 Dobro, povedaj do vseh vseh štih poslednjih sešnjih. V seštih za primer, v seštih konferencij, kaj smo imali, nekaj ekonomiji sešnji, vse smo se vse nekaj prav, da smo vsešnji sešnji v programi, ker smo se načinati, da smo sešnji, da pozvomno, da se spasjo očečne od exporta iz svih ekonomistov, je zelo izgleda, da je kajovosti izgleda, izgleda je prijevno, kaj je vsevbi izgleda, zelo izgleda, ne ne ne ne ne ne ne ne ne ne ne, And then we said, why don't we, together with the usual program that as you have seen, as we have all seen during the last two days, as many established economies, we also had the idea to have for the first time a young economy session. We called for paper. We received around 100 applications, In so many application. We had select only three and I'm happy that the three young economists of this year are on stage with me. The first is Nastasia Antonova from Exmercell School of Economics. Stefano Picca from Banccov Itali and Fabian Seidrich from Berlin School of Economics. Zato, da je vsega vsega vsega, da je vsega vsega, da je prisega vsega. Vsi je tudi tudi 30 min. in 20 min. Presentacije in 10 min. za vsega vsega vsega, da je vsega vsega. Zato, da sem počutila, da sem počutila Anastasia. Moj slijdič nekaj? Dobro. Zdaj sem. Zdaj je tukaj, da je pripravljena pripravljena in pripravljena na produkciju ekonomi. Zdaj, če pripravljena je pripravljena na pripravljena in pripravljena na pripravljena v standarstvih nekajske vjeletnih pripravljena so že vedimo več nas begipnoke terap外 incega termu apellstx, espetu so termu in se spTVcome. kinak корatخرimo posjeli s disposition kraj tako tip ni s poznikom sperm, a z pačen grams to kooj taj sprevers berej Motion iz brkih spets. zato je zelo se načo možne energije. Zelo, da počkaj smo izvene sektor o nukengijanju model, počkaj je stajnje taj zelo te 3 objekta. Zato je to sektor spesifiku izvene, zato je načal sezavstav o ekonomi in je to sektor spesifiku izvečenča. in v zelo v projegu zelo sektor spesivnik prizajsa regiditi. Zelo, da je tudi moždi prizajsih modelov, prizajsih modelov. Prvno, so to so, pa ničnji vzelo vzelo vzelo, in bodo da je tudi prizajsih modelov, in prizajsih modelov, prizajsih vzelo je vzelo. In potem je ina prizajsina, framework, če je stavljene prišličnje modelov, in najbolj izglednje izglednje je menjukost modelov, in na stavljene prišličnje prišličnje režiditi moždje izglednjenje, kako je na zelo šokov, kako izglednje ekonomi. If we take the existing literature that looks into a cost-push effect in multisectual-nukengian models, we will see that this literature largely relies on non-state dependent pricing. But at the same time, state dependent pricing, first, it is intuitively, it is a more intuitive framework, and second, it is confirmed empirically by numerous studies. So in this project, what I do, I look at how state dependent pricing shapes cost-push effect in a multisectual-nukengian economy with a production network. So to do so, I use multisectual-nukengian model with input-output network, and the distinctive feature of my model is that price rigidity is state dependent. And I use this model to do a three-step analysis. First, I look into the theoretical role of state dependence, then I use the model to empirically estimate the degree of state dependence for each sector of the U.S. economy, and finally, I look at the quantitative role of this state dependence in shaping cost-push effect in the U.S. So the main results are as follows. First, state dependent pricing is important theoretically. It may lead to amplification or deamplification of cost-push effect compared to a non-state dependent pricing model, but moreover, it can also lead to a sign reversal of cost-push effect. Then empirically, I find that the majority of sectors in the U.S. economy have some statistically significant degree of state dependence in their pricing. And finally, I show that state dependence is quantitatively important for the U.S. and that the implications of state dependence are different during different type of crisis. So let me skip the literature. So to start with, I will give you some idea of how I introduce state dependence into this New Keynesian production network model while at the same time keeping the analytical tractability of the model. So to model state dependent pricing, we need to ask a question what is a good state variable in this economy, given that we have many sectors. And to define a good state variable, what I do, I look into the expression for the vector of sector-specific marginal cost in the model, and this sector-specific marginal cost, they are a function of sector-specific productivities and sector-specific markups. So productivities in the model, these are total factor productivities in each sector and they are assumed to be exogenous. So given this expression for marginal cost, I define the relevant productivity state in each sector as a combination of sector-specific productivities that affect the marginal cost in this sector either directly if it's the own productivity or indirectly through the input-output network. And then, given this definition of the relevant state variable for each sector, what I do, I model a state dependent price rigidity by combining a famous sticky information framework with heterogeneous inattention framework. More specifically, I assume that firms in given sector track fluctuations in their sector-relevant productivity state. And then, I also assume that firms have heterogeneous degree of inattention distributed across firms within a sector according to some sector-specific distribution. And only those firms who have low enough inattention compared to the current size of the relevant state fluctuation, only those firms update their information in any given period. And finally, those who update their information, they receive the full information about the economy. So, what this framework gives me is that sectoral price flexibility is equal to the share of firms who update their information as you would have in sticky information framework. And in turn, this share depends on the change in a sector-relevant productivity state. So, next, since I want to study the cost-push effect, I will define formally what is the cost-push effect. So, to do so, I derive the Phillips curve for consumer price inflation in the model. And as a normal Phillips curve, it has a demand component, expectations, term, and cost-push effect. So, I'm interested in the cost-push term of this Phillips curve. So, my cost-push effect here is expressed in terms of a vector of sectoral price gaps. So, sectoral price gaps are just the differences between the efficient prices in each sector and the true prices that were present in each sector in the previous period. So, intuitively, these sectoral price gaps capture the desirable degree of price adjustment in each sector. And besides the sectoral price gaps, cost-push effect also depends on the details of production network and on sector-specific price flexibilities, which are now state dependent. So, in this initial form, the cost-push effect is very difficult to analyze because of the presence of production network. So, to facilitate the theoretical analysis, what I do, I decompose the cost-push effect into the two components, the main component and the input-output component. So, the idea behind this decomposition relies on the notion of the reset prices. So, reset prices are prices chosen by those firms who decide to reset their price in a given period. And the main component of the cost-push effect is the cost-push effect that you would have if all those who reset their prices would choose their new prices to be at their efficient level. And the input-output component captures the effect of propagation of nominal rigidity through the production network and this propagation of nominal rigidity leads to a real rigidity because now marginal costs are not their efficient level. So, the reset prices are different from the efficient prices. So, main effect or main component is very useful for us for two reasons. First, it gives a very clear intuition about the first order effect of state-dependent pricing on cost-push inflation. And second, I show in the paper that this main effect is quantitatively much more important, at least in the US, than the input-output effect. So, to understand why state dependence matters for cost-push inflation, let me give you a following three-sector example. So, imagine that we have a three-sector economy and in each sector we have a relative price gap. So, sector one and sector two want to decrease their prices and sector three wants to increase its price. And by construction, because these price gaps are relative, their sum is zero. Now, consider the case of non-state-dependent pricing. So, what does non-state-dependent pricing mean? It means that in each sector price flexibility is assigned in advance and it does not change. So, imagine that we have sector one with fully flexible prices and sector two and sector three with fully rigid prices. In this case, what happens is that prices in sector one adjust while prices in sector two and three do not adjust. And if we compute the main component of the cost-push effect, we will see that this main component is negative and this negativity is driven by the downward adjustment in sector one. Now, consider what happens if we have state-dependent pricing framework instead. So, under state-dependent pricing is the case that price flexibility adjusts to the size of the shock. In my example here, sector three is hit by the largest size of the shock which is reflected in its largest size of the price gap. This means that prices in sector three should become flexible. And in this case, what we would have is that prices in sector one and sector two will remain rigid and prices in sector three become flexible. And in terms of the main component of the cost-push effect, what we would have is the positive cost-push inflation driven by the upward adjustment in sector three. So, as you can see, depending on the pricing framework, we can have different sign of the cost-push inflation. So, let me give you another example of the importance of pricing framework. So, this example is about the effect of commodity shocks. So, imagine that we have an economy where we have two commodities, oil and grain, and we have two final goods. One is oil intensive and another is grain intensive. And also imagine that we have a commodity shock. It is either oil shock or a grain shock. And this economy has a very simple expression of the main component of cost-push effect given each of the shocks. So, of course, this component depends on the shock itself, but also it depends on the difference between price flexibilities in the two final goods sectors. So, now what happens if we have non-state dependent pricing? And the non-state dependent pricing, price flexibility in one of the sectors is always larger than the price flexibility in another sector. And without the loss of generality, we assume that prices are more flexible in the oil intensive sector. And what happens in this case is that if we have a negative oil shock, this will lead to a positive cost-push effect. And if we have a negative grain shock, this will lead to a negative cost-push effect. And this is very counter-intuitive because when we think about the negative commodity shock, we expect that it will lead to a cost-push inflation and not to a cost-push deflation. However, state-dependent pricing model actually remedies this problem. So, under state-dependent pricing, if we have oil shock, oil intensive industry becomes more flexible and if we have grain shock, the grain intensive industry becomes more flexible. And in both cases, both negative oil shock and negative grain shock would lead to a positive cost-push effect in line with our basic intuition about the effect of commodity shock. So, next, I'm interested in whether this state-dependent pricing exists empirically in the U.S. economy at a sectoral level. So, to estimate empirically the state dependence, I impose a functional form on my sectoral price flexibility. And so the sectoral price flexibility is a sum of the average price flexibility and state-dependent component. So, the average price flexibility, think of it as a kalvo parameter. It can be different across sectors, but it doesn't change over time. And the state-dependent component captures how price flexibility adjusts to the changes in the size of the relevant productivity state fluctuation. And then I... So, to estimate the average price flexibility and state-dependent parameter in each sector, what I do, I take the model and I derive the contemporaneous response of sectoral markups to sectoral productivity shocks in the model. And this response, it contains exactly the information about price flexibility. But the problem is that I do not absorb neither sectoral markups, nor sectoral productivity shocks at good enough desegregation at a good enough frequency to do my estimation. So, what I do, I construct the model implied sectoral markups and sectoral productivity shocks from the data that I have. And this is the data on sectoral prices and wages that I absorb in monthly frequency. And I use the model links between prices and wages and markups and productivities. And this is how I estimate my parameters for average price flexibility and state dependence. So, these are the estimates I get. So, these are the distribution of these two parameters across sectors. So, on the left, this is the distribution of average price flexibility. As you can see, it is quite heterogeneous. And here one thing to notice is that those sectors related to commodities, they have higher price flexibility on average, which is in line with what the previous authors have found. But then, on the right, we have the distribution of state dependence parameters across sectors. So, here what I do, I push all the insignificant estimates to zero. And this gives us 70% of sectors. And sectors here are weighted by consumption share. So, 70% of sectors have some statistically significant degree of state dependence. So, next to understand my estimates a little bit better, I look how they correlate with the volatility of my relevant productivity state that I defined before. So, average price flexibility correlate positively with this relevant productivity state volatility, meaning that the more volatile conditions are in a sector, the more flexible are prices in the sector on average. But the state dependence parameter correlates negatively with the state volatility, meaning that the less volatile are the conditions in a sector, the more state dependent pricing is in this sector. So, finally, given that I've estimated this price flexibility framework in each sector, what I do, I compute model implied monthly cost push effect for the US, which is the blue line. I also compute a counterfactual cost push effect where I switch off the state dependent component of price flexibility and this is the gray line. And on this plot there are two historical periods of interest. So, first period is the period after the great recession. So, in 2009 you can see that both state dependent model and non-state dependent model predict quite high positive cost push effect and state dependent pricing here just played an amplification role basically. But if we look at the more recent period starting from COVID crisis and other events that followed, we can see that here often it is the case that state dependent pricing reverses the sign of the cost push effect. So, as you can see depending on the type of the crisis we can have different implications for state dependent for how state dependent pricing shapes cost push inflation. And final remark I want to make about this graph is about the actual recent inflation in the US. So, as you know in 2021-2022 there was an unprecedented observed inflation in the US and the observed monthly inflation is plotted as a gray line here. However, as you can see neither state dependent nor non-state dependent pricing models do not attribute this inflation to cost push effect. So, state dependent pricing model produces transitory cost push effect transitory positive cost push effect around the supply chain crisis problem but then this cost push effect fades away. This means that this recent persistent episode of inflation may be attributed more to the demand inflation or to expectations inflation than to the cost push inflation. So, in conclusion so it is very important to understand how cost push inflation is shaped and in this paper I show that pricing framework is really important when we are trying to compute aggregate cost push effect in a multisectual models and it's important not only theoretically but also quantitatively at least for the US. So, thanks a lot, yeah. Thank you. Maybe we can directly open the floor for questions. Any questions? Wolfgang? Yeah, Wolfgang Lemke, one of the organizers. Just one quick question. So, if you would run on the same sample standard model plain vanilla and identify cost push shocks versus demand shocks how would that differ from your sophisticated model where you would consider state dependent pricing? Or put differently if your world would be the DGP the data generating process how would standard simple models perform in that type of environment? Are there questions? Thank you. Very interesting paper. I was wondering on this cross sectional differences in the state dependence across sectors how much that is actually a manifestation of the fact that in the end the pricing at the sectoral level we still have an aspect of dyadic dependence not between one sector and the other just as these sectors are operating in a production network so I would expect in the flexibility of the prices in the given sector to actually reflect the position of that sector inside the network. So I was wondering how much of the differences you actually uncover are just it's not intimately a higher flexibility of the given sector but just of the position of the sector in the network. Thanks. Giacomo? Just a quick one. I was wondering whether there is also some implication for somehow the effectiveness of monetary policy in this context is also state dependence. I mean if I step back and I look at the standard Nucanism Philips curve the frequency of price adjustment has something to do with the slope of the Philips curve meaning that if it's steeper in stabilizing inflation do we have something like along this line? David, first though. Thank you. This is a very interesting paper, congratulations. The question that I have is can you use some information about the frequency of price changes to identify the difference between productivity and markups component it's unclear, you didn't have time to explain but how do you actually identify what's changing markup and what is the contribution of productivity using the data that you have because my understanding is that you're using only data on prices, right? And wages, yes, thanks. Okay. I'll go back to you. Yes, first your question about the simpler model, right? So to my view there are like two types of simplicity one model would be like one sector New Keynesian model and the other type of simplicity would be just a network model but with colorful parameters. So that would be a good exercise to compare to compare that model with what my model gives I didn't do that but so I tried to model the price rigidity so that it has like two components and one of them can be interpreted as Calvo so when I shut down the state dependent component of price rigidity I interpret this model as what the Calvo model with heterogeneous degrees of price rigidity across sectors would give me so so yeah, if this answers your question so yeah, then about your question about cross-sectional differences and what actually do I capture with these different price flexibilities yeah, that's a very good question so what I think is that so even the average price flexibility it reflects the position on the network and how exposed you are to volatility in different sectors to which you are connected so I allowed for ex-anti-heterogeneity in price flexibilities just so that my model is comparable to the Calvo model with heterogeneous price flexibilities the more advanced or proper way to do that would be to have a full-fledged cost model where everybody has the same pricing framework and then exposed because they face different they are different in the network and they face different volatility exposed they will have different price flexibilities but so my design is just to compare with the standard Calvo model I guess if this answers the question about the effectiveness of monetary policy yeah, this is a very good question so I thought because price flexibilities are time varying in my model and slope of the Philips curve depends on these price flexibilities in my model indeed the slope of the Philips curve is time varying as well but so and I saw that it might have some implications for effectiveness of monetary policy but empirically when I compute this slope it's almost constant so I don't find any quantitatively important results on the slope even though this is theoretical possibility and yes about the identification of how do I disentangle between sexual productivities and sexual markups so I have the data on sexual prices and sexual wages so labor in the model is sector specific and so in each sector wage is different and then in the model the expression that they have is that wages they are a function of some aggregates and then of a sector specific markups so basically and prices are a function of productivities and markups so because I have both prices and wages and in one so in the relationship between wages and markups there are no productivities and between prices and productivities and markups there are productivities that allows me to disentangle this yeah so it's would be easier to show with equations yeah but sectoral wages are a key here yeah thanks a lot for equations thanks a lot again and now we move to Stefano please Stefano alright alright well thank you so much for having me into the program great conference today I'm gonna be talking about housing markets and monetary policy in the euro area and the usual disclaimer applies so the question I ask in the paper is well how do housing and market affect the monetary transmission mechanism in the euro area now we know that monetary policy is heterogeneous effects across countries in the euro area what do we mean by that just some countries respond more strongly than others in terms of aggregate consumption for example a bunch of other variables now what I do in this paper is I investigate the role of some housing and mortgage market characteristics into this heterogeneous transmission mechanism of monetary policy and in particular I'm gonna be focusing on adjustable rate mortgage shares ARM shares and home ownership rates HOR now since there are non-euro area people in here usually click on the hyperlinks but will be diffused here so there are big heterogeneities of adjustable mortgage shares and home ownership rates across countries in terms of adjustable mortgage shares some countries go from 10% relative to fixed rate all the way up to 90% if we think about Spain and the same for home ownership rates there are huge differences across countries so in this paper I'm gonna be essentially leveraging and empirically I shown that there are strong correlations between the local impact of monetary policy so how much is monetary policy effective in each given country and the adjustable mortgage shares and home ownership rates in those specific countries and then I'm gonna set up a new Keynesian model I'm gonna be showing you no equation today but just you know some intuitions of the model of a currency union to think about the euro area featuring long term mortgages and home ownership rates and the idea here is that through the lenses of the structural model I'm gonna be able to separately quantify the contribution of adjustable mortgage shares and home ownership rates to aggregate consumption responses to monetary policy and then I also the model is gonna be able to give me the possibility to do some counterfactual so I'm gonna be asking well what's the effect on the monetary transmission mechanism of a unified mortgage market and I'm gonna be specific what I mean by that in the paper I also look at the consequences of introducing house prices into the euro area price index it's something that was discussed in the strategy review but I'm not gonna be able to do this on time today ok, so let me just so we're all on board just I'm gonna be giving you some preview findings in the paper I show the countries that have stronger empirical responses in terms of consumption, price to rent ratios, the volume of mortgage issuance and mortgage interest rates are actually those countries that have higher adjustable mortgage shares and higher home ownership rates however I also show that empirically you see that it just so happens that countries that have higher adjustable mortgage shares and higher home ownership rates are actually on average the same countries so this sort of identification problem that's why I'm gonna be moving to the structural model and I'm gonna be having two countries there and they are calibrated to Spain and the euro area sort of a it's on one side of the spectrum in terms of these characteristics of the housing market lot of adjustable mortgage shares and higher mortgage share rate and the euro area which is sort of the average so I'll be the model to the key institutions and I show that the consumption in Spain actually increases more than twice as much as in the euro area and this is in line with the data then I show in the paper how adjustable mortgage shares and home ownership rates actually interact to amplify the effects of monetary policy there's countries with higher adjustable mortgage shares where that means that there's more pass through from the policy rate to the mortgage interest rates and that's a pure cash flow effect on borrowers essentially on mortgage homeowners but then there's also higher home ownership rates meaning on average higher fraction of mortgage homeowners in any given country that's more of a level effect through as effect of many more people so that's sort of the mechanism why we see what we see and then I have this counter fraction on the euro area why mortgage market decreases the heterogeneous effects of monetary policy if countries have a more similar fraction of contracts that are adjustable relative to fixed rate and that goes through the pass through of the policy and I'm not going to be talking about this part on house prices in the euro area price index ok, I'm going to skip the literature review in the interest of time but today in this very short presentation I'm just going to be giving you some empirical motivation some impulse response functions monetary policy shocks and then I'm just going to be highlighting some insights from the currency union eukensia model give you some results and then I'm going to conclude so what's the empirical specification I have in the paper right now is a georda 2005 standard local projection specification for 11 euro area countries and for each given country I'm going to be running the local projection where on the left hand side you have the variable y which is say variable of interest say aggregate consumption or average mortgage interest rates or newly issued mortgages or house prices then you have the epsilon mp that's the monetary policy shock I'm going to be using some proxy surprises from the database on altaville et al here and you know these monetary policy surprises I'm going to be using the two-year overnight interest whoops around policy announces so those are from my frequency identification and obviously the quantity of interest here is the beta hc so that's going to tell you essentially the impulse response for any given horizon for any given country versus the percentage change of the variable y to one standard deviation expansionary monetary policy shock then there are some controls and what I'm going to be doing here is looking at this impulse response function looking at the peaks of the troughs and that's how I define the effectiveness of monetary policy and then I'm going to be correlating those with adjustable mortgage shares and homeownership rate so this is just one of the results I show you today the results on mortgage interest rates so you have the y axis the response of the average mortgage interest rate to the monetary policy shock across the different counties and you see here that to the same monetary policy shock some counties just respond much more than others some counties like Germany or Netherlands barely respond while some others like Italy or Ireland so on just respond much more so essentially what I'm going to be doing next is looking at in this specific case the troughs so the minimum points of these impulse response functions those are going to help me define the effectiveness of monetary policy in terms of the average mortgage interest rate correlating those with housing and mortgage market characteristics and so here is the here I'm taking the troughs from the previous graphs I'm going to be plotting them on the y axis and then again on the x axis different characteristics of the housing and mortgage markets so here for example in terms of the share of adjustable remorgages the very first graph you see here that on average counties that are more on the right of the graph so counties have higher share essentially as those countries that respond the most in terms of average mortgage interest rate to monetary policy shock and this is going to be true also for different measure of monaship rates the fraction of outright homeowners and fraction of homeowners with mortgages that's not only true for the mortgage interest rate even if we look at the volume of newly issued mortgages then again you see that this is an expansionary monetary policy shock so these quantities are going up instead and so you see that the counties that respond the most are again those of the higher levels of this housing and mortgage market characteristics and this is going to be true for price-to-rent ratios and aggregate consumption so I told you I showed you that there are some correlations between monetary policy responses and the characteristic of the housing markets that I am interested in but again there is this sort of identification problem I discussed a little bit in the intro which is you know some countries so the countries that have higher adjustable mortgage shares if you look at the first graph for example are actually those that have higher homeownership rates as well so if you think about Spain there is 90% adjustable mortgage shares as of 2015 and you know it also has a very high homeownership rate of around 80% so there are these correlations across countries and that's why I'm going to be switching to a structural model to kind of tease out what are the where are the effects that we looked at in the correlation with monetary policy coming from, coming from adjustable mortgage shares or homeownership rates and what's the intuition behind so I'm going to be very brief on the model because it's a very big model with a lot of equations but I'm going to be just giving you the structure and some intuition so this is a currency union New Keynesian model and there's going to be rich household balance sheets ok so there's going to be two countries home and foreign so this is the style of this international finance literature and I follow Fireman Acelli in 2008 in most of it so there are these two countries home and foreign one is small open economy relative to the other so in this case I will calibrate them to Spain and the euro area and then on top of the structure international finance type of structure I'm going to be enrich the model with some details of the housing and mortgage markets a la Greenville 2018 essentially so there's going to be there's going to be homeowners and outright homeowners and mortgage homeowners and renters and the idea here is that then I can move around the structure of the housing market and I can compare the responses to a common monetary policy shock when moving around these characteristics of the housing market and the first characteristic that I add on top of this structure is adjustable rate mortgage shares and fixed rate mortgages and this I, you know, take a stand here these are essentially exogenous in my mind so there's a parameter governing how much a country there are in a country relative to the fixed rate there are some examples that show actually how in some countries actually these were like institutions behind these characteristics and not necessarily endogenous and there's an example for Spain there and the second characteristic I include because I'm interested in the homeownership rate is I introduce these endogenous homeownership rate homeownership rate does not need to be endogenous actually in this model it can be exogenous but I have it as endogenous so that we can draw a consequence it is also in the movement over the business cycle for homeownership rates and the idea is that essentially some countries on average borrowers are happier than other to live in a house and so there are these shocks each period which aggregate out so that you have endogenous homeownership rate the differences in ownership utility across countries basically are reflective of a bunch of institutions which are no model explicitly such as renter market quality subsidies and renter market quality for instance Spain is not great and subsidies there are a lot in Germany and so on so those are some reduced form for all those things ok let's go to one of the main results which is essentially following a monetary policy shock in the model Spain is essentially more responsive relative to the euro area and so this is like a very persistent 1% fall in the nominal rate so together with the nominal rate obviously the adjustable mortgage rate is essentially the same in the model fall and that's a movement in the interest rate in the euro area but you see Spain is an open economy so essentially it's going to inherit the same because we are in a currency unit the same interest rate and then you see that essentially following this interest rate shock then the average mortgage interest rate in Spain respond more this is also what I was showing empirically and this is really just due to the fact that there is a higher pass through from policy rate to the mortgage interest rate is mostly going to be the fact that in Spain in the paper I do the composition shows how much of this that we see here is due to adjustable mortgages relative to a monetary price I don't have time today but you see that this is true for many other variables as Spain respond more in terms of newly issued mortgages in terms of price to rent ratios and that's because there are many bigger movements in a monetary price so many more movements across people that change their tenure status bigger movements in the price house prices and rent and then there's a stronger response also to in terms of aggregate consumption this is merely based on these differences in the housing market that's where the different calibration across countries come from ok, so then now since I've set up this big model where I can think about differences in the housing and mortgage markets and transmission mechanism in terms of monetary policy what I do in one of the two counterfactual studies in the paper is thinking a little bit more of a euro area wide mortgage market there's a lot of discussion on the benefit of a European fiscal union very limited as of now but mortgage markets are really one of the few types of markets that are really different across countries ok, there's just institutional differences for example you will never get a mortgage in one account in Germany just to buy a house in Italy and your German bank will tell you look, go get it in Italy it's just very different and banks don't do cross issuance of mortgages so very different institutions and so what I imagine in this model is that in a euro area wide mortgage market at least this financial regulation becomes more similar and what I mean by that is that essentially the countries are gonna issue mortgage contracts that are in a more similar provision in adjustable and fixed rate and so I eliminate this big heterogeneity across countries in this case across Spain and the euro area in terms of the share of adjustable rate mortgages. In particular I consider two additional economy yes, when it's paying the adjustable rate mortgage share gets decreased from 90%, which is a very big fraction as of now to 70%, and then it gets shifted in the second exercise I shift the adjustable rate mortgage share of Spain to 47%, so essentially the same as the euro area level. So there are these intermediate cases where still the countries are very different in terms of the ownership rates, how many people own a house and how many people own a mortgage but at least now in this calibration this exercise they're gonna have very similar adjustable rate mortgage shares and this is going to be the result on top of the figures that I had before so essentially in red you still have Spain and in blue you still have the euro area but now there's going to be some intermediate cases and then you see that essentially with these intermediate cases with countries being more similar in terms of the fraction of adjustable rate mortgage shares then essentially the heterogeneity gets reduced across the euro area consumption gets reduced by 40%, the difference of consumption responses in Spain and the euro area gets reduced by 40% in the first exercise and by almost all of it in the second exercise of same adjustable rate mortgage shares across countries. There is a trade off however because yes we are reducing heterogeneity but of course there is also a distribution of resources in Spain there is a different distribution of marginal propensity to consumes because there are more mortgage homeowners constrained houses in this model and so during an expansion of the economy the loss of resources for these borrowers is very demanding in terms of borrow and welfare and it's here essentially the savers who are those that are issuing these contracts are benefitting from this expansion of the economy of reducing the interest rates so I was doing in the next version of the paper I'm going to switch I'm going to do increases in interest rate which is a bit more similar to what's happening nowadays. Ok so in conclusion I showed you some correlations between the local impact of monetary policy how much is monetary policy effective across different countries and some institutional aspects of those countries especially in terms of housing and mortgage markets. In a calibrated transparency union nukeincial model I showed that essentially given some characteristics difference in terms of housing and mortgage markets consumption responses in Spain is more than twice as much as in the euro area and this is similar to the data and then a euro area wide mortgage market when we think about having more similar contracts that are issued across countries in this specific example in Spain but this can be generalized to any country in the euro area and it is effective in reducing this heterogeneous transmission of monetary policy although there is going to be some tradeoff in terms of redistribution of resources and then in the paper if you are interested you can have a look at it I showed that including house prices into the euro area price index is going to lead a tradeoff between output which is going to be more stabilized but then a loss of stabilization in terms of goods and services inflation thank you so much for listening we can open the floor now for questions on the paper thank you Stefano interesting paper as well I was wondering whether you have tried to see also the state dependence of these adjustable rate mortgages given that we did observe that the shares changed abruptly depending on the stance of monetary policy and the duration also of the stimulus so I wonder how much of that plays into your story thank you Katrin thank you for the nice presentation so I was a little bit surprised to see that you choose to model Spain as a small open economy and large foreign economy so I see why you are doing that but I was wondering when you would say the small open economy is the Netherlands if you can also match the data in your impulse responses quick one by the way did you try also with QE shocks namely a shock that leave a short term rate and change and move long term I mean the intuition would be if your story is correct then you should not see a lot of these Spain reaction versus alternative hypothesis that Spain economy is reacting stronger to monetary policy of all sorts it's like a more responsive economy other question was very much interested in the all the facts that you showed as well and I'm wondering do you see any interaction between this home ownership shares over adjustable rates as you said the share of adjustable rates the higher the homeowner chip have we seen any changes in that since there was a move after 2015 especially in Spain but also in Italy to fix more fixed rates Thanks Jasper Just a very brief technical question when you run the regression say for your area and each member country do you sort of impose you're not imposing I guess the aggregate of the individual country sum to the your area aggregate results it would be good to see that if you aggregate up the individual countries impulses with the proper weights do they come close to match your aggregate your area results or is it possible that sometimes differ maybe you're not conditioning on the right stuff for each individual country not just interesting Thank you The floor is yours back Stefano Thank you so much for all the question try to go in order so Lorenzo here the it's so in terms of at least ownership rates I saw that looking at the series they are various across euro area countries different euro area countries ownership rates are pretty sticky there's a lot of momentum in ownership rates and they don't get measured in high frequency in my knowledge unfortunately so you don't really see well over the business cycle these movements but they're pretty sticky so I just used the snapshot in 2014 from the HFCS that's where I measure ownership rates and adjustable remorges shares the asset finance and consumption survey so they're not state dependent in my model essentially sorry adjustable remorges shares are not state dependent it's just a parameter so essentially you're saying what if you would endogenize this so that it would move over the business cycle I don't do that and you know I take a stand in the paper it would be a little bit more complicated I take a stand and this is due to institution they don't move much over the business cycle do adjustable remorges shares move more and you're right especially in Spain we've seen a reversal now there are many more fixed rate mortgages issue since 2015 even was on the slide as well but I could think about it more for sure as of now I don't do that for adjustable remorges shares so Catherine, in terms of the Spain as an open economy you're right we don't really think of Spain as an open economy there are two things here so first you know it's for the economy of the euro area in my model essentially I showed this in the calibration that there's a parameter that determines how much an economy is open to the other so essentially Spain is not affecting Spain or any other country that I would calibrate does not affect the euro area but the other way around is true and for what I'm thinking here I'm not thinking about big shocks or financial crisis in Spain or announcing bubble in Spain I'm just thinking about moving adjustable remorges shares over 20 points so I don't think that should have such a big effect to the euro area so thinking about Spain as an open economy in this context I think it would be fine so thank you for the question so I don't do QE in the paper so I'm just I have this as of now so I'm rewriting the paper and right now I have conventional policy so I again not doing QE but I mean you're right you should see different responses because you're changing the slope of the id code essentially and mortgages are about long rates although in Spain not exactly so because there's a lot of adjustable remorges there's still a lot of short rates so so yes it's something that could think about having an extension with QE as of now I'm doing this persistent interest rate movements and in the new version I'm doing just conventional policies or movement in the short rates I mean it's pretty remarkable although you just shock the short rate you still have movements in the long rates because there are these people that are the savers in my model are trading the bonds which is the short rate in one issue in the model so there's the expectation hypothesis and they're gonna have this pass through coming from them but doing more specifically QE would be a nice extension to do so thank you for the and then the question about changes in adjustable remorges shares and home ownership rates since 2015 so it's true that so I really don't look at that no wave of the HFCS which I think is measured in 2020 so I could look at that but even in the new paper I'm really just looking at the period till 2014 so I'm not gonna be able to say anything even empirically to what happened after I haven't really looked at it so my suspect is that the ownership rates hasn't changed much but maybe adjustable remorges shares the way we were discussing them before are changing so I should develop credit and then in terms of aggregating inputs response functions so I suspect so as of now I'm only doing Spain and the euro area but actually in the new version I'm doing all countries and I'm showing inputs response function for all countries and I don't have them yet so I cannot tell but since I log linearize the model essentially when solving it I should have aggregation because I'm not considering non-linear perturbation solution so you should have aggregation of the inputs response essentially that I have the average which is the euro area I have the single countries and they will aggregate exactly to the average that will be my take as of now but I'm working on it these days so I will get back ok thank you thanks a lot for the organizers for putting together such a nice conference and also for putting our paper into the program so the paper is joint work with oliva pojti and it's titled a Behaful heterogeneous agent nukeinsen model and our paper is motivated by some recent empirical facts about both the transmission mechanism and also the effectiveness of monetary policy first there is more and more evidence that monetary policy at least to a large extent affects private consumption through general equilibrium or so called indirect effects for example after an expansion in monetary policy shock the income of households increases households spend a substantial fraction of that extra income and this adds to the monetary transmission yet not all households are equally affected by monetary policy and a statistic that has been found to be particularly important in the literature is that high marginal pregnancy consume households they tend to be more exposed to output fluctuation induced by monetary policy and this unequally exposure of households it tends to amplify the effectiveness of monetary policy announcement about future monetary policy so called forward guidance shock on the other hand seem to have only weak effects at least on real economic activity and before inflation made its great comeback the advanced economies or a lot of them have been stuck at the effective lower bound for quite some time and we didn't see at least large instabilities arising from that and if you look at these four facts you will see that they are hard to square with instant and macro models if you think about a representative age in the Keyneson model it won't be consistent with any of these four facts so what we do in the paper is we develop a new Keyneson model with household heterogeneity and a form of behavioral friction and cognitive discounting under cognitive discounting households underreact to aggregate news as can also be shown in survey data and our behavioral heterogeneous age in the Keyneson model then hence the title can account for all these four facts and importantly it can account for them simultaneously meaning within one version one model calibration doing so our model overcomes attention that has been found in the Hank literature because Hank models that are calibrated are consistent with the first two facts the unequal exposure and the importance of indirect effects they tend to make foregains even more powerful compared to a representative agent model and also they tend to aggravate the instabilities at the effective floor bound our model can account for all them for simultaneously we then show that it actually matters that our model can account for them simultaneously visiting inflationary supply shocks so when a given supply shock hits our model or through the lens of our model this predicts a much stronger increase in inflation so in inflation response is amplified in our model compared to a model without these model features and that's actually in contrast to demand shocks with at least persistent demand shocks tends to be dampened in our model so the prediction for supply shocks in demand shocks is the other way around in our model something that I won't have time to talk about today but I still want to flag here is that this amplification channel in supply shocks also suggest that there is a more pronounced tradeoff between traditional targets for central bankers and side effects what do I mean with that well, given that there is this amplification channel of inflation reacts much stronger to supply shocks as central banker that wants to tame inflation needs to react much stronger needs to increase the interest rate much stronger which however in economies that heterogeneous agents come as side effects because it changes the return on assets which are not equally spread across households of course and secondly it increases the debt payment or the debt cost of government and so it leaves a fiscal footprint and these fiscal footprint is larger in our model than it is in standard models ok, I also skipped a literature in time for the time sake and I will start telling you how we write down the model so the model consists of three blocks and two out of these three blocks we kept deliberately stylized so on the firm side we employ a tax book, standard nuke engine setup monopolistic competition and some nominal rigidities and you can think about the firm side ending up in a standard nuke engine Phillips curve we also assume that fiscal policy simply issues bonds which households use to self insure and they raise some taxes because they have debt payment on these bonds but fiscal policy doesn't do anything else in our model for monetary policy we assume that monetary policy follows a rule subject to some monetary policy shock both contemporaneous monetary policy shock but it could also be of course foregoined shock because we always also analyze foregoined and for the first part I will assume that monetary policy directly controls the real interest rate to make everything a little bit more illustrative the main innovation however is of course on the household side where we combine this incomplete market setup in the style of bully haket ayagari with this behavioral friction of cognitive discounting and will be now precise on how we model that ok, there is a continuum of X under identical infinitely lift households and they maximize the lifetime utility subject to a budget constraint and subject to a borrowing constraint now there are two departures from the tax book model first of all is this incomplete market setup so each household faces an idosyncratic risk process which is labeled EIT and this idosyncratic risk process think about the Markov chain which pins down the productivity of a household and therefore its labor income and it also pins down the tax payments household needs to do and also how much dividends the households get from the monopolistic competition of firms on the firms side we actually use this function and calibrate it such that we can get this second effect into our model which is that the correlation of the marginal propensity consumer for household and the change in her income after a monetary policy shock this correlation is positive in our model and there is actually like a recent estimate in a recent AER paper by Patterson which estimates exactly this correlation and we target this correlation which she finds the second departure is that the expectation here is not derived rationally but boundedly rational in the following sense we assume here this cognitive discounting setup which has been proposed by Gabe in a representation framework and we extend it now to a to a economies heterogeneous household so the idea is that the expectation of a future value of some variable X or a vector of variable X we first break it down into an anchor value X bar that a household might have in mind plus a deviation around that anchor value and if we then employ this expectation operator it will be the anchor value itself plus M bar which is a parameter times the rational expectation of that deviation the first thing to note here is that if M bar is 1 we are back in the rational age in the rational fully rational world so this is a we can always compare our model to the rational counterpart of the model by simply setting M bar to 1 if M bar is however below 1 this implies that households cognitively discount expected deviations there is some form of myopia or as can be seen in survey data households underreact to aggregate news the anchor value that the households have in our economy is the stationary equilibrium outcome of that variable for example if that variable X is wages that is simply the steady state wage but this variable X might also be marginal utility that the households have in mind to have in some idiosyncratic state that might be tomorrow in then the anchor value for that is the marginal utility that she has in that state in stationary equilibrium this implies that absent aggregate shocks everybody is fully rational in our economy households fully understand their Markov pro-set with stri-status and credit risk but as soon as an aggregate shock hits the economy pushes the economy out of stationary equilibrium household anchor the expectation to stationary equilibrium we also do an extension later on we also assume that households underreact or overreact with respect to the idiosyncratic risk and it doesn't change our results ok, how can we discipline this parameter well as I said you can look at survey data and then you will find that households underreact to aggregate news and then you can compute it back to this mbar parameter and these empirical estimates they somewhere are in between 0.6 and 0.85 and for most what I'm going to show you keep the parameter 0.85 as a conservative choice at the upper bound of the empirical estimates ok given these two new departures from the textbook model what does this imply for monetary policy how does monetary policy work in our model how can we reconcile the model with fact 1 to 3 I won't talk about the effective flow bound today for time reason but about facts 1 till 3 and we will do so in two ways first I will show you a very stylized version of our model in which we essentially there will only be limited heterogeneity why? because we can then derive an analytical is equation and so we can learn from the is equation how these two model features how they interact with each other where do they enter the model secondly I will then also show you the results for the full model where there is a full blown heterogeneity there is a household heterogeneity who have different wealth and so on and so forth let us start with the simple model so the idea here is that we reduce the idosyncratic risk process to only have in two states and we assume that the government does not issue any depth so there is no outside asset supply in the economy and so in equilibrium no household can save so this is kind of a zero liquidity economy you can think about that model or the heterogeneity part of the model as being a two agent model but there is still type switching so agents still switch between the one and there is still this precautionary savings motive that households have in hang models and when we then derive the is equation you will see that the is equation looks kind of familiar but for two innovation showing up in the form of two new coefficients the first one being in front of the expected output tomorrow and the second is in front of the real interest rate today and the first thing to note here is that the coefficient the contemporaneous coefficient in front of the real interest rate today it only depends on coefficients that come from the household heterogeneity setup this lambda is the share of hand to mouse households this chi is the exposure of the hand to mouse households to fluctuations in output and cognitive discounting does not enter that coefficient yet in the forward looking coefficient it depends on the interaction between the cognitive discounting parameter and bar and stuff that comes again from the household heterogeneity setup basically which describes the dynamics of the precautionary savings motive of households and in when you compare it to a rank model in rank these two coefficient would simply be one so we can compare our model to the rank model we can also compare it to the rational version of that and analytical rational hang model and I will do this on the next slide when I look at the implication of that new is equation for conventional monetary policy shocks but also for forward kind shocks and actually I will start with the rank model and what I am going to plot here is not an impulse response function but it will be a set of monetary policy experiments so idea is always the following today monetary policy announced that it will decrease the real interest rate for one period at some horizon in the future so on the x axis the zero means at contemporaneous monetary policy shock decrease in the real interest rate today and then it jumps back to steady state and stays at steady state for all time 5 implies that today's monetary policy announces to decrease the real interest rate in 5 periods from now so it's forward kind shock on the y axis is always the effect that this have on today's output and a well known result of the rank model is that this is a flat line the effectiveness of the forward kind shock on today's output does not depend on its horizon so no matter whether the real interest rate is decreased today or in 5 quarters from now is always have the same effect and people have found this puzzling and have dubbed this the forward guidance puzzle and the reason is that in the is equation there is no discounting because it's a f parameter is 1 a second shortcoming of this model when it comes to the monetary policy transmission is that all of this is due to intertemporal substitution so direct effects play the large role of the monetary policy transmission where there is no role basically for indirect effect in that model if you look at the rational version of our model, the rational HANK model you will see that now indirect effects play a huge role in the monetary policy transmission that has been a major theme of if you think about Kaplan model in the Orlando's paper major theme of this early HANK literature in our model these indirect effects they are particularly strong why? because a monetary policy shock set a respite to a high marginal propensity consumed household which was fact 2 and they spend a lot of their extra income thereby making this pushing up the strength of the indirect effects and also thereby pushing up the overall effectiveness of contemporaneous monetary policy what you can however see is that in that model the forward guidance puzzle is actually aggravated compared to the rank model why is that the case? well when households expect a future boom because there has been a forward guidance announcement, households expect a boom in the future they know that in the high marginal propensity consumed state when they end up being hand to mouse they are actually relatively better off and since the precautionary savings against exactly that state they cut back on their precautionary savings already today which pushes up the effectiveness of the forward guidance today and this becomes if you go further on the horizon this becomes a really, really large this is however not the case in our model when there is also cognitive discounting households both cognitive discount that there will be a decrease in the real interest rate in the future and they also cognitive discount what is implied for their precautionary savings motif and both of these effects pushes down this psi f parameter and for all reasonable calibrations this is below 1 so there is actually discounting in the Euler equation at the same time and our model still indirect effects this important role in the monetary policy transmission because still monetary policy, contemporaneous monetary policy redistributes towards high marginal propensity consumed households and therefore these indirect effects they are very important in our model and so our model can be consistent with these two effects for no forward guidance puzzle and there is this major role for indirect effects in the monetary transmission now you might say that okay this has been basically a two agent model in the hang model do we only get the results because we use this simplified version of the model so what we also do in the paper is we use a more standard calibration where there is a lot of idiosyncratic states and where there is an actual wealth distribution and which we need to keep track of and which we need to then solve numerically and you see now this is the result for our full model for our full behavioral hang model and you see that on the blue dash line that is again our behavioral hang model the importance of indirect effects pushes up the effectiveness of contemporaneous monetary policy whereas the full rational hang model still suffers from these forward guidance puzzles so it also holds in the full model our results so departing from this full model in the paper we do then a lot of extension which I don't have time to talk about in detail today for example we look at heterogeneity in this behavioral friction because we find some evidence that there is some heterogeneity in this behavioral friction also underreaction with respect to idiosyncratic risk there is some evidence for overreaction with respect to idiosyncratic risk we look at sticky wages instead of sticky prices and we look at different calibrations of this unequal exposure and the results basically they go through with all of these extension or model changes so given now that we have a model which is consistent with these four facts about the monetary transmission we then use the model to revisit inflationary supply because we all know we are ended in our world where there is a lot of inflation and the idea here is that what I'm showing you today here is it's a simple TfP shock so think about as total factor productivity which follows an AR1 process with some persistency and TfP decreases such that potential output drops by 1% on impact now monetary policy follows a simple Taylor rule with a standard feedback coefficient and I again compare our model to these other models and I start with the rank model so as typical the nominal interest rate increases because the central bank leans against this supply shock therefore output falls but output falls by less than potential output so there is a positive output gap the economy is overheating and there is inflation in the rational version of our model this is a little bit amplified and the reason here is simply because the positive output gap redistributes towards high marginal propensity consumption households a positive output gap can also be interpreted as a too loose monetary policy R is below R star so there is kind of expansionary monetary policy expansionary monetary policy redistributes towards high marginal propensity consumption households this setterist fibers dampens the fall in output and therefore output increases by a little more and therefore there is also a little bit higher inflation in our behavioral rank model this is even much more amplified and the reason it is now there is a second channel going in exactly the same direction what is that? well, if you look at the normal interest rate again you see that they are increased and also the real rate is increased for quite some time and rational households they fully understand that they understand as soon as they see the supply shock real interest rate will stay up for quite some time and these expectations bring them or incentivize them to cut back further on consumption already to date so rational and irrational household will cut back their consumption already drastically to date simply by the expectations of higher interest rates this is not so much the case in our model because now households cognitively discount expectations or this higher interest rates in the future and therefore the output gap increases by even more redistributes even further towards high marginal propensity consumption households and so on and so forth until the economy ends up in an equilibrium where inflation is almost two and a half times as strong as it would be in their representative agent model we can do a little bit composition of where this extra amplification comes from the part that comes from the unequally exposure of households, the part that comes from discontinued discounting and the part that comes from the interaction of these two model features and the last thing I want to show you is that remember that I said the embar to the upper upper bound of the empirical estimates is 0.85 what if however we go to lower empirical estimates at the lower bound of the empirical gets 0.6 and you see that now the amplification gets even stronger so it's now almost three and a half times or more than three and a half times as much as it would be in the rank model and now the interaction really takes a leading role in this amplification channel ok, I'm out of time, let me quickly conclude and what we do in the paper is we develop this new Keyneson model, this householdation and this behavioral friction in the form of cognitive discounting we then show that our model can simultaneously account for these four empirical facts that I've talked about about the transmission of monetary policy so that it has implications for inflationary supply shocks because it amplifies inflationary supply shocks and in the paper we then also show that it suggests a more pronounced tradeoff between price stability and distributional consequences and I'm out of time, thanks a lot thank you thank you Fabian I open the floor for questions for this paper any question a clarification, you have cognitive discounting only on households and not on firms now, suppose that something like that would apply also to firms doesn't mean that the new Keyneson Phillips curve will have this discounting on expectations and so also the amplification of inflationary shocks might be put in perspective of who is discounting are there questions? that looks very interesting and also very relevant regarding the forward guidance puzzle I'm not an expert in these type of models maybe you can clarify also for others like me does household heterogeneity show up in the welfare of the of the aggregate welfare and then related how would monetary policy optimally behave in such a model because you had a simple tailor rule that was quite clear the results would do the best that can thank you very stupid question since you alluded to side effects fiscal policy and distribution can you say a few words on that? David? very interesting paper congratulations as well so again it's a similar question I think it's very intuitive what's going on in the model what I'm interested in is what's your view on how do you identify this how do you pin down the value of M what's the statistic that you're looking in the data what kind of expectation you're using because you're not presumably not looking at inflation data you're looking at output forecast to pin it down how do you figure that out ok how do we floor back to you Fabian? ok thanks let me start in reverse so we pin it down basically we look at the Michigan consumer service and we pin down about unemployment expectations and inflation expectations in both this kind we find that there is like when households revise their forecast we can find that this predicts actually a forecast error in the future and this means that households when they learn about news they don't go all the way in revising their forecast already and that's why this predicts in the future but there is also we are not the first one who did that over the comment is underreaction towards aggregate news there is also others people in the literature that have done that and they come to similar results and also similar ball parks of where the parameters should be but just to clarify you're looking at the sort of kovil on each angle regression ok then there was a question about so was it referring to what I showed last to this after supply shock ok so the idea here is so in the paper we also look at the case in which monetary policy can prevent inflation from happening after such a supply shock so there is kind of divine coincidence if you want in that case however interest rates need to increase much more forcefully than it does in a representative agent model or also in the rational hang model and this comes with some side effects in the real interest rate it changes the return on assets which has then a distributional consequences and at the same time as the government issues the debt and has interest rate payments also changes the interest rate payments so debt increases much more strongly in our model compared to in a rational hang model where debt increases by much less and this could be some considerations that monetary policy could be taken into consideration from a normative perspective that goes then back also to welfare question so the house of data tonight would enter the welfare function we don't have a welfare the social welfare function in the model we just positively show that there is this pronounced tradeoff and it is pronounced, this tradeoff is very pronounced in our model now if you want to do optimal policy we would first of all need to take a stance of is the steady state is the steady state efficient and is the distribution that we have in steady state is this optimally what we can say is that given assuming that that would be optimally a change in the real interest rate that is even stronger now than compared to other models would play around with this distribution even further and so would push it even further from its optimal steady state level and the same is maybe true as the interaction with fiscal policy because given that it increases the debt level fiscal policy at some point needs to raises taxes and therefore deviate from maybe the optimal steady state at tax level but we haven't done it fully the optimal fiscal policy optimal monetary policy in that model but that would be something interesting to going further welfare and optimality and then was something about the firms we have an extension where we also assume firms to be behavioral and we kept it out of the main part of the paper because we wanted to focus on the household side and how these two frictions interact on the household side we have done it also for the supply for the supply shock and then it would dampen the inflation a little bit it would also change a little bit the trade of between output not output gap but output itself and inflation and but still inflation would increase stronger than in a representative agent in this model and counteract this force when you discuss the amplification results it's clear from the input responses that the table rule is suboptimal it's a pretty bad policy because you create a positive output gap and inflation is going up you want to just actually close the output gap what is also very clear is because you have this quantity approach to rational expectation the table rule with this principle 1.5 that you're using with this value that you're using is even more suboptimal so you might want to have a discussion on how changes in that parameter in that policy parameter is actually affecting the amplification that you're trying to emphasize because in Gavage paper he shows that the table principle depends on this n parameter that you have so you might want to introduce some discussion to just emphasize that the amplification that you're getting is conditional on certain rules in particular the rule that you're using that is clearly suboptimal that could help that's true and actually if we would push the table coefficient to let's say infinity or close to infinity then we would close the inflation output but then we get this large side effects coming from because the real interest rate still needs to react much stronger than it needs in standard models that's correct ok, thanks a lot then let me again thanks Anastasia, Stefano and Fabian for the great presentation on behalf actually of the entire organizing committee would be a great success for your respective paper hopefully that will be published in a very good journal and again thank you for having joined us today