This book didn’t need to be written. Having chosen to write it, it needed extensive chopping at the hands of an editor, which it didn’t get. Having been published, it need not be read. Gordon has been writing about this topic for decades, so I can only assume that he’s published the meat of this book in various papers. My advice, should you find yourself interested in reading this book, would be to seek out some of those earlier papers and run away from the book with all deliberate speed.

Gordon is famous as a technological pessimist, arguing that the productivity gains of the mid-twentieth century are not coming again. Since we live in a pessimistic era, I wager that he’s gotten more traction now than he might otherwise have; but he has good reasons for his argument. It’s essentially one argument: most of the major leaps in innovation can only happen once. We nearly eliminated infant mortality. We brought clean water to nearly everyone’s home. We cleaned up cities by building sewage systems, and by replacing horses with automobiles. We were, in fact, so successful in eliminating infectious disease that Americans are now much more likely to die of chronic disease. We electrified virtually the entire country. We connected everyone with a network of high-quality interstate highways. We built first the railroad, then the telegraph, then the telephone.

Compared to this, the last quarter century doesn’t look all that interesting. What we have we got? The Internet is amazing, yes. We have somewhat larger televisions (movies and television were both pre-1950 inventions). Smartphones are remarkable. We dream of self-driving cars. We dream of nanotechnology. We dream of gene therapy. It’s hard to imagine, though, that any inventions on the horizon can compare to those “one time only” inventions that Gordon mentions. Progress in certain areas of public health seemingly can’t go anywhere else: once we’ve driven infant mortality to zero, we’re done. What’s the life-expectancy dream? Do we think we’ll ever be able to get the average human to live to 100 years? 120? And after you’ve electrified the whole country and gotten everyone on the Internet, is bringing average broadband speeds up to 10 megabits per second really a game-changer? So Gordon’s point is intuitively pretty clear. It would take some argument, of course, but the overarching thesis seems legit.

His main analytical tool to demonstrate the slowdown of American productivity growth is Total Factor Productivity, which seems to also be known as Solow’s Residual, after the Nobel-laureate economist Robert Merton Solow. It’s known as the “residual” because it’s what’s left over when you try to explain output by counting labor (that is, humans) and capital (that is, machines) as inputs. First you count all the machines going into GDP; then you count all the people; then you assume that output is x% labor and y% machines. Whatever you haven’t explained is TFP.

TFP seems to capture “making the best use of the resources you have.” If people learn how to use, let’s say, computers better, then the Solow residual goes up if the quantity of computers remains fixed. I imagine that the same is true of labor: if you use a given quantity of labor better, the Solow residual rises; hence to the extent that living in cities makes people more productive (a fairly well-established economic result, as I understand from Ed Glaeser), the Solow residual rises for a given quantity of labor.

For all the analytical importance that he places on Total Factor Productivity, Gordon spends shockingly little time explaining it; I feel like I spent more time explaining it in the preceding paragraphs than he did in his extremely long book. He asks us to take it for granted that TFP is the appropriate way to measure innovation. Maybe it is, but he doesn’t argue that sufficiently.

That’s the deeper problem: I don’t really know whom his book is aimed at. At least 70% of it is a grindingly detailed walk through American consumer history, much of which illustrates that we mismeasure various economic statistics — as when he notes that the Consumer Price Index may not realize that Wal-Mart exists. Hence the CPI may not count the significantly cheaper goods that people buy there. Or, citing William Nordhaus, Gordon writes that “the value of increasing life expectancy in the first half of the twentieth century roughly doubles the growth rate of the standard measure of consumption expenditures”.

These are interesting measurement problems; if Gordon were aiming at an audience that he expected to have some economic curiosity and some quantitative skill, he would focus on these problems. Instead, before he’s spent much time on any one of these issues, he’s off and running with more trivia — as when he spends the time to tell us that

There was still room for a critical masterpiece such as The Godfather to spur millions of Marlon Brando impersonations and for Star Wars to burst onto the scene and become a cultural icon and box office success, to this day trailing only Gone with the Wind in inflation-adjusted domestic gross revenue.

I still have no idea why that’s there. And I don’t know what I’m supposed to conclude from Gordon’s observation that

So rapidly were new versions of WordPerfect released that by 1992, only a decade into the PC era, it reached version 6.0.

Firefox these days increases its version number by 12 every time it corrects a typo. What does that prove?

Gordon has interesting things to say, and he has trivial things to say, and he systematically buries the interesting things beneath the trivial things. In one of his many asides, he writes that “The patent office was fair, respected, impartial, and not subject to bribes or corruption”. Well, that’s great, and that’s interesting. How did the Patent Office get there? It seems much more relevant to American economic history that we have an incorruptible bureaucracy than that our word processors have high version numbers. Or when Gordon writes that “Running water had been achieved by the Romans, but it took political will and financial investment to bring it to every urban dwelling place,” he’s certainly correct; politics is important. It’s far more important than Francis Ford Coppola’s masterwork, as much as I love Coppola. Yet he spends no time at all on the politics. (Here’s the place to recommend that you read Francis Fukuyama’s Origins of Political Order, whose whole point is to understand how countries ever come to have stable governments with functioning, efficient bureaucracies.)

Of course a book like this is obliged to end with a chapter explaining how we can get out of this mess, if at all. Gordon doesn’t seem to think we can. We have demographic problems, with an aging population. Our debt is mounting. Etc. What I find odd is that Gordon never actually explains why we should care. If productivity growth is slowing because certain things can only happen once, then … well, they’ve happened, so why aren’t we happy about that? Give yourself a pat on the back, humanity: you invented penicillin. That’s only going to happen once, so be proud of it.

Gordon does seem to appreciate that we may today be suffering from an excessively robust standard of living. When the U.S. is as obese as it is, the problem might well be that we have too much stuff; the entire advertising industry exists to convince us that we need to shove more of that stuff in our faces. As far as I can tell, though, Gordon doesn’t go down a Galbraith-like road of considering post-scarcity economics. He just seems to take it for granted that productivity growth is good, and that a slowdown in productivity growth is bad.

He then goes beyond a mere slowdown in growth: toward the end of the book, Gordon says that “America’s children will almost certainly not be as well educated, healthy, or wealthy as their parents.” This seems to imply that GDP will not only grow more slowly than it has, but that it will actually decline. The book doesn’t really provide support for this.

So the book is a hot mess. I’m given to understand that many great books are great because they had great editors slicing away patiently in the background. Gordon’s editor could have made this book much shorter, much more focused, and much more concerned with fundamentals than with trivialities. But then it would be a much different book.

I was listening to the Freakonomics podcast episode about Robert Gordon on the walk home. A lot of people (including my beloved Weeds podcast) have been talking about Gordon’s new book on how American productivity has changed over the last hundred-plus years, and I’m probably going to read it soon; seems like the sort of thing I’d (and you’d?) be interested in. Also, it’s one of those books that a lot of people talk about but probably most people don’t read cover-to-cover, and I enjoy reading such books cover-to-cover out of spite.

A lot of the discussion is necessarily going to revolve around what you count and how you count it. I think I need to understand the National Income and Product Accounts, or at least brainstorm the problems with the NIPA, before I dig too much into Gordon. For instance, how does taking care of your own kids rather than hiring out childcare factor into the productivity measurements? My understanding (see 6.21 et seq. in System of National Accounts 1993 for a very succinct explanation) is that most domestic production is not counted. So if I employ a nanny to take care of my kid, that’s going to count as income for the nanny and an expense for me; if I instead choose to stay home and take care of my own kid rather than work, I am subtracting from GDP. I wonder how much GDP we added from 1945 to 1990, say, just by women shifting from home to the office. That much GDP is arguably artificial, in the sense that we should have been counting their work at home as well.

(If I buy my home rather than rent it, the BEA counts what they call “owner’s equivalent rent of primary residence”. Why don’t they likewise count “mother’s equivalent sale of child-care services”?)

Likewise, consider Nordhaus’s history of lighting efficiency. Consider, specifically, the chart on page 11, registering a 30,000-fold increase in the efficiency of lighting from the beginning of recorded history to the advent of the compact fluorescent light bulb. Is it possible that we’re just under-counting productivity increases?

Or consider the humble shipping container (my review; Krugman’s hat tip). Where exactly does it show up in the productivity numbers? I assume/hope the BEA is counting an abstract cost of “shipping services”: the cost to ship the same quantity of the same goods from point A to point B; that cost has gone down, so the productivity of “shipping services” has gone up.

How about public health? As people stopped smoking and lung-cancer deaths decreased (see page 2), how did that show up in the GDP numbers? We lost some money from the unsold cigarettes, and we lost the money that we would have spent in hospitals. Yet clearly we’re better off than we were. Hopefully that shows up in GDP. Or, to take another public-health number: does water delivered to your tap from a tax-funded municipal water system count for less than water you buy in the store?

These public-health questions point to the broader question: what is the point of economic growth? Isn’t the point of economic growth that it makes us happier, healthier, and longer-living? Are we measuring that? To the extent that economic growth is connected to what matters, we should care about it; but otherwise, maybe we needn’t.

I think most of the usual argument for economic growth is that a rising tide lifts all boats, or a larger pie leaves more pie for everyone, or some such metaphor. But what if insufficiently small pies are not the problem? What if the problem is that we’re not getting pies to the right people? That is, what about distribution rather than growth? And what sort of improvements to the human standard of living could come about without anyone needing to spend another dime? What if we managed to get every last cigarette smoker to quit? What if we ended automobile deaths? We needn’t necessarily develop technological solutions to the problems of cigarette smoking and car crashes; we might be able to solve them without adding a dime to the GDP.

To be clear, I’m not suggesting that the GDP numbers are mismeasuring things. But I think these are the sorts of questions with which one should approach a book like Gordon’s: are they measuring the world correctly? And even if they are, should I care? Will increasing GDP per capita halt the rise in death rates among non-Hispanic whites? (The rise is seemingly driven by women in the South.) These are the sort of questions that I’ll have in the back of my head as I read Gordon.