Explain This

Jaison R. Abel, Richard Deitz, and Yaqin Su write,

the broader V-shaped pattern in the underemployment rate over the past two decades is also consistent with new research arguing that there has been a reversal in the demand for cognitive skills since 2000. According to this research, businesses ramped up their hiring of college-educated workers in an effort to adapt to the technological changes occurring during the 1980s and 1990s. However, as the information technology revolution reached maturity, demand for cognitive skill fell accordingly. As a result, during the first decade of the 2000s, many college graduates were forced to move down the occupational hierarchy to take jobs typically performed by lower-skilled workers.

They refer to a paper by Beaudry, Green, and Sand.

This is an important observation. Some possible explanations:

1. The empirical finding is mistaken. If you do the analysis correctly, the demand for cognitive skills has not reversed. Perhaps jobs that are classified as not requiring a college degree in fact have become much more cognitively challenging. [note: I am not claiming that the empirical analysis is incorrect. I just want to include that as a possibility.]

2. Assume that the distribution of jobs in terms of skill requirements has remained approximately fixed. Assume, perhaps reasonably, that college selects students on the basis of cognitive ability, but it does nothing to change the distribution of skills or ability. As more students go to college, this means more low-ability students go to college, and they wind up in the same jobs that they would have obtained without going to college.

3. There has been an actual deterioration in the quality of college education, at least relative to the needs of employers. An individual who graduated in 1995 was more likely to have gained skills in writing and thinking than if that same individual had graduated in 2005.

15 thoughts on “Explain This

  1. 2 sounds right (though I wouldn’t say college does nothing, just not much).

    Our society just cant deal with the idea that cognitive ability is real, heritable, and explains a lot of variance in economic outcomes.

  2. But of course, the study’s implication may be stronger than that implied by reaching maturity being akin to simply the demobilization of the labor resources needed to iplement a ramp up. It could also be that reaching maturity is when technology is then able to begin replacing other labor resources.

  3. How about some comment or considerations on what factors influence the so-called “demand” for “cognitive skills.”

    What conditions “create” or “limit” the so-called “demand?”

    Can more people do stuff or themselves (and have the time to do so)?

    Take the example of bookkeeping software and the related levels of skill for its use and reductions of time inputs.

  4. Seems the simplest explanation is missing. It is correct. Cognitive skills are among the most easily automated and standardized so once done, the demand for them falls leaving the demand for more difficult to automate and standardize jobs like janitors.

    1 is mixed, information has probably increased but so has accessibility so is probably a wash
    2 is wrong as years of education has not increased and more are not going to college
    3 is not the quality but the utility, by the time you see schools advertising schooling in various specialties, they are already in decline, the programs being fed by supply of instructors unable to find employment in the field

  5. I worked for a software development company in a job that required cognitive skill, but not a college education. The job was unique to the industry and had to be learned, so the company would bounce back and forth between hiring CS majors and young guys with a demonstrated interest in computer stuff. A few of both types worked out well and many of both types did not work out at all. The key factors for both types were job fit, personality fit, and most of all, a desire to be good at the job. Over the years I was there, the company was gradually leaning more towards the young guys as they could get them to start much cheaper than the college grads.

  6. As anyone who has been to college will remember, college degrees are no measure of cognitive skills. All kinds of idiots get them. What college degrees do measure is conscientiousness and socioeconomic class.

    Only a small minority of majors demand cognitive skills. Almost all require math. Demand for these majors has gone up. Supply of these majors has gone down.

    It is the other majors that demonstrate conscientiousness but not cognitive skills that has had demand and salary decrease, as computers replace huge swaths of sales, inventory, management, and other white collar positions. The friendly, if median-IQ people that used to sell us insurance, phones, etc. are now much less employable.

  7. If you unpack “cognitive skills”, there are really two very different skill sets. There are those who create or engineer, and then there are operations that need people who can handle technical tasks.

    My guess is that we too many college grads targeting the creative/engineering segment, which doesn’t need numbers so much as elite talent, and too few people who can manage technical operations jobs.

  8. There was a lot of informal “automation” going on in the ’80s and ’90s. At least for the first 2/3rd of that period, a college degree was a proxy for computer literacy and a tendency toward not accepting the old ways. A lot of the small processes that are site or entity specific got automated. Much of that became incorporated into the 3rd party commercial products but it took a while. I’m sure a lot of what I did has been replaced, perhaps by commercial products, but I did do the small work that showed the integration path and utility.

    We tend to think of the big things that made a splash, but even those usually started as a guy/group who knew a process then figured out how to get the computer to do the boring part or make it easier to get to the gold in the data stream.

    Look at it this way. Steam engines were costly and only used in limited tasks. They usually required the designer to be on call. But they got smaller and cheaper. It wasn’t long until they were providing power across the site via rotating shafts. Then workers not in the mainline started throwing belt over the shaft automating tasks not perviously considered possible or feasible. Then things settled into operators with only a few making smaller innovations.

  9. 4. As the information revolution matured, it made a large supply (tens of millions) of workers (from India, the Phillipines, China, etc.) with above average (more than 1 std dev above?) cognitive skills (especially math and logical reasoning) available to and accessible by US employers at costs far below those of US-based workers of similar or lesser ability. US workers on the margin of cognitive ability are necessarily shifting to jobs that must be performed in-person and are difficult to automate (e.g. janitor, retail sales, physical therapist) or are protected from competition by government regulation. These same US workers cannot lower their cost to US employers because of federal, state and local labor taxes and mandates.

    Factor equalization explains everything in the labor market for the past 20 years.

  10. 1. The measure of “cognitive” is too coarse. (e.g. several posters above are correct.)

    2./3. One wonders if the pool of people that in fact learn a great deal in college (as opposed to earn a signaling credential) are actually still in high demand? So “college” as a proxy for “can do complicated difficult tasks” has broken down?

  11. A different more complicated thought that may or may not apply to the broad issue.

    In the late ’70s to say early ’90s, computer programming was something most people got competent at in college or some other formal course of study. The typical person one met on the street couldn’t write a spreadsheet macro let alone anything very difficult.

    By the late ’90s to say the present, many people, via the web and related social constructions, could learn to program well enough for most tasks. Some acquire a very advanced understanding.

    At the same time, the environment most work is done has matured greatly (sort of comparable from going from steam powered line shafts to per machine electric motors in factories.) And so for an ever larger fraction of the “programming task market” ever lower skill levels are required. (This has been a pretty continuous trend at all levels since the beginning of digital computing.)

    So most jobs are in some ways easier, and there are way more ways to acquire the necessary skills. In such a circumstance “comp sci degree” will be a less powerful indicator for the market as a whole.

    But that means that trying to measure “the market for high cognitive ability labor” by “demand for college degrees” will be dubious or even downright bogus instrument.

  12. ” There has been an actual deterioration in the quality of college education”

    There is anecdotal evidence to this effect (e.g. from employers’ organizations) and it fits with the grade inflation story, i.e. actual ability for a given grade has declined/is declining.

  13. There is a fourth possibility, or a variant on 3.

    “There has been an actual deterioration in the quality of college education, at least relative to the needs of employers. An individual who graduated in 1995 was more likely to have gained skills in writing and thinking than if that same individual had graduated in 2005.”

    It seems at least as likely that there has been a deterioration on average of all levels of education, K-12 at least as much as college, so that the feedstock for the college “factory” is subpar in the first place, so the factory is shipping subpar widgets no matter how good a factory it is. Further, there is plenty of anecdotal evidence that colleges are not as good as they were even as recently as ten years ago. The decline is across the educational board, K-12-College-Grad School. All of it.