estimator of the employment variable is bigger than 1. This results support the
agglomeration effects discussed in the first section.
Another two conclusions can be extracted from the estimation: firstly, the right election
of the method used to correct from sample attrition, Heckman’s methodology, since the
test for independence of the equations has shown they are related, independently of the
regions considered; secondly, innovation, process or product, is the key factor for firm
survival over the years considered. The rest of the variables are not significant.