«Moving to opportunity voluMe 14, nuMber 2 • 2012 U.S. Department of Housing and Urban Development | Office of Policy Development and Research ...»
In both exhibits, regional patterns of low unemployment rates and insurance compensation levels clearly overlap in the Great Plains states, but unemployment compensation extends farther down Exhibit 2 County Shares of Unemployment Insurance Benefits as a Percentage of Personal Income in 2009 for the Contiguous 48 States––(manual classification of location quotient breaks)
Because of the fat upper tail of the distribution, a class break was added to the map in exhibit 2 to better describe it. Note that the highest value in exhibit 1 is 2.75, but the highest value in exhibit 2 is 7.37.
to Texas and Louisiana. Exhibit 2 reveals that many states in the Mississippi River Valley have high unemployment compensation levels around city centers but have lower levels in rural areas.
Exhibit 1 indicates unemployment rates are lower or similar in rural areas. Unlike the patterns in exhibit 1, the Northeastern states in exhibit 2 show a number of counties in New York, New Jersey, Massachusetts, and, in particular, Pennsylvania with higher levels of unemployment compensation than the national level.
Author Ron Wilson is a social science analyst in the Office of Policy Development and Research at the U.S.
Department of Housing and Urban Development and an adjunct faculty member of the Geographic Information Systems Program at the University of Maryland, Baltimore County.
References Bureau of Economic Analysis (BEA). 2012. “Regional Economic Accounts.” Available at http:// www.bea.gov/regional/.
Bureau of Labor Statistics (BLS). 2012a. “Labor Force Statistics From the Current Population Survey.” Available at http://bls.gov/cps/.
———. 2012b. “Local Area Unemployment Statistics.” Available at http://www.bls.gov/lau/.
302 Graphic Detail Data Shop Data Shop, a department of Cityscape, presents short articles or notes on the uses of data in housing and urban research. Through this department, the Office of Policy Development and Research introduces readers to new and overlooked data sources and to improved techniques in using well-known data. The emphasis is on sources and methods that analysts can use in their own work. Researchers often run into knotty data problems involving data interpretation or manipulation that must be solved before a project can proceed, but they seldom get to focus in detail on the solutions to such problems. If you have an idea for an applied, data-centric note of no more than 3,000 words, please send a one-paragraph abstract to firstname.lastname@example.org for consideration.
Introducing the Ohio New Establishment Dynamics Data Joel A. Elvery Ellen Cyran Cleveland State University Abstract The Ohio New Establishment Dynamics (O-NED) data set tracks the number of establishments that first started employing people between the second quarter of 1997 and the first quarter of 2008 and measures the employment and payroll data for these new establishments. O-NED enables researchers to measure the growth trends of cohorts of new establishments for up to 5 years after the cohort’s birth. These data are the first publicly available data that document the growth rates of new establishments at the substate level. The finest unit of geography O-NED measures is a county. This article describes how O-NED is constructed and defines the variables included in the data. It closes with two examples of how researchers can use the data.
Cityscape 303 Cityscape: A Journal of Policy Development and Research • Volume 14, Number 2 • 2012 U.S. Department of Housing and Urban Development • Office of Policy Development and Research Elvery and Cyran Introduction Publicly available data are insufficient to answer a number of questions about the growth patterns of new establishments. The Ohio New Establishment Dynamics (O-NED) data set fills some of those gaps for Ohio. An establishment is a single physical location of a firm; firms can have one or many establishments. O-NED focuses on establishments during their first 5 years, a crucial shakeout period for new establishments. New establishments are grouped based on their year of birth and these groupings are called birth cohorts. O-NED provides annual tabulations of the number of establishments, employment, and wages for each birth cohort. The tabulations cover the period from April 1, 1997, through March 31, 2008. The data include 11 cohorts—7 for 5 full years and 4 for less than 5 years. O-NED provides separate tabulations for births unaffiliated with preexisting firms (entrepreneurial births), births affiliated with preexisting firms (other births), and fast-growing births (gazelles) of each type. O-NED is an outgrowth of the work of Knaup (2005), Knaup and Piazza (2007), Talan and Hiles (2007), and the U.S. Bureau of Labor Statistics (BLS) Entrepreneurship Team. This article describes the construction and structure of the data and provides two examples of research that the data enable.
Few data measuring business dynamics are publicly available. In 2004, BLS introduced the Business Employment Dynamics (BDM) data series. The quarterly BDM data series enables data users to measure the job creation and destruction and the establishment birth and death numbers that underlie the employment totals published in the Quarterly Census of Employment and Wages (QCEW). The BDM quarterly update is released 7 months after the quarter it covers. BLS continues to improve the BDM data by adding new features. One limitation of the BDM data is that the finest level of geographic detail is the state and the finest level of industry detail is the major sector.
Furthermore, it contains only measures of establishment births and deaths and does not shed light on the growth patterns and survival rates of new establishments over time.
In December 2008, the U.S. Census Bureau began releasing Business Dynamics Statistics (BDS), which tabulates annual job creation and destruction statistics by firm age and either firm size or initial firm size. The tabulations are available for the United States as a whole or by sector or state.
BDS and O-NED are similar because both measure the employment and number of establishments in businesses aged 1 to 5 years and enable analysts to track how employment and the number of establishments change over time. Several important differences exist, however. O-NED is designed for tracking the growth of new establishments in Ohio for their first 5 years, whereas BDS provides a more comprehensive set of job creation and destruction statistics for the nation as a whole. O-NED uses establishment age and measures firm age only for entrepreneurial establishments, for which firm age equals establishment age. BDS uses firm age, not establishment age, and includes more firm age categories. O-NED provides more geographic detail than BDS, but BDS includes tabulations by firm size and decomposes changes in employment and the number of establishments into the portions due to new entrants, continuing establishments, and exiting establishments. BDS covers the United States as a whole from 1977 through 2009, whereas O-NED covers only Ohio from 1997 through 2007.
304 Data Shop Introducing the Ohio New Establishment Dynamics Data Construction of the O-NED Data The microdata we used to create O-NED is a combination of the longitudinally linked QCEW microdata from BLS and the edited ES202 data housed at the Maxine Goodman Levin College of Urban Affairs (Levin College) at Cleveland State University.1 Both data sets cover only Ohio and are provided through a special partnership between Cleveland State University and the Bureau of Labor Market Information of the Ohio Department of Jobs and Family Services (ODJFS). BLS provided the longitudinally linked QCEW microdata to ODJFS for this project.
We combined the microdata sources to take advantage of edits that researchers at Levin College made over a period of years. In particular, O-NED took industry and geography codes from the edited ES202 data. One challenge in creating tabulations for cohorts of establishments is that the industry and geography codes of establishments can change over time. New establishments are especially likely to have code changes because some enter the data set with incomplete information, and BLS and ODJFS assign those establishments codes after they have gathered more information.
To minimize the effect of these code changes on our tabulations, we applied the last valid codes we had for establishments to the data for all quarters.2 The O-NED data cover establishments born from April 1, 1997, through March 31, 2008. The sample is restricted to private establishments that did not experience any identifiable splits or consolidations during their first 5 years. This restriction greatly reduced the volatility in the data, because most splits are not truly new establishments but are continuing establishments that changed how they report their data.
Most splits are identified using relevant comment codes, but some are identified based on substantial changes in the number of establishments affiliated with a single employer identification number (EIN). EIN is used to determine which establishments belong to the same firm. Based on careful exploration of the data, we developed a set of rules to identify these splits by finding EINs that simultaneously have increases in their number of establishments and unusually large decreases in the average size of their establishments. Most cases treated as uncoded splits are those for which the EIN had employment of more than 50 people in the birth quarter, the number of establishments grew from the quarter before the birth to the birth quarter, and the average employment of establishments affiliated with the EIN fell by 80 percent from the quarter before the birth to the birth quarter. It is harder to identify splits for EINs with few units or little employment, and, based on exploring the data, we developed a conservative formula to identify these small splits.3 To further reduce the problem of false births, we examined the data for about 250 large EINs in which it was unclear whether the EINs experienced splits or had an unusually large number of The ES202 is Ohio’s version of the QCEW microdata and is based on establishment data collected as part of the unemployment insurance system.
If an establishment has an invalid code, such as a county code of 999, we use a previous, valid code when possible. We assign cases that have only invalid codes that change a single invalid code for all quarters. Based on our work verifying a subset of the code changes, we believe that no more than 25 percent of the code changes lost by pushing back codes were valid changes.
See Elvery and Cyran (2010) for more details on this and other topics.
Cityscape 305Elvery and Cyran
births. We examined the data to see if the new establishments affiliated with the EINs had predecessors or if ownership changed, which would suggest they are false births. For most cases, we did not find conclusive evidence that they were false births and treated them as births. Even with the careful use of the data and hand checking of large births suspected to be false, it is likely that some false births remain in the data. Uncoded splits and ownership changes, which can appear to be births, are more prevalent for new establishments affiliated with preexisting EINs than for other new establishments. Therefore, we believe that entrepreneurial births are less likely to be false births than are nonentrepreneurial births (Elvery and Cyran, 2010).
One goal of O-NED is to demonstrate what can be created with existing BLS microdata. As such, we use definitions that are consistent with those BLS uses for BDM and those proposed by the Organisation for Economic Co-operation and Development (Ahmad, 2006). Exhibits 1 and 2 provide the precise definitions. A birth is any new establishment in the state, regardless of its ownership.
An entrepreneurial birth is the birth of a new firm, not an additional establishment of an existing
AWVSPY 91 ANNLWAGE as a percentage of the cohort’s employment in quarter 1 of 2001.
O-NED = Ohio New Establishment Dynamics.
Notes: O-NED tabulates the variables for each cell. A cell is a combination of a unit of geography, a unit of industry, a type of birth or gazelle, a cohort, and an age. The values come from the cell of Akron CBSA, manufacturing sector, entrepreneurial births, 1997 cohort age 4.
firm.4 A nonentrepreneurial birth is a new establishment affiliated with an existing firm. Year of birth is the year the establishment first entered the data. Survival to age t is defined as having positive payroll and employment for at least one quarter of the year that is t years after birth year.
Employment and wages are those covered by the unemployment insurance system of Ohio.
Although the QCEW data are updated quarterly, we annualize the data. We define a year as a set of four quarters, starting with the second quarter of a calendar year and ending with the first quarter of the following year. For example, establishments that first enter the data from the second quarter of 1998 through the first quarter of 1999 would be counted as part of the cohort born in
1998. This unit of time is chosen because the QCEW microdata register a disproportionate share of establishment births in the first quarter of the year. A portion of these establishments were likely actually born earlier, so keeping them with those born in the previous three quarters groups establishments by cohort more effectively than using calendar years would. Using four quarters enables more geographic and industry detail by increasing the number of establishments per data cell. Focusing on annual data also keeps the data tractable for a broad group of potential users.
An EIN is treated like a firm. An EIN of all zeros is sometimes given to new establishments until they report their permanent EIN. Therefore, we treat a birth with an EIN of all zeros as an entrepreneurial birth.