Productivity Growth: The Good, the Bad, and the Ugly

Bonsack cigarette machine, 1888, capable of producing more than 120,000 cigarettes a day. Durham Public Library, public domain.

If you care about innovation and material progress, you learn to care about productivity growth — because it functions as our best available statistical proxy for those desiderata. “Productivity growth isn’t everything,” Paul Krugman famously observed in The Age of Diminished Expectations, “but in the long run it’s almost everything. A country’s ability to improve its standard of living depends almost entirely on its ability to raise its output per worker.”

In this essay, I want to focus on just how rough a proxy productivity growth actually is. First of all, and this is a familiar point, the productivity growth that everybody cares most about isn’t even something we can measure directly. We start with the rough indicator known as “labor productivity,” or output per worker. But when that figure rises from year to year, the increase may simply reflect a change in the productive “quality” of labor (workers on average are a little better educated and trained, or are slightly older and more experienced), or it could be due to additional purchases of machinery that allows expanded production. What we want to isolate is what economists refer to as “total factor” or “multifactor” productivity growth: increases in the output achieved by given inputs of labor and capital.

It is a “residual,” a leftover figure: it’s the difference between actual measured growth and those parts of growth that can be accounted for by increased labor or capital inputs. After we back out of output growth everything that we can reasonably ascribe to input growth, whatever’s left we chalk up to improvements in productivity. Thus, as the economist Moses Abramovitz memorably put it, total factor productivity is a “measure of our ignorance about the causes of economic growth.”

Once we’ve done our best to isolate productivity growth from output gains that simply reflect additional inputs, it becomes clear that this phenomenon is far broader than the occasional breakthroughs that emanate from founders’ garages and corporate research labs. Rather, fully realized innovation consists of invention plus diffusion: after a new product or production method is introduced, it must gradually attract investment, workers, and of course customers before it can register its full impact in raising output. Accordingly, productivity growth consists not only of new and improved things to do and ways of doing them, but also the successful reallocation of capital and labor resources toward these more productive uses.

And indeed, we can see that in the United States today, the problems of flagging dynamism we suffer from are primarily problems of blocked diffusion. The United States now pays a sizable cost in lost growth because NIMBY restrictions on housing supply are preventing people from moving to the most productive parts of the country; the cost of producing solar and wind energy has plummeted in recent years, but we struggle to take advantage due to delays in building out new production capacity and transmission lines; Operation Warp Speed succeeded in developing effective covid vaccines in record time, yet hundreds of thousands died unnecessarily because they bought into anti-vax conspiracy theories or were simply reluctant to get a shot.

So if we’re going to measure economic progress properly, we need productivity growth to be a broad catch-all category that encompasses both invention and diffusion, as both are needed to keep economic output expanding and material living standards moving upward. But as we shift our attention from the “economic problem” to humanity’s “permanent problem,” and consider the relationship between innovation and living “wisely and agreeably and well,” we find that for this purpose measured productivity growth is significantly overbroad.

It’s still a useful proxy. Rising to the challenge of the permanent problem will require major technological advances — to put our energy and food systems on an environmentally sustainable footing, to push down the cost of living and swell the ranks of the independently wealthy, and to enable abundant self-sufficiency at the household and local levels. To create the material conditions for mass flourishing, capitalism needs to be roused out of its decades-long malaise of faltering dynamism — a malaise brought about by mass loss aversion in the form of a timid safetyism, a meddling vetocracy, and an anti-Promethean backlash against technological mastery over nature. There are encouraging signs at present that a turnaround may be in the offing, with promising developments in solar, wind, batteries, advanced geothermal, alternative fission design, fusion, CRISPR, mRNA vaccines, space launch capability, and of course artificial intelligence. Since the calling card of capitalism’s crisis of dynamism over the past half-century has been stubbornly feeble productivity growth, we will be monitoring this indicator closely for evidence of an innovation rebound.

But when assessing productivity growth as a proxy, not of strictly economic progress, but of progress in human flourishing, we have to look past the top-line, aggregate number. We need to distinguish among three distinct types of productivity growth: let’s call them the good, the bad, and the ugly. And as with productivity growth as an economic indicator, the productivity growth that serves unambiguously as an indicator of human flourishing is a residual: good productivity growth is what’s left when the bad and ugly productivity growth are stripped away.

So what constitutes bad productivity growth? Here we’re talking about innovations that inflict harm on the people who buy and use them. Such innovations duly pump up GDP per capita, thus making us richer in a strictly economic sense, and even create consumer surplus in the sense that consumers would have been willing to pay even more for the goods in question. But in so doing, they exact a heavy toll on physical and mental wellbeing.

How do harmful products gain willing customers? First of all, their harmful effects are often not apparent when the product is first introduced. Furthermore, they are able to pass the market test by exploiting a wrinkle in human nature: the distinction between first-order and second-order desires. Bad productivity growth gives us more of what we want — but we wish we didn’t want it.

For a classic example of bad productivity growth, consider the handiwork of James Bonsack, who in 1881 patented a cigarette-making machine that could produce over 70,000 cigarettes in a 10-hour day, with later models churning out more than 120,000 a day by the end of the decade. By comparison, at that time the most highly skilled manual workers could roll around 3,000 cigarettes a day. Thirty such machines would have sufficed to satisfy nationwide demand for cigarettes in 1885 – but suffice it to say that price declines made possible by soaring productivity led to dramatic increases in demand. U.S. cigarette consumption skyrocketed from a mere 54 per capita in 1900 to 4,345 in 1963 (happily, that number has now fallen to a little over 1,000); lung cancer deaths followed with a lag, jumping from 4.9 per 100,000 in 1930 to 75.6 per 100,000 in 1990.

Of course, it took a long time before people really understood the health risks associated with smoking: health warnings only started going on packs in 1966. And beyond that, most people who ever smoked cigarettes never ended up getting lung cancer or emphysema or heart disease. Lots of people pick up the habit when they’re young, enjoy the taste and the social cachet of smoking, and figure out a way to quit before anything obviously bad happens to them.

Cigarettes are both extremely unhealthy and highly addictive, so they’re an especially virulent example of bad productivity, but there are a number of industries that follow the same basic model: making products that are enjoyable in moderation but harmful in excess. Matthew Yglesias, my Niskanen colleague, refers to them as “akrasia industries,” after the Greek word for weakness of will:

These companies are selling stuff that’s fun but that we know is bad for us — cigarettes, booze, junk food, gambling, probably social media feeds as well. A key thing about all of this is that it’s actually totally fine in moderation. If you go to Vegas one weekend, get drunk, smoke a pack of cigarettes, lose a little money playing blackjack, and then eat a bunch of garbage the next day because you’re hungover, that’s just having fun. But if you do it a lot, you’ll end up ruining your life. And basically everyone knows that. But consumption of these kinds of goods is very skewed, and most of the revenue comes from the small minority of customers who are problematic overconsumers. These customers almost always know that they are problematic overconsumers, but almost all of us suffer from weakness of will in at least some areas of life and some companies in some industries are very good at exploiting that.

The distinctive pattern followed by these industries — most customers consume the product moderately and responsibly, a minority abuse the product and mess up their lives, but that minority account for the bulk of total consumption and determine the profitability of the industry, so companies face strong incentives to prey on that minority — makes for one of the nastier hazards of life amidst mass affluence. The perils of overindulgence weren’t something most people needed to worry about during the long reign of material scarcity; “indulge as much as you can whenever you can” was simple prudence under the straitened circumstances that most people faced. With the stinginess of nature’s gifts replaced by endless artificial bounty, our instincts have been turned against us.

We are forced to compensate for our misleading appetites with reflection and individual willpower — remedies that may suffice for most of us in the face of most of life’s temptations, but whose limitations are starkly revealed by our bulging waistlines and incessant phone checking. It is possible, of course, to come to the aid of those caught between their first-order and second-order desires with collective action — namely, paternalistic policies such as taxes, zoning and time-of-purchase restrictions, or even outright bans, designed to discourage excessive consumption and abuse. But the same mass affluence that multiplies our temptations also leads us to value personal choice above all else, and thus makes us increasingly averse to limits set on us for our own good. Why spoil the fun for everybody just because a weak-willed few can’t handle it? Libertarian pushback can go even further, arguing that paternalism isn’t just unfair, it’s also futile: people with a self-destructive itch are going to figure out a way to scratch it whatever laws you pass.

The libertarian complaint about fairness carries real weight; the argument that intervention is futile, rather less so. And that’s because we need to take account of the power of bad productivity: people in akrasia industries have strong incentives to circumvent all your internal defenses against overconsuming their products, and that leads to a great deal of time and energy being plowed into figuring out how to make those products as irresistible and addictive as possible.

For a case in point, check out Matthew B. Crawford’s recent Substack essay “Misanthropic uses of the nudge,” in which he draws at length from Natasha Dow Schüll’s Addiction by Design: Machine Gambling in Las Vegas. Once upon a time, as recently as the 1980s, card tables and roulette wheels provided the main action at casinos — but those days are long gone, as slot and poker machines now account for more than 85 percent of industry profits. And those machines have been repeatedly modified over the years with one goal in mind: extending “time-on-device” — sound familiar, Facebook and Twitter users? “Player-centric design” is the industry euphemism for addictiveness-über-alles. As one industry insider put it, “The more you tweak and customize your machines to fit the player, the more they play to extinction; it translates into a dramatic increase in revenue.”

“Play to extinction” — that’s another industry term of art, meaning playing until all funds are exhausted. Here’s what it looks like, according to Crawford:

It is not uncommon for heavy users to stand at a machine for eight or even twelve hours at a stretch, developing blood clots and other medical issues. Paramedics in Las Vegas dread getting calls from casinos, which usually turn out to be heart attacks. The problem is that when someone collapses, the other gamblers won’t get out of the way to let the paramedics do their job; they won’t leave their machines. Deafening fire alarms are similarly ignored; there have been incidents where rising flood waters didn’t dislodge them. The gamblers are so absorbed that they become oblivious of their surroundings.

Schüll interviews one woman who makes sure to wear dark clothing when she goes to gamble, so it won’t show when she urinates on herself.

So the idea that we’re respecting autonomy when we allow akrasia industries a free hand is kind of a joke: these industries exist to make a mockery of autonomy. It’s not autonomy, or sober and deliberate self-rule, that our culture prizes so highly, but rather the illusion that unchecked personal choice and autonomy are the same thing. And so our regulation of akrasia industries has grown more permissive even as temptations have multiplied, with restrictions on gambling, pornography, and marijuana unwinding over recent decades. The most notable exception to this trend, the clampdown on cigarette smoking, was able to overcome our cultural aversion to paternalism by asserting the need to protect third parties from the unpleasant odors and (probably overhyped) negative health effects of second-hand smoke. And if there is a new exception in the works with growing momentum for raising age limits on access to social media, well that just goes to show that our cultural libertarianism still carves out an exception for kids.

So bad productivity growth despoils our society with the deliberate infliction of human suffering — it’s hard to conclude that any society could be said to be flourishing, that its members are living “wisely and agreeably and well,” when the systematic exploitation of human weakness is being carried out to such an extent. Yet a flourishing society must also be, in my view at least, a free society, and there are limits to what a free society can do to save people from themselves. We could do a better job of limiting the kind of diabolical ingenuity on display in Las Vegas, forcing the akrasia industries to blunt or dumb down their appeal, but I don’t see how we can eliminate that appeal altogether.

And meanwhile, on the other side of the ledger, you have comfortable majorities of satisfied, moderate customers — not to mention the huge pot of spending power created by the monetization of vice. For one thing, there’s an excellent university in North Carolina now because an entrepreneur named James B. Duke saw the promise in James Bonsack’s new machine and went on to revolutionize the American tobacco industry. And beyond the great fortunes created by akrasia industries and the good uses to which they’ve been put, there are also all the wages paid to all the employees and all the payments made to all the suppliers and all the good, decent lives supported by this economic activity. Yes, capitalism draws energy from base motives — and in the akrasia industries it does so in especially unlovely ways. But capitalism also practices alchemy, transforming what is base into something useful or even lofty. With bad productivity growth, both aspects of capitalism are on display.

And what about ugly productivity growth? Here we’re talking about innovations that show up in the GDP statistics and which make their customers unambiguously better off — but which also inflict harms on third parties along the way. I see three distinct sources of ugly productivity: industries that pollute, industries that mistreat animals, and industries that mistreat their workers. The dark side of these industries cannot obscure the fact that they have played a central role in modern economic growth: we could never have gotten rich without lots of ugly productivity growth. But since we have become rich, it has made sense for us to focus regulatory efforts on reducing capitalism’s collateral damage. And as we set our sights on higher levels of technological and social development, we can aspire to big reductions in ugly productivity growth as we shift economic activity to cleaner and more humane industries.

There’s no way to think sensibly about the tradeoffs posed by ugly productivity growth unless you place them in historical context. In the early days of industrialization, when mass poverty was still the norm, it was inevitable that rapid progress and ugly side-effects went hand in hand. Yes, factories pumped toxic filth into the air and water, but life expectancy was in the process of doubling. Yes, working conditions in those factories were dangerous and brutal, but immigrants flocked to America’s shores for a chance to work in them.

From our contemporary standpoint, the tradeoffs look very different — but they still exist. Fossil fuels powered humanity’s escape from mass poverty: I for one am deeply grateful for this state of affairs, and I have no patience with condemnations of that escape for having been so messy (condemnations invariably issued by people cossetted with every modern comfort and convenience). That said, fossil fuels are not capable of sustaining our prosperity over the long term: they are changing the composition of our atmosphere, and consequently our climate, so securing and expanding human flourishing requires that we move to new, clean energy sources. Meanwhile, completing the clean energy transition successfully will require us to navigate a host of other environmental risks and tradeoffs: dealing with the toxic waste created by producing solar panels, and the impingement on habitats by large solar arrays; minimizing the risk of earthquakes caused by advanced geothermal; coping with the environmental impacts of lithium-ion batteries; and managing the safe operation of nuclear plants and the handling and storage of nuclear waste.

In 1970, when I was a little boy, world population totaled 3.7 billion people and some 35 percent of them were undernourished. Today the population has more than doubled, to over 8 billion, and the share of the undernourished now stands below 10 percent. This is a magnificent accomplishment — one that the conventional wisdom of my boyhood said was flatly impossible. Indeed, obesity worldwide has tripled since the mid-1970s, and more people today are obese than undernourished.

As circumstances have changed, we can now afford to look at our food systems in a new light. And if we can resist turning our gaze from the stomach-turning sight, we will be confronted by the appalling ugliness of factory farming. I’m not a vegetarian, I don’t feel the faintest ripple of guilt over the fact that human beings are omnivores, but I don’t see how anyone can seriously argue that the industrial-scale, systematic cruelty meted out to living creatures every day is anything less than morally horrific. Not to mention factory farming’s sprawling global footprint (agriculture now consumes half the world’s habitable land, and three-quarters of that is for livestock) and the mass extinctions being driven by habitat loss. Or how jamming together humans and domesticated animals in close proximity combined with globalized travel patterns renders future pandemics all but inevitable.

It's clear enough, I think, that further boosting agricultural productivity through ever-more-diabolical forms of animal torture does not constitute progress in human flourishing. On the contrary, we need to continue to expand regulation of livestock production to limit the worst abuses, and we should avidly pursue the good productivity growth of more delicious and less costly cultivated meat.

I doubt any Americans my age or younger can really wrap their minds around how brutal and exhausting working conditions used to be. Consider the fact that in 1913, when Henry Ford first instituted the moving assembly line, he had to hire over 52,000 men just to maintain a work force of 14,000. In other words, the typical worker lasted barely three months before walking off the job — and keep in mind that the typical auto worker in 1913 could hardly afford to be too choosy.

Ford solved the turnover problem by instituting a whopping pay hike to five dollars a day in 1914 — a dramatic illustration of the fact that rising living standards for industrial workers in part reflected “combat pay” for enduring torturous conditions. Over a half-century later, things were much better than they had been back in the Model T days — but they were still pretty terrible. Here’s a reminiscence from someone who started work at General Motors in 1976:

The next day I went in after school and worked ten hours. I thought I had gone to hell. I couldn’t believe what people were doing for money…. I couldn’t believe how hard it was. But I thought of myself as a tough guy, and I was determined to stick it out.

So it’s fair to say that virtually all workers today have a much easier time of it than did their parents or grandparents. But that’s how it should be: since we spend something like half our waking lives at work (or at least we did before the recent spike in working from home), improved working conditions are an important component of overall rising living standards. And therefore, for anyone who cares about human flourishing, it ought to be genuinely dismaying that new digital monitoring tools are making working life decidedly worse for a growing number of employees. It’s now possible to keep tabs on what websites employees are viewing, how fast they’re typing, how much time they spend on calls with customers and the time between calls, the emails they’re sending and receiving, when they go to the bathroom and how long they stay there, where they are now and where all they’ve gone over the course of the day. Needless to say, all this snooping ramps up workplace stress and reduces job satisfaction.

There are two basic ways to make workers more productive: (1) give them new tools or processes that amplify their skills and allow them to produce more with the same effort, or (2) “sweat” them, i.e., push them to work harder and faster. For workers in advanced economies, now well into the 21st century, it’s hard to see a good argument for continued reliance on (2) as an avenue for future productivity growth. Whatever minor efficiencies are gained, and whatever pennies are shaved off the prices of goods and services as a result, these paltry gains cannot under present circumstances justify the intentional infliction of emotional and physical distress.

To continue making progress toward a world of mass flourishing, productivity growth remains indispensable – but we have to be discriminating. We’re not rich enough that we can give in entirely to loss aversion and safetyism — that way lies stagnation and, sooner or later, catastrophe. But we are rich enough that the gaps between what is good for GDP and what is good for human wellbeing have grown large enough to take notice of. We’re rich enough that some forms of productivity growth are no longer worth the candle.

Rapid gains in artificial intelligence, and increasing hype about the imminent arrival of superhuman artificial general intelligence, will surely put our powers of discrimination to the test. Boosters tell of a coming millennium of good productivity growth, while doomsters warn of bad productivity growth that could end the entire human race. Meanwhile, the potential for all manner of AI-aided surveillance and manipulation ensures that there will be plenty of ugly side effects to worry about. We cannot know yet how things will shake out and what the actual tradeoffs will look like, but we should know enough to dispense with false, easy extremes — both “the end is nigh” and “what, me worry?”. As usual, reality is going to be frustratingly complicated.