Types of Entrepreneurship

Types of Entrepreneurship and Economic Growth DOI:10. 1093/acprof:oso/9780199596515. 003. 0004 Abstract and Keywords This chapter is an empirical exploration of types of entrepreneurship and their impact on economic growth in developing and transition countries. It relates indicators of entrepreneurship to average rates of economic growth in the period 2002–5. For this the chapter utilizes a dataset on entrepreneurship in thirty? six countries from the Global Enterprise Monitor (GEM), collected in 2002.It finds that indicators of young business activity have a significant impact on growth in high? income countries and transition countries, but not in developing countries. The chapter explains the lack of significant effects in developing countries by pointing to the lack of complementary physical and human capital and the scarcity of larger companies that can act as a training ground for SMEs. Keywords:  Ã‚  Ã‚  entrepreneurship,  growth? oriented entrepreneurship,  econ omic growth,  global entrepreneurship monitor 4. 1  IntroductionEntrepreneurship has long been considered a crucial mechanism of economic development (Schumpeter  1934; Landes  1998). However, empirical studies on the role of entrepreneurship in economic growth show mixed evidence (Stam  2008). This is not remarkable because there is much heterogeneity in both the kinds of entrepreneurship and the kinds of economic contexts in which economic growth takes place. Until now studies have not sufficiently accounted for this heterogeneity on the micro- and macro-level, which limits our insight into the contingent role of entrepreneurship in economic growth.Important questions in this respect are: ‘How does the role of entrepreneurship differ between high-income, transition, and medium-income countries? ’, and ‘What kinds of entrepreneurship are most crucial for economic growth? ’. The objective of this chapter is to provide insights into the role of dif ferent types of entrepreneurship in economic growth, and on how this role differs in poor and rich economies. In this chapter, we empirically investigate the effect of entrepreneurship on economic growth at the country-level.We use data from the Global Entrepreneurship Monitor (GEM), which provides comparative data on entrepreneurship from a wide range of countries. An important element of this chapter is that we compare the effects of entrepreneurial activity on economic growth in high-income countries, transition countries (China, Hungary, Poland, Russia, and Slovenia), and medium-income countries (Argentina, Brazil, Chile, India, Mexico, South Africa, and Thailand). This dataset also enables us to make a distinction between the effects of entrepreneurship in general and  (p. 9 )growth-oriented entrepreneurship in particular. We present empirical tests of the impact of entrepreneurial activity on GDP growth over a four-year period for a sample of 36 countries. Our empirical anal yses suggest that entrepreneurship does not have an effect on economic growth in medium-income countries, in contrast to transition and high-income countries where especially growth-oriented entrepreneurship seems to contribute strongly to macroeconomic growth. 4. 2  Entrepreneurship and economic development Development is a broad concept entailing the raising of human capabilities (Sen  1999).One of the central challenges in improving economic development is to increase the standards of living for individuals and growth of the economy as a whole. Even though economic growth in itself is a rather narrow target, it is probably one of the most important targets for development policies. It is also one of the measures that is most easy to access for analysts, and probably the best measure to make cross-national (Barro  1991; Sala-i-Martin  1997) and historical (Maddison  2001) analyses of the development of economies.Traditionally the economic output of a country is seen as a function of capital and labour inputs, combined with technical change (Solow  1957). Of course, conflicts and wars might interrupt this function (Sala-i-Martin  1997), but these are ‘just’ contingencies. The standard production function used shows that economic output (Y) is a function of the sum of labour and capital inputs, and the level of technological knowledge (i. e. productivity). This means that economic growth—the growth of economic output—is a function of the growth of labour and capital inputs and technological progress.In traditional models of economic growth investment in capital, labour, and technology is sufficient to realize economic growth. New models of economic growth see these investments as a necessary complement to entrepreneurship/innovation, but not as a sufficient explanation for economic growth in its own right (Nelson and Pack  1999). One could even argue that high rates of investment in human and physical capital are them selves stimulated by effective innovation, and cannot be maintained in the absence of innovation.Recent studies emphasize entrepreneurship as a driver of economic development and some authors include entrepreneurship as a fourth production factor in the macroeconomic production function (Audretsch and Keilbach  2004). Entrepreneurship is the factor that creates wealth by combining existing production factors in new ways. Entrepreneurs experiment with new combinations of which the outcomes are uncertain, but in order to make progress, many new variations have to be tried in order to find out which ones will  (p. 80 )  improve (economic) life (Rosenberg and Birdzell  1986).Other authors have argued that entrepreneurship will only unlock economic development if a proper institutional setting is in place (Baumol  1990; Boettke and Coyne  2003; Powell  2008). This institutional setting comprises informal as well as formal institutions (North  1990). An essential formal in stitution for welfare enhancing entrepreneurship is property rights. Insecure property rights have been an important constraint on the investments by entrepreneurs in transition countries, even more so than capital market constraints (Johnson, McMillan, and Woodruff  2000).A specific example regarding property rights is the fact that until 1988 private firms with more than seven workers were not even allowed to operate legally in China (Dorn  2008). One might say that the production factors capital, labour, technology, and entrepreneurship are the proximate causes of economic development, while institutions are a fundamental cause of economic development (Acemoglu, Johnson, and Robinson  2004). Next to productivity growth and technological change in established sectors, the development process in less advanced countries is largely about structural change (Gries and Naude  2010; Nelson and Pack1999; Rodrik  2007).It is a process in which an economy finds out—self-dis covers—what it can be good at producing, out of the many products that already exist. The role of entrepreneurs in developing countries does not equal innovation and R&D as commonly understood in advanced economies. Their role is to discover that a certain good, already well-established in world markets, can be produced at home at low cost (Hausmann and Rodrik  2003; Rodrik  2007).   Examples of this are the entrepreneurs that figured out that Bangladesh was good in the production of T-shirts, Colombia in cut flowers, India in software services, and Taiwan in bicycles and display technologies. Even if entrepreneurs cannot appropriate all these gains for themselves, their discoveries generate large social gains for their economies. Spurring entrepreneurs to invest in their home economy is said to be one of the most important aspects of stimulating growth in poor countries (Rodrik  2007). Investing refers here to innovation (e. g. mploying new technology, producing new products, searching for new markets) and expanding capacity. These investments trigger the combination of capital investment and technological change. In advanced capitalist economies, innovation and structural change take place through the combined efforts of small (independent inventors) and large innovative (organized R&D) firms, which complement each other in changing the economy (Nooteboom  1994; Baumol  2002). In developing countries the role of large firms is relatively small (Ghoshal, Hahn, and Moran  1999).In transition countries there are relatively many large organizations but these are largely in a process of restructuring and dismantling. This means that  (p. 81 )  small firms will be the prime movers in the process of structural change in developing and transition economies. We expect that the level of growth-oriented entrepreneurship in a country is a more relevant driver of economic growth than the mostly used indicators of entrepreneurship like self-employ ment and new firm formation.In contrast to rich countries, entrepreneurship in medium-income countries is mainly driven by necessity (Bosma et al. 2008). 2  Most entrepreneurs in these economies do not start a firm because they desire independence or because they want to increase their income as compared to being an employee, which are the dominant motives in rich countries. Most new businesses in medium-income countries are started out of necessity, in contrast to high-income countries, where entrepreneurship is most often opportunity-driven.This is reflected in the finding that in poor countries self-employed persons are less happy than employees, while the reverse is true in high-income countries (Blanchflower and Oswald  1998; Graham  2005). Entrepreneurs in medium-income countries most often start a business because they have no other way of earning a living. These entrepreneurs are not likely to be involved in a process of self-discovery; their actions are not likely to have an effect on the restructuring and diversification of the poor economies (Rodrik  2007). . 3  Data and research methods It is generally acknowledged that there are differences in the distribution of entrepreneurship across countries. Studies exploring differences in entrepreneurship across countries often focus on the incidence of new firm registration or self-employment, which may not be reliable indicators when applied to transition and developing countries with significant informal economies and fewer alternatives to self-employment.For these reasons we have used the Young Business (YB) indicator, defined as the percentage of adult population that is the owner/manager of a business that is less than 42 months old. Many studies have used the total entrepreneurial activity index, but that also includes the more speculative category of nascent entrepreneurs (individuals preparing a new business). In the current study we investigate whether the presence of growth-oriented en trepreneurs is a more important determinant of national economic growth than entrepreneurial activity in general.We will perform regression analyses with the YB  high-growth expectation  rate and the YB  medium-growth expectation  rate as independent variables and compare their impact on economic growth with the impact of the general YB index. The data and model used in this study are described below. We use a sample of 36 countries participating in the GEM in 2002. Data on six basic variables are used in our model: YB rate, YB medium-growth, YB  (p. 82 )  high-growth, growth of GDP, per capita income, and the growth competitiveness index (GCI). YB indexYB is defined as the percentage of adult population that is owner/manager of a business that is less than 42 months old. The YB high-(medium) growth expectation rate is defined as the percentage of adult population that is owner/manager of a business that is less than 42 months old,  and expects to employ 20 (six) emplo yees or more within five years  (YB6 and YB20). The YB medium-growth rate has some similarity to the entrepreneurship indicator used by Djankov et al. (2006), which includes owner-managers of a business with five or more employees. Data on the YB rate are taken from the GEM Adult Population Survey for 2002.Growth of GDP (? GDP) (Real) GDP growth rates are taken from the IMF World Economic Outlook database of the International Monetary Fund from September 2005. In equations (1) and (2) below variable ? GDPit  refers to the period 2002–5 (average annual growth) while the lagged GDP growth variable (? GDPi,t-1) refers to the period 1998–2001. Per capita income (GNIC) Most studies on GDP growth include the initial level of income in their analysis and find it to be significant (the conditional convergence effect, cf. Abramovitz  1986). Gross national income per capita 2001 is expressed in (thousands of) PPP dollars.These data are taken from the 2002 World Development Indicators database of the World Bank. Growth Competitiveness Index (GCI) In order to cover some aspects of the state of technology and institutions in a country (see Section  4. 2) we used the GCI for the year 2001 of the World Economic Forum (see McArthur and Sachs  2002). Given the low number of observations we are forced to use a combined index in our model. Even though there are huge problems in measuring technological capabilities and institutions (see Lall2001), the composite GCI is probably the best combined index available that covers these two factors simultaneously. p. 83 )  We investigate whether (growth-oriented) entrepreneurship may be considered as a determinant of economic growth, alongside the well-known determinants technology, institutions, and the macroeconomic environment, which are captured by the GCI. As both entrepreneurship and the factors underlying the GCI are assumed to be structural characteristics of an economy, we do not want to explain short-te rm economic growth but rather growth in the medium-term. Therefore we choose average annual growth over a period of four years (2002–5) as the dependent variable in this study.Following van Stel, Carree, and Thurik (2005), we use (the log of) initial income-level of countries to correct for catch-up effects, and lagged growth of GDP to correct for reversed causality effects, as additional control variables. 3 We allow for the possibility of different effects for high-income, transition, and medium-income countries. In addition we also test whether the effect of YB is different for transition countries. 4  YB rates may reflect different types of entrepreneurs in countries with different development levels, implying different impacts on growth.This is tested by defining separate YB variables for different groups of countries (high-income, transition, and medium-income countries). Our model is represented by equations (1) and (2). These equations are estimated separately by or dinary least squares. The expectation that growth-oriented YBs contribute more to national economic growth than YBs in general corresponds to b2  (c2) being larger than b1  (c1). In these equations sub-scripts t and t-1 loosely indicate that the independent variables are measured prior to the dependent variable.The exact years and periods for which the variables are measured can be found in the variable description above. ?GDPit=a+b1YBrichi,t? 1+c1YBtransitioni,t? 1+d1YBpoori,t? 1 +e? log(GNICi,t? 1)+f? GCIi,t? 1+g GDPi,t? 1+? it (1) ?GDPit=a+b2YB_high-growthrichi,t? 1+c2YB_high-growthtransitioni,t? 1+d2YB_high-growthpoori,t? 1+e? log(GNICi,t? 1)+f? GCIi,t? 1+g GDPi,t? 1+? it (2) To illustrate the data at hand, Table  4. 1  provides the YB rates and the YB medium- and high-growth rates in 2002 as well as the average annual growth rates of GDP over the period 2002–5.From Table  4. 1  and Figures  4. 1  and  4. 2  it can be seen that the ranking of countries in terms of YB or YB high-growth may be quite different. For instance, while China ranks fifth in terms of YB, it ranks first in terms of  (p. 84 ) Table 4. 1 Young business rates (2002) and GDP growth rates for 36 countrieshigh-growth YB. In contrast, Thailand ranks third in terms of YB, but only tenth in terms of high-growth YB. Figure 4. 1  Young business rates Figure 4. 2  Young business 20 rates Figure 4. 3  Correlation of young business rates and GDP growth rates Figure 4.   Correlation of high growth-oriented young business rates (20+) and GDP growth rates When we regress the rate of GDP growth on the YB rate and the YB20 rate, the YB20 rate reveals to have a stronger correlation with GDP growth (see Figures  4. 3  and  4. 4). (p. 85 )(p. 86 ) 4. 4  Entrepreneurship and national economic growth 4. 4. 1  Regression analyses The results of our empirical exercises are in Table  4. 2. Model I presents the regression results of the impact of the general YB i ndex (see equation (1)), while Models II and III show the results using the YB6 and YB20 rates as main independent variables (see equation (2)).The results presented in Table  4. 2  show that the impact of entrepreneurial activity is significantly positive for rich countries, but effectively zero for poor countries. The presence of growth-oriented entrepreneurs seems to be more important for achieving GDP growth than general entrepreneurship. Comparing the coefficients of the various YB rates, we see that the impact of YB6 is greater when compared to the impact of YB in general. Meanwhile the impact of YB20 is even greater, but not always statistically significant. Having more growth-oriented entrepreneurs seems to be particularly important in transition countries.Both the magnitude and the statistical significance of the estimated coefficient point to a stronger impact compared to high-income or medium-income countries. There are many reasons that could(p. 87 ) Table 4. 2 Regre ssion models average annual growth of GDP over the period 2002–5 (N=36)explain the importance of growth-oriented entrepreneurs in transition countries (Smallbone and Welter  2006). First, there are many entrepreneurial opportunities in formerly state-dominated sectors. Second, many highly qualified individuals lost their jobs at state-financed organizations (e. . universities, enterprises, government services). Third, there are many highly qualified (potential) entrepreneurs in these countries (especially in Eastern European countries), who do not face the opportunity costs of working for large public or private organizations. Fourth, those highly qualified (potential) entrepreneurs are also well connected to the power networks that were, and to a large extent still are, important in the political and economic arena of these countries, which takes away some barriers for high-growth firms in these countries.Summarizing, it may be argued that in transition economies high-grow th opportunities are more widely available and hence, a higher number of growth-oriented entrepreneurs willing to act on these opportunities may be particularly fruitful for achieving growth in these countries. However, we should be aware of the large diversity in the group of transition countries, which comprises countries like Russia and China, as well as Hungary and Slovenia. (p. 88 )  Our regression results should be interpreted with care as the analysis is based on a limited number of observations (36 countries).As a test of robustness we estimated the models leaving out one country at a time, i. e. we computed 36 auxiliary regressions, where each regression uses 35 observations (each time leaving one of the 36 countries out). Although t-values sometimes dropped a little, coefficients and t-values were generally in line with those reported in Table  4. 2. The country that matters the most for the results obtained in Table  4. 2  is China. This is not surprising as China combines high YB/YB6/YB20 rates with high GDP growth rates (see Table  4. 1). When leaving this country out of the sample, the coefficient (t-value) for the transition countries is 0. 2 (0. 5) for the YB rate, 1. 47 (1. 2) for the YB6 rate, and 1. 72 (1. 1) for the YB20 rate. The low t-values are in part due to the low number of observations. Note, however, that the coefficients are very similar to the full sample estimates reported in Table  4. 2. Furthermore, the Jarque–Bera test on the normality of disturbances is passed for all models reported in Table  4. 2, indicating that it is not necessary to remove individual country observations. Therefore we feel that our results are quite robust to the potential influence of outliers.Nevertheless, given the low number of observations, the results should only be seen as a first illustration of how the impact of different types of entrepreneurship may differ between groups of countries with different levels of development. 4 . 4. 2  Medium-income countries Within the groups of transition and developing economies there are substantial differences in entrepreneurship rates. Chile stands out because of a particularly high rate of growth-oriented entrepreneurship, while Mexico has a particularly low rate of growth-oriented entrepreneurship.In contrast to high-income countries, entrepreneurship in medium-income countries is mainly driven by necessity: self-employment is often the only occupational choice given a paucity of other sources of employment (necessity-based entrepreneurship; see Acs and Amoros  2008; Bosma et al. 2008). The actions of most of the entrepreneurs in medium-income countries are not likely to have an effect on the restructuring and diversification of the poor economies. This would be the whole story if the rates of growth-oriented entrepreneurship would also be marginal in these economies.This is only the case for Mexico. Next to Chile—where opportunity-driven entrepreneurshi p is dominant—Brazil, India, and Argentina perform quite well with respect to growth-oriented entrepreneurship. This means that there still is a substantial group of entrepreneurs in medium-income countries that might get involved in a process of self-discovery. The problem in practice is that in contrast to rich and transition economies, growth-oriented entrepreneurship is less likely to  (p. 89 )  be realized in developing economies, due to constraints on the provision of capital and (skilled) labour.An additional constraint in medium-income countries is that there are relatively few (foreign) large companies, which could act as a training ground for prospective growth-oriented entrepreneurs, and could open up distribution channels for new fledgling enterprises (Knorringa  1996). This is also reflected in the finding of Bosma, Stam, and Wennekers (2010) that the incidence of intrapreneurship (i. e. employees developing new business activities for their employer) is mu ch lower in medium-income countries than in high-income countries.In addition, one should make a distinction between large firms with productive (manufacturing) and resource extractive (mining, oil) activities here, as the former will be more useful for the development of entrepreneurship than the latter. 4. 4. 3  Transition countries New firms in transition countries not only displace obsolete incumbents but also fill in new markets, which were either non-existent or poorly populated in the past. Our study suggests that in transition countries, growth-oriented entrepreneurs make an important contribution to economic growth.They create new jobs with relatively high incomes which the small incumbent population of private firms cannot provide. This entrepreneurial growth process is facilitated by the relatively high levels of human capital in combination with relatively low opportunity costs of self-employment of the adult population. The high degree of environmental dynamism in the se countries—which is likely to positively affect the level of growth expectations and realizations of entrepreneurs in these countries—requires ambitious and well-connected entrepreneurs in order to translate these abundant opportunities in economic growth.There are considerable differences within the group of transition countries. Hsu (2005) shows that the role of these connections differs considerably between China and Russia: in China it was a tool which could be used to build enough trust to allow business transactions to succeed (‘capitalism without contracts’). In contrast, in Russia these connections devolved into corruption, and faded in importance for ordinary citizens. Without a way to build trust or extend networks, Russians retreated into defensive involution, and engaged in predatory behaviour against those outside their small circles of friends.Instead of capitalism without contracts, Russia suffered the depredations of ‘capitalists wi thout capitalism’. There are also substantial differences in entrepreneurship rates within the groups of transition economies. China stands out because of particularly high rates of growth-oriented entrepreneurship (cf. Hsu  2005). Even though the YB  (p. 90 )  rate is below the average of transition countries, the growth of self-employment has been enormous, not only in the richer coastal provinces, but also in rural areas (Mohapatra, Rozelle, and Goodhue  2007).Research by Djankov et al. (2006) also shows that entrepreneurs in China are more risk-taking and more committed to an entrepreneurial career than entrepreneurs in Russia. In addition, Russia has (and had: see Hsu  2005) a particularly low rate of entrepreneurship in general as well. The striking difference between entrepreneurship rates in China and Russia can be explained by their different paths from socialism to capitalism: gradualism and a shock therapy (see Burawoy  1996).In China the gradual transf ormation started with a policy of decollectivization (decentralization of property relations) in the late 1970s and the promotion of small-scale industry, with a focus on promoting independent entrepreneurship. Experimentation with new economic arrangements, for example privatization of small state-owned enterprises, has led to a favourable accumulation of productive capabilities in China. In contrast, Russia underwent a shock therapy in which the old communist regime was liquidated, with a focus on rapid privatization of the state sector.However, the Russian state failed to organize a market economy, which led to a coordination and entrepreneurial vacuum into which have stepped conglomerates, banks and mafia, siphoning off surplus from production to exchange (Burawoy  1996). 4. 5  Discussion of policy implications In this section we will briefly discuss the potential implications of our exploration of the relationship between types of entrepreneurship and economic growth for en trepreneurship policy and industrial/cluster policy in medium-income and transition countries. 5 4. 5. 1  Entrepreneurship policyOur empirical analyses suggest that entrepreneurship does not have an effect on economic growth in medium-income countries, in contrast to transition and high-income countries where both growth-oriented entrepreneurship and entrepreneurship in general seem to contribute strongly to macroeconomic growth. Does this mean that stimulating entrepreneurship in medium-income countries is bad policy? The least we can say is that stimulating entrepreneurship alone will be insufficient as it is likely to attract necessity entrepreneurs with low human capital levels who do not contribute to economic growth.The non-significant effect of entrepreneurship on economic growth in medium-income countries might point at a shortage of large firms in these countries. By exploitation of economies of scale and scope and by  (p. 91 )  adopting and diffusing technology devel oped elsewhere, large firms are important in transforming a developing economy into a developed economy (van Stel, Carree, and Thurik  2005). In these economies local workers are more productive working as wage employees than as entrepreneurs. Nevertheless stimulating growth-oriented entrepreneurship might be an additional element of transforming a developing economy into a developed one.Attracting investments by large (possibly foreign) firms, stimulating growth-oriented entrepreneurship, investing in labour and capital, and improving the institutional framework may be the recipe for growth here. On the one hand this is old news, in that it provides a plea for the traditional role of governments to invest in education and physical infrastructure, and to build and maintain a set of institutions that enable the development of the private sector (cf. Rosenberg and Birdzell  1986). On the other hand, the addition of growth-oriented entrepreneurship in development policy for medium- income and transition countries is a new element.One must be careful to target the right group of entrepreneurs though, i. e. governments should avoid that resources made available through government stimulation programmes are absorbed by necessity entrepreneurs with low human capital levels. 4. 5. 2  Industrial/cluster policy The focus of this chapter has been on the country-level, which disregards the sub-national level of analysis, and what is of particular relevance here, the level of regional clusters (regional concentrations of particular industries). These regional clusters have proved to be important drivers of economic development in, for example, Taiwan, India, and Brazil.These clusters are both driven by and drive growth-oriented entrepreneurship. Growth-oriented entrepreneurs that start to invest in a particular industry are needed in order to reach a critical mass that is needed to reach certain agglomeration economies (Braunerhjelm and Feldman  2006). If the build- up of capacity to this level of critical mass is not reached due to the lack of complementary investments, there might be a role to play for governments to overcome coordination failure, for example by providing investment guarantees for entrepreneurs (see Rodrik  2007).Such industrial policy is not about ‘picking winners’ or comprehensive planning, but encouraging experiments with new types of economic activity (Rodrik2007). Since it is impossible to judge winners and losers in advance, competent and growth-oriented entrepreneurs should be encouraged to try, success should be rewarded, and failure should not be coddled (Nelson and Pack  1999). These clusters do not have to be close to the technology frontier (as in advanced capitalist economies). The real policy implications arise from thinking carefully about the particular sources of advantage for a nascent cluster  (p. 2 )  and why that source might yield short-term complements with the potential to become l ong-term substitutes (Bresnahan, Gambardella, and Saxenian  2001). Cooperation of clusters in developing countries with existing richer economies is not ‘colonialist’. Take for example the linkages with the US. India and Taiwan are linked to the US (especially Silicon Valley) via outsourcing of software services and manufacturing (due to low labour costs), but also by a returning group of expatriates who have worked there, and who see the benefits of long distance collaboration (Saxenian  1999).There is a flow of people—the so-called Argonauts (Saxenian  2006)—and ideas back and forth between rich and emerging economies. Migrant workers tend to be among the most entrepreneurial in society. Governments of developing countries should not only look at these expatriate workers as a source of remittances. Given their entrepreneurialism, skills, and exposure to business in the developed world, as well as the desire of many of them to return home, they may be very important as a source of self-discovery in their country of origin (Rodrik  2007).In addition to developing the private sector, these return migrants may provide the new elite needed for building up a civil society. Only a fraction of the money spent on attracting FDI would be needed to target nationals abroad. This would attract more knowledgeable human capital and durable investments than most FDI will do. Once critical mass is reached within a regional cluster, it is likely to generate or attract growth-oriented entrepreneurs (e. g. Argonauts), who in turn stimulate further macroeconomic growth. 4. 5. 3  Limitations and further researchThe regression analyses in this chapter are of limited value: they have not only simplified the range6and (linear) effects of determinants for economic development, they have also dumbed down economic development to economic growth over a short-term (four-year) period. We know that sustaining growth is more difficult (and caused by diff erent factors) than igniting it (Rodrik  2007). This also connects to one of the other shortcomings: sustaining growth probably requires much more extensive institutional reform than can be properly taken into account in linear regression analyses.Next to better measures of institutions, future research should take into account samples with low-income countries and a larger number of medium-income and transition countries, and multiple years in order to achieve more robust empirical analyses. In addition, our data did not allow for testing the multiplicative effect of entrepreneurship, so we only analysed the additive effect. A larger number of cases would enable the inclusion of the more traditional indicators of capital and labour in the analyses, and allow for testing the multiplicative effect.