[CTPP] Reconciling Total Employment and Workers-at-Work (long post)

TO: CTPP-News listserv FR: Chuck Purvis RE: Reconciling Total Employment and Workers-at-Work One of the many data issues relevant to the release of the county-to-county total worker flow data is the reconciliation of independent estimates of employment (jobs) with the Census 2000 workers-at-work data. My executive director saw a recent article in the Los Angeles Times about the mismatch between census data and employment and labor force data, and was concerned if this is an issue in our region. My short answer to my ED was, no, the data problems faced in New York and Los Angeles are not that serious in the Bay Area. I also provided him the "long answer" which may prove useful to other metro areas interested in reconciling different data sets. The bottom line is that estimates of TOTAL EMPLOYMENT (i.e., jobs at area of work) SHOULD BE about 7 to 9 percent HIGHER than Census 2000 estimates of workers-at-work (i.e., workers at area of work). It is important to understand that there are important definitional differences between total employment and decennial census-based workers-at-work. Our indepedent estimate of total employment, year 2000, is 3,753,700 total jobs; Census 2000 data on workers-at-work is 3,396,800. This shows that our total employment, unadjusted, is 10.5 percent higher than our census-based workers-at-work. This is a really big difference! AFTER ADJUSTMENTS, the difference between our total employment (adjusted) and census-based workers-at-work is 1.1 percent. There are three main ADJUSTMENTS that are needed to make TOTAL EMPLOYMENT data comparable to census workers-at-work data: 1. Seasonal fluctuations in employment adjustments; 2. Multiple jobholding adjustments; and 3. Weekly absenteeism adjustments. 1. Seasonal fluctuations in employment adjustments. Typically, total employment data is reported on a monthly average as well as an annual average basis. Our independent estimates of total employment are annual averages, and could be adjusted to approximate the number of jobs in the region as of March 2000. In the Bay Area, our state Employment Development Department shows 3,479,100 jobs (wage and salary employment) in March 2000, compared to 3,540,800 jobs for the year 2000 annual average. So, a factor of 0.983 (ratio of March 2000 to Annual Average) is used to adjust the annual average jobs to a March 2000 estimate of jobs. (As a sidenote, employment levels in the Bay Area peaked at 3,636,500 jobs in December 2000, and we lost 318 thousand jobs in 24 months, and we are now at 3,318,700 jobs.) [Note that our state EDD data is based on wage and salary employment data collected in ES-202 records (employment securities, unemployment insurance data provided by employers to every state). Our sister regional agency, ABAG, adjusts the wage and salary employment data to include the "not covered" employment, including: self-employed, unpaid family workers, private household workers, and individuals on unpaid leave from work. EDD shows 3,540,800 Bay Area wage and salary jobs; ABAG shows 3,753,700 total jobs. The California EDD web site is: http://www.calmis.cahwnet.gov/] 2. Multiple jobholding adjustments. The decennial census only collects data on the primary job and the primary workplace location of every worker answering the census long form. It does not collect data on secondary (moonlighting) jobs held by workers. The Bureau of Labor Statistics (BLS) does, however, collect data on multiple jobholding from the Current Population Survey (CPS). Two reports of interest to our community are available from the BLS web site, at: Multiple jobholding by state, 1999-2000: http://stats.bls.gov/opub/mlr/2001/07/rgtrends.pdf New Data on Multiple Jobholding Available from CPS http://stats.bls.gov/opub/mlr/1997/03/art1exc.htm Nationally, 5.6 percent of jobs are the second, or "moonlighting" jobs of resident workers. These multiple jobholding shares range from 3.9 percent in Florida to 10.3 percent in Nebraska. [Can someone check my interpretation of BLS statistics? Is it 5.9 percent of jobs that are second jobs, or is it 5.9 percent of employed residents who have second jobs?] So, the secondary adjustment is to divide your March 2000 total employment estimate by 1.056 (or state-equivalent value) to prepare an estimate of "March 2000 primary jobs." 3. Weekly absenteeism adjustments. The decennial census asks whether the worker was AT WORK during the census reference week (typically thought of as the last week of March 2000.) The difference between employed residents (employed civilians plus armed forces workers) and worker-at-work (persons reporting a workplace and a means of transportation to work) is typically called "weekly absentees." Nationally, 2.0 percent of employed residents were absent from work during the census reference week. This "weekly absenteeism rate" is easily computed using readily available data from Census 2000 Summary File #3. The disconnect in this methodology is that the weekly absenteeism rate is for area-of-residence, and that adjustments are basically needed for area-of-work data. My method was to use the county-of-residence weekly absenteeism rate (which ranges from 1.87 percent in Santa Clara County to 2.48 percent in Solano County) and apply these to our estimates of total jobs by county-of-work. Or, you can use a regional or statewide weekly absenteeism rate as the basis for this adjustment. So, the third adjustment is to REDUCE the "March 2000 primary jobs" by about 2.0 percent to account for weekly absentees. The results is "March 2000 workers-at-work." The adjustments for the Bay Area region are as follows: 3,753,700 = ABAG Total Employment, Annual 2000 3,688,300 = ABAG Total Employment, March 2000 Adjusted (Step #1) -177,000 = Estimate of Secondary Jobs (Step #2) -77,300 = Estimate of Weekly Absentees (Step #3) 3,434,000 = Estimate of ABAG Total Workers-at-Work in Bay Area, March 2000. 3,396,800 = Census 2000 workers-at-work. Difference between Independent Job Estimates and Census Journey-to-Work is effectively 1.1 percent. Not too bad at a regional level, but differences range from -10.2 percent to +6.3 percent at our county level. Additionally, adjustments to independent employment estimates may be needed to prepare more refined estimates of the self-employed and the unpaid family workers once data from the 1-percent and 5-percent PUMS data is available. **************************** Another data issue in preparing Census 2000 workers-at-work data for comparison purposes is the treatment of out-of-town workers (i.e., business persons on business trips to or from your region.) In the Bay Area we have 20,900 resident workers who are out-of-town on business during census reference week; and 40,200 out-of-town residents working in our region during the census reference week. My assumption is that the 40,200 out-of-town residents (from Los Angeles, Maricopa Counties, etc.) DO NOT HAVE REGULAR JOBS in the BAY AREA, and I do not include them in my "workers-at-work in Bay Area" estimates. I DO consider the Bay Area resident workers who are on business trips to Los Angeles, Phoenix, D.C., New York, etc., TO HAVE REGULAR JOB LOCATIONS IN THE BAY AREA, so I include these 20,900 resident workers in my "workers-working-in-Bay Area" estimates. [You may or may not agree with this interpretation. That's OK.] **************** The last data issue is to make certain you have a chance to read the Census Bureau's concerns about group quarters labor force issues. Please read the "Summary File 3: Data Note #4 - Updated December 2002" on page 960 of the SF3 documentation: http://www.census.gov/prod/cen2000/doc/sf3.pdf The Bureau is finding a "college town" issue in terms of patterns of missing responses to variables on the individual census form (group quarters questionnaire) and that the possible impact is a nationwide underestimate of 235 thousand employed persons, out of 129.7 million employed persons. ****************************************** Well, I hope this e-mail piques your interest, and I hope that I'm not too far off-base in terms of how to adjust and interpret all of this data. [Perhaps we can re-write this and include it in a future edition of the CTPP Newsletter?] If and when responding to my e-mail, be sure NOT to include my full e-mail in your response! Thanks! Chuck Purvis, MTC ************************************************************** Charles L. Purvis, AICP Principal Transportation Planner/Analyst Metropolitan Transportation Commission 101 Eighth Street Oakland, CA 94607-4700 (510) 464-7731 (office) (510) 464-7848 (fax) www: http://www.mtc.ca.gov/ Census WWW: http://www.bayareacensus.ca.gov/ **************************************************************