More on Unemployment Duration
Earlier I had posted a graph showing the *mean* duration of unemployment spells, and I used it to buttress my claims that something was screwy with the labor market. But in the comments Kevin Erdmann seemed to be disagreeing with me when he wrote:
I think most of this is due to the extreme extended unemployment insurance policy that was put in effect during the crisis. It’s hilarious that people tried to argue that extended UEI didn’t increase unemployment. It’s so extreme and it’s specific to the duration range that was covered by insurance. Interesting that some of that remains. I think the remaining average is greatly influenced by a small contingent of workers with VERY long durations. I bet the median figure is back to normal.
https://fred.stlouisfed.org/series/UEMPMED#
Yep. Close. I think most of the remaining drift up is demographic. Older workers tend to have lower UE but longer durations. Again, you can get age specific numbers from [BLS], and those will be more stable. [Bold added by RPM.]
OK, so even on his own terms, that wouldn’t blow up the “structuralist” story, to say that the mean (but not median) unemployment duration shot way way up, because there were a segment of workers–how many? 2 million?–who just got ejected from employment prospects after the housing bust.
But is Kevin right when he says that if we look at *median* duration, we’re “back to normal”?
Not at all:
So I understand why Kevin eyeballed that and thought we were out of the woods; it’s a natural reaction to that graph. But it’s wrong.
Look more closely. Yes, the median duration has fallen way down, but its current level is still higher than it was, going back to the 1960s, except for two brief periods, one of which was the depths of the early 1980s recession that at the time was the worst since the Great Depression. So no, we are not at all “back to normal” or “Close” as he says, specifically.
(If you slide the right bar to the left, so that the x-axis ends in 2008, then the y-axis rescales and you can see how high 9.6 weeks is, historically speaking. It just seems pretty low when the whole y-axis gets crunched down to showcase the incredible awfulness of the Great Recession.)
Now let’s try something else. I’m going to plot the same series–median duration of unemployment–against the unemployment rate:
Now do you see it, boys and girls? Something is screwed up with the labor force market. The official unemployment rate is masking deep wounds, and the “structuralist” story makes a lot more sense than “there wasn’t enough spending” story.
P.S. Kevin and I agree about extending unemployment benefits. Where we disagree is in him thinking we’re back to (close to) normal on unemployment duration, or that this doesn’t bear on the structuralist vs. demand interpretations. (Note that Krugman mocked the if-you-pay-people-not-to-work-then-they-won’t-work idea at the time.)
I think we agree on more than you think, because of my inelegant verbiage. Your graph of mean durations showed long term ranges between 10-20, with the current level still above 25. The median has run between 5-10 and is now back down to 10. I don’t know if this is the nadir or not. Maybe it will drop a few more points. By close, I meant we are at least back in the long term range.
As I said, aging accounts for some of the rising trend – maybe a point or 2. There seem to be additional frictions in the marketplace that also are pulling up the number, and I didn’t mean for any of my comments to be a denial of that.
The main point I was trying to make was that emergency UEI was a huge structural influence. You can see that in the difference between median and mean duration. The rate of exit clearly shifted at the 26 week mark when emergency UEI was implemented. There is a lot going on here, but clearly UEI pushed up durations and the stated unemployment rate quite a bit. Even today, there appears to be a set of workers claiming to be unemployed for a very long time – more than 2 years, so that there appears to be a somewhat normal unemployment market plus this contingent of workers with very long durations. That pulls up both the mean and the median, but it pulls up the mean by more. So, the shorter durations are closer to normal, but the 26 and over group remains much larger than previous.
The average duration of workers unemployed for less than 26 weeks had ranged between 7-9 weeks for many years. It bottomed out at more like 7.5-8 weeks in the last expansion. It’s back under 8 weeks now. A little high, but in the bottom half of the range. The average duration of workers unemployed for more than 26 weeks had ranged from 50-60 weeks for many years. After emergency UEI, it moved up to more than 80 weeks, and last I checked, earlier this year, it was still over 80 weeks.
Even though emergency UEI has been gone for a few years now, the unusual exit behavior of long term unemployed workers remains. The anomaly is so extreme and so parallel with the implementation of emergency UEI, it seems to me the burden of proof against EUEI as the cause. In fact, since much lesser versions of EUEI were implemented in the previous two recessions, it might be that some of the persistent shift up in durations even before 2008 is related to that.
I was originally convinced that EUEI had induced workers to extend their non-employment. But then I saw the study I refer to here:
http://idiosyncraticwhisk.blogspot.com/2014/08/long-term-non-employment.html
It claims that their hasn’t been much change in the incidence of very long non-employment. This would suggest that while EUEI caused stated unemployment to rise, it could be mostly a reporting issue, that the availability of long term insurance induced workers to report unemployment, but that it didn’t change their employment behavior significantly in either direction. Maybe the use of EUEI has just led to a cultural shift where workers are more likely to identify as unemployed instead of as out of the labor force.
In any case, it seems clear to me that EUEI had a massive effect on durations that seems to persist even after the program is ended, and that political inclinations are leading to its implementation earlier in recessions, with more generous terms, that are held until further into the recovery. Look for calls to start pushing it to 50 or 75 weeks even if the unemployment rate gets above 5% or 6%.
By the way, your last graph is partly a product of demographics. Older workers tend to have lower unemployment, but they tend to be unemployed for longer durations when they are unemployed. This doesn’t explain the entire separation, certainly only a small part of it after 2008. But, as with so many measures these days, some of the odd behavior is simply a reflection of how old people behave. Ballpark, durations are probably a percentage point higher when adjusted for age and unemployment a percentage point lower, compared to 1990. Don’t quote me on that, but it’s probably a decent ballpark estimate.
Oops, I mean durations would be lower and unemployment higher, if population was distributed more smoothly.
The wages data series in FRED are all over the place and difficult to work with… but I rummaged around and found some representative items from a range of industries. In general there’s divergence when plotted on log-scale the right hand side is noticeably wider than the left hand side so the unskilled wages are becoming a smaller percentage of skilled wages (Waaa! Waaa! Da Ineqalideee!!) but also we can see the price mechanism is working within certain industries e.g.:
* IT workers took a wage hit after the 2000 “dot com” crash (no surprises there).
* Airlines took a dip around 2002 to 2007 but have recently rebounded.
* Automobile manufacturing is doing relatively worse than other industries at a similar skill level.
* Wages in real estate took a big dip around 2006 – 2008 although construction wages did not (maybe more commission sales in real estate and more contract wages in construction).
* Farm wages are highly variable… no idea why that happens, maybe depends on harvest.
https://fred.stlouisfed.org/graph/fredgraph.png?g=gSOR
So it makes perfect sense to me that with a lot of manufacturing (especially automobiles) getting heavily internationally competitive and more imports coming in, the manufacturing workers are losing relative wage position. Also with lots of unskilled immigration (e.g. Mexico, Pakistan, Philippines, Bangladesh , Brazil, etc) there’s competitive downward pressure on the wages of the least skilled workers while much less pressure on the most skilled workers.
Specific events such as the dot com crash end up making a significant impression over the 10 years following that event, although they eventually sort themselves out.
Paying people to be unemployed probably is suboptimal, but it has been said the standard is not perfection, the standard is the alternative. Now, some of these unemployed people would probably sign up for the military if they could not survive unemployed. Paying people to occupy foreign countries, or, worse yet, to kill and bomb other people, is clearly much worse, from the pacifist perspective, than paying people to be unemployed. So, before abolishing unemployment insurance, soldiers should be helped to transition to non-violent professions. Or even just less violent professions. A truly non-violent profession is probably nearly impossible to find in a first world country, anyway, so we would probably have to settle for less violent professions.
If you were a real pacifist, which you aren’t, you would be more concerned about military spending than about unemployment insurance.
Isn’t it kind of odd to assume that if someone is unemployed without sufficient benefits, then he will become a killer?
I agree with you about the importance of reducing military spending, but it’s an unfair criticism in this case, for the following reason: Just because I don’t want to eat waffles for breakfast, that doesn’t mean I do want to eat pancakes. I don’t have go around saying, “By the way, I don’t want pancakes, either!!” just so I can say with credence that I don’t want waffles.
Not a specific person, but a statistically significant number of people. That is part of the point of unemployment, after all, from the perspective of war hawks. Just how many people would sooner starve than take up arms, anyway?
During World War I, for example:
That’s a start.
The pages and pages and pages of pro-war propaganda on Institute for Energy Research is sufficient to prove that Bob Murphy is nowhere close to being a pacifist.
Just look at what Shell did to the Nigerians.
https://imgur.com/a/DjuNc
Bob Murphy’s failure to make any distinction between violent and nonviolent employment, instead focusing on the unemployed as if they’re the major problem, just makes it even worse.
First of all, just because some striking trolley workers joined the military during WWI doesn’t mean that taking away welfare benefits from modern Americans will drive them to enlist. It’s a conceivable story, but that doesn’t make it an empirically relevant story. Unless you can cite empirical evidence of this happening today, under current circumstances, then it is mostly just wild speculation.
Second of all, you concede that your objection to Bob mainly involves other work of his, elsewhere. That’s precisely what I meant when I said “it’s an unfair criticism in this case.” In this case, it comes out of nowhere as a massive non sequitur.
How about this?
Or this?
Many people who receive some sort of welfare, charity, or insurance are, no doubt, spending a portion of their benefits on things that were made by slaves, probably without their knowledge. Or simply receiving things made by slaves, in the case of charities that hand out food and other items rather than money. Many employed people are doing the same. But occupying foreign countries or, worse yet, participating in the killing and bombing of foreigners, are among the worst things someone could do. Incentivizing violent professions is far worse than incentivizing unemployment.
You do have a point. However, we cannot suddenly pretend that we really believe Bob Murphy is a pacifist whenever we read something of his not written on Institute for Energy Research.
It would be like trying to read the writings and speeches of Mao without remembering all the executions and other horrible actions taken by Mao.
Take this, for example.
If we didn’t know Mao said that, or knew nothing of Mao’s atrocities, we might see it as an inspirational speech. But we do know Mao said that, and we know he committed many atrocities. And that makes it impossible to see it in the same light as if it were just a speech by an unknown person. What was Mao trying to inspire people to do? Execute each other?
“Also with lots of unskilled immigration (e.g. Mexico, Pakistan, Philippines, Bangladesh , Brazil, etc)”
Are there really lots of unskilled migrants from these countries?
There are 23.2 million Mexicans and their families in the USA, these are economically disadvantaged than average. From Pakistan there are nearly 500,000 from Pakistan and 277,000 from Bangladesh, but education levels are higher than average as is household income in both cases. 2.9 million from the Philippines, but again the educational attainment is relatively high.
https://www.migrationpolicy.org/research/select-diaspora-populations-united-states?gclid=CjwKCAiAsejRBRB3EiwAZft7sOJ5BXO_lnIIKGTYWjsX779HaUSh8sIjJcSEv9A6ppJZ6Z0ZGzXynRoCgpYQAvD_BwE
Pakistan, Bangladesh, and Brazil are all countries with significant slavery problems. The term unskilled may have historical origins in being used as a slur to justify slavery. For example, picking cotton is often though of as unskilled labor.
But in fact, picking cotton fast enough to meet ever-rising quotas took quite a bit of skill under slavery. The increase in skill level of cotton pickers was driven by torture. According to Edward Baptist, “Using torture, slavery’s entrepreneurs extracted an amount of innovation virtually equal in numerical measure to all the mechanical ingenuity in all the textile mills in the Western world.”
Those are interesting statistics Harold, but they are dealing with families who have been in the USA a long time (not detailed exactly how long). Even second generations are included there.
Comparing the average education levels of select groups (younger on average) with the general population (older on average) is dodgy because a lot of older (retired) people in the USA will have learned skills on the job (e.g. manufacturing, trades, military, etc) but not have university qualifications. It’s not really comparable with the modern generations who rush to university (rightly or wrongly).
Also, modern paper qualifications are without a doubt watered down. Even with comparing income, they would need to match up age cohorts because the general population contains a lot of retired people on low income (but usually with assets like a house paid off, so they don’t need such a lot of income either).
Anyhow, I never claimed that immigrants just sit forever in unskilled jobs. Of course they work their way up the ladder over time. The pressure on unskilled wages comes from the rush of NEW immigration, even if they do have qualifications from overseas often local employers are reluctant to recognize these, so they are left climbing the ladder from the bottom. If immigration is kept at a level where they can trickle up through the skills and jobs market then it comes to a steady state. Here’s a chart of overall numbers (ignoring origin).
https://www.migrationpolicy.org/programs/data-hub/charts/immigrant-population-over-time
You can see a big turnaround in 1970 where the actual flow itself increased, and has continued increasing. So each year the rate of intake gets larger, and to be fair the population as a whole gets larger too, but does not appear to have reached any steady state just yet.
It isn’t well documented but I personally suspect that many illegal immigrants (i.e. without work permits) usually sell their labour below minimum wage, severely pushing down the wages in unskilled areas and disadvantaging those who would want to work legally. These guys probably don’t get studied a whole lot because not many people put their hand up to admit they are breaking the law (and the employers are also breaking the law so they keep quiet about it). Because of the difficulty caused by illegality, these guys have more trouble working their way up via skills, education, etc and they get excluded from studies so they don’t get counted in the statistics.
Not just illegal immigrants. It can happen with legal immigrants too. And the existence of subminimum wage workers has been studied, some, and not only in the context of forced labor.
One thing employers sometimes do is pay per pound picked, and doctor the hours after the fact to make it look like they are paying minimum wage. Additionally, child workers often do not receive their own pay stub. Instead, their work is simply counted on their parents’ pay stub.
This one specifically mentions “guestworkers”, which implies some sort of legal work permit.
And then there’s the problem of farmworkers being raped. Welcome to capitalism, the rape economy.
https://www.pbs.org/wgbh/pages/frontline/social-issues/rape-in-the-fields/female-workers-face-rape-harassment-in-u-s-agriculture-industry/
https://www.npr.org/2013/11/05/243219199/silenced-by-status-farm-workers-face-rape-sexual-abuse
https://www.huffingtonpost.com/2013/06/24/female-farmworkers-rape_n_3491939.html
There is an enormous number of unskilled emigrants from Bangladesh. They don’t typically find their way tot he USA. Instead, they go to Dubai, Bahrain, Saudi Arabia, Kuala Lumpur, and Singapore.
Indeed, lots of emigrants from Pakistan (for example) does not equal lots of immigrants to the USA.