Long-term persistence in entrepreneurship: the case of post-socialist Czechia (1930–2011)

ABSTRACT This paper explores the long-term persistence of interregional differences in self-employment rates. The analysis, based on spatial Durbin error models, was performed for 206 Czech regions from 1930 to 2011. The results obtained supported the ‘persistence claim’ because regional self-employment rates in 1930 were found to be positively and significantly associated with those obtained in 1991, 2001 and 2011. Moreover, this study contributes to the knowledge about transmission channels of historical entrepreneurial experience. First, a new mechanism related to the career and educational paths of individuals with a pre-socialist entrepreneurial family background is suggested. Second, the newcomers’ capacity to absorb knowledge and skills is emphasized as crucial for the functioning of pre-socialist ‘entrepreneurial memory’.


INTRODUCTION
The last three decades have witnessed a growing interest in evolutionary economic geography. The significance of history has been demonstrated in understanding recent regional differentiation which implies that regional development is path dependent (Fotopoulos & Storey, 2017). The ideas of evolutionary economic geography have been applied in different contexts and settings, including research on entrepreneurship, which indicates that regional entrepreneurship rates generally persist over time. The arguments based on slowly changing regional determinants of entrepreneurship and path dependence in regional entrepreneurship have been used to explain the time persistence of entrepreneurship rates across regions (Fotopoulos & Storey, 2017;Koster & Hans, 2017).
Some research has explored persistence in regional entrepreneurship. At least four studies have attempted to do this by analysing persistence in regional entrepreneurship over a long period of approximately one century. Fotopoulos and Storey (2017) confirmed the time persistence of regional entrepreneurship rates in England and Wales between 1921and 2011 and  obtained similar results for East and West Germany , while Fritsch et al. (2019c) found the same for the Kaliningrad region , as did Fritsch et al. (2019b) for Polish regions that belonged to Germany until 1945Germany until (1921Germany until -2017. Fritsch et al. (2022) also identified the long-term effect of the historical regional level of self-employment in knowledge-intensive manufacturing industries on recent new business formation in Poland. The latter three studies are of particular interest in that they imply that the effect of pre-socialist entrepreneurship rates was not entirely destroyed even under the hostile conditions of socialism. Two transmission channels of historical entrepreneurial experience are emphasized in these studies: (1) entrepreneurial values, attitudes and experience passed down from previous generations; and (2) pre-socialist 'entrepreneurial memory'. This paper extends the previous work for another post-socialist country, namely Czechia, which has not yet been the object of such research (e.g., see Dvouletý, 2017;and Lukeš, 2017, for some recent studies dealing with regional entrepreneurship in Czechia, but not over a long period). The main research question explored in this study is whether the interregional differences in selfemployment rates in 1930 have, after controlling for other influences, survived the hostility of socialism towards entrepreneurship in Czechia. If so, this is another argument for the persistence of pre-socialist entrepreneurship in a country that followed the German model of delayed industrialization with a strong entrepreneurial sector in the interwar period (Cieślik & van Stel, 2014). However, this sector faced suppression after the communist takeover in 1948 (Clark, 2000), forcing Czechia to create a new institutional framework for the rapid redevelopment of entrepreneurship after the fall of socialism in 1989 (Cieślik & van Stel, 2014).
This study also contributes to the existing knowledge by further elaborating on the two given transmission channels of historical entrepreneurial experience for East Germany, the Kaliningrad region and Poland. Concerning the intergenerational transmission of entrepreneurial values, attitudes and experience, data from social stratification research in Czechia enables us to further develop the issue by looking at the career and educational paths of individuals with a pre-socialist entrepreneurial family background. These individuals were significantly more likely to achieve high occupational and educational status in the late 1980s. Subsequently, the acquired relational and human capital, and the 'entrepreneurial-like skills', increased the likelihood of them running businesses in the post-socialist period.
Concerning the pre-socialist 'entrepreneurial memory', the expulsion of approximately 3 million Germans from the Czech Sudetenland after the Second World War offers a chance to discuss whether a more or less complete exchange of local population negatively influenced the persistence of pre-socialist entrepreneurship in Czechia. Fritsch et al. (2022, p. 398) proposed a mechanism based on 'the presence of a place-based collective memory about historically successful (knowledge intensive) entrepreneurship that does not require a persistent population'. However, this mechanism is weak in our study, likely due to a low capacity of new settlers in the Czech Sudetenland after the Second World War to absorb this kind of knowledge, that is, 'entrepreneurial memory'. Hence, further research in the changes in the population is desirable to better understand the functioning of this transmission channel in different contexts.
The remainder of this paper is structured as follows. Section 2 reviews the literature concerning previous work, while the methodology is provided in section 3. Section 4 presents the empirical results obtained, followed by a concluding section.

LITERATURE REVIEW
Persistence in regional entrepreneurship, especially that of new business creation, has been noted in several studies from Western European countries (e.g., see Fritsch & Mueller, 2007, for Western Germany;Andersson & Koster, 2011, for Sweden;Fotopoulos & Storey, 2017, for the UK;and Koster & Hans, 2017, for the Netherlands). To explain this persistence, arguments regarding the longterm stability of regional determinants of entrepreneurship and path dependence in regional entrepreneurship are invoked (Andersson & Koster, 2011;Koster & Hans, 2017). The underlying mechanism of the latter source of persistence comprises two elements: (1) regionally embedded entrepreneurial activities and (2) positive feedback and self-reinforcing processes (Fotopoulos & Storey, 2017).
Several arguments have been made regarding the first element on the regional embeddedness of entrepreneurship. Many of these are based on a regionally embedded entrepreneurial culturelocalized values, attitudes and other informal institutions leaning towards entrepreneurshipthat supports the social acceptance of entrepreneurs and the motivation to start a business (Fritsch et al., 2019a). Fotopoulos and Storey (2017) add the importance of the spatially bounded human and social capital needed to identify and exploit entrepreneurial opportunities, and the role played by the entrepreneurs' emotional affinity to their relatives and friends. Positive feedback and selfreinforcing processes are then created through: (1) the intergenerational transmission of entrepreneurial values, attitudes, knowledge and experience; (2) intergenerational transfer of entrepreneurship (e.g., through inheritance); (3) new entrepreneurial opportunities created by previous acts of entrepreneurship; and (4) entrepreneurial supporting services and institutions (Fotopoulos & Storey, 2017;. The theoretical discussion on persistence in regional entrepreneurship differs for post-socialist Central and Eastern European countries. This is because the entrepreneurial culture of these countries, which developed before the Second World War, was exposed to a hostile environment during socialism, and self-employment was allowed only in a limited number of occupations (Wyrwich, 2012). Consequently, positive feedback and self-reinforcing processes directly related to entrepreneurs and continual acts of entrepreneurship were missing in socialist countries .
Nevertheless, Fritsch et al. ( , 2019a found a positive and significant association for East Germany between pre-socialist self-employment rates and post-socialist entrepreneurial activity, after controlling the influence of slowly changing regional determinants of entrepreneurship. A similar conclusion was reached by Fritsch et al. (2019c) for the Kaliningrad region. These findings are a confirmation of the persistence of regional entrepreneurship despite closed traditional transmission channels. Moreover, Fritsch et al. ( , 2019a emphasized the entrepreneurial values, attitudes and experience passed down from previous generations as the transmission channel of pre-socialist entrepreneurship in East Germany. Subsequently, the influence of pre-socialist 'entrepreneurial memory' on historically successful entrepreneurshiptransmitted by stories, statistics and other channelswas accentuated in the case of the Kaliningrad region and Poland, where the pre-socialist German population was replaced after the Second World War (Fritsch et al., 2019c(Fritsch et al., , 2022. This paper explores the persistence of entrepreneurship in Czechia over a long period, which includes 40 years in a Long-term persistence in entrepreneurship: the case of post-socialist Czechia (1930-2011) 713 centrally planned economy dominated by large stateowned enterprises and cooperatives andsimilar to the case of Polandalso the expulsion of Sudeten Germans after the Second World War. Notice that Smallbone and Welter (2001) point out the importance of pre-socialist entrepreneurial history for the rapid development of entrepreneurship in Czechia since the fall of socialism, and Illner (1996) also writes about the reproduction of presocialist regional patterns of entrepreneurship after 1989, without, however, an explanation of the underlying mechanisms of these associations. Consequently, the mechanism suggested by Benáček (2006) and embedded in the insights from social stratification research may be of interest. Benáček (2006) cites two sources of 'entrepreneuriallike skills' under socialism in Czechia. First, these skills were required for engaging in barter trade, moonlighting and do-it-yourself services, that is, in strategies helpful in obtaining scarce consumer goods and services under socialism. Second, managers of state-owned enterprises and cooperatives acquired 'entrepreneurial-like skills' when they developed their own initiatives and innovations, 'however absurd they were in both process and outcome' (Benáček, 2006(Benáček, , p. 1157. Therefore, Benáček speaks about three social groups at the dawn of socialism's demise: (1) marketeers as holders of financial capital stemming from the socialist 'shadow' economy; (2) nomenklatura as holders of financial and relational capital stemming from their political background; and (3) entrepreneurial outsiders divided into two subgroups of 'latent outsiders', possessing human and entrepreneurial capital, and 'passive outsiders', without such capital. The marketeers and nomenklatura were more likely to become entrepreneurs in the early phase of post-socialist transformation; however, the importance of more efficient 'latent outsiders' has increased substantially since the mid-1990s (Benáček, 2006;Clark, 2000).
In this context, the question is whether people with a pre-socialist entrepreneurial family background were significantly disadvantaged in their lives during socialism. It is well known that one of socialist strategies aimed at reducing social inequalities was based on discrimination against people of 'bourgeois origin', for example, through suppression of private property and a quota-based system in education (Andrle, 2001;Ganzeboom & Nieuwbeerta, 1999). This strategy was expected to lead to a downward shift in entrepreneurs' social status towards manual jobs and to improved access to education for children from workingclass families (Andrle, 2001;Hanley, 2003;Hanley & Treiman, 2004). However, the consensus in the literature is that strict measures against the bourgeois were applied only at the beginning of socialism and that their influence steadily eroded later (Andrle, 2001;Ganzeboom & Nieuwbeerta, 1999;Hanley, 2003;Hanley & Treiman, 2004). Consequently, at the end of socialism in Czechia, the pre-socialist social background lost its negative influence even on opportunities to work a managerial position (Hanley & Treiman, 2004) or to become a member of the Communist Party (Hanley, 2003). Moreover, parents' social background remained positively associated with the educational level of their children throughout socialism (Ganzeboom & Nieuwbeerta, 1999;Hanley & Treiman, 2004) due to a small differentiating effect of education on income (Andrle, 2001). Overall, it is likely that people with a pre-socialist entrepreneurial family background were able to transform their entrepreneurial capital into other forms during socialism.
As Illner (1996) points out, the post-socialist selfemployment rates of Czech regions are influenced by both 'pre-socialist and socialist legacies' and the postsocialist transformations. He notes the weakening effect of socialist legacies, while the importance of post-socialist transformations, which relate to the changing political, economic, social and cultural conditions of post-socialist societies, naturally increases. The question is whether pre-socialist self-employment rates influence post-socialist self-employment rates, despite both the hostile socialist environment and the effects of post-socialist transformations. Several components of socialist legacies and postsocialist transformations are worth underlining here because of their role as 'general'and, typically, slowly changingregional determinants of entrepreneurship: (1) changing industrial structures, deindustrialization and reindustrialization, and the rising importance of services; (2) the changing nature of international migration; (3) the growing importance of agglomeration (dis)economies, interregional migration, human capital, innovations and unemployment for socio-spatial differentiation; and (4) changing cultural modes, values, norms and behaviour (Smallbone & Welter, 2001).

METHODOLOGY
The data used in this study have been primarily gathered from the censuses conducted in 1930, 1991, 2001 and 2011. The data pertain to 206 districts that are administrative regions between the LAU-1 and LAU-2 (local administrative unit) levels. However, a difficulty arises because the 330 administrative regions for which data are available for 1930 do not correspond to the boundaries of the 206 recent districts (Jíchová et al., 2014). Therefore, the 1930 data were recalculated for the recent districts, using the methodological approach described in Appendix A in the supplemental data online.

Dependent variables
The self-employment rate is the dependent variable in the empirical analysis. It is defined as the number of ownaccount workers and employers per 100 economically active persons, and, in accordance with other studies (e.g., Fotopoulos & Storey, 2017;, own-account workers and employers in agriculture are excluded from the numerator. The variable is calculated for three years, 1991, 2001 and 2011. The Czech Statistical Office is the source of information.
Besides the self-employment rate, the employer rate in 2001 and 2011, that is, the number of employers per 100 economically active persons, is used as a dependent 714 Jirí Novosák et al.
variable in the robustness check. This is motivated by the findings of Hanley (2000) that own-account workers and employers in Czechia substantially differ in their income and property, and that nomenklatura members were more likely to become employers, particularly in the early phase of post-socialist transformations (Benáček, 2006).

Explanatory variables
Explanatory variables relate to pre-socialist entrepreneurship. The first variable is the self-employment rate in 1930. It is just this variable that has been used for the evaluation of the persistence of regional entrepreneurship in several studies (Fotopoulos & Storey, 2017;Fritsch et al., , 2019c. Fritsch et al. ( , 2019b nevertheless point out that the pre-socialist self-employment rates in manufacturing influence post-socialist self-employment rates the most in the former East Germany and western Poland. Therefore, three other explanatory variables are considered in this study: (1) the self-employment rate in manufacturing in 1930; (2) the self-employment rate in services in 1930; and (3) the self-employment rate in agriculture in 1930. Two other explanatory variables are used to check the robustness of our results. The first is the average business size in 1930, which is considered as a proxy for pre-socialist entrepreneurial culture. The second variable is the number of businesses per 1000 inhabitants aged 15-64 years in 1930. The number of businesses with at least one employee per 1000 inhabitants aged 15-64 years in 1930 is used as an alternative for the models with the employer rate as the dependent variable.
Additional information on all explanatory and control variables, including the sources of information and summary statistics, are provided in Appendix B in the supplemental data online.

Control variables
The choice of control variables is based on a general theoretical framework that integrates supply-and demand-side factors, and the institutional environment and industrial structure, to explain regional differences in entrepreneurship (Delfmann et al., 2014). The supply-side factors, which include, among others, human capital and foreigner status, are related to people's resources, traits and attitudes to entrepreneurship. The demand-side factors are related to entrepreneurial opportunities, which are created by agglomeration economies and innovations. Socio-cultural factors are important to the institutional environment of entrepreneurship (Wyrwich, 2012).
The following control variables, which refer to 1930 to avoid the 'bad control problem', are included in this study. The first is the population density as a 'catch-all' variable (Fritsch et al., 2019b) related to several supply-and demand-side factors of entrepreneurship. Following the approach used by Fotopoulos and Storey (2017), and to account for the influence of industry structure and industrial restructuring, the expected self-employment rate is included in the analyses.
Two other variables, namely the share of Germans and that of other national minorities in the total population, relate to the influence of national minorities. The inclusion of the former variable is motivated by the fact that Germans were the most numerous minorities in Czechia before the Second World War (about 30% of the total population); however, most of them were expelled shortly after 1945 (Gerlach, 2010).
The remaining two variablesthe number of patents per 100,000 people aged 15-64 years and the number of motor vehicles per 1000 inhabitantsare intended to measure the technological level of districts in 1930. They are compiled into an innovation index. While the former variable is a common measure of innovation, the latter variable measures the importance of the automobiles in the fourth technological revolution, dated between 1908 and 1971 (Perez, 2010).
Additionally, another variable is created to control the resettlement of the Czech Sudetenland after the expulsion of Germans in the late 1940s. It is created on a socio-cultural profile of the population of the region in the early 1990s with the intent to augment the models with the 1930 control variables when the impact of the expulsion of Sudeten Germans from the Czech borderlands on self-employment rates in 2001 and 2011 is discussed.
To verify the robustness of the results, the control variables that refer to 1991, 2001 and 2011 are included in this study as alternatives to the 1930 variables. The more recent variables are intended to better control the socialist legacies and post-socialist transformations. The list of these variables is given in Appendix B in the supplemental data online.

Methods
Spatial autocorrelation is a relevant issue in our study because the spatial units used in the analyses are rather small (see also Kelly, 2019, for the importance of using spatial data in persistence studies). Therefore, spatial regression models are utilized here to account for spatial dependence in the data. The econometric strategy of this paper builds on the recent research of Fotopoulos and Storey (2017) that speaks about the local nature of entrepreneurship. Assuming this claim to be correct, LeSage (2014) recommends estimating the spatial Durbin error model (SDEM) that has been incorporated in this paper.
Five groups of SDEM models are specified. The first group comprise the SDEM models for the self-employment rate in 1991, 2001 and 2011 (dependent variables). In these models, all explanatory and control variables refer to 1930 to avoid the 'bad control problem'. The long-term persistence of self-employment is then discussed with reference to the relationship between the self-employment rate in 1930 and the dependent variables.
Three groups of SDEM models verify the robustness of the results from the previous SDEM models. In the second group of SDEM models, the self-employment rate in 1930 is consecutively replaced by the two other explanatory variables and the SDEM models are re-Long-term persistence in entrepreneurship: the case of post-socialist Czechia  estimated. The third group of SDEM models replicates some of the previous models using the employer rate as the dependent variable. The fourth group of SDEM models replaces the 1930 control variables with their 1991, 2001 and 2011 alternatives, thereby controlling the influence of socialist legacies and post-socialist transformations.
The fifth group of SDEM models focuses on the presocialist variables, with the self-employment rate in 1930 as the dependent variable. The intent of these models is to investigate the significance of pre-socialist variables in this group of SDEM models. Additionally, multinomial logit regression is used to evaluate the career and educational paths of individuals with a pre-socialist entrepreneurial family background. 1

EMPIRICAL ANALYSIS
We begin our empirical analysis with a cross-district investigation of data on self-employment rates in 1930 and 2011. These data are presented in two maps (see Appendix C in the supplemental data online) allowing a comparison of spatial patterns. Possible factors in both the persistence of and change in self-employment rates can be found from this comparison. The former include especially the persistently high self-employment rates in the largest city districts, and the persistently low self-employment rates in the heavy industry districts in north-eastern Czechia; while the changes in self-employment rates relate, in part, to the influence of the expulsion of Sudeten Germans from the Czech borderlands after the Second World War, socialist heavy industrialization in north-western and northern Czechia, and also the metropolitan development of Prague and other large cities.
The joint influence of persistence and change is reflected in the correlation coefficients between selfemployment rates in 1930 and self-employment rates in 1991, 2001 and 2011 (see Appendix D in the supplemental data online). These coefficients are positive and significant; however, their magnitude decreases with increasing time delay. This is true for self-employment rates in both manufacturing and services. However, further examination is needed to understand the specific correlation coefficients for self-employment rates in agriculture. The presence of spatial dependence in self-employment rates is confirmed by the Moran's I test for spatial correlation for all the years under consideration (see Appendix D in the supplemental data online).
The relevance of the initial findings was further investigated by using SDEM models with the self-employment rates in 1991, 2001 and 2011 as dependent variables (Table 1). Model I provides the SDEM model estimates for 1991, that is, soon after the fall of socialism in Czechia. Therefore, low self-employment rates in 1991 can likely be a manifestation of emerging entrepreneurship at the dawn of the post-socialist transition. The idea is to explore the connection between self-employment rates in 1930 and self-employment rates in 1991, while simultaneously controlling the effects of other variables in the models.
Suggestive evidence has been provided on the persistence of inter-district differences in self-employment rates between 1930 and 1991, because self-employment rates in 1930 associate positively and significantly with selfemployment rates in 1991.
The next question is whether this result also holds true for self-employment rates in 2001 (see model II) and 2011 (see model III). The proportion of self-employed people in the districts has grown rapidly since the early 1990s, and the effects of post-socialist transformations intensified during this period. Despite these changes, however, the coefficients of the self-employment rate in 1930 remain positive and significant in the SDEM models with the self-employment rates in both 2001 and 2011 as dependent variables. Moreover, it is remarkable that these associations can be observed for the 1930 self-employment rate in manufacturing and services but not for the 1930 self-employment rate in agriculture. Therefore, an early sectoral restructuring away from agriculture to non-agricultural pursuits (see Appendix E in the supplemental data online) has laid the foundations for high long-term self-employment rates.
Three control variables related to industry structure in 1930, population density in 1930, and the share of Germans in the total population in 1930 are significant at the 0.01 significance level in the 2001 and 2011 SDEM models. Moreover, the effect of industry structure, which is one of the slowly changing regional determinants of entrepreneurship, is reinforced by its interaction with innovations (see Table F2 in Appendix F in the supplemental data online). This is in line with the arguments of Fritsch et al. (2022) that the regional persistence of entrepreneurship can be explained by the historical success of regions, particularly in knowledgeintensive industries. The negative sign of the population density coefficient relates to the influence of socialist heavy industrialization in urban areas and the suburbanization processes since the 1990s. These results confirm the importance of socialist legacies and post-socialist transformations. Finally, the expulsion of Sudeten Germans from the Czech borderlands after the Second World War is negatively associated with self-employment rates in 2001 and 2011.

DISCUSSION
The differences in self-employment rates across the districts of Czechia appear to be persistent over time . The results also support the claim that this persistence is related to the past self-employment rates in manufacturing and services, and not in agriculture. The reasons for such persistence are to be explored further.
The existing literature explains time persistence of regional entrepreneurship rates through the following arguments: (1) the slowly changing regional determinants of entrepreneurship; and (2) path dependence in regional entrepreneurship (Fotopoulos & Storey, 2017;Koster & Hans, 2017). The influence of the slowly changing regional determinants of entrepreneurship is controlled in the study by simply including control variables in the estimated models. Despite this, pre-socialist self-employment rates remain a significant predictor of both self-employment rates and employer rates after 1989 in the models. The question of the transmission mechanisms behind the time persistence of regional entrepreneurship rates emerges again.
Entrepreneurial values and attitudes passed from previous generations, and 'entrepreneurial memory' from the pre-socialist past are the two transmission mechanisms suggested by Fritsch et al. ( , 2019aFritsch et al. ( , 2019c. These mechanisms also seem to be relevant for Czechia, as indicated by the significance of the explanatory variables related to pre-socialist entrepreneurship. Our intent here is to extend the understanding of the transmission mechanisms by looking at the career and educational paths of individuals with a pre-socialist entrepreneurial family background and by discussing the impacts of the expulsion of Germans from the Czech Sudetenland after the Second World War.  Long-term persistence in entrepreneurship: the case of post-socialist Czechia (1930-2011) 717

Career and educational paths of individuals
Concerning the career and educational paths of individuals, we build on the research into social stratification in socialist Czechia, and on Benáček's distinction between socialist social groups according to their propensity to start new businesses (Benáček, 2006). One of the wellknown socialist doctrines, which is based on discrimination against individuals with an entrepreneurial family background, expects a downward shift of the social status of these people towards manual jobs, and worsened access to education for their children (Hanley, 2003;Hanley & Treiman, 2004). If strictly applied, this could reduce the likelihood that individuals with a presocialist entrepreneurial family background will start a new business after socialism, simply due to their career path being oriented towards Benáček's group of 'passive entrepreneurial outsiders'. This is the core idea that we pursue here, using data from a representative sample survey conducted in 1993 and 1994 as part of the project 'Social Stratification in Eastern Europe after 1989' (Szelenyi & Treiman, 2017) (see Appendix I in the supplemental data online for additional information on the data used in this study). A multinomial logit model is estimated to test the idea under investigation because our dependent variable is nominal and has three possible outcomes derived from the 10-category version of Erickson, Goldthorpe & Portocararero's occupational class categories (Erikson et al., 1979). These outcomes are: (1) managers, proprietors, professionals and high-grade technicians; (2) skilled manual and non-manual workers; and (3) semi-skilled and unskilled manual and non-manual workers. The main question is whether individuals with a pre-socialist entrepreneurial family background, that is, whose parents or grandfather(s) owned a non-agricultural business in 1948, are more likely to be categorized as belonging to the third category than to the other two categories in 1988, also including the last previous occupation for those not employed in 1988 and controlling the influence of other variables. It is worth noting that it is just the category of semi-skilled and unskilled manual and non-manual workers that best corresponds to Benáček's group of 'passive entrepreneurial outsiders'. Table 2 provides the results of the multinomial logit model with semi-skilled and unskilled manual and nonmanual workers as the base category. It shows that people with a pre-socialist entrepreneurial family background are significantly more likely to be categorized as belonging to the former two social classes than to the social class of semi-skilled and unskilled manual and non-manual workers. Moreover, individuals with a pre-socialist entrepreneurial family background are more likely to attain significantly higher levels of education than those without this background (see Appendix J in the supplemental data online).
It is obvious from the results that the core idea under investigation does not hold, and that, under socialism, individuals with a pre-socialist entrepreneurial family background were significantly more likely to achieve high occupational and educational status, including managerial positions. Consequently, the acquired relational and human capital, and the 'entrepreneurial-like skills' necessary for the management of state-owned enterprises and cooperatives, has facilitated the recovery of pre-socialist self-employment since the fall of socialism in the late 1980s. This is in accord with the claim of Hanley and Treiman (2004, p. 248) that: continuity in ownership of business … was enhanced by the propensity of the descendants of pre-communist business owners to acquire high status positions during the communist era, and for those in high status positions to be more likely to hold business … in the post-communist period.
Finally, it is important to note two other spatially embedded factors that contributed to the recovery of pre-socialist self-employment rates across Czech districts in the 1990s. First, the data from the representative sample survey indicate that at least 70% of respondents with a presocialist entrepreneurial family background had the same LAU-1 region of residence at age 14 as in 1993. Second, as Hanley and Treiman (2004) note, pre-socialist property rights were restored in the early 1990s through restitution, which also provided the funds necessary to start a business (Hanley, 2000).

Expulsion of Germans from the Czech Sudetenland
Roughly 3 million Germans lived in the Czech Sudetenland before the Second World War; however, most of them were expelled from Czechia after the war (Gerlach, 2010). The impact of this expulsion was very uneven across districts: some were hardly affected while others lost almost its entire population. Moreover, as Table K1 in Appendix K in the supplemental data online indicates, the share of Germans in the total population was positively associated with the self-employment rate in 1930, and the districts with a high share of Germans were characterized by their stronger focus on manufacturing and innovations. Considering this background, the findings presented by Fritsch et al. (2022) are remarkable. They claim that 'the positive effect of high historical levels of knowledgeintensive entrepreneurship on current new business formation is independent of changes in the composition of the local population' and explain this finding 'by the presence of a place-based collective memory about historically successful (knowledge intensive) entrepreneurship that does not require a persistent population ' (p. 398). Keeping this in mind, attention is now turned to the Czech case.
Our attention is primarily focused on the variable related to the share of Germans in the total population of Czech districts in 1930. As Table 1 indicates, the share of Germans in the total population is negatively and significantly associated with self-employment rates in 2001 and 2011. Hence, despite their favourable conditions for entrepreneurship before the Second World War, the exchange of population due to the expulsion of Germans from the Czech Sudetenland negatively influences current regional entrepreneurship. It seems that the proposed mechanism of collective memory about historically successful entrepreneurship is relatively weak in the Czech case. But what are the reasons behind it?
The newcomers' capacity to absorb the benefits of historically successful entrepreneurship might provide an answer to the question. This capacity may be influenced by a variety of factors, such as knowledge, skills, or entrepreneurial mindset and values. However, the migration of low-status groups from the Czech interior and underdeveloped Slovakia prevailed in the Czech Sudetenland after the Second World War. Consequently, the newcomers' capacity to absorb the benefits of 'entrepreneurial memory', and their capacity to pass entrepreneurial values, attitudes and experience from one generation to another, are low.
The proposed mechanism was verified by augmenting our basic models in Table 1 with an additional variable capturing the resettlement process in the Czech Sudetenland after the Second World War. The variable is based on a profile of the population of the Czech Sudetenland in the early 1990s characterized by a low share of people living in their birthplace, a high share of people with no religious affiliation, low electoral turnout, and a high share of Slovaks and Roma (Daněk, 1995). Table 1 confirms that the newly constructed variable related to the population exchange after the Second World War in the Czech Sudetenland is negatively and significantly associated with self-employment rates in 2001 and 2011. Additionally, the variable related to the share of Germans in the population is not significant after controlling for the resettlement process in the Czech Sudetenland after the Second World War. Altogether, the composition and stability of local population matters in explaining the long-term persistence in regional entrepreneurship across Czech districts. Fritsch et al. ( , 2019bFritsch et al. ( , 2019c point out that interregional differences in self-employment rates can persist over long periods, even despite 'historical shocks'. This study provides further evidence for this by showing that for the 206 districts of Czechia, pre-socialist self-employment rates in 1930 positively and significantly associate with post-socialist self-employment rates and employer rates in 2001 and 2011. This is while controlling for the effects of slowly changing regional determinants of entrepreneurship and despite the influence of socialist legacies and post-socialist transformations. Moreover, the high pre-socialist self-employment rates of districts in 1930 are associated with early sectoral restructuring away from agriculture to non-agricultural pursuits and technological advancement (e.g., see Fritsch et al., 2019bFritsch et al., , 2022and Fritsch & Wyrwich, 2018, for the importance of industrialization and historically rooted knowledge).

CONCLUSIONS
This study deals with a country where the natural development of the entrepreneurial environment was interrupted by both the Second World War and more than 40 years of socialism. Therefore, the transmission channels based on historical entrepreneurial experience are emphasized. These channels particularly involve spatially embedded entrepreneurial values and attitudes being passed from one generation to another, and 'entrepreneurial memory' from the pre-socialist past. This study contributes to the existing knowledge about these transmission channels in two ways.
Concerning the former transmission channel, this study suggests another, more tangible, mechanism by building on social stratification research and looking at the career and educational paths of individuals with a pre-socialist entrepreneurial family background. The results show that under socialism individuals with a presocialist entrepreneurial family background were significantly more likely to achieve high occupational and educational status in the late 1980s. Subsequently, the acquired relational and human capital, as well as the 'entrepreneurial-like skills' necessary for the management of state-owned enterprises and cooperatives, increased the likelihood that these individuals would own businesses in the post-socialist period (Benáček, 2006;Hanley & Treiman, 2004). The link between restitution in the 1990s in Czechia and this mechanism is also noteworthy (Hanley & Treiman, 2004).
Concerning the latter transmission channel, this study points to a negative relationship between the exchange of the local population after the Second World War and selfemployment rates in 2001 and 2011. Unlike the cases of Poland (Fritsch et al., 2022) and the Kaliningrad region (Fritsch et al., 2019c), the channel of 'entrepreneurial memory' or 'collective memory about historically successful entrepreneurship' is relatively weak in the Czech case. This study explains the difference by the prevailing migration of low status groups from the Czech interior and from underdeveloped Slovakia into the Czech Sudetenland after the Second World War. The capacity of new settlers to absorb 'entrepreneurial memory' is then limited. Hence, a more detailed scrutiny of new settlement in the previously German part of contemporary Poland and in the Czech Sudetenland after the Second World War may be beneficial to better understand the functioning of 'entrepreneurial memory'.
The results of this study provide further support for policy arguments that have been repeatedly mentioned in the literature. Hence, it is difficult, or almost impossible, to change interregional differences in self-employment rates significantly through short-term policy measures (see also Fotopoulos & Storey, 2017;Wyrwich, 2012). This fact must always be considered when evaluating the goal of mitigating regional inequalities through fostering entrepreneurship. Moreover, the relationship between self-employment and regional development is not straightforward. As this study indicates, there are technologically dynamic districts with strong entrepreneurial cultures and a disrupted entrepreneurial culture is common in other regions. A detailed scrutiny of these regions may reveal more fruitful policy implications (see also . The results of this study suggest that this scrutiny should particularly focus on the role and enhancement of human capital, technological capabilities and industry restructuring in relation to entrepreneurship (see also Fotopoulos & Storey, 2017;Fritsch et al., 2022;Fritsch & Wyrwich, 2018).
The study has limitations, of course. First, it is geographical in nature, and the representative sample survey which we refer to in the text was primarily focused on social stratification in the early 1990s. Therefore, fruitful additional insights can be gained from qualitative or quantitative research designed to understand entrepreneurs' perceptions and perspectives. Second, a different specification of variables (e.g., a highly disaggregated industry structure) can be used to gain a more in-depth understanding of the results. Third, the robustness of the results can be checked for data referring to more aggregated spatial units. These areas, and a comparison of the results with those of other post-socialist countries with a different historical development (e.g., Slovakia), are also directions for future research.