Ernie Tedeschi on Output Gaps, Labor Markets, and the State of the Economy

Ernie is a Policy Economist, and the head of fiscal analysis at Evercore ISI, a macro advisory firm. He is also an occasional contributor to The Upshot section at The New York Times. Previously, Ernie was a senior advisor and an economist at the US Department of Treasury. His research interests include the federal budget, monetary policy, and labor markets. Ernie joins the show to talk about output gaps, full employment, labor markets, and the state of the economy. Specifically, Ernie and David discuss Ernie’s recent articles titled “Participation in the Hot Labor Market” and “Pay is Rising Fastest for Low Earners, One Reason? Minimum Wages.”

David Beckworth: Our guest today is Ernie Tedeschi. Ernie is a Policy Economist, and the head of fiscal analysis at Evercore ISI, a macro advisory firm. He is also an occasional contributor to The Upshot section at The New York Times. Previously, Ernie was a senior advisor and an economist at the US Department of Treasury. His research interests include the federal budget, monetary policy, and labor markets. Ernie joins us today to talk about output gaps, full employment, labor markets, and the state of the economy. Ernie, welcome to the show.

Ernie Tedeschi: Thank you, David.

Beckworth: We're glad to have you on the show, and that's particularly nice for me since this is the first time we've met in person. We've interacted a lot on Twitter, so I can actually see the face as opposed to the little picture you have up there.

Tedeschi: Nothing like Calvin, right?

Beckworth: Yeah, no. My image of you was Calvin until today. Unfortunately, our listeners can't see your face. They'll hear your voice. It's great to have you on. Like many of my guests, I love to know how did you get into economics? What's your career path that led you here?

Tedeschi: I could tell a story that's a fairy tale about how I always wanted to be an economist, and that was my life's work. The real story is, it wasn't accidental. I had a sideways path to economics. When I was in high school, I wanted to be a flight surgeon for the US Navy. When that path was X'ed off for me, when I entered undergrad, I was sure I wanted to be a lawyer. In fact, I was so sure I wanted to be a lawyer that I convinced the parents of one of my friends that they wanted her to be a lawyer. She is now a lawyer, and I'm not. She blames me to this day for talking her parents into it. I think that there were a couple of things growing up that influenced me.

Tedeschi: I think primarily public service. My dad was in the Navy, 30 years in surface warfare. He retired as a Rear Admiral. My mom was a nurse, and she spent most of my early life taking care of my sister, who was mentally and physically disabled. She dedicated her whole life to taking care of my sister, who needed 24 hour care. It was only when I went to college... My family lived in California. My sister was eligible for Medicaid, and California changed Medicaid so that it would give more funding to in-home supportive care. That allowed my sister to have 40 hour a week nursing care paid for by Medicaid. That allowed my mom to go and pick up on her career as a nurse.

Tedeschi: She completed her Bachelor's, she got her Master's. To this day, she's a research manager at a local hospital chain in San Diego, studying diabetes. Public service and public policy, I think that sense came from them early on. One other thing was, believe it or not, speech and debate in high school. The type of debate I did in high school, Lincoln-Douglas debate, is very heavy on political philosophy. You get exposed to John Stuart Mill, John Rawls, John Locke, Rousseau, Marx. I think in the process of just studying up on political philosophy to have values debates, you get exposed to a lot of economics. Then if I'm being perfectly honest, the third thing I would cite is video games growing up.

Beckworth: Really.

Tedeschi: While my friends were playing Mario and Zelda, I played those too, but my favorite video games were SimCity and Civilization. Which were a sort of phantom way of exposing me to... I mean what is SimCity, if not a microeconomic or a regional macroeconomic model that you get to interact with, and budget for, and build infrastructure all the time? That had more of an effect on me than I think I realized until my undergrad years.

Beckworth: Interesting. How did you go from in college studying economics, to the policy world where you are today?

Tedeschi: When I was in college, one of the last classes I took was like a practicum course that put you in an internship. I was a public policy major in college. I chose public policy because, I had taken some economics classes, and this is interesting, people rag on the teaching of economics 101. Economics 101 was great. It was after economics 101, when I started taking like the microeconomics core and econometrics. Microeconomics at the undergrad level, at least at Stanford where I was, was just entirely Lagrangians and solving for equations.

Tedeschi: I came out of that micro class as an undergrad really not having any appreciation or sense of what microeconomics really was. Macro was a little bit better, and then econometrics was exactly the same way. The year I took econometrics was the year of the 2000 election. Literally, the last day of the class, the professor saved his first applied example of econometrics for the last day of class. It was all matrix algebra, linear-

Beckworth: That sounds awful, as an undergrad.

Tedeschi: As an undergrad. Finally the last day, he's like, "Why don't we apply what we've learned in this course to the ballot measures in Florida, and whether Pat Buchanan got more votes than we would have expected, given other aspects?" It was like this revelation.

Beckworth: Way to turn off students.

Tedeschi: I was like, "Oh, this is what econometrics is about?" Whereas the public policy degree was much more practical, and drew on economics, politics, sociology, and was applied and showed you how to use it. Anyway, the last course I took in public policy was a practicum that landed me an internship at county government, the County of Santa Clara in California. I worked for a local County Supervisor, and that was like nitty-gritty. That was like going from the ivory tower to nitty-gritty monitoring changes in social services that real people receive, infrastructure, public safety, courts, jails, all of the actual services that local governments provide.

Tedeschi: There was exposure there that made me appreciate economics more. I think too, my wife and I got married around that time. Then for our first year of marriage, we did AmeriCorps in eastern Kentucky building houses. That made me think more about economics and development. Then after that year was done, my wife got a Fulbright. We spent a year in the former East Germany, which made me think even more about development and economics, and how that all interacts. To this day, we have the debate about which was more of a culture shock to us coming from the San Francisco Bay area. It was definitely Germany, but-

Beckworth: Eastern Kentucky wasn't far behind.

Tedeschi: Yeah, it was interesting. We had a great experience in both places. I think it's impossible to be in places like that, with such divergent economic paths from what we were used to, and not think about economics.

Beckworth: Right. This is the Appalachia part of the country?

Tedeschi: Exactly, exactly, yeah, coal country. While we were there, we were hearing the first conversations about opioids and how that epidemic was beginning. It was like, I didn't even know what opioids were. People were talking about family members that they knew getting addicted. This was in 2003, and I had never heard of that before. Yet locally, everybody knew of it as a burgeoning problem that was spreading. Yeah. I think that all of those life experiences pushed me that way.

Beckworth: Now, how did you end up at the Treasury Department?

Tedeschi: My first job in DC was at the Pew Charitable Trusts, which you can think of as like a think tank. I was doing federal fiscal policy for them. The story of how I got to Treasury was very much a story about institutional constraints. Pew is a great place. They hire great people. They have a lot of money. They do a lot of good work. But you could write a whole book about the institutional and political constraints of family foundations. When your family name is on a foundation, whatever scrutiny an academic might give to a paper that's going out, a family foundation is going to double or triple that scrutiny.

Tedeschi: Pew was very much against doing anything in the written language that was controversial, or could be misconstrued. This often ended up being an issue with third party scholars that we would contract with. Their work would go through a gauntlet of red ink and several different rounds of edits, because Pew was very, very low risk, risk averse. Conservative with a lower case C about what it put out. What that did, is we realized that while written reports got scrutinized that way, charts and data visualization often didn't. If we could say something with charts rather than words, that was a much easier approval process internally in the building.

Tedeschi: Right after the Budget Control Act of 2011 was passed, that was the first time that people had heard the word sequester. They were wondering, well what is this sequester? What are the implications of this? I said to my boss, "Why don't we just do 10 charts in a chart deck about the sequester?" My recollection is that was one of the first visualizations and analyses of the fiscal implications of what they were talking about as part of the Budget Control Act. We released it, and from what I'm told second-hand, Secretary Geithner, Secretary of the Treasury at the time, saw this and said, "We need to be able to do that sort of thing in-house." Then the Chief Economist, Jan Eberly, said, "Well, why don't we just hire the guy that did this?

Beckworth: Oh really, wow.

Tedeschi: Jan basically cold called me and was like, "Do you want to come in and talk about your piece?" I said, "Sure," just thinking I was going to talk about the analysis. She ended up pitching a job to me.

Beckworth: That is an amazing story.

Tedeschi: Yeah.

Beckworth: It's usually the other way around, right?

Tedeschi: Yeah, usually, yeah.

Beckworth: You're begging for this job. Here's my great piece.

Tedeschi: Yeah. She was doing the hard sell, and I was not prepared for that, so I folded.

Beckworth: That's awesome. That's amazing. You spent some time in there. Something we were talking about before we went on the show, I thought it would be good to bring up here. You learn a lot in grad school, but when you get out into the actual policy world, there's a big difference between Lagrangians, for example, or doing sophisticated macro models. We had a previous show talking about HANK models. Doing something like that, and then actually doing policy work. Do you have any examples of that? I mean, I got plenty I could talk about. I'm sure you have some rich stories you could tell too.

Tedeschi: Yeah. The one that comes to mind is... The hardest memo I ever had to write when I was at Treasury. Technically, I worked for the Office of Economic Policy, but really I was part of a team of three. We called ourselves an analytical SWAT team. Secretary Geithner, or the Chief Economist, or the Head of Public Affairs could basically come to us and be like, "Here's an interesting question. Can you guys analyze this and write this up?" That worked really well for economic questions. How is economic growth doing? How is the labor market doing, that sort of thing.

Tedeschi: The hard questions came with the look back on the bailouts, TARP, and how to account for the extraordinary measures that Treasury and the Federal Reserve had taken during the crisis. There was no internal consensus in the building, even about how to account for those programs. You had many different sides of the debate. You had Secretary Geithner, who had opinions but also wanted buy-in from all of his different assistant secretaries and undersecretaries. The background on my phone, this was the White House picture of the day from I think 2012, or so, sometime in 2013. Obviously, the listeners can't see it. It's Secretary Geithner showing a memo to President Obama in the West Wing. I'm nowhere in the West Wing. It's unlike a lot of other Washington backgrounds. It's my memo that I wrote, that was like four months of going around the building, and basically trying to negotiate. All right, what are we going to show the President about how these different programs have done? It was literally a compromise. Coming into Treasury, I had always thought, oh, economic projections, forecasts.

Tedeschi: Coming up with point estimates of different programs, there's a right answer and there's a wrong answer, and you're either right or you're wrong. Then you come into these internal debates in a real life policy setting, and then you realize quickly, no, it really is a negotiation like any other. It's not that you're honest or dishonest. We definitely thought that the numbers were honest. But you try to get all the buy-in from the stakeholders that you can for the analysis that you're doing. That was eyeopener as well.

Beckworth: Yeah, no doubt. I saw a lot of that as well. It's not an easy yes or no question. It's a question of, how do we answer a question? Do we have buy-in, as you mentioned, from all the people above you, and other competing parties in the US government?

Tedeschi: Exactly.

Beckworth: Also, I mean I found little things, like how do you properly measure data? Do you use year-on-year, quarter-on-quarter, month-on-month, seasonally adjusted, not seasonally adjusted? Things that at least I didn't think a lot about in grad school that became a big deal when you actually had to look at a country's economy. That was a big deal. Then there's some operational mechanisms you don't think about. Something more recently in this job I've become aware of and we've had on this show, debates over the Fed's operating system.

Beckworth: I'll just give that as an example. In grad school, you talk about a macro model. What do you want to do? We talk about inflation, the price level, what type of framework you set up. Then you actually have to go do it in practice. It's a very different thing. How do you actually implement it? A lot of us don't think about those issues until we're put into the position where we have to actually implement them, or advise people on them. It's good to have someone on the show who's been through this process.

Tedeschi: Many scars.

Beckworth: Many scars to show for it.

Tedeschi: Yeah, exactly.

Beckworth: Something else that's interesting about you, Ernie, is that you have an amazing title. You didn't know I was going to bring this up, but I'm going to. You are a Fed whisperer.

Tedeschi: Oh, okay.

Beckworth: By that I mean, you had been cited twice in speeches by Chair Powell. That means he's watching you. Just like Geithner was watching and reading your work, you have Powell now reading your work. Tell us about that. How did that come about?

Tedeschi: I hope Powell likes video games then, if he reads my work. I think the ones he cited, I think he cited the disability piece that I wrote, for the-

Beckworth: Yeah, The New York Times.

Tedeschi: Yeah. I can just launch into that. One of the themes of my work is, what are the nuggets of insights that we can glean that are in these rich, wonderful, government official data sources that aren't necessarily well-publicized? In the case of this particular piece for The New York Times on disability non-employment, the story was disability non-unemployment had been rising for more than two decades. If you had stopped the clock at say mid-2014 and just looked back, it would have met just about any definition of a secular rise. Here I'm talking about prime age, that is 25 to 54-year-olds, who are not employed, and in this case not in the labor force either.

Tedeschi: Also saying they don't want a job, and the reason that they cite in the current population survey is because of disability or ill health. Been rising for more than two decades. Then it reaches this peak in 2014, and then it starts falling. It starts falling for basically the first time over at least that 25 year period. I've gone back, and I've looked at other data sources. It's robust to even looking back 50 years, that beginning in mid-2014 that rate are falling. When I wrote the piece in 2018, it had fallen by 7% from its peak. It has continued falling since then. As a matter of fact, I should probably write an update of this, the disabled non-employed are one of the biggest sources of hiring in America today.

Tedeschi: That is, people who transition out of disability non-employment are basically making up the bulk of the added employment among prime age workers today. I think that the reason why that was interesting, it was interesting to me and probably the reason why it was interesting to Chair Powell, was because the story that economists told themselves around 2013-2014 was, okay, labor force participation is low. It's really not coming back, maybe a little bit, but it's low because there has been this secular increase in people who are detached from the labor force, and they're not coming back. One of the prime examples given in that narrative was disability.

Tedeschi: Some people blamed our disability policy programs, Social Security disability insurance, SSI. Some people just blamed widespread health issues. I don't remember this term being present in 2013-2014, but I think deaths of despair is the way we might talk about it now. People who were plugged in knew about the opioid crisis. I think there's much wider appreciation. Basically for whatever reason, those people who were out due to disability were not coming back. We can talk about this more later. There were a litany of forecasts of labor force participation that came out in 2013 and 2014. They basically all agreed that labor force participation was not coming back.

Tedeschi: Obviously, that was wrong, but to be totally fair, it was wrong for the right reason. Up until mid-2014, they had no reason to think that labor force participation was going to come back, except for some theoretical belief that, if we can sustain the recovery, then it will eventually reach people on the lower rungs labor market attachment. I think that the conclusion of the piece that people who were disabled were coming back into not just the labor force but employment, I think led to the conclusion that, okay, maybe our estimates of slack in the labor market were actually underestimated this whole time. Really effectively there is more labor supply that we can bring in from a hot economy.

Beckworth: Yeah, that's an amazing story, I mean on two levels. One, it speaks to what disability is doing and providing. Maybe it's making up for a flawed social safety net system, so people are using it as a backup safety net, number one. In terms of our conversation today, it speaks to the importance of how hot can you run the economy. The Fed could have tightened sooner, could have tightened more than it has, but it didn't. Maybe in part because Jay Powell was reading your work. I'm sure there's other considerations that went into the decision. If the Fed had given up back in 2013-2014, because everyone had concluded that structurally we were at full employment, then we wouldn't have seen the change in the disability numbers. That's an important point.

Tedeschi: I think too that our sense of slack before... Slack is a very loaded term, and it's much debated. I think it's fair to say broadly that our sense of slack was a very short-term flexible sense of slack. I think that the fact that this long secular trend in increasing disabled non-employment reversed itself in 2014. It really turns that on its head. It opens up the possibility that there is longer lagged, what you might call medium-term slack. That if you can actually get the recovery sustained for a long enough period of time, then effectively you've taken something that looks structural, and maybe under some definitions was structural, and turned it into something that's cyclical.

Beckworth: Yes, I completely agree with that. Here's the analogy I've used. I'm going to use this as a segue into our next point of discussion. I've used this on the show before, and it's been some time so I'll invoke this example again. Imagine, Ernie, you're bench pressing 500 pounds.

Tedeschi: That's a lot of imagination. Okay.

Beckworth: You get sick, and you don't hit the gym for six months, a year. You go back, and you can only bench press say 200, 250, because you've lost muscle mass. You're pushing, you're exercising, and you can go in and exercise and just maintain that and be in decent shape. Your full capacity is 250. What if you pushed yourself harder? What if you ran yourself hot? You really strained, you really pushed the limits, and slowly you build up to 300, 325, 350, 450, and eventually you get back. There's a difference between your true potential of 500, your medium long-run potential, and your short-run potential. I think that's the point here. We get caught up maybe in a short-term potential level of economic activity, versus what's possible if we really pushed the limits. I think we both agree there's some point where you wouldn't want to go past that, because it would become inflationary and it would be too much stimulus. Let's move in that direction. That's the first big thing I want to talk about. What is a hot economy?

Beckworth: That's something that's very pertinent to discussions of Fed policy, as well as fiscal policy. I want to begin by talking about this from the perspective of potential real GDP. We'll come back to the labor markets we were talking about a few minutes ago. When we talk about potential real GDP, a term that often comes up is the output gap. Help my listeners understand, how do we know what potential real GDP is? What are the typical approaches to measuring it? Again, are we doing an adequate job?

Potential Real GDP in Theory and Practice

Tedeschi: That's a great question. I think probably the best way to talk about potential GDP is, you think of it as the level of GDP that you would be operating at under full capacity. Now, there are a lot of debatable qualifiers that you would add to that. I like to think about it as full capacity under current policies. I think that listeners of yours, for example, who engage in the nominal GDP level targeting debate can very easily see that potential GDP is in many ways whatever the Fed says it is. Something that's consistent with the Fed's mandate. I think it's also full capacity that's consistent with non-monetary policies as well. Obvious one is immigration. How many immigrants are we allowing into the country? Our tax policy at any given time, how much it incentivizes investment, that sort of thing.

Tedeschi: I think that we could get into the nitty gritty, and we could talk about a lot of things that go into that. Basically, I think it's full capacity output. It's measured in a lot of different ways. There are simple ways of doing it. One way that you can measure it is you can just use statistical filtering techniques over a long period of time, that basically take historical information and use that to infer what potential would be at any given point in time. The classic example of that is, you take the log of GDP or maybe GDP per capita, you can go back to 1870 with some amount of reliability. You just graph that and you draw a line, you draw a best fit line.

Tedeschi: You see that, oh, there's a more or less constant growth rate of GDP over the last 100 plus years, that were more or less mean reverting to over time. That's a good I think if not econ 101, maybe econ 102 approach to illustrating the point. There are a couple of problems with that. One is the classic policymaker problem, which is that when you're a policymaker responding to a downturn in real-time... Let's not think about the perspective of 2020. Let's think about the perspective of 2008-2009, and the fog of war ahead of us. We're not sure, we have some theories about what's going on. Obviously the financial system is a big part of the downturn, but we're not sure how bad it's going to get, what the best policies are for getting it.

Tedeschi: The issue was that those sorts of naive statistical approaches will be biased toward the downside of potential GDP at the end of their forecasting period. This is not inventing fire. This is a mathematical point. If we're at the beginning of a downturn, or at the trough of a downturn, a statistical average is going to be biased to the downside, because those are the most recent reads. That was a lot of I think, even with more sophisticated techniques at the time, techniques using HP filters or more built from the ground up techniques that I can talk about in a second. That was a common issue I think with a lot of them. Okay, we've had this massive shock to the level of real GDP. How much of that is structural, and how much of that is coming back?

Tedeschi: Of course, you did have real structural issues at the time. I mean first and foremost, you had demography. You had an aging population. My pet theory, I say this on Twitter a lot, is that demography is behind the curtain in just about everything going on right now. Even if it's not the primary factor in some things, it's lurking in the shadows behind the curtain. I think it's affecting productivity, overall growth, wage growth, labor markets, lots of different things in ways that we don't always fully appreciate. That whole process had begun around the same time as the great recession, but it was hard in real-time to dis-aggregate those things. That's one way of doing it.

Tedeschi: The other way of doing it is to work from the ground up with fundamentals. Just from an accounting perspective, growth has to be equal to productivity, plus hours, plus the employment rate, plus labor force participation, plus population growth. If we can nail down our assumptions for each of those things going forward in a well-informed manner, then we can build up potential GDP growth from that. That's basically the way the Congressional Budget Office does it. I tease them, because they basically have a monopoly on potential GDP growth estimates out. It's so funny, even at the Fed, whenever they use the unemployment gap or output gap, they seem to always use CBO numbers for that.

Beckworth: It's the benchmark.

Tedeschi: It really is. To CBO's credit in a lot of ways, because number one, they have very smart people putting it together. Two, they make it free and public, and they update it twice a year, sometimes once a year, but with some frequency. That I think is a better approach. That can help you avoid some of the pitfalls of a statistical filter, because it forces you to confront your assumptions about, let's say population growth. Whereas an IE filter might lead you to say, "Oh, population growth is going to be 2%." You look at that, and you say, "Wait a minute. No." If we're restricting immigration, or if the birth rate is down, that's an implausible assumption. I think it forces more rigor on some of those things.

Tedeschi: It also runs into one of the problems of the, not problems, but one of the issues of the statistical filtering techniques. What do we mean by full capacity? To take my original example, you draw that best fit line since 1870. That line is not going to be at the maximum of GDP in any given cycle. That's going to be straight in the middle of GDP. It's going to be literally the best fit. Roughly speaking, the average on both sides of that line will wash each other out. Is that really what we mean by full capacity? Are we past full capacity when we're above the historical trend average? I think a lot of prior potential GDP estimates have implied that.

Tedeschi: There's a whole new class of, not totally new, but it's received a lot more attention, models called plucking models. They basically say, "No, no, no, no. That's the wrong way to think about it." To oversimplify, think about potential as literally a maximum, and that deviations from potential are like plucking down on that maximum. Then like a rubber band, they bounce back, maybe not that fast, but they eventually revert to that maximum, not the mean over time. That tends to be just common sense the way I think about potential GDP. I'm not saying that's necessarily the most statistically robust or economically robust way of thinking about it, but that makes more sense to me as a way of conceptualizing it.

Beckworth: Yeah, when I look at different estimates of potential real GDP, I look at the CBO. As you mentioned, it's easy to get, everyone else references it, uses it. It's the first place you go. I've also looked at the IMF. The IMF has output gap measures for all countries, but I'm looking at the US here. The OECD has one as well. Those results from those measures tend to fit my priors better than if I do a pure statistical exercise. The one with the statistical extras like AP filter, it quickly puts you down like, oh we were at full capacity pretty soon after say 2008.

Tedeschi: Exactly.

Beckworth: Where these ones have taken longer. I was just looking at these, getting ready for the show. The CBO right now has us a little bit above full capacity, but the OECD still has us a little bit below capacity. That gets us to where I want to go with this. How do we really know if we're at a hot economy? I think your answer probably has to do more with labor markets. Is there any way to salvage that question, or to answer that question using output gap measures?

How to Identify a Hot Economy

Tedeschi: There probably is from someone more clever than I am. I have seen analysis done where they've looked at historic, they've looked at a real-time dashboard of where the economy is. They've tried to adjust for persistent errors in output gap estimates like from CBO. If CBO is persistently on the downs, or CBO persistently says that we're at an output surplus above potential, when there's not as much inflation as we might think there would be running a hot economy, then those analyses try to adjust those forecasts down. I much prefer taking a step back, and doing a reality check on where the economy is in a given time. First of all, again, the whole concept of potential GDP is policy contingent.

Tedeschi: First and foremost, it's monetary policy contingent. It is full capacity consistent with the Fed's dual mandate, one of which is price stability, which they interpret as 2% growth in the PCE deflator over time. That right there tells you something about... Regardless of any other definition, what is the policy contingent, policy consistent read on potential GDP? There I take a step back and I say, "Well, where is inflation running? Where have the errors around the Fed's inflation target run?" It's one thing if the errors are truly symmetric around the target, and we spend just as much time above 2% as we do below 2%. That definitely doesn't describe the last 10 years, or the last five years, of the last couple of years.

Tedeschi: The errors have been to the downside. As you said, I do look to the labor market. I say, "Okay, what else would we expect in a hot economy?" We might expect employment gains to be slowing to the point where they're just keeping up with population growth, or maybe labor force growth. Payroll gains have slowed down. I think on a 12 month basis they peaked in 2015 or 2016, and they've gradually slowed down since then. But they're still above the levels necessary to keep up with the labor force and with population. Then I look at wage growth, and I say, "Is wage growth consistent with past experiences, where we thought we were at a hot economy? Can we easily explain why it's not at those levels?"

Tedeschi: Wage growth has been accelerating. It's wrong to say that wages have not been growing. They're not yet at nominal levels, at least consistent with where they were back in 2000. I look at all these things and I say, "Well, okay, from a dashboard reality check standpoint, we're not checking many of the boxes of where we would expect to be in a hot economy. The way that we've defined hot economy." If we were talking about a 1% inflation target, we could have a different conversation about that. That's not where policy is right now.

Beckworth: It's useful to step back, is what you're saying, and not look at any one indicator. Look at a broad range of indicators that would give you a big picture, a broad picture of what's actually happening. What we see is not a hot economy.

Tedeschi: I think the takeaway here, and I hope I've conveyed this without saying these exact words, is that all of these estimates of the output gap are so uncertain, particularly in real-time. They're a useful benchmark for policymakers. Again, put yourself in the shoes of that policymaker facing the fog of war. Doesn't know what's happening in the future. You're presenting her or him with these real-time estimates of the output gap that may be overly sensitive to the most recent reads of the data. That's why I think particularly in the real world, when you have to make policy decisions that affect real people's lives, it's always good to take a step back and do a reality check. Are we really seeing what we think we would be seeing?

Beckworth: That's great advice. Had the Fed followed that in the 1970s. Someone who's come up on the show a lot is Orphanides. He did some work on the 1970s, and he found that if you use a Taylor rule, which a Taylor rule has as a part of it an output gap turn and an inflation turn. What happened is, he argues that the Fed was in fact following something like a Taylor rule. A lot of people disagree with this, but he found it was. The problem was the output gap measure was highly flawed in real-time, so the Fed was just giving bad information. Maybe that's one of the contributing factors to why inflation took off. That's his argument.

Beckworth: Your argument is, well if that's the case, step back and look at a wide range of indicators to get a better sense of what's going on. Just to be clear to our listeners, what we're really dancing around here is the Phillips curve. The Phillips curve gives us a relationship between slack in the economy and in inflation. More slack, we'd expect lower inflation. Less slack, the economy's running really hot, higher inflation. Personally, I'm someone who doesn't find the Phillips curve that useful. I like to think in terms of money supply, money demand.

Beckworth: That's the monetarist in me. I think it's the flip side of the same coin. On one side, you can think of a money supply, money demand relationship. The other side you can think of as a Phillips curve relationship. They should be telling similar stories. Any thoughts on Phillips curves? I mean some people like Adam Ozimek says, "Hey, they're still salvageable. Just use a wage measure or the prime employment rate." What are your thoughts on using Phillips curves for policy analysis?

Tedeschi: I think what's interesting is that the original Phillips curve from the 19, I think it was the '50s, was the original paper?

Beckworth: Yeah.

Tedeschi: Was a wage Phillips curve. It was Phillips relating unemployment in the UK to growth in the wage. I think that the wage Phillips curve is actually very robust if you make the proper adjustments for, on the one hand demography is again the big change over time. One way of showing that is by using prime age employment rates, so just people 25 to 54. This also works if you do something I like to do, which is demographically adjust the employment rates. That you keep the proportion of each age group constant over time, rather than have all people make a gradually greater share of the employment force. Then if you look at wages that are adjusted for compositional effects. The best measure for that is the employment cost index.

Tedeschi: That's wage growth, but it's wage growth that isn't affected by say some of the effects we might expect later in a recovery, where more low wage workers are coming in. When you look at a naive average, like average hourly earnings in the payrolls report, that growth rate can be brought down, because low skill workers are coming in and they're pulling down the average of wages. That's actually a good sign, that's not a bad sign. ECI adjusts for that. When you combine those two measures, ECI on the one hand and prime age employment on the other, I think the R squared is close to 90 since 1994. It's a very good, robust indicator of accelerating wage growth.

Tedeschi: The price Phillips curve of course is what the Fed pays a lot of attention to. I'm sure they pay attention to the wage Phillips curve, but the price Phillips curve gets I think a lot more press. There, I'm convinced that the evidence is that it has flattened. The effect of any change in unemployment or labor market utilization on prices has gone down over time. I don't think it's disappeared entirely. I think that when you make proper adjustments on expectations... I've seen exercises from say Olivier Blanchard, writing at the Peterson Institute, and other exercises that I think have convinced me that, no, it's still there. It's harder to see, and it's much more flat and shallow, but the effect is still there.

Tedeschi: I think a lot of what's going on is actually a parallel to what happened in the late 1990s. In the late 1990s, Greenspan was comforted by the fact that an increase in productivity was taking price pressure off of the increase in labor market utilization and the increase in wage growth. I think that we're seeing a similar phenomenon now, but not with productivity. We're seeing it with labor share. Literally, just from an accounting standpoint, again... Accounting identities are not causal models, but I think that they're a good way to conceptualize what's going on a lot of times. Wage growth literally has to be the sum of productivity, inflation, and the labor share.

Tedeschi: I look at the increase in wages that have been happening recently, and the action is not in productivity. Productivity is still very slow. Some theorize that an increase in productivity might be around the corner, because businesses are going to try to invest more in labor augmenting technologies. That could be. We haven't seen a lot of that yet. The action is certainly not in inflation, either. Inflation has been more or less steady at 2%, or just a little bit under. Most of the acceleration in wages that we've seen has been on the labor share side. Labor share could be acting as that same dampener that productivity did in the late 1990s. Now the question is, why is it doing that? What's the story there? There, we get into a lot of... The short answer is I don't know the answer. There are a lot of possibilities. There are a lot of hypotheses about the labor share that are being thrown around. One is that it's because of lower labor bargaining power. Over time, you have fewer workers and unions. You have more concentration of firms, and so they have more wage setting power. This could be yet another effect where demography is having an effect, an aging population over time.

Tedeschi: I'm not sure. There are a lot of possibilities. Whatever those possibilities are, it could be that the tight labor market is one way to counteract those effects that we've seen, particularly since 2000. Some analyses have pointed to a declining labor share since the early '70s. I think that once you adjust for things like depreciation, you don't see that as much since the '70s, but you still do see it since 2000. I'm sorry, and obviously trade openness I think is maybe-

Beckworth: Played a big role.

Tedeschi: Exactly, may also be having an effect there too. Yeah, so that could be explaining part of why the Phillips curve has been dampened over time.

Beckworth: Yes, the Phillips curve hasn't been the most useful, at least the price Phillips curve hasn't been the most useful for the Fed. We can still look at these labor market indicators, as you've said. I want to move to your work on this, because you've written for The New York Times. Again, we mentioned Jay Powell has read these pieces, and cited them.

Beckworth: I want to motivate your articles by recalling an infamous, and I stress the word infamous, piece written in 2014, a Brookings paper by some Fed staffers. The title was “Labor Force Participation, Recent Developments and Future Prospects”. In it, they concluded that this was it. We've reached the point of no more gains. They projected that the labor force participation rate might go down to 61%. Instead, it's actually risen. I believe I've read somewhere that you estimate about two million additional people have entered, if you translate the change in the labor force participation rate into actual numbers. Is that right?

Tedeschi: I think that's right, yeah.

Beckworth: About two million more jobs. That's a pretty big amount. Again, it goes back to the earlier point we made that, had the Fed given up in 2014, had the Fed concluded this is it, like the analysis suggested, this world would be a far worse place.

Tedeschi: Right.

Beckworth: You have done follow-up work, and I want to move to your New York Times piece that was titled “Participation in the Hot Labor Market”. One of the points you make in this piece is whether the change that we have seen, despite this Brookings paper prediction, is it due to people leaving the labor force or coming in. Tell us what you found.

Participation in the Hot Labor Market

Tedeschi: That particular piece was part original research that I had teed up for something else. Then part what ended up being a response to a paper from the San Francisco Fed arguing that the heat in the labor market was entirely because of a fall in labor force exit rates, not a rise in labor force entry rates. For listeners, you can decompose a change in labor force participation by a change in the rate of people coming in and the rate of people leaving. That's called the bathtub model, because it depends on the flow coming in and then the flow of water coming out, in this case.

Tedeschi: I found the conclusion from the San Francisco Fed paper provocative, because it would imply that the way the hot labor market is working is by making things so that perhaps for existing workers conditions were good, so they didn't feel like leaving the labor force. But it wasn't necessarily reaching people outside of the labor force. I wrote this piece earlier this year. I had already written the piece about disability. The San Francisco Fed piece struck me as provocative, and it also struck me as counter to what I had found in my other pieces, which is there are these-

Beckworth: The disability piece, right?

Tedeschi: Yeah, exactly. I'm sorry. There were these what we thought were tenacious margins of non-employment that were coming back. I wanted to figure out what was going on here. It was not necessarily a contradiction. It could have been that. For example, I looked at prime age workers in that disability piece. It could just be that disability had been falling because they were aging out of the prime age category.

Beckworth: Good point.

Tedeschi: Right? Part of a lot of this research is you hear smart arguments from other people, and you just go back and you double check your work, and see what more you can say. It turned out that in this case, this is actually more of a wonky data issue. The San Francisco Fed piece was based on month-to-month transition data between people going from, not in the labor force. Which for listeners that means that not only do you not have a job, but you're not looking for a job, and into the labor force. Which can mean that you're unemployed, meaning that you don't have a job but you're looking for a job, or obviously that you're employed, you have a job.

Tedeschi: When you look month-to-month, there's actually been a lot of research showing that in the current population survey, which is after all a survey of human beings and asks them what they think, the distinction between the labor force and not the labor force... When you're talking about people who don't want a job, and they're therefore not labor force participants, or do want a job and are therefore unemployed, oftentimes that distinction is lost on normal human beings. When you look at the same person over time, they switch back and forth from technically not in the labor force to in the labor force, because they're unemployed, and back again.

Tedeschi: It's an issue that we've known about for a long time. So I said, "Oh, that's interesting. What happens if we try to correct for obvious instances where somebody has a random month where they're unemployed, but really the other months they were not in the labor force, and we correct for that?” Or what happens if maybe we give them a little bit more time? So rather than looking month-to-month, we look over 12 months, and say, "All right, maybe they need 12 months to come to a new equilibrium." The conclusion I came to was, the effect of the rise in the labor force goes from being mostly due to labor force exit, to being about half and half due to labor force exit and labor force entry.

Tedeschi: I came up with a bunch of different adjusts. I made adjustments to the CPS to look at how that affected the month-to-month numbers. I looked at the unadjusted year-on-year numbers, without getting my grubby hands in the data. I did it a whole lot of different ways. The finding was robust to those changes. Basically, what I concluded is, yeah, reductions in labor force exit are an important part of the story. It's not just drawing people in, it's keeping people in once they're in. But drawing people in is also a very important part of this recent labor market story.

Beckworth: It confirms the finding in your disability story.

Tedeschi: Right.

Beckworth: That people are entering the labor force that previously were not in the labor force.

Tedeschi: Exactly.

Beckworth: Again, going back to the bigger point, is the economy hot or not? It indicates that there's still room left for more people to come in, right? There's still slack out there.

Tedeschi: Like we talked about earlier, going back to disabled workers, they're still one of the biggest sources of flows into employment, in 2019 this year. When I see people that were what we used to think of as a tenaciously disconnected margin of the labor force coming back in such large numbers that on my dashboard, I say, "Okay, that's probably a good canary in the coal mine that there is more room to run, and we have more potential if you will, to grow further."

Beckworth: Yeah, you have another article that also speaks to the strength of new entrants coming into the labor market. This is an article titled, “Pay is Rising Fastest for Low Earners, One Reason is the Minimum Wage.” We'll get to the minimum wage part, but you also cited in that article that low wage growth has been relatively strong due in part to the ongoing recovery, keeping it going. Speak to that, and then also speak to the minimum wage finding you had.

Determinants of Strong Low-Wage Growth

Tedeschi: Sure. The money chart literally in this piece is wage growth by tercile. I divided up the American population into thirds, just because I could call it highest, middle, low, I found that simple. Remarkably, this does not happen often, the wage growth among the lowest tercile, the lowest third, is now greater than it is for the other two categories of workers. All three groups have basically accelerated since about 2013-2014, but it's just been over the last couple of years that we've seen people at the low end really rise above where the other two groups are.

Tedeschi: I think that speaks to what we were talking about earlier, that when you let a recovery last long enough, you start bringing in the lowest skilled, the most marginal workers outside of the labor force and giving them opportunities to come in. This is something that Chair Powell has cited a lot now too as one of the reasons why we want this recovery to go as long as possible.

Beckworth: Where does the minimum wage play a part of the story?

Tedeschi: Minimum wage of course is like this long-lasting perennial, very spirited debate in economics. It's a great debate to have. It's an important debate to have in public policy. Going back to one of the themes in my work of trying to find interesting nuggets of insight that aren't necessarily broadcast, I think we get so caught up in the debate over whether the minimum wage is a net benefit to society, which again is a very important debate to have. Then we miss, like there are less bold but still important conclusions that we can come to about how the minimum wage is affecting the economy right now.

Tedeschi: I had actually written an earlier piece for The New York Times that just described how, even though the federal minimum wage has been stuck at $7 and 25 cents nominal for more than a decade now, state and local governments had stepped up in a historically extraordinary way. And raised many of their minimum wages to the point where the average minimum wage worker in America was facing a minimum wage of really more like $12 an hour on average. Which is a far cry from the $7 and 25 cents that you would see if you just looked at the federal minimum wage. Then for this piece I said, "Well how are those state and local minimum wage increases affecting the wage data?"

Tedeschi: I emphasized in the piece, this is not coming down on either side of the net benefit debate. By definition, only the workers that win from a minimum wage increase are counted in the wage data. But I wanted to know how it was affecting the wage data. What I basically found was that for workers in the lowest third, obviously minimum wages only affect workers in the lowest third, it was increasing their wage growth by about 80 basis points from where it otherwise would have been. That basically accounted for all of their premium over say where middle wage workers were.

Tedeschi: I'm very careful to make this distinction. While we can attribute most of the acceleration in wage growth since 2013-2014 to the hot labor market. When it comes to that little bit at the end, the premium of low wage workers over some other groups, that we can attribute to these minimum wage hikes at the state and local level. Because I'm a Fed watcher, I had to connect it to monetary policy at the end. The thing I pointed out was, we don't want minimum wage hikes to give us a false signal about the health of the labor market.

Tedeschi: Obviously in my dashboard, I would look to wage growth as one of the signs that the economy, of the health of the labor market, minimum wages weren't a giant effect on overall wage growth. When you look at overall wage growth, they are adding about 40 basis points when you look across everybody in average. That can be about a year's worth of acceleration. The point I made was, we don't want that to lead us astray, to make us think that the labor market is tighter than it really is.

Beckworth: That's a great point. You've got to be careful in the interpretation.

Tedeschi: Exactly, exactly.

Beckworth: Yeah. Well let me ask this question. I know you said earlier, you can't look at this finding as shedding light on the net effect of minimum wage.

Tedeschi: Right.

Beckworth: Still, I have to ask a question along those lines, because it's very fascinating. One of the maybe arguments for raising the minimum wage is that it's not really a binding constraint. It hasn't kept up with inflation. Even if you did raise it, you still haven't reached the marginal productivity of that minimum wage worker. You could increase it without going above what they really would cost you, or the return you'd get on that worker. I've thought that's a reasonable argument, given that many of these wages, at least the federal hasn't adjusted for inflation.

Beckworth: Now you've mentioned that the state and local adjustments have increased. The fact that they were able to pass through, might that not suggest that they're not binding yet? In other words, if the real minimum wage got to a point that it was higher than the marginal product of the minimum wage worker, at that point you think a firm might substitute into capital or find some way around it. Since we see it being translated into higher wages, could it be interpreted as a sign that the minimum wages still haven't reached the binding constraint level?

Tedeschi: Again, I would not call this conclusive, but one of the things I was worried about when I was doing this analysis... Basically the way I did it was, if you're an economist listening, it's a variation of a shift-share analysis where I keep the share of workers earning the minimum wage constant, and then I look at the change in the effective minimum wage over time. Then I use that weight to infer its effect on overall wages. One of the things I worried about in doing that calculation is, well what if minimum wage hikes do have a negative effect on employment, and the share of minimum wage workers is gradually falling over time? I kind of panicked one night. I went back and looked at my data. I did it a couple of different ways.

Tedeschi: I found out that, well number one the way I did it you would still see a smaller and smaller effect from the minimum wage over time if that hypothesis were true. Another thing is that the share of minimum wage work has been gradually rising over time. Again, not conclusive, because it could be that there is a gross downward effect on the share of minimum wage work. But that's over-washed by the amount of people you're bringing in by raising the minimum wage, and capturing all those workers. That's at least consistent with the idea that in some places, in some states and localities, the ones that probably feel comfortable raising their minimum wages, we haven't reached that threshold yet. Like I said, it's an open question where that threshold is and how far we want to go.

Beckworth: Yeah, it's a great question. I mean some places like Puerto Rico, minimum wage is too high relative to the marginal productivity and what they can get paid. It is a binding constraint there. In other places, maybe it's not.

Tedeschi: Right. I didn't write the original piece about state and local minimum wages to make this point. One thing that was often mentioned to me, and I understand this argument, is in some ways maybe it's a better policy to devolve minimum wages to the state and local government. They know what real labor market conditions are on the ground. They can make choices. I get that. I worry sometimes that when you have geographically limited minimum wages, arbitrage for employers is to, let's think about the DC area of Maryland, set a very high minimum wage. It would be very easy to cross the border and avoid it, even though you might be entirely capable of paying it by making other adjustments. Some people would say that's a feature, not a bug. I don't know what the answer is. I do see the appeal of having geographically localized minimum wages.

Beckworth: Okay, Ernie, we're near the end of the show. I have one last area I want to cover before we end. That is, what has happened to the stance of monetary policy say over the past five, six years? Has it been too tight? Has it been too loose? The reason I ask this question is because, one, the Federal Reserve has lowered its natural rate of unemployment estimates. If you go back to 2013, it's above 5%. Now, it's down to 4.1, I believe.

Beckworth: That's that measure of unemployment that they believe is the lowest you can go without generating inflation, even though we're below that. Maybe they'll adjust that down as we go forward. Even Chair Powell has admitted they've had to adjust that downward. Going back through our discussions about the output gap measures not doing a good job, is there slack or not? I want to ask that question, because I've been someone who has been critical of Fed policy. I've said they've effectively been too tight, not intentionally, but they unintentionally have kept it too tight.

Beckworth: I have a colleague here, a good friend, someone who's really smart, who's argued on the other side, Alex Tabarrok. He's a previous guest of the show. Had a great show with him about long-run economic growth. He had a blog post where he asked, "What is full employment?" He argued the other direction. He made a claim that we can't really say the Fed was too tight. Let me just read one sentence, where he uses his children's height as an analogy. This is what he said, and it's a great piece. I would encourage listeners to take a look at it.

Beckworth: He said, "My children are taller this year than last year, but that doesn't mean I could have accelerated their growth by feeding them more last year." His point is, we see unemployment continue to go down. This doesn't mean we could have stimulated the economy more in previous years. Maybe alternatively, maybe the Fed shouldn't have raised rates in previous years. What is your response to that, based on everything we've talked about today?

Recent Monetary Policy: Too Tight, or Too Loose?

Tedeschi: I agree with you. Listeners should definitely read Alex's post. I think it's an interesting argument. I was struck though by that analogy about his children. I think we know that how tall they can potentially get is affected by diet, and how much we feed them or the quality of what we feed them. I mean certainly every development metric that I've seen about lower income countries versus advanced countries has focused on average height of adults. I think that the better way to put it would be, and I'm going to put this in terms of what I think the Fed error was.

Tedeschi: What if you thought your kid was done growing, and you fed them less or fed them less quality food? Then you found out later that you were wrong? That in fact they were still growing, and through nutrition, you could have had an effect on how big they got, how tall they got. I think that summarizes what we see as the error over the last four years in Fed policy. As Alex argues, it's not that unemployment has continued falling, and that is ipso facto an error. Of course the unemployment rate, labor market utilization takes time to recover. The issue was that the Fed thought that we were at or close to full employment in late 2015.

It's not that unemployment has continued falling, and that is ipso facto an error. Of course the unemployment rate, labor market utilization takes time to recover. The issue was that the Fed thought that we were at or close to full employment in late 2015.

Tedeschi: There were Fed governors and presidents, mostly presidents, who said so in 2015, and they were wrong. They were clearly wrong. The evidence has borne them out since then. They tightened too early, and I think more and more Fed officials are admitting this publicly. I know Governor Brainard basically said this in a speech she gave in New York a couple months back, that that was premature. When the Fed makes a premature policy mistake, it can certainly affect actual employment. I think we all agree on that. Hopefully we agree that the Fed can affect actual employment.

Tedeschi: I also believe, going back to my research, talking about how hot labor markets are bringing back people that we thought were gone forever, like the disabled, I think too the evidence shows that the Fed can somewhat affect the natural rate of unemployment as well. That if the Fed sustains a recovery for long enough, then the full employment, consistent unemployment rate can fall over time as it brings more people into the labor force. I think that was the nature of the mistake made in 2015 and 2016. Which they've corrected since then, and I think admirably have faced up to, if not their mistake, then the uncertainty around where full employment is, and how far they have to go.

Beckworth: This goes back to your earlier point about a short run, full potential real GDP versus a medium, to long run. That there may be two different things. My analogy of bench pressing 250 pounds and that being your limit here and now, but you could push yourself and get back to 500. Again, there would be at some point some true limit. 500 pounds is your true limit. There's nothing you can do short of taking steroids. We know that's not the option you want to go down. I think that's a fair critique. It's one I've been making, and I think we both agree on this point.

Tedeschi: I agree with that. I would add that the nature of the Fed mistake in 2015 and 2016 was they were acting on faith that inflation would show up, and it didn't. I would hope that in the future that they set a higher bar for when they tighten policy. The other thing too is, not all people share this view. Alex may not. I don't know if you do, David. I also think in terms of the asymmetry of the errors here. Especially after we've been running inflation, where inflation has come so persistently below the Fed target for so long. I know that this is a matter of debate in the Fed's framework review, about how to make up for that in future strategies.

Tedeschi: I certainly understand for economists who grew up with the specter of inflation in the 1970s and '80s, how they are still guarding against that. That's a mistake that they don't want to make again. On the other hand, I think about in terms of the asymmetry of the risks here, if we go a little bit over on the 2% target, and we're making up for some of the past shortfalls, and that means that more people are employed. Maybe inflation gets a little bit out of the comfort zone, where it's at like maybe 2.5. I see that as a more acceptable risk than being too safe on inflation, being on the low end of say like 1.5, and not employing enough people.

Tedeschi: At the end of the day, one is a measure of prices, and I understand that we want to guard against inflationary pressure getting out of control. But on the other hand, when we look at the labor market, that's the wellbeing of families and Americans. To me, this just goes fundamentally to what I see as the point of an economy. We have an economy to deliver and distribute resources and wellbeing to actual people. At the end of the day, I would rather take the risk of going a little bit over on the inflation side, but employing more people, rather than undershooting where we could be.

At the end of the day, I would rather take the risk of going a little bit over on the inflation side, but employing more people, rather than undershooting where we could be.

Beckworth: Well said. That actually is an argument for level targeting, but I'd say it's an argument for nominal GDP level targeting where you take away that guessing game.

Tedeschi: Right.

Beckworth: You do correct for past mistakes, and you allow the chips to fall where they may based on real price level adjustments. That's a whole different discussion.

Tedeschi: There you go.

Beckworth: We are out of time. Our guest today has been Ernie Tedeschi. Ernie, thank you so much for coming on the show.

Tedeschi: David, this was great. Thank you so much for having me.

Photo by Alastair Pike