Economic Outlook Symposium Lunch with the Fed Experts I

SPEAKER: OK. We're going to force your lunch hour to go into the 1 o'clock hour here. And we're going to wrap it up with Dan Sullivan, who is going to be talking to us about the labor market. So I mean, we've already talked a lot about it, but I think it's an evergreen topic that we care a lot about at the Fed and that everyone who is in the room I'm sure cares a lot about.

So Dan is an executive vice president here at the bank, and he leads our Economic Mobility Project, which is a really cool project that has a lot of fascinating seminars that they put on that most of them are viewable online so that you can watch them from anywhere and learn a lot of great stuff. So definitely go look up the Economic Mobility Project if you can.

And he was also director of research for a number of years here at the bank. And so I'm just going to stop there and say, Dan, thanks for coming and for being our second great Fed expert and looking forward to hearing about the labor market.

DANIEL SULLIVAN: Thank you. It's cool to be called an expert on the labor market. Almost everybody here is probably working. So they all have a lot of expertise in the room.

And so I'm probably an expert in how the Fed thinks about measuring the labor market. I'll claim that at least. But we're very interested in getting other people's perspectives. And that's why sessions like this are really valuable.

So disclaimer, it's the usual one. I work for the Fed. I'm not Austin. I'm not Jay Powell or anybody else but myself. And so if I say something crazy, it's probably because I'm the crazy one not anyone else.

So these are definitely very interesting times for thinking about the labor market. I would say the overall characterization is that things are quite healthy, but things actually have softened. So there's different perspectives. Some people think that the economy's been running wild and the Fed raised interest rates. And it didn't do anything.

I want to say that's not right. I think that there's a measure of softness that's emerged, at least relative to the very strong labor market we had, say, 2 or 2 and 1/2 years ago. But it's kind of changed in an unusual way, not by the unemployment rate going up so much, but by job openings at employers dropping.

And then another factor people are looking at is wages. And, obviously, that gets LinkedIn various times to inflation. We've made great progress in getting close to our inflation target, but we're not quite there. And some people kind of worry that maybe wage growth is the issue.

And, in fact, wages, I'm going to show you, is still growing faster than would be consistent with 2% inflation. But I don't actually think that's the thing that's holding us back on our inflation mandate. And I kind of expect that once the inflation gets fixed, the wages are going to affirm that afterwards.

And then, of course, we all follow the labor market. And usually, the data that came out this morning, the one that grabs the bond market the most-- of all the statistics that come out every month, the one that moves the bond market the most is nonfarm payroll employment.

So I want to start by talking about what's going on with that, which I think has actually been confusing and actually hard to interpret and is going to become even harder to interpret as we go forward. So I'm going to suggest that maybe we should pay slightly less attention to this really key statistic and maybe a little bit more attention to some other things.

So there's the data on nonfarm payrolls. The data this morning for November, big number. What was it? 227 I think was the exact number. This is a funny series where they give you way more decimal places than BLS is usually comfortable doing.

What you see is the really big numbers back in 2021 at 600,000 and 800,000 in certain months. And the red line shows a moving average, the last three months. And you can see it's come down quite a bit. Although the very most recent thing, it ticked up a little bit.

But the topic I kind of want to talk about is are these numbers good? What is a good number for nonfarm payrolls? And if you asked me that before the pandemic, I would say 90,000 is a good number. And that's like what that trend gold line there is at the bottom. And that's based on how many jobs do we need to deal with the new workers that are coming into the economy, people entering, some people leaving.

But on net, there's been more people entering. And so we need some jobs to absorb them and keep the unemployment rate the same. So before the pandemic, you could argue whether it's 90 or 100 or 80 or something like that, but there's pretty much agreement that some number-- 100 are a little bit less is probably where we would be.

Things have changed. And the big thing that's changed has been immigration. And so it's complicated. I was going to talk a detail about that, but I'm probably just going to give you my kind of bottom line take is that probably if you look at immigration, the standards for what's normal, how many workers we need to absorb to keep the unemployment rate constant, have gone up a lot.

For 18 months, from 2023 until the middle of 2024, I would say it probably took about 200,000, twice as many jobs to keep up with the growth in the labor force. Halfway through this year, it's kind of come down a touch. There's been some changes in policy, and that sort of slowed things down. But it's still elevated, especially given there's probably some lags in terms of how these things work.

So by that measure, the red line, the sort of average, is actually not very different than the benchmark for what we would need to keep the unemployment rate constant. So that would say we're like doing average. But, actually, I think that's actually not quite the full story.

It's often not quite the full story because the numbers, the nonfarm payroll numbers, while they are great numbers, they're based on survey. Actually, they're based on survey plus modeling. And there's good reason to believe that the survey has probably, over the last year and a half or so, been too optimistic.

Getting into the details of this, the unemployment insurance system collects from employers how many workers do you really have, so it's kind of like a census. And so those numbers come out with fairly long lag. But there's pretty good evidence that they're going to show that the numbers over this period from March 2023 to March '24 were too high by probably 50 or 60,000 a month.

So I think we had to drop all those blue bars down by a little bit. And then this is, by the way, a bigger than-there's always a revision. And this happens all the time. But usually, the revisions are 0.1% or something like that. Whereas this time, it's going to be like probably 0.5% or 0.6%.

So it's a bigger than normal revision. And when bigger than normal revisions happen, what typically happens is the data afterwards also are probably running too high. So if you make that assumption, then probably the right numbers are the blue bars in this graph.

And if you look at that, well, then probably we actually have not been generating enough new jobs to keep up with the growth and the very rapid growth of the labor force that's been largely due to immigration. So probably, that's consistent with the fact that the unemployment rate has actually gone up a little bit over this time period.

So that's my take. This is a really important statistic, but it's a little bit hard to interpret right now. And it's going to get really hard to interpret given that policy on immigration-- I don't know what it's going to be. But it's likely to change in a very big way, in a way that's going to dramatically lower the benchmark for what it takes to stay with the unemployment rate.

So there, in fact, is the unemployment rate. It's still kind of low, 4.2%. And the red line is the Congressional Budget Office's guess of what the natural rate of unemployment would be, what the economy should be sort of delivering. And it's basically right there, just a touch below.

So my personal view is probably a little more optimistic. I actually think that-- if I was going to draw a red line but had Dan's name on it rather than the CBO's name on it, I would basically say probably it's half a percentage point lower than that even. But that's my particular take.

But still, historically, these are quite good numbers. But it has moved up, which is an indication that probably the tight monetary policy that we had over the last year and a half or two years has probably had some effects and slowed things down.

And in fact, there's this interesting thing. Claudia Sahm used to work at the Fed, knew her a little bit. She has became famous for this, what she called the Sahm rule, which is-- it goes back to actually another similar idea that-- I had a colleague, Lou Siegel, back in the old days, who there was something called Okun's Law. He had what he called Siegel's law, which was if the unemployment rate is not falling, it's rising. And first reaction is, well, no.

But the point of that is it's hardly ever do you see the unemployment rate kind of constant. It's like it's either going up or it's going down. We've actually been in a period where it's been fairly flat. And what Claudia noted is that you hardly ever see-- whenever you have a half a percentage point increase in the unemployment rate, you usually get a big increase in the unemployment rate. So this is plotting that kind of like-- it's slightly complicated, the three-month average minus the lowest point in the previous 12 months.

And so, usually, when that gets to 0.5, it's kind of shot up into recessions. And her idea was when you see that happen, you ought to be a little cautious. You ought to think about changing fiscal policy, et cetera, et cetera. So we've basically been sort of flirting with that view right now. People view that as being on the edge of potentially things.

But of course, there's lots of reasons to think that this time is different and, in fact, that things are probably stronger than this would indicate. But at least, it's a little bit of a worry. So that unemployment rate only went up a little bit. But the thing which is changed in a more dramatic way is the level of job openings.

If you ask workers do you need a job, that's what the unemployment rate is. When you ask firms, do you need some workers, this is what this job openings rate is. And it doesn't have a super long history in the US. It only goes back to about 2001.

But you can see that, shortly after the period right after the pandemic, it got really high, 7.5% by far the highest level that's ever had. And it's come down pretty dramatically. Some would say it's only down to just before the pandemic, when we thought of the labor market as being quite hot.

I actually have a slightly different perspective on this, which is that I think this thing has got a trend. With the unemployment rate, We talk a lot about the natural rate. I had that red line in the graph of the unemployment rate. People don't talk so much about this, but I actually think it's got an upward trend.

And in fact, if you compared, say, a period in 2005 with a period in 2016 that were kind of similar in most respects, the job openings rate is quite a bit higher in that latter period. And if you extrapolated that-- this is a very loose thing-- you would see that basically we would be expecting to see higher job openings. And what could be driving that? The cost of putting a vacancy out there is dropped over time.

There's a lot of strategizing about this. And especially when the labor market got tight and it got hard to find workers, lots of firms were basically saying, hey, let's go put the job openings out because somebody's probably going to quit, even if they don't have openings at the moment. So there's, I think, a lot of reasons to think that the job openings may be-- the high number is probably exaggerated, the tightness. But you don't want to ignore the fact that they actually have come down a large amount.

So this is-- I don't want to go too long on this. But it was very unusual to have this small change in the unemployment rate versus a big change in the openings rate. So this is the so-called Beveridge curve, from a British economist from the old days, who did it with British data a long time ago. But this is the data before the pandemic.

And what you normally see is-- this a scatter plot, instead of a time series plot-- is you normally see-- in good times, the unemployment rate is low, and the job openings rate is high. Bad times-- the opposite. You move this upward, diagonal version.

And in fact, maybe if you look at that very right edge of the graph, where the unemployment rate got close to 10%, you saw it shift out, actually. And that was actually a big deal at the time. That led a lot of people to think we had a rise in structural unemployment. And there was lots of fights between people within the Fed, and across the Fed, et cetera, et cetera about that small change, where the curve shifted. But it eventually went back up into the left, in a way you might have expected, given the natural-- the typical correlation between these things.

Well, what happened when the pandemic came along was something quite dramatic. Not a small shift, like happened in 2009 or '09 and '10, but a huge shift. And then we slid back the same way. But it's much more-farther from the origin rates of unemployment and vacancies, until we got to March of '22, when the vacancy rate peaked and then started to come down.

And what happened is it came down almost vertically, which was very surprising to a lot of people. And I have a few thoughts on what was probably going on there. But normally, you would have expected more of a diagonal shift here, but it was basically close to straight down.

What's going on with that? A couple of things. People often mention the fact that workers were in such short supply for a long period of time that firms were quite unwilling to let them go. So they're not going to lay people off, but they don't really need as many. And maybe they'd also been saying they had some vacancies, but didn't really mean it because they were expecting people to quit.

And in fact, the quit rate is, I think, a probably, important interpretive way of interpreting this. The quit rate dropped dramatically starting about that time as well. So the quit rate also had this huge increase. People were calling it the Great Resignation. But it's not really quitting work. It was basically quitting your crappy job and trying to get a better job, was probably the most typical kind of thing.

So when people were quitting like crazy, there were lots of openings and firms needed to fill them. Once people got over that and the quit rate started to drop, then we had that-- I think that's a reason why the Beveridge curve was able to shift back in and get close to where we used to be, without very much increase in unemployment.

But things have things have definitely gotten worse. And one way to view it is through-- how easy is it for an unemployed worker to find a job? And in fact, there's several ways of measuring that. Here's one way I kind of like, which is-- take all the people who are unemployed in one month, and ask what fraction of those people are employed the next month.

So when things were great, that was, like, 30% of unemployed people had a job the following month. But that's now dropped to 25%. Actually, the moving average of that has dropped to 25%. So one more-- a number that surprised me a lot this morning is this number for November really dropped quite sharply. And it's kind of bouncy. So I'm not taking too much signal from that. But that's a big change. So if still relatively few people are unemployed. But when they become unemployed it's a bigger problem to get out of unemployment, and that's a bit of a worry.

All right. What else here? Oh, this is-- I was hoping there were some people from the temp industry. Usually, I would have come to this group, and there's usually two or three people that work in the temp services industry, which is one of my favorite industries. I've written several papers about it.

But one weird thing that happened this business cycle is that temporary employment declined, even though regular employment continued to grow quite rapidly. And so if you know anything about this and you read my paper from 1995, you knew that, but many people think of temporary employment as a leading indicator.

And in fact, temporary employment has never come down on any kind of noticeable basis without there being a recession. But in fact, this time, it's been dropping, basically, since March of 2022 without a recession. And that's been a puzzle. But I'm not sure if this is an answer to the puzzle or just a way of linking it up with a different puzzle.

But there is a connection, which I noticed just recently, which is that if you look at the job openings numbers, the ones I showed you before-- the job openings rate-- that also peaked in March of '22. And it's been coming down, basically, the same as temp employment. And as I think about that, it maybe makes sense. If you're a firm and you've got lots of open spots, you probably also turn to the temp agencies.

So maybe one thing that's going on here is that the temp agencies aren't as busy as they used to be because there's fewer job openings than there used to be. And in fact, if you do this in growth rates, it actually lines up quite well. The scales are different. The job openings have twice as much variability in them as the temp help, but they have the same basic pattern. And it's been going on like this for a while.

So this is an answer, in some sense, to the question that Scott Horsley asked Austin about, what does it mean for there to be a better balance between-- in the labor market? And this is-- Austin basically gave the same kind of answer. But here's a picture of this, which is-- this is essentially the number of unemployed workers, minus the number of job openings.

So you can think of, like, game of musical chairs, where there's a certain people who want to sit down in a chair. How many people, how many chairs and musical chairs. Usually, there's more workers than there are chairs and than jobs. And that's typically the fact that this number is usually a positive number. It's expressed here as a percentage of the population.

So typically, on average, there's 1% more of the population who's unemployed than there is a number of jobs out there for people trying to get them. That dramatically changed right after the pandemic where it swung to the other side. So that was considered a very unbalanced labor market, where there were more job openings than there were workers who were trying to fill those things, or unemployed workers, anyway,

It's come back to close to 0, so you could view that as balanced. But historically the average has been not zero. It's been more like plus 1% But related to the point I tried to make, that the job openings has got this trend that people don't necessarily account for, I think it's actually closer to balance than if you just look at this graph. If you take into account that the world has changed to where there's more job openings, typically, than normal, then maybe this isn't as far from being balanced as you would say.

So that's some employment stuff. Onto the wage question. The number this morning was on the high side. I forget exactly what it was-- 0.37% or something like that increase on a monthly basis.

That's probably faster than would be consistent with 2% inflation because we think of-- wage growth would be, typically, inflation plus productivity growth. And productivity growth is probably, typically, less than 2% So probably, wages are still a little bit high. The wage growth is still a little bit higher than you would expect in the long run.

But I think this graph tells you a story about that, which is that while nominal wage growth is fast, it really wasn't fast enough at the beginning of-- right after the pandemic-- to keep up with inflation. So real wages actually had a period where they were dropping. So their growth in number were below zero.

So because of that, we need a period of time, essentially, for workers to catch up. And so if you think about-where we eventually get ourselves back to our 2% target for inflation, the way we sustain that-- we have to have wage growth come down. But historically, the way that sort of connection works is, typically, it's the inflation gets back to where it should be, followed, somewhat later, by what happens to wages. So I think that's eventually going to happen, but I would not expect wage growth to quickly get back to where we would normally expect until inflation has been at the right place for a while.

And then the bonus-- so again, this is another question that came up in Austin's conversation with Scott, is that productivity has been exciting lately. So the red line in this graph is an extrapolation of the trend level of productivity from the 2007-to-2019 period, which is kind of called the, second crappy period-- essentially-- of productivity growth in US history.

So we had a period of good productivity growth from 2000-- or from 1995 until 2007. Before that, there was another crappy period where people thought we got into this bad situation again. And again, there was that super jump up that's misleading right after the pandemic. And that's simply because the composition of the labor force changed in a way to eliminate people who work in the lower productivity sectors of the economy.

But that seems like that's over. And in fact, we've been having this sort of burst since then. And it's basically all speculation at this point because you wouldn't want to count on this. But there's at least some hope that maybe things are going to get better, especially if you thought that the artificial intelligence world was going to be able to generate cool new productivity growth.

And then what this graph does is it kind of combines the wage figures, along with the productivity. So it's no big deal if you have 5% wage growth, if you also have 3% productivity growth to get, because that would allow the full labor costs only to go up by 2%. So this is a graph which does that, which adjusts nominal wage growth for the level of productivity growth, called unit labor costs.

And it's also-- so if you look at, actually, the last bar, which is for the third quarter of this year, the number was 1.9% at an annual rate for the quarter. And this number should, basically, be roughly the same as inflation. Given that this is a nonfarm business sector versus the whole economy, it tends to be, like, two or three tenths below. So we would probably like this number to be 1.7 or 1.8. But 1.9 is about where you want it to be. The average over the full year is higher at 3, but it does seem like things are headed in the right direction, so I think that's a good sign.

And then another topic that came up with Austin and Scott was the-- and it, actually, very much relates to what Susan talked about, in terms of the labor force of parents-- is that there's been a burst of labor force participation of prime-age workers going up, relative to this trend. This is a trend based on an old statistical model that me and a few colleagues put together. Actually, going pretty far back in time.

And it worked pretty well until, basically, the late teens when it-- the labor force participation was stronger than we expected. And we were wondering about whether the model, needed some big changes or not. But it's had a burst where we're-- for prime-age workers, 25 to 54, as you can see in the graph. Us older people-- the trend still seems to be working.

So this is, again another hopeful sign. And I think it probably-- we don't know. But it's likely related to the increased flexibility that certain kinds of workers have, given the work-from-home trends.

So the other thing, which goes back to my first set of graphs, is immigration has really changed how the labor market has needed to perform. And so this is a graph of population growth-- the growth of the population of the people who are 16 and over and not living in institutions.

And what you see is that while it's slowed a lot, especially during the pandemic, it had a big boost. And I was going to lead you through some details of how we calculate numbers like this. By the way, these are taken from the Congressional Budget Office. But they're looking heavily at what's going on at the border, the number of encounters at the border, the number of people who are released with a date to come to court, et cetera, et cetera. And that's really shot up.

And so instead of having growth, like, maybe, two or three tenths of a percent for the population during that period of time, it's been more than 1% for the most recent period. Well, I'm going to add a caveat to that in a second.

But what I think it's important to note is that what makes a lot of confusion-- including for myself sometimes-- is that the numbers here-- these are coming from an independent estimate done by the Congressional Budget Office, where they're looking heavily-- trying to understand immigration-- looking at what's going on with border statistics.

The Census Bureau, which is our normal sense of how many people live in the country, comes up with much lower numbers for the population growth. And the reason is because they do it in a different way. They don't really pay a lot of attention to the numbers from the Homeland Security people. They do it mainly on the basis of their surveys. And if they do a survey and they find out a lot of people say they're new immigrants to the country, then that figures into their calculations of what the population growth has been.

And so I think most people who are in this business of looking at high-frequency data think that it's underestimated by the Census Bureau, how fast population has been growing. Their methods should eventually work, I think. But it's going to take-- it takes a few years for changes to show up in their data.

And this is going to get immensely confusing going forward because I think we're on the cusp of, basically, scaling back immigration in a big way. And so the population is probably not going to be growing very fast. But the Census numbers are going to be lagged and they're going to basically be showing-- so there's going to be confusion, I think, which is a reason why, probably, looking at some of the normal statistics about number of people working and things like that is going to get a little bit dicey.

And this is-- the second bullet point down here is a footnote about the-- there's a second measure, which sometimes gets quite a bit of attention of late, on how many jobs we have. And it comes from the household survey, the same one that we do to get the unemployment rate, where we ask people, are you working or not?

And that-- actually, this morning that dropped by 350,000. So instead of having a 200 increase, we had a 300 decrease. I think that number is always very volatile, which is a reason not to look at it in a month-to-month very much. But it's also highly dependent on the Census Bureau's guesses about the-- or estimates of how the population is growing.

So what the household survey really does is it tells you what fraction of people have jobs. It doesn't tell you the number. It tells you what fraction. But then multiply that by how big you think the population is, and that gives you the number of people who are working.

Because the population growth numbers from the Census Bureau are probably way too low, the numbers from the household survey are probably way too low in terms of the increase in employment. So I think over the last year before this-- before this morning, the yearly increase in household employment was about 160,000, whereas the increase from the payroll survey is, like, 1.6 million. So a factor of 10 difference. So I think probably not paying too much attention to the household survey is a good idea right now.

So the footnote about immigration is that halfway through this year, there were some policy changes. Also, the Mexican-- some by us and some by the Mexican government. They've pretty dramatically curtailed the flow of immigrants into the country. And this is for the whole year. So really, if you went with steady state, you'd go down about twice as much as this. And that would make me sort of lower my estimate of the break-even level from closer to the old days, but not all the way.

But again, some dramatic changes are coming. We don't really know exactly what they are, but they're likely to be quite dramatic. So this is more speculative. Sometimes, we try to get in-- do two decimal places and try and figure some things out. This is on the edge of trying to get some rough numbers.

And I took this from two scenarios. Some people at the Brookings Institution, including Wendy Edelberg, who was my RA way back in the old days, and who's, like, I think my most successful RA out there. I don't want to offend anybody because I have a bunch of really successful raise.

She and Tara Watson put together some scenarios looking at what the Trump administration has said they want to do, versus what's probably feasible to do. And they have-- the so-called high immigration scenario is one that would be a lot like the previous Trump administration, but maybe just a little bit tougher because they've got-they're, maybe, doing that.

If that was the case and if you had the typical 0.2% decline in the labor force participation rate, that would drop the break-even for the monthly employment growth to 50,000. So that's why I would say, typically, we thought 90 was the number. So in this higher-end scenario, with slightly dropping labor force participation, you'd expect to get 50 instead of 100 every month or 90 every month.

Now, I think-- to be fair, I think they're hoping that this tighter immigration policy will lead to greater labor force participation by the people who are here. So maybe instead of dropping, we would get a flat participation rate, which would be unusual, given what we've had over the last 25 or so years. If we had that, then we might get-we'd get about 90,000 per month-- would be the break even so that. So those are-- again, that would be back to the old days.

However, if you thought about the more tougher scenario, where immigration is lower, they came up with-maybe instead of a 1.3-million increase in the population, we'd have a 0.6-million decrease in the net immigration. If that was the case, we'd be looking at-- and I went with my 0.2% drop in the labor force participation. We'd be looking at a benchmark, which was a negative benchmark. Instead of expecting to have 100,000 growth a month, we'd be expecting to have-- 45,000-per-month drops would be the typical thing.

If we kept the labor force participation flat, then it would be basically-- we'd end up with about a 0%-- or a 0 thousand-per-month change because flat labor force participation is not going to-- would be, basically, stuck-would have a drop in the labor force, given that the net immigration was going down.

So the previous graph I showed you had two parts to why the population group would be going up. Births minus deaths. And then the rest of it is immigration. But most of it is the immigration these days. And in fact, if you have negative immigration, it would more than offset the births minus deaths.

So those are radically different numbers. It's going to make interpreting the data hard. So I think the thing to do is to focus a little bit less on those nonfarm payrolls and look-- try to see what's going on with the unemployment rate. Is it staying flat? Is it going up? Or is it going down?

Just a word on wages. So I think a lot of people are thinking, wow, the cut to immigration is going to go-- it's going to be reducing labor supply, and reducing labor supply should boost wages. That's right. But it's not-immigration, actually, also boosts labor demand because people come here. They need to consume things. They need to buy things. And so the people need-- somebody needs to produce those things.

And so I think that, probably-- again, a big guess, because we don't have experience with stuff exactly like this before. But probably-- there's probably not going to be a big effect on wages. If I had to guess the sign, I would say, yeah, probably it's going to raise wage growth a little bit.

But I think the experiment that's in the back of my mind, on the back of a lot of labor economists' mind, is what happened after the Mariel boatlift. If you remember this-- back in the-- I think it was in the late 1980s. Fidel Castro sent all sorts of uneducated, poor people out of jails, et cetera, into the Miami labor market. And so it was a huge increase in immigration.

What did it do to wages? Not very much. And there's some disputes about that. But I think the takeaway I have from reading this literature is that these big changes in immigration don't necessarily lead to big changes in wage growth, but we'll see.

Last-- I mean, tariffs? I don't have much to say about tariffs except for they're likely to create some need for reallocation. So I showed you that picture of the Beveridge curve, which I mentioned. Many people think that the reason it shifts in or out has to do with, how much reallocation do you need to do? If you're in a very static economy where everybody's kind of staying in the same industry and the industries are all kind of growing at the same rate, then you don't need to have as much reallocation. And so unemployment and vacancies can both be a little bit lower.

But if you're in a really-- world in which lots of things are changing and some industries are growing fast, some industries are growing slow or shrinking, then you need to move people between industries. That's the kind of thing which leads to both higher unemployment and higher job vacancies.

I think that's something which could easily happen because of tariffs, if they're done in a big way, because, obviously, tariffs could help some industries. If you're an industry that's producing a final output that's competing with final outputs that are being brought in from other countries and their prices go up, that's probably going to be good for those domestic industries.

But there's also domestic industries which buy inputs from the rest of the world, like steel. If the costs for them go up, this could actually be detrimental to them. And certainly, if you're also an exporter, it's likely the case that higher tariffs-- while they might reduce our imports, they'll probably also reduce our exports by roughly the same amount. That's just a proposition in international economics.

So you could be facing a situation where there's a lot of churn, where some industries are growing because of tariffs; some industries are shrinking because of tariffs. And that's going to be, potentially a reason for a lot of mix-up in the world.

That's what I got. So bottom line-- labor markets softened, but it's still relatively healthy. Wage growth is still higher than we would expect in the long run, given 2% inflation, but we're headed in the right direction. The benchmark for thinking about what unemployment growth should be from month to month is going to be highly uncertain. It is highly uncertain now. It's going to continue to be highly uncertain.

And I think that means looking at the unemployment rate itself and asking whether it's going up or going down is the right thing to do. And then last bullet-- there's just a lot of uncertainty about what's going to be happening next. So that's what I have. Happy to take a few questions.

SPEAKER: All right. So we are a few minutes over. So don't feel shy about, I guess, leaving right now if you do have to--

DANIEL SULLIVAN: I do have a 4 o'clock airplane, too, so I'm--

SPEAKER: OK. Well, we're not going to keep you long then. Let's see here. So we got one-- I'll ask one question here. Let's see if we can make sense here. So people talk a lot about a, quote, "tight labor market." But many also express frustration about difficulties in finding jobs. In what areas or demographics is the labor market tight, and where is it not? So maybe this is about different unemployment rates by, say, industry, or occupation. Do you have much to say about that?

DANIEL SULLIVAN: Sure. So I think both things can easily exist-- sort of tightness on average. Unemployment's low. But once you become unemployed, because there's not as many job openings as there used to be, I think it's natural that some workers are going to find it difficult to find jobs once they do it.

I don't have a great sense of-- certain kinds of workers are in demand more than usual. I mean, obviously, the unemployment rates tend to be lower for well-educated people. People with certain kinds of backgrounds-- it's tougher, with of less sort of preparations.

I don't think anything-- I mean, this shift between industries that we had because of the goods sector, I think, is a little bit of that. So I think the goods sector is kind of recovering. Or the opposite, rather-- the good sector had its share of increase. It's now falling. So probably, that area is going to be a little bit less in demand than the services areas. And so we see that a little bit in terms of-- a lot of the job growth comes from leisure and hospitality and certain other sort of service industries like that.

SPEAKER: OK. And we've got one other question that I'm just going to phrase a little more generally, is-- I think you've put up some uncertainty about 2025. Do you have any forecasts for 2025, acknowledging the uncertainty in terms of-- so in terms of labor force [INAUDIBLE].

DANIEL SULLIVAN: So this the week where we do the so-called Survey of Economic Projections. We put together a forecast which we're going to turn in, and you'll get-- put in the dot plot in a few-- and it's-- it's probably the most confusing one I've been involved in. I've been at the Fed, now, for 30-something years.

And so I think a lot of people, actually, don't want to make the forecast yet because there's two parts to the uncertainty. One is we don't know what the policy is going to be. And then, secondly, given the policy, we don't know exactly what's going to happen because of the policy. So it's really tough at the moment.

But certainly, it seems likely that employment growth is going to be lower going forward, given the direction of policies from the new administration. I think it could-- so extra churn, could also push up unemployment. But I no way of quantifying that at the moment. So I'm going to punt on that question.

SPEAKER: Fair enough. That's a good Fed answer right there. So [LAUGHS] with that, I think we'll end the event. Thanks, everyone, so much for coming. And I had a great time. I hope you all did, too. And we'll see many of you next year, I hope. Thanks.

DANIEL SULLIVAN: Thanks.