So, the first time I was on the policy panel at Hoover, I came and I had I think one slide and a long text. And the person who was on the policy panel with me, like this one was Jim Bullard, and he had no text and all slides. And I thought, there's a revelation. This is a place where I can do that, so I'm going to follow suit. So, I'm going to talk about something that is on our minds. When I called Valerie or emailed Valerie and I said, "What do you want us to talk about?" She said, "Something that's not narrowly focused on policy, something that's big in thinking about how we do macro and monetary policy strategically." So, I'm going to talk about shocks and monetary policy, and importantly, how do assess the conventional wisdom.
So, let me fix ideas by telling you what I mean by conventional wisdom. Oh, where do I point? Okay, I'm going to need help. Oh, there it is. No, that's Now you have to go backwards. Okay, now I got to go forwards. I did it. Hallelujah. Okay. So, the conventional wisdom I'm talking about is when you get a shock to the economy, then you want to understand as policy maker whether it's a supply or demand-driven shock, because you want to understand what to look for, and then you want to understand if it's persistent or if it's temporary. And if it's temporary, the standard conventional wisdom is you look through it. And if it's persistent, well, then you react. You respond in some way, or at least you consider a respond.
So, that all works until it doesn't. And so, I'm going to do what Marvin did in, you know, many slides. I'm going to show this one slide where you can see that we had the pandemic inflation. And in the pandemic inflation, we continued to say that it was transitory, and thought it was transitory. It wasn't just we're saying it and hoping. There was a forecast that was transitory, but it wasn't correct. And so then, the subsequent run-up in inflation, which doesn't look very much, it's another shock. It's two shocks, actually. We had the tariff shock and then now the oil price shock. And so inflation picked up bit, and it doesn't look very different than some of the run-ups you saw in the Great Moderation or in the decade after the financial crisis. But in point of fact, it's got a lot more attention because the concern is is this going to turn into another post-pandemic surge or is it going to be the normal hey, look through an oil shock, look through a tariff shock.
So, how do we know? How do we assess that? That's the question. And early on in my career, I was told by Alan Greenspan and then followed along by Janet Yellen that you can't really use models to beat models when you're trying to forecast because you can use their history, but you don't know how they're going to forecast in the go-forward period. And you only know exposed if they're if they're wrong. So, then what do you do? Well, we had a chance to challenge that that logic when we went to the GFC. So, this is a picture from after the global financial crisis. And the debate at the time, if you like to read transcripts at the Fed, that and many of the policymakers who had to grapple with that are are here or usually come here. You know, the labor market question was did the enormous shock to the labor market that pushed unemployment considerably higher than we had seen in a long time? Did that result in through hysteresis or just displacement a persistent shock that meant it was secular and the Fed couldn't offset it or was it simply a long-lasting cyclical shock that we could offset?
And there was there were estimates at the time. Narayana Kocherlakota estimated that the natural rate of unemployment had actually risen to 8.9%. And John Williams, who was the research director at the time and I was on his team, we we somewhat argued about this a lot, but he said it was as high as 7% from his models. And the problem was that these were models that were predicting this. Narayana was saying you can't make construction workers into nurses, and John was saying but Laubach Williams tells you that, you know, this is it. And the And the problem is you can't settle that debate without more information. So, we started producing more information.
Now, you're not going to be able to read a word of this. I don't think reading is the point. The point is we started making what ultimately was called labor market dashboards, heat market heat maps. I've only got one part of the heat map. There's actually two, three pages of this. But, the point was then you take lots of data. You take as much information about the labor market as you can possibly take. You figure out whether it's below or above its historic norms. And if you're really thinking about this hard, which we did, you're you're putting This is a from a briefing I gave, but you can do this. Many banks did this, many Reserve Banks, the board staff, and many academics. You're actually trying to understand which indicators lead, which indicators lag. When would I know we're on a cyclical upswing? How would I know?
But, even that's dissatisfying because then you just have more data. And what would happen is the debates would come to, you know, well, I like this indicator better than that one, but we're still just guessing about which indicators are going to tell us the truth. Not guessing, but but anticipating or expecting, estimating, and arguing about that. So, the the next thing we did is we talked about and many of them did this. I'm just going to focus on this since I know it well. So, I wrote a paper with Bart Hobijn, Ayşegül Şahin, and Rob Valletta where for the JEP where we tackled this idea of we've got a lot of data, how do we discipline in models? So, the question we wanted to know is what would move the job creation curve and what would move the Beveridge curve and when they intersect, what would we learn about where the natural rate of unemployment is.
And, you know, we were able to separate it with theory into cyclical factors, structural factors, and then score them as whether they would be normally in conventional wisdom transitory or more persistent. And then we could take that to the data, and if you're interested, you can read the paper on that, and it will tell you all the different ones, but I didn't have time to show everything. But the point isn't really about this paper. The point is about you take the analysis, you take the theory, you take the conventional wisdom, and it points you to what to look for in the other sources of information you had. And ultimately, we are estimate with something closer to 5.6 for the natural rate of unemployment.
The Wait, the policy makers used this and other analysis and all the different things that that academic research and policy research at the Federal Reserve was bringing to bear to change their views. And this is a picture of how, you know, the unemployment rate evolved, but also how the CBO's interest natural rate of unemployment evolved, and how the summary of economic projections measure of U* evolved. And it they were lower than they'd originally thought, and in part that we learned that from just seeing how the economy could go, but mostly we learned it from understanding that there was not a lot of evidence to suggest that it was going to be as high as it was, and there was more evidence suggesting that people can change industries. You might not make construction workers into nurses, but construction workers can definitely do other things. And that was a lot more mobility. So those were the pieces of information, and the point in this exercise is to say that when you don't know the answer between models, and you don't know if it's persistent or transitory, you start digging for more information.
So then the question is, can we do the same thing for inflation? So let me remind you of the problem we had on inflation, which again Marvin kind of showed, but I'm going to show you in more detail using the a of economic projections. So, the same debate came up is it temporary or persistent? I'm trying not to use the word transitory, so you know that temporary means that, but I'm not using it anymore. So, we So, this is the basically what we were grappling with was inflation's coming in. If you use conventional wisdom, it's a supply shock. This before it was really clear that people weren't going to come out of their homes right after the COVID and then they were going to stay out of their homes and buy a lot of things. And they were going to be supported by accommodated monetary policy and supportive fiscal policy. But for a minute, put that to the side and say a standard supply shock you look through unless you think it's going to be persistent and I don't think anyone knew at the time, most people didn't, that this was going to be, you know, one supply chain disruption after another. But that said, here's the picture.
Headline inflation PCE inflation is the blue. The The March 2021 year-end projections for inflation, these are the year-end projections that And this is the median of the SEP. And just out of respect for other policy makers, there were other people who had different dots, but this is the median of the SEP. And then it So, what you can see is that the FOMC put in March a marker on where inflation was when we had to make the projection. And then we also And I was on the committee at the time, so I I own this too. And but we And we had a little bit of an increase. But then by September, we moved it up, but we still thought it's going to come down rapidly. And so, obviously, reality often is a forcing function for trying to figure out where you didn't see something. And so, we dug into what we didn't see and how much longer would it last.
And again, many people in the Federal Reserve system did this work. I'm just going to show you things that I know very well because I was We were doing them in San Francisco. So, we We recognized that these traditional inflation gauges they weren't giving us anything. And if you look at inflation, you're getting backward answers. And so, you're still in the forecast of how long this will last. It's very unsatisfying to try to extrapolate from what's happened. You also have inflation expectations, but again, those were moving around. It It was not as reassuring to say that medium run isn't really moving and long run staying stable. You want to make sure you're really sure of that. So, we started making a dashboard. Really looking with the material that we had learned in the GFC. Many of us still had that experience of the of the unemployment rate and trying to grapple with that. And so, we started building an inflation dashboard or heat map.
Now, I've put the whole thing up there, but what I want to start with is the the columns. So, the columns you'll notice don't look like traditional measures of inflation. They've got different pieces of analysis in them. These are ones that the San Francisco Fed did. Like trying to decompose inflation into demand-driven versus supply-driven, cyclical versus acyclical responses, sector responses. We have a momentum shock momentum index, which just asked historically, what how persistent are shocks that hit one sector versus another, and so, what should we extrapolate from that? And then, things like the vacancy to unemployment rate ratio. You know, that's not common wasn't commonly used, if you remember, before the pandemic to forecast or think about labor market tightness, but it was very material in this period. So, we didn't have this dashboard when we were doing the March SEP or the the September SEP, but we had it afterwards. And so, I put it here just to ask the question, would it have helped us? And from this red that was very present in 2021 and certainly by September of 2021, it indicates that a lot of the elements of inflation that that be persistent were flashing red or showing red. It just gives you an idea that this dashboard what it would likely have been helpful in just understanding there was more there than what was obvious from the conventional wisdom or was obvious from, you know, taking the categories of inflation and trying to figure out if used cars were going to stay high or and keep rising or or go back down. So, those are the kinds of things that I think are relevant for assessing conventional wisdom and checking our work.
So, then the final thing I'll say in terms of the slide is let's apply this to the current moment. And it turned out to be helpful, at least in my own thinking, in how I thought about the inflation shock of I'll start with tariffs. So, the conventional wisdom is tariffs are a one-off. You you raise them and then you go out you then they go they raise inflation, but they don't keep raising inflation, so it goes back down as they roll through. So, we were looking at all these different measures and many others. I'm just giving you the stylized dashboard here so that I can fit on a page, but looking at this measure this dashboard and many others and only a small fraction of that dashboard was showing red or even showing pink. And it was just checking the work of is it really going to be a look-through. And you do that and then you know, we go out and talk to businesses too and ask them about how much they're passing through, etc. But this was the part that made it easier to have more confidence, not complete confidence, but more confidence that looking through was a reasonable strategy to have while you're being watchful that something else might change.
So, now we have the oil shock on top of that and this is you know, we only have data through April. So, there's still more data to watch, but it's you're starting to see red in the areas that you'd expect. And one of the things I'm really keeping an eye on myself is not only the momentum index, but the New York Fed's global supply chain index, which which also developed during the pandemic. And if an oil price shock turns out to limit the supply chains, and then it takes you know, what we've learned about supply chains is once they get clogged up, it takes a long time to bring them back. And so that's the things that could lead to more persistent pressure on inflation. And this doesn't say you're going to have more persistent inflation. It says you're going to have more persistent inflation pressure.
So, I I offer this here and I'll conclude by saying that, you know, the the message from all of this isn't that you can do perfectly by looking at more data. The message of this, in my judgment, is that the Federal Reserve and any policy maker really needs to think about how do you resolve the the models are our benchmarks. They give us a a starting point. Conventional wisdom is critical, I think, in terms of what has history taught us. But then history has to be disciplined in what we understand from history with the incoming information. And if we do that well, and keep looking, you know, under every rock for things, and use analysis and models to discipline that, then we have a much better chance of seeing things before they get there. We won't see everything. I mean, that's the unfortunate part of being a central banking. You don't know everything until you know it. But you can know more than we would. And I think we've had a history of this. That history for me started in the GFC when we really dug in, but it's extended now to the to the inflation shocks that we face. So, thank you very much. I think I'm ready.
