Revised Employment Numbers Show More Job Growth!
by: Mike Evans, BLS Program Supervisor

"Employment growth in 2000 was revised upward to 2.7 percent. The best showing since 1994, 1986, and 1981."

Wyoming’s job growth is looking better due to major revisions in employment numbers released in April 2000 (see article "Wyoming's Labor Market in 2000"). According to the new numbers from administrative data, job growth was much stronger in 2000 than originally estimated. Employment growth in 2000 was revised upward to 2.7 percent, the best showing since 1994, 1988, and 1981.

Reasons for Revisions

Every month, the Current Employment Statistics (CES), based on an employer survey, and Local Area Unemployment Statistics (LAUS), based on a household survey, provide preliminary data for the current month, revised data for the previous month, and benchmarked data from the previous year.¹ Yearly benchmarks are performed to adjust the data with more current information than was available initially.² Each month’s CES and LAUS estimates are eventually revised two more times in subsequent years.

Actual Revisions

Figures 1 and 2 show the difference between the preliminary and initial benchmarked estimates. If the estimates match exactly, the difference is zero (0). If the preliminary estimates were underestimated, the difference is above zero (0). Overestimated estimates are below zero (0). As reflected in Figure 1, the CES estimates are typically underestimated, so the majority of the differences are above zero (0). Initial benchmark LAUS estimates are usually closer to the preliminary estimates than CES estimates, but frequently tend to be overestimated (see Figure 2); however, Decennial Census numbers will push the numbers upwards and closer to the preliminary estimates. Overall, preliminary estimates for both CES and LAUS were revised and initially benchmarked by as much as 3.0 percentage points downward and as much as 3.5 percentage points upward. Normally, when the economy is slowing down, preliminary estimates are overestimated due to the time lag between the time of the actual turn in the economy and the time the data reflecting the turn are received. If the economy is growing, preliminary estimates are underestimated and are revised upward. Keeping that in mind, Figure 1 indicates that the economy was growing in all years except 1995 and 1996.

One major problem with using preliminary data, especially at the county level, is the comparison of over-the-year numbers to determine percent changes because preliminary numbers are usually compared to benchmarked numbers (i.e., apples to oranges). Figures 3 and 4 illustrate the difference between the revision from preliminary to initial benchmark and initial benchmark to final benchmark for both CES and LAUS. Making a comparison between the preliminary numbers from one year and the final benchmarked numbers several revisions later would result in substantially incorrect growth data. However, if you compare benchmarked to benchmarked or preliminary to preliminary data, you get a more accurate picture of the actual percent growth occurring. For example, labor force and employment were substantially revised up for Albany County for 2000 (see Table 4), along with 1998 and 1999, so over-the-year growth comparisons have changed dramatically.

CES and LAUS are Working To Reduce Revisions

The Bureau of Labor Statistics (BLS) is currently updating the CES sample survey to eliminate biased and inaccurate estimates, and LAUS is changing the way benchmarks are done. Trying to predict growth accurately is difficult. Therefore, to improve the accuracy of the benchmarks for the CES estimates, we revise the preliminary estimates by using covered Unemployment Insurance (UI) data (see related articles "Covered Employment and Wages for Third Quarter, Part 1: Standard Industrial Classification" and "Covered Employment and Wages for Third Quarter, Part 2: North American Industry Classification System), which has a time lag of only five months.

The LAUS program is also benchmarked every ten years to the Decennial Census numbers. Preliminary Census data for 2000 recently showed population in Wyoming had exceeded original estimates by more than 13,000 people. This population gain is approximately equivalent to the addition of another Lincoln County to our state in a ten-year period. In the next year or two, BLS will adjust labor force and employment numbers up substantially with the final benchmark, showing even more growth. As shown in Figure 2 during 1991 to 1993, the 1990 Census adjusted the LAUS employment numbers down, while total nonagricultural employment (CES) was adjusted up (see Figure 1). Part of the reason for the change in both sets of numbers is that one of the elements used to calculate the LAUS estimate is the CES estimate. In other words, revisions to the LAUS estimate are further compounded by revisions to the CES estimate. However, changes to the CES estimate only affect the monthly growth and seasonal patterns of LAUS, not annual average employment.

Conclusion

An unanswered question is what effect the revisions have on the conduct and understanding of economic and labor policy in the state.³ The preliminary data available provide neither an accurate nor an unbiased prediction of the revised and benchmarked data (i.e., at the time of the first release of the data, they do not provide an accurate picture of how the economy is actually performing). Using preliminary data rather than final benchmarked data misleads anyone trying to understand the historical relationship between the economy’s performance and economic policy decisions. In order to get more reliable numbers, users may want to wait for the final benchmarked data.

¹Both the Current Employment Statistics (CES) and Local Area Unemployment Statistics (LAUS) programs are managed and funded by the U.S. Bureau of Labor Statistics (BLS).

²2Yearly benchmarks on Current Employment Statistics (CES) data are performed using employer tax records from the covered Unemployment Insurance (UI) database. Benchmarks for Local Area Unemployment Statistics (LAUS) are calculated using a variety of sources and methods including the Current Population Survey (CPS), the Current Employment Statistics (CES) estimates, and population controls.

³David Runkle, "Revisionist History: How Data Revisions Distort Economic Policy Research," Federal Reserve Bank of Minneapolis Quarterly Review, Fall 1998, p. 3.

Table of Contents | Labor Market Information | Wyoming Job Network | Send Us Mail

These pages designed by Gayle C. Edlin.
Last modified on by Valerie A. Davis.