Unander et al., 1999; Mulder and De Groot 2003), to the best of our knowledge this literature
lacks a systematic convergence analyses from a macroeconomic perspective. Hence, we add
to the existing literature a systematic comparison of energy- and labour-productivity
convergence. Finally, we do so at a detailed sectoral level. By looking at cross-country
convergence patters within sectors, our analysis differs from virtually all convergence studies
in the empirical growth literature, since they employ aggregated data. Important exceptions
are sectoral studies by Dollar and Wolff (1988, 1993) and Bernard and Jones (1996a,b) who
- using (partly) the same data source (OECD’s ISDB) - conclude that a convergence analysis
of aggregate productivity levels masks substantial differences at the sectoral level. An
important underlying reason for this result is that productivity levels, measured as the ratio of
value added over a unit of input (energy, labour), can substantially differ among sectors
because some activities require inherently more capital, higher labour skills and/or
technology than others. Aggregate productivity trends are therefore not directly attributable to
technological change in individual sectors, as they can also be the result of changes in the
distribution of production factors among sectors. Our sectoral approach corrects for most of
the impact of such changes in the structure of production on aggregate productivity
developments and, hence, establishes a closer link to issues concerning international
convergence of technology driven productivity performance. Our analysis differs from the
previously mentioned sectoral convergence analyses in comparing labour- and energy-
productivity convergence, in further disaggregating the manufacturing sector into 10 sub-
sectors,2 in using more recent data and in carrying out a more extensive search for country-
and sector-specific factors to explain productivity convergence patterns.
The paper proceeds as follows. Section 2 contains a brief review of the convergence
issue in the economic literature, in order to provide some theoretical background to our
empirical analysis and to introduce several concepts used throughout the paper. Section 3
describes the data used. In section 4 we analyse the development of cross-country differences
of energy- and labour-productivity levels within sectors over time. In section 5 we use a panel
data approach to test the proposition that sectoral growth rates of energy- and labour
productivity are inversely related to their initial levels of energy- and labour productivity,
indicating possible patterns of catching-up. In addition, we try to identify the country- and
2 Although Dollar and Wolff (1988, 1993) distinguish 28 sectors, they only present a labour-productivity
convergence indicator for a few years and did not perform a regression analysis to test for convergence patterns.