Emsi Logo

High-Tech Job Growth and the Value of EMSI’s Unsuppressed Data

May 1, 2013 by Joshua Wright

In December, the Bay Area Council Economic Institute published a report called “Technology Works: High-Tech Employment and Wages in the United States” and contributed to a county-level tech jobs map on behalf of Engine Advocacy. The report goes into detail on the growth of tech industries by state and metro, and how tech jobs lead to other jobs in other sectors of the economy. Economist Enrico Moretti, author of The New Geography of Jobs, hailed it as a “useful contribution to our understanding of job creation in America today,” and the Bay Area Council’s work received widespread media attention.

After one such news story, one of our clients was asked by a media outlet to comment on the surprising high-tech growth in their area. Immediately, the numbers in the Bay Area Council report didn’t look right, so our client asked us to look into the source data from the Bureau of Labor Statistics.

As it turns out, the job growth for our client’s metro area and many others was drastically overstated, and the root of the problem was suppressed data.

Background on Suppressions

Those who work with raw government employment data probably have felt the same frustration: For industries with a small number of establishments or regions with small populations, the employment or wages are not disclosed for confidentiality reasons. Instead, the job total or earnings show up as zero or “ND.”

Suppressed data can be a big pain, and it’s why EMSI has developed sophisticated algorithms to unsuppress government data, including the BLS’s Quarterly Census of Employment and Wages (QCEW).

EMSI’s algorithms use insights from multiple data sources and certain assumptions to replace suppressions with educated, bounded estimates. Many workforce development, economic development, and higher education practitioners who work at the local level have come to value and trust our unsuppressed estimates for specific industries and specific geographies (counties, MSAs, ZIP codes). Also among our diverse client base are state labor market offices, which can’t publish these figures but have regularly vouched for our estimates.

A Data Comparison

The Bay Area Council’s report showed the top metros for high-tech job growth from 2006-2011 and 2010-2011. For the more recent time frame, the council’s analysis ranked Greensboro-High Point, N.C., as the No. 1 metro for tech growth, at 36.3%. EMSI’s QCEW Employees dataset — which is an enhanced, unsuppressed version of the Quarterly Census of Employment and Wages — shows Greensboro’s tech industries growing by 14.4% during that time, a difference of 22 percentage points from the council’s numbers.

All told, 1,751 net jobs in these five tech industries can be explained by suppressions — not new growth. That’s a quarter of the total Greensboro tech workforce.

Why are the growth percentages so different? It boils down to suppressions.

In its QCEW dataset, the BLS suppressed the employment number for five tech industries in Greensboro in either fourth quarter 2010 or fourth quarter 2011, meaning those industries were reported as having zero jobs in the given quarter. For some small industries, this didn’t make a huge difference. But for semiconductor manufacturing (NAICS 3344), the employment total went from zero in Q4 2010 to 1,804 in Q4 2011.

GREENSBORO-HIGH POINT, N.C. -- SELECTED TECH INDUSTRIES
RAW QCEW DATA
NAICS CodeIndustry Description2010 Raw QCEW Jobs2011 Raw QCEW EmpDifference
3341Computer and Peripheral Equipment Manufacturing590-59
3344Semiconductor and Other Electronic Component Manufacturing01,8041,804
3345Navigational, Measuring, Electromedical, and Control Instruments Manufacturing099
5179Other Telecommunications170-17
5182Data Processing, Hosting, and Related Services01414
Total761,8271,751
Source: Bureau of Labor Statistics, QCEW

All told, 1,751 net jobs in these five tech industries can be explained by suppressions — not new growth. That’s a quarter of the total Greensboro tech workforce. In contrast, the same industries added 31 net jobs from 2010-2011, according to EMSI’s unsuppressed QCEW dataset, which reports job figures as annual averages.

From 2006-2011, the list of metros yields even more divergence between raw QCEW data compiled by Bay Area Council and EMSI’s unsuppressed QCEW dataset. Boise-Nampa, Idaho, with 82.9% job growth, tops the Bay Area list but is 345th (-26.1%) using EMSI data.* Augusta, Ga. (81.9% growth) is No. 2 according to Bay Area — and 15th (37.5%) according to EMSI.

Two South Carolina metros – Columbia and Charleston – also have wildly different outcomes between the suppressed and unsuppressed data. Bay Area Council reports both metros’ tech sectors grew around 40% from 2006-2011, good for fourth and fifth in its metro ranking. But, according to EMSI data, Columbia’s tech sector scarcely expanded over that time (0.5%) and Charleston’s grew by 67.5%.

*Note that Bay Area Council ranked the largest 150 metros, whereas the EMSI ranking used the complete list of 383 core-based statistical areas (CBSAs) with high-tech industries.

The following table provides the top 25 metros for tech growth from 2006-2011, as ranked by Bay Area Council, and how the percentages and rankings stack up with EMSI data.

MetroBay Area Council Change (2006-2011 % Growth)EMSI Change (2006-2011 % Growth)% DifferenceBay Area Council RankEMSI Rank
Boise City-Nampa, ID82.9%-26.1%-109%1345
Augusta-Richmond County, GA-SC81.9%37.5%-44%215
Peoria, IL41.0%20.7%-20%341
Columbia, SC40.1%0.5%-40%4147
Charleston-North Charleston, SC39.2%67.7%29%55
Little Rock-N.Little Rock-Conway, AR34.7%7.5%-27%686
Albany-Schenectady-Troy, NY29.9%9.6%-20%773
San Francisco-San Mateo-Redwood City, CA27.8%13.9%-14%856
Anchorage, AK27.2%8.5%-19%978
Ogden-Clearfield, UT25.6%25.7%0%1032
Madison, WI25.4%23.0%-2%1135
Lafayette, LA24.2%9.4%-15%1274
San Antonio, TX23.6%8.1%-16%1383
Sacramento--Arden-Arcade--Roseville, CA23.4%-4.3%-28%14191
Charlotte-Gastonia-Concord, NC-SC22.3%8.1%-14%1582
Davenport-Moline-Rock Island, IA-IL20.2%5.5%-15%16100
Mobile, AL20.0%4.6%-15%17108
Green Bay, WI20.0%13.4%-7%1858
Seattle-Bellevue-Everett, WA17.1%15.8%-1%1951
Dayton, OH16.0%-0.6%-17%20163
Evansville, IN-KY15.6%-7.7%-23%21233
Columbus, OH14.7%12.1%-3%2265
Canton-Massillon, OH13.0%-2.6%-16%23173
Raleigh-Cary, NC12.6%7.2%-5%2488
Wilmington, DE-MD-NJ12.4%2.6%-10%25125
Sources: BLS calculations, Bay Area Council Economic Institute; QCEW Employees - EMSI 2013.1 Class of Worker

Conclusion

Suppressed employment and earnings data can be a big stumbling block for researchers and local practitioners. The above examples show just how much year-over-year job change can be distorted if suppressions aren’t accounted for, which can lead to the wrong assumptions about a specific industry or region – or, worse, faulty decision-making. This is why EMSI has gone to great lengths to develop unsuppressed data estimates that can be trusted.

The EMSI data shown in this post comes from Analyst, our web-based labor market data and analysis tool. For more information on EMSI or our unsuppressed data, contact Josh Wright (jwright@economicmodeling.com)Follow us on Twitter @DesktopEcon.

Want to get notified?