What Is NAIRU? The Natural Rate of Unemployment, Explained

NAIRU stands for the “non-accelerating inflation rate of unemployment.” In a model that relates inflation to labor-market conditions, it is the unemployment rate at which inflation is not expected to keep rising or falling because of labor-market tightness alone. If actual unemployment is below that estimate, the model may indicate upward pressure on inflation; if it is above, it may indicate downward pressure. Those are conditional signals, not mechanical rules. {source:frbMishkinPotentialOutput2007}

“Non-accelerating” describes the direction of inflation over time, not its level. A model’s NAIRU does not say that inflation must be zero, that the price level is stable, or that inflation cannot move for another reason. Energy prices, productivity, expectations, supply disruptions, and policy can all affect inflation. The Phillips curve guide explains that broader relationship; this article focuses on the unemployment benchmark and what its uncertainty means.

The natural rate of unemployment is a related idea: the unemployment that remains when cyclical changes in overall demand are set aside. Job search takes time, workers and vacancies do not always match immediately, and people change jobs. A 2007 CBO working paper discusses job creation, job destruction, turnover, and worker-job matching as influences on the natural rate. CBO’s 2023 methods presentation separately describes its noncyclical rate as unemployment arising from factors other than fluctuations in aggregate demand. The working paper is a dated research analysis, not a current universal estimate. {source:cboNaturalRate2007} {source:cboNoncyclicalRate2023}

Economists and institutions sometimes use “natural rate,” “NAIRU,” “equilibrium rate,” and “noncyclical rate” in overlapping ways. They are not guaranteed to refer to the same statistic. A natural-rate estimate may come from a model of labor-market structure; a NAIRU estimate is tied to an estimated relationship between unemployment and inflation; and an agency’s noncyclical rate is an operational measure with its own method. The horizon and assumptions matter, so compare the definition before comparing the numbers. {source:frbMishkinPotentialOutput2007}

None of these terms means zero unemployment. People may be between jobs, entering the labor force, or looking for a better match. “Full employment” is not a synonym for zero unemployment or for one observed NAIRU estimate; these labels do not identify a single universal percentage. U-3 and U-6 are U.S. survey measures; the underlying NAIRU or natural-rate idea is broader and would need to be estimated with a country-specific definition and model. For how the U.S. measures count people, see U-3 vs. U-6 unemployment rates.

Statistical agencies can count people who meet a survey’s definition of unemployed. They cannot directly count the unemployment rate that would prevail if cyclical demand pressures were removed, or the rate at which inflation would stop accelerating under a particular model. That benchmark is a latent, counterfactual quantity: it must be inferred from observed outcomes and assumptions about how the economy works.

Inflation does not move with unemployment alone. Expected inflation, input costs, productivity, labor-force composition, and policy can change at the same time. If a model leaves out a relevant factor, it may attribute too much of an inflation change to the unemployment gap. Different reasonable choices about the time period, data, and equation can therefore produce different estimates. Federal Reserve officials have emphasized that estimates of the natural rate are imprecise and can change when new information or methods become available. {source:fedPowellMonetaryPolicyChangingEconomy2018} {source:frbMishkinPotentialOutput2007}

There is no single universal calculation. Some approaches estimate a relationship between inflation and slack, including a time-varying NAIRU. Other approaches focus on labor-market structure, demographic groups, job flows, or the trend and cycle in unemployment. An agency’s reported number is the output of its selected model, not a directly measured constant.

For example, CBO’s 2023 methods presentation described its noncyclical rate using unemployment patterns across demographic groups and also discussed an alternative trend-cycle approach it was evaluating at that time. That is one agency’s method and one dated description, not a recipe every economist follows or a current benchmark that applies to every country. {source:cboNoncyclicalRate2023}

Labor-market conditions can also change the estimate. A shift in the age mix of the workforce, job-search behavior, hiring practices, or how efficiently workers find suitable vacancies may alter the amount of unemployment that persists apart from demand cycles. The Beveridge curve offers another way to examine how vacancies and unemployment move together, but it does not reveal one exact natural rate by itself.

The estimate may change as evidence and methods change. New observations can alter the pattern researchers see; the model can also change when researchers adjust the time period, variables, or assumptions about how quickly the underlying benchmark moves. A real-time estimate may therefore differ from one made later with additional evidence.

Federal Reserve Chair Jerome Powell has described natural-rate estimates as uncertain and subject to revision, including the difference between what policymakers could estimate at the time and what later data suggested. That history is a reason to read an estimate as a range or model-based reference, not as a fact that was obvious in real time. {source:fedPowellMonetaryPolicyChangingEconomy2018}

The benchmark can move even when the latest unemployment rate does not. Changes in demographics, matching efficiency, or labor-market institutions can shift the noncyclical component over time. Conversely, a revised estimate can change the calculated unemployment gap even though no newly counted worker has entered or left unemployment. That is why the date and data vintage matter when comparing charts.

A common sign convention is:

Unemployment gap = observed unemployment rate − estimated reference rate

With this convention, a positive gap means observed unemployment is above the selected benchmark; a negative gap means it is below. Some publications use the opposite subtraction order, so check the definition before interpreting a plus or minus sign. The gap is a comparison of two rates, one observed and one estimated. It is not itself an inflation rate or a forecast.

The benchmark can also be called the NAIRU, natural rate, or noncyclical rate in a particular publication. Use the term that source actually defines, and do not swap in another agency’s estimate without checking whether it uses the same concept and horizon.

Suppose the observed U.S. U-3 unemployment rate is 4.6%. For illustration only, compare it with two hypothetical reference estimates:

• If the estimate is 4.2%, the gap is 4.6% − 4.2% = +0.4 percentage point.
• If the estimate is 5.0%, the gap is 4.6% − 5.0% = −0.4 percentage point.

The observed rate is unchanged, but the sign of the estimated gap flips because the benchmark differs. The example shows why the uncertainty around NAIRU matters. It does not say that either estimate is correct, that inflation must accelerate or decelerate, or that a particular policy response is warranted. The numbers are invented to demonstrate subtraction, not current data or a forecast.

When a report gives a natural-rate or NAIRU number, ask four questions:

1. Which concept is being estimated? Check whether the source means a price-stability threshold, a broader noncyclical labor-market rate, or another equilibrium measure.
2. What model and horizon does it use? A short-run estimate and a long-run estimate can answer different questions.
3. How uncertain and revisable is it? Look for ranges, confidence intervals, data vintages, and notes about revisions rather than relying only on a point estimate.
4. What does the gap convention mean? Confirm the subtraction order and the observed unemployment series before interpreting the sign.

Use the estimate as context for a labor-market assessment, alongside job openings, participation, wage growth, inflation expectations, and other evidence. The Phillips curve explains why slack may matter for inflation in some models; neither that relationship nor a NAIRU estimate alone tells you what inflation will do next. A careful reading separates the observed unemployment rate from the uncertain benchmark used to interpret it.