Foreign Direct Investment and Entrepreneurship: Gender Differences Across the World Rajeev K. Goel Department of Economics Illinois State University Normal, IL 61790-4200 USA [email protected]Keywords: Entrepreneurship; Gender; Foreign Direct Investment; Institutions; Government JEL classification: M16; F6 Draft for Asian Development Bank 2016 Conference on Foreign Direct Investment, June 2016 Abstract Using recent cross-national data this paper examines the impact of foreign direct investment (FDI) on entrepreneurship activity. The impact of FDI on entrepreneurship is not clear a priori with the possibilities of both a negative effect (crowding out) and a positive effect (synergy or complementarity). Results find support for the crowding out effect; however, this effect varies across nations with different prevalence of entrepreneurship. Another focus of this work is on gender differences. The crowding out effect is stronger for the full sample, rather than the subsample of female entrepreneurship and this finding stands up to a battery of robustness checks. Policy implications of the findings are discussed.
16
Embed
Foreign Direct Investment and Entrepreneurship: Gender ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Foreign Direct Investment and Entrepreneurship: Gender Differences Across the World
transmission of information via formal and informal networks and this could aid entrepreneurial
activity. Again, the gender influences of these factors might vary and the current work will prove
informative in this regard.
1 There could potentially be a bi-directional causality between FDI and entrepreneurship. However, our focus on a
single year (2015) somewhat alleviates concerns about reverse feedbacks. Given appropriate time series data, this
issue merits additional attention in the future.
Rajeev K Goel, Illinois State University Page 5
The institutional setup is very important to entrepreneurial activity and we consider several
dimensions (see Knack and Keefer (1995) for a general study in this context and Acs et al.
(2007) and Estrin and Mickiewicz (2011) for specifics). Institutions dictate the rules of the game
and a “fair” set of rules instills confidence in new entrepreneurs. The degree of democracy
(DEM) involves freedom of the press and civil liberties which have bearing upon transparency of
the government and due political process. So we would expect more democratic nations to have
greater entrepreneurship.
Greater economic freedom would also promote entrepreneurship. However, the degree of
economic freedom involves freedom of trade, protection of property rights, freedom from
overbearing regulations, freedom from corruption, etc. Given the multi-faceted nature, we
consider several dimensions. First, an overall composite index of economic freedom (EconFree)
is employed. Second, we consider government’s involvement in the economy, alternately via an
index (EconFree-GSP) and a more direct measure of government consumption (GovtCons).2
Government involvement can be a bottleneck when it adds red tape and bureaucratic delays, but
it can be beneficial when it promotes checks and balances. Third, the (lack of) corrupt activity is
captured by EconFree-Corr.3 Corruption undermines due process and creates inequities as
individuals and firms with resources are able to buy favors. This dissuades new entrepreneurs,
especially since they are resource constrained. Finally, the property rights protection aspect of
economic freedom is accounted for by including the index EconFree-Prop. Stronger property
rights instill confidence regarding protection of investments and appropriation of profits and thus
encourage entrepreneurship.
2.2 Data
We restrict our analysis to the year 2015, with the main constraint being the availability of
comparable time series data on female entrepreneurship (and due to the corruption index (see
Footnote #3)). Most of the other data are also for that year or the closest year available (see
Table 1). Although the sample size with FemaleENT is smaller, the correlation between
GeneralENT and FemaleENT is 0.93 (see Table A1 in the Appendix).
There was considerable variation in the FemaleENT scores across nations for 2015. The United
States was at the top of the list with an index score of 82.9, while Pakistan was at the bottom
with a score of 15.2. The average score for FemaleENT was 44.94 (Table 1). In contrast, the
2 As noted by Gicheva and Link (2015), there might be gender differences in government support for
entrepreneurship. Also see Rosa and Dawson (2006). 3 The corruption measure used is based off the well-known measure of corruption perceptions from the
Transparency International (www.trasparency.org). This measure has good cross-national comparability of
corruption; however, its time series comparability is limited.
Rajeev K Goel, Illinois State University Page 6
overall index, GeneralENT, was also led by the United States with a score of 85 with Bangladesh
ranked at the bottom with a score of 14.4. The sample mean for GeneralENT in 2015 was 39.07.
Turning to FDI, the mean value (as a percent of GDP) was 34.22; with the range being 76.37 and
-32.93, respectively (with the negative value signifying FDI outflows). The correlation between
FDI and GeneralENT was 0.17 and that was also the case for the correlation between FDI and
FemaleENT. The formal analysis below will test the validity of these relations when other
relevant factors have been accounted for.
The data employed for this study are in the public domain (e.g., Global Entrepreneurship and
Development Index, Freedom House, Heritage Foundation, etc.) and appropriate estimation
techniques to address the above questions are employed (e.g., OLS and quantile regression). The
use of indices for many variables addresses cross-national measurement/comparability issues.
Details about the variables are provided in Table 1 and a correlation matrix of key variables is in
Table 1A.
3. Results
All the models in the baseline results are estimated using Ordinary Least Squares with t-statistics
based on robust standard errors reported. The overall fit of all the models is decent as shown by
the R2 that is at least 0.75. As a test of the overall specification, we ran the RESET test. This test
showed that, relatively speaking, the fit of female entrepreneurship models was better than that
of the overall entrepreneurship. To address specification issues, we conducted a set of
robustness checks (see Sections 3.2 and 3.3).4
3.1 Baseline models
The baseline models in Table 2 show that the crowding out effect of FDI holds for general
entrepreneurship in all the models – the coefficient on FDI is statistically significant in Models
2A.1-2A.7 in Panel A. This is consistent with the idea that foreign investment dissuades or
replaces some domestic entrepreneurial activity. However, with the sample of female
entrepreneurship, the negative impact of FDI is statistically insignificant (Panel B). It could
either be the case that female entrepreneurs are not easily deterred or the focus of foreign
investments is generally in areas in which female entrepreneurs tend to not specialize. The
crowding out effect supports earlier findings of Da Backer and Sleuwaegen (2003) for Belgium
and those of Danakol et al. (2013). In terms of magnitude, the elasticity of GeneralENT with
respect to FDI is -0.025 (based on models 2A.1-2A.5). The differing effects of FDI on
4 While it is possible that there is a bi-directional causality between FDI and entrepreneurship whereby the level of
entrepreneurship in a nation reverse-causes (or invites) foreign investment, this issue is somewhat mitigated given
the cross-sectional estimation employed.
Rajeev K Goel, Illinois State University Page 7
GeneralENT and FemaleENT are noteworthy given the similar pairwise correlations of FDI with
the two entrepreneurship measures (Table 1A).
Greater economic prosperity, greater economic freedom (EconFree), freedom from corruption
(EconFree-Corr), stronger property rights protection (EconFree-Prop) and greater democracy
(note that higher values of DEM imply lower democracy – Table 1) increase entrepreneurial
activity in all cases and this is true for both overall entrepreneurship and for female
entrepreneurship. These findings are in accord with intuition. On the other hand, neither type of
entrepreneurship is significantly affected by the country size (denoted by population) and the
unemployment rate.
The effect of urbanization is positive and mostly statistically significant, with the relative
magnitude of the effect being greater on overall entrepreneurship than on female
entrepreneurship. Government size, whether denoted by GovtCons or EconFree-GSP, has a
significant (positive) impact on overall entrepreneurship but not on female entrepreneurship.5
The positive effect of government size is consistent with the checks and balances story, rather
than with the bottlenecks story.
3.2 Robustness check1: Effect of FDI on entrepreneurship across varying prevalence of
entrepreneurship
To examine another aspect and to answer one of the questions posed in the introduction, we
examine the effect of FDI on entrepreneurship across nations with different prevalence of
entrepreneurship. For this purpose, we employ the quantile regression (see Koenker and Hallock
(2001) for background on the quantile regression), and report regression results for both overall
and female entrepreneurship across q25, q50 and q75 in Table 3. Here q50 can be seen as
focusing on the median regression, while q25 and q75, respectively, capture low and high
entrepreneurship climates.
Results, across two model setups, show that the overall pattern of findings is similar to that in
Table 2. However, while FDI has a negative impact on entrepreneurship across all quantiles, the
negative or crowding out effect has statistical significance in the median regression for overall
entrepreneurship. Thus, crowding out is not significant in the tails (or in the case of female
entrepreneurship). This suggests that nations with mature (or nascent) entrepreneurship activity
have to be less concerned with the negative effects of FDI.6
5 Note that higher values of GovtCons and low values of EconFree-GSP both imply large government spending (see
Table 1). 6 An important caveat, however, that has to be kept in mind is that there are finer disaggregations of data that might
yield some differences in results. For instance, entrepreneurship may be classified as need-driven, opportunity-
driven, academic or underground (informal) entrepreneurship, and FDI may be inward or outward (see Danakol et
al. (2013), Goel et al. (2015a, 2015c)). We leave these extensions for future work, pending comparable data.
Rajeev K Goel, Illinois State University Page 8
In other findings, similar to Table 2, greater democracy and greater economic prosperity promote
entrepreneurship in all cases and both for the overall and the female sample. Also, like Table 2,
the effect of unemployment is statistically insignificant. Interesting differences, however,
emerge with regard to the effect of urbanization. Whereas greater urbanization promotes overall
entrepreneurship in all except high entrepreneurship prevalence nations (q75), the reverse is the
case for female entrepreneurship – urbanization promotes female entrepreneurship in nations
with high entrepreneurship climates. While this issue merits further research, these findings are
suggestive of gender differences in information exchange (see Goel et al. (2015b)) and perhaps
some minimum threshold level of entrepreneurship climate for females to benefit from
urbanization.
3.3 Robustness check2: Employing similar samples for general and female
entrepreneurship
There is substantial difference coverage in the number of countries represented in the overall
sample and the female entrepreneurship sample. Specifically, the overall sample has 127
countries, while the female entrepreneurship subsample has 76 countries. To check the validity
of our findings, we restricted the full sample so that the same number of countries were covered
in both the samples.
Rerunning the set of regressions from Table 2 for the full sample showed the pattern of results to
be similar. In particular, with regard to the main variable of interest, the coefficient on FDI was
again negative in all cases, confirming the crowding out effect of foreign direct investment.
However, given the reduced sample size, the statistical significance was more modest. The
results with regard to the other regressors were similar to what is presented in Table 2. Further,
the RESET test showed that the specification of the general entrepreneurship models in the
restricted sample was better.7
4. Concluding remarks
This paper examines the effect of foreign direct investment (FDI) on entrepreneurial activity
across a large sample of nations, with a focus on related gender differences. Research on this
aspect is important because it is not clear a priori whether FDI necessarily fosters
entrepreneurship. Results find support for the crowding out effect. Another focus of this work is
on gender differences. The crowding out effect is stronger for the full sample, rather than the
subsample of female entrepreneurship and this finding stands up to a battery of robustness
checks. Numerically, a ten percent increase in FDI (as a percentage of GDP) lowers overall
7 Details are available upon request.
Rajeev K Goel, Illinois State University Page 9
entrepreneurship by about 0.2 percent. We also find some differences in the efficacy of the
crowding out effect across nations with different prevalence of overall entrepreneurship.
Specifically, nations in the middle prevalence of entrepreneurship face the negative effects of
FDI and such consequences are not borne in the context of female entrepreneurship.
Several implications for public policy emerge from our findings. First, the obvious and
important policy implication is that gender differences warrant special efforts to foster
entrepreneurship among women. Second, the negative spillovers on entrepreneurship from FDI
should go into the cost-benefit calculations of efforts to encourage foreign investments. Third,
the importance of good institutions, including economic freedom, corruption control, and
property rights protection in fostering entrepreneurship is recommended. Fourth, a large
government size is not necessarily an impediment to entrepreneurship but could rather be helpful
by strengthening checks and balances. Fifth, the existing prevalence of entrepreneurship activity
should be kept in mind in framing policies. Sixth, the differing effects of urbanization across
gender have some implications for knowledge spillovers. Finally, given the positive effect of
economic prosperity, rich and poor nations perhaps need to approach entrepreneurship-
enhancement somewhat differently.
Rajeev K Goel, Illinois State University Page 10
References
Acs, Z.J., E. Autio, and L. Szerb, 2014, “National systems of entrepreneurship: Measurement
issues and policy implications”, Research Policy, 43, 476-494.
Acs, Z.J., E. Bardasi,, S. Estrin, and J. Svejnar, 2011, “Introduction to special issue of Small
Business Economics on female entrepreneurship in developed and developing
economies”, Small Business Economics, 37, 393-396.
Acs Z.J., D.J. Brooksbank, C. O'Gorman, D. Pickernell, and S.A. Terjesen, 2007, “The
knowledge spillover theory of entrepreneurship and foreign direct investment”, working paper,
Jena Economic Research Papers, #2007-059, www.jenecon.de.
Acs, Z.J., S. Desai, and L.F. Klapper, 2008, “What does “entrepreneurship” data really show?”,
Small Business Economics, 31, 265-281.
Albulescu, C.T., and M. Tămăşilă, 2014, “The impact of FDI on entrepreneurship in the
European countries”, Procedia - Social and Behavioral Sciences, 124, 219-228.