The widely published unemployment number only measures the dynamics of layoffs (how many people became unemployed in the window of the past 6 months). The actual labor force participation is what counts. And therein lies the answer to the "why isn't pay rising" question: http://www.tradingeconomics.com/united-states/labor-force-pa... (switch to the 10-year time view)
How does a declining labor force participation explain stagnant wages? As participation (supply) goes down, wages should start to go up (to entice people back into the labor force). What we have is declining participation AND stagnant wages.
If you don't agree with that, then there aren't any good stats, hence the use of the labor participation number, as the "default" goto for unemployment.
If you want alternatives, may I recommend:
https://www.bls.gov/news.release/empsit.t15.htm
This does suggest that U-5 at least is a more realistic figure, or perhaps the average of U-5 + U-6. That would make real unemployment somewhere between 6 and 7.8, and indicates plenty of slack in the labor force.