and the firms that survive are largely innovators, while non-innovators dominate those
firms that have closed. Therefore, it could be suspected that these variables should have
some influence in the survival equation.
The main characteristic considered in this study is geographical distribution. Firms have
been classified in four different groups depending on technological degree of
development of the region they are located. The distribution of the sample and ANOVA
analysis of its main characteristics depending on this geographical classification are
included in Tables III and IV5.
Table III show that firms behave differently depending on the region: those located in
provinces 2 and 4 have a higher rate of survival, specially comparing to the most
developed region. In Madrid and Barcelona (province 1) almost half of the firms have
disappeared during the 12 year period. On the contrary, more than 86% of province 2
firms survive.
The ANOVA analysis of Table IV shows significant differences in process innovation
and sector’s technological development depending on province’s classification for the
whole sample and surviving firms, but those differences disappear for the sample of
firms that close. Employment, age, legal liability and product innovation are not
significantly different for any sample.
The model
In order to test Gibrat’s law we use a typical equation in which employment in the last
period (2001) is dependent on employment of the first period (1990) and the rest of
variables. The equation is:
5 The distribution of the variables in the sample depending on location characteristics are included in the
Appendix. Tables A1 and A2