relatively small - although seemingly persistent - difference in the degree of cross-country
variance. Moreover, convergence patterns turned out to depend on the level of aggregation,
with different sectors displaying varying behaviour: some show reduction in variation, some
increasing variation and others neither a clear reduction or increase in cross-country
differences over the whole period.
These results suggest that different mechanisms may be at work in the different
sectors. For example, the observed patterns of divergence might be the result of increasing
international specialization while the tendency to converge might be caused by technology
spill-overs from ‘leaders’ to ‘followers’, allowing lagging countries to catch-up. Moreover,
our results suggest that determinants of energy-productivity growth and labour-productivity
growth might differ from each other, since we found no clear-cut (and sometimes even an
opposite) relationship between cross-country convergence patterns in terms of energy
productivity and labour productivity. Finally, even in those sectors showing evidence of
convergence there remains substantial cross-country productivity differences, in particular in
terms of energy productivity. This suggests that convergence does not pertain to a uniform
steady state for all countries. In order to further examine this issue, we continue in the next
section with a search for empirical regularities in the productivity improvements over our
cross-section of countries by testing for sectoral patterns of β-convergence. As part of that
analysis we will also try to explain (differences in) energy- and labour-productivity growth.
5. β- CONVERGENCE
This section deals with the notion of convergence in terms of growth rates. In this case
energy- and labour-productivity convergence implies that energy- and labour-productivity
growth rates tend to decline if countries reach their steady states. Obviously, it is not an easy
task - if possible at all - to judge whether a country is in its steady state or not. It is possible,
however, to analyse the correlation between growth rates and levels, assuming that a negative
correlation between these two provides an indication for convergence, because it suggests
that countries with relatively low initial energy- and labour-productivity levels catch-up to
more advanced economies. The rationale behind this is that those countries lagging behind in
terms of productivity levels can benefit from relatively high marginal factor productivity or
from the experience and technologies developed by the countries operating at the forefront. In
this section we adopt a panel-data framework to regress average energy- and labour-
productivity growth rates on initial productivity levels, generating an estimate of the
10