1 Introduction
Many countries have seen an increase in government size and taxes (Mueller,
2004; Tanzi & Schuknecht,
2000). A robust stream of primarily economics research has examined the influence of government size on economic development and social welfare (see Alesina et al.,
2004; Asimakopoulos & Karavias,
2016; Barro,
1990) and entrepreneurship (Nyström,
2008; Audretsch & Lehmann,
2017; Audretsch et al.,
2006,
2019a,
b). However, there is still a paucity of knowledge in identifying how these components interact with each other to influence different types of entrepreneurial activity (van der Zwan et al.,
2016). A critical gap remains in the literature: how government size and institutional environment influence different types of entrepreneurship. To address this gap in the literature, we ask the following question: How do government size, tax policy, and corruption influence entrepreneurial allocation towards necessity or opportunity entrepreneurship?. This question is an in-depth look into how government, the regulatory environment of a country, and broader institutions can influenceentrepreneurial climate. Additionally, the combination of policies and the institutional climate produces the pull and push factors as the two opposing forces that contribute todifferent types of entrepreneurial activity (opportunity and necessity entrepreneurship) in diverse institutional contexts (Audretsch et al.,
2019a,
b; Block et al.,
2015; Block & Sandner,
2009; Nikolaev et al.,
2018; Stenholm et al.,
2013; van der Zwan et al.,
2016; Welter et al.,
2019).
Different types of entrepreneurial activities are essential to consider because the quality of entrepreneurship dictates the technological change, structural transformation, and economic development (Amorós et al.,
2019; Belda & Cabrer-Borrás,
2018; van der Zwan et al.,
2016). Compared to necessity entrepreneurship, opportunity entrepreneurship occurs when entrepreneurs have other work options but still decide to pursue entrepreneurship (Acs et al.,
2008; Vivarelli,
2004,
2013). While entrepreneurship contributes to economic development and social welfare, not all entrepreneurship activity contributes equally. Therefore, determining the context that promotes desired entrepreneurial activity is essential (Baumol,
1990; Baumol & Strom,
2007; Fredström et al.,
2020).
The overall economic environment and governmental policies play a crucial role and govern these opposing forces (Fairlie et al.,
2011). Opportunity entrepreneurs may spend less and delay investments during high taxation and recessions, while they are also more likely to start a business when tax policy and government size are more conducive; necessity entrepreneurs, on the other hand, are forced out of the labor force and into entrepreneurship because of adverse economic conditions (Shiller,
2017).
Government size is a critical component to consider because larger government may mean a greater capacity to provide services and fund public governance activities. These supports could encourage entrepreneurship activities by providing safety nets to offset risks, high-quality inputs related to labor, and efficient capital markets. Larger government could also mean more inadequate governance if it allows overreach or inefficiency (see Afonso et al.,
2005; Hauner & Kyobe,
2010), which could discourage entrepreneurship activities by complicating the business environment and putting entrepreneurs at risk for predatory behavior by government agents.
Along with government size, tax policy directly affects entrepreneurs’ incentives to enter the market (Friedman et al.,
2000). Government is a core dimension of any country’s regulatory setup, and taxes provide revenues for public spending. Taxes, government size, and corruption are important to consider together becausegovernment requires tax revenue to operate and maintainservices, which corruption can affect (see Belitski et al.,
2016; O’Higgins,
2006; Pandey,
2010).
We contribute to the existing literature on entrepreneurship and public policy by demonstrating that government size, corruption, and tax policy do not have predictably consistent effects ontwo types of entrepreneurship. To test this article’s hypotheses, we use a panel of 52 countries from 2005 to 2015. The results suggest that the direction and magnitude of government size, corruption, and tax policy on different types of entrepreneurship depend on the form of entrepreneurship.
Second, we contribute to the literature on entrepreneurship and institutions by demonstrating that engagement in opportunity and necessity entrepreneurship is shaped by both the regulatory environment (tax), informal institutions (corruption), and government size. The study results suggest not all regulatory and institutional settings have a similar effect on entrepreneurship.
The remainder of the paper is structured as follows. In Section 2, we present the relevant literature and develop our hypotheses on the direct and moderating effects of government size, tax policy, and corruption on entrepreneurship. In Section 3, we discuss our data, method, and identification strategy. We report results in Section 4, followed by discussing implications and next steps for scholars, entrepreneurs, and policymakers in Section 5.
4 Empirical strategy
We first check correlations among our variables: Necessity and opportunity entrepreneurship are negatively correlated (− 0.75). Government safety expenditure, expenditure on housing, and economic affairs are positively associated with necessity entrepreneurship, while they are negatively associated with opportunity entrepreneurship (see Table
2).
Table 2
Correlation matrix
Corruption is positively correlated with necessity entrepreneurship (0.68) and negatively with opportunity entrepreneurship (− 0.60).
Our first concern was the covariance structure of the control variable matrix. We are unlikely to estimate partial effects without bias, as several variables included in the model are causally related; this is not a multicollinearity issue but an endogeneity issue. First, government size can be a function of the tax base’s size and government programs, including those serving businesses and the unemployed.
Second, corruption can be associated with entry regulation, taxes, licensing, and many other regulations (Belitski et al.,
2016; Méon & Sekkat,
2005; Rose-Ackerman,
2007). Third, government expenditure could create more opportunities for corruption, such as through public procurement. Unobserved factors that affect government size could also affect corruption and taxes.
Because these variables are causally related to each other, the effects of interrelated variables are being mediated (moderated) by others, and they cannot be included together in the same regressions (Angrist & Pischke,
2008), this makes it difficult to assess the independent effect of any one of the variables. We control for this interdependency effect using an alternative modeling strategy by fitting four models jointly.
First, we combine four equations in a mixed-process model, incorporating both continuous responses (entrepreneurship type, government size) and ordinal responses (corruption). We modeled a simultaneous system of equations: first for necessity entrepreneurship, second for opportunity entrepreneurship, third for government size, and fourth for corruption. Second, the cmp framework in Stata 15 allows for the different observations to enter each equation in the model, with available observations being used to estimate each equation parameter.
Given the potential interdependence between the four equations, they should be estimated simultaneously. The FIML estimates produced by cmp estimation can handle this form of simultaneity (Baum et al.,
2017). A maximum likelihood estimator of a seemingly unrelated equation (SUR) system (Zellner,
1962) can consistently estimate parameters in an essential subclass of mixed-process simultaneous systems: those are recursive and simultaneous (Roodman
2011). Government size is likely to influence corruption and vice versa and is associated with several explanatory factors hypothesized to influence government size, corruption, and entrepreneurship. There is potentially a more efficient approach as it allows estimation of possible cross-equation correlations for each country between equations with entrepreneurship types as dependent variables and equations with the institutional environment as dependent variables. Besides, we controlled for the year and country-specific effects.
Given the panel structure of the data, the equation was estimated using the observations from the various datasets that could be matched by country and year. The entire sample includes 272 observations for 52 countries. We used the fixed-effects (FE) estimator as it concentrates on differences that, over time, characterize a country. This is why the FE estimator is also referred to as the “within” estimator. It explains to what extent a given country’s change in a variable of interest affects its entrepreneurship rates. Our estimates are “within” effect, allowing us to identify the factors that explain the differences between the countries in the panel and control for country-specific unobserved characteristics over time. Thus, the FE estimates should provide a more exhaustive scenario of the drivers of entrepreneurship than random effect estimation, which follows a stronger assumption. The use of the FE estimator also resolves a simultaneity bias induced by unobservable factors and is preferred.
The model with country and time-fixed effects is as follows:
$$ {E}_{i,t}={\beta}_0+\sum \limits_{i=1}^n{\beta}_{11}{x}_{i,t-1}+\sum \limits_{j=1}^n{\beta}_{12}{z}_{i,t-1}+{\rho}_{1i}+{\lambda}_{1t}+{u}_{1\left(i,t-1\right)} $$
where
Ei, t is a necessity (opportunity) entrepreneurship in a country
i at time
t. We deal with panel data for each dependent and independent variable.
Xi, t − 1 is a vector of explanatory variables—the size of government (variety of government expenditures), corruption, and tax rate of country
i at time
t.
Zi, t − 1 is a vector of control exogenous variables for a country
i at time
t. Moreover, we include two additional vectors of fixed country effects:
ρi, controlling for unobserved heterogeneity of a country
i over time
t, and
λt is a vector of time-fixed (entity invariant) effects over each period
t across all countries
i. The error term is denoted by
ui, t − 1 for a country
i, at time
t.
We lagged all explanatory and control variables by 1 year for robustness checks and to rule out possible endogeneity in all equations. There could be a time lag for tax policy, government spending, and corruption to affect and shape entrepreneurs’ behavior. For example, a new tax rate could be passed in 1 year but enforced in the following year, or it could take several months for entrepreneurs to learn about new government programs. It is plausible to assume that regulatory changes and government expenditure will affect outcomes in the next financial year, as they cannot be immediately applied.
5 Results
The results of our fixed-effects panel data estimations are reported in Table
3. Results are grouped as follows: specifications 1–4 are for necessity entrepreneurship, and specifications 5–8 are for opportunity entrepreneurship.
Table 3
Fixed-effects estimation results
DV | Necessity entrepreneurship | Opportunity entrepreneurship |
Stepwise analysis | Basic | Government expenditure | Interaction of govt. expenditure and | Basic | Government expenditure | Interaction of govt. expenditure and |
Profit tax | Corruption | Profit tax | Corruption |
Rich | 1.20 (7.22) | 1.56 (6.84) | 1.21 (6.81) | 0.89 (6.71) | − 2.35 (8.75) | − 0.71 (7.88) | − 1.87 (7.72) | − 0.60 (7.78) |
Private credit bureau | − 0.02 (0.09) | − 0.41*** (0.14) | − 0.54*** (0.16) | − 0.40*** (0.14) | 0.30*** (0.08) | 0.53*** (0.11) | 0.48*** (0.12) | 0.52*** (0.11) |
Human capital | − 0.14 (0.28) | 0.23 (0.33) | 0.22 (0.35) | 0.18 (0.32) | 0.37* (0.24) | 0.89** (0.38) | 0.69* (0.39) | 0.92** (0.37) |
Tax rate (H3) | − 0.19 (0.13) | − 0.43** (0.20) | − 0.48** (0.21) | − 0.25** (0.11) | 0.10 (0.15) | − 0.45** (0.22) | − 2.75*** (1.02) | 0.22 (0.24) |
Corruption (H2) | − 4.87 (3.39) | − 5.87* (3.27) | − 8.31** (4.16) | − 8.96** (4.53) | − 10.13** (4.02) | − 10.49** (4.37) | − 6.74** (3.56) | − 24.59 (15.98) |
Tax time | 0.01 (0.01) | 0.01 (0.01) | 0.01* (0.01) | 0.01 (0.01) | 0.01 (0.01) | 0.01 (0.01) | 0.01 (0.01) | 0.01 (0.01) |
Credit | 0.01 (0.03) | 0.02 (0.04) | 0.03 (0.04) | − 0.01 (0.05) | − 0.03 (0.04) | − 0.08** (0.04) | − 0.10** (0.04) | − 0.05 (0.05) |
Entry density | − 0.67* (0.36) | − 0.27 (0.49) | − 0.22 (0.50) | − 0.61 (0.51) | − 0.02 (0.40) | − 0.24 (0.52) | − 0.48 (0.52) | 0.04 (0.54) |
Unemployment | 0.67*** (0.23) | 0.49 (0.33) | 0.48 (0.34) | 0.51 (0.33) | − 1.31*** (0.26) | − 1.17*** (0.34) | − 1.17*** (0.35) | − 1.34*** (0.35) |
Government support | − 1.44 (2.44) | − 5.22 (3.36) | − 6.33* (3.61) | − 7.01* (3.61) | 5.32* (2.91) | − 1.28 (3.79) | 3.10 (3.99) | 0.30 (4.03) |
Labor force | − 0.27 (0.19) | − 0.08 (0.34) | 0.05 (0.36) | − 0.20 (0.35) | − 0.03 (0.22) | − 0.38 (0.37) | − 0.45 (0.39) | − 0.01 (0.38) |
Safety expenditure (H1) | | − 12.50** (5.72) | − 7.94** (4.06) | − 12.15* (6.14) | | 5.57 (5.47) | 13.33 (74.73) | 13.35** (6.33) |
Economic affairs expenditure (H1) | | 0.25 (0.27) | − 0.20 (1.22) | 2.56** (1.14) | | − 0.77** (0.31) | − 1.37 (1.30) | 0.04 (1.16) |
Housing expenditure (H1) | | − 1.24 (1.95) | 13.07** (6.28) | − 3.71 (4.39) | | − 2.86 (2.16) | − 8.93 (7.03) | − 8.18* (4.79) |
Culture expenditure (H1) | | 3.63 (4.32) | − 7.21 (13.03) | 4.97 (7.32) | | − 8.79* (4.84) | − 3.72 (14.58) | − 14.80* (7.86) |
Education expenditure (H1) | | 0.61 (2.08) | 5.75 (6.99) | − 0.18 (3.16) | | − 0.85 (2.12) | − 0.49 (7.59) | 0.08 (3.19) |
Social security expenditure (H1) | | 1.55** (0.76) | − 0.30 (1.53) | 2.41*** (0.89) | | − 2.20** (0.87) | − 4.31*** (1.59) | − 2.31** (1.02) |
Public administration share (H1) | | 0.09 (0.43) | 1.30 (1.21) | − 0.29 (0.63) | | − 1.51*** (0.48) | − 5.01*** (1.35) | − 1.26* (0.71) |
Interaction: safety expenditure (H5/H4) | | | − 0.06 (0.33) | − 7.17* (4.03) | | | 0.54 (0.37) | 18.39*** (6.15) |
Interaction: economic affairs expenditure (H5/H4) | | | 0.02 (0.04) | 1.31** (0.62) | | | 0.02 (0.05) | 0.33 (0.63) |
Interaction: housing expenditure (H5/H4) | | | − 0.44** (0.19) | − 1.79 (2.85) | | | 0.17 (0.21) | − 3.24 (3.20) |
Interaction: culture expenditure (H5/H4) | | | 0.30 (0.33) | − 1.72 (5.06) | | | − 0.13 (0.37) | − 1.21 (5.47) |
Interaction: education expenditure (H5/H4) | | | − 0.10 (0.16) | − 1.27 (1.96) | | | − 0.01 (0.17) | − 0.04 (2.22) |
Interaction: social security expenditure (H5/H4) | | | 0.05 (0.04) | 1.34*** (0.45) | | | 0.0610 (0.04) | − 1.04** (0.52) |
Interaction: public administration share (H5/H4) | | | − 0.02 (0.03) | − 0.25 (0.43) | | | 0.11*** (0.03) | − 0.10 (0.48) |
Constant | 45.01** (19.62) | 31.35 (40.72) | − 130.7 (116.15) | 21.32 (55.67) | 16.71 (23.22) | 60.70 (45.46) | 291.10** (127.78) | − 29.57 (55.99) |
N | 274 | 274 | 274 | 274 | 274 | 274 | 274 | 274 |
r2 within | .17 | .36 | .41 | .42 | .24 | .47 | .53 | .5 |
r2 overall | .12 | .14 | .18 | .05 | .45 | .06 | .01 | .26 |
r2 between | .09 | .23 | .28 | .15 | .45 | .03 | .07 | .20 |
F-stats | 3.42 | 2.83 | 2.50 | 2.60 | 6.25 | 4.85 | 4.33 | 4.36 |
Loglikelihood | − 795.12 | − 508.47 | − 500.80 | − 499.10 | − 869.88 | − 556.48 | − 545.51 | − 545.13 |
F test for fixed-effects | 7.07 | 5.56 | 4.76 | 4.67 | 6.09 | 6.85 | 5.6 | 5.99 |
Sigma u | 11.29 | 25.68 | 29.91 | 22.86 | 13.38 | 22.34 | 26.19 | 23.16 |
Sigma e | 5.06 | 4.69 | 4.65 | 4.61 | 6.14 | 5.43 | 5.29 | 5.27 |
Rho | .83 | .96 | .97 | .96 | .82 | .94 | .96 | .95 |
In hypotheses 1a and 1b, we posited that government size and both types of entrepreneurship have a positive relationship. Our results do not support H1a that government size increases opportunity entrepreneurship; an increase ineconomic affairs expenditure by 1% to GDP reduces opportunity entrepreneurship by (
β = − 0.77,
p < 0.05). An increase in culture and recreation expenditure by the government reduces opportunity entrepreneurship by 8.79% (
β = − 8.79,
p < 0.01), while an increase in social security expenditure reduces opportunity entrepreneurship by 2.20% (
β = − 2.20,
p < 0.01). Interestingly, an increase of public administration share in total government expenditure reduces opportunity entrepreneurship by 1.51% (
β = − 1.51,
p < 0.001), which could be associated with a negative effect of bureaucratic government (Aidis et al.,
2012; Estrin et al.,
2013). With regard to hypothesis 1b, while we find that social security expenditure is positively associated with necessity entrepreneurship (
β = 1.55,
p < 0.05), supporting H1b, we find that safety-related expenditure reduces necessity entrepreneurship (
β = − 12.50,
p < 0.01).
We find strong support for H2a and H2b—corruption has a negative effect on both necessity (
β = − 5.87,
p < 0.05) (spec. 2, Table
3) and opportunity (
β = − 10.49,
p < 0.05) (spec. 6, Table
3). Given the confidence intervals’ distribution, corruption’s negative effect on opportunity entrepreneurship is twice as strong as on necessity entrepreneurship, demonstrating that opportunity entrepreneurship is worse affected by corruption.
In H3a and H3b, we posited that the tax rate has a negative impact on both necessity and opportunity entrepreneurship. We find support for both H3a and H3b: an increase of 1% in tax will reduce necessity entrepreneurship by 0.43% (
β = − 0.43,
p < 0.01) (spec. 2, Table
3) and opportunity by 0.45% (
β = − 0.45,
p < 0.01) (spec. 6, Table
3). The effect of the tax change on both types of entrepreneurs is very similar (see Braunerhjelm et al.,
2021).
In H4a and H4b, we posited that an increase in government size and tax rate increases opportunity entrepreneurship and reduces necessity entrepreneurship, respectively. When we consider opportunity entrepreneurship (spec 7, Table
3), we note that only an increase in tax rate and public administration expenditure share adds to opportunity entrepreneurship (
β = 0.11,
p < 0.01), while the effect of public administration expenditure share remains negative and statistically significant, partly supporting H4a. Other interaction coefficients in spec. 7 (Table
3) are insignificant, which means that an increase in tax rate and government size does not change opportunity entrepreneurship. We found a negative effect of housing government expenditure on necessity entrepreneurship (
β = − 0.44,
p < 0.01) (spec. 3, Table
3) supporting H4b. Other interactions remain insignificant (spec.3, Table
3).
Finally, we found that an increase in social security expenditure in countries with high corruption increases necessity (
β = 1.34,
p < 0.05) (spec. 4, Table
3) entrepreneurship and reduces opportunity entrepreneurship (
β = − 1.04,
p < 0.05) (spec. 8, Table
3). An increase in government size related to economic affairs expenditure facilitates necessity entrepreneurship by 1.31%, which contrasts our hypothesized relationship. Safety expenditure reduces necessity entrepreneurship (
β = − 7.17,
p < 0.01) (spec. 4, Table
3) and increases opportunity entrepreneurship (
β = 18.39,
p < 0.01) (spec. 8, Table
3). The evidence for H5 is mixed and demonstrates that not all government expenditure has a ubiquitous effect on the type of entrepreneurship activity.
Overall, our findings illustrate the importance of considering the heterogeneity of entrepreneurship. We find mixed support for all our hypotheses.
The results for our control variables are as follows. Economic development proxied by “rich” is not associated with necessity and opportunity entrepreneurship, highlighting that both types of entrepreneurs can exist in countries with different levels of economic development. The private credit bureau is negatively associated with necessity entrepreneurship but positively with opportunity. A share of the population with tertiary education increases opportunity entrepreneurship and is not associated with necessity entrepreneurship. Government support has no direct effect on the type of entrepreneurship activity, while the effect changes when we control for interaction effects. We find that unemployment is negatively associated with opportunity entrepreneurship (spec. 6, Table
3) and is not associated with necessity entrepreneurship (spec. 2, Table
3).
6 Discussion and conclusion
Entrepreneurship activity is vital for economic development, so policymakers and scholars are interested in determining components that positively influence these activities and understanding entrepreneurial motivation to enter the market. This paper has examined how tax policies, government size, and corruption influence necessity and opportunity entrepreneurship. We use a 2005–2015 country-level panel data matching five distinctive datasets for 52 countries and show that a high corruption and tax will negatively affect both types of entrepreneurs, while government size effect on necessity and opportunity is distinctive and conditional on different types of government expenditures. Our results suggest that different types of entrepreneurship require different types of policies. Not all kinds of entrepreneurship contribute equally to society or in the same way (Block & Sandner,
2009; Reynolds et al.,
2005; Vivarelli,
2004; Vivarelli,
2013) and policy makers can consider how to support the type of entrepreneurship they want to prioritize. Our empirical approach enables us to provide detailed new insights. This approach is appropriate for research on the complex institutional environment when one-size-fits-all government expenditure is not enough (Levie & Autio,
2011).
Our results suggest that the impact of the same institutional settings (corruption) on necessity and opportunity entrepreneurship is not uniform (see Levie & Autio,
2011) as the size of the effect varies between types of government expenditure. Additionally, while corruptionmay not discourage a share of necessity entrepreneurs, the overall resultremains negative. Entrepreneurs are constantly engaging in activities to help their ventures and in some cases may seek to leverage corruption to their advantage, but boththe necessity and opportunity entrepreneurs are adversely affected by corruption.
When it comes to the interaction effect of corruption and government size on opportunity and necessity entrepreneurship as the direction of the relationship depends on the type of government expenditure (i.e., safety, economic affairs, housing, education, social security, public administration). Our results indicate that investing in economic affairs and increasing social security can support necessity entrepreneurs, andsafety expenditure can supportopportunity entrepreneurs. We are not able to test if there is a shift or transition between people who become necessity and opportunity entrepreneurship when government spending changes. Our study results also reveal that, in many cases, the effects of taxes and corruption in combination with heterogeneous government expenditure are not the same for opportunity and necessity entrepreneurship. For taxes and corruption, the effect depends on what government invests in and whether it targets to increase the supply of entrepreneursor increase the demand for entrepreneurs. To increase entrepreneurial activity in a country, government plays an important role. However, government investment may have a different effect on necessity entrepreneurs and opportunity entrepreneurs, who may also leverage the government’s services differently.
6.1 Study limitations and future research
While we included time lags in our model, there may be an endogeneity between government size, corruption, and entrepreneurship activity which in the future research may require instruments and longitudinal data with longer lags for robustness check. The GEM entrepreneurship-related measures also have their limitations, notably the entrepreneurship measures’ low comparability between developed and developing countries (Reynolds et al.,
2002). It should be noted that we controlled for time-invariant systematic measurement errors using country-level fixed-effects, which helps reduce concerns about these measurements.
We provide fresh empirical insights to the importance of considering heterogeneous institutions and heterogeneous entrepreneurship outcomes. We add to recent studies that found institutional conditions can play a role in shaping the nature and quality of entrepreneurship activity. Chowdhury et al. (
2019) and Sutter et al. (
2017), for instance, assert that regulation matters for net entrepreneurship productivity score, and it can play a role in shaping cognitive and informal economy dimensions. Future research can examine how policymakers can formulate policies to target specific types of entrepreneurship outcomes that are salient in their contexts, for example if they wanted to support more opportunity entrepreneurs.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.