Experience, Innovation and Productivity - Empirical Evidence from Italy's Slowdown

Appendix

We control for sample selection that could actually come up when Capitalia-Unicredit IX survey
data are matched with AIDA balance sheet of firms present in 2007. Not all Capitalia firms exist in
AIDA register. Nonetheless, we manage to retain almost 86% of the Capitalia sample. Therefore,
we check in what type of characteristics do firms in-sample and out-of-sample differ.

Figure A 1. In and Out Sample distribution of Capitalia firms by size

Firm distribution by number of workers in and out-sample

∣^~^~∣ out-of-sample            in-sample

Figure A1 shows the distribution by class of workers of the firms falling in and out of our final
panel. The panel tends to maintain medium size firms mainly (87%), while keeping around 79% of
the medium-large and large firms. As far as the very small firms, our panel keeps 82% of them.
Formally, the test for independence hypothesis rejects the null (Pearson chi-square(4) = 25.7455, p-
value = 0.000) meaning that being in or out of sample depends in a certain way on firm size.

We lose 15.6% of firms located in North-West part of Italy (Lombardia, Piemonte, Liguria, Valle
d’Aosta), 13.9% of the firms located in the North-East (Trentino A.A., Veneto, Friuli V.G., Emilia
Romagna), 13.5% of the firms located in the Centre (Toscana, Umbria, Marche, Lazio) and 15.8%
of the firms located in the South. The Pearson chi-square(4) = 3.4150 with p-value = 0.491 says that
there is statistical independence between the regional distribution and being in or out of sample.

29

<<   28   29   30   31   32   33   34   >>

More intriguing information