Questions about unemployment are common news items in the wake of this recession. The subject is also fertile ground for explaining the difficulties economists face in precise measurement and forecasts. Let me explain.

Unemployment comes in three main flavors. The first is frictional unemployment which is, as its name suggests, associated with the movement of workers to different locations and the opening and closing of businesses. Think of someone moving with their spouse or a bad restaurant closing. The second, structural unemployment, occurs when workers with antiquated skills (like my acuity at computer punch cards) lose their jobs. Cyclical unemployment, the third flavor, occurs in a recession as workers lose jobs due to lower demand for goods. A purist would include seasonal unemployment in this list, but I won't. Some in my profession would dispute this taxonomy. Those disagreements matter to national economic policy and in economic research journals but not so much elsewhere.

Full employment is said to exist when only frictional and a small and constant level of structural joblessness occurs. This type of job churning is sign of a growing, healthy economy. Economists used to call this the natural rate of unemployment, which did little to improve our hard-hearted image. It was later called the non-accelerating inflation rate of unemployment—an even more unfortunate moniker. We have settled on full employment to describe this condition, but describing it and measuring are two different matters.

There are literally thousands of estimates of the rate of full employment. These involve use of government statistics of various quality in different times and places. Economists write computer programs (no longer on punch cards) to estimate what this rate might be, given a myriad of other factors. The difficulty lies not in the statistical analysis or in the quality of the data, but in the fact that people change. Technology changes at different rates in different places, workers with different skills take different periods of time to retrain, and it is at some times easier to move to a new job than at other times.

I estimate full employment to be 5.7894513 percent. This is wrong, but rounding it to 6 percent is probably just fine for thinking about the new labor market normal. I wish that the number were lower, and that more Americans would be working as the full effects of this recession are done. I am afraid that will not be the case for two reasons. First, the very low unemployment rates in 2007 (4.7 percent before the recession started) were part of the bubble. Try as we might, that bubble ain't gonna re-inflate. Second, structural unemployment has risen. In three years the country has lost more than 2 million manufacturing jobs, but Americans are producing more manufactured goods than in 2007. These folks are structurally unemployed because they are no longer needed in their old occupations. Structural unemployment is a byproduct of healthy technological progress, and those who can learn new skills flourish. Those who cannot learn new skills will not flourish, and that is a problem that is not going away.