3.1 Data and methodology
Data for the total early-stage entrepreneurial activity including necessity- and opportunity-driven entrepreneurship is obtained from the Global Entrepreneurship Monitor (GEM) while information about Greenfield is extracted from the World Investment Report (UNCTAD). We assess the role of Greenfield investment in determining the rate of entrepreneurial activity in the period from 2005 to 2011.
The panel data includes 110 countries for the period 2001-2018 (695 observations). Data related to entrepreneurial activity are produced by Global Entrepreneurship Monitor (GEM) while the dataset on macroeconomic variables including Greenfield investment are obtained from UNCTAD. Table 1 provides the descriptive statistics of the variables used in the paper.
3.1.1. Fixed and random effects
The fixed-effect panel regression model is the preferred choice in evaluating the explanatory variables of entrepreneurial activity because the within model (or fixed effect) is able to disentangle the variations among different cross sections of countries (Albulescu and Tămăşilă, 2013). However, we also include random model in the regression results because the number of countries (T=110) is much higher than the number of years (N=18). A Hausman test is also utilized in order to compare and contrast between fixed-effect and random-effect equations.
The fixed effect panel model is usually used for assessing the entrepreneurship determinants. Fixed effects underline disparities between countries. A new development of this classical model is the panel negative-binomial model, which accounts for violations in the assumption of homoscedasticity and, in the same time, provides the flexibility to address unobserved heterogeneity (Hausman et al., 1984). Nevertheless, as Allison & Waterman (2002) show, this method does not, in fact, control for all stable covariates. We then test a simple fixed effect model, but also a random model, having in mind the fact that the structure of our sample shows a N<T situation (the number of countries is higher that the number of periods). In addition, for the robustness check we do not have strongly balanced panels (lack of data for the beginning of the period) and the random-effects models address these aspects. In order to avoid the broken panel problem, when entrepreneurship data were missing (Germany - 2007; Ireland - 2009; Sweden - 2008, 2009 and Switzerland - 2006, 2008), we have used the linear interpolation. A Hausman test was performed in order to select the most appropriate model between the fixed and random effects. The general tested equations for fixed and respectively for random effects are:
Yi,t = β0 + β1 Greenfield + β2Xi,t + αi + ei,t (1)
In which: Yi,t is the explained variable (tea, necessity, opportunity); β0 is the constant; αi represents all the constant time-invariant characteristics of the countries; Xi,t represents the vector of independent variables; β1 and β2 are the coefficients; ei,t is the error term.
Yi,t = β0 + β1 Greenfield + β2Xi,t + µi + ei,t (2)
In which: µi represents between-entity errors; ei,t is the within-entity error.
3.1.2. GMM estimated method
Most of the previous studies conducted on the influence of foreign investment on domestic entries use panel data fixed/random effect models or least square dummy variable (LSDV) models. These kinds of models are based on a strong assumption that the effect across the panel is homogeneous. With the estimation of the random and fixed effect panel models, some specification issues are expected, the first and foremost of which is the endogeneity issue of Greenfield investment. A wide variety of techniques and different variables have been adopted to mitigate the endogeneity issue of foreign direct investment. Lagged value of Greenfield will be used as an instrument for Greenfield to deal with the problem (Alfaro et al. 2004). The main reason is that foreign direct investment is expected to reinforce itself from time to time (Wheeler and Mody, 1992). Therefore, the endogeneity issue would be best dealt with by the dynamic panel model that includes lagged value of FDI as an instrumental variable. The GMM estimated method will be used on data for the total entrepreneurial activity and the opportunity- and necessity-based components.
This paper uses a dynamic panel data model and generalized method of moments (GMM) estimation method to model the impact of Greenfield investment on domestic entrepreneurship. The dynamic panel data model and GMM estimation deal with the potential FDI endogeneity problem.
The following dynamic panel data models is estimated to model the impact of Greenfield on domestic entrepreneurship.
Yi,t = β0 + β1 Yi,t-1 + β2 Greenfieldi,t + β3 Xi,t + ei,t (3)
where Yi,t is the total entrepreneurial activity, Yit–1 is the lagged value of the total entrepreneurial activity. Greenfieldi,t is the total of net Greenfield investments inflow to the host country. Xi,t represents the control variables for the determinants of total entrepreneurial activity including gross domestic product growth rate, entrepreneurial intentions and the fear of failure rates and ei,t is the random error term. Adopting from existing similar studies (for example Albulescu and Tămăşilă 2013), we used fear of failure rates, GDP growth rate and entrepreneurial intentions as explanatory variables of total entrepreneurial activity.
3.2 Empirical results and discussion
We have conducted one main overall test and three sets of robustness checks. The main test has been performed in the total dataset of 110 countries over the 18-year period from 2001 to 2018 (Panel A). The other sets of analysis consist of similar methodology over the same number of panel, but the sample period has been split into pre-crisis (Panel B), during crisis (Panel C) and post-crisis (Panel D). For each set of tests, both fixed-effect and random-effect models have been shown along with the Hausman test (to determine whether the fixed-effect or random-effect is more preferred). In addition to the Panel on total entrepreneurial activity, we have also made distinction between necessity-driven entrepreneurship and opportunity-driven entrepreneurship.
Table 1: Descriptive statistics
|
Mean
|
Sd
|
Min
|
Max
|
gdpgrowth
|
3.82
|
5.17
|
-62.08
|
123.14
|
gdpcapita
|
8.46
|
1.54
|
4.72
|
12.15
|
lngreen
|
6.81
|
2.65
|
-1.61
|
12.41
|
fof
|
34.13
|
9.24
|
12.3
|
75.4
|
ei
|
18.73
|
15.28
|
0
|
90.95
|
opportunity
|
7.9
|
5.35
|
0.81
|
31.89
|
necessity
|
2.96
|
2.93
|
0.09
|
19.55
|
tea
|
11.45
|
7.66
|
1.48
|
52.11
|
Table 1 above presents descriptive statistics for all the variables. The summary indicates a great deal of variations within each variable with GDP growth rate (gdpgrowth) ranging from -62% to 123% while its average only at 4%. The same is also true for Greenfield investment so we take log of this variable (lngreen). Fear of failure rate (fof) represents the percentage of population who indicates that fear of failure leads them not to open a business while entrepreneurial intentions (ei) refers the percentage of population who intend to open a new venture and this figure varies significantly from 0% to 90%. As predicted, the sum of opportunity-based entrepreneurship (opportunity) and necessity-based entrepreneurship (necessity) are equal to total entrepreneurship (tea) in every single statistic, which is actually observed in Table 1.
Table 2: Correlation matrix
|
gdpgrowth
|
fof
|
ei
|
tea
|
lngreen
|
gdpcapita
|
gdpgrowth
|
1.00
|
|
|
|
|
|
fof
|
-0.10
|
1.0
|
|
|
|
|
ei
|
0.25
|
-0.27
|
1.00
|
|
|
|
tea
|
0.24
|
-0.31
|
0.82
|
1.00
|
|
|
lngreen
|
-0.13
|
0.26
|
-0.52
|
-0.45
|
1.00
|
|
gdpcapita
|
-0.33
|
0.17
|
-0.65
|
-0.56
|
0.57
|
1.00
|
Table 2 correlation matrix shows no indication of multicollinearity and there is no significant correlation between lngreen and tea. However, the correlation between tea and ei is strong and statistically negative.
Table 3: Impact of Greenfield investment on necessity driven, opportunity-driven and total entrepreneurial activity.
|
tea
|
opportunity
|
necessity
|
|
(1)
Random
|
(2)
Fixed
|
(3)
GMM
|
(4)
Random
|
(5)
Fixed
|
(6)
GMM
|
(7)
Random
|
(8)
Fixed
|
(9)
GMM
|
teat-1
opportunityt-1
necessityt-1
|
|
|
0.2***
(0.03)
|
|
|
0.2***
(0.03)
|
|
|
0.31***
(0.03)
|
lngreen
|
-0.37***
(0.11)
|
-0.50***
(0.13)
|
-0.26***
(0.09)
|
-0.19**
(0.1)
|
-0.22*
(0.12)
|
-0.03
(0.08)
|
-0.07
(0.05)
|
-0.14**
(0.06)
|
-0.04
(0.04)
|
gdpcapita
|
-0.28
(0.32)
|
1.62***
(0.51)
|
-1.54
(2.53)
|
0.30
(0.26)
|
2.16***
(0.42)
|
-0.66
(1.97)
|
-0.80***
(0.1)
|
-0.48**
(0.22)
|
-1.7*
(0.001)
|
gdpgrowth
|
-0.04
(0.04)
|
-0.05
(0.04)
|
0.002***
(0.004)
|
-0.004
(0.03)
|
-0.01
(0.03)
|
0.001***
(0.004)
|
-0.03*
(0.02)
|
-0.04**
(0.02)
|
0.005**
(0.002)
|
fof
|
-0.04**
(0.02)
|
-0.03*
(0.02)
|
-0.07***
(0.02)
|
-0.04**
(0.02)
|
-0.03*
(0.02)
|
-0.05***
(0.01)
|
-0.02***
(0.007)
|
-0.01
(0.008)
|
-0.01*
(0.007)
|
ei
|
0.30***
(0.02)
|
0.22***
(0.02
|
0.3***
(0.02)
|
0.19***
(0.02)
|
0.14***
(0.02)
|
0.2***
(0.02)
|
0.09***
(0.007)
|
0.05***
(0.008)
|
0.08***
(0.008)
|
Constant
|
12.94***
(3.01)
|
-3.56
(4.85)
|
14.09
(10.8)
|
4.52*
(2.45)
|
-12.8***
(4.0)
|
7.02
(8.39)
|
10.09***
(1.15)
|
7.92***
(2.04)
|
8.28*
(4.31)
|
Observations
|
695
|
695
|
571
|
555
|
555
|
447
|
555
|
555
|
447
|
R-squared
|
0.53
|
0.24
|
|
0.44
|
0.18
|
|
0.55
|
0.11
|
|
Adjusted R2
|
0.52
|
0.12
|
|
0.44
|
0.03
|
|
0.54
|
0.05
|
|
F Statistic
|
767.22***
|
37.57***
(df = 5; 604)
|
163.55
(df = 6, 564)
|
438.01***
|
21.04***
(df = 5; 466)
|
105.22
(df = 6, 440)
|
661.76***
|
11.84***
(df = 5; 466)
|
129.72
(df =6, 440)
|
Note:
|
|
|
|
*p<0.1; **p<0.05; ***p<0.01
|
Table 3 shows the results of Panel A, which is the main model of our paper. As can be seen in the table, Greenfield investment exerts negative influence on the total entrepreneurial activity with the coefficients lngreen being negative in all settings except for the random-effect model on necessity entrepreneurship. Another noteworthy feature is that the lngreen coefficient shows lesser degree in terms of both magnitude and statistical significance in model (7), (8) and (9) that measure the impact of Greenfield on necessity entrepreneurship. Correspondingly, the inflows of Greenfield investment would have damaging impact on the rate of opportunity-based entrepreneurs while not necessarily decreasing the rate of necessity-based entrepreneurs. This is referred to as the crowding-out effect of FDI. The crowding-out effect of Greenfield could be explained by the replacement of entrepreneurship by employment. More specifically, setting up a new Greenfield venture in the foreign country may involves employing potential entrepreneurs, contributing to a reduction in the rate of potential future entrepreneurs (Goel, 2017). Along with another related dimension, foreign investors might also lure some potential entrepreneurs as employees. This would result in a reduction in domestic entrepreneurship (as employment replaces entrepreneurship). The overall impact of FDI on entrepreneurship might then be negative or positive and might vary across demographic groups. For example, female entrepreneurs might face special challenges in competing against foreign investors (Goel, 2017). In addition, Ashraf et al. (2016) and Calderón et al. (2004) both state that Greenfield investment bears crowding-out effect and hamper long-term economic growth. Since productive entrepreneurship is positively related to economic growth (Salgado-Banda, 2007), Greenfield investment would damage the long-term economic potential, which in turn decrease the rate of opportunity entrepreneurs.
Another interesting result is that high GDP growth rate and high GDP per capita are negatively associated with necessity-driven entrepreneurs while we find no significant evidence of the relationship between GDP growth rate or GDP per capita and opportunity entrepreneurship. This results are consistent with the current literature. Necessity-driven entrepreneurs often have poor education, run smaller businesses and their firms tend to lag behind others (Poschke, 2012). As soon as necessity-based entrepreneurs find better opportunities elsewhere, they will abandon their businesses (Rissman, 2003). When people have higher standard of living and sufficient livelihood, they will be less motivated to pursue entrepreneurship out of necessity (Poschke, 2012). Secondly, the direction and significance of the coefficients fof and ei are generally consistent across all models. More specifically, fear of failure (fof) would prevent entrepreneurs from setting up new ventures out of both opportunity and necessity. Next, entrepreneurial intention is the conscious state of mind that precedes entrepreneurial actions and positively direct people towards entrepreneurial behaviors (Moriano et al., 2012).
Table 4: Relationship between Greenfield investment and entrepreneurship in pre-crisis period
|
tea
|
opportunity
|
necessity
|
|
(1)
Random
|
(2)
Fixed
|
(3)
GMM
|
(4)
Random
|
(5)
Fixed
|
(6)
GMM
|
(7)
Random
|
(8)
Fixed
|
(9)
GMM
|
teat-1
opportunityt-1
necessityt-1
|
|
|
0.41***
(0.07)
|
|
|
0.53***
(0.09)
|
|
|
0.51***
(0.08)
|
lngreen
|
-0.38**
(0.18) |
-0.12(0.20) |
-0.24**
(0.12)
|
-0.35**(0.14) |
-0.18(0.15) |
-0.22*
(0.13)
|
0.002(0.07) |
0.14*(0.08) |
-0.02
(0.09)
|
gdpcapita
|
-0.76(0.60) |
-1.45(1.27) |
-0.02
(0.1)
|
-0.11(0.48) |
0.18(0.95) |
0.04
(0.08)
|
-0.57***(0.20) |
-1.20**(0.51) |
0.06*
(0.04)
|
gdpgrowth
|
0.15(0.12) |
0.18(0.14) |
-0.00**
(0.00)
|
0.13(0.09) |
0.10(0.11) |
0.014*
(0.08)
|
0.01(0.05) |
0.07(0.06) |
0.01*
(0.11)
|
fof
|
-0.0003(0.04) |
0.01(0.04) |
-0.09***
(0.03)
|
-0.02(0.03) |
-0.01(0.03) |
0.07***(0.02)
|
0.01
(0.01)
|
0.02
(0.02)
|
-0.005
(0.675)
|
ei
|
0.26***(0.04) |
0.16***(0.05) |
0.24***
(0.04)
|
0.14***(0.03) |
0.09**(0.04) |
0.12***
(0.03)
|
0.11***(0.02) |
0.06***(0.02) |
0.06***
(0.02)
|
Constant
|
15.25**(6.14)
|
|
6.91***
(1.39)
|
8.55*(4.89)
|
|
5.53***
(1.43)
|
5.52**(2.15)
|
|
-4.2*
(2.43)
|
R-squared
|
0.53 |
0.18 |
|
0.39 |
0.09 |
|
0.61 |
0.4
|
|
F Statistic
|
141.77***
|
3.56*** |
60.14
(df = 6, 75)
|
78.04***
|
1.65
|
27.66
(df = 6, 74)
|
195.85***
|
|
84.62
(df = 6, 74)
|
Observations
|
82
|
82
|
82
|
82
|
82
|
82
|
82
|
82
|
82
|
Note:
|
|
|
|
*p<0.1; **p<0.05; ***p<0.01
|
Similar findings have also been found in the pre-crisis period as can be seen in Table 4. The inflows of Greenfield investment significantly lower the rate of total entrepreneurship and opportunity-driven entrepreneurship. The distinct finding is that Greenfield may actually increase necessity-driven entrepreneurs with coefficient lngreen being significantly positive in model 8. This could be explained by the fact that the entry of Greenfield venture might drive up competition pressure for domestic firms, forcing them out of market and thereby decreasing employment in host economy (Karlsson, 2007). Giant MNEs may raise the average wages in the local narrowly-defined industry and then qualify job growth in the host countries. Subsequently, there would be a higher number of potential entrepreneurs who start up new business upon losing their jobs.
Table 5: Relationship between Greenfield investment and entrepreneurship during crisis
|
tea
|
opportunity
|
necessity
|
|
(1)
Random
|
(2)
Fixed
|
(4)
Random
|
(5)
Fixed
|
(7)
Random
|
(8)
Fixed
|
lngreen
|
-0.03(0.22) |
0.14(0.23) |
0.07(0.16) |
0.24(0.18) |
-0.06(0.10) |
0.02(0.13) |
gdpcapita
|
-2.50***(0.85) |
5.80**(2.46) |
-1.10*(0.62) |
5.27***(1.86) |
-1.07***(0.33) |
0.80(1.42) |
gdpgrowth
|
-0.15**(0.07) |
-0.02(0.07) |
-0.11**(0.05) |
-0.01(0.05) |
-0.06(0.04) |
-0.04(0.04) |
fof
|
-0.002(0.04) |
-0.08*(0.04) |
0.01(0.03) |
-0.05(0.03) |
-0.01(0.02) |
-0.03(0.02) |
ei
|
0.10*(0.06) |
-0.09(0.06) |
0.07*(0.04) |
-0.06(0.05) |
0.06**(0.02) |
-0.05(0.04) |
Constant
|
33.36***(8.41) |
|
15.75**(6.16) |
|
13.15***(3.41) |
|
R-squared
|
0.39 |
0.37 |
0.27 |
0.44 |
0.50 |
0.24 |
F Statistic
|
41.86*** |
2.78** |
23.45*** |
3.76** |
63.97*** |
1.51 |
Observations
|
73
|
73
|
71
|
71
|
71
|
71
|
Note:
|
|
|
|
*p<0.1; **p<0.05; ***p<0.01
|
Regarding entrepreneurship during turbulent times, we see no significant relationship between entrepreneurship and Greenfield investment during crisis in Table 5. Entrepreneurial activity is also unrelated to fear of failure and entrepreneurial intentions. Meanwhile, the other coefficients such as GDP per capita and GDP growth rate show contradictory results in between fixed-effect and random-effect model. We are unable to carry out GMM estimation method in the crisis period due to insufficient observations. Thus, we cannot generalize the results regarding the impact of Greenfield investment during the turbulent times. Future research could resolve the data issue by conducting studies at the country level to examine the distinct impact Greenfield investment may have on different country settings.
Table 6: Relationship between Greenfield investment and entrepreneurship after-crisis period
|
tea
|
opportunity
|
necessity
|
|
(1)
Random
|
(2)
Fixed
|
(3)
GMM
|
(4)
Random
|
(5)
Fixed
|
(6)
GMM
|
(7)
Random
|
(8)
Fixed
|
(9)
GMM
|
teat-1
opportunityt-1
necessityt-1
|
|
|
-0.42***
(0.08)
|
|
|
-0.39***
(0.1)
|
|
|
0.47***
(0.12)
|
lngreen
|
-0.24*(0.13) |
-0.53***(0.16) |
-0.42**
(0.12)
|
-0.06(0.12) |
-0.21(0.16) |
-0.24
(0.28)
|
-0.02(0.06) |
-0.20**(0.08) |
-0.12
(0.14)
|
gdpcapita
|
-0.93***(0.35) |
0.30(1.10) |
-0.15**
(0.07)
|
-0.33(0.29) |
2.53**(1.12) |
-0.06
(0.08)
|
-0.94***(0.14) |
-2.17***(0.57) |
-0.06
(0.04)
|
gdpgrowth
|
0.01(0.05) |
-0.03(0.05) |
0.34***
(0.18)
|
0.05(0.04) |
0.01(0.04) |
0.06***
(0.09)
|
-0.02(0.02) |
-0.01(0.02) |
-0.37***
(0.1)
|
fof
|
-0.07***(0.02) |
-0.05**(0.02) |
0.11***
(0.05)
|
-0.06***(0.02) |
-0.02(0.02) |
0.14***
(0.04)
|
-0.03***(0.01) |
-0.02(0.01) |
0.04**
(0.02)
|
ei
|
0.28***(0.02) |
0.21***(0.02) |
0.18***
(0.05)
|
0.18***(0.02) |
0.10***(0.02) |
0.06
(0.05)
|
0.09***(0.01) |
0.05***(0.01)
|
0.03
(0.03)
|
Constant
|
19.36***(3.37)
|
|
7.82***
(1.89)
|
10.24***(2.82)
|
|
5.71***
(2.22)
|
11.59***(1.33)
|
|
9.14***
(3.11)
|
R-squared
|
0.54
|
0.21
|
|
0.47
|
0.12
|
|
0.57
|
0.12
|
|
F Statistic
|
576.76*** |
21.11***
|
|
308.91*** |
7.12***
|
|
452.06*** |
7.50*** |
|
Observations
|
491
|
491
|
373
|
354
|
354
|
253
|
354
|
354
|
253
|
Note:
|
|
|
|
*p<0.1; **p<0.05; ***p<0.01
|
Finally, the results in post-crisis time is generally in line with the findings from our main model as Greenfield investment illustrates negative impact on total entrepreneurial activity. However, Greenfield appears to no longer adversely influence the rate of opportunity-based entrepreneurship. This empirical finding may suggest that, after crisis, policy makers have re-considered the strategies to attract FDI and offset the unfavorable effect of Greenfield investment (Danakol et al., 2013). Major developed countries have taken a big hit from FDI response during the crisis period and policy makers have since shown enhanced focus on “sustainable FDI” (Poulsen and Hufbauer, 2011).