Last week's surprisingly sharp decline in the unemployment rate from 9.4% to 9% and equally surprising anemic job growth -- 36,000 new jobs -- left a lot of investors scratching their heads. How could the unemployment rate plummet so significantly while a such a trivial number of new jobs were created?

If we simply extrapolate those numbers, we get some nonsensical results. If adding 36,000 jobs to the 139 million jobs in the U.S. economy lowers the unemployment rate by 0.4 percentage points, then adding just 720,000 jobs should lower the unemployment rate by 8 points -- from 9% to only 1%.

Yet the Bureau of Labor Statistics data shows that 812,000 jobs were added in the year from January 2010 to January 2011 (138,511,000 vs. 139,323,000). Based on the unemployment rate announced last week, we could expect that those 812,000 additional jobs would have lowered the unemployment rate to near-zero. But of course, we know they didn't.



What gives?

The basic reason why the numbers don't add up is that the BLS is constantly adjusting the variables of this basic equation:

Number of people in the workforce (civilian labor force) – number of people with jobs (employed) = number of unemployed people.

The BLS tracks the "civilian noninstitutional population" -- everyone not in the Armed Forces, school, prison, etc. -- and the "civilian labor force." The category "not in labor force" includes everyone else, including "discouraged workers" who want a job but who have stopped seeking one.

This ongoing adjustment of who gets counted as part of the labor force leads statisticians to lower the unemployment rate -- even though the number of employed people has barely ticked up.



To understand this, let's consider a labor force of 100 people, of which 20 are unemployed. The unemployment rate is 20%. But if 10 unemployed people drop out of the labor force, that reduces the total labor force to 90, and the number of unemployed to 10. As if by magic, the unemployment rate is now only 11%, even though the number of people with jobs remains unchanged.

This is the "magic" behind last week's astonishing decline in the unemployment rate.

Using the BLS data, we can reconstruct exactly what has happened over the past few years of recession and mild recovery.

The U.S. gains about 2 million new residents every year from births and legal immigration. For instance, the civilian noninstitutional population rose from 236.5 million in October 2009 to 238.5 million in October 2010.



Given this substantial increase in population every year, we might reasonably expect the civilian labor force to expand proportionally, as students graduate and new immigrants enter the workforce.

Yet according to the BLS, the civilian labor force was 153.8 million in January 2008 and 153.2 million in January 2011 -- a decline of 600,000 while the population increased by some 6 million. And the not-in-labor-force category expanded by 2 million from January 2010 to January 2011, from 83.4 million to 85.5 million.

How is this possible? When unemployed people stop looking for jobs at their local unemployment office, the government no longer counts them as unemployed. That's how the number of unemployed can drop from 15 million in November 2010 to 13.8 million in January 2011, a decline of 1.2 million, even though the economy created only about 400,000 jobs in those three months.

Over a longer time period, the not-in-labor-force group rose from 78.8 million in January 2008 to 85.5 million in January 2011 -- an increase of almost 7 million.

How do you drop the unemployment rate? Simple: remove 7 million people from the labor force.

The Real Unemployment Level Is. . .

If the labor force reflected the growth in population, then we might expect it to have increased by almost 2 million people a year. Rather than decline by 600,000 over three years, the labor force should have increased by 6 million to about 159 million.

If we take the number of unemployed as roughly 15 million (the BLS number from November 2010) and the true labor force as 160 million (out of an estimated total population of 310 million), then the true unemployment rate would be about 9.4% -- right where it was before the recent adjustment to 9%.

Other government statistical adjustments make equally little sense. Analyst Mike "Mish" Shedlock has revealed the inconsistencies of the BLS "birth-death model," which guesstimates the number of jobs created by small businesses being "born" and "dying." Depending on what the "black box" issues every month (the BLS does not reveal its methodology), the government may report that the economy has created hundreds of thousands of new jobs -- that are often revised away in estimates a few months later.

Job Growth Is Still Weak

For context, let's look at some other employment numbers. According to the ADP National Employment Report, which I reported on for DailyFinance in late December, the economy lost about 9 million private sector jobs during the recession -- about 7.75% of all private -sector jobs.

Rather than get bogged down in the legerdemain of the BLS unemployment, birth-death model and not-in-labor-force numbers, we can assess U.S. job growth just by looking exclusively at the number of jobs. Isn't that the meaningful gauge we all seek?

According to the BLS, the U.S. had 138.2 million jobs in October 2009, a few months after the official end of the recession. Sixteen months later, the country had 139.3 million jobs -- a gain of 1.1 million. That's a growth rate of about 880,000 new jobs a year. While that's certainly encouraging, this rate of job growth lags far behind those of previous post-recession recoveries as this chart depicts.



Bottom line: The numbers that matter in the U.S. economy are the total number of jobs and the number of jobs created, not the constantly massaged unemployment rate and not-in-labor-force numbers.