Opportunistic employment filtering

The press reaction to the Georgetown Center on Education and the Workforce report Help Wanted was muted (and one could even say the press ignored it after a day or two in mid-June), but the Pope Center's George Leef took a crack at the economic analysis and a crack at the policy implications more recently. Leef's superficial criticism of the labor-market analysis is close to criticism you'll find in the sociological literature, so when the report came out earlier in the summer, I read it, asked the center's staff for a copy of the supporting appendices, and then corresponded a few times with coauthor Nicole Smith. I just haven't had the time to blog about it before now, but it's important.

Two other bits of context: First, the Help Wanted report appeared the year after the publication of Claudia Goldin and Larry Katz's The Race between Education and Technology (2009), which comes to similar conclusions based on different analysis. The Goldin and Katz argument is a long-sweep-of-history piece, very different from the close-trends analysis of the Georgetown report. Closer to the Georgetown report but not as well-read as Goldin and Katz is the work of Michael Handel on "job content," such as a 2008 report available in PDF.

There's also last month's New York Times story on job training and the recession, together with a vague reference to a single evaluation study whose author or title never appears in the article. It's quite irritating (though perhaps not surprising) that the Times would be so sloppy, but there is a long line of job-training criticsm on both empirical and theoretical grounds, going back to Berg and Gorelick's Education and Jobs: The Great Training Robbery (1971). Most if not all skepticism of the human-capital argument focuses on the role of credentials separate from productivity gains one could attribute to attainment or training.

The gist of the Georgetown report's intellectual argument is that the Bureau of Labor Statistics' projection methods for future education requirements in the labor market underestimate those requirements because the BLS ignores within-occupation changes. So the report's authors conducted two different types of analyses in a fairly detailed way. The guts of the technical appendix is the detailed comparison of the projection method(s) against the past few decades of data, not only with regard to the overall U.S. labor market but disaggregated occupational categories. There's the usual question of how far forward one can project the past, but at the very least, it's a fascinating model of how labor-market patterns have changed in the past two decades with regard to education. If you do nothing else, skip the main report and head directly to page 34 of the technical appendix, where the attainment-by-occupational-category figures are. (The vertical scale in each figure is the percentage share of the labor-market fragment occupied by people with that level of educational attainment.) You will find some surprises in both the patterns and where the "noise" of the past two decades (changes in the data that the report's final model doesn't track well).

Because the occupational categories are fairly broad–"education, training and library" includes more specific jobs–the increase in associate-degree jobs for the category in Figure A8 (p. 41 in the technical appendix) is not an increase in associate-degree jobs teaching elementary and secondary schools or college (fairly large groups within the broader category). Without additional information, my guess is that the increase in associate-degree jobs in the category reflects changes in who works in child-care and their educational attainment (with child-care teachers being more likely today to have associates' degrees and less likely to be high school dropouts or graduates without additional education). (Yes, there's probably a small increase associated with the shift within teachers in private primary schools, but it's probably a very small contributor to the total change.)

The important question within an individual occupation is why the educational distribution changes. Die-hard human capital advocates would say it's predominantly a response to real job needs or at least a real productivity advantage of greater educational attainment. Die-hard critics would say it's predominantly the consequence of shifting attainment within the labor market overall and changed use of credentials by employers largely unrelated to productivity (i.e., employers use credentials to screen applicants because they can, not because the extra education of workers does much). Fortunately, not only is the world more complicated than either position, but we have some research exploring that complexity. (Michael Handel argues that the available data is limited, something he's trying to remedy.) One of my favorites in the area of education and young-adult employment is James Rosenbaum and Amy Binder's 1997 article, Do Employers Really Need More Educated Youth? ($), which explores why 51 employers they interviewed engaged in hiring practices that looked irrational from one or another simple perspective: hiring and keeping people with a range of educational backgrounds and paying them in ways that didn't make sense from either a human capital or credentialist perspective.

Rosenbaum and Binder find that there can be some fairly idiosyncratic (and interesting!) decisions based on an individual employer or manager's judgment of business strategy (e.g., hiring "overqualified" young adults and paying them well in hopes of training them to fit other jobs in ways that's a little more complicated than standard segmented labor market theories). How do we make sense of these idiosyncratic decisions? Maybe one way to do so is to think about the credentialist argument as one type of opportunistic employment filtering. In the credentialist view of the world, when the applicants include a lot of people with a range of educational backgrounds, you can make your hiring decision a little easier by setting minimum educational qualifications. But that's not the only way employers can take advantage of changing circumstances of the job market. Today, you'll find job-application kiosks in a lot of chain stores. Not only does that remove the lobbying that a job applicant can do when carrying a piece of paper to a manager, it also provides some filtering based on computer skills (even the basic one of maneuvering around a badly-designed online form).

Of course, this is a two-way street. In an economy with low unemployment, workers can demand higher wages. Or in a sector with limited job access, skilled workers can hop between jobs to suit preferences for work environment as well as wages. Individuals benefit when they stay ahead of the general labor-market curve in some way (including educational credentials but not entirely), if they can persuade employers that their package of skills and experience provides a unique advantage. Regarding our current circumstances, when there's a horrid recession and its flat-employment aftermath, you can expect employers to be patient with job searches, because the pain of going without a particular worker may be more than overbalanced by the likelihood of finding someone better when combined with the risk that you may not be able to afford to hire anyone.

Thus far, I am fairly sure, the passage above is not only obvious but also less sophisticated than the existing economic literature on labor markets. Or as economists who read this might say, so what else is new, yadda yadda? One thing: both bosses and workers have an incentive to use whatever educational skills exist in the labor market to do more. The language of "job needs" doesn't capture what happens when a secretary produces a report in half the time that a manager had expected, so the manager now knows what might be possible with someone who knows what they're doing. (Real-life example one could trace: how did word-processing documents shift from having tab-generated tables to tables formatted in a more efficient way?) Or what happens when a boss realizes that because everyone in a particular shop has a certain skill or educational level, she can add a little more to what the job involves.

What an historian brings to the debate is a sense of how sloppy real labor-market changes can be. Getting people to come to a factory "on time" required a technology to have a public time, a change in some part of a culture to make on time a normative social concept, and a very messy conflict over those norms. From one perspective, it looks like employers won that long-term battle… but again, that pressure is more general. Once you have a system of public time and on time, then you can have a concept of work week and 40-hour work week.

Eight days ago, I ordered replacements for the washing machine and dryer in our house, appliances that I know are probably 20 years old, and you don't need to know why it's a very good idea to find a replacement quickly. Here's what happened in the store: I came in with a model number of the washing machine I wanted to buy. The sales rep told me that they didn't have the item in stock but that they could order it and deliver it "fairly quickly." Through a combination of maneuvering through 20 screens on his point-of-sale device, flipping through a reference list of appliances, and finding a sheet with discount bar codes, he sold me the washing machine and dryer I wanted, applied the relevant discount for the (energy-efficient) washing machine, explained some options, applied my choices, did his part to schedule delivery, and took payment, all while casually chatting with me. As Mike Rose would point out, he's obviously mastered the cognitive requirements of the job.

I see in this interaction the application of subtle shifts in educational attainment as well as changing technology. Yes, the supply-chain and logistical arrangements of the business are in much better shape than anything would have been 15-20 years ago. And, yes, the point of sale (POS) software and systemwide connections are more intuitively responsive and faster than they would have been just a few years ago (or are in other businesses). But the fluidity with which this man finished the sale spoke volumes about his comfort with computerized inventory systems and ordering software. Looking over his shoulder, I thought, "I bet he's flying through this because he's worked on prior systems, and this is a breeze by comparison." The system was built on behalf of and to take advantage of people like him. This is what Goldin and Katz (among others) mean by "technology-skill complementarity."

Yes, there are POS systems that deskill jobs (picture-based cash registers at fast-food joints, for example). If technology allows large businesses to accommodate lower skills, it is part of a general repertoire of attempting to fine-tune the match of technology to either the skills of existing workers or the next generation a business intends to hire. And, again, that maneuvering is not one-sided, even in this economy. People take the time to learn coding in new software languages and as individuals make their cases to bosses to keep them on. And it's workers who can and must engage in certain initiatives to save businesses on the edge of survival (and I'm not thinking of concessions).

So how can we disentangle the opportunistic employment filtering that is related to education and "productivity" from the filtering that is just opportunistic? As an historian, I think you can do that best in retrospect: where was a change in minimum credentials followed by workflow or technology use that took advantage of those changed attainment levels? If your local fast-food joint required everyone to have a high school diploma, for example, how could we tell whether that was just raging credentialism? See what happens after the change in requirements. If the POS machines still are picture-based, that suggests credentialism. If the job changes in a different direction (I don't know — maybe the purchase of computerized equipment in the back that requires frequent attendance by someone–cue the Steve Martin routine about how the same stuff in one chain is used for "shakes," buns, and cardboard boxes), then it suggests the technology-skill complementarity that Goldin and Katz claim has been the staple of the American economic dynamic over the past 125 years.

Incidentally, what looks like "productivity" in a microeconomic perspective may not be what we need as a society. I'm sure that from the technology-skill complementarity perspective, all the math and physics folks going to Wall Street in the past 20 years is consistent with the larger story. After all, those highly-educated analysts were able to come up with a lot of analysis that their predecessors couldn't do: great productivity from that perspective. But it just didn't turn out to be all that good for the society.

Or maybe it's not that incidental at all. One of the strengths of the Goldin/Katz book is (somewhat obscured) contingency involved in growing skill requirements within their story. In the 19th century, they say (correctly), the development of factory systems reorganized work without increasing the skill level. So, as I tell my classes, shoe factories didn't mean that the shoes were made by machine. That changed around the turn of the 20th century with the development of true mechanization and interchangeable parts.

Goldin and Katz make an assumption that we should expect a continuation of the 20th century pattern of increasing skills on the job, but the underlying history is not so sanguine. If that were true, I'd be more optimistic about the future. But businesses do not have to improve their prospects by becoming more skill-intensive.

I'm getting ahead of myself. I need to write more about the Goldin and Katz, but for now, I'll just say that I have the same point to make about the Georgetown report: it stakes a reasonable claim to growing skills requirements in the past 20 years. Whether that continues in the near and medium-term future is less secure.

If you enjoyed this post, please consider subscribing to the RSS feed to have future articles delivered to your feed reader, and sign up for my irregular newsletter below!

5 responses to “Opportunistic employment filtering”

  1. Glen S. McGhee

    Thanks for posting the technical appendices. If there are any trends in the pages following 34 that stand out, it is the upward acceleration of credential inflation spiral. Unbelievable.

    As I mentioned previously, Germany is holding steady its level of college graduates, whereas France is ramping their rate sky-ward like we are. But the German economy is strong, whereas France is weakening. I know the underlying factors are more complex, but the comparison challenges the human capital argument. Comparative studies are badly needed to round out perspective.

    The other thing that Randall Collins would add is that, even in the case of the medical profession (or, especially so), most jobs are “learned” on-the-job. On-the-job training is the rule, even for higher echelon positions with high levels of expert knowledge.

    Now, this supports the human capital perspective, but only at the expense of formal higher education — the human capital investment occurs informally, still. Outside the sorting-institutions we have set up and support.

    BLS numbers for projected new jobs by occupation need to be taken with a grain of salt, or maybe with some laudanum. Local community college leaders take a yearly increase in 4 jobs and say this documents the need for a new program in XYZ, and then go on to argue for a 4 yr bachelor-lite degree as well. But the fact is, jobs are disappearing in Florida. Too bad that this doesn’t show up in the BLS numbers.

    Even Florida’s famous longitudinal survey results have not correlated major-related human capital gains with major-related occupations — and far, far too few are the school-to-job pathway studies that I know about (Steve McDonald, FSU is the exception).
    See for yourself.

  2. Glen S. McGhee


    Interesting counter-example. Hm. What about all the teachers the commit grievous harm against their students — how do the human capital theorists factor that in? It happens, sadly. Yeah, it happens.

  3. Glen S. McGhee

    Driver Ed story shows, I think, that we have lost our humor. This was an attempt to bring some back in.

  4. Glen S. McGhee

    Some vintage Ivar Berg, Randall Collins and David Labaree thrown in —