Testing Gibrat’s Law: Empirical Evidence from a Panel of Portuguese Manufacturing Firms

Abstract

The purpose of this paper is to use Dynamic Panel Data (DPD) models with serial correlation in the error term to see if Gibrat’s law holds and to analyse the empirical determinants of firm growth. This paper makes significant contributions to the empirical literature on the dynamics of firm growth, since it updates the work carried out by previous researchers in this field using micro panel data, dynamic firm growth models with serial correlation in the error term, panel unit root tests and GMM‐system estimator. To conduct this study we used an unbalanced panel of Portuguese manufacturing firms over the period from 1990 to 2001. The main implication of our findings is that firm growth is not quite random since there are some determinants which exert influence on firm growth.

Key Words:

• Firm Growth
• Gibrat’s Law
• GMM‐System Estimator
• Panel Unit Root Tests
• JEL Classifications: L11, C23

Notes

1. With β < 1, in the short run it is possible for the variance of the cross‐sectional distribution of firm sizes to either increase or decrease. In the long run, however, this variance converges and stabilises at its equilibrium value.

2. Extensive discussion on these issues can be found in Lang et al. (1996) and a thorough review of theories of capital structure in Harris and Raviv (1991).

3. These estimators are linear and estimation software is freely available (DPD package for OX).

4. This was assessed by comparison with alternative estimators such as OLS levels, which are known to produce biased estimates of autoregressive parameters.

5. Although a more efficient two‐step GMM estimator is available, the asymptotic standard errors for the two‐step estimator can be an unreliable guide for inference in finite samples. The system GMM estimates that we report are computed using DPD for OX (see Doornik et al., 2002).

6. See, for example, Hart and Oulton (1996).

7. Unfortunately, since only a small number of our firms are quoted in the Lisbon Stock Exchange, it has not been feasible to calculate Tobin’s q, as an index of a firm’s known growth opportunities.

8. Official Journal of the European Communities N^○ L 107/04 of 30.4.96.

9. Bechetti and Trovato (2002) drew the same conclusions.

10. This result confirms the stability condition (|| < 1).

11. See for example Jovanovic (1982), Evans (1987), Dunne et al., (1989), Dunne and Hughes (1994), Almus and Nerlinger (1999).

12. These results are given on request.

Additional information

Notes on contributors

Blandina Oliveira

We are grateful to Stephen Bond, Pedro Portugal, Paulino Teixeira, João Sousa Andrade and the participants at the International Industrial Organization Conference (Boston, 2003), 11° Encontro Nacional de Economia Industrial (Viana do Castelo, 2003), and 30th European Association for Research in Industrial Economics (Helsinki, 2003) for useful comments on a previous draft. We are also grateful for the helpful suggestions and comments of the editor and two anonymous referees of this journal. All errors and omissions remain our responsibility.