Job Creation, Firm Creation, and <I>De Novo</I> Entry

Abstract

Firm turnover and growth recorded in administrative data sets differ from underlying firm dynamics. By tracing the employment history of the workforce of new and disappearing administrative firm identifiers, we can accurately identify de novo entrants and true economic exits, even when firms change identifier, merge, or split-up. For a well-defined group of new firms entering the Belgian economy between 2004 and 2011, we find highly regular post-entry employment dynamics in spite of the volatile macroeconomic environment. Exit rates decrease with age and size. Surviving entrants record high employment growth that is monotonically decreasing with age in every size class. Most remarkably, we find that Gibrat's law is violated for very young firms. Conditional on age, the relationship between employment growth and current size is strongly and robustly positive. This pattern is obscured, or even reversed, when administrative entrants and exits are taken at face value. De novo entrants' contribution to job creation is relatively small and not very persistent, in particular for (the large majority of) new firms that enter with fewer than five employees. The authors are grateful to the Belgian National Social Security Office (NSSO), in particular to Peter Vets, for providing the data and for helping to develop the employee flow record linking method; and to Statistics Belgium, in particular to Youri Baeyens and Antonio Fiordaliso, for linking the NSSO data to the Belgian Business Register and for providing firm record linkages created for the Eurostat Structural Business Statistics. Financial support from ERC grant No. 241127 and KU Leuven project financing is gratefully acknowledged.

The Dynamics of Employment Growth

“…The idea underlying this data construction exercise is to exploit payroll data to track workers across different firm identifiers; if a high percentage of workers in an exiting business at time t is found to be recorded under a new firm at time t+1, then the two observations are considered to be from the same business. Similar techniques have been used in other countries, also to identify outsourcing and insourcing patterns: in the United States as described in Benedetto et al (2007), in Belgium (see Geurts and Van Biesebroeck, 2013), in Denmark (Ibsen and Westergaard-Nielsen, 2011) and in Germany (at the establishment level: Hethey-Maier and Schmieder, 2013). 12 The US Census uses a different approach to compute establishment and firm age from the data in the Longitudinal Business Database (LBD) (see Becker et al, 2006;and Davis et al, 2007).…”

Cross-country evidence on start-up dynamics

“…However, small start-ups are generally more likely to be genuinely new firms (de-novo entry) as compared to larger entrants, which are relatively more likely to be the result of mergers or acquisitions or legal rather than economic changes in the life of a firm (i.e., de-alio entry; see Geurts and Van Biesebroeck, 2013). For this reason, all graphs reporting statistics relative to all entrants are replicated in Annex B limiting the selection only to small units (e.g., below 50 employees) for those countries for which data availability allows it.…”

How Sensitive is the Analysis of Firm and Employment Dynamics to Longitudinal Linkage Problems?

“…for firms involved in an event in a given period t-1 to t. In the next period, we restart from registered employment in t and impute employment in t+1 for events in that period. Geurts and Van Biesebroeck (2014) have extended the imputation method over a five-year period. They found that firm ID numbers involved in a linkage event are more likely to be involved in another event in one of the following years.…”

“…Spurious entrants and exits are eight times larger on average than real entrants and exits (Table A.2 in the Appendix). Other studies have demonstrated that entry and exit from ID changes, firm restructurings or diversifying firms also differ in characteristics other than size, such as in the determinants of entry (Acs and Audretsch 1989;Storey 1991), post-entry growth patterns (Mata et al 1995;Geurts and Van Biesebroeck 2014), or the profitably and productivity at exit (Baldwin and Gorecki 1987).…”

“…Distinguishing these spurious entrants and exits from de novo entrants and exits by closure has implications for firm-level analysis that reach beyond the set of empirical measures considered in this paper. Several studies have highlighted that both populations fundamentally differ in characteristics other than size (Baldwin and Gorecki 1987;Acs and Audretsch 1989;Storey 1991;Mata et al 1995;Geroski 1995;Geurts and Van Biesebroeck 2014).…”