Figures 1-6 show the results from this simulation for each of the six economies for three years
as available in the late 1970s, late 1980s and the second half of the 1990s. Three major results
emerge. First, in each sector and country the number of jobs generated/required has been falling
over time. This reflects productivity growth (bearing in mind that, in the vertically integrated
industry, productivity growth reflects gains in all the supplying industries as well as within the
sector itself). Second, there is considerable heterogeneity across sectors, with the most
employment-intensive sectors generating twice, even three or four times, the number of jobs of
the least employment-friendly. Equivalently, the production of certain outputs is achieved at
three or four times the efficiency, in terms of economy-wide labour use, of the least efficient.3
Third when using the VIS approach a manufacturing/services divide in either productivity levels
or productivity gains is by no means evident. Some manufacturing industries generate high
numbers of jobs, as do some service industries; on the other hand some service industries
generate surprisingly few jobs. Spectacular productivity gains have been achieved in
manufacturing, notably in the production of electronic goods and medical equipment, while
some services, such as posts and communications, also show major gains. The laggards likewise
are not exclusively the preserve of the service sector. 4
See Figure 1 - Figure 6
To summarise the pattern of employment-intensities over time and between countries Table 1
gives the Spearman correlation coefficients for the sectoral rankings of the employment
generated by the unit injections of final demand; the upper panel shows the correlations within
each country over time, and the lower panel the correlations across countries in the latest
3 Implicitly we have computed for each industry a productivity index that relates output to the inputs directly and
indirectly involved in its production. Our computed productivity index (the employment effect of one unit of final
demand) is closely related to the deflated index of total factor productivity used in Baumol and Wolff (1984).
In terms of our example in Table A1, to produce one unit of final demand of X1 takes 1.59 units of X1, 0.35 units
of X2 and 0.46 units of X3. Replacing the diagonal FD matrix in the third panel by the identity matrix I, the matrix
multiplication B*I gives the associated output requirements. To obtain the corresponding labour input
requirements, these quantities are multiplied by (the inverse of) a value measure of productivity (7.77/110 = 0.07
labour units in industry 1, 3.60/110 = 0.03 units in industry 2, and 0.25 units in industry 3), giving a total of 0.35
labour units across the whole economy. Thus productivity, as the value of output per unit of factor input
employed, would be 1/0.35 for industry 1, and similarly 1/0.25 in industry 2 (0.01 employment units in industry 1,
0.12 units in industry 2, and 0.12 in industry 3).
4 The seeming outlier for the Netherlands in Figure 5 is the industry: Other Manufacturing and Recycling. This is in
part a residual category. Its share in gross output in 1977, 1986, 1997 was 0.2, 0.2, and 1.2% respectively, while its
share in employment was 3.2, 2.5 and 2.3%. These relatively large changes in small numbers cause the spike in
Figure 5.
10