5  Changes in the value of human capital, 1986–2001

In this section we consider more explicitly the income generating value of qualifications, which we refer to as the “value” of human capital, and how this has changed over the period. We focus on changes in the population-shares and relative incomes associated with the different qualification groups. In particular, given there has been an increase in the proportion of skilled workers in the population (as measured by the index of skill intensity discussed above), we are interested in the contribution of such a change in population shares to the value of human capital. A key factor for this question is whether there was a change in relative incomes across qualifications, and how this reinforced or counteracted the increase in the skilled population share.

A positive relative income contribution would occur if average incomes increased most for groups that expanded their employment share. If the change in employment shares reflected solely an increase in supply for highly qualified workers, we would expect that average incomes would decline for them, as marginal productivity is reduced, and that average incomes would rise for other workers, as their reduced share induces a rise in marginal productivity. In this case there would be a negative contribution from relative income changes. Conversely, if there is an increase in demand only for highly qualified workers, this would lead to an increase in their relative income, and a positive relative income effect. The greater the supply response in the form of an increasing share of workers having high qualifications, the smaller would be the relative income effect.

The analysis in this section takes the form of a simple decomposition of the change in the value of human capital.[22] The measure of the value of human capital used here is the average annual income across the relevant population (alternatively employed workers, labour force, or the full population).[23] This measure of human capital is closely related to the index of skill intensity used previously, and the two measures coincide in the base year 1986.

Specifically, we express the value of human capital in year-t () as the qualification-share weighted sum of qualification average incomes:

(4)    

where s_qt denotes the share of the population with qualification-q in year-t, and y_qt is the average income of qualification-q in year-t. Given expression (4), we can then express the change in the value of human capital from year-0 (eg, 1986) to year-1 (eg, 2001) in terms of changes in the qualification shares in the population and changes in economic returns to qualifications (measured by average incomes) as follows:

(5)    

The first term in this decomposition, , measures the contribution of changes in the qualification skill-mix in the population, holding qualification average incomes constant at their initial level. This is simply the change in the skill intensity measure described above. The second term, , measures the contribution of changes in the average incomes returns to different qualification groups, holding constant the qualification shares at their initial level. Finally, the third term in expression (5), , measures the contribution of the interaction of the changes in both qualification shares and incomes. For example, if relative incomes increased for qualifications whose share of the population also grew, this interaction effect will be positive.

Table 7 presents the results of this decomposition for employed workers, the labour force, and the full population aged 20-59. Panel A of the table pertains to the full period, while the subsequent panels present results for the three intercensal subperiods. We discuss the full period changes here first, and return to the subperiod effects subsequently. The first row of panel A shows that, between 1986 and 2001, the value of the human capital of the employed New Zealand workforce and the population both rose by 20%, while the value of human capital of the labour force increased 18%. The skill-intensity changes described previously imply that the change in the skill-mix over the period has favoured those qualifications that were more highly rewarded in 1986, and thus contributes positively to the observed increase in the annual incomes. Row 2 documents that this upskilling in the population accounts for between 13% (for the full population) and 18% (for employed workers) of the increase in annual incomes over the period.

The third row of Table 7 presents the contribution to the overall change in annual income that is due to changes in qualification-specific average incomes at constant qualification shares. This component accounts for most of the increase, ranging from 72% for the employed population to 78% for the full population. This factor confounds two effects: a change in average incomes shared by all qualifications, and a change in relative incomes that may cause the incomes of some qualification groups to rise more rapidly than others. For example, a broad increase in productivity across qualifications is likely to result in an increase in the incomes of all qualification groups and, hence, increase the value of human capital. On the other hand, a change in the relative productivity of different skill groups with no broad increase in productivity (eg, via a skill-biased technological change) would likely result in a change in relative incomes of different qualification and, depending on the qualification shares, may increase or decrease the value of human capital. We will return to this issue shortly.

The final row of Table 7 presents the contributions of the interaction effects between changing qualification shares and incomes. For each population group this interaction effects accounts for about 9% of the observed increase in annual incomes. These positive combined contributions of income and relative employment effects confirms that there were some income gains associated with the upskilling, although these may reflect average income gains common across skill groups rather than skill-specific relative income gains.

The complexity of the relative employment and income effects is not evident from the summary measures generated from the decomposition in Table 7. In order to understand more of what drives these aggregate changes, we next delve into the changes that occurred at the qualification level. First, we examine the qualification specific contributions to the component factors of the changes in human capital and, second, we examine the relationship between changes in qualification-specific relative incomes and shares over the period.

Table 8 presents the qualification specific contributions to the decompositions described in Table 7 of the changes in human capital between 1986 and 2001. The first column in the table (for each of the population groups) presents the qualification-contributions to qualification share component of the decomposition in Table 7, the final row sums these into the overall contribution of this component (from Table 7). Obviously the sign of a qualification’s share change determines whether that qualification contributes positively or negatively, while the magnitudes of the share change and the average income affect the magnitude of its contribution. With these points in mind, it is apparent that the shares associated with the “post-school” and “no qualifications” groups fell and this made large negative contributions to the overall share effect on human capital change. On the other hand, all the other qualification groups increased their shares over the period, with the “bachelor’s degree” group making a particularly large contribution to the overall qualification-share effect. The results are very similar across the employed, labour force, and full populations.

The second column in Table 8 shows the corresponding results for the component due to the changes in qualification specific average incomes over the period. In this case, the sign of the change in a qualification’s average income determines the sign of its contribution, while the magnitude is determined by the size of the income change and the qualification share. That each of the qualification contributions is positive implies that the average incomes of all qualification groups increased over this period: that is, there was a general increase in incomes across qualifications. Furthermore, the “post-school” and “no qualifications” groups made the largest contribution here. As we will see soon, this is largely due to these groups having by far the largest shares of the population – each approximately one-third of the various populations. However, the relative contribution of the “no qualifications” group (18-21%, depending on population) is substantially smaller than its initial share (33-36%: see Table 1) implies that the increase in incomes for this group was relatively less than the average increase across qualifications. On the other hand, the relative contributions of the University-level and “post-school” groups exceed their initial population shares, implying their relative incomes increased.

The third column shows the qualification contributions to the interaction effects. Given that average incomes increased for all qualification groups, the sign of a qualification’s share change determines the sign of its contribution to this interaction effect, while the magnitude its affected by the magnitudes of both its share and average income change. Thus the signs in this column are the same as those in column 1; the relative magnitudes also largely mirror those in column 1.

We now consider the changes in the value of human capital over each of the three intercensal periods. The results for the aggregate decomposition of human capital analysis are presented in panels B to D of Table 7, while the qualification group contributions to the changes in the value of human capital in the intercensal periods, shown in Appendix Tables A6a to A6c, also help shed light on the aggregate changes.

First, comparing the results across the panels in Table 7 suggests the aggregate 1986-2001 changes tends to disguise quite a lot of variation, both across time and across alternative populations defined by labour force status. For example, the change in average incomes of employed workers increased most strongly between 1986 and 1991 (8.1%), was slowest over 1991-96 (3.2%), and increased strongly again from 1996 to 2001 (7.2%). In contrast, the growth in average incomes of labour force participants was around 5% in each of the first two periods, and was 7.2% between 1996 and 2001, while the growth in average incomes across the population rose steadily over the period from 3.8% from 1986-91 to 9.3% from 1996-2001. The relative temporal differences in magnitudes for the employed workers and the broader populations are consistent with the changing nature of the employed and labour force populations over the business cycle. That is, the unemployed and not in the labour force constituted larger shares of the population in 1991, which lowered the income growth of the labour force and wider population relative to that of employed workers between 1986 and 1991, and raised it during the subsequent periods.

Second, the relative contributions of changes in the qualification mix and incomes also varied substantially across the subperiods and populations, although the relatively small changes in the human capital during any particular period make the individual numbers somewhat unreliable. For instance, while the change in skill-shares contributed positively to the change in average incomes of the employed population between 1986 and 1991, this effect was negative for the broader populations. These differences probably reflect differential employment selection effects over the business cycle, causing a relative increase in high-qualifications’ shares and, consequently, decrease in the low skill shares of employed workers compared to the population.[24]

Third, although we find that the changes in incomes generally account for the majority of the overall change in average incomes over the period, the increase in skills between 1986 and 1991 contribute relatively strongly to the change in incomes over that period – eg, for the employed workers the changing qualification mix accounted for 32% of the increase in incomes and 47-67% across the labour force and whole population. During the subsequent period, 1991-1996, we find that the interaction effects between changing skill shares and changing average incomes was negative, implying that, on average, incomes fell for qualifications whose shares increased over this period and vice-versa. Appendix Table A6b shows this occurred for all qualifications except the “higher degree”, “6^th form”, “school certificate” groups. The “bachelor degree” qualification group had a negative interaction effect for employed workers and the labour force, but a positive interaction for the whole population. One interpretation of this finding is that increases in the supply of skilled workers outpaced demand causing their incomes to fall.

The final analysis we report on involves the relationships between the changes in relative incomes and the shares of different qualifications over the 4 census years. Figure 5 presents a graphical summary of the movements in incomes (relative to the “no qualifications” group) and shares for each qualification group over the period, adjusted for changes in age composition[25]. For the “no qualifications” group we also show (as the dashed line) the relative changes in income over time (relative to its 1986 level), in order to give a sense of the absolute change in incomes over the period. Horizontal movements measure changes in qualification shares in the population, while vertical movements measure changes in incomes. On such a graph, increases in relative demand generate movements in a North-East direction, assuming that the aggregate labour supply schedule is upward sloping. Similarly, assuming downward sloping demand, an increase in relative supply generates a South-East shift.

First, the dashed line shows a small drop in average incomes of this group occurred between 1986 and 1991, followed by modest increases to 1996 and 2001. The average incomes of each of the other qualification groups increased between 1986 and 1991, and over the full period between 1986 and 2001, both relative to the “no qualifications” group and in absolute terms.

Second, it is apparent from Figure 5 that both University-level and school-level qualifications’ shares have generally increased over the period, at the expense of “other post-school” and “no qualifications” shares.[26] However, the increase in incomes of these groups, relative to the “no qualifications” group, was not monotonic over each 5 year interval of the period: in fact, only the average incomes of the “school certificate” group increased in each 5 year period.

The observed movements for the university-level groups are consistent with a net increase in relative demand between 1986 and 1991, followed by a subsequent net increase in relative supply between 1991 and 1996, and roughly equal increases in demand and supply between 1996 and 2001. The patterns of change for the school-level qualifications are somewhat similar, bearing in mind the caveat of the 1986-91 change associated with the apparent rise in the “post-school” share: the changes are roughly consistent with a steady increase in supply of these qualifications over that period which, with the exception of the 1991-96 period, was generally matched by an increase in demand.

On the other hand, employment movements dominate the changes for both the “other post-school qualifications” and “no qualifications” groups. The relative incomes of these groups changed very little, but the groups’ shares fell steadily (except for the 1986-91 change for the “other post-school” group, mentioned earlier) from 30 and 36%, respectively, in 1986 to 22 and 21% in 2001.

The corresponding figures for employed workers and for the labour force, presented in Appendix Figures A2a and A2b, are qualitatively similar to those discussed here for the full population. The real incomes of employed workers with no qualifications actually increased, and was essentially static for this qualification group in the labour force, between 1986 and 1991 suggesting possible income-based selection effects on the composition of these respective groups, in comparison to the overall population, over the business cycle.