[NLA] RE: NLA Discussion Wages For Men and Women: New Evidence For a Growing Literacy Surplus

[John Comings]

I sent Tom's posting to John Tyler, who is a NCSALL researcher at 
Brown University, to get his comments.  The sections below that have 
a > in front of them are Tom's posting, and the paragraphs that start 
with *** are John's comments. These comments do not refute the 
findings, but they do point out that social science research findings 
should always be discussed with a focus on how the methodology might 
lead us astray.  I thought NLA listserve readers might find this 
interesting.

John Comings


>In a new report, Carnevale and Desrochers (2001) present new analyses
>that suggest that there is a growing surplus of literacy skills in the
>workforce of the United States. In one analysis, they present data
>showing that  from 1973 to 1998 the percentage of prime-age workers
>(both male and female) with some college/AA degree more than doubled,
>from 12 percent to 27 percent.  But from 1979 to 1998 earnings (in
>constant 1996 dollars) of prime-age male workers (30-59 years old)
>declined for those who failed to finish high school, those with a high
>school diploma, AND those with some college or an AA degree. For men
>with a bachelor’s degree or graduate degree, earnings rose from 1979 to
>1998.

*** It might be worth "reanalyzing" this data from a different perspective 
a la the way that Dick, Frank Levy, and I did on a similar question in a 
1995 Montly Labor Review article. The issue there was that there was an 
article that had been published in that journal stating that there was an 
oversupply of college graduates. The "evidence" was that supposedly more 
college grads were taking high school type jobs. We found that looking 
within age group (i.e., 25-34, 35-44, and 45-54) was important. Looking 
within age group may be important here, as well.


>In this case, as the percentage of men with some college or an AA degree
>increased, their wages decreased.

*** Would want to control for potential compositional changes here. For 
example, if more and more AA's were going to minority group members, and if 
they earned less than whites for other reasons, then you might get a 
spurious relationship between education and changes in mean earnings. Also, 
if more and more AA's are working part-time over time, then this 
would affect earnings levels. But neither story (if true) makes the 
case that there is an "oversupply" of AA's.


>But the same was not true for women.
>Across all education categories, women’s wages grew from 1979 to 1998.
>However, they started from low baselines and ended up at low levels of
>wages. On average, women with bachelor’s degrees in 1998 earned only
>about as much as men with high school diplomas. This suggests that women
>may be in lower paying jobs though they may possess education and
>literacy skills that far exceed the demands for education and skills in
>the jobs that they hold. That is, they may possess a literacy surplus.

*** Trying to draw conclusions about gender gaps in pay is notoriously 
tricky, because it usually requires assuming that everything you've 
included in the regression picks up all gender difference, so that if you 
then see a difference in mean wages, you conclude that it all has to do 
with the labor market (e.g., gender discrimination in the market). The 
trouble is that alternative stories can always be told. For example, if on 
average women need more flexibility in their work schedules because they 
are still (on average) the primary child care person in the family, then 
women may be choosing jobs that offer lower pay, but more flexibility, than 
do men with the same education level. I think everyone believes that there 
is a certain amount of gender discrimination in the market. The question 
is, when we see gender differences in pay, controlling for education level, 
then how much is due to discrimination and how much is due to unobserved 
gender differences? Bottom line, I would say that it would be hard to make 
a really convincing case that there is a literacy "surplus" for women.


>In analyses of data from the National Adult Literacy Survey of 1992, the
>authors show data indicating that in 1998 dollars and for workers aged
>16-64, women with high school diplomas and the highest literacy skills,
>NALS levels 4/5, earned less than men with high school diplomas but the
>lowest literacy skills, NALS level 1. Women with a bachelor’s degree and
>NALS level 3 literacy earned about the same as men with NALS level 3
>literacy skills with just a high school diploma.

*** Same concerns here as above...we are worried above the unobservable 
differences between men and women that may cause them to make different 
choices about jobs.


>Carnevale and Desrochers (2001) also present data showing that though
>the more education one has the better the average income, 83 percent of
>workers with associate degrees have earnings that fall within the same
>distributions as workers with bachelor’s degrees. This suggests that
>many workers with bachelor’s degrees possess literacy skills for which
>they are not being compensated in wages. That is, they may possess a
>surplus of literacy.

*** I guess this is one of the most problematic paragraphs. The first part 
just says that the earnings distribution of those with an AA and those with 
a bachelor's degree overlap, with the AA distribution shifted to the left. 
To go from here to the statement that this suggests a literacy surplus is a 
big stretch. Assume that there were, in fact, no literacy surplus 
whatsoever, and that earnings were tightly tied to education level. Would 
we expect to see an overlap in the earnings distribution of AA's and BA's? 
Of course we would, because earnings depend on lots of things other than 
education, and also because earnings are very noisy anyway. We would expect 
to see a fair amount of overlap in the AA-BA distribution even if we looked 
at one industry/occupation cell and controlled for age. That is, we would 
expect to see some AAs and some BAs in this industry/occupation cell 
earning the same amount. Why? One story that could explain this is that 
because of oversupply, BAs have to work in AA-type jobs. However, another 
story is that in this industry/occupation cell, the AAs and the BAs are, on 
average, working in different **jobs**, with the BA doing work that 
requires a BA level of education and the AA working in a job, in the same 
industry and occupation, that requires less education.


>Why do women with high school diplomas and with the highest literacy
>skills on the NALS earn less on average than men with high school
>diplomas but the lowest scores on the NALS?  Is it because women are
>working in jobs that just pay less regardless of one’s education and
>skill levels?

***Possible, but as I argue above, there are other plausible explanations.


>  If so, why are women working in these low paying jobs? Why
>aren’t men working in these jobs to the same extent as women?
>These data lead to questions as to the likely outcomes of educating more
>and more people to higher and higher levels of education and literacy
>Is it possible that as more and more people acquire higher education
>credentials and higher literacy skills wages will fall because employers
>have a larger pool of better qualified workers from which to recruit?

***Of course, this depends on what is happening to the relative demand for 
these workers. If the demand curve for more educated workers shifts out at 
the same rate as the supply curve, then we would expect no change in 
relative wages.


>Is that what happened from 1979 to 1998 as the percentages of men with
>some college or AA degrees increased, and wages for these men decreased
>(in constant 1996 dollars)? 
>It seems clear from data in this report and from earlier work that
>something other than literacy skills and education credentials is at
>work in determining wages in the workplaces of America. This would seem
>to demand a closer scrutiny as adult education and literacy development
>policies are wed more and more closely to workforce investment policies.

*** I agree.



----------------------------------------
John Comings                     Phone: 617.496.0516
NCSALL -- 106 Nichols House      Fax:   617.495.4811
7 Appian Way                     Email: john_comings at harvard.edu
Graduate School of Education     Web:   http://ncsall.gse.harvard.edu
Harvard University
Cambridge MA 02138