Firm ranking by size using multiple classifier systems (MCS) techniques, by Daniele
Porcheddu, Gianfranco Atzeni, Manuele Bicego
Firm size is a topic often studied in the economic theory and business economics,
both under a normative and positive perspective. In this work we apply the multiple
classifier systems (MCS) approach in order to extract univocal information from five
indicators of firm size. Starting from various firm ordering obtained by using a single
dimensional variable, we get a ‘consensus’ firm ordering. The empirical exercise is
carried out on the Capitalia Survey of Manufacturing Firms employing Highest Rank
and Borda Count methods. We show that total assets is the most suitable variable to
summarise the information obtained from other measures, such as employment, sales
and value added. We also show that the relative sectorial position of the firm based on
its dimension is maintained shifting from the sector to the inter-sector level. These empirical
properties make this methodology more suitable for firm comparisons at various
level, such as antitrust surveys on dominant position.
JEL classification: C43; M21
Keywords: Firms’ size ranking; Multiple classifier systems; Borda Count method; Highest
Rank method