Calculating unemployment seems like it is always going to be a challenge no matter how it is done. For example, the current system in the US does not track unemployed graduates, as they have not been laid off and are not filing for unemployment benefits.
How is this information collected? When I was looking for work after grad school I didn't report to anyone as such. The school had no idea whether or not I was actively looking. I was denied unemployment. I'm not sure what my info might have looked like on the IRS end if that is what is being abrogated.
I see, so they don't capture that data well as they admit. household survey has a high margin of error. You have to also consider signal to noise ratio of some of these data and whether the survey even has the power to detect the signal. My grad school cohort for example was 15 people for the entire department. That is about as large as they get for PhD in most STEM fields. What are the odds this sort of employment sector is captured well in the household survey data?
This is perhaps the most well studied and validated data set that any government has ever produced. Their methodology and techniques are documented to the nth degree. "they" will admit whatever you want them to admit because they are stone cold professionals that are scrutinized by every political hack in the country and their datasets are poured over by teams of financial analysts across the globe every time they are released.
You are an internet commenter that couldn't be bothered to read the summary of the article outlining some of the issues with their data collection, where the very first bullet point summary mentions its survey based. Perhaps you shouldn't try to argue from first principles without reading the _extensive_ documentation of methods and reports the BLS puts out. Lots of us have.
