 Wonderful. Well, I warmly welcome you to our second session. As you heard, today is about microeconomics of price and wage setting. And we all in the recent decades, witnesses, quite structural, significant changes, talking about e-commerce, talking about changes advances in communication technology, talking about the lengthening in international supply changes. So with all this, we are asking ourselves, what is the impact of these changes on price and wage setting? And alongside these developments, so many countries experienced stagnating real wages and weak productivity. As a central banker, obviously, we follow such price and wage developments and we become increasingly interested in how these developments can be understood at the micro level and how that might affect our task and functions in terms of monetary and supervisory policy. The papers and discussions in these sessions will explore these and related themes with the help of outstanding scholars. First, we have Aviv Nevo, professor in business institutions and marketing at Northwestern University and former economist at the U.S. Department of Justice. Aviv will talk about how current global and technological trends and advances in big data can shed light on price setting and inflation developments. His paper will then be discussed by Michael Weber, assistant professor of finance at the University of Chicago. Michael has worked in the fields of asset pricing and household finance. And then we will have a short round of questions. And after that, our next speaker is Uta Schoenberger, a professor of economics at University College London. Uta will consider how institutional settings affect different countries' performance in terms of productivity and wages. Uta's paper will be discussed by Michael Borda, professor of economics at the School of Business and Economics Humboldt Universität in Berlin. And Michael has worked in many fields, including labor and European integration. After that, we will have time for another round of questions I hope for a very lively debate. So we have a great set of presenters and discussants. And I would like Aviv to take the floor first. Thank you. Thank you very much for having me here. Although I do have to say I'm still trying to figure out what exactly I'm doing here. Hopefully I will by the end of the talk. This is based on joint work with Arlene Wong in terms of division of labor. Everything you like about the talk, please credit Arlene. Everything you don't like about it, please blame me. I do hope that by the end of the talk, there's going to be much more in Arlene's column than in mine. Okay, so what am I going to be talking about? I think we kind of touched about it. So there's been a lot of trends in, I call it advanced economies in the global economy. Obviously globalization, a change in competition and concentration, a change in the structure of many industries, increased high fixed costs, a lot of innovation or marketing or whatever you want to call it upfront and maybe lower marginal cost industries. And obviously a growth in e-commerce and more generally just the sharing economy, Uber, Airbnb and things of that sort. And then finally what I will claim and this is maybe less is that there's actually been a change in price setting behavior. The way that firms approach pricing, both in terms of the data that's available to them. Everyone likes to talk about big data but firms really do have big data. Enormous amounts that even 10 years ago they did not have and now not only they have it, they're actually trying to figure out how to use it or starting to figure out. All of these together means that there's going to be changes and what I'm going to be talking about really the micro changes from an I.O. slash marketing point of view. There's questions about what are the aggregate implications. I'll actually mostly stay away from it and leave that for the discussion kind of thinking, you know, working with my comparative advantage, which is really on the micro side. So overview, I'm going to organize a discussion along two dimensions. One would be an issue of measurement. I'll touch back on some good old themes of substitution bias but say, you know, try to make the claim, maybe not too convincingly, but I'll try to make the claim that some of these issues are actually getting magnified these days. And then I'll talk a little bit about some conceptual issues, including cost pass-through, decreased competition, you know, quote-on-quote new pricing models and heterogeneity. I'm probably not going to have enough time to get to all of them, but at least touch on some of them and set that up for the discussion. Now, to really be concrete and to touch about these, we need to have a more concrete question. So I'll focus somewhat, although not really in great detail, about the question of low inflation during the recent recovery. And I'll have that as a kind of theme in the background, although I don't want this to be a talk where I'm going to try to explain from a micro point of view why we had low inflation, because if I set it up that way, you'll be very disappointed. I do not have an answer. Okay, I might have some points to think about, but I don't really have an answer. So speaking of answer or bottom line, this is not a talk that I think you'll walk away from saying, okay, what was the elevator pitch, the 30-second, what did we learn from this talk? I don't think I have a clear bottom line. To the extent that I want to emphasize or have sort of two main takeaways, I have the two points and two bullets here that in some sense are not all that original. How many times have you heard a talk when someone ends up saying, we need more and better data or we need more research? I would say about 90% of talks. So this is going to be one of those 90%. But I am going to be a little bit different, hopefully, in the sense of saying we do really need more data, especially when it comes to online commerce, because I think there was yesterday a discussion that was actually a very nice slide of looking for a needle in a haystack. Well, the claim that I will make is that not only are we looking for needle in a haystack, we might actually be looking at the wrong haystack. We're focusing on the old economy. There's a whole new economy. We don't even have that haystack, let alone try to find a needle there. So that's going to be one, the first point. The other point is in order to look at these questions, we really do need not just more research, but more collaborative research. Okay, not collaborative of one macroeconomist talking to another macroeconomist, nothing wrong about that. It's great. You need to reach out. If you really want to understand what individuals are thinking about, whether they care about, you know, you re-talked about yesterday about inflation expectations, whether they care about that, whether they care about other things, you probably want to talk to folks who study households, consumer behavior. If you want to understand what firms are thinking about, you probably want to talk to people who, that's their business, talking to firms, working with firms, understanding how firms make decisions, and we really need a lot more of that. Now, that being said, I think there's been great progress made. I mean, there's the type of work that macroeconomists are doing these days. I mean, you wouldn't have people like, you know, Michael or Lee and my co-author doing this work, maybe 10 or 15 years ago. So there's been great strides along that, but I think there's still a ways to go and still bringing a lot more of the true micro foundations into macro research. Okay, so I said I'm going to talk a little bit about why inflation is low. So I'll take it as a fact that inflation is at least measured inflation. Let me be clear about that. I stayed persistently low post a great recession. Part of what we've been asked to give is to say, well, from an IO marketing point of view, why is this? Okay, I'm not going to have a clear answer, but I'll give you at least some ideas, some of the ways that an IO economist might start thinking about this. So why do we expect inflation to rise post recession? This was really more a slide for me, trying to understand, okay, what's the question before I try to answer it? So I went to our lead and asked, you know, why do you macro guys expect inflation to rise? So I've been given at least one reason is a tightening job market would lead to higher wages and those should be passed through to prices. That's not necessarily the only mechanism, but I think that's, I see a lot of heads nodding, so at least I didn't get that one wrong. That's, I think, kind of the simpler mechanism. So I guess to ask, would this lead to higher prices? You first have to ask, did wages go up? Right? I mean, because if that's the beginning of the mechanism, it seems like maybe not, but I'm actually not going to touch much about that. I'm going to leave that to the labor economist to say much about that. I think the only thing I can say as an IO competition point of view is to say, we're going to talk a little bit later about monopoly power, but you could also wonder whether monopsony power, as we see more concentration, does that actually lead to more monopsony power in the labor market and could that be part of the reason why we're not saying wages increase? So that's kind of the competition aspect. I'm not going to say much about it beyond having a reference to some work by some of my colleagues at Penn. There's additional reason, additional mechanism that could go on. I'm not going to be really focusing on that. So what I'm going to sort of be looking at is saying, let's assume for a second that wages did go up or went up to some extent, question is, will these be passed on to prices? What will we expect in terms of prices? So for an IO economist, it's not obvious that even if wages went up that we would necessarily see an increase in prices. I mean, we'll see some, but it doesn't necessarily have to be significant. And there are several reasons, and I'll talk about some of them. One is just a pure measurement. So it could be that actual prices are going up, but the measured inflation is not. So that's going to be the first part of the talk. Then I'll talk about the issues of imperfect pass-through and how at least micro-economists, IO economists think about it. And then I'll try to kind of weave in the longer term secular trends that might be bringing to some sort of structural change. Again, I'm just going to provide some exploration of this, and hopefully the discussion later in the panel after the break will touch into this. So measurement of inflation. Good old substitution bias. We all know it. We've talked about it. I'll talk a little bit more about it. Although what I'm going to be talking about is not exactly the traditional substitution bias. It might actually be something a little bit different. New products, online shopping, I would actually say that is a change that we're constantly seeing. There's more and more introduction of new product. They're easier to do. Used to be that you wanted a new product. You had to come up with a distribution chain and whatever. Now all you have to do is to get Amazon to sell it for you and you can sell that new product. Online shopping obviously is increasing as well. And the issue of measuring prices online is going to be a big issue. So let me touch on each of these very shortly. Substitution bias, we all know when prices change, how do consumers react, what do they do. Typically people talk about either products, outlets, new products as biases that give. I actually think that the same ideas should be looked at much more broadly. And part of the work that Arlene and I have done in the past is actually looked at the substitution bias across products, that is the traditional way, but also across sizes. So you might say, okay, we're the macroeconomist. Do we care what size of a laundry detergent people buy? Well, you do if you care about what the decisions that people are facing at the store. So when people are looking at prices at the store, they're not thinking about the inflation expectation. What they're thinking about is should I buy the larger size versus the smaller size? Should I use a coupon? Should I buy the product because it's on sale? Because if I buy it on sale, it's 20% less. Okay, this kind of temporary price reduction. They're not thinking about is the inflation going to be two or two and a half percent. Okay, now, what implications does have for the macroeconomy? I have no idea. But I'm telling you, that's what consumers think about when they make choices. So we're going to look at some of these effects and I'll actually show you that there's an interaction between these effects and macroeconomic conditions. Okay, so there's several papers preceding ours. There's a long list here that I don't have time to cover that has talked about how shopping behavior changes with economic conditions. Okay, and this is across different countries, the US, the UK. There's a paper here looking Argentina and others. Okay, people have looked at it. We've looked at it specifically in the context of the Great Recession. So we extend some of this work here. So what we looked at is how shopping patterns change with macroeconomic conditions. So what I have here, these are the fraction of purchases that are on sale. Okay, so a sale is a temporary price reduction. Then just think of it, the price goes down for a week. What fraction of the purchases are made there? So what you see here is you see the shaded area, that's the time of the recession. And you see that sales were actually on a downward trend slightly before that. That stopped during the recession. And actually there was mid recession, a big jump. That's something we documented in our previous work. And what we see is post recession, there's kind of a return to the pre-recession trend. We haven't actually got there in terms of levels. Okay, but there is a decrease. Now you might sort of say, look, in terms of numbers, isn't that big? We went from 25% of purchases to about, you know, 26.5, 27. But in terms of fraction, I mean, that's an increase of, you know, quite significant, almost a 10% increase in the faction of purchases that are on sale. Looking at coupons, same type of behavior. You see a increase, a jump during the recession. And then what seems to be a decrease. By the way, in all of these, I didn't say the very jagged line, that's just literally the data, the green line, that's a flitted linear spline, and then a cubic spline just to kind of try to get, tease out some of the trends. So that's two types of shopping behavior about products. This is looking at generic products. Again, here you see kind of maybe less pronounced of a trend during the recession or a little bit of an increase during the recession and a clear decrease afterwards. And this is generic products, that's talking a little bit more product substitutions, less of an effect than you did in the other two products. How about stores? So this is looking at discount stores. Here there's just a general increase that it's kind of hard to tease from the data. Basically, this is US data. So this is the increase of Walmart. And Walmart was growing before the recession. If you squint very hard, it seems like it slowed down a little bit during the recession. But you have to squint really hard to look at it. And then it seemed to have picked back up post recession. But it's actually all swamp by just the general increase that Walmart is growing. And there's more of it. So it seems to be that there's a change in the behavior. And this is not necessarily your traditional substitution bias as more of a change in consumer behavior due to the economic conditions. It could be because people have more time to shop or take under activities. So could this have an impact on inflation? We didn't actually compute it a formal analysis, but just the back of the envelope calculation suggests that taking this into account could actually mean that the measured inflation post recession is about 20% or the actual inflation is 20% higher than measured inflation. So measured inflation is too low. So it directionally does go in the effect of we think inflation, measured inflation is lower than what we'd expect. This directionally could go in that direction. I'm not actually saying this is a number that I would stand by. This is just really a back of the envelope calculation. It's actually pretty hard to compute an exact index. I think Jim alluded to yesterday and how sometimes computing an exact index that accounts for all these effects is quite hard. It's the case here as well. Okay, so this is just kind of to give you an idea that there might be something here. Okay. So overall, did chopping behavior revert after the recession? Yes. Could that explain directionally low inflation? Maybe. Is this large enough to explain what we're seeing? Again, I'll leave it up to you, but probably not. Okay, so this can't be all of the story. And then a more general question here which I'll just pose and not really try to address is to say, okay, so you saw these kind of behaviors and you can ask whether they're big or not during the recession or after the recession. But is there kind of a change, a secular change over time? I would say maybe there is. Okay, because you could say, okay, whatever happened in this recession, why didn't we see it in previous recessions? Okay, I mean, what's changing? I would say maybe there is a change in chopping behavior. So I don't have time to cover that fully, but let me give you a little bit of an idea. So here's a paper by a couple of young folks at Chicago. They looked at, you know, product variety. So they're actually showing an increase in concentration in the products that households are consuming. So what they measure here is over time, herfinal index of the concentration, a household level, concentration. And they're actually saying consumers are concentrating more of their purchases in fewer brands or fewer products. So that's a change over time in shopping behavior. Might not actually go in the way that we'd expect, but it's just showing there is a change and that's really all that I'm trying to say here. The other big change, of course, is online shopping. We didn't really touch on it there, but that's kind of if you want the elephant in the room. That, you know, big change. Here is actually showing overall commerce in the U.S. over the last 15 years, basically e-commerce as a fraction of total retail sale. So it's only a 10% now, but you look at how fast it's growing. If you look at what's happening to Amazon stock, you realize this trend is here and it's here to stay. Okay? So what does this actually have to do with inflation? Well, Jerry Hausmann wrote a very interesting paper a little bit more than 10 years ago that was talking about CPI bias from supercenters and I highly particularly part of it, he asked, does the BLS know that Walmart exists? Okay? So this was in 2004. You could ask that, write that same title and instead of Walmart, substitute Amazon, eBay, Uber, Airbnb, and the list goes on and on. Okay? So there is a question as to whether the BLS or generally statistical agencies even measure online prices. Okay? That's a little bit unclear what they do. You say, well, you could give the excuse, yeah, they don't, but it's just a sampling issue as long as it's a random sample and they're not behaving any differently than, you know, why do we care? But there's a lot of good theory to sort of suggest that it would act differently. You have a lot more tools to price discriminate and to segment consumers in very different ways online than you would in retail and offline and that might suggest that you will see separate prices and indeed a recent paper that just came out a few weeks ago by Goolsbee and CleanOut suggests that when you look at online prices you do see different inflation. They actually find lower inflation which again maybe isn't consistent with this story of what's happened post possession. They're just saying, look, there's something else going on there. There's a different haystack. We've got to start looking at it. Okay? That's kind of as simple as that if you want. Okay. So that's kind of in terms of measurement. I kind of went through it quickly but I'm already kind of way behind. So let me just kind of quickly go through some of the more conceptual issues. Costs pass through. So it seems intuitive if wages go up or any cost goes up so should prices. But looking from a microeconomic perspective it doesn't have to be. So if we just write the simplest pricing model that we can saying, okay, there's a firm try to maximize profit where profits are priced minus marginal costs, variable profits times quantity minus a fixed cost will give us the following first order condition. Basically says price is equal to marginal cost plus a markup term. Now it's very easy to so look at this and say, hey, marginal cost is on the right hand side so if it goes up price should go up one by one. Well, that's not right because that term that semi-elasticity that's the markup term depends on price. So if you want to compute what the optimal price is that will vary. And generally it depends really on the shape of the demand curve and here this is a residual demand curve so competition and the effect of it will be in there as well. So depending on the shape of the demand curve the pass-through should not generally would not be complete. It could actually be less than one sometimes even more than one. Typical demand curves that we work with will give us pass-through less than one. A linear demand curve for example gives us a pass-through of 0.5. Measured pass-through when people have looked at it are significantly less than one. I mean when you look at micro studies there's only one industry where I've seen consistently pass-through rates that are greater than one. And I'll have a poll later if you guys can guess which it is. All right, I'll tell you it's tobacco, cigarettes. So whenever there's taxes on it they usually pass-through than more than 100%. There's actually good economic reasons why that might be the case having to do with who are the people that stay with you once prices increase. So I think to sort of realize even if wages go up it's not clear that price should go up. Taking empirical measures should be that this pass-through could actually be quite low depending on the circumstances could be pretty close to zero. And I kind of mentioned that one of my colleagues at Penn, one of my macro colleagues said I don't really, I mean it's an interesting theory but I don't buy it. There's no way firms are eating the cost. I said this is not eating the cost. This is a profit maximizing response. As your cost increase what is your optimal price response? It's this straight off between margin and quantity and sometimes you're better off saying look I want to sell more volume, less margin. This is not eating the cost. This is the optimal response. So the question is here going to low inflation could be if indeed our labor colleagues are going to tell us wages haven't really gone up for whatever reasons the labor market they've gone up but maybe not as much. Maybe the shift of price is also low because of these lower cost pastor. So now the question is what's happened? Why is this sort of change? Okay this kind of pass-pass through mechanism could have been there previous recessions what has changed? Okay so this is where I get kind of briefly to the long term trends. So one long term trend that's kind of in very popular in especially in IO circles is to talk about the fact that there's a rise in market power, rise in margins, rise in concentration. Yan, one of the co-authors that's actually of a paper that's quite highly cited and is sitting right there and there's a question is could be impacting pass-through and it could in two ways. One is the fact that most of this document what it really is documenting is a decrease in the share of labor. Okay so share of labor is decreasing and really what you know this paper in particular what it does is it translate that into a markup using kind of various assumptions about the production function side. Now as you have even holding pass-through constant as you have a lower labor being a lower share you just can think of I don't know price of labor goes up by 10% it's just proportionally it's going to impact prices much less as it's a smaller share. Okay so this is just almost kind of just pure arithmetic. The other thing is if it really is a decrease in competition competition has an impact on that pass-through. Okay because that pass-through is a residual demand curve. Now we know that in the case of perfect competition okay the textbook case of perfect competition pass-through is one because prices are equal to marginal cost marginal cost increase will be a pass-through of one. Generally though as you have less and less competition you know there isn't a monotonic relation it will go either way all that we know is we can no longer guarantee that it's one but it's one of those things no one will tell you there is a theorem that tells a particular way but there is a belief among many IO economists that as competition decreases you're going to get lower pass-through rates. Okay some kind of believe in it more than others some actually even use as a test of imperfect competition but there is that belief. So going along with that belief if you actually believe that the economy is becoming less and less competitive the pass-through rates might actually become lower. Okay more complex supply chain so there actually has been there's again a recent paper that's talked about the fact that there's globalization leads to a much more complex supply chain right so instead of actually getting your supplies from someone down the road it could be much a much longer supply chain. What does this have? Two implications one is there's no longer kind of a link or a weaker link between local market conditions and the output market that's one and more importantly for the point that I'm trying to make is the fact that there's more levels mean that there's more of these pass-throughs right so if you just think of one level of there's just a seller they have an increase in their cost they have to pass it through but if you have these multiple levels then there's this imperfect password multiple levels right so you take 0.5 in each of them and now it's 0.5 you know the power of 10 because there might be 10 different levels you're going to get a very small number okay so that could be kind of one implications this paper that I mentioned one of the things that they point out I think it's an interesting fact is a what seems to be a divergence between the PPI and the CPI and if we have more time we can actually talk about it but let me kind of move ahead and then finally the last point I'm going to talk about is really in the change in how firms set prices and part this has two components they have more data they can do things more efficiently I actually think especially with the growth of business schools over the last 20 or 30 years you see more and more companies are actually pricing according to that simple pricing model that I had as opposed to a simple cost plus a cost plus model and that has kind of a long term trend that could actually have implications to how we're going to get passed through in the economy the other thing is up to now I've talked about just the linear pricing but we can think more and more price discrimination and you can see a discussion this is where the big data comes in okay so this is one quote Uber talks about how they can actually set route specific prices and they do if you actually want to have an experiment sit in a restaurant with your friends pull out your phones try to order an Uber and you'll get five different quotes going to that same place again there's a lot of discussion that historically we couldn't have perfect individualized prices perfect one first degree price discrimination now we can do that so what does that what does that mean well two things first is I think there is a change in how firms are pricing they have better data they have different models quant now is the that's the buzzword in all business schools okay now do they know how to apply our models no not quite yet but they're getting there okay and they're changing the price and that's part of what we're seeing these prices these changes and for example let's say in the markups that we're seeing part of it is not certainly a change in concentration it's a change in firms realizing we've been leaving money on the table and if I had more time I could actually give you many examples of this what this does is it means that there's gonna be fancier pricing models okay and fancier is that you know stuff that we as economists have been talking about for 30, 40 and even more years markups okay but they're now actually getting implemented in concentration and it brings the whole issue of realizing a whole new level we've been leaving money on the table and if I had more time I could actually give you many examples bundled and all kinds of stuff like that it gets it's a whole different levels that even theoretically I'm not sure we know how to do and we definitely don't have the the data and then on top of that there's gonna be a lot of heterogeneity the fact that different people might actually be seeing different inflation levels and we have some measurement of that and again that brings it to a whole new level so with that in mind let me conclude what I talked about are a couple of measurement issues conceptual issues I hope I kind of made you know two points that came across A we need more data both in terms of online individualized data to really get at the heart of these questions and I think to really get at them we need to work together we being all marketing guys together with macro economists and central bankers so let me stop here I'm actually 40 seconds behind but ahead of my expectations so perfect Chinese I'd be right okay perfect so thanks a lot for the organizers to allow me to think about those topics so what I want to do is actually also try to provide maybe five themes in 15 minutes and see whether I'll be successful in sticking to that so first I want to look at a little bit of historical perspective and possibly argue that we might be able to actually learn from past periods of low inflation whether we have maybe some insights what might have cost them then secondly I want to follow up on two themes of we for us alluding to and lastly actually on on two of the topics here we was actually talking about yesterday so how possibly one could manage expectations I don't want to talk about manipulation it sounds so bad and maybe actually try to figure a little bit out how actually people form inflation expectations so here we all know this figure this is just over time annualized CPI inflation in the US and so Larry Summers talked two days ago about the great inflation so if you would start there and then would compare the last 10 years to the historical mean you would indeed conclude possibly by looking at the last 10 years definitely below the red line so clearly we have historically low inflation but who says actually that we should actually start in 1970 we could go back 10 years the mean already the red line goes down a little bit but who says we should maybe start in 60s maybe we might be concerned that there's measurement our data quality might be worse and so on so we could actually maybe start in 1990 and you already see that actually the red line goes down and down so I guess bottom line here would be to say that depending on what we would argue is actually the normal we would actually possibly draw different conclusions but I think what I want to take away from from those figures is actually go back maybe to here and say well possibly in backwards over 30 over time series 10 years request one on the other and say well now we have actually low inflation but possibly we might be able to learn something from taking a more historical perspective and maybe just look at possibly getting data in the past when we also had low inflation to possibly even distinguish between competing mechanisms is it lower pass through more concentrations and things like that and just for completeness let me also directly look at the eurozone and here possibly could also say we have historically low inflation so bottom line one possibly let's learn from the past then let's follow up on two of the themes Aviva is actually talking about it's on the one end you could now say well what are the secular changes we've seen in the last three four decades I want to highlight one specific mechanism automation increased compute use of computers and how this might affect it possibly some patterns we see in the data on the other hand follow up a little bit on increases in market content power look at one specific data set of easy access to and then as I said a little bit thinking about how people form expectations and whether they actually are indeed as we would model then typically in our preferred New Keynesian model okay so if you're now actually look at overall inflation what is actually tells you is maybe not that much because if you drill down at the industry there's lots of heterogeneity we see there are lots of pattern differential movements and actually looking at overall inflation camouflage is maybe the most interesting piece of evidence what we can learn but at the same time actually if you think about starting in the 1980s people started to use more computers firms started to use more computers there's been this increase automation use of robotics we've also increased on in let's say import competition and then actually there's this recent interesting paper by Otto and Don and what they actually argue is if you look at this figure what you see on the x-axis this is just a proxy for skill to the far left low skill for a right high skill and here on the y-axis you see the change in the average age of people by skill over time from 1980 till today and what you actually see is that low skilled workers and high skilled workers on average so this way lower increase in the average age in this occupation relative to a middle class jobs and so they argue this job is getting old so middle class jobs on average so a larger increase in the median age of people working there and so the question now is could it possibly be the case that larger workforce, lower bargaining power in terms of wages, lower wage growth could that possibly lead to an observed lower inflation exactly in those industries where you saw increase in computerization, robotics and so on and so I once actually went to the data and let you try to see whether there is a little bit of evidence for that mechanism so you now can use data from the sensors every five years you get average hours worked by age for each industry and you can create what I call old to young ratio old to all this is just the total hours worked by people above 55 and not inference on people in the room of course relative to all the hours worked in this occupation and then you can actually relate that to the average annual inflation rate at the industry level and just actually see whether maybe there's an association about the beginning of the five period you saw a higher share of old workers relative to young workers do actually those industries experience lower inflation subsequently and then also drilling a little bit down on the mechanism is it due to lower wage growth exactly in those industries and of course you might be concerned about international factors so you can be proxying for shipping costs which actually possibly tell us something about import competition of course if you have actually a lower unemployment you would also expect maybe lower wage growth but crucially actually also you would expect that is mechanism if it's at play is actually more important in industries with higher labor intensity so you can also proxy for that and so what you see here in the first row in red is actually in the raw data column one controlling for common shocks in the time dimension for industry specific shocks meaning having different degrees of fixed effects but also actually sucking up different parts of the variation in the last columns you see actually that indeed in industries where you have a higher share of old workers at the beginning of the five year period you see lower average inflation and also actually I think it can explain a meaningful part of the variation and crucially I think in terms of the mechanism in blue you see that this effect is actually larger the more the higher labor intensity you see in that industry maybe actually looking at those numbers it's not too elusive so let's actually look a little bit also at some figures so what you see here on the x-axis you see actually binned ratios of all too young far left, low all too young ratios far right high all too young ratio on the y-axis you have industry inflation and then by five year period you see that actually that's across the board this negative association maybe a little bit less than in the inter-meeting years in the 90s but actually you see this very tight clustering around the regression line especially in the last 10, 15 years or possibly indeed an aging industry workforce and lower wage bargaining power might play a role and in terms of actually exactly the mechanism you indeed see that in those very industries where you see higher ratio of all too young workers you also see subsequently lower wage growth and again you can explain a meaningful part of the variation okay so now let's follow up a little bit okay bottom line two bars so the first one we can learn from history the second one is okay there's potentially some change in the age composition of the workforce which might help explain some patterns we see in the data but I want to follow up on the part are we first raising maybe this increase in competition this might, how this might affect price setting of course like young accurate and co-authors are using is great you have actually all publicly listed firms the downside is that you don't really observe the pricing patterns unless you go to the placement of the BLS what you and I have done in the past and I can promise you that's not too much fun so instead what I'm trying to do here is actually use data from retail scanners also a little bit leveraging the idea of using big data so here you see trillions of observations for individual goods over time and you can actually say do we see at the local level increases in competition and so the way I did that I said well let's now define the market based on goods which are on the one that's substitutable but also in areas which are very similar in terms of local economic shock so for example one area you can think of is like the East Bay, San Francisco, Oracle and San Jose and in terms of actually type of granularity so you don't want to look at Duplo and Danuta individually because depending on price you might substitute between Duplo and Danuta so overall we want to actually define maybe chocolate candy bars as one product category and then actually the interaction of the two is defining my market and then what I can do is just actually look at trillions of price observations and argue and see whether there is any pattern in concentration over time so I calculate the standard Hervindal, Hirschmann index but also Leifard version and then actually this was a little bit surprising if you actually look just at concentration over time so a very specific market so only groceries and actually what you see here only the last 10 years this is due to data availability you see actually if anything in groceries this is actually holds across definitions you actually see a decrease in concentration over time in that data but again so we have to keep in mind this is only a very specific cut of the data and then you can ask why is there actually an association between locked prices and concentration you can also run it in differences and here what you see there's this negative association so if you see higher concentration this tends to be associated with lower prices at the local level and at the local category level and so I think this I wanna actually reiterate what Aviva's saying we ultimately definitely have to understand way better the IO, the industrial organizations how do actually now let's say chains like Aldi or Walmart price at the local level and I think this is definitely also something I would encourage more research on and now the last two bullet points or the last two points I wanna come back is a little bit what you already got Echenko and Oli Korb, you were alluding to yesterday so to the end of the day to the extent you believe in I think the standard model at central banks and New Canes and model policies are effective because inter-temporal substitution is at play and so whatever shifts or moves the real interest rate should lead to changes in consumption patterns and so one way to possibly move real interest rates would be due to changes in inflation expectations and so I guess we should better understand whether indeed on the one hand how people form inflation expectations whether people act on inflation expectations but crucially also then possibly why in the data some policies are less effective than others and I would just want to allude to this forward guidance puzzle so the first thing I wanna look at is actually how people even form inflation expectations and this is as I promised yesterday and this goes back to a speech you gave in 2011 when you argued that we as central bankers typically focus on core inflation but instead actually maybe we make systematic policy mistakes because if you look at the data it looks like the most salient experience of prices is actually from shopping trips you pass by the pump of course those are very high frequency movements but now actually if it's the case that individuals form expectations looking at those very transient price movements and if they act on those expectations maybe keeping them out from our measure of inflation might lead to systematic mistakes and so of course the question now is can we actually learn anything from the data whether people indeed form expectations by looking at prices and supermarkets and again this goes back to our response we can actually use big data so again use the AC Nielsen data trillions of observations at the individual level and then use actually a tailored surveys where you ask people in this AC Nielsen data set 90,000 households you just ask them what do you think is inflation what do you think actual inflation be over the next 12 months and Kushi also lots of things how they actually form inflation expectations and when I started actually working on that topic of one of my colleagues Anil Keshia who is now external committee member in the UK he told me actually Michael you shouldn't really work on those inflation expectations of individuals because they are all over the place and particularly until you can explain to me the most salient pattern in the data you shouldn't do that and what did he mean by that he said well in the data across countries sample periods and cuts it's always the case that women have higher inflation expectations than men which doesn't make any sense to me so we run our own server and we see the very same pattern column one you see that male in the data on average have inflation expectations that is 1.3 percentage points lower than women and again this doesn't really make too much sense why should there be a gender effect in inflation expectations but then actually you can actually ask tons of other questions and one of the questions we ask who is the main cross-over shopper in your household and what you see in column two within household if my wife does the cross-over she actually has high inflation expectations instead if it's me you see I actually have high inflation expectations so the gender effect in column two completely is subsumed by this cross-over dummy which equals one if within household he or she is making the cross-overies now of course you can also do the same exercise just look at men and women separately within men, within women is there a cross-over effect and if anything actually if you look at column four relative to column three it looks actually that the cross-over effect is even stronger among men than among women you in the aggregate you don't see it because on average more women make the cross-overies relative to men but if anything actually also men seem to extrapolate from observing prices in the supermarket on their decisions and on the expectations in the paper we actually also provide evidence that they would actually behave in line what you would expect from lawyer equations potentially we might want to keep in mind things like that and so the fifth point I want to briefly allude to a little bit is actually maybe providing some empirical evidence to what Mike Woodford recently said I think it's fair to say that he has been the intellectual mind behind what we've seen the central banks doing in the last decade so he 2003 already argued that well why should we care about the CLB we don't really have to worry about it if we are able to manage inflation expectations and actually raise inflation expectations from the CLB banks we don't have to worry at all about the slowest bound constraint more recently however it looks like he updated his own beliefs a little bit because in his recent NBR macro annual paper the second sentence of his abstract reads we assume unrealistic cognitive abilities on part of our decision maker in the model so if you promise to keep interest rates at zero until the end of the liquidity trap this generates inflation and we fully understand that today we should update our inflation expectations upwards and that's why we should consume way more you could possibly argue that limited cognitive abilities might hinder the effective net of that policy and we know empirically there is what's called the forward guidance parcel in the data it seems less effective than we might have thought based on our representative H. and New Caneson model so of course if you now want to see whether limited cognitive abilities play a role you somehow have to observe measures of limited cognitive abilities ideally for a large cross section of people and so we have this recent work which only thanks to data by the Finnish central bank was possible so we got for all men in Finland due to mandatory military entrance test IQ data from AD1 until 2002 and then you can link it actually based on social security numbers based through the household balance sheet inflation expectations and what I just wanted to show you here is just look at the x-axis one means low IQ lowest 4% of the population nine means high IQ highest 4% of the population and it's approximately a normal distribution if you look what people say in service what do you think is inflation over the next year you subtract what is exposed to realize you take the absolute value and you average by bin you see actually that low IQ men at least in Finland have forecast errors for inflation which is two and a half times as large relative to the forecast error for high IQ men now you might wonder does it matter so you can run Euler equations do people typically in our Euler equation if you expect higher inflation we would also actually start consuming more do that run it for within men of high IQ within men of low IQ you see that only men which have high cognitive abilities indeed would react in line with their Euler equation I can actually show you in the paper this is not driven by lower IQ men being of their Euler equations so meaning by financial constraints or things like that and I think lastly often times you know as central bankers we try to actually maybe lower interest rates conventional policy to stimulate investment and demand through maybe a credit channel but what I'm plotting here on the right Y axis you see over time just a short term normal interest rate now let's actually see whether there's any interesting heterogeneity in the propensity to take out loans by low and high IQ and the way I actually mean low and high IQ which has split your sample in the middle low IQ top 50% high IQ a bottom 50% and top 50% so if you would focus on high IQ men at least again this is specifically for Finland that I want extrapolate when interest rates go down their propensity to take out loans go up interest rates are flat propensity is flat interest rates go up propensity goes down so this is not just graphical evidence you can also do in a regression framework that's highly statistically significant now instead actually look at low IQ men there's no change in the propensity to take out loans over time as a function of interest rates now you might argue okay maybe if you don't don't want to stimulate consumption of low IQ men but to the extent that we use common policy and some people react and others don't react we might be concerned at least through about implicit redistribution mechanism so what I wanted to do in this talk is actually on the one hand argue that possibly we can use historical data to learn more about why we might see low inflation we actually have some evidence that possibly changes in the age of the labor force might be relevant for trends in inflation in terms of cross-re-data at least we don't see increases in concentration and lastly I think we should actually better understand how people perform inflation expectation my understanding is that the ECB is also trying to actually purchase the shopping data and so I think this would be very useful and also I think we might think about what we label human frictions that maybe parts of the population don't react to policies and so bottom line is we need to think about policy salience, policy communication how they actually together shape the effectiveness of policy thank you well many thanks Michel many thanks Aviv that was quite interesting I have all kinds of questions and I would love to have the data for women and IQ yeah and how they fare with loans yeah I can tell it's much higher it's much higher okay good my wife has told me repeatedly it's much higher for women okay and my second comment would be to being in charge of statistics in the ECB I see a lot of work coming to us yeah with regard to the big data yeah and I I never experienced something else than economists being very greedy with regard to data yeah so and the floor is open for questions and I would ask you to to tell me whom you would like to ask the question yeah so that we know who is addressed please I'm Livio Vojna from Romania I'd like to ask Professor Nevo because he mentioned about about post-crisis wages rising wages post-crisis should lead to higher inflation but empirically we don't see it and I would like to challenge you to think in terms of stock instead of flow wage gaps instead of simple wage dynamics so if you consider that recession is a game changer and there is uncertainty of future income then the reference for a worker is not ahead of him because he doesn't know about his future employment or wage the reference is behind him so the past peak income maximum peak income in the past therefore until he recovers what he lost after a recession he's not willing to spend more basically this would be a accumulated wage gap I wouldn't make reference to this if you didn't mention that's fair but indeed post-crisis Phillips curve I think is completely different than let's say in a normal situation and I actually recently published a paper at CEPS which explains this link between wage gap and inflation so thinking in terms of stock instead of flow in relation to wages versus inflation if you could comment on this many things let us collect three or four questions please sure Ricardo Ries from the LSE this is a question to Aviv you focused almost entirely on cost of living measures of inflation what you call true inflation which are of course dominated by changes in relative prices and thus a substitution response of consumers to those as well as the relative price difference between input prices and output prices I passed through a very important concern of central banking or macro more generally is not with the relative price part but with a pure inflation part that is the homogeneous degree one side of prices that is to what extent when the unit of account changes all the prices move together or not and that's really central to what anchoring of inflation is or not with regard to that I would call it equally true inflation but that part of the inflation process to what extent do you think that these patterns you've identified bring as new insights that is to what extent when all prices go up or even relatedly when monetary policy changes in that sense the unit of and the anchor in unit of account change do we observe that because of more information online prices we'd observe a much faster pass through if you want the nominal part pass through should be won unambiguously no matter what your theory of demand is insofar as the unit of account has changed in demand all demand and profit function homogeneous degree one do we think that with the more information of online we see that nominal pass through of one to everything or actually less because the confusion between nominal and real becomes more severe in this new world or not in the same role just moving three yeah I know you call the university college London just to follow up on Aviv's point about the relationship or the puzzle between the rising prices because of market power and low inflation do you have an idea of what the role of technology could be because if you know if the only thing we consume is a microchips for computers we would see decreasing prices okay so they would be negative inflation now the world is much more complex and we consume many other things but one of the things we see is that there's a big change in the distribution of markets not just the average okay why we see the increase in markups is because really the top 90th percentile is going up and that seems to suggest that there's something underlying the distribution of tfp productivity of firms that's changing and I wanted to know if you have any idea of how that might be explaining the relationship between the slow inflation and at the same time rising prices at the firm level and what the last thing we see that the GDP deflator and CPI are not moving jointly so the GDP deflator is about 20 points lower over the last four decades many thanks and then Janis and then we'll make a stop otherwise we're gonna have to start again exactly and Janis short please very short thank you Janis from the bank of Greece those of us who have grown up loving the microfinders of macroeconomics I would like to use your two papers to explain why the philips curve has become flatter so can we say from your pricing equation that we are in an era now that marginal costs are fixed and constant and the production structure is that is such that from your paper Michael the concentration in retail has fallen a lot so it approximates perfect competition so the combination of very low concentration in in retail and the constant marginal cost produce a flat philips curve okay so Aviva I think you should start okay so let me start actually let me start with the last one which the truth is I have no idea I mean I think that's the kind of discussions that we should have of folks that are used to working with them with philips curves and the implications of it and you know because for me I have to go back to my you know days in graduate school to to think about and then try to couple them with realistic assumptions so for example uh... you did make an assumption in there and you know maybe competition and retailing is increasing that could be one sector maybe things are changing although I say maybe that's the key uh... I think what we want is to really start with good macro economics folks who really understand the macro implications but making sure that we have good foundations in terms of the assumptions that go in so let me give you an example actually something that Michael was talking about when he was talking about different men and women before you even gave the answer and I wish you would have actually stopped and turned to me because I would have told you women are the main choppers okay I mean that's the fact that those of us that have been working with the Nielsen by the way it's not AC Nielsen it's Nielsen they've changed the names about ten years ago that have worked with that know that so I think that's the fact that I had no idea that's relevant to anything that macroeconomists talk about and you can ask me the question I would have told you right away that's where to look now you found it on your own which is great I think that's a type of interaction that we can have so that's pretty much all that I I have to say uh... working back let me now can I go to the no no let us perhaps finish first the last question because Michael was asked to and then we go back to the other three if you don't mind Michael anything to comment on Janice you know Janice I totally agree with you that I think we have to understand so because of that it's definitely the case that retail from as I also argue in terms of inflation expectations it's crucial for the behavior of individuals in terms of the expectations how they actually behave and so ultimately I definitely think given that it's a substantial fraction of their consumption expenditure that it might help explain some of the patterns we actually see in terms of movements of of the philip who have a flattening of the philips curve in the in the last decade but to the extent that it's the Nielsen data only Nielsen not AC Nielsen only makes roughly ten fifteen percent of the overall expenditure for individuals it might be a part but I don't think it can actually fully explain all of the flattening we have we've observed okay so let me take a intern first of the question about the you know I call the wage dynamics so I think you bring up an excellent point actually we're talking over it uh... dinner uh... last night I do think there's a very interesting question and dynamics almost if you think of kind of through a search a wage search uh... model and just interaction if you want to know search slash bargaining between uh... firms and workers uh... as to whether what could happen in a recession as firms use that opportunity to cut costs cost that they maybe should have got even before the recession but in something that gives them an excuse in that bargaining for whatever way i mean you can model in a bunch of different ways cut the cost and then it really changes dynamic post post recession right so there's kind of the whole question of now when you're bargaining about the uh... the wages you can a set in some sense of different expectations because you really lower the level and i that's kind of maybe i'm interpreting your question but that's one way to look at things and i agree with you there's a lot to look there and look at the relationship and i touched just briefly on that one point about the increased monopsony power you could imagine that we want to really look seriously the way setting behavior where it's really kind of a bargaining process with maybe changing bargaining weights before and after the recession i mean that's at least the way that i would interpret your question so i'm not sure i answered it but i think it's a it's an interesting point to look at uh... rickardo to your point i mean i agree that you know when we measure inflation there's different things right from a micro point of view we always try to kind of look at more as a cost of living type of measure and then really care about the welfare implications and we care about the uh... you know is this a welfare measure is it not and maybe that's not what's really kind of being interested in looking at uh... maybe maybe not i mean you'd have to uh... to tell me but i think some of that same dynamics i mean there was something that you said in the middle that didn't quite understand you said that i think that you're working assumption that at some level the pass really is one and maybe quite understand what okay okay yeah so i agree with you that takes away that first part of the talk and we're talking about substitution bias and uh... and changes and maybe you say well maybe we don't want to sort of see you want to hold a fixed basket like i agree with you completely on that uh... but some of the other issues in terms of secular trends in the past with the actual raw prices of the individual commodities uh... i think are still in play right so i mean it could be that you know forget you know the basket we just literally had a single commodity that we were following that was what we were targeting uh... you could imagine that at that point there's going to be just less of pass through from underlying cost either because of the longer supply chain either because of a change uh... in competition maybe other things as well uh... that would impact it as well so i agree with you and uh... to the extent that they know that's why i said i don't actually have an answer of this is what's causing the low inflation i think i was trying to be very clear about that there is no at least no bottom line that i'm willing to stand behind just here's a bunch of things to think about and then finally uh... yeah yon's um... the ons point i mean i agree with you i mean there was actually a bullet point didn't get to at the end bottom of the slides i think there's some really interesting things that we need to kind of put together in terms of suggested a very sort of basic level uh... this goes back to actually when philip contacted me to to give this talk uh... i thought that what he was going to ask me is why a price is so high not what inflation is so low because that's a discussion that you know we're having an i o circles that markups are high right and markups i don't necessarily translate the high prices because you could say well maybe it's about pushing the cost down again going back to the bargaining models and um... and driving that down but then he kind of pointed you know weird at the central banker level worry about low inflation yeah i guess i've been hearing that the news but somehow is a different part of my brain i never put it together with the you know the research part uh... and i think we need to we need to kind of start thinking of why is it that one one hand i o economist are talking about i'm going to bunch it with us i o economy i o economist are talking about higher markups which would lead to you know think naturally higher prices but there's also low inflation now you can reconcile say one is a long-term trend the others a shorter term trend i mean the last few years you know you can still reconcile that but you want to start thinking about these together together with productivity uh... and together with welfare right i mean at the end of sort of saying maybe higher markups are not that bad because it's really about reducing costs and increasing efficiency and yes amazon and other firms are getting some of those returns because they're coming more efficient but actually might be good uh... might be welfare enhancing overall so i think those exactly the kind of discussions that we should be having okay well please two three short questions and then we have to stop going to the second part so uh... uh... america grossing with cornell and most recently commissioner bs so i want to reassure everybody here that bs knows that there is online shopping in twenty sixteen more than eight percent of prices were collected from online retailers and by twenty eighteen i'm sure that percentage is even higher so everybody here can rest assured uh... but uh... then the other comment is just to add to the list of quite full collaborations would be researchers and central bankers with national statisticians because they are i think they're huge gains to be made so i would just push that too i think christine was already asking so a good question for abib you talked about how the pass-through from car shocks largely due to wages is less than one which we well know have you looked at other differences from the pass-through if the car shop comes from commodity prices or oil prices versus wages because you could see there is a trend over the last decade more of the car shocks have come through very volatile oil prices commodity prices largely linked to emerging markets if the pass-through is different that might have affected inflation dynamics and i could just quickly think arguments are critical both ways on one hand you might have firms pass-through less of a car shock due to oil prices or commodities because you think it's temporary so don't adjust prices if you've uh... sort of sticky price model but on the other hand the supply chain argument could work the other way because if you've got a car shock that's oil prices or commodity prices it affects every stage of the supply chain so then you might get more pass-through from that type of car shock so i was wondering if you looked at that or had any evidence thanks christine benoit i wanted to come back to Ricardo's question because i agree it's uh... it's very important for us central bankers and i would like to try it in a different way uh... which is uh... to relate it to uh... to lucas seventy two somehow that is we know that information asymmetry is an explanation of why prices are sticky so the question is uh... in what you see in market structures and in the evolution of market structures uh... do you see more does it make it more or less easy for consumers to extract information on the general level of prices which was Ricardo's question then would imply that uh... uh... you would see more or less nominal rigidity i'm very sorry i have to stop here we will have a second chance for questions later on so i'll make my answers very uh... uh... very brief yes i agree no i don't know the answers to commodities i don't know i think it's the short answer uh... i haven't had it it's actually getting a little bit harder again because prices are individualized and more complex it's just a gut feeling i can't actually point to any pacific research that is very honest and i know that in the second row there is a gentleman and who wanted to ask a question for quite a while i'm steve chickey from the brandice international business school uh... the uh... your focus is seems to me to be primarily on goods prices and i'm wondering whether which which account for about twenty five percent roughly of consumption and so i'm wondering whether or not you have anything to say about the parts of uh... service prices especially if i look at if i link this back to jim stock's paper the the primary uh... items that we are bad at measuring are actually not goods prices uh... but their service prices so can you use any of what you're doing to help us with service price measure theoretically yes until we get the data until we get that haystack or all you know we're all drunks looking under the lamppost okay so let us uh... we will have a second chance uh... so let us move to the second part about wage setting with uh... your turn we were too quick so thank you very much for giving me the opportunity to present my work here on productivity growth wage growth and unions this is joint work with my colleagues alice küchler and ragnel schreiner at university college london and cream in the first part of my presentation i will provide you with an overview about what actually happened to label productivity about wage growth but unemployment and wage inequality in a number of advanced countries over the past two decades more precisely between two thousand and nineteen ninety five and two thousand sixteen my focus will be on uh... germany and france the two largest economies and of the eurozone and the data here come from the only cd economic indicators so let me begin with germany so this black line here plots label productivity measured as gdp at fixed prices divided by the number of hours worked so as you can see germany experienced over this time period relatively robust productivity growth averaging one point five percent so now what happened to wages so this red line shows hourly wages more precisely total labor compensation divided by the number of hours worked deflated by the consumer price index so as you can see cp i deflated wages in germany barely increased between nineteen ninety five and about two thousand seven but more recently wage growth has actually picked up in germany so this blue line here shows wage growth but now deflated using the gdp price index and in germany this actually matters so gdp deflated wages increased more in germany then cpi deflated wages but even gdp deflated wages did not increase as much as labor productivity so this implies that in germany over this time period the labor share declined so what happened to unemployment and employment in germany germany actually had quite high unemployment rates up until two thousand five eleven percent in two thousand five but since then unemployment has steadily gone down and employment rates have gone up in two thousand sixteen and unemployment rate was four percent that is a near record low a level not seen since the later the early nineteen eighties so what about france france experienced a similar growth in labor productivity to germany but in stark contrast to germany wages increased in france in line with labor productivity regardless of whether we deflate wages using the consumer price index or the gdp price index what about unemployment and employment rates in france unemployment has been persistently high in france over these two decades it varied between let's say eight percent and eleven percent here's the picture for italy and spain so neither italy nor spain experienced much of a growth in labor productivity over two decades and wages regardless of whether we deflate them using the consumer price index or the gdp price index did not increase much in either spain or italy and although unemployment has come down in these countries recently it continues to be high and it's above the pre-recession years here's the picture for the united states which in fact looks very similar to germany so in the united states labor productivity has increased more than wages in particular if wages are deflated by the consumer price index rather than the gdp price index so as in germany the labor share has declined in the united states the same is not observed in the united kingdom so the united kingdom is generally considered also a labor market but that is very flexible just like the united states but in the united kingdom over these twenty year period wages have actually increased more than labor productivity but let me come back to france and germany the focus of this presentation so as i've already mentioned labor productivity increased in the two countries at very similar rates wages on the other hand but there there is the big difference they increased in line with productivity in france but didn't increase much less in germany than in france so this means that over this time period unit labor costs have actually declined in germany relative to france and competitiveness has improved in germany relative to france according to this measure here just briefly coming back to unemployment as of 2016 unemployment the unemployment rate is six percentage points higher in france than in germany so that is a very large difference so now what about wage inequality and here the differences between germany and france are frankly really striking so this orange line here shows the evolution of daily wages for full-time workers deflated using the consumer price index at the median in germany okay so in line with what you've seen based on a different data source median wages have not increased between 1995 and 2007-2008 but more recently there has been an increase now what happened at the top of the wage distribution at the 90th percentile well in germany wages at the 90th percentile have increased considerably more than wages at the median so wage inequality at the top of the wage distribution has increased in germany over this time period now what about wages at the bottom of the wage distribution at the 10th percentile up until well just before the recession wages at the 10th percentile declined in germany both in absolute terms and relative to the median but importantly since 2010 wages at the bottom of the wage distribution have picked up again and it seems that the wages at the bottom of the distribution increased more than median wages so let's compare that to France so this orange line shows the evolution of hourly median wages again for full-time workers in France in line with what we've seen before median wages increased more in France than in germany but what about wages at the top of the wage distribution at the 90th percentile now in stark contrast to germany wages at the top of the wage distribution in France have actually decreased relative to the median and although wage growth on average has been a lot higher in France than in germany it looks like that wages at the top of the distribution have increased more in germany than in France what about wages at the bottom of the wage distribution the 10th percentile well once again in stark contrast to germany wages at the bottom of the wage distribution have increased more in France than in germany okay so wage growth in France was higher than in germany particularly at the bottom of the wage distribution wage inequality has increased in germany but not in France now what can possibly account for these divergent experiences between France and germany so i will focus here on one specific factor the role of trade unions and many arguments i will bring forward here based on the joint work i've done with christian dusman bernfitzenberger and alexandra spitzöner published in the journal of economic perspectives and while i believe this is a very important factor of course there are many other factors at play many of them have been have discussed here such as changes in the competition the rise of superstar firms, technological change etc but let me move on to trade unions so trade unions have historically played and continue to play an important role in the wage setting process in both france and in germany and in fact in the eurozone more generally so if you want to understand wage setting in the eurozone we need to take trade unions seriously in both france and germany wage negotiations predominantly take place at the industry level where then the trade unions and the employer federations bargain over pay but also about working conditions more generally in particular working hours in both countries union wages are typically differentiated by skill and they act as wage flaws so firms can pay higher wages but not lower wages so despite these similarities there are also some very important differences in the systems of industry relations between germany and france so most importantly in germany only those firms that belong to an employer federation are bound by union agreements and whether a firm is a member of the employer federation well that's voluntary it's up to the choice of the firm so this actually means a firm that is currently a member of the employer federation and has recognized union agreements in the past can actually opt out of these union agreements and in that case it would then either negotiate wages with the work council so the representative body of the workers at the firm level or after a transition period it could set wages individually with its workers at the same time firms that enter the labour market they may not join the employer federation to begin with and hence not recognize the unions that's not so in france in france the state steps in and declares the union agreements to be binding to virtually all firms in the industry a second important difference is in germany more and more often union agreements include so-called opening clauses so opening clauses allow firms that are members of the employer federation and hence in principle bound by these union agreements to nevertheless deviate downwards from the union agreements if the firm is not doing well so up until very recently this was not possible in france and finally up until very recently there was no national minimum wage in germany there is however a national minimum wage in france that by international comparison is set at a high level by the way the minimum wage in france is directly indexed to inflation so here we have a direct link actually from inflation to wage increases and so this means that the government the state plays an active role in the wage setting process in france but actually not in germany and because firms in germany can say I'm out negotiations between trade unions and employer federations tend to be more consensus-based and less confrontational in germany than in france so let me tell you what actually happened in germany over the past two decades so after the fall of the iron curtain germany was burdened with the coast of reunification but at the same time moving production abroad to the former communist countries in central and eastern europe became a credible threat and in fact should you get hold of newspapers at that time you will realize that this was extensively discussed in the newspapers so paying these high union wages became increasingly costly for firms and indeed firms started to opt out of the union agreements and this is illustrated in the figure here which shows the share of workers who are covered either by an industry-wide agreement or by a firm level agreement which do not matter too much over time separately in west germany the blue line and east germany the red line so in 1995 about 82% of workers in germany were covered by these union agreements by 2016 this had declined to about 60% in 2016 in east germany about half of the workers were covered by union agreements so this de-unionization led to a decentralization of the wage-setting process away from the industry level down to the firm level or even individual level and it has helped to make wages in particular at the bottom of the wage distribution to be more downward flexible it also contributed to the low wage growth we've seen in germany over that time period in particular at the bottom of the wage distribution this is illustrated in this figure here the black line shows the actual wage growth that actually happened in germany between 1996 and 2012 deflated using the consumer price index along the distribution of wages so at the bottom of the wage distribution the 10th percentile wages declined by about 6% whereas at the bottom at the top of the wage distribution the 90th percentile they increased by about 11 or 12% illustrating once more the rise in wage inequality that happened in germany over this time period now the red line shows the counterfactual wage growth that would have occurred if de-unionization had not happened and we computed this using the decomposition methods proposed by dynado, forter and lemieux so this is probably best seen as an accounting exercise but nevertheless this suggests that wage growth indeed would have been higher if de-unionization had not happened by about 4% at the median 2% at the top and 6% at the bottom but this is actually not all that happened because trade unions in germany adopted and they were willing to make concessions that also had a profound impact on wage setting within the unionized sector so first trade unions more and more often agreed to these opening clauses that I've already talked about that allow firms to pay that are members of the employer federation and in principle bound by union agreements to nevertheless pay wages below the union wage so these opening clauses have led to a further decentralization of the wage setting process now within the formal unionized sector and they have made wages more downward flexible even within the unionized sector and at the same time trade unions in germany simply showed extraordinary wage restraint so already in 1995 Klaus Zwickl then the leader of one of the largest trade unions in germany said that his union is willing to accept wage increases based on inflation and the cost of living rather than productivity increases in exchange for more jobs so here by the way we have another direct link from inflation to wage increases but let's take a look what type of wage increases trade unions and employer federations actually agreed upon over this time period so this blue line shows the increase in labor productivity and that's what we've seen in earlier figures this red line here shows the cumulative wage increases deflated by the consumer price index that trade unions and employer federations agreed upon so as you can see that these wage increases are smaller than the increases in labor productivity over this time period and moreover you can see what you can see is that this red line is flat over a number of years so the time period from about 2003 to 2008 is particularly striking here because year after year over five year period trade unions accepted a nominal wage increase just equal to the consumer price index the inflation according to the consumer price index even though over this time period labor productivity had increased and even though over this time period unemployment had started to come down the green line shows the wage growth that was actually realized in Germany as you can see this was less than the wage increases agreed upon by the trade unions and the employer federations and that's first because not all firms are bound by these union agreements and second because of the opening clauses so to recap in Germany over the past two decades we saw a remarkable decentralization of the wage setting process away from the industry level to the firm level or the individual level and this has helped to make wages at the bottom of the wage distribution more flexible and importantly this process occurred without the intervention of the German government rather the system, the Germany system of industrial relations proved to be flexible enough to allow for this change at the same time the differences in the system of industrial relations in France most importantly the automatic extension mechanisms as well as the high minimum wage have prevented France from responding in a similar way and these differences in industrial relations well help us to understand why wage growth was higher in France and in Germany in particular at the bottom of the wage distribution but also why measured in terms of the decline in unit labor costs Germany has become more competitiveness a competitive and quite possibly also why unemployment is now lower in Germany than in France that is the increased downward flexibility of wages in combination with the wage restraint that unions have shown over this period have helped to bring down unemployment in Germany so let me conclude with some more reasoned experiences so I want to emphasize here that wage growth in Germany has picked up since the Great Recession, since 2010 wage growth has in particular picked up at the bottom of the wage distribution and since 2010 wage inequality has not increased anymore in Germany so if I had compared Germany and France in terms of wage growth and wage inequality from 2010 onwards they actually would have looked fairly similar of course they are very different in terms of unemployment at the same time it's interesting to note that the French and the German system of industrial relations seem to become a bit more similar so on the one hand Germany introduced a national minimum wage in 2015 for the first time in its history and elevate not quite at the high level scene in France at the same time both President Hollande and Macron have recently introduced some labour market reforms in France aimed to bring down unemployment in France that have moved the French system of industrial relations a step closer to the German model so one important component of these labour market reforms essentially was a shift of wage negotiations from the industry level to the firm level similar to the decentralization of the wage setting process that happened in Germany over the past two decades but here it's again important to emphasize an important difference between Germany and France because in Germany this decentralization of the wage setting process was not triggered by labour market reform it happened completely outside the political process and gradually occurred over time and for that reason it may actually have been less salient for the German worker than the labour market reforms are for the French worker so more it's important to keep in mind that the decentralization of the wage setting process in Germany well that was achieved in a rather consensus based way because the trade unions were actually on board it was in part supported by trade unions the labour market reforms in France on the other hand have been very controversial have been met with a lot of resistance and trade unions are not on board at all in fact trade unions have been actively involved in organizing demonstrations against these labour market reforms and well let me end my presentation with a statement by Romain Altmann the head of one trade union in France there will be no grace period, no truce now that is a very different rhetoric from that of Klaus Zwickl the former German trade union leader wage restrained in exchange for more jobs thank you thank you Euter please Michael okay so thanks for inviting me to discuss this very interesting excellent paper I think it's fitting that France and Germany came together yesterday and made some very important agreements that we're still learning about and part of this process of understanding how France and Germany can get along for the next hundred years is about understanding how there's a convergence of institutions and how there are also differences in institutions across these two interesting countries and they're also very important not to exclude any other European country but I think if the Franco-German motor works then everything else is going to run better so as I said I'm on board with most of this paper it's an excellent paper based on very careful analysis of micro data coupled with some macro data from the OECD it's relevant because understanding how Germany accomplished an internal devaluation without an exchange rate is probably very important for understanding how Portugal and Ireland have accomplished their adjustments and perhaps how Italy and Greece still need to do some adjustment possibly more in the direction of Italy than in the direction of Greece my comments will be kind of a nuanced view of what's going on in this paper so let me just discuss review what it does it's looking at divergent pay trends in Germany and France in particular but also in nine other countries, in nine countries in total and it looks at both functional capital versus labor the distribution of income in that respect which I think is very important looking at the labor share as part of understanding the story and individual wage inequality within the group of workers that are earning pay it looks at a great decoupling of wages and productivity that you can see in Germany quite clearly in the data this is especially true if you look at the CPI rather than the GDP deflator, I think the GDP deflator is the right way to look at this if you want to look at competitiveness but if you want to understand why workers are dissatisfied you do want to look at the real wage in consumer in consumer goods and deflate by the the CPI or some consumption deflator Uta and her colleagues attribute the recovery of Germany's competitiveness as a result of give-backs what Americans call give-backs or unions making big concessions which we saw in the United States in the 1980s so the United States is always like 10 years ahead maybe 15 years ahead we're very proud of being ahead of the Europeans the Germans just tend to do it right the Americans do it in a chaotic way and the Germans have a controlled it's a controlled increase in inequality but I'm gonna actually make a couple of interesting comments I think you know Uta's I think completely on target and work with Christian that the the unions have been part of this formula but it's not the only part so I'm gonna I'm gonna put a nuance view on this by saying that I think the German labor market reforms of the of the two thousands between two thousand five three and two thousand five were absolutely essential as a as a collateral condition as a as a as a necessary condition for this to work and it's based on simple economics okay so we'll get to that in a second so again overall I'm very happy with this paper and I think it goes back to some older literature that we know on on the difference of of wage-setting institutions across countries and how important this is in understanding so this is a review of what's happened over the past forty years forty five years one of the most important things I got at this was looking at the period up until unification how similar Germany and France are and how important it is for Germany and France to get their act together in the European Union and the monetary union because if the if this axis works then the monetary union will also work German unification was a big disturbance as you can see a big increase in the employment ratio because a lot of East Germans were working at the beginning and subsequently lost their jobs labor force participation jumped for the same reason because East Germany had a higher labor force participation and the last panel shows that GDP and this is GDP normalized on a hundred in two thousand ten hasn't really behaved differently so you really can't say it's because Germany had a great export run uh... had some other reasons for being so successful it's really something is going on within the economy within relative prices and this is the interesting the interesting period is that Germany turned the corner uh... around two thousand five and and France but didn't cyclically they're very similar uh... you can you know one of the most interesting points of the of the paper is that that hourly productivity has grown in a almost tandem way sense uh... for those who are working and this is known this has been known by for a long time uh... solo in blanchard had a nice mckinsey paper a long time ago where they documented this at the at the industrial at the industry level so my comments are basically two-fold i'm gonna look at i'm gonna look at both uh... the macro side and the micro side i'm gonna have a bunch of comments on the macro side and most of my comments are on the micro side and i'm gonna get back to this point that institutions uh... the reform of the labor market constitution if you like the the treatment of unemployed people and the uh... the duration of unemployment benefits are essential to understand why the german uh... labor market adjusted the way it did having the right union attitude is also important they're both necessary conditions for this to work so there's been a well-known well-documented fact that germany has been able to have a real depreciation vis-a-vis its european trading partners is one of the reasons why europe has done probably less well but your germany's done better in terms of uh... the past twenty years uh... the old idea that it's the price uh... stupid it's whether you look at the consumer price index or the or the product price index or a gdp deflator is very very important and i actually have a paper with jeff on this jeff sacks who's left the field but back then we talked a lot about the wage share in germany going in the opposite direction when the wage share was rising uh... with every oil shock the wage share was going up okay so this is kind of an old thing so the question is we want to look at competitiveness in terms of the product deflator or the the price of the stuff that the country's selling and this is an endogenous variable it also depends on on the uh... you know what the what the country's selling and whether it can pass through or not and whether workers if you look at the markup you see this right away this is a counter cyclical variable uh... the the uh... the wage share rises in recessions wage share is kind of the inverse of the markup under certain conditions you can think that over the trend this is because of uh... maybe increasing competition or increasing monopoly power and product markets which is increased the wage the workers get less of the pie could be directed technical change it also just before competition so if you write down the if you decompose the wage share which is an important variable to look at you see that it's got this wedge term uh... which i i just call the terms of trade for lack of a better word you could think of holding wages constant you could have a change in the terms of trade that can make a difference or you could have changing productivity and as uda points out it's really not productivity uh... and it's if you look at the wage share it's they're both kind of declining since the nineteen uh... nineties so it's very slight this is from the from the european uh... union database and i also find that the germans lost a little bit on the terms of trade sense uh... unification so really must be that germans made wage concessions is just confirming what she said wage concessions across the board not just to the median we'll see later uh... if you compare what happened in the u.s completely different story big huge loss of terms of trade this is mostly china probably but as well as other things you see that norway which also figures in her paper actually goes in the opposite direction i think that has something to do with oil prices so on the micro side this is where i differed a little bit i think in a nuanced view would actually look at labor market institutions more generally not just a unionization or de-unionization normally wages are flexible in germany for many reasons uh... i think the fallback position in wage determination is just as important as the willingness of unions to make wage concessions directly simultaneously determined so the i'm gonna ask the question the market clearing wage is really the question uh... is that union concessions and what is that market clearing wage that's gonna have to do with basically what unions have to face when their members become unemployed case of the level of unemployment insurance is essential to understanding this and uh... in germany this was adjusted in two thousand three to two thousand five to two thousand five was the fourth parts law and other things happen so this has got to be part of the formula and you know i'm gonna this is my other work with a with a phd uh... student has actually confirmed this that if you look at the the distribution of work in germany uh... the total number of hours work since the nineteen nineties hasn't changed at all it's been pretty much constant the germans have managed to do a work sharing uh... program through the private sector okay and it basically uh... the way they've done this is to part-time work so the adjustment on quantities is going to be as i agree just as important as the adjustment on wages and if you look at the the difference in uh... growth rates in part-time work to get that huge increase in part-time work it basically occurred during the two thousand decade between two thousand five and two thousand ten so this growth is kind of what's interesting this is taken from her work with christian which is an excellent paper uh... they look at the same data in this paper again with a different slightly different mutation for the wages you see the two things are happening in the nineteen nineties wage distribution starts to open up from above uh... at the highest level and i think she's absolutely right that this is all about unions making concession bargaining and allowing the the companies in trouble to push down wages for their workers who have an increase in dispersion at the upper end but look at the lower end the real drop in wage uh... at the lowest at the lowest this is cumulative wage growth by the way to make sure you understand this is accumulated from the year-by-year from the from from nineteen ninety-nine equals a hundred uh... the big opening of wage dispersion comes in nine in two thousand three that's basically year after the part-time law was passed in anticipation of cuts in unemployment benefits which were implemented in two thousand five okay so you really have two different developments going on and that's part of the story and i think they could stress that a little bit more my co-author uh... doctoral students stephanie sale and i've actually replicated their results we've also looked at part-time workers by making an imputation for part-time workers because they look at full-time workers and you see that it's even more extreme in germany in part-time so understanding this increase in labor supply in the part-time uh... and this is just west germany east germany looks similarly bad uh... increase in low-pay dispersion after two thousand three is kind of the part of the formula so if you look at the slides you can see that it's actually true for east and west albeit a little bit different uh... east germany has a different trend to catch up to the still catching up to the west and uh... part-time you can see is much more extreme the wage the hourly wage imputed drop for hourly wages this larger now this is amazing this also comes in with her but there are other remarks france has actually had wage compression during the same period so those who had jobs and that's why this crazy high growth rate for the for the tenth percentile actually exceeding the the nineties percentile that's you know that's really showing up in the u-s-c-d-a this is full-time workers i just pulled this from the from the economic uh... outlook so they're not they're not exclusive but they should be sort of contrasted and i think you know that it's almost they both have to to be true to have uh... to understand what happened in germany and uh... in our paper we we take an old paper by larry catson kevin murphy we look at just basically trying to understand where these wage uh... uh... employment correlations come from across cells of the data set and we we say okay this is the supply side is uh... act is operative in the in the hearts reforms you just see an increasing negative correlation of wage growth and employment growth across these cells in that period which would not be true before and after in in our work we basically look at that some of the kind of zip through this if you're interested you can see uh... logically if supply curve is moving you should observe across groups and wage uh... demographics you should observe a negative correlation of of wages with uh... employment on the other hand you know obviously this could be true in both in both hypotheses if it's about the supply side and marshals view of the labor market is right then you actually see unemployment rates uh... falling and participation rates rising instead of falling because if you're forcing workers to accept wages they don't want to work for them you'd expect to expect wage labor supply to decline again this is a diagrammatic idea so the the shun bag this month's story will be on the right-hand side if it were just wage rigidity and if in fact you've got a an increasing uh... pressure on labor supply holding demographics otherwise concede expect a participation actually to to rise so we look across uh... some cells and we actually get during this period the comparison two thousand five plus minus two years two thousand ten plus minus two years is negative and uh... in the years afterwards uh... it's looking more positive so if you look at the if you look at the post if you look at the pre the pre two thousand the hearts reforms you get no correlation at all and if you look at more sort of a sort of a finer grid uh... you still get the same the same finding so the the participation rate is is negatively correlated with the wage uh... across these cells in the post hearts period so let me just finish by talking about some of the implications for inequality at the personal income level this is not the same thing as as wage inequality i think a lot of people tend to conflate those two concepts uh... the german magic has been trying to redistribute income using the social welfare system topping up low wages so even if you work for a low wage even if you work for the minimum wage and if you can't feed your family you get a top up it's like the american earned income tax credit program this is what has made uh... germans less dissatisfied possibly more willing to accept these these rather uh... tough nominal changes okay so if you if you compare this with the united states and britain u k you have you still have an increase in the income genie across across countries across uh... periods excuse me uh... whereas in uh... in germany this has been fairly fairly constant it's because the heart system the hearts reforms also allowed for topping up of people working at low wages this is the last slide i really like this one this was sent to me by pia cauc who showed me that basically the you know you everyone likes to work for a job that pays well but it's better to have a job than not to have one and this is based on survey evidence uh... in a sort of successive years showing that in uh... france things uh... people could be made possibly a little bit more happy with a little bit more inequality as long as uh... people take care of the income redistribution aspects so i like this paper a lot i'd really like this paper and we've cited it in our own work i think uh... they show that that nominal wage adjustment was the way it happened uh... time partially by an absolute nominal wage adjustments and this was crucial for understanding what happened in germany and perhaps understanding why germany was able to pull off this internal devaluation which is something that we used to think was impossible it happens in the united states all the time maybe if europe wants to move in that direction they better better get used to it in both ways which also means that german wages have to start rising a lot uh... to wipe out their uber competitiveness that they have right now uh... other margins of flexibility were important so i already mentioned the quantity margin part-time work is extremely important uh... in the paper i my own paper that i say we actually try to do it an attribution sort of an imputation of wages for these people because the germans don't actually get part-time hourly wages and you can see that those adjustments are even larger uh... but the incentives to work part-time actually were implemented before the heart's reforms that was the part-time law that was passed in two thousand two which increased incentive benefits incentivized working part-time just to keep in the pension system and get credit for the the time work the germans have still have a fairly defined benefit system which is it's like that the dutch uh... in many respects kind of it understanding the germany reform you just need to go back to the netherlands in the nineteen eighties and understand what they did and i think you can't avoid saying that the heart's reforms did a lot of stuff but they most importantly reduce the reservation wages and increased labor force participation for people who had been outside the labor force old people especially women uh... who had been uh... outside maybe raising a family and had an education and i think the correlation of relative wages and employment across the cells that one would like to look at are actually going to support a nuanced view of what the germans did uh... since since nineteen nineties thanks or if you wish to for sure i will give uta some minutes to reply to michael and then we will make uh... the round of questions thank you michael for the very nice uh... introduction and i uh... for the very nice discussion and i actually agree with you the heart's reforms method as well so to put it simply the heart's reforms provided incentives for workers to accept low wage jobs but now imagine this reform had been implemented in france with the high minimum wage and the automatic extension mechanisms it's very unclear that in france it would have created new jobs right because wage levels are too high so i think there is an argument to be made here that there is an important interaction between the system of industrial relations and the unemployment benefits system so uh... if wage laws are very high firms have no incentive to create low wage jobs if unemployment benefits are very high workers have no incentive to accept low wage jobs in that sense we totally agree i will start with this side and i know there was one gentleman in the second row who already tried and the next then louis then daniel growth from the center for european policies in brussels my question was actually related more to the previous session but i think it fits also this one we have discussed a lot the details of price setting and measurement of various aspects of the cpi and in this last session more about wages which then later presumably via markup and go into the cpi but if you ask ourselves why are we interested in inflation and the general price level then i think uh... if you go back to what larry summer said at the beginning of this conference we're concerned because we have very high debt levels and therefore we're concerned about that deflation that has not been able to serve as a debt but if you think about this way then we're not so much concerned about the cpi but we're concerned about actually nominal gdp meaning the nominal the gdp deflator yes that determines the average revenues tax revenues governments can get and the revenues firms can obtain to service their own debt if we have that perspective we want to combat deflation because we are fearing debt deflation then should be so much concerned about whether cpi is properly measured the half a percentage point here and there goes into profits or markups because all of this then goes into the gdp deflator and that should be our summary statistics two conclusions one should be actually then look less at the details of the cpi and more whether the cpi and the gdp deflator evolves along the lines that suggest no debt deflation and secondly whether central banks perhaps also change their target the target was done was chosen looking backwards to periods of very high consumer price inflation and low debts today we have the opposite combination and therefore should be not ask ourselves whether we should switch to a different target many things perhaps Luis thank you very much thank you very much for these excellent contributions I will go back to the dynamics of the labor market wages with two remarks the first one is that unit labor costs according to your paper matter and the relative performance of unit labor costs especially in a monetary union matter more because you cannot resort to the depreciation or devaluation of the nominal exchange rate so I think that at the end of the day in a monetary union what determines the relative performance of the different economies is the relative performance or evolution of unit labor costs and this is something that is especially relevant and I think that all the problems that we had some years ago in the monetary union were cost not as much because of fiscal policies divergences but much more because we had different levels of competitiveness among the different countries of the monetary union and the second remark is that there is a clear trade-off according to your to your papers to your contributions between wage equality and employment creation you have put forward to concrete cases Germany and France Germany you know with a better performance relative performance outperforming France in terms of unit labor costs and having a better performance in terms of employment creation but simultaneously in the case of France wage equality is better than in the case of Germany and I think that there is something that is especially relevant is the starting position of the different countries for instance Spain implemented in 2013 a labor reform and that what finally pursued was the decentralization of the wage bargaining process with a 25 percent unemployment rate as Spain had in 2011 the only possibility that we had was to create jobs so that the initial starting point of the economies is going to be extremely important in terms of the labor market reform and the regulation that you want to implement I don't know whether you can elaborate a little bit more on that Thank you. Krishna Guha, Evercore Partners, question for Uttas you described wonderfully the process by which Germany was able to facilitate this internal devaluation a lot of the changes you described of course would also be associated potentially with a lower natural rate of unemployment so I wondered if you do you have any quantitative assessment of the decline in the natural rate associated with these changes in Germany and even more ambitiously would you speculate on what kind of impact this could have in France and other countries adopting these reforms now Perhaps just one row back Thank you very much. I enjoyed very much the presentations I just wanted to make a very general question that it even might apply to the previous panel and is your assessment of how much global trade has affected the wage determination in advanced economies and also in emerging market but you know having China and India be more present in day-to-day economic activity globally might have an effect on wage setting and that might be an explanation also why we we see so much quarrel with open trade Thank you Thanks and perhaps the gentleman at the back Charles Weplo's Graduate Institute I want to go back to the point that was raised by Mr. de Guindos I thought the presentation would both actually were fascinating and you hit this apparent trade-off between wage inequality and unemployment or employment whichever you want to set it now my question is in principle is this a theoretical condition that there is a trade-off you can increase one and not the other or improve one and not the other and how much empirical evidence we have are these just two strange cases that you put face-to-face or is this a general principle that we find in the data Well many things let us finish with the first round of questions and Uttar perhaps you would like to start Okay so the first question was about the CPI index rather than GDP index so I don't consider this my expertise as labor economists we typically focus on the consumer price index simply because that's viewed as a better measure for the cost of living so let me come back to the other questions that was raised so first part of the question over there was well what did the internal devaluation that Germany actually achieved have implied for the other countries for the other EU countries and yes that is an important point on the other hand I mean Germany started with a high unemployment rate and some devaluation was probably necessary to bring down unemployment there have been a lot of questions about the trade-offs between wage inequality and increase in wage inequality and the level of unemployment so now if the unemployment level is high and we want to bring that down yes probably there is a price to pay and that price is increased wage inequality on the other hand but it also should be pointed out here that part of the declines in wages at the bottom of the wage distribution seen in Germany is precisely because more workers have entered the labor market right so the composition of workers who are actually working has changed and that in itself might have brought down wages another aspect that is really very important here is the dynamic aspect so it makes a big difference where the workers are stuck at these low wages for year after year over 10 year period or whether these low wages are actually stepping stones to better jobs so I think we do need more research on this dynamic aspect and that's also what my current research agenda is partly about so the question over there was also about the trade-off between wage inequality and unemployment so there are countries which have relatively low unemployment rate and not too much and relatively low wage inequality like the Scandinavian countries Germany since 2010 had very low unemployment rates and actually wage inequality declined so maybe there is not necessarily a trade-off but we start at very high unemployment levels and want to bring this down as I've mentioned before the price probably is wage inequality at the bottom goes up so another question was to try to quantify how much of the decline in unit labor costs will increase in competition to what extent it brought down the natural unemployment rate now that's a very tough question to answer and I'm a labor economist trained in program evaluation methods to get causal effects I cannot answer that question because all we have here is variation over time so it's close to impossible to disentangle but what's the impact of the decline in the unit labor costs what's the impact of the increase of the labor market reforms from just general time trends that have happened over time there also was a question about the impact of global trade on wage setting so research for Germany actually suggests the research that in general has been done in this area was more about what it did to employment levels and less so how it affected wage inequality so in the US there has been some very influential work that increased competition with China has reduced as the decline drops in the manufacturing sector that did not happen to the same extent in Germany simply because Germany has a trade surplus overall and even though not exactly with China I guess it is interesting to note that wages in China have more recently come up the manufacturing sector in China has started to go down and service sector has gone up as we would expect in a developing country so maybe this is actually now decreasing the wage pressure in the advanced economies Michael you'd like to comment I just had to answer Charles's point I think it's not a necessary trade off but I think some countries like Ute said are better equipped if you educate your less skilled workers or have continuing reeducation like Denmark or Sweden I think you have a better chance of picking up the people who lose their jobs at the lower end of the distribution and I think if these people are out of the labor force for a long time or out of work they lose skills and they become even less productive so the wage dispersion is kind of a the necessary wage dispersion could actually get larger and I wanted to say one more about your point about the Nairu there are some people who have tried to do this with models and one of the hearts reform aspects that people don't talk about very much is the hearts three reform which increase the efficiency and the efficacy of the employment agencies which Germany put on the computer it's all nationwide you can find out whether vacancies everywhere and there's now this increased pressure on people to take the jobs makes a big difference I think that probably has knocked at least one or two percentage points off the if you believe in the Nairu or the band it's gone down and I think this explains why German wage pressure is still rather you know manageable at the present Many thanks I have a gentleman at the back who already tried first I'm a journalist from Italy from Corre della Seja Federico Fubini thank you it was a fascinating presentation I wish Italy could have applied some of the same arrangements at firm level but I have one question Mark on what is you define as competitiveness or social inclusion increasing the employment rate because for instance my question about German competitiveness is to what extent it depends from the arrangements that you have illustrated and how much supply chains with central and Eastern Europe matter so it would be interesting to separate the two things and how much maybe Germany was better but I don't think there is a single complex industrial product in Germany that is fully made in Germany so to what extent the value out that it has been produced somewhere else in the European Union to what labor costs so I would I wish I could see something that makes this difference and the other point it is very interesting what you showed on wage inequality but however you mentioned here we are talking about wage inequality not inequality per se and when you look at OECD data on Gini Germany is a less unequal country than others in the EU in terms of income inequality but it is more unequal in terms of wealth inequality and increasingly so and if you think about the fact that many companies are family owned and as you showed the labor share has declined probably there is a measure of inequality that doesn't show up in your graph because it's wealth inequality of family shareholders so I was wondering whether you have any comments on that well I will move to this part but ask please for short questions because you know we are already a little bit late yeah yeah yeah I know short questions two short questions I'm allowed to they're short Richard Portas love the business school the first one among the many striking charts you showed in your presentation one struck me particularly and that was the participation rate chart the enormous increase in German labor participation rates what's the explanation for this especially since you know you might say oh it's part time jobs maybe that's maybe that's the answer but we tend to think of obstacles to female labor participation in Germany that doesn't seem to have played the kind of role that you might expect the second question is a bit more directed towards Michael but not entirely we talk and the 90% of the talk about you about competitiveness however interpreted is about unit labor costs but isn't that shouldn't we be thinking more in terms of the real exchange rate and you come close to it talking about P over CPI but shouldn't we really be talking about the price of traded relative to non-traded goods that seems to me to be the central and if you do look at that as for example Sophie Pito has done in her research you find that the story is really quite different anything and in the back there the gentleman the first on the in the second row you are exactly correct there yeah Gunther Wolf Brügel I have a question also on this fascinating paper as regards the role of capital if capital and labor are complements which I think most estimates suggest then wage developments and the labor share also depend on investments and the capital that the economy has and one of the interesting features of the German economy is that both the public sector as well as the private sector have had very very little investment throughout the last 20 years the corporate sector has had investment below that of France and Italy for the last 20 years the public sector as well the private sector has the corporations have massively deleveraged basically starting 2003, 2004 around that time so I guess my question is what role does this scarcity of capital play in the wage developments that you documented is it a cause or an effect? thank you and the last one, the gentleman on the side Jonathan Hazel, I'm a PhD student I was wondering because you have this great German data where you can track individual workers on time one can potentially disentangle these two sources of wage variation between worker compositional shifts and within worker growth over time and of course these are conceptually very different and one could have slower overall wage growth or simultaneously having rapid wage growth within individual workers and I was wondering if you've done that and if so what happens and if that changes our results in some way thank you many thanks well I have to finish the round of questions because we are late not as late as yesterday, Christine, yeah? so please, quick, quick answer very quickly, yes it's very important to distinguish between wage inequality, income inequality and wealth inequality that is often muddled up in the discussion it depends on the question you're addressing which is the more relevant one there was a question about the striking increase in labor force participation rates in Germany Michael probably knows more about this than I do but it was to a large extent because women started to return to the labor market although they had been out of the labor market for a while many returned part-time and the Hartz reforms probably had something to do with that so the last question on wage, on compositional shifts and wage increases or decreases within individuals you are absolutely right, Jonathan, in Germany we have longitudinal data to exactly analyze this question and this is what we are working on right now so I can't give you answers but it's very clear, very clearly, very important if we want to understand the impacts of the increase in wage inequality and finally there was a question about the role of capital in the wage-setting process let me be honest here I don't know the answer to that question okay, Michael I can try to pick up on Gudram's question because it's simply an isoquant the low wages in Germany have actually encouraged firms to move away from capital and I think that's the best way of explaining what's happened and if you look at the level of technology and meat-packing plants in Germany I've heard the description of this from people who are actually from Belgium and France they actually complain that German methods are more labor-intensive because they're importing workers from Eastern Europe these are insourced workers from Eastern Europe to work in these meat-packing plants which probably will go out of business now with the minimum wage so thank goodness the economics seem to be working I have one last comment I really liked the last slide of Ute with this that you showed with the picture of the worker burning tires the French love to burn tires and strikes and this to me is kind of a pinnacle of the whole discussion if you want to get conversions between Germany and France or France and Scandinavia you need to create trust between workers and management and the Works Council for all its criticism in the United States is actually a very effective way of sharing data with the workers via management and it has led to a lot more accommodative outcomes if the French have the consulate entreprise but it's not really effective and obviously it's not effective either because you wouldn't have people burning tires it's a complete waste of time and resources so let's try to figure out a way to get collective bargaining on track in France and Italy to get the outcomes that we have talked about Okay, well, many things I do see that we have still many many questions I'm very sorry before closing the session I'd like to come back to Daniel's question on the GDP deflate I mean the central bank's mandate is all about price stability so it might make sense to look on price setting apart from all the questions around the GDP deflate but this would be perhaps a new session in a new environment next year who knows, yeah? Many many thanks many thanks to my presenters, to the speakers and to the lively debate and now you will have a shorter a shorter call break, yeah? Drink a little bit quicker talk a little bit faster and you will be called in for Benoit's panel then afterwards and don't forget to vote for the young economists