The relatively large fall in UK GDP partly reflects better data
27 August 2020
SUGGESTED
Transport
26 August 2020
Economic Theory
1 September 2020
Government and Institutions
There is a lively debate among economists about the ways in which UK statisticians have estimated the impact of Covid-19 and the lockdown on the public sector’s contribution to GDP. This is a relatively arcane topic, but important. As David Smith has put it, “did the official statisticians exaggerate the UK’s coronavirus recession by incorporating big falls in health and education output?”.
These issues have also led to some confusion about an alternative measure of inflation known as the GDP deflator. That’s where I’ll begin. (There is a longer version of this blog on my website, with some charts and tables.)
The GDP deflator is used to convert nominal data into real data, or cash spending into the volume of goods and services actually bought. Implicit deflators can be calculated by dividing an aggregate measured in current prices by the same aggregate measured in constant prices.
Changes in the GDP deflator are therefore usually seen as a measure of general inflation. The ONS itself describes the implied GDP deflator as “the broadest measure of inflation in the domestic economy, reflecting changes in the price of all goods and services that comprise GDP”.
It is therefore striking that the UK GDP deflator jumped by 6.2% quarter-on-quarter (q/q) in the second quarter (Q2). What’s more, the UK was the only major economy to report such a large increase.
However, it would be misleading to interpret this jump as “inflation” in the normal sense. For a start, the GDP deflator covers the whole of the economy, not just consumer spending. If you are looking for an “inflation” measure to compare to the more familiar consumer price index (CPI), you should look at the deflator for “household consumption expenditure”. This was little changed in Q2.
Instead, the big mover was the deflator for “government consumption expenditure”, which jumped by 33% q/q. In the words of the ONS “this notable increase occurred because the volume of government activity fell while at the same time government expenditure increased in nominal terms”.
So, what’s behind this? The short point is that UK statisticians have put a lot more effort into estimating how much activity actually took place in sectors such as public health and education. In contrast, statisticians in other countries have typically based their estimates here on the fact that staff costs and other bills were still being paid as usual.
I believe the UK approach is the right one. If you think that GDP should be a measure of economic activity, or value-added, or contribution to wellbeing, the fact that schools are at least partially closed and that hospitals are treating fewer people must be relevant. The standard assumption that output can be proxied by inputs is surely wrong in these circumstances.
There are still some valid concerns about the size of the adjustments that the ONS has made. According to the official UK numbers, total government consumption expenditure rose by 14% q/q in cash terms in Q2, but the volume of goods and services that this spending paid for was assumed to fall by the same percentage. In particular, the ONS has estimated that government healthcare consumption fell by 30% in volume terms, while education fell by 25%.
My own view, though, is that the ONS numbers are at least in the right ballpark. It seems perfectly plausible that the amount of “education” that children received fell by as much as a third in the second quarter of the year. Online teaching has filled some of the gaps, but coverage is patchy: one study found that children locked down at home in the UK spend an average of just 2.5 hours each day on schoolwork, with many (particularly in state schools and in poorer areas) doing significantly less. Children are also still missing the personal contact with teachers and their friends, and all the social and sporting activities that are part and parcel of normal school life.
In the NHS, the numbers of “elective” (non-emergency) operations, A&E admissions and regular GP appointments all fell sharply in Q2. There is particular concern about the number of cancer patients who may be missing treatment.
Obviously, many staff and resources were diverted to deal with the pandemic and this should be recognised when measuring the output of a service whose role is to protect the nation’s health. However, the NHS employs more than 1.7 million people in total and 1.5 million on a full-time equivalent basis, representing about a third of total public sector employment. Only some of them have actually been fighting Covid-19. (Another million or so work in state education.)
The ONS has also made a decent attempt at allowing for the different ways in which health and education services have been provided in the lockdown. For example, the estimates for education output take account of remote learning by making use of survey data, adjusted for teachers hours and parental input.
In summary, the ONS is right to measure the actual outputs of the public sector during the pandemic, rather than simply assuming these can be proxied by the value of inputs. It is vital that official data properly reflect the impacts of the pandemic and the lockdown on the wellbeing of children and the health of all.
This may well mean that the UK reports a bigger fall in GDP during the lockdown and a faster recovery when the restrictions are lifted – especially when all children return to school and the NHS starts to catch up on the backlog of non-Covid work. As a result, the official GDP numbers for Q3 are likely to be even stronger. But this is a more accurate reflection of what is actually happening.
It may also mean that the differences between the official GDP data in the UK and those in other countries have been exaggerated by the differences in the ways in which the numbers are calculated. For example, the German GDP data for Q2 included an increase in the volume of government consumption expenditure of 1.5% q/q, in contrast to the slump of 14% reported in the UK.
If we compare nominal GDP rather than real GDP, the fall in the UK in the first half of the year, while still huge, was about the same as that in France and probably in Italy too (Italy has not yet released comparable data, but the fall in real GDP there was similar to that in France). Either way, Spain has been the European economy worst hit by coronavirus so far, not the UK.
Nonetheless, as the ONS also acknowledges, there is more than usual uncertainty here and scope for revisions. The statisticians have done a pretty good job of explaining what they have done. However, it would be helpful if more people were aware of these issues.
These issues have also led to some confusion about an alternative measure of inflation known as the GDP deflator. That’s where I’ll begin. (There is a longer version of this blog on my website, with some charts and tables.)
The GDP deflator is used to convert nominal data into real data, or cash spending into the volume of goods and services actually bought. Implicit deflators can be calculated by dividing an aggregate measured in current prices by the same aggregate measured in constant prices.
Changes in the GDP deflator are therefore usually seen as a measure of general inflation. The ONS itself describes the implied GDP deflator as “the broadest measure of inflation in the domestic economy, reflecting changes in the price of all goods and services that comprise GDP”.
It is therefore striking that the UK GDP deflator jumped by 6.2% quarter-on-quarter (q/q) in the second quarter (Q2). What’s more, the UK was the only major economy to report such a large increase.
However, it would be misleading to interpret this jump as “inflation” in the normal sense. For a start, the GDP deflator covers the whole of the economy, not just consumer spending. If you are looking for an “inflation” measure to compare to the more familiar consumer price index (CPI), you should look at the deflator for “household consumption expenditure”. This was little changed in Q2.
Instead, the big mover was the deflator for “government consumption expenditure”, which jumped by 33% q/q. In the words of the ONS “this notable increase occurred because the volume of government activity fell while at the same time government expenditure increased in nominal terms”.
So, what’s behind this? The short point is that UK statisticians have put a lot more effort into estimating how much activity actually took place in sectors such as public health and education. In contrast, statisticians in other countries have typically based their estimates here on the fact that staff costs and other bills were still being paid as usual.
I believe the UK approach is the right one. If you think that GDP should be a measure of economic activity, or value-added, or contribution to wellbeing, the fact that schools are at least partially closed and that hospitals are treating fewer people must be relevant. The standard assumption that output can be proxied by inputs is surely wrong in these circumstances.
There are still some valid concerns about the size of the adjustments that the ONS has made. According to the official UK numbers, total government consumption expenditure rose by 14% q/q in cash terms in Q2, but the volume of goods and services that this spending paid for was assumed to fall by the same percentage. In particular, the ONS has estimated that government healthcare consumption fell by 30% in volume terms, while education fell by 25%.
My own view, though, is that the ONS numbers are at least in the right ballpark. It seems perfectly plausible that the amount of “education” that children received fell by as much as a third in the second quarter of the year. Online teaching has filled some of the gaps, but coverage is patchy: one study found that children locked down at home in the UK spend an average of just 2.5 hours each day on schoolwork, with many (particularly in state schools and in poorer areas) doing significantly less. Children are also still missing the personal contact with teachers and their friends, and all the social and sporting activities that are part and parcel of normal school life.
In the NHS, the numbers of “elective” (non-emergency) operations, A&E admissions and regular GP appointments all fell sharply in Q2. There is particular concern about the number of cancer patients who may be missing treatment.
Obviously, many staff and resources were diverted to deal with the pandemic and this should be recognised when measuring the output of a service whose role is to protect the nation’s health. However, the NHS employs more than 1.7 million people in total and 1.5 million on a full-time equivalent basis, representing about a third of total public sector employment. Only some of them have actually been fighting Covid-19. (Another million or so work in state education.)
The ONS has also made a decent attempt at allowing for the different ways in which health and education services have been provided in the lockdown. For example, the estimates for education output take account of remote learning by making use of survey data, adjusted for teachers hours and parental input.
In summary, the ONS is right to measure the actual outputs of the public sector during the pandemic, rather than simply assuming these can be proxied by the value of inputs. It is vital that official data properly reflect the impacts of the pandemic and the lockdown on the wellbeing of children and the health of all.
This may well mean that the UK reports a bigger fall in GDP during the lockdown and a faster recovery when the restrictions are lifted – especially when all children return to school and the NHS starts to catch up on the backlog of non-Covid work. As a result, the official GDP numbers for Q3 are likely to be even stronger. But this is a more accurate reflection of what is actually happening.
It may also mean that the differences between the official GDP data in the UK and those in other countries have been exaggerated by the differences in the ways in which the numbers are calculated. For example, the German GDP data for Q2 included an increase in the volume of government consumption expenditure of 1.5% q/q, in contrast to the slump of 14% reported in the UK.
If we compare nominal GDP rather than real GDP, the fall in the UK in the first half of the year, while still huge, was about the same as that in France and probably in Italy too (Italy has not yet released comparable data, but the fall in real GDP there was similar to that in France). Either way, Spain has been the European economy worst hit by coronavirus so far, not the UK.
Nonetheless, as the ONS also acknowledges, there is more than usual uncertainty here and scope for revisions. The statisticians have done a pretty good job of explaining what they have done. However, it would be helpful if more people were aware of these issues.
1 thought on “The relatively large fall in UK GDP partly reflects better data”
Comments are closed.
SHARE
Fair enough, although in other European countries the output of their healthcare systems has probably not fallen as much as in the UK. This is because many European countries have ‘social insurance’ systems and a much higher proportion of private sector providers of medical care – as opposed to direct government-run provision as in the UK (i.e. the NHS). These providers only get paid if they actually provide care, unlike the NHS.