Nit: the ADP numbers are not survey based. They come directly from adp’s payroll data set which is immense but biased. So you don’t see hiring/firing delay in the birth/death data like you do with the bls stats, but actual dataset bias is a factor. The more murky part of their numbers is how they calibrate to the bls numbers.
Sure, that's 1/5 of US jobs, but it's also going to be concentrated in some industries more than others -- would be nice to see how that lines up with where job adds and losses landed.
