Globalization: What Did Paul Krugman Miss?

This is a very nice short framework-for-thinking-about-globalization-and-the-world piece by Paul Krugman: Paul Krugman (2018): Globalization: What Did We Miss?

It is excellently written. It contains a number of important insights.


I have, unusually, a number of complaints about it. I will make them stridenly:

First, Paul Krugman claims that, in Heckscher-Ohlin models at least, from the early 1970s to the mid 1990s international trade put only a little bit of downward pressure on the wages of American “unskilled” and semi-skilled workers. I think that is wrong. I think that from the early 1970s to the mid-1990s international trade, at least working through the Heckscher-Ohlin channels, put less than zero downward pressure on the wages of American “unskilled” and semi-skilled workers.

As I see it, it is important to note that “emerging markets” and “global north” are not static categories. Japan, Spain, Italy, Ireland were low-wage countries in the 1970s. From the early 1970s to the mid-1990s the relative wage levels of the then-current sources of America’s manufacturing imports were rising more rapidly than new low-wage sources of manufacturing imports were being added. The typical American manufacturing worker faced less low-wage competition from imports in the mid-1990s than they had faced in the early 1970s.

As I see it, where manufacturing workers came under pressure (and they did) it was not from increased low-wage competition from abroad but rather from:

  1. fiscal policy failures that produced the Reagan (and then Bush II) deficits as Republican governance redirected dollars earned by foreigners from buying our exports to buying our bonds
  2. managerial failures in Detroit (and elsewhere in the U.S.) and successes abroad
  3. technological failures in Pittsburgh (and elsewhere in the U.S.) and successes abroad

As I see it, yes, we could have protected Detroit and Pittsburgh from the consequences of their managerial and technological failings—but it would have been at immense cost for the rest of the economy, a very unfavorable benefit-cost tradeoff. And we should not have elected Republicans and given them the keys to the economic policy car: that rarely works. But given that we did give the Republicans the keys, and given Detroit’s and Pittsburgh’s managerial and technological failings, globalization from the early 1970s to the mid-1990s was a wonderful thing for America as a whole: it provided us with enormous benefits in every scenario, and in the unfortunate scenario we were dealt by the Reagan Democrats and the Big Three auto executives of Detroit, globalization greatly reduced the damage.

Second, I agree with Paul Krugman when he writes as though the “hyperglobalization” from the mid-1990s to the financial crisis was a big deal (which it was):

This huge surge… Containerization was not… new… [but] t took time for business to realize… [the] possibilities…. [Plus] a broad move… toward outward-looking policies…. China made a dramatic shift from central planning….

But I disagree when he writes that “hyperglobalization” was in some sense a threat to blue-collar Americans’ economic and social position:

It’s clear that the impact of developing-country exports grew much more between 1995 and 2010 than the 90s consensus imagined possible, which may be one reason concerns about globalization made a comeback…

Why? For reasons that Paul recognizes and summarizes:

A fairly novel form of trade… break[ing] up value chains, moving labor-intensive parts of the production process overseas…. The factor content of North-South trade hasn’t risen nearly as fast as the volume…

Let’s unpack this. In the age of widely-separated intercontinental value chains, we can see that there are actually more types of “blue collar” manufacturing jobs than the skilled-craft, semiskilled-assembly line, and unskilled traditional classification. Most importantly, we can see that the blue-collar jobs that are traditionally called semiskilled-assembly line are actually divided into two. The first are those jobs that require relatively literate workers with substantial experience and tacit knowledge who plug into sophisticated and highly productive divisions of labor supported by very productive communities of engineering practice. The second are those jobs that plug into those divisions of labor supported by those communities of engineering practice, but that actually do not require relative literacy or involve a great deal of tacit knowledge or experience—jobs that are doable by virtually everybody with the standard mental structure and eye-brain-hand loop of the East African Plains Ape, and that we thus call “unskilled”, even though they involve tasks that are currently regarded as very hard AI problems.

Before the coming of intercontinental global value chains, the distinction between these two types of semiskilled manufacturing jobs was of relatively little importance. Both paid relatively well for jobs requiring little formal education: both benefited from the requirement that workers be located near to engineers (and marketers, and executives) and from their participation in highly productive production processes, so both shared in the rents produced therein. But the truly unskilled portion—even though they were called “semiskilled” were not truly good jobs: they were boring, repetitive, and not very productive. An economy that could figure out a way to offshore those jobs would find that it had a global competitive advantage, and that would strengthen its truly valuable communities of engineering practice and ability to productively employ those relatively literate workers with valuable experience and tacit knowledge.

This was brought home to me most strongly in the years after the NAFTA debate. Opponents of NAFTA from Harley Shaiken and Thea Lee to Ross Perot had claimed it would be very damaging to the American automobile industry. Not so. And not just the firms executives, the shareholders, and the marketers were better off as a result than they would have been otherwise: the blue-collar workers with tacit knowledge and experience were better off as well from Detroit’s larger market share, and the truly unskilled portion—perhaps we should call them “polyester uniform”?—did not have jobs in the auto industry but had jobs about as good outside of it.

So, at least as I see it, the coming of “hyperglobalization” strengthened opportunities for U.S. workers without formal education to find jobs where their skills, experience, and tacit knowledge could be deployed in ways that were highly productive. What “hyperglobalization” did do was provide the top 1% and the top 0.1% with another lever to break apart the Dunlopian labor relations order, break the Treaty of Detroit, and redistribute the shared joint product from highly productive mass production backed by valuable communities of engineering practice upward in the income distribution. But there were many such levers in the U.S. from the 1970s to today. And “hyperglobalization” was, as I see it, one of the weakest and shortest of them. It gets blamed not because it was an important driver of the process, but because it allows one to blame others: brown people, yellow people, and, of course, the rootless cosmopolites.

Third, I quarrel with Krugman’s—and with Autor, Dorn, and Hanson’s (2013)—assessment of the China shock. Paul writes:

[While] trade deficits explain only a small part of the long-term shift toward… service[s]… soaring imports did impose a significant shock on some U.S. workers…. Fights over tariffs look very much as if they come out of a specific-factors world…. This is where the now-famous analysis of the “China shock” by Autor, Dorn, and Hanson (2013) comes in. What ADH mainly did was to shift focus from broad questions of income distribution to the effects of rapid import growth on local labor markets, showing that these effects were large and persistent. This represented a new and important insight…

Put me down as believing that, as I see it, Autor, Dorn, and Hanson’s focus on the stable absolute number of U.S. manufacturing jobs before the China shock of the 2000s and its drop as a result of the China shock is substantially misleading. One might look at the share of the workforce who have—and the share of those entering the workforce who get “good blue-colllar” jobs, in which we see not stability but rather a smooth decline in the proportion. One might look at individual towns, cities, and regions, in which case one sees patterns of regional industrial growth and collapse: the defense cycles, the collapse of New England textiles and leather, the rise of the Carolinas, the shift out of the Midwest to the falsely-called “right to work” states, plus the general desire of people after air conditioning to live in places where the winters are not so dire. It is not a new insight that such shocks to regional labor markets had effects that were large and persistent: anybody who had ever driven through Lowell or Fall RIver, MA knew that before Paul Krugman had published his first paper. It is, however, a very important insight.

Yes, the reduction in the share of the U.S. workforce in tacit knowledge and experience semiskilled blue collar jobs has been a big deal. But the overwhelming bulk of that is due to technology, not trade. Yes, there has been an additional reduction beyond technology. But the bulk of that has been a second-best compensation and adjustment for the disastrous Republican habit of running large budget deficits at full employment. Yes, the U.S. government should have done much more to support communities and workers who found themselves under the hammer. But for that blame the legacy influence of social darwinism on American politics: the U.S. government did little for Lowell or Fall River back in the day. And complaints about the failure to properly manage a process that is, globally, overwhemlingly positive-sum should be mailed to the address of the Reagan and Trump Democrats of Michigan, Pennsylvania, and Wisconsin, not to poorer brown and yellow people in Mexico and in China.

Moreover, from the perspective of the country as a whole and from the perspective of many of the communities affected, the China shock was not a big deal for local labor markets. Yes, people are no longer buying as many of the products of American factories as Chinese imports flood in. But those selling the imports are turning around and spending their dollars investing in America: financing government purchases, infrastructure, some corporate investment, and housing. The circular flow will it: the dollars are of no use outside the U.S. and so the dollar flow has to go somewhere, and as long as the Federal Reserve does its job and makes Say’s Law roughly true in practice, it is a redistribution of demand for labor and not a fall in the demand for labor.

And here is the kicker, as I see it: the types of people and the types of jobs funded by the imports of the China shock looks very much like the types of people and the types of jobs displaced from the tradeable manufacturing sector. Yes, some local labor markets got a substantial and persistent negative shock to manufacturing, often substantially cushioned by a boost to construction. Other local local labor markets got a substantial and persistent positive shock to construction. And on the level of the country as a whole the factor of production that is (truly) semiskilled blue collar labor does not look to me to have been adversely affected.

Until 2008.

Now we get to my fourth quarrel: the play is Hamlet. But where is the Prince of Denmark? Zero references to “recession”, “finance”, “financial crisis”, or “recession”. Yet, at least as I see it, the key thing that we missed about globalization was not its impact on factor prices in some Heckscher-Ohlin model or an shared rents in some specific-factors model but rather that when a big financial crisis and depression came “globaization”—and poor people elsewhere—would provide an excuse to distract blame. There was a lot of blame: Blame financiers who had no control over their derivatives books because they failed to manage. Blame financeirs who had control over their deiverative books but who thought, like Charles Price of Citigroup: “you have to keep dancing as long as the music is playing”. Blame Federal Reserve Chair Alan Greenspan. Blame Treasury Secretary John Snow. They were at the heads of the agencies responsible for controlling systemic risk when the vulnerabilities emerged. And they did not no—control it, that is. Blame Federal Reserve Chair Ben Bernanke. Blame Treasury Secretary Henry Paulson. They were at the heads of the agencies responsible for controlling systemic risk while there was still time to shore up the system—and they did not.

The Prince of Denmark here is the Greenspan-Snow shock, not the China shock. What we missed about globalization was not its impact on blue-collar semiskilled workers with experience and tacit knowledge and communities, but how it would interact with attempts to shift resoponsibility and blame off of the appropriate properties.

And then, of course, ther is 2010: Barack Obama’s declaration in his State of the Union Address that the time for bold action to boost employment was over:

We took office amid a crisis, and our efforts to prevent a second Depression have added another $1 trillion to our national debt…. Families across the country are tightening their belts and making tough decisions. The federal government should do the same. So tonight, I’m proposing specific steps to pay for the $1 trillion that it took to rescue the economy last year…. Like any cash-strapped family, we will work within a budget to invest in what we need and sacrifice what we don’t. And if I have to enforce this discipline by veto, I will…

I have never found anybody working in economic policy in the Obama administration who thought that this large a shift this quickly was a good idea. Some have admitted to believing that it was a meaningless rhetorical nothingburger—after all, it excepted “spending related to our national security, Medicare, Medicaid, and Social Security”, and you can do anything macroeconomic you want on the spending side in those categories. They were wrong. Others were strongly opposed. Others say that they were quiet, but certainly not boosters.

And, indeed it wasn’t a good idea.

If the Greenspan-Snow shock is the Prince of Denmark in this play, the idea that the crisis was over and the need for stimulative policy was at an end as of early 2010—call it the Obama-Geithner shock, perhaps—is King Claudius, or at least Queen Gertrude here.

And this gets me to my fifth quarrel with Paul Krugman here. As I see it, the most important thing we missed about globalization was how much it required support from stable and continuous full employment. That, I think, ought to have been the focus of his talk to the IMF.

It is now 81 years since John Maynard Keynes published:

Whilst… the enlargement of the functions of government involved in the task of adjusting to one another the propensity to consume and the inducement to invest would seem to a nineteenth-century publicist or to a contemporary American financier to be a terrific encroachment on individualism. I defend it… as the only practicable means of avoiding the destruction of existing economic forms in their entirety and as the condition of the successful functioning of individual initiative….

If effective demand is deficient… the public scandal of wasted resources… the individual enterpriser… is operating with the odds loaded against him. The game of hazard which he plays is furnished with many zeros…. The authoritarian state systems of today seem to solve the problem of unemployment at the expense of efficiency and of freedom. It is certain that the world will not much longer tolerate the unemployment which, apart from brief intervals of excitement, is associated and in my opinion, inevitably associated with present-day capitalistic individualism. But it may be possible by a right analysis of the problem to cure the disease whilst preserving efficiency and freedom…

True. Now as much as ever.

Determining Bargaining Power in the Platform Economy: Reinvent Full Transcript

Reinvent: Determining Bargaining Power in the Platform Economy: Our political system has been hacked by time, circumstance, chaos, and disaster…

…The failings of the electoral college, the fact that small states hacked the constitution in 1787, so we now have a world in which the minority in the Senate represents 175 million people, while the majority represents 145 million people, and the gerrymandering after the 2010 census are primary examples of this dysfunction.

Fixes for the economy?:

  • A 4 percent inflation target from the Federal Reserve, * Incentivizing businesses to invest in workers,
  • Reinvigorating the idea that technology should be used to augment workers, not replace them.

The possibilities for positive human flourishing from the platform economy are immense, provided the platforms actually work. Uber’s investors are currently paying 40 percent of Uber’s costs. What happens when these investors start wanting their money back? The platform economy moves bargaining power away from the service providers and from the customers, and into the hands of the platforms. This is a problem for both consumers and independent workers. What bargaining power workers will have will be correlated to the time and resources devoted to training them: when you walk, you disrupt a general production value chain, and it is expensive to figure out how to replace you, even if there’s someone else who certainly could do the job just as well. But if it is not very expensive, you have little power.

Nevertheless, here in California it is hard not to be a techno-optimist—especially if you are an curious infovore…says…

Full Transcript:

Pete Leyden: Hello. I’m Pete Leyden. Today, we have Brad DeLong with us. He is an economics professor here at U.C. Berkeley. He’s also the recently installed chief economist of the Blum Center for Developing Economy.

Pete Leyden: It’s good to have you here.

Brad DeLong: Great to be here.

Pete Leyden: We’re here at this moment with all these folks from the OECD, the economists from the United States here, and the technologists. If you had to think about the kind of moment we’re in right now, how would you characterize where we are as far as the evolution of the global economy and our technologies are? Is there anything special about this moment? Anything critical about it? Any ways you think about this juncture?

Brad DeLong: Let me give you a three-part answer to that: a 50-year horizon answer, a 15-year horizon answer, and then a 0-5-year horizon answer.

Brad DeLong: The 0-5-year horizon answer is: America has been deeply scarred by the financial crisis that started in 2007, the deep recession that followed, and the extraordinarily anemic recovery since. That still leaves us with four million people fewer in the labor force and looking for jobs than we ought to have. We are not sure what all of them are doing. Many are living in their sisters’ basements playing video games. The economy and people’s expectations of how it works have been shocked. How well our society functions is still deeply scarred. We are recovering only slowly, if at all, back to what we used to think was normal. that is the 0-5-year horizon answer.

Brad DeLong: The 50-year horizon answer is: Expect the collapse of the need for people to do a great many tasks that people used to do and are still doing that provide value. In the next 50 years an awful lot of paper-shuffling tasks are going to be taken over by software bots. An awful lot of blue-caller, traditionally male, tasks are going to be taken over by robots. Few occupations will disappear. But many occupations will be transformed. And many will shrink. The income and wealth distribution will be upset—either in a positive or a negative direction. These 50-year horizon processes are what most of us upstairs at this conference are worrying about.

Brad DeLong: Then there’s a 10- to 15-year horizon. How much of this transformation is going to happen in the next 15 years, as opposed to the rest of the next 50? How fast is the 50-year horizon process going to come upon us? Where exactly will be the first sectors and first places in which the coming of—call it the “Rise of the Machines”—the replacement not just of blue-color manufacturing but also construction and transportation and distribution and warehouse workers will be hit by technology, and who? Where and when will a good deal of standard white-color paper-shuffling work start to disappear as expert systems and software bots take it over?

Pete Leyden: In your career, is this about as momentous a time as you’ve seen? Or is that overhyping it?

Brad DeLong: That we are recovering from the macroeconomic catastrophe that started in 2007 has definitely made it very, very fraught. The failure of our Electoral College to deliver us a competent president in 2016 has definitely made it very, very fraught. Our political system was hacked partially by malevolent people, but mostly by time, circumstance, chaos, and disaster. It has been hacked in three ways.

Brad DeLong: The first way it has been hacked is the Electoral College failure; that gave us a president who really is not up to the job in practically any dimension. The second way is that small states hacked the constitution in 1787, so today the 49-seat minority in the senate represents 175 million people, while the 51-seat majority represents 145 million people. The desperately minority congressional government understands it is a minority government. It is acting oddly as a result. Third, the state-level Republican gerrymandering after the 2010 censushas given us House of Representatives is extraordinarily unrepresentative of the median American voter. These have created a time of great political fraughtness. We clearly have a very badly broken political system. That greatly deepens and increased the dangers of managing what I call the 50-year transition. And on top of that is the economic fraughtness left from 2007.

Brad DeLong: The 50-year transition has been going on since Steve Jobs and Steve Wozniak began building personal computers in the garage. It has been going on at a more less constant pace. We see a little bit more about where it’s heading with each passing year. It really has not sped up much. What has made this moment fraught is the political disaster of 2016 and the echoing effects of the economic disaster of 2007.

Pete Leyden: There are three challenges we’ve been wrestling with here. You mentioned one of them—the robots and AI. But there’s this idea that the economy is moving towards more and more independent workers. There’s also this rise of the “platform economy”. How do you think of those other two challenges? How important developments are they? How much do they concern you—or actually encourage you as good thing?

Brad DeLong: The platform economy has a number of dimensions. One is what Hal Varian was talking about at the conference yesterday—the “end of the need for scale”. With Amazon Web Services and with Google anyone with a good idea can launch their website at scale for pennies, providing through the web whatever service or commodity they want to provide. And if demand is there they can scale up as far as they need to using very cheap world class-efficiency systems to support their businesses. You no longer need a large initial lump of capital of any. You just soft launch and look for demand. You use the web and search to attract customers. You use AWS and Google to provide your back end. This should be the cause of an enormous upward surge in entrepreneurship and enterprise, and a great flourishing of creativity. Whether it’s individuals with an extra four hours a week making extra money by driving for Lyft, or whether it’s writers saying, “I don’t want to have to sell books at 20 bucks and get only a $1.50 in royalties. I want to establish my own Patreon and have my fans pay me directly”, or any of a whole bunch of other things.

Brad DeLong: The possibilities for positive human flourishing from the platform economy are immense—if the platforms actually work. Right now we find ourselves in a world in which riders are paying essentially 60% of Uber’s costs. Uber’s investors are paying 40% of Uber’s costs. What happens when the investors begin wanting their money back? Do we find that Uber has enough economies of scale, and scope, and enough of a brand and a first mover advantage, that it’s a profitable business? Or do we find that the business gets taken over by somebody else? What if Google puts a little taxi ride button on every Google Map screen, saying: “We’ll only charge you a $1.50 as a handling fee”? Uber will have charge you considerably more if it wants to repay its investors. Uber may turn out to have done the trail-breaking thing. Uber may suffer the fate of most pioneers—arrows in their back, and face down. Or perhaps the platform economy will not be a good thing. Perhaps it moves bargaining power away from the real producers, who are doing the work, and also away from the customers, and into the hands of that one large company in the middle that controls the information. That’s still up for grabs.

Brad DeLong: Most people who fear that we are, as the extremely sharp Zeyneb Tufekci of Duke University says, “building a dystopia one brick at a time in order to trigger people to click on ads”, greatly fear that individual humans, given our cognitive disabilities, will be no match for the informational middleman organization using deep learning and information to figure out how to trigger and control us. Others are much more optimistic—although not necessarily much more optimistic about the prospects of individual platform pioneers like Uber. They are, however, more optimistic about the prospects of the large companies that have entrenched dominant positions: companies like Apple, Google, and Facebook—not that they are optimistic about any one of them, but rather they are optimistic about the prospects for profits for all of them put together, because what opportunities one of them fumbles another one is likely to recover.

Pete Leyden: They will do well for everybody, or do good for themselves?

Brad DeLong: They will well, and they will do good.

Pete Leyden: Do good. And where do you find that, actually?

Brad DeLong: I’m basically a techno-optimist. It’s hard not to be a techno-optimist in California—especially if you’re an intellectual, a data loving infovore.

Pete Leyden: You think even though there are all these challenges, the platform economy is something we could get behind?

Brad DeLong: Yes.

Pete Leyden: What about speaking as an economist now? This other thread: independent workers playing an increasing role in the economy. There’s a positive way to see that. There are also challenges in that. I’m curious to how you see that challenge.

Brad DeLong: Independent workers have, by their nature, very little bargaining power over what economists call “rents embedded in the system”. Your bargaining power is limited by what you could charge if you walked away from the relationship and went out on your own, and by what your counterparty would have to pay in order to get someone else to step up in your place. You have bargaining power if, when you walk, you disrupt a complex and valuable general production value chain, and your counterparts finds it expensive to figure out how to replace you. You have bargaining power even if there is someone else who could fill your job just as well if and only if it is difficult to find that person. A lot of successful middle-class societies have been based on situations in which relatively low-skill workers have bargaining power and share big time in economic rents, either because there aren’t that many replacements in the area or because they threaten to walk as a group.

Brad DeLong: Yesterday at the conference, ex-governor Jennifer Granholm of Michigan told a heartbreaking story about a town in central Michigan: 8,000 people, of whom 3,000 work edin the refrigerator plant. Those 3,000 workers made a very good living for Michigan at the start of the 2000s—some 35 buck an hour, I think, in wages and benefits. But the refrigerator plant goes. And after it leaves they are lucky to make $12 or $15 an hour. And then all those who worked to satisfy their demands find their markets have halved in value. The refrigerator plant workers’ skills, machines, lifetime of experience bashing metal and operating things that form refrigerator coils—there’s really not much demand for those skills in central Michigan, and they find that they can’t transfer their skills to do anything else of great value. The principal source of their income was sharing in the rents created by the refrigeration value chain: their dominant market positions and imbedded technology. That kind of danger faces a lot of people who become independent workers. And the middleman firm does not have to close. All the middleman firm has to do is say: “I am altering the deal. Pray I do not alter it any further”.

Brad DeLong: On the other hand, an independent worker economy is not bound to be destructive. As long as we have a middle-class society, these are immense amounts of work to be done for one another. And replacing any of us with somebody else will be expensive. Think of the typical job in the economy going from being a manufacturing worker to being a barista at a coffee shop or a teacher at a yoga studio. As long as you have a middle-class society, there’ll be a lot of people who will be willing to pay handsomely for yoga lessons or for a particular espresso beverage brewed exactly the way they like it with a smile, a handshake, and a friendly three-minute conversation about how they’re doing, while the thing brews.

Brad DeLong: On the other hand, if we have plutocracy, in which the only people who have a lot of money are the rich, then the potential customers for the yoga studio won’t be able to pay very much and your fancy expresso drink won’t command very much. The only people who’ll have middle-class lives will be those who control resources that are useful for making things for which rich people have a serious Jones. There’ll be relatively few of those as well. Thus most of it is the shape of what the income distribution will be. That is ultimately a political choice about the distribution of wealth. An unequal distribution of wealth will drive an unequal distribution of income. That will then reproduce itself. And an equal distribution of wealth will drive a more equal distribution of income, which will also reproduce itself.

Pete Leyden: You’ve kind of given—this is probably right that could go this way, could go that way, could be positive, could be negative—I get that and that’s good choices here but…

Brad DeLong: …This is why we economists want not to have just two hands, but a prehensile tail as well…

Pete Leyden: But in that respect, what do you see is the most promising ways forward to kind of deal with this juncture we’re in—tip the balance? What are the things that you’d like to see happen soon here that would evolve this economy in the direction to make it healthy?

Brad DeLong: Am I allowed to say a 4% inflation target from the Federal Reserve? A Federal Reserve that is less focused on keeping inflation very low, more focused on keeping employment high, and more focused on making sure that there’s enough inflation in the system that the Federal Reserve can maintain interest rates at a level at which it will have the power to stabilize the economy? Businesses are not going to want to train their workers unless workers are scarce. Workers tend to be scarce only in what we call a “high-pressure economy.” The first and most important thing would be to change the calculations of those businesses that might be willing to invest in training workers. They need to feel that workers are valuable commodities under their control that they need to boost the value of. It has been totally the case since 2008, and largely the case since 2001, that businesses think there are plenty of workers out there, and we really don’t care about them, because the bottlenecks keeping us from being more profitable are elsewhere. Making labor a serious bottleneck for business so that business focuses on helping workers become more productive and more useful is, I think, the first thing we could do.

Brad DeLong: The second thing would be to repeat what Silicone Valley did in the 1980s and 1990s, that is you’re old enough to remember the coming of the Macintosh computer in 1984…

Pete Leyden: …Indeed…

Brad DeLong: …Apple bought Super Bowl time for a dystopian 1984 TV commercial. It was all about how important the Macintosh computer would be as an engine of freedom. They really believed that the personal computer was an engine of freedom. It allowed you to control and access your own information, rather than having to rely on some human resource or IT department backed by some large mainframe that had lots of data that you weren’t allowed to access and controlled your life. The fear back then was that people would become information serfs: the valuable parts of the enterprise and the value chain would be kept under lock-and-key in the hands of the priests of IT and HR. It was a world in which people would take their draft cards and burn them, in which your information would come on five IBM cards which would all say, “Do not fold, spindle, or mutilate”, and which you would feed into the machine face down, nine-edge first. The vibe was that those cards produced decisions over which you had no control or knowledge. Thus making information technology tools to augment people’s abilities to figure out and maneuver in the world that they were in, rather than information technology being a tool for supervision and control—“you’re 15% less productive at processing claims, so we don’t need you around anymore, and you have no skills or information that can be transferred elsewhere.”

Brad DeLong: The idea of Apple Macintosh 1984 was of information technology as a way of augmenting human intelligence and boosting productivity—rather than information technology as a way that we can substitute capital for labor, and getting these annoying workers out of the factory or office while still producing as much. That was a key and a revolutionary social goal of Silicon Valley as it existed in the 1980s and 1990s, from the coming of the personal computer to the flourishing of the internet. In some sense, Silicon Valley has to figure out how to do this again. Organizations like, say, Berkeley’s engineering school have to help. Large companies tend to be much more interested in figuring out how to use information technology to shed annoying and expensive workers, rather than how to give those annoying and expensive workers more control over their lives.

Pete Leyden: Well, that’s fascinating. It’s a big challenge to the tech world as well as a challenge to policy makers. I wish we had more time to kind of go deeper into many, many possible solutions there. But just to wind up here, big challenges, possible big solutions shift, how confident are you that we’re going to manage this transition here? Whether it’s the five-year, the 15-year transition, how confident are you going to do it and how worried are you?

Brad DeLong: I would say I’m not confident at all.I would say that our income and wealth distribution now has managed to tilt itself in a bad way. If you want the economy to pay attention to you, you better have money. And the money is too concentrated now. If you want the polity to pay attention to you, you better have a movement. Yet somehow it seems that the age of the internet and of the decline of manufacturing has made it harder rather than easier to create durable social movements. At the same time it has made it much easier to create the appearance of a social movement via software bots controlled by some server in the Former Yugoslav Republic of Macedonia. And that crowds the information flow. What is the cartoon? “1980: my incandescent light bulb produces ten times as much heat as light. 2017: my LED light bulb has been taken over and is now running a button out of 50,000 Twitter followers controlled via a server on another continent.” That that seems to be the world we’re moving into. It’s not terribly a reassuring one.

Pete Leyden: Well, that’s a good place to end. At least that’s a sobering thought, and thanks so much for joining us here and giving us your thoughts.

Brad DeLong: You’re very welcome. It’s a great pleasure. Bring me back again.

Pete Leyden: I will.

Reinvent: Determining Bargaining Power in the Platform Economy

Reinvent: Determining Bargaining Power in the Platform Economy: Our political system has been hacked by time, circumstance, chaos, and disaster…

…The failings of the electoral college, the fact that small states hacked the constitution in 1787, so we now have a world in which the minority in the Senate represents 175 million people, while the majority represents 145 million people, and the gerrymandering after the 2010 census are primary examples of this dysfunction.

Fixes for the economy?:

  • A 4 percent inflation target from the Federal Reserve, * Incentivizing businesses to invest in workers,
  • Reinvigorating the idea that technology should be used to augment workers, not replace them.

The possibilities for positive human flourishing from the platform economy are immense, provided the platforms actually work. Uber’s investors are currently paying 40 percent of Uber’s costs. What happens when these investors start wanting their money back? The platform economy moves bargaining power away from the service providers and from the customers, and into the hands of the platforms. This is a problem for both consumers and independent workers. What bargaining power workers will have will be correlated to the time and resources devoted to training them: when you walk, you disrupt a general production value chain, and it is expensive to figure out how to replace you, even if there’s someone else who certainly could do the job just as well. But if it is not very expensive, you have little power.

Nevertheless, here in California it is hard not to be a techno-optimist—especially if you are an curious infovore…says….”

Do “They” Really Say: “Technological Progress Is Slowing Down”?


Consider the 256 GB memory iPhone X: Implemented in vacuum tubes in 1957, the transistors in an iPhoneX alone would have:

  • cost 150 trillion of today’s dollars: one and a half times today’s global annual product
  • taken up a hundred-story square building 300 meters high, and 3 kilometers long and wide
  • drawn 150 terawatts of power—30 times the world’s current generating capacity


  • 4.3 billion transistors in the A-11
  • 2,199,023,255,552 bits in the 256 GB memory—each of which needs a transistor (and a capacitor)
  • Let’s say 2.5 trillion transistors…
  • And you can buy 256 GB of memory for $100—of which, say, 1/4 is the cost of a transistor.
  • So, say, 125 dollars’ worth of transistors in an iPhoneX

How much would it have cost you to buy a vacuum tube sixty years ago, back in 1957?

  • Well, in 1959 you could buy a one-byte—8-bit—Phister 366 for 65 dollars
  • So, say, 8 dollars a bit.
    • 8 dollars in 1957 is 60 dollars today via the GDP deflator
    • 8 dollars in 1957 is 160 dollars today as a share of U.S. nominal GDP per capita
    • 8 dollars in 1957 is 320 dollars today as a share of U.S. nominal GDP
Vacuum Tube Assembly

The transistors in an iPhoneX would, back in the late 1950s, implemented in vacuum tubes, have:

  • cost 150 trillion of today’s dollars, which is:
    • one and a half times today’s global annual product,
    • more than seven times today’s U.S. annual national product
    • forty times 1957’s U.S. national product
    • fourteen times 1957’s global annual product
  • taken up 100 billion square meters of floor space
    • that is (with a three-meter ceiling height per floor): a hundred-story square building 300 meters high, and 3 kilometers long and wide
  • drawn 150 terawatts of power—30 times the world’s current generating capacity

Oh. And clock speed. The AN/FSQ-7 operated at 75khz. The A-11 is a 6-core 24 mhz processor:

  • 2000 100 billion square meter buildings, each a hundred-stories—300 meters high—and 3 kilometers long and wide
  • 3000 times today’s global annual product
  • 300 petawatts of power—60,000 time sthe world’s currnet generating capacity

for the late-1950s vacuum tubes to match one iPhoneX…

And we haven’t even gotten started on the hardware architecture, or on the software and maintenance support necessary to emulate an iPhoneX at speed back in the late 1950s…

Google Image Result for https omicron aeon co images f8ba0a79 8a5e 4d3c ada0 aa928fc72422 header essay sage 104624883 master jpg Google Image Result for https videogamehistorian files wordpress com 2014 06 ibms 10 billion machine jpg Google Image Result for https viceland assets cdn vice com viceblog 26715348sage computer robert duvall JPG

Information Technology and the Future of Society (Hoisted from 2001)

Hoisted from 2001: Information Technology and the Future of Society (My Bekeley CITRIS Kickoff Talk) For perhaps 9000 years after the beginnings of agriculture the overwhelming proportion of human work lives were spent making things: growing crops, shearing sheep, spinning yarn, weaving cloth, throwing pots, cutting down trees, copying books, and so on, and so forth.

Technology did improve enormously over those 9000 years: contrast the clothes-making technology at the disposal of Henry VIII of England with that of Rameses II of Egypt three thousand years before; contrast the triple-crop paddy-irrigated rice- and water-control-based agriculture of the Yangtze Delta in eighteenth-century China with the scratch-the-soil-with-a-hoe agriculture of two thousand years before.

But as Thomas Robert Malthus first wrote in the 1790s, rising populations had put enough pressure on scarce natural resources to offset the benefits of better technology and keep living standards nearly constant for the people if not for the elite: American President Thomas Jefferson in 1803 A.D. certainly enjoyed a higher standard of living than Roman Consul Marcus Tullius Cicero in 63 B.C. But did Jefferson’s slaves enjoy a higher standard of living than Cicero’s? A large amount of archeological evidence has not yet found significant differences.

For the past two hundred and fifty years, since the start of the Industrial Revolution, the productivity of those workers who make things has exploded. Hand-spinners in the eighteenth century took 50,000 hours–20 full work-years–to spin 100 lbs of cotton into thread (Freeman and Louca (2001), and spinning of one sort or another took up perhaps 5% of total labor-time. Today it takes 40 work hours to spin 100 lbs. of cotton: a more than thousand-fold amplification of productivity in this one task.

As our productivity at growing crops and making things has exploded, demand for the things we make has grown too, but not fast enough to keep the crop-growing, food-cooking, mineral-extracting, clothes-making, box-carrying, and other goods-producing share of our economy’s labor force from falling. Today those who in any earlier age would be classified as “production workers”–and would have been the overwhelming majority of the labor force–are perhaps 20% of our economy, and the bulk of them are better characterized as machine-watchers and machine-fixers. According to Stanford’s Robert Hall, as early as 1980 there were twice as many salesmen in Ford-selling auto dealerships as there were assembly-line workers employed by Ford Motor Company.

So what are the rest of us–the other 80%–doing? In a sense, we all–from U.C. professors to chief technical officers to xerox operators, Ford Salesmen, cashiers, and parking-lot attendants–are and have long been information workers: people whose jobs are, if we examine them closely, largely concerned with determining what exactly the goods-producing sectors should make, how it should be made, where it should go, and to whom it should be distributed–and that is leaving aside the large chunk of our economy that is symbolic communication as an end in itself.

Today we see–not yet sharply, not yet clearly, but no longer dimly–the prospect that the ongoing technological revolutions in data processing and data communications will do for the “information” sectors of the economy something like what the Industrial Revolution did for goods-producing sectors like cotton spinning. As Steve Cohen over in the City Planning department here likes to say, you are now building the equivalent of the industrial-age tools for shaping and handling matter, but you are building tools for thought (Cohen, DeLong, and Zysman (2001)). And if we can figure out how to make these tools for thought fulfill their promise, they should produce a quantum jump in our technological power, economic productivity, and–we hope–quality of life of as many energy levels as the jump of the Industrial Revolution itself.

But there are major problems of social engineering and organizational design that stand in our way. A century or so ago, at the height of the Industrial Revolution, the market economy turned out to have an extraordinarily good fit with the developing industrial technologies of goods-making. It provided a framework of social organization that was extraordinarily effective in providing people with incentives to carry on activities that generated rapid technological development, capital accumulation, and economic growth.

An effective form of social organization faces decision makers with incentives that mirror the impacts of their actions on society as a whole.

Because the goods produced by industrial technologies were rival–that is, could only be of use to one person at one time–each person’s use of such a good diminished the supply available to the rest of society. Thus it made sense from the viewpoint of efficient distribution to require that users pay a price–diminish their ability to acquire and use other resources–for commodities. And those prices paid then gave producing organizations the resources to carry on and expand their activities.

Because the goods produced were excludable–that is, it was by-and-large straightforward to limit control over use to those authorized–it was easy and straightforward to push decision-making outward from the clueless bureaucratic center to the periphery where people on the ground might actually have a good sense of the situation, and of what should be done.

These three advantages–earmarking additional resources for successful and efficient production organizations, providing users with incentives for economically-efficient distribution, and decentralization of decision-making to where the knowledge was likely to be–were delivered by accident by the trade-and-market economic structure of Adam Smith.

But now as we try to realize the technological promise of information technologies, the old forms of economic organization no longer have a natural fit with the requirements of technological development and economic growth. Once an “information good” has been produced, sharing it with another person doesn’t reduce the rest of society’s resources and opportunities. So there is no efficient-distribution reason to charge a price for it.

But where then does the flow of signals to assess which production organizations are efficient come from?

In an earlier age we would be more inclined to rely on government funding, but these days we have a keen awareness of the advantages in applied development at least of semi-Darwinian competitive mechanisms, where investigators are responsible to investors seeking profits and not to committees seeking whatever committees seek.

Moreover, it is only with difficulty that information goods are excludable. But if their use can’t be restricted to authorized users, then the entire market-as-a-social-calculating-and-signalling mechanism simply breaks down. Unfortunately, attempts to make information goods “excludable” by various forms of use protection waste valuable time and energy: I shudder at the memory of having spent two hours on hold during three phone calls, and having spent another two hours of my time rebooting and reading installation error messages the last time I tried to upgrade one of the Adobe programs–GoLive–on this laptop. I doubt I’ll ever be able to face the prospect of buying another Adobe program again.

Two things, however, are clear. First, caught between “government failures” in applied research and the ever-larger “market failures” that will be created as the characteristics of information-age goods clash with the requirements for market efficiency, intermediate forms of organization–like large publicly-funded research universities–need to play an even larger role in research and development in the future than they have in the past.

Projects like CITRIS promise the benefits of government research–the wide distribution of knowledge and the acceleration of cumulative research–and the benefits of private entrepreneurship–the willingness to take risks and investigate large numbers of potential development projects rather than just those that have won the stamp of approval of a single central committee. It is the task of chancellors and deans, of course, to make sure that projects like CITRIS don’t wind up producing the drawbacks of both forms of organization: the strangulation by bureaucratic red-tape and committee infighting of government, combined with the restrictions on the distribution of information and the use of products that make a large share of private-sector development work duplicative of what has already been done.

Second, realizing the promise of the Societal-Scale Information Systems that are the Holy Grails of this quest will turn out to be a problem of social engineering as well as computer science. I have long wondered just why it was that the first half of the 1980s were the era of the IBM PC rather than of the DEC VAX–when the hardware cost of a VAX was, as best as I can guess, no more than 1/5 that of the equivalent number of 8086 machines, and when thanks largely to Berkeley UNIX there was no comparison at all in software. The answer lies somewhere in social engineering–that somehow paying out five times as much for inferior software was worth not having to wrestle with established MIS bureaucracies. But what the answer is I am not sure.

So let me turn this into a sales pitch for the social scientists at Berkeley interested in information technology–from Manuel Castells in sociology to Pam Samuelson and Mark Lemley at the law school to John Zysman and Steve Weber in political science to Hal Varian and his simians to Suzanne Scotchmer at public policy to the industrial organization and antitrust barons of the business school and the economics department–Glenn Woroch, Rich Gilbert, Dan Rubinfeld, Mike Katz, Carl Shapiro–and a host of others. I do not know of a place with a more vibrant and smarter community of scholars interested in the social engineering aspects of information technology.

And I do not know of a better place than this to assemble the resources to build the Societal-Scale Information Systems that can make information technologies realize their promise.

Fifteen Theses on “The Wealth of Humans” and “After Piketty”

Notes for the July 11, 2017 Research on Tap event:

  • Ryan Avent (2016): The Wealth of Humans: Work, Power, and Status in the Twenty-first Century
  • Heather Boushey, J. Bradford DeLong, and Marshall Steinbaum: After Piketty: The Agenda for Economics and Inequality

Meditations on Ryan Avent:

Ryan Avent: What will happen to ‘The Wealth of Humans’? “This really dramatic technological change… the digital revolution… is adding hugely to the amount of effective labor that’s available to firms…. A lot of routine tasks in factories and in offices… [to] be automated…. High-skilled jobs… use these new technologies to do work that used to require a lot more people to do and in the process are displacing workers… enormous, abundant labor…. Employer[s] with… huge reservoir[s] of willing workers at very low wages… say…. “I don’t need to invest in this labor-saving technology…. replace my cashiers with automated checkout… replace the people moving boxes in the warehouse with robots”. And so you get this sort of self-limiting technological change…. The more powerful the digital revolution… the more people… looking for low-wage work… the less of an interest firms have in using machines to replace them…”

  1. For the past thirty and the next thirty years—but probably not more—we are in all likelihood facing the increasing drift toward inequality driven by the rise of the Overclass as identified by Thomas Piketty. As long as the Overclass has enough control over the political system to manipulate it to reap enough rents to peg the rate of return on wealth—not physical capital, wealth—at 5%/year, we will see much if not all of the benefits from economic growth flowing to this Overclass, which will increasingly be an overclass of heirs and heiresses, rather than one that can claim that its wealth is due to some sort of meritocratic chops.

  2. For the past ten years and the next ten years—if not more—our biggest and principal problem has been an economy in secular stagnation afflicted by slack demand, and that in a high -pressure economy like we had under Clinton in the late 1990s or Kennedy-Johnson in the 1960s, most of what we see as our economic problems would not melt completely away but be much reduced. Robots and artificial intelligence were overwhelmingly seen not as problems but as opportunities in the high-pressure economy of the later 1990s.

  3. A generation ago we feared. But then we feared not the robot but the mainframe—and our fears of the mainframe then were like our fears of the robot now, save that while we now fear that robots will leave us with no work to do, we feared then that mainframes would leave us with no meaningful work to do and no work to do save being a mainframe-controlled dumb robot. As the Apple commercial said, we feared that 1984 would be like 1984: Those fears were vastly overblown: we did not become robots subordinated to mainframes; instead, microcomputers and the internet became our personal intelligent tools.

  4. The human brain is a massively parallel supercomputer that fits inside half a shoebox. It draws 50 watts of power. It is an amazing innovation, analysis, assessment and creation machine. 600 million years of proto-mammalian and mammalian evolution coupled with the genetic algorithm means that almost every single human can solve AI problems far beyond our current engineering reach—so much so that much of what our machines find impossible our brains find so trivially easy that we call such capabilities “unskilled”. When combined with our brains, human fingers are amazingly fine manipulation devices. And human back and leg muscles—especially when testosterone soaked—are quite good at moving heavy objects. Thus back in the environment of evolutionary adaptation, we used our brains, our big muscles, and our fingers to lead cognitively interesting—if stressful and short—lives.

  5. Back in the environment of evolutionary adaptation, we used our brains, our big muscles, and our fingers to lead cognitively interesting—if stressful and short—lives. Short: life expectancy at brith of 25 or so. Stressful: watching relatively young people die around you all the time is a significant source of stress. And in order for the average woman to have two children who survive to reproduce, the average woman would have had to have three reach adulthood, about four reach the age of five, about six live births, and about nine pregnancies—that’s the average. Up until 250 years ago, the average woman spent about six years pregnant and eighteen years breastfeeding. Some more. Some less.

  6. History has rolled forward since the hunter-gatherer age. And as history has rolled forward, we have figured out other things to do to add economic and sociological value than using our backs and legs to move things, our fingers to grasp things, and our brains to decide what to hunt and gather. Using backs to move heavy objects and our fingers to perform fine manipulations in cognitively-interesting ways has, relatively, declined.

  7. As our use of our backs and fingers guided by our brains to create value has declined, we have turned to: (1) turning many of us into robots ourselves, performing simple routinized repetitive and vastly boring tasks to fill in the gaps in value chains between the robots that we know how to build; (2) jobs as microcontrollers for domesticated animals and machines—the horse does not know what plowing the furrow is—(3) finding jobs as relatively simple accounting and software bots, keeping track of stuff, what it is useful for, and how its use is to be decided; (4) becoming personal servitors; (5) becoming social engineers—trying to keep all those things and all those people—especially, perhaps, trying to keep those brains soaked in testosterone—somehow working in harmony, somehow pulling together, although admittedly with limited success; and (6) remaining innovators, analyzers, assessors, and creators as well.

  8. Backs started to go out with the domestication of the horse. Fingers began to go out with the invention of the spinning jenny. But humans-as-microcontrollers, humans-as-accounting-‘bots—paper shufflers—and humans-as-the-robots we cannot yet build—took up all the job slack. Every horse needs a microcontroller. And a human brain was the only possible option. Even today, to a large amount every textile machine needs a human watching it at least part of the time. It doesn’t know when it’s gone wrong. It has no clue how to fix itself. It no more understands the idea of “fixing” any more than Alpha-Go understands that it is playing Go, and not just solving a problem of outputting a two-element vector in response to a 19 x 19 matrix of inputs with the additional structure that the output changes the matrix and that the possible matrices have a value-function structure.

  9. Now, however, we can finally peer into a future in which the microcontrollers and the accounting bots are on their way out in a manner analogous to the backs and the fingers. But this is our future. This is not our present. For the past ten years and the next ten years—if not more—our biggest and principal problem has been an economy in secular stagnation afflicted by slack demand, and that in a high -pressure economy like we had under Clinton in the late 1990s or Kennedy-Johnson in the 1960s, most of what we see as our economic problems would not melt completely away but be much reduced.

  10. What do we see when we peer into a future in which the microcontrollers and the accounting bots are on their way out in a manner analogous to the backs and the fingers? Fortunately, one thing this brings with it isthe forthcoming extinction of the the jobs that treat humans as simple robots: simple cogs in the machine that is Henry Ford’s River Rouge assembly line. Many occupations that vastly underutilize the massively parallel supercomputer that fits in half a shoebox are on the way out—and good: for those are not properly “human” jobs at all.

  11. Not yet, but starting soon, and continuing for perhaps the next hundred years, we face the deep problem of the obsolescence of human brains as resources that can be employed—or, rather, underemployed—to create substantial economic value. Over the past six thousand years, ever since the domestication of the horse, we have seen the erosion, at first slowly and in the past two centuries rapidly, of the obsolescence of human muscles as resources that can be employed to create substantial economic value. But we have benefited because human brains underemployed—as microcontrollers for domesticated animals and machines, and as relatively simple accounting and software bots—have nevertheless been of great and increasing value. But now our microcontrollers are better microcontrollers than human brains, and our software accounting ‘bots are becoming better accounting ‘bots than human brains. Not next year, and not next decade, but further out by some unknown time, humans’ jobs will be as: personal servitors, social engineers, and innovators, analyzers, assessors and creators. Here we might well, someday, have a huge problem.

  12. The market economy will amply fund AI research that replaces workers in capital intensive production processes by machines. Such industries have mammoth returns to scale. They thus tend to be characterized by large oligopolies. And so the firm that funds such labor-replacing research will capture with its own scale and in its own value chain a substantial part of the benefits of such R&D. That means that the combination of coming AI with a market economy might well be absolute poison for equity and equitable growth. It will race ahead with shedding workers in capital intensive production processes.

  13. The market economy will not amply fund AI research that assists and amplifies workers in labor intensive production processes. Such tend to be small scale. The inventors and the innovators cannot capture even a small part of the benefit in their own production processes and value chains. And intellectual property is a very weak reed indeed to rely on to fix the problem—in fact, intellectual property is more likely to be the problem than the solution, cf. Nathan Myhrvold, and Intellectual Ventures.

  14. The combination of coming AI with a market economy might well be absolute poison for equity and equitable growth. It will race ahead with shedding workers in capital intensive production processes. There will be—as Laura Tyson and Mike Spence pointed out in their contribution to Heather, Marshall, and my After Piketty book—a synergy between the dangers posed by the Rise of the Robots on the one hand and the inequality generating forces analyzed by Thomas Piketty in his Capital in the Twenty-First Century on the other.

  15. Technological progress could rescue us from Pikettyian dystopia. Robots could be intelligent tools. AI could be gold for equity: amplifying the capabilities of workers in labor intensive production processes would, as John Maynard Keynes once said, bring us vastly closer to economic El Dorado. Recall how a generation ago we feared not the robot but the mainframe—and our fears of the mainframe then were like our fears of the robot now, save that while we now fear that robots will leave us with no work to do, we feared then that mainframes would leave us with no meaningful work to do and no work to do save being a mainframe-controlled dumb robot. As the Apple commercial said, we feared that 1984 would be like 1984. But we are unlikely to see a repeat of the microcomputer revolution. Firms will not invest on a large scale in AI that amplifies the capabilities of labor in labor intensive industries. It will not happen unless some NGO does. How about an engineering school? How about an engineering school at a public university?

The Future of Education and Lifelong Learning: DeLong Opening DRAFT

Harvard Class of 1982 35th Reunion :: Science Center B :: Saturday, May 27, 2018, 10:45-12:00 noon

  • Seth Lloyd, MIT: Moderator
  • Brad DeLong, U.C. Berkeley
  • Ivonne Garcia, Kenyon
  • Noel Michele Holbrook, Harvard
  • William Sakas, CUNY
  • Carol Steiker, Harvard

In the spring of our freshman year, then-young economics professor Richard Freeman came to Ec 10 to tell us that going to Harvard would not make us rich.

He was wrong.

Up until 1980 America was winning, and Richard Freeman expected it to keep on winning, the race between education and technology: Thus there were ample numbers of people to take the increasing number of jobs requiring formal education for first class performance. Thus the amount the market paid you extra for taking a college requiring rather than a high school requiring job was modest: 30% or so–not enough to make up for the income you would’ve earned, had you taken the tuition you would not have spent and the extra wages you would have made from working, and put them into some reasonable investment.

But after 1980 America began to lose the race between education and technology.

The expansion of American higher education slowed massively. Higher education for native-born males simply froze in its tracks. As a result, in the world in which we have worked for the past 35 years employers have been betting up the relative price of college graduates: Rather than making 30% more than our counterparts who went straight into the job market after high school did, we have on average received double.

The freezing and of the relative numbers of native born American males taking advantage of hire education as demand, supply, and heterogeneity components.

On the demand-side, states withdrew tuition subsidies. Public college ceased to be free. Those whose parents were not rich worried about their student loans: what if they didn’t succeed and finish and could not get one of those high paying jobs? How were they going to pay back their loans? Americans almost surely over worry about this. But people are who they are, and not who economic theory dictates they should rationally be.

On the supply side, states stopped building campuses. Getting the courses you wanted and needed at public universities became iffy: five or six years rather than four.

And on the heterogeneity side, our colleges are designed for those who take to print literacy and to Arabic mathematics like ducks to water–if you do not have that, or are not trained to have that, learning the way we are taught to teach becomes much more difficult. We economists see this every semester, as even Ec 10 requires great facility in reading, in arithmetic, in algebra, and in algebraic geometry. The extra slice of the population that we would have been sending to higher education in a better counterfactual world in which America had not lost the race between education and technology would have been less well prepared and less suited to benefit.

What is the balance between these supply, demand, and heterogeneity considerations? That, we say, is a research problem.

How important is all this? I would say that about 1/3 of the problem is with America that have developed over the past 35 years–1/3 of the ways in which I see America today falling far short of what I confidently helped America would be by now–are due to our losing the race between education and technology.

Let me make one final point: Over the past generation, Harvard has not helped. We had 1600 in our class. Last week’s graduating class was essentially the same size. Worldwide, between five and ten times as many people are well-qualified to join my niece as freshmen this fall. In our class there were perhaps four times as many people well-qualified to attend as Harvard admitted. Today there are between twenty and forty. Yet Presidents Bok, Pusey, and Rudenstine seemed to have little interest in helping America and the world in the race between education and technology. Contrast that with the University of California, which, under Chancellor and President Clark Kerr and California Governor Pat Brown, set in motion the plan to clone itself across the state and increase enrollment tenfold.

If you are thinking about giving money to help America win this race with education and technology, I would not recommend Harvard. U.C. Berkeley, Columbia, and MIT for moving people whose parents’ were in the bottom quintile into the top 1%. And for overall bottom fifth to top fifth mobility? CUNY. U.T.-Pan American. TCI. SUNY Stonybrook. Pace. and Cal State-LA. That is what Yagan, Turner, Saez, Friedman, and Chetty say…

Inclusive AI: Technology and Policy for a Diverse Urban Future

Inclusive AI: Technology and Policy for a Diverse Urban Future Wed, May 10, 2017 10:30 AM – 5:30 PM

Cursor and 1024px The Garden of Earthly Delights by Bosch High Resolution jpg 1 024×583 pixels

Panel 3: The Future of Work: Automation and Labor

  • Ken Goldberg
  • Brad DeLong,
  • James Manyika
  • Costas Spanos
  • Laura Tyson
  • John Zysman

NewImage NewImage NewImage NewImage NewImage NewImage NewImage NewImage NewImage

Since I get to go first, I will preemptively take the hyper-Olympian and very long-run historical point of view…

The human brain is a massively parallel supercomputer that fits in half a shoebox. It draws 50 W of power. Human fingers are amazingly fine manipulation devices. Human back and leg muscles—especially when testosterone soaked—are quite good at moving heavy objects. And so, back in the environment of evolutionary adaptation, we used our brains, big muscles, and fingers to lead interesting, if stressful and short, lives.

But as history has enrolled we have done other things to add economic and sociological value than use our backs, our fingers, and our brains to innovate and create. Over the long historical sweep, backs and fingers have declined and we have turned many of us into, instead:

  • robots performing repetitive tasks,
  • microcontrollers for domesticated animals and machines,
  • relatively simple accounting and recording software bots,
  • personal servitors,
  • social engineers trying to keep all those things controlled by brains—especially by the testosterone soaked ones—working together harmoniously. With limited success.

while remaining innovators and creators.

Backs started to go out with the domestication of the horse. Fingers with the invention of the spinning jenny. Microcontrollers and accounting ‘bots, we can see, are now on the way out too. So, fortunately, are the jobs that treat humans as simple robots.

That leaves us with a future of work made up of:

  • personal servitors,
  • social engineers,
  • innovators and creators.

Utopia or dystopia? Heaven or hell?

Over to you, James. And, in a broader sense, over to all of you—in the audience, and out there in Internet land…

Trade, Jobs, and Inequality

Eduardo Porter: Good evening. Thanks for coming. There has been a lot of sturm and drang over the last—what?—98 days. Today I read that the administration is preparing an executive order to start the process for the U.S. to leave NAFTA, which was one of Mr. Trump’s campaign promises. However, over the past 98 days I have come to realize that perhaps President Trump did not really mean all of the things he said on the campaign trail.

He has left China off the hook: “not a currency manipulator”. His first trade battle was against… Canada. And the first battle against Mexico he lost, today at the WTO. And yet I do not think that you would be right to discount the “Trump phenomenon”. There were tens of millions of voters who actually bought into his argument that immigration and trade have done them wrong, have taken their jobs, have weighted on their wages, have caused some of the social dysfunction that we now see in many American communities.

Here we have this group of pretty remarkable pros to talk about whether there is a case against trade. To what extent is Trump’s diagnosis correct? Is there an argument for some sort of protectionist policies to support the workers who have been displaced from their jobs, or whose wages have not improved due to competition from other countries? Perhaps Trump’s prescriptions have been pretty outlandish—slapping a 35% tariff against the world’s second largest economy, or withdrawing from NAFTA and undermining the value chains that have been built over the last 25 years, or even abandoning the WTO. This is something I would like the panel to address. But also softer stuff: like talking about using safeguards against import surges or other tools used in administrations in the past. Our upcoming trade representative thirty-five years ago was negotiating voluntary export restraints with Japan in the way of using protectionism to assist workers.

Without belaboring this further, I would like to give each of you guys an opportunity to address these topics and get this going. If I may start with you, David Autor, you have written some very important papers addressing the impact of trade on workers and your last name begins with an “A”, so we can do this alphabetically:

David Autor: Thank you very much. It is a pleasure and an honor to be here, to be on stage, to be a guest of Paul Krugman and CUNY, and I see many of my friends far more distinguished than I in the audience as well. I am going to try to set the stage. I will not talk about trade policy per se. But I will put it in the context of the dramatic changes in the U.S. labor market that have occurred over the past 35 years.

From 1980 to the present we have had a rapid rise in inequality, and a lot of that rise in inequality is divergence in earnings by education level, and a lot of that divergence is not simply higher earnings for higher education adults but falling real earnings levels for lower-education adults, particularly non-college men, particularly men with just as or with less than a high school degree.

A lot of that phenomenon, in our consensus understanding, has to do with technological change that has increased the demand for highly educated workers and those with judgment, expertise, and creativity, and reduced the demand for repetitive physical and repetitive cognitive labor. So over the last 35 years technology has been more consequential for the divergence we have seen. And up until the mid 1990s I think many economists would have said: end of story.

But things really did change in the 1990s and 2000s, and that change had a great deal to do with China’s economic development which, let us be clear, is a fabulous thing from a global perspective. This has been the best thing for the global middle class in a millennium. It has created prosperity throughout the world not just in China, but in sub-Saharan Africa, Latin America, and so forth. It has been a great thing. But it has been extremely disruptive for U.S. employment, especially U.S. manufacturing employment.

I myself am actually startled when I look at the figures. Many of us have in mind the picture that U.S. manufacturing has been in perpetual decline since 1943, when it was 38% of U.S. employment toward the end of the Second World War. It is true: if you look at the share of manufacturing employment, it is just down, down, down. You cannot even see the early 1980s recession. You can barely see China’s rise.

But if you look at the number rather than the share of U.S. manufacturing workers, you see something very different. In 1943, in the Second World War, there were 16.6 million U.S. manufacturing workers. By 1979 there were 19.7—not that much change. In 1999 there were 17.7—again not that much change. But by 2007 U.S. manufacturing employment had fallen by three and a half million jobs. And in the subsequent three years between 2007 and 2010, it fell by another one and a half million jobs. Five million jobs lost—almost a third of U.S. manufacturing employment—and that is a really traumatic shock.

It really falls off a cliff. And it does have a great deal to do with the rising position of China. Not the fault of China. All of a sudden we have a very competitive low-wage country making high-quality goods that are a better deal for consumers. Consumers start substituting toward them. That raises consumer welfare—it makes lots of goods cheaper—but it also has a very concentrated impact on the places that make those things. Manufacturing by its very nature is geographically concentrated. When we are talking about commodity furniture we are talking about Tennessee. When we are talking about textiles we are talking about the Carolinas.

Five million jobs in a labor market of 150 million really isn’t very many. But when it happens in a few places in a very short period of time—when all the firms in an industry go under at once—it has a seismic impact. We have always known that more trade has diffuse benefits and concentrated costs. That is true and theory. In the 2000s we really saw it play out in practice It is understandable that people do feel like this was unexpected, this was unwelcome, they were not warned, they were told that trade was win-win. Things did not work out as expected.

Brad DeLong: Certainly I have learned a huge amount from David Autor. Certainly I have had to substantially revise upward my own views about the negative U.S. domestic consequences of trade and the China shock that hit the U.S. starting the late 1990s as China’s industrialization shifted from something impressive for a very poor country to something that was of world historical consequence. But I would like to lay down a marker and put David’s remarks a little bit in perspective. And I would like to erase what I think might be somewhat of a misapprehension.

When you talk about the number of manufacturing jobs going from 19 million to 17 million from 1979-1999 and then down to 12 million, those aren’t the same manufacturing jobs. There is enormous churning within manufacturing. There are enormous regional shifts as well. My grandfather had to close his shoe factory in Brockton, MA in 1933 at the nadir of the Great Depression and move up to and reopen in a place where wages were even lower: South Paris, ME. The workers of Brockton were killed by this. This was taking pace at the same moment that the rest of manufacturing in southeastern Massachusetts was beginning its move down to the Carolinas. The Lord Brothers Shoe Company was a bonanza for the workers of the South Paris until 1947, when they hit the wall again.

They took the capital and split up to become real estate developers.

That was not a big shift in terms of the kinds of people employed—the net factor content of domestic employment. The Wellman-Lord Construction Company that built turnkey phosphate plants and other things in Florida employed very much the same kinds of people in terms of skills and education as had the Lord Brothers Shoe Company in South Paris, ME. But these were people in Florida rather than Maine.

There is always churn. That’s the reason I think that looking at the share—which shows a smooth decline—yields much more insight than looking at absolute numbers. Absolute numbers seem to say things were stable for a long time, and then they collapsed . What was actually happening was that the jobs were opening up and closing down the economy at all times in different regions. People were moving and shifting about. Manufacturing was not stable. New England manufacturing got killed—and New England workers along with it. Carolinas manufacturing boomed—and that was a bonanza for Carolina’s workers.

Looking at the share, you see a long decline in the share interrupted by sudden short-term sudden shocks which cause considerable distress. To throw out some numbers:

  • We were never going to permanently have 38% of our nonfarm labor force in manufacturing: that was only if we were building little but bombs and tanks.

  • The post-World War II normal was about 30%.

  • if the United States had been a normal post-World War II industrial powerhouse economy like Germany and Japan, by now the evolution of technology would’ve carried us down from 30% to about 12%.

  • We have gone from 12% to 9.2% because since the inauguration of Ronald Reagan, the United States has on the whole followed very strange and dysfunctional macroeconomic policies, policies that have made us a savings deficit rather than a savings surplus country. Instead of the United States as a rich country financing the industrialization of the rest of the world and developing economies using that financing to purchase US manufacturing exports, we have been instead acting as… a money laundering center? a provider of political risk insurance? a place to put your money so that if the balloon goes up in your country and you have to flee suddenly in the Learjet or the rubber boat you will be much happier if a large chunk of your wealth is on deposit at J.P. Morgan Chase in New York.

  • Morever, if you are not a rich person in a developing country but a developing country, and if you want to avoid being subject to the tender mercies of Christine Lagarde of the IMF, it is nice to have large dollar assets in your government’s hands to use in an emergency.

  • Then from 9.2% to 8.7% is because of changing patterns of trade, primarily the rise of China.

  • From 8.7% to 8.7% because of the result of NAFTA: NAFTA is a nothingburger as far as the fall in the manufacturing share of the workforce is concerned.

  • From 8.7% to 8.6% because of bad trade deals we have made with China, with Mexico, with Canada. They’ve taken us to the cleaners, I tell you! Over three generations they have taken us the cleaners! Bigly, they have taken us to the cleaners!

And of course, there are substantial and powerful countervailing gains in other sectors that we have gotten from the trade negotiations that have led us to shed the extra 0.1%-point of our manufacturing labor force.

I would sharply distinguish that 0.1%-point from the 0.5%-point that is due to trade, primarily the rise of China, and from the 2.8%-points that are due to the trade impact of dysfunctional macroeconomic policies, and from the 18%-points that are due to technology.

Eduardo Porter: So, Ann, is trade not it? Is trade a red herring?

Ann Harrison: Let me step back.

Let me try to negotiate between these two men to either side of me. Donald Trump won the US election by convincing voters in states like Michigan, Ohio, Pennsylvania that he would make America great again. What was his platform?

Brad DeLong: If I may say: by convincing three million fewer voters.

Ann Harrison: Yes. Let us recall that he did not win the popular vote.

He promised to impose 20% tariffs on imports. He promised to build a wall to keep out Mexican immigrants. He promised to renegotiate and perhaps throw out NAFTA. On Sunday we saw an almost-repeat of the Trump campaign in the first round of the French presidential elections. Marine le Pen has made it to the final two on a far right platform. Her platform was to take France out of the European Union. So the question is: why are such extreme anti-globalization platform so appealing?

I am going to throw out my own numbers:

Since 1984, when we had 25 million manufacturing workers, half of those workers have disappeared from manufacture. We have gone from a country where one quarter of workers are in manufacturing to one in ten workers. One of the major reasons that this decline is important is that those jobs are in fact”good jobs”. In my research I follow workers were kicked out of the manufacturing sector and move to services. Typically that will lose them 5% of their wages. However if that worker loses his or her job in manufacturing because of the trade shop, he or she loses 15 to 20% of their wages. You are losing jobs, and you are losing good jobs. The important thing is that the pain is real. We are seeing a shrinking middle class. We are seeing stagnant wages for less-educated middle-age workers. And we are seeing an enormous rise in inequality—higher than we have seen in the last 100 years.

In addition to that—and this is very important—we are seeing a rise in insecurity. Even people who still hold on to jobs are afraid for the future. They are afraid about whether their Kit is going to get into college. They are afraid about whether their kid is going to get a job at all. They are afraid about whether they are going to lose their job. That rising insecurity is really important.

Then we have to ask ourselves: somebody like me, an economist who has always supported free trade, did we miscalculate the gains from globalization? I think we made two big mistakes

  1. The first mistake we made was that we thought it would be really easy for people to retool and find new jobs. It turns out that is not so easy. We have a record number of people who have left the labor force.

  2. The second mistake is that we thought it would be easy to compensate the losers. The idea behind globalization is the winners, the exporters, the consumers, are so much richer that it is easy and straightforward to redistribute some of the winnings to compensate the losers. That turned out not to be true. Our package for helping the losers, Trade Adjustment Assistance, helped only about half those that it should have helped. But, much more important than that, Americans do not want handouts. What they really want are jobs.

That leads me to the next question: Is barricading ourselves against China and throwing out NAFTA part of the solution? This is where I am going to deviate from Professor Autor and support my former Berkeley colleague Professor DeLong. Us Berkeley types stick together.

China is a convenient scapegoat. It reminds me of Japan in the 1980s, when Michael Crichton described a “yellow menace”—very racist—in his novel The Rising Sun. I think that is what we are facing.

Why do I think protectionism won’ help? I have a new book, The Factory-Free Economy, with a coauthor and friend named Lionel Fontagne. I want to tell you a story from the very first chapter written by Richard Baldwin. Go down to South Carolina and go into a textile mill, and the only thing you will see is a man and a dog. The dog is there to protect the machinery. The man is there to feed the dog.

That tells you—this is just an anecdote—that China is not the enemy. The enemy is the machine. Let me give you some more facts to bolster my argument and this is where I’m going to support my colleague from Berkeley. My own research has tried to decompose what has happened to manufacturing jobs. My research shows that maybe 10% of the loss is due to firms going offshore. Three quarters of it can be attributed to the fact that the cost of machinery has fallen, and firms are substituting machines for people.

Let us look at the long-run trends here. Let us go back 50 years, 50 years, Long before China came on the scene, the share of manufacturing value added in the economy has stayed completely flat. It has not moved. What has changed very gradually is the share of manufacturing workers, and here again I would agree with my colleague to the left. You do not see a big decline around 2000. There is a decline but it is not enough to materially shift the trend. You see a gradual decline in the share of manufacturing workers. How can we have Constant share of manufacturing value added in GDP while the share of manufacturing workers in employment falls? The answer, once again, is machines. Rising productivity as machines are replacing workers.

Let me summarize:

  • The pain is real.
  • The solution is wrong.
  • It’s not that winter is coming. It is that the robots are coming.

David Autor: I just want to make it clear. I am not advocating trade barriers of any sort. That maybe what you took for my remarks. The China shock was real. It had a large effect. That does not mean we should try to or can turn back the hands of time. Paul wants to say more about this.

Paul Krugman: Instead of arguing with my fellow panelists, I am going to argue with my past self.

Two propositions:

  1. Economists were much too complacent about trade even up until the fairly recent past.

  2. The political system and journalists are way overreacting to trade now.

These are not contradictory. Both need to be in there now.

There was a late 1990s consensus about trade and the economy—that trade was a contributing factor but a very distinctly secondary factor and that if you got the numbers right it was not that big a deal. We should really be focusing on inequality via other programs. Trade was not the issue. If you have to ask: Who was guilty of what turned out to be a bad judgment? The answer is: me.

I was taking standard trade models. I was looking at factor content. I was concluding: look, it’s just not big enough to be having the effects non-economists are ascribing to it.

A couple of things happened. One was that trade went up a lot—and a different kind of trade. It was not just China. If you take a long time series of trade as a share of world GDP, you find that there is a big dip in the interwar period, and by 1980 it does not look that different from 1913. And then something happens. Hyperglobalization. There is this take-off of trade, this geographic deconcentration of value chains, and it is not just China.

Thus a lot of estimates that trade was just not that big a deal were using data from 1993. And guess what? Data from 2015 shows that trade with low wage countries is a much bigger thing.

And then there is the specific falling-off-a-cliff thing of the 2000s. Partly this is a statistical illusion because the rate of growth of the labor force slows a lot. The baby boomers are all in the workforce. The movement of women into the labor force was over. Thus what had been a continuing decline in the share became an absolute decline in the number of manufacturing workers.

There was a big move into trade deficit in the 2000s. A lot of it was emerging market money flooding into the United States in the wake of the financial crisis. In economics, everything is connected to everything else in at least two ways. There is this funny backwash from that. So trade becomes a much bigger deal than it was.

The other thing—which I really kick myself for missing—is what David Autor and coauthors have pointed out: workers are a lot less fungible, and a lot less mobile, a lot less able to shift jobs then we had imagined or would like to imagine. Manufacturing industries are geographically concentrated. When you get a shock you can’t just think as if all workers in the economy are the same. When all the export-oriented workers in some midwestern town have lost their jobs, that matters, the effects are bigger.

On the other hand, it is a dynamic economy. Stuff is always happening. I used the number 75000 for the number of Americans every working day. I got some internal mail from the Times—people saying: that’s ridiculous, that can’t be right, that would mean that 20 million people lose their jobs every year. Yes. 20 million people lose their jobs every year. There is a lot of churn. Things are happening all the time. More retail jobs were lost in the last two months than coal jobs were lost since 2000.

Manufacturing, because of the geographic concentration, does matter more. But disruption of manufacturing centers has been happening for a long time, that is not a new phenomenon, that does not have to be related to international trade. Walk around here and there is even a statue of a garment worker at his sewing machine. This is called the “Garment District”. There is no Garment District here anymore. There are just memories of the Garment District. That disappears, but—and I am not sure where I come down in how we deal with it—we can do the numbers. We can look at objective facts. We can say that it is only one of a number of factors. But it always has a psychological significance that makes it much bigger.

The fact that foreigners are involved makes people always emphasize trade shocks much more than corresponding alternative shocks coming in. People get upset about the disappearance of steel plants in a way they never got upset about the disappearance of the buggy manufacturers. Somehow, the sense of unfairness—and I think to a little bit economists have a reverse problem. Comparative advantage is one of the prides and glories of our profession. And so we overemphasize the good sides of trade as well. That puts us in an awkward position.

Last thought—and I am not offering any answers here—if we think that a large part of the solution is a stronger social safety net, think about the fact that France has a welfare state, a social safety net, that is beyond the wildest dreams of American leftists. Nonetheless le Pen made it into the second round of the election.

Eduardo Porter: Thanks. Just a word in defense of David and of China. I did my last column about whether China should have been labelled a currency manipulator. I was speaking to Brad Setser. He said that there was a “before David Autor” and an “after David Autor”. Because of your work identifying the local impacts of competition from China.

Paul Krugman: I just want to say that I brought multiple columns saying that China was a currency manipulator, which was true in 2010 and 2011, but is not true now…

Eduardo Porter: Part of this conversation should also be about a set of actions that the Trump administration. If trade is at best overblown as a cause of distress, what would be the costs of some of the actions that the Trump administration is proposing?

Brad DeLong: Abrogating NAFTA, for example? I would say full NAFTA abrogation—returning us to 1992—would cause a world of hurt to the American auto industry, as well as to other sectors. The argument back in 1992—Ross Perot on the right and my various labor friends on the right—was that the auto industry was in the crosshairs and gravely threatened by NAFTA. That turned out to be simply wrong.

Sherman Robinson of IFPRI and PIIE have convincingly demonstrated that that was simply not so. NAFTA enabled the creation of one of the first important sophisticated global value chains. After NAFTA, Detroit could reconfigure its work processes and so shift out the “bad”, low-productivity manufacturing jobs to Mexico—the jobs in which we use people as essentially robots because the real robots we can build are not quite good enough—while keeping the high value skilled and semi-skilled manufacturing jobs. Without Mexico there as a resource to draw on in the value chain, Detroit would have a greater cost disadvantage vis-a-vis Nissan, Toyota, Honda, Subaru, and BMW, Mercedes, Audi.

If you abrogate NAFTA, that goes away.

GM, Ford, and Chrysler are then in a much worse position.

Paul Krugman: I do not have an estimate of what Trumpist trade policies would do…

Brad DeLong: And neither does he…

Paul Krugman: We don’t even know what the policies are.

There has been a fair bit of work on BREXIT. We have a better idea of what BREXIT will actually involve. Something like 2% poorer in the long run. There is a dirty little secret: the benefits of trade and the costs of protection are not trivial, but they are not as big as the rhetoric might tend to make you believe. An abrogation of NAFTA would be something like that.

However, here the Autor point cuts in the other direction. NAFTA is not a new thing. The North American economies have adjusted to it. We have an integrated North American production system. A car assembled in Ontario is made from stuff produced all over the continent. Disrupting that would do the same kind of damage as the China shock. The old joke about the motorist who runs over a pedestrian, says “I’m sorry”, and then backs up and runs over him again. The big changes in trade are all in the rear view mirror. The China shock ended around 2008. The big changes from NAFTA ended a decade before that. Now we are talking about taking an established industrial order and disrupting it.

Eduardo Porter: Guys, folks are going to come around to collect the question cards. The cards will then be brought forward here.

David Autor: Let me try to bring this around to trade and automation, which Ann correctly emphasized. Automation has been more important. In the future, automation will likely be more important again. The China shock was hugely disruptive, but it is much closer to equilibrium now. With no change in policy we will not see anything like what we saw over the past 15 years in the next 15 years.

But we ought to learn some lessons about this. One is about our social safety net. As emphasized, trade adjustment assistance policy is woefully inadequate. But a deeper point is, jobs have their own value. You cannot make someone whole… What if you said, “Hey, Paul, we are going to take away your identity. You are no longer an esteemed economist. You are just retired”. Would you say: “Oh, great! I have all this money and I do not have to do anything!”? Of course not. For most people, work is central to organize who they are, how they perceive themselves, how others perceive them, their social identity. A better social safety net is not sufficient. We would like to actually have good jobs.

Are there other ways to go about that? Tax reform. The whole Border Adjustment Tax was a Trojan Horse for a Value Added Tax, which most other industrialized countries have. That treats imports and domestically produced goods symmetrically. Our current tax system levies a lot of taxes on domestically produced goods and only sales taxes on imports. Moving toward a better form of taxation could be beneficial for investment and for the repatriation of profits to the United States. Similarly, we could continue to invest in things that are innovative. We should be concerned about manufacturing not just as a jobs program but as a center of innovation. A lot our patents and a lot of our IP occurs. That creates a lot of wealth and employment. We would be a poorer place and we are going to be a much poorer place if we cut the National Science Foundation and the NIH, if we stop investing in our great public universities. We are really at risk of making that mistake. Severely.

The one thing I disagree with: I don’t think we should fear the robots. Automation has been with us for 200 years. It has made us rich. It has made us more productive. We lead safer, more interesting, longer lives with much higher standards of living because of the advances we have made. It is also disruptive, like trade. But it does not happen as fast. We are living in an era where people are very concerned that automation may all of a sudden change a lot of things very quickly. So far that has not happened at all. In the time since 2000, the U.S. has added 15 million jobs—as many stories as have been simultaneously written about how the robots are taking all of our jobs.

The productivity data do not support the view that we are in this productivity transformational surge. We may be—we may be at an inflection point. But it has not happened yet. We should be preparing for how we can benefit from the complementary and productivity. It is coming. We want to make sure that we are making robots here, and not buying them from China or Germany, because we are going to be using them 20 years from now. We would also like to be making them. We want to be thinking about how we take advantage of these opportunities, not how we make them go away.

Paul Krugman: I also wanted to raise this issue. You could argue that there are too few robots around—that is what the productivity statistics seem to be saying.

Ann Harrison: I want to make just two points about what would happen if we transitioned to a more protectionist state of things. One is that 95% of the world’s markets are outside US borders. A company, any worker who works for an exporting company, anybody who has to think about the rest of the world—for all of them, closing in is really not a good idea even from a narrow business perspective. That is something Trump understands.

The other more important point is that the 2 billion poorest people, the people who live on less than two dollars a day, they live outside our borders. These people in the rest of the world, particularly the Chinese, have benefited massively from the opening up of world trade. A prominent economist sitting in the front row here, Branko Milanovic, has written a very important book showing that even though inequality has risen in every single country, global inequality has fallen. Why is that? Because the poorest countries, particularly China, have been able to use access to world markets to catch up. I just do not see how we can turn our backs on the 2 billion poorest people. That is what I wanted to say about the costs of protectionism.

I want to say one more thing: the hard part is what we will do going forward. It would be great if we could discuss a proposition: Both Bill Gates—whom I think of as middle—and by somebody more on the left, the late Anthony Atkinson and his fifteen points about what should be done. Coming from very different perspectives, one a billionaire and one an Oxford Don, they both agree that what we should do is to tax machines—or find other ways to promote labor-intensive growth.

David Autor: That’s a terrible idea.

Paul Krugman: There are two things I want to say. First, in saying “don’t fear the robots” you are missing “Skynet kills us all”, but let us pat that aside.

Second, in raising this, Ann is making a point that I very much agree with but I am not sure what to do with. If we step back from a U.S. or an OECD perspective and take a rootless cosmopolitan global perspective, hyperglobalization has been an incredible force for good—and it is not just China. In fact, in some ways China is the old story. You want to be thinking about those that are not as far along. If you go to Bangladesh you are horrified: they are very poor, miserable conditions, they have industrial disasters that make the Triangle Shirtwaist fire look like nothing, and yet they are a country that was on the edge of starvation when they achieved independence and has achieved a doubling of real incomes, and it is all because of their ability to export labor intensive manufactures. They are not a banana republic, they are a pajama republic. That is terribly important. That is a reason to keep markets open.

But can you imagine running a U.S. national campaign saying: “Look, your communities are being gutted, but we have to keep these markets open for the sake of the people in Bangladesh”? I don’t know how we deal with that. I really don’t know how to think about that.

Eduardo Porter: If we could think about trade policy into the future. Is there a role for reviewing trade policy, in both the domestic and the global perspective. On the domestic side, you have folks like Larry Summers—hardly a crazy radical—writing pieces about how we should rethink the institutions. We should stop focusing on trade deals, and should focus on other things, like labor rights and institutions…

David Autor: Like the TPP, actually…

Eduardo Porter: And from the global perspective, the issue that globalization has helped a lot of countries move up from poverty is certainly true, but now there is a question about whether that will still be true in the future. You have the work of folks like Dani Rodrik wondering whether the manufacturing economy will be able to do the trick that it did for countries like China.

Brad DeLong: China may well be the last country able to use the standard strategy of export-led industrialization to a complaisant importer of last resort to build up its communities of engineering practice and so become much richer faster. A huge question. Let me say that everybody should read Richard Baldwin’s new book, The Great Convergence.

Paul Krugman: This ties into a huge number of issues. We used to say—I used to say—when we had the original Gang of Four industrializers, Korea, Taiwan, Hong Kong, and Singapore, that: “OK, you can do this, but how much labor intensive manufacturing production is there in the world?” Would it be possible once China tries to do this? And it turned out that it was possible.

Brad DeLong: Barely…

Paul Krugman: But it was possible, and that was because we learned to divide up the value chain, and take labor intensive segments and hive them off. So far—there are still a few countries, Bangladesh, Vietnam, behind China in this—there may be limits. And it doesn’t seem to work everywhere. Why has not Mexico had the takeoff we expected it to have?

Brad DeLong: One thing that it definitely is our job to do as academics and economists is to very proudly fly our rootless cosmopolite freak flag whenever the situation and the evidence calls for it. We need to be saying that a century from now historians will be writing that one of the glories of the American age was that the United States did not view its proper role as hobbling industrialization, growth, development in the rest of the world but rather in encouraging and supporting the making of a much richer world as fast as practicable. An open economy, a liberal polity—that has made the era since 1945 so great.

I find myself wondering—Pascal Lamy, at a conference I was at last November, quoting China’s sixth Buddhist patriarch: “when the wise man points at the moon, the fool looks at the finger”. There was a Hungarian Jewish sociologist 70 years ago, Karl Polanyi, who wrote that a market society was bound to destroy itself. In a market society and the only rights that matter are your property rights, and your property rights are only worth something if they give you control of scarce resources useful for making things rich people have a serious Jones for. But people Believe that their rights extend far beyond their property rights. The way Polanyi put it, people believe that they have rights to Land whether they own the land or not—that the preservation and stability of their community is a right they have. People believe that they have rights to Labor—that if they play by the rules and work hard the market owes them that they should be able to earn the standard of living they expected. People believe that they have rights to Finance—that the firms they work for and the jobs they have should not suddenly blow up and disappear because financial flows have been withdrawn at the behest of some sinister gnomes of Zürich or other tribe of rootless cosmopolites.

This is a huge problem. People believe that they have much stronger rights than the ones a market society gives. But then you try to move beyond a market society to a social democracy with a safety net. And the response is: we don’t want welfare programs. Back in the 1920s “welfare” was a good word. When Edward Filene in the 1920s talked about “welfare capitalism”—firms providing health, accident, and pension benefits to their workers—the “welfare” was in there to make his readers think that the idea was a good thing. But because people want respect, over the past century the word “welfare” has been poisoned.

Dealing with this is one of the major political-rhetorical-economic challenges of our time. As an economist, I want to pass this buck to some sociologists or political scientists somewhere.

David Autor I want to say a little more about why I am not in favor of taxing robots. Unlike carbon emissions, robots do not create direct negative pollution externalities. This would be taxing something that makes us more productive. This would be like taxing computers because they eliminate all of those clerical jobs, or taxing cars because they are displacing all these equestrians end for farriers. It would’ve been a terrible idea—not just making us poor in the short run but keeping us poorer in the long run by retarding economic and industrial development and innovation.

Ann Harrison: Did you just say it was a bad idea to tax cars?

David Autor: I did. To treat them specially and differently. I think that our tax system favors labor over capital in an unfortunate way. They should be treated symmetrically. Rapid expensing and so on is actually a poor idea, but to single out an area of technological advance and say that we are going to punish it because it could be disruptive—that is a way of making us poor in the long run. And I want to reiterate: Robots will be a really large part of production. We had a technological advantage: a lot of the technology behind robot started in the United States. It is now migrating elsewhere. We should be investing in that as a country as we have done at a societal level with so many other technologies. We should not be waving goodbye because we are scared of what it may do.

Brad DeLong: We used to have what, 500,000 women working at telephone switch boards plugging cables into sockets. Now there is nobody in the United States saying: “Gee. I wish I could get my grandmother’s job at the telephone switchboard at Con Ed! That was such a great job!” That was a job that took human being with a very sophisticated brain—one that has evolved to do everything that a hunter gatherer does plus everything else we have built on top of that—and that uses that brain only—and uses her as a massively underemployed robot because making circuits by inserting plugs in the sockets uses only a very small proportion of the intellectual capacity that is the massively parallel supercomputer that fits in the bread box and draws 50 W of power that is her brain for an eight hour shift.

Paul Krugman: Part of the question is: what is a robot? Is a machine learning algorithm that these days makes Google translate so startlingly good a robot? These days are self-driving vehicles robots? If I wanted to think about a technological change the could displace lots of workers in the near future, we have 5 million people in the United States employed as drivers. That could go away in a heartbeat. But on the other hand—third hand, fourth hand, amongst my hands—the long-run optimism about the impact of technological change is born out by history, but the long run can be very long indeed. Wages stagnated for at least 40 years during the Industrial Revolution, and 40 years from now—a lot can go wrong in 40 years.

Eduardo Porter: This conversation was focused on manufacturing. But automation and whatnot are making inroads into the service economy, which employs a lot more people.

David Autor: Software is a much bigger deal than robots already. It will probably continue to be. Software is the hamburger that eats the world.

Eduardo Porter: We have got 15 minutes for questions. I would like to start with this, which seems a great topic for you people to take on. Let me start with this one. The belief is that service sector jobs will always pay less than manufacturing jobs. But is that guaranteed? Should we put the government’s energies into making it so the service sector jobs are highly paid and good jobs rather than focusing on manufacturing, which will always be a small part of employment in the future?

Brad DeLong: I don’t think that “manufacturing” and “services” are the way to partition it. If you asked me to partition it, I would say that there are people who add value with their strong backs, but that began go go out with the domestication of the horse. And there are people who add value with their nimble fingers, but that began to go out with the invention of the spinning jenny. However, because every horse, every steam engine, every spindle, every piece of machinery needs a microcontroller, human brains added an awful lot of value as bosses of the machines that had replaced their backs and fingers. And there is accounting: keeping track of stuff, and who owns the stuff, what the stuff is good for, who is allowed to use the stuff, where the stuff is, information, organization. The problem now is that robots proper are getting rid of the microcontroller jobs and software ‘bots are about to start getting rid of the accounting another white collar Jobs keeping track of the stuff and what it is good for. The graduate admissions committees whose jobs could be better done by an algorithm…

Paul Krugman: We doing know that live job interviews make a fundamental contribution: they destroy information…

Brad DeLong: You’ve seen economists go on the job market. People have long records of what they have done and what they have worked on and what their advisers and peers say about them. And that gets an interview in a hotel room for 30 minutes. And then with the five people in the hotel room for 30 minutes say about you wipes out all previous information. And if the five people turned from’s up then you get if fly out and get to give a seminar. And then what happens in that seminar wipes out all previous information…

David Autor: I think that what Eduardo was asking about was not “service sector”—which is 80% of everything—but rather personal services, helping, assisting. Those are rapidly growing. Those are low paying. Those are low paying in every advanced economy…

Brad DeLong: Sheryl Sandberg and other managers—it’s a social-engineering-organizing job…

David Autor You added “social engineering”: I was talking about food service-security-cleaning-home health aides. 15% of employment. Very low wage. They are intrinsically low wage because they use a very generic skill set. You can be productive in those jobs in a couple of days without a lot of training because labor is not intrinsically scarce for them, they cannot be high wage jobs. They tend to be at the bottom of the ladder. You can cause there to be higher wages for those jobs with subsidies, or through regulations, and then you will tend to have less of them. It is a very challenging problem.

Paul Krugman: Healthcare jobs include a substantial number of middle-class jobs. I have been obsessed since the last election with West Virginia which went three to one for Trump. He promised to bring back the coal jobs. But there are no coal jobs. There will be no coal jobs. It is about 3% of the state workforce now. 15% of the state work force is health and human services. Some of those are low wage—but it is not clear they have to be, you could have more labor unions. Those are the jobs of the future. Of the twelve occupations predicted to go fastest over the next fifteen years, ten are nursing in one form or another. “Services” is everything from hedge fund managers to janitors. Some of those can be middle class. And where we get into things—if Walmart had been unionized, which could’ve happened in a different political environment, we would be looking at quite a different landscape situation for those jobs.

Brad DeLong: So everyone needs a personal shopper and information provider, a personal psychotherapist, a personal trainer, and a personal life coach—and at an hour a day, that’s 50% of employment right there.

Paul Krugman: If you had taken a French physiocrat who believed that the only true creators of value are farmers and showed them modern America, where there are fewer farmers than people playing World of Warcraft, they would be horrified. Everyone must be doing make-work.

Eduardo Porter: Here is a question about unions, but you just brought up. Let me broaden it a little. What institutional changes going forward—you guys have decided that the safety net is not going to do it. Trade policy seems not the right way. So where is a policy set that has some traction? And let me throw in the arguments about the fissuring of the workforce. I’m wondering whether it might have even more impact than trade, and as large a scale as technology—you sorting off workers by their marginal productivity. The janitor who used the work for GE and benefited because he worked for GE is now an employee of the janitorial services subcontractor. Relegated to a much more cutthroat labor market than he would have been in the past.

Paul Krugman: We won’t know until we try it. The decline of unionization—we said that that was inevitable as manufacturing shrank and so we are left with essentially no private sector unions. That is by no means universally true across the advanced world. Canada still has a unionization rate of greater than 20%. DeMarcus something like 70% unionization. A lot of it has to do, I think, this is an argument one could have, with the political environment. The big reason that Walmart, which is our biggest employer, it is not unionized, while GM did is that Walmart became the biggest employer in a time when the legal and political environment was very hostile to unionization. Things like card check—it is possible you could change this. There is lots of reasons to believe that there is substantial wiggle room—that wages are in no wise pinned down by something called marginal productivity to the extent Econ 101 says. Is that enough to sustain the kind of society we want in the face of the robots? I do not know.

Eduardo Porter: you guys did not talk about education. But there is a question: how about spending more on reskilling?

David Autor: Education is the most important investment to make over the long-term. To adapt. In the late 1800s and and early 1900s when agriculture was shrinking very rapidly, agricultural states like Iowa made massive investments in secondary school education to get out ahead of this. They mandated that everybody stay in school until 16. This was a radical step. Not only were those kids hitting the books, but those kids could not provide labor to the farm while they were in school. It was extremely expensive. It was incredibly forward looking. It gave the United States the most productive, the most flexible, the most skilled workforce in the world for most of the 20th Century. It did not happen automatically. It was a societal choice. Imagine taking the labor force of 1900 and plopping them into the 21st century right now. In spite of their strong back some good characters, many of them would not be employable. They would be innumerate and in many cases illiterate as well.

It is challenging. We have definitely fallen behind in this. We are now 14th in the world in getting people through secondary education. We used to be first in the world. I do not want to say “make America great again”, but we are under investing there, and the way we are gutting our great public colleges and universities is particularly tragic, because it is an area in which we are doing very well. In many parts of the world public universities are distinctly mediocre, and that has not been true in the United States.

Brad DeLong: May I concur with and endore everything David Autor has said, and then strengthen it?

Looking out at an audience that include some people who may be thinking at some future time of giving money to some private university like NYU or Columbia, let me just say that from my perspective: don’t. It won’t work. That is not how our private universities work.

Back in the 1950s Harvard educated 1200 undergraduates a year, Yale 800, in a time when the University of California—UCB and UCLA—educated 4000. Today Harvard educates 1600 undergraduates a year while the University of California educates 50,000. At Harvard has in the meantime received some $30 billion dollars in private gifts.

If you want to make a difference as far as education is concerned—it’s organizations like CUNY that carried the serious load back in the days when Harvard and Yale had hard Jewish quotas of 10%. Given the way that our privates are set up, they are unlikely to use further gift money wisely to expand education in the future.

Paul Krugman: We had an event with Raj Chetty here. He did a study of systems that do produce major amounts of upward mobility. Public universities do it. They do it in varying degrees. There is one that stands out—off the charts: CUNY.

Eduardo Porter: One more question. What’s the role of a minimum wage?

Paul Krugman: The minimum wage is clearly something we can do. It won’t transform the situation. But it will make a material difference. There is no subject in economics I know of where careful empirical work has done more to change people’s minds than the question of the minimum wage. People do dig up things I wrote 25 years ago about the minimum wage about how it would lead to a higher unemployment rate. They say: Why do you now say something different? I reply: Because Card and Krueger did the incredible work. It turns out that the evidence is overwhelming that when a minimum wages are at the levels that they are in the United States today there is no detectable employment effect of raising the minimum wage. There is clear poverty reduction. It is one of the things we do know how to do. We should be doing it.

There is a question about the $15 dollar an hour minimum wage. Fine for New York City. Is it fine for Alabama? Maybe not. But a big push on minimum wages—definitely.

Brad Delong: May I ask you to say that to David Card—that it has changed your mind? He was lamenting three weeks ago that I was the only person who had told him it had changed my mind…

Paul Krugman: I always say that. You can tell him this. I like to be open-minded. This is shockingly good work. It moved me significantly left on labor market policy, because it says that you can do much more than I thought.

Brad Delong: And it created a great mystery. It says that, somehow, the manger at the local Burger King have substantial amounts of market power at the low end of the labor market, where you would think he would not.

David Autor: We have another good policy tool in this realm. The EITC. It makes work pay better. It has had positive effects on labor force participation, earnings, and well-being in female headed households. It is almost unavailable to men who cannot claim children as dependents on their tax forms. That does not mean that they are not fathers, by the way, but they are not claiming dependents. But if you are a single mother you can get up to $6000 a year in wage subsidies through the EITC. If you are a father with two children who are not your dependents you get about $400. The greatest declines in employment and falling wages are low-education men.

Brad DeLong: And that is my fault. In 1993 we needed to get the Clinton deficit reduction package up above $500 billion over 5 years, and the EITC for “childless” workers was the last thing we in the Treasury Department had to throw out of the sleigh to the wolves…

David Autor: It’s an idea that is a good one. But it is an expensive one. And the minimum wage appears free—you are making employers pay. For the EITC you would have to pay for it…

Brad DeLong: But the minimum wage is free in an allocative-efficient sense! You are reducing a market power-generated deadweight loss!

Paul Krugman: Let me just say that there is a tone—we worry about the politics, about how hard it is to do stuff. But if you look at how successful safety net programs were in the Great Recession, the answer is: incredibly successful. Increase in child poverty: negligible. Increase in adult policy: barely there. We have the tools to make life a lot better. Do we have the tools to convince people who benefit from these programs that they should vote for politicians who support them? That is another issue.

David Autor: We haven’t done that. The tax share of GDP has been at 20%…

Brad DeLong: But three million votes…

David Autor: But all in California. You can run it up in California—it’s not real America, so…

Brad Delong: Let me make a plea for UBI. The argument against Universal Basic Income is always that it is a “handout”, and people don’t want handouts. You would have to sell it as something that is not “welfare”.


After my grandfather Bill Lord moved to Florida and became a construction company boss and real estate speculator, he did well: he was briefly the richest man between Tampa and Orlando. Ever since the money from the Lord trusts has boosted my consumption by about $10000 a year. That has been my UBI. It has been quite welcome at times, and always convenient.

I don’t see this as “welfare”, as making me a “loser”, as offending my “dignity”—even though I had nothing to do with the invention of the Wellman-Lord desulfurization process or any other value created by his accomplishments, and I would accept the argument that he was overcompensated for them. Similarly, Mitt and Ann Romney do not think in any way that they are loochers—losers and moochers—from their UBI, which came from the stock American Motors had given Mitt’s father George to incentivize him, which stock they could sell to boost their standard of living when they went to BYU.

The problem with UBI and with the welfare state is one of perception and of perception of who is deserving and undeserving. Inheritance in America today is not tainted. I think UBI should not be tainted either—and I think Mitt and Ann Romney would agree, if only they would step back and think a little…

Notes: Working, Earning, and Learning In the Age of Intelligent Machines

The key seems to me to build intelligent machines that will assist workers in labor-intensive industries, rather than build intelligent machines that will eliminate workers in capital-intensive industries. The first is a clear win. The second can be a major loss if the things made in capital-intensive industries are close enough substitutes for the products of labor-intensive industries to greatly drop their value.

But what I have to say so far is limited.

It is simply made up of: Five Disconnected Points:

  1. Cast your mind back to 1999. All of this was then viewed not as a threat but as an opportunity. Few things can turn a perceived threat into a graspable opportunity like a high-pressure economy with a tight job market and rising wages. Few things can turn a real opportunity into a phantom threat like a low-pressure economy, where jobs are scarce and wage stagnant because of the failure of macro economic policy.

  2. Those historical cases in which technological progress has been genuinely immiserizing have been relatively few. They have been confined to situations in which technological progress takes the form of greatly amplifying labor productivity in capital-intensive occupations. Those then shed labor massively, as those tasks in which human beings act like robots—filling in the gaps that machines cannot yet do—vanish. But at the same time technological progress must do next to nothing to equip workers in labor-intensive occupations with better tools to assist them. Thus the canonical case is the 19th century handloom weaving industry in Britain and India. That suggests a focus on building robots to serve as tools and assistants for workers in labor-intensive industries, as opposed to further mechanizing and thus replacing workers in capital-intensive industries.

  3. Let me endorse the observation that for the past 200 years the mechanization of manufacturing has to a great degree involved treating humans as if they were robots. We can do better. At least, we hope we can do better.

  4. Let me try to satisfy Barry’s demand for less abstraction and also endorse Nils’s emphasis on human customization by asking you to cast your minds back to the days of Metropolis, Henry Ford, and Brave New World. Henry Ford wanted to satisfy real human needs in the cheapest and most effective fashion by taking massive, mammoth, and total advantage of all possible economies of scale. You can have a car in any color you want: as long as it is black. You can have whatever kind of car you want: as long as it is a Model T. You can wear any clothes you want: as long as they are identical blue overalls. You can play any sport you want: as long as it is Centrifugal Bumble Puppy. That was not the world people wanted. Alfred P. Sloan and General Motors drank Henry Ford’s milkshake by finding a sweet spot, in which you sold everybody mass produced Chevy parts in different, near personalized configurations. We can argue about whether people should value such human touch salesmanship and customization—whether it is a cognitive behavioral mistake stemming from our origin as hunter gatherers seeking to gather the most useful objects. But the fact is we do value such human touch customization. All the evidence suggests that it makes us very happy. That is a very large set of potential labor-intensive occupation that will last for a very long time.

  5. Never forget that back in the environment of evolutionary adaptation we were sociable toolmaking hunter-gatherers—constantly interacting with the complex environment where we would choose and modify objects to advance our purposes—in which we turned ourselves into an anthology intelligence under a geas to learn as much as possible, and immediately tell it to—gossip about it with—everybody else. With the coming of first the Agrarian and then the Industrial Age our jobs became overwhelming boring. Only humans, with our brains being supercomputers that fit into a bread box and draw only 50 watts of power, could be the necessary microcontrollers for animals and machines necessary for first Agrarian Age and then Industrial Age production. But those jobs vastly underutilized the human brain. Do not overromanticize looking at the hind end of a mule or tightening bolt number six on every object coming down the assembly line for four hours without a break.

Reading List: