Official statistics are the standard source of information about the economy. But when, as now, economic activity is falling sharply, alternative indicators of what’s happening to households and businesses are invaluable. Where else should decision-makers look?
Since the start of lockdown, the need for timely economic and social statistics has become all the more important. All the available evidence points to a dramatic plunge in many areas of economic activity. For example, as the latest data show, UK GDP fell by 10.4% in the three-month period from February to April, as government restrictions on movement dramatically reduced economic activity.
Figure 1. UK GDP fell by 10.4% in the three months to April 2020
Yet it has become extremely difficult to collect data and construct the information so essential for policy-makers. This ranges from the number of furloughed and unemployed workers to the fragility of small businesses’ finances to the impact on international trade.
There has been a three-pronged effort to fill the information gaps:
- Collection and construction of official statistics with necessary adjustments, and new nationally representative surveys from the Office for National Statistics (ONS).
- New economic and social surveys commissioned by others.
- Use of alternative economic indicators for real-time tracking of activity.
Official statistics are the gold standard source of information about the economy, given their reliable methodologies: sampling techniques, large sample sizes and careful weighting. They are also subject to governance arrangements that guarantee their impartiality and integrity.
The drawback is that they are available only with a delay and even then subject to revisions. The ONS has provided monthly GDP estimates since 2018 but even these, published about six weeks after the relevant month, are too slow for current needs.
A further challenge is that bringing the economy to a sudden stop has raised data collection challenges and some tricky methodological questions. Production and expenditure in some sectors, such as hospitality and travel, have plunged. How should the incomes of their furloughed workers, paid by government, be counted in GDP? What will be the impact of Covid-19 on healthcare output? Or education? If a business does not respond to an ONS survey, is this because the office is closed or because the firm has gone out of business? The range of issues relating to GDP estimates is addressed by the ONS here.
Price information is affected too. Data collectors cannot gather prices in the usual way because stores are closed; and online prices are not a perfect substitute, although they do give an insight. For example, experimental ONS web-scraped data showed prices of some high-demand goods such as nappies, pasta and cough medicine increased substantially in the last week of March.
Related question: How can we measure consumer price inflation in a lockdown?
Some items – such as hand sanitiser, flour and yeast – were unavailable for some weeks, or seen even steeper price increases. Consumers are not spending on certain types of item such as haircuts and meals out, so the expenditure weights used to construct a price index like the CPI (consumer prices index) ought to reflect this. Again, the ONS has explained how it plans to calculate the CPI; while the Institute for Fiscal Studies has summarised the challenges.
Despite the measures taken to continue these key economic series, future empirical work will need to take careful account of the extraordinary movements in growth, inflation, unemployment and the government’s finances in 2020/21.
The ONS has also introduced rapid response statistics using new two-weekly surveys covering respectively business impacts and social impacts new data sources and experimental methods. Links to latest data can be found here.
Economists and other social researchers have sprung into action by extending existing surveys or commissioning new ones. Some of these are large scale, continuing surveys. One example is the Decision Maker Panel covering business practices.
Related question: Which firms and industries have been most affected by Covid-19?
Another is the Understanding Society survey, which regularly interviews around 40,000 UK households and which, in its Covid-19 questionnaire, is covering issues ranging from employment and income to loneliness or home schooling.
Other new surveys cover a range of issues. And the European Economic Association has established a registry of Covid-19-related projects, inviting research teams working with real-time data during this crisis to share their work.
Alternative indicators for real-time tracking
The urgent need for rapid information has boosted efforts to use real-time indicators to track the economic downturn. These lack the comprehensive coverage and appropriate weighting of official aggregate statistics, but nevertheless give an invaluable snapshot of some areas of activity.
Interest in ‘nowcasting’ the economy – making real-time predictions of current data ahead of the publication of official data – has been on the increase for some time.
Related question: How can we measure what is happening in the economy now?
A number of papers have used Google Trends, for example, to measure unemployment. There are other social science applications of the technique, although there are well-known limitations, including the tendency of observed statistical relationships between Trends searches and other variables to break down after a period of time.
Nowcasting GDP is a well-established phenomenon, producing estimates of GDP and the corresponding range of uncertainty well in advance of official ONS data. The National Institute of Economic and Social Research (NIESR) has been nowcasting monthly GDP since 2004 and continues to provide a monthly estimate one month ahead of that provided by the ONS.
The Bank of England’s Monetary Policy Committee considers nowcasts of GDP. There are commercial providers and the financial markets have a keen interest in nowcasts.
Economists in a number of other countries are of course engaged in similar real-time tracking efforts. One example is a real-time dashboard for Swiss data from KOF/RTH Zurich. In the United States, economists have built a number of trackers such as Opportunity Insights (described in this report) or the New York Federal Reserve Bank’s Weekly Economic Index.
Many of the pandemic-related efforts involve using private data, which means they provide sporadic insights rather than the big picture. The big tech companies Apple and Google have been releasing data on the level of activity around the world. These provide geographical mobility estimates as a proxy for activity, although it is not clear how long the companies will continue to provide the data. Mobile telephone records can also help track movement, as in this research on the Italian lockdown.
Other potential sources of information include bank and credit card transactions, and electricity usage. Other tech companies hold potentially useful data, including Airbnb, LinkedIn, Amazon and Uber.
While these options are all promising, the challenge is in ensuring access to private data sources – as companies are naturally concerned about privacy, security and commercial confidentiality – and also joining up different sources to provide a reasonably comprehensive picture. This is an issue that pre-dates the pandemic and will continue to be a challenge for statisticians and economists afterwards. There has been too little progress on making such data sources available either to official statisticians or researchers.
Related question: Do we make informed decisions when sharing our personal data?
Economists are nevertheless working on a range of big data sources and machine learning techniques. Access to the large-scale datasets held by private companies, as well as joining up official ‘administrative’ data sets and other surveys, is one of the key challenges for economics to deliver insights to policy-makers in future.
Related question: Can textual analysis be used to track the economy during the pandemic?
Data from market research firms are allowing us to track prices of grocery items.
There is considerable interest in understanding how the pandemic is playing out across the UK. Sub-UK data are not as robust or as widely produced – for example, a number of the surveys mentioned above are only available at the UK level.
We do have Scottish GDP data produced (with a lag of three months) and a composite indicator produced for Northern Ireland. The ONS produce experimental data for the English regions and Wales with a lag of about six months. Some work as part of the Economic Statistics Centre of Excellence has developed quicker estimates of quarterly regional GDP – you can read more about these here.
We also have regional breakdowns from the Labour Force Survey, and from the benefit system via the Department for Work and Pensions (DWP) – although some of the real-time tracking information from the DWP does not provide a regional breakdown. In terms of wider indicators, the monthly PMI measures (purchasing managers’ index) are available for the NUTS1 regions of the UK, as are measures of consumer confidence. Development of new innovative measures at a regional level remains relatively limited so far.
Where can I find out more?
Coronavirus and the latest indicators for the UK economy and society: Early experimental data from the ONS using rapid response surveys, novel data sources and experimental methods.
Saving economic data from Covid-19: Claudia Biancotti and colleagues sum up the challenges facing all economies with regard to measuring the economy.
Bringing an economist’s perspective to data science with Susan Athey.
The Covid-19 crisis through the lens of 1.4 billion transactions: Vasco Carvalho and colleagues analyse data from the Spanish bank BBVA.
Consumption patterns in the UK in real time: Elias Papaioannou in conversation with Paolo Surico.
Who are UK experts on this question?
- Michaela Benzeval, Director of Understanding Society
- Diane Coyle, University of Cambridge & ESCoE
- Ana Galvao, University of Warwick
- Martin O’Connell, IFS
- Paul Mizen, University of Nottingham, Decision Maker Panel
- Elias Papaioannou
- Rebecca Riley, NIESR/ESCoE
- Martin Weale, KCL
- Stephen Hansen, Imperial College London
- Graeme Roy and Stuart McIntyre, Fraser of Allander