Fallacy of regression towards the mean.4 In other words, β-convergence is a necessary but not
a sufficient condition for σ-convergence. In this paper we will explore both patterns of σ-
convergence and β-convergence. Moreover, we test whether convergence is conditional or
unconditional and which are the factors explaining (cross-country differences in) labour- and
energy productivity growth.
3. DATA
The analysis presented in this paper is based on a newly constructed database that merges
energy data from the Energy Balances as they are published by the International Energy
Agency (IEA) and economic data from the International Sectoral Database (ISDB) and the
Structural Analysis Database (STAN), both published by the OECD. The main idea behind
the construction of this database is to establish a link between economic and energy data at a
detailed sectoral level. This results in the sector classification as described in Table 1.
(Table 1, Page 27)
The database covers the period 1970-1997 and includes the following countries: Australia
(AUS), Belgium (BEL), Canada (CAN), Denmark (DNK), Finland (FIN), France (FRA),
West-Germany (WGR), Italy (ITA), Japan (JPN), the Netherlands (NLD), Norway (NOR),
Sweden (SWE), United Kingdom (GBR) and the United States (USA).
We measure energy productivity by gross value added per unit of final energy
consumption and labour productivity by gross value added per worker (in full time
equivalents). Value added is the net economic output of a sector, measured by the price
differential between the price of output and the cost of input and comprises compensation to
employees, operating surplus, the consumption of fixed capital and the excess of indirect
taxes over subsidies (OECD 1998). Following the IEA, energy use is defined as final energy
consumption in kilo tonnes of oil equivalence (ktoe),5 with sectoral data excluding
transformation losses. Total employment is measured in the full-time equivalent number of
persons, including self-employed.
4 See Bernard and Durlauf (1996) and Durlauf and Quah (1999) for further discussion of empirical
methodological issues of convergence tests.
5 Hence, we do not analyse explicitly the impact of changes in fuel mix on overall energy-efficiency
improvements.