4.6 Sample Properties
4.6.4 Establishment Size and Growth
Studies of industrial organization stress the importance of size as an indicator of an establishment’s potential for long-term survival. Size represents the establishment’s access to critical resources, financial or otherwise, productive efficiencies attainable through internal scale economies (Audretsch and Mahmood 1995; Mata et al. 1995; Audretsch et al. 1999), the entrepreneur’s level of uncertainty in undertaking a new business endeavor (Jovanovic 1982; Pakes and Ericson 1998), and potential ‘sunk’ costs in non-recoverable assets that discourage exit (Caves and Porter 1976). Starting size also correlates strongly with the adoption of advanced production technologies, which has also been shown to increase the
likelihood of new firm survival (Doms et al. 1995).
Establishment employment s is also an important indic an establishment’s potential relationship with its external environment, namely its reliance on external reso to compensate for its limited
internal production capacity (Sweeney and Feser 1998; Feser 2001a). Industry At Birth At 8 Years Incumbent (1994)
Manufacturing (ex. SIC 21) 7.1 18.4 45.2
Drugs 12.4 23.3 132.0
Farm & garden machinery 7.3 12.9 51.1
Metalworking machinery 5.1 11.5 20.2
Electronic components & accessories 11.5 39.9 79.7 Motor vehicle parts & accessories 9.4 41.6 95.4 Measuring & controlling devices 8.7 25.5 54.0
Professional services (SIC 73 & 87) 3.9 12.9 14.2
Advertising 2.9 7.3 9.5
Computer & data processing services 3.0 14.9 13.5 Research & testing services 4.5 16.2 23.3
Source: US BLS Longitudinal Database (LDB) New Firms
Establishment Size - Average Employment Table 4.7
ize ator of
urces
Most new firms are very small. Keeping in mind that entrants with over 250 employees have been excluded from the sample, the average new manufacturing firm has only 7.1 employees per establishment at birth, roughly 1/6th the size of the average incumbent in 1994
(see Table 4.7). The bulk of new manufacturing firms (70 percent) have fewer than five employees at birth, and less than 1 percent are born with more than 100 employees. For manufacturing as a whole, 43 percent of the establishments have less than five employees, and a considerable number of these are new and young establishments.23 Similar trends are
observed within the specific study industries. At birth, new firms in drugs, electronic components and motor vehicle parts are somewhat larger than the other study industries, but they also face a larger absolute size gap with incumbents. New firms in metalworking machinery face the smallest gap, perhaps explaining their higher survival rates.
New firms in professional services are considerably smaller than their manufacturing counterparts, but then so are professional service incumbents (see Table 4.7). The average new professional services firm has approximately four employees per establishment, compared to an incumbent average of only 13 employees. New research and testing firms are the slightly larger than the other study industries. The size distribution for new
professional services establishments is also more skewed than manufacturing, with roughly 85 percent of new firms having less than five employees and only 0.3 percent larger than 100 employees.
There is a notable tendency for the relative size of new firms to vary systematically with the average incumbent size across industries. Industries where the average incumbent size is large tend to have larger entrants.24 This suggests that the structural forces that condition
incumbent size across industries also affect new enterprises. The drugs industry has the largest new firms (12.4 employees per establishment) and the largest incumbents (132 employees per firm). But new firms in the drugs industry also have the largest size gap with incumbents, suggesting that these firms may have greater difficulty matching scale
efficiencies if they were to compete directly against incumbents.25 Metalworking machinery
has the smallest new firms (5.1 employees per establishment), the smallest incumbents (20.2 employees per establishment) and the smallest new firm size deficiency. The average establishment in metalworking machinery devices is only 4 times larger than the average entrant. Among professional services, research and testing has both the largest new firms (4.5 employees per establishment) and the largest incumbents (23.3). New firms in
24 The pairwise correlation coefficient between new firm size at birth and incumbents in 1994 is 0.81 when measured across all manufacturing and professional services sectors.
25 A large new firm-incumbent size gap may also signify industries where entrants compete on the industry fringe (i.e. niche markets).
advertising have the smallest size gap with incumbents, on average they are only one-third the size of incumbents.
While informative, a firm’s size at birth is only a static indicator of its potential relationship to the local environmental context and potential for survival. Mata (1995; 1996) and Wagner (1994) argue that
establishment growth, rather than size at birth, is a more appropriate indicator of scale and information advantages. Growth also reflects organizational learning if, in the face of uncertain market prospects, entrepreneurs choose to start small and expand if their market experience is favorable (Jovanovic 1982; Pakes and Ericson 1998).
Figure 4.4 New Firm Size by Age
Source: US BLS Longitudinal Database (LDB)
0 5 10 15 20 0 4 8 12 16 20 24 2 Quarter E m pl oy ee s pe r E sta b lis hm en t
All New Firms - Manufacturing Survivors - Manufacturing
All New Firms - Professional Services Survivors - Professional Services
8
Measuring the relationship between age and average establishment size must account for selective attrition of small firms (Evans 1987; Dunne et al. 1989). If small employers exit the market early, average establishment size increases even if the actual size of the surviving plants remains constant. To account for selective attrition, I measure both the average size of all new firms as well as the average for only entrants that survive the entire study period. The average size of survivors will not be influenced by sample attrition.
Figure 4.4 presents the average entrant size by quarter for the duration of the study period for manufacturing and professional services. Supplemental charts for each study industry are provided in Appendix C. Three patterns stand out. First, the average size of
surviving establishments almost always exceeds the overall average size, reflecting higher propensity for exit among smaller firms. The drugs industry is a notable exception where eventual survivors are slightly smaller than non-survivors in the first few years. Second, even after seven full years new manufacturing firms are still much smaller than incumbents (see Table 4.7 for incumbent size in 1994). After seven full years the average manufacturing plant born in 1994 or 1995 has only 18.6 employees, compared to a 1994 incumbent average of 45.2 employees per plant. This suggests that eight years is hardly long enough to establish oneself as a dominant manufacturer in most markets. For professional services, the size gap with incumbents is nearly eliminated by the end of the study period, 13 employees per establishment for new firms compared to 14 for incumbents. In computer and data processing the average size of new firms after seven years is slightly larger than the
incumbent size in 1994. Third, there is evidence that new firm employment growth plateaus in the final quarters of the study period. This is expected in services where the average entrant size approaches the incumbent average. But even in manufacturing it appears that survivors approach an equilibrium size considerably smaller than incumbents, suggesting niche market production as opposed to direct competition.
The preceding analysis suggests that new firms, particularly those in manufacturing, face serious scale disadvantages if forced to compete directly against incumbents on a pure efficiency basis. But the existence of a significant size rift between new firms and
incumbents throughout the study period suggests that new firms can survive at a sub-optimal level of efficiency for quite some time. Presumably they do so by occupying a strategic niche market where competition might be on the basis on service, quality or innovation (Caves and Porter 1977, Agarwal and Audretsch, 2001; Porter 1979, 1980). For this study,
the key implication is that post-entry dynamics of new firms may be very differently than established competitors in the same industry. The results found by this study could be considerably different than similar work conducted on a broader population of firms.