With seventy-five months of job growth in the private sector, the recovery since the Great Recession has been longer than most Post War economic recoveries.

The economy added a respectable 255,000 jobs in July. But by all measures, job growth has begun to slow. Jobs growth has averaged 190,000 jobs over the past three months compared to 282,000 jobs in the fourth quarter of 2015. Still, the 255,000 jobs added in the month of July are more than enough to absorb additions to the working age population. It marks the second month of respectable job growth after a hiccup in the month of May. It is realistic to say that we are approaching full employment, as the unemployment rate has dropped significantly from the high of 10 percent at the close of the Great Recession.

So are we at full employment? The economics jury is still out. Some journalists called it back in February. This seems premature, given lackluster wage growth and the puzzle of low labor force participation rates and employment to population ratios, both of which held steady this month at 62.8 percent and 59.7 percent, respectively.

Others are less optimistic. They claim that more evidence of wage growth and further decreases in unemployment are needed before we can begin to consider full employment. By one recent estimate, the unemployment rate would be at least 1.5 percent higher if many working age Americans had not given up their search for work.

Full employment is a slippery term in economics that basically boils down to a point in the labor market at which anyone who is willing and able to work is employed. This does not mean that the unemployment rate reaches zero; it can still be positive as people move from job to job, sorting to where they are best suited (call frictional unemployment) or from structural unemployment due to skills mismatch, technical change or industrial shakeups (think the decline of coal mining leading to coal miners losing their jobs). Though it can be seen as a moving target, progressive economists generally agree that somewhere around 4 percent unemployment is the sweet spot, with labor force participation rates at historic norms given the age of the population.

Care for a Beveridge?

A tried and true approximation can be found using a figure called the Beveridge Curve. Named after the English economist and progressive reformer William Beveridge, the curve plots the unemployment rate of an economy on the horizontal axis and the job openings rate on the vertical axis, measured by dividing the number of job openings by the sum of employment with job openings. The logic of the curve is simple: if the rate of job openings is equal to the rate of unemployment there will be just enough workers available to fill all the job openings without causing inflation.

While Beveridge himself thought that full employment was around 3 percent, the curve has more flexibility. The above figure shows the relation between the two rates over time via a rather untidy curve. Full employment exists when job openings and the number of unemployed equal one another, or when there are more job openings than unemployed people. As a stylized fact, there is a negative relationship between the vacancy rate and the unemployment rate.

It should be noted that the economy can overheat when job openings increase relative to unemployed people. This can lead to wage growth and increasing bargaining power for workers, but it can also lead to inflationary spirals as employers raise prices to counteract increase wage demands. There is plenty of literature discussing this from various political-economic perspectives, but the evidence lately is waning in a time when inflation is so low.

Thus, during expansionary periods or stints of full employment, the economy will center around the left side of the graph above, where unemployment is low and job vacancies are relatively high. Recessionary periods will be located to the right of the curve when unemployment is high and job vacancies are low. Recovery periods show a steady slink back to the left of the curve. Based on the Bureau of Labor Statistics (BLS) data from December 2000 to May 2016 plotted on the figure above, this relationship holds, demonstrating the increase in unemployment during the Great Recession and the dot-com recession of the early 2000s. Shifts in the curve can give us an idea of the strength of recoveries relative to other periods in time, but we should be cautious in jumping to the conclusion that the shift is precipitated by structural factors in the labor market.

Plotted over time, it is evident that the unemployment and job openings rates are converging since the recession (see figure below). When the two lines meet, we are at full employment.

There remains a gap between the openings rate and the unemployment rate in May 2016. In December of 2000 the rates were nearly equal: The unemployment rate of 3.9 percent was only 0.3 percentage points away from the job openings rate of 3.6 percent. At that time, the labor force participation rate was 67 percent, compared to 62.7 percent in July, meaning the gap is larger today. December 2000 was certainly a period of full employment.

Today it is harder to find work, with many giving up looking altogether, decreasing the labor force participation and employment to population ratio. The gap between job openings and unemployed becomes even wider when these missing workers are factored into the numbers, meaning we are further from full employment than the conventional unemployment measure suggests in the Beveridge curve. If missing workers were factored in, the curve since the end of the Great Recession would shift to the right.

Got Skills, Need Demand

There is a debate between whether cyclical unemployment or structural unemployment is the main driver of this gap. The kernel of the argument is whether unemployment in an economy can be identified as a result of the business cycle in which firms are laying off workers and not hiring new ones (or not), or if it is related to structural changes in the economy that mismatch workers and firms (skills gaps and the like) such that unemployment remains high in spite of many job openings. Based on a general Beveridge curve, structural unemployment would be evident if unemployment was high while the job opening rate was also high, meaning that employers are having a hard time filling jobs; conversely, cyclical unemployment would be evident when unemployment was high while job openings were low. While an outward shift in the curve suggests structural unemployment, cyclical factors could still dominate. Nobel Laureate Peter Diamond’s excellent 2013 paper on the Beveridge Curve discusses the difficulties of pinning down structural changes in the labor market.

We seem to be somewhere in the middle at the moment. The job openings rate in May was 3.7 percent while the unemployment rate was 4.7 percent, meaning there are still more unemployed people than job openings. A supply of people willing to work is there, but the demand for workers (job openings) is not.

What is more, we have a problem of missing workers, those that have given up looking for work because of weak opportunities. If these missing workers are included in the conventional unemployment numbers, the unemployment rate in May would jump from 4.9 percent to 6.5 percent. There is also evidence that people are becoming newly employed a month after being out of the labor force entirely, meaning that pool of available workers is clearly larger than those counted as officially unemployed. The long-term unemployed are dropping out of the labor force because of job shortages, while labor mobility is sluggish. This means that workers are not switching employers for better jobs and higher pay, and could point to slack demand for workers.

In a recent paper from the Roosevelt Institute, economists Mike Konzcal and Marshall Steinbaum find that wages and earnings have decreased in metropolitan areas and industries where mobility has declined the most. Workers that are currently employed are getting fewer offers to switch jobs for higher pay and better benefits. Relative to owners and managers, workers are not seeing the increased bargaining power associated with full employment.

Skills mismatch could remain a factor, but given that the job openings rate now is similar to the openings rate in the early 2000s, the skills shortage does not adequately explain the rightward shift in the curve relative to the low-unemployment era of the early 2000s. At most, the evidence is mixed and could be valid in some industries more than others. Given that some occupations could face labor shortages, job training programs could help bring some missing workers back into the labor force with the skills they need to find work, particularly for middle-skilled workers.

Looking to the Recent Past: Full Employment in the 1990s

The last time we reached full employment using this definition was from the late 1990s to 2000. The unemployment rate dipped to 3.8 percent in April of 2000 and sustained wage growth was seen throughout the economy. African-American workers and other people of color saw their incomes increase during this time, suggesting that full employment does indeed raise all boats. In order to address economic inequality exacerbated by structural racism and racial inequities, we must approach full employment. There is real reason for concern that economic growth will slow before we get there. Economists had hoped for a strong second quarter rebound in GDP, but data disappointed with an increasing number of economists worried about secular stagnation. At current rates of labor force participation we would need at least another year of job growth to return to pre-recession employment levels.

The Way Forward

What is the upshot of all of these conjectures about job openings and unemployed workers? It is that a meaningful definition of full employment is tied up with wage growth (average hourly earnings have risen 2.6 percent over this year), shared prosperity through widely distributed income growth and the idea that anyone who wants a job and is willing to work can find employment.

When the job openings rate and unemployment rate approach equalization, we know that people are more able to find work, workers are rewarded with higher wages because they are in high demand, economic growth proceeds and productivity can be enhanced.

The key is to look at policies that can push us toward the full employment goal. Discouraged workers can be brought back into the labor force through a variety of policy options. Skills training and youth apprenticeships can deal with some of the skills mismatch kicking around in the economy, particularly among at-risk youth, but bolder fiscal policies will be necessary to generate demand for workers.

Monetary policies of cheap money and low interest rates are not enough to transmit economic growth to workers in the form of higher incomes. In a time when money is cheap via low interest rates set by the Fed, we cannot afford to pass up this opportunity to generate demand through infrastructure spending, subsidized jobs programs, and expansionary fiscal spending to get us over the hump to full employment. Deficit spending could do this economy some good, now that deficit hand-wringing has taken a backseat in economic policy circles.

We have the opportunity to reap the gains of full employment by sowing a little fiscal spending now, when interest rates are low and the cost of borrowing money is cheap. Government spending is as much of a bargain now than it can ever be. The recovery since the recession has not been swift; a little demand priming would go a long way to get more people employed and more working families the income they need to thrive.