What do small businesses do?
Hurst, Erik ; Pugsley, Benjamin Wild
ABSTRACT We show that most small business owners are very different
from the entrepreneurs that economic models and policymakers often have
in mind. Using new data that sample entrepreneurs just before they start
their businesses, we show that few small businesses intend to bring a
new idea to market or to enter an unserved market. Instead, most intend
to provide an existing service to an existing market. Further, we find
that most small businesses have little desire to grow big or to innovate
in any observable way. We show that such behavior is consistent with the
industry characteristics of the majority of small businesses, which are
concentrated among skilled craftspeople, lawyers, real estate agents,
health care providers, small shopkeepers, and restaurateurs. Lastly, we
show that nonpecuniary benefits (being one's own boss, having
flexibility of hours, and the like) play a first-order role in the
business formation decision. Our findings suggest that the importance of
entrepreneurial talent, entrepreneurial luck, and financial frictions in
explaining the firm size distribution may be overstated. We conclude by
discussing the potential policy implications of our findings.
**********
Economists and policymakers alike have long been interested in the
effects of various economic policies on business ownership. In fact, the
U.S. Small Business Administration is a federal agency whose main
purpose, according to its mission statement, is to help Americans
"start, build, and grow businesses." Researchers and
policymakers often either explicitly or implicitly equate small business
owners with entrepreneurs. Although this association could be
tautological, we show in this paper that the typical small business
owner is very different from the entrepreneur that economic models and
policymakers have in mind. For example, economic theory usually
considers entrepreneurs as individuals who innovate and render aging
technologies obsolete (Schumpeter 1942), take economic risks (Knight
1921, Kihlstrom and Laffont 1979, and Kanbur 1979), or are
jacks-of-all-trades in the sense of having a broad skill set (Lazear
2005). Policymakers often consider entrepreneurs to be job creators or
the engines of economic growth.
In this paper we shed light on what the vast majority of small
businesses actually do and, further, what they report ex ante wanting to
do. Section I highlights the industrial breakdown of small businesses
within the United States. By "small businesses" we primarily
mean firms with between 1 and 19 employees; firms in this size range
employ roughly 20 percent of the private sector workforce. However, we
also define alternative classifications, such as firms with between 1
and 100 employees. We show that over two-thirds of all small businesses
by our primary definition are confined to just 40 narrow industries,
most of which provide a relatively standardized good or service to an
existing customer base. These industries primarily include skilled
craftspeople (such as plumbers, electricians, contractors, and
painters), skilled professionals (such as lawyers, accountants, and
architects), insurance and real estate agents, physicians, dentists,
mechanics, beauticians, restaurateurs, and small shopkeepers (for
example, gas station and grocery store owners). We also show that
although firms within these industries are heterogeneous in size, these
industries account for a disproportionate share of all small businesses.
This composition of small businesses foreshadows our empirical results.
In section II we study job creation and innovation at small firms,
both established and new. First, using a variety of data sets, we show
that most surviving small businesses do not grow by any significant
margin. Rather, most start small and stay small throughout their entire
life cycle. (1) Also, most surviving small firms do not innovate along
any observable margin. Very few report spending resources on research
and development, getting a patent, or even obtaining copyright or
trademark protection for something related to the business, including
the company's name. Furthermore, we show that between one-third and
half of all new businesses report providing an existing good or service
to an existing market. This is not surprising when one thinks of the
most common types of small business. A new plumber or a new lawyer who
opens up a practice often does so in an area where plumbers and lawyers
already operate.
Most existing research attributes differences across firms with
respect to ex post performance to either differences in financing
constraints (for example, Evans and Jovanovic 1989, Clementi and
Hopenhayn 2006), differences in ex post productivity draws across firms
(for example, Simon and Bonini 1958, Jovanovic 1982, Pakes and Ericson
1989, Hopenhayn 1992), or differences in the owners'
entrepreneurial ability (for example, Lucas 1978). In section III we use
new data on the expectations of nascent small business owners to show
that these stories are incomplete. When asked at the time of their
business formation, most business owners report having no desire to grow
big and no desire to innovate along observable dimensions. In other
words, when starting their business, the typical plumber or lawyer
expects the business to remain small well into the foreseeable future
and does not expect to innovate by developing a new product or service
or even to enter new markets with an existing product or service.
If most small businesses do not want to grow and do not want to
innovate, why do they start? We address this question in section IV. The
same new data set that we used to explore the expectations of nascent
business owners also specifically asks about motives. Over 50 percent of
these new business owners cite nonpecuniary benefits--for example,
"wanting flexibility over schedule" or "to be one's
own boss"--as a primary reason for starting the business. By
comparison, only 34 percent report that they are starting the business
to generate income, and only 41 percent indicate that they are starting
a business because they want to create a new product or because they
have a good business idea. (Respondents could give up to two answers.)
Exploiting the panel nature of the data, we show that those small
businesses that started for other than innovative reasons were less
likely to grow in the ensuing years, less likely to report wanting to
grow, less likely to innovate, and less likely to report wanting to
innovate.
Collectively, these results suggest that the first-order reasons
why most small businesses form are not the innovation or growth motives
embedded in most theories of entrepreneurship. Rather, the nonpecuniary
benefits of small business ownership may be an important driver of why
firms start and remain small. Additionally, some industries (such as
insurance agencies) may have a natural scale of production at the
establishment level that is quite low. In section V we discuss how our
results challenge much of the existing work on entrepreneurship and
small-firm dynamics. We highlight how our findings suggest that the
importance of entrepreneurial talent, entrepreneurial luck, and
financial frictions in explaining the firm size distribution may be
overstated. In section VI we discuss the policy implications of our
results. Section VII concludes.
More research into the diversity of motives and expectations among
small businesses has been done in developing economies than in developed
economies. (2) Recent work by Rafael La Porta and Andrei Shleifer (2008)
and a review of the literature by Abhijit Banerjee and Esther Duflo
(2011) show that most small businesses in developing economies do not
grow or innovate in any observable way. We discuss in section V how the
qualitatively similar outcomes we observe in the United States are
driven by different forces than in developing economies.
Overall, our results reveal substantial skewness among small
businesses within the United States, in terms of both actual and
expected growth and innovative behavior. Although growth and innovation
are the usual cornerstones of entrepreneurial models and the usual
justifications for policy interventions to support small business, most
small businesses do not want to grow or innovate. Our results suggest
that it is often inappropriate for researchers to use the universe of
small business (or self-employment) data to test standard theories of
entrepreneurship. More specialized data sets, such as those that track
small businesses seeking venture capital funding, may be more suitable
for this task, because these firms have been shown to be more likely to
actually grow or to innovate than other small businesses. (3) For their
part, policymakers who want to promote growth and innovation may want to
consider more targeted policies than those that address the universe of
small businesses.
I. Industrial Composition of Small Businesses
This section intends to show that most small businesses are
concentrated in a small number of narrowly defined industries
(industries at the four-digit level of the North American Industry
Classification System, or NAICS) that mostly provide standard services
to local customers. This context is important when interpreting our
findings that the majority of small businesses do not intend to grow or
innovate in any substantive way.
To examine the types of small businesses that exist within the
United States, we use data from the Statistics of U.S. Businesses (SUSB)
compiled by the U.S. Census Bureau. (4) To create these statistics, the
Census Bureau compiles data extracted from the Business Register, which
contains the bureau's most current and consistent data for U.S.
business establishments. (5) The data cover most U.S. firms with at
least one paid employee. (We also discuss how our results would differ
if we included information from firms that do not hire paid employees.)
We focus our attention on the statistics from 2003 to 2007, all of which
are coded using the NAICS 2002 industry definitions; additional data
from the Economic Census are available for 2007. However, our results
are nearly identical if we pick any year between 1998 and 2008.
Throughout the paper, to avoid contamination by large firms operating
many small establishments, we classify business size by total firm
employment. (6) For most purposes in this section, we refer to
"small businesses" as those with between 1 and 19 employees,
although we also consider alternative definitions based on different
employment size cutoffs.
As is already well known, small businesses account for a very large
fraction of the population of employer firms. Figure 1 uses the SUSB
data from 2007 to construct the cumulative distribution function for
firm size using several different measures of economic activity. In 2007
roughly 6 million firms had paid employees; the 90 percent of these
firms that had fewer than 20 employees accounted for about 20 percent of
aggregate paid employment and about 15 percent of sales receipts and
payroll. These numbers change only slightly when one looks at firms with
fewer than 100 employees: firms with between 20 and 99 employees
represent an additional 8 percent of all employer firms and 15 percent
of aggregate employment.
Next we study the concentration of small businesses with paid
employees at very fine levels of industry classification. These results
yield two important messages. First, most small businesses are
concentrated in a few detailed industry classifications. Second, within
these few detailed industries, the distribution of employment across all
firm sizes differs from the overall distribution for all other
industries. Most of the industries in which small businesses reside are
also industries in which a disproportionate amount of economic activity
takes place in small firms.
[FIGURE 1 OMITTED]
We start by taking the universe of all employer firms with fewer
than 20 employees. Within this group of small firms, we rank the
represented four-digit industries by a crude measure of concentration,
namely, each industry's share of all firms in this universe. (7) We
define this share [x.sub.j] as
[x.sub.j] = [s.sub.j] / [summation over j] [s.sub.j],
where [s.sub.j] is the number of small businesses in industry j.
This measure gives the importance of a given industry out of the
universe of businesses with fewer than 20 employees. There are 294
four-digit NAICS industries in the SUSB data; we rank these industries
from 1 to 294, with the industry with the largest [x.sub.j] ranked 1.
[FIGURE 2 OMITTED]
Figure 2 shows the cumulative sum of [x.sub.j] across four-digit
industries by rank in 2007. The first 20 industries accounted for just
about half of all firms with fewer than 20 employees in that year, and
the top 40 for about two-thirds. The employment shares for the top 20
and the top 40 industries (not shown) were similar, at nearly 50 percent
and 65 percent, respectively.
Table 1 lists those top 40 four-digit industries ranked by
[x.sub.j]. The table shows that most small businesses are either
restaurants (full service, limited service, or bars), skilled
professionals (physicians, dentists, lawyers, accountants, architects,
consultants), skilled craftspersons (general contractors, plumbers,
electricians, masons, painters, roofers), professional service providers
(clergy, insurance agents, real estate agents), general service
providers (auto repair, building services such as landscaping, barbers
and beauticians), or small retailers (grocery stores, gas stations,
clothing stores).
These results are robust to alternative cuts of the data. If we
extend our classification to the top 60 four-digit industries (which
account for over 80 percent of all firms with fewer than 20 employees),
the broad types of industries in which most small businesses reside are
not altered. Rather, the firms ranked 41 to 60 are similar in spirit to
those in the top 40: they include dry cleaners, office supply stores,
hardware stores, jewelry stores, automobile dealerships, liquor stores,
furniture stores, and the like. Additionally, when we extend the
definition of small business to include all firms with fewer than 100
employees, our results are very similar to those under the narrower
definition: the 40 industries listed in table 1 also represent 66
percent of the firms and 61 percent of the employment in this group.
One concern may be that the important small business industries may
reflect the overall size of the industry rather than the role of small
businesses within the industry. In fact, the bulk of small businesses
are concentrated in industries where a disproportionate amount of
employment is concentrated in small firms. For example, within the
skilled crafts industries, 48 percent of all employment (on average) is
in firms with fewer than 20 employees. This figure is much larger than
the 20 percent of economy-wide employment that is in firms of this size
(figure 1).
Figure 3 attempts to better document the relationship between the
importance of an industry within the universe of small businesses and
the amount of activity that takes place within small firms within that
industry. (8) The figure groups four-digit industries into deciles based
on [x.sub.j], the share of small firms within a given industry out of
all small firms in the economy. As in figure 2 and table 1, we define
small firms as those firms with between 1 and 19 employees; however, the
patterns are broadly similar if we instead define small firms to have
between 1 and 99 employees. The figure then plots for each decile the
within-industry share of employment in small firms, averaged across the
industries in the decile, again using data for 2007. Formally, we define
the within-industry share of employment in small firms as
[y.sub.j] = [e.sup.s.sub.j]/[e.sup.n.sub.j],
where [e.sup.s.sub.j] is the number of employees in small
businesses within industry j and [e.sup.n.sub.j] is the number of
employees in all businesses, regardless of size, within industry j. The
results show that the industries that make up the bulk of small
businesses (that is, that have a high [x.sub.j]) are also industries
where more of the employment within the industry is in small firms (have
a high [y.sub.j]). The top decile of industries with respect to
[x.sub.j] consists of the first 29 industries listed in table 1. These
industries account for about 60 percent of small businesses by number
and about 60 percent of employment within small businesses. For these
industries about 40 percent of employment within the industry, on
average, is in small firms. Again, only about 20 percent of employment
across all industries is in small firms. Thus, the high-[x.sub.j]
industries are skewed toward small firms. Across deciles, as [x.sub.j]
falls and the component industries become less important as a fraction
of all small businesses, the scale of these industries, for the most
part, monotonically increases.
[FIGURE 3 OMITTED]
A few other comments can be made about figure 3. First, the top
three deciles contain roughly 90 four-digit industries, which together
account for roughly 85 percent of all small businesses. Even the
industries in the second and third deciles have within-industry
employment ([y.sub.j]) that is skewed toward small firms. Second, the
differences between the average [y.sub.j] for the industries within the
first decile and the average [y.sub.j] for the industries within each of
the other deciles are all statistically significant. For example, the p
value of the difference between the first and the second deciles is
0.017, and that of the difference between the first and the fourth
deciles is < 0.001. Likewise, the p values of the differences between
the average [y.sub.j] for the second and third deciles and that of the
fourth decile are both about 0.03. This suggests that it may not be
surprising that most small firms neither grow nor report wanting to
grow, given that most small firms are in industries where the observed
scale of production is on average lower.
Our analysis in this section focuses on employer firms, which are
defined as firms with at least one paid employee. Most U.S. firms,
however, are nonemployer firms. In 2007, for example, there were 21.7
million zero-employee firms, representing roughly 78 percent of all
firms. Often these are second businesses or independent consultants who
report self-employment income on their federal income tax returns. As a
result, despite their importance in the number of firms, nonemployer
firms collectively represent less than 4 percent of all sales or
receipts during a given year. (9) Because many of the existing data sets
exclude the nonemployer firms from their analysis, it is hard to
systematically analyze their composition. Recently, however, the Census
Bureau has released data that sort these firms, in terms of both numbers
and receipts, by broad industry classification. (10) Appendix table A1
summarizes these data for 2007. The patterns documented in table 1 carry
through to nonemployer firms. Most are in a handful of industries where
a larger share of production takes place in small firms. As a result, we
see our broad results as extending to the inclusion of nonemployer
firms.
To summarize, most small businesses operate in a limited set of
narrowly defined industries in which a larger share of economic activity
takes place in small firms than is true of other industries. As we
discuss in later sections, these industries usually do not match the
theoretical models of "entrepreneurship" put forth in the
literature.
Before proceeding, we wish to acknowledge that even within the
industries where most small businesses are located, many firms are still
quite large. John Haltiwanger, in his discussion of our paper that
follows, emphasizes this point. Any theory focusing on the distribution
of firm size thus needs to account for several facts: that most small
businesses are concentrated in a small set of industries, that the
fraction of total employment in small businesses within these small
business-intensive industries is higher than in other industries, but
that even these small business-intensive industries contain many large
firms. We emphasize the first two points whereas Haltiwanger emphasizes
the third.
II. Ex Post Small Business Growth and Innovation
In this section we explore the extent to which small businesses
actually grow or innovate by observable measures in surveys of small
business activity.
II.A. Small Business Growth
It is well documented that small businesses are heterogeneous in
the extent to which they grow, even when one controls for observable
factors such as firm size or firm age. Most recently, Haltiwanger and
others (2010) find little relationship between firm size and firm growth
conditional on firm age. Nearly all of employment growth is driven by
young firms, which also happen to be small. In this section we use some
new and existing data sets to illustrate some additional facts about the
distribution of growth propensities across both small and young firms.
We show that even among young firms, and even among only those young
firms that survive, growth is still rare overall.
Table 2 shows data from the 2005 Business Dynamics Statistics
(BDS). The BDS is produced by the Census Bureau from longitudinal
(annual) establishment-level administrative data similar to the source
data for the SUSB discussed above. It provides measures of gross job
creation and destruction by firm size and age for 1977 through 2009.
Sector-level measures are available for the United States as a whole,
and overall measures are available by state. Again like the SUSB, the
database tracks the employment patterns of employer firms only. The top
panel of table 2 shows the share of all businesses within different firm
age categories that have fewer than 20 employees, both for the entire
economy and within broad BDS sectors. In 2005, of all operating firms
within the economy that had survived less than 10 years, 92 percent had
fewer than 20 employees. The figures for some individual sectors are
quite similar: within the construction sector, for example, 94 percent
of operating "young" firms had fewer than 20 employees. The
bottom panel of table 2 shows the share of employment by sector in these
same small firms. The patterns are similar to those in the top panel:
for example, firms with fewer than 20 employees accounted for 45 percent
of the total employment of all firms that had been in existence for less
than 10 years.
Table 2 highlights two other important facts. First, among mature
firms (firms in existence between 10 and 25 years), most have fewer than
20 employees, and much of total employment is in firms in this size
range. Across the economy as whole, small firms represent nearly 90
percent of all firms and nearly 25 percent of all employment by firms
that have been in existence between 10 and 25 years. Thus, even well
into their life cycle, the overwhelming majority of firms remain small.
Second, and consistent with the results in the previous section,
there is substantial variation among sectors with respect to employment.
In construction, for example, 38 percent of employment within mature
firms is in small businesses, but the figure for manufacturing is only
16 percent. Other sectors in which the employment of mature firms is
concentrated in small businesses include finance, insurance, and real
estate (FIRE), retail trade, and wholesale trade. The heterogeneity in
the firm size distribution across sectors implies differences in
dynamics by sector.
To shed light on employment dynamics for firms of different ages
and industries, we use data from a variety of additional sources,
starting with the 2003 Survey of Small Business Finances. (11) The SSBF,
conducted by the Board of Governors of the Federal Reserve System,
surveys a random sample of businesses with fewer than 500 employees. The
survey is designed to measure the financial position of these
businesses, but it also contains other background questions. In 2003
firms were asked to state whether in the past year, and over the past 3
years, their total employment grew, remained the same, or contracted.
Table 3 summarizes the responses to these questions by firms with
fewer than 20 employees. We break down the responses by firm age to
highlight differences between newer and more established businesses. As
the table shows, the overwhelming majority of small firms do not grow by
adding employees from year to year or even over 3-year periods. (12)
Only 14 percent of these small businesses added an employee between 2002
and 2003, and only 21 percent did so between 2000 and 2003. Thus, by
this measure, roughly 80 percent of surviving small firms did not grow
at all even over a relatively long period. The percentages are slightly
higher among small firms that had been in existence between 1 and 10
years, but only 19 percent of these grew between 2002 and 2003, and only
28 percent grew between 2000 and 2003. These data show that although
most aggregate employment growth may come from small (new) firms growing
big, the vast majority of small (new) firms do not grow, even over
longer horizons.
Within the modest share of growing firms, the SSBF data do not tell
us by how much the firms grew. To address this question, we turn to the
Kauffman Firm Survey (KFS), a panel study administered by the Kauffman
Foundation of 4,928 businesses that were newly founded in 2004. (13) As
Haltiwanger and others (2010) show, it is new firms that contribute the
most, on average, to job growth. Yet as we have just shown, job growth
is rare among typical small businesses: it is not true that most new
businesses generate employment growth. To create the KFS sample,
researchers began with a sample frame of nearly 250,000 businesses
started in 2004, from a database created by Dun & Bradstreet, a firm
that collects and manages commercial data. From these data, the KFS
oversampled businesses in high-technology industries and businesses that
reported high employment in research and development in the
business's primary industry. The final sample of 4,928 firms is
resurveyed annually in follow-up interviews. As of this writing,
public-use data are available on these firms through 2009. For the work
below, we consider only the 2,617 firms in the sample that survived
through 2008. We use the survey weights provided by KFS, which are
designed to make the firms in the sample representative of all new firms
in the economy.
Because the KFS is a 4-year panel, we can assess the growth rate of
employment for new businesses within the KFS over 4 years. In each wave
of the survey, the KFS asks firms to report their number of employees.
Table 4 shows that between 2005 and 2008, 42 percent of the surviving
firms in the KFS reported an increase in employment. Very few, however,
added more than one or two employees: 89 percent added 5 or fewer
employees, and 96 percent added 10 or fewer.
The results from the KFS hold more broadly in the United States. We
find that small businesses within the top small business industries
(those listed in table 1) actually have lower than average job creation
rates. To see this, we pool employment change data from the SUSB from
2003 to 2006. These data are released as a companion to the levels
reported in the SUSB annual data. Using the same administrative data,
the Census Bureau measures the number of jobs created (by either
expanding or new establishments) or destroyed (by either contracting or
exiting establishments) at the establishment level and aggregates these
into annual measures of gross job creation and destruction by industry
and firm size. (14) Examining industries at the four-digit level, we
compute for each size category the gross job creation rate (jobs created
at continuing establishments), the gross job birth rate (jobs created at
newly opening establishments), and the gross job destruction rate (jobs
lost at both contracting and exiting establishments). Following Steven
Davis and others (1996), we define these rates as follows:
[g.sup.Ms.sub.jt] = [M.sup.s.sub.jt]/([e.sup.s.sub.j,t] +
[e.sup.s.sub.j,t+1]/2,
where [M.sup.s.sub.jt] represents a measure of job creation or
destruction (either jobs created from expansion, jobs created from
births, or jobs destroyed from contracting and exiting establishments)
by small businesses within industry j between period t - 1 and t, and
[e.sup.s.sub.jt] is defined as above to be the number of employees in
small businesses within industry j during period t. We again define
small businesses to be those firms with fewer than 20 employees. Davis
and others (1996) show that this specification of the growth rate has a
number of desirable properties: it accommodates entry and exit, and it
is equivalent to a log difference up to second order.
We use these growth rates to ask whether or not having a large
fraction of small businesses in an industry can predict the degree of
job creation or destruction in small businesses within that industry,
conditional on aggregate industry characteristics. To do this, we
estimate the following:
[g.sup.Ms.sub.jt] = [[gamma].sub.0] + [[gamma].sub.1][x.sub.j] +
[GAMMA][Z.sub.j] + [[mu].sub.t] + [[eta].sup.s.sub.jt],
where [g.sup.Ms.sub.jt] is either the gross job creation rate, the
gross job birth rate, or the gross job destruction rate for small firms
in industry j, depending on the regression. As above, [x.sub.j]
represents the share of small businesses in industry j out of all small
businesses across all industries. [Z.sub.j] is a vector of
industry-level controls, which include industry-wide measures of the
gross job creation rate, the gross job birth rate, and the gross job
destruction rate, and p, is a vector of year dummy variables. The sample
for this regression is all four-digit industries with nonmissing
measures of [M.sup.s.sub.jt] during 2003-06. This results in 929
observations for the job creation regressions, 666 observations for the
job birth rate regressions, and 656 observations for the job destruction
regressions. The sample sizes differ because more data at the four-digit
industry level are missing for the measures of job birth and job
destruction than for the job creation measure.
Table 5 reports the estimation results. We estimate each
specification first giving each industry equal weight (top panel), and
second weighting each industry in proportion to the share of small
businesses in the industry (bottom panel). The weighted estimation is
similar to one using a grouped-data estimator and would deliver the same
point estimates as firm-level data if the employment shares of small
firms within an industry were equal. (15) The results support our
earlier claim that the "typical" small business does not
create jobs. The small business share of an industry has little to say
about small business job creation through new small businesses, or about
small business job destruction (fourth and fifth columns of table 5).
However, it is a powerful predictor of weaker than average small
business job creation for existing firms (first three columns). Small
businesses in industries with the largest shares of all small businesses
(those with a high [x.sub.j] as shown in table 1) grow more slowly than
the average. These results hold even when we control for each
industry's overall characteristics (compare the first two columns
of table 5). One might be concerned that the difference between the
strong effects for job creation and the absence of effects found for job
births and job destruction could stem from differences in the samples
across the regression. The third column of table 5 shows that such
concerns are unwarranted. In this job creation regression we restrict
the sample to those industries with nonmissing job birth and job
destruction data. The results are unaltered from those in the first
column.
According to the weighted results, for each percentage-point
increase in an industry's share of small businesses, that
industry's small business job creation rate falls by a little less
than three-quarters of a percentage point. To provide greater context, a
1-standard-deviation increase in [x.sub.j] (1.1 percentage points)
reduces the job creation rate by roughly 0.8 percentage point. The
average weighted job creation rate for the sample was 14.6 percent. So a
1-standard-deviation increase in an industry's share of small
businesses reduces the industry's small business job creation rate
by about 6 percent (0.8 divided by 14.6). When industries are treated
equally, a 1-standard-deviation increase in [x.sub.j] reduces the
industry's small business job creation rate by roughly 8 percent.
All these results are robust to alternative specifications of industry
controls.
It may be surprising at first that so little job creation comes
from the industries that most small business owners are likely to enter.
However, this finding is consistent with an understanding of the
important heterogeneity among small businesses. Most small businesses
(those highlighted in table 1) start small and stay small throughout the
life of the business.
We draw three conclusions from the results in tables 2 through 5.
First, there is substantial skewness across firms in the extent to which
they grow over time. Although some firms do increase their employment
over time, most do not. Only a small proportion of small firms add more
than 10 employees over the life of the business. Reflecting this, the
bulk of firms still employ fewer than 20 employees when they are mature.
Second, even among new or young firms, most firms do not grow by any
meaningful amount, even conditional on survival. Finally, some of the
heterogeneity in employment growth for small firms is explained by
industry. Although many mature businesses in manufacturing are quite
large, the vast majority of mature businesses in other sectors, such as
construction, remain quite small. The industries in which firms tend to
remain small are those that tend to make up the bulk of small
businesses.
We again stress that even within these small business-intensive
industries there is considerable heterogeneity and skewness. The focus
of this paper is on the behavior of small firms, which in our selected
small business-intensive industries account for nearly half of
employment. John Haltiwanger, in his discussion that follows, looks
closely at the other half. It turns out that in many of these industries
where employment in small firms is overrepresented, the firms within
these industries can also be more dynamic than average. Retail trade,
for example, is composed largely of small local shopkeepers and big-box
stores. Job creation in this sector was almost 10 percentage points
higher over 2003-05 than the average for other sectors excluding retail
trade. However, almost all of this job creation was due to a relatively
small number of firms. The high degree of skewness, especially in these
industries, thus makes analyzing the averages very deceiving. As we have
shown, the typical (modal or median) small business is not creating
jobs. In section VI we will return to the potential implications of
these findings for public policy.
II.B. Small Business Innovation
In this subsection we document that there is also substantial
heterogeneity across firms in the extent to which they successfully
innovate along observable measures. Again, although some authors have
shown that a large share of measured innovation (patent applications,
for example) is attributable to small businesses, the converse is not
true. (16) Most small firms do not seem to innovate along those
observable margins. Before proceeding, we stress that it is hard to
measure all aspects of potential small business innovation using only
the surveys we are analyzing. As a result, we focus on some broad
measures of innovation about which the surveys do provide information.
We begin by documenting that very few new firms obtain patents,
trademarks, or copyrights during the first 4 or 5 years of their
existence. For this we use two data sources. First, we again use the
Kauffman Firm Survey, focusing on the same sample as above. The KFS asks
respondents to report separately whether they have already applied or
are in the process of applying for any patents, copyrights, or
trademarks. In 2008, when the firms in the sample had been in business
for 4 years, only 2.7 percent of the businesses in the sample had
already applied or were in the process of applying for patents (table
6). Larger shares had applied for copyrights and trademarks, but still
most firms were not innovating, at least according to these crude
observable measures. According to the KFS, nearly 85 percent of small
businesses did not acquire a patent, trademark, or copyright during
their first 4 years of existence.
We augment our analysis of patents and other measures of innovation
using data from the Panel Study of Entrepreneurial Dynamics II (PSED).
(17) The PSED started with a nationally representative sample of 31,845
individuals. An initial screening survey in the fall of 2005 identified
1,214 "nascent entrepreneurs." To be considered a nascent
entrepreneur, individuals had to meet the following four criteria.
First, the individual had to currently consider himself or herself as
involved in the firm creation process. Second, he or she had to have
engaged in some business start-up activity in the past 12 months. Third,
the individual had to expect to own all or part of the new firm being
created. Finally, the initiative, at the time of the initial screening
survey, could not have progressed to the point that it could have been
considered an operating business. The goal was to sample individuals who
were in the process of establishing a new business.
In the winter of 2006, after the initial screening interview, these
1,214 respondents were surveyed about a wide variety of activities
associated with their business start-up. They were asked detailed
questions about their motivations for starting the business, the
activities they were currently undertaking as part of the start-up
process, the competitive environment in which the business would
operate, and their expectations about the desired future size and
activities of the business. Follow-up interviews occurred annually for 4
years, so that the data also have a panel dimension.
When analyzing the PSED data, we use three samples. The first
consists of all 1,214 respondents. The second includes only the 602
respondents who actually had positive revenue at the time of their first
detailed interview in 2006. This sample distinguishes people who only
said that they were planning to start a business from those who actually
followed through and engaged in some market business activity. The third
sample consists of the 162 respondents who had positive revenue from the
same business venture in 2010, 4 years after the first interview.
With respect to innovative activity, the PSED asks three different
types of questions. The first is similar to the questions in the KFS
about patent, trademark, and copyright applications. However, instead of
being asked about the three measures separately, in the PSED they are
asked one question covering all three. As table 7 shows, only 5 percent
of the new firms (6 percent of those with positive revenue) applied for
patents, trademarks, or copyrights during their first few years in
existence. By the fifth year of operation, surviving firms appear
similar to those in the KFS, with roughly 18 percent having obtained a
patent, trademark, or copyright.
Of course, patents, copyrights, and trademarks are imperfect
measures of innovation. Many firms can innovate without applying for a
patent, and many firms can trademark their company name without doing
any real innovation. We have focused first on these measures because
they are easily observable in both the KFS and the PSED. The PSED,
however, also provides data on broader measures of innovation. In a
separate set of questions, businesses are asked directly whether they
have "developed any proprietary technology, processes, or
procedures." This question results in a slightly broader measure of
innovation than patent, trademark, and copyright applications in that it
conceivably covers a more fluid set of activities related to innovation
in production or in the firm's business model. Yet only 7 or 8
percent of new businesses (depending on the sample) reported that they
had developed any proprietary business practices or technology during
their first few years in business. Even conditional on survival 5 years
later, 80 percent of firms still reported not having developed any
proprietary technology, process, or procedure. (18)
The PSED asks one last broad question about the potential
innovation taking place within the firm. This question asks about how
the product or service produced by the business compares with the
products and services of other producers in the firm's market:
"Right now, are there many, few, or no other businesses offering
the same products or services to your [intended] customers?" The
answers to this question are informative in that they indicate whether
the firm is providing a new product or service to existing customers or
an existing product or service to potentially new customers. Across the
three samples, between 36 and 43 percent of new business owners reported
providing a service similar to that of many existing firms in the market
to an existing customer base; these businesses, more often than not,
provide a standardized service (such as plumbing) to existing local
customers. Fewer than 20 percent of respondents reported that no other
business was providing their expected product or service to their
expected customer base.
The responses to this question varied substantially across business
owners in different industries (results not shown). For example, owners
who reported starting a business in the professional, health,
construction, and real estate industries were between 7.5 and 9.5
percentage points more likely to report that they were starting their
business in an area where many current providers already served their
expected customer base. Owners in these same industries were nearly 10
percentage points less likely to report that they were providing a new
product or service or were targeting an underserved customer base.
III. Ex Ante Expectations about Growth and Innovation
In this section we document that many business owners have no
expectation or desire to grow or innovate when they start their
business. One of the strengths of the PSED is that it asks the nascent
business owners about their expectations for the business, their desired
future business size, and their motivations for starting the business.
For example, owners of all new firms are asked which of the following
two statements best describes their preference for the future size of
the new business: "I want this new business to be as large as
possible," or "I want a size I can manage myself or with a few
key employees." The top row of table 8 shows the responses to this
question across our three PSED samples. For the sample consisting of
those whose businesses lasted to 2010, we report their expectations when
they were first asked in 2006. Around three-quarters of all respondents,
regardless of sample, reported that they wanted to keep their business
limited to at most a few key employees.
Of course, the meaning of keeping the business to "a few key
employees" could vary across respondents. In a separate part of the
survey, the respondents are asked to state their expectation as to the
firm's employment 5 years hence. Again, we report the responses for
each sample when they were first asked in 2006. The median number given
was either 3 or 4, depending on the sample. Even respondents at the 75th
percentile expected to employ only between 6 and 10 employees. These
findings indicate that not only do very few small businesses grow, but
most small business owners do not want or expect their business to grow,
when asked at the time of its formation.
The PSED also asks about expected innovative activity: business
owners are asked, at the inception of their business, whether they
expect to innovate in the future. These results, also reported in table
8, show that only roughly 15 percent of all new businesses plan to
develop proprietary technology, processes, or procedures in the future.
The numbers are slightly higher with respect to expectations about
future patent, copyright, and trademark behavior. A likely reason is
that many firms trademark the name of their business even if they do not
apply for patents or copyrights.
Business owners in the PSED are also asked if they expect research
and development to be a major priority for the business. As table 8 also
shows, nearly 80 percent of all owners of new businesses reported that
they had no plans for R&D to be a major priority.
The results in table 8 suggest that the observed lack of innovation
and growth may be deliberate: when starting their business, most
business owners have no plans to grow or innovate in the future.
Interestingly, despite these expectations, new business owners remain
committed to starting and running a firm. In the next section we examine
the stated motives of nascent small business owners and explore how
these motives correlate with expectations.
IV. Motivations for Starting a Business
To explore heterogeneity in founders' motives, we again turn
to the PSED data. As part of the initial survey of the PSED, the
business owners were asked, "Why do [or did] you want to start this
new business?" Respondents could report up to two motives. The
respondents provided unstructured answers, which the PSED staff coded
into 44 specific categories. We took the raw responses to the question
and created five broad categories of our own: nonpecuniary reasons,
reasons related to the generation of income, reasons related to the
desire to develop a new product or implement a good business idea,
reasons related to a lack of better job options, and all other reasons.
The main responses in the nonpecuniary category include "want to be
my own boss," "flexibility/set own hours," "work
from home," and "enjoy work, have passion for it/hobby."
The main responses in the generating income category include "to
make money" or "need to supplement income." The main
responses in the new product or business idea category include
"satisfy need," "there is high demand for this
product/business," "untapped market," and "lots of
experience at work." Appendix table A2 lists the 44 specific PSED
categories, grouped into our five broad categories. For each specific
category, the table reports the number of PSED respondents citing a
motive in that category either at first or at second mention.
The columns labeled "First response" in table 9 show the
distribution of the first responses given by each respondent by broad
category, and the columns labeled "Either response" show the
distribution of all responses, for each of the three PSED samples. Three
things should be noted before we discuss the results. First, only 60
percent of respondents provided a second response. Second, given that
the respondents could provide any answer they wanted, the first and the
second response often fell into the same broad category. (For example,
many respondents answered "want to be my own boss" and
"flexibility/set own hours," both of which we record in the
nonpecuniary benefits category.) Third, the sum of the numbers in the
first column exactly equals 100 percent whereas the sum of the second
column exceeds 100 percent, given that respondents could offer a second
response.
The main result from table 9 is that although there is substantial
heterogeneity across respondents in their reported primary reason for
starting a small business, nonpecuniary benefits play a leading role for
most respondents. This result is consistent across all three PSED
samples. For example, between 35 and 37 percent of first responses
across all samples referred to nonpecuniary reasons for the business
start-up decision. Combining the first and second responses, we find
that over half of all respondents in all samples stated that
nonpecuniary benefits were an important component of their decision.
The second most commonly expressed motivation was having a good
business idea or creating a new product. Roughly 30 percent of first
responses and roughly 38 percent of combined responses fell in this
category. Many respondents also reported that they wanted to generate
income: answers in this broad category represented roughly 20 percent of
first responses and 34 percent of combined responses. Finally, very few
respondents reported starting the business because of a lack of other
employment options. (19)
In the remainder of this section, we explore to what extent the
respondents' reported motives predict their expected and actual
values on the growth and innovation measures. We focus on two motives in
particular: the desire to implement a good business idea or create a new
product, and nonpecuniary motives. The first is the motive most closely
associated with the traditional role of an entrepreneur, whereas the
second is typically ignored or considered only anecdotally. We define
for each motive a dummy variable that is equal to 1 if that motive was
identified in either the first or the second response. Then we run a
simple regression of the various measures described in tables 7 and 8 on
the two dummy variables. (20) Since a motive can be identified in either
mention, it is possible for both dummies to equal 1.
Table 10 presents the results of these regressions. We show results
for the first two PSED samples only, in the top and bottom panels. Given
the small size of the third sample (those still earning revenue in
2010), statistical significance is an issue in interpreting the
coefficients. However, even in this sample the signs of the coefficients
display patterns very similar to those for the other two samples.
In the top panel of table 10, which shows the results for all
respondents, the first column reports the constant from each regression.
This represents the unconditional mean for those individuals who did not
report starting their business for either nonpecuniary motives or
motives related to creating a new product or implementing a new business
idea. The next two columns show the coefficients on the two dummy
variables defined above. These coefficients can be interpreted as the
percentage difference in probability of the outcome (or difference in
the employment forecast) relative to respondents that mentioned neither
the new product or business idea motive nor nonpecuniary motives?'
The fourth column shows the difference between the coefficients on the
two dummy variables. This can be interpreted as the difference in
probability of the outcome (or employment) for those business owners who
mentioned exactly one of those motives. For example, respondents who
mentioned creating a new product or implementing a business idea and did
not mention nonpecuniary motives were 13 percent less likely than those
that specified nonpecuniary motives but not a new product or business
idea to enter a market already offering the same product or service. The
final column shows the p value of a two-sided test for equality of the
two coefficients. The bottom panel of table 10 reports results for the
same regression using the second PSED sample (respondents with positive
revenue in 2006).
The results of these regressions show that individuals who start
their business because they think they have a good business idea or
because they want to create a new product are much more likely to want
to grow, to want to innovate, and to actually innovate. Conversely,
those who start a business for nonpecuniary reasons are less likely to
want to grow, to want to innovate, and to actually innovate. As
mentioned above, those reporting nonpecuniary motives were much more
likely to enter an already crowded market than those with a new business
idea or product. Likewise, they were 5.1 percentage points less likely
to report that they had already developed some proprietary technology or
processes as part of their business start-up, and 9.4 percentage points
less likely to report expecting to get a patent, trademark, or copyright
in the future. The p values on both these differences are 0.01.
As can also be seen in table 10, those who reported starting their
business because they had a new business idea or product were much more
likely to want the business to have a higher number of employees in 5
years, and to want to grow their business, than those who started for
nonpecuniary reasons. For example, those who started because they had a
good business idea were 8.3 percentage points less likely to report
wanting to keep their business to a few key employees.
We wish to highlight a few additional results not shown in table
10. First, there is little statistical difference in survival rates to
2010 between those business owners who reported nonpecuniary benefits as
a primary motivation and those who reported a new business idea or
product as the reason they started. If anything, in some samples and
specifications, those who reported nonpecuniary benefits as a primary
motivation had a higher probability of survival. (22) Second, there is
no statistical difference in actual firm size in 2010 across the
different groups. The reason is that nearly all firms had only 1 or
fewer employees even 4 years after the business started. There is not
much variation across the firms in this small sample of survivors. This
is consistent with the results in table 3 showing that most surviving
firms remain very small. Finally, there is some variation across
industries in the relative importance of nonpecuniary reasons versus
wanting to implement a new business idea or create a new product: owners
of businesses in the finance industry were statistically much more
likely, relative to other industries, to report nonpecuniary benefits as
an important motive, and a similar pattern appears among those starting
businesses in retail trade. Two industries where the dominant reason to
start the business involved a new product or business idea are
manufacturing and wholesale trade. The results lack enough statistical
precision to allow decisive conclusions about the other industries.
The results in this section show that there is substantial ex ante
heterogeneity across individuals in their reasons for starting a
business. Only a fraction of firms are started because the owner has a
good business idea or a new product to bring to market. However, these
business owners at the time of inception are more likely to report a
desire to grow and innovate and to achieve higher actual realizations of
innovation. Many owners, in contrast, report nonpecuniary benefits as an
important driver of their behavior. Incorporating such ex ante
heterogeneity into models of small business dynamics will almost
certainly alter conclusions about the importance of ex post measures of
heterogeneity such as stochastic productivity draws or binding financial
constraints. We turn to this discussion in the next section.
V. Why Heterogeneity in Starting Motives or Expectations Can Matter
There are a number of reasons why ignoring the ex ante
heterogeneity in motives and expectations may matter. Here we sketch how
this ex ante heterogeneity confounds inferences in a number of relevant
contexts. We consider two literatures: the literature on firm dynamics,
and the literature on measuring the private equity risk-return
trade-off. Finally, we assess how our work relates to recent papers
documenting the nature and growth patterns of small businesses in
developing economies.
V.A. Firm Dynamics
In theoretical models, differences in employment growth across
firms are attributed to either differences in entrepreneurial ability
(for example, Lucas 1978), differences in realized productivity draws
(for example, Simon and Bonini 1958, Jovanovic 1982, Pakes and Ericson
1989, Hopenhayn 1992), differences in access to capital markets (for
example, Evans and Jovanovic 1989, Vereshchagina and Hopenhayn 2009), or
some combination of the above (for example, Clementi and Hopenhayn
2006). Although all of the above are potential drivers of firm dynamics,
the results we have documented here suggest that these stories are at
best incomplete. Differences in luck, talent, or credit market access
are not the only determinants of firm size. As we have shown, there is
also substantial ex ante heterogeneity in the desires and expectations
of new business owners with respect to the future growth of their
business. Some firms do not grow or innovate simply because they do not
want to grow or innovate.
What drives these differences in ex ante expectations and desires?
Our results point to at least two potential channels. First, many small
business owners start their business, in part, because of the
nonpecuniary benefits associated with small business ownership. As seen
from the PSED data, many small business owners report starting their
business because they value the control and flexibility that small
business ownership offers. If these benefits diminish with firm size,
individuals who start for these reasons will prefer to keep their
business small. We do find evidence of such correlations in the data:
those business owners who reported starting their business in part for
nonpecuniary reasons were more likely to want to keep their firm small
well into the future.
Second, some businesses may stay persistently small because they
are in industries where the naturally efficient scale is small. Many
small business owners are dentists, plumbers, real estate and insurance
agents, small shopkeepers, and beauticians. Within these industries, the
productivity of the firm is directly linked to the individual's
skill set. Given that the fixed costs of production may be small
relative to the variable costs, optimal firm size may be quite small. As
a result, firms in these industries may start with no expectations of
growth. (23) These firms may be particularly attractive to business
owners driven by nonpecuniary motives.
Pugsley (2011a, 2011b) formalizes the insights put forth in this
paper by writing down models of small business formation and small
business dynamics where individuals are allowed to differ in the utility
they derive from small business ownership and where industries differ in
their natural returns to scale. In these models, many of the predictions
of the standard models of firm dynamics can be replicated in a model
with no differences across firms in entrepreneurial ability and no
difference across firms in their financing constraints. Two important
results emerge from these papers. First, Pugsley (2011a) shows that the
existence of nonpecuniary benefits can generate a positive relationship
between wealth and starting a business, by making business ownership a
normal good that wealthier individuals "purchase" as their
marginal utility of consumption diminishes. Second, Pugsley (2011b)
shows that there is not a one-for-one mapping between the distribution
of firm size and productivity draws (like the ones emphasized in much of
the literature outlined above) when industries differ in their fixed
costs and owners have a preference for keeping their business small.
This finding cautions against using unconditional firm-level dynamics to
estimate a process for entrepreneurial productivity.
Finally, much of the empirical work on firm dynamics proceeds
either by studying the universe of firms or by focusing on a sector
thought to be representative of that universe (typically manufacturing).
It is in this empirical context that the applicability of Gibrat's
law (for example, Sutton 1997), which states that firm growth rates are
on average independent of size, or of Zipf's law, which states that
the distribution of firm sizes appears to follow a particular power law
(for example, Gabaix 2009), is frequently demonstrated. Why these
empirical regularities appear at the aggregate level is an interesting
question. However, consistent with Pugsley (2011b), that is not to
suggest that imposing this structure on a particular industry, or
assuming a representative industry typified by manufacturing, is
appropriate. The concentration of small businesses varies considerably
across industries, and the heterogeneity we consider is especially
important for the industries we highlight in this paper. There is
considerable cross-industry variation in the distribution of firm sizes,
even conditional on average firm size, as figure 3 illustrated for broad
industry groupings.
V.B. Understanding the Risk-Return Trade-off
A separate literature assesses the risk-return trade-off of small
business owners. For example, Tobias Moskowitz and Annette
Vissing-Jorgensen (2002) document that the returns to investing in
private equity (business ownership) are no higher than the returns to
investing in public equity, despite the poor diversification and higher
risk. Their focus is only on the pecuniary returns of private business
investment, and their corresponding analysis spans a large class of
businesses, many of which are the small businesses we study here.
However, even among venture-backed start-ups, which are a tiny traction
of small businesses, the risk-return trade-off looks poor. Robert Hall
and Susan Woodward (2010) show that even among the highly skilled
population of venture capital-backed entrepreneurs, potential
entrepreneurs would be roughly indifferent between salaried employment
and launching a venture-backed start-up, considering the high
idiosyncratic risk of the payoffs to entrepreneurship, at modest levels
of risk aversion and wealth.
Not surprisingly, a model with nonpecuniary benefits can help to
explain these findings. If there are private benefits to small business
ownership (relative to allocating effort to the labor market), the
measured pecuniary return could be lower than the total return. Our
results above suggest that for many individuals, nonpecuniary benefits
are an important motive for starting their small business. Although the
results above are based on survey reports, they are consistent with the
work of Barton Hamilton (2000) showing that the median small business
owner receives less in accumulated earnings over time than he or she
would in paid employment. (24)
Overall, our results suggest that for many individuals,
nonpecuniary benefits could be an important factor driving their small
business formation. Incorporating such preferences into models of small
business formation can alter the assessment of the risk-return trade-off
of small business ownership.
V.C. Small Businesses in Developing Economies
Recent work has emphasized the fact that most small businesses in
developing economies do not grow, do not innovate, and are started
because of a lack of jobs in the larger firms within the economy. For
example, La Porta and Shleifer (2008) examine the importance of the
informal sector in developing economies. They conclude that, on average,
the small firms that populate the informal sector in developing
economies are much less productive than similar small firms in the
formal sector. Given the low quality of the inputs (including human
capital) into their production, it is not surprising that these small
informal firms do not grow or innovate in any observable way. Banerjee
and Duflo (2011) document the existence of "reluctant
entrepreneurs" in developing economies. They find that most
individuals who own a small business in the developing countries they
analyze do not grow, are not profitable, and often enter because of the
lack of jobs in larger, established firms.
The results in the present paper both complement this literature
and show that different mechanisms are at play in a developed country
like the United States. As in developing economies, it is true that most
small businesses in the United States do not grow. However, the reasons
for starting small businesses and the nature of small business owners
seem quite different in the United States than in developing countries.
Many U.S. small business owners are highly skilled (lawyers, doctors,
dentists, and others). There is little relationship between formal years
of schooling and either the propensity for small business entry or small
business survival. Additionally, as shown above, very few U.S. small
business owners (fewer than 4 percent) report starting a business
because of a lack of employment options. In other words, it does not
appear that U.S. small business owners are "reluctant
entrepreneurs."
Overall, our results showing that most small businesses in the
United States do not want to grow or innovate is consistent with
others' findings for small businesses in developing economies, but
the underlying reasons may be very different. A more formal analysis of
the similarities and differences between small businesses within
developed and developing economies would be a worthy area for future
research.
VI. Policy Implications
Economic arguments for subsidizing small businesses hinge on the
claim that small businesses are important contributors to aggregate
innovation and growth but that market forces alone fail to allocate
sufficient resources to the sector. These market failures may stem from
technological spillovers ignored by entrepreneurs, or from financial
constraints that inhibit an optimal quantity of capital from reaching
the small business sector. The subject of entrepreneurship and
technological spillovers is well studied in the endogenous growth
literature (for example, Audretsch, Keilbach, and Lehmann 2006, Acs and
others 2009). If a substantial portion of economy-wide R&D occurs in
small firms, the social returns to their entrepreneurship could far
exceed the private returns. Charles Jones and John Williams (1998), for
example, find the optimal level of investment in R&D to be two to
four times the observed level. Additionally, subsidizing small
businesses may be appropriate if liquidity constraints or other
financial market imperfections prevent small businesses from securing
the financing they need to bring their innovations to market (Evans and
Jovanovic 1989, Evans and Leighton 1989).
In the belief that there are social spillovers from small business
innovation or that small businesses face liquidity constraints, many
developed economies have enacted policies that favor small businesses
relative to established firms. The subsidies to small businesses come in
two potential forms. The first are direct subsidies, where the explicit
intent is to promote small business activity. Within the United States,
for example, small business subsidies include subsidized or guaranteed
loans, access to special lending programs, exemption from various
regulations, preferential treatment when awarding government contracts,
and preferential treatment through the tax code. (25) Adam Looney, in
his comment on our paper that follows, discusses a number of these
public small business subsidies in the United States. These subsidies
are often linked explicitly to firm size: only firms with fewer than a
certain number of employees are eligible. As a result, many of these
subsidies promote small business entry but do not promote small business
growth, because if the firms grow beyond a certain size, the subsidy no
longer applies.
Subsidies of the second type are indirect. For example, because
nonpecuniary benefits are not taxed, sectors where such benefits are a
larger fraction of total compensation are effectively tax preferred
relative to other sectors. To the extent that small business ownership
offers larger nonpecuniary benefits relative to owning a larger business
or being a wage worker, the small business sector would be tax preferred
even if there were no other direct subsidies. Additionally, a large
literature shows that small business owners are much more likely than
wage and salary workers to underreport their income to tax authorities.
(26) If it is easier to underreport income if one owns a small business,
the small business sector again would be tax preferred relative to other
sectors even if there were no additional direct subsidies. The important
point is that although policymakers and researchers often invoke the
potential benefits of direct small business subsidies, there is very
little quantitative research documenting the actual benefits and costs
of these subsidies. The results in this paper suggest that the potential
costs may be nontrivial.
These potential costs come from two sources. First, as we show
above, the bulk of small businesses report ex ante that they do not want
either to grow or to innovate. And, as anticipated, most small firms do
not grow or innovate. Linking small business subsidies to firm size may
support the handful of firms that eventually turn into Googles or
Microsofts, but it also encourages the entry of real estate agents,
small law firms, and small construction firms, among others, for which
the social spillovers and growth potential may be much smaller. To the
extent that these subsidies alter the behavior of firms that start for
reasons unrelated to growth and innovation, the policies can be
distortionary by allocating more resources to the small business sector
than is optimal. Second, if the benefits associated with subsidizing
small business activity come from the small businesses actually growing,
yet the subsidies are linked to firm size, they may actually prove
counterproductive by inhibiting firm growth. If a firm grows beyond a
certain size, the small business subsidy no longer applies, and
therefore the firm has an incentive to remain small.
A companion paper (Pugsley 2011a) illustrates the potential costs
of small business subsidies in a simple static general equilibrium model
of small business formation and occupational choice. Within the model,
industries differ in their natural return to scale. Households differ in
the size of the nonpecuniary benefit they receive (in flow utility) from
starting a business. To highlight the potential costs of subsidizing
small business activity in the model, it is assumed that there are no
differences across individuals in terms of talent, no social spillovers
from small business formation, and no liquidity constraints preventing
firm formation. These extreme assumptions allow one to focus on the
potential costs of subsidizing small businesses in a world where
individuals get nonpecuniary benefits from small firm ownership.
Individuals in the model can allocate their labor either to running a
business or to working for some other business. Household-run businesses
cannot grow to their efficient scale without forfeiting the utility
flow.
The model makes many predictions that can inform researchers and
policymakers about the potential costs of small business subsidies.
First, subsidizing small business (and funding those subsidies with
taxes on labor income) will distort the allocation of production within
the economy toward small businesses. Individuals choosing to start a
small business trade off the size of their nonpecuniary benefits from
owning a small business with the loss in wages they incur from forgoing
the benefits of increased returns to scale in production. When small
business activity is directly subsidized, the economy as a whole becomes
less productive, given that individuals respond by choosing to work in
small (subsidized) self-owned firms rather than establishing larger
firms that can produce at lower average cost by exploiting the returns
to scale. Notice that such distortions could occur even in a world where
there are no direct subsidies to small businesses, only the indirect
subsidies discussed above.
Moreover, in such a model where the nonpecuniary benefits of small
business ownership are a normal good, the subsidies to small businesses
are regressive. The reason is that, in a world without small business
subsidies, high-wealth individuals are much more likely than others to
start a business, because they are more able to afford (in utility
terms) forgoing the benefits of increased returns to scale when they
start their business. For the wealthy, a small business subsidy is
simply a transfer tied to activity that they were more likely than
others to do anyway. Not only does the existence of nonpecuniary
benefits to small business ownership result in subsidies being welfare
reducing, but lower-wealth households suffer more from the subsidy than
do higher-wealth households.
To our knowledge, no empirical work evaluates whether subsidizing
small businesses results in positive net present value. Addressing this
question would seem to be a very important area for future research. Our
work suggests that subsidies may be less distortionary if they are
targeted at growth and innovation rather than mostly linked to firm
size. Such policies could address the concerns raised by our results in
at least two ways. First, we have shown that most small businesses
operate in industries with potentially smaller natural scales of
production. Business owners with little intention to grow or innovate
may select into these industries for that very reason. If the subsidy is
focused on the intensive margin, it is more likely to be taken up by a
business owner focused on growth or innovative activity. Subsidies could
lower the cost of credit for existing firms, and by increasing their
value entice productive entrepreneurs with high wage employment
opportunity costs. Second, if nonpecuniary compensation is independent
of the scale of the firm, the incidence of an expansion subsidy would be
undistorted by nonpecuniary benefits. If anything, nonpecuniary benefits
may help separate businesses that want to grow from businesses that
would prefer to remain small. Of course, there may be other social
virtues to noninnovative small businesses, such as supporting
communities and neighborhoods, which are aided by subsidizing the entry
and exit margins. However, when targeting job creation or innovative
risk taking, our findings suggest caution when supporting businesses
purely on the basis of size. At a minimum, more research is necessary to
better understand both the costs and the benefits of subsidizing small
business activity.
VII. Conclusion
In this paper we have shown that there is substantial skewness in
the desires and expectations of individuals who start small businesses.
The vast majority of small business owners do not expect to grow, report
not wanting to grow, expect never to innovate along observable
dimensions, and report not wanting to innovate along observable
dimensions. We show that there is also substantial heterogeneity in
individuals' reported reasons for starting their business. In
particular, only slightly more than one-third of new business owners (on
the eve of their start-up) reported that they were starting their
business because they had a new product or service that they wanted to
bring to market. Instead, the most common reason given for starting a
business was the existence of nonpecuniary benefits. Individuals
reported that they like being their own boss and like the flexibility
that small business ownership provides.
Our results suggest that much of the current literature has
overlooked an important component of many small businesses. Essentially
all of the current literature on firm dynamics explains the ex post
distribution of firm size with models emphasizing differences in
entrepreneurial talent, differences in entrepreneurial luck, and
differences in access to credit markets. The results in this paper,
however, suggest that another factor may be at play: many small business
owners simply do not wish to grow big or to innovate along observable
dimensions in any meaningful way. The paper shows two potential reasons
for the ex ante differences in desires and expectations with respect to
future growth. First, the natural scale of some industries may be quite
small. For example, the fixed costs to become a plumber, barber, lawyer,
or insurance agent may be small relative to the variable costs, making
the returns to scale quite small. Second, the existence of nonpecuniary
benefits of owning a small business, such as increased flexibility and
control, may induce individuals to forgo some natural benefits of
increased scale in exchange for higher utility. Regardless of the exact
reason, most individuals who start a small business have little desire
or expectation to grow their business beyond a few employees.
Recognizing these characteristics common to many small businesses
has immediate policy implications. Often subsidies targeted at
increasing innovative risk taking and overcoming financing constraints
are extended to small businesses generally. Our analysis cautions that
this treatment may be misguided. We believe that these targets are
better reached through lowering the costs of expansion, so that they are
taken up by the much smaller share of small businesses that do aspire to
grow and innovate. In fact, the U.S. Small Business Administration
already partners with venture capitalists whose high-powered incentives
are aligned with finding these small businesses with a desire to be in
the tail of the firm size distribution. We also think that a missing
component of the academic and policy discussion is a formal cost-benefit
analysis of small business subsidies. To do this, more work is needed on
the potential frictions limiting small business growth and on the
externalities associated with small business growth.
Lastly, our results suggest that it is often inappropriate for
researchers to use the universe of small business (or self-employment)
data to test standard theories of entrepreneurship. Most small
businesses do not match the standard conceptual measures of
entrepreneurship, which focus on the desire to innovate or grow.
Researchers interested in testing such specific theories of
entrepreneurship may need to use more specialized data sets such as
those that track small businesses seeking venture capital funding.
APPENDIX
Table A1. Nonemployer Firms by Industry, 2007
Percent of all
nonemployer
Industry firms (a)
Professional, Scientific, and Technical Services 14
Other Services (Except Public Administration) 14
Construction 12
Real Estate and Rental and Leasing 11
Retail Trade 9
Administrative and Support and Waste Management 8
Health Care and Social Assistance 8
Arts, Entertainment, and Recreation 5
Transportation and Warehousing 5
Finance and Insurance 4
Education Services 2
Wholesale Trade 2
Manufacturing 2
Information 1
Accommodation and Food Services 1
Agriculture, Forestry, Fishing and Hunting 1
Mining, Quarrying, and Oil and Gas Extraction 0
Utilities 0
Source: Authors' calculations from Census data at www.census.gov/econ/
nonemployer/index.html.
(a.) Percents do not sum to 100 because of rounding.
Table A2. Reasons Reported by Nascent Entrepreneurs
for Starting a Business (a)
No. of respondents
giving indicated reason
First Second
Benson response response (b)
Nonpecunic?ry reasons
Be own boss; tired of working for 80 75
others
Flexibility; more free time; set own 26 22
hours
Stay home with children; work from home 33 12
Enjoy the work, have passion for it; 122 96
hobby
Job security/financial independence 34 14
Try new career; charge career; do 24 10
something new
Creative; do creative work; creative 9 5
outlet
Better life 3 0
Lifelong ambition 24 10
Challenge 3 3
Personal growth 2 8
To do more fulfilling work 2 3
Other lifestyle references 20 7
Other work desirability references 20 7
To generate income
Income; to make money 117 93
Extra income 50 20
Need supplemental income 8 6
Retired-need to supplement income 8 3
Income for educational expenses 1 3
Income for retirement 11 8
To leave business/money to children 5 4
Unlimited income potential; good money 22 19
Potential to make more money working 7 12
for self
Other income references 23 22
Had a good business idea or to create
new product
Take advantage of opportunity 23 17
High demand for productsibusiness: 75 30
satisfy need
Market opportunity; untapped market; 42 17
shift in market
New technology/product/service 110 3
Good product faith in product 13 5
Expansion of old/current business 23 2
Good business opportunity 1 2
Lots of experience at this type of 129 25
work; backgroundin field; knowledge
Have formal training/education in field 21 13
Have talent in field, area of expertise; 23 23
ability to do it
Other business opportunity references 33 21
Lack of employment options
Cannot find employment elsewhere; lost 18 8
job
Disabled/injured/sick and cannot work 18 12
elsewhere
Retired 14 8
Friend/family member had an idea and 25 9
started a business
Inheritance 5 1
Believe in value of work; think 4 1
business is important
Help others; help community 32 31
Aid in economy; economic development 9 1
Other reasons 51 20
Source: PSED data.
(a.) Respondents in the initial wave only of the PSED were asked,
"Why do [or did] you want to start this new business?" and could
give up to two responses in their own words. PSED staff then coded
the responses and classified them into the 44 specific categories
above and 6 broader categories. For purposes of this study the
responses were reclassified into the above 5 categories; the
reclassification is similar to the original PSED classification,
which can be found in the PSED codebook. The table reports the
number of respondents, out of all 1,214 PSED respondents, who gave
the indicated reason as either their first or their second
response.
(b.) Responses sum to fewer than the full sample because some
respondents did not provide a second response, and some gave two
responses within the same category.
ACKNOWLEDGMENTS We would like to thank Mark Aguiar, Fernando
Alvarez, Jaroslav Borovicka, Augustin Landier, Josh Lerner, E. J. Reedy,
Jim Poterba, Sarada, Andrei Shleifer, Mihkel Tombak, and the editors, as
well as seminar participants at Boston College, the 2011 Duke/Kauffman
Entrepreneurship Conference, the Federal Reserve Bank of Minneapolis,
Harvard Business School, the Institute for Fiscal Studies, the 2011
International Industrial Organization Conference, the London School of
Economics, the Massachusetts Institute of Technology, the 2010 NBER
Summer Institute Entrepreneurship Workshop, Pennsylvania State
University, Stanford University, and the University of Chicago for
comments. We gratefully acknowledge the financial support provided by
the George J. Stigler Center for the Study of Economy and the State as
well as the financial support provided by the Brookings Institution.
Additionally, Erik Hurst is grateful for the financial support provided
by the University of Chicago's Booth School of Business, and
Benjamin Pugsley expresses thanks for the financial support from the
Ewing Marion Kauffman Foundation. Certain data used in this paper are
derived from the Kauffman Firm Survey release 3.1 public-use data file.
Any opinions, findings, and conclusions or recommendations expressed in
this material are those of the authors and do not necessarily reflect
the views of the Ewing Marion Kauffman Foundation. The authors report no
conflicts of interest.
References
Acs, Zoltan J., and David B. Audretsch. 1990. Innovation and Small
Firms. MIT Press.
Acs, Zoltan J., Pontus Braunerhjelm, David B. Audretsch, and Bo
Carlsson. 2009. "The Knowledge Spillover Theory of
Entrepreneurship." Small Business Economics 32, no. 1: 15-30.
Ardagna, Silvia, and Annamaria Lusardi. 2008. "Explaining
International Differences in Entrepreneurship: The Role of Individual
Characteristics and Regulatory Constraints." Working Paper no.
14012. Cambridge, Mass.: National Bureau of Economic Research.
Asoni, Andrea. 2010. "What Drives Entrepreneurship?"
University of Chicago.
Audretsch, David B., Max C. Keilbach, and Erik E. Lehmann. 2006.
Entrepreneurship and Economic Growth. Oxford University Press.
Banerjee, Abhijit, and Esther Duflo. 2011. Poor Economics: A
Radical Rethinking of the Way to Fight Global Poverty. New York:
PublicAffairs.
Bhide, Amar V. 2000. The Origin and Evolution of New Businesses.
Oxford University Press.
Clementi, Gian Luca, and Hugo A. Hopenhayn. 2006. "A Theory of
Financing Constraints and Firm Dynamics." Quarterly Journal of
Economics 121, no. 1: 229-65.
Davis, Steven J., John C. Haltiwanger, and Scott Schuh. 1996. Job
Creation and Destruction. MIT Press.
Davis, Steven J., John C. Haltiwanger, Ron S. Jarmin, C. J. Krizan,
Javier Miranda, Alfred Nucci, and Kristin Sandusky. 2007.
"Measuring the Dynamics of Young and Small Businesses: Integrating
the Employer and Nonemployer Universes." Working Paper no. 13266.
Cambridge, Mass.: National Bureau of Economic Research.
De Rugy, Veronique. 2005. "Are Small Businesses the Engine of
Growth?" Working Paper no. 123. Washington: American Enterprise
Institute.
Evans, David, and Boyan Jovanovic. 1989. "An Estimated Model
of Entrepreneurial Choice under Liquidity Constraints." Journal of
Political Economy 97, no. 4: 808-27.
Evans, David S., and Linda S. Leighton. 1989. "Some Empirical
Aspects of Entrepreneurship." American Economic Review 79, no.
3:519-35.
Gabaix, Xavier. 2009. "Power Laws in Economics and
Finance." Annual Review of Economics 1: 255-93.
Hall, Robert E., and Susan E. Woodward. 2004. "Benchmarking
the Returns to Venture." Working Paper no. 10202. Cambridge, Mass.:
National Bureau of Economic Research.
--. 2010. "The Burden of the Nondiversifiable Risk of
Entrepreneurship." American Economic Review 100, no. 3:1163-94.
Haltiwanger, John C., Ron S. Jannin, and Javier Miranda. 2010.
"Who Creates Jobs? Small vs. Large vs. Young." Working Paper
no. 16300. Cambridge, Mass.: National Bureau of Economic Research.
Hamilton, Barton H. 2000. "Does Entrepreneurship Pay? An
Empirical Analysis of the Returns to Self Employment." Journal of
Political Economy 108, no. 3: 604-31.
Holmes, Thomas J., and John J. Stevens. 2010. "An Alternative
Theory of the Plant Size Distribution with an Application to
Trade." Working Paper no. 15957.
Cambridge, Mass.: National Bureau of Economic Research.
Hopenhayn, Hugo A. 1992. "Entry, Exit, and Firm Dynamics in
Long Run Equilibrium." Econometrica 60, no. 5:1127-50.
Hurst, Erik, Geng Li, and Benjamin Pugsley. 2010. "Are
Household Surveys Like Tax Forms: Evidence from Income Underreporting of
the Self Employed." Working Paper no. 16527. Cambridge, Mass.:
National Bureau of Economic Research.
Jones, Charles I., and John C. Williams. 1998. "Measuring the
Social Return to R&D." Quarterly Journal of Economics 113, no.
4:1119-35.
Jovanovic, Boyan. 1979. "Job Matching and the Theory of
Turnover." Journal of Political Economy 87, no. 5: 972-90.
--. 1982. "Selection and the Evolution of Industry."
Econometrica 50, no. 3: 649-70.
Kanbur, S. M. 1979. "Of Risk Taking and the Personal
Distribution of Income." Journal of Political Economy 87, no. 4:
769-97.
Kaplan, Stephen N., and Josh Lerner. 2010. "It Ain't
Broke: The Past, Present, and Future of Venture Capital." Journal
of Applied Corporate Finance 22, no. 2: 36-47.
Kihlstrom, Richard E., and Jean-Jacques Laffont. 1979. "A
General Equilibrium Entrepreneurial Theory of Firm Formation Based on
Risk Aversion." Journal of Political Economy 8, no. 4:719-48.
Knight, Frank. 1921. Risk, Uncertainty, and Profit. Boston:
Houghton Mifflin.
La Porta, Rafael, and Andrei Shleifer. 2008. "The Unofficial
Economy and Economic Development." BPEA (Fall): 275-363.
Lazear, Edward P. 2005. "Entrepreneurship." Journal of
Labor Economics 23, no. 4: 649-80.
Lucas, Robert E. 1978. "On the Size Distribution of Business
Firms." Bell Journal of Economics 9, no. 2: 508-23.
Moskowitz, Tobias J., and Annette Vissing-Jorgensen. 2002.
"The Returns to Entrepreneurial Investment: A Private Equity
Premium Puzzle?" American Economic Review 92, no. 4: 745-78.
Pakes, Ariel, and Richard Ericson. 1989. "Empirical
Implications of Alternative Models of Firm Dynamics." Working Paper
no. 2893. Cambridge, Mass.: National Bureau of Economic Research.
Pastor, Lubos, Lucian A. Taylor, and Pietro Veronesi. 2009.
"Entrepreneurial Learning, the IPO Decision, and the Post-IPO Drop
in Firm Profitability." Review of Financial Studies 22: 3005-46.
Pugsley, Benjamin. 201 la. "Wealth, Tastes, and
Entrepreneurship." University of Chicago.
--. 2011b. "Selection and the Role of Small Business Owners in
Firm Dynamics." University of Chicago.
Puri, Manju, and Rebecca Zarutskie. Forthcoming. "On the
Lifecycle Dynamics of Venture Capital and Non-Venture Capital Financed
Firms." Journal of Finance.
Schumpeter, Joseph. 1942. Capitalism, Socialism, and Democracy. New
York: Harper and Row.
Simon, Herbert A., and Charles P. Bonini. 1958. "The Size
Distribution of Business Firms." American Economic Review 48, no.
4:607-17.
Sutton, John. 1997. "Gibrat's Legacy." Journal of
Economic Literature 35, no. 1: 40-59.
Vereshchagina, Galina, and Hugo A. Hopenhayn. 2009. "Risk
Taking by Entrepreneurs." American Economic Review 99, no. 5:
1808-30.
ERIK HURST
University of Chicago
BENJAMIN WILD PUGSLEY
University of Chicago
(1.) Haltiwanger, Jarmin. and Miranda (2010) show that, when one
controls for firm age, there is no systematic relationship between firm
size and growth. They conclude that those small firms that tend to grow
fast (relative to large firms) are newly established firms. We discuss
in later sections how our results add to these findings. In particular,
we show that most surviving new firms also do not grow in any meaningful
way.
(2.) Two notable exceptions include Bhide (2000) and Ardagna and
Lusardi (2008). Bhide (2000) examines the attributes of the founders of
many successful firms and concludes that their actions and behaviors are
an important determinant of firm growth. Ardagna and Lusardi (2008) use
survey data from the Global Entrepreneurship Monitor to show that
individuals who report starting a business because they had a good
business opportunity differ demographically from other business owners.
(3.) Some papers in the literature take this approach. See, for
example, recent work by Kaplan and Lerner (2010), Puri and Zarutskie
(forthcoming), and Hall and Woodward (2010). As shown by Puri and
Zarutskie (forthcoming), firms that seek venture capital funding are
much more likely to grow than the universe of remaining firms.
(4.) For a complete description of the data, see U.S. Census
Bureau, "Statistics of U.S. Businesses,"
www.census.gov/econ/susb/.
(5.) The Business Register is updated continuously and incorporates
data from the Census Bureau's economic censuses and current
business surveys, quarterly and annual federal tax records, and other
departmental and federal statistics. The data include information from
all NAICS industries except crop and animal production; rail
transportation; the U.S. Postal Service; pension, welfare, and vacation
funds; trusts, estates, and agency accounts; private households; and
public administration.
(6.) A firm (termed here an "enterprise") may consist of
many establishments, which are distinct locations of business activity.
For example, Starbucks Corporation is a large firm that operates
thousands of small establishments. Given our focus on total firm
employment, we do not treat the individual Starbucks establishments as
small businesses.
(7.) The national SUSB data are available at the six-digit NAICS
level of aggregation. Without much loss of generality, we aggregate
these data to the four-digit level.
(8.) We also performed a different set of robustness results, based
on a measure of the importance of small businesses in industry j out of
all small businesses that adjusts for the importance of industry j out
of all firms regardless of size. The patterns in table 1 and figures 1
through 3 were robust to this adjustment.
(9.) Even though they are currently small, the nonemployer firms
are an important source of future paid-employee firms. Many eventual
employer firms start out as nonemployer firms. See Davis and others
(2007) for a more detailed discussion.
(10.) See U.S. Census Bureau, "Nonemployer Statistics,"
www.census.gov/econ/non employer/index.html.
(11.) The SSBF was formerly known as the National Survey of Small
Business Finances. It was a quinquennial survey that began in 1983 and
was last conducted in 2003.
(12.) We exclude firms that are unable to answer the employment
change question because they did not exist in the base year. Thus, the
firms responding to the I-year change question are at least 1 year old,
and the firms responding to the 3-year change question are at least 3
years old.
(13.) The Kauffman Foundation (www.kauffman.org) is an organization
whose goals are to study and understand entrepreneurship.
(14.) As before, the distinction between firms (referred to as
enterprises by the Census Bureau) and establishments is important. The
SUSB data report expansions and contractions by firm size, by measuring
employment changes at the establishment level. If Starbucks Corporation
opens 100 new stores in a year and closes 50, the gross job creation and
destruction from these establishment births and deaths (as well as from
continuing establishments) would be attributed to the 2,500+ firm size
category.
(15.) This is a reasonable approximation since all the small firms
have fewer than 20 employees, so there would be very little variation in
the employment share within an industry if this regression were
estimated with the underlying administrative micro data.
(16.) See Acs and Audretsch (1990) and the citations therein.
(17.) The initial wave of the PSED (PSED I) was a test run for the
bigger PSED II. We do not use the initial data in our analysis. All data
and documentation for the PSED can be found at the University of
Michigan's PSED website, at www.psed.isr.umich.edu/psed/data.
(18.) One should be wary of putting too much emphasis on
self-reports of innovative behavior by small businesses. However, most
behavioral stories of how business owners might respond to such
questions would likely lead one to believe that the innovation numbers
are upper bounds on actual behavior. This would occur if the respondents
were more likely to report that they were innovative even if no actual
innovation was taking place within the business.
(19.) Ardagna and Lusardi (2008) show that the lack of employment
options is a much more important motivation for starting a small
business in developing economies.
(20.) Estimating the saturated version of this regression with an
interaction term had almost no effect on the point estimates and p
values shown in table 10. We also estimated the same regression with
each category that could be named in either response represented. This
also did not change the results.
(21.) Respondents that mentioned neither motive would have
specified either income reasons (the vast majority), lack of other
options, or some other motive.
(22.) This would be consistent with a model in which nonpecuniary
benefits are a large part of the return to small business formation, as
shown in Pugsley (2011b). In that model, individuals will be willing to
stay in business even if they get a bad productivity draw, because for
them the pecuniary returns are only a small portion of the total returns
to business entry.
(23.) This idea is consistent with recent research by Holmes and
Stevens (2010) which attributes the variation in firm size within
narrowly defined manufacturing industries to differences between large
plants that produce standardized goods and small plants that make custom
or specialty goods. Similar differentiated-product stories can also
potentially explain within-industry size variation in other narrow
industries such as retail trade (big-box stores versus mom-and-pop
stores), health care (small physician practices versus hospitals), or
lawyers (small law offices versus big corporate law firms). Explaining
the variation in within-industry firm size is an interesting avenue for
future research.
(24.) Hamilton's (2000) analysis does not take into account
the possibility of underreporting of income by the self-employed. Hurst,
Li, and Pugsley (2010) show that such underreporting is important.
Although Hamilton's results are weaker when underreporting is
accounted for, it still appears that the median self-employed individual
takes a pecuniary earnings loss when becoming self-employed.
(25.) See De Rugy (2005) for a detailed discussion of the various
ways the U.S. government provides subsidies to U.S. small businesses.
(26.) See Hurst, Li, and Pugsley (2010) for a recent discussion of
this literature.
Comments and Discussion
COMMENT BY
JOHN HALTIWANGER This paper by Erik Hurst and Benjamin Pugsley is
an interesting and informative study of small businesses in the U.S.
economy. If one wanted a short summary answer to the question asked in
the title--"What do small businesses do?"--the paper suggests
that the authors might answer, "Not much." They base this
conclusion on an array of interesting qualitative and quantitative
evidence. They document evidence suggesting that many small businesses
are clustered in industries dominated by small businesses, which they
argue are associated with inherently small-scale business operations.
They show that the "typical" (that is, median) small business
does not exhibit much dynamism in terms of growth or innovation: the
median growth rate is low, and in response to qualitative surveys about
their ambitions, many small business owners indicate that they have
little intention of growing. Related qualitative evidence shows that
small business owners often suggest that their motive for being a sole
proprietor is to "be one's own boss" rather than a profit
motive per se.
The authors use these findings to argue that the standard models
that economists have developed to characterize and analyze the dynamics
of firms in general, and small businesses in particular, are missing
important elements. Moreover, they argue that policymakers'
excessive focus on promoting entrepreneurship and small business
creation and support is misguided. Simply put, targeting small
businesses that are neither growing nor innovating much nor planning on
doing so in the future is unlikely to contribute much to aggregate
productivity and job growth.
There is much that I agree with in the paper in terms of
interesting qualitative and quantitative facts. Moreover, in giving a
detailed characterization of small businesses, the authors acknowledge
that there is enormous heterogeneity and skewness in the growth rates of
businesses, including small businesses. They recognize that, because of
this, it is critical to distinguish between averages and medians. On
this I agree completely, but I have a different perspective on the
implications. For example, the authors state that "the high degree
of skewness, especially in these [small business-intensive] industries
... makes analyzing the averages very deceiving." A more accurate
statement would be that it is of critical importance to keep in mind the
differences between averages and medians. More generally, I think that
it is precisely by exploring and understanding the heterogeneity and
skewness of businesses that one can obtain much richer insights into the
nature of job creation and productivity growth. Put differently, the
authors seem to suggest that one should mostly think about what small
businesses do by focusing on the typical (median) business, whereas I
think that the evidence suggests that focusing only on the median
business misses much of the interesting and important dynamics that
contribute to job and productivity growth.
To give a more solid foundation to these points, I have generated
some descriptive statistics about the distributions of size and growth
of U.S. firms drawn from my collaborative work with Ron Jarmin and
Javier Miranda (see Haltiwanger, Jarmin, and Miranda 2010) as well as
earlier work with Steven Davis and Scott Schuh (Davis, Haltiwanger, and
Schuh 1996). Using the same comprehensive firm-level data used in
Haltiwanger and others (2010), (1) figure 1 presents information about
the skewness of the size distribution in the small business-intensive
industries identified by the authors.
Two measures of the size distribution are included in figure 1. The
first is the simple average firm size as measured by the number of
employees. Consistent with the results in the present paper, I find that
the average firm size in these designated small business-intensive
industries is small. However, examining this statistic by itself is
misleading. It turns out that even in these industries, the average
worker is employed in a relatively large firm. The second statistic
reported in figure 1 is the co-worker mean (a measure developed by Davis
and others 1996). This measure is the employment-weighted average firm
size and is interpretable as the average firm size for the average
worker. The figure shows that the co-worker mean is above 1,000 in more
than half of the top 20 small business-intensive industries. For
example, the average accountant works for a firm that has more than
15,000 workers. This evidence alone raises questions about the
authors' argument that these industries are such that the scale of
activity is inherently small. Big businesses exist in each of these
industries and indeed account for much of the activity in these
industries.
A related point is that a firm's industry is not a
particularly good predictor of either its size or its growth. The
following table reports the [R.sup.2]s from simple regressions relating
indicators of size and growth to detailed (six-digit) industry effects:
[R.sup.2]
Probability that the firm has fewer than 20 employees 0.12
Net firm growth rate- (all firms) 0.06
Net firm growth rate (small firms) 0.06
Probability that the firm is a high-growth firm 0.04
(net firm growth rate > 20 percent)
Probability that the firm is a high-growth firm 0.03
(net firm growth rate > 20 percent and net change
> 10 employees)
The detailed industry of a firm accounts for only 12 percent of the
variation in the probability that the firm has fewer than 20 employees.
Detailed industry accounts for even less variation in the dispersion in
firm growth rates and in the probability that the firm is a high-growth
firm. In this respect, I think the authors overemphasize the role of
industry in accounting for the nature of the size distribution of
activity as well as the distribution of growth. The thrust of the paper
is that firms in these small business-intensive industries have little
growth potential, and that it should not be surprising that little
growth occurs in these firms because firms in these industries are
inherently small. Figure 1 and the above table show that this is a
misleading characterization of these industries.
In my work with Jarmin and Miranda cited above, we have shown that
the job creating prowess of small businesses is better attributed to the
contribution of business start-ups and young businesses. Start-ups and
young businesses are also usually small, but the evidence shows that
mature small businesses have negative average net growth rates. We also
show that young businesses exhibit an "up or out" dynamic:
they have a high probability of exit, but those that survive exhibit
rapid growth on average.
[FIGURE 2 OMITTED]
How can one reconcile these findings with the present authors'
evidence that small (and in some related findings in the paper, young)
firms exhibit little growth? This is one of many cases where the median
and the mean patterns differ substantially. Figure 2 shows the
employment-weighted mean and median net firm level growth rates (similar
patterns hold for the unweighted mean and median growth rates). It is
apparent that young surviving businesses have high average net growth
rates but not high median growth rates. Underlying this pattern is a
very skewed distribution of growth rates. Figure 3 shows the
employment-weighted 90th and 10th percentiles of the firm-level growth
rate distribution. The 90th percentile is very high and monotonically
declining in age, whereas the 10th percentile is negative but
monotonically increasing in age. High-growth young firms grow very fast,
but there is also enormous heterogeneity among young firms: the 90-10
differential is very large. The skewness is also apparent from the fact
that the absolute value at the 90th percentile for the young firms is
much larger than that at the 10th percentile. It is this skewness that
drives the difference between the means and medians in figure 2.
The high-growth firms in figure 3 contribute disproportionately to
job creation. Figure 4 shows the patterns of job creation for both the
average industry and the average small business-intensive industry (from
among the top 40 small business-intensive industries listed in table 1
of the paper). Several observations are worth noting. First, the average
(gross) job creation rate for the small business-intensive industries is
higher than the average for all industries. (3) Second, start-ups plus
high-growth firms (defined here as firms with growth rates exceeding 20
percent in a year) account for over 80 percent of job creation in the
small business-intensive industries and just under 80 percent in all
industries. Start-ups are important contributors to job creation here,
but it is also apparent that other high-growth firms contribute
substantially (over 40 percent of the total) to job creation.
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
In sum, start-ups and high-growth firms are the biggest
contributors to job creation. Start-ups tend to be small (Haltiwanger
and others 2010), and high-growth firms tend to be young and small
(Haltiwanger and others 2010 and figure 3). It is true that the median
small firm and even the median young firm is not a high-growth firm. But
does this imply that one should ignore the contribution of start-ups and
high-growth young firms? Obviously this would be a mistake, since they
account for such a large fraction of job creation.
The authors suggest that the standard models of firm heterogeneity
yielding insights about the size distribution of the level and growth of
employment are "at best incomplete." Although this is
undoubtedly true, I do not think the standard models do as badly as
their discussion suggests. The canonical model of the size distribution
of activity within industries describes firms that face idiosyncratic
shocks to their profitability (from productivity, demand, and cost
factors). The evidence suggests that these shocks follow a Pareto
distribution and that they exhibit considerable persistence. To justify
an equilibrium distribution of size, it is standard to assume that the
existence of curvature in the profit function derives either from
economies of scope and control (see, for example, Lucas 1978) or from
product differentiation (see, for example, Melitz 2003). In these models
the most productive and profitable firms will be the largest, but the
curvature of the profit function implies that these firms will not take
over the industry. In addition, some fixed operating costs are typically
assumed, so that firms with very low profitability and productivity will
choose to exit. Dynamics are introduced into this class of models by
recognizing, first, that firms are continually subject to new
idiosyncratic shocks, and second, that there are a host of frictions
that slow adjustment (such as capital-labor adjustment frictions, but
also information and learning frictions).
This class of models is fully consistent with the paper's
evidence as well as with the evidence discussed above. Given the Pareto
distribution of profit shocks, one should expect a very skewed
distribution of both firm size and growth. The median firm will be small
and not growing. But some firms will be large, and some (the high-growth
firms) will become large. Even in the small business-intensive sectors
that the paper focuses on, both large and high-growth firms exist, which
is again consistent with this class of models. Also, the
differentiated-product versions of these models provide a potentially
rich way of capturing the many small firms that fit into specific
niches, even within industries.
One piece of evidence in the paper that might seem at odds with
this interpretation is the qualitative survey evidence suggesting that
it is not the profit motive but rather nonpecuniary factors (such as
"owning one's own business") that drive most small
businesses. I agree that nonpecuniary factors are potentially of
interest, but I also have some skepticism about the evidence presented.
For one thing, the evidence comes from the Panel Survey of
Entrepreneurial Dynamics (PSED), which is a small survey of
"nascent" entrepreneurs (the sample consists of 1,214
potential entrepreneurs). I think the objective of the PSED is of
first-order importance, but it is inherently a challenge to find a
sample of potential entrepreneurs and to obtain high response rates from
them (Curtin and Reynolds 2008). I have related questions about the
representativeness of the Kauffman Firm Survey (KFS), the paper's
other source of qualitative (and quantitative) evidence on small
businesses. I think the KFS has a better sample frame than the PSED, but
it still faces limitations. Among other things, it has been shown that
the KFS has much higher survival rates than nationally representative
databases (Robb 2009). The underlying sample frame for the KFS is the
Dun and Bradstreet data set, which is well known to miss most start-ups
initially and to underrepresent young firms (Acs, Parsons, and Tracy
2008).
An alternative way to state this skepticism about the data sets
used is that all of the evidence from nationally representative
firm-level studies (for at least employer-based businesses) shows,
first, that exit rates are very high for young and small businesses, and
second, that the probability of exit is much higher for businesses with
low profitability or productivity (Foster, Haltiwanger, and Krizan 2001,
2006). It may be that business owners, in response to a survey, will
express the nonpecuniary benefits of being a business owner, but at the
end of the day, unproductive and unprofitable businesses exit. I do
think the importance of nonpecuniary benefits is likely greater for
nonemployer businesses (the self-employed who do not hire any
employees). There are many more nonemployer businesses in the United
States (between 15 and 20 million) than employer businesses (about 6
million). But the present paper focuses on employer businesses, which is
a considerably more important group to consider when it comes to job and
productivity growth.
To emphasize my main comment on this paper: I think it is critical
to understand the role of start-ups and high-growth young firms (and
high-growth firms more generally), not only to deepen our knowledge of
the job creation process, but also so that the most appropriate policies
will be implemented. Many distortions can impede the dynamics discussed
above. Of particular relevance in recent years are distortions in
financial markets that may be impeding business start-ups (which have
fallen dramatically since 2007; see Haltiwanger forthcoming) and
high-growth young firms. Consistent with the paper' s findings, it
may be that financial market distortions have relatively little impact
on the median young and small firm, but they may still be very relevant
for high-growth young firms, and in particular those likely to
contribute significantly to job creation.
The challenge, of course, is that any policies that apply broadly
to small or even young businesses will fail to take into account the
enormous heterogeneity and skewness in the distribution of firm-level
growth rates, especially for young (and small) businesses. Most young
businesses will either fail or not grow--that is the basic message of
this paper. But again, focusing only on the median young and small
business misses much of what contributes to job creation in the United
States. It is the impact of financial market and other distortions on
the margins of particular relevance for job creation (that is,
distortions affecting start-ups and high-growth firms) that we need to
understand, recognizing that the typical young and small firm may be
much less affected by such distortions.
REFERENCES FOR THE HALTIWANGER COMMENT
Acs, Zoltan, William Parsons, and Spencer Tracy. 2008.
"High-Impact Firms: Gazelles Revisited." SBA Report
SBAHQ-06-Q-0014. Washington: Office of Advocacy, Small Business
Administration.
Curtin, Richard, and Paul Reynolds. 2008. "Business Creation
in the United States: Panel Study of Entrepreneurial Dynamics Ii Initial
Assessment." Foundations and Trends in Entrepreneurship 155-307.
Davis, Steven, John Haltiwanger, and Scott Schuh. 1996. Job
Creation and Destruction. MIT Press.
Davis, Steven, John Haltiwanger, C. J. Krizan, Ron Jarmin, Javier
Miranda, Al Nucci, and Kristin Sandusky. 2009. "Measuring the
Dynamics of Young and Small Businesses: Integrating the Employer and
Non-Employer Universes." In Producer Dynamics: New Evidence from
Micro Data, edited by Timothy Dunne, J. Bradford Jensen, and Mark J.
Roberts. NBER Book Series Studies in Income and Wealth. University of
Chicago Press.
Foster, Lucia, John Haltiwanger, and C. J. Krizan. 2001.
"Aggregate Productivity Growth: Lessons from Microeconomic
Evidence." In New Developments in Productivity Analysis, edited by
Charles R. Hulten, Edwin R. Dean, and Michael J. Harper. NBER Book
Series Studies in Income and Wealth. University of Chicago Press.
--. 2006. "Market Selection, Reallocation and Restructuring in
the U.S. Retail Trade Sector in the 1990s." Review of Economics and
Statistics 88, no. 4: 748-58.
Haltiwanger, John. Forthcoming. "Job Creation and Firm
Dynamics in the U.S." In Innovation Policy and the Economy, vol.
12, edited by Josh Lerner and Scott Stem. Cambridge, Mass.: National
Bureau of Economic Research.
Haltiwanger, John, Ron Jarmin, and Javier Miranda. 2010. "Who
Creates Jobs? Small vs. Large vs. Young." Working Paper no. 16300.
Cambridge, Mass.: National Bureau of Economic Research.
Lucas, Robert E. 1978. "On the Size Distribution of Business
Firms." Bell Journal of Economics 9, no. 2: 508-23.
Melitz, Marc. 2003. "The Impact of Trade on Intra-Industry
Reallocations and Aggregate Industry Productivity." Econometrica
71, no. 6: 1695-1726.
Robb, Alicia. 2009. "Kauffman Firm Survey Overview Follow-Up
Report." Kansas City, Mo.: Kauffman Foundation.
(1.) Details about the data can be found in that paper. For
purposes of the tabulations reported here, I focus on 2003-05, which are
the overlap years between those that Hurst and Pugsley focus on and
those in the Haltiwanger and others (2010) sample, 1990-2005. Note that
the underlying firm level data I use for these tabulations contain about
6 million observations per year.
(2.) The net firm growth rate is the measure developed by Davis,
Haltiwanger, and Schuh (1996) and now used as a standard summary measure
of firm growth, for example in the Business Dynamic Statistics. The same
measure is also used in the paper.
(3.) This finding is related to another somewhat misleading finding
in the paper. The authors show that their x measure is inversely
correlated with job creation by small businesses in their top 40
industries as measured by x (their table 1). One could interpret this as
saying that among these industries, those with a larger share of small
businesses have lower rates of job creation by small businesses. I am
not convinced that this is an interesting or robust characterization of
the data. In unreported tabulations using the firm-level data underlying
figures 1 through 4, I show, first, that I can verify their finding;
second, that even among their top 40 x industries, those where small
businesses account for a higher share of employment have higher job
creation rates; and that across all industries, those where small
businesses account for a larger share of employment have higher job
creation rates. The last two findings are consistent with the
well-established finding (see Davis and others 1996 and Haltiwanger and
others 2010) that small businesses have higher job creation rates and
higher job destruction rates than large businesses within the same
industry. I think the present authors' finding raises some
questions about whether the x measure is the most appropriate and
meaningful way to characterize small business-intensive industries. It
is correlated with the share of employment in the industry accounted for
by small businesses, but that correlation is far from I. Part of the
explanation is that counts of the number of firms in an industry are
inherently noisy, for both measurement and conceptual reasons. For
example, many very small employer businesses (those with 1 or 2
employees) are often crossing the line between being an employer
business and being a nonemployer business (see Davis and others 2009). I
think a more robust measure of the small business intensity of an
industry is the share of employment accounted for by small businesses in
the industry.
Figure 1. Average Firm Size and Co-Worker Mean for Top 20 Small
Business-Intensive Industries, 2003-05
Average Co-worker
firm size mean (a)
Full-Service Restaurants (7221) 28 12,538
Offices of Physicians (6211) 11 1,347
Limited Service Restaurants (7222) 29 12,788
Religions Organizations (8131) 10 267
Build. Equip. Contractors (2382) 5 98
Dentists (6212) 7 53
Auto Repair (8111) 6 185
Legal Series (5411) 7 286
Res. Bldg. Construction (2361) 4 597
Service to Build. (5617) 12 5,725
Build. Finishing Contractors (2383) 4 103
Build. Exterior Contractors (2381) 6 104
Insurance Agents (5242) 7 6,217
Other Health Practitioners (6213) 6 648
Arch./Engineering Services (5413) 14 2,390
Accounting Services (5412) 13 15,560
Personal Care Services (8121) 7 4,082
Consulting Services (5416) 8 2,277
Gas Stations (4471) 17 4,381
Child Day Car Services (6244) 16 2,324
Source: Author's tabulations using Business Dynamic Statistics data.
a. Employment-weighted average firm size. The set of industries
is that in the conference draft of the Hurst and Pugsley paper and
differs slightly from that in the version in this volume.
Note: Table made from bar graph.
COMMENT BY
ADAM LOONEY Innovation and entrepreneurial activity are crucial for
raising the productivity of the workforce, expanding job opportunities,
and developing and introducing new and better consumer products, and
therefore for increasing living standards over time. The potential
spillover benefits of such activities for aggregate living standards
make it natural for policymakers to search for opportunities to promote
innovation and entrepreneurship. Relying in part on the premise that
small businesses are responsible for a large share of entrepreneurial
activity, innovation, and job creation, policymakers have directed
considerable attention to small businesses and have enacted a broad
array of tax breaks, direct spending programs, and regulatory relief
specifically to promote such businesses.
In this paper Erik Hurst and Benjamin Pugsley reassess the role of
small businesses in the economy and find, in contrast to the
conventional wisdom, that most are neither particularly innovative nor
entrepreneurial. Their data suggest that few small businesses engage in
measurable activities that correspond to research and development, such
as obtaining patents and copyrights. Most of the small business owners
in their survey data do not report bringing a new idea to market and do
not report a desire to expand beyond a few employees. Perhaps most
surprisingly, a majority of nascent entrepreneurs surveyed for the Panel
Study of Entrepreneurial Dynamics do not cite income generation as one
of their top two reasons for starting a business, and over 50 percent
instead mention nonpecuniary benefits, such as flexible hours or being
one's own boss.
Hurst and Pugsley also find that most small businesses stay small;
on net, small businesses are not disproportionate job creators. Indeed,
their evidence suggests that most small businesses are small not because
they are new and growing, but because they operate in industries or
occupations where small is the appropriate and efficient scale of
business and production. Examples include restaurants, doctors' and
law offices, and electricians. Such evidence corroborates previous
findings, for example by John Haltiwanger, Ron Jarmin, and Javier
Miranda (2010), showing that, conditional on firm age, there is little
relationship between firm size and firm growth.
Together these results suggest that most small businesses are not
involved in activities that generate innovation spillovers or other
positive externalities, nor are they disproportionate generators of
economic activity or jobs, and thus they call into question whether
small businesses, broadly defined, are the appropriate target of
policies to promote innovation. But are there other reasons why it may
be efficient or desirable to tailor policies specifically to small
businesses?
One alternative motivation for such policies is to further equity,
fairness, or redistributive goals, out of a desire to raise the
well-being of either small business owners or their employees. However,
evidence from tax data suggests that small business owners, on average,
are likely to be financially better off than the typical employee at
either a small or a large firm. Individuals who earn money from small
businesses tend to have other sources of income as well, and the average
total annual income for individuals with a significant portion of income
coming from small businesses is over $76,000. (1) As a result, small
business subsidies are likely to be regressive. Moreover, Hurst and
Pugsley point out that small business owners tend to enjoy a larger
share of nonpecuniary benefits from their work than other workers do.
Because these benefits are not taxed, the existing tax structure already
implicitly subsidizes small business, since a smaller fraction of total
compensation is taxed. These considerations indicate that equity or
distributional considerations focused on owners are unlikely to justify
the transfer of public resources to small businesses.
A similar hypothesis is that small businesses create employment
opportunities that provide unusually favorable pay, benefits, or
stability and thus reduce the fiscal burden on all taxpayers while
raising the living standards of their workers. However, statistics do
not appear to support this idea. For example, employees of small
businesses are less likely to receive health benefits. In 2011, 99
percent of firms with 200 or more employees, but only 59 percent of
firms with between 3 and 199 employees, offered health benefits to their
workers, according to the Kaiser Family Foundation and the Health
Research & Educational Trust (2011). Similarly, employees of small
firms are also less likely to receive retirement benefits: Irena Dushi,
Howard Iams, and Jules Lichtenstein (2011) estimate that in 2006, 84
percent of employees of firms with 100 or more workers were offered some
kind of retirement plan, compared with only 50 percent of employees of
smaller firms. Small firms also have higher job turnover rates than
larger firms. Experimental data from the Bureau of Labor
Statistics' Job Openings and Labor Turnover Survey show that
between May 2010 and April 2011, employees of businesses with between 10
and 49 employees separated from their jobs at an average monthly rate of
3.5 percent, compared with only 2.1 percent at businesses with between
1,000 and 4,999 employees. (2) Thus, jobs at small businesses appear to
be less stable or enduring jobs, on average.
A final important consideration is the cost of complying with
complex tax and regulatory requirements. Learning the relevant features
of the tax and regulatory system, undertaking the necessary
record-keeping, and filing tax returns and complying with the tax system
all include some fixed-cost component and are thus disproportionately
burdensome to smaller businesses. This is undoubtedly a concern to small
businesses and is an important argument for differential and favorable
treatment of such businesses.
In sum, the data on small business activity that Hurst and Pugsley
analyze suggest that the scope for advancing innovation and
entrepreneurial activity or for supporting job growth or the creation of
"good" jobs by focusing on small businesses is limited.
Although an analysis of equity considerations is beyond the scope of
their analysis, a brief review of the evidence suggests that support for
small business neither disproportionately improves the welfare of
disadvantaged business owners nor promotes unusually high quality
employment at those businesses. Moreover, Hurst and Pugsley identify
several reasons why policies encouraging or subsidizing small businesses
may introduce additional costly distortions into the economy beyond
their direct budgetary impacts. Small business subsidies may, at the
margin, encourage businesses to operate at a smaller than optimal scale.
They may also subsidize economic activity in industries in which small
businesses are naturally concentrated--such as legal services, real
estate, and carpentry--at the expense of industries in which businesses
are naturally larger.
Under current policies based, in part, on the conventional wisdom,
small businesses benefit from a number of legal, administrative, and tax
provisions that provide them both implicit and explicit support. The
Small Business Administration (SBA), whose main mission is to help
"Americans start, build, and grow small businesses," has an
annual budget of almost $1 billion. With about a third of this budget,
it guarantees over $20 billion worth of bank loans taken out by small
businesses, to make it easier for these firms to borrow (Small Business
Administration 2010, 2011 a). The agency also works to enforce a
government-wide goal, established by the Small Business Act, of awarding
at least 23 percent of the value of prime procurement contracts to small
businesses (Small Business Administration 2011 b).
Small businesses receive exemptions from many workplace
regulations. For example, certain provisions of the Family and Medical
Leave Act of 1993, which requires employers to provide job-protected
unpaid leave for qualified family and medical reasons, do not apply to
businesses with fewer than 50 employees. Some provisions of the Age
Discrimination in Employment Act of 1967 do not apply to businesses with
fewer than 20 employees. Neither the provisions of the Americans with
Disabilities Act of 1990 prohibiting employment discrimination based on
disability nor the employment provisions of the landmark Civil Rights
Act of 1964 apply to firms with fewer than 15 employees. The
Occupational Safety and Health Administration (2011) prescribes lower
penalties for violations of health and safety regulations that occur
within small firms and subjects such firms to reduced filing
requirements.
In addition to regulatory relief and subsidies and loans from the
SBA, small business owners receive favorable tax treatment relative to
other businesses. Gary Guenther (2009) identifies six major tax
preferences for which the congressional Joint Committee on Taxation
(JCT) estimated revenue costs in fiscal 2009: (3)
--Small corporations face lower tax rates than larger businesses.
On most income under $10 million, corporations are subject to marginal
tax rates lower than the usual corporate rate of 35 percent. (4) The
lowest marginal rates apply to firms with the least income, so that a
firm making $100,000 pays an average rate of roughly 22 percent. The JCT
estimates that the revenue loss associated with this provision was $3.3
billion in fiscal 2009.
--In tax years 2008 and 2009, businesses were permitted to treat
many purchases (up to $250,000) of depreciable assets as expenses rather
than as capital expenditures. This allowed the purchases to be deducted
from taxable income immediately rather than over several years, enabling
businesses to defer taxes into the future. The JCT estimates that this
provision cost the Treasury about $6.0 billion in 2009. (A temporary
provision enacted for 2011 allows all businesses to expense capital
purchases.)
--Business owners may deduct up to $5,000 of start-up and
organizational costs in the year they start their business. (The maximum
deduction is reduced by the amount by which start-up costs exceed
$50,000.) Eligible expenses that cannot be deducted may be amortized
over 15 years. The cost of this deduction and amortization was about
$0.8 billion in 2009, the JCT estimates.
--For tax purposes, many small firms are permitted to use
cash-based accounting instead of the more complex and rigorous
accrual-based accounting, at a revenue cost of about $0.9 billion in
2009, according to the JCT.
--Many taxpayers may exclude 50 percent of capital gains from the
sale of qualified small business stock, subject to certain limits. The
JCT estimates the cost of this provision at $0.5 billion in 2009.
--Taxpayers may deduct losses from the sale of qualified small
business stock as ordinary losses rather than capital losses, at an
estimated cost of $50 million in 2009.
The estimated revenue cost of these provisions alone was thus about
$11.5 billion in 2009, and this total does not reflect a number of other
small business tax preferences for which the JCT did not estimate costs,
including a tax credit for expenses necessary to comply with the
Americans with Disabilities Act, and special treatment of losses on
stock in small business investment companies. This number was likely
much higher in 2010 and 2011, as the Small Business Jobs Act of 2010
temporarily expanded many of these tax preferences (Internal Revenue
Service 2011). For example, it doubled the limit on expensable capital
expenditure for 2010 and 2011, and it raised the capital gains exclusion
to 100 percent on small business stock acquired within a certain time
frame. In addition, the Affordable Care Act, also enacted in 2010,
created a new permanent tax credit for small businesses that provide
health insurance to their employees, which the Office of Management and
Budget (2011a) estimates will reduce the tax liabilities of small
businesses by $21 billion between 2012 and 2016.
As significant as these explicit tax incentives are, they are
dwarfed by a variety of other tax breaks that, while not applied
according to firm size, in practice end up benefiting small firms
disproportionately. For example, firms legally organized as S
corporations, sole proprietorships, partnerships, and limited liability
corporations (LLCs) generally receive more favorable tax treatment than
businesses organized as C corporations (which disproportionately are
large, publicly traded companies), and in certain cases their owners
face lower tax rates than shareholders in C corporations or ordinary
employees.
For instance, owners of S corporations and LLCs can reduce Social
Security and Medicare payroll taxes on much of their income by
classifying that income as business distributions rather than wages--a
choice that is not available to regular employees, the self-employed, or
owners of partnerships. Although the tax system includes anti-abuse
provisions to ensure that these businesses owners pay themselves
"reasonable compensation" in wages, in practice they often do
not. As a result they often pay lower effective tax rates than do
employees, the self-employed, or owners in partnerships: a report by the
JCT (2005) estimated that applying the same rules used for partnerships
to owners of S corporations and LLCs would raise roughly $60 billion
over 10 years, and about $7 billion per year from 2011 to 2014.
Similarly, self-employed workers face lower payroll tax obligations
than non-self-employed workers with comparable incomes because of the
formula for calculating payroll taxes under the Self-Employment
Contributions Act. One result is that self-employed people who earn more
than Social Security's taxable maximum have lower Medicare tax
obligations than comparable workers who are not self-employed. The JCT
(2005) estimates the cost of these rules at about $500 million per year.
Other favorable rules fall at the intersection of tax and labor
laws. For example, the rules that determine who is counted as an
employee rather than as an independent contractor allow many employers
and sole proprietors (and the larger businesses they do business with)
to avoid many labor regulations, employment taxes, and tax withholding
requirements by claiming that they operate in a business-contractor
relationship rather than an employer-employee relationship. The Office
of Management and Budget (2011b) has estimated that clarifying the rules
for classifying workers as independent contractors would generate more
than $7 billion over 10 years.
Although legal tax preferences for small businesses provide them
with billions of dollars in support each year, small businesses derive a
much larger implicit tax subsidy from a tax compliance system that
allows for significant underreporting of small business income on tax
returns. Small businesses underreport income on tax returns at a much
higher rate than larger businesses or ordinary employees, largely
because they are subject to less third-party reporting on their revenue
and expenses. The Internal Revenue Service conducted a detailed analysis
of the "tax gap" in 2001--taxes that should have been paid but
went unpaid because individuals or entities did not file required
returns, did not report their full tax liability on a return, or did not
pay the full amount required on a return. According to updated estimates
from that analysis, small businesses and self-employed individuals were
responsible for $257 billion, or about 75 percent, of the $345 billion
tax gap in 2001 (Internal Revenue Service and U.S. Department of the
Treasury 2007).
Despite the size of this small business tax gap, policymakers have
frequently shied away from efforts to close it. For example, a provision
of the Affordable Care Act requiring business and real estate owners to
report payments to other businesses in excess of $600 per year was
repealed soon after the law was passed, because it was seen as overly
burdensome to small businesses (White House Office of the Press
Secretary 2011). The provision had been projected to raise $17 billion
between 2012 and 2019 by deterring tax cheats (JCT 2010). A similar
provision requiring withholding of taxes for federal contractors was
also recently repealed, reducing tax revenue by an estimated $11 billion
over 10 years.
Of course, many of the tax, regulatory, and compliance measures
that exempt or provide favorable treatment to small businesses arise
from a desire to reduce the burden of understanding and complying with
many complex and frequently changing rules and regulations. Tax
provisions such as cash accounting and immediate expensing of
investments simplify record-keeping and filing and reduce other fixed
costs of dealing with the government. An important challenge is to
identify the appropriate balance across several objectives: mitigating
the burdens of compliance, maintaining tax revenue, promoting an equal
playing field for all businesses and their employees, and upholding and
enforcing important workplace rules.
To conclude, Hurst and Pugsley have established that, for
businesses, "small" is not synonymous with
"innovative" or "growing," and thus that policies to
promote innovation spillovers or otherwise encourage economic activity
based on business size alone may not be particularly efficient or
effective. Of course, a few new firms that start off small do attempt
the kind of innovation that creates large spillover benefits--and some
succeed. If policymakers could identify this subset of firms that
innovate the most, it might be far more efficient to target just those
firms. Although that is obviously a challenging task, one approach might
be to focus on specific industries or sectors that invest heavily in
research and development. Alternatively, government could continue or
expand support to the basic inputs to innovation, such as basic R&D
and financing of education and research at universities, which could be
useful to innovative businesses regardless of size. Such changes could
improve the efficiency of support for innovation and, ultimately, job
creation and would not necessarily preclude policies that treat small
businesses differently to address the greater compliance and
administrative burdens they may face.
REFERENCES FOR THE LOONEY COMMENT
Congressional Budget Office. 2005. "Budget Options."
Washington. www.cbo. gov /ftpdocs/60xx/doc6075/02-15-BudgetOptions.pdf.
Dushi, Irena, Howard M. Iams, and Jules Lichtenstein. 2011.
"Assessment of Retirement Plan Coverage by Firm Size, Using W-2 Tax
Records." Social Security Bulletin 71, no. 2: 53-65.
Guenther, Gary. 2009. "Small Business Tax Benefits: Overview
and Economic Rationale." CRS Report for Congress no. RL-32254.
Washington: Congressional Research Service.
Haltiwanger, John C., Ron S. Jarmin, and Javier Miranda. 2010.
"Who Creates Jobs? Small vs. Large vs. Young." Working Paper
no. 16300. Cambridge, Mass.: National Bureau of Economic Research.
Internal Revenue Service. 2011. "Small Business Act of 2010
Tax Provisions." Washington.
www.irs.gov/businesses/small/article/0,,id=230307,00.html.
Internal Revenue Service and U.S. Department of the Treasury. 2007.
"Reducing the Federal Tax Gap: A Report on Improving Voluntary
Compliance." Washington,
www.irs.gov/pub/irs-news/tax_gap_report_final_080207_linked.pdf.
Joint Committee on Taxation. 2005. "Options to Improve Tax
Compliance and Reform Tax Expenditures." Washington.
www.jct.gov/publications.html?func=startdown &id=1524.
--. 2010. "Estimated Revenue Effects of the Amendment in the
Nature of a Substitute to H.R. 4872, the 'Reconciliation Act of
2010,' as Amended, in Combination with the Revenue Effects of H.R.
3590, the 'Patient Protection and Affordable Care Act
("PPACA"),' as Passed by the Senate, and Scheduled for
Consideration by the House Committee on Rules on March 20, 2010."
Washington, www.jct.gov/publications.html?func=download&id=3672&chk=3672 &no_html=l.
Kaiser Family Foundation and Health Research & Educational
Trust. 2011. "Employer Health Benefits 2011 Annual Survey."
Menlo Park, Calif., and Chicago, Ill. ehbs.kff.org/pdf/2011/8225.pdf.
Knittel, Matthew, Susan Nelson, Jason DeBacker, John Kitchen, James
Pearce, and Richard Prisinzano. 2011. "Methodology to Identify
Small Businesses and Their Owners." Department of the Treasury
Office of Tax Analysis Technical Paper no. 4. Washington.
Occupational Safety and Health Administration. 2011.
"Occupational Safety and Health Administration's Field
Operations Manual." Washington. www.osha.
gov/OshDoc/Directive_pdf/CPL_02-00-150.pdf.
Office of Management and Budget. 2011a. "Analytical
Perspectives, Budget of the United States Government, Fiscal Year
2012." Washington. www.gpo.gov/
fdsys/pkg/BUDGET-2012-PER/pdf/BUDGET-2012-PER.pdf.
--. 2011b. "Budget of the United States Government, Fiscal
Year 2011." Washington. www.gpoaccess.gov/usbudget/fy11/index.html.
President's Economic Recovery Advisory Board. 2010. "The
Report on Tax Reform Options: Simplification, Compliance, and Corporate
Taxation." Washington.
www.whitehouse.gov/sites/default/files/microsites/PERAB_Tax_Reform_
Report.pdf.
Small Business Administration. 2010. "Recovery Loan Incentives
Spurred Continued Rebound in SBA Lending in FY2010." Washington.
indus.sba.gov/idc/ groups/public/documents/sba_homepage/news_release_
10-54.pdf.
--. 2011a. "Fiscal Year 2012 Budget Summary." Washington.
www.sba. gov/content/fiscal-year-2012-budget-summary.
--. 2011b. "Goaling Guidelines: Statutory Requirements."
Washington. archive.sba.gov/aboutsba/sbaprograms/goals/gg/index.html.
White House Office of the Press Secretary. 2011. "Statement by
President Obama on H.R. 4." Washington.
www.whitehouse.gov/the-press-office/2011/04/14/
statement-president-obama-hr-4.
(1.) This figure is based on data from table 16 of Knittel and
others (2011): under their "narrow" definition of small
business owners, there were 9.4 million small business owners who earned
a collective $716 billion in 2007.
(2.) Because they are experimental, data on turnover by firm size
from the Job Openings and Labor Turnover Survey are not yet published,
but they are available by request from the Bureau of Labor Statistics.
(3.) Guenther also highlights a small business tax preference
concerning net operating losses that cost around $4.7 billion in fiscal
2009, but it is not included in the list because it was a temporary
provision of the American Reinvestment and Recovery Act.
(4.) Firms in certain sectors, including health care, law, and
engineering, are ineligible for these lower rates and always pay a fixed
rate of 35 percent.
GENERAL DISCUSSION Erik Hurst began the discussion by summarizing
some comments he had received from Antoinette Schoar. A professor of
entrepreneurial finance at the MIT Sloan School of Management, Schoar
had been scheduled as a formal discussant but at the last minute was
unable to attend the Panel meetings.
Schoar reported that the misconception that all small businesses
want to become big businesses was also widespread in the development
economics literature. This idea has been used to support many
questionable investments, by developing countries, aid organizations,
and the like, in entrepreneurship training and other small business
support programs. In her own work she had stressed the importance of
better targeting these interventions to those small firms that are
actually likely to grow. Several papers-including one by Suresh de Mel
and coauthors--have found evidence that these firms do look different
from other small businesses: for example, their owners tend to be better
educated and to have greater tolerance of risk. They are also less
likely to be female, because women in these societies are often expected
to put family ahead of any entrepreneurial ambitions. Related work--for
example, by Camilo Mondragon-Velez and Ximena Pena Parga--has found that
few self-employed individuals in developing countries ever make the
transition to taking on employees and expanding their business.
All that said, Schoar observed that there were also real
bottlenecks to small business growth in much of the developing world.
Esther Duflo and others had found that capital constraints are
important, and work by Nicholas Bloom pointed to a shortfall of
managerial know-how among small business owners in these countries.
An intriguing finding, from work that Schoar had done with Dean
Karlan and others, was that management training and consulting aimed at
small business owners in developing countries had the potential not only
to increase their output, but also to change the owners'
expectations and desires. This suggested that for some small businesses
at least, the expressed disinterest in growing was endogenous to the
constraints they perceived rather than a permanent feature.
Finally, Schoar cited work by Augustin Landier and David Thesmar
indicating that those entrepreneurs who do want to grow their businesses
tend to be overoptimistic about their prospects, which suggested that
some of Hurst and Pugley's findings might be biased upward.
Diego Comin suggested that a worthwhile next step for the
authors' research agenda would be to focus on those small companies
that do grow, to try to understand why they grow. Even more interesting
would be to examine whether the factors that drive their growth vary at
a business cycle frequency. Laurence Ball agreed that closer study was
needed of the firms that achieve the transition from small to large. Are
they concentrated in some industries than others? Do they innovate more?
And is it possible to better target to these growing and innovating
companies the policies that now benefit small businesses across the
board?
Edward Lazear remarked that he routinely asks his MBA students
whether they are seeking a business degree because of the money they
expect to earn, and typically only a handful say yes. The implausibility
of this response--if business students are not in it for the money, who
is?--suggested to him that questions of that nature, including the
authors' question about nonpecuniary benefits, are not well framed.
Perhaps one should ask instead how much income respondents would be
willing to forgo in order to be their own boss. Or one could compare the
actual incomes of self-employed and salaried workers, but those data are
not unequivocal--indeed, it is unclear that the implied trade-off even
exists. Although data from the Current Population Survey indicate that
self-employed workers earn less on average than salaried workers, those
data do not control for individual characteristics. Lazear's own
data, which do control for such differences, yield the opposite result.
Jesse Rothstein added that workers in some occupations, such as
hairdressers, have little choice but to operate as a small business.
These workers' answers to the question of why they chose to start a
small business might show up in the "nonpecuniary" category
even if they entered primarily to make money.
Robert Gordon noted that whereas the paper focused on innovation by
small businesses, it was also worth thinking about the impact of
innovation on small businesses. Much innovation is by large rather than
small businesses, and work by David Autor and Lawrence Katz finds that
much of that innovation, together with globalization, is rendering many
U.S. jobs obsolete. How many of those jobs, Gordon wondered, are in
small businesses? He cited an article in that morning's Wall Street
Journal reporting that small business start-ups today are being launched
with only about two-thirds as many employees, on average, as those
launched in the 1990s. It could well be that information technology is
what is allowing these new businesses to start out with fewer employees.
A concrete example of this phenomenon, Gordon suggested, was the
auto repair industry. The introduction of online parts catalogs has
reduced the labor cost of running an auto repair shop; on the other
hand, the greater technical complexity of today's automobiles has
shifted competitive advantage to the larger dealerships, which can more
easily acquire the necessary capital equipment and hire and train the
more specialized workers needed. A contrasting example, Gordon observed,
could be found in the health care industry: while the large health care
organizations are making the change to electronic record-keeping, many
small medical practices still have the traditional roomful of file
cabinets and seem resistant to change.
Robert Shiller wondered whether the real value of many small
businesses lay not so much in their innovative potential as in their
ability to provide more individualized personal service. The success of
so many big restaurant franchises notwithstanding, few people, he
imagined, would want to see the chains entirely supplant the familiar,
owner-operated neighborhood restaurant. The same could be said of many
personal service providers, such as clinical psychologists and
accountants. Shiller noted that the Small Business Administration offers
little support to this type of business, instead catering mainly to
other types for which it can play a more venture capital-like role.
Although small businesses of the first type do receive quasi-subsidies
in the form of lighter regulation and lower taxes, these seemed to him
justified if they allow such businesses to continue providing the
attentive, one-on-one service that customers value. Ricardo Reis added
that although such businesses might not be innovative in the sense of
introducing wholly new products and services, they do often innovate in
a broader sense of providing greater local variety--adding a Thai
restaurant where formerly there were only pizza and Chinese, for
example. They also provide local competition--actual competition, as
opposed to just the threat of competition as would suffice in a
frictionless economy--and thus help keep prices close to marginal cost.
Reis also pointed out that if the authors were right that the bulk
of the benefits of small businesses are largely nonpecuniary, one should
ask whether this also determines the marginal benefits when one chooses
to start a business. If so, subsidizing small businesses is not so
costly because they are supporting something of value without distorting
incentives. Small business owners are doing what they would have done
anyway, because it is the nonpecuniary benefits that are driving their
behavior. The question for Reis was whether this proposition could be
tested: the effective subsidies to small businesses have varied greatly
over time and space--could one exploit this variation to measure their
effects?
The argument over small business subsidies reminded Refet Gurkaynak
of the debate in Europe over the Common Agricultural Policy: the French
have long resisted the elimination of subsidies to small family farms,
arguing that these farms are essential to maintaining the traditional
French way of life. Gurkaynak felt that although such subsidies are
indefensible on efficiency grounds, if they promote something that the
public wishes to preserve, whether it be national identity or variety or
something else, policymakers should respect that wish, even at the cost
of some efficiency loss. In other words, whereas Reis had argued that
small business subsidies are justifiable because they create no economic
distortions, Gurkaynak was arguing that they may be justifiable even if
they do create distortions. The question then becomes how to subsidize
them in the least inefficient manner.
Justin Wolfers objected to the previous speakers'
beatification of non-entrepreneurial small businesses. After all, he
noted, these businesses create few jobs and often mistreat those they do
employ; they are exempt from or can circumvent workplace safety, family
leave, minimum wage, and antidiscrimination laws and regulations, among
others. Many of them cheat on their taxes as well.
Alan Blinder said he had been about to comment that the Panel was
surely in agreement on not subsidizing small businesses, but clearly
that assumption was no longer tenable. In his view, the effective
absence of taxation on small businesses in the United States created an
enormous distortion that could only be justified by some overwhelming
rationale in its favor--something far beyond having a different ethnic
restaurant on every corner. Blinder accepted the paper's finding
that only a small fraction of small businesses innovate and grow, but
one thing economists still agreed on, he thought, was that government is
incapable of picking those winners ex ante, which meant that not even
those few small businesses should be subsidized.
Robert Hall saw the paper as undermining the whole notion that
"small business sector" is an interesting or useful concept;
such a classification, the paper had shown, lumps together businesses
that are in fact quite different from one another. He agreed that only a
few small businesses are really entrepreneurial. Indeed, work that he
had done with Susan Woodward suggested that entrepreneurial effort is
dramatically undersupplied in the U.S. economy--a phenomenon that they
believed was associated with the extreme burden of idiosyncratic risk
that entrepreneurs bear. It would be better to redirect attention--and
the activity now conducted by the Small Business Administration--away
from the bulk of small businesses and toward those that engage in real
entrepreneurship.
Frederic Mishkin urged a distinction between types of small
business subsidies: while agreeing with Hall that formal subsidies like
those offered by the SBA should be shifted to those businesses that
could make better use of them, he argued that subsidies that take the
form of exemption from regulation are different, and indeed are not
"subsidies" at all in the usual sense. If, as seems likely,
there are high fixed costs to complying with regulations, then a strong
argument can be made for exempting small businesses from them.
Following up on Mishkin's observation, Randall Kroszner
conjectured that for many small businesses, the deterrent to growth was
precisely the prospect of higher regulatory costs if they grew beyond a
certain threshold size. It would be interesting to see, given Adam
Looney's documentation that many regulations have a size threshold
for exemption at x employees, whether there is a clustering of firms at
x--1.
Benjamin Friedman reminded the Panel that small firms are not the
only recipients of government subsidies. General Electric, as is well
known, paid no federal income tax in 2010. Many other large firms,
including many that invest in intensive lobbying, likewise pay no income
tax. The oil industry, which is heavily subsidized, is hardly dominated
by small firms, nor is the defense industry, although small firms do
exist in both. The U.S. Export-Import Bank is an important source of
federally subsidized lending, and although it now has a dedicated small
business program, the bank still mostly caters to large companies. More
than a hundred Boeing aircraft a year--nearly a third of the
company's output--receive Ex-Im Bank subsidized loans, and the
financing it provides to this one company's sales dwarfs its entire
small business loan program in value.
Christopher Carroll responded to Hall that it does make sense to
consider small business as a class, precisely on the grounds that all
are exempt from certain regulations to which other businesses are
subject and from certain taxes that other businesses pay. If one wants
to analyze the effects of those regulations and tax provisions, one
cannot avoid making the distinction between small and large. On the
other hand, Friedman was correct to point to the favors that the largest
businesses receive. It may be that it is the firms in the middle that
are at a disadvantage relative to the extremes. Carroll also suggested
that anyone who sees few economic consequences from the implicit subsidy
inherent in weak tax enforcement should look at Italy. The Italian
marketplace is full of small shops that almost surely would not survive
were they not effectively exempt from taxation. Some Italian economists
believe that the country's recent economic stagnation owes much to
the political system's refusal to rationalize the tax burden across
the different size categories.
Susan Woodward noted that if the owners of small businesses start
them for mainly nonpecuniary reasons, one might expect those businesses
to be relatively immune from business cycle fluctuations. The experience
of the last recession, however, suggested otherwise. Whereas payroll
employment had fallen 5 percent from its 2007 peak, self-employment,
taken as the sum of incorporated and unincorporated self-employed
persons, had fallen 13 percent. Likewise the number of firms with 100 or
fewer employees was down 10 percent, compared with 5 percent for all
firms. Those who would like to see fewer small businesses thus seemed to
be getting their wish.
Annette Vissing-Jorgensen asked whether the authors could sort
their data by industry or in some other way that would reveal which
small business owners are more likely to choose that form of
organization for nonpecuniary reasons.
Responding to the discussion, Erik Hurst agreed with Blinder that
the way to frame the subsidy issue was to say that there are costs as
well as benefits from subsidization, and one of the aims of the paper
was to point out that the net costs could be substantial, especially if
the targeted firms do not respond in the way the subsidizer had hoped.
In reply to Reis, Hurst suggested that the presence of private benefits
from having more ethnic restaurants and the like was not sufficient
grounds for subsidization--one also derives private benefits from having
flat-screen TVs, for example--and he thought that subsidizing them might
well lead to distortions.
Hurst agreed with Lazear that some available evidence indicated a
difference in average pay between self-employed and payroll workers that
compensated the latter for the disamenities of not being one's own
boss. He also agreed that surveys were not the best instrument for
measuring that differential. What interested Hurst in the survey data
was that there seemed to be some correlation between respondents who
reported nonpecuniary benefits and the actual growth of their
businesses.
Finally, Hurst also agreed with the panelists who had called for
zeroing in on the small firms that do grow, to see how they differ from
other small firms, and in fact he and Pugsley were presently seeking to
obtain Census data to do exactly that.
Table 1. Four-Digit NAICS Industries with the Largest Shares of All
Small Businesses, 2007 (a)
Percent of
all small
Rank Industry (NAICS code) businesses
1 Residential Build. Const. (2361) 3.5
2 Offices of Physicians (6211) 3.2
3 Legal Services (5411) 3.2
4 Build. Equip. Contractors (2382) 3.1
5 Religious Organizations (8131) 3.0
6 Svcs. to Build. and Dwellings (5617) 3.0
7 Auto Repair and Maint. (8111) 2.7
8 Limited-Svc. Eating Places (7222) 2.6
9 Full-Service Restaurants (7221) 2.6
10 Mgmt./Sci./Tech. Consult. (5416) 2.5
11 Insurance Agencies (5242) 2.3
12 Build. Finishing Contractors (2383) 2.3
13 Offices of Dentists (6212) 2.2
14 Other Health Practitioners (6213) 2.0
15 Found./Struct./Build. Comr. (2381) 1.9
16 Accounting Services (5412) 1.9
17 Real Estate Agents and Brokers (5312) 1.8
18 Computer Systems Design (5415) 1.8
19 Personal Care Services (8121) 1.7
20 Lessors of Real Estate (5311) 1.7
21 Arch. /Engineering Serv. (5413) 1.7
22 Oth. Spec. Trade Comr. (2389) 1.7
23 Activ. Rltd. to Real Estate (5313) 1.3
24 Gasoline Stations (4471) 1.3
25 Oth. Prof., Sci, Tech. Svcs. (5419) 1.1
26 Grocery Stores (4451) 1.1
27 Bus./Prof./Labor/Political Orgs. (8139) 1.1
28 General Freight Trucking (4841) 1.0
29 Wholesale Electronic Mkts. (4251) 1.0
30 Amusement and Recr. (7139) 1.0
31 Child Day Care Services (6244) 0.9
32 Spec. Freight Trucking (4842) 0.9
33 Drinking Places (Alch.) (7224) 0.8
34 Other Fin./Invest. Activities (5239) 0.8
35 Health and Pets. Care Stores (4461) 0.8
36 Clothing Stores (4481) 0.7
37 Build. Material Dlrs. (4441) 0.7
38 Nonres. Build. Constr. (2362) 0.7
39 Mach./Equip./Suppl. Whsle. (4238) 0.7
40 Other Misc. Retailers (4539) 0.7
Cumulative
Rank percent
1 3.5
2 6.7
3 9.9
4 13.0
5 16.0
6 19.0
7 21.7
8 24.3
9 26.9
10 29.4
11 31.8
12 34.1
13 36.2
14 38.2
15 40.1
16 42.0
17 43.9
18 45.6
19 47.3
20 49.0
21 49.0
22 50.7
23 52.0
24 53.3
25 54.4
26 55.5
27 56.6
28 57.6
29 58.6
30 59.6
31 60.5
32 61.4
33 62.2
34 63.0
35 63.8
36 64.5
37 65.2
38 65.9
39 66.6
40 67.3
Source: 2007 Economic Census data.
(a.) Businesses with fewer than 20 employees in the 2007 Economic
Census are classified by their four-digit NAICS industry codes
(in parentheses), and those 294 four-digit industries are then
ranked by their share of all small businesses thus defined.
Table 2. Shares of Small Businesses and Small Business Employment
in Major Sectors, by Firm Age, 200S,
Firm age (years)
Sector 0-10 10-25 All firms
Small businesses as percent
of all firms
All industries 92.0 85.7 87.2
Finance, insurance,
and real estate 95.5 91.8 91.9
Agriculture 94.8 88.1 91.6
Construction 93.7 86.0 88.9
Wholesale trade 93.0 83.2 84.1
Services 92.7 88.4 89.1
Transportation,
communications, and
utilities 92.3 82.2 86.0
Retail 88.6 81.8 84.6
Manufacturing 85.5 71.5 72.4
Small businesses employment
as percent of all employment
All industries 44.8 24.7 19.4
Finance, insurance,
and real estate 50.8 31.7 19.0
Agriculture 57.7 47.1 50.1
Construction 59.1 38.4 39.4
Wholesale trade 52.8 30.6 21.7
Services 40.7 23.1 20.8
Transportation, 44.2 14.7 11.8
communications, and
utilities
Retail 46.9 24.8 18.8
Manufacturing 34.6 16.0 8.5
Source: 2005 Business Dynamics Statistics data (www.ces.census.gov/
index.php/bds/bds_database list).
(a.) Small businesses are defined as firms with fewer than 20
employees. Sector classifications are those provided by the
Business Dynamics Statistics (BDS). Like the Statistics of U.S.
Businesses data, the BDS data include information only on firms
with paid employees.
Table 3. Change in Employment at Existing Small Businesses,
by Firm Age, 2003 (a)
Percent
Firm age (years)
Direction of change in employ- 1-10 11-20 21+ All firms
ment
Over last year
Increase 18.9 10.6 9.1 13.9
No change 74.3 79.7 84.0 78.4
Decrease 6.8 9.8 6.9 7.7
Sample size 1,163 817 727 2,707
Over last 3 years
Increase 27.6 19.4 15.3 21.3
No change 61.0 64.9 72.5 65.6
Decrease 11.3 15.7 12.2 13.1
Sample size 847 814 725 2,386
Source: 2003 Survey of Small Business Finances data.
(a.) Firms with fewer than 20 employees only. See text for further
description of the sample.
Table 4. Changes in Employment at New Businesses over Last 4 Years,
2008 (a)
Change in employment
>1 employee >5 employees >10 employees
Percent of firms 41.9 10.8 3.6
Source: Kauffman Firm Survey data.
(a.) Sample consists of the 2,617 surveyed firms of all sizes that had
remained in business for the 4 years since the survey started. Firms
in the sample had a median employment of 1 employee and a mean
employment of 3; the number of employees at the 90th percentile was
14. All data are weighted using the sample weights provided within the
survey.
Table 5. Regressions Using Industry Shares of Small Businesses to
Predict Small Business Job Creation and Destruction (a)
Dependent variable
Gross job Gross job
Independent variable creation rate creation rate
Industry share of all small -2.14 -1.98
businesses
(0.25) (0.23)
No. of observations 929 929
[R.sup.2] 0.077 0.093
Weighted by number of small
firms within industry (c)
Industry share of all small -0.73 -0.73
businesses
(0.27) (0.26)
No. of observations 929 929
[R.sup.2] 0.42 0.421
Industry job creation controls? Yes Yes
Industry job births controls? No Yes
Industry job destruction controls? No Yes
Dependent variable
Gross job
creation Gross job
Independent variable rate (b) birth rate
Unweighted
Industry share of all small -2.05 0.10
businesses
(0.29) (0.17)
No. of observations 652 666
[R.sup.2] 0.205 0.351
Weighted by number of small
firms within industry (c)
Industry share of all small -0.73 0.03
businesses
(0.30) (0.24)
No. of observations 652 666
[R.sup.2] 0.437 0.588
Industry job creation controls? Yes No
Industry job births controls? No Yes
Industry job destruction controls? No No
Dependent variable
Gross job
destruction
Independent variable rate
Industry share of all small -0.23
businesses
(0.17)
No. of observations 656
[R.sup.2] 0.353
Weighted by number of small
firms within industry (c)
Industry share of all small 0.04
businesses
(0.26)
No. of observations 656
[R.sup.2] 0.531
Industry job creation controls? No
Industry job births controls? No
Industry job destruction controls? Yes
Source: Authors' regressions using pooled 2003-06 Statistics of U.S.
Businesses data.
(a.) Small businesses are defined as those with fewer than 20
employees. The unit of observation is four-digit NAILS industries.
All regressions include fixed time effects. Robust standard errors
are in parentheses.
(b.) Sample is limited to industries that have nonmissing gross job
birth and destruction rates.
(c.) Estimates can be viewed as grouped-data estimates of firm-level
job creation or destruction if employment shares of small firms within
an industry were equal across industries.
Table 6. Innovation-Related Behavior of New Businesses, 2004-08 (a)
Measure of innovative activity, at
year 4 of firm's existence
Have or are Have or are
applying for a applying for a
patent copyright
Percent of all 2.7 8.9
new firms
Sample size 2,581 2,550
Measure of innovative activity, at
year 4 of firm's existence
Have or are Have either a
applying for a patent, a trademark,
trademark or a copyright
Percent of all 12.3 17.3
new firms
Sample size 2,546 2,510
Source: Authors' calculations from 2008 Kauffman Firm
Survey data.
(a.) Sample sizes differ because of different response
rates to different questions. All data are weighted
using the provided survey weights.
Table 7. Innovation-Related Activities of Nascent Entrepreneurs, 2006
and 2010 (a)
Percent of firms
First year Fifth year
business (2006) of business
(2010):
Positive positive
revenue revenue
Indicator All only only (b)
Firm had applied for patent, 4.9 6.0 17.6
copyright, or trademark
Firm had developed proprietary 6.5 8.3 20.3
technology, processes, or
procedures
Owner stated that many existing 35.7 43.3 39.6
firms already offer same
product or service to
expected customer base
Owrier stated that no existing 19.2 13.3 17.3
firms already offer same
product or service to
expected customer base
Sample size 1,214 602 162
Source: Authors' calculations from PSED data.
(a.) All data are weighted using the sample weights from the indicated
survey year.
(b.) Responses are those given in the 2010 survey.
Table 8. Ex Ante Expectations and Desires of Nascent Entrepreneurs
about Future Growth and Innovation, 2006 and 2010 (a)
First year of
business (2006) Fifth year
of business
Positive (2010): po-
revenue sitive re-
Indicator All only venue only (b)
Percent reporting 24.3 23.0 28.3
that they want firm
to be "as large as
possible"
Expected no. of
employees when
firm is 5 years old
25th-percentile response 1 0 0
Median response 4 3 3
75th-percentile response 10 8 6
90th-percentile response 29 24 25
Percent expecting to 14.6 9.2 12.2
develop proprietary
technology, processes,
or procedures in future
Percent expecting to apply 26.0 17.9 24.9
for patent, copyright,
or trademark in future
Percent expecting R&D 25.7 19.5 22.8
spending to be a
major priority for
the business
Source: Authors' calculations from PSED data.
(a.) All data are weighted using the sample weights from the indicated
survey year. Sample sizes differ slightly from those in table 7
because not all respondents provided responses to all the questions.
(b.) Responses are those given in the 2006 survey.
Table 9. Distribution of Reasons Offered by Nascent Entrepreneurs for
Starting a Business,
Percent
Positive revenue
All only, 2006
Either
First response First Either
Reason for starting response (b) response response
business
Nonpecuniary reasons 35.3 50.5 37.6 53.9
To generate income 19.5 34.1 21.4 36.6
Had good business idea 32.2 40.6 28.3 34.9
or to create new
product
Lack of other 2.2 3.8 2.6 4.0
employment options
Other 10.8 15.7 10.2 15.5
No. of respondents 1,214 602
Positive revenue only,
2010
First Either
Reason for starting response response
business
Nonpecuniary reasons 35.0 52.4
To generate income 17.6 32.4
Had good business idea 33.8 37.5
or to create new
product
Lack of other 2.6 4.3
employment options
Other 11.0 14.7
No. of respondents 162
Source: Authors' calculations from PSED data.
(a.) The table uses the same data set and sample construction as in
table 7. All data are weighted using the sample weights from the
indicated survey year. Respondents were asked, "Why do [or did] you
want to start this new business?" and could give up to two
responses in their own words. All data are weighted using the
sample weights from the indicated survey year.
(b.) Percentages sum to less than 200 percent because about
one-quarter of respondents did not provide a second response and,
of those who did, some provided a response that was classified in
the same broad category as the first. See appendix table A1 for a
more detailed classification of responses.
Table 10. Regressions Explaining New Businesses' Expectations and
Innovative Behavior by Differences in Motivations
Dependent variable Constant
All res-
pondents,
2006
Owner stated that many existing firms already offer 0.380
same product or service to expected customer base
Owner stated that no existing firms already offer same 0.190
product or service to expected customer base
Firm has developed proprietary technology, processes, 0.084
or procedures
Firms expects to apply for patent, copyright, or trade- 0.230
mark in future
Percent of firms reporting that they want to be "big" 0.288
Expected no. of employees when firm is 5 years old 8
(75th percentile)
Expected no. of employees when firm is 5 years old 25
(90th percentile)
Owner stated that many existing firms already offer 0.508
same product or service to expected customer base
Owner stated that no existing firms already offer same 0.130
product or service to expected customer base
Firm has developed proprietary technology, processes, 0.093
or procedures
Firms expects to apply for patent, copyright, or trade- 0.161
mark in future
Percent of firms reporting that they want to be "big" 0.262
Expected no. of employees when firm is 5 years old 5
(75th percentile)
Expected no. of employees when firm is 5 years old 20
(90th percentile)
Coeffi-
cient on
"create
new
product"
Dependent variable dummy (c)
All res-
pondents,
2006
Owner stated that many existing firms already offer -0.082
same product or service to expected customer base (0.035)
Owner stated that no existing firms already offer same 0.037
product or service to expected customer base (0.029)
Firm has developed proprietary technology, processes, 0.010
or procedures (0.019)
Firms expects to apply for patent, copyright, or trade- 0.104
mark in future (0.034)
Percent of firms reporting that they want to be "big" 0.036
(0.035)
Expected no. of employees when firm is 5 years old 4.0
(75th percentile) (2.3)
Expected no. of employees when firm is 5 years old 15.0
(90th percentile) (5.3)
Owner stated that many existing firms already offer -0.112
same product or service to expected customer base (0.053)
Owner stated that no existing firms already offer same 0.059
product or service to expected customer base (0.041)
Firm has developed proprietary technology, processes, 0.027
or procedures (0.029)
Firms expects to apply for patent, copyright, or trade- 0.067
mark in future (0.045)
Percent of firms reporting that they want to be "big" 0.042
(0.049)
Expected no. of employees when firm is 5 years old 5.0
(75th percentile)
Expected no. of employees when firm is 5 years old 20.0
(90th percentile)
Coeffi-
cient on
nonpecu-
niary
motives"
Dependent variable dummy (d)
All res-
pondents,
2006
Owner stated that many existing firms already offer 0.049
same product or service to expected customer base (0.035)
Owner stated that no existing firms already offer same -0.049
product or service to expected customer base (0.028)
Firm has developed proprietary technology, processes, -0.041
or procedures (0.018)
Firms expects to apply for patent, copyright, or trade- 0.010
mark in future (0.033)
Percent of firms reporting that they want to be "big" -0.047
(0.033)
Expected no. of employees when firm is 5 years old -2.0
(75th percentile) (2.3)
Expected no. of employees when firm is 5 years old -10.0
(90th percentile) (5.1)
Owner stated that many existing firms already offer 0.020
same product or service to expected customer base (0.051)
Owner stated that no existing firms already offer same -0.024
product or service to expected customer base (0.036)
Firm has developed proprietary technology, processes, -0.051
or procedures (0.028)
Firms expects to apply for patent, copyright, or trade- 0.054
mark in future (0.041)
Percent of firms reporting that they want to be "big" -0.028
(0.046)
Expected no. of employees when firm is 5 years old 0.0
(75th percentile)
Expected no. of employees when firm is 5 years old -10.0
(90th percentile)
Difference
between
coeffi-
Dependent variable cients
All res-
pondents,
2006
Owner stated that many existing firms already offer 0.131
same product or service to expected customer base
Owner stated that no existing firms already offer same -0.086
product or service to expected customer base
Firm has developed proprietary technology, processes, -0.051
or procedures
Firms expects to apply for patent, copyright, or trade- -0.094
mark in future
Percent of firms reporting that they want to be "big" -0.083
Expected no. of employees when firm is 5 years old -6.0
(75th percentile)
Expected no. of employees when firm is 5 years old -25.0
(90th percentile)
Owner stated that many existing firms already offer 0.132
same product or service to expected customer base
Owner stated that no existing firms already offer same -0.083
product or service to expected customer base
Firm has developed proprietary technology, processes, -0.078
or procedures
Firms expects to apply for patent, copyright, or trade-
mark in future -0.013
Percent of firms reporting that they want to be "big" -0.070
Expected no. of employees when firm is 5 years old -5.0
(75th percentile)
Expected no. of employees when firm is 5 years old -30.0
(90th percentile)
p value
of dif-
Dependent variable ference
All res-
pondents,
2006
Owner stated that many existing firms already offer <0.01
same product or service to expected customer base
Owner stated that no existing firms already offer same 0.01
product or service to expected customer base
Firm has developed proprietary technology, processes, 0.01
or procedures
Firms expects to apply for patent, copyright, or trade- 0.01
mark in future
Percent of firms reporting that they want to be "big" 0.03
Expected no. of employees when firm is 5 years old 0.03
(75th percentile)
Expected no. of employees when firm is 5 years old 0.01
(90th percentile)
Owner stated that many existing firms already offer 0.02
same product or service to expected customer base
Owner stated that no existing firms already offer same 0.05
product or service to expected customer base
Firm has developed proprietary technology, processes, 0.02
or procedures
Firms expects to apply for patent, copyright, or trade-
mark in future 0.77
Percent of firms reporting that they want to be "big" 0.17
Expected no. of employees when firm is 5 years old 0.01
(75th percentile)
Expected no. of employees when firm is 5 years old 0.01
(90th percentile)
Source: Authors' regressions using data from the PSED.
(a.) Robust standard errors are in parentheses.
(b.) Represents the mean for respondents who did not report starting
their business for either nonpecuniary or new product motives.
(c.) Variable equals 1 if the respondent stated a new product motive
as a reason for starting the business.
(d.) Variable equals 1 if the respondent stated a nonpecuniary motive
as a reason for starting the business.
