Spatial disaggregation patterns and structural determinants of job flows: an empirical analysis

Abstract

The paper investigates the changes in job creation and destruction flows at a very disaggregated level of analysis. We analyse whether job flows at lower levels of spatial aggregation display regularities that are in line with national ones in a bid to disentangle the role of labour market institutions. Using a unique database of the population of firms in Trentino (a north‐eastern province of Italy) from 1991 to 2001, we find that: (1) job flows show a ‘fractal’ tendency, i.e. many regularities appear to be scale invariant (magnitude of flow and persistence) and that job flow magnitude is in line with the average values for Italy; (2) there are some qualifications to their ‘fractality’: the contribution of entrant firms to the job creation process is lower than the corresponding contribution at national level, as is the share of job destruction accounted for by exiting firms; and (3) firm size and age influence job flows.

Keywords:

• labour reallocation
• job creation
• job destruction
• spatial disaggregation two‐stage Heckman estimator

JEL classifications:

• C34
• J23
• L11

Acknowledgements

We would like to thank Rob Alessie, Giovanni Dosi, Giorgio Fagiolo, Christopher L. Gilbert, Paola Villa, Enrico Zaninotto, participants in the EARIE Conference, Porto, Portugal, 2–5 September 2005, and in the seminar held in Trento University for interesting comments and helpful suggestions. We also thank Giampaolo Sassudelli (ISTAT, Trento) for support in creating the database, and Karen Klomp and Mihaela Ghita for research assistance. The financial support of the University of Bergamo (grant ex 60%, n. 60CEFI08, Dept. of Economics) and of Provincia Autonoma di Trento is gratefully acknowledged. The usual disclaimers apply.

Notes

1. INPS: Istituto Nazionale della Previdenza Sociale.

2. ISTAT: Istituto Nazionale di Statistica.

3. See Blanchflower (1994) for an analysis on job creation and destruction in which the concept of fractality is used.

4. Details about DM/10 model can be found in Gallo (2003).

5. One of these examples is the IDA database in which information about the Danish economy is collected. More information about it can be found on the Denmark statistic website at the address: http://www.dst.dk/HomeUK/Guide/Varedeklarationer/emnegruppe/. See also Lee (1983) for a theoretical extension on the issue and for an application.

6. See Frazis (1993) for an application of the same procedure on the effects of different degrees on performances of students. Alonso‐Borrego and Collado (2002) apply the method to study the role of innovation in job creation and destruction flows.

7. We recall that we do not have data for the period from January 1997 to December 1997.

8. See Ashenfelter and Card (1999) for a complete picture of the state‐of‐the‐art of labour market literature. Pissarides (2000) provides an exhaustive survey of matching models that appears to be the main theoretical approach in labour market studies.

9. See Blanchard and Diamond (1989) for a discussion about the issue and for a good example of matching model that try to account for job flows.

10. See Kirman (1992) for a discussion about the fallacies emerging from using a representative agent in the economic analysis.

11. See Leombruni (2003) and Fagiolo, Dosi and Gabriele (2004) for a discussion about the use and an example of the interpretative power of evolutionary modelling in the labour market. In particular, Fagiolo, Dosi and Gabriele (2004) propose an evolutionary model able to interpret most of the labour market dynamics based on less restrictive assumptions in terms of the number of firms and workers and in terms of their homogeneity restrictions in behaviour. They demonstrate that once the usual models of the labour market are abandoned, it is possible to reproduce and explain different regularities of the labour market in a dynamic perspective. In this framework, any restriction in the steady state is not necessary in order to solve the model. They analyse the labour market in its dynamic evolution over time and map different institutional setup with different growth regimes for the system.

12. See Baldwin, Dunne and Haltiwanger (1998), Barnes and Haskel (2002) and Davis, Haltiwanger and Schuh (1996).

13. The linkage between firm size and job creation and destruction is negative. Although, the descriptive evidence contains some bias due to the choice of size classes. The regression analysis discussed below clarifies the point.

14. The regression analysis undertaken earlier takes into consideration a finer level of disaggregation to account for sectoral differences. The sectoral controls are included at 2 digit ATECO2002 level.

15. Results obtained using current average size methodology are similar.