I gave the methodology a read, and it’s a real head scratcher. I haven’t sat down and done the math myself, but if I were going to, I would first investigate all the assumptions they had to make, such as:
using the Personal Consuption Expenditures PCE measure of spending instead of the Consumer Expenditure Survey (CE) , since one includes military and government spending on behalf of households and the other doesn’t
Using 2019/2020 aggregate spending estimates at the census division level as part of a formula to estimate Jan 2021 spending data. (2020 was not a normal year for spending)
Using a Gross CE/PCE ratio at the census division level to estimate consumer spending in each census division’s constituent states
using the PCE growth nationally (remember what it includes) as a substitute for statewide , statewise expenditure growths.
Again, using census division aggregate data to estimate state level consumer units per household
providing no validation that their methodology for transforming PCE values into CE values works (it should have been as simple as calculating it for other years, say 2015,2016, and checking the accuracy)
using Census division CPI instead of state data.
Anyway, this is where I would start picking this mess apart. If I were so inclined.
From this little bit, I’d be concerned about the analysis using suppressed spending in 2020 as the starting point in 2021, as well as including spending from outside the CPI basket and non household entities in CPI estimates. I’d be concerned about over generalizing the states by estimating using their census units. Basically, I’d consider this report skeptically at best if no additional validation can be provided.
I gave the methodology a read, and it’s a real head scratcher. I haven’t sat down and done the math myself, but if I were going to, I would first investigate all the assumptions they had to make, such as:
Anyway, this is where I would start picking this mess apart. If I were so inclined.
From this little bit, I’d be concerned about the analysis using suppressed spending in 2020 as the starting point in 2021, as well as including spending from outside the CPI basket and non household entities in CPI estimates. I’d be concerned about over generalizing the states by estimating using their census units. Basically, I’d consider this report skeptically at best if no additional validation can be provided.