I don't think it's anywhere near that easy to make a statistical inference about unknowns, as in, how many people have been hired this month at newer, smaller and likely more agile companies who use a competing software platform. If they know about the needs of those potential client companies, they'd probably have had more success getting them as clients.
And yes, I integrate call center workforce management software with payroll platforms, so the call center micromanagement infrastructure knows who has the day off, or so that call center supervisors can do the approvals from within the workforce management software without having to log into the payroll system, etc.
So yes, I'm making a larger point about an expectation that ADP numbers will be getting less and less relevant over time.
I understand what your saying but it's largely irrelevant to sampling of large data sets. There are like 30 some million small business in the country if I recall, we could miss entire industries and still land on a usable data for the whole. The process of refining how we collect data , what data we collect etc are the best efforts to account for changes in a dynamic system. With most systems you can arrive at relevant data with a surprisingly small sampling. Not the place to discuss sampling/polling methodologies but your point is not lost that it's completely plausible for a company that may be losing it's position would over time have less and less insight into the macro trend.
For example if we only relied on Comcast to understand TV viewership in 2022. To bring our back and forth to some use, economic statistics in general must adapt to a changing economy. I think BLS , FRED etc are obviously aware of this and I do believe there is some credible academic rigor that is applied by professionals attempting to keep data relevant. They are always fighting against improving data of current trends vs losing correlation to data of the past, and the study of those relationship is a principal part of economic analysis.
The basket of goods for CPI undergoes change, maybe slower than some would like to see, but I believe it's disingenuous to always assume that the hesitancy to change formulas is rooted in the nefarious.
Your previous statement is correct, this is a forum on the internet full of thought vomit and for the most part it's ill-suited to a deeper discussion of policy when bite sized twitter responses are so much more palatable to the masses.
Reminds me of young professionals that are really beginning to understand their field run to the more wizened superiors to show them how they discovered a better way or an inefficiency, and sometimes they have but for most they will then learn the missing part of why the process is what it is, and their discovery is not new but simply something that requires more understanding to see why that inefficiency simply remains the lesser of two evils rather than an oversight. So tldr;
Economic Statistics are universally flawed, but that does not make them useless, the more we understand the data they are inferred from and how they are derived we can make a best attempt to weigh their value. And as always if a particular set of data is not useful then simply set it aside and seek something better.
