Remember Biden's advice to everyone just to be patient on the economy?

Job creation using the Establishment (employer) and Household surveys (click on figure to make it bigger).


As more people give up looking for work and leave the labor force.


WH economic adviser Austan Goolsbee told CNBC:
"The trend is clearly going the right way. You know, we have the unemployment rate. So far in the last five months is down 8/10 of a point. We added more than a quarter million jobs a month over the last quarter and we've got more than 2 million over the last 14 months. That's -- where we are moving the right way."

The right direction? Well, one thing that is interesting is that Goolsbee picks the lowest employment point 14 months ago in February 2010 for his comparison to get the 2 million job private sector growth using the Establishment survey? Why that month besides the fact that it gives him the largest number? Why not the beginning of the recovery? It is then about 900,000 jobs. If he looked at all jobs (probably the most relevant number), it would drop to 535,000. Why not from the beginning of the stimulus? It is actually a minus 6.6 million. Why not the beginning of last year in January 2010? Why not the number of permanent jobs since the recovery started? It is only 35,000. 535,000 total nonfarm jobs - 500,000 temporary service sector jobs. Someone in the media needs to call the Obama people on these weird definitions of changes in the number of jobs.

This is from a discussion back in 3/10/2010.

BIDEN: What I keep saying in the White House -- patience. Have a little patience here. Things are beginning to turn around. They're beginning to turn around not only in fact in the economy, they're beginning to turn around in figuring out the Republicans are for nothing. What are they for? What have they offered?
And it is true that when you see in the news every single night X hundred billion dollars for banks, no matter who started it -- no matter who started it -- it's awful hard for the guy sitting at the kitchen table in North Philly saying, Hey, man, I don't have a job, and they're giving all this money to banks. This is a hard thing to sell until we start actually every month seeing 100,000 created, 200,000 jobs created, the economy moving. It's going to move. Patience. . . .

UPDATE: A friend of mine, Steve Bronars, wrote me a note about the smoothness of the Establishment survey numbers. Unlike the Household survey numbers, the Establishment survey numbers have a lot of adjustments and imputations made by the government before the numbers are released.

The establishment survey has many adjustments built into it.

Its a voluntary program and in each month some companies may be slow to get their payroll numbers in. Consequently the BLS reports preliminary numbers for 2 months before reporting the final totals. They must have an algorithm to impute the companies that have not yet responded.

The BLS oversamples large companies because with a sample of size N, if you want a more precise estimate of total employment, you want large employers included in N. They do this despite the fact that most companies are small.

The BLS has an algorithm for imputing the jobs due to firms being born. The have no way of getting an unborn firm into the survey but they know that this is an important phenomenon for employment changes. I think they have the same problem with deaths of firms. The firms they sample may have a lower death rate than most firms, which they must adjust for. Even if one of their firms does die, it is unlikely that the firm will send in a report to the BLS saying that they closed down.

Look at the magnitude of the annual revisions to the BLS establishment numbers over the past few years. The apparent smoothness is because the numbers have been processed.

The CPS sends people out to a given set of addresses each month and does their best to interview someone in each household and come up with the number of employed persons. There are no revisions, and in theory it is benchmarked to the population.