 I would now like to introduce the wider annual lecture for 2022. Daryon Asimoglu is Institute Professor of MIT and an elected fellow of the National Academy of Sciences, the American Financial Society, the British Academy of Sciences, the Turkish Academy of Sciences, the American Academy of Arts and Sciences, the Economic Society, the European Economic Association, and the Society of Labour Economists. He's also a member of the Group of 30. Daryons are also of five books, including The New York Times' Best Seller, Why Nations Fail, Power, Prosperity and Poverty, with joint with James Robinson, a book that has become compulsory reading to all the different scholars and practitioners. I can't imagine any of us here haven't read the book. He's also, of course, there's a wonderful textbook that he has, Introduction to Modern Economic Growth, and along with that also, The Narrow Colleges, which was published more recently, State Society of the Fate of Liberty, again with James Robinson. Daryon has published over 140 journal papers and several of his papers had huge impact on the profession. His 2001 AER paper, The Colleges' Origins of Comparative Development, had received a stunning 16,000 citations and one of the most cited papers in economics. Daryon has made path-breaking contributions to a wide range of areas, including world economy, economic development, economic growth, ecological change, inequality, labor economics and the economics of networks. He's elected today is in the interface of ecological change, inequality and labor economics. He's also highly influential public intellectual with regular columns and part of the syndicate and other media outlets. Daryon has received the inaugural D.W. Schultz Prize for the University of Chicago in 2004, the inaugural Sherwin-Rosen Award for Outstanding Contribution of Labor Economics in 2004, the John Bates Clark Medal in 2005, the John Newman Award and the Jean-Jacques Lapland Prize in 2018, among many other prizes. It's a great honor for us to have Daryon Achimoglu provide the wider annual lecture, 26. I would also like to welcome the discussion, Marcella Eslava. Marcella Eslava is the professor and dean of economics at our host institution here, Nividat Los Angeles, a fellow of the Economic Society and vice president, 2020 to 2303, as well as president for 2024 to 25, of the Latin American and Caribbean Economic Association. Her current research interests were all from the lack of inclusive growth in Latin America. We heard wonderful insights from Marcella yesterday with a focus on firm dynamics, productivity, as well as informality. Marcella Eslava's research has been published in leading academic journals. She's been involved in multiple initiatives, research projects and missions led by multilateral organizations and the Colombian government to contribute to understanding the origins of the deficit inclusive growth in the region and to devise solutions to it. Now moving on to the annual lecture. The annual lecture is based on Darren Archimogal's newest book in the name of progress our thousand year struggle over technology and prosperity, co-authored by Simon Johnson, challenges the technological optimism of our age and of our academic profession, which maintains that technology advances ultimately benefits society at large. As he moved to ever increasing use of digital technologies, especially in the wake of the pandemic, the lecture asked what is perhaps the most crucial question of our times. Will technology deliver shared prosperity to the citizens of the world or will it simply accentuate inequality within and across countries? I would now like to invite Darren Archimogal to deliver the wider annual lecture, 26. Darren, over to you. Thanks. Thank you very much for those wonderful introductions. I think after those, everything I do will be downhill, so I'm not sure I'm looking forward to that. You don't wanna build up expectations too much. It's my true pleasure to be here to give the annual wider lecture in this wonderful conference and this wonderful university and lovely town. And thanks to all of you for being here. So I'm going to talk about power and progress our thousand year struggle over technology and prosperity. It is very much on the topic of inequality, but much of the emphasis in the conference has been rightly on persistent inequality in the developing world, for example, Latin America. I'm gonna change the focus a little bit and talk about increasing inequality and non-shared nature of prosperity in the developed world. And there's a good reason for that. This is actually the theme of a new book by myself and Simon Johnson. It used to be called In the Name of Progress, but the book has not changed. It's exactly the same, but the title has been changed by the publishers to Power and Progress, so that's the title. So without further ado, let me just jump in. In 1791, Jeremy Bentham proposed the Panopticon as an efficiency-enhancing monitoring system. In a well-lit circular room, essentially placed guard can look over and monitor what the prisoners are doing, either in terms of work or security. Thanks to its resonance in popular culture, Michel Foucault and the Guardians of the Galaxy included, you know, Panopticon is an ever-present contraption in our lives, but actually the person who dreamt of it was Jeremy Bentham, was his brother, Samuel Bentham. He was working in St. Petersburg, in a factory owned by the Tsar, and he came up with this idea as a way of monitoring the workers. Jeremy's contribution was seeing this as a broader efficiency-enhancing device and publicizing it. For example, I won't read the quote, but it sort of captures his view. This is such a wonderful thing. It's going to be revolutionary in all sorts of places, hospitals, schools, manufacturing facilities, and of course prisons. To Bentham, this was all in the name of progress. If something improved efficiency, society should welcome it with open arms. There was an undercurrent that if something improved efficiency and he thought that better monitoring would certainly do so, more information for the employers, why not? This would somehow be beneficial for everybody, but Bentham's philosophy, as you know, the utilitarianism, even if some people were losers, not that important, we sum across their utilities and still something like this efficiency improvement must be accepted. And in this, Bentham was not alone. Many of the famous philosophers of the late 18th, early 19th centuries were of the same view. Adam Smith, of course, much more famous among economists for many path-breaking contributions, was very much of the same opinion, even though in some domains his thinking was more nuanced. For example, Adam Smith, although he didn't talk about industrialization per se that much, he talked about many components of it and he felt compelled in the wealth of nations to take on the idea that better technology might be bad for workers. Again, I'm not going to read the full quote, but he goes on through the argument that, well, you save labor, that might be bad for workers, you can produce the same amount with fewer employees, but at the end of the day, the efficiency improvements are going to be good for the real price of labor and it should rise considerably. David Ricardo, the other great towering figure, was of the same opinion. He argued so in the first two editions of his monumental principles. He argued in parliamentary debates, he was an MP against concerns related to the Luddism that he couldn't imagine how technological improvements would be bad for workers on the whole, but later on he actually changed his mind and that's sort of interesting. So perhaps we're following a little bit in David Ricardo's footsteps. He wrote, if machinery could do all the work that labor now does, there would be no demand for labor. So he thought there were some concerns into the third edition of his principles was quite different. Modern thinking completely turns back, turns its back to Ricardo and it's very much along the lines of Adam Smith's thinking. Of course, in macroeconomics, labor economics, development economics, there's a huge literature inequality, but many of the approaches that we have developed, at least in the medium run, sort of confirm Adam Smith and Bentham's ideas. If something is efficiency improving, it should be generally useful. In fact, this is completely ingrained in the models that we use. In Cobaglis models, by fiat, unless you go into labor market imperfections, which I'll go a little bit later on, you're going to have labor and capital share constant, so if something improves the size of the pie, labor is a whole benefit. Perhaps within labor, there might be some small groups, but most people think, well, that might be a transitory phenomenon, but again, on utilitarian grounds, yeah, things probably are going to be good. In economics, we of course don't contain ourselves with Cobaglis production functions. We can go to other neoclassical constant returns to scale production functions with competitive labor markets, competitive factor markets. You get more or less the same sort of conclusions that generally technological change of any sort is going to be beneficial for labor. And in economics, we tend to sort of think of Luddism as a sort of a derogatory term. In a seminar, if you accuse somebody of, well, if you say that somebody's a Luddite, that's not a good thing. So the view that progress, some sort of efficiency enhancing progress is going to be beneficial, and we shouldn't sort of question it. I think it's not completely accepted universally, but it's fairly, fairly general. In fact, so interesting that these effects in aggregative macro models, labor models, are so pervasive, they don't even have a name. It's so ingrained that we didn't even think, as economists, to give them a name, but for a broad audience, Simon and I had to come with a name and we came with the name of a productivity bandwagon. You improve the productivity, and then that as a bandwagon pulls most people who are on it. There could be short-term disruption, but most people will ultimately benefit. And a good summary of that view is given not by an economist, perhaps, but by LinkedIn and PayPal co-founder, Reed Huffman. Could we have a bad 20 years? Absolutely, but if you are working towards progress, your future will be better than your past, your present, sorry. So this is the view. But this view, very much Benthamite and Smithian, was not shared by workers in the midst of the Industrial Revolution. Fortunately, we know Britain has very good records of petitions, letters, parliamentary commissions, so we have very detailed information of what the workers thought. And they thought nothing like Bentham and Smith. They thought this was a horrible time, they were being exploited, the cards were stacked against them, they complained about the fact that machines were taking their jobs, Luddites were not alone. They complained about wages declining, hard work, work getting harder. They complained about every aspect of the new production system. So for example, here is one from one petition from a weaver. No man would like to work in a power loop. They do not like it. There is such clattering and noise, it would almost make some man mad. Next, he would have to subject to a discipline that a hand loom weaver can never submit to. Automation, or what we call today automation, or many contemporaries would have called mechanization, was a particular point, sore point. I am determined for my part that if they will invent machines to supersede manual labor, they must find iron boys to mine them. We don't have perfect data from this period, but by and large they were right. From 1750 to about 1840s, there is almost no growth in the real earnings of British workers. Some regions show growth, there's some movements up and down, but overall no growth. But during the same time period, industrial workers' working hours increased by about 20%. So their real wages almost surely fell by quite significant amounts. Working conditions completely changed, and for the worse. Workers had a lot of autonomy. It wasn't a great system for them in the putting out system or other home-based production structures were not great, but at least by their own accounts, the additional discipline, what Bentham wanted, but did not actually achieve in terms of the Panopticon, was pretty bad for them. And of course the concentration of people in cities, much worse health conditions, much worse pollution, all of these were adding up. This is not an isolated incident. There are many other defining transitions over the last several hundreds of years, or 1,000 years, which paint a very similar picture. The dark ages were not dark in most senses. They were actually quite innovative. There were many new agricultural methods, new technologies introduced, windmills especially, and productivity increased quite significantly during many episodes. But for much of this period, peasants had almost no improvements in their living standards, except during episodic events. Advances in the Indo-European ship design were critical for growth commerce during this time period, but of course they were also the same technologies that were fueled and enable the transportation of millions of people as slaves from Africa to the New World. So another very clear example of major efficiency-improving technological innovations from which not everybody benefited. In fact, some very clearly delineated groups became clear losers. The introduction of steam engine was absolutely critical, as critical as the textile machinery, but the steam engines for the first 80 years or so were used mostly in coal mines for pumping water, and their immediate effect was that you could dig into deeper coal mines, and if you go into deeper coal mines, that's very expensive, and employers came up with a perfect way. You're gonna use children to do that job. Children as young as five were sent to incredibly dangerous conditions, and at the end of the 18th century, the about 20% of most coal mines workforce were underage children, and they had tremendous health hazards, complete lack of education, and often death out of that. The cotton gin, let me mention one more, was another completely revolutionary technology. US could not even produce its own cotton, let alone export cotton, and then after Eli Whitney's cotton gin, although there were other prototypes before Eli Whitney's, they could essentially apply it to the tougher southern cotton, led to a huge cotton boom. US became the world's leading cotton exporter, and was a critical part of US economic development in the 19th century, but the cotton gin also meant that slaves who were in more dispersed and working under less harsh conditions were moved to the more delta regions, deeper south, other much, much harsher, much more greater punishment-based production methods, again, very clear losers from efficiency. So this, I would say, is not just of historical interest. It's not something that we can turn our back on and say, today in industrialized nations, these things are just memories. Today we are going through a period of rapid efficiency-enhancing technological changes, led by digital technologies, but also biotech and other high technology, high investment sectors. But in the midst of this, we are seeing a complete remaking of the distribution structure of industrialized nations, led by the US, but not confined to the US. For example, the US labor share, which hovered around two thirds for 80 years at least, perhaps for longer, shows a remarkable decline, almost 10 percentage point decline. And this is not composition, this is not because some industries are, new industries are more capital intensive, the red line here is cap, composition adjusted, you see exactly the same. Even more striking is the concurrent changes in wage structure in the United States. This picture shows the real wage evolution of 10 demographic groups, distinguished by gender and education, all the way from workers without a high school degree in dark red, to those with more than ecology, postgraduate degree in dark blue. And you see that from the 1960s onwards, for about 15 years, you have this period of shared prosperity, where all 10 groups are experiencing more or less the same growth in their real wages. This picture starts in 1963, because before then you cannot distinguish college graduates and non-college graduates, and that distinction is rather important for certain things, but if you extend it back by just looking at college graduates, you see the same thing going back to 1945. Remarkable wage growth, and if anything, actually faster at the bottom of the distribution than the top. But then from 1980s, around 1980 onwards, you see a complete sea change. Inequality increases quite significantly, but most jarring is the fact that while wages at the top continue to increase, those for low education workers, especially men, but to some degree women as well, is declining. There's a quite significant decline in the real earnings of, say, high school graduate men. If you are a high school dropout in the United States today, if you're a high school dropout man, you're in the United States today, you are almost surely earning quite a lot less than your father did. This type of bottom of the wage distribution collapsing is essentially unique to the United States because this is the country where you don't have much protection for workers, minimum wages, union bargaining, or norms against very high wage inequality, but wages at the bottom have been stagnant pretty much in all industrialized nations and inequality in various dimensions of inequality here summarized by the Gini coefficient has increased pretty much everywhere. Moreover, the nature of work has changed in quite parallel fashion. So here again, I'm showing it for industrialized nations. This, the previous picture, of course, is not the same for Latin America where inequality is higher but declining in a few places. This one is actually similar also in Latin America. Jobs, you would identify as the middle class and here in this classified as those in the middle, occupations that are in the middle third of the wage distribution, they have been disappearing pretty much everywhere in the industrialized world. And of course, no surprise, these are blue-collar jobs and clerical office, back office jobs. There are many fewer of those in some places replaced by more technical jobs, especially with the educational upgrading. In some places replaced by much worse quality jobs, security, food preparation and so on. Now, in interpreting all of these facts and the broader optimistic view of what technology does, I think the perspective of economists is shaped by two factors. One is what I've already mentioned, the sort of idea that if something is efficiency improving, it's gonna increase the pie. Somehow the market process is going to ensure that the gains are shared. The second is that of course, we have many models that recognize the importance of choice, endogenous technology model or other endogenous choice models. But I think on the whole in many of these debates, how malleable technology is, how manipulable its direction is, is often ignored. This is the basis for, again, a reaction that I have been working on this topics for over 10 years now. I mean, they go back to things I was working on in my dissertation, but much more actively over the last 10 years. And the criticism that I hear often is, but being critical of the current direction of technology is essentially a way of being anti-technology and that's dangerous and that's bad. It's a form of laddism. And I think the basis of that sort of view is that, well, because you cannot manipulate and change the direction of technology very much, criticizing technology is the same as saying, criticizing the current direction of technology is the same as saying, well, perhaps we don't want this technology. So choice is a very important element in how we should form a worldview about technology. So perhaps criticizing the current direction of technology should be not resisting technological change in general, but a plea for changing its direction. And why is that important? Well, that's particularly important because I'm going to argue the productivity bandwagon has its limits. And I want to highlight two of them here, although there are others that we can talk about. One, going back to the views of the workers in the midst of industrial revolution, automation, meaning use of machinery and increasingly algorithms for substituting for tasks that previously labor performed is very different than increasing efficiency of production uniformly or finding new things, new ways, new productivity enhancing devices for workers. Automation has very different distributional effects. I'll also mention in a second, show you has very different productivity effects. And it's not a type of technology that by itself going to undergird shared prosperity. Second, efficiency improvements will increase the size of the pie, but then institutional factors are going to determine how we share those gains. We can increase the same size of the pie, but if the environment is coercive, for example, some things that I studied with Alex Wolitski, that greater productivity might actually lead to higher coercion and lower wages or rent sharing patterns might change or they may be bad enough that the gains are not going to be shared. If the technological changes that you're talking in mind are like what Bentham imagined and what Amazon applies today, much better monitoring devices, perhaps you wouldn't be so surprised that those may not lead to sharing a lot of those gains with workers. So those are the two elements that I want to emphasize, but time is short and I want to sort of spend a little bit of my time with you today on the first point on automation. So this is an integral part of the book, but of course the book is written for a broad audience so we don't do the technical stuff. So here I'm gonna do just a little bit of the technical stuff. So think of this as a way for me to propose a somewhat broader way of thinking of production structure. And to do that, I'm going to, mea culpa, I'm gonna use two slides of math, but most of it, I can just do it with diagrams. So in the classic way of, we think about production, especially at the aggregate level, we have an aggregate production function, something shifts and then that increases output and perhaps wages and employment. Instead of doing that, I wanna think of production as consisting of a range of tasks. If you want to produce a piece of garment, you need to do a bunch of spinning tasks, weaving tasks, design tasks, then chemical processes, then you have to do all the non-production tasks, et cetera, and all of these are done, then you get a piece of garment. A key decision is which task is gonna be allocated to which factor of production. In general, think of many different types of skilled and unskilled labor and different types of capital, but let me simplify it here, just two, capital and labor. And I'm going to assume, I'm gonna put the tasks on the horizontal axis, I'm gonna assume that only those to the left of some threshold I are technologically automatable. So these are the ones that we know how to automate, okay? And then on the horizontal, on the vertical axis, I'm plotting the cost of production of that task, cost of accomplishing that task with the two factors of production. The orange one is with capital, the rental rate of return of capital, one type of capital divided by the productivity of capital of that task, and the blue one is the same wage divided by labor's productivity. And then the firm's cost minimization problem is easy, choose the lower envelope. Now, in this framework, let me think about what different types of technological changes might be. So the one that economists love, Uzava theory and balanced growth, all things work so beautifully is labor augmenting technological change. So labor augmenting technological change would look like labor's productivity increasing everywhere. So great, this is a wonderful thing if it only existed. So I don't know of any technological change that makes labor more productive in everything, but perhaps certain things might approach it. What it does is that because it's making labor more productive in everything, it creates a huge productivity effect that's shown in that shaded blue area. What about its impact on the allocation of tasks to factors? Here, no impact. In fact, generally second order. And as a result of that, it has some distributional effect, but it's very tamed, distributional effects. This is the reason why in the neoclassical world, technological change, efficiency enhancements do improve everybody's welfare. If you like capital augmenting the same thing, big integral, not much effect on the allocation of tasks. But now, none of this looks like automation. Automation would be we introduce spinning machines or weaving machines. Now, handloom weavers are replaced by machinery. Well, that would be a shift of I to some I prime there. When you do that, now you get something very different. First of all, the productivity effect instead of that big integral, you get a small triangle. That triangle could be as small as possible, as small as you want, if orange and blue lines are very close together. So the efficiency implications of automation could be very mild, but it has huge distributional implications. Why? Because it displaces a lot of workers from the tasks that they used to perform. It impacts not just those workers, because those workers then go and compete for tasks that other workers were performing. So it has a generally depressing effect on wages. Now, if history was one of automation and nothing else, we would not be talking of shared prosperity and things that actually were pretty good after say, 141850s and 19th century and certainly after 1945. We wouldn't be talking of labor share being constant. In fact, it's not because of labor augmenting technological changes either, because labor augmenting technological changes don't even have the power distributionally to undo the effects of automation. Something else, this framework suggests and our empirical work and our historical work sort of backs up, that something else is new tasks. New technologies have the capacity to introduce tasks that did not exist before in which labor can specialize. So if you think of modern economy, think of most people around you. Many of them performed tasks that did not exist 50 or 60 years ago, even those like us professors, accountants, we are performing tasks that professors and accountants 50, 60 years ago couldn't dream of. So that's the new task that is so important for labor demand. Okay, so this is the informal thing, but I wanna now tax you for two minutes by putting a little bit more flesh on the bones. So here is the math for it. There's a purpose for why I'm doing this. So think of the way that you have to perform these tasks as some CES aggregator, some aggregator, any aggregator would do. So they have these task services, you aggregate them, you produce some output. And what is the task production function that I summarize in that figure? Well, for tasks less than I, you can produce them by capital or labor. For tasks above I, you have to use labor, that's it. That was the model. The only other thing I'm adding is this N here, which is going to be this index of new tasks. You're adding things to the right, okay? You solve this model and express everything as a function of the labor supply and the capital stock of the economy, and you get something that looks a little bit familiar. So AK times K to something like elasticity of substitution. So you might say, oh, you know, Daron just gave a micro foundation for the CES production function. But actually it's not, appearances can be deceptive. This thing is not a CES production function. In fact, it couldn't be farther away from it. The thing is that Sigma here is endogenous. It's not a fixed parameter. It changes depending on the endogenous substitution between capital and labor. And most of the action doesn't come from AK or the Sigma. It comes from these orange terms, which normally are dropped in the CES or are taken to be as given. And one way of seeing that is to focus on the key economic object that this model highlights, factor shares. So since I have two factors just I can focus on the labor share, that thing that was heading south before. So SL, labor share, you can write it in this form. So the blue terms are exactly what you would get from CES with a fixed Sigma. It's wage relative to factor productivity. All the action has to come from this Sigma. And if that Sigma is not too far from one, you have to work very hard to get anything out of CES. But here all the action is gonna come from these orange terms. What are these orange terms? That's what I'm gonna call the task content of production. How much of the tasks are being performed by labor? How much by capital and automation directly impacts that? That orange term looks complicated, don't worry about it, think of it as N minus I, approximately N minus I, that's what it is. So that orange term, the labor enhancing or labor friendly part of it gets declines when there is automation, I increases and increases when there are new tasks. So now let's look at the effects of automation on wages. So if you are a believer in the efficiency enhancing view of technology will be spread as you can say, well, as long as automation is going to improve output, it will ultimately get shared in one way or another with labor. So let's look at that. So let me look at the measure of the delays. Okay, all right. So let me look at the measure of labor like wage bill that the corporate sector paid and see what happens to it when I change I. Well, don't look at the math, just look at the label. This is a productivity effect, that's that triangle. And this is the displacement effect, the displacement effect completely about this orange term, that gamma that I introduced going down. And which one is larger? We don't know. In general, productivity effect could be larger, displacement effect could be larger, but let's go back to that figure. If I make the orange and the blue close enough together, the green triangle is small enough so that necessarily the displacement effect is gonna be smaller. So therefore with automation technologies, you can have productivity improvements, capital business owners, perhaps high skilled labor, we'll see that in a second, may benefit, but workers, especially line workers may actually end up fairly worse off. Okay, is this all theory? Well, it is, but I hope not just theory. So one way of understanding that is to look at a archetypal example of an automation technology, robots. Actually, robots are stacking the cards against funding something like a large displacement effect because they're actually very productive. I'll argue if I have time later on that many automation technologies that we have been enthusiastically adopting over the last 30 years are not so productive, but robots are the exception. They're actually quite productive. They have revolutionized manufacturing, especially heavy manufacturing electronics, chemicals and cars. But if you look at their effects, they look nothing like a rising tide lifting all boats. So let me not go into the details a lot, but what Pasquale Restrepo and I do is we construct a measure of exposure to robots based on which sectors are having most robotic innovations and different local labor markets in the United States that specialize in different industries in the 1970s. So on the basis of this, we have this exposure to robots measure. We show that higher exposure to robots predicts more robot adoption and then we look at what it implies for workers. So here are three graphs that summarize it. Each circle here is one commuting zone or a local labor market. And you see a very strong negative effect. It's showcased by places like Detroit, Lansing City, Defiance City, heavy manufacturing hubs in the Midwest. But if you leave those guys out, you get the dashed line rather than the solid line. So they're not driving the relationship, they're just bang on the regression line. This is for employment, lower employment to population ratio. This is for wages, quite significantly lower wages in places that have been adopting robots. And if you look at the distribution, say for example with a quantile regression, you find that negative effects are borne by the low wage workers. So this already gives you a hint that it's not just between capital and labor, but the broader inequality trends perhaps are going to be related to automation as well. That's something that Pasquale and I turn in a more recent paper. And we changed the focus away from robots to all automation technologies, so software automation which is a very important part of it. And we look at the national labor market because a lot of inequality is not just local, but at the national level. Again, let me summarize everything with one figure. So what we do is that we construct a task displacement measure, which is loosely speaking, in the 1980, your demographic group, which could be women with low education, who are young and native born. Your demographic group, what fraction of the tasks that you have been performing have been taken over by automation, which could be software automation, non-robotics, manufacturing automation, or robotics automation. And then we put that on the horizontal axis and then change in the real wage of that demographic group for 500 demographic groups on the vertical axis. And this is the cloud of points. If you look at the R square of this the width or without control is about 70%. So a huge fraction of the inequality between groups in the United States seems to be very strongly correlated with automation. And all of those groups that you saw losing real incomes, those are below the real wages, I should have said, zero. They are all those that have heavily lost tasks to automation. So automation, of course, is nothing new. So if there is such a negative effect of automation on inequality, on the labor share, how come we don't see them in the 40s, 50s, 60s, 70s? Well, in a different paper, Pasquale, Restrepo and I look at what's the overall rate of automation and displacement caused by automation and a proxies for new tasks. So here is a summary. So here you have to make more assumptions. So this is a little bit more assumption based than the figures that I showed you before. But this is data from 1947 to 1987. So you have a lot of displacement and everything I've put in the labor share units because that's the key. You have a lot of automation, but at the same time you have a lot of workers being reinstated into the production process because of new tasks. So much so that some of those two is hovering around zero, so the two are balanced. So it's like a balanced growth path. Same thing we did in manufacturing. But now fast forward to the post 80s period. You see a very different picture. Automation accelerates, but even more remarkably reinstatement slows down. And in manufacturing, there's almost no reinstatement. It's all automation. So there is something technological going on that I think is very important for the new, non-shared nature of economic growth. But I have not talked about red sharing in institutions because I don't have time, but I wanna also argue that these technological trends are themselves not separate from institutional changes. And to do that, I'm gonna use my last 10 minutes to do a whirlwind tour of going back to the British Industrial Revolution. So let me skip these. So things looked like, worrying like today during the Industrial Revolution. Many workers not benefiting capital business owners benefiting and workers not doing so. Why was that happening? Well, looked at it from this perspective, two major reason. One is institutional workers couldn't organize. In fact, trade unions were illegal. There were still master and servant acts that made it illegal for workers to quit their employers. And Britain was an aristocracy oligarchy without any democratic counterweight. And early technological change had a strong automation bias. This is what most employers were interested in textiles and in some of the associated industries. That's what some of these workers were complaining about, as I said. Why did things change from 1840s onwards? Well, I think both of these elements started changing. First of all, technology. New industries and new technologies even in some of the existing industries took a very different direction. There are many reasons for that. I don't have time to get into it. Railways is emblematic of one of them. Railways were very worried about low productivity of workers. So they, from early on, experimented with many methods, but Panopticon and things like that wouldn't work. They went for efficiency wages. So railways are much more into sharing the benefits with workers. But even more importantly, many industries went in a direction of trying to increase the productivity of low-skill workers. This is most emblematic in the US and associated with the Habakkuk hypothesis. US had a very steep shortage of skilled labor. And that was what people kept talking about in the 1830s, 1840s. And then they came up with a solution. New technologies, interchangeable parts being one part of it, which were trying to reorganize work in order to increase the productivity of unskilled labor. And then these technologies then spread from the US very quickly to the UK, Canada and Europe. So this is the Habakkuk force that changed the nature of technology in a much less automation, more worker-friendly direction. But the institutional changes were no less sweeping. Democratization was a huge deal. Civil service reforms, you know, British cities were places you wouldn't want to be in. And they all got cleaned up, but part of it because the civil service got on the case and civil service got on the case because state capacity was much higher already in the 1850s. But also worker power increased. So collective bargaining, trade unions became, becoming legal was one of the major part of the chartist reform questions that happened in 1871. But even after 1871, they were cases against collective bargaining attempts under the master and servant laws and those were repealed in 1877. The broader forces are actually captured by an early radical, John Tellwell, who is often associated also with the Luddite movement. And in fact, it shows that the Luddite movement wasn't just break the machines, they were asking for a different social contract. And what Tellwell said, you know, this accumulation of capital that you're seeing, all the inequality, but it is bringing something else. It's bringing workers together and new ideas. Hence, every large workshop manufactures sort of a political society which no act of parliament can silence and no magistrate can disperse. These were, of course, imperfect steps. Britain, the US don't look like happy, shared prosperity societies in the late 19th century, but the sort of the Habakkuk path of technology continued. Again, this is a large part of the book, but I don't have time to go into it today. But there was a lot of effort at increasing worker productivity, creating new tasks throughout British and American, especially American industry. And again, it was sort of bolstered by worker power. And worker power took a very interesting form. In many places, you know, it went wrong. Unions sometimes blocked technology, but many union leaders, not all, many union leaders also understood that what they had to do was not to block technology, to resist technology, but to actually try to steer the direction of technology. UAW in its heyday was very successful in this, against GM, against Ford. We offer our cooperation a common search for policies and programs that will ensure that greater technological progress will result in greater human progress. And what they emphasize is you wanna create new tasks, new training opportunities for low-skill workers. And there were many successes, but then things went wrong. Why did they go wrong? Many reasons, this failure has many fathers, but I think no part of it can be told without digital technologies. And again, this is both a qualitative account, but also quantitatively, I've showed you again, if you believe it, that 70%, 50 to 70% of the inequality is related to a subset of digital technologies, the automation technologies. So how come? Why, why digital technology? In fact, if you look at the ideology of digital technologies early on, this is not what you would have expected. The early pioneers of digital technology, some of them at Berkeley, some of them at MIT, were very much opposed to big corporations. This is quoting from one of them. Industrial approach is grim. It doesn't work. The design model is designed by geniuses for use by idiots. And the watch word for dealing with the untrained, unwashed public is keep their hands off. This is what they were against. They thought digital technologies would be liberating for workers, liberating for citizens. It would destroy IBM, IBM was the organization that they hated most. And it would sort of create much more basis for actively participating citizenry and shared prosperity. At the end, the opposite happened. Institutional changes, a new ideology among businesses in the shape of Friedman doctrine, union movements completely being decimated, lack of countervailing powers. But very importantly, the next generation of digital leaders completely swallowed the hook line and sinker, the idea of IBM. We're gonna produce big machinery so that companies can monitor workers, automate some part of it, gives the control to managers rather than the workers. So that's the direction in which the digital technologies went. Now again, I can get going to the details with some more evidence, but let me just show you one thing because I think this might be fun and people may not have seen it from some new work that I have done Alexi and Daniel Lomar. One place where you can see the Friedman doctrine, we argue, is in business schools. And so what do business schools do? Well, we argue, it turns out that if you have a manager who graduates from a business school in the US and in Denmark, this is for the US, what they do is they cut wages. So nothing before, a new CEO with a business school degree comes in, wages decline, labor share decline. Well, perhaps they are doing that while at the same time increasing the pie. No, output doesn't increase, employment and investment doesn't increase, exports don't increase. You see the same pattern in Denmark. Nothing before the business school CEO comes in, thereafter, wage declines. This is just the tip of the iceberg. I don't think business schools explain much, much of what's going on, but it's emblematic of a new ideology without countervailing powers. That was, I think, part of how the direction of technology got dislocated. Well, I'm almost at the end of my 50 minutes, so let me just say two more things. One is artificial intelligence, all of the things I've shown so far is be free AI, but artificial intelligence is continuing and perhaps amplifying the trends that we're seeing with digital technologies. And I think it's doing so again because it's embedded in an institutional structure that elevates automation rather than improving worker productivity, and it's also embedded in an ideological structure. And in fact, the very word machine intelligence and artificial intelligence are, in my view, just tell-tale sign of that. What you should want is not machine intelligence, what we argue is machine usefulness. What you want are machines that are useful to people, not intelligent machines. And in fact, this is not one criticism that I get, including from some of my closest co-authors, or this is all well and good, but can you really redirect technological changes? There really is a possibility that things like artificial intelligence or digital technologies, more capable machines, could actually be used for improving what human productivity does. Well, both historically and today, the answer is yes. Some of the early luminaries of the field, I think those that had a much better idea than the ones that are followed by the current AI field, actually had exactly that view. People like Norbert Wiener, Douglas Engelbar, J.C.R. Licklider, and some of the most important digital innovations came out of that approach of machine usefulness. And the opposite of that is exactly what the early hackers described as the IBM approach. You use digital technology to control humans, to monitor them, to do the things that they used to do. That's a very different vision. Now, since the crowd here is very interested in development, actually let me say this is bad news for development as well. I think for developing countries, as Jacob is here, Jacob's wonderful work is on this also, on inappropriate technologies, I think have been a very major problem. And in some sense, artificial intelligence by automating white-collar work, again, moving in that direction, slowly albeit, moving in that direction, I think could be the mother of all inappropriate technologies because it's going exactly in the direction of using things, capital, skilled labor, programming that are scarce in the developing world and not using the resource that the developing world has, which is abundant labor. So it will be a force for within country and between country inequality. And then finally, all of these technologies make things much harder because they have also impacted the nature of democracy. And the way that they have done that is actually not a separate phenomenon, we argue in the book. It's part of the same ensemble. If you have a business model of top-down technologies and technologies for control, it lends itself very easily for monetization based on ads and data collection. And that is the nexus that we argue is at the root of a lot of the democracy troubles. And redirecting technological change, again, is feasible. The examples I have given are telling on that, but another one that we see is renewable technology. This is, I think, not much for us to be happy about in the environment climate change realm, but there's one thing we should take pride in, which is that even a very small amount of government and societal pressure led to a huge redirection of technological change. More than 20, 30-fold declines in solar and wind costs. And you have to look at the book if you're interested for arguments of why the same is feasible in the realm of digital technologies. Let me conclude here. Thank you very much for spending this 50 minutes with me. Thank you. Well, let me start by thanking Daron for one, for being here. Two, for what really was a wonderful lecture. Illuminating is always inspiring as well. And also thank Leopoldo Carlos and Cuma as, because as you guys already said, that discussion led me to me being here. So what I wanna do today is to place the light a little bit more on the final words that Daron gave us and try to think a little bit more from the standpoint of developing economies. Of course, what Daron just gave us is a very convincing and fair challenges to this idea of what he called the productivity bang wagon idea. The idea that productivity growth will bring all the good edges that we need. He plays the light on two main things that talk about the limitations of productivity growth. One is the fact that the prevalence of automatization places challenges for the absorption of the less privileged workers, but also the fact that there is poor rent sharing. Of course, all of these worsened in the era of digitalization for all the reasons that Daron mentioned, but let me also add because of the great gains that digitalization brings for the scale of corporations through the fact that they rely on networks and the need for these big, let me call them, mutualities. Of course, this does place a challenge for the idea that technological adoption and progress may be inequality reducing and does help us answer the questions that we're all together here to answer. But of course, as Daron said, his focus has been especially in these technologically leading countries. And at the end, he pointed that this may be even worse for economists such as the ones that are of most interest to people here because of their labor abundance and the great pearls that automatization does bring. The other feature that I wanted to highlight is how crucial labor and social protection in general, the welfare state, has been to lessen the problem. So let me pick up on some of these points and ask what's in there for developing economies. I would say the problem is even more complicated for developing economies because as all of these things matter for these economies, exactly in the way that Daron said, they may matter even more. We also have the problem that rather than the main issue being that this developing slash adopting firms exploiting their workers or shedding out their workers, the main problem in developing economies is that we never had the absorption of workers by firms that had already adopted or the technologies that allow them to even think of paying high incomes to their workers. The problem let me emphasize is not that rent sharing is better or the feature that I wanted to emphasize is not that rent sharing is better in these countries and that's the reason we don't have to worry that much about automatization of the rent sharing problem. In fact, markdowns are high in our countries. Here's some data for China, here's some data for Chile and in fact the larger corporations also pay lower wages that is to say they have larger markdowns. But rather the problem that I wanted to highlight is that rent sharing becomes less relevant because as I said in the beginning, less workers are being absorbed by firms with which they have the chance to bargain. So the numbers I wanted to bring here to emphasize this point are these. If you look at this more developed economies in this chart, the vast majority of workers are in firms here exemplified by firms of 10 or more workers, the typical technology adopters will be in the higher end of the firm size distribution. Instead, in Latin American countries here, the only countries for which I have numbers in the developing world, that is the vast minority of workers that are absorbed by these types of firms. Instead, a very large fraction of workers here are self-employed and that is even worse for the workers for which the inequality problem is more problematic in these countries. Because, sorry, I'm also having trouble here, okay. Because the other interesting feature of inequality in this country, the other distinguishing feature is the fact that we not only are more unequal but inequality is more dominated by poverty rather than just by a very far and dense upper tail of the income distribution. Sorry, there's a little bit of a delay. So, not a problem as you may think that in the late years we have adopted technologies that are very automatized and therefore we have shed a lot of the workers out. This has been a problem that has been prevalent for very many years. There has been an anti-job creation bias by firms. Even pre the new era of automatization and digitalization, here's some numbers for what the manufacturing share of employment and value added has been over the process of development. It's very hard to distinguish the two lines but in all countries the line that is further above is the share of total value added that's represented by manufacturing. And the other line is the share of employment that is represented by manufacturing. And whereas in richer economies especially at the peak of industrialization they share that manufacturing meant for employment and for value added was very similar. Developing countries here again exemplified by Latin America here, Colombia in particular have generated much less employment from the start. So what I wanna bring with these numbers is the additional issue that we have to deal with in these economies that while all the perils that are on mention are very much there and very much represent a danger there is an additional danger of an anti-technology bias in a set of regions where technology has not even gotten there and has not been able to generate the value that at some point did create the prosperity that the more developed economies saw. Let me also mention a similar or a related point regarding the welfare state social protection more generally understood and this is of course an issue that was raised yesterday in the policy panel. While labor protection is indeed central and central when we have the rent sharing problems that have been pointed at today. There is also the question of how we combine social protection with technological progress and more general with the kind of growth that developing economies definitely need. And so there is the question of how do you fund social protection when you have not gone through the wealth increases that allowed for that social protection to arise in the countries that they're all told us about some minutes ago. So I leave you with that open question and let me move now to the Q&A panel I think. What's going to go on? Sorry. So we've had an amazing lecture and really interesting observations from Marcella and now I know that a lot of you asked one to ask questions both to Darren and also to Marcella. So let's get started. And what I want to do is take three questions and that's okay Darren and I'll say that. Take three questions from the audience here. One from here, one from here and one from there. So you get a parody into the questions. So let's start from this side and just put your hands up. Wait for our colleagues with microphone. Sanjay, if I can see the first person to ask questions. Go ahead. Yeah, go ahead. Maybe you want to introduce yourself. I'm Santel Levy. Thank you so much for the lecture. So on the empirical part that you show, which is showing the latter years, the reinstatement effect being so negative that in the end the labor sure is coming down. The question is, how do you entangle the productivity and technology part that you're emphasizing vis-a-vis international trade? What could I say? No, no, no, you're not really measuring that. What you're really measuring is NAFTA plus China plus whatever, I don't know that. And then one observation which is connected to Marcella. I really like the emphasis on, at the end of the day, the sharing depends on the political institutions by which workers have power and no power. But in our countries, with more than 60, 50% of labor force unorganized, the political power of labor is very weak because they're very, very dispersed. So that's yet another argument. If in the future the gains from productivity can be shared more equally, we need different political institutions by which workers need to be organized. So a question in an observation, thank you. Thank you, Santiago. Question from here? There's a question there, keep your hand up. Thank you, very nice as always. So theoretical comment first. So regarding the choice of technology, you have like two optimality issues. The first one is if you have a planner, the planner is going to choose the technology which maximizes output and then you share the output. But when you have firms choosing the technology, maybe they choose the technology which maximizes capital income. So then you have one problem. And the second problem is then by this choice you change income distribution. So I would want to hear an analysis of that. The second one is you may have, let's say you talk about the purely technological distributional issue and then the institutional issue. But in my view, and I'm actually on the right, these labor saving innovations have the effect of reducing bargaining power of the workers. Basically because they are less needed. So I want to know if you have studied these issues. Thank you. Much, let us have a question from this side of the audience, on my right. Anyone from that side? Okay, Leopold, I think. Go ahead. I'm gonna ask a very short but obvious question perhaps. I want to know if those business school managers increase the manager wages. You should tell us that. The manager wages, yeah. Okay, well let me first say thanks to Marcela as well. I think those are exactly right, your points. Let me make two comments. And that's why the book is focused on the technologically leading countries. Because I think the problem for the emerging world or middle income world is much harder. So both the type of social protection that is needed is becoming harder and harder, especially when these countries are being squeezed more and more by both technology and trade. And second, most emerging economies don't have control as much control over the choice of technology. So even if you buy everything, I've said, and some people may not, you know, the extent to which say Colombia is a country or the Colombian business has a choice over the direction of technology is significantly less than American or German firms. And that's gonna be a really complicating factor. Put that together with the difficulty of building missing institutions, I think it's a huge challenge. The three comments were all excellent. And I think they all require detailed answers, well, except Leopoldo's, I gave it already, not detailed. They increase shareholder values. They increase share repurchases dividends and managers wages. So that's a first order question that your question applies both to that displacement and reinstatement figures. It applies to the robots evidence and it applies to the inequality evidence on the task displacement. And from a statistical point of view, it turns out that trade with China is fairly orthogonal. It's a curious pattern, but when you think about it, a lot of that was in the low value added sectors, such as apparel, furniture, toys. Whereas much of automation, both software and equipment robots is much more in the high value added sectors. Now, in terms of the inequality patterns that I showed you, there is the other aspect of globalization, offshoring. So I took credit for offshoring there. So actually what I showed you on the horizontal axis was both offshoring and automation. And I see offshoring very related to automation. But quantitatively it's about one tenth of automation. So if I didn't do that, the numbers wouldn't be that different. But it is, it can become even more important in the future. NAFTA was very important for offshoring. Much less, I think the WTO accession of China was much more for the final goods. And there's a curious pattern there which I don't understand. Which is when you look at the China effects, they're very well identified on employment. David Outer and Colleen's work and some work that I have done finds very large effects. But inequality effects are more limited. Whereas when you look at automation, both the employment and the inequality effects are very commensurate with each other. So there's some puzzle there that I don't fully understand. And now let's a wonderful question. And you know, I'm going to give a lecture at the AEA that's exactly on these issues is when is it that the distortion, when the direction of technology is distorted. And you know, if markets are perfectly competitive and all of the other neoclassical assumptions hold, even if firms are making the decision and they're looking at it from the point of view of the capital return, the market's gonna work out. But exactly once you deviate from these things, then there are gonna be distortions. Distortions can be technological in nature. They can be because of this automation versus other things. Again, in settings in which things are not fully competitive. And it also will be because of differences in markups. And in general, what I, going to your imperfect labor market comments, bargaining and so on, that's actually a topic of a new work that Pasquale Restrepo and I have. And I think there are lots of interesting things. One thing that we also emphasize in the past is that there's already a pecuniary channel that leads to excessive automation. I even had a slide on that, but I skipped it. In any labor market in which there is a difference between observed wages and the social opportunity cost of labor, efficiency wages, bargaining, whatever, firms are gonna have automation incentives based on the wage. Social planner would like to do it on the lower social opportunity cost of labor. So there's gonna be an excessive. But there are other interesting effects that come much amplified rent reduction effects of which tasks you automate. In principle, there's another channel, which is exactly the one that you pointed out, which is that you may also choose to automate in order to reduce. Not to reduce the rent directly, not to remove the rent, but you still keep the, you use the threat of automation as a way of increasing your bargaining power. I believe in that effect, but empirically it's hard to identify. Theoretically, it turns out to be also very tricky. Many models don't give that for a variety of reasons we can talk more about. Thanks, Darren. We have a very large virtual audience for this lecture. Around 500 are registered. So I'm just curious whether there's a competition for the virtual audience, not so far. So let's turn again to all of you and let's have a set of another round of three questions. There's one for Rackle there. A little bit, go ahead, yeah. Maybe we'll just wait for the mic to answer. Rackle Fernandez. And you know, I, you didn't mention a 19th century figure economist also in some sense. Who would have said, well, it's obviously a question of who owns the means of production. And I never would have disagreed with you that technological change is endogenous. But why don't you think aside from what government can do in terms of thinking about, you know, taxing certain types of technology, what do you think about sharing those rents that come? Not at the firm level, but at the society level. I actually don't agree with myself in thinking that social protection is going to be very helpful here because I think this is just a problem that's going to be accumulating over time. So what schemes could you come up with, not by giving people stocks, but directly via automation, which is what's lowering labor's bargaining power to rectify this problem, or at least help rectify the problem? Thank you, Rackle. Oh, no, I think I'm supposed to take more questions. Sorry. A couple of more questions. There are any more? Go on, see. Okay, yeah. There's one there, right? Actually, yeah. And, yeah. Put it back. Put it back, okay. It's very important that I don't see it here. Yeah, do correct it. Thanks for a super interesting talk. And so as it has been mentioned before, the political economy problem is that firms mostly make this decision, this call of what technologies they want to develop with some bargaining in the story. And I really like the paper on green technologies that you mentioned that really shows that it is feasible to provide some incentives that affect which technologies tend to be developed. I'm just afraid that the parallel for, in this situation, is not as easy, and it requires a whole lot of information. We are in not only information, but ability to predict as one of the quotes says, maybe it will be a harsh time for 20 years, but it will be worth it in the longer term. And some of it is true. So how do you, in the meantime, I guess one doesn't want to become China and as a government saying, you have to develop this kind of R&D and not this. So how do you see a feasible way of reconciling the positive forces of market decisions for what R&D, where R&D should put its efforts? And this rectification, it seems like you convinced us that it's necessary, but you didn't give us what you think is the best solution now and feasible and politically feasible, I'd say. Thank you. There is a question at the back. I can't see where clearly the talk, but yeah, sorry, go ahead. Is that right at the back, yeah? Thank you for the presentation. So I remember in your presentation, you said many times that like, eventually everyone will better off for the advancement of technology. But I think everyone's bettering off is not the same as reducing equality because if the bottom better off for a little bit and the top better off for a lot, then actually inequality increase even though everyone is bettering off. And another comment of mine is, I think, how do I put it? So inequality, if in the words of the British, it's the haves versus the have nots. So when we say will technology reduce inequality? So whose technology, right? So if the technology is eventually, say in the hands of a few number of people, then probably that wouldn't reduce inequality. If that sounds too much like a Marxism economist, then let me put it into like an international perspective. For example, 60 years ago, United States probably make everything from television to telephone to toys, right? But it's not because other developing countries, 60 years ago they don't want to be, it's because they don't have the technology. But in the past few decades, first Japan then China and then now Vietnam, they, well for one way or another, they start to have this technology to make all these manufacture things. So I think probably democratization of technology is what's eventually going to reduce inequality. That's just what I think, thank you. Wonderful question, so let me start in sequence. Thanks, Raquel. So in the book we discuss the other half of that duo, Engels a lot, because he had excellent analysis of the living and the working conditions. Marx actually did talk about automation, but his writings on that are a little confusing and confused, so we don't get into it. The other sort of issues you raise are absolutely right and important. So I know you're asking this partly to be provocative and in an intellectual way, so I don't wanna sort of say this in a negative way, but what you asked is exactly the Silicon Valley view. So there are many sort of semi-progressive Silicon Valley leaders, including Reid Hoffman, who I mentioned, Eric Schmidt, and many others, who are very worried about automation and their solution is we should do all the automation that we can and then we have to find a way of sharing the benefits. And I think the problem with that is, first of all, it's not an equilibrium in a political economy sense. Once you disempower labor, and I think taking away the most important things that people do in their jobs or their labor market presence, you're really disempowering them. Once you disempower labor, then it becomes very difficult to have very generous redistribution. And second, there is growing evidence, although controversial, that this isn't actually going to lead to a healthy society either. People, when they don't have a job, they feel very disenchanted, alienated, and so on. So it seems like the most immediate thing, if it's feasible at all, and that's where the feasibility issue and Rashid's question comes in, steering the direction of technology in a way that keeps on creating jobs, meaningful jobs, better jobs, and that's the way to redistribute is much better. So the alternative, yet another alternative, but I think is not that different from the Hoffman-Schmidt Zuckerberg solution, which is, well, you give some ownership of the means of production, and that's where your Marx reference comes in, and to workers and they become renters. But I don't think that's really fundamentally different from social protection and redistribution doing it, because if they do that and they don't work, it's gonna be the same thing. And even if you promise that, are you gonna be able to implement it? Even if you implement it, once they start losing those shares, they sell them in the market, are you gonna give it back to them and what type of, so the political economy cannot be bypassed through those mechanisms. Rashid, that's an excellent question. So, first of all, let me start by emphasizing, I don't want any of my sort of comments to be misunderstood in that I am, everything I said is not opposed to technology. It's all in the spirit of we want the technological change, but we want the direction of technological change to be much more worker friendly and consistent with shared prosperity. And moreover, again, showing sort of my neoclassical affinity, I don't think that you can have that technological progress by a very heavy state hand. So the private sector has to play a critical role. So then, put the two together, which is, I'm saying the private sector right now in the current institutional environment is generating highly distorted technology, but at the end, you need technological change, you need the private sector, so the only feasible solution, you need a better institutional framework and that institutional regulatory framework to change the incentives for the direction of technology. Now, your question was extremely well put. It has two layers. One is, is it feasible technologically to steer the direction of technology? And yes, I 100% believe that. Now I don't have hard evidence, but digital technologies are so versatile. And so in the book, there is a half a chapter devoted by many things that you can do with digital technology that create new tasks, new platforms, new information for decision makers at every level of skill with examples that tries to bolster that case. But then you were raising the second layer. In energy, it is possible to go to renewables or it's possible to go to fossil fuel. In digital, it's possible to go to automation and to work a friendly technology. But you were raising the second issue in energy, it's easy to find which one is renewables. Although I can tell you big oil companies are doing an amazing job of masquerading some fossil fuel technologies as renewable, but that's a different matter. But generally, I think it's feasible. And we admitted the book and I admit here, it's much harder, AI technologies, are they empowering managers to buy new tasks or are they empowering managers to cut the rents of workers? So there are many gray areas. So for that, I think specific technological choices, I think would not have worked in the energy field, it would not imposed by the government, would not have worked in the energy field, it won't work in technology. But there are many other things that can be done. So one of the other chart slides that I removed, for example, is that right now, it's not that we are creating a level playing field, we're actually in the US, and the same is true in most industrialized nations, the tax system provides a 20, 25% subsidy for automation relative to hiring workers or increasing worker productivity. So the first thing is that you would create that, you would remove that bias. So that if that's not enough, I think what you have to do is set broad goals for labor-friendly technologies and the role of workers is much greater. So one model there is Germany. So emphasizing that this is not just that global trend and choice is relevant, Germany is far ahead of the US in terms of robots, but it's embedded in a very different institutional structure. So work councils are consulted with robot adoption, unions just like UAW did, are very engaged in training for workers after automation. So firms can introduce robots, but then they choose or they're induced to retrain many of the workers for fewer layoffs and they also introduce other technologies that are complementary to humans in design, inspection, planning, where internally trained workers take many of these jobs. So again, the sort of combination of government oversight, civil society pressure and workplace coordination, much harder in the US, but I think those would be solutions that at least can get us started. And finally, first the clarification point, I'm sure you did not misunderstand me, but when I said ultimately everybody will benefit, I wasn't saying my own view, I was giving a caricature of not every economist, but a model economist view and a view that follows from some economic model. And why do I focus on people benefiting rather than inequality? Well, actually, I think inequality matters as well, but even Reed Hoffman's quote and other sort of Silicon Valley and many economists would agree that many of these technologies are going to increase inequality. So the claim isn't that Zuckerberg and Elon Musk are gonna benefit as much as the workers, but the workers are gonna benefit as well. And I'm taking exception to that view. So that's a sort of a stronger claim and that's why I focused on that one. I'm not sure whether I understood the last part of the comment, but if you mean by democratization of technology, technology going in a direction that's sort of broader in its applications and is useful for workers, is useful for citizens, consumers, users, I think that's exactly what I am sort of talking about and that's why just like you said, a very important part of the emphasis is about the control of technology. But there's a lot of talk of democratization in Silicon Valley that was Facebook and Twitter were gonna democratize communication and news sharing. Some of that is not clear what democratization in general means in that context. Thank you. Just very quickly in response to Raquel's comment, just a clarification that I don't think I said it, but I didn't mean to say that social protection was going to be the answer. Just picking up on the round's point that labor protection, social protection have been so important to keep control on the power of firms on employees. But I completely agree with you that it has limitations. One of the limitations is whether we can pay for it and how we can pay for it, as I mentioned, but of course it is also a very limited tool to start with. I think all of the comments actually go in the way of pointing at the importance of generating income to start with before providing social protection from the state. And most definitely the design of the tech system is going to be crucial. Thanks, Marcella. It's time to bring this other lecture to a close. I know we would like to carry on a little bit longer, but it's already almost six o'clock. I don't, I don't absolutely agree with you that we have to, as economists, not take technology, not in the sort of optimistic way we do. But I think one key takeaway from what you said, and that's one of the optimistic view on this, is that how do we develop the worker-friendly technologies that both create jobs or good jobs and are also productive, right? This sort of balance is so important. I think we would like, I would like all of us to thank Darren and Marcella for the wonderful lecture and the wonderful comments. So let's thank them. So before I end, I should mention that UNEWIDER has a very exciting project we just finished that looks at the role of technology, utilization, automation on workers, on inequality, and also on the changing nature of work. It's a project that we have several research outputs on our website, led by Carlos Graden and others, Simone, who's also here. And do take a look at our publications because in our view, the first project one has done where we have evidence on the effect of automation and automation on inequality and job polarization, so on. So that's something I just wanted to mention. So let me stop here and say that I know many of you are here with UNEWIDER even for the first time. I do hope you're going to sign on to our newsletter so you're gonna get to know more about the work we are doing. And hopefully we'll see you again in a future UNEWIDER event with us not just. Thank you.