The major result of the crude adjustment for
development levels in the variable LKDEV is that
INV/OUTPUT becomes more efficient, whereas the
estimated coefficients of PREQS and ED3RD become
less significant, although their signs remain as
expected. In equation (5), the coefficient of
ZNVIEMPL is significant at the one per cent level,
whereas those for PREQS and ED3RD are not significantly
different from zero. In equation (6), the coefficient
of the alternative competition proxy INVIOUTPUT
is significant at the 5% level, but the confidence
levels for requirements and education are below 10%.
The reason why the proxies for education and requirements
are not significant in these two equations is
probably that their effects have already been captured
by the adjustment for development levels in the
dependent variable.