TY  - RPRT
PB  - DEM Discussion Papers
ID  - eprints2247
EP  - 23
A1  - Bee, Marco
A1  - Riccaboni, Massimo
A1  - Schiavo, Stefano
M1  - working_paper
SN  - 2282-2801
KW  - Firm size distribution; multi-product firms; Pareto; Zipf's law; lognormal    -    JEL Codes: C46
KW  -  L11
KW  -  L25
AV  - none
N2  - The proper characterization of the size distribution of business firms represents an important issue in economic literature, with the most common reference distribu- tions being the lognormal and the Pareto varieties. This analysis is related to some methodological issues that are rarely properly addressed in applied work, and may significantly affect the results: the major difficulties arise from low power of the tests caused by limited sample size and the common practice of binning the data. In this paper we contribute to this body of literature by analyzing the size distribu- tion of all French companies, strongly rejecting the hypothesis that it is a Pareto distribution. Moreover, we argue that the lognormal distribution is a more reason- able first-cut benchmark for the entire population of firms. This is especially true for single-product firms, while we show the emergence of a Zipf tail for the class of multi-product companies. Our findings are in strong agreement with some recent theoretical contributions, which predict that the size distribution depends on a set of industry specific determinants.
Y1  - 2014///
TI  - Where Gibrat meets Zipf: scale and scope of French firms
UR  - http://eprints.imtlucca.it/2247/
ER  -