 Hello and welcome back. Some say it's not the job of monetary policymakers to provide jobs or care about inequality, that those tasks lie with governments instead. Yet, while wealth and income distribution may well lie outside the policymakers' ambit, they can have a substantial impact on monetary policy transmission. Whether or not they're responsible for it then, central banks need to care about these topics. That is going to be the theme of this coming session. I now want to introduce the chair, Philip Lane of the UCB's Executive Board. Mr Lane, over to you. Good afternoon and welcome to this panel. The connection between monetary policy and employment has to be one of the longest running topics in economics. And this is really fundamental, I suppose, to our day-to-day business in understanding how monetary policy works. This is for several reasons. One reason is the way we think inflation dynamics are maintained over the medium term especially is really through the labour market. It's very difficult to have inflation at 2% in the medium term unless we have a sufficient rate of wage inflation which, taken into account productivity, is consistent with a 2% inflation target. So that's a very basic reason. A second reason and it's the most basic reason for our medium term perspective is we know there can be adverse supply shocks in the economy which push inflation and employment in opposite directions. And to allow us to manage that conflict, that trade-off, we take a medium term perspective. Now, the reason why we've brought together the topics of monetary policy, employment and inequality into this session is that once you spend any amount of time thinking about it is that the relation between employment and monetary policy has to depend on the nature of the very many different types of inequality we have in society and in the labour markets. So people are different along many dimensions. Skill level, education level, the sector they work in as we know from the pandemic by age group, by gender, by income, by wealth and all of those dimensions may have a role to play in understanding the interconnections between monetary policy, employment and employment. So at the very least we need to take this into account in order to have a good model of how monetary policy works. Of course, when we take monetary policy actions, regardless of their effect on inflation and employment, these policies may have an effect along some of those inequality dimensions and for transparency and accountability we should be clear about those side effects. In addition, in relation to this topic, inequality may also matter for other reasons. For example, we think inequality of income and of wealth may play a role in determining the underlying equilibrium real interest rates, which of course is a big topic for a central bank. Now, I'm very pleased today that we've brought together a really excellent panel. All of the panel members are really working at the intersection of macroeconomics and labor economics, which is what you need for this topic. Let me emphasize also all three contributors are also very much dedicated to a type of research, which we find very helpful in central banks, which is to focus on quantitative implications that it's not enough to have some kind of partial understanding or qualitative understanding of these forces. We need to be able to understand the quantitative contribution of different dimensions to weigh up the different elements. So our panel today has three very distinguished academics. Juan de Lado will speak first, from Costa Terceira in Madrid, then Antonella Trigari from Bocconi and Farni Jean-Lucca Violante from Princeton. So each will have 10 minutes for opening presentations and then we'll turn to a discussion. And of course, let me remind the audience in terms of raising your hand, you can join the queue to ask questions or make comments in what I'm sure is going to be a very good discussion. So with that, over to Juan. Thank you, Philip, for your kind introduction and good afternoon to all participants. The topic I wanted to address in my introductory remarks is whether inequality should be a concern for monetary policy. This topic is becoming increasingly relevant and increasingly popular in the academic literature. Basically, of course, because of the long-standing trend in the concerns for inequality since the 80s, which were aggravated by the unequal outcomes of the recovery of the Great Recession and in macro modeling by the growing influence of micro-level heterogeneity acknowledging that we are not all the same and the search and matching friction framework which introduces endogenous risk faced by different individuals. So this has put inequality at the center stage. For instance, just to give you an example, the recently released ECB Workstream Report on Employment devolves, as Philip was saying, a large section to the implication of inequality for the conduct of monetary policy. So here we have two views. The traditional view is that distributional issues shouldn't be considered as relevant for central bank policies. They are side effects because the goal of central bank policies to stabilize the economy as a whole, and I emphasize that. The alternative view, which has been put forward more recently, is that monetary policy, which I denote by MP, could have non-negligible direct effects on inequality at the business cycle frequencies, which interacts with the other channels of the monetary policy mechanism. To keep in line with the current stance of monetary policy in the sequence, in the following slides, I'm going to focus on the impact of expansionary monetary policy and inequality rather than contractionary, which is what most people believe costs higher inequality. So to remind you of the main channels through which expansionally monetary shock think of a sudden cut in interest rates affects inequality, I'm going to summarize them in five categories. The savings redistribution channel, which benefits borrowers and hard lenders, so one can think that that will reduce inequality. The interest sensitivity channel, lower interest rate, higher asset prices, higher inflation, the former favors the richer, the second harms the poorer. That increases inequality. The household firm has a united channel, so expansionary monetary policy may ease access to financial markets, may affect intertemporal rates of substitution, may benefit mortgages, small young firms, so we think that that tends to decrease inequality. The income composition channel, so it's how different sources of income, wages, salaries, profits, transfers, change with expansionary monetary policy. Here we have ambiguous effect, and the labor earnings at the united channel, which depends on the skill, education levels, and occupations. And I'm going to argue that that one could be worse by expansionary monetary policy. Just to give you a simple representation of the findings which have been reported in the recent literature, this is, let me just look at the point to this graph, which is from a paper by Anderson et al, which uses very highly granular information. They have individual level tax records and balance sheets for the entire adult population in Denmark over a long period. And what they do is they simulate the changes in the income shares according to the exact position in the income distribution for a 100 basis points cut in interest rates, which in Denmark is sort of an exogenous event since the crown is pegged to the euro. So they inherit whatever happening, what is ever happening with the euro rate. So what you can see in this graph very clearly is that there is what I call a positive income gradient. The upper part of the income distribution, of the exact income distribution, gains in share about 3.5%, whereas the lower part loses about 2%. So this is the phenomenon that I want to draw your attention to. So let me just summarize a very, I think a very logic channel which has been somewhat unexplored in the Monetary Policy Mechanism, which is the role of investment. So I'm going to rely here on results from a recent paper with Gergo Motioski and Evi Papa. Well, just to simplify matters, think of a world tank model to agents so that we have high skill individuals and less skill individuals. So a key issue here is that capital, which is of course increased by a reduction in interest rates or by quantitative easing, exhibits skill complementarity. So think that capital equipment and high-skill workers are complements and capital equipment and low-skill workers are substitutes. For instance, this is the famous production function advocated by Yaluca and Kothos. But the reasoning I'm going to give you is very similar. If we think of investment in artificial, intelligent robots and the decline of routine tasks along the lines of multimodal and restricted people. So that's one element. Capital increases the marginal productivity of some type of labor and decreases the marginal productivity of the other type of labor and vice versa. And second is that the search and matching frictions are not the same for both types of workers. So typically low-skill workers have higher separation rates. They have lower matching efficiency and they have lower bargaining power. So what is the insight of the fact that I'm going to highlight? You have an expansionary monetary policy shock that increases investment and other components of aggregate demand. That because investment is a complementary factor of high-skill labor, that increases the relative demand for complementary and more fluid high-skill labor, which of course increases the marginal productivity of capital so that increases investment, which in the second round increases against the demand for high-skill labor. So there is a multiplier loop, a demand amplification effect. Of course, capital-skill complementarity, CSC plus asymmetric search and matching friction, the outcome is going to be an increase in the relative income of high-skill workers versus low-skill workers. That is, can be decomposing to a change in the skill premium, relative wages, and a change in the relative employment rates, a change in quantities. The main result is that the interaction of these features, capital-skill complementarity and asymmetric search and matching frictions, yields much stronger effects on the relative income shares than the sum of the two separate forces. So for instance, in our theoretical model, we simulate again the effect of a one percentage point cut, an expected cut in the annualized interest rate, and we compare different outcomes. I want to highlight the purple dash line is when both features are relevant, capital-skill complementarity and asymmetric search and matching. You see that the relative income share of the high-skill relative to the low-skill increases by 1.5 percentage points and then goes down. Whereas when we consider these effects separately, only capital-skill complementarity or only asymmetric search and matching, the effects are much smaller and the sum of the two is lower than the interactive effect. When you confront this with the data, for instance, with the U.S. data, using information on real wages and employment rates of people with workers with college degree, non-completed college degree, and so on, what you find is that the predictions of the model are supported by the data. So for instance, this is again a cut in interest rates by 1 percentage point. The employment ratio increases by 0.4 percentage points. This is here in after 40 months, whereas the wage premium also increases by about 0.2 percentage points even after 40 months. Finally, and I want to stop here, I just want to raise an issue for the subsequent discussion which is also related to the issue of job polarization and capital-skill complementarity, but this time with the slope of the price-fillings curve. The idea is the following. Look at this, the left-hand side graph. This is the slope of the fillings curve where inflation is a regress on unemployment. So you should find a negative relationship. And this is eight-year periods from 2002 to 2020. And as you see, starting around 2010, the elasticity becomes zero. So the fillings curve gets completely flat. On the right-hand side graph, what you have is the decline in the share of routine jobs which have fallen by almost 7 percentage points from 2002 to 2018. So the basic idea I was conveying before is that the market for non-routine jobs is more fluid than the labor for routine jobs. Routine jobs, critical jobs, production jobs, et cetera. Fluidity is going to make the curve much flatter. And here I quote a couple of papers which explore this possibility. And I think that's all from my side. Thank you very much for keeping to the time and for the excellent first contribution. So now I turn to Antonella. Can you hear me now? Perfect. Thank you. So first of all, thanks for the invitation and for the great opportunity to participate in this panel. So what I would like to do today is to share some remarks about how to best assess slacking the labor market and being inspired by the greatest emphasis that has been put in both academic circles and policy circles in the US regarding, you know, measuring slack. And this is likely due to the explicit dual mandate of the FET. So... Oops. Sure. Sorry about that. Okay. So... I'm phrasing what Philippe said at the very beginning assessing labor market under-routineization is key to monetary policy. And the four are time and the reliable measures of labor market slack are key inputs to monetary policy for two main reasons. First, because they provide a measure of the cyclical position of the economy and thus permit the assessment of whether the short-run inflationary pressures are indeed tolerable in presence of a trade-off. And they also provide a signal of demand-related inflationary pressures. Now, historically, policymakers and academics focus on the unemployment rate as the primary measures of under-utilization in labor market. However, it is well known that the unemployment rate may not capture all the margins of labor market slack. And indeed, in its statement of the longer-run goals of monetary policy, the FET explicitly writes that assessing its maximum employment, it considers a wide range of indicators and the famous Yellen's labor market dashboard is an example of this strategy. Now, the first point is that the unemployed are not the only job seekers. The fact that we observe large flows from non-participation to employment and even largest flows from employment to employment is a signal of some unmet demand for employment or effective labor supply that is not captured by the unemployment rate. The second point is that the pool of job seekers is heterogeneous in particular in terms of their likelihood to find jobs which is then a proxy for their effective labor supply or search intensity. Marginally attached workers, for example, discouraged workers likely to find jobs then the short-term unemployed. And the unemployed themselves are heterogeneous in terms of demographics, duration and history. For example, long-term unemployed workers are well known to have lower job finding rates than short-term unemployed workers. So, these implies that simple counts of the number of job seekers, for example, as some measures of alternative labor market underutilization that are computed by the BLS, the U4, U5, U6 unemployment rates will which simply sums different categories of job seekers from unemployment and from outside the labor force, we failed to capture this heterogeneity in propensity to search and propensity to work. This illustrates the importance of accounting for search intensity. Let me compare the evolution of the unemployment rate in the U.S. and in the Euro area during the COVID processions. So, if you look at the first plot, you see that the unemployment rate in the U.S. increased by more than 10 percentage points while in the Euro area the unemployment rate at the onset of the COVID crisis only increased by 1.5 percentage points. Now the main reason for this is not a different fundamental behavior of the labor market is simply that official statistics report temporary unemployed differently in the U.S. in the Euro area in particular temporary layoffs are employed absent from work in the Euro area while they are counted as unemployed in the U.S. Now, so essentially they are given a zero weight in the Euro area as if their search intensity was zero while they are given a weight equal to one in the U.S. unemployment rate as if they were searching with the same intensity as other unemployed workers, permanently unemployed workers. Now if we construct two contractual rates that bring these two measures closely together in terms of definition essentially by subtracting temporary laid off from the U.S. unemployed rate or adding temporary laid off to the Euro area unemployed rate then we observe a quite striking result because in that case the increase in the unemployed rates becomes extremely close in the two cases. Now what is the key takeaway is that actually neither of these scenarios is likely to be appropriate because what would be appropriate is to weight temporary laid off workers by their actual search intensity which is not going to be zero but definitely is going to be below that of other unemployed given that these workers have the option to go back to their previous work. So how can we construct a measure of effective job seekers a comprehensive and synthetic measure we can do it by doing two things first of all to consider the entire population of effective job seekers and second by weighting them by their search intensity now the challenge of course is measuring the search intensity and different approaches have been proposed in the literature but one particular approach is the one followed by Abrams, Vangaren, Brandelina in a recent paper where they infer the relative job the relative search intensity from exposed outcomes from exposed transition rates to employment from job finding rates and using CPS data to track the flows to employment by the initial state or suggesting by demographics. So what they do they estimate relative search intensity for 22 groups 13 among the unemployed, 7 among the non-participants and 2 among the unemployed and here is the results which show a very wide range of relative search intensity across the different 22 groups. So what I did is to build to construct in the spirit of Abrams and Quarters an admittedly very rough measure of effective job seekers for the Euroator using the available data I don't have as of now the micro data I will in the future and in particular I consider so the available data is 2006 quarter 1 to 2021 quarter 1 I consider only 6 labor market states 2 among the unemployed 3 among the non-participants short-term and long-term unemployed among the non-participants seeking but not immediately available but not immediately seeking and then others and finally the employed and then I compute the relative transition rate and weight these stocks and finally normalized by the working age population to obtain a rate of effective job seekers. Let me show you the results and let me mention that of course there are some Euroator-related caveats in the construction of these measures we already mentioned job retention schemes but on top of that also distinguishing between fixed term versus open-ended contract which workers in these two contracts might be extremely relevant. Now in terms of results what we find is the following. So in the figure you see the rate of effective job seekers against the official unemployment rate shift in the scale so as to you know align the needs and what we find what I find is that the two series are highly correlated and both counter cyclical but they have very different volatility in particular the job seekers rate is much less volatile with the standard deviation which is 35% of the standard deviation of the unemployment rate. Now why is the volatility dumping? There are many reasons for that one reason is that and here I mentioned the two principle reasons one reason is upsetting changes in the cyclical composition of searchers during a recession you are going to have more unemployed job seekers but less employed job seekers and the opposite in expansion. The other important component dumping in the volatility it's the down-weighting of the long-term unemployed which essentially because they have much lower weight in the short-term unemployed is going to and both components are you know accounting the standard unemployment rate this is going to reduce mechanically the volatility from this component. Now the next question I ask is whether the unemployment rate is indeed you know an imperfect signal of the true measure what you know I take to be the true measure of lack of the effective job seekers rate and the point is that if job seekers of each of the you know the components of the job seekers rate would move together with the unemployment rate then it would not matter to account for them separately because the implication of course for wage and prices through the lines of a Phillips curve would be exactly the same but of course so the first thing is that this is not possible because in this measure if anyone category increases then this is going to translate into a decrease in the other categories but still to explore whether you know quantitatively whether the unemployment rate is an imperfect signal of the job seekers rate I standardize the two measures and compare them and what I find is that you is indeed an imperfect signal in a quantitatively meaning manner and in particular it underestimates lack during the recession while it overestimates during books so while let me not go into the reason for this but just mentioning that the story during the great financial crisis is a very different one than the story during the COVID crisis so the last thing I wanted to do but maybe I can even if I'm out of time I can talk about that during the discussion later on so let me stop here thank you Antonella and no doubt there will be time to come back to this final dimension so with that let me turn to Jean-Lucas so over to you Jean-Lucas all right thank you thank you very much for the invitation Philip I'm delighted to be part of this panel so both one and Antonella mentioned in their presentations Hank model so what I want to do in this 10 minutes that I have available I want to reflect a little bit on what we have learned from these new cluster models so far so I want to start from the observation that central banks can deploy sweet or very different models to inform monetary policy decisions mainly with two objectives the first one is that of forecasting in the short or medium run and the main tool is vector or progressive models but they also need models to interpret the data and run policy counterfactuals and the main tool that they use are structural DSG models now traditionally these DSG models are very intelligent models so models with a representative household or models with very limited heterogeneity so models with like two types of agents for example a spender or a saver or a borrower and a saver so I was very glad to read in one of the excellent background papers prepared by the ECB staff for this strategy review the review of macaroni model in the Euro system that given the achievements in the academic sector of banks should venture into this new era of modeling and advancing the empirical validation of those models so what are these HANK models so oops sorry so the HANK stands for heterogeneous agent so the NK stands for the New Keynesian block of the model which is essentially two equations to equilibrium relationship the first one is the Philips curve that summarizes the production structure of the model and the second one is the monetary policy rule that summarizes the sort of the behavior and choices of the central bank the HA stands for heterogeneous agents and here it's the novelty because these models replaced the aggregatorial equations of the IS curve with modern theory of consumption and saving essentially the buffer stock model so also these models are heterogeneous either ex ante because of different preferences for example different discount factors with conversion or skills and occupation as in one presentation and exposed because of different labor market trajectories as in Antonila's presentation where you have some workers who are employed that are unemployed that are out of the labor force and so on but the key thing is that there is imperfectly sharing among workers of the labor market the aggregation results that leads to the representative agent representation so what emerges is an equilibrium distribution of income consumption and wealth that kind of resembles in the data and at the cost of oversimplifying a little bit these models are basically three groups of households with kind of endogenous mobility among them you have the hand to mouth households where kind of low liquidity and high margins to consume a kind of a middle class for which the precautionary saving model is very strong because they want to stay away from the credit limit to smooth consumption and then you have the wealthy which are kind of the high net worth individuals so what what do we learn from this, what do we learn from this kind of class of models so far so I'm going to argue that we learn a four main lessons the first one is about the transmission mechanism of monetary policy so the transmission mechanism of monetary policy to the real economy is different in this class of models in the representative agent models the key channel of transmission is intertemporal substitution so the central bank cuts interest rates and households they consume more and they save less that's kind of the standard intertemporal substitution channel in heterogeneous agent models there are a host of additional effects so intertemporal substitution is still there but this effect is dominated by a number of other indirect effects that take place through the general equilibrium forces so changes in labor income and employment asset prices, mortgage rates revaluation effects on debt and even fiscal policy as I will explain later so what we learn is the centrality of indirect general equilibrium channel and this is even truer for unconventional monetary policy which by definition works through basically asset prices so what does it mean for the practice of monetary policy it means that there are a variety of intermediating factors that are sort of outside the direct control of central banks and that depend on institutional design market structures and a number of other factors which are very country specific just think about the housing market how different housing market is across countries so that makes from the point of view of bank models that makes the job of the monetary authority much harder because most of the effects of monetary policy occur exactly through market equilibrium in various markets the second lesson that we learn is that there are several sources of amplification of monetary policy related to representative of monetary policy the first one is what one called the redistribution channel so let me explain this channel consider a monetary policy shock that expands how the GDP the key observation is that the incidence of such expansions varies across the distribution so the micro data as you can say here you can see in this plot the micro data tell us that the elasticity the sensitivity of individual income to aggregate income is much higher at the extremes of the distribution to the middle so at the bottom it's basically about unemployment and at the top is about the performance-based pay which is very sensitive to the cycle so amplification occurs when sort of this redistribution of income occurs from households with no margin of risk to consume towards households with high margin of risk to consume because this distribution increases and expands aggregate spending but also when the distribution occurs from households with no propensity to take risk because that reduces the risk premium makes capital cheaper and increases investment in capital that also in turn increases aggregate demand so this is the first channel the second channel has to do with the cyclicality of labor market risk so basically unemployment risk as you can see in this figure in a recession unemployment risk goes up clearly across the population and that leads to a stronger precautionary saving motive so kind of saving for the rainy day so this increase in precautionary saving further generates a further fall in spending and employment and sort of induces a kind of a downward spiral by reverting this downward spiral the effects of monetary policy can be amplified relative to say a representative agent model where there's no precautionary saving motive no heterogeneity kind of nothing of this sort if you like sorry it's a bit flickering my mouth today the third source of amplification has to do with the fiscal response of the fiscal authority so these are here what I did I just wrote the budget constraint of the government and you can see that a decline in the interest rate kind of frees up some resources because reduces the the cost of servicing debt for the government so how the fiscal authority responds to a monetary policy shock is going to matter in a non-recordian world in particular the size of these extra resources are going to be a function of many things in particular the maturity structure of debt the shorter the maturity structure and the extra resources which are available for the government to possibly distribute to the household sector and where in the income distribution these extra resources and that is going to determine amplification again if they're targeted to the high margin of risk to consume households that is going to increase aggregate demand okay third lesson monetary policy has an impact on income and wealth inequality what do the Hank models tell us they tell us that monetary policy is redistributive and that matters for the transmission so at the very least it's important for central banks to understand and take into account heterogeneity, inequality and the distribution to deepen their understanding about the transmission mechanism but more to the point in this class of models every stabilization policy has redistributive implications and any redistributive policy can stabilize or destabilize the economy so there's really like a very intimate relationship between stabilization and redistributive policies but now another thing that you learn and you see immediately from these models is that fiscal policy is much better suited to redistribute and ensure risks on the household side because it can be targeted much more accurately right and this is something that emerges very clearly very clearly from this model now the key problem of monetary policy is that it is often it's sort of plagued with institutional delays and political compromise and so because of these delays often the first response falls on the shoulder of the monetary authority in this sense for example if the monetary authority has implemented automatic stabilizers and improved their design the monetary authority could fulfill its mission better and speaking of mission going back to one's point that he made in one of his slides should the monetary policy authority be concerned with inequality but here you have kind of these two almost polar opposite responses from the two major central banks in the world and the answer is almost it's like a no unless we interfere with price stability now I don't feel like dissenting with the ECB view in a way for a number of reasons the first one is that institutions we narrowly define missions I think many strengths specialization we know that leads to higher productivity so specialization is a good thing also in institutions in my view the second thing is that there are many strengths in the income distribution which are shaped by very strong market forces like technology and globalization that are impossible to revert with monetary policy or even fiscal policy you can do small deviations there are interesting interactions like in one's paper but at the end of the day the secular forces are going to determine the long run trends so perhaps one little compromise could be given the many tools now more than ever that are available to central banks a compromise could be to aim for price stabilization in the most equitable way so kind of choosing the instruments that as the smallest implications for inequality in my last 30 seconds literally let me talk about the fourth lesson so the fourth lesson is very important I think and it is that what the what the Hank literature tells us is that central banks are kind of new data requirements so granular micro data are essential to the mission of the ECB they're really essential in the any central bank so the House of Finance and Consumption Survey is a great first step is an awesome really data set but it's small and ideally in the larger sample size to explore all the dimensions of heterogeneity that Philip mentioned at the beginning and try to understand which ones really matter in your section and you really need the longitudinal dimension and you also need better coverage of the top of the distribution and these surveys have a hard time reaching the the high net worth individuals so one can think of complementing the HFCS with additional data sources like administrative data, public or proprietary data for example bank transaction data and link them to social security number also I think what is needed is time or high frequency data you know I think that the COVID recession maybe very clear how important is to have like real time high frequency data to really monitor the evolution of the crisis in real time and try to understand you know who monetary and fiscal policy should target if possible you know concluding what is needed I think that the ECB has great data on banks balance sheet on first balance sheet but I think what it's lacking is on a rich data set providing a comprehensive household finance polls for the Europe so on. Okay thank you Jean-Lucas I mean the way I think having listened for the last half hour it's been a great introduction to the complexity the richness but I also think the advances that the academic world has made in addressing this topic of how to take a kind of heterogeneity along these many dimensions in understanding the interaction of monetary policy and employment in passing I'm not going to develop this topic but as a side comment someone somewhere should deeply think about whether in fact the difference in mandates across central banks really is fundamental to this question but now in large I think the divine coincidence element where typically under a lot of configurations what is needed to bring inflation from below towards 2% typically is also pro-employment maybe goes a long way and also in terms of with the correct measure of slack taking into account the different search intensities the different participation rates of different groups is a lot more similarity than dissimilarity now I again encourage the audience to raise your virtual hands to join in the discussion while that is taking place let me come back to the panel members and maybe again because we went through a lot there there's a lot covered but it might be helpful to look for some bottom lines but it's a bit reductive here but it's helpful maybe sometimes to look for bottom line so my bottom line for each of you is really a two-parter one is directional let's say I'm traditional, I'm conservative I appreciate this work but I don't want to learn too much how far wrong am I going to go by taking a representative agent approach where will I get the direction wrong will I be loosening policy when I should be tightening or vice versa so do this approach ever come up with scenarios which tell you not only are you calibrating wrong but you're actually moving in the wrong direction so that's one basic reductive question a second reductive question is Jean-Lucas mentioned some amplification forces but is everything an amplification force or are the other forces move as a kind of dampening force so is there any kind of universal principle that allowing for heterogeneity amplifies the effectiveness of monetary policy or diminishes it or I'm going to guess maybe the answer is going to be it all depends but with that whether again with each panel member maybe take one or two minutes either to take that on or maybe if you kind of felt you didn't have time to make all the points you wanted to make you can come back to some of the your interventions that you would like to get across so over to you Juan but if you can wait for the microphone to catch up with you sorry, yeah so regarding your first question about the implications of ignoring heterogeneity I think this is going to be very costly because of course the propagation mechanism is different in the in the representative agent model interest rates are going to change equilibrium interest rates are going to go up following expansionary monetary policy monetary shop whereas in the in the hand model is going to go down because all this middle class who is worried about precautionary saving will save more in a recession sorry so the implications change and I think ignoring them it's problematic with respect to the amplification versus dampening effects my impression is that there are forces that pushed towards the dampening effect and forces that pushed over the amplification effects and identifying those sources it's absolutely crucial in the presentation we several illustrations were pointed out so the nature of what sort of capital how does it combine with the force skill labour skills etc etc also I think in Antonella's presentation there was this issue of how to count the such intensity which could be a dampening if we do it correctly it could be a dampening effect rather than an amplification effect like for example when you're unemployed which is counted as an employment but of course they don't accept any down-ward pressure on wages so I don't have a single response to both of your queries but my point is that we should not disregard our models, they are rich and they bring new issues on to the table and that the amplification forces are probably more important than dampening forces. Thank you. Over to you, Antonella. So, again, if you can wait for the microphone. Thanks. So, my take on your two questions is that, yes, the chance of moving in the wrong direction is definitely there given the difference in terms of the mechanism. And the answer to the second question, you know, whether, you know, amplification relative to dampening, it's also, yeah, it depends. It depends on which aspect are we focusing on. And so what I want to to argue is, you know, I want to sort of raise the following question, which is related to this question, Philip, that you raise about, you know, what we learn and how this model compared to representative agent model. And the point is whether, you know, these models are ready for, you know, to be used for policy analysis, you know, to be incorporated into the toolkit of central banks. And there I, you know, and I would like to also note the opinion of the other panelists. I think that, you know, one issue is whether we should go for this, you know, fully fledged, continuous agent, you can, as your mother, rather, you know, going step and develop more tractable version or do this in tandem so that we can, we can learn from the more tractable version, which, which, you know, aspect, despite being, you know, realistic, they may not matter too much for the transmission of shocks and for the transmission of policy. So, so I think that's one important, you know, if I, if I look at, you know, at the, at the slide that Januka presented to us clarifying all the possible transmission mechanism and, you know, additional transmission mechanism associated to inequality in, in the presentation of, of one, I, I, you know, my, my impression is first that no model, I guess, is going to incorporate for the same time, all these, you know, mechanics that the, the, the interplay of those different mechanics going to depend very much on, you know, the calibration underline assumption about distribution of incomes. And so the point is, you know, do we know enough or do we have a, can we use, are these models ready for, to be used, or rather, we should go in step and use, you know, develop other models that are more tractable to get intuition to, but also to, to, to spread the knowledge to be used for teaching. So I think, which is something that it's, it's extremely important. And, and, you know, the last thing in this direction, I would like to say is that I, I suspect, you know, there has been so much progress in terms of, of solving this model, you know, in American methodologies to solving these models, even though I think the literature hasn't settled on one yet, but I suspect that very soon there will be some, you know, available code, like it happens for the Bayesian, you know, estimated DSG models where you can, you know, put into the code a hand model and get, you know, your, your, so are we ready for that or should we first, you know, first develop some knowledge? Yeah. Thank you, Jean-Lucas. Yes. I think Juan and Antonella were very exhaustive, so I'm going to be sort of brief. Let me start from your second question, Philippe, the question about amplification versus dampening, I think, you know, to see whether there would be a possibility of dampening this model, the easiest example is the following. Okay. So think about like a cut in the interest rate and think about like two major effects that these are on the macroeconomy. The first one is that we're kind of expanding aggregate demand and employment wages, reducing employment and therefore kind of affecting the lower end of the income distribution, which is the, which is sort of the distribution, the partial distribution with households that have a higher marginal price to consume. And that type of distribution goes towards amplification, but a reduction in policy rates also leads to an increase in asset prices and that's then to increase the wealth or, you know, use capital gains mostly for those at the top distribution, which are individuals with low marginal price to consume, very low marginal price to consume, much lower than the hand-to-mouth households. And that would go in the direction of dampening the shock. So try to understand whether, you know, conventional, especially unconventional monetary policy mostly affects one tail or the upper tail or the lower tail of distribution is key to understand whether at any date there would be amplification or dampening. To your first question, I agree with Antonella that, you know, the step from representative agent to like a two agent model. So a model where you have basically some hand-to-mouth, high marginal consumption households, and then the rest of the model is essentially a representative agent model. So models that have been used in central banks for for decades is a small step, even computationally, also conceptually is a small step. But that is certainly a step that is not costly and that captures many of the forces that are in more general heterogeneous agent models. What these models, there are a number of things that these models don't capture. Okay. So two important elements, maybe one is that in the data, the share of households that are hand-to-mouth, so they're constrained, they're very low liquidity, it's very cyclical. It's very cyclical. It's counter cyclical. It goes up in recessions. So all the effects associated with that group of households is sort of endogenous to the evolution of the cycle, which is something that you don't have in two agent models. The other, which is probably the most important, is that these models don't have a precautionary saving mechanism. Or if you want to be really precise, they would have it, if you solve the model globally with respect to aggregate fluctuation, aggregate shocks, but they don't have a precautionary saving motive with respect to idiosyncratic risk, which is the main source of precautionary saving. And that, I think, is really an important economic force that must be taken into consideration when thinking about aggregate fluctuations and the effects of monetary policy. So there is, yes, something that we lose. Whether the policy prescription in terms of directions would be wrong, that I don't know. And honestly, I would be surprised if a two-agent model would get it completely wrong. But fundamentally, I think that central banks should use an array of models. They should keep using representative agent and two agent models. But also, they should add to their toolkit these class of models because you can always learn. And I think that what also is important is to kind of converge between academia and institutions on a language, on a common language. And using these sort of models in central banks and institutions would help us converge on a common language and that would, I think, would be like a ground for very, very prolific interaction between institutions and academia. Very good. Thank you, Sean, Luca. Thank you, all of you. And now I'm going to turn to a question from the audience. I'm very pleased that Kristen Forbes has volunteered to make a comment or ask a question. So Kristen, over to you. Thanks very much, Phillip. So these recent presentations provided a really nice overview of the different channels by which monetary policy could affect inequality and employment. But I was wondering if the panelists could discuss whether the different forms of monetary policy had different effects in their framework. A couple of you hinted that they might, for example, Jean Luca, you just talked about conventional monetary policy to differentiate the effects from unconventional. But more specifically, I was hoping I could try to nail you down on whether the frameworks we use could give us information on whether, if say, a central bank wanted to provide a certain amount of stimulus is measured by the impact on inflation. Would providing that stimulus do lower rates have more or less impact on inequality than providing the same stimulus through QE? Or particularly relevant for today, if a central bank decides it needs to remove a certain amount of stimulus because of concerns about price stability, would removing that stimulus through shrinking the balance sheets have a different effect on inequality and unemployment than removing a comparable amount of stimulus by tightening to raising interest rates? Thank you, Kristen. So I guess, Annie or all of you could take that, but let me just see if there's Juan or Antonella or Jean Luca. Okay, Juan. So, Juan, if you go ahead. I think the implications are going to be the implications will be different. For instance, in terms of access to financial markets and so on, the effect of quantitative easing and conventional monetary policy is going to be quite different from conventional monetary policy. And therefore, in your language, the dampening effect of quantitative easing, of course, eventually corrected when recovery is underway, should be much stronger than the amplification effects of standard monetary policy. That would be my short reply. Thank you. So let me just cross check. Peter, Antonella or Jean Luca wants to add? Just very quickly. So, yes, definitely, there's been a debate on whether unconventional monetary policy has contributed to raising inequality through the effects it has on asset prices. And I think that, I mean, I will let Jean Luca talk about that because actually that was the last point of his life, you know, that one way to care about inequality is precisely by, you know, which while keeping, you know, the goals limited is to choose tools, different tools in taking into account that they do have in certain dimension the same effects. In a quality, I would say labor markets, but in other dimension, they have different effects asset prices. Okay. Thank you. Jean Luca, do you want to add any, please? Yeah, I know just very quickly. I was going to say precisely what Antonella said. So let me just repeat it. So I think that, you know, within technically, just speaking technically within the technical framework of this class of models, I think it is possible to, say, design a certain amount of quantitative easing, so say purchases of certain less liquid assets, say, or even stocks or equity, that has exactly the same effect on, say, aggregate GDP, aggregate income or aggregate employment, then a certain, say, you know, a 25 basis point cuts in the interest rate, that is possible. What is going to be different is precisely what are the implications across the distribution. So who gains and loses? That those, the winners and losers are going to be different across these two, these two policies. And again, I, you know, I do think that not as a kind of a main objective, maybe as a, you know, as a secondary sort of precept, if you like, of the conductor monetary policy, you know, one could think that, you know, price stability could try to be achieved in a way that is the most equitable possible with respect to, you know, the distribution of income and wealth. So choosing the instrument that achieves the objective with the limited, most limited impact on, say, inequality, that could be, you know, a compromise if, you know, such a compromise is possible. Very good. So I'd like to thank all of the panelists. The way I think about this session is it's very much a progress report. I've no doubt that central banks, especially here at ECB, will indeed be making more use of these models. And as the academic literature progresses, so will we. I just want to make one point about big data. Actually, just before the pandemic, we launched a consumer expectation survey. It doesn't deliver everything John Nuka would like, but we've been learning a lot. And actually it passes through to policymaking pretty quickly every month now when we find out that the attitude of consumers to the income windfalls they've received, to the fear of unemployment differs so much across different groups in a systematic way. So that very basic dimension of understanding consumer behavior is now greatly enriched at the ECB through this pan-European consumer expectation survey that our staff now are leading. So with that, let me turn back over to Claire. Thank you, Mr. Lane, for some excellent moderation. And thank you to all of the speakers for their contributions too. I think it will have been very well received by the participants. Participants, you should have just received a link to a very short survey. We'd really value your feedback. So if you could take a few minutes to fill it in now, it would be much appreciated. We're going to take a break. We'll be back at 1745 Central European Time when we'll be joined by four of the world's leading central bankers. See you then. Bye. Climate change is one of the biggest economic risks confronting Europe this century and the rest of the world for that matter. In responding to the crisis, the pandemic has fueled. Now, hopefully, we can begin to think about what the economy might look like after. Bonjour. I'm Mel Valle. I'm 24 years old and I'm from France. Salut. My name is Adrian Eifrin. I'm 31 years old and I am from Romania. Guten Tag. My name is Ilya Kantorovich. I'm 29 years old and I am from Germany. I'm interested in bank risk and my work explores the design of bank regulation to build a more financially stable world and to prevent future financial crisis. My research interests are in macroeconomics with a focus on energy, climate change and inequality. I'm interested in the differences countries exhibit in the share of non-performing loans and the varying size and duration of the increase in non-performing loans during recessions. My current research focuses on the corporate bond market. I study how investment heterogeneity impacts market liquidity, bond pricing and ultimately firms' financing conditions. In my work, we show that in the eurozone, bond finance firms react more strongly to monetary policy action than bank finance firms. My paper explores fire sale risk in contagion originating from collateralized loan obligation or CLO managerial contracts. In this project, I study how to accurately measure monetary policy from the European Central Bank considering the relevance of government bond yield spreads in the eurozone.