What the ideal is

Inflation or Unemployment? by Megan McArdle. Emphasis added.

More to the point, this rather begs the question of what the ideal is. You could define "full employment" as a situation in which everyone who wants a job has one. But on any level of economic organization above the size of a small village, this will not be true. At any given time, there will be what economists call "frictional unemployment", which reflects the fact that it takes a little time to find a job after you've entered the labor force. As long as companies are constantly creating and destroying jobs, the ideal level of unemployment is not zero.

To me, full employment is probably best defined as the situation where everyone's going to get a job in a relatively comfortable period of time--not instantly, but pretty briskly. The time to find a job will vary with skill and income level (in my full employment world, we understand that a laid-off marketing executive is going to take longer to find another job than someone who works retail for $9 an hour, because it takes longer to find work that suits specialized skills).

Over time, I'd say full employment will probably correspond pretty closely with what Milton Friedman and the monetarists called "the non-inflation accelerating rate of unemployment", or NAIRU. NAIRU could be 4%, but it's not particularly likely to be--and it's even less likely to stay at 4%. NAIRU changes along with changes in the economy and the labor force. Equilibrium unemployment rates are very low when most men are unskilled labor who can pick up new jobs very easily--and need to, because they can never save up enough of a cash cushion to see them through an extended job search. They will be higher when jobs are more specialized, when people are rich enough not to have to take whatever's offered, and when there are regulatory barriers to hiring and firing.