Measuring the Impacts of COVID-19
Thursday, 30 April, 2020
Briefing to the Australian Business Economists
I also thank Stephen Walters, Chair of the Executive Committee of the Australian Business Economists (ABE), for inviting me to speak to you today.
When I first received this invitation, I’d planned to talk about innovations in labour statistics, including the ABS Labour Account, and other key labour market developments.
But like so much else, this was turned on its head by COVID-19. Since the arrival of this virus, it has been hard to think or talk about anything else.
So today I will be focusing on how the ABS is adapting and responding to meet the challenge of measuring the impacts of this pandemic. I think I can confidently say that we all suddenly find ourselves in a rapidly evolving environment with the inevitable uncertainty that brings.
It’s certainly not what I was expecting when I took the helm at the ABS only four months ago.
For those of us seeking to support the economic policy debate, particularly the ABE with its mission to inform, connect and influence, responding to the exigencies of COVID-19 manifests in different ways.
Today I’m going to discuss:
I want to start by taking you behind the scenes at the ABS, into our statistical war room, if you like.
On Friday 28 February a group of senior ABS staff gathered for a brainstorming session on how we could best provide new, and more up-to-date, statistics to inform governments and the community about the impacts of the spread of COVID-19.
Looking back, that day was the calm before the storm. Events overseas, particularly the early signs of explosive growth in confirmed cases beyond Wuhan, in Italy and Iran, as well as the PM declaring the previous day that the world would soon be entering a pandemic, were making it clear that the COVID-19 storm was coming and, when it arrived, it would likely generate enormous disruption to our way of life and our economy.
While the ABS is widely acclaimed for our gold standard, high-quality, accurate surveys, we knew there was more we could offer. We were in a unique position to provide enormous value by collecting and publishing near real-time information about how individuals and businesses were responding to the rapidly changing circumstances.
So, planning for new, rapid-turnaround outputs was one of the first parts of the ABS response that began in late February.
At the same time, like organisations and businesses across the country, large and small, we were also focusing on protecting the health of our staff and respondents, while ensuring our ongoing operating capacity.
By mid-March, given the risks associated with travel and gatherings in confined spaces, we reduced the number of agencies and individuals attending lock-ups for market sensitive statistics. We also suspended surveys that required close face-to-face contact, including those within Aboriginal and Torres Strait Islander communities, to protect vulnerable populations.
Some of our regular activities were paused to free up resources to ensure our key economic series weren’t at risk, or to make way for the collection of COVID-19 related data.1 Critical teams were identified, and surge capacity and shadow teams put in place to keep production processes going in the event that infection rates of ABS staff rose to critical levels.
As Australians were encouraged to work from home where possible, the ABS enjoyed a relatively smooth transition to home-based work, due to our existing flexible working arrangements, which already incorporated large-scale teleworking capacity.2
It will be interesting to see to what extent regular work patterns undergo a permanent shift as a result of the current huge rise in the number of people working from home, and the associated use of videoconferencing facilities.
So, from the early stages of the pandemic, we were confident we could continue to deliver the key statistics Australians were depending on.
But clearly maintaining our business as usual approach was not sufficient.
So, on the 16th of March, and then twice again over the following fortnight, I announced the ABS would publish new data, to respond to the hunger for up to date information about the economic responses of individuals and businesses to the pandemic.
To illustrate this, Chart 1 shows the progression of cases in Australia, key events, including the announcement of government support packages and social distancing requirements, and the timing of ABS COVID-19-related data releases.
Additional analysis using existing data
The first extension to our regular products included additional analyses of the data we had already collected.
From 16 March, extended analysis of short-term visitor arrivals to Australia and, for each state and territory, the flow of international students, was published in Overseas Arrivals and Departures, Australia (cat. no. 3401.0). These figures will continue to be published each month.
In addition, Labour Force, Australia, Detailed – Electronic Delivery (Cat. no. 6291.0.55.001), now includes additional monthly analysis on hours worked, including reasons for working fewer hours, and quarterly analysis by industry.
But these new analyses were just the beginning.
Additional confidentialised data
Analysts were keen to investigate pre-COVID-19 business data disaggregated by industry and region, so confidentialised microdata for Australian businesses is also being made available today for researchers to produce tables, graphs and maps.
These data relate to financial year 2018-19, and support analysis of the Australian economy prior to coronavirus impacts. These data are available free via the TableBuilder facility on the ABS website.
In addition, our DataLab has provided a platform to provide more confidentialised data to trusted users to conduct extra analyses, even though these data can’t be released publicly.
A number of changes to our processes – both in obtaining permissions from data custodians to approve users and ABS processes to on-board users have enabled much faster access to these data – all while maintaining stringent standards to ensure confidentiality.3
New interactive maps
Another of part of the early ABS response to COVID-19 was the development and release of new, interactive employment distribution maps.
These maps, released initially to Commonwealth Treasury and then to the public in mid-March, support a regional assessment of the potential impacts of the coronavirus on employment. The maps include age, industry and employment dimensions, and are based on data from Jobs in Australia (Cat. no. 6160.0).4
Take for example Chart 2.
It provides the distribution of all employed persons aged 50 years and older across areas of Brisbane and Sydney before social distancing and other measures were implemented.
The map shows that Sydney had a much higher concentration of areas where older workers live – those who might be at greater risk from COVID-19. Images like this help decision makers target policy and strategies according to local conditions and risks.
Other interactive maps released by the ABS in early April showed small area modelling of chronic health conditions. They show the geographic distribution of people with a range of health conditions that make infection by COVID-19 more life-threatening, for several age groups, by residential location.
Chart 3 shows people aged 60 years and older in Melbourne with three or more chronic health conditions. You can see there are distinct clusters of areas in northern and western Melbourne, as well as Geelong, where large numbers of people have these characteristics.
In NSW, the government is using similar aggregated, small area data from the ABS and other sources to identify these kinds of at-risk communities. This will help plan more localised methods of COVID-19 containment, as/when national or state-wide measures are relaxed.
The wider community is also making use of these maps. Since their release, over 6,400 people have viewed the employment distribution maps, and more than 6,800 have accessed the health conditions maps.
Another new set of ABS COVID-19 related products has been the range of preliminary data we’ve released. In publishing these figures, we have trusted that analysts would understand the caveat that goes with them – that revisions would necessarily follow the preliminary figures.
As we know, the social distancing restrictions introduced in late March had major impacts on consumption behaviour and retail turnover, which is why we introduced the preliminary retail trade publication to provide faster information on these impacts.
Preliminary retail turnover estimates are being published two or three weeks after the end of each month, with the final monthly estimates published two weeks later in Retail Trade, Australia (Cat. No. 8501.0).
Chart 4 shows the latest Preliminary Retail Trade data, which saw an estimated increase in retail turnover of 8.2 per cent in March 2020, seasonally adjusted. This was the largest seasonally adjusted monthly rise in the nearly forty-year history of the Retail Trade series.
Analysis of supermarket and grocery store scanner data backed up stories of some consumers stockpiling for the pandemic and television images of empty supermarket aisles. It revealed a doubling in monthly turnover for products such as toilet and tissue paper, flour, rice and pasta.
As anticipated, sales were also strong in retail industries selling items related to home offices, while falls occurred in cafes, restaurants and takeaway food services, which were particularly affected by social distancing measures.
A preliminary estimate for Australia’s international merchandise trade is also being published three weeks after the end of each month.
The first preliminary figures, released last week, showed merchandise trade rebounding in March, following declines in January and February. Among the key movers were exports of iron ore to China, Australia’s largest trading partner. The final monthly estimates will be published in International Trade in Goods and Services, Australia (Cat. no. 5368.0).
We’ve received a great deal of interest in these new statistical products, and we have also adapted to deliver other types of new products.
A five-year transformation investment in the ABS has made possible a range of new products and approaches.
These include the quick delivery of new survey instruments and outputs, and we have seized the opportunity to get from the field to publication in record time.5
Over the past two months, we have introduced two new rapid-response surveys, the first of which is the Business Impacts of COVID-19 Survey.
This survey was introduced in March to measure the incidence and nature of impacts of COVID-19 on businesses as they do their best to navigate their rapidly changing environments. It provides information from around 1,200 responding businesses, including changes to workforce arrangements, staffing levels and location of work.
The first cycle started collection prior to the Government’s announcement of Stage 1 social distancing restrictions and the second stimulus package ($66bn).
It found almost half (49 per cent) of Australian businesses had experienced an immediate adverse impact as a result of COVID-19, and 86 per cent anticipated adverse impacts in future months.
The second cycle started collection after the announcement of the Stage 2 social distancing restrictions, on the day the JobKeeper package was announced.
It found nine in ten businesses were continuing to operate, but were doing so with significant operational changes.
Just over half (55 per cent) had temporarily reduced staff work hours, while a similar percentage (52 per cent) had changed staff work locations (including working from home).6
The Business Impacts of COVID-19 Survey will continue to run at least once a month while it continues to generate valuable information. The results from the third cycle will be published next Monday, 4 May, and will provide information about business sentiment, trading status and business utilisation of JobKeeper payments.
Response rates for our surveys, particularly business surveys, have fallen. I want to acknowledge and thank those businesses that have made the effort to keep responding in these challenging times. Their continued participation is both appreciated and extremely valuable. The information they provide helps us understand the economic and social impacts of the pandemic, and in turn helps governments and communities decide how best to respond.
I also want to thank the individuals who continue to respond to our household surveys, including the second of our new rapid-response surveys, the Household Impacts of COVID-19 Survey. (In-house this survey is known simply as the ‘RAPID’ survey, an acronym which stands for Rapid Acquisition of Population Information and Data – but for clarity I’ll continue to use its public name).
The response rate in the first week of this survey was a healthy 91.5 per cent. One of the upsides of running a household survey in a pandemic is that it’s easy to find people at home who are willing (some even eager) to spend a few minutes on the phone answering survey questions!
This survey captured information from a sample of over 1,000 households, with data collected from 31 March to 6 April, 2020, and published on 20 April. It provides a quick snapshot of how people are faring in response to the changing social and economic environment caused by the COVID-19 pandemic. Responses to the survey are weighted to reproduce the actual distributions in the wider population, to ensure the survey responses are representative.7
While it is as well to recognise that people breaking the restrictions might not reveal that to the interviewers conducting the survey, the results are nevertheless encouraging. The survey found that social distancing restrictions were being taken seriously, with almost everyone reporting that they were keeping their distance from other people (98 per cent), and the vast majority also avoiding public spaces and events (88 per cent) and cancelling plans to gather with friends and family (87 per cent).
Given the seriousness for older people of being infected with the virus, it is not surprising that a higher proportion of people aged 65 years and over reported that they were self-isolating than in younger age groups. At the time of the survey, 2.2 million Australians aged 18 and over had had a flu vaccination this year, with another 12.2 million intending to be vaccinated.
Respondents were also asked questions about changes to their job situation.8
The second iteration of the Household Impacts of COVID-19 Survey has collected information on changes to job situation; feelings of emotional and mental well-being; extent of contact with family and friends; financial stress; and whether stimulus payments have been received and, if so, how they were used. Results from this second iteration will be published tomorrow.
New insights into the labour market
Let me return now to the topic I was originally going to talk about today – insights into the labour market.
Understanding the impact of the spread of the coronavirus on the labour market – from the perspective of both people and businesses, as well as the responses to government restrictions and government support packages – will be critical to understanding the evolution of both the economy, and people’s wellbeing, over coming months.
The monthly Labour Force release provides the most reliable and authoritative source of information on the Australian labour market. It collects data from 25,000 households a month (approximately 50,000 people), and therefore provides estimates of changes in key labour market indicators with a high degree of precision, in line with international standards.9
But it takes time to conduct such a survey, and faster publication of more frequent data on the labour market was clearly needed.
For some time, the ABS and the ATO have been discussing the opportunity to extract higher-frequency information from the Single Touch Payroll, or STP, system. This system is used by employing businesses to report to the ATO information about wages, superannuation and tax payments for their employees.10 The arrival of the coronavirus meant that providing public access to this rich vein of detailed near real-time information became an urgent priority.
I want to take this opportunity to record my gratitude to the ATO for expediting this process, at a time when they have been extremely busy on many other things, including delivering the JobKeeper package.
New COVID-19 labour market products
This joint ATO-ABS effort has evolved into the Weekly Payroll Jobs and Wages in Australia publication. The new data (Chart 5) show that between 14 March and 4 April (the three weeks after Australia recorded its 100th confirmed COVID-19 case) the number of employees in paying jobs fell by 6.0 per cent.11
Chart 6 shows that the Accommodation and food services industry saw the largest fall in the number of employees in paying jobs (down by 25.6 per cent), followed by the Arts and recreation services industry (down by 18.7 per cent).
Total wages paid by businesses fell by 6.7 per cent over the period.12
Together with the Household and Business Impacts of COVID-19 surveys, Weekly Payroll Jobs and Wages in Australia (Cat. no. 6160.0.55.001) will continue to provide up-to-date labour market information.
It will allow analysts to see evidence of labour market changes before they appear in the more detailed Labour Force Survey results.
The feedback from those who have accessed our new data offerings has been overwhelmingly positive. Some of those people are no doubt with us in this virtual room today, and I thank you for your support, and the positive feedback you have provided to me and my team.
We will continue to look at what more we can do to provide valuable information in the months ahead.
The next phase of the COVID-19 crisis – determining how, when and to what extent to lift the level of restrictions placed on individuals and the economy, is arguably the phase with the greatest reliance on data.
The international statistical community is collaborating closely as we grapple with measurement issues including:
Policymakers, businesses and many other sectors in the community have already begun discussing the timing of relaxing social distancing and other restrictions, and how this might best be achieved.
At the ABS, we will continue to provide data to provide insights across a range of economic and social dimensions to support these important discussions.
A dedicated taskforce has been established to identify and secure new data sources to supplement the production of existing ABS products, address emerging policy questions and data needs in response to COVID-19, and deliver novel and innovative products in the future.
Questions, reactions and insights
In the midst of confronting economic times, the decisions that will need to be made will require the input of experts in many fields, including business economists and statisticians, among others.
In the time allocated for our discussion, I’m happy to take your questions and I’m particularly interested in your views on the road ahead, and what more the ABS can do to support your efforts to gauge the health of Australia’s economy and businesses as we navigate the road out of the COVID-19 crisis.
* I am grateful to Kristen Stone and Meg Dixon-Child for much help preparing these remarks, and to Martin Brady, Cheng Chen, Shaun Copley, Nicholas Gruen and Jenny Wilkinson for helpful comments and suggestions on an earlier draft.
2From 2014 to 2018, all ABS staff were provided with laptops, Skype messaging and video-calls were embedded as part of our standard communications and, with 10 offices around Australia, we increased our teleconference capability. By May 2019 two-thirds of ABS staff were working away from the office at least occasionally. At present, our average office occupancy is around 15%.
7The survey was voluntary and conducted with the person in the household willing to answer it. As it turned out, respondents were disproportionately female with higher levels of education. Responses were therefore reweighted to reproduce the actual distributions in the wider population, to make the survey representative. On a separate point, if the success of a statistical product was measured by its media coverage, the Household Impacts of COVID-19 Survey would have scored well. In the 24 hours after its release on April 20, it generated over 70 items of media coverage, with a potential reach of almost 5 million people.
8Given the nature of the survey, respondents were not asked the full suite of employment-related questions that are included in the Labour Force Survey. Therefore, the results of the two surveys are not directly comparable.