COVID-19 impact on the collection of labour market statistics

Table of Contents

ILO support

Through the process of developing this note, the ILO has been in close contact with countries and fully recognises the many difficult challenges being faced. The ILO is committed to maintaining this close contact and is ready to provide support, including the continued sharing of country practices and the development of additional guidance that will be made available incrementally as soon as available.

Last update: 11 May 2020

The COVID-19 crisis is having a substantial impact on all aspects of our lives. The immediate focus and ongoing priority is inevitably, and correctly, on public health, and is likely to remain that way for the coming weeks and months.

In the world of statistics there is equally a focus on timely information on the spread and impact of the virus. In the first instance, this of course relates to statistics on the number of cases and the outcome of those cases.

However, there is a high degree of interest in the many other impacts of COVID-19, including the many economic and labour market impacts, which have been immediate and very significant, and likely to continue in the near future or potentially beyond. In the case of the labour market many millions of workers across a large number of countries have been directly impacted by lockdowns. Some are able to continue their work through teleworking or remote working arrangements. Many others have seen a reduction or complete loss of their livelihood. Others still, for example workers in health or public security, will experience a different type of change, namely a huge increase in working burden in the face of the crisis.

Trying to track and describe all these changes is a huge challenge for official statistics across the world. The restrictions necessary to combat COVID-19 are creating a huge obstacle to normal data collection approaches and operations, exactly at the moment when there is a massive increase in demand for information. Furthermore, the situation is rapidly evolving, making normal planning impossible.

The ILO has reached out to National Statistical Offices (NSOs) to understand the impacts of COVID-19 on their statistical operations, in particular in the domain of labour statistics, mostly the operation of the national Labour Force Survey (LFS). The purpose of this note is to share the information gathered so that all countries can learn from the experiences of others. Perhaps the summary message is that all countries are facing a major challenge in maintaining continuity and quality, while simultaneously attempting/needing to be flexible and react to changing circumstances. Some patterns and commonalities can be seen in the challenges and reactions across countries and these are summarised in figure 1 and the following pages.

Figure 1: Maintaining the balance – the challenge for NSOs and the LFS

Key impacts and reactions to date

The impacts and responses to date can be summarised under a number of headings. The most obvious impact is on field operations, with the current design/mode being a major determinant of the impact, but many other issues are faced. The issues covered below mainly refer to the operation of the survey and ability to collect and disseminate data. Without question there will also be a substantial impact on the level of estimates which is discussed briefly later. In addition, there is the complex challenge of being responsive (both to the practical challenges and extra demands) while maintaining quality and continuity, this being perhaps the over-riding concern of many NSOs.

Impact on field operations

The large majority of countries have unsurprisingly reported a major impact on their field interviewing operations. The most substantial impact has been on countries using face-to-face interviewing – either Computer Assisted Personal Interviewing (CAPI) or Pen and Paper Interviewing (PAPI). An ILO review of national practices from 2018 found that approximately 80% of the 120 countries that provided information, relied on these methods as their primary data collection mode for the LFS (see figure 2). Only in Europe was it relatively common for non face-to-face interviewing to be the main mode, namely Computer Assisted Telephone Interviewing (CATI), along with a selection of countries in other regions. Computer Assisting Web Interviewing (CAWI) is a growing practice but as part of a multi-mode approach, typically alongside CATI (and CAPI in some cases).

Figure 2: Main mode of LFS data collection by Region

For many of the countries relying on face-to-face interviewing, restrictions on movement have led to a suspension of all face-to-face interviewing activities. For those countries with regular or continuous survey operations, interviewing is often being moved to telephone interviewing (computer assisted or not) or a combination of CATI and Computer Assisted Web Interviewing (CAWI or online self-completion surveys). However, within that change there are many different types and levels of impact, to a large degree depending on the capacity of the statistical office and existing fieldwork organisation, such as:

  • For many countries in Europe, and a selection of countries elsewhere, mixed mode interviewing and multiple visits over time are common, for example with CAPI interviewing for a first visit to the household and CATI or CAWI as the main mode for subsequent rounds/waves. The face-to-face first interview is important to obtain contact details, usually telephone numbers, and ensure co-operation with the survey in later rounds. For these countries, in general, the shift is to move to 100% CATI or CAWI, either by attempting to contact all sampled households by this mode, or restricting the sample only to those due to answer by these modes (e.g. drop the first interview). This, however, creates challenges as good contact information may not exist on sampling lists, creating an additional task of developing a comprehensive register of telephone numbers or emails in the absence of a face to face first interview. Some countries are attempting to deal with this through internet searches, linking of different data sources or different ways of making advance contact with households to capture contact information or promote participation (e.g. introductory letters).
  • Some impact on response rates is expected in many countries, which will depend to some degree on their ability to create and maintain a comprehensive register of contact details if they are switching to CATI/CAWI. Even if the register can be created and maintained, the loss of advance letters and lack of face-to-face first contact is a concern, as it plays a role in introducing the survey, comforting respondents that it’s official (as opposed to the many telemarketing operations which make first contact by phone or email) and thus maintaining high response rates. If response rates are impacted, this can evidently cause volatility in estimates generated and create a difficulty in distinguishing the genuine impact of COVID-19 and increasing survey error due to falling response rates. One country is planning tests of different approaches to handle very low response rates in case this occurs.
  • Countries with existing CATI operations are not exempt from impacts. For example, some countries operating call centres are being forced to shut the centre and decentralise the interviewing to interviewers homes. This can be a major challenge depending on available technology and systems used. A similar challenge can be noted for countries looking to introduce CAWI for the first time, an approach planned by a few countries, some of which were already in the process of CAWI development allowing implementation relatively quickly. For others starting this process a development and testing process will be required before implementation, as well as related IT systems, contact methods and details etc.

In cases where CAPI or PAPI are the only current modes of collection, impacts generally are more substantial. In many of these countries the capacity or resources do not exist to completely switch to another mode of data collection either due to the absence of a register of contact details or the lack of infrastructure needed to undertake telephone or web interviewing. Nonetheless most of the countries in this situation have stated an intention to continue with surveying by telephone and are experiencing varying degrees of success in terms of response rates.

For those using PAPI and now moving to telephone interviewing, electronic data capture applications often do not exist, either requiring them to be quickly developed, or requiring paper questionnaires to continue to be submitted to central office staff for entry and processing. Registers of phone numbers may not exist, or may be very limited, creating a necessary process of collecting that information. Several countries in this position are exploring the use of administrative data to identify contact numbers for the selected households, or retaining minimal field operations to deliver letters to households giving details for the completion of the survey (i.e. contact phone numbers for the NSO or website details if CAWI). Recognising the possible impacts on response rates, several countries are planning a reduction in survey content in order to lower respondent burden and limit non-completion of surveys. These reductions needs to be made carefully to maintain core necessary.

An idea being implemented by some countries is to return to old samples of households, for which contact information is already available. While this would not be ideal under normal circumstances, for example due to attrition, it can offer a plausible solution in the current situation. This can be true both for countries who already have panel designs or those who do not but have regular survey operations. For example, multiple countries are considering to add an additional interview (e.g. a 6th wave) in place of the introduction of new households. If the disruptions related to COVID-19 are relatively short-lived this could be an effective measure to bridge gaps, but will become increasingly challenging the longer the crisis continues.

For a number of countries data collection has been suspended in the middle of the survey, creating varying degrees of difficulty in producing results, depending on the point in the cycle at which the suspension occurred. Some countries are considering options such as resuming the operation later and asking questions retrospectively (e.g. for the originally planned reference period) for the remaining sample, the pooling the microdata for analysis and publication. Other countries are looking at ways to assess then already captured data to determine if it can be used as is or with some adjustments. This challenge of incomplete datasets or low response is a something many countries are considering, potentially requiring an update to estimation methods used to generate results. This may not be straightforward as there may be differences between the sample covered and the sample not covered, potentially requiring additional analysis and adjustment. This can only be ascertained through a review of the available data. In all cases it will be essential to provide accompanying metadata and be clear in dissemination so users can understand any limitations or impact on the interpretation of results.

Evidently for those countries where data collection operations must be entirely postponed or cancelled the impact is most severe. The challenge created will depend on how frequently data collection occurs. For example, for those with regular (e.g. quarterly) data collection the suspension of operations will take the form of a temporal break in series with one or multiple periods of data unavailable or biased. This will impact series based on different periods (i.e. monthly, quarterly, annual) and the challenge will be different depending on the series involved. For example trying to create an annual average with one missing quarter etc will be an additional challenge to dealing with missing responses in a single period. Efforts will subsequently be needed to find ways to bridge any gaps in time-series and many possible approaches could be imagined, such as additional retrospective questions when operations restart, modelling or interpolation of results, making use of auxiliary (e.g. administrative) data as a supplementary source of analysis etc.

For those with irregular periodic surveys (e.g. once every 2 or 5 years) the best case scenario is a delay of some months in the operation. However, there may be a high danger of complete suspension of the round of the survey as reported in some of the countries in Africa in particular, potentially creating a very long gap in the availability of labour market information. Timing of survey operations will be an important consideration for comparability with any previous rounds if the sample is not spread over a longer period of time (e.g. a full year).

While the majority of countries are reporting significant impacts on their field operations, for some there is minimal or no expected impact, particularly those already fully utilising CATI or both CATI and CAWI. In fact, in some limited cases an increase in response rates is being observed, due to the greater ease in making contact with people whose movements are restricted, who may also have additional time available to participate in surveys. In at least one country some respondents have expressed a greater willingness to participate due to an increased awareness of the need for information on the impacts of COVID-19 on their lives. For these countries the greater impact will be on support activities (management, supervision etc) and the results themselves.

Support activities and other fieldwork related impacts

In many cases, in addition to field operations, the operations of NSOs are being suspended or heavily impacted. For most of the NSOs that provided information, teleworking arrangements are being imposed for some or all staff.  Depending on the country in question, this can make the continuation of any survey operations impossible, and this is relevant for both developed and developing countries. For example, if a NSO does not have the technology to allow people to remotely manage field operations while at home, process collected data or access and analyse survey microdata, the continuation of interviewing or other statistical operations would become impossible.

An additional impact on field operations mentioned by countries is the training and support given to interviewers. Interviewers are often being asked to change their role, moving from face-to-face interviewing to telephone interviewing (computer assisted or not), requiring different skills and techniques to gain co-operation with surveys. Furthermore countries are having to provide additional guidance to interviewers on how to handle cases arising which were not envisaged when questionnaires or guidance was designed, for example many people being supported by Government supports due to loss or reduction of employment. NSOs are also being required to put in place contingency plans in case of the unavailability of interviewers or their inability to complete the work, e.g. to lack of availability of necessary technology or systems. Furthermore, NSOs are increasingly feeling the direct impacts of COVID-19 through illness to interviewers or other key staff (or indeed respondents), which inevitably has consequences for continuity of operations.

Attempting to summarise the above, it is clear that the impact on operations is heavily related to current mode of data collection and current systems and capacities of NSOs. These things vary quite substantially across countries meaning the impact will also vary. Those countries with pre-existing fully CATI and CAWI based systems may experience relatively little impact on their operation or survey response rates. For all others there will be varying degree of impacts and challenge in retaining response rates at desirable levels. In the most extreme cases survey operations will be postponed completely for an undetermined period. For all countries the disruption is important in some way or another, creating pressures such as the need to review survey content, the need to create additional systems and processes, the need for training and support for changed operations, the need for IT development and support, the need to review weighting and estimation approaches etc.

Impact on data analysis

Evidently any impact on response rates or ability to undertake interviews can have a major impact on the use of the data. This may necessitate changes to the process of imputation and estimation used to generate estimates. While many countries are clearly concerned about that challenge and achieving some continuity, several countries also expressed an intention to supplement existing analysis or change focus. In particular, an additional focus will be placed on information that can shed light on the COVID-19 impact on jobs/businesses and labour market engagement. There are several dimension through which this impact could be shown, or variables which are of even higher priority than before, including:

  • Temporary absences from employment, potentially of unknown duration, with or without pay. In some cases continued pay will be from the Government rather than the employer and may either be made through the employer or directly to the worker. Some information may already be captured on a typical LFS on reason for absence, duration and continued receipt of income, all of which can become of high interest. However, even greater value could potentially be generated through limited updates to questionnaires if possible, as being attempted by some NSOs (discussed further below).
  • Changes in working time, including reduced work hours and excessive work hours, and reasons for working non-typical hours.
  • Working arrangements, particularly working from home, other changes in location of work.
  • Degree of attachment to the labour market. For example looking at those who are not identified as employed and may be seeking but not available for work, or available for work but not seeking. Both these groups are included in the potential labour force as defined in resolution I of the 19th International Conference of Labour Statisticians ICLS). For these groups, it will be particularly important to analyse the reasons for not seeking or not being available, particularly reasons such as: waiting to be recalled, discouragement (no jobs in the area/for my skills), own illness, increased family responsibilities, government curfew/shutdown
  • Recent job loss, reasons for recent job loss, or changes in job or type of work done
  • Informality – given that this identifies cases generally not benefitting from social protection of different types this becomes of very high relevance as Government schemes are less likely to be able to target these workers.
  • Changes in income from employment, including reduced pay, owed income, etc.

Some questions on the above elements will likely exist on all LFS questionnaires. NSOs will need to assess what useful information is already collected on their LFS, and highlight this data alongside headline statistics such as employment and unemployment and/or in additional publications1See for example : https://www.abs.gov.au/ausstats/abs@.nsf/Latestproducts/6291.0.55.001Main%20Features3Feb%202020?opendocument&tabname=Summary&prodno=6291.0.55.001&issue=Feb%202020&num=&view=. For example, measures of labour underutilization to supplement the unemployment rate (as defined in 19th ICLS Resolution I) could be particularly relevant in highlighting groups of people who are without employment but not seeking because the availability of paid employment or their ability to search for work is completed restricted (potential labour force), or those required to work shorter hours but still wishing to work more hours (time-related underemployed).  The size of the groups involved are likely to increase very substantially in many countries, perhaps to an even greater degree than the number unemployed.

In addition flows analysis, possible when households are interviewed multiple times, can be particularly powerful in tracking impact on individuals and the rate of movement in and out of the market (churn) now seen. For many this will be a loss of employment. However, for some sectors increases in demand are leading to an increase in vacancies (e.g. online retailers, supermarkets). The pace of job churn will have increased very rapidly and add a very ‘temporal’ element to the analysis, namely answering the question ‘what has changed recently’ which will create a distinction between long-standing situations and those which have changed recently.

These new analytical approaches can be tried out with existing data from periods before the COVID-19 outbreak, allowing the impact to be assessed relatively quickly when updated information becomes available.

Changing questionnaire content

A minority of countries have made changes to their questionnaire content but several others are considering this possibility. As with many issues there are conflicting pressures including the important need to maintain consistency in current measurement approaches (i.e. not changing the wording or flow of core questions used to derive employment etc.). Several countries are actively assessing a reduction in questionnaire content to maintain core information and indicators while reducing burden in the hope of maintaining response rates.  However others are seeking to add questions to allow an assessment of COVID-19 impacts, such as
  • Some are adding questions very directly targeted to COVID-19 related issues, for example, asking people absent from work if the reason for their absence is directly related to COVID-19. This would be particularly relevant for some reasons for absence, for example: own illness, family responsibilities, mandatory leave, temporary or permanent layoff, government shutdown, lack of inputs, resources, clients, strike or lockout.
  • In other cases the questions may not very directly ask about COVID-19 but could be designed to yield additional information not previously collected, for example questions about teleworking arrangements, reasons for differences between usual and actual hours.
  • Other questions could be very temporal in nature – in other words asking how the employment situation or working arrangements have changed since a certain date. This is a supplement to the type of stock analysis typical for LFS results, and the type of flow analysis already mentioned which is only available in countries with longitudinal panel designs and which requires a different analytical approach.
  • In other cases it may only be necessary to add response categories to an existing question if it is considered likely that the question would lead to COVID-19 relevant information. For example questions on reasons for temporary absence from employment are almost universal in the LFS and additional response categories may be useful. Alternatively a common practice can be the inclusion of a ‘Other, specify’ response categories in cases like reasons for absence. This could be used as an opportunity to capture relevant information but needs to be done carefully and consistently to ensure clear interpretation of the results. For example, if a category already exists for temporary lay-off clear instructions would be needed to interviewers when to use which category.
All countries should consider which additional information could be useful to capture. If this is considered to add unmanageable burden, NSOs should consider if some existing topics are of lower priority, and thus can be temporarily removed to create space for the new questions. A careful balance will need to be maintained between the range of issues covered and respondent burden. Different approaches could be imagined to dealing with this, such as using some questions for subsets of the full sample (generally only possible for computer assisted methods). In all cases where changes are made, care will be needed to avoid unintended impacts on the core elements of the survey, such as the classification of the respondents labour force status.

The issue of adding questions to the LFS highlights a limitation or feature of large scale survey operations running on a continuous or regular cycle. The introduction of changes in methodology or questionnaire content is a major undertaking, which has to be done alongside the regular survey operation. Many countries have a development cycle of many months for the introduction of new topics or questions, with good reason giving the need to ensure this is done correctly. However, in a time when the labour market is changing so rapidly this can make the LFS unable to provide the type of rapid information demanded. Modular approaches to survey content can be helpful as they promote some flexibility in survey content, adding and dropping some elements as needed, but this still requires a development, testing and implementation process which takes some time. All of these related challenges (content, mode, new analytical focus) create a delicate balance of issues to be resolved and it is clear that all countries have serious concerns about the maintenance of quality and continuity of data, while at the same time being flexible enough to make necessary or desirable changes. This balance is discussed further later. A further point to note, both for changes in analytical approach and the addition of new variables, is the issue of sample size of the survey. This is always a necessary consideration in deciding the appropriate use of data from any household survey. As valuable as some information may seem, if it relates to something very rare in the population, it will be unlikely that robust estimates will be generated from any household survey, even the LFS which typically has a larger sample than many other household surveys. This consideration needs to be borne in mind when engaging in new analysis or considering the addition of new questions/survey content.

Use of administrative data

For those countries with access to good administrative data (for example registered unemployment, taxation records covering employment etc), the intention to make extended use of this data was referenced by several countries. Even if the use was unchanged it was noted that the administrative data is likely to provide a good leading indicator of reactions in the labour market and thus receive even higher levels of attention given its high frequency and comprehensive coverage of the populations involved.

The usual restrictions with administrative data need to be borne in mind, mainly the degree of correspondence between the administrative data and relevant concepts (e.g. are the criteria for registration as unemployed consistent with those envisaged in the labour standards). Metadata presented with statistics generated from administrative data should explain clearly the conditions and rules relating to the administrative data source and any implications of this for analysis and interpretation, including comparability with data from the LFS. The same could be said of big data sources, although there are few obvious big data sources for labour market related issues.

There will be a particular interest in tracking the take up of any new Government schemes or services being activated, which may be entirely separate to existing schemes and services. Contact should be maintained between NSOs and Government Ministries responsible for any new schemes to ensure data can be made available regularly for analysis and dissemination.

In more advanced cases, administrative data is already used to supplement or populate some of the information on the LFS. This practice could potentially be extended but many countries will lack the administrative data required to achieve this. Nonetheless, it is very powerful where possible.

However, one country did note a possible COVID-19 related impact on availability of data from administrative sources. If workers in Government ministries are required to telework, or cannot continue working, there can be restrictions on access to timely data from administrative sources. Furthermore, if the Government services involved are disrupted the ability of people to register may additionally be impacted, meaning that administrative data sources may not be exempt from practical impacts.

Other issues

Countries reference several other relevant issues including:

  • There are impacts on other surveys which can be a source of labour market statistics including the Census of Population, Economic Census or other types of household surveys which capture labour market information (household income and expenditure surveys, integrated household surveys etc). For example, planned Census of Population field operations have been postponed in some countries who were due to start their activities as of April/May 2020, while a limited number have had to suspend their operation in the middle of the field phase. For other countries preparatory activities for the Census have been disrupted with potential knock-on effects on the implementation phase. Given the scale and importance of the Census operation this is clearly a major concern for the countries involved.
  • Some countries have taken the step of indicating a clear prioritisation of their survey activities and providing clear published information on ongoing survey activities. The prioritisation generally indicates the LFS as a high priority activity with the significant that it would have higher priority on available resources. This approach has been mostly adopted in developed settings but that type of prioritisation and planning is useful for all. The provision of clear published information on ongoing activities can be useful to keep the public informed and be a reference for interviewers to show that the survey is indeed official when attempt to obtain co-operation to surveys.
  • A few countries referenced the idea of undertaking a parallel survey with different approaches mentioned. At least one country, where field operations have been suspended, is planning a parallel survey to the regular LFS when operations recommence. This additional survey will have a retrospective focus covering the originally selected sample and period of lost data, allowing the LFS to continue without major change. The intention is to create data to bridge the gap created by the suspension in operations. At least one country was already developing a parallel CAWI only survey with a more limited set of core LFS type questions. The idea would be to gather supplementary data for separate reporting from the LFS and has not been planned specifically as a reaction to COVID-19 impacts. While such an activity can certainly yield interesting supplementary information it will not be a replacement for the LFS or be likely to yield directly comparable information due to differences in population coverage, survey mode and content etc. These types of options highlight that NSOs are attempting to flexibly respond to the challenges they are facing in order to fulfil their mandate.

Maintaining quality and continuity of data

An underlying theme of many of the responses received from countries is the very delicate balance between the disruptions now being created, the flexibility being required/desired and the continuity and quality of current data.

In trying to achieve this balance sight cannot be lost of the major challenge for many NSOs of maintaining the continuity and quality of data. Continuity of statistics on employment, hours worked, unemployment, potential labour force, and many related labour indicators in the long-term is a very important objective and should be a priority. It will be a major achievement by any NSO to have accomplished this objective, either through the avoidance or minimisation of breaks in time series. Being able to supplement this with additional analyses of existing data can also be an important contribution which does not create an additional burden or impact on existing processes.

With this is mind it is perhaps unsurprising that the primary focus of most countries is continuity, maintenance of response, clarifications of treatment of individual cases and supplementary analysis, either using data already on the LFS or secondary data sources.

While the option to update questionnaires or doing parallel data collection could be very valuable, they should not be taken lightly, nor their potential impact on existing series or data be underestimated when these approaches are being planned and tested. The ILO is engaging in work to develop some additional survey content for the LFS to supplement existing data, reflecting on some of the changes already being made or planned by countries. This will include among other things some retrospective questions to capture the impact in hindsight, once survey operations can be continued, in case they were disrupted. This will be published in due course.

Impact on estimates

There is no question that the impact on labour market estimates will be substantial. Decreases in employment and increases in labour underutilization seem inevitable. However, they are not all predictable. For example, it is unclear if unemployment will increase, as people do not actively seek work given a restricted ability to do so and a collapse in labour demand. This requires us to widen our focus beyond the very limited number of key indicators, which are often a primary focus of labour market analysis. These indicators, such as employment and unemployment, will remain critical, but will be insufficient to fully describe the impact of COVID-19 on the labour market, workers and their households. This is not a failing of the concepts of employment and unemployment, rather a re-emphasis of something evident, the labour market is too complicated and diverse, as well as changing too rapidly, to be summarised into two statistics. A new, wider focus can be built into existing labour market statistics publications or additional publications to ensure wide dissemination and as comprehensive an understanding as possible of the impacts of COVID-19 on this aspect of life and the economy.

Notwithstanding this, there is a major communications challenge faced by all countries in the coming months to translate the various impacts on the labour market into meaningful publications, while attempting to be clear about data quality and any limitations on interpretation which may arise from the many disruptions now being faced.

Summary

As is evident from the above, the impacts of COVID-19 on labour force surveys and labour market statistics are very broad, being realised very rapidly and continually evolving. This creates a major challenge for NSOs and they are using various means to manage this impact in order to continue to providing accurate and timely information on the labour market. Nonetheless, the challenge created by COVID-19 is unprecedented and is likely to lead to a variety of impacts on the availability and quality of labour market statistics, not all of which can yet be anticipated.

The full impacts of COVID-19 on labour market statistics operations and time series will only be known over a longer period of time. Maintaining a flow of data will be an important achievement, with any supplementation to existing data series a very valuable contribution. Given that the context differs so heavily across countries there will be no single right solution to achieve these things.

One message that can be taken from the responses of NSOs to date is the need for flexibility of approach in various ways, analytically, operationally etc. Approaches will most likely have to be adapted over time given the fluidity of the situation. Some approaches, for example the use of CATI and CAWI, modular questionnaire design approaches, use of administrative data, will likely be better able to react to the disruption created by COVID-19. While countries cannot instantaneously alter their operations, creating greater flexibility for the future is a valuable, perhaps longer term, objective.

Annex: Summary of country responses

NOTE: The table below reflects the information provided as of 5th June. Only data for countries which has been collected information are included (i.e. countries for which no information is available are not listed).

Given the rapid evolution of the COVID-19 crisis countries responses are also changing rapidly. As such the table below is likely to be out of date for some countries and will not be exhaustive as different countries provided different levels of detail. Furthermore, countries will undoubtedly adapt their approach over time.

The impact on fieldwork is highlighted in particular in the table below because this is the most significant impact being reported by most countries and the one for which most comprehensive data has been reported.

The different impacts and responses have been summarised in the main note to which this table has been attached.

In the case of European countries the primary source of information has been data collected by Eurostat through a survey of EU/EEA/EFTA countries. The other information has been captured by ILO through direct contact with countries.

COUNTRYMode or fieldwork ImpactOther impacts/changes
Africa 
South AfricaField operations stopped 20th March impacting ongoing LFS. Next steps unclear.None indicated
MauritiusFace to face interviewing stopped. CATI will continue for repeat interviews (households are interviewed 4 times over 16 months). Decision awaited on new households which would have been interviewed by CAPI.None indicated
KenyaLFS is ongoing with face to face interviewing. Change to smaller groups of interviewers to lower levels of exposure.None indicated
NamibiaLFS cancelled due to Census of Population which is due in 2020. This decision was taken prior to the COVID-19 crisis.Impact on mapping exercise for Census but no decision yet on alteration to plans.
SeychellesFieldwork for Q12020 LFS (including STWS) cancelled in March (normally 3rd week of the month). Plans for Q22020 LFS unsureNone indicated
Burkina FasoCensus and SWTS already completed. Some knock on effects on analysis and processing but minimal impact.None indicated
Cabo VerdeLFS finished in December 2019. Census planned for 2020 and will be delayed, new date not confirmedNone indicated
Cote d’IvoireLFS was completed in 2019. Publication of results may be delayed due to impact on operations of the NSO.

Census planned for April/May 2020 – may be delayed, preparatory work already complete.

Work to collect data from administrative data sources is being suspended.

GambiaOngoing Integrated Household Survey (includes a Labour module) suspendedNone indicated
GhanaCensus planned for 2020 – delays are expected.None indicated
LiberiaCensus 2020 has been suspendedNone indicated
NigeriaQuarterly LFS has been delayed.Socioeconomic survey was completed in December 2019 – data being processed – no impacts yet.
SenegalUsing telephone interviewing in place of face to face.None indicated
AlgeriaLFS for April has been suspended. Impact on September collection is unclear so far. Possible modification of the questionnaire (i.e. more questions on working time)Changing the mode of data collection may need time for development, testing and additional funding.
BurundiLFS activities continued as there is no lockdown at this stageNone indicated
CameroonLFS was planned for May 2020 – now suspendedNone indicated
MoroccoLFS field activities stopped. Alternative modes being studied to replace direct interviews with households and businesses.None indicated
MozambiqueAll field surveys stopped. Possible delay of Census mapping preparation and planning the Household Budget Survey.None indicated
EgyptAll field collection activities suspended. Alternatives being considered. Commencing telephone interviewing.Adding some questions on COVID-19 impact.
TunisiaLFS field operation stopped from 16th March, which result in partial executed sample. Only CPI data collection continued beyond that date.Hoping to capture the current labour market situation using retrospective questions in the next round hoping for a post-estimation of reduction in employment or increase in labour underutilization.
EthiopaLFS has been planned for April/May 2020 – operations impacted. Next steps unclear.None indicated
ZambiaPlans were being developed for LFS in 2020 but plans now unclear.None indicated
NamibiaNo LFS planned for 2020.None indicated
MauritiusMoved to telephone interviews using smaller sample.None indicated
AngolaSuspended quarterly survey for Q2 2020.None indicated
ZimbabweReleased LFS results on 17 March before lockdown. Stopped all field operations after the lockdown was announced.None indicated
Asia and the Pacific
South KoreaShifting away from face to face interviewing – offering email, telephone, and web-based interviewing.  Face to face interview has been completely prohibited for the most badly affected regions, Deagu city and near provinces. For other regions, it is only allowed if necessary.No delay of sample surveys, but one major field operation (Census on Establishments) has been postponed. Intends to make increased used of administrative and big data (e.g. mobile phone data for population migration trends). 
ChinaMove from CAPI to CATI.None indicated
FijiLFS postponed to begin in August 2021 (was due to start in July 2020 with fieldwork for 12 months)None indicated
MaldivesHIES fieldwork suspended and unknown date of recommencement.None indicated
Sri LankaLFS temporarily suspendedNone indicated
MalaysiaTemporary postponement of LFS field interviews, moving to telephone and web interviewing (to be developed). Considering reference period to apply (currently one reference week for employment etc).Using administrative data to supplement LFS
PakistanLFS planned to commence in July 2020 – no alteration to plans yet.None indicated
BangladeshLFS planned to start in July 2020, no alteration to plans yetRecruitment cost survey completed before COVID-19 crisis.
IndiaField operations suspended until End March – then to be reviewed.None indicated
NepalLSS has been suspended (includes labour module) – temporary for the moment but likely to be extendedCensus pilot suspended
ThailandSuspension of monthly data collection by CAPI, and move to quarterly as from Q2. Plans to recommence as soon as the lockdown is over, and cover full quarterly sample in the remaining period of Q2. No decision on proposed mode yet.None indicated
MongoliaContinuing as normal in rural areas. In the capital city dropping off and collecting forms.Has added questions on COVID-19 impact on income
PhilippinesApproximately 3 week delay in publication of data. Now using mixed mode (CAPI/CATI/CAWI – previously all CAPI face-to-face).Introduced additional questions on telecommuting, absences and short-working hours due to COVID-19
IranPlanning to move from CAPI to telephone interviewing.None indicated
IndonesiaNo changes to operations reported yet; February data collection completed as normal.None indicated
VietnamNo changes to operations reported yetNone indicated
JapanNow allowing online response as well as postal response in some regions.None indicated
SingaporeFace-to-face interviewing has been temporarily suspended.  Interviews are now conducted via telephone and via internet.Questions on responses of companies/employers to COVID-19 have been added to regular surveys.
New Zealand

CAPI suspended – continuing with CATI. Exploring means to initiate contact with ‘First time in’ respondents, in the absence of CAPI.

Contact Centre staff redeployed to support Government’s wider COVID-19 pandemic response. Survey Interviewers operating as virtual Contact Centre.

Included Wave 9 for the June 2020 quarter (i.e. Wave 8 from the March 2020 quarter), so as to help ensure the sample size is fit for purpose

Planning to add a short supplementary questionnaire to the June 2020 quarter HLFS on different impacts of the crisis (e.g. wellbeing, income, and poverty-related data).

AustraliaOnly around 10% of response was CAPI before COVID-19. CAPI was suspended at the end of the March survey, with strategies put in place to further increase CATI and CAWI (both already in use) in future months. CATI has always been delivered using a decentralised workforce.

Producing additional hours analysis (focusing on categories of hours of work and reasons for working fewer hours than normal), using existing data, on a monthly basis.

Begun releasing weekly administrative data on paid jobs and wages, on a fortnightly basis.

Introduced a new business survey and new household survey to measure COVID-19 impacts as a supplement to existing surveys.

Europe and Central Asia
DenmarkInterviewing already done by CATI/CAWI – some disruption to CATI operations but not majorGreater use of administrative data – e.g. daily data on newly registered unemployed.
Sweden100% CATI. However some disruption due to temporary inability of some interviewers to work from home.Possible increased analytical focus on data on absences. LFS nominated as one of the priority activities of the NSO meaning resources may be reallocated from other activities if needed to ensure it continues.
FinlandNo further CAPI interviewing (already small so no major impact). Plans for introduction of CAWI from 2021.Some tests being done in case of very low response rates and how to handle it. Will publish additional information such as home working, numbers laid off, underemployment. LFS identified as priority meaning resources may be reallocated to it if need.
GermanyCAPI cancelled. Some Lander (regions) have suspended activities completely meaning no LFS data collected. Plans for future collection undeterminedNone indicated
NorwayNo impact so far – all interviews completed by CATI and interviewers are able to do interviews from home.Have produced an information note on how COVID-19 is likely to impact unemployment and lay offs data in LFS.
FranceSwitched to CATI for all waves (some issues with incompleteness of phone number registers). Was previously using CAPI for 1st and last interview Also lack of field listing activities has impact on sample (inability to identify vacant households etc). Anticipating an impact on response rates.Additional information being issued to interviewers and additional training provided on how to convince respondents to answer, how to deal with certain types of absence due to COVID-19 etc.
LuxembourgNo impact to date, already using CATI/CAWINone indicated
SwitzerlandNo impact to date, already using CATINone indicated
CyprusMoving from CAPI to CATI (from interviewers homes – same interviewers)None indicated
LatviaMoving CAPI interviewers to CATI (first interviews were CAPI previously) – challenge to get phone numbersNone indicated
IcelandImpact undetermined so farNone indicated
PortugalMoving CAPI interviews (first interview plus some for other waves) to CATI. Concerns about response rates due to lack of contact information for telephone interviews. Efforts being made to update sample of newly introduced households to ensure contact information available using matching to other sources.Some additional questions being planned for introduction in Q2 2020 covering work from home, use of technology and COVID-19 impact.
BelgiumFirst interview was CAPI (plus some of other waves) – moving to CATI if phone numbers found. Most interviews done by CATI/CAWI alreadyNone indicated
TurkeyMoving CAPI interviewing to CATI from regional centres. Response rate concerns for first wave.None indicated
MaltaMoving interviews from CAPI to CATI. CATI already used for follow up interviews so main concern is contact information and maintaining responseNone indicated
EstoniaChanged CAPI interviews to CATI (same interviewers).Some impact on response rates due to lack of contact details but not major so far.None indicated
CroatiaUsing CAPI and CATI up to now. Now CAPI suspended entirely with interviews moved to CATI. Also some impact on CATI operations as fewer interviewers working in the call centre.None indicated
PolandMoved CAPI interviews to CATI (at interviewer’s home). Additional information included on letters to new respondents to get contact information. Proposed to continue with CATI only for Q22020 and some updates to sample being planned to attempt to maintain response rates.Adding some questions to assess COVID-19 impact.
BulgariaWas using PAPI, attempting to collect information by telephone (same interviewers) – low response rates so farNone indicated
SlovakiaMoving to CATI from CAPI/PAPI. Also some loss of response due to absent interviewers.Planning a 6th wave (used to be 5 waves) instead of a new wave 1
RomaniaNo field interviewing taking place. All interviews will be conducted by phone or self-completed onlineNone indicated
IrelandSwitched first interview to CATI (other interviews were already CATI) – no other changes yet.Will consider changes in questionnaire content and change analytical approach.
CzechiaCAPI suspended. Wave 1 lost but attempts being made to make contact by mail to get contact information for CATI.Intentions to introduce additional questions in the questionnaire to discuss the recent situation, endangered jobs as well as impact of Government rescue plan.
ItalyCAPI for first interview suspended. Also CATI company stopped operating. CAPI interviewers will do the interviews of all waves by phone (when telephone numbers are available) using the CAPI questionnaire.Considering allowing longer recall period/data collection period – 5 weeks from the reference week.
MoldovaCurrent PAPI interviewing is being switched to telephone still using pen and paper forms.None indicated
AlbaniaFace-to-face suspended. CATI continuing for wave 2 to 5. Was using CAPI for first interviews and to collect phone numbers, so unsure how to deal with first contact for future quarters.None indicated
AustriaWas using mixed mode CAPI/CATI/CAWI. All CAPI now moved to CATI from interviewers homes, with existing call centre and CAWI continuing as before. Some impact on response rates being experienced following the Government lockdown.Adding some questions on teleworking to Q2 2020 questionnaire
HungaryWas using mainly CAPI – moved to telephone interviewing from interviewers homes. Unable to complete first wave interviews due to prohibition on face to face interviewing, now using letters to try to collect contact information for telephone interviewing. Impact on response rates expected.Planning an independent short telephone survey from April to cover additional information about recent labour market changes.
SpainSuspension of CAPI, moved fully to CATI with a small element of CAWI but concerns for response due to lack of introduction letter or first CAPI visitIncreased analytical focus on reasons for absence (layoff) and other elements.
UkraineSuspension of face-to-face interviewing. Doing telephone interviews with previously interviewed households with contact information available. Will be reviewed for April data collection.None indicated
United KingdomCAPI suspended. Fully CATI. Could be an impact on response rates for wave 1 data (first interview). Interviews were temporarily suspended to allow IT updates but now up and running again.Additional questions being added about absences and changes in work hours and whether changes are COVID-19 related. Also launching a web only survey in parallel to normal LFS on ‘core’ labour market issues for supplementary information.
Americas
MexicoFieldwork suspended. Attempting to move to telephone interviewing and considering alternative sampling approaches.None indicated – currently in the middle of Census of Population so impact could be major for Census operations also.
ParaguayField interviewing postponed. Exploring potential to use telephone (without CATI system in place) to reach selected households (due to panel design 50% of the sample was previously interviewed, so contact information available – for 50% new sample assessing options to get contact information)Planning to reduce questionnaire content. Considering creating a specific panel for information before / after break in operations.
UruguayTemporary postponement of field interviewing, now using CATI (high contact levels so far)Considering implementing a panel design
EcuadorExpected to stop field interviewing and evaluating possibility to reach selected households via telephone (no CATI system in place). Using administrative records to identify contact information. Concerns with uneven telephone coverage in the populationConsidering reducing questionnaire content.
Costa RicaCAPI interviewing suspended, moving to telephone interviewing (23rd March).Census mapping exercise still in the field.
ArgentinaAll field operation suspended from 26th March. Evaluating alternative modes of data collectionNone indicated
ChileField operations suspended. Moving to telephone interviewing (no CATI system in place), no phone contacts for selected households, planning to send/deliver NSO contact information to households requesting to call NSO to complete interview or use CAWI option (launched for testing in January). Also exploring use of administrative records to generate telephone contacts for selected householdsMay reduce questionnaire content to improve response.
BrazilSuspended field operations. Proposing to introduce alternative modes but no details yetNone indicated
PeruField operations suspended outside capital city. Moved fully to telephone interviewing to maintain the survey in the capital.None indicated
Dominican RepublicMoved interviewing to telephone using a fixed panel.None indicated
MontserratField operations suspendedNone indicated
St. LuciaMoved from PAPI to telephone interviewing (paper form). 80% response by phone so far.Census has been postponed (due
DominicaLFS planned during 2020 but planning on holdNone indicated
GrenadaSwitched to telephone interviewing to finish fieldwork for Q4 2019 – significant impact on response rates so far. Unsure of plans for upcoming quartersNone indicated
AntiguaLFS planned for Q3 2020 – no impact on plans yet but will be kept under reviewNone indicated
St VincentLFS finished before the pandemic. Next LFS likely to be 2022Census planned for 2021. No impact known yet.
BermudaAttempting to move to telephone interviewing. Considering reusing Q42019 sample for which contact details are availableNone indicated
ColombiaEvaluating a postponement of field interviews. Considering telephone interviewing and planning tests.May also use reduced questionnaire with telephone interviewing.
BelizeLFS planned for September 2020, planning to complete this using telephone interviewing (was originally planned as face to face).Census 2020 has been postponed to 2021.
CanadaAll surveys to be done by CATI/CAWI, now remotely rather than from a central location. CAPI interviewers now doing telephone interviewing from home.Launching a new ‘Disaster – Catastrophe Module’ in March to gather additional relevant information. Also developing a dashboard or social impacts of COVID-19. LFS is identified as one of the ‘critical’ programs which will continue in case resource prioritisation is needed.
United StatesChanging to all CATI/CAWI or email. CATI from interviewers home as call centres closed.ATUS suspended. Additional instructions to LFS interviewers on treatment of absences due to COVID-19 etc. Reviewing some updates to estimation if response rates fall. 4 new questions to be added to CPS on effects of COVID-19 on labour force activity. Adding information to the National Longitudinal data in September 2020.
Arab states
JordanForced to halt LFS approximately 50% through field collection. No indication yet of plans to cover remaining sample and consideration being given on how to use data already collected.None indicated
IraqLFS postponed – proposed to commence early JuneNone indicated
LebanonLFS not planned for 2020MICS suspended until at least June (child labour module included).
United Arab EmiratesCurrently processing LFS 2019. LFS 2020 proposed for October 2020 so no impact yet.Has administrative data which may be used for analytical purposes.
Saudi ArabiaFirst quarter data collection completed. No confirmed plans for Q2 data collectionNone indicated
OmanNo LFS or household surveys planned for 2020. Has CATI capacity already and will use for future exercisesNone indicated
YemenNo LFS planned in 2020None indicated
Syrian Arab RepublicNo LFS planned in 2020None indicated

 

Frequently asked questions

Both countries with regular Labour Force Survey (LFS) operations, and those with periodic surveys (e.g. every 2 or 5 years) are facing disruptions of different types. For NSOs with lower capacity, running periodic surveys, the surveys have generally been entirely suspended with an unknown date for the resumption of field operations.

The most substantial impact is being faced by countries using face-to-face interviewing – either Computer Assisted Personal Interviewing (CAPI) or Pen and Paper Interviewing (PAPI).

READ MORE

Evidently any impact on response rates or ability to undertake interviews can have a major impact on the use of the data. This may necessitate changes to the process of imputation and estimation used to generate estimates.

Moreover, breaks in series are likely in many contexts for many inter-related reasons. 

READ MORE

The ILO is committed to continue its support, particularly to those with lower levels of capacity to attempt to navigate these difficult times and fulfil the mandate given to us and all NSOs by our stakeholders and users of our data.

The ILO has started the process of identifying possible new content for LFS and developing guidance for countries on potential responses to minimise the impact of disruptions related to COVID-19, both operationally and on data. Some flexibility is needed in approaches and some longer term lessons can be learned about the conditions which can lessen the impact of disruptions on operations, for example the mode of data collection, or the approaches to questionnaire design to allow content to be added or removed with some flexibility. Among other things, some thought is needed on the possibility to retrospectively fill data gaps disrupted by a break in interviewing and some countries have already started to plan parallel surveys to the LFS with retrospective questions for this purpose. The ILO will be considering such approaches in its content and guidance development activities and will release this guidance in due course.