5.2 Sector switching and exit
We present the first set of results from the random effects probit models of sector
switching and exit in Table 9, including firm specific explanatory variables only.28 The
first columns in both the sector switching and exit specification include the weighted
relative efficiency score together with province, sector and time dummies. As expected,
we find a significant negative relationship between a firm’s relative efficiency and
sector switching as well as exit. Less (more) productive firms are more (less) likely to
switch sector or close down.
Controlling for firm size and age does not qualitatively change the relative efficiency
result, and both control variables have the expected (and well-determined) sign in the
sector switching specification. Older and larger firms are more likely to remain non-
switchers in accordance with the predictions of traditional firm life-cycle theories. Firm
size has the expected negative sign in both the switching and exit specifications, and
firm age affects switching negatively, so younger firms are more likely to switch. Yet,
firm age is positive and significant (contrary to the hypothesized effect) in the second
exit regression, where ownership is not controlled for. The ongoing reform process (and
the general legal restructuring of the business sector) is key in explaining this. Older
firms tend to be SOEs, and many are exiting in the reform process. This corresponds
with the fact that the effect of firm age is no longer well-determined in the exit
specification once we control for legal ownership type (column (3)).
Ownership structure matters for both the sector switching and exit decision, and - as
hypothesized - state and foreign owned enterprises are less likely to switch than
domestic privately owned firms. Similarly, in the exit specification foreign owned firms
are as expected less likely to exit, whereas state owned firms show the opposite
tendency. Exit is significantly associated with state ownership, and this is, as just noted,
rooted in the ongoing reform process discussed by CIEM (2003).
In the final specification the efficiency variable is only well-determined in the exit
specification. This suggests that relative efficiency is critical in the exit decision, but is
less of a driving factor behind switching, once other characteristics are accounted for.
28 Results of the conditional fixed effects logit are presented in Appendix Table D.
19