Official statistics in the search for solutions for living with COVID-19 and its consequences¹

2.1The consequences for the economic base and social conditions

In many countries, the main action to prevent a catastrophic rise in death rates from COVID-19 was a dramatic shutdown of economic activity. The immediate closing down of economies will have severely damaged critical forms of wealth creation. It is most likely that such a large fall in national income will lead to a rethink of the economic, political and public health structures, with consequences for the balance between central and local governments. A permanent decline in demand for many goods and services will shape economic activity in all countries even after a vaccine for COVID-19 has been found. A rethink of globalised value chains and value networks will occur. In the case of New Zealand, a large share of economic activity and employment depend on international tourism. The New Zealand Treasury noted on April 14.

The New Zealand Government has established four levels of restrictions on activity, the most restrictive level 4 was put in place on 25 March, for four weeks. On April 14, the NZ Treasury estimated the economic impacts of each of the levels as below.

While in both economic activity and social welfare the consequences of a policy inflicted period of closure of economic and social connections will be large, there will be longer-term falls in demand as the global economy falters that are more difficult to anticipate fully. These two effects will reduce at different rates. Looking past the immediate lockdown period, we can expect quite huge and differing consequences for tax and transfer systems. The scope of potential change is outlined in a paper from Koi Tū: The Centre for Informed Futures at Auckland University [1]. There is an agenda for responding to COVID-19, to fit the implications of the consequences identified in the paper. At the time of publication, the optimistic tenor of the paper is founded on the opportunities that New Zealand has through its resource endowments and recognition of the integrity of agricultural products. This is based on a rapid return to normal arrangements while the country itself gradually relaxes its isolation from high levels of international contact. A recent paper by the Grattan Institute [2] in Australia focuses on the economic implications for Australia. Both papers present assessments that will have relevance in most countries with a similar population and economic structure.

The lockdown of economic activity and limiting personal interactions is unsustainable for long periods. Without lockdowns, health facilities could have become overwhelmed, and in some countries, lockdowns have not prevented this. New Zealand and Mexico have the lowest capacity of intensive care beds per 100,000 people of the OECD countries, just over 1/3 the ratio per 100,000 population of Australia and England. In order to be better placed to treat large numbers of hospitalised people, all hospital activity that was not critical ceased in New Zealand, as well as other places. Ironically, for some a lockdown is certain to have increased the inequities that reduce the personal capacity to fight COVID-19, through differential access to even the most basic goods, and delays in diagnosis of treatable life-threatening health conditions. New Zealand has been fortunate in the quality of its nation-wide broadband infrastructure, which provides very high-quality broadband access to most but not all. This online connectivity has enabled a huge range of activities to continue, by expanding possibilities for continued employment, retail sales, entertainment, health care and personal connection on a scale that might have been unimaginable only two months before. A significant share of this expansion may become permanent and should have downstream effects of entities that met these needs before the lockdown.

For the immediate and medium term, when governments decide when the time is right to wind back the restrictions on economic and social activity, the much heightened economic and health risks that we expect to be normal for several years need to be judiciously balanced. Public trust in government actions will be essential and trade-offs will involve balancing the limited evidence we might have, the policy perspectives of Ministers of the day and the capacity to challenge them in Parliament and by interest groups as well as the scientific community. The limits to achieving full public compliance in physical distancing could become an important influence on the speed and size of increments in return to a post COVID-19 norm.

Recognition of what a post COVID-19 norm will be like is important early in the economic recovery, as countries need to assess the potential for recovery of demand among the industries that make up their economic base. For those countries where international tourism is a significant contributor to GDP, as in New Zealand, the factors that will be important in how to cushion a very long lasting drop in demand will include the potential for domestic tourism to replace part of what is lost. We need to become clear on the other sectors which will be similarly affected. In those cases, there will become a time when transfers to those out of work, retraining and other forms of personal support may be preferred to payroll support at an enterprise level.

A good understanding is needed of what now makes up competitive advantage against other countries with its businesses, labour force and natural endowments. This will inform understanding of the opportunities that New Zealand has, and will be relevant in other countries. It will inform any efforts to stimulate innovation. It is also important to know as early as possible where forms and industries will have ceased permanently, so that they are not included in business recovery measures, rather than individual welfare transfers. Government actions in taxation, regulation, business support, welfare transfers and societal solidarity will essentially be shaping the structure of the New Zealand economy as it re-emerges from its predominant state of quarantine.

Lockdown rules will have temporarily damaged the economic viability of many local enterprises through the concentration of services through national chains, where overseas ownership is predominant. For businesses, we have seen nation-wide chains becoming more profitable during lockdown while many small local businesses will suffer losses, which for many will be irrecoverable. In this respect, the latest Treasury scenarios which assume that the decline in economic activity in 2020/21 will be reversed by 2022 may be misleading, and underestimate the economic predicament that COVID-19 has placed us in. The likelihood of otherwise viable small businesses recovering may be overestimated, while the impact of a significant rise in inequality will be difficult to recover from, given the experiences of many in New Zealand after the 1980s economic restructuring and privatisation that occurred then.

On an individual level, lockdowns will most likely have amplified the inequalities that exist in New Zealand through the curtailing of universal public services, the reduction in targeted support services, the reduced connectedness with critical services, and the reliance of many employees without jobs on the economic viability and business practices of employers.

2.2Information imperatives

Official statistics enable governments and communities to get some context against which to judge what is immediately in front of them. For example, in the UK, the Office for National Statistics has provided weekly information about what is happening to deaths from all causes. This has enabled a debate not just about deaths attributable to COVID-19 in NHS hospitals, as widely reported from administrative data on a daily basis, but also what is happening in care homes and deaths that may be due to reductions in medical attention for other (often more prevalent) reasons such as cancer. The use of frameworks, especially the Sustainable Development Goals (SDG) framework, to put health, economic, social and environmental questions alongside each other is a critical asset of the official statistics system that should become a big new priority.

With the right access to data, statistics offices can also readily analyse geographic effects (to answer questions about differential policy responses in different areas) and issues of different impacts for different groups (are ethnic minority groups, the disabled, women, older/younger people being left behind). The priority should be about the distributions rather than the averages as has often been the case in the past.

Similarly, the pandemic has reinforced the priority for international cooperation. What do we need to do to be confident that the figures from one country can be compared with another? For all the UN guidance we have the pandemic has shown that as soon as you try and compare one country with another you still often end up with more questions than answers.

In New Zealand, an urgent decision to bring in a lockdown and restrict economic activity and public interaction was one of few options available to Ministers. It was significantly influenced by the limited readiness of testing and health resources, and strength of the accumulating epidemiological evidence both in New Zealand and globally. Much more information will be required for a staged return to business when almost all enterprises and most of the population face an uncertain path to a new normal.

Until there is a vaccine, all countries care at risk of a resurgence of COVID-19. Screening can be more strategic, and consequently more adaptive to the constraints of testing on one hand, and the knowledge that exists of differences in predisposition to infection, and the consequences of infection both in terms of mortality by age, and also potential to infect others. There is much uncertainty about the quality of much of this information, especially where it is derived from screening approaches for testing that vary between countries, and over time in the same country. The time-lag between having a detectable condition, showing symptoms and then seeking medical attention could be up to ten days. References provided by Professor Nick Wilson, University of Otago, noted:

Ways of effectively using the resources used in testing by selecting groups for screening in advance of them becoming fully aware of their infection would need to be sufficiently efficient for that delay to be of value in signaling potential shifts in infection in the population. At an individual level testing in this way would increase the possibility of placing in quarantine infected individuals much earlier than before. The continued absence of a vaccine means that future screening strategies need to be more adaptive, and bring together expertise on statistical sampling, epidemiological, medical practitioner and public health.

Much of our current statistical capability is anchored in information sources which enable regular comparisons of levels of activity over common periods of time and between sectors. Such measures are usually designed with a fixed frequency and report with a lag. Coherence is usually strong within types of statistics (e.g. economic statistics, demographic measures), but much less strong between types. There are questions we need to address in each of the five stages. They are not those for which official statisticians will have planned. Where we need to speedily develop additional indicators that provide more frequent, more immediately available information, sometimes this will not be possible without compromising on representativity, quality measures or coherence. This must be transparent if long term trust is not to be put at risk. We need sources to be more granular, to be able to pinpoint emerging concentrations of need, and measure extra-ordinary changes to rates of change almost in real time. Developing ad hoc responses requires highly intense methodological oversight because the range of issues that need to be considered does not change despite the need for a rapid response. Understanding the interdependence between the health of the public and the strength of the economy will challenge statistical systems that are compartmentalised, not only in concepts and frameworks, but in the modalities of information gathering. Government administrative data sources shape such compartments, with their prime focus on administering existing statutes. Statistical surveys often miss the very groups who are not able to connect to public or community services, even when they are intended to be universally available.

The scale of policy change will require still unforeseeable changes in public information which will test the role of the state in requiring information. We should see a heightened trust in experts. Public acceptance of when and to what extent governments can place whole populations in quarantine has varied around the world. Acceptance of strong responses to knowledge of a pending catastrophe has been stronger in countries with recent experience of pandemic risk (e.g. Taiwan, Singapore, South Korea, Hongkong). Other countries including New Zealand have associated major decisions with the evidence base that scientists have provided and discuss freely in the media. COVID-19 has required a heightened scientific and statistical competence among the media that when faulty can amplify distrust. It is vital that the capacity to understand not only the incidence in any given day found by testing, but information about the evolution of testing practices is also critical.

3.1New Zealand’s situation at the point of beginning nationwide quarantine

NZ adopted an elimination strategy starting at midnight on 25 March. At that time no deaths had occurred, and the number of diagnosed cases of COVID-19 was 318. At that time, almost all cases involved transmission as a result of overseas travel. Supporting this decision were NZ epidemiological data and modelling of community interactions. At that time the consequences of not taking precautionary action had been demonstrated by the speed of infection in Italy and Spain. Deep concern about a weak policy response in the United Kingdom and United States naturally spilled over to politicians and the media in New Zealand. New Zealand had some advantages:

As in many countries, New Zealand initially had a limited stockpile of necessary equipment and little test capability. This must later bring about a rethink of the quality of strategic planning for likely national emergencies and the monitoring of sentinel events. The obvious costs visible so far would be initially reflected in delay in starting lockdown and requiring health staff to face avoidable risks of infection. This meant that New Zealand was slow to scale up testing – but once expansion began it reached the current levels quite quickly, extending beyond those judged as a carrier or otherwise diagnosed in the community. The first need for testing was to have a diagnosis of those who presented some recognised symptoms. Initially, with few testing resources, this has been the key priority, with those identified as being potential cases also able to be tested. After these needs were met, there has been less certainty about how best to use limited testing resources and to effectively build confidence that we are moving to eliminate all viral transmission in the community. A John Hopkins paper (April 2020) summarises why tracing in cases of COVID-19 is more complex than that of other infections.

It became clear quite early on from experiences with the limited capacity of intensive care units to handle huge volumes of people that ensuring the resource needs implicit in their operation (ventilators, nurses) needed a greater understanding of the value chains for health supplies. Their ongoing monitoring needs to become more extensive. For the materials needed to fight future pandemics, there may be a need to have a systematic approach to monitoring global value chains, and formalising regional alliances.

3.2Testing to inform the urgency and scale of government services responses

In summary, up until the ending after nearly five weeks of a nationwide lockdown at the most severe level on April 28th, epidemiological modelling has been one of the mainstays of Ministerial decision-making. Scientists have been able to determine the path of infecting the population with COVID-19 in New Zealand through reacting to individual medical conditions as they presented.

The information needed to determine whether an individual needs treatment will not provide the information needed to understand the predisposition of the population at large to COVID-19. What we have some certainty about is:

After South Korea, Taiwan, China and Singapore, Iceland and New Zealand were the two non-Asian countries cited for the success of tracing methods in a recent John Hopkins plan [5] for tracing in the USA. They noted that:

In Iceland [6] two forms of testing were carried out. The first form was to test all who were regarded as infection risks, and the second was to test a sample of the population whose situation was not known. At the stage of informing the government response to COVID-19, the testing of at-risk persons was much more informative than the samples of the population for assessing the risk from COVID-19. The population screening by sampling the population was in two parts – 80 percent was of people who self-selected, and 20 percent or 2283 people were selected at random in the population. There were 33.7 percent who responded to a request to participate sent by text. The population sampling was carried when infection was at high levels, yet the prevalence found by population sampling remained constantly below one percent throughout the trial. The sampling results might suggest that sampling strategies may need to evolve as the risk of infection from COVID-19 changes, until elimination is guaranteed by a vaccine. The study was able to identify the different origins of the infection by examining the distribution of variants across the SARS-CoV-2 Genome of those tested.

3.3Monitoring the potential for infection of the population at large

Epidemiological models have been able to inform the public through their ability to project the speed and scale of COVID-19, under the conditions when it was able to develop before it was finally noticed in those who required medical care. The epidemiological models are based on a mix of transparent and less obvious assumptions, as is the case with all models. Key parameters of the epidemiological models are anchored in measures with which the public have been able to empathise, in particular:

Testing [7, 8] of those who present to the health services involves a form of self-selection that cannot provide adequate information to connect the experiences of those being tested, to the predisposition and prevalence of the population at large to COVID-19. It is essential that diagnostic testing occurs for anyone who attends a primary care setting or a hospital with acute respiratory symptoms – this is a given and it is especially important to minimise risk to precious healthcare workers at this time [7]. Tracing of contacts of individuals tested positive at a health clinic after self-selection for testing would be immediate. Guaranteeing the elimination COVID-19 or even suppression to a very small threshold requires sufficient certainty that no residual level of COVID-19 has been detected among people from broad age and ethnic groups and places and those with known predisposition. Until that stage is reached, public health experts advise that the testing of the population for diagnosis of acute illness and to protect healthcare workers needs to have first claim on testing resources. While testing could be expanded to include those, who self-select by seeking confidence that they are not potential cases, there are no wider benefits in this beyond the personal comfort a tested individual will probably gain. While it was vital for the health services to properly diagnose those who come before it, measuring the predisposition of the population at large is a different measurement problem, for which self-selected testing is inadequate.

The prevalence of COVID-19 in the population is not measured by such processes. Having additional resources beyond that needed for testing those identified as a potential risk creates the opportunity to use statistical science to estimate the unidentified level of infection in the population at large. Government responses need sufficiently reliable estimates of the prevalence of COVID-19 in the population, and the exposure to the conditions that cause it to spread. Statistical science through the well-established science of statistical survey sampling exists to connect COVID-19 experiences to the population at large.

As lockdown levels loosen, we need to know much more about the prevalence in the population at large, and testing resources have now become available for this. This is a vital next step, as from what we measure now, and what we learn from other countries, we know there is a differential impact on different age groups. Gluckman and Bardle note “A starting point would be sentinel screening and testing of supermarket workers and healthcare workers in areas such as Queenstown where there is a very high prevalence. Our current testing regime effectively is the equivalent of looking for lost keys only where there is light under the lamp post.”

The quality requirements of COVID-19 tests are different for diagnosis than for population monitoring. Counts of those proven to be infected prove a lagged measure of the infection risk in the population, as noted earlier. In addition to the measurable sampling errors arising from random sampling the population there is uncertainty associated with the tests. There are two aspects to be considered. One is the sensitivity (true positive rate) of the diagnostic test and specificity (true negative rate). This refers to the RT-PCR (reverse transcription polymerase chain reaction) test performed with throat swabs, which has low sensitivity. This can be accounted for by increasing the sample size. Although the PCR test has high analytical specificity, it has lower diagnostic specificity. This means increasing the sample size or re-testing will not necessarily avoid the problem of having a high false positive rate. This problem will dominate when the prevalence of COVID-19 is very low. The analysis required to develop a randomised sampling process relevant to the diverse characteristics of distinct population subgroups and the practicalities of making contact with them will involve agreeing on testable assumptions, and such scrutiny would determine the feasibility of any randomisation process.

3.4Random sampling schemes for monitoring infection risks in the population

It may well be that the criteria and means of screening people to test for the presence of COVID-19 infection once low levels or elimination have been reached will vary country by country. The preferred means will probably change at each stage in opening up economic activity as well as depending on the size of the country, its degree of isolation from high risk countries, as well as the prevalence to date of COVID-19 and the outcomes within the infected population. The path taken to open up economic activity and extent of social connection permitted will also be important, as will the extent and scale of high intensity urbanisation. It can be expected that those in public health, medical practice, service providers, carers and employees will weigh the health and economic risks differently. There is a growing opportunity to draw on international experience in determining design assumptions. As of this writing, Iceland [6] has one of the highest per capita rates of coronavirus testing of any country in the world. The number of tests carried out by the Icelandic health authorities and deCode Genetics up to 20 April totals33 43,800 or 120,300 per million. Long before the development of an antibody test, Germany, for example, began conducting as many as 50,000 diagnostic tests a day to help trace and isolate cases.

The widespread testing has revealed that as many as half of the people who test positive for the virus have no symptoms at all. Early reports [6] cited “The number of individuals tested by the country’s health authorities and the biotechnology firm deCode Genetics – 3,787 – roughly translates to 10,405 per million, which compares to about 5,203 in South Korea, 2478 in Italy, and 764 in the UK”.

Population sampling will not be perfect given the current constraints for interacting with people and the invasiveness of the current PCR test. Deciding exactly what we need to know from a sample of the population at large needs to be determined by epidemiologists and other public health experts. We need the capacity to assess an individual’s predisposition, willingness to be tested, and exposure rating. Where there is no (up-to-date) list frame of the population (population register) the most preferred and generally used approach to select a population to measure the prevalence of some condition is to construct an areal frame of clusters of dwellings, sample those clusters by some random (probability based) sampling scheme, enumerate the sampled clusters and randomly subsample the enumerated dwellings. Generally, the people usually resident in the dwelling are considered to be a household. Depending on potential efficiency gains, or the need for subpopulation (regional) estimates, the clusters may also be assigned to strata. A random stratified (cluster) sample of the population provides both estimates of what we want to measure but also probability-based estimates of the errors in those estimates, which is not the case with purposive sampling.

3.5Some practicalities of random sampling

There is a need to ensure that any complex randomised sampling method itself can be put in place by expediting the necessarily demanding requirements for introducing new information sources. This could be achieved by establishing a team of the top experts in statistical sampling who can work with surveillance-orientated epidemiologists and laboratory-based microbiologists to determine the best approaches to testing in the community. There will be features of population sampling additional to those raised in this paper. There will be a need to use existing science leaders in government to work with policy experts to determine the appropriate a priori thresholds and develop sampling strategies for determining with 90%, 95% and 99% confidence that no more pandemic virus is circulating. Such measures of confidence need to take account of the sensitivity (true positive rate) and specificity (true negative rate) of the test.

What would those tested get as a benefit? Quite simply, were they to test positive, they would receive treatment earlier than would be likely if they waited until their symptoms became obvious, and if they are asymptomatic they would know the risk that they were placing on others, so that they could take appropriate action. When people refuse tests, that in itself would give us an indication of the limits to public compliance. This would prove a useful measure in informing the public and assessing the risks of infection from a random outbreak.

3.6Comparing sampling alternatives

The main sampling options are compared below.

3.7The importance of an adaptive approach to screening

Until there is a vaccine, all countries are at risk of a resurgence of COVID-19. Screening needs to be more strategic, and consequently more adaptive to the constraints of testing and of limits to the relevance of past knowledge. This includes the differences in predisposition to infection, and the consequences of infection both in terms of mortality by age, and also potential to infect others. Using this additional knowledge, it is possible to purposively select clusters of people and apply a different sample design and approach possibly for each cluster and learn a lot about that cluster. The expert judgment needed for purposive selection of cohorts for random sampling needs to be based on medical knowledge of different risks and outcome costs of clusters.

It may be that where elimination or even containment has been reached, that measuring the prevalence in the population overall is of less importance than testing for changes in the prevalence in preselected subpopulations who are in a situation The time-lag between having a detectable condition, showing symptoms and then seeking medical attention could range from four to ten days [3, 4]. Screening by random sampling of key clusters needs to be sufficiently efficient for that delay to be of value in signaling both potential shifts of infection levels in the population, but also at an individual level recognising the possibility of placing in quarantine such individuals much earlier than before.

The continued absence of a vaccine means that future screening strategies need to be more adaptive, and bring together experts in statistical sampling, epidemiological, medical practitioner and public health expertise to determine the information most important in selecting clusters, and how best to contact and select a sample for testing. There is a need to determine what inferences can be made about the population at large from individual cluster results. When to reserve testing resources for clinical needs and contact tracing will be an important factor in the scale of sampling. Ensuring that institutions that are supposedly under full lockdown are sampled such as care homes and prisons need to be included in the potential mix of clusters. When a situation of elimination or containment has been reached, the form and scale of testing needs to reflect the huge cost of a return to lockdown. Once full lockdown has ceased, screening strategies need to be more adaptive, there is a need to bring together quite frequently statistical sampling, epidemiological, medical practitioner and public health expertise.

In comparison with the purpose of a broader monitoring regime for monitoring the predisposition to infection of various kinds, the purpose and function of sentinel monitoring of communicable disease initiated during a crisis risks being seen as an alternative rather than supplement. It is not clear that there is strong scientifically driven oversight of the various approaches to screening and managing their inter-dependence. The full potential knowledge gained from screening as a whole will not be realised.

4.1Immediate information needs during the transition

Until there is certainty of the elimination of COVID-19, there will be a staged removal of restrictions on economic activity, an iterative process that may last many months. New Zealand moved out of its highest-level phase on April 28 but will enforce high levels of distancing. Balancing the public’s welfare through ensuring the effective oversight of COVID-19 risk while reversing the lockdown on economic activity brings difficult political choices. Building confidence in these needs to be founded on trustworthy statistics based on measurement processes that have to be both more frequent and timelier than we are used to. The inability to make personal contact has led to a degradation of statistical survey capabilities fundamental to official statistics which will create major information gaps. Filling these gaps will require a much more intense collaboration in a rapidly expanding network of organisations who monitor activity that can reduce the emerging gaps.

Official statistical institutions need to recognise both the fundamental importance of their strengths, and the extent to which traditional weaknesses will now be amplified. The longstanding strengths of official statisticians remain vital, in the intellectual coherence of diverse sources, the fundamental building blocks of integrated statistics, and the key elements of system stewardship including developing coordination, governance and quality management capabilities [11]. This includes classifications and frames of both area and industry, and comprehensive protocols of good practice. These strengths will be vital for strengthening the integration between economic, health and social statistics. They are a long-term investment that is readily able to be drawn on.

It will be critical for all sectors of official statistics that key external users are tightly integrated with experts from the statistical office. The need for strong collaboration across government and with business organisations, as well as non-government organisations has become essential. For many statistical measures and sources, the original purposes which justified their initiation have become comparatively minor, while the most significant uses may be unrecognised. It is often only after longstanding statistics have ceased that those who produce them discover how incomplete is their knowledge of how the statistics were used, and what they meant.

In developing new measures, the time frames involved will be tight, few will have all the necessary knowledge, and there is little room for the time lags in the experimentation and testing as are usual in introducing new statistical measures. If past experience is any indication of how the current crisis will be managed, statistical offices will receive only some of the resources needed for this new work. Part of user collaboration will be to identify where existing statistics can cease, reduce in frequency or granularity, or be deferred. Across government, resources are going to be very tight over the period of such change.

The New Zealand government has chosen to act in a way that was amongst the most decisive and strongest lockdowns in the world at the time it was put in place. This does not make unravelling of the lockdown any less dependent on the information that is essential for determining the next steps, which could occur any time from a few weeks or several months. When the level of lockdown changes, new information needs must be recognised and provided for. During the level 4 period, the New Zealand department (Ministry for Business, Innovation and Employment) responsible for employment policy estimates that some 510,000 essential workers had been continuing to work in places of employment, while another 130,000 essential workers have been working from home. This has risen by another 400,000 from April 28.

Countries which chose a more staged approach to entering quarantine have had to be highly effective at monitoring, but some (United Kingdom, United States) have simply moved iteratively to strengthen lockdowns possibly more as a result of regret rather than thoughtful anticipation.

Protecting access to food during a crisis is a fundamental element of government action while it lasts. Community access to food outlets is concentrated in supermarkets, which themselves are located in well populated areas of the country.

4.2Official statistics “Fit for Purpose” in a post COVID-19 world

The arrival of COVID-19 is changing not only the type of statistical information government is beginning to need but it is likely to require significant change in the way information is gathered.

This severely tests statistical survey models whose strength is in measuring small amounts of incremental change with high precision, a good while after the reference period. For businesses to be able to trade as before, there may remain some collectively imposed restrictions, as well as a range of personal ones until there is certainty that COVID-19 will not arise again once it is believed that it has been eliminated, Measuring changes in the extent of interpersonal interactions may be of importance in monitoring the wellbeing of the population.

COVID-19 does not change the criticality of developing statistical measures to take account of environment capital, climate change, energy sources, urbanisation and water. A comprehensive framework already exists for these vital issues, but commitment is varied. The COVID-19 priorities will expedite the methodological and technological advances needed to expand the scope of regular statistics and increase their granularity with respect to place. Priorities that are driven by the need to manage the consequences of COVID-19 can reinforce the importance of the Sustainable Development Goals. All of those listed below can be achieved within the Sustainable Development Goals framework.

A preliminary summary of the priority areas for statistical office response to COVID-19 and its aftermath is presented below.

4.3Adapting contemporary technologies

The potential scope for recording measurements of digital technologies is bounded by the limits on accessibility to the appropriate devices and connectivity. New Zealand has a wide penetration of high-quality broadband which could be a stronger platform for redeveloping the means of contact with people and business, as well as monitoring environmental conditions. Where information is captured by digital means, it can usually provide a degree of granularity about place, frequency of transactions and immediacy of availability that traditional means do not. However, for such data, the means to ensure the statistical integrity of the population covered needs to be provided. Universality of coverage will be constrained by selection and rationing practices inherent in the processes in question. The readily available counts that summarise administrative processes have no conceptual coherence, as is the case of the administrative data of the State. The definitions describing the content of such data collections is usually not consistent with the concepts around which economic, demographic and social statistical systems have been based. For example, in analysing Goods and Service Tax receipts (GST is New Zealand’s version of VAT), it is not possible to distinguish GST payments on capital from operation expenses, making GST aggregates a volatile measure of trends in consumption. Despite the high volumes of data that are now available at low cost, there remains a need for this type of data to be tested for statistical integrity, coherence and sample bias. The ability to see interactions from monitoring mobile phone location typifies the potential of digital technologies. Experiences and plans with these need to be shared across countries.

The options of obtaining business data from commercially available web based accounting services offer considerable scope for automating the coverage of a large share of small business, while it is now realistic to consider the option of influencing some of the content to bring some information into line with economic concepts. In applying the same thinking to the conduct of household surveys, statistical offices could consider giving some households mobile phones where they do not exist, to ensure that connectivity constraints do not seriously bias the respondent population mix.

Prior to the large-scale expansion of household surveys from the 1960s, many important measures were obtained by proxy counts. For example, in New Zealand:

The urgency of obtaining information that at least for some period will be completely or partially unavailable does not obviate the need for the statistics that are produced.

4.4The urgency of measures of the impact of economic lockdown

Separating the prospects from recovery after losses from the lack of capacity to trade is important for policy responses after the immediate lockdown period ends. We need to distinguish:

Governments cannot reimburse all businesses. Where government is to invest in sustaining a sector of business through what are essentially capital injections, it will need to have the same confidence that any other investor would have in the return on that investment, the form it will take, and the ability to manage any potential; capital loss. The welfare of employees of businesses that do not meet such tests need to have access to the income transfers that are available, in the current or future form. The capacity of business to provide employment for the labour force has changed to an unknown permanent extent. The effect of business temporary closure on long term employment levels now needs to be assessed. Other consequences of importance to understanding the scale of economic loss nationally will include stock losses, capital loss from closures, and capital that has become redundant.

COVID-19 has placed new demands on health services as well as government economic leadership, social cohesion and community solidarity, and the reach of redistributive programmes. The methodological expertise in statistical offices has become a scarce national resource, and it is important that there is a good understanding of where it needs to focus beyond on making the large array of existing sources able to continue in the face of COVID-19. Many immediately critical new information forms will shape government decisions.

4.5Meeting the challenge through international collaboration

Official statistical institutions around the world will be a pivotal resource for policy makers and service providers as countries work to enable their citizens, businesses, governments, communities and health services to contribute fully again in their country and around the globe. The more countries can share new ways of thinking, innovation and practices the faster this will happen. The UN Department of Economic and Social Affairs Statistics Division [13] through its Governance Lab has created a repository for collaboration with the aim to build a responsible infrastructure for data-driven pandemic response. There is an updated survey of the numerous data collaboratives and partnerships that are happening beyond the official UN system.

Collaboration on the Sustainable Development Goals provides a well-established and intellectually strong focus to give momentum early to this collaboration and connect the diverse regional and professional statistical organisations whose members span many countries. It could also strengthen the ownership within countries where commitment has not extended beyond statistical offices, diplomats and aid agencies.

5.1Indicative areas of policy uncertainty

Economic activity and social welfare in both New Zealand and elsewhere are in the midst of storms whose effects will probably abate at different rates. The need to have confidence in the health, economic and social outcomes of any new policy by governments will challenge the measuring tools of national statistical offices. Redistributive policies have to meet the challenges from the transitional lockdowns and long-term consequences of COVID-19. Looking past the immediate lockdown period, we can expect quite huge and differing consequences for tax and transfer systems. Accumulating knowledge of the consequences of both the COVID-19 and the policies needed to contain it is an important task for official statistics and NSOs. Statistics Norway55 has published an article and documentation of the calculations of the economic loss from the COVID-19 and closedown. It shows that the main driver will be changes in employed persons and loss to unemployment. The reduction in GDP from before COVID-19 to the present is estimated to be 10–15%. Some of the serious changes can be readily foreseen in income effects as well as demand falls and capital losses. These are:

Because the policy context varies across countries, the extent of issues looked at in this section will reflect the political perspectives of the governments of the day, and recent economic history. Those listed in the sections following are similar to those which were under active consideration during the last global financial crisis in New Zealand. Ironically, the need to expedite methodological and technological capabilities may enable statistical offices to meet information needs of climate change through increasing the core infrastructures of the office that can be applied to all areas.

5.2Economic management

It is possible for national income to drop by some 20 percent, which will reduce the weekly income of many households to such an extent that to ensure their wellbeing there will need to be a fundamental shift in the level of transfers and the means of allocating them. This will require a reassessment of the balance between direct and indirect taxes, the progressivity of the income tax scale, and company tax payments. The alternative of raising the level of hardship among households would have consequences for the scale of severe health conditions and potential public unrest. An excessive number of people could be displayed from proper housing, at a time when housing is already a point of stress for many families. Early understanding of the shifts in the level and sources of national income will provide a measure of the scale of the impact to wage earners and investors of the lockdowns and losses from international connections of New Zealand businesses. It will also indicate the potential for recovery, and the scale and duration of the transitional support considered necessary.

Government can ameliorate the effect in the short term of business losses of various sorts. Where business losses are the result of falls in demand that will remain when lockdowns have ceased, governments will need certainty that such support influences job retention and the associated benefits compared to income transfers for unemployment.

The response to COVID-19 opens up concerns about the means by which fiscal pressure will be managed, and how those on different parts of the income distribution will be affected. Given the scale of the fiscal impact, how its costs will be shared across generations will determine the explicit and implicit changes to the nature of the tax base. However, many of the costs of recessions or depressions that fall on particular generations are not financial but result from lost opportunity or permanent harms. It has not been the practice in other crises for governments to seek a return for all to previous income levels when that has become possible. The economic crisis of the 1980s led to a fall in real incomes for a time at all levels of the income distribution. As the economy recovered, those on lower incomes did not recover to the same extent as middle or higher earners. Past reductions on the level of redistribution by governments at the time of crises have only rarely been reversed, resulting now in little leeway to reduce income transfers without having a severe impact on poverty levels.

5.3Economic viability of households

Those households with little discretionary capital will have most likely exhausted it during the lockdown. For some of these people, the lockdown may have placed their health at greater risk than before. We need information on the extent to which the capital accumulated by people will be at risk of being severely reduced or lost. The distribution of basic needs through supermarkets has enabled the supply of food to be managed very well. Being a customer of this supply chain is highly dependent on financial resources and transport. As a consequence, an increasing share of households will lack access to this supply chain through lack of cash. The increased inability of households to pay utility bills will be an important measure of cash shortages, and also the effect of limiting the means of payment to web-based connections. The impact on those with low incomes and more precarious employment of the lockdown means that the range of households who will need income transfers will increase significantly. The means test elements used in the selection processes that determine the level, form and conditions of benefits by MSD generate considerable public distaste. The means tests will need to be simplified and could be replaced by simple income tests that are wholly transparent being of vital importance, now. Benefit applications and approvals need to be available quite expeditiously with the capacity to analyse individuals by ethnicity, place, age, education and former employment.

It will be vital to reduce the rigidities in connecting those now out of work with employment opportunities for all forms of work. In earlier recessions prior to the late 1980s, the government agency responsible for employment had responsibility for matching any individual who wanted work to its list of available jobs. For some three decades, this service has now been restricted to those who are recipients of unemployment benefits, so that the transition from one job to another is broken for many who simply need pointing to options. Information on the contact people have with the benefits services needs to be published weekly. Such information is now essential given the likelihood that the labour force survey will not be able to operate as usual for some time. The hiatus in information can be readily filled by the administrative measures that filled this role before the Labour Force Survey began in 1984. The advantage of administrative records is that statistics can be quickly produced on a weekly basis. There are many examples of systems that enable communities to draw on the information held by the state and others.

In addition to the capability to publish administrative counts from tax and welfare systems, Statistics New Zealand has some information sources that could potentially enable a rapid estimate of the scale of the lockdown on firms and people. New Zealand has a list from the claimants for emergency payments of firms in distress.

5.4Reducing the connectivity of people to health and other services

Connectivity is a vital part of household wellbeing. As it has become more narrowed to technology-based connections, then for those without Facebook, internet access or a mobile phone, many of the measures put in place by governments to ameliorate the consequences of lockdowns will not be available. For these people, access cannot be assumed. Governments need information to assess the forms of connectivity that people in diverse circumstances have, including those who do not have access to those forms of engagement with which basic services and information are delivered. The remedial action taken after the flawed enumeration stage of the 2018 Census of Population in New Zealand could provide valuable information for assessing the levels of connectivity of people to services in New Zealand.

Health services in New Zealand have become one of the few remaining universal elements of New Zealand’s earlier welfare state. Barriers to access exist but unlike other elements of the government social services mix of housing, income support, education and care, there are fewer explicit barriers that deliberately reduce access and cost to government. Furthermore, the myth of universal access to health care has remained. The elimination of COVID-19 makes it vital that there is unqualified accessibility to medical services for all. The diverse nature of the barriers to universal access are only partly countered by the services provided by the community sector. As barriers become more complex, then resolving them will require more intense enquiry into peoples’ lives than governments’ have cared to undertake up to now. The forms to barrier to access that we need to understand include:

The means of access to universal public services is increasingly restricted by transport, reliance on technological pathways, limited regional distribution of service centres or by rationing methods that can be subverted by private means. Part charges, centralised office services and overloaded call centres all build barriers for services that are purportedly universal. The elimination of COVID-19 makes it imperative that the impact of restricted access to health services is understood and countered. Connectivity is dependent also on the economic position of households, which is also a probable influence on the predisposition to COVID-19. Maori and Pacific communities contain a larger share of people who are at risk of being excluded by rationing devices, as are the very old and those with low incomes. Consequently, the elimination of COVID-19 will put a spotlight on the contradictions in public policy unless improving the universality of access to health services has an impact also on the multifaceted means by which rationing takes place for services that affect welfare more generally.

The capacity of primary health services to reach all in the population will require new means of rationing other than by affordability. This will require an integrated approach to balancing the demand for and supply of all health services, many of which require long term investments. It may be an appropriate time for primary health services to become integrated into the existing public health system, and completely rethink the structure and scope of public health services. Improving the capacity of smaller hospitals to operate when staff have had to be placed in quarantine is just one example. This is even more so with aged care institutions.

The almost complete shutdown of health services for activity not related to COVID-19 highlights the need for a contingent capability to be recognised across the health services. The COVID-19 pandemic has put a spotlight on the need to recognise the range of health-related activities that are part of vital national infrastructure, failures which bring disproportionately large consequences to households and other sectors of the economy. This includes population health, the ability to conduct campaigns to prevent disease and to respond to outbreaks of disease including epidemics. The latter includes scientific and modelling capability able to rapidly absorb the latest scientific findings, understand and model their significance for New Zealand and make policy implications clear. There is also a role in communicating these findings to the public. At present carried out largely by public spirited academics, many of whom do not have ‘epidemics’ in their job descriptions. The cost of the lockdown of New Zealand and the graduated return to a new maximum capacity can be used to determine a measure of the scale of future risk to test the affordability of contingent activity.

Community organisations that have well established systems for delivery of support in various forms to individuals and households are already experiencing significant increases in demand at a time when volunteer capability is constrained by the quarantine. Changes in the level of demand for the services of community sector organisations including foodbanks provide a measure of the pressure on those in precarious income situations. Area profiles could integrate information from the Business Directory with population statistics to identify those communities where the exclusions for business in the lockdown and later losses have left whole communities without services.

5.5Inflation measurement

Monitoring consumer prices for some items including food on a weekly basis would provide the public with more confidence in how inflation is changing their food costs, whether it be as a result of temporary supply constraints, seasonality or temporary or permanent structural changes in food distribution. The practice of statistical offices of explaining changes to consumer price indexes would provide the public with a more reliable measure than selective reporting by media outlets of items they chose for their own interest.

Just as globalisation of supply chains has had a huge effect on keeping inflation at low levels in many countries for some two decades, the same supply chains are likely to be significantly disrupted by COVID-19. There may be both short term and permanent rises in inflation rates, the causes of which will need to be understood. The weights of some indices including the Consumers Price Index will need to be reassessed in order to prevent index bias when large elements of expenditure or trade have disappeared.

Given the strong interest in the economic capacity of households, the preparation on a weekly basis of a price index of selected consumer items including basic foods would complement individual experiences of price level change. Measuring the volumes of basic items being sold each week could complement this. In New Zealand, such information should be easy to obtain from the two main supermarket chains and other large food trading organisations. Field staff could do on the spot validation checks for quality assurance.

5.6Rethinking contingency provisions for macro-economic risks

COVID-19 and the prospect of further pandemics in the face of climate change and a predisposition to earthquakes are likely to lead to a rethink of the range and scope of contingency provisions that need to be resourced by government. Three significant areas have already become apparent. Financial contingency, resource contingency and institutional structures.

5.7The differential viability of economic sectors from loss of future demand

Statistics of demand will provide a steer as to the industries immediately affected, but the full effects will depend on the complexity of supply chains. Understanding the indirect impact on the production of other sectors by industries that will now be significantly smaller can be analysed by inter-industry models, which can also be used to build value networks. Producing a current inter-industry study is a mammoth exercise at the best of times. Where the scale of change in industries that will be highly influential on future has been large, it is possible that there are sources of connections between industries that come from data outside the official statistical sources. Tourism is the most obvious sector where it is expected that the reduction in activity will impact directly and indirectly on national income and employment, with variations across regions. We will also want to measure the impact on global warming and pollution on a long-term reduction in international tourism. For countries whose education systems have become dependent on large numbers of overseas students, not only educational institutions but also ancillary services such as accommodation will face major uncertainties about any return to normal.

Changes in employment of different industries will provide further confirmation of the focus of change, but it may be difficult to differentiate the employment losses from the permanent damage done to firms through experiencing the immediate closure in March 2020, from the effects of those who will in the future have faced a permanent fall in demand. These two different reasons will require different transition responses once the immediate full quarantine period is over. While existing official statistical surveys can put a spotlight on the immediate and likely future condition of industry sectors that operated up to the quarantine date, the need for new, more timely and frequent measures of business activity has already been recognised by the UK Office for National Statistics, and the Australian Bureau of Statistics. Their websites already report the results of such surveys. The chart below is a map of the different rates of job loss for sectors of Australian Industry, by the Grattan Institute [2].

It is important to be able to identify from Business directories as well as retail trade and other business surveys where small business have ceased trading, as a one-off urgent analysis.

5.8Climate change, biodiversity and water

The rigidity of systems focused on reporting on the market economy has meant that change has involved a disproportionate cost when extending concepts and definitions that have been long embedded in statistical infrastructures and sources.

The results of international collaboration on matters which shift long established boundaries are often determined by the most well-resourced countries. The Sustainable Development Goals have challenged this dominance and provide a ready-made and relevant basis for collaboration among countries in order for statistical offices to maintain their relevance in a post-COVIOD-19 world.

The OECD [14] notes:

The report outlines a comprehensive range a of policy initiatives relevant to environmental health that they argue need to be part of a COVID-19 policy agenda.

5.9Intergenerational consequences

Every policy has consequences for the size and composition of the population and will be influenced by it. Where the demographic dynamism of the population is high, those consequences will be greatly exacerbated. Demographic impacts have a long lingering life, which makes ignoring them potentially more serious than might appear. In New Zealand, the different demographic structure of Maori and Pacific communities has shaped their history with the State. For Maori this has been occurring for 200 years.

Much of the information we use in policy decision making comes from sources that do not distinguish reliably Maori and Pacifica characteristics, and therefore, Maori and Pacifica individuals appear as outliers when the rules to implement policies are put in place. The demographic vibrance is also coming from these communities, as they are both younger and more fertile. While equity in access to educational and other resources could shift their place in the income distribution in the long run, the type of austerity policies seen in times of fiscal constraint invariably impact much more heavily on these communities.

Fiscal constraint will bring a need to reassess the tax and transfer system, which will have implications for both inter-generation transfers and environmental protection. In many countries there is already a concern about the extent to which the benefits received by the growing share of the population which is above retirement age are funded by taxes on a younger workforce. The tax system provides opportunities to recognise the true impact on the environment of economic activity in business costs. The appetite for this may be greater when for one of the larger sectors of tourism, activity will be severely impaired anyway by international travel restrictions and the resulting damage done to international airlines.

5.10Human rights considerations

The protection of those who are institutionalised in some form of custody, either for their care because of infirmity or imprisoned as a result of being sentenced for some criminal activity is the direct or indirect responsibility of the State. Without independent oversight, the rights of such populations can be treated differently from the general public. The paper earlier referenced Heneghan and Jefferson [10] who note that:

In New Zealand the Solicitor-General [15] advised all Crown Prosecutors to ask judges to consider denying bail if they think a defendant has a possibility of COVID-19. Placing someone into any prison because they may be COVID-19 risk brings risks to all prisoners and prison staff in what are already among the high-risk environments in the country. Such an action would seem to fly in the face of knowledge of managing infections we have had for more than 150 years.

5.11Statistical reporting

It is not only government, but businesses and citizens need to have some way of establishing their future prospects for work, customers and suppliers. That information is now needed more frequently by governments and the public is a matter for the immediate attention of statistical offices. There are now many websites containing global and regional comparisons. The OECD [16] has placed on its website one of the more comprehensive surveys of what is available. Changes in websites are now occurring very fast and the examples cited below will be very much changed by the time readers check themselves for their content. The UK Office for National Statistics, Statistics Canada, Statistics New Zealand and the Australian Bureau are statistical offices that have transformed the content and frequency of selected statistical outputs to meet the needs for more timely reporting in these areas.

International comparisons of the range of testing prevalence of COVID-19 and the outcomes of infection among different population groups, health workers and others have been made much less useful by the varied testing practices, definitions of the relevant populations and errors and revisions. It has been a poor example of statistical collaboration. Case fatality comparisons are flawed because of these definitional differences.