only a limited role in explaining energy- and labour-productivity growth and, hence, in
determining cross-country productivity differentials. The results shown in Tables 4 and 5
might, however, suffer from an omitted variable bias if the included explanatory variables are
correlated with the unspecified country-effects μi, which are excluded from equation (6).
Hence, to correct for this potential bias, we add the unspecified individual country-effects to
equation (6), according to
g1t = a + β ln (y X t _1 + ∑ 7 jx + μ1 + εrt (7)
j=1
The results of this regression model for energy-productivity growth are presented in Table 6.
(Table 6, Page 32)
It can be seen that adding the unspecified country effects again affects the estimates
substantially. The estimation results in Table 6 show that except for Food, Non-Ferrous
Metals and Wood all sectors yield a statistically significant estimate of β, with regression
equation (7) displaying a much better goodness of fit in most sectors than equation (6) (see
Table 4). Except for Food and Non-Ferrous Metals, the speed of convergence has increased
considerably as compared to Table 4 with the half-life to be between 1 and 5 years in all other
sectors. The statistically significant energy-price effect is robust in Iron and Steel and in
Paper, while in Chemicals the null hypothesis of no effect is only just rejected at the 10%
level. In addition, energy prices also seem to have a statistically significant positive effect on
energy-productivity growth in Services and Textiles. The impact of the other explanatory
variables on energy-productivity growth remains mixed, with economies of scale having the
largest statically significant effect on energy productivity growth, being positive in Services,
Chemicals and Transport Equipment, while negative in Textiles. We found the vintage effect
and openness to have a statistically significant positive effect in Transport Equipment only,
while for specialization this is only the case in Iron and Steel.
In Table 7 we present the regression results of equation (7) for labour-productivity
growth.
(Table 7, Page 33)
17