3.2 Industrial districts: definition and algorithms
Let us now analyse the definition of industrial district. As is well known, the empirical
definition of this productive system is a quite complex operation. For this reason, the
literature on Italian industrial districts has devoted considerable attention to devising
definitions that could be useful for the purpose of empirical research. In this regard, we
adopt the definition used in the so-called Sforzi-ISTAT methodology (ISTAT, 1997). This
procedure identifies 199 Italian industrial districts, starting from the information on
commuting provided by the 1991 Population Census. In particular, this statistical
procedure is articulated in two steps. First, it divides the national territory into 784 Local
Labour Systems (LLS) on the basis of the degree of commuting that characterises each
Italian municipality. These LLSs are groups of contiguous municipalities characterised by a
certain degree of commuting to work. Secondly, it defines as industrial districts those LLSs
which satisfy the following three requirements: (i) in the LLS the percentage of
manufacturing employees compared to the total of non-agricultural ones must be higher
than the national average; (ii) the LLS is specialised in one particular manufacturing
industry; (iii) in the LLS the percentage of employees working in firms with less than 250
employees must be higher than the national average (ISTAT, 1997). In this way, 199
industrial districts are identified.
3.3 The data set
After this discussion on business groups and industrial district definitions we are now
ready to illustrate the main characteristics of the statistical information used in our
empirical investigation. The latter is in fact based on a new and original data set at the firm
level built up by matching two different statistical sources: namely, (i) the Italian Business
Register (ASIA — Archivio Statistico delle Imprese Attive) and (ii) the 2001 edition of the Italian
Statistical Business Register on Business Groups (Archivio sui gruppi d’impresa). In particular,
from the first source we draw information on all the Italian firms operating in the
manufacturing industry according to geographic location, productive activity (at three digits
level), and number of employees. Instead, from the Register on business groups we draw
information on control links among resident businesses and — to a minor extent — among
resident and non resident ones. Using the latter information, we are able to classify the
manufacturing firms in our sample in three different types: (i) firms without a known