top of page

The Phillips Curve

EconBuff Podcast #15 with Andrew Keinsley

Dr. Andrew Keinsley talks with me about the Phillips Curve and the hypothesis that it no longer holds. Dr. Keinsley gives a brief description of what the Phillips Curve is and how it works. We discuss the link between unemployment and inflation, what mechanisms might be at play, and why it is important for monetary policy. We explore the various explanations others have offered for the apparent disappearance of the relationship, and Dr. Keinsley lays out why he thinks the Phillips Curve still exists. Dr. Keinsley argues his work show the relationship still exists but is more complex than traditionally though. Finally, we explore how Dr. Keinsley uses non-linearity, lags, expected inflation, and nominal wage rigidities to clear up the Phillips Curve conundrum.


Transcript

Stitzel: Hello and welcome to the Econ Buff Podcast. I'm your host, Lee Stitzel. With me today is Dr. Andrew Keinsley. Andrew is a Professor of Economics at Weber State University. Andrew, welcome.

Keinsley: Thanks for having me.

Stitzel: So our topic today is one that I've been interested in of for a while. I teach graduate macroeconomics here at West Texas A&M, and I have actually tried in the past to build a module. I teach my classes kind of a seminar-style class instead of straight out of a textbook. And I've been interested in this idea of what we call The Phillips Curve. And I, through mutual colleagues, Andrew was connected with you. And they had kind of mentioned what your work was. And I immediately said about getting you on the podcast, because this is one of my pet-macro topics.

Keinsley: Good.

Stitzel: And you actually have a paper that you're working on now that's going through the peer review process. And you were kind enough to share that with me. The --- I want the readers/the listeners to know I was very impressed with the paper.

Keinsley: Thank you.

Stitzel: It was a beautiful paper, well done, well laid out, [and] extremely well written. So it's really wonderful when you get an opportunity to read somebody's work. And it just blows you away. I can tell you guys this probably isn't the first draft that I've got here in my hands, because it's really polished.

Keinsley: [Laughs].

Stitzel: And so I want to read a paragraph out of here before we jump into this topic. So you and your co-authors write quote: “the unemployment inflation relationship described by Phillips has faced an existential crisis in recent years. This negative association, which serves as a pillar of macroeconomic theory, seems to no longer hold. Specifically both core and headline inflation have only recently settled around the targeted rate on a sustained basis, despite a 4 percentage point reduction in the unemployment rate since 2012. The accumulation of low unemployment, low inflation observations over the past few years has led many in the economics community to abandon The Phillips Curve. But what has been driving this apparent change.” That hits it right? Oh sorry, end quote. That hits it right on the nose. So I want to lay this out. But essentially what you've done in that paragraph is encapsulated what drives somebody like me, who teaches macroeconomics, to be interested in this topic. For the listeners, who may not be as familiar, I kind of want to walk them through and set this up really nice and clean. So I'm going to start off our topic today is The Phillips Curve. Andrew tell me what The Phillips Curve is.

Keinsley: Yeah. So The Phillips Curve is essentially just a an empirical representation of the relationship between unemployment and inflation. So The Original Phillips Curve, back in 1958. dealt with unemployment and wage inflation. So basically saying that as unemployment goes down, as the pool of the unemployed gets shallower and shallower, the only way you're going to be able to hire them (get them to come work for you) is if you pay higher wages. And then since then, we have we've really just pushed that to say: O.K. we have wage inflation, [so] how does that translate into, you know, price inflation (you know, what we see at the grocery store [and] what we see online)? So that's the basics/basis of The Philips Curve today --- is the relationship between the unemployment rate and inflation. And traditionally we would assume that this is a negative relationship --- that again as unemployment goes down, inflation should rise. And that's really important for policy makers, especially monetary policy. If you think about what The Fed does, The Fed has its dual mandate. It's shooting for high employment and price stability. So understanding that there might be a connection between those two is really important for The Fed, in terms of what they're going to do with interest rates, [and] what they're going to do with monetary policy going forward.

Stitzel: So I didn't realize it's intuitive, right, that your relationship between wages and unemployment --- that's something that's completely intuitive. I guess I didn't realize that it was grounded in that, [and] which in retrospect is silly of me. Can you talk a little bit about how we got from an obvious (I would say obvious and maybe it's not obvious), an obvious relationship between well, if you if your unemployment is low, then you're gonna have to pay more to attract workers. That's just straightforward supply and demand in some way.

Keinsley: Hmm mmm.

Stitzel: How do you get from that to where there's a relationship between unemployment and inflation in the economy just, sort of, broad price inflation, like you were talking about?

Keinsley: Sure. So really if we're thinking about The Original Phillips Curve --- that relationship between unemployment and wage inflation --- basically we just have to make the jump from wage inflation to price inflation. And that comes in just looking at the composition of inputs for a company, right? So you're gonna have raw materials that you're gonna bring in. You're gonna have machines that you use to make whatever you're making. And you're gonna have labor. And in a lot of industries, labor is one of the biggest costs associated with producing whatever it is you're producing. And so it's not too much of a jump to say: hey, if you have significant wage inflation within your, you know, within your company [then]/that you might have to pass some of that along to your prices as well.

Stitzel: So you're thinking that it --- so there's two ways to, sort of, view that. One is that resources when they're scarce [that] they often move together and they're, sort of, economy wide, right? We're thinking about aggregating up from individual markets --- that the scarcity of inputs move together. Or you're also saying: well, as a company we're thinking about how to structure our production and therefore cost. And a part of that is if the wage, if the scarcity of employees and potential workers increases, then that's going to also translate into the products that I'm producing being [at a] higher price. Do I have that right -- that there's, sort of, two mechanisms there? Are you focusing on one more than the other?

Keinsley: Yeah there's definitely two mechanisms there. Our focus is primarily on that second one where, you know, we can look at the scarcity of all the resources. But in this case, if we're looking at unemployment, you know, we're [focusing]/the focus is pretty much just on that scarcity of labor. And if you need to expand your business, if you need to continue to produce, [then] you're going to need to hire more people. Or on you know on the intensive margin you might, if you need to produce more, you might have to make your workers work overtime. That's automatic time and a half. That's automatically having to pay higher wages. And again, you know, that's going to hurt profits. And the only way to make up for that is to pass that along to prices.

Stitzel: So I think at some point today we'll probably talk about wage stickiness. And maybe now would be a time to, sort of, talk about what that is and for our listeners. But do you think there's a role for the idea that wages are sticky? And you're, sort of, placing the idea of wages in that cost structure of a firm as having a particularly important point --- maybe more even more important than some of the other costs that would be in a firm. And from a micro-perspective, there are reasons to think that. I'm happy to lay those out in a moment, but I want to see if you think there's any connection there. And maybe start by defining what it means for wages to be sticky.

Keinsley: Sure. So when we talk about sticky wages, or sticky prices, or sticky information, [or] any of that stuff, basically what we're talking about is that some of these variables just take time to adjust. That prices aren't going to change rapidly for everything. And especially wages -- I mean, if you think about your own personal wage (I don't know about you guys that are, you know, at West Texas A&M), but if I'm gonna get a raise, I get it once a year. And so my wage is gonna be sticky for the for the duration of a year. I get one wage bump and, you know, at the end of the year, you know, whenever the end of our fiscal year [is]. And that's it. So in terms of, you know, even though prices might be going up around me, if I go to the grocery store, you know, if I can find toilet paper [then] it's probably pretty expensive.

Stitzel: [Laughs]

Keinsley: You know, I I'm gonna have to, you know, my wages are not going to have adjusted accordingly yet. We also see a lot of wage stickiness in terms of wages going down. Wages don't go down, at least in nominal terms, very often. They, you know, it's really hard to cut people's pay. You know, if you're in a business school, you'll take those classes that talk about the, you know, the/your business environment. And you have to incubate that environment. And cutting people's pay is not exactly what you want to do if you want happy, you know, productive workers. And so we tend to see even more so wages being sticky on the downside than on the upside.

Stitzel: And so do you think the idea that I posited a moment ago --- do you think these two things are related --- that wages are sticky for the reasons that you just eloquently laid out? I'm glad that you talked about the fact that wages don't tend to fall very much. Actually on a previous episode, I talked about that idea. You're tying it obviously to the morale of workers. I won't repeat the story because I/I've told it before. But Arnold Kling has this idea that wages are often sticky because if you lower the wages the people you will lose first are the best people. But yeah. That's your experience, my experience, [and] probably everybody's experience is [that] I'm happy if I get a raise once a year.

Keinsley: Hmm mmm.

Stitzel: It definitely doesn't happen any more frequent than that. And of course in today's educational environment, sometimes happens less often than a year. But I'm never worried that they're gonna say: O.K. we gave you a 2% cut. And your idea about morale, I think, is very important there. But do you think that makes the way that firms see wages and then translate into prices --- do you think that's more important than the other inputs that they're taking there? Or is that a completely industry by industry or maybe even firm by firm case?

Keinsley: I think it's probably not firm by firm. But I would say definitely industry by industry. You know, certain firms within certain industries you're going to have kind of the same structure of production. But you're going to have industries that are more labor-intensive. You're going to have more that are capital-intensive. So, you know, I think it does it is industry by industry. And that's one of the things that in this current paper we're kind of digging into now. Part of our revision process is to kind of explore O.K. where, you know, the results that we're getting --- where are they coming from? What explains them? And the idea of, you know, labor intensity being a big key is, you know (we're still working on the analysis of it), but it's looking like more and more that labor intensity is a big driver of this.

Stitzel: So define for us what labor intensity is.

Keinsley: Sure. So essentially if you were to, you know, look at all the inputs that you need in producing whatever you're producing --- how much of those inputs is labor, right, versus the machines that they're using, versus the raw materials that you have to bring in? So, you know, more labor-intensive industries might include like textiles (or something like that) where you actually have to have people do/sitting there by hand. A lot of agricultural ventures are much more labor-intensive where you actually have to have people out there picking the crops. So yeah. So those are, you know, I think the best examples of extremely labor-intensive industries.

Stitzel: So at the risk of jumping way ahead into this episode --- the fact that you just talked about labor intensity, and you laid out this idea that, you know, textiles or agriculture --- it would be my intuition that labor intensity in those industries has dropped; and [therefore] that is one potential explanation for why we might see The Phillips Curve not work the way that we've seen it in the past? Do you, do any -- I mean obviously you're talking about it so you're thinking about it. What do you think of that notion?

Keinsley: I definitely think that that's a an interesting channel to get into. Our initial analysis, you know, it essentially just says right now that the more labor-intensive you are, the more likely you are to have to raise your prices in the face of falling unemployment. But that being said, if you have more and more industries that are becoming more capital-intensive and less reliant on individual workers, then that ([coughs] excuse me) that mechanism is definitely going to diminish.

Stitzel: Yeah. So some decades ago agriculture would have done/been done largely by hand. My father-in-law is a farmer. He has a tractor that's a (tractor's not the word) a combine [harvester] that is GPS guided to within a clearance of like a, I want to say, a foot. Don't quote me on that, but some incredibly small margin given the size of his field. And so obviously that's in a lower labor intensity higher capital intensity. If that's an overall trend in the macro-economy, [then] that has implications for The Phillips Curve? No?

Keinsley: Absolutely. Absolutely. And even then in some sort, you know, I’m originally from Indiana. So, you know, I know combines. And, you know, that's something I grew up around. You know, in some of those kind of cash crops there (where you have corn and soybeans primarily) you know, you're gonna get a lot more capital intensity out of that because we have the machines to do it. There's a lot of agriculture out there that is that is still hand-picked. If we get into, you know, especially fruits if they grow on a tree, [then] you can't runover it with a combine. [Laughs]. So, you know, it's still a big thing in part of/in some of these industries. But we are definitely making that move more towards, you know, mechanized production; which could [therefore] erode some of that wage to price mechanism, simply because wages are less of a factor for a lot of these industries.

Stitzel: So now we've laid out what The Phillips Curve is. And you've already hinted at this, but expand on why it might be important that there's a relationship between unemployment and inflation.

Keinsley: Yeah. So the big driver here, you know, the big reason why it's important is basically the policy idea. So especially if you're talking about monetary policy, you can use unemployment to kind of get an idea of where inflation is going. That's if we're using The Phillips Curve. That's what we're going to do. And this is --- if we go back to 2014-2015, interest rates were on the/at the zero lower bound. They were talking about raising interest rates, despite the fact that unemployment was still upwards of, you know, six, I think somewhere between 5-6% at that point. And, you know, one of the big reasons why they were talking about raising interest rates at that point was because unemployment continued to fall. And with this relationship, they were expecting inflation to rise. So, you know, that's so it's definitely a big driver in terms of policy; again because they have that dual mandate --- that high employment and price stability dual mandate. If those two things are connected, then that makes, you know, that makes it [an] extremely important part of their job.

Stitzel: We actually see The Fed talking about both of those things in basically everything that they do --- for obvious reasons the dual mandate that you've mentioned several times. And so you're saying that it's important, because if that relationship exists, [then] it shapes policy.

Keinsley: Exactly.

Stitzel: So tell us then --- what are some implications of a link being between those two things? Like if there were no link [and] if we thought: well, these two things just operate independently, [then] how might policy be different? How about the choices that The Federal Reserve may be different?

Keinsley: Yeah. So basically if there's no link, then we really have a hard time trying to figure out what policy to implement. At that point, we're not sure that raising interest rates is going to lead us to the goals that we want. You know, if we want price stability [then] we want inflation to be around 2%. Or if we want low unemployment, [and] if unemployment takes up, you know, will raising interest rates lower unemployment? What is the connection there? Is it more impactful on inflation? Or is it more impactful on unemployment? Essentially we see the same thing happening if you have your stereotypical, you know, aggregate supply shock. You know, a negative supply shock (in like an AD-AS model) will give you higher prices and higher unemployment; which is really, you know, kind of a predicament for monetary policy makers. Because if you raise interest rates, [then] you might bring inflation back down, but you might push unemployment up further. If you lower interest rates, you might bring unemployment back down, but you might push inflation up further. So, you know, dealing with those supply shocks is already a big issue. And if this connection between unemployment and inflation is really gone, [then] that basically puts you in a, you know, dealing with supply side shock (type of, you know, type of paradigm) all the time, because you're just not sure which way (to send, you know, which way) to push interest rates.

Stitzel: So the quote that I read from your paper talks about this, you know, talking talks about The Phillips Curve potentially no longer holding. Of course, there's some very high profile economists such as Larry Summers that are out there arguing The Phillips Curve is broken -- that there is no longer a link. How do you view this particular claim?

Keinsley: It's one of those things that if you're looking at the aggregate level, we see a lot of very shallow (if we think about this in terms of a slope) ---- like we're basically seeing very flat lines (but, you know, in terms of the relationship between these). And the reason I started getting into this paper was, you know, I kept seeing this argument and kept seeing this argument unemployment's getting down into the 4s. It's getting down into the 3s. We're not seeing inflation. The Phillips Curve must be, you know, must not exist anymore. And to me, you know, digging into this was more of a --- this channel just makes too much sense to not exist. There's got to be something behind why when we estimate this very --- you know in most literature they're estimating a very --- simple linear expression, we know that the economy is not simple. It is not linear. That there are all sorts of, you know, hoops that we have to jump through to really lock down the relationship. And so that's why we got into this paper into the first place, because I just/I did/I refused to believe that this wasn't the case. There's got to be something else driving this.

Stitzel: So I think it's important then for the listeners to, sort of, see [that] we see a claim like that (and for non-economists we might look at that) and be like: O.K., the link between unemployment and inflation is broken. And your intuition immediately --- and I'm hoping that we can talk about this here in just a moment, but I'll call it intuition for this moment --- is that just cannot be. Is that a fair summary of what you're…?

Keinsley: Yeah. I mean, I'm, you know, just like any of my old professors. I'm stuck in my ways. This is, you know, this is a relationship that, you know, that I grew up learning. And I'm like: O.K., this makes a lot of sense. And there's a lot of other relationships out there --- that theoretically we expect two variables to move in a certain way relative to each other, and they don't. So we go and we dig into: O.K., why is that? There's got to be some, sort of, outside, some other outside force, that is kind of muddying the waters in terms of this relationship. And that's what we get into with this paper.

Stitzel: This is why science exists. This is why economics exists, in particular, is we see things and we think there's something here, and yet there's something unexplained. What is that? I always tell my students that, you know, economics isn't hard --- it's just counterintuitive.

Keinsley: [Laughs].

Stitzel: And so often we see things that we can't explain. I actually use this as a teaching technique. I'll introduce topics, and I'll say: here's this relationship, and here's this behavior here, [and] something that we see, and it just doesn't make any sense. And then we use economic tools, and investigate incentives, and [investigate] different relationships. And then by the end of that module, chapter, [or] whatever it is, [then] O.K. now we can see this is actually a rational behavior or this explains that. So what I'd like you to do is talk a little bit about if it's even theoretically possible --- based on your understanding, and your past education, and even your work is it even theoretically possible --- for there to not be a link here, or we just have to look at other factors?

Keinsley: Yeah. I mean, when you look at the theoretical models (you know, the work horse models that we have), you know, they basically all break down just using simple, you know, intuitive [reasons] (you know, what we call micro-foundations which is basically this idea that: hey, people have preferences [and] they, you know, companies are trying to maximize profits). And when we when we put those [intuitive reasons] in there, you know, we always get what we call this New Keynesian Phillips Curve, which is, you know, essentially just a Phillips Curve that that's forward-looking. And, you know, and so theoretically it's really hard to kind of take that out of there, and, you know, keep it from being, you know, being a big part of the theoretical models. There have been some that have basically said: O.K., we're gonna take, you know, take this derived formula, and we're going to replace it with something that's a little bit more in line with the aggregate data that we're seeing (which is I think is a really interesting way of going into this). I think sometimes we try to rely too much on the micro-foundations. And sometimes, you know, we're not going to be able to capture everything. And sometimes we need to have a little bit more reduced form (as they call it in the model), which is basically here is a relationship. When we look at the correlations in the data, [we understand] this is what exists, [and] so that's what we're going to put in the model. And that's going to get us, you know, get us the result that's that we're seeing.

Stitzel: So I very intentionally worded the last two questions to include the word theoretical (which I think is interesting) because you kicked off whenever I asked what The Phillips Curve is; and I think you use the word empirical or observational at least. And so that is kind of the conundrum here.

Keinsley: Hmm mmm.

Stitzel: So you're saying: if tomorrow I decide I'm done being an applied microeconomist, [and] I'm gonna go write down my macro-models, and I'm gonna change the world. And I try to write my macro-model (my simple little model) to explain how the macroeconomy works. And I try to do something that doesn't include this idea of some relationship between unemployment and inflation. [You are saying] that that's just not going to be theoretically viable?

Keinsley: Yeah. You're gonna have basically --- the way that some economists remove, you know, keep that out of there is basically --- they derive the entire model with The Phillips Curve intact, [and] then they just simply remove that equation and plug in a new equation. So, you know…

Stitzel: Yes.

Keinsley: So, you know, theoretically speaking if you are, you know, working with some of the basic models that we work with right now it's --- it comes with it. It's not something that we plug in there because we think it's true. It's actually something that is derived from the model setup itself. When we work out all the math, that's what we get.

Stitzel: So we've, sort of, set [this] up from a modeling and theoretical standpoint. There's always people that want to pushback against theory. And so it's a little hard for in a podcast --- and this kind of setting for us to describe what it means to write down a macro-model --- [to] say [that] this is the relationship, and this is how these behave, and then use mathematical principles, you know, to solve that model [in order] to get conclusions to derive other equations. So I don't want to spend too much time doing that. I want the listeners to understand, you know, [that] this is your field, [and] this is something that you do, and that there is a strong theoretical link. What I want to turn to then is: there's a reason that good well-renowned economist like Summers (and he's not alone --- even you say there's some questions around this holding --- do you think the evidence is there for somebody) to make the claim that The Phillips Curve is broken?

Keinsley: Absolutely. If we're looking at (if you're just looking at) the data, and you're looking at movements and unemployment, and you're looking at movements in inflation, [then] it's not hard to sit there and go: O.K., what am I looking at? Am I --- you know, if I squint a little [then] maybe I can see a relationship coming through. But for the most part, when you look at the --- you know, you're just looking at the --- data on a spreadsheet, [then] it's hard to decipher that initial relationship.

Stitzel: So you mentioned the word “a flat relationship.”

Keinsley: Hmm mmm.

Stitzel: So [a flat relationship] --- this is the idea that the relationship between unemployment moving around, [which is] causing inflation to move around, is pretty small in the data (if you, of course, --- spoiler alert) if you use a linear relationship.

Keinsley: [Laughs].

Stitzel: And this is part of why, again, the field exists. But let's take that at face value for just for just a moment. There are some potential explanations that you don't [and] your paper doesn't advocate for. But I want to get your thoughts on them, because this is what we see in the broad economic public. If the listeners at home there ---as much as we would like them to read the Journal of Macroeconomics or something [but] they're not doing that --- they're reading The Economist. And some of the some of the things put forward in publications like The Economist would be that increased foreign competition has ended the relationship. How do you view that as a potential explanation? And maybe even briefly lay out why that [or] what mechanism that would even serve (for those of us that's not immediately obvious to).

Keinsley: Yeah. Yeah. So essentially what we might be seeing is, you know, think about like increased international competition, or even, you know, even increased competition, you know, within The U.S. If you are being more competitive, [then] that basically makes you less of a price setter and more of a price taker. So you build whatever it is, and you sell it at the market price. You don't have a whole lot of control. And at that point then, you know, fluctuations in wages you just, you simply just, might not be able to pass those along because you'll lose too much market share. You'll end up, you know, you'll end upgoing under either way. You know, you can't raise your price because you won't sell anything. And you can't pay higher wages because, you know, you'll lose those profits and you'll go under as well. So that increased competition can cause problems with that. So international (increased international) competition is definitely something that can [and] that would (you could see) [cause a] flattening out [of] that curve.

Stitzel: So I think there are two important things there. The first is: this argument only works if somehow the increased competition breaks the link between wages being driven up and our ability to pass that along to prices…

Keinsley: Hmm mmm.

Stitzel:…i.e. your original Phillips idea must still be in place. But that second part that is: well, there's a tightening labor market. Prices get driven up for labor because unemployment is very low; and yet competition essentially forces the cost structure of the firm to take on that increased cost without passing it along in the form of price to consumers. We could see that as a mechanism. But to be convinced of that, I would need to understand why. And I think there's actually an intuitive reason, but I want to get your feedback on this first. Why would it be the case that somehow the end price of the product --- we can't pass along our labor costs to it, and yet the foreign competition for the labor itself isn't a factor in the same way?

Keinsley: So I guess maybe I'm having a hard time understanding the question. But it's essentially --- my take on this is especially if you have, you know, you got to think --- that you're not only competing pricewise for what you're selling, but you're also competing in terms of, you know, you're trying to find workers wherever you can find it. So we might see a decreased relationship like that in The U.S. especially from the labor. And remember that this is, you know, this inability to pass on higher prices goes, not only just to labor, but also to capital and to intermediate inputs as well. But, you know, we see a lot of this in terms of companies leaving The United States and shifting, especially if they're labor intensive, shifting their production into other countries where the labor costs are cheaper. And so at the international level, you know, basically we might not see wage increases here or that those wage increases passing on to prices in The United States because if there are major increases in the price of those types of inputs they might source out those inputs from somewhere else where they are cheaper. This is the same idea, you know, going into back into why prices are, sorry, wages are sticky. On the downside, you know, if you look at price inflation, and you look at unemployment --- unemployment in a recession just skyrockets every time. it goes really high. Why? Because, you know, if we, you know, getting back into the workplace environment it's much easier. So I told my, you know, joke with my students. I was like: it's much better for your workplace environment to fire someone than to cut their pay and have to deal with them on a regular basis, right? And so that's why we see a lot more flexibility in terms of labor hours than we do in terms of in terms of the wages. This is simply the structure of the system.

Stitzel: So despite my having set that question poorly up, you hit it right on the head. The other thing that I want to bring in here is this idea of labor mobility. And the reason that I want to talk about that is one: because that's a pet topic of mine. Listeners to the show, and students of my graduate classes, will know this is something that I'm very interested in. And we're seeing this downward trend in labor mobility. And so one of the things that's interesting to me, right, is you're saying: well, foreign competition can come in and that potentially even tightens up the labor market. Or we can substitute away from labor in The U.S. to labor overseas. How do you view the role of: I can go and buy a product from almost anywhere. And so as an American company, you're competing with people all over the world. And yet, those firms are not competing in the same way for the labor in this particular market, right? So like, I go and I buy a laptop. I can buy one that's produced in The U.S. or the one that's overseas, right? But it's difficult for a firm that's overseas to, like, hire away my labor out of The U.S. market. And then actually we see labor mobility, even within the country, falling. What role do you see that interaction playing?

Keinsley: Yeah. So labor mobility is also --- kind of gets into something that we're looking into as a cause of this relationship. And labor mobility, and the declining labor mobility that we've seen, is really interesting in that it makes it to where even if, you know, there are jobs to be had --- the workers are going to have less bargaining power, all right? So you can start getting into like monopsonies and stuff like that, which is where you have one, you know, a single firm hiring all the labor. Well if that's the only job in town, then you take whatever price, whatever wage, sorry, that you can get from that company. And so the decreasing labor mobility keeps individuals from being able to kind of bounce around The U.S. in search of the best job for them. And this is one of the reasons why we're seeing even more like, you know, we break it down in terms of regional trends. We tend to see wage stickiness as more of a problem in your rural areas in the middle of the country; versus, you know, if you're in New York City there's/there are, you know, hundreds of companies that you can go work for doing all the same thing. So at that point, the companies within those major cities don't have as much, you know, power over your wages. You have a little bit more [power], you know. I, you know, if you don't pay me more, [then] I'm going to leave for another company who will. There's more competition for that. So yeah. So decreasing labor mobility is also something that a lot of researchers have linked to this as well. Just the sheer fact that we're not seeing (you know, we're not seeing) when unemployment goes down, we're [also] not seeing inflation rise because wages aren't rising. So it's not even that connection between wages and inflation that we're seeing break down. We're actually seeing a breakdown of the unemployment to wages relationship there.

Stitzel: And so I'm glad you laid it out the way you do, because that's exactly what I was going for on this question is that...

Keinsley: [Laughs].

Stitzel:…two-part length there.

Keinsley: Glad I could help.

Stitzel: So you brought it around and tied it together beautifully. So the question that I want to ask -- the second thing that I wanted to bring up and ask you about on this topic --- is: one of the ways I think we would test whether the foreign competition has ended the relationship is by looking at different industries. Is that a thing that the field is doing? Or maybe should they even be doing that? Talk to me about whether you think that's a good way to test this idea. And if so, is it being done? Or maybe why not?

Keinsley: Yeah. One of the big new fields in macroeconomics is essentially (and this is I'm gonna butcher some names here) like [Gabriel] Chodorow-Reich and some of his co-authors; [whereas they] have been looking at sectoral or regional data [in order] to try to find better tests of what we're seeing. So that maybe what we're seeing is just: there's a lot of noise in the aggregate data, so we can't decipher what these relationships are. So they're basically trying to break it down by sector and, you know, industrial sector (or in terms of in terms of regions around the country) and trying to basically start smaller and then aggregate your way up. And so that's a fun new field that's really emerged in the last, I'd say, five years or so especially which is, you know, trying to see ---O.K., we're looking at the data, [and] it's becoming increasingly hard, [and] we're having to come up with all kinds of new econometric tricks to try to parse out what's going on, and the answer might just be hey --- let's think a little bit smaller and then try to, you know, make implications for the aggregate level from that.

Stitzel: So before we leave this potential explanation --- do you think it has merit? Now maybe the jury is still out. But if somebody just asks you: so do you think foreign competition could have ended the relationship? Would you say: yes, no, [or] maybe? What would you say to that?

Keinsley: I definitely think foreign competition could flatten out the general relationship that you're looking at. And that's so --- in part of what we're [doing or] what we do in our papers, we also try to control for things like import prices and, you know, other things like that. So if you have other factors that are driving prices down, that increase competition [and] that increase globalization, [then] that is maybe making things cheaper [because] we're seeing more specialization around the globe. And so that could/those could be other factors. So that when we control for those within the model, we can kind of say: hey yes, there is a relationship still between unemployment and inflation. But that, you know, negative relationship is kind of being washed out by these other five factors that they go into it.

Stitzel: So that's good science, right? So the idea is [that] we observe: hey, unemployment moves around inflation, [then] unemployment goes up, and inflation ends up going down. We see that negative relationship, and but those are not the only things determining inflation. Those are not the only --- right?

Keinsley: Hmm mmm.

Stitzel: And so if we're seeing that, [then] maybe there're just other forces that are factoring in inflation and yet --- unemployment. Would the unemployment dropping, would be pushing inflation up? But we have these other factors pushing it down. And I think that is a big theme here. And it leads us perfectly to my next question is: who fights inflation? Well central banking does.

Keinsley: Hmm mmm.

Stitzel: And some people have posited -- just --- central banking has just gotten so effective that they've/they as another factor push the inflation down. And I think an extreme version of this argument, and I won't assign this to anyone because maybe nobody would go quite as far as that, [but] they've just kind of decoupled those two things. And of course I think based on what you said a minute ago, it could just be: well, there [is] more than one factor that goes into inflation. Unemployment might be putting upward pressure, but we just might have other things that are….

Keinsley: Yeah.

Stitzel:…counteracting that.

Keinsley: Yeah. And the other thing to consider here is that inflation is a very behavior driven phenomenon. A lot we see a lot of what we call self-fulfilling prophecies in inflation --- where if I think prices are going to go up by, you know, going to go up ---- well, then I'm going to go out and spend my money today [and] buy that thing, but then that increased demand actually causes the inflation. And so when we talk, you know, people bring up, you know, Olivier Blanchard and others have brought up this idea of anchored inflation expectations. And it's basically this idea that The Fed has done such a good job of keeping inflation kind of in that, you know, 1.5-2.5% window for so long that, you know, I just, I trust that they're going to keep it there. And so I'm going to keep behaving in a way that that reflects that. So if I'm a company, I can now plan ahead thinking that: O.K. well, it's going to be 2%. So now I plan to have a 2% cost of living adjustment for all my workers every year. And I just automatically pass that along to my prices. And so we basically get this idea that, you know, if you have these expectations are widespread enough, that you have everybody raising their wages by 2%. And everybody raising their weight, you know, their prices by 2% to keep up with that. And they're, in that case, you get 2% inflation without The Fed doing anything. And so, you know, the idea of anchored expectations in it has definitely been a big part of this as well. We do, like I said, that in the model we do control for inflation expectations and they get very, very stable starting in about the mid-1990s.

Stitzel: I, my parents were switching financial advisors/financial companies and there was some survey that the company had sent out; and I think trying to like get a feel for what their customers were expecting from them. And one of the questions was: what do you think inflation is going to be for the (I forget) next year, next five years, or something? And so, of course, parents know I'm an economist. And so they call me up and they say: well, what's inflation going to be for the next year [and] the next few years? And I said: well, 2% seems like a good number. And, of course, I didn't think about that very long. And that ties into, I think, your well-anchored expectation. So maybe --- even gets to us professional economists from time to time.

Keinsley: Yep. Exactly.

Stitzel: So I want to transition now, because I think we've hit on the two big ideas that other people are proposing. And I think you see some merit in them. And your work suggests some really important modifications for The Phillips Curve. And you just teased us with well-anchored expected inflation. I really would like to dive into that. But one of the things you suggest is non-linearity --- which we've mentioned. Not all of us (even somebody like me as a microeconomist knows what that means in general) may not know what that means in this context. Describe why non-linearity is important for the less technical part of our audience.

Keinsley: Sure. So at least how we use non-linearity in our paper, is this idea that when that the economy might behave differently [and] that the dynamics might be different in different environments. So in ours, we separate it out between a high unemployment environment in a low unemployment environment. And basically what we find is that, you know, this relationship kind of disappears in that high unemployment environment. That when, you know, when times are bad, you know, whether it's, you know, anchored inflation expectations --- or it's input prices, or it's, you know, import prices, [or] whatever it might be we, you know, --- we basically see that this this kind of breaks down. But when you get down into very low unemployment rates in that low unemployment environment, that this relationship really does kind of work its way out. We see this show up a lot better in the data when the economy, when the labor market is what we call tight, [and] when unemployment is particularly low. So yeah. So that non-linearity is something that we see. You know, it's one of those things that we all kind of just know. That like there's no way that everything is simple. That this is complex. That it's non-linear. That people react in different ways, in different situations. But then for most of, especially this literature, you know, the estimation is always built (is typically built) around, you know, a single linear equation, that they're trying to fit to a complex economic environment. And that's really what drove us to start this paper in the first place --- was, you know, we take all kinds of relationships and we run them through the econometric ringer trying to parse out what's going on. But for some reason, this particular model has stayed true to tradition. And yeah. So we're just kind of trying. That's one of the reasons for the title of the paper. We start with The Phillips Curve in reality. That's actually a play on Chris Sims's Nobel Prize winning paper in 1980, I think. But basically this, you know, he was the first one to start with vector auto regressions really saying: hey, instead of estimating linear equations, why don't we try to pull all this together and into a more thorough empirical investigation? And so that's kind of what we, you know, how we titled this paper is kind of a play on words there.

Stitzel: So tell us what vector auto regression models are.

Keinsley: Yes. So if you have a single linear regression, essentially a vector auto regression is going to take a bunch of, you know, take all of those variables in that single equation. And it's basically going to estimate multiple of those equations at one time (trying to get at not, you know), trying to pull in the idea of this, you know, built-in endogeneity that that one variable affects another variable, [thus] affects another variable, [and] that they're all interconnected. And so that's what that basic model does --- is it tries to estimate multiple linear regressions at one time.

Stitzel: So I want to return to what you said. The non-linearity here is this idea that the relationship between unemployment and inflation might be different in a low unemployment environment versus a high unemployment environment. And that has some a lot of intuitive sense, right? That we would see the relationship might be much stronger [and] might be much more noticeable in a low unemployment environment. What's interesting to me about that is just, sort of, as a stylized fact, right, is that we've had a lot of time periods of low unemployment. So we would expect to see the relationship in that case. Do/is my intuition correct on that front? And if so, what does that mean for your non-linearity idea?

Keinsley: Yeah. And so that's, you know, and again getting into why we're doing it. You know, I'm looking through Twitter. And, you know, analyze, you know, reading up on the conversations that other economists are having. And they're saying: hey, unemployment just keeps going down, and yet we're having a hard time getting inflation back up to its 2% target. Like so this obviously cannot, you know, still cannot hold. And so what we get into this is also the potential of say, you know, import prices falling. If we get into, you know, part of our analysis is that 2014-2015 period where oil prices fall. We add that in there as well. So there's a lot of other factors that could be kind of, you know, muddying those waters right now, in terms of that relationship. And that's essentially what we're trying to parse out here. And really the most recent phenomenon is not really new. Every time we get into a very good spot in terms of the expansion, you know, we assume that inflation is going to stay around its 2% target. And The Fed's trying to keep it around a super-set target. And at that point, we're expecting this relationship. But, you know, it doesn't necessarily showup in the data. And that might be because The Fed is not letting it show up in the data as well. So we try to try to control for, you know, everything that we could think of. And I'm sure referees in the future will come up with more things for us to think of. So…

Stitzel: So I like that intuition there which is: well yeah, low unemployment is pushing inflation up. And then The Fed is doing everything it can to fight the inflation. So yeah, we're not observing that.

Keinsley: Hmm mmm.

Stitzel: Now one comment I wanted to make --- you said: well, we're clinging to this linear relationship idea. And again, I'm not a macroeconomist and so I could be totally off base on this. But I could almost see a reason for that which is: when it's first discovered, we see this relationship. And the linear model there early on did it justice.

Keinsley: Hmm mmm.

Stitzel: That was strong and it popped out. That's why researchers found it in the first place. That's why it gained such notoriety. That's why it's in every textbook that you and I went through…

Keinsley: Yep.

Stitzel:…all the way through school. And then we observe this breakdown of it. And people, you know, they're (you said) stuck in --- we might be stuck in --- our ways, right? And so people say: well, it's a linear relationship. We have, whatever, 40 years worth of evidence that shows that. So why all of a sudden would we think it's non-linear? And it takes, you know, young up-and-coming researchers like yourself to kind of break that out. You think that has any merit?

Keinsley: Yeah I def [short for definitely] don't know if it necessarily takes young and up-and-coming researchers like myself. I do appreciate the thought that you think I'm up, you know, up and coming. But, you know, I think it just, I think sometimes we just need to take a step back and say: O.K., this we thought this was true. It no longer seems to be. Like it doesn't really pass the eye test. And maybe it's not that it doesn't hold anymore, but maybe there's something else out there. So we talked about increased globalization. So like I said in the model, we control for import prices. We also control for like the share of imports coming in terms of those intermediate inputs. And so, you know, when you think about that, you know, this basic relationship could be breaking down; because there's other things, you know, that are becoming more and more important to inflation rather than unemployment. But that doesn't necessarily mean that there's no longer an unemployment to inflation channel.

Stitzel: So when it first started out --- well let me go back and comment about one other thing. I think the history of science is that somebody new has to come along. Because when I…

Keinsley: [Laughs].

Stitzel:…develop a theory, and then I'm very old, you know, you actually mentioned this --- it's the me in my work actually --- which is you know we look at things at a level of data that's aggregated up too high. And so of course, we don't see anything because there's many other things happening at that level of aggregation. So it is a theme of my work to disaggregate --- and then to find trend at a level that you might not be able to observe at a higher level of aggregation --- because there's so many other factors, right? You mentioned that in the sectoral research, that you were talking about earlier. You know, so maybe someday I, you know, I'm an old crotchety professor hopefully (you know, crossing my fingers) and some, you know, young people are questioning the things that I came to. And it could be as simple as the environment changes, right? You just said that. Like, maybe it's the case that it was close to linear (probably never was truly linear). But maybe it was really close to linear, and there just -- there wasn't as much central banking, and it wasn't as effective, and they weren't as worried about controlling inflation quite so tightly.

Keinsley: Hmm mmm.

Stitzel: And there's fewer imports, and on and on. And we just live in a different environment. So that could be part of it. And then maybe there are things that we observe. And of course, any time we get into these kind of situations, sometimes it can look linear. But maybe it's much more complex than that. It is non-linear. it just happened to be the case of the data, that you were looking at the time, that forces cancelled out or something, right?

Keinsley: Hmm mmm.

Stitzel: And so, you know, I say that because I think it is difficult from outside academics to understand that when someone (you and I are like at the exact same point in our career, so I kind of feel like an interesting connection there)…

Keinsley: Hmm mmm.

Stitzel:…when we go through a graduate program, what's the point of the Ph.D.? The point of a Ph.D. is [that] they force-feed us literature. They equip us with lots of tools. And they say: you go find things that don't make sense [and] you go find new questions and new interesting things to do. The reason --- one of the reasons --- I love doing my job is I have papers, and I write them, and I get them published, and my peers give me feedback (you know, sometimes very harsh feedback…

Keinsley: [Laughs].

Stitzel:…because it's anonymous), and I contribute something that nobody has observed before, right? And that's what I want to do. That's what this paper does. That's why I started off this podcast saying like this: I love it when I read a paper. And I go: not only is it well done, not only is it thoughtful, you know, but it has this contribution. Like it's, you know, maybe the listeners don't, sort of, grasp that, because they don't read the kind of things that you and I read --- where it's like we're forging into this new territory. And so I think it does take young up and coming people, because we're the ones that are going: O.K. well, The Phillips Curve has been here for (what -- I'm not going to try to do the math on the fly) 60 years or whatever.

Keinsley: Hmm mmm.

Stitzel: And now all of a sudden, it looks like it breaks down. Why? Is it still there? It seems like it should still be there. What are the factors? I think it does take those kind of people. You know, and I have the advantage of, sort of, knowing what your _____ @53:19 is, so I can talk about the way that your work has progressed, in a way that the listeners are not going to be privy to. But I do want to make that, sort of, as a general point. I think this is how science works. And I think it takes people that are in a position like you and me to go through and say: you know, I'm going to look at this Phillips Curve thing. And I'm gonna say: (well, because I started off the podcast saying I'm an economist who is an applied microeconomist on the research side) I'm interested in macro things a lot, because I teach them. and because macro is just interesting. And so I see these people out there like Larry Summers. And they're saying: well, The Phillips Curve is broken. Like it's done. It's no more, right? And that doesn't ring true to somebody like me that has had to read (I don't know how many) textbooks that teach me about The Phillips Curve.

Keinsley: [Laughs]. Exactly.

Stitzel: And it doesn't ring true to you for, sort of, the same reason, [but] just a much more nuanced and detailed understanding of it. And so I'm interested in why. And of course, you're interested in why. And you're equipped to answer that question, which is why I brought you on today. So I, kind of, don't want to let that that go unnoticed, that there is some meta science there for us to talk to our listeners about. And so I appreciate that you're very humble, and I can assure the listeners that you are somebody that does good work, and is being recognized by your field. So I don't want to kind of let that slip away --- that this is how science works, right…

Keinsley: Hmm mmm.

Stitzel:…is that people like you come along and say: this is a really important theory. Let's not just shudder it at the first sign of trouble. And you're equipped with the econometric things, which I want to talk about a little bit later. The next thing that I want to talk about is (you talked about) well-anchored expected inflation. You did a good job of that. So I want you to tell us why economists might use expected inflation instead of just inflation --- which is the original relationship [and which] is just unemployment and current inflation, right?

Keinsley: Yeah. So like I said, inflation is very prone to changes in behavior. Or yeah, I'll just leave it with that --- changes in behavior. And we will change our behavior based on our expectations. You know, individuals are, you know, are forward-looking. And so like I said, if I think prices are going to go up in the future --- I have limited resources today, [then] that means that my limited resources going into tomorrow will not go as far. So if I want to take advantage of those limited resources, I need to do it today. So we talk about inflation expectations and being well-anchored, that's a key aspect of The Phillips Curve, because it really brings in that behavioral aspect that we see constantly. If we/if you take this to the extreme, you know, and you get into say hyperinflationary episodes (whether it's Venezuela today or, you know, Brazil back in the 80s, or Germany in the 1920s, you know, Zimbabwe in the 2000s) you know, all of these start off with some, sort of, you know, start off with some, sort of, bad policy --- where they're spending, you know, they're spending too much money. Things are a little bit out of control. They're not listening to some of the issues that you see out in the data. But then it becomes this self-fulfilling prophecy that I need to go, you know, --- my money's not going to be worth anything tomorrow, so I need to go --- spend money today. I need to --- whatever my paycheck was today I need to go spend it today --- spend the whole thing. Well that huge increase in demand across the entire economy --- that's going to fuel even more inflation. So we see this this self-fulfilling prophecy all the time, that just changes in my expectations can, you know, actually result in the change that I expected. And so that's why, you know, having expected inflation in your equation is important in the first place. And it's also really important in terms of say, you know, The Federal Reserve anchoring those expectations [and] making sure that people believe that you're gonna hit your 2% target. That makes it easier to conduct monetary policy, if their behavior is gonna cause it to be 2% anyway. So you basically just have to sit back and watch out for those other aspects that could throw inflation off.

Stitzel: So I'm going to ask sort of a macro 101 type question here.

Keinsley: Good work. [Laughs].

Stitzel: I think fits in really well. So how is it that we can use an expected inflation --- which is one thing for people like you and I who, you know, talk to people or do it ourselves, read people who are keeping on top of what The Fed is doing, and it's literally part of our job…

Keinsley: Hmm mmm.

Stitzel:…right? Not a lot of people do that. I guess it would be hard for us to guess a percentage of the population low enough that keeps track of what inflation actually is and what The Fed is thinking about doing with it. So how is it that you can use expected inflation when something less/way less (I would guess) than 1% of the population is even thinking about inflation?

Keinsley: Ah. That's a great question. And that is actually on, you know, the front lines of macroeconomic research today. Actually, I saw a great paper presented (not yeah maybe it was the last AEA [American Economic Association] meetings) where there's a guy at The University of Texas, and his co-author at Berkeley, are really just diving into where do households derive their, you know, their inflation expectations. And they find that if (actually one of their papers if they find that if) you include household expectations it (the model) actually does better than if you include --- you know, because we do surveys [and] The Philadelphia Fed does its Survey of Professional Forecasters and so they forecast out what inflation is going to be, and they find out that the models tend to do better if you ignore --- what the professional forecasters are saying…

Stitzel: [Laughs].

Keinsley:…(and [then just] focus on what the households are doing). Because those are the ones [and] those are the people who are actually spending [and] who are actually acting on those beliefs. And so they're really diving into right now what drives those expectations. And some of it is, you know, well, you know, and honestly, I think, a lot of it is, you know, The Fed's done a good job with keeping inflation around 2%. So, you know, it might be an adaptive -- what we call adaptive expectations --- which is where I say: O.K., what do I think is it going to be tomorrow? Well what was it yesterday?

Stitzel: Yeah.

Keinsley: O.K., it was 2% yesterday. That's probably a good enough guess. And so if we anchor those inflation expectations --- [reiterates] anchor --- and if we keep inflation within that window (around 2%), then essentially you just, kind of, get default answers from a lot of people who aren't paying attention; because they just look back and say: well, it's been 2% for 20 years (or there abouts). So, you know, why would I say it's gonna be 5% you know?

Stitzel: it's simultaneously like unnuanced and very nuanced at the same time for somebody to say: well, it's going to be 2% because it's/it is that feedback mechanism that you say. So let me get your thoughts on this as a non-macroeconomist, a self-identified non-macroeconomist.

Keinsley: We try not to hold it against you.

Stitzel: Yeah.

Keinsley: [Laughs].

Stitzel: Yeah. You --- the micro/macro divide is getting wider by the minute. What's wrong with the notion that individuals don't have to think about inflation as an aggregate? But instead, I think most people have a pretty solid sense of what the prices of things that they're actually buying are doing. Like, I don't know what the price of milk is off the top of my head, because I'm blessed to have a wife who does a lot of the grocery shopping. That's the way that we've divvied up our household responsibilities. And my guess is if I if I got her on the phone right now and I said: darling, what's the price of milk [and] that she would be able to quote it within a couple of pennies. And my wife and I are needing to get a new vehicle. And so, I know quite well what the price of the type of vehicle --- minivan, nothing sexy here…

Keinsley: [Laughs].

Stitzel:…because I have small children --- is gonna be and even, sort of, like how that's moving around. So is it not the case that as, sort of, (I don't really think it's difficult to write into your model) but as difficult as that would be to write into your model --- expected inflation is an aggregate of prices? Or sorry, inflation is an aggregate of the prices and people know what prices are? Tell me what's wrong with that.

Keinsley: Oh. Yeah. I honestly, I think I don't think there is a whole lot, you know, wrong with that notion. Just, you know, the idea that, you know, in our day-to-day basis, like, we go to the grocery store. We go, say, once a week. And we see: here's --- this is the price, this is the price, this is the price. And we might see, you know, we know when the sales are. And, you know, with inflation --- if you think about inflation at 2% annually, then, you know, we're talking about super slow creeping up of prices; so that even if you really know what's going on it you don't feel the pain on a week-to-week basis typically. You know, we've been, you know --- think about what's going on right now in terms of, you know, with the coronavirus and everything like that, you know, what we're ---- seeing, you know, as a household is essentially a supply side shock. What we're seeing [are] prices just going through the roof, because there is, you know, there's just not a lot of supply. Supply chains are having a hard time keeping up. And we notice this jump, because all of a sudden, where they don't have the generic brand that we want. So we have to buy the name brand, and that's going to be more expensive. So that's the type of stuff that you would actually feel. But on a week-to-week basis, even if they started to raise prices, you know --- prices tend to go up so slowly, that we don't even really notice it on a day-to-day basis. And as long as wages are keeping up with that, then you shouldn't, you know, honestly, you shouldn't realize it at all.

Stitzel: That's an interesting point, because when I grocery shop I very rarely price shop. My wife gave me a list and it says: milk, butter, [and] eggs. And I go get: milk, butter, [and] eggs.

Keinsley: Hmm mmm.

Stitzel: And now maybe if I got there and milk was, instead of 4 bucks, was 8 bucks --- maybe I wouldn't notice. Hopefully I would notice. My wife would be upset if I didn't.

Keinsley: [Laughs].

Stitzel: But maybe that's an…. I want to say another maybe that is --- [another] part of the self-reinforcing mechanism of expected inflation. So let's [turn]…. That is, that's fascinating. That's some fun stuff to think about. ….what I want to turn to now is [that] you say nominal wage rigidities shape The Phillips Curve. So [the] first thing I want you to do is: tell us the overlap between nominal wage rigidities and the wage stickiness that we've talked about before. And then tell us what you mean that they shape The Phillips Curve.

Keinsley: Sure. So basically, you know, like nominal wage rigidities are just, you know, go hand in hand with wage stickiness in that, you know, wages are slow to adjust. You know, whether, you know --- we can talk about just in normal times when you expect to get that that pay raise. Like, you know, it might take a year, you know. If we talk about, even we talk about contracts, you know --- when you're signing, you know, if you sign a contract with your company, or you're sending a contract as an independent contractor itself, you know, [then] you're not gonna be able to renegotiate that for some time. And that's really gonna lead to, you know, in terms of shaping The Phillips Curve, this idea that you might have a drop in unemployment but the corresponding increases in wages might be slow to adjust. So in our data, basically what we find is --- that we find about a, you know, on average (and we run all sorts of robustness checks on this thing), you know, and typically what we're finding is --- about a three to four month lag in the movement of overall inflation to unemployment. So you can assume that there's some sort of disconnect between unemployment going down --- maybe wages rising. And then there's also, maybe a company tries to hold off for some time before raising their prices because, you know, there's competition. There's, you know, all the stuff we talked about before --- and which is/which really I think speaks to the Phillips Curve not passing the eye-test recently --- because, you know, typically we get, say, you know, certain measures of inflation on a quarterly basis. Or, you know, we get unemployment on a monthly basis. Even if we get the inflation measures that we get on a monthly basis, you know, we get a decrease in unemployment today; [however], the inflation won't show up for another three to four months. And so by the time --- I mean three four months is a pretty long time for an individual. So when we look at that we're like: O.K., well obviously we're not seeing increased inflation right now; so therefore it must be broken. But basically what we find is that this, you know, these wage rigidities, this price stickiness, [and] the wage stickiness --- this can really, you know, shape how that dynamic works out; because in the traditional sense of The Phillips Curve, it's a contemporaneous relationship like it is, you know, unemployment goes down [and] today inflation goes up today. And if, you know, especially, if we talk about contract work or, you know, if we get increase increases in wage rigidity --- that can make it look like that relationship that contemporaneous relationship, yes, definitely has broken down. But what we show in this paper is that: yeah, unemployment's going down. And it just it takes some time for these things to work through the system and to show up as inflation.

Stitzel: So one of the things that I teach whenever I'm teaching, especially a principles course, (because when I learned principles macro, I wasn't taught this, and I got to grad school it became a real eye opener) is to just talk about the lags in policy formation --- for The Federal Reserve lag, to get a data lag, to recognize a pattern lag, to make a decision lag, [and then] to implement said decision. This is going to go in my lectures for this Fall. And I'm going to add a fifth lag. And the lag is going to be, you know, something happens. And then there's a lag in actually how The Phillips Curve plays out. And then there's a lag in data collection. And then there's a lag. Talk to me about the implications of this new element of The Phillips Curve adding this lag in there and saying: well, this is what it seems to be happening in reality. What are the implications of that for policy formation?

Keinsley: I definitely think it's important for policy formation, because if you understand that inflation is not going to be coming for three to four months, you know, in that situation --- you're gonna have, (you know, you could) have three more, you know, FOMC Federal Open Market Committee [meetings]. You could have three more Federal Reserve meetings until then. And it might be one of those situations where they started raising it. For instance, they started raising interest rates in 2015 because they assumed that inflation was going to come. And they're trying to, you know, be proactive about it. But basically this says that maybe that lag is a little bit longer; and if you really want to try to control inflation, you probably have a little bit more time to get to that. You don't need to look at unemployment right now and say: hey, this is going to cause inflation, you know --- just next, you know, tomorrow or --- next month. [And] that, you know, it might not be just around the corner, [but] it might be a quarter away. And so that's definitely something to think about. And in terms of The Fed policy, there's some people calling for, you know, this, you know, after this next [recovery and] during this next recovery is basically [calling to] not raise interest rates until you actually see inflation happen --- then you can start to raise interest rates and bring it back down --- because we're seeing that lag just might/it just takes a little bit longer than what we initially thought.

Stitzel: So there's two things there. One is --- that [it] actually might make inflation fighting more effective, because you have time to counteract it before it even hits.

Keinsley: Hmm mmm.

Stitzel: But of course, maybe you're right, and we should see it before, or before we fight it. And then two --- that might also explain [why] that’s sort of endogenous. That might explain part of why we don't see it. If The Fed has more time to fight it, but in that lag that would flatten out that relationship when we're looking at it linearly, right?

Keinsley: Hmm mmm.

Stitzel: So we've said this several times --- that labor markets are tight. And we talked about that in the context of unemployment. Can you tell us what that means for it to be tight, and what the role of labor markets is in driving The Phillips Curve? And I'm sort of setting you up for a question here --- which is we love to talk about unemployment, we love to talk about unemployment numbers, what is in unemployment, [and] on and on. And there are some economists out there, who I follow and admire, and say: well, unemployment has a lot of problems. You and I teach this in our principles class…

Keinsley: Hmm mmm.

Stitzel:…like all the different things [of] what it means to be searching for a job, and people that are marginally attached, and on and on --- all these different unemployment things. So talk to us about the role of labor markets driving The Phillips Curve. And then maybe is unemployment the measure the only way that you would look at this? Or are some of those other employment measures important too?

Keinsley: Yeah. So one of the things that I, you know, that me and my co-author really tried to dig into with this paper, was this idea that unemployment doesn't capture everything. That unemployment is a --- you got to think about, you know, basically every statistic that measures the economy is a human construct trying to make sense of a complex, you know, complex world. And so unemployment obviously has (you know, the basic unemployment rate) has its obvious flaws; whereas, you know, if you might want a job, but if you haven't looked for one in four weeks, then they don't even count you in in the labor force. So one of the other measurements that we look at pretty heavily in this paper, is the spread between the broadest measure of unemployment (what we call U6) and the headline measurement of unemployment (which is U3) --- and that [the U-3 measurement] includes people who are marginally attached to the labor market; which basically extends that four weeks out to a year [and] says, hey have you looked in the last year? Then O.K. yes, you are part of a part of the labor force. And it also includes (which I think is really interesting, but I think really speaks volumes when you're looking at the data it includes) those who are part-time for economic reasons basically as unemployed, you know, when we're thinking of this as an unemployment measurement. But really what I think it's getting more at is this idea of underemployed that --- you know, I would love to have a full-time job and I don't because the economy is bad. Now let's think about wage dynamics. Let's talk about bargaining in terms of being part-time, right? If you're a part-time worker, you have zero bargaining power. So, you know, what we saw in in the recovery after the financial crisis, or at least the one back in 2008 (we'll see how this one shakes out whether we give it a fancy name like that or not) but, you know, we saw that the unemployment rate --- so there were people who had jobs. The quantity of jobs was going up, but the quality of jobs wasn't always rising at the same rate. That the people who were part-time for economic reasons, you know, stayed pretty high for a lot longer period of time. And that could explain why we're not seeing, you know, increases in inflation; because a lot of companies are still relying on part-time labor who are not having to be paid [benefits]. You know, there's no insurance, there's no wage bargaining, [and] there's no anything like that. And so what we find is that in these tight labor markets, we tend to see when the number who are part-time for economic reasons comes down (so those people are getting full-time jobs), [then] all of a sudden we see, you know, that inflation relationship really start to take off. And so it adds another dynamic --- that we maybe shouldn't always be looking at the quantity of jobs, but also we need to try to parse out the quality of jobs. Because people who are better at their jobs, [then] they should be getting, you know, better pay. If they're just stuck in a part-time job because there's not, you know, [and] they can't find anything better, [then] they're not going to have that bargaining power. They're not going to be able to bid up their price/their wages, which is then going to lead to price pressure. And so we find that particular channel can be very revealing in terms of why we haven't seen inflation, you know, during this latest recovery as much as we would think. You know even though unemployment's going down, this/these part-time individuals are still around --- which means that we're not seeing a whole lot of wage growth and therefore no price growth either.

Stitzel: So when we're thinking about this problem [and] when we're thinking about any problem --- we want to think like scientists and say: where could things be changing? We spent a lot of the first part of this episode talking about prices and inflation and what the breakdown in the mechanism might be between that. Of course, what you've just gotten into very beautifully is this idea that: yeah, the way that we measure employment really matters.

Keinsley: Hmm mmm.

Stitzel: I think there's another factor there that I'd really like you to take on which is: it can be the way that we measure unemployment. And there's also this idea of the natural rate of unemployment. And people argue about whether that moves around or not. And you feel free to weigh in on that in general. But definitely weigh in on it in specific in that one of the things that could be happening is: it looks like unemployment is moving around more than it is, or less than it is, (either way) if it's relative relationship to the natural rate. So start by defining what [the] natural rate of unemployment is for people [and] why that might be important. And then [tell us] whether you think that moves around and what role that has to play in explaining The Phillips Curve.

Keinsley: Sure. So one of the things we notice when we look at measures of (you know, when we look at the data of) unemployment is that unemployment never goes to zero. We never have everybody working. And there's a few reasons for that. We generally define unemployment and, you know, as three types. We have frictional unemployment --- which is basically you're going to have people who just leave their jobs. It could be normal times. You know, they might have quit. They're looking for another job. And it's just simply that time it takes to match a person with a new job. [There’s] nothing inherently wrong with the economy, just simply that time kind of job hopping, whatever. You know, there's a variety of different things that go into that. You have structural unemployment --- which is basically this idea that the company you're working for now requires a new skill set that you don't have; so you're going to require new training before you can go back to that job. And so that particular point in time you're going to be unemployed. Again, [there’s] nothing structurally wrong with the economy. It's actually --- we tend to consider this to be a good thing that, you know, we're forcing new technologies [and] where we're, you know, forcing new skills of our, you know, to, you know, within the labor force. So when you combine those two types of unemployment, that's what's going to give you your natural rate of unemployment. And it's basically just unemployment outside of what the economy is causing. So your the third component is cyclical unemployment. We don't include that in the natural rate, because that's usually just the ebb and flow with the economy itself. So the natural rate of employment just considers, you know, if all things are good, [then] you're going to have some unemployment and this is what it is. And sorry, I think I forgot the last part of that question.

Stitzel: I was… Yeah. I set you up poorly there because I poured on a whole bunch of questions. So you've defined what natural rate of unemployment is.

Keinsley: Hmm mmm.

Stitzel: Do you think that moves around at all, or very much, or not?

Keinsley: O.K. Yeah, sorry. So our --- the thing about the natural rate of unemployment is that we can't measure it. You know, if we were to go and ask someone if they were frictionally, or structurally, or cyclically unemployed --- it's really hard to separate each individual into one specific category. So it's an estimate. We estimate what the natural rate of unemployment is. And recently we've been consistently lowering our estimate of the natural rate because unemployment keeps going down, you know, and inflation's not going up. Well maybe some people are saying: well maybe the natural rate is lower. And this idea that that The Phillips Curve, you know --- when you think about unemployment --- don't think about it in terms of its absolute level. Think about it in terms of its relationship to the natural rate. So only when we're below the natural rate of unemployment will we see inflation. And so this idea that since unemployment's continued to, kind of, trickle down without that inflation, basically is a suggestion that maybe our estimates are wrong. And we need to be revising our natural rate of unemployment down because we're not seeing that inflation. That the natural rate could be, you know, (some I've seen some estimates and it be that) it could be as low as, you know, as low as 3 or 2.8% or something like that. Like something, you know, extremely low. And that's why we're still, you know, still not seeing that those inflationary pressures.

Stitzel: So we're over an hour now. So I want to bring this in for a landing. And to do so, I want to this because we've set up The Phillips Curve. We sort of talked about the history and what that is. We talked about the fact that we're not observing it in the way that we've observed it in the past. We've put out some of what the more popular potential explanations for that and your thoughts on those. And then we turned to your paper. And I think working through your paper, we kind of did it bit by bit.

Keinsley: Hmm mmm.

Stitzel: And so to close us out today, what I'd like you to do is kind of give me a brief summary of what your paper does, and what the findings are, and if there's anything that, like, I didn't draw out with the questions. And if you want to interject there, now is the time to do it. So give us a quick summary and tie it together from the perspective of your paper, because I've broken that down into all these little pieces.

Keinsley: Sure. Sure. The elevator pitch of my paper. Sure. So, yeah. Basically what our paper does is: we look, you know, look at this idea that The Phillips Curve is broken [and] that The Phillips Curve no longer exists. And basically we say: O.K.: this --- maybe it's not that The Phillips Curve basic philosophical relationship doesn't exist anymore. Maybe it's just that there's other factors that are contributing to this. So essentially what our paper does is: [we] try to take a step back, and try to add the flexibility and the rigor of most basically of all empirical exercises today, and [then] apply it to something that's still rooted in a very traditional sense. So what we do is except we take that step back and we add you know we run it through a basic econometric model that a lot of people use it's a Jordá Projection Method [Local Projection Method (LP)] to estimate these things. We won't get into that. But, you know, what we're trying to do here is say: O.K., what has really caused this? So we add non-linearities. We separate the data between good times and bad. We say: maybe there's a difference between how the economy reacts in bad times versus when times are really good. Maybe there is a better connection there. Maybe there's --- you know, we have to take in into consideration how integrated we are globally. We have to take into consideration things like import prices. We have to take into consideration oil prices. What The Fed is doing ---we have to, you know, include interest rates in there. We have --- you know, there's just a variety of other aspects, and allowing them to be dynamic, [and] allowing, you know, these variables to be determined by where they were before is also important. So we basically just lay this out. And essentially what we find is that when you control for a lot of the other stuff that's happening out there --- that there still is this connection between unemployment and inflation --- that when unemployment goes down, inflation will rise. However it's lagged. Like we mentioned before, it's gonna be, you know, three to four months before we start to see that inflation, everything else held constant. [So] obviously, everything else looked constant, we would still expect to see a substantial lag in the data --- especially when data comes out monthly. And you're talking about a full quarter worth of, you know, worth of lag. That's, you know, that's an important aspect of that as well. We also find that that the quality of the labor market match is important. We kind of proxy that with this U-6 [and] U-3 spread --- which is looking at mostly (most of that is) part-time for economic reasons. So basically saying: hey, there's some wage bargaining issue here. There's some, you know, some issues with --- you know, even though people have a job and the unemployment rate is really low --- maybe they're not in the best jobs. Maybe they're working two part-time jobs, so even though they're technically employed, it's not really the situation we want them to be in. So [when] we're thinking about the labor market -- the health of the labor market --- this, kind of, adds another dimension into assessing the health of that labor market. And so we find that that is also fairly important as well in terms of deriving this Phillips Curve relationship. So it's not --- so not only is it non-linear, not only is it, you know, very dynamic, but we also need to broaden our view of that labor market mechanism going from just the unemployment rate to something a little bit more to really parse this out. So, yeah. That's the, you know, the gist of it.

Stitzel: My guest today has been Andrew Keinsley. Andrew, thank you for joining us on The EconBuff.

Keinsley: Well thanks for having me. This has been great.

Stitzel: Thank you for listening to this episode of the econ buff you can find all previous episodes on YouTube at EconBuff podcast you can check out our website at econbuffpodcast.wixsite.com that's wixsite dot com you can contact us econbuffpodcast@yahoo.com.

Comments


bottom of page