Ceris-CNR, W.P. N° 10/1997

Assuming that λi are log-linearly dependent on the explanatory variables, we can write:
lnλi =β0 + Σjβjxij

(2)

Maximum likelihood estimators of (2) are readily available in most specialised econometric
packages9.
In entry/exit in foreign countries/primary industry equations, basic explanatory variables can
be summarised as follows:

Σjβjxij = β1Ci + β2GRi + β3MSi + β4CRi + β5SEMi

(3)

where:

C*i = Number of countries (in logs) firm i was not present in 1987 in its primary industry. This
variable is used only in entry equations;
Ci = Number of countries (in logs) firm i was present in 1987 in its primary industry. This variable is
used only in exit equations;
these two variables capture the number of entry/exit options available to the firm. In the entry
equation, it is expected to be positive if firms operating in a limited number of countries are more
likely to enter new foreign countries than firms whose operations are already widely scattered
throughout Europe. Analogously, in the exit equation it is expected to be positive if firms operating in
a large number of countries are more likely to close foreign plants than firms operating in a small
number of countries.
GRi = 1991-1987 EU Growth Rate in firm i’s primary industry. This variable measures expected
growth prospects and is expected to enter positively in entry equations and negatively in exit
equations.

9

For the ML estimates of (2) we used Limdep 7.

12