Scholarly article on topic 'Regional determinants of firm entry in a developing country'

Regional determinants of firm entry in a developing country Academic research paper on "Social and economic geography"

Share paper
Academic journal
Papers in Regional Science
OECD Field of science

Academic research paper on topic "Regional determinants of firm entry in a developing country"


Regional determinants of firm entry in a developing country*

Carla Daniela Calá1, Miguel Manjón-Antolín2, Josep-Maria Arauzo-Carod2'3

1 Facultad de Ciencias Económicas y Sociales (Universidad Nacional de Mar del Plata), Funes 3250; 7600 - Mar del Plata, Argentina (e-mail:

2 CREIP and Quantitative Urban and Regional Economics (QURE), Department of Economics (Universitat Rovira i Virgili), Av. Universitat, 1; 43204 - Reus, Catalonia, Spain (e-mail:

3 Institut d'Economia de Barcelona (IEB), Av. Diagonal, 690; 08034 - Barcelona, Catalonia, Spain (e-mail:

Received: 22 October 2013 / Accepted: 21 April 2014

Abstract. We analyse the determinants of firm entry in developing countries using Argentina as an illustrative case. Our main finding is that although most of the regional determinants used in previous studies analysing developed countries are also relevant here, there is a need for additional explanatory variables that proxy for the specificities of developing economies (e.g., poverty, informal economy and idle capacity). We also find evidence of a core-periphery pattern in the spatial structure of entry that seems to be mostly driven by differences in agglomeration economies. Since regional policies aiming to attract new firms are largely based on evidence from developed countries, our results raise doubts about the usefulness of such policies when applied to developing economies.

JEL classification: R12, R30, C33

Key words: Firm entry, Argentina, count data models

1 Introduction

The entry of new firms, while varying considerably across regions (Fritsch and Mueller 2004; Audretsch and Keilbach 2005), contributes to the growth and welfare of nations (Powell 2008). Moreover, the creation of new firms is an important mechanism for economic development that may help to reduce inequalities between poor and rich countries (Acs and Amoros 2008; Acs

* We are grateful to A. Gennero, E. Sarmento and participants at the 12th EUNIP Conference (Reus), the XLV Reunión Anual de la Asociación Argentina de Economía Política (Buenos Aires), the Demography of Firms and Industries Workshop (Créteil) and the 9th World Congress of Regional Science Association International (Timisoara) for helpful comments. Two anonymous referees and the editor also provided useful suggestions. We also thank the Employment and Business Dynamics Observatory from Argentina for kindly providing the dataset. This research was partially funded by CICYT's SEJ2010-19088/EC0N and SEJ2010-16934/EC0N, the 'Xarxa de Referencia d'R+D+I en Economia i Polítiques Públiques' and the SGR Programme (2009-SGR-322) of the Catalan Government. The second author further acknowledges the hospitality of the Center for Economic Studies at KU Leuven. The usual disclaimer applies.

© 2014 The Author(s). Papers in Regional Science © 2014 RSAI

et al. 2011; Naude 2011). However, there is very limited evidence on what determines firm entry in developing countries and on whether these determinants differ from the ones typically found in developed countries.1 This paper aims to contribute to this literature by analysing the determinants of firm entry in the Argentinean provinces during the period 2003 to 2008.2

In particular, we analyse annual provincial data on the number of new manufacturing firms with employees registered with social security to find that most variables that typically determine entry in developed countries (such as the evolution of economic activity, population density and industrial structure) are of similar importance here. However, we also find that some explanatory factors that are never considered when developed countries are studied (such as the extent of poverty, the size of the informal economy and the existence of idle capacity) turn out to be statistically significant. Lastly, we show that the spatial structure of entry fits quite well a core-periphery pattern in which firms in central areas benefit from a number of advantages (e.g., better access to markets, more skilled workers, and more and better external services). In the Argentinean case, this centre-periphery structure seems to result in differences in the impact of the agglomeration economies but not in the impact of idle capacity.

Of the developing countries, Argentina has a number of features that are worth noting. First, it is a country with important regional differences in terms of wages, labour skills, growth rates and natural resources. Second, firms and people are highly concentrated around the main cities, particularly the capital. Third, Argentina covers a vast territory that is organised in large administrative units. Interestingly, many other developing countries (e.g., South Africa, Brazil, Russia, Mexico and Vietnam) share these features to some extent. This means that although it may not be possible to generalise our results to all developing countries, they are likely to hold for a number of them.3

With this in mind, our results suggest that entry-promoting policies in developed countries cannot be automatically transposed to developing countries. Rather, the design of such policies should be based on studies that take into account the specificities of the data (e.g., certain variables may not be available and others may be defined in a non-standard way, as pointed out by Thompson 2010) and the institutional setting (e.g., macroeconomic instability and financial crises, as argued by Caballero and Hammour 2000). There is, hence, a risk that regional policies aiming to attract new firms in developing countries will fail to accomplish their goals if they are based only on evidence from developed countries.

Our empirical strategy is similar to that of Fritsch et al. (2006) in their study of the determinants of firm survival in East and West Germany. This means that we do not make any prior assumptions about the existence of differences in the determinants of entry between developed and developing countries. Rather, we will argue that these differences may exist and (indirectly) test this hypothesis by comparing results from Argentina with those typically found in studies on developed countries. In the case of Fritsch et al. (2006), they find that only a few of the factors that have a statistically significant effect on survival in West Germany are also

1 Evidence from developing countries can be found in Lay (2003) and Wang (2006) for Taiwan and Gunalp and Cilasun (2006) and Ozturk and Kilic (2012) for Turkey, all of whom used industry level data; see also Naude et al. (2008) for South Africa and Santarelli and Tran (2012) for Vietnam, who used regional level data.

2 Previous studies of firm entry in Argentina are merely descriptive (Bartelsman et al. 2004; MTEYSS 2007; Katzand Bernat 2011; and Cala and Rotondo 2012 - the only one to adopt a regional perspective). Regression analyses can be found in Castillo et al. (2002) and Gennero et al. (2004), but the former studies the rates of employment creation and destruction using firm-level data and the latter new business ideas using regional-level data.

3 The size of the administrative units, as well as the degree of heterogeneity and urban concentration, are all considerably smaller in developed countries. To illustrate, Argentina's surface area is roughly four times that of France (the largest EU country) and the smallest province (Tierra del Fuego) is roughly two-thirds the surface area of Belgium. Likewise, while in the US 40 per cent of employment is located in counties constituting around 15 per cent of the land area (Scott and Storper 2007), in Argentina more than 65 per cent of the employment is located in a similar share of the land area. This kind of 'macrocephalic' urban systems, consisting of a few abnormally large cities, is typical of developing countries (Lipton 1977).

statistically significant in East Germany.4 They then interpret this result as evidence that the survival of new businesses in East Germany is subject to more erratic influences than in West Germany and associate this to distortions in the market structure and institutional setting. However, an alternative explanation might simply be that survival depends on factors that are not included in their model specification. This criticism may also apply to our study, since the omission of relevant variables might alter our conclusions. Still, it is worth noting that our set of explanatory variables is fairly comprehensive and that our estimates are largely robust across different variable definitions (e.g., demand for goods, urbanisation economies and poverty) and model specifications (Poisson and negative binomial fixed and random effects models, with and without including centre-periphery differences in the impact of the agglomeration economies).

The rest of the paper is organised as follows. Section 2 reviews the empirical literature on firm entry in developed and developing countries. It also discusses the empirical strategy. Section 3 describes the data set. Section 4 discusses the econometric model and the main results. Section 5 summarises the main conclusions.

2 Regional determinants of entry: An overview

2.1 Evidence from developed countries

A number of studies have shown substantial differences in regional entry rates.5 Also, most of the observed regional differences in entry rates arise from differences in regional characteristics (Davidsson and Wiklund 1997; Fritsch and Schmude 2006). However, this regional variation in start-up rates is consistent with different (and often competing) theoretical frameworks (Spilling 1996). This probably explains that most empirical studies tend to use econometric specifications that are derived ad hoc (Arauzo-Carod et al. 2010). In particular, following Bosma et al. (2008), we can group region-specific determinants of firm entry into three categories: (i) demand for goods and supply of factors; (ii) agglomeration effects; and (iii) cultural attitudes and policies towards entrepreneurship.

First, proxies for demand include variables that affect firm's profits, such as the size of local markets (typically using population measures) and consumers' purchasing power (measured by income, (un)employment and output measures such as regional GDP). All these variables can appear in the models in levels and/or in growth rates. As for the supply of factors, the focus is on labour and capital. Labour refers to the number of people endowed with the ability to start new firms, usually proxied by the composition of the labour force (age, gender, ethnic and geographical origin, etc.) and human capital characteristics (education, skills, etc.). Also, wages are the usual proxy for the price of this factor. Capital refers to infrastructures (e.g., accessibility) and financial resources (both in terms of the extent of financing, for example, bank loans, and the constraints that may exist to access credit, particularly on SMEs). In addition, it is common to consider proxies for the industrial structure such as the weight of SMEs, the number of incumbents and the number of exits (lagged one or two periods to avoid endogeneity concerns).

4 See also Ghani et al. (2014) for an analogous result when comparing the effects of incumbents' employment on the spatial distribution of entrepreneurship in India and the US.

5 As far as we know, these include Audretsch and Fritsch (1994) and Fritsch and Falck (2007) in Germany; Davidsson et al. (1994) in Sweden; Garofoli (1994), Carree et al. (2008) and Santarelli et al. (2009) in Italy; Guesnier (1994) in France; Keeble and Walker (1994) and Fotopoulos and Spence (2001) in the UK; Hart and Gudgin (1994) in Ireland; Reynolds (1994). Campbell (1996), Rigby and Essletzbichler (2000), Armington and Acs (2002), Sutaria and Hicks (2004) in the US; Spilling (1996) in Norway; Fotopoulos and Spence (1999) in Greece; Kangasharju (2000) in Finland; Arauzo-Carod et al. (2008) in Spain; and Tamasy and Le Heron (2008) in New Zealand.

Notice that the definition of the supply and demand categories is not self-excluding, for some variables may affect both supply and demand. Higher real wages, for example, mean more purchasing power but also higher labour costs and higher opportunity costs for self-employment. Similarly, unemployment can push individuals to start their own business. However, it may also reflect the poor economic situation of the region.

Second, having other firms close by may increase market opportunities and firms' efficiency. However, there is no general agreement on what is the ultimate driver of agglomeration. While some claim that it is the location of firms operating in similar industries (i.e., localisation economies), others argue that it is the location of firms operating in different industries (i.e., urbanisation economies). Whatever the case may be, it is important to stress that there are potential diseconomies in the agglomeration process. Congestion and the rise of input prices (e.g., land and wages, but also housing) can make a region much less attractive for new ventures.

Third, although it has been widely acknowledged that it is important to include proxies for cultural attitudes and policies towards entrepreneurship in the analysis of regional entry (see, however, Davidsson and Wiklund 1997), its empirical implementation has not been a complete success. The problem, of course, is that it is very difficult to find good proxies. Since data on specific entry-promoting policies is generally not available, for example, Sutaria and Hicks (2004) and Reynolds et al. (1994) advocate using public spending. Cultural attitudes are even more difficult to measure, so the proposed solutions are even more debatable. Garofoli (1994, p. 388), for example, argues that 'areas exhibiting social mobility (...) will have higher rates of new firm formation'. He also tries to capture the 'political climate' by using the percentage of votes obtained by communist and socialist parties.

As for the empirical evidence, the main findings can be summarised in the following way:6

• Demand: population and GDP growth have a positive effect on entry, while the effect of income levels is ambiguous (both positive and negative estimates have been reported).

• Supply: the unemployment rate has an ambiguous effect on entry, while the change in the unemployment rate and the level of wages negatively affect entry; capital and bank deposits have a positive effect on entries: likewise, the proportion of small firms has a positive effect on entries, the effect of establishment size tends to be negative, and the effects of the industry specialisation levels are unclear (both positive and negative estimates have been reported); exit rates have a positive effect on entries.

• Agglomeration: population density, localisation economies and population living in urban areas affect entries positively; dwelling prizes and the share of owners also have a positive effect on entries.

• Cultural attitudes and public policy: immigration tends to affect entries positively, while public policies and political ethos have non-significant or ambiguous effects.

2.2 Evidence from developing countries

The entry process exhibits certain regularities (see Geroski 1995). However, the intensity of entry differs with the level of development of the country, being higher/lower in less/more developed economies (see Wennekers et al. (2005) for a thorough discussion on this topic and

6 It is important to stress that these findings come from the studies listed in footnote 5 that focus on the manufacturing sector (as we do). Namely, Audretsch and Fritsch (1994); Armington and Acs (2002); Carree et al. (2008); Davidsson et al. (1994); Fotopoulos and Spence (1999); Fritsch and Falck (2007); Garofoli (1994); Hart and Gudgin (1994); Keeble and Walker (1994); Reynolds (1994); Reynolds et al. (1994); Santarelli et al. (2009); Spilling (1996); Sutaria and Hicks (2004); and Tamasy and Le Heron (2008).

empirical evidence). In fact, there seems to be a U-shaped relationship between entrepreneur-ship and development (Acs et al. 1994, 2008b). Acs et al. (2008b) also show that developing countries generally exhibit higher turnover rates - especially when the informal economy is included.

Several factors may explain the differences in the patterns of entry (and exit) between developed and developing countries. First, developing economies are generally characterised by macroeconomic instability and intense cyclical variations. The recurrent crises inevitably result in obstacles to the 'creative destruction' process: human capital attrition (Stiglitz 1998), tighter conditions in the financial market (Caballero and Hammour 2000) and higher expected rates of return on the firms' projects due to the shortening of planning horizons (Katz and Bernat 2011). Second, innovation systems in these countries suffer from important deficiencies. This makes innovative entry less frequent, regardless of the technological regime (Burachik 2000). Third, underdeveloped factor markets may restrict access to the resources needed to start a business (financing, skilled labour, raw materials, technology, infrastructure, etc.). They can also negatively affect the supply of entrepreneurs by reducing the share of people with access to information, education, business networks and/or financial resources. Fourth, the political economy of developing countries may cause distortions in the allocation of resources. Bartelsman et al. (2004), for example, argue that governments may give incumbents preferential treatment, artificially increase barriers to entry and/or make exits for some type of businesses more frequent (e.g., SMEs). In addition, government programmes are usually inefficient at promoting entrepreneurship and supportive institutions are largely underdeveloped (Carbonell 2005).

Moreover, these differences not only arise in the intensity of entry but in the profile of the entering firms. For example, the underdevelopment of factor markets may reduce not only firm entry but also their initial size (Kantis et al. 2005), thus decreasing the likelihood of survival (Audretsch 1995a). Also, the number of nascent ventures under the model of 'entrepreneurial economy' tends to be smaller in developing economies (Amoros and Cristi 2008). Similarly, the weight of the necessity-based entrepreneurs is usually higher because of the difficult economic conditions (Acs et al. 2008a). Lastly, Acs et al. (2008a, 2011) argue that the number and type of public institutions influence the allocation of entrepreneurs between formal and informal activities.

2.3 Empirical strategy

The question here, however, is whether there are also differences in the regional determinants of entry between developed and developing economies. The answer is not obvious. Although there is extensive empirical literature on regional firm entry (see footnote 5), the evidence from developing countries is scarce (see footnote 1). Moreover, the heterogeneity of cases (databases, institutional settings, etc.) makes it very difficult to compare results across countries. Lastly, there is no well-established theory that can provide guidelines on what the expected differential effects of a particular determinant of entry are.

The empirical approach we propose is both motivated and limited by these issues. We take as the starting point a set of determinants that are generally found to be statistically significant in regional entry studies using data from developed countries (e.g., demand, education, density and industrial structure). This provides our first (admittedly, indirect) test on the differences between developed and developing countries (see e.g. Fritsch et al. 2006; Ghani et al. 2014). However, we also acknowledge that there are factors that, while potentially important in developing countries, are never considered by studies on developed countries (Bruton et al. 2008). This is the case, for example, of the size of the informal economy and the extent of

poverty (Gerxhani 2004; Schneider 2005; Acs et al. 2008b). This provides our second test on the differences between developed and developing countries.7

In light of the aforementioned differences in the patterns of entry, we expect our first test to show that (most) variables that explain firm entry in advanced countries have either weak statistical significance or show the opposite sign to that typically found in developed countries. We also expect the second test to show that (most) variables that are meant to incorporate some of the specificities of developing countries have substantial explanatory power. We discuss these expectations in more detail in the next section, where we provide a description of the data and the variables we use.

3 The data

3.1 Entry

The Employment and Business Dynamics Observatory (EBDO) of the Ministry of Labour and Social Security of Argentina has drawn up an annual database on firm demography since 1996. Data is available for the 23 Argentinean provinces and the Capital Federal city. However, the Buenos Aires Province is further divided into Gran Buenos Aires and the rest of the province. This is why there are 25 jurisdictions in the database, which we take as our units of observation.

The database includes information about the number of entries, exits and incumbents based on all manufacturing (formal and private) firms with at least one employee.8 Moreover, the EBDO handles changes in firm codes that do not reflect true market entries and exits. In particular, spurious entries and exits caused by the displacement of the whole firm's workforce from firms that 'exit' to become 'new' firms are identified and excluded from the database. Lastly, we restrict the analysis to firms that declare that most of their workforce is located in the assigned jurisdiction. This means that we concentrate on 'local firms' (about 90% of the total firms in 2008), while branch offices or subsidiaries located in other jurisdictions are excluded.9 We report the resulting number of entries, exits and incumbents in Argentina in the years 2003 to 2008 in Table 1.

According to the MTEYSS (2007), in the period 2003-2005, entry rates reached the highest values in a decade. Of course, this was closely related to the recovery of the Argentinean economy after the severe crisis of 2001-2002. Table 1 shows that the high entry rates (around 11%) persisted the following years (2006-2008), although the increase was not so sharp because entry rates dropped in the last two years of our sample (to values of about 7%). As for the exits, after the first two years of stability (2003-2004), they followed the opposite trend, with an average yearly-variation rate of 21 per cent. All these figures indicate that our period of analysis roughly covers a cycle of the Argentinean economy: from recovery (with net entry rates above 5% in the period 2003 to 2006) to progressive decline (with net entry rates of 3% and 0.5% in 2007 and 2008, respectively).

7 One limitation of our approach is that the econometric specification is not directly derived from a set of theories explaining firm entry (in developing/developed countries). This means that we cannot discriminate among conflicting theories and/or test whether one of these theories has empirical support. Notice, however, that this is not the goal of the paper and that this limitation does not invalid our empirical strategy.

8 This means that our data set does not contain information on either public or informal employment. In fact, no statistical source in Argentina can distinguish between informal and formal entries, exits or incumbents. At the aggregate level, the National Household Survey reports that the unregistered workforce in the manufacturing industry was 26.9 per cent in the last quarter of 2008.

9 This constraint was suggested by the EBDO staff so that new offices or branches of large firms that are opened in another province with only one or two people were not regarded as new entries. Moreover, new branch offices may be driven by factors that are different from the ones that influence the creation of 'local' firms.

Table 1. Number of entries, exits and incumbents in Argentina (2003-2008)


5,486 6,264 5,886 5,389

3,623 4,358 5,103

2,330 2,326

42,754 45,234 48,317 49,987 51,796 52,417

Source: Authors (from EBDO data).

In particular, our dependent variable is the number of annual entries in each of the 25 jurisdictions previously described over the period 2003 to 2008. We start our analysis in 2003 to avoid the structural break caused by the economic and political crisis of the end of 2001 that led to the devaluation of the Argentinean peso in January 2002. Including these years of turmoil would completely distort results. We finish our analysis in 2008 because this was the last available year in the EBDO dataset when this investigation was initiated.

Figure 1 shows the evolution of our dependent variable over the period of analysis in Argentina and each of the jurisdictions considered.10 In developed countries, this evolution closely follows the upswings and downswings of the business cycle. That is, entries tend to be pro-cyclical and exits tend to be anti-cyclical. In developing countries, however, Figure 1 shows how macroeconomic instability, financial crisis and/or changes in the prices of raw materials make economic cycles more pronounced. By including the number of entries in Argentina, Figure 1 also shows the extent of heterogeneity in the provinces considered. First, although entries in each province follow the same evolution, some provinces seem to start the cycle later. Second, there are substantial differences in the number of entries across provinces. In particular, the Capital Federal city, the provinces of Gran Buenos Aires, the rest of Buenos Aires Province, Santa Fe and Córdoba stand out as the most attractive provinces in which to create new firms.

Figure 2 displays the spatial distribution of entries and shows that differences in the number of entries across provinces cannot simply be explained by the size of the regions. What is most striking about this figure is the high spatial concentration of manufacturing in Argentina. Notice that most activity clusters in the capital of the country and the surrounding provinces. In fact, according to the EBDO database about 80 per cent of workers and firms in manufacturing are located in the Capital Federal city and the provinces of Gran Buenos Aires, the rest of Buenos Aires Province, Santa Fe and Córdoba. However, these five jurisdictions cover just 22 per cent of the surface of the country. This uneven spatial distribution of the economic activity is quite characteristic of a developing economy (Scott and Storper 2007).11

3.2 Explanatory variables

We used data from the EBDO and the National Household Survey (NHS) to construct our vector of explanatory variables (the size of the provinces in km2 comes from the Military Geographical Institute). The distinction is important because the information contained in the EBDO database refers to the whole province, while the NHS is performed by the National Institute of Statistics and Census (INDEC) on samples of families in 31 urban areas. Nevertheless, we were obliged

10 It is important to note that the interest of this figure is not to identify regional patterns but to show that there are important differences between the provinces with high/medium/low levels of entry.

11 These five provinces also concentrated 62 per cent of the population, 75 per cent of expenditure on science and technology activities, 77 per cent of university degrees, 62 per cent of universities, 85 per cent of the exports of manufactured products, 71 per cent of the GDP and 80 per cent of the manufacturing added value in 2003.

Capital - GBA

Bs As - Córdoba

Santa Fe -Argentina

Chubut - Corrientes

Entre Ríos - Mendoza

Misiones - Neuquén

Río Negro - Salta

San Juan - Tucumán


Catamarca - Chaco

Formosa - Jujuy

La Pampa - La Rioja

San Luis - Santa Cruz

Santiago del Estero - Tierra del Fuego


Fig. 1. Firm entry in Argentina and provinces with high/medium/low entry levels (2003-2008) Source: Authors (from EBDO data).

Santa Cruz

Tierra del Fuego



12-19 20-24



40 60 I 61 - 64 65-146 I 149-235 236 - 669 ■ 670- 1468

Fig. 2. Number of entries by province (average 2003-2008) Source: Authors (from EBDO data).

to use the NHS data because there is no statistical source providing yearly information on demographic and/or socioeconomic characteristics of the Argentinean provinces (population censuses, for example, are performed every 10 years).12

12 Data from the more populated provinces (Capital Federal city, rest of Buenos Aires, Gran Buenos Aires, Córdoba, Chubut, Entre Ríos and Santa Fe) comes from several urban areas of the province (called aglomerados in the jargon of the NHS), while data from the other provinces comes almost exclusively from the capital of the province (small close-by towns are added in some cases, like 'Gran Mendoza', 'Gran Salta', etc.). Because of the inherent error in this procedure, the NHS staff recommends that only variables with a variation coefficient of less than 10 per cent be used (INDEC 2003). All the variables used in this paper have variation coefficients below the 10 per cent. In particular, these variables

In particular, we were able to construct variables related to the evolution of economic activity, the labour market, the level of education, input prices, the industrial structure, the industrial tradition, the existence of agglomeration economies and cultural attitudes. As discussed in the previous section, these factors are widely used in studies on developed countries.13 Moreover, we were able to construct variables related to the level of poverty, the informal economy and idle capacity. As pointed out in Section 2, these variables are usually not included in studies on developed countries. They are included here in an attempt to capture the economic and structural singularities of a developing country. We have also included the square of these variables to account for possible non-linear effects.

Table 2 reports the definition, statistical sources and descriptive statistics of the explanatory variables used in this study. It also contains a column with the expected sign of the associated coefficient. In this respect, it is important to stress that the reported sign correspond to the one commonly found in the empirical literature. This means that, as pointed out above, the inclusion of some of the explanatory variables may be justified on different grounds. That is, in some cases the expected sign may be consistent with alternative or conflicting theories. With this in mind, below we briefly review the arguments and evidence supporting these expected signs.14

3.2.1 Demand for goods

We use the rate of variation in employment in all formal firms (alternatively, the rate of variation in unemployment) to proxy for the evolution of economic activity. The coefficient of this variable is expected to be positive (negative for the rate of variation in unemployment), thus reflecting the procyclicality of entries. As previously pointed out, however, its statistical significance in developing countries may be hampered by the shape of the business cycle and/or the heterogeneity of the geographical units used.

3.2.2 Supply of factors

• Labour. We use the unemployment rate to assess the impact of the labour market on firm entry. In developed countries, the impact of the unemployment rates on entry is ambiguous (Storey 1991; Tervo and Niittykangas 1994; Spilling 1996; Hamilton 1999; Delmar and Davidsson 2000; Ritsila and Tervo 2002). According to the so-called 'push hypothesis' the impact should be positive: the unemployed are more likely to become self-employed and unemployment should push down the cost of labour in the jurisdiction. However, in developing countries the informal sector provides a less costly option to the unemployed and is not reflected in official firm entry registers (like the one we use here). On the other hand, the 'pull

were constructed under the assumption that the ratio of the population with a certain characteristic (e.g., with primary, secondary or university studies) in the aglomerado to the population of the aglomerado is also applicable to the whole province.

13 Notice that we have not included variables related to the capital factor. Unfortunately, there is no reliable information about public and private spending in infrastructure in Argentina. As for measures of credit access, we have explored the amount of loans granted (i) to manufacturing, (ii) per firm, and (iii) per employee. However, these variables showed a negative and statistically significant coefficient that became statistically non-significant when covariates that are characteristic of the developing countries were included. In any case, these results did not differ substantially from the ones reported in Table 3. We consequently decided not to include these variables in our final specifications.

14 The descriptive statistics reported in Table 2 do not include the data from the Río Negro province because the NHS data for this province has only been available since 2006 and the aglomerados surveyed actually cover both urban and rural areas (which are also partly in the Buenos Aires province). This is why the final number of provinces considered in this study is 24 and the total number of observations is 144.

Table 2. Main explanatory variables: definition, sources, expected signs and descriptive statistics




Expected sign Mean St. Dev. Min.

Employment variation

Unemployment rate Primary education

Secondary education

University education

HH index SMEs

Exit ratet-i Industrial tradition






Non-registered/registered Idle capacity

Rate of variation in employment in all formal

firms Unemployment rate

Active individuals with primary education (in thousands)

Active individuals with secondary education (in

thousands) Active individuals with university-level

education (in thousands) Average monthly wage of private registered

workers in manufacturing Hirschman-Herfindahl Index Ratio of small and medium industrial firms to

total industrial firms (formal) Number of exits in the previous year Incumbent firms 7 years ago (3-years moving average)

Log (population/area) (in thousands) Incumbent firms

Private employees/public employees Migration from other provinces (number of

individuals, in thousands) % of households below the indigence line Ratio of non-registered workers to registered workers

Rate of variation in employment in all formal manufacturing firms

Own calculations from EBDO Own calculations from NPS*

Own calculations from EBDO

Own calculations from Military Geographical

Institute and INDEC Own calculations from EBDO Own calculations from NPS*




8.19 191.36




24.36 39.92

135.74 1.916.31

5.20 -6.97

3.81 297.19




12.00 5.77

238.87 3.396.97

Own calculations from EBDO

1.999.11 3.32 206.16

8.87 0.81


8.06 27.27

676.91 2.732.61

3.472.29 1.64 294.16

6.15 0.31

18.20 1554.53




62.90 57.03

1.112 14.550.00

0.83 13.739.75

88.00 1.22 29.93

0.40 0.16

6.79 -15.75

15.107.00 9.14 1.506.10

29.80 1.51

Notes'. * Data refer to 3rd quarter of every year, except for 2007 (4th quarter). Source: Authors.

hypothesis' suggests that the impact should be negative because the unemployed lack entrepreneurial abilities and capital. In developing countries, the negative impact may be higher because of the attrition of human capital in economic downturns (Stiglitz 1998).

• Education. Our proxies for education include the active population with primary, secondary and university-level education. In developed countries, the evidence is mixed and both negative and positive effects have been found (Garofoli 1994; Reynolds 1994; and Reynolds et al. 1994 find a negative impact, while Fotopoulos and Spence 1999 and Davidsson et al. 1994 find a positive effect). This ambiguous impact may be explained by the regional specialization in industries that require different shares of skilled labour (Spilling 1996). Thus, one should expect that in developing countries where firms typically operate in earlier life-cycle stages and tend to specialise in natural resource-intensive goods and scale-intensive industrial commodities, the entry of manufacturing firms show no relation with high educational levels.

• Input prices. Wages correspond to the average monthly wage of private registered workers in manufacturing, in nominal terms because official inflation rates in Argentina have not been reliable since 2007. We expect a negative sign for this variable, in line with what is typicially found in developed countries (Audretsch and Fritsch 1999; Fotopoulos and Spence 1999; Santarelli et al. 2009).

• Industrial structure. The industrial structure of the province is approximated using the Hirschman-Herfindahl Index, the share of micro firms, the share of small and medium firms and the number of exiting firms in the previous year.15 All these variables should impact positively on entry, except for the Hirschman-Herfindahl Index, which measures lack of diversity. First, businesses are more likely to be started in a more diversified environment (Guesnier 1994; Reynolds et al. 1994). Second, entry costs may be lower in areas with a dense network of small and medium-size firms, because these firms pay lower wages (thus reducing the opportunity cost of self-employment). Also, SMES may serve as role models for new entrepreneurs and help their workers to develop the skills required to create a new business (Audretsch 1995b; Ashcroft et al. 1991). Third, exits in previous periods may leave room for newcomers (Arauzo-Carod and Segarra-Blasco 2005; Sutaria and Hicks 2004). Still, studies by Lay (2003) and Gunalp and Cilasun (2006) on Taiwan and Turkey do not support this, which indicates that there is no displacement effect in the (largely) unsaturated markets of developing economies.

• Industrial tradition. We control for the previous industrial activities carried out in a province using the average number of incumbents 7, 6 and 5 years before (i.e. a 3-year centred moving average). Following Rocha and Sternberg (2005), we expect past incumbents to boost current entrepreneurial activities. However, the high macroeconomic volatility of developing countries may mitigate this effect. Changes in the conditions that determine profitability (exchange rate, credit conditions, tax policy, etc.) and the lack of continuity in the industrial policies prevent the consolidation of national firms from which new entrepreneurs can emerge.

3.2.3 Agglomeration

Density and its square have been widely used as proxies for agglomeration and disagglomeration economies, respectively (see e.g., Davidsson et al. 1994; Nystrom 2007; Tamasy and Le Heron 2008). Thus, a positive sign for the density coefficient and a negative sign

15 Firms are distributed by the EBDO in four size levels depending on total employment: micro, small, medium and big. These roughly correspond to the following intervals: micro: 1-5 employees; small: 6-25; medium: 26-100; big: more than 100. However, these intervals vary by industry taking into account sectoral differences in average labour productivity (MTEYSS 2005).

for its square are the expected outcomes in our models. The number of incumbent firms is also included as an additional measure of the agglomeration of economic activity (and as such its impact on entry is expected to be a positive). Lastly, we explored the existence of a core-periphery structure by including the products of a dummy that identifies the richest provinces (the Capital Federal city, Gran Buenos Aires, the rest of Buenos Aires Province, Santa Fe and Córdoba) with the variables of density and incumbents. We expect these products to have a positive effect on entry (Krugman 1991), since this unbalanced geographical pattern means that entering firms in the core and the periphery face quite different conditions (e.g., access to markets, skilled workers and services to the firm). That is, positive agglomeration effects are expected to arise only in the 'central' areas.

3.2.4 Private-to-public sector

Cultural attitudes towards entrepreneurship may be captured by the ratio of private-to-public employees. In particular, we expect entries to be higher in jurisdictions with a higher private/ public rate (Spilling 1996).

3.2.5 Migrants

As Lee et al. (2004) and Tamásy and Le Heron (2008) show, there are more entries in communities with higher inflows of migrants. We have consequently included among the regressors the number of individuals from other provinces.16

3.2.6 Poverty

We proxy the extent of poverty with the percentage of households below the indigence line. This threshold is given by the capacity to afford a basic food basket, which is estimated to be about US$38 per adult in 2003. We expect this variable to show a negative coefficient in our models for two reasons. First, low income markets do not attract the entry of new firms. Second, the proportion of entrepreneurs who have access to resources for backing up their business decisions should be lower in low-income areas (Casson 1982; Hamilton and Harper 1994).

3.2.7 Informal economy

The instability and insecurity of informal jobs are factors that may push individuals to start their own business (Storey 1994). Likewise, the informal sector may encourage entry by acting as a 'stepping stone' (Bennet 2010). That is, entrepreneurs may first enter the informal sector to 'test the water' before deciding on whether or not to enter the formal sector. Lastly, informal suppliers can offer lower prices, thus making formal entries cheaper. We use the ratio of non-registered workers to total workers to incorporate this positive effect on entry. However, this variable may also reflect the productive structure (e.g., the seasonality and/or low productivity of certain activities may facilitate the growth of the informal sector) and/or the lack of government controls on the informal economy in certain provinces, and thus have a negative or nonsignificant effect on formal entry.

16 Note, however, that these studies refer to international migration. Our dataset contains information on the number of individuals coming from other countries. Unfortunately, the contents of this variable turned out to be flawed.

3.2.8 Idle capacity

The idle capacity caused by the economic recession of 2001-2002 may have slowed new firm creation to the extent that the subsequent demand for new goods (from 2003) may have been satisfied by existing firms rather than by new firms. In this respect, Cala and Rotondo (2012) show that during the period of analysis provinces with higher (lower) industrial intensity had lower (higher) net entry rates. This suggests that the impact of idle capacity may have been more intense in more developed provinces. We seek to capture this effect by including the rate of variation in employment in all formal manufacturing firms and the product of this rate by a dummy for the five most developed provinces (see footnote 8).

4 Econometric modelling and estimation of results

Given the definition of our dependent variable (yearly number of entries in the 24 Argentinean provinces considered), we rely on panel count data models to estimate the impact of entry determinants.17 Panel data models were preferred to cross-section estimates on the grounds of two empirical tests. First, likelihood ratio tests on the variance of the random effects always yielded statistically significant results, thus rejecting the validity of pooled estimates (Cameron and Trivedi 2009). Second, we computed the covariance matrix of the year vector of Pearson-residuals from the pooled Poisson regression model (see Hausman et al. 1984 for details). We found large values in the off-diagonal elements of the matrix in all the specifications, which supports the assumption of independence of the observations across the years studied and justifies the use of panel data models.

It should be noted that there are no zeros in our dependent variable. That is, in each jurisdiction-year pair of our sample we have a strictly positive number of entries. This is why we concentrate on the estimation of Poisson and negative binomial models (Cameron and Trivedi 1998). This contrasts with the typical outcome of studies of developed countries, which tends to be constructed from the inflated versions of these models to account for the 'excess of zeros' (see e.g., List 2001; Basile 2004; Manjon-Antolfn and Arauzo-Carod 2011). The size of our administrative units, much larger than the municipalities, counties and metropolitan areas studied in developed countries, lies behind this important difference.

In particular, Table 3 shows the results from the negative binomial fixed effects model.18 Our choice is based on the results of a number of tests (see the bottom rows in Table 3). First, the Pearson goodness-of-fit test from a Poisson model with province dummy variables provides evidence of over-dispersion in the data (Allison and Waterman 2002).19 This means that the Poisson estimates are not efficient (and may even be inconsistent if the conditional expectation of the entry rate is not correctly specified, as shown by Hausman et al. 1984). Second, the Durbin-Hu-Hausman test rejects the null hypothesis of no correlation between the covariates and the individual effect, which means that the random effects model yields inconsistent estimates. Lastly, negative binomial fixed effects estimates provide the best fit according to the Akaike information criterion (AIC).

17 See for example, Chappell et al. (1990); Ilmakunnas and Topi (1999); Barbosa et al. (2004); Barbosa (2007); and Fritsch and Falck (2007) for analogous applications in developed countries. Panel data methods not only allow us to increase the number of observations, thus improving the efficiency of the estimates, but also to capture unobserved heterogeneity across provinces over time. Cross-sectional studies are becoming less frequent in the analysis of entry, if any because panel datasets have become more available in recent years (Arauzo-Carod et al. 2010).

18 Coefficients' estimates can be interpreted as semi-elasticities. We do not report marginal effects because of the difficulties in integrating out the unobserved heterogeneity in non-linear models (Cameron and Trivedi 2009).

19 Only the ratio between the individual effect and the over-dispersion parameter is identified in the negative binomial model, which makes difficult to construct an equidispersion test (Cameron and Trivedi 1998).

Table 3. Firm entry determinants

[1] [2] [3]

Employment variation 0.0159*** 0.0159*** 0.0131**

(0.0039) (0.0046) (0.0055)

Unemployment rate 0.0061 -0.0040 -0.0118

(0.0087) (0.0092) (0.0113)

Primary education -0.0008 -0.0002 0.0001

(0.0005) (0.0006) (0.0007)

Secondary education 0.0004 -0.0000 -0.0000

(0.0005) (0.0005) (0.0005)

University education 0.0011** 0.0004 0.0006

(0.0005) (0.0004) (0.0004)

Wages -0.0001 0.0001 0.0000

(0.0001) (0.0001) (0.0001)

HH index 0.0101 0.0200** 0.0067

(0.0074) (0.0084) (0.0108)

SMEs 0.0290*** 0.0350*** 0.0243*

(0.0090) (0.0094) (0.0124)

Exit ratet-1 -0.0005*** -0.0003** -0.0003*

(0.0001) (0.0002) (0.0002)

Industrial tradition -0.0002*** -0.0002** -0.0003***

(0.0001) (0.0001) (0.0001)

Density 4.6742*** 4.0241** 0.8770

(1.6046) (1.6658) (1.6838)

Density2 -0.4610*** -0.4169*** -0.4074*

(0.1736) (0.1466) (0.2204)

Incumbents -0.0003** -0.0002 -0.0013***

(0.0001) (0.0001) (0.0005)

Private/public 0.0036 -0.0151 -0.0110

(0.0165) (0.0170) (0.0186)

Migrants -0.0004 -0.0001 0.0001

(0.0004) (0.0004) (0.0004)

Poverty -0.0474*** -0.0566***

(0.0124) (0.0134)

Poverty2 0.0013*** 0.0013***

(0.0004) (0.0004)

Non-registered/registered 1.8110*** 1.6745***

(0.4911) (0.5898)

Non-registered/registered2 -0.9865*** -0.8862***

(0.2336) (0.3010)

Idle capacity -0.0114* -0.0152**

(0.0067) (0.0072)

Idle capacity2 0.0000 0.0000

(0.0003) (0.0003)

Idle capacity x rich provinces dummy 0.0024 0.0359

(0.0231) (0.0248)

Idle capacity2 x rich provinces dummy 0.0006 -0.0004

(0.0011) (0.0012)

Density x rich provinces dummy 3.2532**


Incumbents x rich provinces dummy 0.0009**


AIC 978.61 973.34 977.54

LR test of joint significance 170.81*** 245.36*** 217.21***

Hausman 25.78** 32.77*** 29.04***

Pearson'GoF test 182.39*** 131.18*** 117.23***

Notes: Negative binomial fixed effects estimates are reported. Standard errors in brackets. Asterisks indicate the statistical significance of the coefficient: *** p-value < 0.01, ** p-value < 0.05, * p-value < 0.1. Year dummy variables are included in all the specifications. The only difference between the results reported in columns (1), (2) and (3) is the inclusion of additional explanatory variables that proxy for the specificities of developing economies.

Let us first consider results from the specification that contains variables which are widely used in studies on developed countries. These are reported in the first column of Table 3. The first thing to point out is that, as previously hypothesised, many of the determinants considered are not statistically significant. To be precise, our proxies for the labour market, education, input prices and cultural attitudes have practically no explanatory power. Only demand, agglomeration economies, and industrial structure and tradition show statistically significant coefficients. From these results we can conclude that entries follow the evolution of economic activity (i.e., they are procyclical) and are positively affected by the number of graduates, the share of SMEs and agglomeration economies (although the negative sign of the squared density and the number of incumbents point to the existence of disagglomeration effects). On the other hand, past incumbents and exits deter entry, which suggests that macroeconomic instability hampers the boosting effect of past incumbents on current firm formation and the rate of exit actually reflects negative expectations about the evolution of economic activity.

We now go on to consider the results obtained when the covariates that are characteristic of developing countries are added: poverty, the informal economy and idle capacity. They are reported in the second column of Table 3. Interestingly, these additional variables and their squared terms are all statistically significant (except for the square of the idle capacity). Furthermore, the coefficient estimates and the statistical significance of the rest of the covariates remain essentially unaltered with respect to those reported in the first column of Table 3 (except the university education and the number of incumbents are not statistically significant and the Hirschman-Herfindahl Index is). Therefore, this evidence is largely supportive of the arguments put forward in Section 2: while most of the determinants typically used in previous studies analysing developed countries are still relevant here, there is a need for additional explanatory variables that reflect the specificities of developing economies. In fact, our results show that including these variables improves the fit of the model in terms of AIC.

In particular, the negative sign of the poverty variable is consistent with low-income people having less purchasing power and entrepreneurs having greater difficulty in finding appropriate resources in poor areas. Also, the positive effect of the squared term suggests that high levels of poverty spur the creation of (possibly small) firms. As for the impact of the informal economy, it seems that it is 'too much of a good thing'. A small informal economy encourages entry, but it becomes a barrier when it grows too much. Lastly, the negative sign of idle capacity suggests that it is existing firms (which increase their number of employees) rather than new firms that satisfy a good deal of the demand for new goods.

Our final specification seeks to analyse the impact of a core-periphery pattern in the agglomeration economies (see the last column in Table 3). Descriptive statistics show that there are huge differences in terms of the location of population and firms between 'central' areas and 'peripheral' areas (i.e., between the five richest provinces and the rest). Our estimates indicate that these differences have an impact on entry. In fact, the positive sign of the products of the dummy of the richest provinces and the density and incumbent variables indicates that firms entering these provinces may have access to better resources and business opportunities. Notice also that including these differential effects of the agglomeration economies has practically no effect on the estimates of the other covariates.

To conclude, it is worth noting that the previous conclusions are largely robust to alternative model specifications. Although some of the coefficients had different values and/or statistical significance with respect to those reported in Table 3, most of the previous results remain unaffected (these are available upon request) but the fit was generally worse. In particular, we explored the robustness of our conclusions to the use of alternative model specifications (random effects and fixed effects Poisson, as well as negative binomial with province dummies)

and a different set of proxies (for the demand for goods, the agglomeration economies and the extent of poverty). Below we briefly discuss the results of these robustness exercises.20

First, alternative model specifications provided essentially the same signs and statistical evidence. The main changes were the negative and statistically significant sign of secondary education and the lack of significance of the product of density and the dummy of the richest provinces. Second, including the rate of variation in unemployment instead of the rate of variation in employment in all formal firms as a proxy for the evolution of the economic activity barely changes the results. However, this variable showed a negative but not statistically significant coefficient (in the negative binomial fixed effects specification). Third, we looked into including the ratio between the population in the main urban areas of the province ('aglomerados') and the total population of the province instead of the density of the province and its square. Since our units of observation are extremely large (see footnote 3), this agglomeration measure may better reflect the uneven distribution of firms and individuals within large provinces (with large stretches of available land with no industrial activity) and the concentration of services in urban areas (Puga 1998; Henderson 2000). Estimates showed that this variable often had a negative impact on entry, thus indicating that jurisdictions with a bigger urban ratio are less attractive than jurisdictions with a smaller urban ratio. However, we faced severe converge problems in many of the specifications considered. Lastly, we explored the use of a variable constructed with a different threshold to proxy for the extent of poverty (US$83 rather than UD$38, which corresponds to the value of the total basic basket per adult in 2003 and includes the basic food basket plus the value of basic household expenditures such as housing, dressing, transportation and education). Again, estimates remained essentially unaltered. However, the square of this alternative variable was not statistically significant.

5 Conclusions

This paper analyses the regional determinants of firm entry in a developing economy. This is a novelty in the context of an empirical literature that largely focuses on evidence from Europe, North America or Japan. In particular, we provide estimates from panel count data models using annual provincial data on new manufacturing firms with employees registered in the Argentinean Social Security files during the period 2003 to 2008. This is the most up-to-date, comprehensive, reasonably long-term and spatially disaggregated data source currently available for firm demography studies in Argentina.

We compare the results obtained using a well-established list of economic and demographic characteristics that explains entry decisions of new firms in the developed economies with those obtained by adding variables that proxy for the specificities of developing countries. We find that most of the determinants used in previous studies analysing developed countries remain relevant when we add variables proxying for the extent of poverty, the size of the informal economy and the existence of idle capacity. Furthermore, we find that the entry process shows significant differences in the richest provinces. In particular, we find evidence of centre-periphery differences in the impact of agglomeration economies but not in idle capacity.

In terms of policy implications, our results stress the risk of 'rubber-stamp' policies that simply follow recipes that work well in developed countries. In other words, policy-makers should take into account country specificities when designing entry-promoting policies in developing economies. In the Argentinean case, for example, the negative effect that poverty has

20 We also computed the cross-sectional correlation between the explanatory variables and found that it was generally low, except for density and its square, idle capacity and its square, and exit and the industrial tradition, which showed values above 0.9. However, it is hard to assess the potential impacts of these correlations in non-linear models. Our results from the robustness exercises suggest that in our case this collinearity should not be a major concern.

on entries is unlikely to be reversed by policies simply aiming at promoting new business creation, for reducing the rate of poverty probably requires a long-term policy of investment in human capital.

Any further extension of this study will be mainly driven by the limitations of our data set and empirical strategy. First, a more disaggregated unit of observation should be used. Given the lack of data on smaller geographical units (municipalities, counties and/or metropolitan areas), exploring a sectorial breakdown will not only reduce the degree of heterogeneity but also incorporate industry-specific variables. Second, the uneven distribution of the economic activity across the country should be dealt with. We have used dummies to control for the huge concentration around the capital and the surrounding regions, which is typical of a developing country. However, this phenomenon may require a more sophisticated approach.

We conclude by noting that the data used to analyse firm entry in developed and developing countries differs considerably. In particular, there are differences in the reliability (e.g., data is based on estimates rather on measures), representativeness (e.g., data is provided only for small, core areas of each administrative unit) and spatial aggregation (e.g., data is only available for large and heterogeneous areas). Addressing these shortcomings is critical if solid and comparative evidence is to be provided on the determinants of firm entry in developing countries. Thus, we leave for future research the question of whether the reported results from Argentina hold for other developing countries.


Acs ZJ, Amoros JE (2008) Entrepreneurship and competitiveness dynamics in Latin America. Small Business Economics 31:305-322

Acs ZJ, Audretsch DB, Evans DS (1994) The determinants of variation in self-employment rates across countries and

over time. Centre for Economic Policy and Research, London Acs ZJ, Bardasi E, Estrin S, Svejnar J (2011) Introduction to special issue of small business economics on female

entrepreneurship in developed and developing economies. Small Business Economics 37: 393-396 Acs ZJ, Desai S, Hessels J (2008a) Entrepreneurship, economic development and institutions. Small Business Economics 31: 219-234

Acs ZJ, Desai S, Klapper LF (2008b) What does 'entrepreneurship' data really show? Small Business Economics 31: 265-281

Allison PD, Waterman RP (2002) Fixed-effects negative binomial regression models. Sociological Methodology 32: 247-265

Amoros JE, Cristi O (2008) Longitudinal analysis of entrepreneurship and competitiveness dynamics in Latin America.

The International Entrepreneurship and Management Journal 4: 381-399 Arauzo-Carod JM, Liviano D, Manjon-Antolin M (2010) Empirical studies in industrial location: An assessment of their

methods and results. Journal of Regional Science 50: 685-711 Arauzo-Carod JM, Liviano D, Martin M (2008) New business formation and employment growth: Some evidence for

the Spanish manufacturing industry. Small Business Economics 30: 73-84 Arauzo-Carod JM, Segarra-Blasco A (2005) The determinants of entry are not independent of start-up size: Some

evidence from Spanish manufacturing. Review of Industrial Organization 27: 147-165 Armington C, Acs ZJ (2002) The determinants of regional variation in new firm formation. Regional Studies 36: 33-45 Ashcroft B, Love JH, Malloy E (1991) New firm formation in the British counties with special reference to Scotland.

Regional Studies 25: 395-409 Audretsch DB (1995a) Innovation, growth and survival. International Journal of Industrial Organization 13: 441-457 Audretsch DB (1995b) Innovation and industry evolution. Massachusetts Institute of Technology, Cambridge, MA Audretsch DB, Fritsch M (1999) The industry component of regional new firm formation process. Review of Industrial Organization 15: 239-252

Audretsch DB, Fritsch M (1994) The geography of firm births in Germany. Regional Studies 28: 359-365 Audretsch DB, Keilbach M (2005) Entrepreneurship capital and regional growth. The Annals of Regional Science 39: 457-469

Barbosa N (2007) An integrative model of firms' entry decisions. Applied Economics Quarterly 53: 45 Barbosa N, Guimaraes P, Woodward D (2004) Foreign firm entry in an open economy: The case of Portugal. Applied Economics 36: 465-472

Bartelsman E, Haltiwanger J, Scarpetta S (2004) Microeconomic evidence of creative destruction in industrial and

developing countries. Institute for the Study of Labor, IZA, Discussion Paper 1374 Basile R (2004) Acquisition versus greenfield investment: The location of foreign manufacturers in Italy. Regional

Science and Urban Economics 34: 3-25 Bennet J (2010) Informal firms in developing countries: Entrepreneurial stepping stone or consolation prize? Small

Business Economics 34: 53-63 Bosma N, Van Stel A, Suddle K (2008) The geography of new firm formation: Evidence from independent start-ups and

new subsidiaries in the Netherlands. International Entrepreneurship and Management Journal 4: 129-146 Bruton GD, Ahlstrom D, Obloj K (2008) Entrepreneurship in emerging economies: Where are we today and where

should the research go in the future. Entrepreneurship Theory and Practice 32: 1-14 Burachik G (2000) Cambio tecnológico y dinámica industrial en América latina. Revista de la CEPAL 71: 85-104 Caballero R, Hammour M (2000) Creative destruction and development: Institutions, crises, and restructuring. MIT Working Paper 00-17

Calá CD, Rotondo JS (2012) Dinámica empresarial en la industria argentina. Un análisis provincial para el período

2003-2008. Revista Realidad Económica 267: 138-170 Cameron AC, Trivedi PK (1998) Regression analysis of count data. Cambridge University Press, Cambridge Cameron AC, Trivedi PK (2009) Microeconometrics using Stata. Stata Press Publication, College Station, Texas Campbell CM (1996) The effect of state and industry economic conditions on new firm entry. Journal ofEconomics and Business 48: 167-83

Carbonell S (2005) The entrepreneurial environment in Latin America. In: Acs Z, Arenius P, Hay M, Minniti M (eds)

Global entrepreneurship monitor. Executive Report 2004. Babson College/ London Business School, London Carree M, Santarelli E, Verheul I (2008) Firm entry and exit in Italian provinces and the relationship with unemployment. International Entrepreneurship Management Journal 4: 171-186 Castillo V, Cesa V, Filippo A, Rojo Brizuela S, Schleser D, Yoguel G (2002) Dinámica del empleo y rotación de empresas: La experiencia en el sector industrial de Argentina desde mediados de los años noventa (Estudios y perspectivas, no 9). CEPAL, Santiago Casson M (1982) The entrepreneur. An economic theory. Edward Elgar, Cheltenham

Chappell W, Kimenyi M, Mayer WA (1990) Poisson probability model of entry and market structure with an application

to US industries during 1972-77. Southern Economic Journal 56: 918-27 Davidsson P, Lindmark L, Olofsson C (1994) New firm formation and regional development in Sweden. Regional Studies 28: 395-410

Davidsson P, Wiklund J (1997) Values, beliefs and regional variations in new firm formation rates. Journal ofEconomic Psychology 18: 179-199

Delmar F, Davidsson P (2000) Where do they come from? Prevalence and characteristics of nascent entrepreneurs.

Entrepreneurship & Regional Development 12: 1-23 Fotopoulos G, Spence N (1999) Spatial variations in new manufacturing plant openings: Some empirical evidence from

Greece. Regional Studies 33: 219-29 Fotopoulos G, Spence N (2001) Regional variations of firm births, deaths and growth patterns in the UK 1980-1991.

Growth and Change 32: 151-173 Fritsch M, Brixy U, Falck O (2006) The effect of industry, region, and time on new business survival: A multidimensional analysis. Review of Industrial Organization 28: 285-306 Fritsch M, Falck O (2007) New business formation by industry over space and time: A multidimensional analysis.

Regional Studies 41: 157-172 Fritsch M, Mueller P (2004) The effects of new firm formation on regional development over time. Regional Studies 38: 961-975

Fritsch M, Schmude J (2006) Entrepreneurship in the region (International Studies in Entrepreneurship). Springer, New York

Garofoli G (1994) New firm formation and regional development: The Italian case. Regional Studies 28: 381-393 Gennero A, Baltar F, Liseras N (2004) Diferencias espaciales en la gestación de ideas empresariales en la Argentina. In: Gennero A (ed) El proceso de creación de empresas. Factores determinantes y diferencias espaciales. Suárez, Mar del Plata, Argentina

Geroski PA (1995) What do we know about entry? International Journal of Industrial Organization 13: 421-440 Gerxhani K (2004) The informal sector in developed and less developed countries: A literature survey. Public Choice 120: 267-300

Ghani E, Kerr WR, O'Connell S (2014) Spatial determinants of entrepreneurship in India. Regional Studies,

doi:10.1080/00343404.2013.839869 Guesnier B (1994) Regional variations in new firm formation in France. Regional Studies 28: 347-58 Günalp B, Cilasun SM (2006) Determinants of entry in Turkish manufacturing industries. Small Business Economics 27: 275-287

Hamilton RT (1999) Unemployment and business formation rates: Reconciling time series and cross section evidence.

Environment and Planning 21: 249-255 Hamilton RT, Harper DA (1994) The entrepreneur in Theory and Practice. Journal of Economic Studies 21: 3-18 Hart M, Gudgin G (1994) Spatial variations in new firm formation in the Republic of Ireland 1980-1990. Regional Studies 28: 367-380

Hausman J, Hall BH, Griliches Z (1984) Econometric models for count data with an application to the patents: R&D

relationship. Econometrica 52: 909-938 Henderson J (2000) The effects of urban concentration on economic growth. NBER Working Paper 7503 INDEC (2003) EPH-continua - Estimación de los errores de muestreo. Dirección Encuesta Permanente de Hogares,

Departamento de Muestreo. URL: Ilmakunnas P, Topi J (1999) Microeconomic and macroeconomic influences on entry and exit of firms. Review of

Industrial Organization 15: 283-301 Kangasharju A (2000) Regional variations in firm formation: Panel and cross-section data evidence from Finland.

Papers in Regional Science 79: 355-373 Kantis H, Aangelelli P, Moori Koening V (2005) Developing entrepreneurship: Experience in Latin America and

Worldwide. Inter-American Development Bank, Washington Katz J, Bernat G (2011) Exit-entry, productivity growth and structural change in response to changes in macroeconomic

policy: Evidence from Argentina. Innovation and Development 1: 227-244 Keeble D, Walker S (1994) New firms, small firms and dead firms: Spatial patterns and determinants in the United

Kingdom. Regional Studies 28: 411-427 Krugman P (1991) Increasing returns and economic geography. Journal of Political Economy 99: 483-499 Lay TJ (2003) The determinants of and interaction between entry and exit in Taiwan's manufacturing. Small Business Economics 20: 319-334

Lee SY, Florida R, Acs ZJ (2004) Creativity and entrepreneurship: A regional analysis of new firm formation. Regional Studies 38: 879-891

Lipton M (1977) Why poor people stay poor: a study of urban bias in world development. Temple Smith, London List JA (2001) US county-level determinants of inbound FDI: Evidence from a two-step modified count data model.

International Journal of Industrial Organization 19: 953-973 Manjón-Antolín M, Arauzo-Carod JM (2011) Locations and relocations: Determinants, modelling, and interrelations.

Annals of Regional Science 47: 131-146 MTEYSS (2005) Dinámica del empleo y rotación de empresas: Nota metodológica. Mimeo, Buenos Aires MTEYSS (2007) La creación de empresas durante la etapa post convertibilidad (2003-2005): Impacto sobre el empleo asalariado registrado. In: Salarios, empresas y empleo 2003-2006, Series Trabajo, ocupación y empleo, 5. National Ministry of Labour, Employment and Social Security, Buenos Aires Naudé W (2011) Entrepreneurship is not a binding constraint on growth and development in the poorest countries. World Development 39: 33-44

Naudé W, Gries T, Wood E, Meintjies A (2008) Regional determinants of entrepreneurial start-ups in a developing

country. Entrepreneurship & Regional Development 20: 111-124 Nystrom K (2007) An industry disaggregated analysis of the determinants of regional entry and exit. Annals of Regional Science 41: 877-896

Ozturk S, Kilic C (2012) Patterns and determinants of entry and exit in Turkish manufacturing industries. International

Journal of Arts and Commerce 1: 107-118 Powell B (2008) Making poor nations rich: Entrepreneurship and the process of economic development. The Independent Institute, Oakland, CA

Puga D (1998) Urbanization patterns: European versus less developed countries. Journal of Regional Science 38: 231-252

Reynolds P (1994) Autonomous firm dynamics and economic growth in the United States 1986-1990. Regional Studies 28: 429-442

Reynolds P, Storey DJ, Westhead P (1994) Cross-national comparisons of the variation in new firm formation rates.

Regional Studies 28: 443-456 Rigby D, Essletzbichler J (2000) Impacts of industrial mix, technological change, selection and plant entry/exit on

regional productivity growth. Regional Studies 34: 333-342 Ritsila J, Tervo H (2002) Effects of unemployment on new firm formation: Micro-level panel data evidence from

Finland. Small Business Economics 19: 31-40 Rocha HO, Sternberg R (2005) Entrepreneurship: The role of clusters theoretical perspectives and empirical evidence

from Germany. Small Business Economics 24: 267-292 Santarelli E, Carree M, Verheul I (2009) Unemployment and firm entry and exit: An update on a controversial relationship. Regional Studies 43: 161-173

Santarelli E, Tran HT (2012) Growth of incumbent firms and entrepreneurship in Vietnam. Growth and Change 43: 638-666

Schneider F (2005) Shadow economies around the world: what do we really know? European Journal of Political Economy 21: 598-642

Spilling OR (1996) Regional variation of new firm formation: The Norwegian case. Entrepreneurship and Regional

Development 8: 217-243 Scott AJ, Storper M (2007) Regions, globalization, development. Regional studies 41: S191-S205 Stiglitz J (1998) Towards a new paradigm for development: Strategies, policies, and processes. Given as the 1998

Prebisch Lecture at UNCTAD, Geneva, 19 October Storey D (1991) The birth of new firms. Does unemployment matter? A Review of Evidence. Small Business Economics 3: 167-178

Storey D (1994) Understanding the small business sector. Routledge, London

Sutaria V, Hicks DA (2004) New firm formation: Dynamics and determinants. Annals ofRegional Science 38: 241-262 Tamasy C, Le Heron R (2008) The geography of firm formation in New Zealand. Tijdschrift voor Economische en Sociale Geografie 99: 37-52

Tervo H, Niittykangas H (1994) The impact of unemployment on new firm formation in Finland. International Small Business Journal 13: 38-53

Thompson K (2010) Data in development: An overview of microdata on developing countries. IASSIST Quarterly

Winter/Spring 2010: 25-30 Wang S (2006) Determinants of new firm formation in Taiwan. Small Business Economics 27: 313-321 Wennekers S, Van Stel A, Thurik AR, Reynolds P (2005) Nascent entrepreneurship and the level of economic development. Small Business Economics 24: 293-309