away. Putting focus on columns (2) and (4) we note that these observations are
consistent with our findings for sector switchers compared with exits only.
In summary, we have identified a series of highly significant variables affecting firm
switching out of a sector or closing down. Moreover, sector switchers and exits tend to
be affected differently by firm and sector specific factors. Younger firms are more likely
to switch sectors while exit is significantly associated with older firms. The latter result
is primarily due to changes in ownership structure as state owned firms are more likely
to exit rather than switch sector as a result of the economic reform process. Foreign
owned firms are less likely to exit and switch sectors, but when the sample is restricted
to firms which either exit or switch the latter group has a significantly higher
concentration of foreign participation. Sector specific differences also exist. A higher
share of foreign firm output, higher levels of sector concentration and higher sector
level efficiency are more associated with sector switching than exit. In contrast, sector
switching is less likely than exits where sectors are either more dominated by SOEs or
heavily protected.
6. Conclusion
Much of the literature on industry evolution has found firm turnover to be an important
source of sector-level productivity growth in the sense that new firms are widely
understood to have higher productivity than exiting firms. In this paper, we approached
industry dynamics in Vietnam from a new angle and asked, first, whether the
delineation of “entry” and “exit” firms matters for the impact of firm turnover on
growth outcomes. We found that the sub-group of exiting firms, which continue
production in a different sector have different characteristics to those that cease
production altogether.
In the manufacturing sector in Vietnam, less productive firms are more likely to switch
sector and to exit as predicted by existing literature. However, while firms which switch
sectors have lower efficiency levels than incumbents, they are often significantly more
efficient than new entrants. This is important given that average efficiency levels are
23