Wednesday, January 11, 2017

Book Review: Rise & Fall of American Growth

Robert Gordon's book The Rise and Fall of American Growth certainly doesn't need another review: see here, here, here, and here, for example. But I'm so inspired by reading it that I can't resist.

Whatever you think about Mr. Gordon's conclusions, the book itself is magnificent. He posits a miracle century from 1870 to 1970, beginning with the invention of the electric light, and concluding with the widespread adoption of central air conditioning. Prior to 1870, Americans lived lives more similar to medieval times than to our own. If not necessarily brutish, life was definitely nasty and short. Men's work was dirty and dangerous, while women's toil was unremitting drudgery. Children often died young.

By contrast, a household in 1970 had electricity, indoor plumbing, clean, running water, a car, television and radio, a refrigerator, washer and dryer, a telephone, and more. Nobody needs to share bathwater anymore. While there have been incremental improvements in all those devices, a modern family could move into an unrenovated, 1970s house and live quite normally. The only significant household appliance invented after 1970 is the microwave oven.

Similarly, workers have it much better off than before. They're employed in air-conditioned offices, in ergonomically-designed factories, and stores with break rooms and washrooms with indoor plumbing. Women don't have to sew their own clothes, or launder their husband's filthy clothes by hand, much less cook over an open fire. Today they can go work in the same climate controlled offices where their husbands are employed.

The result is a huge increase in productivity! Mr. Gordon documents it meticulously, and discovers that the largest productivity gain occurred during the decades from 1920 to 1950. That despite the Great Depression and World War II, or perhaps, even because of those events. For example, US manufacturers learned how to build one B-29 bomber every hour. That skill was not forgotten at the end of the war.

The central thesis of the book is that the miracle century can only occur once. The electric light has already been invented--that can't happen again. And likewise with the equally important internal combustion engine. Households can be networked (electricity, water and sewerage, telephone) only once, and while the network can be upgraded, the fundamental productivity change can't happen again.

In other words, we've eaten the low-hanging fruit. Productivity improvements such as occurred during the miracle century are once-off, never to be repeated. Hence economic growth will shrink from ~4% annually during the miracle century, to something around 1% today.

But Mr. Gordon's thesis does not just rely on anecdote and statistics. He brings some solid economic reasoning to the task as well. The growth in GDP is typically decomposed into three terms, which are then added together. Those are 1) the growth in the size of the labor force; 2) the growth in the total capital stock; and finally 3) the growth in TFP, which concerns most of the book.

TFP stands for total factor productivity, but that's really a misnomer. A more apt name is the Solow residual, named after Robert Solow, who invented the concept. But residual is the important label, and that means everything that is left over after you've accounted for the principal factors.

So economic statistics are always a bit dismal--hard numbers are hard to come by. For example, Mr. Gordon demonstrates conclusively that government statistics systematically overstate inflation--not because they're evil, but just that inflation is really hard to measure. Simple concepts such as real GDP growth are fuzzy--beset with uncertainties such as the inflation rate. And likewise with growth in labor and capital investment.

Nevertheless, within some error margin one can estimate growth as a function of labor. Add one additional worker x, and y amount of additional output will be produced. Similarly, buy a new machine for that worker to use, and output will increase all the more. Thus GDP growth as a function of the size of the labor force and as a function of capital investment can be reasonably estimated.

But there are some factors in GDP growth that are not a function of either labor or capital. For example, fine weather will produce a much larger agricultural output, regardless of how many workers or tractors the farmer employs. A change in government regulations may make it cheaper and easier to manufacture widgets, as can also a change in global trading networks. None of these can be expressed by the principal components (labor and capital), but are instead are left over as a residual, aka a fudge factor, known as TFP.

Among other things, TFP collects any errors that accrue in measuring labor and capital investments. I propose that it be renamed TFF: total fudge factor.

But the total factor productivity moniker is longstanding, and not irrational. While nobody denies any of the items that I've designated TFF, most of the residual is attributed to new technology. Buying a new abacus may improve the productivity of your new accountant at the margin, but replacing the abacus with a computer makes for a whole new ball game--productivity will make a giant step upward.

And Mr. Gordon is certainly on solid ground when he attributes to the miracle century to new technology. The rest of the fudge factor is either small or it cancels out. For example, despite the Great Depression, TFP growth continued unabated throughout the 1930s. There is no other reasonable explanation for this other than technology.

He also makes a strong argument about why TFP growth has decreased since 1970: apart from the digital revolution there's been very little new technology. While digital technology (computers, internet, etc.) caused a large spurt in TFP from 1994 to 2004, since then it's petered out--yet another revolution that can no longer be repeated.

So his book is a tour de force in economic history, and well worth reading just for that reason. The problem, insofar as there is a problem, is when he starts predicting the future. He's what might be called a techno-pessimist. I think he's too pessimistic--I believe AI and driverless vehicles will have a much larger impact than what he predicts. But then my crystal ball isn't any better (or worse) than his--we'll just have to wait and see.

Nevertheless, in addition to declining TFP growth he details four headwinds to American economic growth. One of these is rock solid, namely demography. The US labor force is not growing very fast, and soon may actually decline. This is partly because the baby boomers are retiring, partly because men are disproportionately leaving the workforce, and even workforce participation by women is declining slowly. Further, immigration rates will slow down (regardless of what Mr. Trump does).

If labor force growth declines, then economic growth will decline with it. No way around that.

The second headwind is debt. Government has run up huge debts. Not just the federal deficit, though that's bad enough, but social security and Medicare are increasingly burdensome obligations. States and municipalities are committed to pension payments that are unsustainable. Debt has added to economic growth today in exchange for reducing growth in the future. And Mr. Gordon's reasonable prediction is the future economic growth will be reduced. I agree with this conclusion as well.

The third headwind is education--we're no longer making progress in increasing the percentage of high school or college graduates. I think he's just wrong here. Our country invests way too much in education. I've written about that elsewhere.

Finally, he mentions inequality. This is a bit of a category error--he doesn't claim that inequality will per se inhibit growth, but rather that the proceeds will not be evenly distributed. "When we consider the future of American growth, we care not just about the growth of average income per capita, but also about growth of income per capita for the median American household" (p. 612).

I think the inequality problem is overstated. The proper comparison is not income, but rather consumption. The CEO may earn 1000x more than the worker, but he doesn't consume 1000x as much. He certainly doesn't eat much more than the average employee, and he can only drive one car at a time. Maybe he's 10 or 20 times richer than the working stiff. Not a big deal.

In summary, whatever you think about Mr. Gordon's conclusions, this is a superb book and well worth your time.

Further Reading:


  1. For what it is worth, I believe that he leaves out another important headwind...regulatory burdens. These have massively increased since the late 1960s, and end up consuming much of what is gained by growth in the various contributing factors to growth.

    1. I think you're right, and apart from demographics I think that's the biggest problem with our economy.