| Περίληψη: | The asymmetric outcomes of firms’ performance due to conceptual heterogeneities
have considerably drawn the attention of scientific community. The aim of this
dissertation is to detect what determines high-growth firms (HGFs). A random
forest approach is then used to predict whether a Greek firm will be HGF in three
periods ahead, i.e 2018, given current period’s information, that is 2015. To this
end a database from ICAP which includes 91999 cases of Greek firms between
2005-2018 is employed. HGFs dummies which constitute the dependent variables
are defined by two indices, founded on OECD-Eurostat’s and Birch’s definitions
of HGFs. The set of explanatory variables includes firms’ financial indicators,
age, sector of economic activity and region. Main results of the study suggest
that Greek firms principally consist of LGFs between 2009-2018, while intense
declines in the number of HGFs imply the weak sustainability of Greek HGFs. It
is further inferred that HGFs are younger and LGFs older. The results of Birch
Index manifest that random forest accurately predicts 16.8 % of HGFs. Instead,
True Positive Rate entails that random forest tends to classify firms as LGFs.
Besides, OECD-Eurostat results suggest that random forest fails to predict HGFs,
whilst accurately predicts LGFs. The prediction of HGFs is heavily dependent on
firm’s size which is captured by the Total Assets and its growth variables. Finally,
most of the significant predictors of HGFs correspond to financial aggregates and
firm’s age.
|
|---|
