Cognitive legitimacy and ownership heterogeneity: impact of local incumbent firms on new private firm formation in China’s manufacturing sector

ABSTRACT This study investigates the impact of local incumbent firms on new private firm formation in China by focusing on the cognitive legitimacy lent by different ownership types. We argue that new firm formation requires legitimacy spillover from local society. Based on the panel data of China’s manufacturing enterprises, we find that local ownership structure plays a vital role in new private firm formation. Incumbent privately owned enterprises enhance the legitimacy of private ownership naturally, and the effect from foreign-owned enterprises to new private firms is positive but weaker than that of privately owned enterprises, whereas state-owned enterprises discourage entrepreneurship due to their entirely different identities.


INTRODUCTION
New firm formation is associated with regional employment, urban growth and economic development (Glaeser et al., 2015). Large numbers of studies in economic geography have verified that region-specific determinants, such as capital, knowledge or local specialization, have significant impacts on entrepreneurship (Armington & Acs, 2002). Recent studies have rethought the focus on economic condition and instead emphasized that entrepreneurship is also embedded in local social context and is largely shaped by local culture and its historical roots (Acs et al., 2017;. Further, a growing number of studies have started adopting the narrative of legitimation to scrutinize the interaction between potential entrepreneurs and local entrepreneurial environment (Frenken et al., 2015;Stuart & Sorenson, 2003).
In this study we regard new firm formation as a process of legitimation of an organizational form. Several recent works have focused on the pivotal role of legitimation process in new firm formation, and tested the way how regulation construction, social capital provision or value conformity help construct legitimacy (de Vaan et al., 2019). According to cognitive psychology, people tend to judge new things by deconstructing and categorizing them based on their cognitive structures such as schemas and stereotypes (Markus & Zajonc, 1985). Kostova and Zaheer (1999) further use the term 'legitimacy spillover' to describe the phenomenon of cognitive categorizing and argue that 'the legitimacy of a given organizational unit in a particular environment is not independent of the legitimacy of other organizational entities with which the unit is cognitively related' (p. 99). Therefore, we pay special attention to cognitive legitimation based on the knowledge about the new firms and the identity overlap between the evaluators/observers and new firms. According to Ahlstrom et al. (2008), individual firms tend to select a more appropriate environment or form an alliance with incumbents to search for legitimacy. However, from the perspective of observers, the cognitive legitimacy associated with an organizational form is more readily enhanced if those observers think they have a higher degree of identity overlap with that organizational form (Li et al., 2007).
To test the effect of cognitive legitimacy spillover from local incumbents with different cognitive categories and identity characteristics on new firm formation, we take the new private firm formation in China as a case. Evidence from China has suggested that the pattern of new firm formation is closely related to institutional change and state involvement (He et al., 2018). China has been undergoing the transition from a central planning system that prohibits private ownership to a market-oriented economy along with a gradually changing formal institution and social value. Private economy is a radical new form characterized by a slow and unsteady growth and strongly affected by the old forms in the early stage of transition (Dobrev et al., 2006;Xu et al., 2014). Different from a market economy in advanced countries that is shaped by free market and privatization, a mixed economy in transitional China is characterized by its diversity and hybrid form of ownership types (Nee, 1992). An ownership type is a form of organization in a specific population that adopts similar strategies, strives for specific resources and shares the same identity (Peng et al., 2004). And incumbent firms with different types of ownership create different but mutually related social norms and regulations in the mixed economy (Xu et al., 2014). We argue that an ownership type represents a distinct collective identity that differentiated one type from the other.
This empirical work focuses on the entry pattern of private enterprises and investigates the heterogeneous impacts of cognitive legitimacy spillover from incumbents of three ownership types: privately owned enterprises (POEs), state-owned enterprises (SOEs) and foreignowned enterprises (FOEs). We identify different levels of legitimacy spillover lent by different ownership types to new private firms based on their cognitive characteristics and identity recognition.
The remainder of the paper is structured as follows. The second section reviews the theory of cognitive legitimacy and develops our hypotheses about the impact from heterogeneous local incumbent firms with various ownership types on new private firm formation in China. The third section clarifies the data sources and methodological issues. The fourth section models and explains the impact of local incumbents with different ownership types on new private firm formation in China. The fifth section checks the robustness of our models. The sixth section summarizes the main findings and discusses their implications.

LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT
2.1. Legitimacy, heterogeneity of local incumbent firms and new firm formation Literature using evidence from all over the world has widely discussed the regional variation in new firm formation rates and confirmed that the pattern of new firm formation is largely shaped by regional characteristics (e.g., Armington & Acs, 2002). Most of the existing studies, however, have focused on specialization, knowledge spillover or competition to unpack the relation between entrepreneurship and economic input, without paying much attention to the role of legitimacy from local society (Acs et al., 2009;Cainelli & Iacobucci, 2012). Legitimacy is an assumption that the actions of an organizational form 'are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions' (Suchman, 1995, p. 574). Zimmerman and Zeitz (2002) argue that legitimacy is a resource for new ventures that is at least as important as other more well-perceived resources, such as capital, labour, technology, customers, networks, etc. From a perspective of evaluators, legitimacy can also be regarded as a judgment on an organization, rendered by individuals, collective actors and external environment at both micro-and macro-levels (Bitektine & Haack, 2015). Compared with the economic-concentric consideration of new firm formation, the theoretical framework of legitimacy pays more attention to formal permission and informal willingness of local actors to contact with each other (Boschma, 2005;Frenken et al., 2015).
As a newly established organization, a new venture needs to gain social acceptance for survival, growth and development because of the liability of newness. First, new ventures experience different types of problems related to their newly established status, and the risk of failure would be higher if they could not gain social acceptance which is related to approval, trust and support from others. Without legitimacy, new private firms would not be well accepted by local society and thus hard to gain investment, recruit employment and win the trust of their customers. The liability of newness would frighten potential entrepreneurs to start their business, which would further lead to an organizational ecology that is harmful to new firm formation . Second, legitimacy is not only a significant resource highlighted by prior research, but also enables organizations to access other resources needed to survive and grow, such as talents, capital and partners. As Li et al. (2007, p. 176) argued, 'legitimacy signals taken-for-grantedness in an institutional environment and is necessary for organizations to obtain normal access to resources, protection from authorities, higher visibility, and enhanced life chances'. Third, legitimacy is viewed as a way to overcome crises, negative press and uncertain environment that may impact the organization (Pollack et al., 2012). However, new ventures tend to be harder to obtain legitimacy because a society judges an organization as appropriate largely based upon its past performance. As it stands, while established organizations can easily persuade external audiences or evaluators with their excellent performance in the past, a new venture has little or no record of past performance on which to base its claim for legitimacy (Zimmerman & Zeitz, 2002). Therefore, the demand of new ventures for legitimacy is more imperative than incumbent firms.
Several studies have identified certain sources of legitimacy for new firms to pursue which are multidimensional and related to multiple actors, ranging from individuals to business organizations, governments, trade associations, networks, etc. (Zimmerman & Zeitz, 2002). There are three elements of the external environment from which legitimacy can be achieved: regulative, normative and cognitive legitimacy (Cruz-Suárez et al., 2014). Whereas regulative legitimacy stresses the regulation and rules made by state agencies, normative legitimacy can be obtained by obeying informally social norms, and cognitive legitimacy that is more relevant to identity and practices that are taken for granted (Ruef & Scott, 1998). In this study, we focus on cognitive legitimacy because it is identified as the most durable and the most efficient among the different types of legitimacy, and it has been found to be especially important for new ventures (Pollack et al., 2012).
The cognitive legitimacy of a new organization is based on the observers' knowledge about this organization. It is determined by those observers who assess its conformity to specific standard or models that are varied across different groups in a locality. In general, cognitive legitimacy can be more easily derived from adjacent actors who knows the organization better and understand it well because of geographical proximity. But established organizations with different identities (i.e. also related to value and ideology) may lend different degrees of cognitive legitimacy to a new venture. In a transitional economy such as China, a hybrid of ownership structure gives rises of a coexistence of SOEs, POEs and FOEs, with each showing a distinct identity, ideology, strategy and behavioural pattern. This may arise a divergence in terms of cognitive legitimacy spilled over to new firms. The cognitive legitimacy from a group of organizations can be further aggregated to a regional level, exerting influences on regional entrepreneurial culture and environment, which in turn impacts new firm formation in the region.
In particular, we argue that cognitive legitimacy spillover usually occurs in three steps when new firms and local incumbent firms have similar cognitive characteristics: deconstruction, cognitive processing and transference. First, people generally deconstruct local legitimate incumbents into a collection of different cognitive characteristics, for instance, SOEs or POEs according to the ownership type, or large-sized or smaller enterprises based on their scale, etc. Second, new firms can be categorized into existing groups according to the similarity between new firms and certain groups. Third, a society would transmit the stereotypes as well as an acceptance of legitimate incumbents to new firms in the same cohort. Furthermore, the degree of cognitive legitimacy spillover differs based on the degree of cognitive similarity and group identification between new firms and incumbents (Kuilman & Li, 2009). On the one hand, the social judgment to new firms is normally based on the cognitively related incumbents, and the recognition of legitimate incumbents may be easily transferred to new firms (Kostova & Zaheer, 1999). On the other hand, it has been argued that new firms tend to actively seek legitimacy not only from incumbents with the same identity but also from cognitively related incumbent firms in the surrounding environment (Li et al., 2007). To Li et al. (2007), new wholly owned foreign subsidiaries (WOFS) learn how to contact with local actors, capture resources and make themselves accepted by the host country from not only incumbent WOFS but also joint ventures which are cognitively related with WOFS.
Therefore, the present study takes China as a case to investigate the relationship between cognitive legitimacy and new firm formation rates at prefecture-level cities.
In China, state ownership has been dominant for a long time, and private ownership as a young economic form which was first legalized in 1988 was still not fully approved by public around the millennium all over the country (Han & Pannell, 1999). The emerging private ownership as a new organization form thus suffered from the lack of legitimacy in local contexts. Potential entrepreneurs tend to gain legitimacy spillover from the incumbents who may fall into different ownership types but have similar value or culture characteristics to the new firm (Kostova & Zaheer, 1999), such as similar value, business strategies or organizational structure as well as similar habits and customs (Tan, 2002;Xu et al., 2014).

Legitimation process of new privately owned firms in China
To better understand the emergence of private ownership in China and the cognitive characteristics of different ownership types, we start from the development of the mixed economy and scrutinize the changes of the ownership structure in three major types of incumbents: POEs, SOEs and FOEs. The classification codes are shown in Appendix A1 in the supplemental data online. China used to be a central planning system after the socialist 'Three Great Transformation' in the mid-1950s, which in principle eliminated private assets through the redemption and establishment of cooperative teams. After the reform and opening-up in 1978, China started to discard 'Soviet-style mandatory planning' and adopted 'guidance planning', and then claimed 'to develop the socialist market economy' in 1992, which is characterized by the triple processes of decentralization, marketization and globalization (Wei, 2001, p. 13;He et al., 2008). Foreign investment was endorsed by Chinese law in 1979, 1 and founding new firms as well as hiring employees by an individual was illegal until the introduction of the Amendment of the Constitution in 1988. 2 After the legalization of private ownership, the proportion of output value created by SOEs underwent a dramatic decline in the 1990s (Wei, 2001). Figure 1 shows the number of new manufacturing enterprises and the proportion of incumbent firms with different ownership types using the data from the China Annual Survey of Industrial Firms (ASIFs), which covers 285 prefecture-level cities from 1999 to 2007. While new POEs experienced rapid growth in terms of both absolute number and relative proportion, FOEs grew in a slow and steady way, and SOEs experienced a sharp decline. The results point to the differential interaction between new private firm formation and different ownership types of incumbents, and they show the process of rapid legitimation on private ownership from scratch in the period 1999-2007. Both entrepreneurial intention and behaviour significantly depend on regional social legitimacy of entrepreneurship (Kibler et al., 2014), which makes the role of legitimation as vital as that of economic factors.
Based on legitimacy theory, we argue that cognitive legitimacy from incumbents to new private firms is various across ownership types, because an ownership type can be 346 Fan Shi and Cassandra C. Wang regarded as 'a type of socially coded identity', and a community with a higher degree of identity overlap exerts a stronger legitimating influence than other communities (Polos et al., 2002). Therefore, firms within the same ownership share similar identity, ideology, value and strategies, and hence lend the highest degree of cognitive legitimacy to each other. Potential entrepreneurs and other audiences tend to develop interests in and evaluate the new POEs by looking at the status or situation of local incumbent POEs. The successful POEs confirm that private ownership is not 'evil' but with a high return and is worthy of trial. And the role models attract audiences to start their own firms or work in a POE and thus increase the possibility of persistence of a high new firm formation rate (Andersson & Koster, 2011). On the contrary, a recent work using Global Entrepreneurship Monitor (GEM) data found that failing role models would lead to a fear of business failure and depress future new firm formation . In the local context, the density of POEs points to the prevalence and success of private ownership, which lend ample legitimacy to potential entrepreneurs. Therefore, Hypothesis 1 can be given by: Hypothesis 1: The share of POEs in a region gives a high level of legitimacy to new private firms and hence has a positive effect on the new private firm formation.
The identity-dependent legitimation process also shows that SOEs with their planned-economy identity and POEs with their free-market identity are ideological rivals and opposing forces, and it therefore is hard to imagine POEs can obtain legitimacy from SOEs (Xu et al., 2014). Recent studies have shown that state ownership has a negative effect on entrepreneurship . State ownership is the old and once dominant form in the Chinese economic system. Although the proportion of SOEs has been undergoing a rapidly decline in recent years, SOEs are still dominant in some cities and some industries in China and they adhere to different value from new private firms. It is suggested that incumbents may hamper new entrants when established norms and values are challenged or vested interests are threatened (de Vaan et al., 2019). In countries undergoing ideological transformation, the inhibition may be stronger when confronting with ideological struggle (Ingram & Simons, 2000). SOEs are usually associated with easier access to state subsidy and the welfare system of social infrastructure such as housing allocation, factory-run schools and endowment insurance in the early 1990s. Skilled laborious with specialized knowledge and experience are so content with the welfare system provided by SOEs that they are not interested in starting their own firm or working in new private firms, even if productivity in the private sectors is higher (Sachs & Woo, 1994).
Besides, the abundance of SOEs in a region implicates the state-affinity social value that is ingrained in the memory of people. In the 'good old days', it was honourable and risk-free to become an employee of SOEs and rest on their laurels because of equalitarian beliefs such as 'one for all and all for one'. The value cultivated by state ownership is not quite the same as the entrepreneurship that prefers taking great risk to exploit opportunities and to pursue profit maximization (Caliendo et al., 2014). Moreover, SOEs are characterized by 'their sources of funding and their mode of social control of the organization' and 'a narrow market domain, a stable customer group, and an established structure managed by older executives', which is different from the organizational structure and business strategies of POEs which focus on capital expansion in the market economy (Uhlenbruck & Castro, 1998, p. 619;Peng et al., 2004Peng et al., , p. 1109. Despite several rounds of reform on SOEs, the situation of resource domination Cognitive legitimacy and ownership heterogeneity by SOEs still might lead to a monopoly or bureaucracy in the early 2000s (Hu, 2017). Therefore, the social norms and values created by SOEs tend to discourage the radical new private firm formation, and the legitimation of new private firms in a region dominated by SOEs is miserably little. Hypotheses 2 is given as follows: Hypothesis 2: The share of SOEs in a region gives little or no legitimacy to new private firms and hence has a negative effect on the new private firm formation.
Compared with SOEs, FOEs are younger organizational forms and viewed as outsiders that have been legalized in China since 1979. After decades of cultivation for legitimacy, FOEs have generally received public recognition through media publicity and joint venture (Li et al., 2007). The legitimacy lent from FOEs is more complicated than other ownership types. On the one hand, FOEs have a much higher degree of identity overlap with POEs because they are both directed by rules of the free-market economy. Both FOEs and POEs have to search for local resources by themselves, which is different from SOEs which comparably are more dependent on resources provision by government. For potential entrepreneurs, managerial knowledge and supply chain from FOEs are more valuable than that from SOEs. Domestic partners are also needed for FOEs to get access to domestic social networks (Liao, 2015). And for foreign managers, it is easier to reach an agreement on mutual benefit with private entrepreneurs who hold the same value with them than overcautious executives of SOEs (Tan, 2002). Therefore, FOEs and POEs are valuable for each other, which makes the legitimacy spillover possible. Moreover, the more intensive are FOEs in a local context, the more taken-for-granted the marketization and globalization would be. It makes it easier for potential entrepreneurs to gain venture capital, recruit employment and obtain market permission in such an environment (Dobrev et al., 2006;Xu et al., 2014). Finally, the growth of FOEs also breaks the obsession for Communism and introduces speculation as well as individualism to China. We thus have Hypothesis 3: Hypothesis 3: The share of FOEs in a region gives legitimacy to new private firms and hence has a positive effect on the new private firm formation.
On the other hand, the legitimacy from FOEs will be mitigated because of cultural differences between FOEs and POEs. First, in a new environment under which the culture and social norms are quite different from their parent country, FOEs have to be flexible and adopt the strategy of a so-called 'analyser' who sometimes takes the conservative approach like SOEs and sometimes switches to the pattern of broadening the market like POEs as needed (Peng et al., 2004). The difference in terms of technology, managerial outlook and mentality reduces identity overlap and hinders the legitimation process between FOEs and POEs. Second, the cultural difference leads to a weak social tie and less cognitive legitimacy between foreign executives and potential entrepreneurs. The obstructions in language and culture between FOEs and new private firms increase the cost and reduce the willingness of contact, which reduces the efficiency of transaction and inhibits legitimacy spillover (Ramos-Rodriguez et al., 2010). A recent research based on GEM data has showed that in countries with highly educated labour forces, foreign direct investment benefits employment in the wage sector rather than entrepreneurship (Berrill et al., 2020). Compared with domestic firms, FOEs not only bring the internationally advanced managerial and operational experiences, but also provide a higher salary and an attractive perk to their employees in China. The generous recruitment package offered by FOEs lures highly qualified talents to select FOEs over private firms, but also directly raises the opportunity costs to start new firms. Over time a region with a high proportion of FOEs will cultivate a culture that worships FOEs and produces a crowding-out effect on private firms. Those people who have the abilities to start a new firm would simply choose an easy life: being an employee in FOEs rather than taking the risk of running their own firms. Under this circumstance, the positive effect of FOEs from identity-based legitimacy is certainly weaker than that of POEs. We thus propose Hypothesis 4: Hypothesis 4: The positive legitimacy effect from FOEs on new private firm formation in a region is weaker than that of POEs.

Data
Two datasets compiled by the National Bureau of Statistics of China are collected to investigate the impact of cognitive legitimacy spillover from different ownership types on new private firm formation in China at prefecture-level cities. The first set of data comes from the China Annual Survey of Industrial Firms (ASIFs) during the period 1999-2007. In this duration, first, state ownership had lost its absolute dominance, and new private firms had started to emerge from scratch and grow rapidly ( Figure 1). Second, after support by the legal framework, the triple processes of decentralization, marketization and globalization significantly characterized the institution in China during this period (Wei, 2001), which lays the foundation of local free competition to some extent. Third, domestic macroeconomic operation and policy orientation between 1999 and 2007 are relatively stable and without an external shock until 2008. The period we choose is therefore appropriate for our studies because the regional start-up rate is sensible to unstable external environment. Fourth, it is generally agreed that the new era of socialism with Chinese characteristics started in 2012. After then, China has been gradually improving the institution for a market-oriented economy and the central government also claimed that the market plays a decisive role in resource allocation. This set of data includes the basic firm-level information of all enterprises in 2004, and SOEs and so-called 'above-scale' non-SOEs with their annual sales over 5 million yuan (about US$0.71 million) in other years in manufacturing industry, which is the most detailed and consistent micro-data released by the Chinese authority. According to the size classification of manufacturing enterprises in China, firms with revenue between 3 million yuan (about US$0.4 million) and 20 million yuan (about US$2.8 million) are small-sized enterprises. In ASIFs, the information for micro-sized enterprises and a small part of small-sized enterprises is omitted, which is fine because these enterprises have a low threshold to start and only account for a very small fraction of economic activity in the whole manufacturing industry (Brandt et al., 2012). Finally, following Brandt et al. (2012), we remove invalid observations and distinguish restructured or privatized firms from new firms.
The information about region-related characteristics is collected from China City Statistics Yearbooks for various years. The geographical units discussed in the present study include 269 prefecture-level cities. Due to changes of administrative division and missing values, the panel does not cover all those units at each year during the period 1999-2007. The panel is thus unbalanced but still effective for our study.

Research variables
New private firm formation (NPF) is generally interpreted as the new POEs' formation rate in a region, which is measured by the number of new POEs over all firms in manufacturing industry in a prefecture-level city.
Local incumbent firms with different ownership types in a prefecture-level city is measured by the employment share in manufacturing firms with a specific ownership type of total employment in all manufacturing firms in a prefecturelevel city. We control for agglomeration economies in terms of urbanization, industrial relatedness and diversification. Urbanization affects new firm formation rates through the provision of public resource. It is measured by logarithmic population density. Industrial relatedness and diversification are measured by related and unrelated variety following Frenken et al. (2007) (see Table 1 for measures).
We also control the following regional characteristics inspired by the previous literature (e.g., Fritsch & Storey, 2014). (1) Gross domestic product (GDP) per capita is a basic variable that assesses the level of regional development. The present study uses logarithmic GDP per capita to lower the order of magnitude. (2) Unemployment rate is another factor that may positively or negatively affect local new firm formation rates. It is measured by the ratio of unemployment to the sum of unemployment and employment. (3) The support of service industries is measured by the output value contributed by service industries over the total output value created by all sectors in a city. All variables and their statistical description are shown in Table 1. The correlation coefficients and variance inflation factors (VIFs) of all variables are shown in Appendix A2 in the supplemental data online. All correlation coefficients are less than 0.8, and the tests on VIF also suggest that variables do not have a serious multicollinearity problem. In the empirical section, we use logarithmic independent variables to eliminate the skewed distribution, and all control variables are lagged by one year to eliminate the possibility of reverse causality.  Figure 2. In general, the new private firm formation rate in eastern China is higher than that of western China. The pattern is similar to the spatial distribution of incumbent POEs, namely, the higher the presence of incumbent POEs, the higher the rate of new private firm formation in a locality. On the contrary, the employment shares of SOEs in eastern China are lower than those of the west, which implies a negative correlation between new private firm formation and incumbent SOEs in a region. As for incumbent FOEs, the cities with a high employment share are mainly along the coast, such as the Pearl River Delta, Yangtze River Delta and Bohai Rim. Figure 2 shows that the majority of cites with the lowest new private firm formation rate have the lowest employment share of FOEs. It is also interesting to note that most coastal cities with the highest employment share of FOEs have a rather low new private firm formation rate. We argue that FOEs would strengthen the acceptance of private ownership in cities where people prefer public ownership, but it may hamper new private firm formation in those cities that have already harboured lots of FOEs because of the fierce competition.

THE IMPACT OF INCUMBENT FIRMS WITH DIFFERENT OWNERSHIP TYPES ON NEW PRIVATE FIRM FORMATION
Fixed-effects regression models for panel data are employed to test the impact of legitimacy spillover from incumbents with different ownership types on new POE formation (NPF) using panel data. For city i and year t, the unobserved effects model is constructed by: where a i represents the individual heterogeneity of cites, which is fixed over time; and u it is the error term that changes with individuals and time. The empirical results of fixed-effects models for panel data are shown in Table 2. Models 2-4 test the impact of incumbent POEs, FOEs and SOEs on new POE formation individually, and model 5 represents the comprehensive impact of all three types of incumbents. Taken alone, the share of POEs and FOEs in a city produces significantly positive

RV
Related variety is the weighted sum of entropy of output at the three-digit level within each two-digit level. The formula of related variety is: For the three-digit sector i in a two-digit sector S g , P g is the share of each twodigit sector g, and is derived by summing the three-digit shares p i ; and H g is given by: effects on new POE formation, while the share of SOEs shows significantly negative effect. Specifically, 1% growth on the employment share of local incumbent POEs and FOEs would produce a 0.0122% and 0.0030% increase in the share of new POEs over local manufacturing firms, respectively, while a 1% increase in the employment share of incumbent SOEs would lead to a reduction of 0.0030% in the share of new POEs over local manufacturing firms. The results indicate that the impacts from incumbents with different ownership types are significantly different. Local incumbent POEs have the highest level of identity overlap with new POEs, which enhances the legitimation of new POEs and increases the new private firm formation rate. Hypothesis 1 is confirmed. Second, the impact of SOEs is demonstrated to be significantly negative on new firm formation, as shown in models 4 and 5. This finding is consistent with prior research that claims that SOEs, supported by local or central government and having an entirely different identity, would cripple entrepreneurship. They also develop collectivist value that is against the private ownership. Hypothesis 2 is confirmed.
Third, local incumbent FOEs have a different cultural background from that of new POEs, but have a certain level of identity overlap with POEs. The coefficients of the share of FOEs in models 3 and 5 indicate that FOEs exert a positive effect on new firm formation. Meanwhile, as shown in model 5, the coefficient of POEs is 0.0112, much higher than that of FOEs (0.0030). It demonstrates that the impact of POEs on new firm formation is stronger than that of FOEs. Hypotheses 3 and 4 are thus confirmed. As a supplement, we add a dummy variable to distinguish those cities who access foreign culture much earlier to further demonstrate the impacts of incumbents with different ownership types on new firm formation. In the first few years after 1978, China chose a few cities as pilots to adopt the open-up policies to attract foreign investment and embrace foreign culture. Four cities were selected to be special economic zones in 1980, and 14 coastal port cities were chosen to be coastal open cities in 1984. These 18 cities are the first group to be in contact with foreigners and have access to foreign culture, and thus would welcome a market economy and capitalism that are favourable for private ownership in our research period. We use a random effect model to estimate the dummy variable that does not change over time. The results are shown in Appendix A3 in the supplemental data online. While the impact of POEs is still positive and significant, the impact of FOEs turns out to be insignificant. This interesting finding implies that in regions where local actors already accept private ownership, the legitimation process has already been completed, and the effects of legitimacy spillover from FOEs and SOEs are significantly reduced.
In terms of control variables, industrial diversification is positive to new POE formation, which is consistent with the existing research of China . The unemployment rate is positive in all models, highlighting the role of labour availability to new firm formation. Finally, it is noted that industrial relatedness, urbanization, gross domestic product (GDP) per capita and the support of service industries are not significant to new POE formation. A possible reason is related to the process of industrial transfer and transformation in the early 2000s. Cities with a high level of industrial relatedness, urbanization and economic development or a high share of service might start to transfer their manufacturing industries to other less developed regions, and the transformation of leading industries from manufacturing industries to service industries neutralizes the benefits from service support (Shi et al., 2016).

ROBUSTNESS CHECK
We check the robustness with the following regression modelling. First, we replace the share of employment with the share of the number of firms to measure NPF. The results (see Appendix A4 in the supplemental data online) are generally in accordance with those in Table  2, except for the impact from FOEs which is not significant anymore. This further points to a complicated role

DISCUSSION AND CONCLUSIONS
The determinants of regional new firm formation have been widely discussed by scholars for a long time, and most of the literature in economic geography has focused on region-specific and economic-concentric determinants such as capital, knowledge or localized specialization to explain entrepreneurship from the perspective of cost reduction and market expansion (Armington & Acs, 2002). In recent years, scholars have started to scrutinize the impacts from locally social as well as institutional contexts on entrepreneurship and emphasize that social legitimacy is fundamental to the emergence of local firms, especially new firms (Acs et al., 2018;de Vaan et al., 2019). However, existing research mainly explores the legitimation process stemming from the growth of firms in advanced countries, but underestimated the role of cognitive legitimacy spillover in regions of transitional economies. We argue that incumbent firms that are already accepted by local society can produce cognitive legitimacy spillover to new firms if the later has an identity overlap with the former. We also point out that the identitydependent legitimation process could be very different among firms with different ownership because an ownership type can be regarded as a type of socially coded identity. Our empirical results confirm that local incumbent POEs would enhance the legitimacy of private ownership and promote new POE formation naturally because of the highest degree of identity overlap. Incumbent FOEs lend legitimacy spillover to POEs and support entrepreneurship, but their effect is weaker than that of incumbent POEs. Moreover, since SOEs and new POEs have entirely different identities, legitimacy from SOEs is miserably little and the effect of SOEs on new firm formation is significantly negative. Finally, we also find that once new POEs completed the legitimation process, the effects from FOEs and SOEs would be weakened.
Recent studies have integrated the theory of legitimacy into the framework of economic geography and argued that potential entrepreneurs require not only knowledge spillover and industrial agglomeration, but also social acceptance and local network to start a new business (Delgado et al., 2010;Frenken et al., 2015). Scholars have scrutinized the legitimacy-building process of new firms and addressed the question of how new firms actively seek legitimacy conformance from local, national and global contexts (Gong, 2020;Vestrum et al., 2017). This study moves one step further by exploring how new firms in a transitional economy achieve legitimation from local incumbent firms in different cohorts. We highlight the role of geography in cognitive legitimacy spillover by confining our geographical unit to prefecture-level cities because new ventures embed in its locality and are affected by the regional environment, and geographical proximity facilitates cognitive processes of local society to new ventures. In emerging economies experiencing a significant transformation, new firms are more dependent upon legitimacy to cope with institutional uncertainty. This study shows that incumbents with different ownership types have produced a different degree of cognitive legitimacy to new manufacturing firms in urban China.
Cognitive psychology shows that people tend to deconstruct the new firms and categorize them into the existing cognitive structure. For example, research has found that people tend to judge the rightness of new firms by looking at the situation of similar incumbent firms (Markus & Zajonc, 1985). New firms can benefit from legitimacy spillover granted by incumbent firms with an identity overlap, but are harmed by existing conservative social norms cultivated by firms with opposite identities or values. Our work also widens the theoretical scope of the impact from incumbent firms on new firm formation and finds that not only competition and crowding-out effect, but also the conformity to the existing informal institution and culture constructed and cultivated by incumbent SOEs would inhibit the formation of new private firms. A region dominated by SOEs is associated with a risk-aversion social attitude that would hinder entrepreneurship.
It points to the importance of legitimation as a crucial dimension of new firm formation and calls for social and institutional aspects of entrepreneurship studies in economic geography.
This research has several policy implications. First, new firms or organizations find it difficult to enter the local market for lack of legitimacy. The differences in the regional social context highlight the importance of region-tailored policies. Based on local social network, government, business association and managers should help to build an environment under which new firms can easily gain legitimacy from local society and cultivate a regional value that supports private ownership and encourages adventurous spirits (Lundström & Stevenson, 2005). Second, the market-oriented reform of SOEs is not a simple reform of SOEs per se, but more about turning around the old value built in the planned economy era. However, the transition from the old value and norms to new ones is not accomplished at one stroke. The promotion of new firm formation therefore requires a longterm effort that should be based upon the existing regional organizational structure and its historical root .
The study also has some limitations that need to be addressed in future. First, ownership heterogeneity is emphasized, but industry heterogeneity is underestimated because it is beyond our scope in this paper. Second, our dataset only allows us to explore the manufacturing sector. The service sectors, which play an increasing important role in the Chinese economy and have distinctive characteristics from manufacturing industries, are worthy of future research. Third, China has hosted FOEs from all over the world with quite different culture backgrounds. The culture differences between FOEs and new POEs could be moderated by the nationality of FOEs. Unfortunately, our dataset does not provide enough information for us to track the nationality of FOEs. A scrutiny on the subtle culture difference among FOEs may offer a deeper understanding of legitimation process of new firms from foreign incumbents with different culture backgrounds.

DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.