Technological Changes
The early evidence on the importance of
technological changes a source of the
shifts in the relative demand for
different types of labor during the 1980s came
from case studies. The Bureau
of labor Statistics conducted several case studies
of the effects of changes
in production processes in particular industries
(Mark, 1987). In an industry
that experienced a significant change in
technology, the usual pattern was a
dramatic reduction in the employment of
production workers with an increase
or no change in the number of skilled
workers in that industry. More recently
there have been several econometric
analyses of the effects of variables like
the (appropriately lagged) rate of
investment in computers and/ or other
forms of "information capital" and the
ratio of expenditures on research and
development to sales on changes in the
skill composition of industries (for
examples, Berman, Bound and Griliches,
1994). The results of these
studies are consistent with those of the case
studies and the hypotheses that
the recent technological change has shifted the
relative demand for skilled
labor to the right. Changes in production techniques
have widened across the
country quickly, especially the multinational firms.
Thus, if
technological change is an important determinant of relative demand
shifts,
one would expect to observe patterns in other industrialized
countries
similar to those in the United States. Some of the recent studies
report results
for a variety of old industrialized (OECD) countries that are
indeed consistent
with the U.S. results (Collechia and Papaconstantinou,
1996; Machin, Ryan and
Van Reenan, 1996). These countries vary a great
deal with respect to changes in
their situations with respect to trade, labor
market institutions (like the
importance of trade unions), and unemployment.
Obviously, the relative demand
for skilled labor in each of them is rising
rapidly. In my view, it is the
fairly strong evidence in favor of the wage
inequality and technological change
story. A factor that is often cited as
the specific issue of the post-1980 is
the widespread adoption of computer
technology throughout the economy. As
mentioned above, the rate of skilled
labor has tended to be greatest in those
industries with the highest rate of
investment in computers. There is also
evidence that workers who use
computers on the job have, other things constant,
higher earnings than those
who do not (Krueger, 1993). In my view, it is
probably too early to determine
that how much of the technological change dues
to computers affect wage
inequality, but we will know more about the answer to
this question in 20
years. wwwwwwwwwwwwwwwwwww Over time, the difference between
the rate of
economic growth and the rate of growth of the quantity of labor
input is
usually attributed to technological change. It is also roughly equal to
the
growth of the average real wage rate in the economy thus providing the
link
that changes in technology or productivity are closely linked over time
to
growth in real wages. Since the mid-1970s, however, the average real wage
rate
in the United States has grown at a very low rate. Returning to the last
two
columns in Table 2, column d shows the slowdown in the growth of real
wages that
started around the 1970s. By subtracting column e from column d,
it can be seen
that the average real wage rate has been essentially stagnant
after adjustment
for the increase in average wages expected because of the
upward shift in the
educational distribution. This fact is troubling for any
explanation of the rise
in income inequality that focuses on skill-biased
technological change. After
all, if there was so much technological change,
why didn’t it cause high
average real wage increases, rather than the
historically unprecedented
stagnation of wages? An answer to this question is
that the effect of
technological change on the average real wage rate depends
on which kind of
change occurs. Technological change that is neutral with
respect to labor
skills—that increases the efficiency of both skilled and
unskilled labor by
the same proportion—will result (after the adjustment of
the aggregate capital
stock has occurred) in increases in the average real
wage equal to the rate at
which efficiency increases. A bout of intensive
skill-biased technological
change—resulting in skilled workers becoming more
efficient in jobs that they
previously performed—means that as skilled
workers become more productive,
their wages rise. This also leads to a rise
in the wages of unskilled workers,
since they are complementary in
production. But as long as elasticity of
substitution between different types
of labor, o , is greater than one, then
employers will not be able fully to
substitute unskilled for the higher-wage
skilled labor. Since the wages of
both types of workers rise, the average real
wage rate rises; however,
because of imperfect substitutability, the
skilled-unskilled relative wage
rate will also rise slightly. However, extensive
skill-biased technological
change—changes in production processes such that
skilled workers are
profitably employed in some jobs that unskilled workers used
to do—is a
different situation. In this case, the wage of skilled workers
rises, but the
wage of unskilled workers fall. The average wage in the economy
rises due to
extensive skill-biased technological change, but it does not rise
very much,
for the increased efficiency associated with skilled workers
performing their
new jobs more efficiently than unskilled workers used to is at
least
partially offset by the decrease in employment in the initially skilled
jobs
and by the lower productivity of unskilled workers in the jobs that
remain
for them. A particular innovation, like assembly lines, railroads,
or
telecommunications, may represent different forms of technological change
at
different stages of its development. The introduction of personal
computers, for
example, may have increased the efficiency of skilled workers
in their initial
jobs, but it is clear from the various empirical studies
that it caused an
increase in the fraction of jobs normally performed by
skilled workers. A lot of
research and development expenditure over the past
20 years probably went into
figuring out ways that various repetitive
functions could be computerized. This
should have increased the relative
demand for skilled labor, but it would not
necessarily have increased output
per person-hour very much. Will Inequality
Continue to Rise? The sharp
rise in earning inequality and the virtual
stagnation of the average real
wage in the United States since the early 1970s
are matters of intense public
concern. The weight of evidence suggests that the
principle cause is an
increase in the rate of extensive skill-biased
technological change. But will
this trend continue? It is possible that as
computer technology becomes
better able to perform more sophisticated tasks, the
effects of technology
will change direction so that unskilled workers can become
more efficient in
jobs that were formerly done by today’s skilled workers. For
example,
insurance agents are reportedly concerned about the implications for
their
future earnings of the new ability of customers to purchase their
policies
through the Internet. This would represent the reverse of extensive
skill-biased
technological change; high school graduates could perform most
of the necessary
labor functions with computerized purchases. If practices of
this sort become
fairly widespread in the future, the relative demand for
skilled labor will not
shift to the right as rapidly as rapidly as it has in
the recent past and might
theoretically even shift back to the left. But
while one can offer possible
examples in particular industries of the
technological bias turning in favor of
unskilled workers, it would be unwise
to stake too much on a movement of this
sort reversing the trend toward
greater inequality. After all, the demand
function for high-skilled labor has
been shifting to the right fairly
consistently for a long time. Based on the
post-World War II experience in the
United States, it is highly unlikely
that it will not continue to do so. The
only realistic question is whether
the pace of change will be as great as it has
been during the past 25 years.
It is difficult to think of policies that would
halt this technological
shift—for example, imposing a large tax on
computers?—and even if such
policies could be put in place, they should not
be. Thus, public policies to
"do something" about earnings inequality have
focused on other proposals,
like a large increase in the minimum wage,
government training programs,
import restrictions, tax rate reductions for rich
old people, and so forth.
While this paper is not the place for a review of
these choices, it is fair
to say that most of the available options either
present potentially
undesirable side effects or may work only very slowly or not
at all, or all
of the above. Will the supply side of the economy adjust to
reduce the amount
of inequality? There has been a fairly steady rightward shift
in the
educational distribution in the United States over the past
half-century.
The trends shown in Table 1 imply that the relative supply
of college/high
school equivalent labor has grown at an average annual rate
of 2.9 percent.
Further, one might suppose that the huge increase in the
private rate of return
to investment in a college education associated with
the rise in the
college/high school relative would cause a large increase in
the relative supply
of college graduates, which could eventually reduce the
wage premium of
high-skilled labor. This is the story implicit in Figure 1,
in which the
vertical short-run relative supply function shifts over time the
same amount as
the relative demand function so that, in the long run, Returns
to R. If this
story were correct, the rise in inequality would be a
relatively temporary
event—that is, lasting perhaps another decade or
two—which might not require
policy intervention. There is some supply
adjustment in the pipeline, judging
from the time series data on the
educational attainment of young adults and of
the proportion of college-age
persons enrolled in school. For example, the
proportion of 20-to 21-year-old
males enrolled in school rose from 31.6 percent
in 1979 to 42.6 percent in
1993, while the value of R increased by 17.7 percent.
The implied
relative supply elasticity is 1.7, which is in line with previous
estimates
of this parameter (Freeman, 1986). This value, however, is not nearly
large
enough to yield the 3 to 5 percent rates of growth of
relative
supply—depending on whether the next 15 years is like the 1950-1980
or the
post-1980 period—necessary to keep up with the likely technologically
induced
shifts in future demand for high-skilled labor. In addition, much of
the past
high rates of increase in the relative supply of educated labor were
due to the
replacement of old people with a low average level of schooling by
young persons
with a high average level. Beginning around 2010, the average
level of education
of labor force exiters will be quite high, reflection in
part the large boost in
the number of students who attended college in the
1960s, probably in part
because it was a way to defer military service. For
the relative supply of
educated labor to grow rapidly after 2010, rates of
college attendance will have
to be extremely large to avoid a new bout of
wage inequality. It would seem,
therefore, that a long-term commitment to
increasing greatly the fraction of
individuals who go to college is the
appropriate public policy response to the
phenomenon of increasing
inequality. But this is more easily written than done
effectively.