Professional employer organizations: what are they, who uses them, and why should we care?
Lombardi, Britton ; Ono, Yukako
Introduction and summary
A growing number of U.S. workers are counted as employees of firms
that they do not actually work for. Some of these workers are from
temporary help services (THS) agencies and some are leased employees,
who are on the payrolls of professional employer organizations (PEOs)
but work for PEOs' client firms. Several studies have looked at
firms' use of THS, but few have examined the use of PEO services.
In this article, we use data from the U.S. Census Bureau to shed some
light on PEOs--how they operate, what types of companies employ them,
and why.
PEOs provide various services related to human resources (HR)
management, such as payroll processing, benefit management, and
regulation compliance. Unlike consultant firms that only provide
recommendations on these functions, PEOs operate in a co-employment
relationship with their clients, by including the clients' workers
on their own payrolls. In such a relationship, PEOs become employers of
record for tax and insurance purposes. PEOs exercise some
decision-making in HR management; at the same time, they share legal
responsibilities as co-employers. By pooling the workers of its clients
on its payroll, a PEO gains economies of scale in performing its
required tasks.
The workers whose payrolls are moved to PEOs are often referred to
as "leased employees" because, on paper, they work for the PEO
and are leased back to the client firm. Since leased employees are not
accounted for on clients' payrolls, the payroll-based labor
statistics underestimate labor used by the industries of PEO client
firms. In the 2002 Economic Census's subject series on
Administrative and Support and Waste Management and Remediation
Services, (1) the PEO industry consisted of about 5,000 establishments.
It employed 1.7 million leased employees. The PEO industry achieved
rapid growth through the 1990s, with a growth rate of 386 percent from
1992 to 2002, subtracting an increasing number of workers from the
payroll records of other industries. (2)
In this article, we first review the history and current nature of
PEO services. Next, we use publicly available data to show that the
distribution of the use of PEO services is not uniform across industries
or geographical areas. We then use confidential microdata from the U.S.
Census Bureau's 2002 Census of Manufactures to examine how
characteristics other than location and industry may influence
companies' use of PEO services and, therefore, why certain types of
companies are more likely to use PEO services than others. This also
sheds some light on the issues researchers face in interpreting
payroll-based labor statistics.
Dey, Houseman, and Polivka (2006) provide a review of the issues
related to payroll-based labor statistics. They also assess the effect
of firms' use of employment services as a whole, including PEO
services and THS, focusing on the manufacturing sector. Between 1989 and
2000, employment in manufacturing reportedly fell by 4.1 percent. Dey,
Houseman, and Polivka (2006) show, however, that manufacturing
employment would have actually increased by 1.4 percent if employment
services workers had been included. (3) They also estimate that the use
of these employment services added 0.5 percentage points to the annual
growth rate of labor productivity as measured by output per worker in
the manufacturing sector between 1989 and 2000, equaling approximately
14 percent of the overall growth. (4) Houseman (2006) also shows that
the multifactor productivity measure for manufacturing would also
overestimate productivity growth as the data do not allow us to fully
capture employment services input to manufacturing.
The existing literature (Houseman, 2006; Estavao and Lach, 1999;
and Segal and Sullivan, 1997) relies mostly on data on firms' use
of THS industries, partly reflecting data availability. We complement
this literature here by examining firms' use of PEOs.
History and activities of PEOs
PEOs started out in the early 1980s, conducting payroll processing
for client firms. Payrolling involved preparing and distributing payroll
checks; depositing wages directly to bank accounts; maintaining payroll
data; filing local, state, and federal government paperwork; and
tracking vacation and sick leave. To perform such services, the PEO
grouped its client firms' workers on the PEO's own payroll and
processed tasks at the same time. Small- and medium-sized companies, in
particular, benefited through cost efficiencies gained from the
PEO's economy of scale.
Outsourcing payroll processing to PEOs, however, caused some
confusion about the employer status of the PEO versus that of the client
firms (Drucker, 2002; and Greening, Barringer, and Macy, 1996). Having
transferred clients' workers to the PEO's payroll, PEOs
appeared as employers on paper. Some client firms took advantage of this
confusion about employer status (5) and tried to reduce their legal
responsibilities (Houseman, 2003). As these practices became more
prevalent, regulatory agencies and insurance companies tried to clarify
the situation by creating new regulations and policies. As a result, the
PEO officially became accountable for the performance of HR
responsibilities as a joint employer under contractual agreement with
its client firms, in essence acting as an outsourced HR department for
client firms (Klaas, McClendon, and Gainey, 2000).
As regulations affecting HR management have increased over time,
the kinds of services that PEOs provide have also expanded. Between 1980
and 2000, the number of employment laws applying to employers grew by 60
percent, and between 1991 and 2001, the number of lawsuits, in
particular sexual harassment cases, more than doubled, according to the
U.S. Equal Employment Opportunity Commission (Drucker, 2002). The growth
in the number of regulations and lawsuits has generated higher time and
monetary costs for firms and increased the firms' liabilities to
both their workers and enforcement agencies. Adding to the complexity of
HR management, some regulations have different rules and enforcement
requirements based on firm size (6) and/or location. (7) In addition to
the changing regulatory landscape, the increasing cost of
employment-based benefits, especially health care, continues to add to
firms' administrative costs (Bodenheimer, 2005). Reflecting these
changes, PEOs began to expand their services to further support the
management of their clients' work forces with such duties as
ensuring compliance with regulatory issues, as well as providing and
administering benefits packages (Cook, 1999).
In addition to the aforementioned HR tasks, these days many PEOs
offer additional HR activities to provide a more integrated overall HR
management service, including relocation administration, employee
handbooks and background checks, physicals, and job descriptions
(Gilley, Greer, and Rasheed, 2004; and Cline, 1997). Some PEOs also
provide potentially high-liability HR management functions to
differentiate themselves from the rest of the market. One of the more
complex tasks that firms outsource to PEOs is the administration of
their retirement plans, which have intricate requirements to be
compliant with the Employee Retirement Income Security Act (ERISA) (8)
(Greet, Youngblood, and Gray, 1999). Some PEOs also support the
administration of Employee Assistance Programs (EAPs), (9) which provide
support services to client firms' workers and their families
(Greer, Youngblood, and Gray, 1999). Some firms outsource the
responsibility of processing drug testing to PEOs to minimize
confidentiality issues regarding personal employee information (Greet,
Youngblood, and Gray, 1999). And some firms also use PEOs to facilitate
the centralization of HR functions (Greer, Youngblood, and Gray, 1999).
Since PEOs manage various HR and regulatory issues as joint
employers, it would be instructive to summarize a typical contractual
relationship between a PEO and a client firm. First, to define the joint
relationship, both the PEO and the client firm enter into a contract to
document which firm takes on the legal and administrative
responsibilities of the firm's employees (Lenz, 2003, p. 10). Under
this agreement, the client firm purchases the PEO's assistance by
compensating the PEO an amount that covers the client's total human
resources costs plus an additional service fee. Many times when a PEO
agrees to administer payroll and benefits to the workers, the PEO also
becomes responsible under the law for the liabilities associated with
these administrative duties. Such duties include issuing workers'
compensation for employees accidentally injured on the job. (10) Health
and pension benefits that some PEOs offer fall under another set of
state regulations (Lenz, 2003, p. 10). (11)
While a PEO plays a significant role as a joint employer as
mentioned previously, its role and responsibilities are limited to those
involving HR management. The PEO does not provide daily supervision to
workers for their production activities, in addition, it does not
typically get involved with interviewing and hiring. (However, it may
offer guidance on job postings and skill matching, and it may take care
of regulatory issues or conduct basic functions such as background
checks and drug tests.) Thus, it would be natural to consider leased
employees on the PEO's payroll as part of the work force for the
production activity of the PEO's client firm. However, the
payroll-based labor statistics do not take this into account, and the
greater use of PEO services by an establishment or an industry would
cause the underestimation of the labor used for its production. In the
next section, we examine the distribution of the use of PEO services
across industries and geographical areas.
Cross-sectional distribution of the use of PEO services
In what follows, we use publicly available data of leased employees
from the U.S. Census Bureau. In particular, we study the distributions
of leased employees versus payroll employees across clients'
industries and geographical areas, using data from the 1992, 1997, and
2002 Economic Censuses. Note that another source often used for
employment data is the Current Employment Statistics. However, there is
a concern about using the CES data for our purpose. The CES's
sampling frame, the Quarterly Census of Employment and Wages, seems to
be undergoing changes regarding whether or not the leased employees are
counted in the PEO industry or client firm's industry, and the
practice varies across states (Dey, Houseman, and Polivka, 2006). Here
we focus on using the Economic Census data for consistency to make
year-to-year, industry, and location comparisons.
Based on the 1997 Economic Census's subject series on
Administrative and Support and Waste Management and Remediation
Services, (12) table 1 shows that the intensity of use of PEO services
varies across industries. The first column shows the number of leased
employees used in each industry as reported by PEOs. The second column
shows payroll employment by industry, which does not include leased
employees or THS workers. (13) The third column shows the share of
leased employees of total workers who work on a regular basis for an
industry; we divide the number of leased employees (first column) by the
sum of the leased and payroll employees by industry (the first column
plus second column). The transportation industry uses leased employees
most intensively. Leased employees represent 4.6 percent of employees
working regularly for this industry. Transportation is followed by
repair services with 2.9 percent, educational services with 2.3 percent,
and construction with 1.8 percent, while mining has a very low share of
leased employees with 0.2 percent. The index values in the fourth column
show the intensity of use of leased employees in each industry relative
to the U.S. average (0.84 percent). The intensity of use of PEO services
seems to vary a lot across industries. The transportation industry uses
leased employees at a rate a little over 20 times that of mining. The
transportation industry also represents the highest share of national
total leased employees with 15.2 percent (fifth column). While various
reasons would explain the intensive use of leased employees in the
transportation industry, one factor may be a high injury rate reported
by the transportation industry; (14) this high injury rate may prompt
firms in that industry to seek more efficient or lower-cost ways to
insure their workers.
To see whether industry distribution of leased employees changes
over time, we make a comparison between the distributions in the 1992
and 1997 Economic Censuses. (15) Taking into account that the intensity
of use of leased employees varies over time nationally, we compare the
index values as calculated in the fourth column in table 1 in both
years. The industry categories included in the questionnaires of the
1992 and 1997 subject series on Administrative and Support and Waste
Management and Remediation Services are not identical, so we only
compare the index values for the industry categories that are common to
both years, These industries include mining; construction;
manufacturing; wholesale trade: retail trade; and finance, insurance,
and real estate. We found a similar pattern in the use of leased
employees in both years. The index values for 1992 show that the
construction sector used leased employees 1.6 times more intensively
than the U.S. average (0.4 percent), whereas in 1997, the multiple for
the construction sector was 2.1 times the national average of 0.84
percent. For manufacturing, the index was 0.72 for 1992 and 0.73 for
1997. For retail trade, the index was 0.48 for 1992 and 0.49 for 1997.
Next, we examine whether the intensity of use of PEO services
differs across geographical areas. Table 2 illustrates the variation
across states. The first four columns show 1997 data, while the fifth
column shows 2002 data. The first column shows the number of leased
employees reported by PEOs located in each state. The second column is
the total payroll employment of all private industries in each state,
which is often used as state employment in economic research. Unlike
industry payroll employment (the second column in table 1), the state
payroll employment number includes both leased employees and THS workers
on the payrolls of PEOs and THS agencies located in each state. The
third column shows the leased employee share of state payroll
employment. Of the top five states with the highest percentage of leased
employees, Florida comes in first with 3.6 percent, followed by Arizona
with 3.3 percent, Utah with 2.3 percent, Georgia with 2.1 percent, and
Texas with 1.4 percent. One might think that these states have higher
shares of industries that tend to use more PEO services. This is not
necessarily the case. We evaluate this potential explanation for Florida
and Arizona. To isolate the effect of the industry mix, we assume that
the use of leased employees by industry for Florida and Arizona is the
same as that for the U.S.; we then take the weighted average of the U.S.
shares of leased employees (the third column of table 1) and weight it
by each industry's share in state employment. Implied shares of
leased employees in Florida and Arizona are similar to the U.S. average
and not nearly as high as the numbers in the third column of table 2.
The industry mix does not explain the high share of leased employees in
Florida and Arizona. There may be some location-specific factors that
could further explain these states' use of PEO services.
In order to view the geographical distribution of leased employees
over time, we again calculate an index, dividing a state's share of
leased employees by the U.S. average, and compare the index's
values for 1997 and 2002. (16) We observe similar patterns of
geographical distribution across years. Out of the states with data
disclosed for both years, seven out of the top ten states with the
highest use of leased employees in 1997 remained in the top ten in 2002.
Seven out of ten of the states with the lowest use of leased employees
in 1997 remained in the bottom ten in 2002.
Finally, we note that the patterns of distribution in the use of
PEOs across industries and states are different from the patterns for
THS workers. Using the contingent work supplement of the 1997 Current
Population Survey, we calculate each industry's share of THS
workers. Among the industries using higher shares of THS workers are
manufacturing (31.8 percent) and administrative and support services and
waste management (21.3 percent). This contrasts with the industry
distribution for leased employees, where the transportation industry
represents the highest share at 15.2 percent and the manufacturing and
construction industries represent about 12 percent each. The different
geographical distributions between leased employees and THS workers are
summarized in table 3. The top and bottom ten states are quite different
for leased employees versus THS workers. Only four states appear in the
top ten and three in the bottom ten for both leased employees and THS
workers. The intensity of use of leased employees is also more varied
across states, reflected in the larger range of index values between the
top and bottom states compared with the results for THS workers.
Characteristics of manufacturing plants that use PEOs
In this section, we summarize how the use of PEO services varies
across establishments depending on their characteristics. In particular,
we use the establishment-level data of the 2002 Census of Manufactures
compiled by the U.S. Census Bureau. In the census's questionnaire,
a representative of a plant (that is, a manufacturing establishment) is
asked to answer "yes" or "no" to a question on
whether the plant uses any leased employees; thus, a plant with a
representative that answered "yes" uses workers whose payroll
is managed by a PEO. Prior to 2002, the U.S. Census Bureau only
collected information about PEOs' client firms by asking PEOs about
their clients as part of the Economic Census. To obtain a more detailed
picture of where PEO "employees" actually work, the U.S.
Census Bureau attempted to collect information directly from PEO users
by including questions about their PEO use in the 2002 Economic Census
for the first time.
In this article, we examine through probit analyses which plant
characteristics are associated with a plant's likelihood to use any
amount of PEO services. (17) Our analyses suggest that some plant
characteristics play important roles, even after controlling for a
plant's industry and the plant's location-specific or
state-specific factors. Note that it is possible that the firm rather
than the plant decides whether or not to use PEO services. Even if a
firm were the decision-maker, however, its decision may be made for each
of its plants based on each plant's individual characteristics, in
fact, based on our data, the use of PEO services tends to vary across
plants within the same firm. We explore the effects of both plant-level
and firm-level variables. Note that a plant or an establishment is the
smallest unit for which individual responses are collected in most of
the Economic Censuses, in the sense that the U.S. Census Bureau
typically creates industry-level or state-level data by aggregating
establishment-level data. Establishment-level analyses inform us of
establishment attributes that may help in the interpretation of such
aggregate data.
In our analyses, we include various plant characteristics. One such
variable is plant size measured by the log value of shipments. Larger
plants seem to face more regulations; for example, the Worker Adjustment
and Retraining Notification Act applies to businesses with 100 or more
employees. The federal law states that a firm must provide written
notice of plant closings or massive layoffs, defined as 50 or more
employees at a single establishment, 60 days in advance. Facing more
regulations, larger plants may rely on PEOs to comply with relevant
regulations. However, larger plants may have more economies of scale in
complying with regulations. We include the squared term of plant size to
allow a quadratic relationship between plant size and the plant's
probability of using PEO services.
Note that, as we mentioned previously, the decision to use PEOs may
be made by a firm rather than a plant. The relevant economies of scale
for performing the HR management services may be at the firm level
rather than the plant level. Therefore, we also examine how firm size is
associated with a plant's use of PEO services. We include the total
value of manufacturing shipments of the parent firm with which each
plant is affiliated. (18) It is not appropriate to use the number of
employees as a measure of plant or firm size because the number of
employees reported in the Census of Manufactures does not usually
include leased employees and is endogenous to the plant's use of
PEO services. (19)
We include a dummy variable indicating newly constructed plants. It
is possible that start-up plants may want to use PEOs to outsource any
noncore HR activities until their businesses take off. We also include
the average rates of work-related injury and illness at four- or
five-digit North American Industry Classification System levels provided
by the U.S. Department of Labor's Occupational Safety and Health
Administration (OSHA). OSHA collects such information in order to
provide reliable data to employers, policymakers, and health and safety
specialists to help determine priorities of workplace safety.
Establishments are asked to report all injuries and illnesses of all
workers on site.
We examine the effects of firm characteristics other than firm size
as well. Two additional firm-level variables are the firm's degree
of diversity across locations and across industries. A firm that has
plants in multiple states would face different regulations in each
state. A firm producing multiple products would also have to deal with
various regulations. A diversified firm may rely on a PEO to take
advantage of the PEO's economies of scale to keep up with all the
regulatory updates within different states and/or industries. For a
plant affiliated with a firm with at least one other plant, we measure
the geographical diversification by the number of states where the firm
has manufacturing plants, as well as the Herfindahl-Hirschman Index
(HHI) (20) based on the firm's manufacturing shipments by state.
The HHI is the sum of the squared terms of each state's share; we
define firm i's HHI for geographical concentration as
[HHI.sup.states.sub.i] [equivalent to] [summation over (i [member of]
[A.sub.i])] [share.sup.2.sub.is], where [share.sub.is] is the share of
state s in firm i's total value of manufacturing shipments, and
[A.sub.i] is a set of states where firm i operates. The HHI is greater
when the market concentration is higher. Analogously, to measure
industrial diversification, we calculate the number of industries
(three-digit NAICS manufacturing industries) in which the firm's
plants operate and the HHI based on the firm's value of shipments
by each of its three-digit NAICS manufacturing industries; we define
firm i's HHI for industry concentration as
[HHI.sup.industries.sub.i] [equivalent to] [summation over (i [member
of] [B.sub.i])] [share.sup.2.sub.ij], where [share.sub.ij] is the share
of industry j in firm i's total value of manufacturing shipments,
and [B.sub.i] is a set of manufacturing industries in which firm i
operates.
As panel A of table 4 shows, our sample contains 145,534 plants
reporting either "yes" or "no" to the question on
the use of leased employees, which is 42 percent of the plants with
positive shipments included in the 2002 Census of Manufactures. While
the response rate for the newly added question is not high, our analyses
using the data of respondents show systematic relationships between some
of their characteristics and whether or not they used leased employees.
Among respondents, the newly constructed plants represent 3.8 percent of
our sample. Plants that responded to the question about their leased
employee use are, on average, larger and more likely to belong to
multi-establishment firms than plants that did not respond to the
question. Of the respondent plants, the average value of shipments is
$3.6 million, 32 percent of them are affiliated with multi-establishment
firms (firms that own multiple establishments), and 28 percent are
affiliated with firms that have other manufacturing plant(s). For plants
affiliated with firms that have other manufacturing plant(s) (panel B of
table 4), the average number of states in which those firms operate is
9.8 and the average number of three-digit NAICS manufacturing industries
is 2.7, which are similar to the numbers based on the overall Census of
Manufactures sample.
Note that it is possible that nonrespondent plants--those without
responses of either "yes" or "no"--are plants that
did not use leased employees. We compare characteristics of the
nonrespondent plants and the respondent plants answering "no"
to see if they are similar. We find that those that indicated explicitly
that they do not use leased employees are more similar to other
respondents answering "yes" than to nonrespondents. For
example, the average log shipments is 8.02 for respondents answering
"no" and 8.7 for respondents answering "yes," but
6.1 for nonrespondents. The percentage of plants affiliated with firms
that have other manufacturing plant(s) is almost the same between those
who answered "yes" and those who answered "no" at
about 30 percent, but it is 20 percent for nonrespondents. On average,
nonrespondents do not seem to have similar characteristics as those that
indicated they do not use leased employees. We also performed analyses
where we treat nonrespondents as plants that did not use leased
employees; we obtain less precise coefficients than those we obtain by
limiting our sample to respondents. Next, we report the results of our
analyses based on the data of respondent plants.
Table 5 shows the results of the probit analyses. The first,
second, and third columns show the results based on the specifications
that do not control for state- and industry-specific effects, while the
fourth, fifth, and sixth columns show the results when we control for
these effects. From the first column, we can see that plant size, on
average, has positive effects on a plant's use of PEO services. Our
results may be capturing a statistical artifact that a plant with more
workers has a higher probability to have at least one leased employee.
As we mentioned before, however, larger plants seem to face more
regulation, which might also lead them to rely on specialists for
compliance concerns. As we see in the fourth column, the effect is
qualitatively the same even after controlling for state-and
industry-specific effects. Based on calculations using the fourth
column, a one standard deviation increase in plant size increases the
plant's probability of using PEO services by 1.9 percentage points,
which is equivalent to 40 percent of the actual percentage of plants
using PEO services (4.7 percent). Note that when we include the squared
term of plant size, the result seems to show that the effect of plant
size is quadratic; the positive marginal effect of plant size is smaller
for larger plants, possibly because of their greater economies of scale
in managing regulatory compliance themselves.
Plants facing a higher potential rate of work-related injuries and
illnesses are also more likely to use PEO services. Such plants may be
able to benefit from better insurance premiums and health care benefits
by using a PEO, since the PEO can pool its injury and illness risks
across all its client firms. Also, PEOs may be the employer responsible
for paying workers' compensation, which would protect both the PEO
and the client firm from lawsuits related to work-related injuries or
illnesses. The magnitude of the effect is small, however. Based on the
fourth column of table 5, a one standard deviation increase in the
injury and illness rate raises a plant's probability of using PEOs
by only 0.3 percentage points. We also investigated whether the effect
of the injury and illness rate changes with plant size by including an
interaction term. Based on our sample, however, we did not find a
statistically significant difference.
Newly constructed plants are more likely to use PEOs than older
plants. This is consistent with the view that new plants, which face
various uncertainties in their business environment, may want to focus
on their core activity first in order to secure their survival. The
magnitude of the effect is large. Based on the fifth column of table 5,
a new plant's probability of using PEO services is greater than
others' by 6 percentage points, which is equivalent to 130 percent
of the actual percentage of plants using PEO services.
We also find that for plants affiliated with multiestablishment
firms, the probability of using PEOs is slightly greater. Of those
plants, the plants whose parent firms have other manufacturing plants
have a much greater likelihood of using PEOs than those whose parent
firms have no other manufacturing plants. The difference in the
likelihood is, on average, as large as 7.0 percentage points. Having
multiple manufacturing plants may make it more challenging for a firm to
comply with the increased number of regulations and laws. This might
have led these firms to be more likely to rely on PEO services.
Some firm-level variables are also systematically associated with a
plant's use of PEOs. For plants affiliated with firms that have at
least one other manufacturing plant, the overall manufacturing size of
the firm is negatively associated with a plant's likelihood of
using PEO services. The negative correlation seems to show the existence
of firm-level economies of scale in the firms performing the HR services
themselves. We also include the squared term of the firm-level size. The
coefficient of this term turns insignificant.
Finally, it seems that more diversified firms are more likely to
use PEO services. In the first, second, fourth, and fifth columns of
table 5, we report the results including the HHI variables, which
represent the degree of a firm's concentration across states and
across industries. As you can see, both HHIs obtain negative and
significant signs in most specifications. Firms that are geographically
diversified across different states are more likely to use PEO services.
Such firms may rely on PEOs in order to make sure they comply with the
different regulations of all the states in which they have plants. We
also find the same tendency for firms with multiple industries. The
coefficients, however, lose significance once we control for state- and
industry-specific effects. In the sixth column, we perform the same
analysis where we measure a firm's industry diversity by the number
of three-digit NAICS manufacturing industries of all of a firm's
plants instead of the HHI. We find that the coefficient for the number
of three-digit NAICS manufacturing industries is positive and
significant--evidence that a firm's industry diversification may
matter for its decision to use PEO services.
Conclusion
Using both public and confidential data, we summarize how the
intensity of use of PEO services varies across industries, geographical
areas, and establishment characteristics. The uneven distribution of the
use of PEO services gives us an insight into how, to varying degrees,
the payroll-based labor measure may be underestimated. Among the
industries, transportation and repair services have particularly high
intensity of use of PEO services. Florida and Arizona are two states
with particularly high intensity of use of PEO services. We also find
that the patterns of use of leased employees across industries and
across states are different from the patterns of use of THS workers.
Finally, our analyses using microdata of manufacturing establishments
suggest that various establishment-level characteristics are associated
with establishments' use of leased employees and thus the degree
that the payroll employment number underestimates the actual number of
workers. We found that, for plants in our sample, the use of PEO
services depends on the size of the establishment and of its parent
firm. The use of PEO services is greater for newly constructed plants
and for plants with a potentially high injury and illness rate. The
greater diversification across industries and geographical areas of a
parent firm may also increase an establishment's use of PEO
services. As the use of PEO services increases over time, in order to
better estimate the amount of labor regularly used for production, it
will become more important to incorporate leased employees into the
labor statistics of establishments or industries for which they work.
APPENDIX: 1997 ECONOMIC CENSUS DEFINITIONS OF NUMBER OF EMPLOYEES
General definition
Paid employees are full-time and part-time employees, including
salaried officers and executives of corporations. Included are employees
on paid sick leave, paid holidays, and paid vacations; not included are
proprietors and partners of unincorporated businesses. The definition of
paid employees is the same as that used on Internal Revenue Service
(IRS) Form 941.
Sector-specific information
Construction and Manufacturing sectors--comprise all full-time and
part-time employees, on the payrolls of establishments, who worked or
received pay for any part of the pay period including the 12th of March,
May, August, and November, divided by four.
Finance and Insurance sector--comprises all employees who were on
the payroll during the pay period including March 12. Excludes
independent (nonemployee) agents.
Information; Professional, Scientific. and Technical Services;
Administrative and Support and Waste Management and Remediation
Services; Educational Services; Health Care and Social Assistance; Arts,
Entertainment, and Recreation; and Other Services (Except Public
Administration) sectors--comprise all employees who were on the payroll
during the pay period including March 12. Include members of a
professional service organization or association that operates under
state professional corporation statutes and files a corporate federal
income tax return. Exclude employees of departments or concessions
operated by other companies at the establishment.
Management of Companies and Enterprises sector--comprises all
employees who were on the payroll during the pay period including March
12.
Mining sector--comprises all employees who were on the payroll
during the pay period including March 12. Includes employees working for
miners, paid on a per ton, car, or yard basis. Excluded are employees at
the mine but on the payroll of another employer (such as employees of
contractors) and employees at company stores, boardinghouses,
bunkhouses, and recreational centers. Also excluded are members of the
armed forces and pensioners carried on the active rolls but not working
during the period.
Real Estate and Rental and Leasing sector--comprises all employees
who were on the payroll during the pay period including March 12.
Excludes independent (nonemployee) agents.
Retail Trade and Accommodation and Food Services sectors--comprise
all employees on the payroll during the pay period including March 12.
Exclude employees of departments or concessions operated by other
companies at the establishment.
Transportation and Warehousing sector--comprises all employees who
were on the payroll during the pay period including March 12.
Utilities sector--comprises all employees who were on the payroll
during the pay period including March 12.
Source: U.S. Census Bureau. 1997 Economic Census.
REFERENCES
Bodenheimer, Thomas, 2005, "High and rising health care costs.
Part 1: Seeking an explanation," Annals of Internal Medicine, Vol.
142, No. 10, May, pp. 847-854.
Cline, Sandra B., 1997, "Prosper with no employees?"
Indiana Business Review, Vol. 41, No. 8, pp. 29-31.
Cook, M. F., 1999, Outsourcing Human Resource Functions, New York:
American Management Association.
Dey, Matthew, Susan Houseman, and Anne Polivka, 2006,
"Manufacturers' outsourcing to employment services," W.
E. Upjohn Institute for Employment Research, working paper, No. 07-132,
December.
Drucker, Peter F., 2002, "They're not employees,
they're people," Harvard Business Review, February, pp. 70-77.
Estavao, Marcello, and Saul Lach, 1999, "Measuring temporary
labor outsourcing in U.S. manufacturing," National Bureau of
Economic Research, working paper, No. 7421, November.
Gilley, Matthew K., Charles R. Greer, and Abdul A. Rasheed, 2004,
"Human resource outsourcing and organizational performance in
manufacturing firms,'" Journal of Business Research, Vol. 57,
No. 3, March, pp. 232-240.
Greening, Daniel, Bruce Barringer, and Granger Macy, 1996, "A
qualitative study of managerial challenges facing small business
geographic expansion," Journal of Business Venturing, Vol. 11, No.
4, July, pp. 233-256.
Greer, Charles R., Stuart A. Youngblood, and David A. Gray, 1999,
"Human resource management outsourcing: The make or buy
decision," Academy of Management Executive, Vol. 13, No. 3. August,
pp. 85-96.
Houseman, Susan N., 2006, "Outsourcing, offshoring, and
productivity measurement in manufacturing," W. E. Upjohn Institute
for Employment Research, working paper, No. 06-130, revised September
2006.
--, 2003, "The benefits implications of recent trends in
flexible staffing arrangements," in Benefits for the Workplace of
the Future, Olivia S. Mitchell, David S. Blitzstein, Michael Gordon, and
Judith F. Mazo (eds.), Philadelphia: University of Pennsylvania Press,
pp. 89-109.
Klaas, Brian S., John McClendon, and Thomas W. Gainey, 2000,
"Managing HR in the small and medium enterprise: The impact of
professional employer organizations," Entrepreneurship: Theory and
Practice, Vol. 25, pp. 107-124.
Lenz, Edward A., 2003, Co-employment: Employer Liability Issues in
Third-Party Staffing Arrangements, 5th ed., Alexandria, VA: American
Staffing Association.
National Association of Insurance Commissioners
(NAIC)/International Association of Industrial Accident Boards and
Commissions (IAIABC) Joint Working Group, 2002, "Summary: Report on
employee leasing and professional employer organizations," Journal
of Insurance Regulation, Vol. 21, No. 2, Winter, pp. 95-101.
Segal, Lewis M., and Daniel Sullivan, 1997, "The growth of
temporary services work," Journal of Economic Perspectives, Vol.
11, No. 2, Spring, pp. 117-136.
NOTES
(1) The U.S. Census Bureau conducts the Economic Census every five
years to profile the U.S. economy, from the national to local level The
subject series on Administrative and Support and Waste Management and
Remediation Services covers employment in the NAICS (North American
Industry Classification System) sector 56 (for details, see
www.census.gov/econ/census02/naics/ sector56/56.htm).
(2) While 2007 Economic Census data are not yet available, the U.S.
Bureau of Labor Statistics' Current Employment Statistics (CES)
data indicate that the PEO share began to level off from about 2004 (For
the years prior to 2003, the CES used a fixed ratio to create the THS
and PEO industry payroll employment data.)A few possible reasons exist
According to Dey, Houseman, and Polivka (2006). the CES's sampling
frame--namely, the U.S Bureau of Labor Statistics' Quarterly Census
of Employment and Wages (QCEW)--somewhat underestimates leased employees
in the PEO industry. At the same time, the CES data could reflect
stricter regulation on using PEO services For example, the State
Unemployment Tax Act (SUTA) Dumping Prevention Act of 2004 requires all
states to enact anti-SUTA-dumping legislation, thereby potentially
decreasing the opportunity to use PEO services lo sidestep tax rate
modification procedures.
(3) Dey, Houseman, and Polivka (2006) use the contingent work
supplements to the U.S. Census Bureau's Current Population Survey
(CPS), as well as the US Bureau of Labor Statistics' Occupational
Employment Statistics (OES) program, for their estimation.
(4) The labor productivity measure here is calculated based on the
U.S. Bureau of Labor Statistics' manufacturing output indexes and
the CES manufacturing employment data.
(5) One possible advantage was the manipulation of the experience
rating modification factor for insurance premiums by basing the
adjustment on the PEO's past claim history rather than on that of
the client firm to receive lower rates (NAIC/IAIABC Joint Working Group,
2002). Also, a client firm might use a PEO to misrepresent its physical
location as being in a state with lower insurance rates. Other concerns
included misrepresented payrolls and misclassified occupations, as well
as contusion about which firm was responsible for providing
workers' compensation (NAIC/IAIABC Joint Working Group, 2002; and
Houseman, 2003).
(6) For many laws protecting workers from discrimination and
harassment, enforcement varies by the number of employees in a firm: for
example, the Worker Adjustment and Retraining Notification (WARN) Act,
the Civil Rights Act of 1991, the Americans with Disabilities Act of
1990, the Uniform Guidelines on Employee Selection Procedures (1978),
and the Pregnancy Discrimination Act.
(7) For example, slate and local taxes and workers'
compensation requirements differ across states. Also, sometimes state
requirements or benefits supersede federal regulations.
(8) ERISA regulates how a pension plan can be funded, vested,
disclosed, and eventually paid out to the employee.
(9) EAPs provide supporting services for substance abuse, work
relationship issues, emotional distress, mental health concerns, and
other issues that may adversely affect an employee's work
performance.
(10) Workers' compensation is considered the employee's
only way (in legal terms, the exclusive remedy) to receive benefits for
a workplace injury; workers' compensation protects the PEO and the
client finn from injury-related lawsuits except under special
circumstances (Lenz, 2003, p 25).
(11) Although not the main PEO function, the PEO reserves the right
to hire, reassign, and fire employees and maintains some control or
direction over the joint employees with its client firm By retaining
such decision-making control, the PEO has an ability to manage its
liabilities and earn protection from some lawsuits under state and
federal laws (Lenz, 2003, p 10).
(12) For further details on this series, see note 1.
(13) Leased employees and THS workers are reported on the payroll
of employment service establishments. To avoid confusion, we do not
include the employment service industry in table 1.
(14) See the data from the US Bureau of Labor Statistics'
Injuries, Illnesses, and Fatalities (IIF) program, available at
www.bls.gov/ iif/oshwc/osh/os/ostb0770.txt.
(15) The industry distribution of leased employees is not available
in the 2002 Economic Census.
(16) A division of leased employees across states is not available
for 1992.
(17) The US, Census also asks a question about the number of leased
employees. However, many plants did not provide the actual number of
leased employees.
(18) The data sets we have access to in this study provide the
information on the value of shipments only for manufacturing plants.
(19) Because of the high nonresponse rate for the number of leased
employees, even in the 2002 Census of Manufactures, it is difficult to
capture the total number of employees.
(20) For details on the HHI, a commonly accepted measure of market
concentration, see www.usdoj.gov/atr/public/testimony/hhi.htm.
Britton Lombardi is an associate economist and Yukako Ono is an
economist in the Economic Research Department at the Federal Reserve
Bank of Chicago, The authors thank Vanessa Haleco-Meyer for excellent
research assistance. The research in this article was conducted while
the authors were Special Sworn Status researchers of the U.S. Census
Bureau at the Chicago Census Research Data Center. The views, research
results, and conclusions expressed are those of the authors and do not
necessarily reflect the views of the U.S. Census Bureau. This article
has been screened to ensure that no confidential data are revealed. In
addition, the authors are grateful for support from the National Science
Foundation (awards No. SES-0004335 and ITR-0427889) for their. research
at the Chicago Census Research Data Center.
TABLE 1
Leased employees, by client industry category, 1997
Payroll
Employees employees (b)
leased to (not including
firms by leased employees
Industry PEOs (a) and THS workers)
Mining 1,065 509,006
Construction 102,123 5,664,840
Manufacturing 104,415 16,888,016
Transportation 134,760 2,811,017
Utilities except
waste management 2,052 702,703
Information services 12,839 3,066,167
Wholesale trade 29,615 5,796,557
Retail trade 57,236 13,991,103
Accommodation and
food services 77,311 9,451,226
Finance and insurance 15,593 5,835,214
Real estate and
rental/leasing 16,243 1,702,420
Professional, scientific,
and technical services 47,987 5,361,210
Administrative and
support services including
waste management 48,304 n.a.
Health and social services 58,363 13,561,579
Educational services 7,565 321,073
Arts, entertainment,
and recreational services 13,316 1,587,660
Personal care and
laundry services 13,447 1,217,185
Repair services 38,016 1,276,389
Leased
employees as
share of leased
plus payroll Index, (c)
Industry employees U.S.=1.00
(percent)
Mining 0.21 0.25
Construction 1.77 2.11
Manufacturing 0.61 0.73
Transportation 4.57 5.45
Utilities except
waste management 0.29 0.35
Information services 0.42 0.50
Wholesale trade 0.51 0.61
Retail trade 0.41 0.49
Accommodation and
food services 0.81 0.97
Finance and insurance 0.27 0.32
Real estate and
rental/leasing 0.95 1.13
Professional, scientific,
and technical services 0.89 1.06
Administrative and
support services including
waste management n.a. n.a.
Health and social services 0.43 0.51
Educational services 2.30 2.74
Arts, entertainment,
and recreational services 0.83 0.99
Personal care and
laundry services 1.09 1.30
Repair services 2.89 3.45
Industry
share of
national
leased
Industry employees (d)
(percent)
Mining 0.12
Construction 11.55
Manufacturing 11.81
Transportation 15.24
Utilities except
waste management 0.23
Information services 1.45
Wholesale trade 3.35
Retail trade 6.47
Accommodation and
food services 8.75
Finance and insurance 1.76
Real estate and
rental/leasing 1.84
Professional, scientific,
and technical services 5.43
Administrative and
support services including
waste management n.a.
Health and social services 6.60
Educational services 0.86
Arts, entertainment,
and recreational services 1.51
Personal care and
laundry services 1.52
Repair services 4.30
(a) Apart from the leased employees, the PEO industry hires 11,409
management and administrative employees in the U.S. See the 1997
Economic Census, subject series on Administrative and Support and
Waste Management and Remediation Services, available at
www.census.gov/prod/ec97/97s56-sb.pdf.
(b) See the appendix for definitions of the number of employees on
payroll by industry. Note that the 1997 Economic Census does not
provide the North American Industry Classification System code of
each industry to which leased employees are allocated. We obtain the
payroll employment of a corresponding industry based on the industry
title.
(c) The index is calculated by dividing the value in the third column
by 0.84 percent (the U.S. average share of leased employees in 1997).
(d) This column does not sum to 100 because we do not include leased
employees in the agriculture, administrative and support services
including waste management, and "other" industries.
Notes: PEO means professional employer organization. THS means
temporary help services. n.a. means not applicable.
Source: Authors' calculations based on data from the U.S. Census
Bureau, 1997 Economic Census.
TABLE 2
Leased employees, by state, 1997 and 2002
1997
Employees of all Leased
Employees private industries, employees
leased to including leased as share
firms by employees and of total
State PEOs (a) THS workers employees
(percent)
Alabama 14,644 1,591,179 0.92
Alaska 459 188,923 0.24
Arizona 55,457 1,701,357 3.26
Arkansas 11,894 925,498 1.29
California 67,804 11,565,015 0.59
Colorado 9,575 1,675,514 0.57
Connecticut 1,535 1,471,970 0.10
Delaware 219 348,009 0.06
District of Columbia n.a. 396,328 n.a.
Florida 197,632 5,550,307 3.56
Georgia 63,730 3,106,872 2.05
Hawaii 5,520 426,129 1.30
Idaho 848 404,670 0.21
Illinois 39,214 5,090 0.77
Indiana 15,497 2,487,609 0.62
Iowa 4,191 1,179,660 0.36
Kansas n.a. 1,049,359 n.a.
Kentucky 1,860 1,422,605 0.13
Louisiana 4,943 1,531,663 0.32
Maine 893 447,063 0.20
Maryland 7,595 1,906,880 0.40
Massachusetts 7,891 2,859,594 0.28
Michigan 39,021 3,844,460 1.01
Minnesota 11,085 2,195,621 0.50
Mississippi 6,135 909,746 0.67
Missouri 6,132 2,281,643 0.27
Montana 204 273,746 0.07
Nebraska 9,493 701,132 1.35
Nevada 3,415 768,708 0.44
New Hampshire 6,641 497,878 1.33
New Jersey 13,617 3,300,923 0.41
New Mexico 4,584 533,858 0.86
New York 25,000 6,895,924 0.36
North Carolina 13,186 3,167,303 0.42
North Dakota 108 242,047 0.04
Ohio 22,384 4,709,180 0.48
Oklahoma 5,921 1,127,734 0.53
Oregon 12,124 1,292,579 0.94
Pennsylvania 10,072 4,840,877 0.21
Rhode Island n.a. 390,914 n.a.
South Carolina 19,548 1,473,831 1.33
South Dakota n.a. 279,187 n.a.
Tennessee 15,327 2,247,944 0.68
Texas 104,533 7,250,925 1.44
Utah 18,788 824,120 2.28
Vermont n.a. 232,476 n.a.
Virginia 9,341 2,626,844 0.36
Washington 2,139 2,081,017 0.10
West Virginia 1,141 542,782 0.21
Wisconsin 4,214 2,277,849 0.18
Wyoming n.a. 161,772 n.a.
Total/U.S. 884,002 105,299,123 0.84
1997 2002
Index, (c) Index, (d)
State U.S.=1.00 U.S.=1.00
Alabama 1.10 1.45
Alaska 0.29 0.39
Arizona 3.88 1.74-3.48 (b)
Arkansas 1.54 0.12
California 0.70 0.52
Colorado 0.68 4.60
Connecticut 0.12 0.12
Delaware 0.07 0.06
District of Columbia n.a. 0.05
Florida 4.24 5.31
Georgia 2.44 1.33
Hawaii 1.55 0.70
Idaho 0.25 1.71
Illinois 0.92 0.51
Indiana 0.74 0.76
Iowa 0.43 0.10
Kansas n.a. 0.37
Kentucky 0.15 0.10
Louisiana 0.38 0.54
Maine 0.24 0.39
Maryland 0.48 0.44
Massachusetts 0.33 0.32
Michigan 1.20 1.16
Minnesota 0.60 0.33
Mississippi 0.80 0.57
Missouri 0.32 0.40
Montana 0.08 0.06
Nebraska 1.61 1.40
Nevada 0.52 0.68
New Hampshire 1.58 1.32
New Jersey 0.49 0.87
New Mexico 1.02 0.88
New York 0.43 0.35
North Carolina 0.50 0.36
North Dakota 0.05 0.005-0.002 (b)
Ohio 0.57 0.57
Oklahoma 0.63 0.96
Oregon 1.12 0.43
Pennsylvania 0.25 0.23
Rhode Island n.a. 0.41-0.81 (b)
South Carolina 1.58 1.73
South Dakota n.a. 0.02
Tennessee 0.81 0.81
Texas 1.71 1.52
Utah 2.71 2.48
Vermont n.a. 0.01
Virginia 0.43 0.41
Washington 0.12 0.08
West Virginia 0.25 0.26
Wisconsin 0.21 0.14
Wyoming n.a. 0.03
Total/U.S. 1.00 1.00
(a) Number of leased employees reported by PEOs located in each
state.
(b) Due to disclosure concerns, only a range of the number of leased
employees was given for these states; our calculated index range is
based on the highest and lowest values of this given range.
(c) The index is calculated by taking the value in the third column
and dividing it by 0.84 percent (the U.S. average share of leased
employees in 1997).
(d) The index is calculated by taking the value of leased employees as
a share of total employees in 2002 (not shown) and dividing it by 1.48
percent (the U.S. average share of leased employees in 2002).
Notes: PEO means professional employer organization. THS means
temporary help services. n.a. means not available (undisclosed). The
total value at the bottom of the first column includes the
undisclosed data.
Sources: Authors' calculations based on data from the U.S. Census
Bureau, 1997 and 2002 Economic Censuses and 1997 and 2002 County
Business Patterns.
TABLE 3
Geographical distribution of leased employees and temporary
help services workers, 1997
Leased
employees
Index,
Top ten U.S.=1.00
Florida 4.24
Arizona 3.88
Utah 2.71
Georgia 2.44
Texas 1.71
Nebraska 1.61
New Hampshire 1.58
South Carolina 1.58
Hawaii 1.55
Arkansas 1.54
Bottom ten
Pennsylvania 0.25
West Virginia 0.25
Maine 0.24
Wisconsin 0.21
Kentucky 0.15
Connecticut 0.12
Washington 0.12
Montana 0.08
Delaware 0.07
North Dakota 0.05
Temporary help
services workers
Index,
Top ten U.S.=1.00
Maryland 1.62
Arizona 1.33
California 1.27
Michigan 1.25
Georgia 1.24
Texas 1.22
South Carolina 1.18
Delaware 1.18
Colorado 1.12
Illinois 1.09
Bottom ten
Idaho 0.59
Iowa 0.56
Nebraska 0.54
West Virginia 0.50
Montana 0.48
Mississippi 0.46
Wyoming 0.35
Hawaii 0.26
North Dakota 0.26
Alaska 0.23
Notes: The index values for leased employees here are the same as
those shown in the fourth column of table 2. Analogously, we
calculate the index values for temporary help services workers by
dividing each state's temporary help services workers percentage
of total payroll employment (not shown) by 2.48 percent (the
U.S. average).
Sources: Authors' calculations based on data from the U.S. Census
Bureau, 1997 Economic Census and 1997 County Business Patterns.
TABLE 4
Summary statistics of variables
Our sample
Standard
Mean deviation
A. All plants
Number of plants 145,534
Plant size: log value of 8.1 1.7
shipments
Injury and illness rate (four- 6.8 2.7
or five-digit NAICS industry
level)
Percentage of newly 3.8
constructed plants
Percentage of plants 32
affiliated with
multi-establishment firms
Percentage of plants 28
affiliated with firms that
have other manufacturing
plant(s)
B. Plants affiliated with
firms that have other
manufacturing plant(s)
Number of plants affiliated 40,251
with firms that have other
manufacturing plant(s)
Firm size: log value of 12.4 2.4
shipments of a firm's
manufacturing plants
Number of states with parent 9.8 10.2
firm's plants
Number of NAICS three-digit 2.7 2.4
manufacturing industries of
all plants owned by a parent
firm
HHI for firm's state 0.5 0.3
concentration (in terms of
value of shipments)
HHI for firm's manufacturing 0.8 0.3
industries concentration (in
terms of value of shipments)
All plants in the 2002
Census of Manufactures
Standard
Mean deviation
A. All plants
Number of plants 348,295
Plant size: log value of 6.9 2.0
shipments
Injury and illness rate (four- 6.7 2.7
or five-digit NAICS industry
level)
Percentage of newly 7.2
constructed plants
Percentage of plants 19
affiliated with
multi-establishment firms
Percentage of plants 16
affiliated with firms that
have other manufacturing
plant(s)
B. Plants affiliated with
firms that have other
manufacturing plant(s)
Number of plants affiliated 56,914
with firms that have other
manufacturing plant(s)
Firm size: log value of 12.1 2.4
shipments of a firm's
manufacturing plants
Number of states with parent 9.1 10.1
firm's plants
Number of NAICS three-digit 2.6 2.3
manufacturing industries of
all plants owned by a parent
firm
HHI for firm's state 0.5 0.3
concentration (in terms of
value of shipments)
HHI for firm's manufacturing 0.8 0.3
industries concentration (in
terms of value of shipments)
Notes: NAICS means North American Industry Classification System.
HHI means the Herfindahl/Hirschman Index, a commonly accepted
measure of market concentration (see www.usdoj.gov/atr/public/
testimony/hhi.htm).
Source: Authors' calculations based on microdata from the U.S.
Census Bureau, 2002 Census of Manufactures.
TABLE 5
Probit analysis
Dependent variable = 1 if a plant uses any leased employees
1 2
Plant size: log value 0.132 *** 0.301 ***
of shipments (21.89) (6.88)
Squared term of -0.00997 ***
plant size (-3.73)
Injury and illness rate
(4-or 5-digit NAICS 0.0147 *** 0.0138 ***
industry level) (4.76) (4.43)
dbirth=1 if a plant
is newly constructed 0.445 *** 0.477 ***
in 2002 (15.80) (16.41)
dmulti=1 if a plant is
affiliated with a firm
with multiple 0.0708 ** 0.0783 ***
establishments (2.54) (2.80)
dmulti mfg=1 if a
plant is affiliated
with a firm with
other manufacturing 0.977 *** 0.727 ***
plant(s) (4.50 (3.39
dmulti_mfg x firm -0.0679 *** -0.0503 ***
manufacturing size (-4.78) (-3.59)
dmulti_mfg x HHI
of a firm's state
concentration
(in terms of total -0.257 ** -0.204 *
value of shipments) (-2.12) (-1.80
dmulti_mfg x HHI
of a firm's
manufacturing
industries
concentration
(in terms of total -0.199 * -0.174
value of shipments) (-1.67) (-1.46)
dmulti_mfg x number
of NAICS three-digit
manufacturing
industries of parent
firm's plants
State dummies No No
NAICS three-digit
manufacturing
industry dummies No No
3 4
Plant size: log value 0.292 *** 0.123 ***
of shipments (6.59) (20.37)
Squared term of -0.00933 ***
plant size (-3.42)
Injury and illness rate
(4-or 5-digit NAICS 0.0139 *** 0.0117 ***
industry level) (4.45) (2.58)
dbirth=1 if a plant
is newly constructed 0.477 *** 0.437 ***
in 2002 (16.36) (15.24)
dmulti=1 if a plant is
affiliated with a firm
with multiple 0.0747 *** 0.0816 ***
establishments (2.65) (2.87)
dmulti mfg=1 if a
plant is affiliated
with a firm with
other manufacturing 0.682 *** 0.838 ***
plant(s) (3.59) (3.78)
dmulti_mfg x firm -0.0645 *** -0.0624 ***
manufacturing size (-4.43) (-4.44)
dmulti_mfg x HHI
of a firm's state
concentration
(in terms of total -0.207 * -0.231 **
value of shipments) (-1.88) (-1.98)
dmulti_mfg x HHI
of a firm's
manufacturing
industries
concentration
(in terms of total -0.138
value of shipments) (-1.10
dmulti_mfg x number
of NAICS three-digit
manufacturing
industries of parent 0.0304 **
firm's plants (2.19)
State dummies No Yes
NAICS three-digit
manufacturing
industry dummies No Yes
5 6
Plant size: log value 0.287 *** 0.279 ***
of shipments (6.55) (6.28)
Squared term of -0.00970 *** -0.00915 ***
plant size (-3.67) (-3.38)
Injury and illness rate
(4-or 5-digit NAICS 0.0103 ** 0.0106 **
industry level) (2.25) (2.34)
dbirth=1 if a plant
is newly constructed 0.468 *** 0.468 ***
in 2002 (15.69) (15.63)
dmulti=1 if a plant is
affiliated with a firm
with multiple 0.0890 *** 0.0855 ***
establishments (3.12) (2.97)
dmulti mfg=1 if a
plant is affiliated
with a firm with
other manufacturing 0.605 *** 0.599 ***
plant(s) (2.76) (3.19)
dmulti_mfg x firm -0.0459 *** -0.0583 ***
manufacturing size (-3.30) (-3.91)
dmulti_mfg x HHI
of a firm's state
concentration
(in terms of total -0.182 * -0.180 *
value of shipments) (-1.66) (-1.69)
dmulti_mfg x HHI
of a firm's
manufacturing
industries
concentration
(in terms of total -0.115
value of shipments) (-0.92)
dmulti_mfg x number
of NAICS three-digit
manufacturing
industries of parent 0.0247 *
firm's plants (1.73)
State dummies Yes Yes
NAICS three-digit
manufacturing
industry dummies Yes Yes
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
Notes: NAICS means North American Industry Classification System.
HHI means the Herfindahl-Hirschman Index, a commonly accepted
measure of market concentration (see
www.usdoj.gov/atr/public/testimony/hhi.htm). Robust z statistics
are in parentheses; errors are clustered for plants in the same
firm. See the text for further details.
Source: Authors' calculations based on microdata from the U.S.
Census Bureau, 2002 Census of Manufactures.
