Evan Soltas
Apr 30, 2012

Locked Out

Labor markets are healing in the US, but not for outsiders

lock doors....forever
I want to be enthusiastic about declines in the most common measure of the unemployment rate, now at 8.2 percent, in the beginning of a decline in more broadly defined measures of unemployment, such as those which include "discouraged workers" and those who are marginally attached to the labor force, and in improvements in the prime-age employment-population ratio.

But I am having trouble being enthusiatic. And this is not only because of recent ugly numbers in initial claims, for which I haven't heard a good explanation that doesn't involve a faulty seasonal adjustment which caused economists to overstate the economic gains of in the last two quarters.

This is because the economy is healing, but only for some people. And for whom the economy is not healing, things remain as bad as they were in 2009-2010.

I am most concerned that labor markets in the US are beginning to work against outsiders.

For those of you not familiar with the insider-outsider theory of unemployment, here is a great overview of the research. In short, the theory says that participants in labor markets can be divided into two categories, "insiders" who are employed, have been so for a while, and have considerable job security, and "outsiders" who are unemployed, have recently been rehired, or anyone who does not have considerable job security. These two classes of labor-market participants exist because of transaction costs which make "insiders" and "outsiders" imperfect substitutes, which gives "insiders" bargaining power and generates a conflict of interest by which hysteresis in labor markets can be explained.

The 2008 recession made a lot of "insider" workers "outsiders," both through layoffs and fires, but also because employment growth was weaker, and so new entrants into the labor force struggled to find jobs, or they chose to not officially join the labor force. The effect was to diminsh the insider-outsider distinction, because, in essence, there were so many people unemployed who were nominally "outsiders," but had such strong employment prospects that there was little doubt they would soon regain the "insider" distinctions. Now, as the normal unemployment rate falls and labor markets, at a superficial level, begin to normalize, we are beginning to see the insider-outsider gap reemerge.

Think about it this way: imagine you're at the beach. With you, you have a sieve. Reach down, and feel the sand with your fingers. The sand is uneven and coarse -- there are many small grains of sand, but there are also quite a few larger grains of sand. Maybe even a few pebbles here and there. You dip the sieve into the sand. In the beginning, it seems as though all the sand will pass through, because with so much sand in the sieve, there are always small grains which can fall through the holes and out of the sieve. But as the amount of sand in your sieve diminishes, so does the stream of sand coming out the bottom. You begin to realize that there is no way your sieve will fully empty. There are big grains and pebbles which will not pass through.

In the recovery from the 2008 recession, we are beginning to see the larger grains and pebbles. The problem is that are way more of them than I thought there were, more than there were before the 2008 recession, and more than this economy should have. In sum, the "sand sieve" approach to understanding hysteresis in unemployment suggests that the dramatic progress in the decline of the unemployment rate will begin to slow, and that structural unemployment may be higher than is currently thought.

(Sidebar: I don't believe that structural problems are an excuse for not taking action, or lesser action, in the short term. The primary cause of all our problems is -- far and away -- a shortage of nominal spending and aggregate demand. The only problem with just seeing the demand shortage, however, is that the demand shortage can generate structural disruptions in supply, through hysteresis, skill atrophy, and similar avenues. In this respect, structural problems become an argument for, and not against, an immediate policy response.) 

We can see the emerging insider-outsider disparities by looking at unemployment data carefully.

First, let's look at the fraction of workers who have been unemployed for 27 weeks (half a year) or more but are still searching for work. This indicator is cyclical and lagging, but what we see is disturbing: first, this recession generated a particularly large change in the fraction of unemployment which is long-term, and second, it hasn't gone away at all.Now, let's look at labor markets in a different way -- I haven't seen this done anywhere else, but I think it tells us an important story. Although the "sand sieve" is emptying -- unemployment rates are falling -- one of two things is happening: (1) either there are far more pebbles than there once were, or (2) our sieve is broken. To move out of metaphor, either the level of structural unemployment has risen, or our labor markets have changed in a way that systematically reduces the value of work from the long-term unemployed. (If the latter is true, plausible explanations include an increase in the rate of skill atrophy.)The graph you see above contains two metrics of labor markets which I've developed, both of which give us a reading on the severity of insider-outsider distinctions in labor markets.

The first, in blue on the left vertical axis, tells us the fraction of the unemployment rate which is new entrants and reentrants to the labor force. It is cyclical because during recessions, cyclical unemployment increases its share of total unemployment relative to frictional and structural unemployment. What I find disturbing about this measure is the rate at which it is rising, and the level it has attained. Both are unusual given the unemployment rate -- we would expect the cyclically unemployed (who are not be new entrants or reentrants to the labor force) to make up a larger fraction. The fact that they don't tells us that the remaining unemployed people in the US are harder to reemploy than the two percentage points we've taken off the unemployment rate already, as the still-unemployed are "pebbles" -- due to their "outsider" status in the labor market, less attractive to employers and less likely to be hired. These are the 20-somethings who couldn't get jobs after college, or those who have been out of work for a very long time but want to give it another shot.

The second, in red on the right vertical axis, tells us the ratio between two measures of the duration of unemployment: mean and median duration. When the ratio is very high, it implies that there is a meaningful fraction of the unemployed who have been so for an extraordinarily long time; they pull up the mean duration of unemployment, but not the median. Coming out of this recession, the median duration of unemployment has been falling, because, fortunately, it is becoming easier for the median unemployed person to find a job. But the mean duration has not fallen -- even as the unemployment rate falls, there remains a contingent of unemployed persons who, despite their job search, have been unemployed for months and months, if not well over a year. Given that the unemployment rate remains elevated, we would expect less of an insider-outsider distinction to exist at this point -- if everything was alright in labor markets, there should still be a lot of fine-grained sand that can pass through the sieve, a lot of "near-insiders" who can find jobs quickly -- and the fact that we see this distinction appearing so sharply and so early on in the employment recovery raises the same question about the number of "pebbles" -- the "outsider" structurally unemployed -- or the functioning of labor markets.

One way we can visualize this is by looking at it in the same way economists look at the Beveridge curve. Recently, there was concern that it had shifted -- reflecting concerns about the efficiency of labor markets -- but these were, I think, largely assuaged when we realized that the counterclockwise movement was normal cyclical behavior. (See here for more.) I found a similar outward shift in the graphs of these two measures against the unemployment rate. It cannot, I fear, be adequately explained by normal business cycle behavior -- I didn't see any distinct outward shifting, especially of this magnitude, happening in other recessions. Here are the graphs:The problem is staring us right in the face -- the curves suggest that if labor markets haven't changed their behavior, the "outsider" structurally unemployed rate may have risen by somewhere on the order of 1.5 to 2.1 percent of the labor force.