Volume 8, No. 3
Fall 2002

Go Home

Featured Articles:

The Economy is Growing But The Labor Market Has Yet To Rebound

Non-Farm Employment Estimates for Small Areas (Part 2)

Personnel Supply Industry Suffers After Years of Prosperity

Archives

Non-Farm Employment Estimates for Small Areas (Part 2)

by Joe Malcolm

The estimating system known as the Small Domain Estimator (SDE) has been used to produce monthly data for smaller, non-metropolitan, areas since January 1999. The SDE was developed to meet our customers’ needs for non-farm employment data in smaller areas. This development stemmed from a cooperative effort by Current Employment Statistics (CES) and the Labor Market Economists (LMEs) of the Economic Information and Analysis (EI&A) Division, in partnership with the U. S. Bureau of Labor Statistics (BLS) and University of Chicago’s National Opinion Research Center (NORC).

A previous article, in the Fall 2001 issue of the Illinois Labor Market Review, dealt with the implementation of the SDE. It is available online at http://lmi.ides.state.il.us/lmr/ieesa.htm. Illinois’ CES has continued the production of employment statistics at the county level for Labor Market Areas (LMAs) and this report will focus on their review.

With over three years of production, the opportunity presented itself to compare and analyze the estimates produced by the SDE with the ES-202 universe employment counts. ES-202 is a Federal/State cooperative program which collects and compiles employment and wage data for workers covered by State Unemployment Insurance (UI) laws and federal civilian workers covered by Unemployment Compensation for Federal Employees (UCFE). ES-202 serves as “a near census of monthly employment¹,” and for that reason is used as a benchmark with which to compare the accuracy of estimated data, such as CES and SDE. The benchmark data used in this article spans a period from April 2000 to June 2001.

Each year the CES estimates are revised to reflect the ES-202 universe employment counts. ES-202 data is collected quarterly and lags about six months behind the CES estimated data. CES is a Federal/State cooperative effort as well, which uses a sample drawn from a subset of the ES-202 universe to calculate employment statistics by industry and area. The estimate period that will be evaluated in this article spans April 2000 through June 2001, the latest ES-202 benchmark period.

Of Illinois’ 102 counties, 23 counties are, by themselves or clustered with other counties, known as Metropolitan Statistical Areas (MSAs). The remaining 79 are, by themselves or clustered with other counties, known as Labor Market Areas (LMAs). Employment estimates for the 79 LMA counties, which are typically smaller than the MSA counties, are being produced by the SDE.

For comparisons of the total non-farm employment of the Non-MSA Counties, we will look at the differences between the SDE and the benchmark data, (see Chart 1). The first month of SDE data, April 2000, represents the biggest employment difference but SDE is within 1% of the benchmark. The differential declines from April 2000 through July 2000 while the over-the-month fluctuations between the two coincide. From July 2000 through March 2001, the differences remain constant as the SDE stays within 1/2 of a percent of the benchmark for each month with the exception of September 2000 and January 2001 where it hits 0.06% and 0.08%, respectively. November 2000 had the smallest percent difference of the entire benchmark period with a value of 0.006%. Again, the SDE’s predictions parallel the benchmark, with two exceptions: the July-August and September-October changes. During the final months, the SDE remains within 1/2 of a percent of the benchmark, until the last month, where it decreases while the benchmark stays flat. The final month, June 2001, has the third largest percent difference of 0.07%.

For an overview of the individual counties, we will look at a comparison between the total employment of each SDE and benchmark County, (see Chart 2). Of the 79 County Employment Estimates:

· 18 were within 1% of the benchmark
· 43 were within 2% of the benchmark
· 71 were within 4% of the benchmark
· All were within 8% of the benchmark

For greater detail, we can look at the distribution of the SDE Counties in a histogram, (see Chart 3). As before, we are comparing the percent difference between the total employment of each SDE and benchmark county. The largest group of SDE Counties falls into the 0.96% to 1.83% interval, consisting of 29 counties. The second largest group falls into the 0.08% to 0.95% interval, consisting of 24 counties. From the cumulative percent distribution (represented by the line graph in Chart 3) we notice that 69.2% of the SDE Counties have a total employment within 1.83% of their benchmarked total employment. In the middle range, we find the 1.84% to 2.70% and 2.71% to 3.58% groups, consisting of 8 and 9 counties respectively. Looking at the cumulative percentage, we notice that 89.7% of the counties are accounted for at the 2.71% to 3.58% difference mark. Moreover 71 of the 79 SDE Counties are within 3.58% of their benchmarked data. Of the remaining 8 counties, half fall into 4.47% to 5.33% interval. The one SDE County in the greater than 6.21% or ‘More’ group differs by 7.09% from the benchmark data. The average difference of the counties was 1.70% with a standard deviation of 1.44% (About three-fourths of the observations are between 0.26% and 3.14%).

Exploring a benchmark period and comparing estimates with universe counts measures the usefulness of the estimation outcomes. Further research investigating the similarities and differences between Non-MSA, MSA, and Statewide areas with respect to employment change may be warranted. With regard to this benchmark experience, the overall results obtained are encouraging.

¹ ‘Covered Employment and Wages’, U.S. Department of Labor Bureau of Labor Statistics <http://www.bls.gov/cew/cewover.htm>, October 18, 2001.