Signaling and the College Bubble

From Bryan Caplan:

Many educators sooth their consciences by insisting that “I teach my students how to think, not what to think.” But this platitude goes against a hundred years of educational psychology. Education is very narrow; students learn the material you specifically teach them… if you’re lucky.

Other educators claim they’re teaching good work habits. But especially at the college level, this doesn’t pass the laugh test. How many jobs tolerate a 50% attendance rate – or let you skate by with twelve hours of work a week? School probably builds character relative to playing videogames. But it’s hard to see how school could build character relative to a full-time job in the Real World.

At this point, you may be thinking: If professors don’t teach a lot of job skills, don’t teach their students how to think, and don’t instill constructive work habits, why do employers so heavily reward educational success? The best answer comes straight out of the ivory tower itself. It’s called the signaling model of education – the subject of my book in progress, The Case Against Education.

According to the signaling model, employers reward educational success because of what it shows (”signals”) about the student. Good students tend to be smart, hard-working, and conformist – three crucial traits for almost any job. When a student excels in school, then, employers correctly infer that he’s likely to be a good worker. What precisely did he study? What did he learn how to do? Mere details. As long as you were a good student, employers surmise that you’ll quickly learn what you need to know on the job.

In the signaling story, what matters is how much education you have compared to competing workers. When education levels rise, employers respond with higher standards; when education levels fall, employers respond with lower standards. We’re on a treadmill. If voters took this idea seriously, my close friends and I could easily lose our jobs. As a professor, it is in my interest for the public to continue to believe in the magic of education: To imagine that the ivory tower transforms student lead into worker gold.

What makes the college bubble so problematic is that it is essentially inflationary. College degrees can be considered a form of currency in the labor market, wherein one purchases a salary with not only one’s labor but one’s college education as well. Obviously, this mechanism is not as direct as, say, buying milk at a grocery store, but the effect is similar.

The labor market, then, relies on college degrees to indicate a prospective employee’s fitness for the salary being offered. Certain types of degrees generally pay better than others, certain colleges’ degrees pay better than others, certain grade point averages are worth more than others, etc. Someone who receives an MBA from Harvard while maintaining a GPA of 4.0 will generally earn more than someone who receives an Associate’s degree from ITT Tech while maintaining a 2.0 GPA. This should make sense, as the quality of student varies by institution, degree, and grade, and there are ways to sort this. The college bubble, then, serves as a form of inflation because it distorts the signal that a college has in the labor market.

Basically, as is well known, the college bubble is the result of massive governmental interference in the post-secondary education market. The federal government offers direct subsidies of education costs (e.g. the Pell Grant), and also makes college loans a very enticing offer to lenders by guaranteeing the loans. With direct subsidies and easy credit, prospective students have a very strong incentive to go to college. Furthermore, with this much money on the line, colleges have a very strong incentive to accept more students.

The effects of this bubble, as noted before, are seen primarily in signal distortion. This occurs because employers now have a larger labor from which to select workers. This generally seems like a good thing, since employers can now offer lower wages, but this is not always the case because some potential workers are perhaps not as well-qualified for their position as others. The problem with using college degrees as a qualification is that, at this point, there isn’t enough data to sort the good from the bad. When there were a limited number of college-educated labor candidates, the quality was considerably better since colleges had an incentive to maintain quality control. This is no longer the case because the federal government is paying colleges, indirectly, to simply pass out degrees to young adults with no regards for their qualification.

Thus, the lower wages that have resulted from the increased pool of labor applicants can be thought of as a risk premium. Because there are more college-educated people in the labor supply coupled with increased variance of abilities without there being an increase in the sharpness of the signal generally associated with a college education, and because American labor is tightly regulated with regards to discrimination (particularly as it pertains to firing employees), there is consequently more risk associated with hiring someone because the chance that person a company hires turns out to be a bust, as it were, is considerably greater. Given the costs associated with firing incompetent workers, particularly if they are in a union or minority, employers have an increased incentive to mitigate that risk by offering lower wages.

As such, the most problematic aspect of the college bubble is the consequences that come with signal distortion. Because the supply of college educated labor has increased with a matching increase in demand for said labor, and because a college degree isn’t nuanced enough as a single, there will be an increase in the number of people who are overpaid and an increase in the number of people who are underpaid. This happens because the signal sent by a college degree is roughly the same for everyone who has one.*

Some people will be underpaid because their aptitude is such that they would ordinarily deserve more pay than they are currently receiving but, because it is now more difficult to tell who has what levels of aptitude, they must take a pay cut. The reverse is true for those who are overpaid. Basically, the inflation in the number of students undermines meritocracy, thereby distorting the pay scale. Thus, the current bubble has introduced not only distortion, but market failure on a large scale.

The irony of the current college bubble is that its existence is largely predicated on the belief that a college education makes one more intelligent. This claim is laughable on its face because it does not begin to account for the self-selection bias inherent in this sort of activity. Do students learn because they go to college or do they go to college because they like to learn? This is a crucial question because if the answer is the latter, then it seems likely that those who do go to college would become just as knowledgeable if they lived in a library for four years.

At any rate, the college bubble has had the nasty effect of giving diplomas to those who have no desire to learn, and have undermined the meritocracy that once was a college education, thereby depriving those who are truly above average from an income that would properly reflect this fact. This, then, is the lamentable effect of the college bubble: The attempt to make everyone equal in education has only led to a diminution of standards. We are all idiots now.

* Obviously, a Harvard diploma is still more valuable than an ITT Tech diploma. However, if Harvard’s business school doubles the number of graduates, year over year, the value of a Harvard degree will decline assuming that there is not a corresponding increase in demand for Harvard grads.

A Tale of Two Recessions

So I wind up often getting into an agument that boils down to me disagreeing with a common belief that the recent recession for Pittsburgh is a slightly milder, but still very similar experience to the recession of the early 1980’s.  Nationally it was actually two recessions officially (January to July 1980 followed by the longer period July 1981 to November 1982), though I think most would agree that the two recessions really were one big recession for the Pittsburgh region.

So as the national news parses the good unemployment numbers yesterday, the nabob version focuses on the drop in the labor force participation the numbers seem to show.  It lead me to making the graph below.  I took the labor force trends in Pittsburgh and made a comparable index from a period early in the two recessions.   So from January 1982 and from January 2008 forward, the graph shows what happened to the national and Pittsburgh region labor forces over the subsequent 4 years.  The graph shows the change from those baseline months.  Lot’s to parse from it, but just take a look:

The extreme differences for Pittsburgh in the two recession (1980s vs recently) go way beyond what that graph shows. That decline in labor force in Pittsburgh actually masks the decline in the male labor force a bit as women entered the workforce in record numbers to replace the men who were out of work. The drop in the labor force clearly correlate with the net migration that spiked from the region and the drop in population it caused. The folks who were leaving both the regional labor force and leaving the region period were predominantly younger workers who were the folks most capable of adapting and changing to new jobs in new industries.  Those who stayed were far more likely to be older workers who had been displaced from the occupations they had had for decades and for many would never find new employment.  Today we know that in recent years we have seen the first net migration into the Pittsburgh region in decades and changes in migration patterns almost entirely reflect changing migration patterns of young workers.   It is folks in their 20’s who dominate migration flows with rates of migration dropping as folks get into their 30’s, 40’s and older.. until there is a bit of a spike in early retirement years. So if net migration for the Pittsburgh region flipped from net negative to net positive just a few years ago, it has to reflect changes in the flow of younger workers into the region.

So, just as the incredibly high unemployment rates of Pittsburgh in the early 1980’s persisted even though so many workers were leaving the region which would have taken a lot of potentially unemployed folks out of the regional labor force…  masking how bad the employment situation really was; today the regional unemployment rates are being impacted by more workers, or those seeking work, flowing into the region and potentially making local labor force metrics look worse than they appear otherwise.

A Broken Market

The pharmaceutical giant, Pfizer, watched its main source of revenue and profits, Lipitor, lose its patent protection this week, and now faces competition from generic equivalents. In 2010 Lipitor was the second highest selling prescription drug with $5.2 billion in sales in the U.S. alone. (source: Drugs.com). Now, in the next year, prices of the generic drug, Atorvastatin, should drop dramatically. The Lipitor saga gives us an opportunity to see market forces in action, but it also points out the problems when insurance coverage is involved.

Lipitor BrandLipitor Brand

Like most first world countries, the United States uses the patent system to encourage research and development. If a pharmaceutical company can develop a new drug, they can maintain a government approved monopoly on the sale of that drug for up to 17 years. Monopolies drive higher prices, which helps the inventor, Pfizer in this case, recoup their research costs, and return a handsome income to their shareholders. Once the patent runs out, other manufacturers can apply to produce the drug. This increased competition then quickly drives down prices. So far, this is a classic example of market forces at work.

Pfizer has been planning for this day for a number of years, and with annual sales figures like those in 2010, this is vital to the company’s fortunes. The company has triggered a number of legal and regulatory efforts to delay the arrival of generic equivalents. For a compilation of news articles on Lipitor, see this page in The New York Times.

Two particular strategies twist prescription drug coverage in favor of the brand name. Many prescription drug plans have incentives to encourage patients and their physicians to use generic drugs. Often this is done with a lower co-payment on the part of the patient. The lower co-payment provides an incentive for the patient to accept a generic equivalent, and the insurance plan saves money by paying the lower, generic price. Pfizer (and other drug companies facing similar out-of-patent challenges) is trying to subvert this incentive. Here’s a hypothetical example.

These figures are illustrative – made up – but make the point.

Typical Brand vs. Generic Comparison for a Drug Plan

Brand:  Patient Copay: $30 – Total Cost of Drug: $200 – Insurance Pays: $170

Generic: Patient Copay: $10 – Total Cost of Drug: $50 – Insurance Pays: $40

Now Pharmaceutical Company Offers a Copay Discount
(Pfizer discounts its price of the brand drug to cover reduced copay)

Discount Brand: Patient Copay: $8 – Total Cost of Drug: $178 – Insurance Pays $170

With this discount arrangement the patient is happy, the drug store doesn’t lose any money, but the insurance company still pays the larger cost. This puts upward pressure on insurance premiums.

Another strategy – Pfizer offers a significant discount on the price of brand name Lipitor to pharmacy chains as long as they agree to not provide generic equivalents. The chains save money, and can pass some of that on to patients, but the insurance plans that pay for the drugs don’t enjoy any savings.

Is this legal?  The second, discounting strategy with pharmacies, smells a lot like restraint of trade/anti-trust concerns to me. The earlier example, offering a discount on copays, seems legal. Are either of these good social policies? Not a chance.

These creative approaches illustrate one of the problems that insurance introduces into a market. In healthcare, patients have enough discretion that they can alter their buying behavior, based on prices they face. Yet the patients don’t see or feel the full price of their purchase decision. In a regular market the patient balances the benefit of the purchase against the price, and makes a good decision on allocating resources. That good decision helps society. With insurance the patient sees only a small fraction of the total price, and may make a decision that is not socially optimal. This breakdown in market forces is one of the challenges our healthcare reform goals face. Ideally we would like patients to be full partners in the decisions made about their care. Insurance blunts that participation.

Economic Events on December 5, 2011

At 10:00 AM Eastern time, the Factory Orders report for October will be released.  The consensus is that there was a decrease of 0.3% in orders from the previous month.

Also at 10:00 AM Eastern time, the ISM non-manufacturing index for November will be released.  The consensus estimate is that increased by 1 point to a value of 53.9, and will continue to signal economic growth as it remains above the mid-point of 50.