1 Introduction

Research on the economic importance of small firms was negligible until Birch (1979) claimed that they generated a disproportionately large share of new net jobs.Footnote 1 Birch’s findings have been criticized by, e.g., Brown et al. (1990), Davis et al. (1996a, b), and Haltiwanger and Krizan (1999), but they sparked small business research. It is now a vigorous research field with a wide coverage, encompassing issues such as the importance of entrepreneurship, firm demography, and firm dynamics for job creation and economic growth.Footnote 2 Van Praag and Versloot (2008) review the empirical literature on the economic contribution of entrepreneurial firms, i.e., small and young firms, which are found to have positive effects on employment, productivity, innovation, and utility. With reference to employment Van Praag and Versloot conclude (p. 135): “Entrepreneurs create more employment than their counterparts, relative to their size. This result is unambiguous. Small and young firms are required to boost employment.”Footnote 3

The purpose of this article is to go one step further and survey the empirical evidence on whether, in fact, net employment growth is generated by a few rapidly growing, not necessarily small and young, firms, so-called Gazelles. The term was coined by Birch some 20 years ago (Landström 2005, p. 170) to denote a small group of high-growth firms that, according to him, generated most of the new net jobs in the economy. This stands in contrast to the few large (often publicly traded) companies, known as “Elephants,” which according to Birch had a large employment share, but generated few new jobs, and to the vast majority of all firms that started out small, grew very little and hence contributed only marginally to employment growth. The latter firms were termed “Mice.”Footnote 4 In addition, we are interested in whether Gazelles, in fact, are young and small, and whether Gazelles are overrepresented in high-technology industries.Footnote 5 Much economic policy has been targeting high-technology firms since politicians have relied on high-technology firms and industries to boost economic growth and job creation. The research questions may be stated as four propositions:

  • Proposition 1: In a population of firms, net employment growth is generated by a small number of high-growth firms, so-called Gazelles.

  • Proposition 2: On average, Gazelles are younger than other firms.

  • Proposition 3: On average, Gazelles are smaller than other firms.

  • Proposition 4: Gazelles are overrepresented in high-technology industries.

In the next section we discuss the definition of Gazelles and the method used in our survey. Section 3 reports the results from the identified studies. These results are analyzed in Section 4, where we also offer our conclusions.

2 Definitions and method

There is no general agreement on the definition of Gazelles. Birch (e.g., Birch et al. 1995, p. 46) defines them as “A business establishment which has achieved a minimum of 20% sales growth each year over the interval, starting from a base-year revenue of at least $100,000.” Hence, the definition is based on firms growing at least at a particular pace (e.g., that firms exhibit a certain annual growth rate or more for a certain number of years). Another way is to use a high-growth threshold and define Gazelles as the x% fastest growing firms. Recently, the Organization for Economic Cooperation and Development (OECD; Ahmad 2006) proposed defining high-growth enterprises as enterprises with an average employment growth rate exceeding 20% p.a. over a 3-year period and with ten or more employees at the beginning of the period. They also proposed that the term Gazelle should only apply to young high-growth firms, or more specifically to enterprises less than 5 years old and with an average employment growth rate exceeding 20% p.a. over a 3-year period and with ten or more employees at the beginning of the period. Consequently, the literature is quite disparate.

Delmar et al. (2003, pp. 192–197) systematize the literature on high-growth firms. Several issues are addressed showing large heterogeneity among studies:

Choice of growth indicator. Employment, market share, physical output, profits, and sales are by far the most commonly used.

Measurement of growth. Growth is measured in several ways, both in absolute and relative terms. Multiple or composite growth indicators and growth measures are also employed.

The regularity of firm growth over time. Firm growth fluctuates substantially over time. The choice of time period over which growth is measured, annual growth, growth between initial and final year etc., therefore affects observed growth rates.

The process by which firms grow, i.e., organic or acquired growth.Footnote 6

Firm demographics. Firm size, firm age, and industry affiliation have been shown empirically to have a large impact on firm growth, and therefore need to be considered.

We have also noted that two different kinds of benchmarks are used to evaluate the job contribution of Gazelles: either by comparing it to the job contribution of non-Gazelles in the investigated population or by relating it to aggregates such as total employment growth, total unemployment, and the job contribution of new firms. The former is preferred in studies investigating a smaller sample of firms. Studies investigating large samples of firms, such as all firms in the private sector, also use the latter. Population refers to three types of firms: continuing firms (also called permanent firms or ongoing firms), i.e., firms existing throughout the studied period; new firms, i.e., one or several cohorts of new firms established during the studied period; or all firms, i.e., continuing firms as well as new firms established during the studied period.

Job contribution can be discussed in terms of gross job creation, i.e., total employment gains in studied units; gross job destruction, i.e., total employment losses in studied units; and net job creation, i.e., the difference between the two during the same time period (e.g., Davis and Haltiwanger 1999, Sect. 2.1). All identified studies measure net job creation, and when nothing else is stated this is what we refer to.

Net job creation is measured at different levels: firm, groups of firms (notably small, young, and industries), and at the aggregate level. This means that net job creation may differ across levels and across groups of firms; for instance, even though total employment may decrease, certain groups of firms, e.g., new ones, may experience net job growth.

Organic growth is supposed to have a larger effect on net employment than acquired growth. Some studies investigate single establishments to deal with this alleged “problem.” It is conceivable that single-establishment firms mostly grow organically. To remain a single establishment when acquiring other firms implies that acquired establishments have to be shut down and employment reallocated to the establishment of the acquiring company. This is not particularly likely; rather, one or several of the acquired establishments are likely to remain in operation.

However, acquired growth is important for reallocating employment and other resources to more productive uses. Hence, Gazelles growing externally may be of crucial importance for productivity growth. Klepper and Simons (2005), for instance, show that growing industries typically experience shakeouts in which the number of firms after some time falls sharply due to exits, mergers, and acquisitions. Hence, a natural pattern in the course of the evolution of an industry is that the number of firms is initially very large, but when the industry grows and matures the selection process rapidly reduces the number of firms. It therefore seems normal that Gazelles in mature industries grow through acquisitions of less efficient competitors. Klepper (2002) provides many interesting examples in this regard. The US automobile industry consisted of 271 firms in 1909. This number was down 60% by 1923, and by the mid-1960s, only four car manufacturers remained in business. The television industry shows a similar pattern.

The studies in our survey have been identified by searching the following databases: the American Economic Association’s electronic bibliography of economic literature (Econlit), Google Scholar, Journal Storage (JSTOR), Research Papers in Economics (RePEc), and Social Science Research Network (SSRN). We first searched for “Gazelle”, “high-growth firm”, “rapidly growing firm”, and similar words and phrases in titles, abstracts, keywords, and, when possible (Econlit, JSTOR, and RePEc), in the main text.Footnote 7 In total, there were thousands of hits. We browsed the hits and selected the papers that investigate the employment contribution of Gazelles, or high-growth firms, relative to one or both of the identified benchmarks during a particular time period. In what follows we will use the terms Gazelles and high-growth firms synonymously. The identified studies were then complemented by references found in the identified studies and studies we know of. We confined the survey to studies published after 1990, partly because we did not find that many studies before 1990 (earlier studies are surveyed in Storey 1994), partly because the quality of data has improved substantially in the last two decades. The primary purpose of some of the papers was not the study of the job contribution of Gazelles per se.

3 Results

In total we identified 20 studies in our search, which was a much smaller number than we had expected, especially given the importance of the issue.Footnote 8 The key explanation for the small number of studies is the lack of suitable data. A systematic analysis of the importance of Gazelles requires data on a large number of individual firms that can be followed over time, preferably covering all ages, sizes, and industries. By their very nature such analyses are time consuming and costly, especially for new and small firms. Appropriate data do not even exist in many countries (Schreyer 2000; Hoffman and Junge 2006). As mentioned, academic research on these issues is also of relatively recent vintage. To begin with, research addressed fundamental issues such as whether Gibrat’s law (that a firm’s growth rate is independent of its size) holds and the impact of the turnover and mobility of firms on employment and economic growth (see, e.g., Sutton 1997; Caves 1998; Lotti et al. 2003; Audretsch et al. 2004 for surveys). Table 1 summarizes the studies in chronological order based on Delmar et al.’s (2003) systematization, and we now offer a summarizing comment and evaluation on each of them.

Table 1 Summary of studies on Gazelles as job contributors

Birch and one of his critics, James Medoff, co-author of Brown et al. (1990), summarized what Birch and his protagonists agreed upon regarding the job contribution of small and large firms. About Gazelles they concluded (Birch and Medoff 1994) that a relatively small number of firms create a disproportionately large share of new jobs.Footnote 9 During the 1988–1992 period, 4% of the firms generated 70% of all new jobs among ongoing firms in the USA. These 4% accounted for about 60% of all new jobs in the whole economy during the same period. They were relatively small; in 1993, the average Gazelle firm employed 61 people. Gazelles were found in all industries and every industry had about the same proportion of rapidly growing firms.

Kirchhoff (1994) studies the job contribution of firms in the 1977 and 1978 cohorts of firms in the nonagricultural private sector of the US economy. The analysis is based on the Small Business Data Base (SBDB) and includes all firms established in 1977 and 1978. Multi-establishment firms and firms with more than 500 employees are excluded.Footnote 10 The remaining firms (95% of the original population) are followed throughout 1984. Adjustments are made to exclude firms growing rapidly as a result of mergers and acquisitions.Footnote 11 Firm growth is calculated as the percentage change of employment between the beginning and end of the period. The firms are ranked according to their employment growth rate, and the uppermost decile is classified as high-growth firms. Kirchhoff finds that a small number of high-growth firms create a disproportionate share of net jobs. Four percent of the new firms produce 75% of employment for the entire cohort during the first 6 years of life. The entire 1977–1978 cohort of firms made a net contribution of 3.4 million jobs in 1984 (20.6% of the 16.5 million net new jobs created between 1976 and 1984). Furthermore, they amounted to 4% of total employment in 1984. However, Kirchhoff (1994, p. 188) also finds that the total job contribution of the 1977–1978 cohort falls by 25% during their first 6 years of life compared with the initial number of employees at the time of their start-up, when jobs lost due to exits and job loss among survivors are deducted from employment growth among surviving firms.

Storey (1994, pp. 113–119) summarizes research on the role and functioning of small firms. Among other things he reports findings from 14 studies investigating the job contribution of rapidly growing firms.Footnote 12 One study concerns the USA and 13 studies concern the UK. Most studies look at manufacturing, but some examine services. Based on the survey Storey (p. 119) estimates that among studied firms “… approximately 4% of firms create approximately half the new jobs over a decade.” In follow-up studies of the characteristics of rapidly growing firms Storey (1996, 1997, 1998, 1999) investigates limited companies, or groups of companies, which in 1996 had achieved an annual compound growth in turnover of at least 30% in the last 4 years. The companies/groups of companies were not subsidiaries and also had a turnover of between 5 and 100 million pounds in 1996. The investigation was based on data from the ICC/OneSource UK Companies database. About 10% of the population fulfilled the criteria, hence they were denoted the “Ten Percenters.”Footnote 13 Young and small firms were overrepresented among the Ten Percenters. They were found in a diverse range of industries and the sectoral variations in the concentration of Ten Percenters were moderate. Storey does not, however, compare the job contribution of the Ten Percenters to that of other firms. Consequently, we do not include these reports in Table 1.

Birch et al. (1995) study the job contribution of different types of firms. They use Dun and Bradstreet data for 1990–1994 covering all size classes and industries in the USA. Despite the fact that Gazelles only represent 3% of the firm population, Birch et al. (1995) report that they account for all employment growth between 1990 and 1994.Footnote 14 In 1990, 82% of them employed fewer than 19 people and only 3.6% employed at least 100 people. However, the 3.6% of Gazelles that start from a base employment of at least 100 are, on the other hand, “spectacular” job creators. They account for more than half (53%) of the jobs created by Gazelles. Birch et al. (1995, p. 8) call them “Superstars.” Some of them were already Fortune 500 companies while others were heading rapidly in that direction.Footnote 15 Gazelles are found in all industries. In fact, the share of high-growth firms is about the same across sectors. Only 1.8% of all Gazelles are in high-technology industries.

Picot and Dupuy (1998) analyze the job contribution by size class in Canadian firms.Footnote 16 They employ a longitudinal data set covering all firms in the business sector for the years 1978–1992 (annual data). Irrespective of the measures of growth, small firms generate a disproportionate share of net jobs in the whole economy. The result is largely due to new firms. Excluding firm entry, the disparity between the small and large firm sector disappears. The job contribution is very unevenly distributed among growing firms. It is heavily concentrated to a few rapidly growing firms. Among continuing small firms (studied for the 1983–1986 period), 5% accounted for 43% of jobs gained. High-growth firms are found in all size classes and a number of large firms create a significant share of new employment. The correlation of a firm’s growth is low between adjacent periods, suggesting that past growth is a poor predictor of future growth.

Autio et al. (2000) study the impact of Finnish Gazelles. Gazelles are defined as independent single establishments increasing their sales by at least 50% during three consecutive years from 1994 to 1997. To qualify as a Gazelle a firm also requires a turnover of at least FIM 1 million by the end of the period.Footnote 17 All establishments meeting the Gazelle criteria are included. They find Gazelles to be important job contributors, especially a few Gazelles showing “ultrarapid” growth. Altogether the Gazelles increased their employment by more than 400% during the studied period. Most Gazelles were found in trade or in services. High-technology firms were not overrepresented among the Gazelles. No information is provided on the age of the Gazelles and the effect of firm size is not discussed.

Brüderl and Preisendörfer (2000) study the employment effects and growth of new firms in Bavaria, Germany. The data are part of the Munich Founder Study. The analysis is based on a stratified sample of firms established in 1985 and 1986, and interviews are made to examine whether there are any factors predisposing a firm to grow rapidly. Agricultural businesses, architects, crafts, lawyers, and physicians, making up about 20% of all newcomers, were not covered by the data and were therefore excluded. To qualify as rapidly growing, firms had to fulfill three criteria: survival until 1990, growth by at least 100% by the end of the period, and an employment increase by at least five employees during the same period. About one-fourth of the initial employment in the studied firms was lost due to closures. Job losses due to contraction of firms were small. The expansion of surviving firms more than compensated for the losses and total employment increased by 20% during the period. Fast-growing firms, constituting about 4% of the initial sample and about 6.5% of the surviving firms, were the main job contributors. By the end of the period, they had expanded their employment at the year of establishment by close to 400%. Their contribution to aggregate employment growth in the whole population of studied firms exceeded 150%.

Schreyer (2000) presents the result from six OECD country studies (France, Germany, Italy, the Netherlands, Spain, and Sweden) and from Quebec in Canada.Footnote 18 The data are not fully comparable and the applied methodology varies somewhat across countries. All studies investigate permanent firms employing 20 or more (10 in Spain and no threshold in Germany) people at the beginning (end in the Netherlands and in Sweden). They investigate different time periods; see Table 1. All studies include manufacturing. Italy, Spain, Germany, and Sweden also include services. Firm growth is measured as a composite index.Footnote 19 Gazelles are defined as the 5% (Quebec, Italy, the Netherlands, and Sweden) or 10% (France, Spain, and Germany) fastest growing firms in a reference population. The Netherlands, Spain, and Sweden define the reference population as all ongoing firms. For France, Quebec, and Germany the reference population is defined as all ongoing and growing firms, while in Italy it is defined as all firms in manufacturing having between 20 and 499 employees. Nevertheless, Schreyer (2000) maintains that a number of general findings emerge: High-growth firms account for a disproportionately large part of net job creation. Among high-growth firms, job creation rates of small firms exceed those of large ones. Large high-growth firms are substantial job creators in absolute terms. However, their rapid growth seems to be due to mergers and acquisitions. High-growth firms tend to be younger than the average firm, and high-growth firms are found in all industries and in all regions of the countries examined.

The Swedish study in Schreyer was carried out by Per Davidsson and Frédéric Delmar, who also conducted three of the other studies identified in our survey: Davidsson and Delmar (2003, 2006) and Delmar et al. (2003). These three studies basically use the same data and draw the same conclusions.Footnote 20 We therefore comment upon them jointly. The data include all commercially active firms in the nongovernment sector, independent as well as dependent (i.e., subsidiaries and branches), with 20 or more employees in November 1996. The data cover the period 1987–1996. The studies investigate the job contribution of high-growth firms in Sweden relating it to: the job contribution of non-high-growth firms, total job creation in the economy, the job contribution of new firms established in 1996, and total unemployment in the Swedish economy.Footnote 21 They discuss the employment contribution of high-growth firms distributed over industries, size classes, firm age, and type of governance (independent firm or belonging to a company group). High-growth firms are defined as the 10% of firms exhibiting the highest average annual increase in absolute employment among all firms in the population, i.e., both continuing and new firms.Footnote 22 The Swedish studies and a new study on Finnish data (Deschryvere 2008) differ from the other studies in our survey in one important respect: they separate organic from acquired growth.

In contrast to other studies, the job contribution of Gazelles is reported to be modest in the Swedish studies. This conclusion is based on relating the contribution of Gazelles to total job growth in the economy, to the job contribution of new firms established in 1996, and to total unemployment. However, there are some special circumstances affecting the results concerning the comparison to total job growth and to total unemployment. Foremost, during the time period, Sweden went through the worst depression in modern times, showing negative GDP growth for three consecutive years (1991–1993). In fact, aggregate employment decreased during the studied time period. As Davidsson and Delmar point out, this sharply influences their results. Also, the large government sector in Sweden, accounting for roughly one-third of total employment, lowers the potential job contribution of private firms.Footnote 23 This is particularly true for the service sector, from which private entrepreneurs were largely barred (see, e.g., Henrekson 2005; Johansson 2007 for a discussion).

As regards the comparison with the employment contribution of new firms, Davidsson and Delmar (2003) report new firms to generate approximately 40,000 new jobs and high-growth firms to generate 45,000 new jobs in 1996.Footnote 24 The result is in line with Halabisky et al. (2006), who report the net employment effect of churning (job gains in entries less job losses in exits) to exceed that of high-growth firms. The results may rather reflect the importance of new firm formation than the modest effect of high-growth firms.Footnote 25

Davidsson and Delmar (2003) and Delmar et al. (2003) do not base their conclusion about the modest contribution on a comparison with the non-high-growth firms in their population.Footnote 26 Relating the job contribution of Gazelles to that of other firms in the population examined, it is clear that high-growth firms are major job contributors; see Table 2. While the other firms lost more than 250,000 employees during the studied period, high-growth firms expanded by more than 180,000 employees (organic and acquired growth). Even if the analysis is restricted to organic growth, high-growth firms make a substantial job contribution. We therefore modify Davidsson and Delmar’s inference and conclude that high-growth firms are most important in the studied population.Footnote 27 This is in accordance with the findings in the other studies reaching the conclusion that Gazelles are important.Footnote 28

Table 2 Employment growth broken down by organic and acquired employment, 1987–1996

Firm size, firm age, and industry affiliation exhibit a significant relationship with firm growth. Studying total growth, large firms are overrepresented among high-growth firms. According to Davidsson and Delmar (2003, 2006) this is expected due to the choice of studying absolute employment growth. Firm age has a negative influence on rapid growth. While high-growth firms exist in all industries, service industries are overrepresented: the professional service sector has twice as high a representation among the top 10% firms compared with its share of the entire population.

Studying organic growth, the result for size is altered, while the results for age and industry are strengthened. Smaller firms grow organically to a greater extent, while larger firms mainly grow through mergers and acquisitions. High-growth company groups even exhibit negative growth in organic terms. Hence, independent firms loom larger when the focus is on organic growth. In high-growth firms younger than 5 years, 80% or more of employment growth is organic, while the corresponding share for high-growth firms older than 10 years is a mere 16%. High-growth firms are overrepresented in young and growing industries with a large inflow of new firms, especially in knowledge-intensive business-to-business services, education, and health care. About two-thirds of the high-growth firms were established during the period covered by the analysis.

Delmar et al. (2003, pp. 210–211) conclude: “In relation to previous research, these results largely support a view that organic growth is more associated with young and small firms, and that acquisition growth is more common among larger and older firms, and firms in stagnant or low-tech industries.” Moreover, they conclude that age, rather than size, determines rapid growth and, hence, that new firm formation and early growth of new firms are crucial for net employment growth, particularly in young and growing industries. Davidsson and Delmar (2006) write that the results indicate that renewal leads to organic growth as well as growth in the whole economy.Footnote 29

The purpose of Littunen and Tohmo (2003) is to investigate the factors involved in the start-up and first years of existence of firms that also experience rapid growth, e.g., characteristics and motives of the founding entrepreneur. They study a sample of Finnish metal-based manufacturing and business service firms founded in 1990. The firms are followed until 1997. To be classified as high growth, a firm has to more than double its turnover in real terms over the 1990–1997 period. In addition, its turnover has to be at least FIM 500,000 by 1997. In 1997, overall employment in the sample studied had increased. Employment of non-high-growth firms was reduced. Employment of high-growth firms increased not just to offset the decrease, but also to increase total employment (Table IV, p. 196). It is noteworthy that Finland was in a deep recession when the study was conducted.Footnote 30

Fritsch and Weyh (2006) study the employment trajectories in the 18 cohorts of start-ups founded from 1984 to 2002 in West Germany. The data are taken from the German Social Insurance Statistics. They cover all private-sector industries and include start-ups with at least one employee. Start-ups that have more than 20 employees in the initial year are excluded; a main motivation advanced for this decision is to avoid including firms that are not genuinely new but a result of the reorganization of large firms. The median start-up only has one employee. Employment growth is measured in absolute terms. A particular definition of Gazelles or high-growth firms is not used, but employment shares of the largest 1%, 5%, 10%, and 25% firms in different cohorts are reported. Fritsch and Weyh (2006) observe that the cohorts have a propensity to start by expanding employment. However, employment growth ceases quite soon; employment in a cohort stagnates or declines after 1 or 2 years. On average, total employment in a cohort has fallen below its initial level after 8 years. This has two causes. First, mortality of individual new firms is high, and after 10 years just about half of manufacturing firms in a cohort survive. The corresponding figure in services is even lower, about one-third. Second, most surviving firms do not grow.Footnote 31 Just a few firms do. On the other hand, these firms generate a significant number of new jobs. The authors conclude that a small fraction of firms dominate job creation. By the end of the period 1% of the firms accounted for about 44%, and 5% of the firms accounted for close to three-quarters of the jobs in the initial cohort. In 2002, employment in the 18 studied cohorts made up about one-fourth of total employment in the private-sector industries. Fritsch and Weyh (2006) also test Gibrat’s law for all studied cohorts for all years. Gibrat’s law is rejected for all cohorts and for all years.

Halabisky et al. (2006) study the job contribution of hyper- and strong-growth firms in Canada from 1985 to 1999. Data from the Longitudinal Employment Analysis Program (LEAP) and the Small Area File (SAF) are used. The data include all firms with employees, except in health, education, and government. Hence, firms with zero employees are excluded. The study focuses on firms in full-year operation in 1985 and still in business in 1989, i.e., continuing firms. A hypergrowth firm is defined as a firm growing by 150% in terms of employment in the 4-year period 1985 and 1989, and a strong-growth firm is growing 50–150% in the same period. The job contribution of hyper- and strong-growth firms and other types of firms is related to the overall job creation in the private sector and to the net employment effect of churning (job gains in entries less job losses in exits). The job contribution in different phases of the business cycle is of particular interest. The hyper- and strong-growth firms, representing 7% of the population, accounted for 56% of net job growth in the private sector (nearly 1 million out of 1.8 million new jobs). Small firms (fewer than 100 employees) dominated among rapid growers.Footnote 32 High-growth firms were resistant to recession and hardly lost any jobs as a group (employment stayed roughly constant) in the downturn that occurred during the period. In particular, high-growth firms contribute tremendously to job creation in provinces experiencing weak overall employment performance. Rapidly growing firms are found in all industries, and high-technology is not overrepresented. If anything, services are overrepresented. The net effect of churning over the period resulted in 1.3 million new jobs.Footnote 33 Hence, entry is found to be of crucial importance for total job growth.

Acs and Mueller (2008) study the employment effect of business dynamics in a regional context. They combine data from the Longitudinal Establishment and Enterprise Microdata (LEEM) with data from the Current Employment Statistics (CES) survey. Every US private-sector (nonfarm) business with employment is covered. The analysis is carried out for 320 US Metropolitan Statistical Areas (MSAs) and covers the 1990–2003 period. Growth is measured as percentage change of employment in the MSAs over a 3-year period. The start-up rates in the MSAs are used as a measure of business dynamics. The start-up rates in 1 year and each of the preceding 6 years are regressed on growth in order to analyze the long-term effect of business dynamics on employment. New firms are differentiated according to their size, measured as the number of employees in the year of entry. Gazelles are defined as new firms having 20–499 employees in the year of establishment and experiencing persistent employment growth over time. New firms have a strong effect on employment in the year of entry. The effect fades after 6 years. Only Gazelles located in large diversified metropolitan areas exhibit pronounced long-term job effects.

Acs et al. (2008) revisit some of the earlier conclusions of Birch’s work on rapidly growing firms with new and better data: The Business Information Tracking System (BITS) and the Corporate Research Board’s American Corporate Statistical Library (ACSL). Rapidly growing firms are referred to as high-impact firms, which are defined as enterprises (p. 4): “with sales at least doubling over the most recent 4-year period and which have an employment growth quantifier of two or greater over the same period.” The employment growth quantifier is, in turn, defined as the product of absolute and relative change in employment over a 4-year period. They look at four sets of questions: high-impact firms compared to non-high-impact firms, location by industry and region of high-impact firms, and what type of firms high impact firms are before and after they become high-impact firms. The data include all firms in all industries and cover the period from 1994 to 2006. The main period of analysis of the high-impact firms is 1998–2002. The 1994–1998 period is used to study the nature of high-impact firms before they become high-impact firms and the 2002–2006 period is used to study high-impact firms after they have become high-impact firms. High-impact firms exist in all industries and regions, accounting for 2–3% of all firms depending on industry and region. High-technology industries are not overrepresented. They are of all sizes and create almost all employment growth in the whole economy. High-impact firms with fewer than 500 employees and high-impact firms with more than 500 employees create about half of the new jobs each. The rate of high-impact firms that continues as high-impact firms is double in the 500+ size class compared with smaller high-impact firms. These are called super-Gazelles. Close to all job losses are due to non-high-impact firms with more than 500 employees. High-impact firms are not young; their average age is around 25 years. This is, however, considerably less than the average age of non-high-impact firms. A conclusion is that a diversified economy enhances the creation of high-impact firms since such firms can be found in all industries and since the growth of industries shift over time.

Deschryvere (2008) studies the job contribution of high-growth firms in Finland. He applies the definition proposed by OECD (2006) and defines high-growth firms as firms with average employment growth greater than 20% p.a. over a 3-year period, and with ten or more employees at the beginning of the period. The investigated period covers the 3-year period 2003–2006 and the analysis is based on the firm and establishment data from the Finnish Business Register. Continuing firms are analyzed. According to the definition, high-growth firms make up 5.4% of the total stock of firms with more than ten employees, corresponding to 750 firms. When the definition is based on organic growth the number of high-growth firms is reduced to 642, corresponding to 4.6% of the population. Most high-growth firms start out small, but medium-sized rapidly growing firms create most jobs. There is a negative relationship between initial firm size and organic growth. Firm age is not reported. During the studied period the Gazelles in total added 62,000 net jobs to the economy, whereof 65% (about 40,000 in absolute terms) through organic growth. High-growth firms generated about 90% of all net jobs created in the Finnish economy during the studied period. High-growth firms are represented in most industries.

4 Discussion and conclusions

It is apparent from our survey that the studies differ in their definitions of Gazelles, measures of growth, and time periods. They also differ regarding industries, firm sizes, firm ages, methods used, and geographical coverage. Sometimes this is a drawback since comparability may be impaired. However, in this case the large variation should be seen as an advantage, since the results regarding the importance of Gazelles turn out to be quite robust. Regardless of definition, method, time period, etc. some findings emerge.

Proposition 1 is supported. A few rapidly growing firms generate a disproportionately large share of all new net jobs compared with non-high-growth firms. This is a clear-cut result. All studies find Gazelles to generate a large share, all or more than all net jobs (in the case where employment shrinks in non-Gazelle firms taken as a whole). It is noteworthy that this is particularly pronounced in recessions when Gazelles continue to grow.

Proposition 2 is also supported. The results regarding age are unambiguous. All studies reporting on age find that Gazelles tend to be younger on average. Super-Gazelles are also relatively young.

As regards proposition 3 the results are ambiguous. Gazelles can be of all sizes, small firms are overrepresented but larger Gazelle firms are important job contributors in absolute terms, in particular a small subgroup of so-called Superstars or super-Gazelles. These are both large firms and major net job creators. It appears that newness is a more important factor than small size.

Proposition 4 is not supported. There is no evidence that Gazelles are overrepresented in high-technology industries. Gazelles exist in all industries. If anything, they appear to be overrepresented in services.

All studies but the ones based on Swedish data and the new Finnish study investigate total growth. Since organic growth is supposed to generate new employment to a larger extent than acquired growth, the conclusions may be altered when organic growth is studied. However, the Swedish and Finnish studies show that the conclusions are similar irrespective of whether organic or total growth is studied. Rather, the conclusions are more pronounced when organic growth is the object of study. It is also noteworthy that the growth of young and small firms is more organic compared with large and old firms. Hence, they make a larger contribution to net employment growth. We have little reason to surmise that the pattern in other similar countries would be different. Moreover, the studies focusing on single establishments—where acquired growth is likely to be insignificant—report similar results.Footnote 34

This survey of existing studies clearly shows that a small number of high-growth firms are particularly important for net job creation. Moreover, it is clear that Gazelles more often than not are young firms.Footnote 35 Nevertheless, this conclusion still rests on a fairly small number of studies. Additional studies would therefore be of great value. There are a number of ancillary aspects that could be dealt with concurrently such as the importance of spatial localization (e.g., Stam 2005) and whether other performance measures such as the growth in sales lead to similar conclusions (e.g., Morena and Casillas 2007). One should also use alternative econometric approaches, such as quantile regression techniques, to examine whether more accurate estimates can be obtained (e.g., Coad 2007). Yet another interesting expansion would be to study the economic significance of firms showing exceptional growth rates in more detail (e.g., Markman and Gartner 2002).

The results also provide additional perspectives on two discussions raised in several papers. The first discussion concerns the question of whether it is the entry of many new firms or the rapid growth of a few firms that generate employment growth, the so-called Mice versus Gazelles debate (cf. Davidsson and Delmar 2003, 2006). Our survey suggests that the two views are complementary. The studies in this survey indicate that employment in new firms is crucial for total employment growth and seems to be at least of equal importance as the net job contribution of continuing (Gazelle) firms. The positive employment effect of new firm cohorts tends to decline over time. Thus, continuous entry of new firms is required to achieve net job creation.Footnote 36 Parker et al. (2005) also report that only a small subset of the Gazelles show sustained growth. Hence, it seems plausible that a high inflow of new firms increases the likelihood to generate young Gazelles (with sustained growth), which tend to have a larger impact on aggregate employment than older Gazelles, since the former are more likely to grow organically.

The second discussion concerns the assertion that small businesses contribute disproportionately to job creation. The critics of this claim have pointed to the quality of data, the narrow emphasis on net job contribution, and the disregard of gross job flows and regression-to-the-mean effects (e.g., Haltiwanger and Krizan 1999).Footnote 37 An often overlooked part of the critique, and in our opinion perhaps the most important one, is that net employment growth has to be viewed in a broader perspective of creative destruction, where net employment growth entails considerable churning and restructuring in a dynamic process of firm entry, expansion, decline, and exit.Footnote 38 Hence, gross job flows are critical for net job growth, since they are part of, and a prerequisite for, the discovery procedure of new business opportunities that create jobs in the long run. It may therefore be misleading to narrowly focus on a particular piece of this process and claim that it alone contributes a disproportionately large share of net employment growth.

This is not to deny that some firms are more important than others in the process of creative destruction, in the same sense that some entrepreneurs are more important than others; cf. Acs’ (2008) discussion of high-impact entrepreneurship. However, a prerequisite for the growth of these firms is also that the process of creative destruction functions so that efficient new and expanding firms can attract resources from inefficient firms, resources that are released through contraction and exits. Without this dynamic reallocation the growth of firms will be hampered, irrespective of their inherent growth potential.Footnote 39 The policy implications are in line with the OECD’s (2007) recent assessment that the evidence of favorable policy impact is more clear-cut for macro/institutional policies than for various types of targeted micro policies.Footnote 40

Further support of our conclusions that different views are complementary in the two discussions is given by the increasing evidence that turbulence in itself, i.e., the entry and exit of firms, boosts job creation (e.g., Bartelsman et al. 2004, 2005; Brown et al. 2006; Birch 2006; Fogel et al. 2008; Caballero 2007). Turbulence is a natural effect of an accelerated search for new business opportunities and a rapid reallocation of resources from unsuccessful to successful firms, and when an industry evolves and becomes more mature it is natural that the market selection process reduces the number of firms, in some cases to a very small number (Klepper 2002). This implies that an employment-enhancing policy should aim at lowering barriers to new firm entry and firm exit to support an experimental process increasing the number of trials (new firms) from which potential Gazelles can be recruited, and not hindering the closure of failures.