from one being due to the standard error of the cumulative survival
estimates. Therefore the probability of surviving a given time period is one
minus the probability of exiting in that period.
The Cox model is estimated for the population of foreign-owned
plants with ten or more employees in 1986. The dependent variable is the
length of duration the plant remained operational (survived) post-1986. We
include dummy variables to account for each plant's scale of R&D activity
over the period 1986-1993, its nationality, the period in which the plant
began operating in Ireland, and the sector in which the plant is operating.
In addition we include the mean annual entry rate of foreign-owned plants
by industrial subsector and the employment size of the plant in 1986. For a
general discussion of the factors affecting survival of plants, see Audretsch
(1995) and Siegfried and Evans (1994). Specifically for Ireland, see Walsh
et al (1997) for the effect of exchange rates on foreign plants and Kearns
and Ruane (1998) for the role of research and development as it affects the
survival rates of indigenous plants.
(Table 7)
Interpretation of the Results
The estimated model (Table 7) for all plants with ten or more
employees in 1986 is statistically significant.22 We report the hazard ratio
for each of our explanatory variables. A ratio of less (greater) than one
confirms that for a one unit increase in a variable, the risk of exit for the
plant is reduced (increased). The larger the firm is in 1986, the lower the
risk of exit and the higher the probability of survival facing the plant over
the period 1986-1996. Plants in the low-tech sectors have a significantly
22 A table of descriptive statistics and correlation coefficients for all exogenous variables are in
appendix A in Tables A3 and A4.
21