Finally, it is to be noted that the ability of the simple regression equation (1) to
explain cross-country productivity growth rates is rather small in most sectors. This is not
surprising since the specification of equation (1) implicitly builds upon the assumption that
energy- and labour-productivity levels converge towards a uniform steady state. However,
economies differ and so do (most likely) their steady states. Contrary to a framework of
single cross-country regressions, a panel data framework is capable of allowing for cross-
country differences in steady states in the form of unobservable individual ‘country-effects’
(Islam 1995). These country-effects might include all sorts of country-specific tangible and
intangible factors that affect productivity growth and which have not been included in
equation (1) or, to state it differently, have been subsumed in its error term. Therefore, in the
next section we test for β-convergence allowing for these ‘country-effects’.
5.2 Conditional β-convergence
We test for conditional β-convergence by including (unspecified) individual country-effects
in equation (1), assuming that productivity levels converge to country-specific steady states.
Hence, we reformulate equation (1) into a panel-data model with individual country effects,
according to:
git = β ln(У)i,t-1 + βi + Zit (5)
with μi representing unspecified country-specific (fixed) effects. In Table 3 we present - for
each sector - the estimated coefficient β obtained from equation (5).
(Table 3, Page 29)
The table shows that allowing for individual country-effects substantially improves the
explanatory power of the regression equations for both energy- and labour productivity.
Moreover, in terms of energy productivity, equation (5) yields significantly negative
estimates of β in all sectors, including now also Total (i.e., the macroeconomic level),
Chemicals, Iron and Steel, Non-Ferrous Metals and Machinery. Also in terms of labour
productivity the estimates of β are higher in several sectors such as, for example, Services
and Food. The evidence on conditional labour-productivity convergence is, however, less
clear-cut than it is for energy-productivity convergence: in some sectors such as, for example,
13