about 40 percent over the last quarter century (women currently earn 15-20 percent less than the
equivalent male).
Since education can also be expected to be a tool for sorting of persons by native ability,
the regressions are likely to overstate the pure return to education. Thus, I have chosen to use a
slightly lower estimate of its contribution in constructing an index of labor quality. The specific
formulation, given earlier in equation (5), assumes that each year of schooling raises individuals’
productivity by a constant 7 percent.14 The resulting index of labor quality is shown in figure 7,
together with that from the TDRI studies. The improvement in education increases the effective
labor input by an average of 0.6 percent per year over the 1977-2003 period. The assumption of
a 12 percent rate of return would raise the improvement factor to one percent per year. This is in
addition to a 2.6 percent annual rate of growth in total employment over the 1977-2004 period.
In contrast, the two TDRI studies obtained a 2.8 percent annual rate of quality augmentation over
the 1977-96 period. The gains in educational attainment in the major sectors follow a similar
pattern. Services employ workers with the highest average level of education and with the
largest gains since 1977. Agricultural workers have the least education and the smallest gains.
Thus, the sector differences in educational attainment have grown over the past 25 years.