Is the U.S. economy at maximum employment? It’s a critical question for the Federal Reserve, tasked by Congress in 1977 to pursue monetary policy that promotes the goals of stable prices and maximum employment—the Fed’s dual mandate.

The Federal Reserve has clearly defined its stable price goal as an inflation rate of 2 percent over the long term. To measure inflation, economists track the prices of thousands of goods and services every month. This is an exercise that many of us informally participate in daily as we encounter prices for apples and eggs, bus fares and T-shirts, electricity and streaming services.

Amanda Michaud Principal Research Economist, Institute

Maximum employment, on the other hand, is both harder to define and more challenging to measure. The Federal Reserve’s definition is the highest level of employment or the lowest level of unemployment that the economy can sustain while maintaining stable prices. The Federal Reserve does not assign numbers to those levels, however. According to its 2025 Statement on Longer-Run Goals and Monetary Policy Strategy, “the maximum level of employment is not directly measurable and changes over time owing largely to nonmonetary factors that affect the structure and dynamics of the labor market.”

To a layperson, “maximum employment” might look like an economy where everyone who wants a job can find one. But even the concept of “employment” can become difficult to pin down in our modern economy.

“Employment is a funny construct because it’s not, Are you doing work? Instead it’s, Are you doing work that involves a market exchange? And we each have lots of ideas of what that looks like,” Minneapolis Fed economist Amanda Michaud said. “Are you working for an employer? Are you being paid a regular salary? And the economy has changed. There’s lots of people who are doing more irregular work, consulting, or working for themselves in the gig economy.”

It’s a definitional challenge that Michaud has faced first-hand. “I took one of the official Census Bureau surveys for my household when my partner was doing gig work, and I had a hard time answering whether he was unemployed or employed.”

Michaud has long studied labor market data and what they communicate about the trajectory of the economy. In 2025, Michaud and fellow Federal Reserve economists Christopher Foote, Shigeru Fujita, and Joshua Montes wrote “Assessing Maximum Employment” as part of the Federal Reserve’s review of the framework it uses to achieve its dual mandate. We talked recently about the indicators economists use to evaluate maximum employment and what they have been saying about the state of the labor market.

We spoke on August 20, 2026.

One concept, many measures

The unemployment rate is often used as a measure of how far the economy is from maximum employment. What does the unemployment rate tell us, and what does it not tell us?

The unemployment rate is one measure of people who don’t have a job and are actively searching for one. But actually when you look at hires from nonemployment, the majority of them are not classified as having been “unemployed.” They are in what we call “nonparticipation” in that they are not actively searching for a job, or at least haven’t searched in the last couple of weeks. They still might want a job, they would be open to working, but they don’t fit the narrow criteria of how we measure unemployment.

The employment-to-population ratio (EPOP) is another indicator economists look at to assess the state of the labor market. Does EPOP provide a better way to capture the number of people who are employed compared with the number who could be employed?

EPOP is a different concept. It doesn’t measure the pool of unutilized potential labor. It is the utilized labor. Certainly, we can start asking questions if we see movements in EPOP. If we see a decline in EPOP, we would ask, “Okay, are we drifting further from maximum employment?” But the answer depends on if the movement is caused by people who don’t want a job leaving employment, or if there are policies in place that induce fewer people to be available to work. For instance, right now immigration is falling and the country’s demographics are changing.

In a paper I wrote with Kathrin Ellieroth, we construct a “want-to-work” measure using the Current Population Survey (CPS). What it does is actually ask the people who say they are not looking for a job if they would want a job. We build an economic model that matches all of the flows in the economy and can replicate what the labor market looks like over the business cycle. And we find that the CPS want-to-work measure matches closely the model’s prediction for unutilized labor, which is the right metric for slack.

So now we’re recommending that people look at the want-to-work measure because sometimes policy can influence whether people declare themselves as unemployed or as a nonparticipant. When unemployment benefits are higher than normal, as they were during COVID for example, more people were unemployed. But that doesn’t mean that more people wanted a job than before.

That’s why we love that this question because you just straight up ask people, “Do you want to work? Do you want a job?” You never know how people will answer a question, but what our research does is validate that their answers track the economic concept of slack, or shortfalls from maximum employment, in our workhorse theoretical models of the economy.

And are there a lot of people right now without a job who want to work?

The nonemployed who want to work peaked in late 2025 and has been falling since. Right now it is tracking the unemployment rate pretty well. In contrast, in 2022 unemployment was falling faster than want-to-work, which suggested that unemployment overstated how hot the labor market was becoming. During that period, the want-to-work measure was particularly useful. (See figure.)

We’ve been discussing employment from the point of view of people supplying labor—who has a job, who wants to work. But it also matters how many jobs are available. For that, economists look at the Beveridge curve, which is the relationship between job vacancies and unemployment. What does this relationship indicate about maximum employment?

This might help clarify why it can be difficult to track the concept of maximum employment. If we see the unemployment rate rise from 2 percent to 4 percent, for instance, we may immediately think that the economy is moving further away from maximum employment. That’s not necessarily the case, and the Beveridge curve can help us see why.

The Beveridge curve is the number of jobs that are available per unemployed worker. If vacancies per unemployed worker go down, typically that means that the labor market is becoming less tight, so we think there would be less pressure from the labor market to prices: When workers are abundant, employers might not have to pay as much to attract new hires or to keep current workers, because workers’ outside options aren’t as good.

Going back to the 2 percent unemployment versus 4 percent unemployment example. Certainly, if unemployment rises from 2 to 4 percent and vacancies per unemployed fall significantly, then more likely than not the labor market has more slack and we have moved further from maximum employment.

If unemployment rises from 2 percent to 4 percent but vacancies per unemployed worker do not change much, then the story is more complicated. One possibility is mismatch. This was the story during the Great Recession. The example people were using was that construction workers were being laid off in the construction industry while the health care industry was doing all the hiring. It’s difficult to move those laid-off workers to the available jobs because different skills are needed. This is a situation where not everybody who wants a job can find a job, but it’s not necessarily a shortfall from maximum employment, because if you increase demand, it’s going to increase demand in a sector where there aren’t enough trained workers, so it’s not going to change unemployment.

This is another reason why it’s so hard to use one number to track the labor market. In this example, the same number [of job vacancies per unemployed person] could happen in two different scenarios with two very different implications. And even with the additional information from the Beveridge curve, it is often the case that both demand-driven and structural factors are affecting the economy simultaneously, and we’re just trying to figure out how much is coming from each.

In your research with Kathrin Ellieroth, you’ve developed a new dataset that looks closely at worker flows, that is, the people moving from one labor market status to another. You identify the reason a person left a job (Did they quit or were they laid off?) and what status they ended up in (Are they looking for a new job or did they leave the labor force?). This is more detail than traditional “stock” data such as the unemployment rate provides. What does this extra level of detail tell us about the state of the labor market?

The stocks in the labor market, like unemployment and employment, are slow moving. Shifts happen very slowly. And what’s changing their trajectories is labor market flows. So we talk a lot about the flows when we want to understand, Is that shift going to accelerate or decelerate?

In particular, when you have turning points in the economy, for example when the economy enters a recession, unemployment spikes. By then, it’s kind of too late. The antecedents of the recession are already well in play. What tends to happen before the recession, as the economy slows down, the first thing that stops happening is people stop quitting their jobs without having another one lined up because it’s going to be harder to find a job. They might be more worried about the financial security of their household. So in the data we see quits fall first.

After that, what we see in our data is layoffs. And we differ a little bit from other measures of layoffs because we see where people go. It’s one thing to have many layoffs when people have another job lined up. What we track is layoffs where people don’t move directly to another job—instead they move to nonemployment.

So usually we see quits start to fall, and right after quits start to fall, layoffs to unemployment start to go up. And then we see unemployment rise.

“A super boring labor market”

Given the current economic environment, what indicators are you paying attention to now to assess the strength or weakness of the labor market?

In 2026, the labor market has been remarkably steady. Unemployment has basically been flat all year. Employment may have fallen a little bit. Flows, which usually tell us where the labor market is going, have been steady. So it’s been an interesting time where I think we are all waiting for the first signal of what’s going to happen next. It’s a funny question because all of the signals look the same right now, in my opinion.

So the labor market has not been changing much. Do you think in its current condition we are relatively close to maximum employment?

From an individual perspective, you might hear very different things. There are still people who are losing their jobs, and there are people who are having great difficulty finding jobs. And obviously for them, it’s not a good labor market. But from an aggregate perspective, again, unemployment is low, employment has been steady and high. Unemployment insurance claims have been very low. So it appears that as a whole, the labor market’s very stable.

It’s actually remarkably stable considering all of the exotic shocks that have hit the economy. We’ve had big swings in trade, immigration, new technologies. In a way, it’s puzzling how boring the labor market has been for the past year.

You have observed remarkable steadiness in labor market indicators, but for a while there was considerable concern that the labor market was weakening and unemployment would go up. Why didn’t that happen?

Kathrin and I have been tracking this for a long time. During the more expansive labor market in the early 2020s, you saw a lot of new people enter the labor force. And the question for us was, If things start cooling, what is going to happen to them? The people who might have had trouble finding a job before who found a job during the expansive labor market—if they lose their jobs, will they enter unemployment and really increase the unemployment numbers? Or are these people whose outside option is nonparticipation, maybe retirement, caregiving, or the type of hard-to-measure work activities we discussed earlier? Because there’s always a lot to churn in the labor market. There’s always job loss even if things are steady.

You asked about this period of pessimism last year. I always called us “the lonely labor market optimists” because we saw the turn in these flows. In early 2025, we saw that people started quitting a little bit more and layoffs were starting to come down. That gave us the early knowledge to be less pessimistic even as unemployment rose a couple basis points. And we were right: Unemployment stabilized and fell in the fall of 2025. I’m not saying we will always be right based on these data, but that was one example where our data really proved useful.

This reminds me of the theory of “the wisdom of the crowds”: To make a prediction about where the economy is headed, look at what workers are doing.

That’s right! But there are a lot of nuances too. It’s a strong signal of worker confidence to quit without a job lined up. It is less of a strong signal when unemployment moves because most hiring is from nonparticipation. And I guess we’re going against the grain with our want-to-work measure. Usually, economists look at what people do, not what they say they want to do, but this question actually holds up for tracking slack in the labor market.

This interview was edited for length and clarity.