On Friday I highlighted the difference between the results of the establishment survey and the household survey.  A 2006 paper from the BLS (pdf) explaining the differences in how jobs are counted in the two surveys shows us why:

Thus, although the household survey is the smaller sample, and thus subject to much more noise, it probably gave us a much truer picture of the labor market for the whole of September.  While the employment gain itself (906 thousand!) was insane, and surely not accurate, the ratios of unemployment, underemployment, and participation in the survey probably picked up the true  trend of improvement.

So let's look at those.  First of all, here is the U6 underemployment rate.  I've subtracted 8.3% from the result, to better show how the present situation compares to the last two expansions:



In the last expansion, the underemployment rate newer got below 7.9%. The late 1990s was a genuine boom.

Next, here is the employment to population ratio for the prime age workforce of 25 to 54:



We are 1.3% below the peak of the last expansion.  Again, the figure for the 1990s shows what a real boom looks like. Since that demographic is about 125 million people, that 1.3% equates to a little over 1.6 million people.

So if we were to get another 1.6 million jobs over and above population growth, plus growth of about 400,000 full time jobs vs. part time jobs to convert the involuntarily part time employed, that would get us to full employment.

Finally, here's a graph of the last 12 months of leisure and hospitality employment from the Establishment survey:

In the 11 months before September, an average of 27,000 jobs were added each month. If we replace the -111,000 jobs lost last month with the +27,000 average of previous months, the Establishment survey would have shown a gain of 105,000 jobs.  Of course, this wasn't the only sector affected by the hurricanes, but that number, while still relatively weak, is probably closer to the actual trend.