Missing Workers: What the Unemployment Rate Isn't Saying About the Recovery

Dan Henry/Chattanooga Times Free Press via AP

In this April 2, 2015, photo, a crowd gathers for a huge 15-county job fair at The Colonnade in Ringgold, Georgia. 

Later this month the Federal Reserve will decide whether or not to begin raising short-term interest rates to slow the pace of economic recovery. If they do, it will be the second major step they have taken to pull back the historically unprecedented support that they’ve been providing the economy since the onset of the Great Recession in 2008. Late last year, they stopped purchasing long-term government bonds in an effort to push down long-term interest rates directly.

A decision to raise rates would be an implicit statement by the Fed that the agonizingly slow economic recovery from the Great Recession has finally achieved self-sustaining escape velocity. This positive view of the economy’s health would likely surprise many Americans; in May 2015 more than half of respondents to CBS/New York Times poll said that the economy was in “poor shape”.

So what is the Fed basing its relatively optimistic view of the recovery on? Largely on the undeniable progress made in recent years on lowering the unemployment rate. The unemployment rate peaked at 10.0 percent in October 2009, but since the middle of 2010 it has steadily declined. In the last month available, it reached its lowest level since March 2008—5.1 percent.

Yet the unemployment rate is outperforming almost every other indicator of how close the economy is to full recovery. This raises the question of just how much weight we should put on this measure alone to make crucial policy decisions—like when the Fed should raise short-term interest rates. Increasingly the answer to this—recognized by almost all serious analysts—is less and less. But even given this growing realization that the headline unemployment rate should be just one indicator of how the economy is performing, it still drives too much analysis of where the economy is and where policymakers should try to take it next.

So why has unemployment become the most-examined economic indicator? And how does today’s rate likely underestimate overall economic weakness? Is this a function of how it is measured; can we estimate a “true” unemployment rate? Or is there some other indicator entirely that should be front-and-center in the minds of policymakers?

Why so Much Attention on the Unemployment Rate?

There are many historical reasons why the headline unemployment rate became so prominent in policy discourse. But a key reason is the role it plays in linking together the two statutory mandates of the Federal Reserve. Attaining “maximum employment” is one part of this mandate. And a long history of economic theory—from Marxian theory to modern macroeconomics—has posited a negative relationship between unemployment and wages. Simply put, when unemployment is high, it is very hard for workers to demand higher wages. So this link between unemployment and wages leads to the second part of the Fed’s statutory mandate: price stability. Falling rates of unemployment should lead eventually to higher wages for workers. Given that the overall cost (and hence price) of most things in the American economy depends mostly on labor costs, this means that falling unemployment that pushes wages up can feed through eventually into higher rates of price inflation.

Since the Fed is mandated to balance maximum employment and price stability, a close monitoring of unemployment makes sense. Much of modern macroeconomics has gone even further, arguing that there is a natural rate of unemployment that policymakers should aim for (sometimes referred to as the Non-Accelerating Inflation Rate of Unemployment, or NAIRU). This natural rate represents the rate below which workers’ wage demands begin outpacing the economy’s ability to deliver them (with this ability-to-deliver measured generally as productivity growth). Stimulating the economy beyond this natural, the theory goes, will spur not just inflation, but ever-accelerating inflation. In the current debate over interest rates, some economists have argued that the labor market is rapidly approaching this and that keeping rates low for much longer could result in a reignition of inflation.   

Policymaking based on strong claims about the natural rate of unemployment has always been a flawed endeavor. For one, there are plenty of determinants of wages besides the unemployment rate, and policymakers do not take these into account strongly enough. For example, the historical episode often pointed to as evidence that the Fed must always and everywhere be a vigilant inflation-fighter when setting interest rates is the wage-price spiral of the 1970s. Sparked in part by soaring oil prices, the 1970s saw a lethal combination of sharply rising inflation, dwindling productivity rates, and high unemployment that together marked the end of the postwar economic boom. Inflation hawks today point to this experience as proof that the Fed must raise interest rates when unemployment reaches its “natural rate.”  

Yet the tinder for the ‘70s wage-price spiral just don’t exist today. American workers going into the 1970s had seen five years when unemployment averaged 3.7 percent, and had seen 25 years of wages across-the-board matching the growth rate of productivity. More than a quarter of private sector workers were represented by a union contract (often indexed explicitly to provide wage increases when prices rose) and the federal minimum wage’s inflation-adjusted value had reached its historical high-point in 1968.

Conversely, today’s American workers have seen unemployment average 8 percent over the previous five years and the vast majority have seen wage-growth lag severely behind overall productivity growth. Unlike the early ‘70s, today only 7.4 percent of private-sector workers are represented by a union contract, and today’s inflation-adjusted value of the minimum wage is more than 30 percent lower than it was in 1968.

And even the tinder for the ‘70s wage-price spiral needed a large spark—the quadrupling of oil prices that characterized the decade. Given all of this, the idea that any natural rate of unemployment should be the same today as in early 1970s seems very hard to credit.

Further, real-world estimates of the natural rate are extremely imprecise. A review of NAIRU estimates by leading econometricians in the late 1990s noted that “…these estimates [of the NAIRU] are very imprecise; the tightest of the 95 percent confidence intervals for 1994 is 4.8 to 6.6 percentage points…they suggest that debating over whether the NAIRU is 4.5 or 5.5 or 6.5 percent does little to enlighten monetary policy”.

All of these shortcomings were well-known before the Great Recession, yet recent years should make us even less confident that recent progress in reducing unemployment will translate quickly into higher wages and increased inflation.

The Unemployment Rate is too Optimistic About the Recovery

The unemployment rate is based on a monthly survey of American households. It is calculated by dividing the number of workers classified as “unemployed” by those adults considered part of the “labor force”. A key twist is that “unemployed” does not simply mean “jobless” and the adult “labor force” is not the same thing as the adult population.

To be considered part of the labor force, one must either have a job or be actively searching for one. And the criterion for this active job-search is pretty stringent: you must have engaged in job-search activity in the same month that the labor force survey is taken. If somebody is jobless but has not actively searched for work in the last month, they are not classified as unemployed—in fact they’re not even considered part of the labor force.

But what if the economy remains so weak so long that many workers who would clearly prefer to work are jobless yet have largely given up on active job-searching? In this case, the unemployment rate can fall because of the shrinkage of the overall labor force. And this shrinkage has been quite rapid over the course of the current recovery.

For example, in the first two years following the unemployment peak in October 2009 fully 60 percent of the decline in unemployment was driven by falling labor force participation instead of job-creation.

More recent declines have been more helpfully driven by relatively robust job-creation, but it’s clear that today’s unemployment rate is low in part by the fall in labor force participation that happened over the cycle. Some have claimed that this fall would’ve happened even without an economic crisis, with the aging and retirement of the Baby Boomers. But simple aging of the workforce does not come close to explaining the entirely of this decline.

Where I work (the Economic Policy Institute), we have used estimates of pre-Great Recession trends within various age/gender groupings to forecast what labor force participation would be without the decline caused by the Great Recession itself—trends like the retirement of baby boomers. We then call the gap between actual labor force participation and what these pre-recession forecasts imply for today’s labor force participation an estimate of “missing workers”—people who would be employed or at least actively looking for work today had the economic weakness of the past seven years not occurred. While this missing workers estimate is pretty volatile month-to-month, there is little evidence that it has yet shrunk reliably to less than three million.

An even more straightforward way to purge labor market trends of demographic influences is to simply look at the employment-to-population ratio (EPOP) of “prime-age” workers—those between the ages of 25 and 54. This is a group with traditionally very strong labor force attachment, and they really should not be voluntarily retiring or sheltering in college in huge numbers. This measure has recovered less than half of the fall it took during the Great Recession and early recovery. More worryingly, it has been flat-lining since January 2015.

The Proof of Labor Market Slack Today? Wages

So, by the measure of unemployment, the American labor market looks much more recovered than by the measure of prime-age employment-to-population ratio or labor force participation. Is there some extra bit of evidence that can give us a clue which of these measures is telling a truer story about the state of economic health?

Yes: wage-growth. Remember that the entire point of monitoring unemployment is to forecast the increase in wage-costs that may threaten to bleed through to overall price inflation. If the unemployment rate says that the labor market is tight, but the prime-age EPOP says it’s not, one can look at the behavior of wages to figure out which seems more correct. If wages are rising smartly, then the unemployment rate would seem to be telling the true tale. If wage-growth is flat, then the slack indicated by the prime-age EPOP would seem more reasonable to assume.

The wage data are so far unambiguous over the recovery. Wage-growth has been slow and flat—running at just over 2 percent (not adjusted for inflation) over the course of recovery. Combine this 2 percent nominal wage growth with steady (if unspectacular) productivity growth, and growth in unit labor costs (how much businesses have to pay in wages for each unit of output they produce) is running at just 1 percent per year. And in fact these incredibly anemic unit labor costs are a prime reason why the Fed is having trouble getting overall price inflation to meet even its own too-conservative target of 2 percent. In fact, the only reason that prices have risen perceptibly at all over the course of the recovery is that corporate profit margins have reached historically thick levels.

Luckily, the Fed realizes that examining only the unemployment rate would be a mistake. They have in fact identified an entire “dashboard” of economic indicators to guide their policy choices. Given that wage-growth is the key economic link between the two prongs of their mandate (maximum employment and inflation stability), many have argued that nominal wage-growth and a long-run target for nominal wage-growth, should be given an especially prominent place on this dashboard.

After all, arguing about whether or not American workers will choose to participate in the labor force more when the economy returns to genuine health for a sustained period of time is fun for academics. But when it comes to policy—and when it comes to how actual Americans view the health of the labor market—focusing on wage-growth seems like a much more tangible way to orient the conversation.

You may also like