Network Survival Strategies of Migrant Entrepreneurs in Large Cities: Analysis of Albanian Firms in Milan

ABSTRACT

This paper addresses the role of cultural bias (preference for what is culturally more akin) in the entrepreneurial choice regarding different types of social networks in the context of urban mixed embeddedness. We test empirically the presence and aftermaths of this cultural bias, drawing on evidence from a natural experiment with regard to Albanian ethnic entrepreneurs in the city of Milan, Italy. Namely, these entrepreneurs are exposed to the same mixed urban embeddedness and, when we control for firm characteristics, the only discriminating component explaining their success is their choice of social network. We focus on the choice over three types of social networks, classified according to varying degrees of cultural distance between the network and the entrepreneur: (a) the indigenous population, (b) the local Albanian diaspora, and (c) fellow citizens residing in the country of origin, Albania (i.e., transnational networking). We employ a novel method for reverse engineering of preferences for networking by using a Kaplan-Meier estimator and a propensity-score matching technique. We find that strategic network liaisons with locals is actually the most beneficial social network for ethnic firm performance. However, it is social networking within the culturally closer local Albanian diaspora that is the most common behaviour.

KEYWORDS:

• Networks
• social capital
• mixed embeddedness
• home bias
• transnational networking ethnic entrepreneurship

Disclosure statement

No potential conflict of interest was reported by the author.

Supplementary material

Supplemental data for this article can be accessed here.

Annexe 1: Data Gathered through Structured Interviews

The interviewed ethnic entrepreneurs were asked about their: status of employment before starting entrepreneurship; time duration of being entrepreneur in the host country’s market; reason for being an entrepreneur; main way of starting the business; sources of capital to found the enterprise; role and status played in the enterprise; relation with employees in the enterprise; the number of the employees since the begining; the place/market segment to sell the products and services; origin of the buyers; size of the market segment; origin of the suppliers; way of communication with the market; weight of the Albanian products in the total budget; origin of the goods and services bought; number of functions of business completed within enterprise; national origin of the competitors; total competitvenes capacity; “level of ties with egos and alters (level of internal/strong ties – 1,2,5,6 – X21a) and (level of external/weak ties – 3,4,7 – X21b); origin of the intermediator for contacts, hiring employees or collaborators; origin of the loan; origin of the organization or networks of membership; impact of the economic crisis on the business; 31-33 are related to social status and social involvement of the entrepreneur; economic conditions of the family of origin compared to the average of the country; national origin of the spouse/husband; employment before immigration; immigration motive.

Notes

^1. The most typical examples of cultural distance is the obvious linguistic distance, as well as distance in internalized norms of behaviour and communication, socialization patterns and expectations (see Bielenia-Grajewska, Carayannis, and Campbell 2013).

The most typical examples of cultural distance is the obvious linguistic distance, as well as distance in internalized norms of behaviour and communication, socialization patterns and expectations (see Bielenia-Grajewska, Carayannis, and Campbell 2013).

2. Focusing on one type of ethnic entrepreneurs, from only one ethnicity and in one specific locality (viz. Albanian small entrepreneurs in the Milan area), allows us to reduce all context-related cultural heterogeneity with regard to opportunity structures and institutions. The entrepreneurs in our dataset also share a common cultural-ethnic origin. This provides us with a group for empirical analysis that is fairly homogeneous and is situated in the same mixed embeddedness context. Therefore, we can clearly identify different effects from using (i) networks of incumbents (Italian network) (ii) networks of Albanians in Milan (diaspora networks) and (iii) Albanians in Albania (transnational networks).

^3. A social network has been defined by Reggiani and Nijkamp (2009) and Kesler and Bloemraad (2010) as structures of connected people.

A social network has been defined by Reggiani and Nijkamp (2009) and Kesler and Bloemraad (2010) as structures of connected people.

^4. Markusen (1996, 2004) recognizes networks as one of the most prominent causes of a firm’s development and survival.

Markusen (1996, 2004) recognizes networks as one of the most prominent causes of a firm’s development and survival.

^5. According to ISTAT, Milan was in 2011 the leading region in Italy in terms of share of foreigners in the regional population. Thus, Milan is only comparable to and competing with Rome, the two cities sharing or exchanging the first place in terms of diversity, depending on the measurement approach and period considered.

According to ISTAT, Milan was in 2011 the leading region in Italy in terms of share of foreigners in the regional population. Thus, Milan is only comparable to and competing with Rome, the two cities sharing or exchanging the first place in terms of diversity, depending on the measurement approach and period considered.

6. https://www.urbandivercities.eu/wp-content/uploads/2017/02/Divercities-City-Book-Milan.pdfhttps://www.rc21.org/en/wp-content/uploads/2014/12/G3-Marzorati.Quassoli.pdfhttps://core.ac.uk/download/pdf/6264162.pdf

^7. The eventual social desirability bias (which would suggest that the interviewee might feel pressed to report preference for networking with the local population in order to be perceived as socially agreeable by the interviewer) is not likely in our case, since the interviewer.was of Albanian origin and we have promised full anonymity to the interviewees. Moreover, the question was phrased around ‘what is more important to you’ (addressing the affect), rather than ‘what do you perceive as the more important network for your business’ (which would address a rationally perceived merit from supporting this network). Thus, we believe to have successfully captured the social value that the entrepreneur bestows onto the three types of networks.The eventual social desirability bias (which would suggest that the interviewee might feel pressed to report preference for networking with the local population in order to be perceived as socially agreeable by the interviewer) is not likely in our case, since the interviewer.was of Albanian origin and we have promised full anonymity to the interviewees. Moreover, the question was phrased around ‘what is more important to you’ (addressing the affect), rather than ‘what do you perceive as the more important network for your business’ (which would address a rationally perceived merit from supporting this network). Thus, we believe to have successfully captured the social value that the entrepreneur bestows onto the three types of networks.

^8. The interviewed 53 small entrepreneurs have been guaranteed full anonymity according to the institutional ethical requirments for data collection. Therefore, we cannot provide any revealing individual firm level information about any of the participants, as it may lead to identifying the entrepreneur. We have however, provided all the summary average characteristics of the interviewed sample, depicting in sufficient detail the heterogeneity of the firms and their mapping against the subgroups in the full sample.The interviewed 53 small entrepreneurs have been guaranteed full anonymity according to the institutional ethical requirments for data collection. Therefore, we cannot provide any revealing individual firm level information about any of the participants, as it may lead to identifying the entrepreneur. We have however, provided all the summary average characteristics of the interviewed sample, depicting in sufficient detail the heterogeneity of the firms and their mapping against the subgroups in the full sample.

^9. Since we have data comprised of only Albanian entrepreneurs, it is difficult to claim that Albanians are typical or atypical per se in order to generalize our results. However, having the opportunity to focus on a representative sample for only one ethnic group, allows us to analyse the factors of interest here, free from the cultural relativity bias when comparing choices made by people from different ethnic groups.Since we have data comprised of only Albanian entrepreneurs, it is difficult to claim that Albanians are typical or atypical per se in order to generalize our results. However, having the opportunity to focus on a representative sample for only one ethnic group, allows us to analyse the factors of interest here, free from the cultural relativity bias when comparing choices made by people from different ethnic groups.

^10. In a total population of 200 firms, a sample of 53 firms is representative with confidence level 95% and margin of error 12%.In a total population of 200 firms, a sample of 53 firms is representative with confidence level 95% and margin of error 12%.

^11. The full range of questions the interviewed answered can be seen in Annexe 1.The full range of questions the interviewed answered can be seen in Annexe 1.

^12. Namely, we measure success through number of employees. The firms have between 1 and 10 employees. We code as survival having more than 1 employee, next having more than 2 employees, then having more than 3 employees, with the last survivors having only 10 employees each. The same logic can be applied when success is measured through generated turnover.Namely, we measure success through number of employees. The firms have between 1 and 10 employees. We code as survival having more than 1 employee, next having more than 2 employees, then having more than 3 employees, with the last survivors having only 10 employees each. The same logic can be applied when success is measured through generated turnover.

^13. Results are robust to different cut-off points. However, choosing 8 or above for a Likert scale with 10 levels approximates choosing those people whose likelihood to really choose a particular network was above 80%.Results are robust to different cut-off points. However, choosing 8 or above for a Likert scale with 10 levels approximates choosing those people whose likelihood to really choose a particular network was above 80%.

Additional information

Funding

Peter Nijkamp acknowledges  grant PN-III-P4-ID-PCCF-2016-0166 (PNCDI III ReGrowEU) of the Romanian Ministry of Research and Innovation (CNCS-UEFISCDI).