Scholarly article on topic 'New business formation and employment growth: some evidence for the Spanish manufacturing industry'

New business formation and employment growth: some evidence for the Spanish manufacturing industry Academic research paper on "Social and economic geography"

0
0
Share paper
Academic journal
Small Bus Econ
OECD Field of science
Keywords
{""}

Academic research paper on topic "New business formation and employment growth: some evidence for the Spanish manufacturing industry"

Small Bus Econ (2008) 30:73-84 DOI 10.1007/s11187-007-9051-4

New business formation and employment growth: some evidence for the Spanish manufacturing industry

Josep Maria Arauzo Carod • Daniel Liviano Solis • Monica Martin Bofarull

Received: 19 October 2005/Accepted: 6 February 2007/Published online: 22 May 2007 © Springer Science+Business Media B.V. 2007

Abstract This paper explores the effects of new business formation on employment growth in Spanish manufacturing industries. New firms are believed to make an important contribution to economic growth but the extent of this contribution is unclear. We consider time lags of new firm formation as explanatory variables of employment change and identify how long the effect of new firm entries on employment lasts. Our main results show that the effects of new business formation are positive in the short term, negative in the medium term and positive in the long term, thus confirming the existence of indirect supply-side effects found in similar studies for other countries.

JEL classifications L00 • L60 • R11 • R12 • L26

Keywords Regional growth • Firm entry • Time lags • Spanish economy

J. M. Arauzo Carod (El) • D. Liviano Solis • M. Martin Bofarull

Department of Economics, Universitat Rovira i Virgili, Av. de la Universitat, 1, Reus-43204, Catalonia, Spain e-mail: josepmaria.arauzo@urv.cat

1 Introduction

What is the incidence of new business formation (or firm entries) on employment growth? We assume that there is a dynamic relationship between employment growth and firm entries as a result of job creation by entering firms, job destruction by exiting firms and both job creation and job destruction due to the interaction of the effects of entering and exiting firms.

The literature on regional economics provides considerable empirical evidence for understanding the relationship between firm entries and employment growth. Some studies consider the employment effect attributed to new business formation to be a dynamic process that is related to the characteristics of firm turnover. These studies report that new firms can have both positive and negative effects on employment. Recent empirical papers on firm entry therefore suggest the incorporation of time lags in order to capture the various effects of new business formation on employment.

Fritsch and Mueller (2004) stress the importance of considering two types of effects that new firm creation can have on economic development: i.e. direct and indirect effects. First we shall discuss the direct effects. The first effect of new firm formation on employment is the creation of jobs. Later, however, the market begins a firm selection process. Net job formation may therefore be positive or negative and will depend on how the newcomers develop. Two types of exits derive from the entry of

new capacities. First, some new firms may have to leave the market after a certain time due to their lack of competitiveness. Second, some incumbents may be forced out of the market by new competitors. With a ''survival-of-the-fittest'' scenario, and if the overall market volume remains constant, a negative net job creation can be expected from the difference between the creation of employment by new firms and the destruction of employment by exiting firms (both newcomers and incumbents).

The indirect supply-side effects derive from the entry of new firms and the more intense competition this creates. These effects can help to increase the competitiveness of an economy and may stimulate economic growth. Fritsch and Mueller (2004) and Fritsch, Mueller, and Weyh (2005) point out the following main types of indirect effects: a greater efficiency of the incumbent firms due to stronger competition from real and potential entrants; a faster structural change, since the turnover of firms leads to the adoption of new technologies; greater innovation, since newcomers are more able than incumbents to introduce radical innovations and are more interested in exploiting the possibilities for potential profit; and innovative entry, which may lead to better-quality and more varied products and a greater probability of finding a better match for customer preferences.

The supply-side effects of new firms are related to increased employment induced by improvements in competitiveness. However, this positive result does not depend only on the success of the newcomers: the greater supply could come from both newcomers and incumbents.

The impact of entries on net employment change in the long term is a major issue in industrial organization because of its implications for industrial policies. If we know how entries affect employment, we can target the most suitable policies for promoting net employment growth and identify what kinds of entries should be promoted. From this point of view a key question is: is it better to promote entries or to help incumbent firms? Audretsch and Fritsch (2002) reviewed the literature on this issue and showed that there are different growth regimes at a regional level and that regional economic development could be caused by both new firms or by incumbent firms. These findings highlight the fact that there are no clear answers as to which kind of development strategies are more suitable—those based on support-

ing incumbent firms or those based on supporting entries—so more work is needed to understand, for example, the role of new firms in employment growth.

This paper explores the incidence of new business formation (or firm entries) on employment growth in Spanish manufacturing industries. For the Spanish case, there is only some evidence of the incidence of firm entries on productivity growth (Callejon & Segarra, 1999; Farinas & Ruano, 2004; Martin & Jaumandreu, 2004),1 though almost nothing is known about the incidence of firm entries on employment change.

The paper is organized as follows. In the second section we present the empirical evidence of firm entry and employment change. In the third section we present the database and the variables used. In the fourth section we develop the model and econometric estimations and present our main results from these estimations. In the fifth section we discuss our findings and draw some conclusions.

2 Firm entry and employment change at a

regional level

The entry of new firms erodes the power of established firms by increasing competition in the market. This leads to a displacement process that causes the least efficient firms to leave (Geroski, 1989). For this reason, the entry and exit of firms are closely related phenomena.

What, then, are the consequences of entries? First, new entries can create a displacement effect (Au-dretsch, 1995) in which jobs are destroyed as less efficient incumbent firms exit the market due to the

1 Callejon and Segarra (1999) demonstrate that entries

contribute positively to the growth of total factor productivity. Farinas and Ruano (2004) show that incumbent firms are the main contributors to the change in the productivity distribution, while entering firms have lower productivity than incumbent firms. Martin and Jaumandreu (2004) also show that more efficient firms have replaced low productivity firms. Generally

speaking, firms that enter the market have higher productivity levels than firms that exit the market. For an overview of the effect of turnover on productivity growth, see Tybout (1996) and Caves (1998). For some empirical evidence in other countries see, among others, Baldwin and Gorecki (1991), Baldwin (1995), Aw, Chen, and Roberts (1997) and Geroski

(1989).

greater competitiveness created by the new entrants. Second, the survival rate of new firms is low and most of today's entries are tomorrow's exits. However, not all new entries exit markets immediately, which imply that the effect of new entries on employment is unclear in the short term. In the long term, however, we assume that the more efficient firms survive (these surviving firms have a faster growth pattern) and displace the less efficient firms (incumbents and/or new firms).2 At the same time, however, there is the conical revolving door effect (Audretsch, 1995), by which the turnover of entries and exits is much higher for smaller firms than for larger ones. This means that the size of entering firms is smaller than the industry average, that many of these firms exit during the first few years, and that those that survive can grow faster than the incumbent firms. Several contributions have analyzed the relationship between the initial size of the firm and the likelihood of survival (Audretsch, 1995; Audretsch & Mahmood, 1995; Dunne & Hughes, 1994; Mata & Portugal, 1995, 1999). The empirical evidence shows that there is a positive relationship between size at start-up and the likelihood of survival.4 However, some empirical papers suggest that, if small firms are to survive, they need to grow at a faster rate than their larger counterparts. Thus, Audretsch, Santarelli, and Vivarelli (1999) point out that the post-entry growth rates of surviving firms are observed to be negatively related to firm size. Agarwal and Audretsch (2001) also find a negative relationship for the mature life-cycle stage.

Despite job destruction in the short term due to adjustment costs, we can assume that there is a job creation process in the long term, though it is not clear (from an empirical point of view) what long term and short term mean i.e. after how many years the entries contribute to an increase in net employment. To know this, we require disaggregated information about the characteristics of entries and

2 See, among others, Dunne, Roberts, and Samuelson (1988), Evans (1987), Hall (1987), and Wagner (1994).

3 Most research on the survival of firms shows that the revolving door effect prevails over the displacement effect (Callejoin & Segarra, 1999).

4 Segarra and Callejon (2002) and Segarra et al. (2002b) show that the survival patterns of new Spanish manufacturing firms are similar to those of other countries.

exits and changes in market structure due to the turnover process.

The dynamic relationship between employment growth and firm entries suggests a time lag structure. The effects of entries on employment are then a result of job creation by entering firms, job destruction by exiting firms and both job creation and job destruction due to the interaction of the effects of entering and exiting firms. There are some examples of employment growth analyzed using lagged variables (see, for instance, Acs & Armington, 2004; Audretsch & Fritsch, 2002; Folster, 2000; Fritsch & Mueller, 2004; Fritsch et al., 2005; Van Stel & Storey, 2004) but most of research is conducted using no time lags or just a short lag.

3 Data and variables

To model the incidence of new firms on employment growth, we needed to know the gross rate of entry (GRE) and employment change (EC) in the Spanish manufacturing sector. So, the basic relationship to be modelled is:

EC = f (GRE)

To obtain the data to model this relationship, we considered two statistical sources: the Encuesta Industrial (the Industrial Survey; EI) and the Registro de Establecimientos Industriales (the Register of Manufacturing Establishments; REI). These databases allow us to use start-up data on the establishment level.

The data on employment (number of workers) in Spanish manufacturing industries came from the EI.5 The number of establishments created every year6 both at a regional and industry level7 came from the REI, which is an administrative register. The vari-

5 The interested reader is referred to, for example, Segarra et al. (2002b) and the references therein.

6 The data on establishments include the set-up of several branches by the same firm (the data is at the establishment level). Besides, there is not a minimum size of new establishments to be included.

7 The REI provides information about all new manufacturing establishments while the EI focuses on those firms with more than 10 employees (it also includes firms with less than 10 employees, but only as a sample). See Mompo and Monfort (1989) for further information about the REI.

ables included in our database are available for each region and manufacturing sector for each year between 1978 and 1996. The sectors and regions included are listed in Table 1.8

Specifically, and to avoid disturbances due to short-term fluctuations, the dependent variable is employment change (EC), which we computed as the growth rate of employment over 2 years, i.e.

(Employment — Employment^ 2)

Employment,

■t- 2

Following the labour market perspective,9 the independent variable (GRE) was calculated by dividing the number of establishments created every year for each region and manufacturing sector (Entriest) by the initial level of employment (Employment^), i.e.

Entries,

Employment

■t-1

As Table 1 shows, in the Spanish manufacturing sectors there are substantial differences in the gross rates of entry both at the sectoral and at the regional level. International empirical evidence shows that, even after controlling for differences in the industrial mix,10 there are substantial differences in the regional rates of entry (Keeble & Walker, 1994; Reynolds, Storey, & Westhead, 1994). This suggests (dis)econ-omies exist at the regional level that directly affect

Because of an excess of zero values, data on mineral extraction activities, as well as data on the Spanish region Extremadura, have been excluded. However, these exclusions do not affect the results of the analysis whatsoever.

9 The independent variable, i.e. the GRE, can be measured in three ways. The first way is known as ''labour market perspective'', where the number of workers is used to standardize entries. The second way is called the ''ecological perspective'', because the number of firms is used to standardize entries. The third way of calculating the entry rate is the ''population perspective'', where the population is used to standardize entries. Given that we assume that agents decide to set up a new firm in the labour market where they come from and where they have previous labour experience (Ashcroft, Love, & Malloy, 1991; Johnson, 1983; Kangasharju, 2000; Keeble & Walker, 1994; Storey & Jones, 1987) we have chosen the labour market perspective.

10 Some scholars use the shift-share procedure to obtain a sector-adjusted entry rate (see Ashcroft et al., 1991, for a more detailed explanation).

Table 1 Sectoral and regional rates of entry (average 19801994)

Sectors (NACE-25)

GRE (labour market approach)

Mineral products .32

Chemical products .14

Metal products .70

Ag./Ind. machinery .41

Electrical goods .26

Transport equipment .10

Food/Bev./Tob. .37

Textiles .42

Paper/Printing .42

Rubber/Plastic .44

Other manufacturing 1.40

Total manufacturing .43 Regions (NUTS-2)

Andalusia (AND) .65

Aragon (ARA) .45

Asturias (AST) .40

Balearic Islands (BAL) .72

Canary Islands (CANAR) .78

Cantabria (CANT) .29

Castile-Leon (CLEON) .42

Castile-la Mancha (CMANCHA) .68

Catalonia (CAT) .32

Valencia (VAL) .65

Galicia (GAL) .40

Madrid (MAD) .40

Murcia (MUR) .65

Navarre (NAV) .24

Basque Country (BASQ) .20

La Rioja (RIO) .35

Spain (Total) .43

Source: Register of manufacturing establishments REI and industrial survey EI

the decision to enter. In the Spanish case, for example, Segarra, Arauzo, Manjon, and Martin (2002a) show that entries are not randomly distributed over the Spanish regions (see also Table 1). Rather, there is a close relationship between economic growth and the rates of entry (positive) and exit (negative). At any rate, the existence of different growth regimes (Fritsch, 2004) suggests that policies that attempt to stimulate growth should take these regional specific growth regimes into account.

An analogous behaviour is present in the evolution of the dependent variable, i.e. employment change over time (EC). Figure 1 summarizes the evolution of this variable, which we will attempt to link to business creation. We found that the economic cycles that characterized production (gross added value: GAV) are also present in employment growth. There is therefore a close link between economic growth (recession) and the creation (destruction) of employment.

However, it is also important to consider the cyclical evolution of the Spanish economy during the period analyzed. Specifically, between 1978 and 1996, and taking into account the average growth, we can distinguish three stages. The first stage covers the period of readjustment in manufacturing between 1978 and 1985. During this period, the Spanish economy suffered the second energy crisis (1979) and stagnation in industrial production and investment (1979-1982). Also, several economic, political and institutional reforms were implemented to restructure production (1982). One of these was a thorough industrial restructuring with important adjustments in employment. The second stage, which includes the country's integration into the EEC (European Economic Community) in 1986, covers a period of growth that ended in the late 1980s. The third stage covers a period of recession characterized by a decrease in economic activity and employment, followed by a slight recovery at the end of the period.

This cyclical behaviour has had several implications for industrial labour and gross added value in the different Spanish regions, while the various manufacturing sectors have experienced repercussions of different intensities. From a territorial and sectoral

point of view, the overall effect between 1978 and 1996 is shown in Figs. 2 and 3, respectively.

Figure 2 analyses the regional effects on industrial labour and gross added value growth. One result that is common to most of the regions as shown in Fig. 2, is the decline of employment due to industrial restructuring in the period analyzed. However, value added, measured in constant prices, experienced a positive growth in all regions (the only exception being the Canary Islands). These results therefore show a positive impact on regional productivity in the manufacturing sector.

Figure 3 shows that for our sample the average annual growth in gross added value was 3% throughout the 1978-1996 period. This general growth hides several sectoral differences, however. For instance, paper and printing products had the highest average annual growth in added value (4.98%), while textiles had the lowest (1.0%). Over the same period, labour decreased. Clothing, for example, had the highest decrease (on average, —2.55% per year). Most sectors experienced a decline of employment but also a positive growth in gross added value. These sectors therefore experienced improvements in labour productivity. On the other hand, three sectors suffered a drop in both employment and gross added value. However, as in these sectors the drop in employment was higher than the recession in activity, the effect on labour productivity was positive.

4 Model and results

As stated earlier, we have used data on firm entry to assess the dynamic relationship between firm entries and employment change in the Spanish manufacturing sector. The model we present is similar to the one used by other similar empirical contributions (Fritsch & Mueller, 2004):

ECit = a + b0GREit + ^GRE^— 1 + •••

+ bs GREit—s + Uit (3)

In this model we consider data at both the regional and the temporal level.11 Each variable therefore has

11 In this model we are interested in the effect of firm entries on

the overall level of industrial employment of a region and do not consider each specific sector.

Fig. 2 Average annual growth rates in regions: labour and gross added value (base = 1990). 19781996. Source: Industrial survey

Fig. 3 Average annual growth rates in manufacturing industries: labour and gross added value (base = 1990). 19781996. Source: Industrial survey

two subscripts: i denotes region and t denotes year. The dependent variable is Employment Change (EC), and the independent variables are the current and the past values of the GRE.12 Since the industrial mix varies between regions and the relative importance of new firms and incumbents varies between industries, we used a shift-share procedure to obtain a sector-adjusted measure of new firm start-ups (Ashcroft et al., 1991; Audretsch & Fritsch, 2002). This procedure adjusts the data by imposing the same industrial mix on each region.

Notice that, in principle, current and past values of the GRE affect employment change. Therefore, the number of lags (s) determines the number of periods during which the effect of the rate of entry on employment change occurs. To estimate the model (3)

12 We also tried to include a control variable for the business cycle, but it did not fit in the estimation. Besides, the basic results obtained here did not change, and therefore we excluded it.

we have to give the parameter s a value. Since we aim to assess the long-term effect of entries on employment, we set the value s = 7, i.e. a model with seven lags,13 in accordance with similar studies. Table 2 shows the results of the estimation of (3), including the contemporaneous regressor as well as the first seven time lags (the impact of each regressor lag is also analyzed separately). The estimation technique used was fixed effects allowing for both regional and time effects,14 and the standard error estimates were obtained using the cross-section White method.

The results of the regression including all gross rates of entry and the separate regressions for each lag of the GRE are very similar, yielding a positive

13 Although we also tried a larger number of lags, we chose to use seven lags because the results are quite similar. However, the efficiency of the estimation decreases as the number of lags increases.

14 For estimation methods of panel data models with two-way error component disturbances, see Baltagi (2001).

short-term effect in years t and t — 1, a negative effect in year t — 2 and a positive effect in years t — 3 and t — 4. In the last years the effect of entries on employment change seems to vanish. The positive short-term effect reflects the direct employment creation caused by the entry of new firms, whereas the negative medium-term effect is likely to be caused by the exit of firms as a result of the previous entry of new firms.15 The positive effect observed for years t — 3 and t — 4 is likely to be related to indirect supply-side effects caused by entrants, i.e. improvements in efficiency, structural change and innovation (Fritsch & Mueller, 2004). These lag structures are shown in Fig. 4.

Models that include several lags of the same variable, as in (3), are likely to suffer from a multicollinearity problem, which makes the interpretation of the coefficients unreliable. To solve this problem, we impose a structure on the lag distribution by applying the polynomial distributed lag model.16 This method solves the problem of multicollinearity in distributed lag models by imposing a structure on the lag coefficients. We assume that the effects of entries on employment change are distributed over 7 years because the previous results show that in this period the effects have already vanished. Table 3 shows the results of the polynomial distributed lag estimation considering a second, third, fourth and fifth order of the polynomial. In this type of estimation, both regional and time effects are accounted for, and the standard error estimates are obtained, as in the previous estimation, using the cross-section White method. Figure 5 shows the graphical lag structures resulting from the different polynomial orders considered.

The lag structure of the second order polynomial is approximately a U-shaped structure, whereas the lag structures of the other polynomial orders (third, fourth and fifth) are quite similar, showing a pattern

15 With regard to this negative effect on the short-term, empirical papers on firm entry suggest that the average size of new firms is smaller than the average size of incumbent firms (the size distribution of new cohorts is more skewed than market structure: see Arauzo and Segarra, 2005, for a detailed analysis of start-up size for the Spanish case). This is why a post-entry size adjustment is very important in manufacturing markets, especially in the first few years when suboptimal size affects a lot of newcomers and selection is very painful.

16 This model is also known as the Almon lag model.

2) in

.2 (2.

f (2. 7 co * .9 m .9 oo 7 .2

* (2. * ' 1 (2.

CO * 00 6 in 2 (2 6

in 6. .7 00 00

<N m I I

in \o I I

on RE CG

S if 2. 2

H In 22

in in 2

( t( t( ( ( ( ( b

E E E E E E E m

OOOOOOOceh, £

Fig. 4 Labour market approach: lag structure of the impact of new business formation on regional employment growth resulting from the joint estimation (left) and from separate estimations (right)

Table 3 Polynomial distributed lag estimation of Eq. 3 (dependent variable: employment change (EC))

Almon method assuming a polynomial of

Second order Third order Fourth order Fifth order

Constant -.04 (.04) -.04 (.04) -.04 (.04) -.04 (.04)

a0 .47 (1.87) -.53 (1.73) -.09 (2.01) -.45 (2.17)

a1 -.63 (.48) 1.69* (.86) 2.03** (1.02) 2.69* (1.38)

a2 .14 (.21) .53** (.19) .19 (.60) .60 (.87)

a3 -.27** (.08) -.32** (.10) -.57 (.36)

a4 .02 (.04) -.01 (.07)

a5 .01 (.02)

GRE (t) 3.70 6.59 6.59 6.29

GRE (t - 1) 2.33 .41 -.32 .25

GRE (t - 2) 1.25 -1.41 -1.57 -2.00

GRE (t - 3) .47 -.53 -.09 -.45

GRE (t - 4) -.01 1.42 1.83 2.26

GRE (t - 5) -.21 2.82 2.58 3.05

GRE (t - 6) -.11 2.01 1.19 .62

GRE (t - 7) .27 -2.61 -2.65 -2.29

R2 .32 .37 .37 .37

F 2.40** 2.75** 2.65** 2.58**

Log-likelihood 279.33 284.39 284.63 285.00

Number of obs. 160 160 160 160

Note: Standard errors appear in parenthesis

** and * mean that the estimated coefficients are significant at a 5% and 10% significance level, respectively

also found in the previous estimations. According to the values of the Log-likelihood function, as well as the significance of the Almon coefficients, the structure of the second order polynomial can be rejected in favour of the structure of the third, fourth and fifth order polynomials. These results confirm the interpretation of the lag structure proposed earlier, i.e. the direct effect of entries on employment change is positive between t and t — 1, becomes negative between t — 2 and t — 3, and is positive again between t — 4 and t — 6. The magnitude of the effect decreases and is negative in the last year (t — 7). The

F statistic is significant in all the estimation, so these results can be regarded as reliable.

Our results are similar to those obtained by Fritsch and Mueller (2004) for the German case and Baptista, Escaria, and Madruga (2005) for the Portuguese case, in the sense that entries cause positive indirect effects in the form of increased occupation. These authors, however, consider a longer period of time in their analysis and find that the indirect supply side effects take place later in time. This result might derive from differences in the industrial structure of the countries, but differences in the aggregation and quality of data

Fig. 5 Labour market approach: lag structures of the impact of new business formation on regional employment change resulting from the Almon lag model estimations

may also have an influence.17 These studies also conclude that the indirect supply-side effects of entries contribute to employment growth more than the direct effects associated with jobs created by entrants. These results show that future research must focus on why the lag effects on employment differ between countries. For the British case, for instance, Van Stel and Storey (2004) conclude that the strongest effect on employment growth occurs about five years after firms enter. Fritsch and Mueller (2004) analyze business formation in West German Kreise (districts) and emphasize the role of the indirect supply-side effects of entries, which are more important (in terms of employment growth) than the number of jobs directly created by the new firms. Specifically, they conclude that these are not short-term effects (the exits of previous entrants could actually destroy net employment) but long-term effects. Also for the German case, Audretsch and Fritsch (2002) show that firm entries between 1983

17 At this point we should insist on the fact that we use data from manufacturing industries only, while most of research in this area uses data for the whole economy, including both manufacturing and services activities. Given the specific industry effects, there can be differences between our results and the other contributions based on the whole economy.

and 1985 could help to explain employment growth between 1993 and 1998. They found that the effects of entrants on employment growth take more than a decade to become evident, though other studies of the German case showed no relation between entries and lagged employment growth (Audretsch & Fritsch, 1996; Fritsch, 1996, 1997).

5 Discussion and implications

In this paper we have explored the incidence of new business formation on employment growth in the Spanish manufacturing industries. We have specified a distributed lag model where both the GRE in level and its time lags affect employment change. We have estimated this model by including all the lags from t to t — 7, then including each lag separately and imposing a structure on the coefficients using the polynomial distributed lag model, which is also known as the Almon lag model. The results from the various estimations are quite similar: a positive short-term effect, a negative medium-term effect and a positive long-term effect. These results are similar to those of other studies that also estimate distributed lag models.

These results are interesting for our analysis of the effectiveness of entry-promoting policies—particu-

larly for determining whether these policies are successful in the short or long term. Also from this point of view, it is important to measure how far new firms contribute to employment growth because the amount of public funds intended for policies aimed at increasing firm entry depend on this contribution. If this is measured according to specific industrial factors or according to individual characteristics of entrants (such as size, capital resources, or the skill level of employees and entrepreneurs), more suitable policies could be designed to concentrate resources on these job-generating start-ups (Fritsch, 1997). The public promotion of firm entries may be justified in order to safeguard the quality of market selection (Fritsch & Mueller, 2004) and promote gross added value growth and innovation, but, as Fritsch et al. (2005, p. 548) point out, ''market exits are necessary elements of market selection and that policy should abstain from subsidizing firms in order to prevent them leaving the market''.

It should be borne in mind, however, that we did not have enough time series data to test the incidence of the business cycle on firm entry and job creation. Further research into the Spanish case should explore the incidence of the business cycle on the lagged effect of firm entry on net job creation. Policy makers need to answer a key question: what are the most efficient entries in terms of lagged job creation— those that correspond to periods of growth or those that correspond to periods of decline?

Despite these conclusions, more work is needed in this area. Here we have worked at a regional level, but these administrative units are maybe too large to cover real economic areas. Given that space is an important determinant of firm entries, one solution could be to use a more disaggregated area, such as provinces, to also allow spatial econometric techniques to enter this analysis. We could then link the topic analyzed to the location of firms, assess the consequences of agglomeration in terms of employment and productivity, and analyze the role of geography in this phenomenon. Another future extension of this work should be to take into account specific industry effects. Here we have considered all manufacturing activities but there are some particular effects within industries. On this

18 See Arauzo, Manjon, Martin, and Segarra (2007) for a more detailed analysis on regional determinants of industry dynamics in Spain.

point, for instance, Acs and Armington (2004) found that the effect of entrepreneurial activity on employment change varies across industries. Specifically, they show that results for the manufacturing industries as a whole deviate from the results of all other sectors, since for the formers there is no impact of entrepreneurial activity over employment change, while this effect is positive for the laters. If those differences are explained in terms of industry-specificities, an analysis at manufacturing disaggregated level could show also those specific industry effects.

Since employment change does not describe the full effect of the efficiency provided by new entrants, a more complete analysis should include both the effect of new entrants on employment change and their effect on productivity. This analysis may show that entries are employment-reducing but productivity-enhancing. This would mean that entries introduce labour-saving innovations or take place mainly in capital-intensive sectors. Such results cannot be obtained without considering a measure of productivity. Several papers have tried to assess the effect of firm entry and exit on productivity for the Spanish case. Martin and Jaumandreu (2004) found that during the 1980s competitive entry accounted for 80% of productivity growth, while Farinas and Ruano (2004) found that the replacement of exiting firms by entering firms made a positive contribution to the dynamics of Total Factor Productivity growth. If we link these results to those of our paper, we may conclude that, in the long term, entries foster both employment and productivity growth. However, more research is needed to confirm these results.

Acknowledgements The authors are grateful to the CICYT (SEJ2004-05860/ECON and SEJ2004-07824/ECON). We also would like to acknowledge the helpful and supportive comments of Michael Fritsch, Miguel Manjon, Enrique Lopez Bazo, seminar participants at the ''Effects of New Businesses on Economic Development in the Short, Medium and Long Run'' Workshop (Max Planck Institute of Economics) and two reviewers. The usual disclaimer applies.

References

Acs, Z. J., & Armington, C. (2004). Employment growth and entrepreneurial activity in cities. Regional Studies, 38, 911-927.

Agarwal, R., & Audretsch, D. (2001). Does entry size matter? The impact of the life cycle and technology on firm survival. The Journal of Industrial Economics, 49(1), 21-43.

Arauzo, J.-M., & Segarra, 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.

Arauzo, J.-M., Manjon, M. C., Martin, M., & Segarra, A. (2007). Regional and sector-specific determinants of industry dynamics and the displacement effect. Empirica.

Ashcroft, B., Love, J. H., & Malloy, E. (1991). New firm formation in the British counties with special reference to Scotland. Regional Studies, 25(5), 395-409.

Audretsch, D. (1995). Innovation and industry evolution. Cambridge: The MIT Press.

Audretsch, D., & Fritsch, M. (1996). Creative destruction: turbulence and economic growth. In E. Helmstadter & M. Perlman (Eds.), Behavioral norms, technological progress and economic dynamics: Studies in schumpeterian economics (pp. 137-150). Ann Arbor: University of Michigan Press.

Audretsch, D., & Fritsch, M. (2002). Growth regimes over time and space. Regional Studies, 36, 113-124.

Audretsch, D., & Mahmood, T. (1995). New firm survival: New results using a hazard function. Review ofEconomics and Statistics, 77(1), 97-103.

Audretsch, D., Santarelli, E., & Vivarelli M. (1999). Start-up size and industrial dynamics: Some evidence from Italian manufacturing. International Journal of Industrial Organization, 17, 965-983.

Aw, B. Y., Chen, X., & Roberts, M. J. (1997). Firm-level evidence on productivity growth differentials and turnover in Taiwanese manufacturing. Journal of Development Economics, 66, 51-86.

Baldwin, J. (1995). The dynamics of industrial competition. Cambridge, MA: Cambridge University Press.

Baldwin, J., & Gorecki, P. (1991). Entry and exit and productivity growth. In P. A. Geroski & J. Schwalbach (Eds.), Entry and market contestability: An international comparison. Oxford: Basil Blackwell.

Baltagi, B. H. (2001). Econometric analysis of panel data. Chichester, England: John Wiley & Sons.

Baptista, R., Escaria, V., & Madruga, P. (2005). Entrepre-neurship, regional development and job creation: The case of Portugal. Discussion papers on entrepreneurship, growth and public policy #0605-2005. Jena: Max Planck Institute for Research into Economic Systems.

Callejon, M., & Segarra, A. (1999). Business dynamics and efficiency in industries and regions: The case of Spain. Small Business Economics, 13, 253-271.

Caves, R. E. (1998). Industrial organization and new findings on the turnover and mobility of firms. Journal of Economic Literature, 36, 1947-1982.

Dunne, P., & Hughes, A. (1994) Age, size, growth and survival: UK companies in the 1980s. Journal of Industrial Economics, 42(2), 115-140.

Dunne, T., Roberts, M., & Samuelson, L. (1988). Patterns of firm entry and exit in U.S. manufacturing industries. Rand Journal of Economics, 19, 495-515.

Evans, D. (1987). The relationship between firm growth, size, and age: Estimates for 100 manufacturing industries. Journal of Industrial Economics, 35, 567-581.

Fariñas, J. C., & Ruano, S. (2004). The dynamics of productivity: A decomposition approach using distribution functions. Small Business Economics, 22, 237-251.

Folster, S. (2000). Do entrepreneurs create jobs? Small Business Economics, 14, 137-148.

Fritsch, M. (1996). Turbulence and growth in West Germany: A comparison of evidence by regions and industries. Review of Industrial Organisation, 11, 231-251.

Fritsch, M. (1997). New firms and regional employment change. Small Business Economics, 9, 437-448.

Fritsch, M. (2004). Entrepreneurship, entry and performance of new business compared in two growth regimes: East and West Germany. Journal of Evolutionary Economics, 14, 525-542.

Fritsch, M., & Mueller, P. (2004). Effects of new business formation on regional development over time. Regional Studies, 38(8), 961-975.

Fritsch, M., Mueller, P., & Weyh, A. (2005). Direct and indirect effects of new business formation on regional employment. Applied Economics Letters, 12, 545-548.

Geroski, P. (1989). Entry, innovation, and productivity growth. Review of Economics and Statistics, 71, 572-578.

Hall, B. (1987). The relationship between firm size and firm growth in the U.S. manufacturing sector. Journal of Industrial Economics, 35, 583-606.

Johnson, P. S. (1983). New manufacturing firms in the UK regions. Scottish Journal of Political Economy, 30, 75-79.

Kangasharju, A. (2000). Regional variations in firm formation: Panel and cross-section data evidence from Finland. Papers in Regional Science, 79, 355-373.

Keeble, D., & Walker, S. (1994). New firms, small firms and dead firms: Spatial patterns and determinants in the United Kingdom. Regional Studies, 28(4), 411-427.

Martín, A., & Jaumandreu, J. (2004). Entry, exit and productivity growth: Spanish manufacturing during the eighties. Spanish Economic Review, 6, 211-226.

Mata, J., & Portugal, P. (1995). Life duration of new firms. Journal of Industrial Economics, 42(3), 227-245.

Mata, J., & Portugal, P. (1999). Technology intensity, demand conditions, and de longevity of firms. In D. B. Audretsch & A. R. Thurik (Eds.), Innovation, industry evolution and employment (pp. 265-279). Cambridge: Cambridge University Press.

Mompo, A., & Monfort, V. (1989). El Registro Industrial como fuente estadística regional: El caso de la Comunidad Valenciana. Economía Industrial, 268, 129-140.

Reynolds, P., Storey, D., & Westhead, P. (1994). Cross-national comparisons of the variation in new firm formation rates. Regional Studies, 28(4), 443-456.

Segarra, A., Arauzo, J.-M., Manjon, M. C., & Martín, M. (2002a). Demografía industrial y convergencia regional en España. Papeles de Economía Española, 93, 65-78.

Segarra, A. (dir.), Arauzo, J.-M., Gras, N., Manjon, M. C., Mane, F., Teruel, M., & Theilen, B. (2002b). La creación y la supervivencia de las empresas industrials. Madrid: Editorial Cívitas.

Segarra, A., & Callejón, M. (2002). 'New firms' survival and market turbulence: New evidence from Spain. Review of Industrial Organization, 20, 1-14.

Storey, D., & Jones, A. M. (1987). New firm formation-a labour market approach to industrial entry. Scottish Journal of Political Economy, 34, 37-51.

Tybout, J. R. (1996). Heterogeneity and productivity growth: Assessing the evidence. In M. J. Roberts & J. R. Tybout (Eds.), Industrial evolution in developing countries: Micro patterns of turnover, productivity and market structure. New York: Oxford University Press.

Van Stel, A., & Storey, D. (2004). Link between firm births and job creation: Is there a Upas tree effect? Regional Studies, 38, 893-909.

Wagner, J. (1994). The post-entry performance of new small firms in German manufacturing industries. The Journal of Industrial Economics, 42(2), 141-154.