemic, helping them to optimize current and post-pandemic safety and stability, in order to maintain the health and economic well-being of their populations and alleviate the collateral damage caused by COVID-19.

Policies implemented during the COVID-19 pandemic

Person-to-person contagion of COVID-19 depends on the characteristics of the virus itself, including how easily it can infect a new host and how long it can survive outside the human body. But it also depends on the number of potential opportunities of transmission provided by social interaction between people. Since contagion can be rapid, and carriers may unwittingly spread the virus, as COVID-19 appears to have a long lag before symptoms manifest themselves, it has turned out to be essential to contain the spread of the disease at an early stage, before it affects larger shares of the population and the number of patients exceeds the capacity of health systems.

Although facing many unknowns about the virus and its transmission mechanisms, governments around the world started implementing containment measures aimed at reducing the probability that an infected person transmits the virus. These measures included, but were not limited to: school closures; limiting non-essential business activity and promotion of remote work; restrictions on public or private gatherings and cancellation of public events; stay-at-home requirements; restrictions on domestic or international travel; obligatory or recommended use of masks, gloves and other physical barriers; and information campaigns. These measures were applied broadly to the entire population or targeted to specific population groups (for example, in highly affected geographical areas or for most at-risk groups).

The curve in figure 9 measures the application of physical distancing measures worldwide since the outbreak of the disease. It is constructed as a population-weighted average of country-level scores on the Oxford COVID-19 Government Response Tracker’s Stringency Index.13 There was a first wave of policies in late January and early February, primarily concentrated on China and other countries in East and South-East Asia that responded to the first cases of the disease. The implementation of such measures was more widely adopted around mid-March, after the number of affected countries passed 100 and the disease was declared a pandemic by the WHO (see section Timeline of a pandemic). Since early May, we see a gradual decrease in the index, as some of the containment measures are rolled back in areas where the disease is considered to be under control.

Figure 9. Stringency of global confinement measures
Source: UNCTAD calculations based on University of Oxford, Blavatnik School of Government (2020) and UNCTAD (2020a).
Notes: This index ranges from 0 to 100, with higher numbers indicating more stringent confinement measures. The global average is calculated as the population-weighted average of country level indices.

The global trend observed in the first months of 2020 hides significant different patterns at the country level. As shown in figure 10, some countries swiftly implemented distancing measures and successfully contained the spread of the disease. In all these countries, there were already strict measures in place by the time there were 100 confirmed cases, with a resulting slowdown in the contagion rate. In some cases, such as El Salvador, New Zealand or the Philippines, some measures were active even before the first case was detected. Other countries delayed the onset of these policies (see figure 11) until the number of cases was already high and rapidly increasing, with a resulting surge in the spread of the disease. It is worth noting the case of Singapore, one of the first countries to put in place containment measures against COVID-19. This resulted in slower infection rates already in February; however, the country was affected by a second wave beginning in mid-March forcing it to scale up their policy response.

Figure 10. Stringency of confinement measures and cumulative COVID-19 cases, selected countries
Source: UNCTAD calculations based on University of Oxford, Blavatnik School of Government (2020) and WHO (2020c).
Note: This index ranges from 0 to 100, with higher numbers indicating more stringent confinement measures. Cumulative number of cases are in logarithmic scale. The red dotted lines indicate the date when the number of confirmed cases reached 100.
Figure 11. Stringency of confinement measures and cumulative COVID-19 cases, selected countries
Source: UNCTAD calculations based on University of Oxford, Blavatnik School of Government (2020) and WHO (2020c).
Note: This index ranges from 0 to 100, with higher numbers indicating more stringent confinement measures. Cumulative number of cases are in logarithmic scale. The red dotted lines indicate the date when the number of confirmed cases reached 100.

In some cases, neighbouring countries chose different policies to contain the spread of the virus. Figure 12 shows the situation in four Nordic countries. While Denmark, Finland and Norway took strict measures (the three of them scored above 60 on the Stringency Index by mid-March), Sweden adopted a more relaxed containment policy. As of 10 June 2020, Sweden had 4 547 confirmed cases of COVID-19 per million people, compared to 2 072 in Denmark, 1 268 in Finland and 1 580 in Norway. In terms of confirmed deaths, Sweden has registered 467 deaths per million people, in comparison with 102, 58 and 44 in Denmark, Finland and Norway, respectively.

Although the pandemic remains active and it is too soon to conduct a full evaluation of the impact of containment measures, early evidence seems to indicate that they were effective in slowing down the infection rate of COVID-19 and reducing the number of deaths. The timing of the measures has also proven crucial, with those implemented faster resulting in stronger effects (Deb et al., 2020).

Figure 12. Stringency of confinement measures and cumulative COVID-19 cases, selected countries
Source: UNCTAD calculations based on University of Oxford, Blavatnik School of Government (2020) and WHO (2020c).
Note: This index ranges from 0 to 100, with higher numbers indicating more stringent confinement measures. Cumulative number of cases are in logarithmic scale. The red dotted lines indicate the date when the number of confirmed cases reached 100.

It quickly became evident that, while the containment measures could be effective in slowing down the rate of infection, they also had serious economic and social consequences. With international trade collapsing, domestic economic activity at a standstill and unemployment soaring, the pandemic could also bring long-lasting harm to the economy. And the detrimental economic effects are not distributed evenly. Because of their lower diversification, more limited capacity to hedge risks and less resources in general, smaller firms were particularly affected. Also, poorer families, households in rural areas, workers in the informal sector and certain population groups were more impacted than others. The health crisis could, therefore, exacerbate existing sources of inequality. Governments proposed and started implementing policy packages covering fiscal, monetary and macro-prudential measures, along with employment preservation, income support and social protection policies.

The Oxford COVID-19 Government Response Tracker’s Economic Support Index provides a quantitative indicator of such measures. Because it only covers policies related to income support and debt/contract relief for households (and does not include fiscal stimulus for firms, for instance), it only provides a partial picture of the full spectrum of economic measures taken as a response to the pandemic. However, it can still give an indication of how reactive the governments were when faced with the supply and demand shocks brought by the pandemic.14 A GDP-weighted global average of this index is presented in figure 13.

Figure 13. Global economic support measures
Source: UNCTAD calculations based on University of Oxford, Blavatnik School of Government (2020) and UNCTAD (2020a).
Notes: This index ranges from 0 to 100, with higher numbers indicating more economic support measures. The global average is calculated as the GDP-weighted average of country level indices.

Figures 14 and 15 show the implementation of economic support against the evolution of the PMI in the manufacturing sector, a timely indicator of economic activity in this sector. The first graph covers developed economies, while the second includes developing and transition economies. We see a strong response since mid-March or early April in many countries, as soon as economic indicators signalled a slowdown. But other countries have implemented more muted economic stimulus. The capacity of countries to implement stimulus policies depends on factors such as the available fiscal space and the degree of development of the financial sector. Because of this, the crisis could also deepen pre-existing inter-country inequalities, affecting poorer or less financially-integrated economies to a larger degree.

Figure 14. Economic support measures and manufacturing activity, selected countries
Source: UNCTAD calculations based on University of Oxford, Blavatnik School of Government (2020) and Refinitiv (2020).
Notes: The Economic Support index ranges from 0 to 100, with higher numbers indicating more economic support measures. PMIs are diffusion indices, with values above (below) 50 indicating an expansion (contraction).
Figure 15. Economic support measures and manufacturing activity, selected countries
Source: UNCTAD calculations based on University of Oxford, Blavatnik School of Government (2020) and Refinitiv (2020).
Notes: The Economic Support index ranges from 0 to 100, with higher numbers indicating more economic support measures. PMIs are diffusion indices, with values above (below) 50 indicating an expansion (contraction).

COVID-19 and the SDGs

COVID-19 has had a dramatic impact on the global economy, environment and society. This section presents a small flavour of developments since the outbreak. One indicator has been selected to represent each of the three key pillars. For economy, developments in international trade are examined, which relate directly to SDG targets 17.11 and 17.13; for social, the likely impact of COVID-19 on extreme poverty, target 1.1, are highlighted; for environment and climate change we examine changes in greenhouse gas emissions, target 9.4.

Other analyses regarding COVID-19 and the SDGs are included elsewhere in this report, related to:

Economy: Severe decline in global trade

The 2020 trade collapse will be big, sudden, synchronised and broad – but it should not be unexpectedRichard Baldwin, Professor of International Economics, The Graduate Institute, Geneva

At the end of 2019, global merchandise trade volumes and values were showing modest signs of recovery. But in 2020, as the world adopted a range of measures to contain the COVID-19 pandemic, the global economy grounded to a halt, and international trade with it. In early May 2020, the monthly UNCTAD Trade Nowcast (UNCTAD, 2020b) estimated that the value of global merchandise trade would fall in the second quarter of 2020 by 27 per cent year-on-year (see figure 16). As economies start to reopen after containment, a rebound in June is anticipated. However, as no data are available yet to reflect this upturn, the nowcast is still extrapolating prior trends. Consequently, the June edition of the UNCTAD Trade Nowcast was suspended, as UNCTAD statisticians were concerned that their models were overshooting, as data picking up impacts of decontainment were not yet available.

Figure 16. Global merchandise trade
(quarter-on-quarter growth rate, seasonally adjusted)
Source: UNCTAD (2020b).
Notes: The shaded area indicates UNCTAD nowcasts as of 5 May 2020.

The UNCTAD nowcasts incorporate a wide variety of data sources to capture the diverse determinants and indicators of trade. To help users understand this, UNCTAD also publishes, alongside the headline nowcast, a time series, showing how the nowcast has evolved on a weekly basis, as the model incorporates new information (see figure 17). For value estimates, one can see a clear deterioration since late April as new data became available.

Figure 17. Evolution of global merchandise trade nowcast, 2nd quarter of 2020
(quarter-on-quarter growth rate, seasonally adjusted)

Social: Worsening impact on global poverty

In May, the World Bank (Gerszon-Mahler et al., 2020) estimated that COVID-19 could push between 40 and 60 million into extreme poverty (CCSA, 2020). Since then, the epicentre of the pandemic has shifted from Europe to the Americas and the Global South, increasing the death toll in low- and middle-income countries. As a result, they have updated their assessment of the impact of COVID-19 on global poverty.

Based on the updated growth forecasts presented in their Global Economic Prospects, the World Bank (2020) has updated their impact assessment on global poverty. They present two scenarios, a baseline scenario (global growth contracts by five per cent in 2020) where the outbreak remains at currently anticipated levels, with economic activity recovering later in the year. The more pessimistic downside scenario (global growth contracts by eight per cent in 2020) anticipates a more persistent outbreak, forcing prolonged containment measures, resulting in vulnerable firms closing, vulnerable households sharply reducing consumption, and several low- and middle-income countries experiencing heightened financial stress.

Figure 18. The impact of COVID-19 on global extreme poverty
(millions of persons)

Based on these deteriorating economic forecasts, the World Bank have updated their assessment of the impact of COVID-19 on poverty. They estimate that the baseline scenario will result in 71 million people being pushed into extreme poverty (measured by the international poverty line of US$1.90 per day), whereas the downside scenario would see this rise to 100 million people.

Environment: Reduction of CO2 emissions due to COVID-19 not enough to reach climate targets

In the first quarter of 2020, global CO2 emissions were more than five per cent lower compared with the same period in 2019 according to estimates by IEA (2020). Depending on the scenario used, 2020 global CO2 emissions are forecast to decline by around eight per cent; the equivalent of 2.6 Gt. This will be the largest reduction ever recorded and will bring us back to levels last seen a decade ago. The last significant decline, caused by the global financial crisis in 2009, only yielded a reduction of 0.4 Gt.

Figure 19. Greenhouse gas emissions and target reductions
(Gt of CO2e)
Source: UNCTAD calculations based on the Netherlands PBL (2019) and UNEP (2019).
Notes: For additional details, see Make or break for green economy.

Early in 2020, global demand for energy fell sharply owing to containment measures taken against the COVID-19 pandemic. Significant contributors to this slump in demand were the fall in demand for air and road travel (see Make or break for green economy). The fall in demand, combined with changes in the global energy mix in favour of renewables, in turn, contributed to notable short-term improvements in air quality, particularly falls in NO2 (Carbon Brief, 2020; NASA, 2020; European Data Portal, 2020; CCSA, 2020).

Although record-breaking, the forecast reduction of CO2 emissions caused by the COVID-19 outbreak will not be enough to achieve even the weakest of the targets set out by the Paris Climate agreement. Global emissions would need to be cut by almost eight per cent every year for the next ten years to keep us within reach of the Paris Climate agreement. Even if COVID-19 has induced fast reductions of CO2 emissions in 2020, it will not be enough to win the fight against climate change. More effective and lasting efforts are needed to reduce CO2 emissions and other greenhouse gases to limit global warming below 2°C or especially below the 1.5°C target by 2100. As populations and GDP per capita continue to grow, a drastic reduction in carbon intensity will be required. Rising energy efficiency serves as an important step in that direction, as well as renewable and cleaner energy.

Trade shocks and gender equality in employment

15

Business cycles are not gender neutral (e.g. Hoynes et al., 2012; Peiro et al., 2012; Razzu and Singleton, 2016), as a consequence of gender-segregation into different industries and occupations (Razzu and Singleton, 2018). Economic downturns usually affect men more than women since men tend to work in industries that are more closely tied to economic cycles (e.g. construction and manufacturing). However, the COVID-19 economic downturn may be different as sectors most exposed to the collapse absorb a sizeable share of female employment (ILO, 2020b). Therefore, women are likely to be more affected, at least in the short-term (Alon et al., 2020). As the economic consequences of COVID-19 unfold, the effects may spread. As outlined above, the latest UNCTAD (2020b) nowcast anticipates that the world trade will fall by 27 per cent during the second quarter of 2020. This will have differing effects on women and men in the labour markets which will be important to consider in the crisis response.

To analyse the link of gender and trade in these conditions, we estimate the response of women’s and men’s employment to changes in international trade. A set of gender balanced indicators in employment, as proposed by Van Steveren (2012), shows how gender equality has evolved during previous economic fluctuations. These indicators also provide early signs of changes in the labour market by gender in response to changing international trade. Data from the EU and the United Kingdom, comparable by EU regulation, make an interesting case study, noting the synchronizing effect of the common economic area on business conditions.

Figure 20 compares year-on-year changes in male and female unemployment rates with year-on-year changes in international trade for EU countries and the United Kingdom. The ratio of female to male unemployment seems to follow similar patterns to international trade, meaning that male employment increases more than female’s as trade increases. From this viewpoint, international trade benefits men more than women (Luomaranta et al., 2020).

Figure 20. Quarterly ratio of women’s to men’s unemployment and trade in goods, EU and United Kingdom
(year-on-year change)
Source: UNCTAD calculations based on the Eurostat (2020).
Notes: Both variables are standardized to have mean zero and standard deviation of one.

To inspect the relationship of trade and employment in selected groups in the labour markets, we estimate a set of panel-VAR regressions using Abrigo and Love (2015) as:

X_{i,t}=\pi_i+\Theta(L)X_{i,t}+\varepsilon_{i,t}

where X_{i,t} = \big[\ln Y_t \ln T_t \ln U_t \big]^{\prime} vector includes the labour force indicator of interest, international trade growth rate, and unemployment growth rate, all in logs. \Theta(L) is the matrix polynomial in the lag operator L . \pi_i captures the country fixed effects and \varepsilon_{i,t} is the error term. As in Clark and Summers (1980), unemployment rate is used to capture the state of the economy, distinguishing overall economic conditions from international trade.

Figure 21. Impulse responses of employment and unemployment rates in selected groups in the labour markets, EU and United Kingdom
(year-on-year change)
Source: UNCTAD calculations based on the Eurostat (2020).
Notes: We apply a set of homogeneous panel VAR models, useful for analysing cyclical responses of the variables, all of which are entered in a stationary form and are de-trended and de-seasonalized natural logarithms of year-on-year growth rates. The impulse-response plots provide a short-term response to one per cent increase in the impulse variable (international trade in goods) on the response variables (selected employment variables) in each country. The lag length is selected by the Akaike information criterion. Notice that the plots should only be used to consider the very short-term responses. The titles placed above each plot provide the impulse variable and the response variable, respectively.

The first two charts in figure 21 illustrate the differing responses of male and female employment rates to a one per cent increase in international trade in goods. Indeed, based on the estimated model, male employment rate reacts more strongly to an increase in trade than female rates: 0.24 per cent increase for men compared with only 0.13 per cent for women. Similarly, male employment will drop by 0.24 per cent for every one per cent decrease in international trade. This reinforces the observation that male employment is more pro-cyclical than female employment.

The remaining four charts review responses to a one per cent increase in unemployment to capture the labour market responses to worsening economic conditions in a number of gender balance indicators. The third chart compares women’s unemployment to men’s unemployment among youth, with a declining development referring to men’s unemployment rate increasing faster than women’s among young workers (20-24 years). In part-time employed, gender balance in employment shifts for the benefit of men, when economic conditions deteriorate. Similarly, women would gain relative to men when the economy picks up.

The opposite is true among employees with a lower education, as gender balance in employment shifts for the benefit of women when the economy deteriorates. Men are relatively more hit in low-skill jobs when the economy plummets. The gender balance in employment in the high-skill category is not strongly responsive to economic shocks.

Taken together, the results provide evidence that international trade has gendered impacts on employment and points out that young, part-time workers and those with a lower education are most vulnerable to shocks, such as those related to the COVID-19 pandemic. According to ILO (2020c), over one in six young people (aged 15 to 24) surveyed have stopped working since the onset of the COVID-19 pandemic, and for those remaining in employment, working hours have dropped by 23 per cent.

Gender balance in the labour markets can be significantly affected by international trade and economic fluctuations and should, therefore, be closely monitored. UNCTAD (2018) provides a conceptual framework for analysing the interconnections of gender equality and trade. Countries should collect and analyse gender statistics linked to trade to inform crisis response and recovery plans, since it looks like the most vulnerable are likely to suffer the strongest effects of the COVID-19 related economic downturn.16

Impact on global statistics

The shutting down of large parts of the economy was not anticipated in the construction of our economic statisticsProfessor Tara Sinclair, George Washington University

The global COVID-19 crisis has disrupted the compilation of official statistics across the global statistical system, throwing up a wide range of methodological, conceptual and data collection challenges. National and international statistical organizations have had to implement a variety of innovative actions to ensure the continuity of key statistical collections and outputs.

COVID-19 has posed challenges for some longstanding statistical concepts, not least, the definition of unemployment. The internationally agreed statistical concepts and definition of unemployment, set out in the 1982 ILO Resolution Concerning Statistics of the Economically Active Population, Employment, Unemployment and Underemployment17, have been strained by confinement. In summary, to be classified as unemployed, a person must be without work, available for work, and seeking work during a reference period. But what happens when an economy closes? Curiously, strict application of ILO rules, despite the difficulties presented for job search amid COVID-19 restrictions, and the variety of government social protection and furlough schemes put in place to protect labour that have fully or partially replaced wages and salaries usually paid by employers, could yield a counter-intuitive result, whereby the numbers employed and unemployed would be little impacted by the pandemic. Consequently, some countries have made special adjustments, in respect the ILO standards, to yield credible results. For example, in Ireland the Central Statistics Office presents their traditional (or standard methodology) monthly unemployment estimates alongside an alternative COVID-19 adjusted unemployment measure that estimates the share of the labour force that were not working due to unemployment or who were out of work due to COVID-19 and were in receipts of special COVID-19 related social protection or unemployment payments. In May 2020, the traditional measure for unemployment was estimated to be 5.8 per cent, whereas the COVID-19 adjusted rate was 26.1 per cent (CSO, 2020).

The compilation of national accounts is also facing similar conceptual challenges, not least how to treat or account for COVID-19 related payments to enterprises, employees and self-employed in the system of national accounts and GDP. The Intersecretariat Working Group on National Accounts (2020) advise that government supports to employers to maintain businesses and keep employees on payroll, and government supports to self-employed to support business, should be recorded in the SNA as ‘other subsidies to production’. Government supports to households to maintain income (depending on whether they are considered as social benefits or not), should be recorded as social security benefits, social assistance benefits or miscellaneous current transfers. For example, the Coronavirus Job Retention Scheme implemented in the United Kingdom, where employers of furloughed staff are paid 80 per cent of salaries by government, will be treated as a subsidy to business, to be netted off the income measure of GDP (Athow, 2020).

COVID-19 has additionally thrown up a whole host of methodological issues. For example, many national statistical offices have had to either temporarily suspend face-to-face interviews or switch very quickly to other modes of data collection, such as telephone or web-based collection, web scraping, or greater use of administrative or privately held data. Important household surveys, such as labour force, consumer price index, household budget, income and living condition surveys have suffered from disruptions. This presents not only logistical and infrastructural challenges but also significant statistical challenges. For example, creating telephone databases or adopting dual or multiple frame sampling (a challenge if surveying both landline and mobile phones) are significant complications. Furthermore, if NSOs switch from CAPI to CATI, then they will also need to adjust for ‘mode’ as each mode of collection has its own inherent biases. They may also need to deal with suddenly reduced response rates (ILO, 2020d). Many traditional imputation and seasonal adjustment procedures, which rely on historic patterns, will have been rendered redundant by containment.

Equally, enterprise surveys too have been impacted as many businesses are closed or have ‘relocated’ to new addresses as business owners and employees work from home. The crisis is likely to pose very particular challenges for the quality of statistical business registers, as enterprise churn, the washing machine of enterprise births and deaths, is likely to be much higher and less predictable than usual. In turn, this will impact both sample selection and the weighting of many other business surveys. NSOs have also had to grapple with the knotty problem of compiling price indices when markets have shut down. For example, how to continue residential and commercial property price indices when there are no transactions, and consequently no reported prices for some products. How do you impute for a market that does not exist? These are important questions for the indices themselves, but also for the derived deflators – the basis for volume and constant price measures. COVID-19 will also disrupt normal seasonal patterns, introducing a set of new challenges for statisticians hoping to present consistent time series and provide timely information.

From an official statistics perspective, COVID-19 hit at a particularly unfortunate time, as 2020 was the beginning of the next round of the decennial census of population. More than 120 countries were scheduled to conduct census enumeration between 2020 and 2021. Censuses are expensive, and if delayed, many of the sunk costs cannot be recouped and may result in cancellations rather than just postponements. By early May 2020, UNFPA reported that already 64 countries had reported adverse impacts of COVID-19 on their population and housing censuses (CCSA, 2020). In a recent survey, ‘Monitoring the state of statistical operations under the COVID-19 Pandemic’ jointly conducted by UN DESA and the World Bank (more below), 58 per cent of the 61 countries who were planning a Population and Housing Census in 2020 reported impacts on their preparatory activities, with more than half (53 per cent) postponing fieldwork to later in 2020 or to 2021 or beyond (UNDESA and World Bank, 2020). If the global census round is disrupted, this will ripple through the entire statistical system, as not only will many minority and vulnerable populations go uncounted, but as the denominator for so many other indicators, the impact will be felt in every statistical domain – social, economic and environmental.

National statistical systems and international statistical offices around the world have risen to the challenge. Like many other industries, they have switched rapidly to working from home, while simultaneously introducing new data collection methodologies, adapting existing conceptual frameworks to incorporate government interventions and yield technically accurate but plausible results. There has also been considerable innovation, with many offices having introduced new data sources, surveys and statistics.

Statistics Canada (2020), for instance, introduced a monthly flash GDP estimate in April 2020 to provide a faster approximation of the scale of economic disruption in March 2020. Statistics South Africa (2020) and the ONS in the United Kingdom (Athow, 2020), among others, have introduced online surveys on the business impact of COVID-19 and surveys to assess the impact on people, households and communities, similarly to the new Household Pulse Survey of the United States (United States Census Bureau, 2020). Many offices have partnered with government and private organisations to access timely data sources, such as big data on ship tracking, road traffic sensors, credit card transactions and mobile phone use. Statistics Netherlands improved the timeliness of many statistics, including mortality, retail trade, use of energy, bankruptcies statistics and introduced new statistics on emergency measures and mobility among others. Statistics Estonia and the Ghana Statistical Services, for example, have been measuring mobility under the confinement period using anonymized mobile phone data (Migration data portal, 2020; Ghana Statistical Services and Vodafone Ghana, 2020). A quick adaptation of data collection methods has also been necessary under confinement, including in South Africa, where a large proportion of price data collection was moved online (Statistics South Africa, 2020), and offices like the United States Census Bureau and Statistics New Zealand have started using credit card purchase and supermarket price data directly for statistical production.

As noted above, the UN DESA and the World Bank's Development Data Group, in coordination with the five UN regional commissions, recently conducted a global online survey to monitor the nature, scale, and scope of the impact of the coronavirus crisis on statistical agencies, as well as to identify new data needs. The survey results, covering 122 responding countries, highlight the tremendous challenges being faced by national statistical offices as a result of the COVID-19 crisis, but also illustrate the range of measures being taken to mitigate negative impacts and meet new data demands. 65 per cent of NSO headquarters offices are partially or fully closed, 90 per cent have instructed staff to work from home, and 96 per cent have partially or fully stopped face-to-face data collection. The results also show that NSOs in low- and lower middle-income countries have been hardest hit, where nine out of ten offices report impediments to their ability to meet international reporting standards and additional funding constraints. Unsurprisingly, the survey has reinforced the importance of technological infrastructure and skills, which has allowed some offices to find substitute modes of data collection for face-to-face interviews. Worryingly, at a time when good quality statistics are needed, 38 per cent of responding NSOs reported funding cuts.

Figure 22. Impact of COVID-19 on NSOs‘ funding by country income group
(Percentage)

UNCTAD Statistics responded quickly by introducing a new quarterly nowcast for merchandise and services trade (UNCTAD, 2020b), providing up-to-date information on global trade (see section COVID-19 and the SDGs – Economy). The online statistical capacity development that UNCTAD provides in cooperation with WTO and UNSD has continued uninterrupted (see UNCTAD in Action TRAINFORTRADE), bringing capacity to developing and developed countries all around the world.

36 international organisations also quickly came together, under the aegis of the CCSA, and assembled a report in May 2020, ‘How COVID-19 Changed the World: a statistical perspective’, providing a wide range of statistics to illustrate how COVID-19 has impacted different aspects of our lives (CCSA, 2020).

There is a lot of work to be done. The fast spreading COVID-19 pandemic shows the interconnectedness of countries and underlines the need for more granular, interlinked and timely official statistics. There is, most likely, no return to ‘business as usual’ for official statistics. The statistical community will need to reshape future official statistics by exploring new partnerships, integration of surveys, registers and alternative data sources for the provision of timely, agile and more bespoke statistics to inform policies with a rich picture of the economy and society – be it on health, employment, production, trade, globalisation, technology, inequality, skills, environment or their interactions. Interesting debates are underway on what this future might look like on the Statistical Journal of the International Association of Official Statistics discussion platform18 and on the UN DESA COVID-19 Response page19.

2020 hindsight

In recent years, there has been much debate surrounding the ethics of using personal data and what are the acceptable trade-offs vis-a-vis privacy. Captains of industry 4.0, such as Mark Zuckerberg (Facebook), Scott McNealy (Sun Microsystems) and John McAfee (McAfee Associates) have all argued that the concept of privacy is extinct (Kirkpatrick, 2010; Noyes, 2015; McAfee, 2015). Many disagree and have voiced concerns over the loss of privacy (Pearson, 2013; Payton and Claypoole, 2014; Zuboff, 2019). New data protection legislation in Europe (EU, 2016) and in California (State of California, 2020) suggest that at least some legislators still see a value in privacy. Nevertheless, it is difficult to see how the concept of privacy can survive unscathed with the relentless drive towards the Internet of Things - smart phones, smart TVs, smart cars, smart homes and smart cities, and harvesting of personal data. Soon it seems everything we do will be monitored. One cannot help but wonder whether privacy as an ‘ideal’ might still be alive and well, but privacy in ‘practice’ is on life-support; day after day, we read about enterprises and institutions failing to protect personal records.

There is a risk that COVID-19 may exacerbate this situation. In a time of crisis, populations expect their governments and public services to adapt and provide new services (and information) without delay. At the same time, populations tend to have elastic ethical frontiers. Thus, social license typically contracts in good times but loosens in emergencies, with the result that populations are less concerned about the how job gets done as long as it gets done. While this is understandable, reactions to recent crises have arguably permanently stretched the limits of the pre-crisis ethical frontiers. For example, following 9/11 many legal barriers to data sharing were quickly swept aside as the political focus shifted from privacy to security (Lyon, 2001); many were never reinstated. The COVID-19 pandemic may do the same. In March 2020, it was reported that 19 countries were accessing citizen data to track the virus (Cozzens, 2020; Doffman, 2020), including Austria, Germany, Italy, the United Kingdom and the United States of America, while Liechtenstein is even planning to electronically tag and monitor its citizens (Financial Times, 2020). Furthermore, Google began publishing detailed statistics, harvested from their applications and platforms, on population movements (Kelion, 2020; McGrath, 2020). Yale’s professor Sudhir neatly sums up the situation: ‘Privacy concerns are on the back burner during this emergency’ (Sudhir, 2020). While this is understandable, it raises the question of what happens after the crisis? Can we put the genie back in the bottle afterwards?

COVID-19 may have additionally unwittingly exposed tensions between community and individual rights. Many will argue that the growth of the Internet of Things and the ability to measure everything is a good thing. But good for who? As Sen (1999, p. 150) reminds us, ‘in judging economic development it is not adequate to look only at the growth of GNP or some other indicators of overall economic expansion. We have to look also at the impact of democracy and political freedoms on the lives and capabilities of the citizens’. There are some who now fear the growth and centralisation of technology as a direct threat to democracy (Reich, 2015; Taplin, 2017; Zuboff, 2019). Data can both be a tool and a weapon; used for good or evil. As noted in the 2019 In Focus of SDG Pulse The many faces of inequality, (UNCTAD, 2019) equal access to data is of central importance to achieving the 2030 Agenda. The growth in proprietary data is exacerbating the split between ‘the data haves and have-nots’ and is creating a new dimension of inequality.

Notes

  1. WHO (2020a, report #132).
  2. WHO (2020a, report #1).
  3. WHO (2020a, report #11).
  4. On 11 February 2020, the WHO, in consultation and collaboration with the OIE and the FAO, named the novel coronavirus as COVID-19.
  5. Africa, Americas, Eastern Mediterranean, Europe, South East Asia, and Western Pacific.
  6. WHO (2020a, report #37).
  7. WHO (2020a, report #133).
  8. WHO (2020a, report #58).
  9. The date of the first stage reopening.
  10. Range around the estimate: 1.8 – 4.7 million people.
  11. WHO (2020a, report #112).
  12. 24 participating countries or territories: Austria, Belgium, Denmark, England, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, Malta, the Netherlands, Northern Ireland, Norway, Portugal, Scotland, Spain, Sweden, Switzerland and Wales.
  13. This composite index provides a quantitative score of the implementation of nine types of containment and physical distancing measures in over 150 countries. It is updated daily from a variety of data sources. It is scaled to a 0-100 range, with higher numbers indicating more “stringent” containment measures. For a complete description of these index and its methodology, see Hale et al. (2020).
  14. For a more complete compilation of economic measures, see IMF (2020) and OECD (2020).
  15. This note is based on ongoing research and should be taken as preliminary. More developed research report will appear as an UNCTAD research paper.
  16. These tests provide only a partial view as we are not capturing the complex interrelationships between the labour market and, for example, policies imposed during the COVID-19 lockdowns. Instead, we measure, in a simple way, the responses of the selected employment variables to trade and unemployment, as a marker of the state of the economy, to provide insights on gendered economic structures which can help anticipate future developments.
  17. Adopted by the Thirteenth International Conference of Labour Statisticians, see ILO (1982).
  18. Official Statistics in the context of the COVID-19 crisis, see Statistical Journal of the International Association of Official Statistics (2020).
  19. UNDESA (2020).

References