The content on this page and other DCITA document archive pages is provided to assist research and may contain references to activities or policies that have no current application. See the full archive disclaimer.
5.8 Factor groups
In order to reduce collinearity problems, in some regressions explanatory variables were amalgamated into three composite factors: technological, education and other institutional. The makeup of these composite factors in terms of explanatory variables is given as follows:
Technological factor
- ICT investment to value-added
- ICT investment per hour
- capital intensity
- OECD labour productivity growth rates
Education factor
- share of university graduates in the workforce
- share of post-school qualified persons in the workforce
- change in share of university graduates
- change in share of post-school qualified
Other institutional factor
- decrease in the share of union members in the workforce
- decrease in the proportion of the workforce covered under award pay
- decrease in the number of days lost due to industrial disputes
- a dummy variable marking sectors that were subject to significant industry-specific competition reform measures
Each of the three composite factors consists of four explanatory variables, which tend to be positively correlated with each other (see tables E.1 and E.2).
A fairly simple approach was followed in constructing the composite variables. First, the arithmetic mean of each explanatory variable across 16 industries was calculated. Then the observation for each industry was divided by that mean. These yielded four sets of indices, which were added up and divided by four to yield the estimated composite factor sets. The single period calculation in respect to the technological factor is presented in table 5.12.
Tables F.1 and F.2 in the appendix show the single-period calculations for composite education and institutional variables. Table F.3 shows the two-period composite factor sets. The application of composite variables reduced the strong positive correlations between explanatory variables but did not eliminate them entirely (see tables F.4 and F.5). The correlation between technological and ‘other institutional' composite variables is high.
Table 5.12 Components of the technological composite factor
|
ICT to VA |
ICT per hour |
OECD prodty |
Capital intensity |
Composite factor | |
|
Electricity and gas supply |
1.64 |
3.21 |
1.43 |
3.86 |
2.54 |
|
Water supply and sewerage |
0.99 |
2.54 |
1.43 |
3.08 |
2.01 |
|
Construction |
0.40 |
0.17 |
0.18 |
0.21 |
0.24 |
|
Wholesale trade |
0.66 |
0.43 |
1.18 |
0.34 |
0.65 |
|
Food retailing |
0.78 |
0.22 |
1.18 |
0.20 |
0.59 |
|
Household good retailing |
0.62 |
0.23 |
1.18 |
0.19 |
0.56 |
|
M/V retailing/service |
0.47 |
0.13 |
1.18 |
0.14 |
0.48 |
|
Accommodation and restaurants |
0.57 |
0.20 |
-0.01 |
0.45 |
0.30 |
|
Road transport |
0.46 |
0.20 |
0.95 |
0.52 |
0.53 |
|
Air transport/travel |
1.09 |
0.91 |
0.95 |
1.09 |
1.01 |
|
Rail, water and other transport |
1.38 |
0.59 |
0.95 |
0.65 |
0.89 |
|
Storage and transport facilities |
1.18 |
1.01 |
0.95 |
1.42 |
1.14 |
|
Communications |
2.07 |
2.10 |
2.83 |
1.56 |
2.14 |
|
Finance |
1.45 |
1.21 |
0.96 |
0.56 |
1.04 |
|
Insurance |
1.50 |
2.30 |
0.80 |
1.34 |
1.48 |
|
Cultural and recreational |
0.76 |
0.57 |
-0.15 |
0.42 |
0.40 |
Source: Tables 5.1, 5.2, 5.3 and D.1.
Back to 5.7 Correlations | Table of contents | Forward to 5.9 Heavy and light ICT users