Just shows how Covid has made inferences much harder to make. This is also true in many other instances, prices for example. But one thing I´m sure of. Being uncomfortable with the recent rise in inflation was a small price to pay to avoid a "deadly" depression!
I too found this very interesting and wondered whether there are a couple of international comparisons you might look at. First, the (new) payroll data in the UK - based on tax data has also been subject to major revisions. But in the opposite direction to the UK. I was wondering if the different features of the two systems might help explain this. In US I was wondering how layoffs (not a featuer of UK) are treated and whether lags in workers being taken back on might be a factor. In UK it is likely to work in opposite direction. There is 20% imputation in the RTI payroll data. I suspect, therefore, that employment in previous month is rolled forward until actual information - including outflows - are brought in in future months.
The other feature is the UK household data which is also showing odd signs. The most common explanation is that population estimates are over-stated as 'migrants have gone home' & this affects levels (rather than rates) and as the fall in population is (supposedly) focused on migrants there are major compositional changes due to the different population estimates.
However, our LFS is based on 5 waves & the 1st wave was essentially cancelled during the UK's 1st lockdown (Apr-Jun 2020) which because the attrition rates are biased is, I believe, a more likely reason for the odd compositional results thant the wrong population. Luckily, as the 1st waves resumed the effect should wear out after a year.
Happy to engage in discussion with you if you would find it useful. Bill Wells
Very interesting! I and a colleague have written a draft working paper (overlong) on the divergence in Australia between our unemployment figures - best since pre-GFC - and the number on working-age income support (twice as high). I could email it if you are interested.
Just shows how Covid has made inferences much harder to make. This is also true in many other instances, prices for example. But one thing I´m sure of. Being uncomfortable with the recent rise in inflation was a small price to pay to avoid a "deadly" depression!
https://marcusnunes.substack.com/p/a-21st-century-us-monetary-story
I too found this very interesting and wondered whether there are a couple of international comparisons you might look at. First, the (new) payroll data in the UK - based on tax data has also been subject to major revisions. But in the opposite direction to the UK. I was wondering if the different features of the two systems might help explain this. In US I was wondering how layoffs (not a featuer of UK) are treated and whether lags in workers being taken back on might be a factor. In UK it is likely to work in opposite direction. There is 20% imputation in the RTI payroll data. I suspect, therefore, that employment in previous month is rolled forward until actual information - including outflows - are brought in in future months.
The other feature is the UK household data which is also showing odd signs. The most common explanation is that population estimates are over-stated as 'migrants have gone home' & this affects levels (rather than rates) and as the fall in population is (supposedly) focused on migrants there are major compositional changes due to the different population estimates.
However, our LFS is based on 5 waves & the 1st wave was essentially cancelled during the UK's 1st lockdown (Apr-Jun 2020) which because the attrition rates are biased is, I believe, a more likely reason for the odd compositional results thant the wrong population. Luckily, as the 1st waves resumed the effect should wear out after a year.
Happy to engage in discussion with you if you would find it useful. Bill Wells
Very interesting! I and a colleague have written a draft working paper (overlong) on the divergence in Australia between our unemployment figures - best since pre-GFC - and the number on working-age income support (twice as high). I could email it if you are interested.
Peter