I used to teach labor economics and try to keep up with the literature. *There is no single measure of employment that is best for all purposes. *For most purposes (especially if you are using an employment variable in a regression equation), you get most of the explanatory power using a combination of U-6 and the employment-to-working age population ratio, both reported monthly by BLS. *If you want to do better than this, you need to measure the "quality" of employment. *The most common variables used are the length of the work week and median weekly earnings. *More recently, adding a measure of part-time employment helps. *This is all econometrics, what is statistically related to what. *
If you are interested in other specific questions, other measures can be important. *The dispersion index of employment changes is a good measure of how widespread throughout the economy changes are. *Unemployment rates for youth may be better measures if your examining crime statistics. *Ethnic breakdowns can help in explaining poverty and regional variations.
So it really boils down to why you want to measure employment. *I suspect that some politically motivated "analysts" misuse statistics and are inconsistent, but the BLS and Census are pretty professional and all of the data is pretty transparent. *Researchers can and do have access to the raw data and can develop statistics outside the reported series and create new measures for specific research projects. *So I don't have much sympathy for critics who trash the agencies but have no alternative measures that can be subject to peer review.
This is an excellent post. *Much of the raw data is available publicly. *I had it in MSAccess.
All by itself, any one stat doesnt mean anything.
It all comes down to the fact that there is no absolute frame of reference.*
The most basic reference is the comparison of a rate to itself, as the months progress. *Is it getting larger or smaller? *Relative to how it has trended before, how *is it trending now?
Other frames of reference are normative, that is comparing it to what it ought to be. *U-5 references the "every one that would take a job should have one." *U-6 references "Everyone that wants full time work should have that".
Using U-3, along with the employment ratio gives a better sense from just two stats. *U-3 is the basic reference to those actively looking. The limitation of U-3s dependency on a variabe labor force size is eliminated with the employment ratio. *The employment ratio references the total population. *And how they trend compared to each other means different things.
If the employment ratio is rising, the economy is adding jobs ahead of pop growth. *We would expect U-3 to fall. If it rises instead, then people are joining the labor force.
For me,
I find the employment to pop ratio the most informative as my "should be" reference point is that everyone should be working. *And, for whatever the "reason" given, they would be and could be, the rest is just social norms. *The more people working, the more gets built and accomplished giving a better standard of living.