David R. Henderson  

Gordon on Growth: A Critique

The Public Goods Model vs. Soc... The Temptation of Discriminati...

On Sunday I laid out Robert J. Gordon's reasons for thinking that growth of per capita income in the future for most Americans will be substantially lower than it has been for the last two centuries. I promised a critique for yesterday but instead I had a good Labor Day with my family. So here's my critique today. It is not complete because to write a full critique would take much research and many key strokes.

My main critique is that Gordon is way too certain of his conclusion. His big argument--not the one about the six headwinds--is as follows:

[T]he central theme of this article is that innovation does not have the same potential to create growth in the future as in the past.

Again, I emphasize that I don't want to say that he's wrong--he could well be right. But I don't think his degree of certainty is justified. By the very nature of innovation, we don't know what it will be. Gordon is aware of this and points to "four classic examples of innovation pessimism that turned out to be wildly wrong": (1) Western Union's claim that the telephone would not be "a serious means of communication, (2) the head of Warner Brothers saying, in the silent-film era, "Who the hell wants to hear actors talk?", (3) IBM president Thomas Watson's 1943 statement, that the world market for computers is "maybe five," and (4) Bill Gates's 1981 statement that 640 kilobytes (of capacity on a floppy disk) "ought to be enough for anyone."

He misleads by saying "There are four classic examples." No. He gave four classic examples but there are more. How about Lord Kelvin's statement 1895 statement, "Heavier than air flying machines are impossible"?

Moreover, the term "innovation" doesn't apply only to new inventions. It can apply to new government policies. What if state and local governments carried out two innovations: (1) allowing jitneys (see here, chapter titled "99 and 44 Hundredths Built" and here) and (2) pricing road use in a roughly revenue-neutral way (cutting gasoline taxes while raising tolls)? If those two measures reduced the average American's time in traffic by even 20 hours a year (which is only 5 minutes a day for a 250-day year), that would be like a 1% increase in real GDP. Of course, that would be a one-time increase and an increase in growth only over the period in which it happens, but it's nothing to sneeze at. But Gordon considers very few policy options. And there are many more he could have considered.

Now to his six "headwinds" against growth.

1. The "demographic dividend" is now "in reverse motion." That is, the baby boomers are retiring. Thus hours per capita are falling and therefore, ceteris paribus, growth of output per capita will fall. That drops growth by 0.2 percentage points per year. I'll grant him that one. Notice, though, that this is a level change stretched over time. 30 years from now, that change will be pretty much complete. Gordon admits as much.

2. Higher education. Because high tuitions are deterring people, especially poor people, from going to college, we will lose 0.4 percentage points in growth. Hmmm. Gordon obviously does not read co-blogger Bryan. (Bryan's posts on this are too numerous to mention.) Gordon takes it as given that education is mainly about becoming more productive rather than mainly about signaling higher productivity. But, as I pointed out in my critique of a similar point by Tyler Cowen:

Cowen's third factor in the disappearance of low-hanging fruit is the high percentage of people who attend college. I found this factor more persuasive. He points out that in 1900, only 6.4 percent of high-school-aged Americans graduated from high school. That number peaked in the late 1960s at 80 percent, he notes, and has fallen to about 74 percent more recently. Also, in 1900, only one quarter of one percent of Americans went to college, whereas 40 percent of people aged 18 to 24 are in college today. In other words, there is not much room for improvement in educational attainment and, as he notes, the marginal college student today "cannot write a clear sentence, perhaps cannot read well, and cannot perform all the functions of basic arithmetic."
This suggests a huge piece of low-hanging fruit right in front of his nose: the number of people going to college. Cut that number dramatically and many of the "marginal students" would get jobs doing something productive -- as plumbers, as electricians, or in any of a number of occupations that do not require a college degree. I say this not as a central planner who wishes to decree who shall attend college and who shall not, but as a defender of people's right to use their income and wealth as they see fit rather than being forced to subsidize the education of others. Cut that subsidy to zero and cut taxes accordingly, and you would increase the after-tax income of many people, including the median family, and substantially reduce the number of marginal college students. If a marginal student still wants to go to college and he or his family or someone else he persuades wants to pay for it, then let him.

3. Rising inequality of income. Gordon notes that this doesn't affect growth of per capita income but would reduce growth of median income. But he takes an era of an unusual increase in income inequality (1993 to 2008), finds that, for that era, the gap between the growth of real household income for the bottom 99 percent and the top 1 percent was 0.55 percentage points per year, and extrapolates that into the future. There's not much basis for that.

4. The interaction between globalization and developments in information technology and communications (ICT). He writes, "Foreign inexpensive labor competes with American labor not just through outsourcing, but also through imports." True. But that makes us better off, not worse off. When we can buy something cheaper, our real income is higher. Ricardo nailed this almost 2 centuries ago. It's true that if the United States is a net exporter of something, then competition from abroad can hurt producers here more than it helps consumers here. But that's not what he's getting at. He's getting at "factor price equalization." The idea is that wages will equalize and, with the U.S. being one of the highest-wage countries, that means wages will fall. Of course, this is premised on the idea that all labor within a country and around the world is the same. This factor gives him a 0.2-percentage-point drop in growth.

5. Energy and the environment. Gordon buys into the idea that we need a stiff tax on carbon. If such a tax is not rebated, we lose 0.2 percentage points of growth annually. Simply solution: rebate it. Interestingly, though, he hints that the purpose of such a tax is mainly to make a statement, not to actually reduce carbon use: he points out that China and India together use twice as much carbon as we do and hints that their growth in carbon usage will be higher than ours.
Interestingly, he doesn't point out that U.S. carbon dioxide emissions in the first quarter of 2012 were the lowest they've been in any January-March quarter since 1992. This is partly due to the slow recovery and a mild winter, but another important reason is the switch from coal to natural gas. Of course, Gordon would be right to point out that this switch can't continue past the extreme case that we get rid of all coal. But it still could go far past where it's gone.

6. The twin household and government deficits. We can reduce debt/GDP, he notes, with lower government spending and/or higher taxes. This gives him a 0.3-percentage-point reduction in annual growth. I can't comment because he doesn't give enough data to allow one to know how he reached this conclusion.

The good news is that even if Gordon is right on six "headwinds," they're all slow changes in levels. The long-run growth rate, looking out about 30 years beyond now, would be pretty much unaffected by any of the six factors.

In a later post, I'll make some summary comments.

Comments and Sharing

COMMENTS (15 to date)

Hmmm. The same Robt. Gordon who authored this paper;

The rise in American inequality has been exaggerated both in magnitude and timing. Commentators lament the large gap between the growth rates of real median household income and of private sector productivity. This paper shows that a conceptually consistent measure of this growth gap over 1979 to 2007 is only one-tenth of the conventional measure. Further, the timing of the rise of inequality is often misunderstood. By some measures inequality stopped growing after 2000 and by others inequality has not grown since 1993. This cessation of inequality’s secular rise in 2000 is evident from the growth of Census mean vs. median income, and in the income share of the top one percent of the income distribution. The income share of the 91st to 95th percentile has not increased since 1983, and the income ratio of the 90th to 10th percentile has barely increased since 1986. Further, despite a transient decline in labor’s income share in 2000-06, by mid-2009 labor’s share had returned virtually to the same value as in 1983, 1991, and 2001.

Bob Montgomery writes:

Just thought I'd mention that Bill Gates never actually said that.

Thucydides writes:

Reading over these "headwinds," the question naturally comes up: is GDP growth really a good measure of our economic wellbeing? To mention just one item, aggregate GDP might grow with a rise in government spending that was a bit larger than a drop in consumption and investment. Or suppose we discovered a way to meet our household energy and food needs that was much cheaper than at present. We would be better off, but have a lower GDP.

Joe Cushing writes:

If growth stagnates it will be for one reason and one reason only. There is only one thing that can hold back growth for an extended time. That one thing is the increased intrusion of government into our lives. The reduction of freedom to do business, to invest, to borrow, and to spend the money you earn by taxation, rules we call regulation, and outright manipulation/takeover of markets is the the only thing that can keep our economy from growing. I can't say with certainty but it looks like we are going to have all of these things for some time and that America will stagnate for a long time, then decline, then collapse. We are on an accelerated path that is in the same direction as Rome's was. It took them 700 years to go from revolution against tyranny, through progress and prosperity, to decline and ultimately to collapse. I don't know when America will collapse but I feel like natural forces are in place that make collapse nearly inevitable.

The only question is, what will we do after the collapse? In eastern block nations, they got freedom back and started on a growth path. I hope that happens to us.

robb L writes:

I don't know about the "headwinds" but I do find your idea that young folks who are incapable of succeeding as college students can be converted into good electricians or plumbers laughable.

I have known many competent and incompetent tradespeople over the years and I believe that almost all of the competent ones would have done fine in college if they had chosen to and/or had the opportunity.

David R. Henderson writes:

@robb l,
You, not I, are the one who made the assertion about various young people being “incapable of succeeding as college students."

Mark Bahner writes:
I don't know about the "headwinds" but I do find your idea that young folks who are incapable of succeeding as college students can be converted into good electricians or plumbers laughable.

In 2-3 decades, I doubt there will be many things that human plumbers and electicians can do that robot plumbers and electricians cannot do.

In only 1-2 decades, a computer with human processing power (approximately 20 quadrillion operations per second) will cost less than $1000. And that same $1000 computer will have more than a petabyte (1000 terabytes) of storage, so that it can instantly recall the cause of failure of every single appliance that ever failed in the U.S. (or indeed the entire developed world) in the previous decade.

Why would anyone ever hire a human plumber or electrician if a robot that worked 24/7/365 would come and do an equivalent job for a tiny fraction of the cost?

P.S. Think of that. Seriously, you could call the robot plumber at 1 AM Christmas morning, and have the problem fixed before you got out of bed to open presents at 8 AM. You would not even have to be awake to see it arrive, do its job, or depart.

Chris Koresko writes:

It's hard to be sure what might drive economic growth over the next century or two, but it's not hard to think of a few things that might:

* Raw materials from space. That means energy, metals, ceramics, and volatiles in quantities that can ultimately vastly exceed what's available to us today. We're probably on the cusp of this now, with the rapid decline in launcher prices (by a factor of around 3-5 in the last few years, probably another 3-10 over the next decade), and the widespread availability of computers, software, and sensors to enable cheap smart spacecraft.

* Personalized medicine. With sensors, processors, and software all making dramatic progress, it should soon be practical for a family to own and use diagnostic and therapeutic tools exceeding the best available in modern hospitals, at a cost affordable to most. That'd free up a good fraction of the GDP, and effectively increase it by improving overall health.

* Virtual teleportation (telepresence). While it's possible to do a fair amount of work from home today, the combination of increasing low-latency bandwidth, cheap robotics, and immersive displays will allow people to routinely visit and interact with real environments almost as if they were physically there... but with the advantages of near perfect safety and instant "mobility".

* Flexible fabrication: Today's primitive "3D printers" may evolve into much more powerful devices capable of creating robust, intricate machines with more or less raw materials as input. Need a new air recycler unit for your Martian apartment? Print one!

* Robots: The perennial science-fiction appliance that takes care of the tedious chores at home and the dangerous jobs at work will bring us a lot of new wealth. We're on the cusp of this now in the physical world. Most U.S. factories already highly automated, and driverless cars having begun to gain regulatory approval. In the virtual world, robotic assistants that can understand and make sense of your words, make your appointments, buy your tickets, and make your travel arrangements are pretty much here now (though not yet in wide use).

We could be on a path to permanent stagnation, but it's not exactly obvious that's true.

Robb L writes:

To David,

excuse me...let me restate. Marginal students would also be marginal plumbers or electricians. Where is the gain?

To Mark Bahner,

Really? I bet you would have said the same thing 1 to 2 decades in the past. But even if you are right about AI That still leaves lots to buy when building your robot. If you look at an industrial robot, the cost of the cpu is a tiny fraction of the cost.

Also, I have to assume that you never hired a plumber. In real life there are usually various ways to solve a plumbing problem. You have to negotiate and think through the cost benefit of repair vs replace and how much to fix....you still would not be able to sleep in.

Poor Bill Gates. No one remembers a thing he actually did say, but everyone remembers him saying something he actually didnt say: http://www.wired.com/politics/law/news/1997/01/1484

Tom Chambers writes:

IIRC the 640 Kb limit was the memory management capacity of the MS-DOS operating system, not the capacity of a floppy disk.

Dano writes:

The the other classic example about innovation pessimism: cell phone use will max out at 6 million phones. I first heard that attributed to the old AT&T (which thought that slapping its brand on computers was better than entering the cell phone market) though my father in-law who is a retired ITT electrical engineer says that was ITT's belief.

David R. Henderson writes:

@Robb L,
You’re still missing my point. It is not necessarily the case that the only people who would choose not to go to college are marginal, in the sense of not being very good at it. That’s your assertion, not mine.

Mark Bahner writes:

I wrote:

In 2-3 decades, I doubt there will be many things that human plumbers and electicians can do that robot plumbers and electricians cannot do.
In only 1-2 decades, a computer with human processing power (approximately 20 quadrillion operations per second) will cost less than $1000. And that same $1000 computer will have more than a petabyte (1000 terabytes) of storage, so that it can instantly recall the cause of failure of every single appliance that ever failed in the U.S. (or indeed the entire developed world) in the previous decade.

RobbL replied:

Really? I bet you would have said the same thing 1 to 2 decades in the past.

No, I doubt I would have said the same thing 1-2 decades ago. In 1997, Hans Moravec predicted that a computer with the computing speed of the human brain (which he estimated at about 500 teraflops, or 0.5 petaflops) would be available for about $1000 (in 1997 dollars) "in the 2020s":

Moravec on when cheap computers would be as powerful as human brains

But even if you are right about AI That still leaves lots to buy when building your robot. If you look at an industrial robot, the cost of the cpu is a tiny fraction of the cost.

OK, let's say the brain of your robot plumber/electrician is $1000, and the rest of the body is $400,000. It still beats the heck out of any human plumber or electrician. Let's say you finance on a ten-year loan at 5% interest. That's $4200 a month, or $50,000 a year. (For 10 years...after that, it's working for free.) In the meantime, it's working 24/7/365. So it's probably doing 4 times the work of your typical human plumber/electrician, for that $50,000 per year.

The *key* thing here, that everyone needs to understand, is that the brain is everything. Cheetahs, horses, and dogs run faster than humans. Elephants and gorillas are stronger than humans. But none of those animals contributes any appreciable amount to world per-capita GDP. Human brains are, to borrow a phrase, The Ultimate Resource.

Also, I have to assume that you never hired a plumber.

You are assuming wrong. I was thinking about times I'd hired plumbers when I made my original comments.

1) I was up at 4:30 am to catch a flight from NC to CT one Memorial Day a few years ago. I noticed my shower was completely cold. I go out to the shed next to my townhome where the water heater is, and find the water heater has sprung a leak. It's an Apollo water heater that also supplies hydronic heat (air blown over a hot water heat exchanger). I shut off the water, then come back a week later. I hire a plumber, because I don't want to deal with the hassle of getting rid of the old heater, and putting in the new one. The old unit is 56,000 Btu/hr. I get some quotes. One guy is substantially less than the others, and will give me an Apollo unit. But when he comes, it's only 24,000 Btu/hr. I tell him, no way...that's just not going to heat my house. He calls around and says he can drive to Raleigh and get a 46,000 Btu/hr unit. I'm worn down and reluctantly agree. He installs it. Now my heating system is basically full-on on any really cold day. No way a computer plumber would have been so careless.

2) I got another plumber to install a kitchen faucet and repair a plastic pipe (long story). He did fine on the kitchen faucet, but misunderstood what I wanted on the pipe, and his boss ended up telling him to re-do it the way I'd asked.

But don't get me wrong. No job will be safe from computers in a couple decades. When computers with the processing power of the human brain get down to about $1000 each (which they should in the next 1-2 decades) that will change everything. No economist should fail to take that into account. And yet 99.99+ percent of economists *do* fail to take that into account.

Dan Hill writes:

I don't see any reason to beleive that innovation can't continue. In fact, all other things being equal, you could make an argument that it ought to accelerate (standing on the shoulders of giants and all that).

The real issue is whether government will stifle either the development of new innovations by hindering the institutions that do the R&D - a small but non-zero risk - or more likely prevent their application to the economy. The real risk is the government continuing to entrench incumbent institutions and practices. Innovation drives growth through the process Schumpeter called Create Desctruction. Too many politicians don't understand you can't have the Creative part without the Destrruction part.

Comments for this entry have been closed
Return to top