Monday, February 13, 2017

Economies in reverse

How can economies forget? How is it that once we have learned to do something better, that knowledge can be lost and economies move backward? How can productivity decline? Viewing productivity as knowledge, it would seem almost impossible for it to do so -- and real business cycle theory was often derided on that point. Yet middle ages eurpoeans lost the recipe for concrete, and time after time we have seen economies get worse. How can our own productivity be growing so slowly overall when so much we see around us is progressing so fast?

Scott Alexander at Slate Star Codex has an intriguing blog post that illuminates these questions (HT marginal revolution). I'll offer my thoughts on the answers at the end.

Scott starts with education:

Inputs triple, output unchanged. Productivity dropped to a third of its previous level.

Scott offers remarkable economic clarity on the issue:
"Which would you prefer? Sending your child to a 2016 school? Or sending your child to a 1975 school, and getting a check for $5,000 every year?

I’m proposing that choice because as far as I can tell that is the stakes here. 2016 schools have whatever tiny test score advantage they have over 1975 schools, and cost $5000/year more, inflation adjusted. That $5000 comes out of the pocket of somebody – either taxpayers, or other people who could be helped by government programs.
...College is even worse. Inflation-adjusted cost of a university education was something like $2000/year in 1980. Now it’s closer to $20,000/year.... Do you think that modern colleges provide $18,000/year greater value than colleges did in your parents’ day? Would you rather graduate from a modern college, or graduate from a college more like the one your parents went to, plus get a check for $72,000?  (or, more realistically, have $72,000 less in student loans to pay off)"
Health care is similarly bloated, though a more complex case.
The cost of health care has about quintupled since 1970. ... The average 1960 worker spent ten days’ worth of their yearly paycheck on health insurance; the average modern worker spends sixty days’ worth of it, a sixth of their entire earnings. 
Unlike schooling, health care is unquestionably better now. Scott notices that lifespan doesn't go up as much as we might have hoped, and other countries get the same lifespan with much less cost. Tell that to someone with an advanced cancer, curable with modern drugs and not with 1970 drugs. Still, it's a good example to keep in mind, as it's pretty clear health care is delivering a technologically more advanced product with a huge decrease in organizational efficiency.

Infrastructure, today's cause célèbre is more telling,
"The first New York City subway opened around 1900. ...That looks like it’s about the inflation-adjusted equivalent of $100 million/kilometer today... In contrast, Vox notes [JC: This is an excellent article worth a blog post on its own] that a new New York subway line being opened this year costs about $2.2 billion per kilometer, suggesting a cost increase of twenty times – although I’m very uncertain about this estimate.
...The same Vox article notes that Paris, Berlin, and Copenhagen subways cost about $250 million per kilometer, almost 90% less. Yet even those European subways are overpriced compared to Korea, where a kilometer of subway in Seoul costs $40 million/km (another Korean subway project cost $80 million/km). This is a difference of 50x between Seoul and New York for apparently comparable services. It suggests that the 1900s New York estimate above may have been roughly accurate if their efficiency was roughly in line with that of modern Europe and Korea."
I have seen similar numbers for high speed trains -- ours cost multiples of France's, let alone China's.

I find this one particularly telling, because we're building 19th century technology, with 21st century tools -- huge boring machines that dramatically cut costs. And other countries still know how to do it for costs orders of magnitude lower than ours.

Similarly, housing. bottom line
"Or, once again, just ask yourself: do you think most poor and middle class people would rather:

1. Rent a modern house/apartment

2. Rent the sort of house/apartment their parents had, for half the cost"
Housing is a little different I think, because much of the cost rise is the value of land, so supply restrictions are clearly at work.

More useful anectdotes, on whether this is real or just a figment of statistics.
The last time I talked about this problem, someone mentioned they’re running a private school which does just as well as public schools but costs only $3000/student/year, a fourth of the usual rate. Marginal Revolution notes that India has a private health system that delivers the same quality of care as its public system for a quarter of the cost. Whenever the same drug is provided by the official US health system and some kind of grey market supplement sort of thing, the grey market supplement costs between a fifth and a tenth as much; for example, Google’s first hit for Deplin®, official prescription L-methylfolate, costs $175 for a month’s supply; unregulated L-methylfolate supplement delivers the same dose for about $30. And this isn’t even mentioning things like the $1 bag of saline that costs $700 at hospitals. 
Where is the money going? It's not, despite what you may think, going to higher salaries:

Scott has similar evidence for college professors, doctors, nurses and so on. What about fancy salaries you hear about?
...colleges are doing everything they can to switch from tenured professors to adjuncts, who complain of being overworked and abused while making about the same amount as a Starbucks barista.
It's also not going to profits, or CEO salaries. Those have not risen by the orders of magnitude necessary to explain the cost disease.
This can’t be pure price-gouging, since corporate profits haven’t increased nearly enough to be where all the money is going.
My thoughts below.

Scott's elegant summary:
So, to summarize: in the past fifty years, education costs have doubled, college costs have dectupled, health insurance costs have dectupled, subway costs have at least dectupled, and housing costs have increased by about fifty percent. US health care costs about four times as much as equivalent health care in other First World countries; US subways cost about eight times as much as equivalent subways in other First World countries.

And this is especially strange because we expect that improving technology and globalization ought to cut costs. In 1983, the first mobile phone cost $4,000 – about $10,000 in today’s dollars. It was also a gigantic piece of crap. Today you can get a much better phone for $100. This is the right and proper way of the universe. It’s why we fund scientists, and pay businesspeople the big bucks.

But things like college and health care have still had their prices dectuple. Patients can now schedule their appointments online; doctors can send prescriptions through the fax, pharmacies can keep track of medication histories on centralized computer systems that interface with the cloud, nurses get automatic reminders when they’re giving two drugs with a potential interaction, insurance companies accept payment through credit cards – and all of this costs ten times as much as it did in the days of punch cards and secretaries who did calculations by hand.

It’s actually even worse than this, because we take so many opportunities to save money that were unavailable in past generations. Underpaid foreign nurses immigrate to America and work for a song. Doctors’ notes are sent to India overnight where they’re transcribed by sweatshop-style labor for pennies an hour. Medical equipment gets manufactured in goodness-only-knows which obscure Third World country. And it still costs ten times as much as when this was all made in the USA – and that back when minimum wages were proportionally higher than today.

And it’s actually even worse than this. A lot of these services have decreased in quality, presumably as an attempt to cut costs even further. Doctors used to make house calls; even when I was young in the ’80s my father would still go to the houses of difficult patients who were too sick to come to his office. This study notes that for women who give birth in the hospital, “the standard length of stay was 8 to 14 days in the 1950s but declined to less than 2 days in the mid-1990s”. The doctors I talk to say this isn’t because modern women are healthier, it’s because they kick them out as soon as it’s safe to free up beds for the next person. Historic records of hospital care generally describe leisurely convalescence periods and making sure somebody felt absolutely well before letting them go; this seems bizarre to anyone who has participated in a modern hospital, where the mantra is to kick people out as soon as they’re “stable” ie not in acute crisis.

If we had to provide the same quality of service as we did in 1960, and without the gains from modern technology and globalization, who even knows how many times more health care would cost? Fifty times more? A hundred times more?
And the same is true for colleges and houses and subways and so on.
Scott points out that many of our intractable political debates -- paying for college, health care, housing, and transportation, are made intractable by this bloat:
What's happening?

I think Scott's post is exceptionally good because it points out the enormous size of the problem. It's just not salient to point to productivity numbers that grow a few percentage points higher or lower. When you add it up over decades to see that while some things have gotten ten times better, other things are ten times more expensive than they should be really strikes home.

Scott tries on a list of candidate explanations and doesn't really find any. He comes closest with regulation, but correctly points out that formal regulatory requirements, though getting a lot worse, don't add up to the huge size of this cost disease.

So, what is really happening? I think Scott nearly gets there. Things cost 10 times as much, 10 times more than they used to and 10 times more than in other countries. It's not going to wages. It's not going to profits. So where is it going?

The unavoidable answer: The number of people it takes to produce these goods is skyrocketing. Labor productivity -- quality adjusted output per number of people involved in the entire process -- declined by a factor of 10 in these areas. It pretty much has to be that: if the money is not going to profits, to to each employee, it must be going to the number of employees.

How can that happen? Our machines are better than ever, as Scott points out. Well, we (and especially we economists) pay too much attention to snazzy gadgets. Productivity depends on organizations not just on gadgets. Southwest figured out how to turn an airplane around in 20 minutes, and it still takes United an hour.

Contrariwise, I think we know where the extra people are. The ratio of teachers to students hasn't gone down a lot -- but the ratio of administrators to students has shot up. Most large public school systems spend more than half their budget on administrators. Similarly, class sizes at most colleges and universities haven't changed that much -- but administrative staff have exploded. There are 2.5 people handling insurance claims for every doctor. Construction sites have always had a lot of people standing around for every one actually working the machine. But now for every person operating the machine there is an army of planners, regulators, lawyers, administrative staff, consultants and so on. (I welcome pointers to good graphs and numbers on this sort of thing.)

So, my bottom line: administrative bloat.

Well, how does bloat come about? Regulations and law are, as Scott mentions, part of the problem. These are all areas either run by the government or with large government involvement. But the real key is, I think lack of competition. These are above all areas with not much competition. In turn, however, they are not by a long shot "natural monopolies" or failure of some free market. The main effect of our regulatory and legal system is not so much to directly raise costs, as it is to lessen competition (that is often its purpose). The lack of competition leads to the cost disease.

Though textbooks teach that monopoly leads to profits, it doesn't "The best of all monopoly profits is a quiet life" said Hicks.  Everywhere we see businesses protected from competition, especially highly regulated businesses, we see the cost disease spreading. And it spreads largely by forcing companies to hire loads of useless people.

Yes, technical regress can happen. Productivity depends as much on the functioning of large organizations, and the overall legal and regulatory system in which they operate, as it does on gadgets. We can indeed "forget" how those work. Like our ancestors peer at the buildings, aqueducts, dams, roads, and bridges put up by our ancestors, whether Roman or American, and wonder just how they did it.


  1. John,

    And your solution is what exactly - something like this?

    "How would you like to work at Hoover for a year, focusing on research with NO TEACHING OR OTHER RESPONSIBILITIES, and soaking up the intellectual climate of Hoover and Stanford?"

    Where does paid research fall into your productivity graph in terms of improved test scores?

    1. But Stanford and Hoover are private not for profit - so technically, that money came from some donor who cares a lot about where it goes(I say that completely tongue in cheek).

  2. I think people are much better off now than they were 40 years ago, so this is really about relative costs of different goods. Lets name three areas that the government is highly involved with subsidizing (education, healthcare, infrastructure) and there is no mystery whatsoever.

    1. But in France, or Korea, or China, each of those systems Cochrane mentioned are run extremely cheaper and more efficient. In each of those countries, they all have far more governmental power and involvement compared to the US.

  3. I think it might be an excellent econometric project (a thesis?) analyzing the line-item costs of a bridge or rail project in the US and one in e.g. France. This is in principle understandable.

    This has been done for medical costs, for example. It is end-of-life costs that drive a lot of the difference. This could be entirely due to factors like culture (e.g. Americans potentially being much less comfortable with the concept of death than other cultures), or Americans just spend more on end-of-life simply because we can afford it. Whatever the reason for it, at least the particular problem is identified in medicine. I don't think we know where the problem lies for infrastructure.

  4. You point out that housing costs are closely connected to land costs. There must be a way to assess their impact of land prices on the other industries you mention, such as education. Most forms of education are not economizing on land, though we do have important innovations like MOOCs. And we can easily see why that is, because universities typically own their buildings and don't pay rent. Likewise for schools and hospitals. There is no reason for innovators to reduce our use of land in crowded locations if its cost simply doesn't enter into business decisions.

  5. Monopoly profits and the quiet life? Maybe, but…..

    Scott is struggling to understand the change in the cost. Have his poster-child industries—education, healthcare and infrastructure—really become that much less competitive during the period in which “THE MOST IMPORTANT THINGS COST 10 TIMES AS MUCH AS THEY USED TO”?

    There is no way we can say that these industries are competitive. I have a lot more choices in the brand of beer I drink than in where my kid will go to school. And since I have really great health insurance, I care more about the price of beer than I do about the price of medicine. But haven’t things been like that for a long time? More competition in those sectors would be great, but there's some really nasty complementaries between market structure, regulation and ...well...something.

  6. Hey Grumpy.

    "The unavoidable answer: The number of people it takes to produce these goods is skyrocketing. Labor productivity -- number of people per quality adjusted output -- declined by a factor of 10 in these areas. It pretty much has to be that: if the money is not going to profits, to to each employee, it must be going to the number of employees."

    1. I get that you are saying the number of people it takes is skyrocketing, and that labor productivity declined. But you also say the "number of people per quality adjusted output -- declined" and that just doesn't sit right.

    2. If the number of people per quality adjusted output increased by a factor of whatever, then why isn't labor share going up?

    1. I think we have uncovered, in these comments, the possibility that the mysterious increased spending is on stuff like classroom equipment, supplies, cafeterias, fancier buildings, and so on.

    2. Non-profit enterprises do have this tricky problem of how to distribute rents. Many have commented elsewhere on the questionable spending of elite universities on fancy amenities. But as Prof. Cochrane points out in the above quote, hiring unnecessary people is the quickest way to to spend your budget.

  7. Great post.

    Worth noting is that military outlays have ballooned also, even possibly worse.

    In WWII, the Brits conceived and had into battle the excellent de Havilland Mosquito bomber, all within a year, despite wartime labor and steel shortages. The pilots loved the plane so much they used as a fighter too.

    Today the F-35 has been in development for more than 20 years, will cost $1.5 trillion over life of plane, is no longer stealthy due to changes in radar, but compromised performance metrics to be stealthy. Drones are cheaper and better.

    Another disturbing stat to the number of hours in a hanger a warplane spends for every hour in the sky.

    The US spends $1 trillion a year on "national security" (DoD, DHS, black budget, VA, debt service) but we are told we have to "rebuild" the military

    How can we spend $16 trillion on national security over the past two administrations and have a military that need "rebuilding"?

    1. It's a new enemy and we have not adapted our military funding stream to it. George W and the neocons went about inviting folks to be our enemy ("if they are not with us they are against us" (but, by the way, the Bible says the opposite)). The folks who accepted the invitation (like ISIS) don't drive tanks and such. But our military, and the industrial momentum of making stuff for it, is adapted for blowing up tanks and such.
      Hence the need to build new military thinking. And maybe new equipment for it. Unfortunately.

  8. How about this for an explanation: the numbers are totally wrong in the first place.

    Not going to discuss public education, as the reasons why that is increasing in price and quality not so, on average, are more complex and have to do with demographics in my opinion (and no one wants to touch the demographic aspect here).

    I'll just talk about college costs: the figures presented are false. They are wrong. Where is this figure of "$20,000" in costs for college coming from? There is virtually no state university in the US that even has that much as a sticker price for in-state tuition.

    NET price of 4-year college is around $6-7,000k/year on average in the US. So lets get the figures right first before attempting to figure out why we can't figure out why things cost more.

    Second, the price of a college education is depended on the value it creates. The more value a college degree can earn in the market, the more expensive it will be. And while bachelor degrees have not increased in terms of salary potential very much (but they have increased), two things have happened:

    1) They have increased a lot more compared to the next best alternative, hence making it a more attractive investment, and hence a more expensive one.

    2) The returns to master degrees and professional degrees have exploded through the roof over the last few decades. Implying that graduate degree programs are growing much faster, getting more "expensive" as they earn greater returns, and more money is put into graduate degrees by universities (and the corresponding research focus). Which means greater salaries for research professors, and more reliance on adjuncts etc to do the much less valuable duty of teaching undergrads (and of course a healthy over-supply of PhDs, creating lots of surplus cheap labor as adjuncts)

    No, the PRICE of something does not represent increased inputs. A UNIT of input is not the PRICE of that unit. Lets also clear that up.

  9. As for public education, the answer is obviously that costs of inputs are NOT inputs, and certainly not all the inputs, into a test score outcome.

    Parents are inputs, hours worked at home on HW are input etc. I think you get the picture.

    Fast forward from 1975 to 2017...and now 30-40% of the kids in public schools are foreign born, mostly from third world countries, or minorities of one sort or another, and mostly in family situations where parent input or HW or other non-school inputs are much lower or non-existent.

    But of course, we're not supposed to talk about what happens to public schools when you import the entire third world into your city. No no!

    There are parts of my city I dare not venture into because not a single person there speaks english and I get assaulted by tamale vendors on every stoplight. I can only imagine how the public schools in those neighborhoods are.

    Or to put it in another way: this ain't the same cities, school districts or students from 1975. It's like you're comparing 2 different firms or countries in terms of productivity, not the same one over time.

    Our cities ARE the third world. So it's not surprising that the same factor inputs are less productive now than in the past (although I also think, we're not looking at the whole inputs here at all in the first place).

  10. Every sector you cite is massively involved with government. Health care is totally entangled with government. Education is almost all government. Housing costs are driven by local government land use regulation and the hell on earth that is "environmental" regulation.

    What you are witnessing is the sclerosis and dementia of the Peoples Democratic Republic of the United States of America. It may or may not outlive me, but my children and grandchildren will live through the collapse. I do not envy them. The next regime is not bound to be very nice.

    1. How does your thesis stack up against the fact that countries like France (hardly a libertarian paradise) seem to have much less of a problem with cost disease? It seems to me than any argument for cost disease relying on this concept of "simply too much government" crashes right into this problem.

    2. I think it has been fairly well established, in these comments, that the school cost example is mainly a matter of changing requirements. The chart, after all, does come from a treatise excoriating the teachers' union. And that the civil engineering examples are largely a matter of differing circumstances.
      Healthcare in USA is dominated by insurance business. Insurance is a terrible business model for healthcare. Incentives for minimising cost exist only if a long term view is taken. Insurance, because business horizon is the next quarter, cannot foresee cost reductions well into the future that can result from preventive care. Whereas government run basic healthcare (e.g. France, UK, NZ, Japan, etc.) does have the ability to take account of long term consequences.

  11. With respect to: "Most large public school systems spend more than half their budget on administrators." - you're drawing a long bow here.

    I'll use CPS as an example, but you'll find this story is reasonably typical.

    Chicago public schools spent $3.87bn on instruction in FY16, with 25,615 employees, and $0.32bn on administrative support with 705 employees. These numbers are from the FY16 CAFR.

    From memory this is pretty typical for a US school district. There is often a reasonably large staff classified as "non-instructional" - often about 30% of staff. Of these many are pretty clearly servicing students, e.g. pupil support (things like guidance services, hall monitors, social work) - 4,415 employees for CPS, facilities support - 1,427 for CPS, and catering/food services - 2,721 for CPS.

    All this depends on the accounting treatment of course - e.g. how do you roll up teacher pensions into instruction services? But I think you would struggle to find a large public school system in the US with more than 50% of the staff and expense in administration and other support services not directly interacting with pupils.

  12. JC, spot on! The dynamic effects of regulation are often missed. Basically this is happening in all of these industries:


  13. And I forgot the punchline. Even if we moved from zero non-teaching staff in 1970 to the approx ratio today of 2 teaching: 1 non-teaching - which must be an exaggeration because there were still bus drivers, cooks, hall monitors and janitors in schools in 1970 - that can't account for a 200% increase in expense in that first cato graph.

  14. Some simple rules I usually follow on the internet: if you're getting information from Marginal Revolution or Vox, you're probably doing it wrong.

    I'm the anonymous from above who says the numbers are wrong in the first place. But I'll add some more detail on that with the other examples Scott brings:

    1) Infrastructure. Everyone's favorite target. This one is easy enough to "debunk" and point out how the numbers are...monstrously...wrong.

    Yes the new NYC subway line costs about $2 billion per mile. Yes, subway lines in some other countries cost much less per mile. Here's the deal: they are WIDELY different miles.

    The NYC line in question is, in fact, about 2 miles long. Oh, and it's actually a station, underneath another station, in very deep rock, and it's actually 4 (FOUR) subway tunnels, not 1.

    The subway lines in other countries used for comparative purposes are in actually trenches dug from the surface, and then filled back in once the subway line is build. So these are absolutely not comparable things, even if they are "subways".

    2) Healthcare. This one is obviously more complicated but it helps to disaggregate and not rely on "averages". On "average", the US spends about the same amount per person as most other "developed" countries...up until about the age of 50-55. At that point, the cost skyrocket in the US compared to other "developed" countries. So the cost disadvantage of the US comes at particular point in time in a person's life...i.e. it has to do with quality of life for the elderly.

    Lets be honest and realize that these are not issues that can be analyzed by looking at "averages" without having ANY understanding of the fundamental inputs or outputs of a system we're analyzing, as is being done in all these blogs you cite.

  15. One additional comment. Adjusting for inflation. Economists, wrongly, use just general CPI or some similar proxy as a an all encompassing deflator for everything.

    That is clearly wrong. Certain things deflate at different rates for very many reasons. Cars change in value differently then books, for example.

    That is why the Fed published about 500 different deflators appropriate for different classes of things. There is no reason to assume CPI is the correct or appropriate deflator to apply to education, to healthcare, to infrastructure or anything of the sort.

    After all, when we look at firm capital investment for example as inputs to figure out firm productivity, we use a deflator appropriate for capital investments. Not CPI!

    You can get widely different figures depending on the deflator you use.

  16. Very interesting post.

    Maybe an additional factor driving this decline in productivity a.k.a. "we need more people to produce the same/comparable goods and services" is the high and low "ability" divide among skilled workers .

    Let me expand on that: 50 years ago there was a hard limit to what a brilliant engineer or doctor could achieve, while today the same engineer/doctor would be able to become a specialised consultant, for instance, leaving the standard job to some less able (but still officially skilled) person. Same goes for administration: the quality in administrative staff may have gone down because those who were more than able to do the job became consultants or found a more lucrative career altogether.

    While organisations are now able to draw on expertise not previously available to them, this mechanism also leaves the day to day running to "worse" and "worse" workers. The lower productivity comes from both the added costs due to experts/top workers, and the larger number of people it now takes to do ordinary tasks.

  17. The 40-year increase in the costs associated with these government run/government controlled industries seems to coincide with the increasing skills premium. Could it be that the government uses increasing competition restrictions in the industries it controls as a form of welfare to compensate “useless people”? I could see how attractive administrative bloat would be to politicians who are looking for a redistribution mechanism. Or maybe it goes the other way – that the government’s creation of all these cushy administrative jobs exacerbates the skills premium by drawing in otherwise talented people. Do you see a connection between this cost phenomenon in government controlled industries and the increasing skills premium?

  18. Doc at the Radar StationFebruary 14, 2017 at 7:14 AM

    I wonder if the ease with which we can electronically document anything has a lot to do with this phenomenon. We humans have a thing for organizing and categorizing stuff beyond what we likely need to. Maybe we have 'enabling' technologies that are pandering to our rather obsessive and relatively unproductive natures.

  19. The CATO number does not mention inflation adjusting, but the teacher salaries are in 2011 dollars. I don't believe those are comparable.

  20. Oops, delete this and my last comment. I see the 2012 CPI adjustment now under data sources.

  21. Isn't the administrative bloat (as well as regulatory problems) testable by looking at the cost/regulatory structure of the other countries sited? Seems simple. Also, wouldn't we want to look at the productivity of the other countries compared to the US to see if they are experiencing the same thing? Testing this hypothesis seems really simple.

  22. Pretty clear that bankers and public unions wish to destroy the middle class...and they are doing a heck of job! Outside bankers outrageous bonuses, tech billionaires and 50 year old teachers and cops retiring on $100k is bleak!

  23. There is a serious problem with this. The output of education is not a test score, it is an educated workforce that is put to productive use elsewhere. It's productivity must be measured in terms of the increases in productivity in those other places.

    Perhaps you could make a go of your argument if you could value the test score in terms of its true productivity. Otherwise, the argument is botched from the beginning.

    1. I agree with the premise, but I doubt our schools these days are teaching in a way thats any different than it was in the past. Whether its 1970 or 2016, a math teacher is still teaching math. With the extra 5k in costs, you would think that increase would show up in the test scores at least a little.

    2. The referenced articles claim that education is exhibiting "Baumol's cost disease." One of the articles directly links to Wikipedia which defines that creature as "a rise of salaries in jobs that have experienced no increase of labor productivity, in response to rising salaries in other jobs that have experienced the labor productivity growth." The claim that education has not experienced productivity growth is spurious. Test scores are not a metric for the productivity of education. Period.

      If you want to make this "cost disease" argument - you have to show that the students put out by schools are no more valuable to businesses than they used to be. I don't think you can do that. The point is that you can't just look at costs - you have to look at the value of the thing produced - and test scores aren't it.

  24. I think one has to be careful on the issue of education. The percentage of high school students who continue from high school to college has doubled during that time, most likely by including students who are further down the distribution of cognitive skills. That could explain why the increase in spending is not associated with better outcomes. Put differently, we have been moving to the right along the production function as increased spending has been shifting the function up. If the marginal product diminishes fast enough (because we are teaching more students who are less capable), average productivity may decline despite the increased spending.

    Regarding schools, the percent of non-English speaking families in the US has increased, which presents additional challenges. For example, the educational outcomes of Hispanic students are worse than average educational, and the percent of students who are Hispanic has doubled since 1980. Again, that could negate any benefits from increased spending, especially if weaker students have a negative spill-over effect on others.

    Finally, it is also possible that students spend less time on educational activities at home as a result so the increased spending did raise productivity which materialized however not in better outcomes but in more leisure (with same outcomes).

    Not saying that increase in administrative costs is not at play, only that one has to be very careful about controlling for other relevant factors.

    1. I agree with all of that. It probably explains why the magnitudes for college are much higher than those for high school. Still, if we focus just on high school - how much of that additional 5k is coming from non native speakers? I suspect its a second order effect at best to the bloated administrators.

    2. Constantine makes a good point.

      The % of college students who require remedial classes and special assistance has skyrocketed over the last few decades. We're pulling in students who are substantially less qualified to handle college level coursework as the students in 1970.

  25. Great post as usual! Let me focus on education.

    I admit that administrative bloat is due to lack of competition, though not entirely. For some part, I think educational bloat is because of the increase in wage inequality. The (rough) logic is as follows:

    "More education leads to higher wages." => "Parents pay more (than they received when kids) for their kids." => "(Kids:) My parents paid a lot for me so that I would earn higher wages. So, I do the same for my kids." => (repeat this process)...over time.

    I think this process is one of the underlying causes of rising cost of education.

    Data on wages by education level here:

    Chad Jones's recent slides seems to be related to this post:

  26. John, interesting post as usual. It reminded me of another entertaining article here:

  27. I am not satisfied with these answers. Some of the cases might be difficult, but certainly it is possible to find a public high school with good records on the budget, etc since 1970. We should be tell exactly how the cost structure has changed over time. We should not accept speculation.

  28. We shouldn't accept speculation about "admin bloat" or any other potential explanation when we have the data. Most high schools that have been around since 1970 must have reasonable records of the changing budget (and number of students). The answer should be clear from historical data.

    1. It's hard to be honest on test scores and other educational achievement issues these days. Id you're have to tackle the elephant in the room: demographics. When you import the third world into your country, you need to spend more on their "education", and your scores will get progressively worse on average compared to a different cohort with very different demographics.

      This is why both simply using averages is silly and pointless, but also why you can't really have an "honest" conversation.

      So other excuses have to be found, like "administrative bloat" or other politically acceptable explanations.

      The explanation is pretty obvious to anyone who wants to be "honest": the HS I went to was in the 1970s mostly comprised of Italian and Polish third generation immigrants. Today that same HS is comprised mostly of recent arrivals from Guatemala and Honduras.

      What a mystery!

  29. Two notes:
    In regards to K-12 spending (2014 constant dollars):
    2002: $10,455
    2014: $11,009

    In regards to spending in public higher education institutions:
    Total educational revenue per FTE (Constant Adjusted Dollars):
    1990: $11,583
    2015: $12,907

    Honestly, these numbers just don't seem that bad.

  30. "Unlike schooling, health care is unquestionably better now"
    Maybe schooling is not better now but it is certainly delivering a different product and is serving a changed need. The education delivered by a 1975 school, for a 1975 workforce, was significantly different that that needed for a 2016 workforce. Therefore using a 1975 test to evaluate the effectiveness of 2016 education is a bit silly. Sure it proves that the students of both periods got adequately trained in basic reading comprehension and minimal mathematics. But what of all the other stuff they have to learn? Both then and now? What if the extra stuff of 1975 was much cheaper than the extra stuff of 2016?
    And what if, perhaps, the underlying cause of greater cost is simply the availability of more pocket money to spend? Thus it gets spent on progressively less cost-beneficial stuff. This, I would assert, is a chief factor in the reduced cost/benefit ratio of healthcare spending. The extra cost of a faster car is way beyond proportional to the extra speed. But if you have that money to spend then why not? But then if it becomes that nobody's interested in faster cars then the recipe, and equipment for making them, will be lost. Who now has the means of reading an 8" floppy disk? Or an 8-track tape? It is, in some ways, easy to forget. But forgetting about burning coal (a move driven, in the short term, by hydraulic fracturing resulting in cheap natural gas) is causing a fair bit of angst.

    1. This is a very good point. Students may be learning the same quantity of "stuff" but qualitatively much different stuff. They have never seen a floppy disk but know how to store memory on the cloud, and those skills are more valuable but also require greater investment from schools on internet lines, email accounts, etc.

  31. There are a lot of product/services that are extremely cheaper and of much better quality in other developed nations in Europe and Asia than in the US: internet, TV, telephone data, air travel, private medicine, office supplies and many more instances. It is administative bloat, government interference and crony capitalism/cartels, all working together to deliver a raw deal to the American people

    1. Yes indeed.
      "extremely cheaper" ... include dentistry in Mexico.
      "government interference and crony capitalism/cartels" ... "conservatives", "the left", and sane business would do well to unite against the capture of government by crony capitalism.

  32. On the case of subway in the US versus rest of the world. Unlike most countries, in the US property rights are much better defined. While around the world governments own drilling rights, that is not necessarily so in the US. How much of the costs of building a subway system come from making sure that the drilling can actually happen? It may not be a difference in terms of efficiency, but simply of transparent rents (government to government in China or Paris is invisible). I would delve into the actual costs for the case of subways before comparing US and rest of the world.

  33. A fascinating post that reads like a suspense mystery. Why have certain sectoral costs risen so disproportionately over the decades?

    We don't discover why here but Cochrane once again focuses his ire on the cost of government. "Rule of law good, regulations bad" has been Cochrane's recent refrain.

    This time, however, Cochrane slips in: "The main effect of our regulatory AND [emphasis mine] legal system …".

    Whoa, there. This is shifty rhetoric.

    In an earlier essay, Cochrane condemned the regulatory state via a divide-and-conquer argument. But that tactic fails when one recognizes that regulations exist only in a legal context.

    Now, when it suits, the entire legal system comes under his ire. Is Cochrane now broadly attacking our legal system? Or was this just a simple slip?

  34. "Labor productivity -- number of people per quality adjusted output..."

    ... you mean quality adjusted output per person employed.

  35. Professor,
    Thought provoking post as always. I'm a physician, but do have an undergraduate degree in econ/finance. I linked in to the post by way of the health economics posts. Just some thoughts I'd like to hear you take on.
    1) Baumol's theories, I assume, can't explain fully the nearly tenfold increase in the costs of the heath care sector. What you expect to see if it were the case, however?
    2) An idea I've read about a few times is that lack of wage/income growth is at least partially or maybe fully explained by health care inflation and having employer based health care coverage. If health care coverage really is a 6-time greater portion of a workers compensation, can that along explain stagnant incomes?

    (from Florida)

  36. Is there evidence to support the administrative bloat idea in cross-sectional variation across schools? The marginal value of an additional administrator is hump-shaped---at some point it is negative value-adding to add an additional administrator.

  37. Professor Michael from EnglandApril 10, 2017 at 11:30 AM

    This is a very good article which quantifies the (obvious) extra cost of public sector bureaucracy, which is the inevitable corollary of increasing public sector spending to GDP. Unfortunately, it does not touch on the crowding out of private sector investment/productivity, or the political consequences of creating a massive class of public sector bureaucrats, with a strong motivation for constant expansion, led by both major political parties in the US (and also my country, the UK).
    Do you think the wrecking ball approach taken by Trump offers a solution?

  38. Another explanation for K12 education is the opening of the labor market to women. In 1970s, most women, including high ability women, could only become teachers, nurses, clerks, or stay-at-home moms (or before they became stay-at-home moms). Now high ability women are doctors, lawyers, managers, etc. There has been huge adverse selection into teaching.


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.