An official website of the United States government

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Bureau of Labor Statistics > Employment Projections > Data

Labor Force Projections Evaluation: 2002-2012

Labor force projections begin with civilian noninstitutional population (CNIP) estimates from the U.S. Census Bureau, to which BLS applies our projected labor force participation rates. Errors in BLS labor force projections can therefore come from either source. For more information, refer to our evaluation methodology.

Limitations

In 2000 the Census added a new race category, "Two or more races," which limits BLS' ability to confidently evaluate the accuracy of our labor force projections by race. BLS projections were developed using the categories from before 2000¹, which are not comparable to the categories published in 2012.²

Measuring accuracy

How often did BLS correctly project growth and decline for labor force segments?

BLS correctly projected which labor force segments would grow and which would decline about 79 percent of the time.³

How much did BLS project the labor force to grow between 2002 and 2012?

BLS projected the labor force to grow 12 percent between 2002 and 2012.

How much did the labor force actually grow?

The labor force actually grew 7 percent between 2002 and 2012.

What contributed to the difference?

After the release of the 2002–12 projections, there was a recession which reduced employment and shifted labor force participation rates. In 2012, the labor force had fewer workers in the younger cohorts and more in the older cohorts than BLS projected.

Accuracy of the CNIP projections

The 2012 actual CNIP was similar to the projected CNIP, with an absolute difference of less than one percent, and relatively low dissimilarity indexes.

Table 1. CNIP dissimilarity indexes
Race	Sex	Dissimilarity index
All	All	1.0%
All	Male	1.3%
All	Female	0.8%

Accuracy of the labor force projections

In most cases the naïve model outperformed the BLS projections, as measured by lower absolute percent error. In almost all cases the naïve model over– or under–projected the labor force in tandem with the BLS projections. The actual 2012 labor force generally had fewer workers in the younger cohorts and more in the older cohorts than BLS projected.

Table 2. Absolute percent error by age group
Age	Actual 2012 labor force	Labor force projection		Absolute percent error		Best performer
Age	Actual 2012 labor force	BLS	Naïve	BLS	Naïve	Best performer
16 and 17	1952	2632	2840	34.8%	45.5%	BLS
18 and 19	3870	5005	4964	29.3%	28.3%	Naïve
20 and 21	5715	6222	6255	8.9%	9.5%	BLS
22 to 24	9747	10519	10407	7.9%	6.8%	Naïve
25 to 29	16792	17599	17779	4.8%	5.9%	BLS
30 to 34	16673	17807	17659	6.8%	5.9%	Naïve
35 to 39	15658	16491	16026	5.3%	2.3%	Naïve
40 to 44	17076	17943	17571	5.1%	2.9%	Naïve
45 to 49	17456	18506	18301	6.0%	4.8%	Naïve
50 to 54	17598	18520	18574	5.2%	5.5%	BLS
55 to 59	14908	15039	14905	0.9%	0.0%	Naïve
60 and 61	4721	4678	4504	0.9%	4.6%	BLS
62 to 64	5081	4900	4813	3.6%	5.3%	BLS
65 to 69	4427	3889	4056	12.2%	8.4%	Naïve
70 to 74	1917	1522	1641	20.6%	14.4%	Naïve
75 to 79	853	607	615	28.8%	27.9%	Naïve
80 and over	530	392	339	26.0%	36.0%	BLS

Examining the mean absolute percent errors and weighted mean absolute percent errors shown below in tables 3 and 4 reveals minor differences in the performance of the BLS projections and the naïve model. The substantially smaller weighted mean absolute percent errors indicate that the BLS projections and the naïve model both made smaller errors in the larger population cohorts.

Table 3. Mean absolute percent error, labor force
Race	Sex	Mean absolute percent error		Best performer
Race	Sex	BLS	Naïve	Best performer
All	All	12.2%	12.6%	BLS
All	Male	11.7%	11.6%	Naïve
All	Female	12.6%	13.8%	BLS

Table 4. Weighted mean absolute percent error, labor force
Race	Sex	Weighted mean absolute percent error		Best performer
Race	Sex	BLS	Naïve	Best performer
All	All	6.7%	6.1%	Naïve
All	Male	5.7%	4.8%	Naïve
All	Female	8.1%	8.0%	Naïve

Dissimilarity indexes provide another measure of accuracy, but as with the measures presented above, BLS projections performed similarly to the naïve model for the 2002–12 labor force projections.

Table 5. Labor force dissimilarity indexes
Race	Sex	BLS	Naïve	Best performer
All	All	1.8%	2.0%	BLS
All	Male	1.7%	1.9%	BLS
All	Female	2.0%	2.6%	BLS

Note

¹These categories are: White, Black, Asian, Hispanic, White Hispanic, and White Non-Hispanic.

²These categories are: White, Black, Asian, Two or more, Hispanic, and White Non-Hispanic.

³Male and female, all detailed age groups.

Return to Projections Evaluation Homepage

Last Modified Date: August 1, 2018