( Figure 2a and 2b, Pages 23 and 24)
From Figure 2a it can be seen that only Manufacturing resembles the macroeconomic pattern
of σ-divergence for energy productivity. Transport, Agriculture, and in particular Services,
display evidence of σ-convergence. Note that the level of cross-country variation is relatively
high in Services, which is to a large extent due to the exceptional and so far unexplained
energy-productivity performance of Finland and Italy.12 Figure 2b shows that the
macroeconomic pattern of σ-convergence for labour productivity is only evident in Services
and to a lesser extent in the Agricultural sector. At the same time, variation in cross-country
productivity differentials remains overall fairly constant within aggregate Manufacturing and
Transport (although with fluctuations over time).
Comparing Figures 2a and 2b shows again that in each sector the cross-country
variation of energy productivity is substantially larger than of labour productivity. These
results do not change when the United States is removed from the sample. Moreover, they
accord well with the findings of Bernard and Jones (1996a), who by means of a conclusion
suggest “that international flows, associated mostly with Manufacturing, may not be
contributing substantially to convergence either through capital accumulation or
technological transfer” (Bernard and Jones 1996a:1230). Our analysis suggests that this
conclusion holds even stronger for manufacturing energy-productivity performance, where
international flows cannot prevent an increase in cross-country differences of productivity
levels.
This raises the question as to what the determinants of these cross-country
productivity differences are. In our search for an answer we subsequently take three steps.
First, we go one step further in the σ-convergence analysis than Bernard and Jones (1996a,b)
by examining productivity convergence for a breakdown of aggregate Manufacturing in order
to see to whether the energy-productivity divergence and the lack of labour-productivity
convergence observed in aggregate Manufacturing is also found within the different
Manufacturing sub-sectors. Second, we perform a β-convergence analysis to test whether a
statistically significant negative relationship exists between the initial level and the growth
rate of productivity performance, in order to gain a better insight in the mechanism behind the
12 Excluding Finland and Italy from the sample for Services reduces the cross-country dispersion by about 40%
while leaving the pattern of σ-convergence unchanged. Note that the Netherlands also exhibits an exceptional
development of energy-productivity performance in Services, but has already been excluded form the sample
used in Figure 3a.