 Just a real consumption of manufactured products prior to 2000 was less. The growth rate was less than for services. That's not been the case since 2000. It's actually been growing faster. And that's consistent with there being a surge of low-cost imports increasing consumer demand in this country. Finally, let's think about productivity growth. Without computers is not much higher or only somewhat higher than that in private industry. There's a very nice paper by Martin Bailey and Barry Bosworth written a few years ago that showed that once you took out computers over the 1987 to 2011 period, productivity growth was the same in manufacturing as it was in private industry. Now, flip that around, if productivity growth is the same, then for this most of manufacturing as it is for the private sector, then it can account for none of the relative employment declines. I do want to say that I've done alternative calculations with different data. And I find it doesn't, productivity is not the same, but it does account for the slower output growth. Does, again, in this accounting sense, make up most or account for most of the relative employment declines. The point, though, that I want to make is that there's just no prima facie evidence for the productivity growth, for the automation narrative, that productivity growth caused the employment declines in manufacturing. Understanding the causes of what was going on really requires a good look at why output growth for most of manufacturing was just plain slower, a lot slower than it has been in the private sector overall. And I'm not going to weigh in on that per se, but before I move on to some research evidence on that topic, I think it's useful to just pause and take a look and think about what productivity measures. Another criticism that I have of much of the discussion that takes place over automation is that productivity growth is often equated with automation. It should just assume that that's what's happening. Productivity measures many things. Here I'm really referring to labor productivity growth. It's picking up many different things, including automation. We've already seen, for example, that product improvements in computers and semiconductor doctors caused tremendous productivity growth in that industry and, by extension, in the aggregate manufacturing statistics. That's not automation. Also, one often sees a rapid increase in productivity or rapid productivity growth during major periods of restructuring, as we saw in the early 2000s. That isn't necessarily an automation of what's going on. In fact, one would expect, for example, if on account of increased imports, international pressure from manufacturers in other countries, there was a decline in the number of factories. The first things that would be pushed out would be the most labor-intensive, and that's going to result in a mechanical increase in productivity, measured labor productivity growth. In international trade, competition can cause a sorting out of what's made in here, a change in the production, and lead to increased productivity. Another thing that can be happening is that with globalization, the stages of production that occur here in the United States can just be changing. In the paper that I wrote for this, I gave some nice examples from actually a case study by Thomas Holmes of the furniture industry. Much of the particularly home furniture industry was decimated in the early 2000s. Entire segments went out of business, and what remained, much of what was produced here, was fundamentally altered. For example, in upholstery, the very labor-intensive, cut-and-sew aspects of that were off-shored to China, and minor assembly continued back in the United States. That's going to result in a measured increase in productivity growth that doesn't have to do with automation per se. It's also been found that international competition may spur investments in automation. In that case, the direction of causality is a little unclear, because the competition from overseas can spur manufacturers in this country to increase investments and automate faster than they might already. Otherwise, I've done so. The bottom line is that productivity growth does not per se cause employment declines. The strong argument that automation is primarily responsible for employment declines and both relative and absolute employment declines is largely based on descriptive evidence. To really understand what's going on, we need to look at research evidence. That's what I'm going to very briefly turn to now. There's been a significant and growing body of literature that in particular has focused on what's happened with the precipitous decline in employment since the 2000s. Exchange rates, industry subsidies, corporate taxes, tariffs, non-tariff barriers, and the like can all affect the relative competitiveness of manufacturing in the United States. And globalization can reduce, may reduce domestic manufacturing output growth through imports, and a lot of the focus has been on sort of looking at the growth of imports, particularly from China. But we should also remember they can also operate through slow export growth in exports. Companies may choose to close factories in the United States and move them to Mexico or to Ireland or somewhere else, or they may just choose to expand their production overseas. Much of what multinational companies, when they manufacture overseas, they often are manufacturing primarily for export markets. Some of that may come back to the United States, but by no means all of it. And I mention that because we don't capture that dynamic very well in the statistics, so it can be a bit hard to study. Before I sort of say a few words about the research literature, I really want to emphasize that no single study captures all aspects of globalization and its effects on manufacturing employment. There's weaknesses, there's big holes in the literature. There probably always will be. But collectively, say, the research find its point to large adverse effects operating through a variety of mechanisms. One of the most widely cited studies by Otterdorn and Hansen concludes that about 25% of the decline in manufacturing employment can be attributed in the 2000s can be attributed to the growth of imports from China. A completely different study using a different data and methodology looked specifically at changes in trade policy after 2000, causing an increase in imports from China and also finds very large adverse employment effects. Interestingly, the sets of authors on these two papers have recently looked at the impact of imports from China and on other things and have concluded that they have led to declines in production, investment, and patent activity in the United States, which can, of course, over time have spillover effects on employment. One study, the final study that I'll highlight, and there are others that I'm not going to cover, I don't have time, looks actually at an appreciation of the dollar, a large appreciation of the dollar in the early 2000s on the manufacturing employment and can account, depending on specification, it can account for roughly half or even more of the decline in manufacturing employment. The reason I like this paper, I think it's important, is that it's not just looking at Chinese imports, it's looking at all, effectively, it's capturing imports and exports, the effect of this exchange rate appreciation on both. One of the very interesting findings from this paper also is that the job losses from a temporary exchange rate appreciation tend to be permanent. That is as if an exchange rate appreciation induces firms to invest overseas or whatever the structure of production shifts globally, you've got those sunk cost investments and when the dollar returns to its original level, it's not necessarily going to be the case that those jobs will come back. So it's a different, often you hear the story, the jobs aren't coming back because of automation, but there's another dynamic operating too, which is that once you lose it, it is hard to get it back because other countries develop expertise, there's sunk cost investments and the like. Oftentimes we want to say that, well, if this particular study explains, I can say a quarter of the decline in manufacturing climate, three quarters is explained by automation, but that doesn't follow. And it turns out that several studies have tried to look specifically in the 2000s at the effect of automation, specifically investments in IT technology on productivity and employment declines and have found none. Okay, one is Fiasa Moglu and others in a sort of companion paper to their China paper, Otterdorn and Hansen, look at industries experiencing increased competition for China and again find that it led to significant employment declines, but they also in parallel looked at industries susceptible to computerization of routine tasks and found that while it changed the composition of jobs within the industry led to no employment declines. In a much publicized paper, working paper that came out last year by Asimoglu and Restrepo, they found that industrial robots have the potential to cause large dislocations to workers, but that's in the future. So far the number of industrial robots that have been invested in can account for no more than a trivial share of the employment decline in manufacturing. And then finally I'll just write so what I mentioned, what I think of some related literature, which is that a couple of several studies actually have found that the rise of markups in the 1980s since the 1980s and offshoring of labor intensive processes account for the rise of capital's share in the economy and the decline of labor's share in the economy. And in other words, it's not capital investment per se. It's these other things, rise and markups, offshoring. And that sort of stylized fact is inconsistent with there being a large technology shock causing the employment declines and capital share. I'm gonna just close with a few thoughts on sort of why manufacturing matters. People sometimes may think, well, manufacturing employment is now under 10% of the economy. We shouldn't worry about it so much. But there's been a lot of domestic outsourcing. Half or more of the employment used in the manufacturing of products actually is employed outside the manufacturing sector. Also it has very large spillover effects on local and national economies. So that you get typically multipliers of manufacturing employment of orders of magnitude of two or more. And one of the most robust findings in the economics literature is that plant closures and mass layoffs have large adverse lasting effects on workers in communities. Workers on average experience large and very persistent, persistent across decades earning losses on average. Some do fine, but many don't. And the size of the shock matters. Large adverse shocks can send regional economies into a downward spiral and depress regions for decades. We've known this for a very long time and it's been documented in a better way recently in the literature. The collapse of manufacturing employment while not the only factor behind the woes and labor markets in the United States in the 2000s has certainly contributed to employment declines in employment rates especially among less educated men and women. Now we have a relatively low unemployment rate today but we still see in the form of very weak earnings growth and very low labor force participation rates among less educated sort of greater structural problems in the economy that at least in some way are related to the client of manufacturing employment and there's been evidence on that. Okay, also as I mentioned the loss of manufacturing production has led to the loss of research and development which has longer term implications for the health of the U.S. economy. So to conclude, my main point is that claims that automation primarily caused the relative and absolute decline in manufacturing employment are just simply at this point in time not supported by the evidence. I'm not saying that automation played no role, right? Certainly automation is happening but there's no prima facie evidence nor is there evidence for the 2000s that a big surge in automation was the primary cause of what we've been seeing. Studies do find trade played a significant role in manufacturing employment declines in the 2000s but again I want to emphasize the research is incomplete, there's big holes in the data foreign competition does burst investments in automation and hear that direction of causality is ambiguous and the precipitous decline in the 2000s has had very large adverse and lasting effects on workers in the economy and this, I'm not gonna talk about trade policy but this obviously raises questions about policies that may have triggered the sudden shift in global structure of production. I am guessing that Carolyn and Josh will weigh in on it but thank you very much. Thank you. Thank you so much to Dr. Susan Hausman from the Upjohn Institute and now we'd like to welcome Dr. Carolyn Freund from the Peterson Institute for International Economics and the World Bank. I'll ask each of the discussants to stick to about seven minutes, thanks. Thanks, it's my pleasure to be here. Let me see if I put a PowerPoint here if I can get it onto the screen. Is someone doing it from there or? It's the, it's this one. Yeah, I did it in Wikipedia. So, you know, let me start by saying I agree with most of what Susan said and I think she gave a really nice presentation. I think I'm gonna quibble about the size of automation and skill bias technological change and demand shifts and things like that versus trade but I do agree that especially in the 2000s increased imports, you know, especially from China played a role in the decline in US manufacturing. I do want to take a step back and just remember why we like to trade because when you just focus on manufacturing you can ignore that part of it. The main reason is actually their consumer gains. So consumers gain when you go to the store and prices are lower. That's a real income gain for people and especially for people who are the most vulnerable the poorest people who spend a really large share of their income on goods. So I don't think we should forget about that when we talk about this because while there are the workers in the manufacturing sector there are also all the people who work in the service sector actually the majority of Americans who are benefiting enormously from this and especially some of the poorest. And if we do want to think about the loss in jobs for manufacturing the right policies are probably about training, adjustment, et cetera, as opposed to restricting trade. And then finally, this was a really special period. It was a period when China was running a current account surplus of around 10% at some point and the US was having a big consumption boom. So I think we have to take macro policies into consideration that in some sense we drove this ourselves not to do with trade policies but to do with our policies that discouraged savings primarily. And then finally in terms of just a broad overview let's also not forget that 500 million people were lifted out of poverty in China from the early 80s until since the early 80s until 2010 or so. And part of that was because of productivity improvements there, reforms there and trade and that's another bonus of global integration that we don't want to forget about and some would also argue that integration, talking with other countries, business links, et cetera have supported a long period of peace. So just to think about those as an overview. In terms of the quibbles with this paper one is I think demand matters a lot and it has to do with the size of the economy and I'm gonna go back to this kind of relative productivity hypothesis that Susan talked about that productivity of manufacturers increases but how many TVs can you have? How many cars can you buy? And that's gonna put a constraint, demand is shifting towards services as countries get richer and so we just can't expect jobs to be coming from there. I also have a quibble that I've seen this before it's been a common thread in this literature of let's take out the computer sector and show it's nowhere else. Well if we're gonna take out one sector we should take out some other sectors too because now we're weighting it towards the sectors that had the lowest productivity growth since we're taking out the high productivity growth one and I'm gonna show you some data that that would matter as well and finally I'm gonna talk about just some other things like macro context and other nitpicks one could have with the data besides the computer price index. So first this is the alternative picture that people show instead of looking at the number of manufacturing jobs let's look at the share in employment and there this period of rapid decline is not there. So it's been kind of a steady decline in manufacturing employment and the difference has to do with the fact that during the early period when manufacturing growth was stable so the left picture's pretty much the same picture that Susan had where you do see the precipitous decline in manufacturing employment part of it was because productivity was increasing but demand was too as workers were coming on the labor force incomes were overall income was going up as a result of that demand as well as there being plenty of workers to hire. In this later period if you look at the working age population it was much more stagnant so the manufacturing sector's competing with other sectors for employment much more which makes it more difficult to maintain employment and also there's less of a demand boost for all those goods that they're producing. And here's what the technology looks like and I just put a few sectors here just to give the idea if we take out computers for some other sectors actually if computers were on this chart on the left of labor productivity it'd be way off the scale it's at like 11 where this scale in productivity growth is much smaller so you'd get the flat lines that Susan showed for other things that computers just dominate but there is a difference even within some of these other sectors and what you really see is that even primary metals, machinery, transport equipment, et cetera had much higher productivity growth than our services and hotel and food services, truck transportation, oh wow I have one minute so let me just quickly go through the rest and you can see it in prices as well that the prices have fallen much more for goods than for services so that's where the consumer gains come in but also this leads to a shift in part because of this demand for services, the prices are rising there but the share of what people are spending as countries get richer they spend less on goods and because prices are falling you spend less on goods and you can see this very clearly in this shift towards consumption of services this is where in this picture you can see again I'm trying to distinguish between industries that if we wanna take out computers and electronics we should take out probably apparel and textiles as well on the other side or something of equal size so the size of the bubbles are the size of the sectors and it shows the employment change which is kind of exaggerated here because of the scale versus the output change they both are going to roughly 0.8 but the point being that one all sectors experienced a decline in employment it wasn't specific to the declining sectors because productivity was rising in part in some sectors where output and employment declined it was probably more related to trade I'll concede that but we also saw really big productivity gains especially in transportation equipment and I think that should be recognized it's a huge sector that had a big productivity increase not as big as computers but big and finally the other issues that I wanted to talk about one and I said it at the beginning the macro context matters a lot average unemployment in this period was very low the consumption share of GDP was increasing if you look at imports to GDP looks very much like other countries over this period what's really different is exports to GDP so something was wrong in the US that as we were as competition was affecting our manufacturing industry we weren't growing our exports in the same way as other countries and finally if we're gonna nitpick the data and just take out one sector and say look then there are other nitpicks as well transfer pricing and profit shifting matter a lot there's incentives out there to over invoice imports as a company or imported inputs which then pull your profits down and your taxes down and to under invoice exports and some have estimated this alone is half of the US trade deficit and then the price index on autos and machinery probably understates advances I lease a car and I have not had a price increase on my car and yet I keep getting tons and tons of more features since 2000 since like for the last eight years or so so there's something going on that if we're gonna argue about the price index on computers then we should probably look into some of these other price indexes that may understate real increases and I'll wrap up there cause I think it's time thank you. Thank you very much Carolyn and next we'll hear from Dr. Josh Bivins from the Economic Policy Institute. Sorry I'm gonna try to find my, okay thank you. So I'm gonna say a couple words I'll start by saying I think Susan's paper and presentation are really thought provoking and aberrably careful in their claims about what economic analysis can really precisely say about the various facets of globalization and manufacturing job loss and so I'm going to take my job to be a little less careful and basically say it's true we can't be super precise about this but we can look at some sort of broad brush data and try to bracket the effect of rising trade deficits on manufacturing employment and it's gonna be really hard to escape the conclusion that those effects are pretty large so before getting to the precise question at hand I also have two quick asides. First one the extent of globalization's pressure on the living standards of American workers is not equal to its effect on manufacturing employment. We could have balanced trade and there might still be large downward wage pressure on non-college wages in the US economy and in fact I think these wage effects cumulatively are probably quite a bit larger in terms of the losses they put on American workers than the damage done by manufacturing job displacement. I think the manufacturing displacements they're the most acute damage felt by individual workers for sure it is much worse to completely use your job than to have your wage sort of chronically pushed down over a long period of time but the wage effects are really big and so that's one aside. Second aside, trade deficits can certainly have effects on overall employment, not just manufacturing employment particularly when the economy's growth rate is demand constrained. I mean, yes when the economy is unambiguously at full employment, trade deficits essentially just change the composition of unemployment pushing jobs out of tradable sectors into non-tradable. I would argue we have not been in an unambiguously full employment economy for a long, long time. The graph I'm showing you here is basically interest rates and if you just concentrate on the far right of that graph we've had interest rates essentially bumping right above zero for about a decade and the standard story for how trade deficits do not harm overall employment is yes the trade deficit reduces employment all less equal but you give this big inflow of capital that pushes down interest rates and that creates jobs in the economy. Well, pushes down, capital inflows push down interest rates, yes. We do not need lower interest rates in the US economy right now for job growth we need something else because we've had zero interest rates for essentially a decade and so I would argue that countervailing effect of capital inflows in pushing up jobs over the past decade has been extraordinarily weak so trade deficits can have influences on overall employment for sure not just manufacturing and then finally this is boring but it's really we're saying if we are going to tolerate a really large trade deficit and we want to be at full employment then we have to tolerate a really large budget deficit or a really indebted household sector like there's no getting around that fact and so someone who wants to say we're at full employment and we have a large trade deficit and that's fine I really don't want to hear them talking about how we need to reduce the budget deficit unless you think there's a one way causality from budget deficit to trade but there's not. So let's get to the real question at hand trade deficits in manufacturing job loss this I realize is just exactly the same graph that Susan put in hers this is to me looking at the right it's the implosion post 2000 that to me is the really interesting thing here it's one thing to say manufacturing's employment share has gone down for a long time it has the overall level was extraordinarily stable for about 35 years between 1965 and 2000 and then it just falls off a cliff and again my contribution to sort of dismissing the really easy was all automation explanation for this if you look at productivity growth between 1965 and 2000 when manufacturing employment was really stable and you look at it post 2000 when it's not stable at all it's really hard to see much difference there productivity growth was really fast in manufacturing from 65 to 2000 and yet we did not lose 33% of the jobs in a 16 year period and so basically the second column is basically hours that that's employment in the industry that's the annual loss over that time period so the way I match up these two things I got best a quote unquote acceleration of productivity of 0.2% in the latter period maybe it gets you 10% of the explanation for what happened post 2000 I think even that's actually a little high so can we be more precise than this if we don't think it's productivity but what can we say that it is and I in the interest of meeting my seven minutes I'll skip one slide focus on the bottom expression here and I'll tell you what that means in one second but the logic is going to be the changes in manufacturing employment are basically just equal to the change in domestic output how much stuff factories in the United States are producing minus the change in productivity so if you produce 4% more stuff but productivity rises by 4% employment doesn't change at all you didn't need to hire anyone else to produce that extra 4% your existing workforce just got more efficient but one thing that we often miss is the change in output is a function of essentially two things I mean a lot of things but two things domestic demand what do US consumers actually want to buy from the manufacturing sector and then the wedge the trade deficits introduce between that domestic demand and domestic output so maybe US consumers want to buy a trillion dollars worth of manufactured output in a given year but if on that 200 billion of that is satisfied by imports you're going to see a wedge between what US consumers want to buy and what is actually produced by US factories and so that's what I'm trying to get at with this last expression with the little hats over it that L with the hat over it that's basically employment in manufacturing and that's going to equal the growth of domestic demand that's that YD with the hat plus the change in the wedge introduced by the trade deficit that's that little D and then minus productivity growth over that time period and this is going to let us do a sort of very simple accounting decomposition first thing I want to convince you though is that little D the wedge that the trade deficit puts between what US consumers demand for manufactured manufacturing output and what producers actually produce it's pretty big this the sort of blue area is output real output in US manufacturing and then the red is domestic demand and I'm defining domestic demand is basically just output plus manufacturing imports minus manufacturing exports and so it is the manufacturing goods demand that sticks in the United States I'm taking out exports because that's foreign but I'm adding the imports in and so this is the wedge the rising trade deficits are putting between output between demand and output and you can see it started to become big and it's grown significantly over the past couple of decades and so if you just take that expression from before that had employment as a function of domestic demand the wedge between output and demand and productivity you can actually get some numbers and the numbers I looked at are between 1997 and 2016 for a couple of reasons choosing these years one actually 1998 is the manufacturing employment peak in the US economy as pretty striking because between 1997 and 2000 the US economy was roaring private sector employment rose by 10% over those three years that is an extraordinarily fast growth rate yet manufacturing employment declines what's going on there in 1997 if those old enough to remember it's the Asian currency crisis we start to see really large trade deficits start to develop because of what's happening elsewhere in the world and that's one reason to choose 1997 because that's when I would argue our modern era of really large trade deficits begins and then two data's much easier from 97 after in these series so that's why I chose it so if you look at ours basically people have shown this before we see about a 34% decline in employment hours in manufacturing over the same period though we actually have domestic demand rising by 34% over this time period there was actually relatively robust demand growth in that time but the wedge between demand and output that trade balance wedge that knocks about 18 percentage points as a share of manufacturing output off of that so that's domestic demand US consumers wanting to buy washing machines and dining room tables and TVs but it's not being satisfied by domestic production it's being satisfied on net by foreign demand that leaves us with output growth of just 16.4% and then you get the productivity growth of 50% over that time period so when people say productivity growth in manufacturing employment they really mean the interplay between demand and productivity growth like no one thinks that employment should have declined by 50% they basically say productivity rose faster than demand and so it's sort of the domestic influences so those domestic influences domestic demand growth minus productivity that only explains about little less than half of the total employment decline in manufacturing over that time that wedge from the rising trade deficit explains about half and so to my mind just looking at this aggregate data it is really hard to not see the rise in manufacturing trade deficits is really co-lead with what has happened in manufacturing employment over that time and then final 10 seconds so why did manufacturing trade deficits rise substantially over this time period it's overwhelmingly misaligned exchange rates the US economy still has a very overvalued dollar did for a lot of that period that should really be sort of the laser focus of people who are really concerned about the effect of manufacturing trade deficits on employment thank you thank you Josh and our final discussant was to be Professor Michael Hicks from Ball State University he wasn't able to be here because of weather Larry Michelle former EPI president is going to read Professor Hicks's prepared remarks thanks Larry thank you Michael was concerned that the the expected snow would allow him to get here but not to leave and I guess he doesn't like Washington um not that okay I'm going to read his remarks and as supplemented by an email from him this morning um he starts out acknowledging the um uh the significant decline in uh factory employment that's been very disruptive uh and he writes at the same time both real GDP and real value added production in US manufacturing likely peaked in 2017 and absent business cycle disruptions have grown steadily across the historical horizon of data likewise the United States has experienced negative net exports since 1980 and the balance of trade with China was roughly 350 billion in 2016 given these data a significant body of research has attempted to explain the causes and consequences of these employment changes Dr. Hausman's paper offers a useful and concise restatement of much of this analysis she does so as her title suggests in two parts which I discuss in turn the causation of job losses is an important ingredient to policy innovations despite common perception data on employment trade and output suffer significant conceptual challenges so such issues as domestic outsourcing of non-production tasks imprecise trade data or BLS adjustments for quality all affect conclusions which can be drawn from the these data Dr. Hausman explains much of the research on these issues but there are two especially notable issues that merit comment first the limited work on BLS bias some of which is not reported suggests more symmetry than this review suggests and just to summarize he's saying Sue says that there are studies which show that the BLS productivity is overstated he's pointing out that he thinks there's studies that say it's understated second Dr. Hausman corrects a common error by economists to two casually attributing productivity gains to simple automation this was covered in her earlier work with Michael Mandel and in papers with Bartik and Sturgeon more research on the sources of productivity improvements may uncover such diverse contributions as better education changes in union strength and robotics my only real critique of this section is that I believe it overstates the disagreement regarding the relative contribution of trade and productivity reported in the literature part of the disagreement is total logical there was a surprising consensus across several studies using different techniques that identify pure trade-related manufacturing job losses during the 2000s as between 750,000 and one and a half million or between about 10 to 30% there's also agreement that some of the productivity related job losses are motivated by competition induced by imports so do we call this a trade or productivity related job loss I think an important contribution of this work is to motivate economists and policymakers to talk more broadly about measurement issues I do think that Dr. Hausman is right and that there is an unlikely to be that there is unlikely to be agreement on the figure but I think there's less disagreement among researchers than appears in either the policy debate or in the media the consequences of many frank job loss are as in Dr. Hausman suggests among the most robust findings in labor economics in conclusion Dr. Hausman has produced another important analysis of the puzzle regarding the effective productivity and trade on manufacturing employment I asked Dr. Hicks to expand his remarks in order to directly comment on the one of the main issues that Sue raises which is can we attribute the erosion of manufacturing employment to automation or to you know rather than anything related to output related matters or trade and so here's what he says well I concur that slow output growth is a contributor to lower levels of manufacturing employment growth I don't think it is necessarily relevant to exclude computers from the discussion Susan may be correct there's a theme here Susan may be correct that BLS weighting biases upward productivity growth but a full accounting of its total effect on manufacturing would require conducting some of the same analysis for non-computer inputs that was the point I attempted to make about earlier criticism of the BLS manufacturing goods are increasingly platforms for computing capacity that alter the value consumers receive from the product that value likely exceeds the value of the products produced separately thank you thank you very much Larry thank you Dr. Ball in absentia before we go to the audience I'd like to give the panelists a little bit of a chance to go back and forth with each other a couple of interesting issues have been raised here and I think one of them in particular for Sue is whether it's arbitrary to exclude computers from the manufacturing output and why it is that we make that choice and what that tells us and invite Carolyn also to respond thanks sure well first I want to make really clear I think is it soft so okay so I want to make really clear that I am not arguing that the adjustment of price deflaters in the computer and electronics products industry is wrong I'm not saying that productivity is biased that the government estimates of productivity growth there is a very robust debate within the small corner of the economics profession that pays a lot of attention to price deflaters okay and it is absolutely you know about whether these adjustments are too big or too little and there is a very large debate going on right now about whether we should be more better adjusting other sorts of products and services goods and services particularly IT goods and services better for changes in product quality my point rather is really two-fold one is that these productivity changes are not picking up automation and too often in the public discussion we're pretty sloppy about that we're assuming that productivity growth is you know implicitly or explicitly we're very often assuming in this debate the productivity growth is picking up automation that's not okay and and secondly the reason I take out I think it's useful to take out computers is because when you are looking at the manufacturing numbers the trends are just so dominated by what's happening in that one industry that it's really hard to interpret I think the statistical agencies will be moving increasingly towards more better adjustment for product quality I think it's going to make things very difficult more difficult for people to interpret the statistics in the way that they're used to interpreting them so Carolyn, you want to add anything? Yeah, I would I would just add that I think there is there tends to be a view when we talk about productivity growth and automation that we're saying the industry is losing jobs because these jobs were automated I think that's part of it but I think it's a small part the point I'm making is the really big part is about price elasticity and income elasticity they're just not big enough so that when productivity rises and prices come down that people want so many more of these goods and then if people want the same amount of these goods or maybe a little bit more and productivity has risen a ton that's going to cost jobs so it's not automation that's costing jobs it's this relative price problem and sort of structural change as well on top of that which has led us to consume more services than goods so I think that productivity still plays a really big role and then my mind automation in combination with this relative productivity hypothesis is the lion's share I still think trade especially in this period you know was probably you know the 20 to 40% people want to allot it I don't have a problem with that and it is important we do have to think about it like Josh I think it is much more to do with the real exchange rate than trade policy and then just on taking out one industry you know why should I take out one industry I have a whole set of data let's do that with all our data take out one industry and I can show you almost any correlation so I do think it's problematic to take out one industry or at least then let's talk about all industries because there's some interesting things going on as I mentioned especially in transportation equipment machinery things that the U.S. is really good at producing that we should also focus on so if we're going to take out one industry let's at least look at all disaggregated data not computers, everything else but Caroline do you agree that there's something unique about the computer hedonic indices and the way we calculate price changes and quality changes so that you have this kind of small very small industry which is actually in decline and I think it is contributing to the counterintuitive sense where I think a lot of Americans certainly manufacturing employment and job and trade was a huge issue in the 2016 presidential election and has been a contributor to growing inequality and wage stagnation and the decline of a lot of communities particularly in the Midwest and so people have this sense and then a lot of pundits come out and say you're all wrong, you're all idiots there's nothing wrong with manufacturing because manufacturing output is in fact growing and there's nothing happening here except that you're all idiots and there's no problem with trade because it's all automation and so I think what Sue is trying to pull out is the idea that there's something actually strange that happens in the data the adjustment that happens which is when you say your computer this year has twice as much computing power as it had last year and yet it costs pretty much the same or $10 more then you all of a sudden have bought twice as many computers so what we see about the health of output of GDP what we see about the growth and productivity is skewed by some necessary adjustments to computer prices and quantities but that they are also misleading us when we try to think sensibly about what's actually happening No, I think that's a fair point I would just say that I think we have to be careful then to do that in other industries as well that doing that only with computers skews the discussion in one direction But Susan, does that fix the problem if you took out textile and apparel, for example? No, I'm just gonna make sure I'm sorry I don't think that the other, you know the productivity growth in the other sectors isn't going to make that much difference in terms of to change the overall picture what really changes the picture and I have experimented with this but it doesn't change the overall picture in fact another industry where there is a lot of quality adjustment is autos I've tried taking out autos but just isn't enough to make a difference computers is, you know, sui generis it's unique among itself in sort of the magnitude of the changes I wanna, while I have the microphone I did wanna reply to one of Carolyn's other comments and it is very common to say well this study, repeating what I said during my remarks, this study found 25% this study, that study found 30% so the rest is probably automation that's a problem right now, I mean what you should be saying is what was the technology shock that occurred in the early 2000s that caused a nearly 20% decline now near 30% decline in employment and people that have tried to look at that have come up short now I emphasize that there's holes in the research debates still open but that's the question that you have to answer it's not sort of saying because research is slow it's hard to kind of, as you know Carolyn it's hard to get sort of very convincing causal evidence so it's looking at this effect of this trade policy on employment and so forth so it's not complete what needs to be done is for researchers to come up if you really believe the automation story is going on for some compelling evidence that there was a technology shock in the early 2000s that caused a collapse and that was specific to manufacturing that caused that decline and then just quickly I completely agree with Carolyn the trade and I perhaps didn't make this as clear as I should have during my talk we all benefit a lot from trade nobody's, at least not I'm certainly not arguing for throwing up walls at the border I think that what was really unique and special or special isn't the right adjective but what was really unique about the early 2000s was just the magnitude of the shock I was trying to get at that in my discussion about dislocated workers the size really matters and we had a very large trade shock then and that's where I think the problem is most serious and it's that it's that policies whether it's exchange rates or anything else that may have led to large sudden shifts in the structure of global production that can be especially problematic for an economy I just wanted to add about this thing that well if it's not trade and we don't have this big productivity growth then it has to, we have to come back to trade that you really look at the data and just the timing of when the job loss happened it's one thing, recessions so in recessions that's when the job loss happens and so then it goes back to the demand the demand suddenly drops you've been holding workers for a long time as productivity has been increasing and then during the recession you let those workers grow so I don't think if we look at the timing of when they fall versus the technology in trade it's the right way to look at it because it's extraordinarily lumpy when that job loss happens it wasn't a continuity over this period when China was increasing exports to the US it was really two distinct periods when a lot of job loss happened that happened because of holding too much labor and then a recession hits and that forces this readjustment Yeah, just on that point because that's related to what I wanted to talk about I think one way to think about this the way I think about it anyway is like productivity growth as we've all shown it's basically been on relatively the same pace for a long time it is the background music to the manufacturing sector and then what changes to cause job losses is what's happening to output growth around that and I think two things post 2000 you basically have a huge decline in output growth relative to the trend and relative to productivity and I think one, there's the obvious answer of rising trade deficits that has put that big wedge between what US consumers demanded and what US factories produced and then it's true we've just had a very slow growing macro economy since 2000 and I think that is we've got the background music productivity going pretty fast in manufacturing and then you downshift the overall level of growth you're going to run into some job problems in that sector I think in regards to the lumpiness of job losses though I think there's two tells that it's not just about recessions per se one that the evidence I talked about in my talk we had a roaring economy from 97 to 2000 in zero manufacturing growth actually a slight decline as trade deficits rise after the 2001 recession which was heavy in manufacturing but mild overall we got none of those jobs back before the great recession it and we got none of them back because even as the overall economy grew and US consumers started wanting to buy things again they were much more likely to buy imports from China over that time so I think on the one hand yes most of it happens during recessions but the complete failure to mount any recovery at all in between those recessions to me is a big tell that sort of the trade deficits were a big player Thanks very much to all the panelists and we do have time for some questions we'd like to open to the audience please wait for the microphone state your name and affiliation if you have one and we'll take maybe two or three questions and then allow the panelists to respond Robert Blecker in the back Hi I'm Robert Blecker from American University and in the interest of full disclosure I'm an EPI research associate so first of all I would actually like to be a little heretical and challenge the idea that we have to accept the productivity numbers for computers and I understand that I'm not talking about the exact numbers but to me the way we measure variables should depend on the purpose of our analysis so if we want to know and I think Michael Hicks put it very well when he said the value consumers receive from something like a laptop or your car yes we should do the hedonic adjustment of course that computer is far more powerful than the ones we used to have but if we want to know how productive are the workers in the factories who put the chips in and put it all together or even the workers in the assembly line somewhere who make the chips they're not actually making more chips or more laptops or more cars than they used to it's just that the things they're plugging in because of knowledge increases and technological innovation are far more create more value for the consumer but they're not I would challenge the idea that in terms of productivity that's really different a second point relates to some of this conversation at the end how do we account for the fact that there's this precipitous decline in the total amount of manufacturing productivity sorry manufacturing employment and yet this continuous decline with no change in the trend in the share of manufacturing employment and the answer has to do with something Josh was hinting at at the end the stagnation of the economy secular stagnation since somewhere around 2000 if you look at the total number of jobs created in the US economy from 1960 to 2000 it was about two million jobs a year since 2000 I haven't got the latest numbers but it's well under one million there was less than half of the annual rate of job creation overall now why are we having this stagnation obviously I don't have time to go into that but the only reason that the ratio is doing the same thing when the numerator has gone off the charts is because the denominator is also not doing very well Thea thank you for putting on this terrific program I want to make two points I was on the banking committee staff from 83 to 97 and I saw the movement from stakeholder to shareholder capitalism which is in there are two chapters in the book called the Golden Passport about the history of the Harvard Business School which describes this phenomenon the second thing I was on the China Commission for 10 years and saw the movement of what happened once we got China into WTO now people say we didn't give China a different tariff after they were in them but what we did we gave it to them permanently in other words we could always take MFN away from them until they came into the WTO without MFN the average good would have faced a 42% tariff in the US with MFN was probably lower than 4% so that's the key thing and then when our companies are driven by shareholder value to increase the profits for their shareholders and the other countries are offering them subsidies, underpriced exchange rates it's gonna cost them they're gonna move your production from here to there so I'm wondering have you guys focused on shareholder capitalism as a problem or a factor in all of this that has happened to the American worker over the last 15 years which has led to results in the last election? Hi, I'm Alan Tonelson and I write the reality check blog and thank you all for this great conference and thank you all for these very high quality exercises in economic analysis and I've got one observation and then one suggestion that I hope you'll all react to one, it's really a shame that it's all necessary and I say that because you all agree that there are big data flaws here that you've gotta try to work around and circumvent and try to explain and yet there is a big body of data that is out there that could explain a great deal that no one looks at because there's no obligation on the part of the major players to make it publicly available and what I mean by that is the major players especially when it comes to job offshoring or of course for the American economy US based multinational companies they know exactly how many workers they've moved all around the world they know exactly what they make because if they didn't have this information they probably couldn't be profitable so wouldn't it be great if every company in this country at least above a certain size had to report every year how many jobs have you moved overseas how many jobs back we do have actually of course labor department data that tells us how many workers have lost jobs from foreign competition but we've got flaws there too because a lot of the folks who lose jobs don't know that they have this option which leads down the road to various retraining programs so it would be great if all the organizations up here and everybody else who studies trade policy started pressing Washington to force multinational companies to release this data and then we could actually know what's happening rather than have to engage in all of this very sophisticated but ultimately unsatisfying economic analysis. Thank you Alan. We have one more quick question and then we'll allow the panelists I'm sorry just wanna squeeze it in. Quick question but a sort of large one my name is Carolina Young I'm a new APSA congressional fellow in Senator Mark Warner's office I believe our office has been very concerned about future of work and supporting workers in the manufacturing industry in particular and so one of the questions that I had for the panel was sort of policy recommendations for the future of manufacturing in particular there are lots of little there are lots of policy recommendations that are part of this debate in terms of wage stagnation in terms of exchange rates in terms of all sorts of things upscaling for workers in particular so I'd love for the panel to talk about those different things. Thank you. Thank you Carolina that's a great last question and with that I'd like to allow the panel starting with Sue going down the road here to respond to all four of those questions which were all excellent thank you. Okay so first on Robert's questions a question about computers and should this productivity growth really be show up somehow in the manufacturing sector? Not to publicize my own work but I've actually argued in an article with Tim Bardic and Tim Sturgeon in MIT my colleague Tim Bardic at the Uptown Institute that probably the great productivity growth in computers is to a large degree being driven by research and development and there probably should be not booked to productivity, to manufacturing rather. The problem is that we're not, we don't know how to adjust yet services prices to book that productivity growth to the services industry. It's really a huge challenge. It's way beyond the scope of this but at bottom line I would agree we don't really know how to develop reasonable measures for R&D services. Interestingly as production in these sectors is moving overseas and you get perhaps some of the R&D staying here you may see it showing up less and less as productivity growth in the US economy simply because we don't sort of measure it very well that productivity growth, that improvements to products coming out of R&D. I do agree with very much with Pat Malloy's yes comments about kind of the change in the way corporations are run them. Movement from stakeholder to shareholder value that that has certainly had a big effect and the pressure on companies which is very real to increase margins. X-Mise profits is perhaps and almost certainly to some degree pushed companies to shift more quickly overseas to increase margins than one would have seen in the past. I think it's an important phenomenon. Some people are looking at it but it's arguably understudied. Alan, I always favor more and better data and I actually do think that kind of the globalization of things has just made it harder to measure stuff and we need to think more broadly about collecting better data to better understand the competitiveness of domestic manufacturing and other industries. We need to rethink the architecture of our statistical system in that way. Finally, Carolyn, I think Carolina from Mark Warner's office policies, wow, there's a whole host and I think that my colleagues will have more to say about that but certainly there's a host of macro policies. I think there's broad agreement that an appreciation or overvalued exchange rates is a huge problem and we should be cautious about that happening, allowing that to happen in the future and as Josh has argued, perhaps they're still overvalued but I'm gonna stay out of that debate. In terms of the nitty gritty, there's certainly a lot of very good vocational training programs and the like that can make a big difference for the skills of workers in manufacturing. There's a big demand for that and I think the sort of vocational education should be expanded. Great, great questions. Let me start with the PNTR one because I really fundamentally disagree with that view that PNTR had a big effect before China was granted PNTR. There was no time when it reverted once it got preferentials or the MFN status. We never reverted away from that so what's to make us think we would have and we did have two really important tools which the US still uses today which are anti-dumping and countervailing duties which were used very heavily against China. So I think that there were policies in place but let me turn to research on the topic. There's a great new paper by Mary Amiti, John Romales, Rob Finstra and one other person and what it shows is that it wasn't about anything other countries granted China. To the extent you wanna argue it was WTO session it was China's own liberalization. So the fact that all of a sudden in China you had access to cheaper intermediates and you could enter into global supply chains and all this had a much bigger effect on China's export growth than anything other countries granted. So we pushed China to liberalize but remember that some of that liberalization is gonna improve their own productivity so it was reforms China made that really had the biggest impact. In terms of policy recommendations before I went on leave from the institute, Chad Bown and I were working on two policy briefs exactly on this question and one compares, looks at the decline in US labor force participation compared to other countries and tries to understand the other one looks at lessons from other countries on labor market policies because while other countries had very similar experiences overall in terms of GDP growth, in terms of declining manufacturing, what they didn't have was some of the declining labor force participation that the US had. And why is that? And one, we spend a lot less on labor market policies than other countries in the OECD in particular, job placement policies, training but that's not gonna be enough if there's really structural change and demand fell. So then you need to turn to policies like wage insurance, potentially public works though I visited the Hoover Dam over the holidays and was amazed that it was finished two years early and employed all these people at a time of great difficulty. So I think given the state of infrastructure in this country, there is a way to boost demand through infrastructure and then on top of that have these policies for labor that could do a lot. And then finally just to say that if we really think it's about and I think most people do that the trade deficit when we're not at full employment is gonna be making it more difficult for workers. I think there's research by Joe Ganyan and others showing that fiscal policy is important. So if you're gonna run a fiscal deficit that means the government's consuming more than it has in revenues, than it's production and some of that is gonna go to imports. And so that's gonna pull the deficit and so forth. So Juan and I stopped there but those were great questions, thank you. Yeah, so I'll say a couple words on policy first because that is a very, very long discussion and so just the headlines. I mean, if you're really worried on sort of the narrower issue and it's a big issue but it's narrower relative to all the other problems in the American economy of manufacturing and employment to my mind dealing with misaligned exchange rates is the really key one. And dealing with them in a way that's credible in the long run. I mean, it's one thing to say, we should make the dollar go down for a couple of years. That won't do a lot. I mean, there needs to be real credibility that the US is now committed to never again allowing the dollar to just annihilate the US manufacturing sector. And that means doing something like we've got a Federal Reserve that monitors this one really important macroeconomic price that is the interest rate. And we say, add us your job to adjust it depending on economic conditions. It seems to me we should probably have something like that for the exchange rate. It's an extraordinarily important macroeconomic price and that's when we just allow to whipsaw around and cause a lot of damage and it's not clear to me why we do that. And so I think big picture ideas on how to credibly commit to never allowing the dollar to really do what it did to manufacturing again is a big part of the story there. I would also agree. I think infrastructure investment would be really key for a couple of reasons. One, I think we need it. I think we've under-invested in it for a long time. I think we still have some slack to be rung out of the economy that it would do. And it's particularly good for manufacturing. It buys a lot of inputs from the manufacturing sector so I think that would be a very good thing. I think turning to Pat's point for a second about these policies are sensible enough, why haven't they happened? I think his point is really interesting because it reminds us that most trade policy, if you think of it as a USV China or USV Mexico, you will get it wrong every time. I mean, trade policy is about people within the United States and people within China and within Mexico and how the interests collide. And to be really clear on this, everyone agrees that the Chinese government was managing the value of their currency for competitive gain in the early 2000s and many people may have been like, why is the US allowing that to happen? Well, which US? I mean, Walmart loves this. Walmart gets really cheap imports from this policy. And so the fact that manufacturing workers in some states are getting hammered and Walmart likes it, well, who do you think wins that policy debate most of the time? And so I think that the really key thing I take from Pat's comment is exactly that these are not nations versus nations conflicts. These are really class-based conflicts within nations. That's the way you need to look at this stuff. That's a great note on which to end. I'd like you to please join me in thanking our panelists and thank you all for coming out today. And I hope this will be just the beginning of an ongoing conversation about policy in the important area of trade, automation, manufacturing, and the future of work. Thank you very much. Thanks to C-SPAN. And I hope to see you again soon. Thank you, guys. Thank you. Thank you, Ryan. I just had reading a paper on your earlier work as well. And your earlier work on this topic as well.