Getting the Facts Right on Born Globals

Policymakers in several countries have recently taken steps to promote the rapid export expansion of small- and medium-sized enterprises (SMEs). The goal of these policies has been to create successful export-intensive startups, which are often referred to as born globals. These measures are motivated by studies claiming that born global firms are disproportionately important for job creation and economic growth. Using detailed register data on the universe of Swedish manufacturing startups founded 1998–2011, we find that born globals are a very small group of firms whose long-run size and growth do not outperform other exporting firms. We also show that removing continuing firms and spinouts from the analysis is crucial for obtaining correct results. Thus, the notion that born globals are superior to firms that follow a more gradual internationalization process, a conclusion largely based on case studies and surveys, does not withstand scrutiny. Policymakers must therefore be aware that encouraging more born globals need not necessarily lead to large benefits for the overall economy, especially in terms of employment.


Introduction
The forces of globalization present both a market opportunity and a competitive challenge for entrepreneurs.
Policymakers have thus become interested in encouraging and accelerating startups' export activity in order to promote economic growth and boost job creation. This has led many countries to adopt policies that assist small and medium-sized enterprises (SMEs) and startups to expand into export markets. The goal of these policies has been to create successful export-intensive firms, which are often referred to as born globals.
The term born globals was first coined in a report by McKinsey (Rennie 1993) to describe enterprises that are able to quickly and successfully engage in foreign exports. Born globals are characterized by an ability to overcome the initial barriers that are associated with entry into foreign markets without first establishing a strong home market presence. The ability of these firms to circumvent a more lengthy process before taking steps to become internationally competitive has piqued the interest of many governments in both developed and developing countries. 1 For example, in 2015 the Swedish government published an export strategy that specifically emphasized the importance of encouraging born global firms. 2 One part states that: "There are many successful examples of Swedish companies that have been international from the start, but there could be even more of these so-called born globals", implying that born globals can be created through government support.
There are few studies, however, that rigorously investigate the performance advantages typically associated with born globals compared to other exporting firms.
Our study has two aims. First, we examine the long-run performance outcomes of born globals compared to other exporters in terms of employment, sales and value added. Second, we examine how removing continuing firms and spinouts affects the results. Given the widespread use of government policies that encourage exporting, we ask whether a born global export strategy leads to larger and faster-growing firms compared to a more gradual export strategy. We focus on the dynamics of firm size in this study because it is a central element of economic growth. Many firms in the data appear as though they are a genuinely new firm but have simply changed their organizational number or been spun out from an existing firm. We show that removing such firms from the analysis yields much more modest results.
Policymakers' interest in born globals is driven by several studies by academics and think tanks from the 1990s and onward. These studies claimed that born globals were growing in numbers throughout the world (Rennie 1993;UNCTAD 1993;OECD 1997). A study by Eurofound (2012), the European Union agency for the improvement of living and working conditions, claimed that as much as one fifth of all new firms in Europe are born globals and concluded from survey data that such businesses are characterized by higher employment growth. The OECD (2013) claimed that such firms played a pivotal role in mitigating the economic downturn of the great recession. If these claims are true it would be difficult not to concur with Eurofound's (2012, p. 3) conclusion: Bearing in mind that born globals' effect on the economy and labour market is not limited to a single country but, due to their international activities and the knock-on effects of these, become apparent at the European level, it is not only up to national, but also to EU policymakers to enhance their potential.
However, these strong statements and policy conclusions are often based on results from surveys and case studies, which focus on a highly selective group of successful born globals that may not be fully reflective of the 3 behavior of born globals in general. Cavusgil and Knight (2015) and Zander et al. (2015) have thus called for a more rigorous approach to study born globals in the fields of Strategic Management and International Business, namely using longitudinal data on all exporters in an economy in order to obtain a proper control group of other startups. In this study, we use detailed annual data covering all Swedish manufacturing startups founded 1998-2011. Since our data spans 1998-2014, this allows us to follow born globals and other exporting firms in the data for up to 17 years. As a small open economy with many export-oriented firms, Sweden is the ideal testing ground for evaluating the performance of born globals.
Our study contributes to a small and recent literature that uses register data to estimate the impact of a born global export strategy on long-run firm-level size. These studies focus on manufacturing, where detailed firmlevel export data that covers the universe of firms in a particular country is readily available. One the one hand, a born global export strategy may lead to firm growth by accessing consumers in as many markets as possible, providing the firm with a competitive edge . Indeed, the very definition of born globals rests on the assumption that these firms "derive a substantial proportion of their revenue from the sale of products in international markets" (Knight and Cavusgil 2004), which may lead to firm growth. On the other hand, however, international expansion is risky (Bonaccorsi 1992;Knight and Cavusgil 2004). Whether the progrowth or anti-growth mechanism dominates is thus an empirical question.
Several existing empirical studies of born globals' long-run performance focus on non-size measures such as sales per employee, profits over sales and labor productivity (Braunerhjelm and Halldin 2019), Tobin's q (Garcia-Garcia et al. 2017), or long-run export performance (Kuivalainen et al. 2007;Hashai 2011). We focus on firm size since we are interested in how a born global export strategy contributes to growth. Our work is most similar to two existing studies that include firm size in the analysis and employ large-scale firm-level longitudinal register data. Choquette et al. (2017) find that Danish born globals in manufacturing have higher sales and employment compared to other exporters. Braunerhjelm and Halldin (2019) mainly focus on non-size measures of firm performance, but they do include employment as a size measure, and find that Swedish born globals in the manufacturing sector have higher employment levels five years after establishment of a born global firm compared to other similarly-aged exporting firms.
Our work is distinct from Choquette et al. (2017) and Braunerhjelm and Halldin (2019) in many respects.
Our data covers the period 1998-2014, which allows us to study the performance of firms after the financial crisis. In contrast, Braunerhjelm and Halldin's analysis was limited to the 1998-2008 period. We study firm size comprehensively in terms of firm-level employees, sales and value added, while Braunerhjelm and Halldin's only size measure is firm-level employment. At the same time, we also apply best practices in terms of removing continuing firms and spinouts from the analysis, which was a major limitation in the Choquette et al. study. Our ability to properly identify startups yields much weaker results compared to Choquette et al.. In contrast to Braunerhjelm and Halldin, we find no statistically significant born global size advantage in terms of employment. We find no statistically significant firm size difference in terms of sales or value added either. Our results thus suggest that the firm size advantages related to a born global strategy previously found in the Swedish data are not robust.
We also include value added as a third proxy for firm size, which is novel to the born global literature that tends to focus on employment and sales. Value-added is a common measure of firm size in the economics literature and is the basis for GDP. Moreover, using value added as a metric for firm size circumvents the 4 problem that some firms are highly engaged in the import and export of intermediate inputs, which inflates sales but adds very little value in-house (Johnson and Noguera 2012).
The rest of the paper is organized as follows. We summarize the literature explaining the size and growth of born globals in Section 2. Data sources, the definition of born globals, descriptive statistics and our regression methodology are provided in Section 3. The regression results are presented and discussed in Section 4.
Conclusions follow in Section 5.

Explaining the Size and Growth of Born Globals
The notion that exporting firms in general may have beneficial effects on employment and growth is often motivated from a vast international economics literature, which has found that exporters tend to outperform their non-exporting peers in terms of productivity, employment, capital intensity, financial resources, and spending on R&D and investments (Bernard and Jensen 1999;Mayer and Ottaviano 2008). But these same studies also show that very few firms export, and even fewer firms are export-intensive. Born globals are distinct from other exporters by their ability to overcome the initial barriers associated with entry into foreign markets without first establishing a strong home market presence. Many studies have found born globals to be small, yet fast-growing exporters (Knight and Cavusgil 1996;Moen and Servais 2002;Knight 2015).
Both market-level and firm-level factors can explain the emergence of born globals. Knight and Cavusgil (1996) hypothesize that structural change and new technologies were important underlying factors that encouraged the development of born globals. In terms of firm-level factors, Hagen and Zucchella (2014) discuss the long-term growth of born global firms and reason that the "openness" of the founders and the early preparation for growth determine both the extent and speed of organizational learning, which in turn drives longrun growth. In the international economics literature, there is broad empirical support for the "selection into exporting" hypothesis, which asserts that only firms that are sufficiently productive self-select into exporting, and that the source of this productivity advantage predates their entry into export markets (Bernard and Wagner 1997;Jensen 1999, 2004).
Since the aim of this study is to provide evidence regarding the claim of born global firms' superior performance compared to startups with a more gradual export strategy, especially in the long run, the theoretical literature on the advantages and liabilities associated with a born global strategy is highly relevant. Starting with the advantages, Autio et al. (2000) argue that the early pursuit of international opportunity induces superior entrepreneurial behavior and leads to faster growth. Autio et al. suggest that young firms can more easily adapt its processes and structure to new markets, which allows young firms to enjoy the "learning advantages of newness." Young firms, for example, may be able to integrate knowledge about foreign markets quicker than  Bell et al. (2003) posit that a rapid internalization provides a first-mover advantage by locking in customers and more efficiently exploiting proprietary knowledge. Almor (2011) shows that Israeli born globals handled the Great Recession of 2008 better than other firms, due in part to their ability to tap customers in many markets. A born global export strategy may encourage firm growth by accessing consumers in as many markets as possible, providing the firm with a competitive edge .
In a related vein, the international economics literature has tested for the presence of "learning by exporting," whereby firms can improve their performance as a consequence of exporting due to a learning process. Aw et al. (2000) posit that firms can also become more productive after expanding to export markets by Electronic copy available at: https://ssrn.com/abstract=3172752 obtaining economies of scale in production. However, empirical support for the "learning by exporting" hypothesis is mixed. Wagner (2002), Andersson and Lööf (2009) and De Loecker (2013) find positive productivity effects using German, Swedish and Belgian data, respectively, including employment growth and wage increases in German firms. At the same time, Clerides et al. (1998) and Aw et al. (2000) find no such evidence of "learning by exporting" effects for Columbian, Mexican, Moroccan, Korean, and Taiwanese firms.
The learning mechanism is sensitive to export intensity, as Andersson and Lööf (2009) show.
The literature also describes many liabilities associated with a born global strategy. Although exporting may lead to higher revenues, it is seen by many firms as a risky strategy (Bonaccorsi 1992;Knight and Cavusgil 2004). Increased exposure to risk may lead to a survival bias among studies that employ case studies and surveys to examine successful born globals. Born globals' need for exporting knowledge from the very beginning, as well as new and more complex risk profiles due to foreign market exposure, can lead to negative effects on employment and growth (Luostarinen and Gabrielsson 2006). Knowledge and skills in navigating foreign markets is crucial, and firms with an aggressive international expansion may face a disadvantage of "foreignness" (Zaheer and Mosakowski 1997;Rugman and Verbeke 2007) and "newness" (Stinchcombe and March 1965;Zahra 2005) compared to foreign incumbents, which may adversely affect the performance of some firms. Young firms are particularly constrained with respect to management competencies and other resources, which can make rapid internationalization risky (Andersson and Wictor 2003;Oviatt and McDougall 2005;Sapienza et al. 2006;Carr et al. 2010). Sui and Baum (2014) find that born globals face increased liabilities of foreignness compared with other strategies, which makes them more prone to failure. Sleuwaegen and Onkelinx (2014) find that survival rates are lower for born globals, which appears to stem from their "liability of newness." Overall, there are arguments for both advantages and liabilities associated with a born global strategy, and it is ultimately an empirical question as to which effect dominates. Observable differences in size or other characteristics between born globals and other exporting firms at founding may be indicative of the selection mechanism. The learning by exporting hypothesis would suggest that differences will become more important over time, although this could also be driven by differences in initial unobservable characteristics, such as entrepreneurial ability, that take time before they are observed in firm performance.
A related factor to consider is that the relative ease of scalability, which is a common attribute among born globals (Kudina et al. 2008;Cannone and Ughetto 2014) and can imply that employment growth does not rise proportionately with output growth among these firms. The ease of scalability is often associated with the presence of economies of scale, whereby fixed production costs are a large component of total costs compared to the variable cost of production. If labor costs are primarily a fixed production cost, then employment growth will lag behind growth in sales and value added as firms' output expands. Therefore, studies evaluating the performance of born globals should look at output-based measures of size in addition to employment, such as value added.

Data sources
We use firm-level register data covering all limited liability companies for the years 1998-2014 in manufacturing (NACE Rev. 2 industries 10−33), obtained from Statistics Sweden. Firm accounting variables 6 such as sales, value added and the number of employees are collected by Statistics Sweden directly from firms' tax returns. We use Statistics Sweden's definition of value added, which is defined as sales minus input use.
We match the accounting data with detailed firm-level data on firms' exports, which is derived from Swedish customs records. In order to avoid erroneously classifying continuing firms that change their organizational number as new firms we use a system developed by Statistics Sweden. Firms are classified as continuing firms even if they change their organizational number as long as at least 50 percent of their workers continue to be employed at the same establishment. Firms are classified as a spinout if the employees that "moved" made up less than 50 percent of total workers at the old firm and at least 50 percent of the total at the new firm.
Instead of following Choquette et al. (2017) and comparing born globals with all other new firms, an approach that would entail the complex task of controlling for selection into exporting in the subsequent analysis, we follow Braunerhjelm and Halldin (2019) and opt to compare born globals with other exporting firms. Other exporting firms are also more comparable to born globals in terms of size, which makes them a more suitable control group. Moreover, using other exporters as the control group allows us to address our research question, which is whether a born global export strategy is beneficial in terms of firm size and firm growth compared to a more gradual export strategy. In order to maintain an identical sample regardless of outcome variable, we restrict the analysis to only those firm-year observations that report positive employment, sales and value added.

Defining born globals
The definition of born globals aims to capture a unique type of export firm with an accelerated export process, in contrast to the traditional internationalization process whereby firms build up a customer base in the domestic market and then gradually expand internationally (Bilkey and Tesar 1977;Johanson and Vahlne 1977;Cavusgil 1980). Research on born globals has mainly focused on documenting their characteristics and understanding the underlying trends that give rise to these types of firms (Moen and Servais 2002). These studies found that born globals are typically innovation-intensive (Andersson and Wictor 2003;Cavusgil and Knight, 2015), human capital-intensive (McDougall et al. , 2003Knight 2001;Melén and Nordman 2007), and are characterized by a production process that is easily scalable (Kudina et al. 2008;Cannone and Ughetto 2014) There is a lack of a harmonized definition of born globals in the literature, although many efforts have been made to define them, both qualitatively and quantitatively. The quantitative definition of born globals is usually defined along two dimensions: the degree of export intensity (exports as a share of total sales must exceed a certain value) and the age of the firm at which this export intensity criterion is met. 3 In the literature, numerous definitions have been applied (see, e.g., Braunerhjelm and Halldin 2019; Gabrielsson and Kirpalani 2004;Oviatt and McDougall 1994;Rennie 1993), such as export activity within 2-10 years and a minimum export share of total sales ranging between 20 to 80 percent. In a literature review by Bader and Mazzarol (2009), they found 12 different definitions across 126 studies, where Oviatt and McDougall's (1994) definition was the most referenced. They defined born global firms as an extreme version of "international new ventures" that are international from inception and derive "significant competitive advantage from the sales from multiple countries". 4 Born globals are startups by definition and are thus not established as a spinout of an existing firm. 7 This important distinction is an important difference compared to many international new ventures  that begin life as spinouts.
In this study, we define born globals as startups with at least 25 percent of their sales in exports within three years of founding since it is the most commonly used definition in the literature. This definition stems from Moen and Servais (2002) and Knight and Cavusgil (2004), and is used in studies of Swedish born globals by Nordman and Melén (2008), Melén and Nordman (2009) and Braunerhjelm and Halldin (2019). Choquette et al. (2017) also use this definition in their study of Danish born globals. As a robustness check we also use a 50 percent export threshold. Given that Sweden is a relatively small and open economy where many firms export, several startups also meet the more stringent 50 percent rule.

Descriptive statistics
The number of firms included in the analysis under various sample restrictions is given in Table 1. The base sample consists of 27,344 firms in the manufacturing sector that appear as a new firm between 1998 and 2011.
Once we remove spinouts and continuing firms we are left with 15,093 true startups that were born between 1998 and 2011. 3,496 of these firms engage in export for at least one year. Of these, only 560 qualify as born globals, as at least 25 percent of their sales are exports within three years of founding 5 . We also report the number of born globals using a 50 percent export threshold, which cuts the number of born globals in half in all cases. Table 1 about here We also restrict our sample to domestically-owned startups that have less than 20 employees at founding in order to further ensure that our analysis is focused on true startups. Once we impose these further restrictions, the sample is reduced to 3,322 exporters. Only 517 firms in this more restricted sample are classified as born globals using the 25 percent rule. The domestic-ownership and employee restrictions help to ensure that spinouts are not erroneously included as born globals despite implementing Statistics Sweden's approach to removing spinouts. The results in Table 1 emphasize that only 3.6 percent of manufacturing firms are born globals once spinouts are removed. Startups with a more gradual export strategy are far more common, exceeding the number of born globals by a factor of six. Moreover, removing spinouts and continuing firms from the sample reduced the total number of startups by 45 percent and the number of born globals by as much as 67 percent, while the other restrictions had little effect on the total number of born globals (merely reducing the share of born globals by 2.6 percentage points).
Descriptive statistics are presented in Table 2, where we report the sample averages for the firm size measures and control variables under different sample restrictions for six-year-old born globals and other exporting firms founded 1998-2008 in the manufacturing sector. Table 2 highlights the importance of removing spinouts and continuing firms from the analysis in order to avoid erroneous conclusions. When spinouts are removed the average number of employees among born globals falls by 95 percent, and average sales and value added falls even more. Firm size in terms of employees, value added and sales suggest that six-year-old Swedish strategy. Oviatt and McDougall (1994) use the term "global startups," and McAuley (1999) uses the term "instant exporters". Knight and Cavusgil (1996) were among the first in the academic literature to refer to such firms as born globals. 5 Firms whose exports reach 25 percent of sales in their first or second year, but not in later years, are thus counted as born globals. 8 manufacturing born globals are small and similar to other same-age exporting firms, averaging less than five employees, and value added and sales of SEK 3.4 and 10.1 million, respectively, once spinouts are removed. 6 The measures of firm size and capital stock decrease only marginally as more restrictions are applied to the sample despite that foreign-owned firms and spinouts tend to be larger than new firms. The share of collegeeducated employees remains stable across different sample restrictions. It is clear from Table 2 that on average the born global firms and the firms in the control group are fairly similar in terms of size. The most noteworthy difference is that born globals have a higher share of educated workers compared to other exporters. The average number of employees per firm among born globals, other exporting firms and non-exporting firms in our final sample is illustrated in Figure 1. The figure suggests that born globals employ roughly the same number of workers as other exporting firms regardless of age. It is only among the very small number of 16-year-old born globals that any discernible difference can be seen. Employment among non-exporting firms is much lower, suggesting that other exporting firms are a more relevant control group than non-exporting firms.
The bars in Figure 1 show that the number of firms in the sample decreases with firm age, which is due to a combination of exits and truncation due to the fact that we only observe firms until 2014, regardless of age.

Figure 1 about here
The average sales and value added among born globals, other exporters and non-exporting firms under our final sample is presented in Figure 2 and 3, respectively. We find that born globals' sales and value added is similar to other exporting firms in terms of sales, and slightly larger in terms of value added in later years, although the sample size in later years is very small.

Regression methodology
We test whether born global firms founded between 1998 and 2011 are larger than other exporting firms r years after birth. We thus perform a cross-section regression analysis at the firm-level using OLS, which takes the following form: where , = is the size proxy for firm i, r years after founding. We regress Equation (1)  which control for the fact that certain industries were more likely to grow than others during the 1998-2014 period.
Since we log the firm size measures, a positive point estimate for the born global indicator variable indicates that born globals are ( − 1) × 100 percent larger in size compared to other exporters. The null hypothesis is that there is no size difference in terms of employment, sales or value added compared to other exporting firms.
Our analysis differs from Braunerhjelm and Halldin (2019) in two main respects. First, Braunerhjelm and Halldin is based on older data, namely firms born during the years 1998−2003 with performance measured during 2003−2008. In contrast, we find smaller and less statistically significant born global size advantages in our longer and more recent sample of startups. Second, we focus on firm size as a measure of performance, whereas Braunerhjelm and Halldin focus mainly on non-size performance, such as sales per employee, profits over sales, and labor productivity.

Results and Discussion
We now present our regression results. We first present the regression results separately for each age group, an approach used by both Choquette et al. (2017) and Braunerhjelm and Halldin (2019). As a robustness check, we also employ a panel regression approach, similar to Choquette et al., and show the importance of correctly removing continuing firms and spinouts. Figure 4, 5 and 6 we present the results of the regression analysis by firm age. All specifications include year fixed effects, 2-digit NACE Rev. 2 industry fixed effects and firm size at age zero. The figures are based on regression Equation (1), which employs the fully-restricted sample of all exporters in Table 1 The first vertical bar of Figure 4 illustrates the regression results when the dependent variable is logged employment the year after the firm is born (age = 1). We report the regressions for each age group in our data up to 12 years after birth. The full set of regression results up to 16 years are reported in Table A1 and A2 in the Appendix. We perform these regressions by firm age in order to study the relative performance of born globals in both the short and the long run. The gradient of these point estimates over firms' lifespans is also informative regarding the relative growth in employment, sales and value added for born globals versus other exporting firms in the long run. The 95 percent confidence intervals spans zero in all specifications, which implies that the point estimate for the indicator variable is statistically insignificant at the five percent level across all age cohorts. The gradient of the point estimates suggests that employment growth among born globals is very similar to other exporters, which corroborates our descriptive findings in Figure 1.   Figure 6 illustrates the regression results by firm age when logged value added is the dependent variable. We find that the estimate for the indicator variable is statistically different from zero only when firms are between nine and ten years old, whereas the point estimates are statistically insignificant for all other firm ages.

Figure 6 about here
In Figure A1 in the Appendix we check whether our results are robust to using a more export-intensive definition of born globals. In this robustness check we define born globals using a 50 percent-three-year rule, which is reasonable given the fact that Sweden is a relatively small economy. Approximately 200 exporters meet this stricter definition of born globals. The regression results by firm age, using logged employment as the size measure, are illustrated in Panel A. We once again find no statistically significant born global employment advantage compared to other exporting firms among any age cohorts. The results using logged sales are illustrated in Panel B, where we find a statistically significant size advantage for born globals that are nine years old. Finally, the regression results using logged value added are illustrated in Panel C, where we find a statistically significant size advantage among nine-year-old born globals. Overall, we find that our results are robust to using this stricter definition of born globals.

The importance of removing continuing firms and spinouts
In Table 3 we perform a panel regression, which combines data on firms of all ages. We include industry and year fixed effects as in Equation (1), and we also include firm-age fixed effects to account for the fact that firm size may differ by age irrespective of export strategy. This exercise allows us to compare our results to the panel approach used by Choquette et al. (2017), who attempt to remove continuing firms and spinouts by dropping the top one percent of firms by employment the first year they appear in the data. Their cutoff corresponds to firms with between 50 and 70 employees in their first year. In the odd-numbered columns of Table 3 we follow Choquette et al. by dropping firms with more than 60 employees in the first year. In the even-numbered columns we use our more sophisticated approach to removing continuing firms and spinouts, which corresponds to the same restrictions imposed on the regressions presented in Figure 4, 5 and 6. Using Choquette et al.'s approach to removing continuing firms and spinouts yields a statistically significant point statistic for the born global indicator variable for all outcomes. In contrast, using our approach to removing continuing firms and spinouts leads to an insignificant point statistic for employment and value added, while the relative sales advantage of born globals compared to other exporters drops from 82 percent to 10 percent, with statistical significance dropping from the one percent to the five percent level. 7 Table 3 about here The point estimates using the panel specification in the even-numbered columns of Table 3 are similar to the average point estimates across all ages in our earlier analysis, and the main difference is the standard errors, which are smaller in the panel specification. Standard errors in a panel setting with a limited time dimension tend to be underestimated (Bertrand et al. 2004), which suggests that our original cross-section specification estimating the effects separately for each age group provides a more conservative and trustworthy estimate of statistical significance.

11
We also show the importance of removing continuing firms and spinouts for our earlier analysis in Table 4, where we focus on 6-year-old firms for illustrative purposes. Panel A, B and C in Table 4 present the results using logged employees, sales and value added, respectively, as measures of firm size. In Column (1) we estimate the size premium associated with born globals without any sample restrictions. In Column (2) we estimate the size premium of born globals after removing spinouts so that the sample includes only true startups.
In Column (3) we present the regression results after removing firms that are foreign-owned in order to ensure that our results are not driven by foreign-owned affiliates. In Column (4) and (5) Table 4 presents the results for employment. In Column (1), prior to removing spinouts, we find a large and statistically significant employment premium associated with born globals. However, the premium vanishes once we remove continuing firms and spinouts from the analysis in Column (2). This non-significant result persists across Column (3), (4) and (5) as we remove foreign-owned firms and firms with unusually high employment in their first year. In Panel B of Table 4 we find a positive and statistically significant size premium in terms of sales before removing spinouts. However, we again find no statistically significant born global size advantage in terms of sales once continuing firms and spinouts are removed and the full set of sample restrictions is imposed. Similarly, in Panel C of Table 4 we find no evidence of a statistically significant size premium in terms of value added once spinouts and continuing firms are removed. Firm size the first year after founding, which we include as a control variable, is positive and significant at the one percent level across all specifications.

Discussion
We draw two main conclusions from our results. First, we find no evidence of a born global size premium in terms of employment, and weak evidence in terms of sales and value added. In the regressions by age, reported in Figure 5 and 6, we find that born globals display a size advantage only among firms aged nine or ten. Second, we find much smaller sales premia compared to Choquette et al. (2017). Choquette et al. find that born globals are between 96 percent and 182 percent larger than other exporting firms in terms of sales. 8 We thus fail to detect the quantitatively large born global size advantages found by Braunerhjelm and Halldin (2019) and Choquette et al. (2017). Cavusgil and Knight (2015) and Zander et al. (2015) have called for more rigor in the measurement of born globals' performance. Our study shows that the methods and assumptions used to identify born globals and a suitable control group are crucial when studying born globals' performance in the context of register data. In particular, our results highlight the importance of correctly identifying startups based on where the workers were previously employed. We also show that non-exporters are too different from exporters in terms of size in order to be included in the control group. Our inability to detect a born global size premium is at odds with the few 12 other studies that examine size dynamics in longitudinal register data, which suggests that a connection between firm growth and a born global export strategy is by no means a settled issue.
Our results also have some implications that might further the theory on born globals. There are two possible theoretical interpretations of our results. One interpretation is that the costs of a born global strategy cancel out the advantages in terms of reaching more customers abroad. In this case, one may conjecture that born globals could potentially be more successful if such costs could be avoided. Another possible interpretation of our results is that both the benefits and costs of a born global strategy versus a more gradual export strategy are negligible in terms of their impact on firm size. Further study with more detailed data would be required in order to determine whether or not there are particular costs associated with the born global export strategy that negatively affect long-run performance, such as "foreignness" (Zaheer and Mosakowski 1997;Rugman and Verbeke 2007) or "newness" (Stinchcombe and March 1965;Zahra 2005). Lastly, the weakly positive result for sales and value added may be due to the presence of economies of scale in production, whereby export growth leads to higher output among born globals but does not result in a corresponding increase in employment in these firms.

Conclusion
Promoting the emergence and growth of born globals is seen in many circles as a desirable policy goal. Such advice is typically motivated by reference to studies claiming that born globals are disproportionately important for job creation and economic growth. As a result, such firms have become the target of policy interventions in many countries. We have analyzed whether born globals lead to higher employment, sales and value added in the long run. We study this question using detailed firm-level data on the universe of Swedish manufacturing firms founded between 1998 and 2011, which allows us to follow firms for a period of up to 17 years.
Overall, our results do not suggest any robust evidence of a size or growth advantage associated with born globals in the Swedish manufacturing sector. Our results stand at odds with other studies using register data to study the performance outcomes of born globals. We show that the measured performance advantage of born globals hinges critically on restricting the sample to true startups, and excluding spinouts. The evidence presented here suggests that there is no clear advantage in terms of long-run employment, sales or value added associated with a born global strategy. This could be driven by the fact that the costs and risks of a born global strategy cancel out any of the benefits associated with reaching more foreign markets, or that the costs and benefits are both too small to detect.
Our results also suggest that a born global export strategy is practiced by a small number of Swedish manufacturing firms; a mere 3.6 percent of all new manufacturing firms founded 1998-2011 were born globals.
Policymakers must therefore be aware that encouraging more born globals need not necessarily lead to large benefits for the overall economy, especially in terms of employment. Our study also holds a valuable lesson for entrepreneurs, namely that adopting an aggressive global export strategy from the start, rather than entering export markets more gradually, does not necessarily lead to higher firm growth.
Our study is subject to certain limitations that deserve mention. First, our analysis is limited to manufacturing due to the availability of data on trade in goods, but many born globals are active in the service sector. Future research into born globals in the service sector would thus be highly relevant. Second, the nature of the register data limits our ability to study the particular management strategies used by each firm. Such studies may shed light on the underlying mechanisms that lead to success or failure. Large-scale detailed survey data based on questionnaires that combine more management strategy detail with large sample sizes may thus be a fruitful 13 avenue for future research. Third, our study does not address the aspect of firm exit, which affects the composition of the sample over time. We thus leave a study incorporating exit into studies of long-run performance for future research.   (227) Notes: Sample restriction criteria use information for the year that the firm is founded.  (3), (4) and (5) (123 firms), therefore the descriptive statistics are identical.  (5) in Table 2 (Continuing firms and spinouts removed, domestically-owned and less than 20 employees in year of founding). Age fixed effects, year fixed effects and 2digit NACE Rev. 2 fixed effects included in all specifications. Odd-numbered columns use Choquette approach to removing spinouts and continuing firms, even-numbered columns use the full set of restrictions used in Table  3, Column (5). Robust standard errors in parentheses, clustered at firm level in all specifications. *** p < 0.01, ** p < 0.05, * p < 0.1.  (5) in Table 2 (Continuing firms and spinouts removed, domestically-owned and less than 20 employees in year of founding), plus restricted to six-year-old firms. Age fixed effects and 2-digit NACE Rev. 2 fixed effects included in all specifications. Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1 Figure A1 Difference in employment, sales and value added by firm age, born globals defined using 50 percent three-year rule.
Panel  Notes: Same sample restrictions as Column (5) in Table 2 (Continuing firms and spinouts removed, domestically-owned and less than 20 employees in year of founding).
Year fixed effects and 2-digit NACE Rev. 2 fixed effects included in all specifications. Constant term included but not reported. Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.01 Electronic copy available at: https://ssrn.com/abstract=3172752 29 Notes: Same sample restrictions as Column (5) in Table 2 (Continuing firms and spinouts removed, domestically-owned and less than 20 employees in year of founding).
Year fixed effects and 2-digit NACE Rev. 2 fixed effects included in all specifications. Constant term included but not reported. Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.01