Survival chances of new businesses: do regional conditions matter?

Abstract

This article analyses the effects of industry-, regional- and firm-level characteristics on the post-entry performance of new businesses by means of an econometric survival time model. First preference is given to an accelerated failure time model assuming a log-logistic distribution. The data involve a representative sample of businesses in the private sector of West Germany during 1993 to 2002 period. The results demonstrate that the regional dimension is most important; whereas firm-level characteristics play a subordinated role.

Acknowledgements

The author is indebted to Lutz Bellmann, who helped make possible a research fellowship in cooperation between the University of Passau and the Institute for Employment Research. The analysis presented in this article was conducted during the author's stay at the IAB in Nuremberg, Germany. Gratitude is also due Charlie Karlsson, Susanne Kohaut, Claus Schnabel and three anonymous referees for making comments on earlier versions of the article that resulted in its improvement.

Notes

¹ Burke et al. (2005) account for the possibility of multi-entrepreneur start-ups.

² For a survey of the literature, see Geroski (1995) and Sutton (1997).

³ Harhoff et al. (1998) and Mata and Portugal (2002) discuss the impact of legal form on business survival. For more on the influence of foreign ownership, see Hymer (1976), Braconier and Ekholm (2000), Mata and Portugal (2002) and Bernard and Sjöholm (2003). The influence of national subsidies on new business survival is analysed by Santarelli and Vivarelli (2002). Persson (2004) analyses the survival chances of businesses that belong to a multi-unit firm. Klepper and Sleeper (2005) discuss the role of spinoffs.

⁴ See Audretsch (1995, pp. 36–64), Marsili (2002) and Winter (1984) for the concept of the technological regime with which innovation activities in industries can be described.

⁵ However, there appear to be numerous papers discussing the impact of regional conditions on new firm formation – especially in Eastern Germany – as well as on the development of newly founded businesses. See Steil (1999), Bellmann et al. (2003), Fritsch and Niese (2004) and Brixy and Niese (2004). Fritsch (2004) compares entry and performance of new business in East and West German growth regimes.

⁶ See Bellmann (1997) for a description of this data source.

⁷ For a detailed description of the concept of the hazard function, see Section Estimation Procedure.

⁸ Regression coefficients are interpreted differently in the proportional hazard model. In the proportional hazard model, regression coefficients proportionally shift the hazard rate as the value of one regressor increases by one unit, all other regressors being constant.

⁹ The agency BIK Aschpurwis + Behrens GmbH provides a classification system for municipalities.

¹⁰ In the case of time-varying covariables, the hazard function is with .

¹¹ Alternatively, a model with unobserved individual γ-distributed heterogeneity was estimated. However, the likelihood ratio test proposed by Nielson et al. (1992) in order to test whether or not theγ-variance parameter shall be considered is not significant at a p-value of 1.000. In this case, the value of the χ²(1)-distributed test statistic equals 0.00.