Bookmark and ShareShare
Monday, March 12, 2012

Unemployment Numbers Suggest U.S. Economic Boom, or Not

A careful look at the government's unadjusted household unemployment data shows a stunning 740,000 jobs added to the economy in February -- three times the 227,000 reported based on the establishment payroll survey. If this is economic reality, then the underlying economy must be growing much faster than most Americans currently believe. If the U.S. economy is surging, and jobs increased at the rate of three-quarters of a million last month, why haven't we heard a lot more about it? And, given a rapidly expanding economy, how can Gallup's nearly 30,000 random interviews with Americans across the nation show a significant increase in the unemployment rate?

According to the government's household survey, the number of employed Americans increased 740,000 to 140.684 million in February from 139.944 million in January. This increase of three-quarters of a million jobs is how the unadjusted unemployment rate was 8.7% in February compared with 8.8% in January, even as the U.S. workforce increased by 629,000 employees and the number of unemployed Americans fell by 111,000.


On the other hand, a comparison of February's household survey results to the government's December 2011 unadjusted unemployment data suggests a much more modest improvement in jobs and the U.S. economy over the past two months. The number of employed Americans increased by 3,000 on an unadjusted basis between February 2012 (140.684 million) and December 2011 (140.681 million). On the same basis, the number of unemployed Americans increased by 738,000 to 13.430 million in February 2012 from 12.692 million in December 2011. The U.S. workforce increased by 741,000 over these two months. The government's unadjusted unemployment rate increased to 8.7% in February 2012 from 8.3% in December 2011.

This morning on CNBC, there was discussion about how the increase in payroll survey jobs is hard to reconcile with economists' growth estimates for the U.S. economy. If the payroll jobs numbers are right, then the economy is growing faster than estimated, or maybe, productivity is plunging. Of course, if there are questions about how we reconcile payroll jobs with other economic data, making economic sense of the household survey surge in jobs is even more difficult.

Friday, March 9, 2012

Untangling the Government’s February U.S. Unemployment Report

The U.S. government announced the seasonally adjusted unemployment rate was 8.3% in February, the same as in January. In a mid-February report Gallup forecast that the seasonally adjusted U.S. unemployment rate would increase in February, given an upward trend in its unadjusted measure of unemployment at that time. Gallup's assessment is that the timeliness of its measure is picking up an uptick in unemployment that is not being seen by the government's survey methodology.

Gallup's measures showed that U.S. unemployment, as measured without seasonal adjustment, was 9.0% in mid-February, and in the full month of February the rate was 9.1%, based on 30,000 interviews with randomly selected Americans. Gallup's survey is conducted throughout the month using live telephone interviews. As a result, the Gallup survey tends to pick up the most recent reality in the job marketplace.

The government finds fewer unemployed Americans in its unadjusted measure in February at 8.7%, down slightly from its 8.8% of January. This slight decline took place despite an increase in the participation rate to 63.9% in February from 63.7% in January. This means more Americans were looking for work in February than January, according to the government. The government uses a version of a panel for its survey methodology and focuses on the middle week of the survey month. Both of these methodological differences can lead to a lag of recognition of quickly shifting market conditions.

Gallup looked to its mid-month rate of 9.0% to anticipate the government's release based on the BLS reference week from the middle of each month. Gallup then applied the 0.5% seasonal adjustment applied by the government to its unadjusted data in February 2011, producing a Gallup estimate of an adjusted unemployment rate for February of 8.5% -- a substantial increase from the 8.3% adjusted rate reported by the government for the month of January.

As it turned out, the government's unadjusted survey results were essentially unchanged at 8.7% in February. This lack of increase in the government's unadjusted rate contrasts sharply with the 0.5% increase in the Gallup unadjusted rate for February.

While the overall pattern of the past two years has been that Gallup and the government's unadjusted unemployment measurements track fairly closely in both direction and magnitude, there have been some instances showing large divergences. This was the case in February 2011 when Gallup's unadjusted unemployment rate increased and the government's rate declined. However, this was not the situation in February 2010 when Gallup's unadjusted rate was 10.6% versus the 10.4% rate reported by the government.

The government applied a 0.4% adjustment factor to seasonally adjust its results this February. This is slightly lower than the seasonal adjustment it applied in February 2011. As a result, the government ended up with an 8.3% adjusted unemployment rate (8.7% minus 0.4%) for February 2012 -- same as the 8.3% in January.

In sum, Gallup Daily tracking of the U.S. unemployment rate -- not seasonally adjusted and on a current time basis -- showed an increase in February from January. The government's unemployment report on Friday, based on a complex adjusted process, and with a tendency to not pick up the most recent changes in the marketplace, reports a stable rate of unemployment. Although the government report is clearly good news, Gallup suggests caution because its state-of-the-art measurement design -- without adjustment -- is signaling that the unemployment reality in the marketplace may not be as good as the government report suggests.   

Thursday, March 1, 2012

Gallup’s Unadjusted Unemployment Data Suggest Increase in BLS Adjusted Figure

Gallup finds U.S. unemployment, as measured without seasonal adjustment, to be 9.1% in February, based on almost 30,000 interviews with a random sample of Americans. When Gallup applies the 0.5-percentage-point seasonal adjustment that the government applied to its unadjusted data for February last year, it produces an adjusted unemployment rate for February 2012 of 8.6% -- a substantial increase from the 8.3% adjusted rate the government reported for January.
The findings provide a preview of what Gallup will report in its monthly employment release next Thursday, March 8. Because Gallup’s data are collected continuously throughout the month, the data are available now, one week ahead of the BLS report scheduled for Friday, March 9.

Three key factors help determine the relationship between Gallup's measurement of the unemployment rate and the unemployment rate reported by the government. The first involves the relationship between Gallup’s and the government’s unadjusted survey results. Data from the past two years show that on an unadjusted basis, Gallup’s and the government's unemployment measurements track fairly closely in both direction and magnitude.
Gallup reported an unadjusted rate of 8.6% for January 2012 and the government reported an unadjusted rate of 8.8%. Gallup’s and the government's unadjusted results also tracked closely in January 2011 (9.9% versus 9.8%, respectively) and January 2010 (10.9% compared with 10.6%).

The results were similarly close in February 2010 (when Gallup found 10.6% versus the government’s 10.4%) but diverged significantly in February 2011, when Gallup reported an unadjusted rate of 10.3% and the government an unadjusted rate of 9.5%. That is, after the government re-based its unemployment data in January 2011, Gallup’s and the government's February results differed in both magnitude and direction.

Relating Gallup’s current findings to the likely unemployment rate the government will report for February depends on a number of somewhat complex key assumptions. Consider the following scenarios:
Based on its full-month measure, Gallup finds that the unadjusted unemployment rate increased by 0.5 points, to an average of 9.1% in February from 8.6% in January.
  • If we assume the government’s unadjusted unemployment rate experienced a similar 0.5-point increase, it would rise to 9.3% in February from January’s 8.8%. Applying the 0.5-point seasonal adjustment (based on the government’s February 2011 adjustment) to the February 2012 unadjusted rate (9.3%) would result in an increase in the U.S. seasonally adjusted unemployment rate to 8.8% in February 2012 from January’s 8.3%.
  • If we instead assume the government's unadjusted rate increased by 0.3 points in February to exactly match Gallup's measurement at 9.1%, this would produce an increase in the seasonally adjusted unemployment rate to 8.6% in February 2012 -- assuming the use of the February 2011 seasonal adjustment of 0.5 points.
Of course, there are lots of other possible assumptions. For example, the government’s unemployment estimate is based on a mid-month measurement. Gallup’s mid-month measurement for the unadjusted unemployment rate in January 2012 was 8.3% and it was 9.0% in mid-February. If we were to apply this mid-month increase of 0.7 points to the government’s unadjusted January rate of 8.8%, the resulting unadjusted 9.5% would lead to a 9.0% seasonally adjusted rate.

On the other hand, the government re-based its household survey once again in January 2012. If the government's unadjusted survey results are unchanged (8.8%) or go lower in February, the government's seasonally adjusted rate is likely to do the same.

A second factor to consider is the way the government seasonally adjusts its unadjusted unemployment rate for February. As noted previously, the government used a 0.5-point adjustment factor to seasonally adjust its results in February 2011. However, in 2010, it used a slightly different 0.6-point adjustment. Applying this larger seasonal adjustment to Gallup's full-month results would produce an 8.5% adjusted unemployment rate for February 2012 -- up from 8.3% in January.

A third factor that affects the unemployment rate might involve shifts taking place in the size of the workforce. If Americans seeking work get discouraged and drop out of the workforce in large numbers, it is likely to reduce the unemployment rate. This appears to have happened at various points in the past. Gallup's measurement of the participation rate -- the percentage of Americans in the workforce -- shows a decline to 67.6% in February from 68.2% in January. If the government shows a similar decline in the workforce, it is likely to lower the U.S. unemployment rate that the government reports and that may not be fully picked up by Gallup's data.

In sum, Gallup's Daily tracking of the U.S. unemployment rate -- which is not seasonally or otherwise adjusted -- shows an increase in February from January. This reflects a consistent measurement of the job-market reality that Gallup has measured since January 2010. It may be that various factors will lead the U.S. government on Friday, March 9, to release a different unemployment rate than that implied by Gallup’s results, but that report will not change the reality in the marketplace.

Copyright © 2010 Gallup, Inc. All rights reserved. | Terms of Use | Privacy Statement