The scale of the problem in comparing GDP falls related to the coronavirus lockdown is nicely illustrated by the 2nd quarter falls in Scandinavian. Sweden – so the story goes – did not have a lock down but has had more coronavirus deaths and been hit harder by the economic impact of the pandemic. In the second quarter Sweden’s GDP fell by 8.6%, Denmark’s 7.4% and Finland’s 3.2%. Norway’s GDP fell by 6.3% – also better than Sweden.
Certainly, when it comes to deaths Sweden has seen many more than its close neighbours. (Measured per 100,000 they are still below that of the UK at 57 compared to 62 and 82 for Belgium.)
There are however two other problems with this story. The first is that Sweden has had a substantial lockdown. It has just had more of voluntary character. Unfortunately, we have no lockdown index that measures the scale of what has happened to make comparisons easy.
Measuring GDP Falls
The second is what happens when we compare GDP falls in close neighbours like the Scandinavian countries? Look at the figure below.
You have to admit it is pretty weird. First, we see a miracle in Finland – there has been no hit in terms of investment or government expenditure. The investment figures may be explained by public investment and construction holding up while private investment has fallen. But this needs more exploration and, if true, an explanation in itself.
The second problem – which we have been writing about and tweeting on for two weeks now – is that the GDP data is not reliable for comparisons because different statistical offices are doing the public sector calculations in different ways. This particularly affects the valuation government services. Some national statistical offices are calculating these as if nothing has changed. Since teachers are paid educational output is holding up. Others take the view that since fewer children are in school it has fallen. See here the contrast between Finland and the others.
Third, we see that there has been a substantial hit in Norway and Sweden. (Denmark’s figures will be out later). But the overall hit is bigger in Sweden than Norway – yet for components it is not. How is this possible? One explanation might be trade since remember in calculating output we calculate
GDP = C+I+G+(X-M)
The value of output in an economy is equal to household consumption (C), investment (I), government output (G) and net exports (X-M).
Trade performance is unlikely to be related to coronavirus but even if it were it would not reflect much on domestic policies. Short term losses (or gains) would not be within the control of governments.
These figures come from the national offices. They are early data and not yet on Eurostat who are hopefully now exploring comparability and other issues.
But what this means for us is that we cannot make simple international comparisons like this one from the Guardian.
The economy of Finland also outperformed its larger neighbour in the second quarter, despite a tougher lockdown. Finland’s gross domestic product shrank by 5% against an 8.6% contraction in Sweden from the previous three-month period.
What do we need to think about?
The chart below shows the sorts of things that need to be thought through before genuine comparisons can be made. The complexity of the calculations is such that if comparisons are to be made it will probably require adjustments to the more sophisticated data of countries like the UK and Sweden to make it consistent with the cruder data of other countries.
Bad Data Drives out Good?
This might be a nice example of a new law – that bad data drives out better data (or at least the aspiration to better data).
This will obviously be comforting too to politicians worried about the effect on their reputations of showing a bigger GDP hit.
Check output chapter 3 of my 2020 book on Productivity for other examples of the rabbit holes we disappear down.