What Is the Beveridge Curve? Job Openings and Unemployment

The Beveridge curve is a way to plot two U.S. labor-market rates together: the job-openings rate and the unemployment rate. The U.S. Bureau of Labor Statistics (BLS) publishes a chart that places the unemployment rate on the horizontal axis and the JOLTS job-openings rate on the vertical axis. Each monthly observation becomes a point, and the path of points over time forms a curve. The BLS chart lets readers see how the two measures have moved together across different periods.

The relationship is usually inverse. When businesses are expanding and recruiting is strong, vacancies tend to be plentiful while fewer people are unemployed; on the chart, observations tend to sit toward the upper left. When labor demand weakens, openings often fall as unemployment rises, taking observations down and to the right. These are broad patterns, not a rule that every monthly move must follow the same path. A typical curve slopes down and is convex: its steepness changes across labor-market conditions. That shape gives a compact picture of labor-market tightness and matching frictions. It does not say that each open position corresponds to one unemployed person with the right location, skills, and availability.

JOLTS counts a job opening only when a specific, unfilled position has work available, could start within 30 days, and is being actively recruited for outside the establishment. A vague plan to hire later is not enough. The BLS JOLTS definitions set out these conditions. Its job-openings rate is the number of openings divided by employment plus openings, multiplied by 100; this expresses openings as a share of filled and unfilled jobs. The BLS rate methodology describes the calculation.

The unemployment rate comes from a different BLS survey, the Current Population Survey (CPS). It is the number of unemployed people divided by the civilian labor force. People generally must be available for work and have actively looked in the previous four weeks; people temporarily laid off and expecting recall can be counted as unemployed without an active search. The BLS CPS definitions explain these classifications. The JOLTS rate and unemployment rate therefore use different surveys and concepts; they are not two sides of one employer-level ledger.

Their reference periods also differ. Job openings are a snapshot on the last business day of the month. Hires and separations count events over the full month. The BLS JOLTS FAQ explains this stock-versus-flow distinction and notes that early estimates can be revised. The Beveridge chart compares rates, but anyone combining it with hires, layoffs, or payroll data should check what period each series covers.

On the standard chart, move right to find higher unemployment and move up to find a higher openings rate. A point high and left suggests many positions are open relative to unemployment; a point low and right suggests fewer openings and more unemployment. The curve bends rather than forming a straight line, so the relationship can look different depending on where the labor market sits.

That bend matters when vacancies fall. At a high openings-to-unemployment ratio, some vacancies may be difficult for employers to fill even in a strong market. Removing a portion of those postings can cool recruiting without immediately producing an equally large increase in unemployment. Closer to a weaker part of the curve, a similar decline in openings may coincide with a larger unemployment change. The U.S. Federal Reserve’s 2022 analysis describes how the curve’s position, its estimated slope, and separations affect such calculations; those are model-based results for that study, not a current forecast or permanent coefficient. See the Federal Reserve analysis.

The plotted path can turn because labor conditions change over many months, and monthly data are noisy. Compare a sequence of observations and a stated window rather than treating the most recent point as a diagnosis. The BLS Beveridge curve chart lets readers inspect monthly paths over time; it does not forecast when the next turning point will occur.

A movement along a given curve describes a changing balance between openings and unemployment when the underlying matching relationship is treated as broadly similar. For example, weaker labor demand can reduce vacancies and raise unemployment, moving a point down and right. Stronger demand can take it up and left. This is a useful first description of the business-cycle mechanism, not proof that every other labor-market condition stayed fixed.

A shift means the relationship itself may have changed: for a similar openings rate, unemployment is higher or lower than before. Lower matching efficiency can push the curve outward because employers and job seekers produce fewer matches from a given set of openings and unemployed workers. Changes in the industries hiring, workers’ skills or locations, search behavior, labor-force participation, or the separation rate can also affect the path. The Federal Reserve’s analysis discusses matching efficiency and separations as distinct influences on the curve.

One chart cannot identify which explanation caused a shift. A change in survey composition, revisions, or timing can also affect observed points. Treat “outward shift” as a description of the pattern; use hiring, layoffs, quits, participation, industry detail, and other evidence before naming a cause.

Fewer vacancies can reflect a cooling in labor demand: firms may post fewer new jobs, pause replacement hiring, or take longer to fill positions. If layoffs remain limited, workers leaving one job may still find another, and unemployment can rise gradually. That pattern can be consistent with a labor market moving toward a less overheated state, but the curve alone cannot show whether inflation will fall or a recession will be avoided.

Separations matter. If layoffs and other involuntary exits rise at the same time that openings fall, more people can enter unemployment while fewer positions are available. The unemployment response may then be larger than a simple movement along a stable curve would suggest. The Federal Reserve study models vacancies together with separations, which is why its 2022 discussion should be read as a dated, assumption-dependent analysis rather than a current prediction.

A recession is a broad decline in economic activity, evaluated with evidence beyond vacancies and unemployment. The curve can add context to employment conditions, but it does not certify a soft landing or date a recession. For that broader question, compare it with output, payrolls, hours, and other labor measures; see the guide to soft landing versus recession.

Suppose three invented monthly observations are plotted. Point A has unemployment of 3.5% and a job-openings rate of 6.0%. Later, point B has unemployment of 3.8% and openings of 5.0%. A-to-B movement is down and to the right: openings declined while unemployment rose modestly. If other conditions were similar, that would be consistent with labor demand cooling along the existing curve.

Now compare point C, also with a 5.0% openings rate, but unemployment of 5.0%. Higher unemployment at a similar openings rate could be consistent with an outward shift: matching may have become harder, or other labor-market conditions may have changed. It does not prove that matching efficiency fell. The values are hypothetical teaching points, not BLS observations, forecasts, thresholds, or a claim that the curve must pass through them.

The practical lesson is to compare both coordinates and the surrounding evidence. A lower openings rate is not, by itself, proof of a recession; a higher unemployment rate at a similar openings rate is not, by itself, proof of a structural mismatch. The curve helps frame questions that other data must answer.

JOLTS and the Current Population Survey measure different things. JOLTS surveys establishments about open positions, hires, and separations. The household survey classifies people as employed, unemployed, or outside the labor force. The concepts differ, and job openings and unemployment do not form a one-to-one matching count. The JOLTS FAQ also explains that its estimates are revised as additional information arrives.

Use seasonally adjusted rates when following the BLS historical curve, and check the time range and data vintage. A revised point can shift the path. Opening rates use employment plus openings as the denominator; the unemployment rate uses the labor force. A percentage-point move on one axis is not a count of workers and should not be compared mechanically with a percentage-point move on the other.

Other indicators can explain what the curve leaves unresolved. Check hires and quits for labor-market flows, layoffs and payroll growth for demand, participation for labor supply, and hours for how employers adjust work. U-3 and U-6 provide different views of labor underutilization; the guide to U-3 versus U-6 unemployment explains their scope. Okun’s law gives a separate way to discuss output and unemployment, with its own limits.

Start by naming the geography, sources, rates, and time window. For the U.S. chart, that means identifying JOLTS openings and CPS unemployment and noting whether values are seasonally adjusted. Then describe whether the points moved along a familiar path or whether the relationship appears to have shifted. Do not infer a cause from shape alone.

Next, examine the factors that could change matching or the number of separations: job mix, worker location and skills, hiring, quits, layoffs, and labor-force entry. Check whether the underlying series were revised. A short stretch of data may not distinguish a lasting shift from normal variation or a change in the mix of industries hiring.

Finally, keep the curve in its role. It organizes evidence about vacancies and unemployment and can help explain why lower openings do not automatically mean a sharp rise in joblessness. It is not a countdown to recession, a substitute for broader activity measures, or a promise that labor-market cooling will be painless. Use it alongside the related guides on soft landing versus recession, U-3 and U-6, and Okun’s law.