annually for about 2000 firms. It includes a very complete questionnaire about each
firm’s strategic decisions, producing good insight into the Spanish manufacturing
industry3.
The ESEE began in 1990 with 2188 firms, but we only have useful information for
1073. From those, 751 have full information for the whole period of analysis, survived
from 1990 to 2001, and 322 can be considered as firms that have exited the market
(closed) during the period. The rest did not respond to the survey anymore, fused with
other firms, or divided into multiple firms.
Firms in the sample are small in size (Table I). Though the mean of the sample could be
considered as relatively high, 266 employed people for the whole sample, 332 for the
firms that survive and 111,3 for those that disappeared, the mode and the 25th and 50th
percentiles are quite small for the three samples. As it could be expected, the skewness
coefficient confirms that the sample is very right asymmetric: most of the firms are very
small, and only a few have a large number of employees.
As we mentioned in the last paragraph, there are big differences in size between those
firms that survived and those that did not: the mean size of the surviving firms is three
times that of those that closed. This difference is corroborated by comparing the size of
the firms at various percentiles of the distribution, especially at the 75th percentile (the
25 percent of the firms with the largest number of employees). Consequently, it is
reasonable to suppose that selection bias should play an important role in the estimated
results.
On the contrary, the differences between surviving and disappearing firms for the age
variable4 are not significant.
Other aspects related to the data are shown in Table II, where we can see the
distribution of the sample for some dummy variables employed in the study. So,
limited legal liability characterizes both samples (more than 70% of them); low
technological development is a predominate characteristic of those firms that closed;
3 The questionnaire and general results of ESEE can be found in www.funep.es/PIE/ESEE/esee1.htm.
4 The variable definitions are included in the Appendix.