U3 is kinda silly but is mostly used because it is the original employment rate definition and therefore is comparable across time.

U6 would not include the original poster unless his new job was "minimally attached or part-time."

>This is captured as part of the "U-6" figure for unemployment.

I always wonder how this is captured. For U-3, you could go with the number getting unemployment benefits. For U-6, you'd have to literally call people and ask them, and I've literally never been called to ask if I'm working in my field and I'd guess most of us haven't either. I have to think if they are sampling, it's a very narrow sample likely biased by geography, industry, age, etc.

U-6 mostly comes from the BLS Occupational Employment Survey.

I don't think the original poster would be considered as part of U-6 unless his new job is defined as "marginally attached to the workforce or part-time for economic reasons." Simply getting a job with less pay doesn't put you in U-6.

If that's true, you've identified a huge hole in the measures. And you can't fix what you can't measure. I think you're right, because despite massive tech unemployment in the last 4 years, I don't see any of the charts spiking. Furthermore, people keep saying "this is captured in Figure X" but if it were really captured, it would be somewhat separable and could produce a chart that shows it in detail. I fear we're giving the Fed too much credit here.

The models are pretty complicated and incorporate several data sources, sampling is just one part of it. I've seen (very dry) documents that go into the methodology in detail. The model does introduce some obvious biases through assumptions of representativeness but how those interact with the headline numbers is not straightforward.

The harm is when people take the metric as literally "what percentage of people who want to be employed can find a job", which U-3 emphatically does not represent.

It doesn't directly measure it, true. But the parent poster was saying it has extremely high positive correlation to measures that do measure it, so it probably doesn't matter.

By that logic should we also further broaden "unemployment" to mean "100% - employment rate"? That's a pretty common misconception as well, maybe even more common than the one you listed.

gig work platforms have definitely skewed the traditional significance of U-3