(Table 4, Page 30)
It can be seen that the regression analysis generates significantly negative estimates of β in
Agriculture, Transport, Iron and Steel, Machinery, Non-Metallic Minerals, Paper and
Textiles. Compared to Table 3 this means that in 6 sectors the estimate of β is no longer
statistically significant once we include the previously mentioned specified explanatory
variables. The effect of investment share, openness, specialization, and economies of scale on
energy-productivity growth is mixed and their impact is statistically insignificant in virtually
all sectors. An exception is the energy-price effect, which has in all sectors the expected
(positive) sign, while the positive impact of energy prices on energy-productivity growth is
statistically significant in Chemicals, Iron and Steel and Paper, which makes sense since
these are energy intensive sectors. Finally, the speed of convergence has slowed down with
the half life increased to a minimum of 5 years (Wood) and a maximum of 41 years
(Chemicals).
In Table 5 we present the results of regressing average labour-productivity growth
rates on initial labour-productivity levels and the five additional explanatory variables,
according to equation (6).
(Table 5, Page 31)
Table 5 shows statistically significantly negative estimates of β for Agriculture, Food and
Textiles only. Similar to energy-productivity growth, the effect of the investment share,
openness, specialization, and economies of scale on labour-productivity growth is mixed
while their impact is statistically insignificant in virtually all sectors (or it is statistically
significant with an unexpected sign such as, for example, in case of the ‘vintage effect’ in
Agriculture, Iron and Steel, Transport Equipment and Wood, and economies of scale in
Services). Finally, the speed of convergence has slowed down with the half life increasing to
in between 12 years (Textiles) and 95 years (Machinery), while for Services the estimate
yields a positive β , implying divergence.
In sum, while there is strong evidence of conditional convergence in terms of both
energy- and labour-productivity growth (see Table 3), we found energy prices and in
particular wages, investment share, openness, specialization and economies of scale to play
16