Self-employment, corruption, and property rights: a comparative analysis of European and CEE economies

This study analyzes the relationship between self-employment, corruption, and property rights in 30 European countries, including 11 Central and Eastern Europe (CEE) economies, across the two decades of 1996–2016. In general, relatively little research has focused on the relationship between entrepreneurship and the protection of property rights. Furthermore, past findings show that corruption may have both negative and positive effects on the level of entrepreneurial activity, either “greasing” or “sanding” the wheels for entrepreneurship. Overall, research on how the informal institution corruption and the formal institution property rights are linked to entrepreneurship in post-socialist/transition economies has been limited. We find that stronger protection of property rights increases self-employment ratios, both in Europe in general and in CEE economies. The relationship between self-employment and the control of corruption is not significant. We conclude that neither higher nor lower levels of corruption control affect the share of self-employment. In comparative perspective, the ratio of self-employment in the group of CEE economies does not respond differently to these two key institutions.


Introduction
How is business activity affected by the control of corruption and the efficiency of property rights? In this study, we analyze these relationships in several European and Central and East European (CEE) (post-socialist) economies across two decades. The dominating paradigm in the entrepreneurship and small business literature maintains that the creation and growth of small enterprises are endogenous components of job generation and economic development (Braunerhjelm et al. 2010;Bruton et al. 2010). Indeed, entrepreneurship is a key element in economic policy today. During recent decades, policy programs for boosting entrepreneurship have been introduced by several international organizations such as the World Bank, the International Monetary Fund, and the European Union (EU). In the EU, promoting self-employment has become an important strategy element, which is maintained as a means of fighting unemployment; furthermore, increasing self-employment rates is seen by European policymakers as an instrument to support general enterprising activity for the benefit of job generation, innovation, and economic growth. Nevertheless, it has been acknowledged that business owners in European countries sometimes encounter different, country-specific conditions and opportunities (European Commission 2015. Both the formal and informal rules and institutions that support or impede entrepreneurial efforts are therefore important to consider when attempting to understand differences and changes in enterprising activity both within and across economies (e.g., Autio and Fu 2015;Baumol 1990). The present article focuses on a European sample and provides an opportunity for a comparative analyses of a setting with substantial institutional variation as well as extensive variation in entrepreneurship.
Both the quality and the quantity of entrepreneurship are shaped by existing rules, regulations, and norms (Baumol 1990;Bjørnskov and Foss 2008;Henrekson 2005). Research on these relationships has increased in recent decades (Autio and Fu 2015;Urbano et al. 2018). For instance, one strand of cross-country research has examined how formal welfare system indicators such as employment protection and taxation affect self-employment (Parker and Robson 2004;Robson 2003Robson , 2010Torrini 2005). Other researchers have studied the role of "national culture" in explaining variations in entrepreneurship across different economies (e.g., Hofstede et al. 2004;Liñán and Fernandez-Serrano 2014).
Another fundamental institution that has received increasing scholarly attention is the quality of government-more specifically, how corruption and property rights are linked to entrepreneurial activity. The (negative) effect of corruption on macroeconomic growth and society has long received attention (Mauro 1995;Shleifer and Vishny 1993), and researchers in entrepreneurship have increasingly come to study how this informal institution is linked to business activity (e.g., Belitski et al. 2016). Similarly, research has investigated whether entrepreneurial activity is affected by the formal regulations of economic freedom and the protection of property rights (e.g., Amorós et al. 2019;Stenholm et al. 2013). A small but growing number of studies have extended these lines of thought, incorporating both of these key regulations in cross-country analyses (e.g., Aidis et al. 2010;Dempster and Isaacs 2017;Estrin et al. 2013). In this article, we follow this emerging strand by studying the linkages between self-employment, corruption, and the protection of property rights in several European and CEE countries across the two decades of 1996-2016.

Background and research setting
European economies have some of the world's highest rankings in corruption control and property rights protection. However, several European economies still score weakly-or have recently exhibited diminishing scores (Charron et al. 2015;Miller et al. 2018). Since self-employment represents a key component in established entrepreneurship theory (Braunerhjelm et al. 2010), as well as an important strategic element in European policy (European Commission 2015), research on how these two institutions are linked to entrepreneurship is highly relevant. Setting out from the extant literatures on entrepreneurship and institutions, and entrepreneurship in transition, we ask the overarching question: how is self-employment affected by corruption and property rights in Europe and in CEE economies? Our study makes a general contribution to the literature on entrepreneurship and institutions-specifically, it contributes to the knowledge on how corruption and property rights are related to entrepreneurship. Particularly the linking of formal institutional property rights to entrepreneurship and small business activity has been relatively uncommon in past research (Urbano et al. 2018). This article thus adds to the understanding of how this formal institution affects enterprising activity.
We make extensive use of earlier research and have strived to control for variables that describe human capital, as well as economic and institutional conditions that have been acknowledged to be related to entrepreneurship in past research. While earlier research on entrepreneurship and institutions has often used rather short panels or cross-sectional samples, this study uses a 20-year-long (unbalanced) panel . Unlike several previous analyses, our study covers a wide variety of European countries. Finally, this article makes a methodological contribution by applying the common correlated effects mean pooled (CCEP) panel model. Investigating previously utilized panel models in the literature on entrepreneurship and institutions, we find the CCEP model to be the most suitable for our purposes.
Corruption is an informal institution, representing customs and traditions that change incrementally (Krasniqi and Desai 2016;North 1990;Urbano et al. 2018). Informal institutions are more or less widely "shared" and corruption may evolve to be a norm. Property rights, on the other hand, are a formal institution that relate to the ability and will of the government to ensure the protection and stability of intellectual, financial, and physical property, as well as to constraints on the executive branch to confiscate this property. It is one of the most fundamental market-supporting institutions in a modern society. The quality and protection of property rights depend on both the regulatory framework in a society and its rule of law (Amorós et al. 2019;Williamson 2000).
Both the informal institution corruption and the formal institution property rights are likely to affect the quality or "variations" of different types of entrepreneurship. In his seminal works, Baumol (1990Baumol ( , 1993 convincingly elaborates on how the institutional setting shapes productive, unproductive, and destructive entrepreneurship. Moreover, institutions also affect the quantity of entrepreneurship-such as the number of new firms created over time or the rate of self-employment. As we will develop in more detail below, the influence of the government's will to ensure property rights-or the efficiency in combatting corruption-on entrepreneurship depends on the stage of economic and social development. The effect of corruption control and property rights protection on entrepreneurship rates in mature market economies may differ substantially from that in other types of economies, such as developing countries or transition economies (Dempster and Isaacs 2017; Desai et al. 2003;Dreher and Gassebner 2013).
Despite a growing number of studies on "entrepreneurship in transition" (e.g., Aidis et al. 2010;Chepurenko and Sauka 2017;Dana 2005;Desai et al. 2003;Grilo and Thurik 2006;Ovaska and Sobel 2005;Szerb et al. 2017), entrepreneurship scholars have noted that more than two decades of systemic change has not resulted in any substantial volume of research (Chepurenko 2015;Chepurenko and Sauka 2017). The present article contributes to this knowledge: following Ovaska and Sobel (2005), one way to assess the "success" of transitional countries is to focus on the ability of their institutions to generate and foster entrepreneurship. In this article, this assessment translates into how corruption control and property rights protection affect self-employment-both in Europe as a whole and in European post-socialist countries.
Current research debate revolves around whether entrepreneurship in CEE economies differs from that in more mature (European) market economies, and whether the "post-socialist culture" may have had long-standing negative effects on enterprising activity in general (Dana 2005;Grilo and Thurik 2006). At the same time, as maintained by Kornai (2006), the transformation of the CEE region since the early 1990s could be viewed as a "success story," since a capitalist system was established within a historically short period of time. Scholars have thus claimed that, in our time, the entrepreneurial transition of the CEE countries is very close to convergence (Szerb and Trumbull 2016). Unlike several other transitional regions, CEE countries and the Baltic states came under the legal and political umbrella of the EU early, and there has generally been a long period of enforcement of change toward a "normal" market economy (Chepurenko and Sauka 2017;Johnson et al. 2000). Therefore, even though many CEE economies may share some "historical" characteristics such as lower levels of opportunity perception among entrepreneurs (Szerb et al. 2017), several of them do not now substantially differ from mature developed economies. In our time, the CEE and Baltic economies are therefore generally on a "normal" capitalist path; their levels of entrepreneurship have been found to correspond to their overall (relatively lower) level of economic development (Cieślik and van Stel 2014;Szerb et al. 2017). Against this background, we find it relevant to analyze the relationship between self-employment, corruption control, and the protection of property rights, in addition to the literature on entrepreneurship in transition (e.g., Sahakyan and Stiegert 2012; Szerb et al. 2017).

Theoretical framework
As summarized by Autio and Fu (2015), there is a vast body of comparative and institutional research on the linkages between a country's legal and regulatory environment and its economic behavior. The quality of social, economic, and political institutions regulates the benefits and costs of economic action. However, it is only more recently that comparative entrepreneurship research has started to analyze how legal and regulatory frameworks in different countries influence entrepreneurial choice. The present article's empirical context assumes that the institutional environments for self-employment differ across time and place. New institutional economic theory offers a useful framework for this type of analysis: potential and existing entrepreneurs are extensively affected by the institutional setup and adapt their activities and strategies according to the "rules of the game." Since institutions affect the level of uncertainty in a society, shaping transactions and opportunity costs as well as the society's reward system (Baumol 1990;North 1990North , 1994, the extent to which people desire and pursue business ventures is likely to be shaped by institutional determinants. Ultimately, variations in entrepreneurship are affected by institutions and institutional change (Baumol 1990;Bjørnskov and Foss 2008;Henrekson 2005).
Institutional economic theory divides institutions into two main categories: formal and informal institutions. Formal institutions affect the uncertainty and transactions costs in a society, ideally reducing them and enhancing market performance-or, in some cases, creating disincentives and obstacles for human action. Thus, the formal structure of the political, legal, and political systemsuch as rules and regulations and bureaucratic and administrative procedureswill affect human behavior. Human behavior is also affected by the second major category of informal institutions, which are represented by belief systems and social norms (or "culture"). The function of norms and belief systems is to give structure and stability to a society; like formal institutions, informal institutions may hinder or encourage enterprising activity by providing an appropriate environment or imposing barriers (Dheer 2017). Due to path dependence, institutional change is mostly incremental-especially changes in informal institutions. While formal institutions can be changed overnight, informal institutions only change gradually. Once established, institutions become locked in through path-dependent self-reinforcement (North 1990(North , 1994. This explains why informal institutions "have a lasting grip on the way society conducts itself" (Williamson 2000: 597).
The informal institution of corruption and the formal institution of property rights are related. This connection implies that countries with high corruption levels often (but not always) have weaker property rights. For example, entrepreneurs are more likely to engage in corruption if the protection of property rights is weak (Tonoyan et al. 2010). Despite this relation, the two institutions cannot be described in the same way and merely partially overlap. Corruption generally increases direct transaction costs and can therefore be regarded as a tax on economic activity and entrepreneurship. Weak property rights will also shape incentives and entrepreneurship; however, instead of affecting direct transaction costs, the weak or absent protection of property rights functions in a different way, relating to the more important threat of arbitrary government and expropriation (Estrin et al. 2013).

Corruption
Corruption is a representation of a society's overall institutional quality and reflects a society's legal, economic, and political institutions; it is an outcome of cultural institutional factors. This phenomenon is not only linked to weaknesses in the formal institutional setting but also, to an even greater extent, to social norms and culture (Aidt 2009;Svensson 2005). Corruption can be private, such as when (individuals in) firms and organizations act contrary to duties or laws, thereby harming the organization, other organizations, or the functioning of the market through, for example, gifts, bribery, or collusion (Argandoña 2005). One common definition of public corruption is the misuse of public office for private gain, which can be seen as a response to beneficial or harmful rules of the game (Svensson 2005). The (negative) effect of corruption on macroeconomic growth and societal development has long received attention (e.g., Kaufmann and Wei 1999;Mauro 1995;Shleifer and Vishny 1993). At the firm level, both comparative and single-country analyses in several different country contexts support the view that corruption is an obstacle to firm performance (Chowdhury et al. 2018;Gaviria 2002;Sahakyan and Stiegert 2012), often finding asymmetries between formal and informal institutions, where firms and entrepreneurs may engage in corruption and "shadow economy" activities due to dissatisfaction with formal institutions (e.g., the tax system) and distrust in the government (Putniņš and Sauka 2011;Tonoyan et al. 2010).
Several empirical studies have analyzed the relationship between variations in entrepreneurship and corruption in a cross-country setting. Estrin et al. (2013) use survey data from the Global Entrepreneurship Monitor (GEM) on the growth aspirations of entrepreneurs in 42 countries during 2001-2006 (ranging from advanced economies to developing countries) and find that corruption has negative effects. Belitski et al. (2016) employ a panel of 72 countries on new limited-liability company entry and find significantly negative effects from corruption on entry. Using a 5-year panel over 44 countries, Chowdhury et al. (2015) come to mixed conclusions and find no relationship between more freedom from corruption and the rate of selfemployment (but positive effects from freedom from corruption on nascent entrepreneurship and start-up rates). Furthermore, Anokhin and Schulze (2009) use a 7-year-long panel on 64 countries to study the control of corruption and total entrepreneurial activity (TEA; the percentage of individuals actively engaged in starting or managing a new business). They find that stronger control of corruption has positive effects on TEA.
The TEA measure is also used by Dempster and Isaacs (2017), who analyzed 47 countries during 2001-2011 in repeated cross-sectional samples. Different from the previously mentioned studies, they find that better control of corruption actually depresses entrepreneurship when economic freedom is low; under such circumstances, corruption seems "necessary" to engage in entrepreneurship. This result supports research that specifically investigates whether (some degree of) corruption "greases the wheels" and actually helps entrepreneurship. The essential idea in this debate is that, in some societies, paying bribes or "rewards," or having personal relationships with public officials, facilitates entrepreneurship and, ultimately, economic growth (Aidt 2009). In a similar way, using large country samples, both Chowdhury et al. (2019) and Dreher and Gassebner (2013) maintain that corruption facilitates entrepreneurship in highly regulated and/or less developed economies. A related comparative perspective, based on a cross section of business founders in 176 countries, is proposed by Avnimelech et al. (2014): corruption generally leads to low levels of entrepreneurship; however, the negative effect is much stronger in developed economies, which implies that corruption would be less harmful in poorer countries. Using GEM survey data from the early 2000s on nascent entrepreneurship in 31 economies (from developed to developing countries), Aidis et al. (2010) find that entrepreneurship is weakest in socialist legal-origin countries. However, there is no statistically significant relationship between corruption and entrepreneurship.
Using cultural context as a moderator, Dheer's (2017) analysis of TEA in individualistic versus collectivistic societies suggests similar patterns. For 84 countries, corruption shows no significant main effect. However, unlike in individualistic (i.e., Western) countries, corruption actually facilitates new business startups in collectivistic nations or cultures. In contrast, using a panel of 130 countries to investigate new limited-liability company entry, Dutta and Sobel (2016) conclude that corruption never helps entrepreneurship: all efforts to reduce corruption are thus beneficial for entrepreneurship, even in countries with the worst business climates. Somewhat similar effects are found by Aidis et al. (2012), who employ repeated cross sections of GEM data on nascent entrepreneurship in 47 countries (1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005). They find that less corruption might have a positive effect on nascent entrepreneurship (the effect becomes even stronger if the richest 10% of the countries are removed from the analysis)-implicitly suggesting that corruption is more harmful in less developed economies.
Research has also applied a European and/or transition perspective. For 17 West-European countries (and the USA), Gohmann (2012) finds that high levels of corruption generally depress self-employment. In a similar fashion, Ovaska and Sobel's (2005) panel analysis of new enterprise creation in ten CEE countries (1995)(1996)(1997)(1998)(1999)(2000) concludes that a high level of corruption has a negative effect on business entry. However, similar to previous studies that use very wide country samples, existing research also suggests different effects depending on context. Differences between transitional regions are found by Hashi and Krasniqi (2011), who analyze small and medium-sized enterprise (SME) growth in advanced European CEE economies (e.g., Poland) and laggard European transition economies in South-East Europe (SEE; e.g., Albania). They find that high levels of corruption are obstacles to firm growth in CEE countries but act as facilitators in more laggard European regions, where they "grease the wheels." In a similar vein, Krasniqi and Desai (2016) use a wide variety of 26 transitional countries from both the CEE and SEE regions and the former Soviet republics for 1998-2009 and investigate whether the rate of high-growth firms is affected by formal and informal institutions; they conclude that corruption may "grease the wheels" for entrepreneurship.
In sum, previously generated results generally support the notion that corruption affects entrepreneurship negatively. Earlier empirical studies hypothesize and find that a better control of corruption (high levels of corruption) is generally beneficial (detrimental) to entrepreneurship-particularly in high-income and mature market economies (e.g., Avnimelech et al. 2014;Dutta and Sobel 2016;Estrin et al. 2013;Gohmann 2012). However, past results on the linkages between entrepreneurship and corruption in transitional countries or in CEE economies are somewhat inconclusive. In (developing and) transitional countries, entrepreneurship may benefit from corruption (Dempster and Isaacs 2017; Krasniqi and Desai 2016). However, for CEE countries in particular, the existing empirical evidence of detrimental effects from corruption calls these conclusions into question, due to these countries' relatively higher level of development compared with other transitional or postsocialist regions (Hashi and Krasniqi 2011;Ovaska and Sobel 2005).

Property rights
A substantial body of literature covers the evolution and significance of property rights institutions (e.g., Acemoglu and Johnson 2005;North and Thomas 1973;Rosenberg and Birdzell 1986). The quality and protection of property rights depend on both the regulatory framework and rule of law, and the absence of threats. A developed property rights system is fundamental to the rise and functioning of a modern market economy. The definition and enforcement of property rights is an important feature in formal institutions (Williamson 2000).
As maintained by Estrin et al. (2013), property rights relate to the fundamental threat of expropriation. Societies that provide secure property rights are seen as being more likely to engage in the creation of new wealth (Amorós et al. 2019;Williamson 2000). Consequently, property rights have distinct effects on both aggregate investment and economic growth (Acemoglu and Johnson 2005). In a similar vein, recent research finds linkages between property rights and entrepreneurship: in a very broad spectrum, property rights are thought to create (dis)incentives for potential entrepreneurs, to promote or prevent firm growth, and to affect business owners' decision to reinvest economic returns (Amorós et al. 2019;Johnson et al. 2000;Sobel et al. 2007;Stenholm et al. 2013). The quality of property rights will thus have direct effects on formal (registered) business activity: weak property rights lower intentions and aspirations for formal entrepreneurship (Estrin et al. 2009). Individuals in environments where their time, efforts, and assets are protected are also more prone to engage in formal business activity. Strong protection of property rights will also discourage rent-seeking activities from predatory entrepreneurs who otherwise would seek to take profit from other (potential) entrepreneurs. Consequently, it could therefore be expected that the level of all forms of entrepreneurship-be it nascent entrepreneurship, new firm start-up, or self-employment-would increase with greater protection of property rights (Chowdhury et al. 2015). Weak property rights are perhaps not an obstacle for entrepreneurial activity per se-under weak protection, unregistered and informal entrepreneurship can be a substitution for formal entrepreneurship-but it is likely that the rate of formal entrepreneurship will be negatively affected by weak property rights (Estrin et al. 2009). Stenholm et al. (2013) use aggregated indicators to capture different regulatory and normative institutional dimensions in which property rights are a subcomponent. They employ a cross section of 63 countries and find support for the hypothesis that the rate of newly registered limited-liability companies is positively related to the quality of formal rules (for a somewhat similar approach, see Thai and Turkina 2014). Furthermore, Sobel et al. (2007) employ a cross section of 23 Organisation for Economic Co-operation and Development (OECD) countries (including Hungary and Poland) and regress the TEA measure against an economic freedom index that, among other things, includes private ownership rights. 1 Sobel and colleagues find that a higher degree of economic freedom positively affects TEA. However, Bjørnskov and Foss (2008) analyze cross-sectional TEA data on regulatory and legal quality (which includes property rights) for 29 developing and developed economies and find neither any discernible effect from legal quality nor any distinct effect for post-socialist countries. In the same way, Estrin and Mickiewicz (2010) discover no linkages between nascent entrepreneurship and the protection of property rights in a wide variety of developing, transition, and advanced economies. Yet other researchers have found more conclusive evidence: Nyström (2008) uses a long-term approach and employs a 30-year-long panel of 23 OECD countries . Nyström finds that the rate of self-employment is positively affected by the protection of property rights.
Some previously mentioned studies have also analyzed the extent to which both corruption and property rights are linked to entrepreneurship. As previously reported, Dempster and Isaacs (2017) find evidence for a link between TEA and corruption but cannot find any conclusive evidence for a link between TEA and formal institutional property rights (however, other formal institutional indicators are significant). Conversely, Aidis et al. (2010) discover no effect of corruption on nascent entrepreneurship but find significantly positive effects from property rights protection. Furthermore, already noticing the somewhat inconclusive effects of corruption, Aidis et al. (2012) find even weaker evidence for different indicators for formal institutional factors (i.e., aggregate measures of market freedom, including property rights). On the other hand, in a study of the growth aspirations of entrepreneurs in a panel of 42 countries, Estrin et al. (2013) find that not only does corruption have negative impacts, but entrepreneurship is also positively affected by efficient property rights. Similarly, as with the effect from corruption, Chowdhury et al. (2015) come to mixed conclusions in their study and find (unexpectedly) a negative effect from stronger protection of property rights on self-employment (but positive effects on nascent entrepreneurship and on start-up rates).
Relatively fewer studies have focused on different (transition) contexts. Ovaska and Sobel's (2005) panel analysis of new enterprise creation addresses both corruption and economic freedom in CEE countries during 1995-2000. In contrast to other studies (e.g., Estrin et al. 2013), variation in "economic freedom"-which indirectly includes property rights-has no discernible effect. Furthermore, Desai et al. (2003) analyze several measures of entrepreneurship (i.e., firm entry, exit, and size) using cross-sectional data on incorporated businesses in European countries and find that greater protection of property rights increases entry rates and decreases exit rates. Nevertheless, these effects are not equally pronounced in all parts of Europe: stronger property rights are unambiguously associated with higher rates of entry in Central and Eastern (and South Eastern) Europe. Moreover, although countries with stronger protection of property rights have significantly lower exit rates, this effect is discernible only in CEE countries and not in Western European countries. Consequently, Desai et al.'s (2003) results show that entrepreneurship in European transition economies is more (positively) dependent on the quality of government (see also Amorós et al. 2019).
To some extent, in comparison with research on corruption and entrepreneurship, previous research results on the relationship between entrepreneurship and property rights are more inconclusive. Yet, when summarizing the rather scant literature, property rights are generally believed to affect entrepreneurship by creating (dis)incentives for both prospective and existing entrepreneurs; weak protection of property rights lowers entrepreneurship and entrepreneurial aspirations (Estrin et al. 2013;Nyström 2008;Sobel et al. 2007;Stenholm et al. 2013). The rather small body of empirical cross-country research on transition economies and/or CEE countries has either produced inconclusive results (Ovaska and Sobel 2005) or found that entrepreneurship positively responds to higher protection of property rights, similarly to (or more strongly than) mature market economies (Bjørnskov and Foss 2008;Desai et al. 2003).

Analytical framework
This article's focus on a European sample makes possible an analysis of a setting with rich institutional variation and ample variation in business activity (c.f., Desai et al. 2003;Grilo and Thurik 2006). Our study analyzes changes in entrepreneurial activity across both time and place. For such an approach, a key advantage is to apply an extended period of analysis-in our case, over more than 20 years. Such a research strategy can increase our understanding of current developments since entrepreneurship encompasses a sequence of events over time (Aldrich 2009;Shane 1996).

Self-employment as a measure of entrepreneurship
There is often a discrepancy between the diverse theoretical definition(s) of entrepreneurship, on the one hand, and available empirical indicators, on the other.
Quantitative research commonly measures entrepreneurship by using various stockand rate-based measures of self-employment, by using data on (new) SMEs, or by measuring attitudes and aspirations for entrepreneurship. Self-employment has been one of the most common indicators used in both policy and research and this measure has some advantages: self-employed persons must handle both risk and uncertainty. While self-employment partially covers some established theories on entrepreneurship, it is questionable whether it can capture, e.g., Schumpeterian entrepreneurship and innovation (Henrekson and Sanandaji 2019;Parker 2009;). Furthermore, different definitions of self-employment are not uncommon, even in the same country. This diversity may lead to considerable measurement errors both across and within individual countries and over time (Bjuggren et al. 2012;Parker 2009). Moreover, self-employment measures a running venture, rather than the process leading up to it. It is often difficult to assess whether (unpaid) family workers should be included, and there is often a "gray area" between paid employment and self-employment, since some employers organize their workforce using self-employment contracts (Bögenhold and Klinglmair 2017;Parker 2009). 2 Finally, the rate of self-employment is a static indicator that includes both new and incumbent firms and, from an occupational choice framework, can be seen as distinguishing between self-employment and wage labor (Van Praag and van Stel 2013). Self-employment does not always include all types of business activities and often excludes substantial parts of the private sector and other legal forms of businesses. Therefore, the institutions that form and affect self-employment might differ from those that form high-growth incorporated ventures (Stenholm et al. 2013). Yet, one of the main advantages of using self-employment data is that such data is often available for several economies and usually spans reasonably long periods.

Data and variables
The dependent variable in the analysis is self-employment, which we define as the non-agricultural self-employment ratio in 30 European countries. The ratio is calculated as the number of self-employed persons in relation to the total work force (Eurostat; see Tables 1 and 2). We have made extensive use of previous research, mainly those with cross-country approaches, to control for variables that may be linked to variations in self-employment. 3 Several earlier cross-country studies on variations in entrepreneurship and self-employment find that the level of female labor force participation has a negative relationship with the rate of entrepreneurship 2 An additional problem in international comparisons might be how to deal with the owners of larger businesses. These are known in the USA as the "incorporated self-employed," and are commonly counted as employees; however, they are included in the self-employment count in both OECD and European statistics (Blanchflower 2000). 3 Past studies have analyzed a set of institutional variables that are likely to affect self-employment; more specifically, different welfare system indicators such as unemployment benefit replacement ratios, employment protection strictness (Parker and Robson 2004;Robson 2003Robson , 2010, and taxes (Torrini 2005). Due to shortages in data coverage, we are currently unable to include or control for these measures; we are aware of that it might affect our empirical results. (Acs, Audretsch and Evans 1994;Blanchflower 2004;Carree et al. 2007;Parker and Robson 2004;Robson and Wren 1999;Verheul et al. 2006;Wennekers et al. 2010). Entrepreneurs often have a long history of employment, and women commonly have shorter employment histories than men due to breaks for getting and raising children. Furthermore, females tend to less risk taking, and to show lower preferences for business ownership and higher preferences for wage employment than males (Uhlaner et al. 2002;Verheul et al. 2012). If more women participate on the labor market this will lower the self-employment rate among women-therefore, higher female participation on the labor market means in itself a lower total rate of self-employment in the economy (Acs et al. 1994;Wennekers et al. 2010). To the best of our knowledge, this factor has seldom been accounted for in earlier studies on entrepreneurship, corruption and/or property rights. As an exception, Dutta and Sobel (2016) find a consistently negative effect on entrepreneurship from female labor market participation when investigating the effect from corruption. Thus, we include female labor market participation rate, measured as the percentage of female to male labor force participation rate (data from World Bank/International Labour Organization, ILO 2022) as a control variable and expect it to be negatively related to self-employment. Government size has regularly been used as both a control and independent variable. The basic idea is that a large government has the potential to crowd out (prospective) entrepreneurs from the market and thereby depress the rate of entrepreneurship (e.g., Belitski et al. 2016;Nyström 2008). Recently, Audretsch et al. (2022) show that it may not be the size of government (government consumption) per se, but the very type of government expenditure that is of importance: a large government in combination with good/sound governance may encourage entrepreneurship. Based on extant literature, we assume a priori that government size is negatively related to self-employment. In the present study, the variable size of public sector is measured using government spending as percent of the gross domestic product (GDP). 4 Furthermore, earlier research has shown a counter-cyclical relationship between self-employment and unemployment (e.g., Blau 1987; Bögenhold and Staber 1991);  (2013) find that the equilibrium entrepreneurship or business ownership rate falls as a result of more individuals engaging in tertiary education. Capable persons that are eager to run large firms will increase with higher enrollment in tertiary education; the demand for employed persons in large businesses will increase, and the rate of business ownership and self-employment will therefore decrease (see also Chowdhury et al. 2015). 5 In line with the existing literature, we expect a negative relationship between human capital and self-employment.
The main independent variables of interest in the study are corruption and property rights. The corruption index (World Bank; Kaufmann et al. 2010) measures the control of corruption and has a very wide coverage of countries. The index reflects perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as the "capture" of the state by elites and private interests. The index originally ranges from −2.5 to 2.5, with higher values indicating higher (better) government performance. We have rescaled the index, so it ranges from 1 to 5. In contrast to other indices on corruption (e.g., Transparency International's Corruption Perception Index), the World Bank index covers both private and public corruption (for a discussion, see Rohwer 2009). The index has been extensively used in previous research on entrepreneurship and institutions (e.g., Anokhin and Schulze 2009;Dreher and Gassebner 2013;Estrin et al. 2013). Based on extant literature on the relationship between entrepreneurship and corruption, we assume that better control of corruption will increase self-employment in European economies.
The property rights variable comes from the Heritage Foundation, which publishes annual data on economic freedom. The property rights index is a sub-index in the Heritage Foundation's economic freedom index (Miller et al. 2018), aimed at ranking countries on several different factors that are averaged equally into a total score. The index is constructed from statistics from institutions such as the World Bank and the International Monetary Fund. A number of studies have used this data when studying entrepreneurial behaviors and patterns (Aidis et al. 2010(Aidis et al. , 2012Estrin and Mickiewicz 2010;Ovaska and Sobel 2005). The Heritage property rights sub-index ranges from 0 to 100. We have rescaled the index to 1-5, with the highest score (5) denoting property rights that are guaranteed by the government, where the court system enforces contracts efficiently and the justice system punishes those who illegally confiscate private property. Thus, low(er) scores indicate weak(er) protection of property rights. In the present study, we expect that higher property rights scores will affect self-employment positively.
There is considerable variation in non-agricultural self-employment in Europe, across both time and place ( Table 1). The CEE states (including the Baltic states) that are included in the present case are Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia. As shown in Table 1, there is substantial variation over both time and economiesincluding within the group of CEE countries. For instance, the Czech Republic already scores relatively high during the first year of observation (in reality, from 1997) and increases its self-employment ratio over time. Somewhat differently, Estonia, for example, starts from a relatively low level (1997) but exhibits an increasing trend. In contrast, Lithuania remains fairly stable over time, while Hungary starts from a reasonably high level in 1996 then shows a clear negative trend. However, the present sample also shows ample variation in and between non-CEE countries. Indeed, variations in self-employment are substantial in several regions across the world, and it is generally difficult to discern any global trend.
Without a doubt, the East European (transition) economies are a heterogeneous group (Chepurenko 2015;Dana 2005). Despite this heterogeneity, in the quantitative analysis, we create a group that consists of the eleven CEE countries in the sample. The CEE group is contrasted against the European sample at large-accordingly, against several North, West, and South European economies. We acknowledge that this procedure may not capture the heterogeneity and rich variation between countries. However, past cross-country research has found that the CEE (and Baltic) economies display several shared or mutual economic, political, social, and cultural characteristics that can be linked to both the quality of and the propensity for entrepreneurship, as well as to economic activity in general, when compared with more mature (European) market economies (see e.g., Achim et al. 2019;Boltho 2020;Liñán and Fernandez-Serrano 2014). It has been a common strategy in the literature to group together country economies with a post-socialist origin or with CEE country status (Aidis et al. 2010;Bjørnskov and Foss 2008;Boltho 2020;Desai, et al. 2003;Dreher and Gassebner 2013;Estrin and Mickiewicz 2010;Verheul et al. 2006).

Empirical model specification
To some extent, the existing literature has ignored the problem of cross-sectional dependence. Investigating previously employed panel models in the literature, we find that the fixed-effect model and the random-effects model are inappropriate due to possible specification errors. An important issue concerning panel data concerns the cross-sectional dependence of individual observations. If independence is rejected, traditional fixed-effect and random-effect models can be inconsistent. The cross-sectional dependence (CSD) test proposed by Pesaran (2004Pesaran ( , 2015, with the null hypothesis of cross-sectional independence, is therefore employed in our study for checking the CSD of all the variables involved, as well as their first-order differences. Except for corruption and property rights, the CSD test results show that all variables are cross-sectionally dependent (Table 2); that is, the variables are highly correlated across individual countries. The results also show that, except for property rights, all first-order differences are cross-sectionally dependent. Interestingly, the difference would not eliminate the union-wide impacts. Furthermore, to highlight the property of the time dimension, we use panel unit root tests to check for panel stationarity. This exercise is complicated by the issue of CSD. When the variables are cross-sectionally independent, it is possible to employ the first-generation panel unit root test by Maddala and Wu (1999). Cases of levels with trend and first-order differences without trend are considered. For the variables that are CSD, the second-generation panel unit root test is employed-namely, the cross-sectional augmented Im-Pesaran-Shin (IPS) test (CIPS; Im et al. 2003;Pesaran 2007). The CIPS is based on the IPS panel unit root test. We only report the test without lags and with the trend for levels and without the trend for first-order differences. The results in Table 3 show that several variables-namely, self-employment, unemployment, female labor market participation, government spending, tertiary education, and corruption-are clearly I(1) processes. GDP is stationary. The null assumption of I(1) for property rights is rejected at the 5% level. Thus, we have mixed stationary and nonstationary variables in our study.
The dependent variable in our analysis is the (non-agricultural) self-employment rate, se. From the previous discussion, we consider two indicators for the institutional setting: corruption, co, and property rights, pr. We also consider the variables (the log of) GDP, y; unemployment, u; and female labor market participation, f; along with gs, the ratio of governmental spending; and tertiary education, te. At the same time, we introduce one dummy variable, cee, to indicate whether a country is a CEE country (with a value of 1). Thus, we construct two variables for CEE countries: cee*co = psco, and cee*pr = pspr.
We start with the basic model: for i = 1, …, N and t = 1, …, T i . Here, we assume that all slopes β j s are common for each country. i is time invariant and captures heterogeneity across countries. u it is the error term. A classical panel data model for which the ordinary least squares (OLS) is the best linear unbiased estimator requires homogeneous and no correlations in both time-series and cross-sectional dimensions. However, these requirements might not be realistic: Pesaran (2006) points out that, when there is CSD in the error terms, the OLS is generally not consistent. Pesaran (2006) further suggests a CCEP estimator, which imposes homogeneous slopes on independent variables across the sectors but allows heterogeneous loading coefficients on a multifactor error structure using the cross-sectional averages of the variables involved to capture unobservable common effects.
where the cross-sectional averages are denoted with the upper bar " − " in the specification in Eq.
(2). The CCEP estimator is consistent when CSD appears, and performs well with a moderate sample size. Although Pesaran (2006) only allows stationary variables, Kapetanios et al. (2011) investigate the case in which nonstationary variables are included. They find that the CCEP estimator works well regardless of I(1) or I(0) variables. They further show that, if the error terms are panel cointegrated, the estimated relation can be regarded as the long-term relation. If the true model (1) consists of heterogeneous slopes, imposing slope homogeneity in (2) results in inconsistent and biased results. Thus, our study carries out the homogeneity slope Delta test (Pesaran and Yamagata, 2008) to ensure homogeneous slopes. In addition, no time dummy is included in (2), since our results show that including time dummies may fail to eliminate cross-sectional dependence. Moreover, (2) is a static model which is useful for identifying the long-run relationship. Nevertheless, (2) can be extended to a dynamic model which can capture persistency of self-employment and control for endogeneity. We also show why we consider the CCEP model to be superior to other panel models, as the latter introduce the risk of producing erroneous or unstable results.
(1) se it = y y it + u u it + f f it + gs gs it + te te i + co co it + pr pr it + psco psco it + pspr pspr i + i + v it (2) se it = y y it + u u it + f f it + gs gs it + te te i + co co it + pr pr it + psco psco it + pspr pspr it + sei nrse t + yi y t + ui u t + f f t + gsi gs t + tei te t + coi co t + pri pr t + psco psco t + pspr pspr t + e it ,  As a robustness check, we also employ panel-corrected standard errors (PCSE) to overcome these possible specification errors. As shown by Beck and Katz (1995), the PCSE estimator performs better than the generalized least squares (GLS), given the size of data commonly used in social science projects.

Empirical results
The empirical results are presented in Table 4. For all models, the CSD test on the residuals according to Pesaran (2004Pesaran ( , 2015 cannot reject the null of crosssectional independence (weak dependence). Thus, the CCEP models fully control the CSD. The panel unit root tests are then based on the first generation of the panel unit root test (PURTest) suggested by Maddala and Wu (1999), which requires cross-sectional independence. The result of the tests shows that, irrespective of the number of lags (up to 2), the unit roots of the residuals can be rejected. According to Kapetanios et al. (2011), this result indicates a panel cointegration relation among the variables involved when some of the variables are nonstationary, I(1). In addition, according to the Delta statistics, all models without CCE dummies do not reject the null of homogeneity of slopes. These results support the use of the CCEP specification (2). We consider seven related specifications. Model 1 does not include the institutional variables of corruption and property rights. As can be observed, government spending does not appear to "crowd out" self-employment. Furthermore, tertiary education does not seem to be related to variations in self-employment. However, GDP, unemployment, and the rate of female labor market participation show significant effects: economic growth has positive effects: a 1% increase in economic growth leads to a 2.4% points increase in self-employment. The positive relationship between unemployment and self-employment indicates a counter-cyclical relationship. Here, a 1% point increase in the unemployment rate leads to a 0.06% points increase in self-employment. Furthermore, the female labor market participation rate has significant effects: in economies for which female participation is one percentage point higher, or where it increases, selfemployment drops by 0.1% points. (Indeed, with but one exception [Model 6], both the variable female participation and the rate of unemployment consistently show significant effects across all models.) Model 2 introduces the effect from corruption without distinguishing the group of CEE countries. Viewed in isolation, stronger control of corruption appears to have no significant relationship with self-employment in European economies: corruption neither "sands" nor "greases the wheels." Model 3 introduces the interaction for corruption in CEE countries. As can be observed, the effect is negative but insignificant. In short, CEE countries do not respond differently to corruption in comparison with non-CEE countries. This result indicates that self-employment-regardless of whether in CEE countries or in other European countries-is not affected by stronger control of corruption; the impacts from corruption are not robust in terms of sensitivity toward the specifications (models). Model 4, introducing property rights, clearly shows that stronger protection of property rights significantly and positively affects self-employment in Europe: a one standard deviation increase in the property rights index (from the mean of 3.63 to 4.57; see descriptive statistics in Table A1 in the Appendix) is associated with a 46% points increase (= 0.490 × 0.94 × 100) in the ratio of self-employment. Model 5 adds the interaction variable for CEE countries and property rights. The coefficient for property rights remains positive and significant. The interaction coefficient is negative but insignificant. Consequently, the group of CEE countries does not respond differently to variations in property rights protectionregardless of whether stronger or weaker, or in an opposite way.
Model 6 includes both the main effects variables of corruption and property rights. We find moderate evidence (at the 10% level) that stronger control of corruption has positive effects on self-employment; again, however, this result is not stable. Furthermore, stronger protection of property rights significantly increases the ratio of self-employment rate with 38% points (= 0.406 × 0.94 × 100). Finally, Model 7 confirms the previous results for the CEE group (Models 3 and 5): neither the variable for corruption nor the variable for property rights reveal any different effects across CEE and other European countries. Once more, we note that the results on corruption are not robust; however, the estimations on property right are robust and consistent across different specifications. In conclusion, during the two most recent decades, the ratio of self-employment in Europe and in CEE countries increases with stronger protection of property rights.
As stated earlier, that the fixed-effect model and the random-effects model are inappropriate due to CSD. Other studies in the literature employ random-effects models (e.g., Autio and Fu 2015;Ovaska and Sobel 2005), fixed-effects models (see e.g., Krasniqi and Desai 2016;Nyström 2008), both random-and fixed-effects models (Chowdhury et al. 2019), or panel-corrected standard errors (PCSE) models (e.g., Dreher and Gassebner 2013). For comparative reasons, emphasizing the advantages of the CCEP model employed here, we have analyzed the data used in the present study using the fixed-effects and PCSE models. The latter could be also considered as an alternative robustness check.
We first estimate the fixed-effect model according to Eq. (1) without considering serial correlations and cross-sectionally dependent residuals. The result is reported in columns 1 and 2 in Table 5. The fixed effects with year dummies fail to detect any significant impact according to the current set of explanatory variables. In the lower panels of columns 1 and 2, the statistic of the test on serial correlation is carried out according to Wooldridge (2002) that is significant at 1%. Thus, we reject the null of no autocorrelation. We assume a common ρ for all countries. The estimated ρ in the AR(1) error terms is 0.95. Furthermore, following Pesaran (2004), the null of cross section independence is rejected at 1%, resulting in CSD residuals. We also carry out the panel unit root test on the error terms. Due to the CSD, we employ the Pesaran (2007) CIPS test. The results from the tests are mixed depending on the number of lags involved in the tests. When there is no lag, the residuals are stationary. However, if we assume one and two lags, respectively, the null hypotheses unit root is not rejected. In sum, we need to deal with specification errors such as autocorrelation and CSD.
A linear regression with PCSE is employed to overcome the issue of autocorrelated residuals. The result is reported in the third and fourth columns in Table 5. As for the main effects (column 3), the PCSE model shows that corruption control has significant and negative effects on self-employment. In disagreement with the CCEP model results, this suggests that less control of corruption positively affects self-employment. Unlike the CCEP regression, property rights shows no significant effects. Turning our attention to the effects in the CEE economies (column 4)-and unlike the CCEP model-the PCSE model again shows a (highly significant) negative effect from stronger control of corruption in CEE economies. Overall, the results in columns 3 and 4 are in line with earlier results (e.g., Dreher and Gassebner 2013) and suggest that weaker control of corruption "greases the wheels" in both Europe in general and in the CEE economies.

Discussion
A growing body of cross-country research has focused on the linkages between corruption, property rights and entrepreneurship. The present article follows this stream of research. However, in contrast to several earlier efforts, we have deliberately focused on a European and CEE country setting. Excluding transcontinental European countries (e.g., Russia and Turkey), our analysis has nonetheless covered 30 out of the 44 existing European economies, including eleven CEE economies. By using a research strategy that deliberately avoids the inclusion of developing and low-income economies, we believe that we have been able to more accurately isolate and analyze the effects of these key variables for countries with (relatively) high levels of economic development. Our research strategy specifically addresses these issues as well as research on entrepreneurship in transition and in post-socialist (European) economies. This strategy has facilitated an assessment of whether entrepreneurship in post-socialist CEE countries responds differently to the institutions of corruption and property rights in comparison with that in mature and long-standing (European) market economies. The independent variables used as controls showed both expected and unexpected relationships with self-employment variations in Europe. The rate of female labor market participation exhibited a systematically negative effect. These results are in line with several earlier cross-country studies (e.g., Acs et al. 1994;Wennekers et al. 2010). This variable has generally not been included in past research on how property rights, corruption and other institutional conditions affect entrepreneurship. We believe that it should be taken into consideration in future research. Economic conditions and aggregate demand also affected selfemployment: GDP growth (Anokhin and Schulze 2009) and in particular unemployment (e.g., Bögenhold and Staber 1990), revealed systematic and expected effects. Surprisingly, and different from much previous research (see for instance Estrin et al. 2013;Dempster and Isaacs 2010;Nyström 2008), we found no significant evidence for a negative effect from government size. One explanation could be the European sample in this study, including several countries with generally high levels of government spending. As discussed by Audretsch et al. (2022), large government size must not automatically affect the level of entrepreneurship negatively. We had also expected that a higher rate of individuals with tertiary education would have negative effects on self-employment (see Chowdhury et al. 2015;Van Praag and van Stel 2013). Here, we found a negative yet systematically non-significant relationship. Future research would perhaps benefit from other human capital indicators.
Using a novel panel analysis methodology (CCEP), our study found that the quality of government is clearly linked to variations in entrepreneurial activity. More specifically, stronger protection of property rights has generally a positive impact on self-employment, both in CEE countries and in Europe at large. Several theories maintain that institutions affect entrepreneurial behavior in terms of both the quality and quantity of entrepreneurship (Baumol 1990;North 1990;Shleifer and Vishny 1993;Williamson 2000). In several areas, the results also align with empirical research in the domain of entrepreneurship. However, a number of studies have found little or no support for the notion that property rights-or other indicators of economic freedom-have an empirical link to entrepreneurship (see Aidis et al. 2012;Bjørnskov and Foss 2008;Dempster and Isaacs 2017). Contrary to these results, we discovered a significant, positive link between property rights protection and entrepreneurial activity: economies with weaker protection exhibit lower ratios of self-employment. When it comes to research on transition or CEE economies, the rather small existing body of empirical cross-country research shows relatively inconclusive results. Existing cross-country research on postsocialist economies has suggested a greater sensitivity to the quality of protection of property rights (Desai et al. 2003;Ovaska and Sobel 2005). Interestingly, the results of the present study show that post-socialist CEE countries do not stand out. Thus, the quality of this fundamental market-supporting institution can be directly linked to variations in self-employment in Europe as a whole. However, it should be noted that our data covers a large part of the 2000s and 2010s, during which several of the CEE countries were on a route toward convergence and adapted to common market rules (see e.g., Szerb and Trumbull 2016). This may explain why variations in self-employment in the present paper reacts in the same manner to this key institution in CEE countries as in non-CEE countries. Our study contributes with new knowledge on how property rights are related to entrepreneurship (Urbano et al. 2018).
Corruption control showed no robust linkages to self-employment since the majority of specifications showed non-significant results. Thus, the assumption in the present paper that higher control of corruption would enhance self-employment ratios received no support. Furthermore, the group of CEE countries did not exhibit any deviating behavior as regards the positive effect from stronger control of corruption -the impacts from corruption were not robust in terms of sensitivity toward the specifications. However, it should also be noted that our results lend no support for the alternative hypothesis that corruption "greases the wheels," discussed at length in the economic development literature (e.g., Aidt 2009) and in entrepreneurship (e.g., Dutta and Sobel 2016)-neither in Europe as a whole nor in the CEE economies. Some scholars maintain that, in less developed economies, corruption may have a positive effect on the supply-side, helping to increase the amount of entrepreneurship (Avnimelech et al. 2014;Dreher and Gassebner 2013). Furthermore, past results on corruption in postsocialist transition countries are somewhat inconclusive, with some studies maintaining that corruption "helps" entrepreneurship (Dempster and Isaacs 2017; Krasniqi and Desai 2016). Others have questioned whether this would be the case when it comes to CEE economies, which have had a higher level of development than other transitional regions (Hashi and Krasniqi 2011;Ovaska and Sobel 2005). This article's results are in line with these studies, finding no positive effects from less corruption control in CEE countries.

Conclusions
Since increasing self-employment is an important goal for European governments, then stronger corruption control and stronger protection of property rights could be viewed as means of achieving this goal. However, a number of individual countries in our analysis reveal relatively low or even falling scores on these two institutional variables during the period of investigation-including some long-standing European market economies and some CEE countries that are newer member states. Therefore, European governments that weaken their protection of the ownership of capital, property, and intellectual ideas-fundamental institutions for a market economy-should consider how this may affect business activity and entrepreneurship in the long run. Similarly, we found no evidence for the notion that (some) corruption "greases the wheels." As found in earlier research, the amount of entrepreneurship in developing economies, or in other post-socialist transition countries in other regions, may benefit from some corruption-but such is not the case for the eleven Baltic and CEE countries analyzed in the present study.
There are some limitations to our study that should be addressed in future research. Future studies on corruption could benefit from a stronger focus on emerging countries and could expand future samples to include more countries. Emerging economies often have more variation in corruption levels. The present paper has mainly focused on a sample of economies that generally are highly developed and included a smaller number of post-socialist countries that are close to convergence. Indeed, it is worth pointing out that several countries in the present study also are OECD countries, which as a rule have lower corruption levels than non-OECD countries. This might explain why we did not find any conclusive links between self-employment and corruption control.
However, since corruption has been found to affect business activity and entrepreneurship also in high-income economies (Avnimelech et al. 2014;Dutta and Sobel 2016;Gohmann 2012), future research on both developing, transition and high-income economies, respectively, could also benefit from employing alternative corruption indicators (such as Transparency International's corruption index). Similarly, future studies would benefit from using alternative measurements of entrepreneurial activity-self-employment is but one of several indicators. One advantage with self-employment is that there are often long series available, but it is less suitable for measuring entrepreneurial aspirations or various forms of high-impact entrepreneurship. It is clear that different forms of entrepreneurial activity respond differently to various institutional conditions (Chowdhury et al. 2015). Here, we have not taken these dimensions into account.
Generally, and pertaining to all different types of national economies, we also find it imperative to expand research on entrepreneurship and property rights in a cross-country setting, since there has been less research on the effects from this fundamental institution (Urbano et al. 2018). Overall, and given the availability of data, the state of knowledge on the dynamics between entrepreneurship and institutions can be further expanded following our suggestions and by following the methodological framework employed in the present paper. The CCEP approach used here overcomes several methodological issues and has several advantages over the fixed-effects and PCSE models that commonly have been employed in cross-country research on entrepreneurship, corruption and property rights.
Author contributions All authors have contributed to the study conception. Research idea, data collection, literature search, and first draft of the article were carried out by MB. MB and KG contributed to the literature framework. Methodological framework and econometric analyses were developed and carried out by XL. MB, KG, and XL have read, revised, and approved the final manuscript. Final editing of the manuscript was carried out by MB and XL.
Funding Open access funding provided by Södertörn University. This study is part of the project Firm demography and entrepreneurship in Eastern and Central Europe and in the Baltic region, and the authors acknowledge support from The Foundation for Baltic and East European Studies (Östersjöstiftelsen).

Availability of data and materials
The datasets generated during the current study are available from the corresponding author on reasonable request.
Code availability Not applicable.