 Let's start with the first presentation by James, Slack and Seekly Quality in Sensitive Inflation. So we have 25 minutes, and then Lucretia, discuss in 15 minutes, and then we open the floor. So first session, please. Thank you. My role is to discipline you. There are plenty of slides I tell you. So James. Okay. Thank you very much. It's a pleasure to be here, and it's a real honor to be able to speak right after President Draghi. It is also a bit of pressure on me to make sure that I set a good example in keeping to my time limits. So I will do what I, I will try to do my best on this one. So I think the challenge, the challenge of understanding inflation dynamics is particularly acute at the moment, given the tightening economies both in the United States and in the Euro area. The United States is, as you know, especially tight with unemployment rate now at 3.8 percent, lowest since the 1970s. And yet despite these very tight economic conditions by a variety of measures, and indeed economic conditions that are facing additional fiscal stimulus, both as a result of the Tax Reform Act that was passed in December, and then the Subsequent Omnibus Budget Reconciliation Act of January, we're facing quite a bit of additional fiscal stimulus in an already tight economy, yet what we've seen is that inflation is just going sideways. It's been, I'm going to focus on quarterly inflation, so my last data are from the first quarter of this year, and as everyone knows, there's been some slight uptick in the monthly numbers more recently, but there's also a lot of noise in those monthly numbers. So the challenge, and the picture actually is very similar in the Euro area as it is in the United States, where HICP, X Energy and Unprocessed Food, which is the green line, has again been going roughly sideways despite the decline in the black line, the unemployment rate since this recovery began. So the question is what's going on here, and there are of course a lot of different possible theoretical explanations, and people point to anchored expectations and changes in monetary regimes and special factors and so forth, I'm going to go back to the basic theme of this conference and investigate two possibilities. First of all, one is just are we measuring slack wrong? And so I'm going to take a closer look at different slack indexes. Here I'm going to focus on the United States simply because we have a longer data span, so we're able to get a little bit better resolution. The answer to that question is going to be no, I don't think that there's not a magic slack measure out there waiting in the wings for us to use, and we just need to transition to using a different measure of slack. This is a cross the board, you can measure this in lots of different ways, and this flattening of the Phillips curve is real, it's really there. So the next question is, well, what if we take a closer look at components of inflation? Are there reasons to think that certain components of inflation might be more cyclically sensitive, that is, is there a slack or an economic activity relationship to changes of inflation that's present in certain measures but not in others? And this isn't something that we normally do, we normally look at an overall index or maybe strip out energy and food so that we can look at a less volatile index or measure of core, but I'd suggest that there's a fair amount of merit to looking at this, both a priori and then as we shall see, by looking at it empirically. A priori, different prices are set in very different ways, so at one extreme, take global commodities such as unprocessed food or wheat or maize or oil, and those are set in global markets based on global economic conditions. And at the other extreme, think about food at a restaurant that you purchase here or purchase at any particular location. Those are local markets, and so there's not really much international, there's the costs are based on some basis of international trade, but the actual price that's going to be set in the markup that's going to be made is one that's going to be very sensitive or potentially sensitive to demand conditions. If you add on top of that, so we might expect different cyclical properties for different types of prices depending upon their exposure to international trade and other factors. If you add on top of that concerns about the quality of measurement of prices, then you can imagine that you could see a much stronger signal of cyclical activity in well-measured prices than in poorly measured prices. And so that's something that's going to be part of the story too. And there are, and I'll talk about that again, mainly in the U.S., and now in the U.S. I don't get to hide just behind the data length. I will actually admit that I know much more about price measurement in the United States than I do in the Euro area, and so I would look to people here to extend some of those comments. Okay, so let me just give you one picture, and this is just the CBO unemployment gap. This is the Phillips relation, just a pure scatter plot, four-quarter changes and four-quarter inflation plotted against the CBO unemployment gap. And the main picture, the main takeaway here is that the blue dots are from the 1960s and 70s, the orange-gold is from the 80s and 90s, and the green is from the 2000s. And this is this flattening of the Phillips curve that is a chronic situation. We look at a variety of different slack measures. Gaps are incredibly difficult to measure. As we know, gaps change ex-post as you get more data because we have a better estimate of what the underlying potential is or underlying Nehru is. It's a challenging problem. So we look at a variety of different gap measures. Some of them CBO gap measures in the Euro area will look at the IMF and EC gap measures, but some of them also are statistical techniques, using statistical techniques and smoothing. When we look across a variety of gap measures, you get different pictures. Here's what our gap measures look like in the United States, and you can see that there's different pictures for different gaps, but they give a broadly similar pattern, which is one that indicates that moves very cyclically and one that especially indicates that we're in an increasingly tight environment right now. I'm going to put a table of numbers up, which is, I hope, yeah, this should be pretty visible. Okay, so these are, the first block of things are correlations, Philip curve correlations, and this is four-quarter changes in four-quarter inflation compared to a particular slack measure. And then the second block is the slope of the Philip's curve. And the main thing to note as you scan across this is if you look across the first three numbers, you'll see this decline in the correlation between this Philip's curve correlation between the unemployment gap and the four-quarter change in PCE inflation from minus 0.52 to minus 0.48 to minus 0.11 over these three episodes. And that is the main message of this table, and then I'll flip some things on the next one, is that that goes across the board. That's the, just changing a slack measure isn't going to change that answer. If I go down and I use other things, capacity utilization or employment population ratios or other measures, short-term unemployment rates, you see the same flattening of this Philip's curve. You can look at lots of different gap indexes, ones that we haven't reported here, and you get the same picture. So this is not, there's not a magic bullet out there in terms of using some other slack measure. Sure they give different pictures of the current circumstances, but the flattening of the Philip's curve is something that's really there. So then the question that we then investigate is whether or not you can learn more by digging into the details of the components of inflation. And there's really two separate and distinct messages here, and this slide relates to one of them. One of the messages is that not all components, not all price measurements are created equal. Some are really well, pretty well measured, quite well, I would argue quite well measured prices and quite well measured rates of inflation, whereas others are really quite poorly measured or even just purely imputed, so not measured at all. So at the one end, and here I'm focusing on the United States, on the one end we have, what we've done is we've characterized, we've grouped the different price components, the 17 components of PCE inflation into three different categories of well measured, okay, and not so great. And the well measured ones include things like housing, excluding utilities. So housing is measured not in housing prices, for those of you who aren't familiar with the US system, we don't measure housing prices, we actually measure rents and then we use those rents either for the rental units to represent the actual rental stock or to impute a owner's equivalent rent for owned homes. But they're all based off of actual market rents and they're based off of market rents at the same unit. So a unit is surveyed and then someone goes back six months later and re-measures it, and if they've painted it or if they've rehabbed it or put it in a new stove, then they're gonna upgrade that and build that into the costs. But this is really very comparable, it's the same good, it's the same apartment in the same location and they're getting the actual rent that's really paid. So these are well measured, the new goods problem is handled in a very sensible way if you get a new stove put in, then you get a new stove put in. Other measures are really quite well done and they aren't just goods, there are a lot of these are services. So there's the services from housing, recreational services, some of these are very well measured objects. Some of them on the other extreme are extremely poorly measured objects. So financial services is a large chunk of financial services are actually not, there are no market prices for, for example, convenient services provided by banks to consumers. So those are based on imputed prices based on interest foregone on some potential alternative asset or alternative investment. That requires a lot of judgment and it imputes some challenges, some major challenges. My favorite here, which is actually part of our consumption basket is not part of HICP, but it's part of per final consumption of is a nonprofit institution serving households. So for example, that's the price index there is the price of religious services. So if you go and go to a church, you don't have to pay for that, but it's a service provided for you and it's part of final consumption, what's the price of that? And of course, there's no price. So it's, this is a completely made up object. So what we do is we are gonna compare that to some various slack measures and we're gonna try to look at the cyclical properties of these different components. The slack measures we use, and I suspect our discussant will talk about this, the slack measures we use are actually band past. So we're gonna look at the cyclical, the business cycle frequencies. So we take industrial production and GDP and a number of other series, pass them through a time series filter so that we're looking at their fluctuations between six and 32 quarters and then we use that as our slack measure. That actually induces somewhat different behavior than gaps and I'm gonna come back to that. Here's four different components. So food services and accommodations, remember these are like, these are restaurants. So this is food, not food off-premises or unprocessed food. This is food at restaurants and this is hotels. And the blue line is our slack measure and the black line is the four quarter change and four quarter inflation. And you can actually see that there's a reasonable amount of co-movement between food services and accommodations and our slack measure. The correlation's about 0.46. Correlation of about 0.48 with rents. So rents are cyclically sensitive if you have hot economic times and if you're anyone who's gone to San Francisco and tries to rent in San Francisco knows that rents are really darn high that it's a very desirable place to be. So that's a local market that's subject to demand pressures and you'll see that and you actually see that in rents as well. On the other extreme you have goods or services in this case like healthcare services to a large extent those are negotiated or managed or administered prices in the United States. They're mainly not prices that are directly paid for by consumers. Plus there's an awful lot of difficulty of measuring what those prices actually are because of the difficulty of measuring a quantity or a unit of consumption in healthcare. That correlation actually happens to be negative minus 0.11 but we'll just call that zero. Clothing and footwear is an interesting one to keep in mind. It's a well measured, the prices themselves are well measured. They're really prices at stores or online. The problem is the new goods problem with clothing and footwear is really substantial and there's always improvements and of course the improvements are like this shirt gets out of fashion and then some other shirt that's exactly like it is introduced at a higher price and how do you handle that and then that's a real mess and I think the people who construct that are well aware of that and so what you might find is a very low correlation but that could just be a measurement problem or it could be some international trade issues. So if you look across these different series components you can see wide variation in the correlation structure. So ranging from 0.48 for housing excluding gas and electric utilities all the way into negative signs. Here's a little factoid. Motor vehicles and parts actually seems to have a strong negative cyclical correlation. That seems really strange. How could the correlation actually be backwards for motor vehicles and parts? Is why is there an upside down Phillips curve and it turns out there's just one observation there which is a program during the recovery act called cash for clunkers and that occurred at the depths of the recession and it completely changed motor vehicle prices in one quarter so don't trust that series. So what we're gonna do is we're gonna construct we're gonna combine these together and we're gonna say which of these if we look at weighted averages which comes in with weights and we're gonna construct an index and we'll call that index cyclically sensitive inflation and we're gonna use a regression to figure out what these weights actually are. You can tech this up and do it in lots of different ways but the basic message of the simplest regression approach gives an answer here. And say when we look at the weights the final column here is now the weights that we get in this regression on our CSI index where we're going to be weighting these up to be giving a cyclical signal and most of the weight goes on housing some of the weight goes on foods and beverages off-premises so that's restaurants I'm sorry that's a grocery stores. Some of it is on recreational services some is on food services and accommodations. A really interesting side note is we had gone a priori just based on conversations with experts in learning about how the quality of measurement almost all of the weight in this index goes on the well measured series. So the poorly measured series get almost no weight which is an interesting side observation that really suggests that these measurement problems could be potentially important especially when underlying we think underlying changes of inflation are small now. This is our index the black line is our cyclically sensitive index and you can see that at least more recently in certain instances we have a stronger cyclical behavior so if you look before the 2001 right around the turn of the century you'll see that there was a sharp rise in CSI inflation prior to the financial bubble recession. You'll see that there was a rise in CSI inflation coming into the 2007 recession a sharp drop in it coming out of the recession it's been rising slightly core PCE which is the red line has been roughly flat for the last three years the CSI index has been rising ever so slightly by about 0.5 or 0.6 percentage points over the last three years but it's a very modest increase. Certainly nothing like the increases that we've seen prior to previous recessions or previous periods of overheating. Let me turn to, so this is a table on these Phillips relations and although these gaps were not used in actually estimating it what's really striking about this table is that a lot of these Phillips correlations and Phillips curve slopes have remained stable or strong over the course of the last 20 years whereas that was not the case with PCE ex food and energy. Let me turn to the Euro area so I'm gonna conclude by looking at the Euro area measures of slack and we look at a variety of different measures of slack there we look at the IMF output gap the EC unemployment gap and then we use three band past series all for the Euro area 19 countries and those are output series. And the data set is shorter of course than for the United States so we do our estimation starting in 1996. I want to thank the ECB staff for helping us sort out some data issues they're extremely helpful. So let's take a look at some of these same components so now it's a little the ECB the Euro stat structure is a little different than the is quite different than the US structure the US structure goes along the goods and services dichotomy the Euro stat structure tends to go along whatever a consumer hedonic uses I guess is one way to think about it or looking at it from the perspective of consumer consumer basket consumer areas rather than through a goods and services breakdown. So these things are not at all they're not directly comparable in terms of their components but we still can take a look. So it turns out that restaurants so again restaurants food and service food food at restaurants seems to be have strongly cyclical in the Euro area housing and rents are not at all so that's quite different than the United States healthcare is just like the United States which it seems to have a very small cyclical component clothing and footwear is like in the United States it seems to be very noisy and also not with much of a cyclical component. So when we look across the different series you see some of them furnishing household items and maintenance has a high cyclical component food and non-alcoholic beverages has a high cyclical component restaurants and hotels has a high cyclical component. When we estimate the CSI weights a fair amount of it actually goes on furnishing households and maintenance this is a combination of goods and services so it includes furnishings like sofas and that and rugs and that sort of thing it also includes some household maintenance items repairs domestic services and a really interesting question is digging underneath this and trying to understand what's driving this correlation whether it's coming from the good sides or the services sides where you would think there'd be fairly different dynamics we have not done that and I think that's an interesting next step. This is our index of CSI inflation for the Euro area it's the black line unlike in the so the most striking feature of this is in the US our index of CSI inflation was rather different than PCE core it shows an uptick now it showed different properties around the 2000 recession it showed different properties around the 2007 recession in contrast in the Euro area at least over the last 8 or 10 years our CSI index looks a lot like PCE ex-energy and unprocessed food that's not true over the entire historical period but that certainly is true recently and that I don't really have any explanation for that other than just saying that's how the numbers worked out and it's a little bit curious because the weights we're putting on the components are really quite different than the consumption share weights that the HICP ex-food and energy is putting on the components so why that? I think there's no particular reason why that needs to be and it's kind of a curious feature that it is if you look at what's going on at the very end of this series over the last year or so these two indexes again looking at through core Q1 of this year have been roughly flat so they're showing very little indication of a pickup of prices so let me just summarize what we've done is we've done one thing we're pretty confident of is that just trying to find a magic slack measure isn't that's not the right way to go I think there's a lot of information in these components the components really have a wide range of measurement qualities it's intriguing that in the U.S. when you look at the cyclical sensitivity that the cyclical sensitivity is the most for the best measured series there is an issue in terms of what the dependent variable here might be one possibility which we've done is to look at these band pass series the cyclical indicators the cyclical measures based on activity a different one is to look at gaps gaps at pluses and minuses what we did is we did an exercise both for the United States and for the Euro area and we said well what if we use on the left hand side we use all through we use an output gap and we use an unemployment gap and we use our band pass cyclical indicator and we let the data choose the weights so that's something that if you want to be technical you can call that a constrained canonical correlation analysis and the weights that it shows it put almost all the weights on the band pass series which is quite surprising to us we thought that the gap measures would come in and say something and the gap measures didn't really come in so it sort of says that these cyclically sensitive inflation series want to be correlated with these cyclical measures based on these band passed indexes in any event what we're seeing on a substantive level right now is basically quiescence over the last over the last year that both in the Euro area and in the United States these measures like the core measures have been pretty stable and are not indicating any particular uptick what one takes away from that for monetary policy is certainly there's no particular reason right now to step to press the panic button rather this seems to be a situation where we're making slow progress towards an inflation that's closer to our targets or could possibly be commensurate with that sort of thing I do want to stress that the actual level of the CSI inflation measure doesn't really it's because it has these different weights on these components you can't really interpret the level of the inflation measure because it excludes other things that were to enter a market basket but you can interpret the changes and for now it seems to be fairly stable and certainly doesn't look anything like what we saw going into the 2000 and 2007 recessions Thank you, thank you That was perfect, thank you guys So, Lucretia, your remarks and there are many questions, James that will come later but I was a bit struck by the fact that you had you put all the you accent on one indicator which is CSI there are many others, as you know I was a bit... but that's for the discussion later Okay I think my presentation will come, hopefully Okay, good So, good morning it's a pleasure to be here since this year again and I think that Jim and I must be putting a good show because this is the third time in a year and a half that I'm discussing stock and Watson papers so at a policy conference and it's a pleasure to do so So, what is this paper about? This paper actually it's about looking for the Phillips Curve Jim has said it is the wrong way to go to look for a magic component reflecting the Phillips Curve but at the end of the story identifying a CSI this cyclically sensitive index is more or less aiming at the same objective and the issue is that it has been very difficult for empirical people like us to extract such cyclical component from raw data and so from that, you know, the usual question is the Phillips Curve flattening has it disappearing or simply is it lost in noisy data and I think the answer that Jim provides is it's actually lost in noisy data actually the answer is to the question of whether the Phillips Curve has disappeared or just is lost in noisy data and can be found it is very important for policy reasons of course or it is a very difficult empirical problem because as you know, inflation is dominated by trends which are actually influenced by expectations and monetary policy so what we are looking for is really a tiny component of inflation which is what is left and part of it is measurement and part maybe is cyclical Another difficulty is that one empirical fact about inflation is that and this is the empirical fact that Jim is emphasizing is that when you look at disaggregated prices there is very little correlation across component of inflation there is much little, much smaller correlation across component of inflation than there is for example across congenital real indicators and this is actually so I think I have very much sympathies with your approach which is try to exploit this heterogeneity in order to extract something cyclical so the message of my discussion is that I like the approach as I said and actually I also agree with the author's conclusion which actually there is such a thing which is cyclical so there is such a thing which is like reflecting the Phillips curve so the Phillips curve is not dead you can rephrase that conclusion in that way I will quarrel on some technical details on how robust is this particular indicator and then I will provide another model on my own in order to read the results through the lens of another model which I would use as a benchmark so three parts close look at the CSI my alternative decomposition and then I will spend in the third part some time just looking at where are we now and what are the data telling us about the Phillips curve in the US and in the Euro area so some fact about the cyclical sensitive indicator so the indicator that Jim and Mark have extracted so this is basically housing, okay so you see it here you see it here this in red you have the CSI in blue you have what I call the core PC housing which is actually the housing variable that they have used in the paper so I think it's basically that's what it is okay when you look at it in Delta and you see that there are situations in which the core PC housing is actually quite, you know this is picking up this is, you know, before the crisis this is actually the housing boom in the US and, you know, overall although it's not very visible in rate of differences it will be much more visible in levels you can see that housing tends to be a bit higher than the CSI now, if you look at their table with their indexes of course what you see is that housing has about 63% of weight in the index while it's only 23% in PC the second most important variable is food and beverage and then everything else is more or less zero so how do we interpret this index this is basically an index about housing it is cyclical but it's a very small component of PC which, you know, then we have to ask the question how can we use it, you know, in policies given the fact that it's such a small component now, if you look at why this is the case why the index pick up housing I think you have to look at the correlation between housing and what they call the CAI the cyclical activity indicator which is this kind of bandpass smoothed version of various real variables and unemployment, GDP and so on so indeed you can see a lot of correlation in housing and this actually would be the difference which you don't find in European data because this is actually mostly rent while, you know, we have a different definition in the HICP in the Euro area so then the question is, okay, there is this correlation but then let's look at the CAI so the indicator of Slack and plot it against the CBO the Congressional Bad Grid Office Output Gap to see what are the differences in order to understand what does regression that picks up the way it actually picks so you see that actually the CAI is not exactly the output gap as Jim emphasized and it has some weird, I would call them weird characteristics for example, it has a peak here I don't know exactly what that is and actually from 2012 it indicates a weakening of cyclical conditions in the US which it seems to me a little bit counterintuitive so given this kind of smooth but volatile component in CAI then I wonder whether when you do the correlation you are capturing something which is not actually business cycle but maybe some component of the housing dynamic which reflects something else probably a part of an asset price cycle since housing is also correlated with asset prices now this is actually more visible when you do it in level you can see here you have the CSI so they're indicator here you have the core PC and in blue you have the PC so you see that the CSI it is higher so Jim is saying we should not interpret the level but the level is just the reflection that also when you see it in first difference is it indicates it's a bit shifted up so this is when you accumulate back you can see this shift and you can see that now this is flat in the last part of the sound and well above the PC so the question here I would have for Jim is are you really picking up just the cycle or are you picking up a slow moving component which is specific to housing and possibly correlated with housing prices so basically how should we interpret the higher level of the CSI with respect to the core PC Jim says we shouldn't care about it does this reflect asset prices and then the third question is why in the last part of the sample this indicator is actually flat although there is a lot of evidence that actually the recovery in the US has gained strength in the last few years now the fourth question is okay let's say that we find a disconnection between the PC and the CSI how should we really interpret it and in particular how should we use it so if for many years so we have seen that the CSI is a very small component of PC so if the CSI is signaling a cyclical strengthening but this does not reflect in headline so what okay so how should the policymaker you know use this thing in policy decision so in order to try to provide a tentative answer to these questions I'm gonna produce another model okay so let's now stop for a minute because this is another model and I will have to spend sometimes discussing it so this is an alternative approach which in principle should deliver the same results of the CSI so instead of pre-massaging the data I am going to put everything in the same box real variables and inflation variables I'm not going to look at all this different indicator of inflation but I'm just going to look at oil and headline inflation in my case the CPI because I think that understanding oil it's you know it will give you a lot of you know mileage in understanding also the cyclical part and okay so I will tell the model so I will do everything jointly without pre-massaging so to avoid these kind of quarrels about what kind of measure of output gap you should use and so on okay so the model will filter I will address identified by saying please give me a cycle which is stationary okay because I mean if the cycle has to have some characteristic you will have to be stationary okay so there will have to be reversion to the mean and the rest is trend and the cycle I will split it into part a part reflecting the real variables which is the common component from the real variable between the real variable and the nominal variables so in a way it's like a CSI okay so it kind of capture the maximum correlation between the real and the nominal variable of that stationary component but I would also tell the model to give me to kind of identify an orthogonal part which is just energy is that part of energy that does not go through the Phillips curve I will call the Phillips cycle the common component between unemployment rate GDP inflation expectation and inflation I will identify the trend through inflation expectations but I mean today I would just want to concentrate on the cycle so here are my results this is the output gap is the common component in real GDP this is just the cycle part of my result this is the output gap common component between prices and real variables in GDP so you see how I will return to that I will show you a picture how it compares to the CBO the yellow part that you see here in unemployment is an idiosyncratic element because we know that unemployment and GDP they have lag lead relations so when you kind of average them out you don't exactly get the same you average variables which are lagging and leading so have different dynamic characteristics but I really want to focus just on CPI the blue here is what I call the Phillips curve component which would be similar in principle to the CSI this is a very smooth component it is actually a cycle which has a duration of about eight years so that corresponds to our notion of the business cycle the red is oil so there is a very large component of oil which is actually orthogonal to real variables probably influenced by expectation or by global economic slack rather than local economic slack and then you have some yellow here which is just measurement error now let me just zoom from 2008 in order to understand why this is the composition is actually important you see that this is the crisis you have the blue the Phillips curve is pushing prices back and also oil is pushing them back but actually from 2011 oil is pushing them up while the Phillips curve is pushing them down so this is actually the missing disinflation why did you have a missing disinflation because oil was working in the opposite direction of the Phillips curve of the real slack of the slack economic conditions as you go up as you move on you see that now oil is moving again in the opposite direction but pushing prices down while the US economy is recovering and actually here my output gap is quite different from GIMS because it's telling us that we have had a steady reinforcement of business cycle conditions since 2014-2015 so here this is the missing inflation the missing inflation is because oil is clouding out the Phillips curve component so this is basically the importance of getting rid of oil so for the euro area you get very similar results but again what I differ here from GIMS that at the end of the sample I have that actually according in line with what Mario Draghi just said that I think that the European recovery is getting steam and actually this is the forecasting period in dashed lines you can see that the price the Phillips curve component for prices has still a little bit to go so it's not neutral but there are still there are pressure for this tiny cyclical component of inflation and we'll put prices up and I'll show you the forecast in a minute so let's compare it now this is our output gap as it comes out from that model the data the model is doing the extraction is quite similar in shape with the CBO our is this green here I mean I don't want to emphasize of course we're all massaging the data so I'm not pushing I'm not saying mine is better or not but I'm just saying what is this thing here in the CI of their paper so I think that we from the 90s to 2001 we had a steady recovery this is what the output gap measures the CI picks this thing which then feeds the CSI the cyclical thing and then what is this thing here that we have picked according to the CI earlier on and now actually the index is pushing pushing the cycle down now in terms of now the comparison between my Phillips core component the blue and the CSI there is quite a lot of correlation others should be but at some point the correlation actually the level is shifted up so that you can see because we are very similar to core so you can see the same chart that we have seen before and and actually you can see that here while we are telling we are indicating that cyclical pressure and inflation are going up they are quite flat okay so there is another component but I think I'm running out of time so that you see here there is a little bit of a difference between the CSI and housing this is housing and actually if I had more time I would show you that this is all related to energy so that actually this is the second component which is weighted in Jim's indicator which is food but food has an indirect effect on energy and when you correlate it with energy you see that you know the you know the indicator is they correlate with housing whenever energy goes up okay so let me this is for the euro area instead okay we are much more correlated with dream and stock with dream and marks indicator and this is just because the housing in the euro area has is a different type of variable is flat is not cyclical and so you know we are at that point we are very correlated which indicates that you know housing is really what matters okay so that and probably you know there is a a cycle of housing which is not entirely correlated with the business cycle now does it matter okay let's see you know after all this is all about a lot of massaging um but if this should if this matters this cyclical pressure that from a James index or from our index should translate a some point on headline inflation okay so this is how this thing should be you know possibly used so now where are we now today the U.S. the CSI is saying positively the positive cyclical pressure for in play inflation more or less constant since 2012 and while the trend model cycle model says upward pressure for 2015 which has now peaked okay so our model will tell us that now if there is no more for a fiscal expansion or other external shocks or oil development mechanically you know the model our model will will bring inflation gradually down because the U.S. cycle has picked for the euro area um well James says neutral pressure we say upward pressure since 2007 gradually building up not picked up not not having picked yet so that's the forecast the first CPI inflation for the U.S. okay so this is we are actually saying this has peaked and is going to go down gradually so we are much more we are much lower than the OCD and IMF forecast this of course there is a lot of uncertainty around that picture consider that this is a tiny movement because actual inflation is dominated by trend okay so that this for good things because you know possibly inflation expectations are anchored okay so this is the HICP and we are actually having quite a different pictures I mean there is a gradual building up and here we are very much similar to the institutional forecast so here we have updated this this is that we started the forecasting in the first quarter of 2018 so you know similar in spirit but quite different results empirically especially for the last part of the sample where we should focus on so conclusions doubts about the model specification however I agree with the paper fundamental conclusion there is a cyclical component of inflation hidden in the data this indicates that the Phyllis curve is alive and well but it has to be found and you know it is like finding a needle in a haystack but is there somewhere okay so go and look for it thank you apparently shall we take a few questions now before because I think that the second presentation is a bit different so I I think maybe you want James to react to especially the the last graph that we had this inflation outlook from Lucretia compared to what you had I think it's would be useful to have a little exchange on this now oh that might work okay excellent okay so I guess the first thing is thanks so much Lucretia I look forward to being a discussant for one of your papers in the near future yeah so that we don't so it would be great to react to those forecasts we don't we actually intentionally didn't try to take this to the forecasting stage I think it's still early the one thing let me just stress one thing though it I think it's possible to take issues with a lot of the particulars of what we would have done what the one thing if there is a take away from here one thing I want to stress I think there's a lot of interesting information in these components uh... there's also a lot of non information in some of the components and stripping away those problems and taking them seriously and modeling them seriously I think is a really important thing to do I know some of that work is actually already ongoing here at the ECB uh... it's not just a matter of constructing the indexes uh... but it's a matter of using our tools to analyze those components and to extract this cyclical information uh... I think what I'd rather do than respond to Lucretia's excellent comments is just get some comments from the questions if we have some time pick up one I had one question I mean uh... James I mean taking the room but why do you come only with that indicator one indicator we have a battery of indicators at the ECB uh... you know one base on the dynamic factor model when you try to see the underlying trends you know in inflation and not uh... you know basically Philips Curve sort of reasoning so we have a number one so I was surprised that you only focused on that one you have others uh... just here for this presentation you decided that oh no I think that the goal here is a very simple one which is if you go and look at the components do you see an indication of Philips Curve type behavior then we decided we are trying to put that together into a single indicator but I think just going in looking at what's happening with food and accommodations that's really an interesting index what's happening with housing in the United States I really would say that it's not as the way that the most models of housing go is that the fundamentals are based on rents and then that drives what the fundamental price of housings would be and then whether you have a housing bubble would is usually measured by how far away asset prices are from the rents underlying and the rents are driven by local economic conditions so I think that that's actually uh... the way it's measured is getting at uh... getting at something of economic interest give name in affiliations because I'm Ben Friedman also from Harvard Jim I think it would be interesting for you to say something about how you see the central bank taking advantage of what you found the reason I ask that is that it strikes me they're too quite uh... opposite interpretations uh... you mentioned two reasons for thinking a priori that different components of the price index would behave differently over cyclical frequencies and although you mentioned them together they have very different implications for how the central bank might use what you found you mentioned on the one hand that some of them are badly measured you gave examples and if that's what's going on then a plausible inference is for the central bank to target the index that you constructed on the notion that if those components that you flag are moving forward rapidly or not then it's plausible to believe that many other prices are moving forward or not also but we just don't observe it because they're not measured well and the implication is that the central bank should target that index and by contrast you mentioned the difference between prices of goods and services that are subject to local within country cyclical pressures versus those that are not and if that's what's going on then the central bank shouldn't target because then if your index is showing some uh... say above two percent inflation there's no reason to think that other prices showing above two percent inflation it's not a measurement issue it's just that they're said they're set very differently and so the fact that the csi is say that's moving forward at three percent might be informative and might help forecast but it certainly doesn't represent any reason to believe that prices as a whole are unstable on the central banks definition so could you say something about how you without without downplaying any of the considerations lucretia mentioned could you say how you would even if everything that you said we take it face value how would you use it okay so i'm going to give the very the very short answer is uh... in the current circumstance with three-part eight percent unemployment in the united states there's a based on the historical evidence if you go back and you think about it over a long period of time when we've had unemployment rates that have been this low there's been a sent inflation is that something that's going to is that something that we see today or is it not so we're gonna this is the way i'm looking at this is this is a way to hone in and see whether we're seeing nascent inflation in areas where we might expect it we don't at the moment see that and that's a very interest that's an interesting finding that tells me that we are not in a position where we need to step on the brakes quite yet uh... president drahi mentioned uh... longer lags and uh... lucrezia uh... mentioned trends uh... i mean these are really serious issues and i'm worried that the acceleration is to be that you've taken up inflation is is misleading uh... if one thinks of inflation relative price developments adjustment to to relative prices is part of the inflation process and then the sargon back in nineteen sixty four wrote this this great paper using those ideas in order to understand the inflation process uh... you know i think the the accelerations view has uh... but these two problems one is the demand curves slip down so if a firm raises prices too much the tendency for prices then to correct downwards and the other one is that central banks after all uh... here and the focused on on dampening inflation so when you have a year of high inflation the central bank then intervenes and that's followed by lower inflation so that's that's the opposite of the accelerations view uh... we've done some empirical work jr and i in which we find for the u s that uh... once you allow some of these long run effects to enter the equation uh... a year of high inflation tends to be followed by lower inflation so that's the opposite of the acceleration is to use i i really think that these long-run uh... it's long-run view inflation needs to be treated more seriously and one other point uh... i think in uh... in two thousand eight two thousand nine the world economy suffered something like a heart attack now that's a structural break you mentioned the cash for clunkers for example as one example of really weird price development during that period so one way of thinking of it is to say well let's let's see whether after that structural break things are back to normal so you instead of treating the period as a whole and trying to explain what happened you might try to take out the period of the crisis and let's say look at uh... the model up to two thousand eight and then from twenty twelve onwards and see whether without window taken out we're back to a more normal set of uh... relationships one of the exercises we did was we estimate the model just to two thousand seven and you get basically the same weights for what it's worth it's basically the same index and very similar cyclical patterns also my mother is very stable question there name and affiliation of things frank smiths i was wondering about uh... whether you've looked into non-linearities uh... i know that in the past you've sort of demonstrated at least for the u.s. that the phillips curve emerges in sort of deep recessions and it's quite dormant in boom periods and i was wondering whether they could also be non-linearities in boom periods that as the output gap closes actually the phillips curve steepens again and and sort of price pressures emerge yeah so we didn't we didn't do that in this exercise and uh... uh... you frank's referring to our jackson hall two thousand ten paper and i think that that's that's that certainly is linked you know sitting in the background and that's another reason for means being concerned about the lags in the current circumstance did you uh... i'm joe hazel from MIT did you look at excluding the bulk of recession so in the early nineteen eighties so it seems to me if you look a lot of the evidence that post nineteen eighty four going to the nineteen nineties the phillips curve is very flat and in some sense the surprising episode is the early nineteen eighties not the subsequent period and i was wondering if that affected your calculations well i mean certainly the there's this enormous change in monetary policy post valker so that do what we would expect structural shifts and uh... indeed whenever you whatever you look at there's a roughly speaking great moderation nineteen eighty four vocal recession whatever you want to call it uh... structural break around that period of time so we've just entirely focused on the post nineteen eighty four episodes uh... jim bullard st. louis fed so uh... just two comments following echoing some of the previous comments it seems like there's a fundamental problem if you go to subcomponents of the price index because when one price goes up you'd expect people to substitute away moved toward other goods and uh... so it seems like that would present bias in your uh... in your results so i'm not sure if religious services are supposed to you're going up or down when these other prices are are uh... are moving and you'd like to be able to so i'm wondering what the biases for that and then the other part is that there there really clear trend components in the in the sub indices medical inflation and especially tech uh... prices but many others and i'm wondering what you're doing are you saying that there are different trends with the bandpass filter are you imposing a common trend or what are you doing with that bandpass filter pulls out the low-frequency trends in each of them so treats them as it's agnostic and streets all those trends is different so it doesn't impose any common trend that but you're absolutely right that that's actually a big a big deal if you analyze these components that's difficult i guess in terms of substitution sure uh... in if we think about the ultimate effect on prices in a market basket you have to take into account substitution will have more discussion about that but uh... but also if you're just going back to this sort of overheating pressure is there are the overheating pressure that's visible i think that you know view this as an indicator as a instead of thinking about it in terms of ultimate consumer price of basket prices other questions i mean still the point of the creature asset prices you know in the link you know the u.s weight in in housing i mean it's uh... i i don't find out that i just yeah i don't think that your presentation i think it's really something but i just don't think that's a big i don't think that's the right interpretation here because the u.s it is rents and the rents are driven by local by various supply and demand considerations now you could have criticized as the rents are driven by both supply and demand so so if there's a increased housing construction then that's going to be something that's going to be okay so related to that and uh... and yes and and this question and that there is a the an idiosyncratic trend that you can see in the levels yes so that's why you are higher than p c for most of the sample and actually since the nineties so that uh... you're picking up not only the cyclical thing but also that kind of smoother component which is that part of housing which is related to asset prices okay i think that's that's something is going on because the level mean you have to interpret that thing so you and you you have to it has to be commensurable to inflation you're consistently high it doesn't it doesn't doesn't has to be related to that has a price is it has to mean we've had a lot of over the last twenty years or thirty years that his the the story of housing in the united states is one of the local zoning restrictions adequate supply and so people are spending more and more of their incomes on housing there is an idiosyncratic trend anywhere in europe there has been a big disconnect between uh... asset prices housing and and rents so i was not surprised by the the low way that you have in the you're interested so comments questions okay if not uh... we we're also uh... so you're so the two parts i like the provocation part that you have about uh... expectation as an instrument so please uh... thanks thanks for putting this paper on the program that's a great pleasure to be here this is joint work with only co-op you know who's also here satan kumar and uh... made you put them on there now why is this a good time to talk about inflation expectations as a policy tool assault larry summers at the stage perfectly for us we know that uh... a key variable for many economic decisions consumption investment pricing employment is the real interest rate and uh... by moving this object around we can decelerate or accelerate the economy and so we can achieve microeconomics stabilization and the way we typically do this in terms of monetary policy is to anchor inflation expectations at some level say two percent and then we're going to move around the nominal interest rate which is going to translate into changes in the real interest rate and so this is going to accelerate or decelerate the economy and this is how we stabilize now if the nominal interest rate is stuck at the zero lower bound we cannot do this policy anymore right we have to use something unconventional but it doesn't mean that in this set of circumstances we're completely powerless we can still have some uh... power to move the real interest rate so if the nominal interest rate is zero we can still move around the nominal interest uh... the inflation expectations which are going to translate into changes in the real interest rate and so again we can accelerate decelerate the economy we can stabilize it mario dragy a few years ago we had a very simple summary of how this works when inflation expectations go up with zero nominal interest rates real rates go down when real rates go down investments in economic activity improves that's the reasoning okay that was in the response to a question from a journalist now we can elaborate on this summary and describe in a various channels of how this inflation expectations are going to work for example by raising inflation expectations we can stimulate households to consume more we can induce firms to invest more we can induce firms to raise their prices workers demand higher wages all of the good things that we need to happen at the zero lower bound to stimulate the economy so it seems like conceptually it's a very straightforward exercise and we should use inflation expectations as a policy tool but before we do this before we make this a conventional tool we should answer some uh... practical questions the first question is going to be whose expectations we should target and obviously we should care about financial markets professional forecasters but the channels I described in the previous slides are really about households and firms and so one answer we provide in this paper to this question is that we should really care about households and firms if we want to make inflation expectations inflation expectations a successful tool now if it's households and firms the next question is what forces influence uh... expectations of this agents what moves them around and the basic conclusion we get in this paper is that salient prices of frequently purchased goods gasoline food in some countries exchange rate are the key determinants of inflation expectations now what is you know this present challenges and opportunities for example central banks cannot run networks of gas stations supermarkets to control this prices so it'd be really unconventional monetary policy but we have some opportunities there as well in principle we can just engage in communication strategies and try to talk people into changing their inflation expectations and the question is is is going to work were you available evidence based on a variety of sources and our answer to this question is yes you can communicate to households useful information they're going to behave like basions they have some priors to give them signals that they believe and uh... you can be very effective in moving their inflation expectations now suppose you're successful in moving inflation expectations you change beliefs then the next question is are people going to act upon this expectations right that's the key question because you know some people think that's very measures of inflation expectations appear on the list they don't matter for innocent for consumption for investment for employment for pricing again were you available evidence i'll show you some details in a bit and answer to this question is yes the survey measures of expectations are useful people act upon this information there are some details to be argued about the basic conclusions is people act on this expectations alright so so far the logical chain is very clear we have all ingredients in place uh... but there are some challenges in using inflation expectations as a tool i'll describe this in in more detail later but basically we identified two issues one is we have to have much better measures of inflation expectations if we want to use us as a tool that's one and second i will argue that uh... consumers and firms don't really pay a lot of attention to monetary policy and low inflation environments and so if we want to move inflation expectations if we want to convince people that they should consume more they should invest more we have to penetrate through i say here away with an attention it's really a wall of inattention it's it's really hard to reach to those people uh... now in the rest of the stock i'm going to elaborate on each of those points in trying to convince you that you know our conclusions here actually make sense so the first question uh... whose expectations we should mirror this is the time series of inflation expectations in the u.s. for professional forecasters and financial markets you see the very very highly correlated uh... sometime in two thousands they stabilize it two percent and so this is what you know we used to argue that uh... central banks were successful at anchoring inflation expectations now when you look at households this is the time series of inflation expectations in the michigan server consumers you see that in the early part of the sample the correlation between the inflation expectations of households and professional forecasters was very very high but sometime in two thousands we start to see uh... big differences we see wedges and sometimes the differences are very big as big as two percentage points right so it's not clear that inflation expectations are very much anchored in the u.s. for households and so we can really use inflation expectations of professional forecasters and financial markets as a substitute for expectations of households now one key question here is where firms are going to be in the spectrum it's conceivable they're like professional forecasters they pay a lot of attention to what what happens with inflation what monetary policy says what they may be like consumers for this paper we're on a special survey on a national representative sample of firms in the u.s. and we found that the average expected inflation in the u.s. for CEOs and CEOs in the sample was three point seven percentage points which is very close to what happens with households they're very far from professional forecasters and financial markets so if we care about managing expectations of households and firms we can rely on measures of expectations from uh... professional forecasters and from financial markets now when we see these differences the key question is then what moves these expectations why do we see those differences we review available evidence and basically group these forces into four beans the first being is perceptions and experiences of inflation basically if i oversimplify the available evidence it would say that if i think inflation was five percent in the previous over the last twelve months i'm going to take inflation is going to be five percent in the next four months that's kind of the basic uh... result in this literature now this is a sort of a backward looking force there is a forward looking force shopping it offers us natural opportunities to observe prices and update their beliefs our beliefs about inflation and to give you a sense of how powerful this force is this figure here shows you time series of inflation expectations in the Michigan sort of consumers the blue line and the black line is the price of gasoline in the u.s. of the monthly frequency now i know i guess all of us know uh... that it's impossible to prove anything beyond reasonable doubt in micro-economics but when you see this kind of correlation you will start to wonder if you have a shot at you know proving something for example we typically think that high frequency variation in inflation expectations in the Michigan sort of consumers is just pure noise but here this figure is telling you that this you know short-term fluctuations are actually very highly correlated with the changes in the price of gasoline alright so there is a very strong correlation and there is something similar for for europe now we can also obviously get information from media and we may be affected by policy when we review the relative strengths of those forces we basically conclude that shopping and perceptions are very strong forces and policy is a very weak force and will give some examples uh... in a bit now so if households are really influenced by their shopping experience what is uh... the option uh... for central banks to influence the inflation expectations one option is just no talk to those people and try to convince them that you know inflation should be such and such number can we manipulate the expectations in this way and that's a key question and to answer this question uh... typically this literature relies on randomized control trials which means you will use it expectations if you get a sense about priors people have then you have an informational treatment for example you tell them about the inflation target and then you're going to measure their expectations again and you see how much they revise their beliefs the two basic results from this literature is that uh... economic agents households and firms uh... basing learners they have priors they have signals dependent on the relative strengths of these two forces they come up with uh... posterior beliefs that's the first result the second result is that this informational interventions are very shortly basically within six months people forget what you tell them okay so it's probably not surprising they're very busy now to give you a sense of how much you can move inflation expectations uh... only core beyond me and michael weber who's going to speak here tomorrow ran a special experiment on households participating in the ac nilson homeskin uh... panel uh... the experiment is very similar to what they describe you know just in a bit basically we have a control group and then we have various treatment groups one of the treatment groups was told that past inflation is such and such number and then we see what was the treatment the fact of that how much people revise their beliefs and we see that the average treatment the fact here is minus two uh... percentage points that is you're up to start with four percent inflation expectations you moved them back to two percent that's roughly the magnitude you can also tell people about the target of the fact again a very powerful revision of uh... in inflation expectations you can also tell them about the four mc inflation forecast again a big revision of their inflation expectations so you can move inflation expectations by talking to people now as i said the next question is how do we know that you know consumers and firms are going to use this information for their decisions consumption investment employment pricing that's a huge identification challenge because we don't know what people know right we don't know what they do because of something uh... the way to get causal estimates is to rely on each of the experiments randomized control trials the design of this randomized control trials is similar to what i showed you before except that you have an extra lag when you measure not just expectations but also actual outcomes you know prices that uh... charged by firms or employment that was made by uh... firms or consumption spanner that was done by households now what is the so summary of this literature that uses randomized control trials in each of the experiments for consumers there is some variation in the results but the basic conclusion here is that if you can convince households that inflation is going to be high in the future they're going to spend more today for firms it's a little bit more nuanced when we did this experiment in new zealand we see that if we convince firms and firms uh... firms that inflation is going to be high in the future they're going to hire more people do more investment and have no effect on prices when we do something similar in italy using a quasi-experimental design we see that moving inflation expectations up generates high prices so firms want to charge high prices but at the same time they want to lower their employment and lower their investment it's going to move in the opposite direction now why we have this differences between italy and new zealand is an active area of very active research for us uh... but you know it tells us that you know the treatment effects are not necessarily invariant they depend on circumstances for example in italy we have a zlb and new zealand we didn't have a zlb maybe this is one of the reasons why we have this differential responses for france we don't have randomized variation but we have a very long time series of matched inflation expectations and price changes and there we see a very strong correlation between inflation expectations and prices if a firm believes inflation is going to be high it's very likely this firm is going to raise its prices now what what conclusions we can have from this literature first is that households and firms act upon their reported inflation expectations so this is not just noise that's good news uh... also we know that manipulating inflation expectations can give us a new channel to influence inflation as if we can convince firms that inflation is going to be high in the future they will start moving prices now okay and that's a new channel for monetary policy now as i told you this is still a relatively young literature and we have some details to iron out uh... but for example we need to understand better mechanisms and generally equilibrium effects but conceptually all the evidence i have presented to you now suggest that we can use inflation expectations in principle we can move inflation expectations people are going to act upon those expectations we can have a new stabilization tool now before we do this as i said we have to address some challenges and one is going to be measurement of inflation expectations for households we have reasonably good measures of inflation expectations even though we found it surprising that many central banks do not have their own inflation expectations they have to rely on somebody else to do those expectations for example in the u.s for a long time i was the michigan server of consumers there was nothing around by the fact now there is a a server run by the fact the e-c-b doesn't have its own server of inflation expectations of households for firms the situation is a little worse okay and the paper we have a long list of complaints what is not quite right was with this service let me give you some examples uh... the service tend to be based on small non-representative samples and if you want to manage the area of the economy you can't rely on fifty firms to say okay this is what we're going to do uh... in terms of monetary policies spend billions and trillions of dollars or euros trying to achieve something another one which is really problematic is that many of the service have qualitative questions you know what prices go up prices will go down prices stay the same again if you want to use inflation expectations as a tool you have to have a lot more precision and granularity in this response so you can just rely on up and down you can use balances another common problem there is priming of responses basically you notch respondents to give you an answer you would like to see and kind of exaggerating this but this is roughly an idea for example bank of canada has those business outlooks which is a multiple choice question we have four possible responses inflation will be less than one percent one to two, two to three, more than three okay so it gives you a very narrow range and obviously firms when look at this they pick something in the middle two percent and so you claim that inflation expectations are stable uh... you can give them one option two percent yes no uh... for some surveys it's not really about irrigate outcomes uh... for example atlanta fed survey has questions about firm specific unit costs and we ran a series of experiments in new zealand to explore whether it's a problem to ask firms about their uh... specific outcomes or about their irrigate outcomes we found that this micro macro uh... expectations have very different properties they're basically uncorrelated and so if we want to measure inflation expectations we should not rely on uh... firm specific prices expectations about firm specific prices or from specific unit costs now surveys of firms are expensive you have to penetrate through layers of secretaries and various filters before you reach out to ceos and ceos especially when you talk to big firms they are expensive but they're also extremely useful and maybe not surprisingly some of the best service are run by countries which have the recent histories of high inflation uroguay and ukraine you know if ukraine and i'm from ukraine if ukraine can do it i think any advanced economy can do the same if not better alright so that was the first challenge the second challenge is breaking through this whale of inattention alright they give you a sense of how problematic this inattention is again we ran a special survey of firms and households in the u.s and ask them to report their perceptions of the inflation target in the u.s what's the target of the fed you know you look at this and we were very surprised roughly twenty twenty five percent of firms and households in the u.s gave us the right answer roughly fifty percent said you know they don't know gave us a nonsensical number right so that was shocking and if we want to manage inflation expectations it's not going to work if we have something like this now this is not just about you know steady state perceptions about targets we can all look at sort of high frequency high frequency evidence of inflation expectations this vertical line shows the time when the fed made an announcement that it's going to have a permanent two percent inflation target and the I guess roughly horizontal line is the time series of inflation expectations in the Michigan serve consumers again it's one year ahead you know before the announcement households thought inflation was going to be four percent after the announcement the soil is going to be four percent you know we look at the financial markets and professional forecasters we see big movements there for households the treatment effect was zero they just didn't know what's going on now this is not about you know the fed or you know inflation expectations you can look at other countries look at the measures of uh... expectations for example this year here shows employment expectations in in european countries around the time when the cb made an important announcement and again you see very little action here now this may suggest to you that okay this is a mission impossible we can influence those people and they just you know hopeless but i would be wrong conclusion uh... we know public campaigns work very well for other policies for fiscal policy for health care policy we can influence expectations of people we can change their choices we also have solid experimental evidence telling us that if we can somehow reach people give them useful information they're going to revise their beliefs and they're going to act upon those beliefs so it's doable the question is how you know design our messages in such a way that we can penetrate through this wall of an attention and here we have some practical advice uh... again as a part of this experiment we did two things we'll ask people to read the fomc uh... statement and uh... you know various estimates suggest that to fully understand the statement you should have eighteen years of education this is just in a regular people in a senior some panel anyway so some of them and uh... generated bigger visions in inflation expectations down but the side of the revision is comparable to what these people will do if they just have one sentence summarizing inflation expectations in our forecast from the fomc so you don't have to have complicated messages you can just have you know one sentence summary and tell them this is what's going to happen it's going to have the same effect of inflation expectations another implication from our analysis is that we know information depreciates very quickly right and uh... typical policy announcements are one-time events right you say something and then stay silent for a long period of time our results suggest that you know if we want to successfully move inflation expectations it should be a sustained informational campaign should repeat message over and over and over again if you want to move inflation expectations so repeat repeat repeat another one uh... which is very interesting you know typically we have this intermediary link between policy makers and the general public typically conventional media right so they should they should do the translation from complicated language to something accessible to the general public we did two cents again ask people to read the fomc statement and then ask uh... people to read uh... basically the same message but in the words of usa today and what we found here was that you will say today treatment is only half as effective as a four mc statement now it may be you know fake news low credibility it doesn't matter at this point what's important is that we can't rely on conventional media to reach out to households where maybe we should use you know social media advertising something else and we have to be more creative here now this in attention is obviously a problem if you want to manage inflation expectations but it can also create uh... you know unique opportunities for us because if people know so little about monetary policy we can educate them in a particular way to our advantage for example we can differentiate messages across uh... audiences hypothetical suppose you have a currency block uh... ways you know north countries doing very well in south countries not doing very well if you have just one tool the nominal interest rate to uh... you know stabilize this block and uh... you know this uh... interest rate applies uniformly across different countries it would be really hard to resolve this regional balances now if you have a tailored message for north that is you should decelerate and the message for the south that you should accelerate you may be able to resolve some of the imbalances it's not going to be a perfect solution complete solution but would be a step in the right direction and the reason why it will work is because people in the north and the south know so little about monetary policy relatedly you know say inflation is too low instead of telling people what the past inflation was you can tell them about inflation target which is higher than the past inflation our evidence suggests that if you do this you're going to raise inflation expectations you're going to induce people to consume more you're going to induce firms to invest more you're going to generate more inflation so it should work right so we can make this communication strategy much more precise much sharper than you know just in a simple nominal interest rate now in the end our conclusion is that inflation expectations as a policy tool have a huge potential right it can help us move consumption and employment investment it can directly influence prices it can give us a new degree of flexibility that we can have with just you know nominal interest rates we can target our messages to specific areas industries various groups of population are ready to use it now answer is not yet as I told you we need to do more research to iron out various details but more importantly we have to do a lot more investment in the infrastructure the first is that we have to have much better service of inflation expectations if we want to use inflation expectations as a tool and I told you this is expensive but we can do it and another one is that we have to have a lot more investment in your communication strategies and as Larry Summers said we you know we can't afford another recession when monetary policy doesn't have any ammunition we have to do it very soon now in the end I just would like to say that Ben Bernanke once said that QE works in practice but we now should make it work in Syria here we have the opposite situation inflation expectations work in Syria now we need to make them work in practice thank you okay thank you very much thank you for having me um that inflation expectations matter for monetary policy I think is a completely uncontroversial uh statement in this audience and that most of the people in this room measure them and follow them pretty closely is also I think something that I'm very happy to state without needing further evidence what is the fact though is that when people talk about inflation expectations and measure them usually they are referring to either the expectations that come from financial prices I'm sure that tonight there will be at least one news report saying what happened to the breakeven rate or the swap rate following Mario Draghi's speech earlier this morning and or secondly we look at the service of professional forecasters there's a series of people who are economists or others and who follow relatively closely what goes on with central banking and the overall economy what this paper does though and when they say inflation expectations in some way they should put survey inflation expectations because what they say what they notice that over the last twenty years and especially over the last ten years has been an enormous amount of academic research on a different form of certations which are the ones that arise from surveys of firms and households and that those do not play as much of a role into the monetary policy process and yet we have learned a lot about how they work and that some of those lessons should perhaps be incorporated or at least discussed and so what the paper does is for the most part survey what we have learned over the last ten years about inflation expectations and so let me give you my survey of their survey of the survey of inflation expectations uh... first that so inflation expectations are consistent with a phenomenon which the theoretical literature is called inattention what we mean by that we mean that first many people are clueless they really don't know what inflation is now in the recent past and the distant past in the future they seem to pay very little attention to it second though that they're not clueless and stupid they'll be a very wrong uh... because they exhibit a lot of smart and rational behavior in other perspectives first they disagree not just about the future as consistent with say just uncertainty but also about the present it's not just that we disagree on who's going to win the world cup if i serve you we disagree also on what was the score of england tenejal last night some people paid attention some people did not third the information seems to diffuse in the population again people aren't stupid you just see that with long and variable acts use a common monetary policy statement you do see people catching up they do figure out when we have large regime changes they are sometimes even quick to catch up but that their past experiences linger in an important way they're not arbitrary but actually what you live through seems to matter third in response to news people update if you tell them if they learn about inflation they do update and moreover that in volatile environments in environments of high and volatile inflation then they can actually update very quickly as usual quip about how taxi cab drivers in argentina are often extremely good at telling you what inflation was just in the last week let alone the last few months available data either from people shopping experiences or the media affects expectations again revealing a lot of if not bayesian updating at least reflection of what news are and finally though that in if you look at certainly the last twenty years many people are unaware of who is mario dragy what is the cb or its targets but that is not to say that at the same time they're not anchored expectations and again the members of the cb should not be depressed about this last bullet point because most people also don't know who the president of the supreme court of justices in europe or of the european commission and in many ways that's fine they have other better things to worry about in the role of a policymaker is to let them focus on whatever their productive on and not draw attention to itself so the conclusion from this is that an important conclusion is that it's not the intention that we observe for the facts from the surveys is not that people don't have enough information it's not just that policy makers don't speak with enough credibility it's not just that they're simply backward looking or stupid in some ways it's simply that there's a lot of inattention as described by the set of facts second result i think from the literature in that the authors emphasize that survey expectations can be measured but it takes effort we have done it for households for many decades we have started only recently doing it for firms it is important and work the effort so it can be done in terms of the effort first it is very important surveys not to mistake disagreement for uncertainty people are concerned about the future they can disagree a lot about the present but not about the future in the academic world if you want an old tradition of say i'm gonna look at this agreement as a proxie for uncertainty is now a sure way to get your paper rejected at journal we know very well that these are very different concepts they move very differently when we measure them they're different theoretically third that the wording of inflation turns out not to be too relevant whether you ask people about the CPI course CPI general absolute level of prices they give you very consistent and similar answers you can take that as for the people in this room who obsess about the difference between these ones just as we obsessed in the beginning of the session you can take that as a as a uh... house a discouraging answer on the other hand you can say that no matter what you ask people they're being very coherent about what they think but a big concept of what is the relevance a unit of account for their measures for the design of questions matters to see the ranges but again we've made enormous progress on exactly designing questions and avoiding most psychological biases in survey design the sampling matters and it matters a lot whether you ask people about individuals versus aggregates but the conclusion is that while dcb has to be involved because it does take effort to this right it is certainly worthwhile because we are getting consistent coherent answers by people when we ask them third and related they seem to be correlated with actual economic decisions we observe and again the fact that has been documented both in the time series in the cross-section is that higher expected interest rates not so high expected inflation lead to higher willingness to spend meaning that when people expect more inflation consistent with them expecting therefore that the real interest rate will be lower they do spend more and survey after study after study has kept on finding this as Yuri noted more recent work has shown that when we look at expectations of higher inflation by firms sometimes they raise prices and wages sometimes the lower hiring and investment the facts are less solid insofar as we get different answers but again so is the theory less solid it will depend on whether high inflation was driven by demand or supply shocks whether you want to raise or lower prices raise or lower hiring and investment the households though or the firms I'm sorry at least giving you very consistent answers that is these surveys are not just noise they're not just opinions like whether whoever you think England Tunisia or is going to win the World Cup but they're very informative for the choices that people are making in terms of their actual economic decisions fourth point that the authors make in their survey that policy announcements have little effect on them now this is harder it's a very hard question to answer why first given the inattention that I told you that one is documented at best only some people are paying attention when you make a policy announcement and that's okay but still you have to really target those people to see the effects of policy announcements given the slow diffusion of information you have to look at the low frequency you have to see whether policy amounts in effect people's expectations in the next week really have to look over a few months quarters if not even years third given the correlation between speeches and actions most speeches announce actions disentangling what was the effect of change in interest rate versus talking about the interest rate or announcing something is hard and fourthly the reverse causality is makes the identification very hard because many speeches are in response to indeed changes in expectations and other events they make it hard as a result it is very hard to see whether policy amounts have a little effect on them or large effect so what does urian colleagues do in some regards they look at essentially plots of event studies that is around some announcements of policy what did happen to the expectations some of their events are really announcements of qe which makes the distinction between policy and announcements very difficult so instead I'm going to focus on the announcement by the two percent inflation target on the left panel you have on the red line what happened to expected inflation for financial markets and on the right the one from professional forecasters and the paper concludes from looking at this that policy announces have some effect on expectations i find it a little bit hard to see meaning if i try to abstract from where i drew the vertical line and draw it somewhere else find it a little bit hard to see that there was a big change following the announcement but perhaps what they emphasize more is that when they look instead of households they say we see very very little that is we don't see much of a change where as we see it before again the evidence here comes from plotting at these and when i look at these pictures versus these pictures i just don't see much of a difference in neither it turns out to be very hard to see where there's much of announcement moreover the frequency here on the horizontal axis has to do with months week or in the case of the financial markets days because we can measure them when you go to the households you have to measure in quarters that's usually where the data comes from months if you plot the two at the same frequency that is you look at the financial markets household inflation expectations it is just very hard to see i think a difference between policy amounts not having an effect that's differential between professionals or financial markets it's just a difficult problem because when you do the low frequency it's just very hard to draw a line and see a big effect it's very hard to know maybe they have a little effect on them i just don't know from looking at higher frequencies i do know though from other work that i look at lower frequencies and i look at large policy regime changes i do seem to see an effect when i look at the valkyries inflation i plot for you from the michigan survey the distribution of inflation expectations what i see is that starting in seventy nine there was the nice bell shaped curve around a high level of inflation and when i play the picture forward all the way to nineteen eighty two where we then end up again with a nice bell shaped line bell shaped curve around low inflation i see that in between you see precisely the low and slow dissemination of information by modality arising in the distribution of expectations consistent with the change in policy regime having had an effect likewise when one looks at the announcement of inflation targeting and looks in this picture at the spread of opinions among professionals unfortunately we don't have evidence for this on households and in comparison non-inflation tariff inflation target we see that inflation target or nothing else really shrinks the amount of disagreement that there exists in a very very significant way and likewise when one looks compares inflation targeting adopters versus not solid line versus the dash line it looks at what extent people make forecast errors that persist one says that within inflation targetters when there's inflation target regime we see that forecasters are much less seriously correlated forecast errors consistent with a lot more anchoring and therefore this matter so policy announces was that this inflation inflation targeting does seem to in many ways i think be effective in communicated to the server respondents fifth the author's note better communication can break in attention here's the evidence that one can bring to that first that if i tell you pass or target inflation that changes their expectation this should not be very surprising if i ask you what's inflationation now or maybe i should ask you that should ask you a different question what's the population of portugal right now you will tell me a number maybe fifteen million then i'll say actually it's twenty million or actually it's ten million no matter which one i say but if i told you ten i'd be surprised if after i ask you what's what is your advice expectation population of portugal i told you ten you would say well then maybe twelve and a half actually or if i told you twenty seventeen and a half if you tell people they will update in that certain direction secondly that once you tell them what inflation was if you then add f m c statements on news reports that adds very little again but that's consistent with the fact that if i lag inflation is a pretty good predict lag or target inflation are pretty good predictors of what inflation is f m c adds a little bit of the decimals to a point that people may not extremely relevant and third that i for reveal information to you again in a survey just told you the publisher portugal turns out to be it's a little shy of ten million that most likely in six months you're gonna forget about that because lots of other things would have happened portugal's population is not the most important thing for you and as a result in six months you just i ask you again and again you tell me fifteen million even though it's a little shy of ten million now from these three pieces of evidence the authors conclude that one should target message to the scenario simple message do better repeat the message and take the message directly to the target evidence all these are very sensible i agree with all of them but the three pieces of evidence that i reported i don't think really support these as being very solid conclusions as opposed to again very sensible things that are consistent with some of the evidence maybe absolutely i think we should target message staff simple messages repeat them but we have evidence that any of these really lead to better communication i don't i just don't think i have the evidence right now to support that on the paper yes in attention yes we can measure them yes they're correlated the pulse in us have little effect on them at the high frequency yes at the low frequency i think no subject to a lot of uncertainty and this better communication breaking attention i think is an open question as of now now aside from what is in the paper from this main point of the paper the authors also go after this provocative question in the title is peter said in introduction let me extend it to justice and the question is our household or firms survey and important to add the contribution it's also in firm survey not so much patience are they a policy tool so let me try to give you in my remaining couple of minutes my answer to this question there's a pedantic answer which is either a of course are not a tool to an endogenous object and not something that you set of course they are something that a policymaker cares about the inflation targeting regime and even if it was a two percent you observe that houses and firms all expect inflation to be accelerating to five ten or twenty percent in a very consistent way you better be worried and i know no central bank that would be worried so for sure they are an intermediate target in the language of what central bankers often use what i think is the more concrete question is rather if you look at a precise exogenous tool speeches by the members of the governing council uh... and when looks at their effect by measuring like you reset using the surveys that we now have developed and we're at your current and well done and what considers that one has to control to understand the effect of the speeches one has to control that is have this being independent of other policies interest rate qe or other policies and finally known the transmission from expected to actual inflation with ultimately with inflation what we really want to know is whether mario drage speech this morning see had an effect on inflation ultimately the target and that's going to be a result of understanding a defective expectations on actual inflation what a lot of mac modern macro is but importantly for the question that the authors asked to expand the speeches affect expected inflation conditional not other policy actions the real question is on what this second object is and this is where i think the effort or that more if you want precise way of stating the question in the previous slide is to be made that is to what extent can speeches affect expectations controlling for other policies and you mean it or beyond answering that question which is should be a is a certainly a priority for research comes the fact that even if we find that it does affect whenever i told you about the difficulty in measuring in the first place from the low frequency from the identification others is this because the communication the speech is telling us something about future policy peace in the future is because it's telling us about what inflation actually is or is it because it's just moving them independently the first one would be just communicating what you will do if what you'll do matters if setting interest rate today matters surely telling you how interest rate i'm gonna set interest rate in a few months should matter as well secondly is to the extent of the central bank is a long staff of people that study economics very deeply in the conditions they better have some information produce some information that may be relevant such that revealing it may make a difference or is a third is really about animal spirits works that i can make people believe and therefore shape their expectations now the evidence for looking for guidance and policy dates using crucially financial markets points to a lot of the speeches that mario dragy and his colleagues around the room make are a lot about revealing fundamental information and some about revealing future policy when it comes to professional fork to start to the households in the firms that's a much more open question like you're saying but i think a very first order one when you speak and people listen are are you is it because you're telling them some about the economy future policy or not what about spirits and that i think is but a dangerous and so i want to mention interpretation of uh... of the question in this paper to what extent can i just announce inflation target to be four percent instead of two and in doing so have inflation magically jump to four now the evidence we have for my pre-inflation says that absolutely not when i pre-inflation's run and i just come and say don't worry if there's going to be low i'll take care of it nothing happens i pre-inflation continues only i do fundamental policy changes that often involve fiscal monitor coordination can i actually do something about it second we fall into the danger of propaganda do to what extent it is very dangerous and certainly dealing with politicians as many of you do show shows you the clear evidence of that is to what extent is that that it's very time for a politician to say look if you just make a speech everything will be solved and nothing will happen without actually having to do anything about it uh... but again outside of some situations are multiple to remove the reality and some of which involve central banks like financial crisis and runs i think that's a very dangerous attitude to have and one that very very very often fails as we see through blasting of propaganda towards us there's two more issues one how far away that is when we talk about the speeches today at tea affecting expected inflation we're talking about expected inflation in the very on the long horizon or short horizon into are we learning about permanent versus transitory components of inflation where many ways we'd like to think about the permanent components on those two a little bit of evidence is that from the paper it is true that households inflation expectations move a lot in response to gasoline prices and others but of course they move in response to transitory price changes and they also move transitorily in response to those price changes so even though it matters a lot those gasless price change in some in the same way that central banks have learned to in many ways ignore these changes in focusing more on court inflation guiding policy also ignoring the inflation expectations responses to those may end up being important when it comes to the transfer versus permanent it's important to note that over the last ten years there's been an enormous success of anchoring inflation expectations which doesn't have not moved much when we look at longer permanent expectations in spite of a lot of variability of inflation it was very useful for the ECB and central banks that people were inattentive and so when inflation fell below zero in 2009 and 10 that people didn't revise inflation expectations but kind of didn't pay attention and thought that inflation was still around too making it easier for the comeback to happen whereas when it comes to fine-tuning what we see from vector order regressions here the response of inflation expectations to monetary policy shocks is that indeed you have these very slow and delayed responses to the shocks so it's my answer to the question oh sorry and so it's my answer to the question of inflation expectations as a policy tool again in the awareness of the authors first that moving animal spirits is hard and experience with big changes is shown that not often effective whereas communicating fundamental future policy is essentially unavoidable it is unavoidable that the central bank would not share the information it has with the public being itself a public information that houses and firms focus on the transitory but the key is that the central banks must extract the signal on the permanent and to do so looking at more moments meaning second moments and distributions and who's paying attention and some others becomes important third that inattention makes anchoring easier but fine-tuning harder precisely because people are not paying attention it's very hard to shift them in high frequency but it's good in low frequency but on that and to finish compare this let me let me try to finish with an analogy compare this with food labeling more important than whether inflation is one percent zero percent three or four for my personal well-being is how many calories or sugars I ingest during this year in terms of my happiness my well-being and my health in many ways as opposed to plus one or two percent of inflation effect that has on my real wage or real investments now if I want to ask the survey of very smart people how many calories you ingested yesterday at dinner my guess is that I would find that most of you are clueless one two there would be a lot of disagreement on how many calories or sugar you ingested yesterday and more if I asked you to forecast how much sugar or calories will be consumed by the Portuguese population over the next 12 months then you'd be very uncertain have very little clue even though this has enormous implications for health policy now that is you are very inattentive about something that's very important to you now that does not mean that we do not invest enormous resources in telling people how bad it is to ingest so much sugars that we not impose a very costly food labeling criteria in terms of how many calories are in different meals of different types and these are not very important to anchor choices and they do not have an effect on people's dieting choices in a low-frequency important uh... important welfare way at the same time no one in the health field will tell you that we should fine tune those messages so that people eat more calories in the winter and less in the summer as they should as is optimal to do but that is not to say that because you would fail at doing so mean thinking about anchoring low-frequency expectations taking account people in attention is not important and so do banks probably have a lot to learn from the way we've done these food campaigns and something that's very important for people so monetary policy in the comes to the stable unit of account using expectations has been a great success as a stabilizing policy is very hard and perhaps unachievable thank you thank you floor is open questions both said the michigan expectations of consumer expectations are very much influenced by oil and they are short-term expectation one-year horizon right it's not already the horizon and it's not clear what the horizon is but it is very like one year right so it's five years but what you should well one or five years but i mean it's very likely that uh... you know consumers would look at what goes in their pockets and so they look at oil which is a very big component of price volatility so a suggestion to look at your uh... to russian attention think that why don't you do announcement that with all the announcement and see whether they move because you know you would separate the you know the inattention story from the old story and i think that would be the obvious thing to try yes first of all i would like to thank rikardo for his discussion was very very nice uh... going back to look at his question uh... so in the michigan sort of consumers we have expectations of different horizons and you know one of the striking uh... results in this series that if you think inflation is going to be five percent next year you take inflation is going to be uh... five percent in five years from now right so there is a very high correlation between short-term expectations and large-term inflation expectations uh... now we didn't do this a pack experiment but this would be something we can easily do you know once in a short mention briefly is that in two thousand fourteen when oil prices collapsed you can see a massive decrease in inflation expectations so people paid attention also not in this paper but in another project with solico beyond we looked at inflation expectations as a function of how much people spend on gasoline well what we found was that the more you spend on gasoline the more sensitive you become to changes in prices of gasoline oil can i add something so lucrezia people have done this indeed just because urie didn't urie has a very nice paper dollar corbina 2012 doing this the way you need to do it is though not look at announcements in an event steady way because of the slow lags but put in the vr as in as a shock and so you know you'll announce exactly these oil announcements oil announcements for instance the way looks killin has built the series and see what happens to first and second moments of of expectations what you find is i showed a picture of that from the more recent paper by melosi there but since you find this very slow uh... response with this disagreement that initially increases and declines consistent against some people figuring out and others catching up slowly so it takes a while and this is from michigan survey in the vr with the shocks in the old prices just like i think you would do the idea that central banks should embark on manipulation my instincts extremely negative extremely negative but it's not only my instincts you have shown that in some cases uh... prices didn't rise after uh... such announcements italy i think was a man so cynically one could say after six months uh... households have forgotten uh... but if they also they have forgotten probably the message uh... on which they reacted have they also forgotten that's uh... what they experienced is different from what's uh... they were thinking when they are increased their purchases of goods and my understanding is that you mentioned only in a side remark that's reaching households that people sort of say uh... central bank should use social media and social media uh... i think work only not the easy be you are any central bank could use social media it must be personalized so should president drage produced with us like trump uh... so uh... but but concerned the key issue is for me is what concerned banks win if they use this media and even for manipulating uh... and what is the risk for the credibility good answers of many questions in the room yeah so manipulation is is the tricky word i agree it gives you a very negative connotation that you know you treat people into doing something uh... i use manipulation in the statistical sense that is not causation without manipulation this was the meaning i didn't want to say that the central bank should give false information to households of firms only truth that's that's reaction number one about communication uh... you know one of the messages we have in the paper is that central bank should learn from from the politicians you know trump speaks in very simple sentences to reach out the wide audience in the u.s. right so we can do the same we don't have to have a complicated statements it doesn't have to be personalized but you know one example of how successful social media could be trump has tens of millions of subscribers to his twitter account i don't know the exact number for the cb but i assume it's a small fraction of what trump has so if we really want to move inflation expectations you have to make a big investment in this now in terms of credibility again if you tell people truth uh... i don't think you're under money credibility if you tell people my inflation target is two percent you don't like you know inflation expectations you just execute what you know uh... what you were mandated to do it's a question it's a question for the names in general the question for professor karenichenko and shaby evid alfolk from el paiz spain if manipulation expectations in this case is so nice and much better than accurate information shouldn't central banks replace their communication directorates and information desks by many manipulator teams that would that be consistent with the democratic values of advanced economies very quick and it's not for me to decide uh... there are the people in the room who should make this decision well mario mario kate it's a question i find it irresistible to uh... just say a couple of words uh... first uh... i think to agree without mar uh... you don't measure credibility of central bank by the fact that it tells the truth you measure it by the fact that it's able to achieve its target so uh... the communication if the matter is to tell the truth shouldn't be the central bank to communicate about inflation expectations should be some statistical office so it's not so much a central banks task central bank makes policy based on the current available information and it's reading of it so it's different things i think yes i agree you know statistical offices can report this information to the public and public should know it but the available evidence suggests that most people pay no attention to what is provided by statistical offices and so if you want to convince people that we know we're successful at uh... meeting our inflation target and so on i believe it's the job of the central bank to convey this information to the public also i don't see an inconsistency between you know trial telling people what they should think about inflation and achieving an inflation target because you know if inflation is too low our job is to generate more inflation we look for means to do this uh... one way to do it is lower interest rates and communicated to the public that's one way to do it another way to do it is to engage in a deeper in a more substantive communication strategies to move their inflation expectations and so they can react to this policies as well but you see that we don't agree match that's why for example we always refused the idea of changing inflation target with the view to change inflation expectations either way by the way the discipline here so next question next question no here first no no first he was before here and then you thank you i'm uh... philip lowe from the reserve bank of australia i really enjoy the presentation so thank you uh... models are really clear here if you change inflation expectations to change inflation and that's really an important part of our intellectual book uh... heritage that we carry when we talk about big changes in regime that's obviously really important shifting inflation expectations but i really wonder in the world that we're living in now with inflation between one and three percent in most countries we can get much more from shifting inflation expectations and i say that because when most firms are setting prices the aggregate rate of inflation is way way down the list of things they're worrying about the worry about competition their costs the market power they have most firms don't even know what the current rate of inflation is they can get it within one or two percentage points they're doing they're doing well and the same is true consumer level if you're telling consumers inflation is going to be two-and-a-half rather than two or one-and-a-half really do you think that changes their behavior i wouldn't even know how to adjust my spare optimal spending plan if someone she could get rate of inflation by half of a cent that doesn't mean we don't we shouldn't try and educate people because i think we should we should do that but to think that we can get a really big policy bang manipulating or kind of engaging in some type of greater communication i'm really very skeptical there's one area that i i think we can make progress and i've been trying this in australia and that's by trying to influence wage expectations because the aggregate rate of inflation both past and expected is often the critical variable into wage discussions either if it's centralized or sometimes at the kind of the bilateral level people really focus on what the current rate of inflation is there so i i think that you can influence inflation outcomes through influencing wage expectations and that's a better way of going about it at the moment i've got a public campaign to try and lift wage expectations in australia in the hope that we'll hit our inflation target so i think it's wage expectations they're really the key variable here rather than trying to manipulate inflation expectations thank you let's take the next one and then see the answers i'm erica gross in cornell and one of the things that i think is uh... we forget is remember before inflation targeting when people talked about what what was success for central bank was well price stability is when inflation is low enough that people don't take it into account in their decisions so this inattention uh... one way to take this turn this around is this is a sign of the success of central banking so far and also that uh... the very similar to to what folk just said that is it really optimal for the central bank to be spending a lot of resources on trying to uh... form expectations about something that really won't make that much difference in the quality of their lives that that really doesn't need to be factored in to their decision so much so uh... i think uh... the result says uh... i'm i'm not sure that it really works very well as a tool when you're talking about very low changes small changes in inflation maybe in a uh... in a regime where you get uh... we're talking about bigger uh... changes in inflation then it would seem to be a much more useful tool but when you talk about a half a percent uh... it's not clear you're gonna get anything maybe at this point you just declare success don't think so let me pick up a few other i see benoit and then then here very quickly so i think i'd love to say something yeah it's a benokorei cv so it's a very short remark which is very consistent with what was just said which is i mean when we when we teach macro and and students ask us what's price stability the easiest answer is just to say price stability is when people don't care about inflation right they don't care about changes in prices so what is called uh... inattention bias is just maybe just be a proof of success one question here maybe we posed in a second and i take not a batch of questions that was the most interesting paper i've really enjoyed the uh... the facts that you brought to light so one of them was that new zealand is very different from italy in the impact of inflation expectations on on the real economy now i have a theory about that which i've been preparing for a long time which has to do with households so new zealand is a very high-debt economy households of heavily in debt uh... inflation expectations higher inflation expectations actually improve their expenditure uh... their ability to to spend because it reduces real debt in italy japan germany you have households we've accumulated very large levels of liquid assets uh... have relatively low levels of debt relative to that that liquid assets high inflation uh... reduces the the value of those liquid assets and means they they spend less so the weight of savers relative to borrowers is is the key thing here so i think we really need to think very carefully about differences between countries in the way that uh... policy works thank you maybe yeah i'd like to respond uh... so there is a there is a question uh... we'll move inflation expectations by fifty basis points is it going to be effective or not i find this discussion a little puzzling if we take that move in nominal interest rates by fifty basis points is going to have any effect on the economy we must reach the same conclusions about inflation expectations because what matters is the real interest rate you move nominal interest rate i we move inflation expectations so if we all organize about moving uh... nominal interest rates by fifty basis points and we expect to have uh... some effect on the real economy we should see exactly the same symmetrical effect when we move inflation expectations the point was mentioned on wage behavior which i think is a very important point uh... in countries with indexation of wages you know people are very well aware about uh... the the index part uh... now they don't necessarily believe in in in in the uh... in the inflation which is in the index uh... but at least that they have they are very attentive to that yeah try to spot differences between these countries and others then i take uh... there is one question there so uh... yes there but maybe you answer i mean yeah i'll answer very quickly we didn't do this in new zealand or in in italy but in other countries for example in europe why there is a formal process to uh... renegotiate wages and this is tied to inflation what we see is whenever firms they uh... renegotiate their labor contracts in different periods of time when they have to renegotiate they pay a lot more attention to inflation if you're not at this stage when you renegotiate your contracts you basically don't care question there i'm a deal from stanford i think your work is not a level and for inflation expectation but also for household knowing if real inflation and uh... target of the central bank because this ever brother uh... affect not only on uh... the eulah equation but also for political outcome and for people perception of the central bank so if people knew the area how the inflation is uh... he's and uh... how was the target and why the central bank makes some policies decent and can affect some political outcome i can affect people perception of the central bank which is very relevant for the current you're in a situation in which we have different elections so i think that i'm just asking if you're thinking about broader implication of your work absolutely you know once and i don't want to mention about credibility is that in the u.s. we have the surrey about perceptions of of the inflation target right and most people have no idea about the the target of the fat so i interpret this as a low degree of credibility and we view this this paper is a suggestion that we should invest a lot more resources in energy and making sure that people know about inflation target the central bank has more credibility you have to penetrate through this wall of inattention uh... to show that you know we have successful central banks they should be trusted instead of thinking that you know kill the fat stop the fat uh... he said he is a bad institution this kind of sense so i agree with you there are huge political uh... implications but our view is that you know having stronger communication strategy should help you know to hurt you question on the left yet with love and i think this week so several panelists said that uh... households are very short memories but that's some great work moment jay and others looking at uh... the great uh... depression depression babies that's people living through those periods actually are scarred for life uh... so they do carry that kind of memory with them and uh... so of course this focus is on some more structural components of inflation expectations but they may be very important and make me kind of uh... think about the following so let's say that central banks are very successful in reaching their targets so inflation hovers around two percent in the aggregate but in the magic and survey we see some stubborn behavior on the part of many of these households and they would keep reporting four percent four percent all over the question for uri maybe rickardo wants to jump in is should central banks really care about that difference maybe the last one then we will conclude tells me close the graduate institute in jenita you're his data if i remember well showed that when inflation is low people tend to over estimate inflation to think it's much higher than what we measure uh... and what that what it means i think to me that the large number of people just disbelieve the numbers and the easily accused whoever produced the numbers of manipulating to making them uh... seem lower jim's presentation was highly disquieting this morning because uh... one way of listening to him is that these numbers are not really worth a lot so with it's another dimensions of things that we need to to worry about to try to reassure people convince people that the numbers are at least honestly produced but acknowledged at the same time that with a large standard there are whatever we mean inflation i have a very quick reaction about why we should care about expectations in the uh... of consumers first these guys account for sixty seventy percent of gdp in advanced economies right they have to be important we have to understand what those people think on a more practical level if i care about for example the philips curve and we want to stabilize the economy around some inflation number right so then uh... there is always a question whose expectations we should put in the philips curve and uh... available evidence that just that if you put professional forecasters or something back where i look in the philips curve is dead but if you use expectations of households or firms you have a lot better uh... kind of understanding between the relate for the relationship between inflation and slack in the economy so it may have a very practical implication for policy yes and a few minutes more lucrecia and ricardo yeah i just want to try to bridge the gap between our session and your session following charles i mean i think that the message uh... this morning uh... was not that uh... you know we shouldn't care about headline inflation is the message is that mostly strength the cyclical component is very small i mean that's your finding is my findings a lot of other people's finding so and for inflation expectation then the question is that if uh... if the if michigan reflects justice a very transient oil cycle then central bankers shouldn't care but if it is actually an idiosyncratic trend something that persist for a long time then maybe we should care so that that's the evidence we you should provide okay so maybe you should correct me i was under impression that oil prices are in the mocks they are no transitory it's true there is a lot of noise no i agree with you because uh... i agree with you but you were not that clear on that in the sense that there is a component that which i mean in terms of our philips car models all price should go to marginal cost okay but then what you so i mean your evidence is that there is another component of oil which uh... goes you know it's probably linked to expectation something else and this what makes the disanchoring now is this transit or or it is uh... uh... like a permanent component and so that's and what's the quantitative importance of this tool with this tool it's very briefly because luke asked i mean it goes back to the two points of my first two points in the summary first people are very attentive and i said experiences linger are exactly in mind malmandeer nagle when i said that uh... there's an enormous amount of attention meaning if i was able to change your afflictions in a permanent way along the lines of what lucrez was saying then it really persist for a very long time the transitory side was different was what Yuri was saying is that though if i just want to tell you something if that just changes this is transitory that dies away very quickly and so both facts are consistent with each other they're not going to be shut and why should we care i mean ultimately that goes to then the second bullet point in my discussion of this work which is that we keep on fine that these extinctions matter for their choices they matter for their choices they're gonna matter for the macroeconomy and so one has to care thank you uh... my my own experience with uh... households is not too good in general when i travel i ask you know about inflation and uh... people have no clue or they complain a lot about the loss of purchasing power in general you know and uh... in most countries where i go i ask questions taxi drivers of course that's the worst of all but even beyond that beyond that and uh... so now the way it works usually is market extremely so you have two two groups there you have markets and they have false holes and the market is extremely sensitive so financial conditions will react very quickly uh... and that will of course have an impact on the behavior in general uh... i think the indexation story is quite interesting people don't believe the figures usually when you ask them uh... not the unions perhaps but uh... the public in general they like it because they take it i mean what is uh... given is given but they're not like they don't believe the figures anyway Peter you should remember that you should remember that taxi drivers taxi drivers maybe the government yeah yeah yeah no i mean i mentioned it on purpose i mean uh... it's it's an information but uh... okay then that has been quite quite fascinating discussion uh... but still the impact of changing expectations via the financial market because that's usually the way it works now the real the real expected rate that's that's remains you know the the difficult part of course okay thank you very much of all was very fascinating discussion thank you and uh... we have the panel just have a look at the panel just have a look