Key Unemployment Indicator Signals Recession Might Be Ending

The 4-week moving average of weekly initial unemployment claims reached its highest point nine weeks ago, on April 8. Since that peak, at 659,500, the metric has fallen into a range between 632,000 and 624,500. The most recent reading, 631,000 yesterday (June 4 report, May 30 data), is still close to the peak. This indicates that there is still a lot of pain, with nearly double the number of people applying for the first time for unemployment benefits as would be the case in a healthy economy. But nine consecutive weeks below a peak has, in six of the seven occurrences, indicated a recession has ended. The one exception occurred in the 1969-70 recession, when the first peak, occurring 30 weeks before the recession ended, was followed by a second peak only five weeks before the recession ended.

The chart, courtesy of the St. Louis Fed FRED data base, is shown below:

Recent articles discussing weekly claims and the end of recessions have been posted by Lounsbury, Iacono and Miller.

Two other employment indicators that could flash signals within the next month would reinforce the weekly claims peak signal. One of these is from the monthly report from the Bureau of Labor Statistics, U.S. Dept. of Labor. A minimum in the weekly hours worked in manufacturing is an early indicator that a recession is ending. This number rose in the early May report (covering April) from the level for March. If the report for May (due today, June 5) and for June (due in early July) remain above the March minimum, a second signal will be made that the recession has ended in the second quarter.

The third indicator to watch is the number of continuing insured unemployment claims (the insured unemployed number). The 4-week moving average of insured unemployed has also peaked nearly coincidentally with the end of recessions. As of this week it is still rising, but at a slower rate. Jeff Miller has cautioned that this number may not peak in as timely a manner in this recession because of the high level of extended benefits implemented.

Unemployment data is too often dismissed as a “lagging indicator.” If you look at the right pieces of data it can actually be concurrent or slightly leading economic turning points. This subject, including all three of the statistics mentioned in this article, as well as some less timely data, is covered in more detail here.

Data can be followed at dol.gov, updated every Thursday at 8:30 am Eastern Time, for weekly initial unemployment claims and insured unemployment totals. Monthly employment data is updated the first or second Friday of each month at bls.gov.

Comments

[Pat C:]

Please keep in mind the timing of the federal extension of unemployment benefits. Mine ran out last week, unemployed then, unemployed now just running out of benefits. Job market is just horrible.

Jun 05 09:50 AM

[traden4alpha:]

Nice chart and analysis. I agree that unemployment can be a leading indicator, especially in this particular recession. Surely, unemployment is a good predictor of future delinquencies and foreclosures with the peak in unemployment-related foreclosures occurring many months after a person losses their job. I'd also imagine that even if the jobless person finds employment, if they are forced to move to get that new job, then they may be forced to allow foreclosure on an underwater home. Thus, initial claims and continuing claims are strong predictors of future foreclosures with all the downstream pain that that causes to banks and housing prices.

One small quibble is that the chart looks like it shows 4 other cases in which a drop after a peak didn't signify the end of a recession (early in the 1974 recession, mid-way in the 1982 recession, 2/3 of the way in the 1982 recession, and in the middle of the 2001 recession). Perhaps these false lulls in initial claims didn't obey the 9-week threshold, though.

My bigger concern with the current unemployment data is in deciding whether to pay attention to the number of folk employed vs. the number of folks unemployed. The unemployment rate jumped quite a bit more than expected. Perhaps the high growth in unemployed people represent optimistic re-entrants. But they could also represent people in financial trouble (e.g. retirees with decimated savings, college students dropping out to earn money, or stay-at-home spouses being forced into the workplace due to job loss by the family breadwinner). In any case, increased competition for employment will hold wages down and decreases the chances that truly needy job seekers find employment.

Finally, I wonder what this means for the market. During the last recession, the stock market suffered nearly a year of declines AFTER the official end of the recession.

Jun 05 10:02 AM

[paintismotion:]

I do not believe things are improving. It is impossible to improve short term. We have lost too many manufacturing jobs - period. We have reached the tipping point and it is a long grade downhill. No one is hiring on multitude of laid of workers running out of UI benefits in this sad economy! I suggest the lower numbers suggest manipulation of the figures by the gov't and when you compound this with the fact there are people coming off of unemployment, who have simply given up, or who do not report statistically, we are in a real collective mess. There are just no jobs. The numbers will decelerate too while the ghost numbers of real unemployed (and homeless) expand because people will simply give up looking and hope and expec a government bailout - of the unemployed, the unemployable abd those unable to find work or not looking any more. Its that bad. Oh, did I mention the government is BROKE!

Jun 05 10:07 AM

[John Lounsbury:]

traden4alpha - - -

Good comments. First, what does this mean for investors? Historically, this point has been a good time to increase investment in stocks, as discussed in this analysis.

You make a valid point about what I call "headfakes" - short-lived drops in weekly initial unemployment claims that reverse and go higher. The examples you pointed out, plus another in the past December and January, are discussed here. You are correct that these "headfakes" are the reason for the 9-week delay before accepting the peak as a signal.

Finally, this morning's numbers indicate that there has been a large increase in the number of people in the "work force". Thus, even though the increase in the number of unemployed was reported today to be less than in previous months, the larger increase in the work force produced a surge in the unemployment rate. The work force is one of the definitions that has caused some discussion. One detailed discussion of this can be found here. The definition of who is in the work force does not count more than 75 million working age people. Obviously, there is considerable room for people to fall through a 75 million people wide crack (or, as a mountaineer, I am tempted to say crevasse).

Thanks again for a very insightful comment,

Jun 05 10:52 AM

[John Lounsbury:]

traden4alpha - - -

Sorry - hyperlinks intended for the preceeding comment were dropped by the SA software. I will give them again:

For the use of the initial claims peak for timing investments is www.thestreet.com/stor...

The "headfakes" discussion and 9-week delay before accepting a peak is www.thestreet.com/stor...

The discussion of how the Dept. of Labor analyzes employment data, including the elephant in the room, the large number of working age people not in the DOL defined "labor force" is seekingalpha.com/artic...

Jun 05 11:00 AM

[John Lounsbury:]

paintismotion - - -

You said: "We have lost too many manufacturing jobs - period. We have reached the tipping point and it is a long grade downhill. No one is hiring on multitude of laid of workers running out of UI benefits in this sad economy!"

Your concern is reinforced by the detailed Dept. of Labor report this morning. A key leading indicator is weekly manufacturing average hours worked, which fell in May to a new cycle low (preliminary reading).

There are three leading indicators I have identified out of all the employment data published by the government: (1) the intitial claims number discussed in this article, (2) the average weekly hours worked (manufacturing) and (3) the number of insured unemployed (still rising). If you have the time to do the reading, these (and other) reported data are discussed in the three references for which I provided links above.

One indicator has given a signal that the recession might be ending right now. The other two indicators have not yet confirmed that.

Jun 05 11:15 AM

[Moon Kil Woong:]

A big concern regarding unemployment showing a recovery is the fact that hours worked keeps falling. This adds to a decrease win wages and points to a general unwillingness of employers to re-hire since they would tend to fully employ their existing surplus labor before spending money hiring and training new labor pools.

I appreciate your articles but I don't agree with green shootists yet. I'll wait until the grass lokls greener. However, I remain in them market for the pure reason that money is moving towards it as an inflation hedge. Now that's a very big shoot. If inflation was a green shoot, I'd be betting it was the fabled beanstalk that Jack will climb into the sky.

Jun 05 12:51 PM

[Donald Ingram:]

If you believe the government numbers as the "truth", then I have some seaside waterfront property in Arizona that I would like to talk to you about purchasing!

Jun 05 01:15 PM

[John Lounsbury:]

A good article today by Dirk van Dijk discusses the employment numbers in great detail: seekingalpha.com/artic...

Jun 05 01:18 PM

[ciel:]

John Lounsbury, Thank you for this balanced analysis, complete with great links. I may have noticed a few small mistakes made in the graphs provided (for example on manufacturing hours - false signals: www.thestreet.com/stor...) you list Sept 1959 as being the "fake minimum" for the 4/60 - 2/61 recession. I presume that either you meant to write in "Sept 1960" - or, I would just add that the "fake minimum" shouldn't even be considered, as it occurred even prior to the start of that recession. Another tiny, perhaps overlooked mistake: In the "Insured Unemployment Peak" Graph: www.thestreet.com/gall..., you list July 1975 as "0" delay from the recessions (March) trough. Small things, but I thought I might ask you to correct or explain them.

But, I nit-pick. The analysis is still excellent!

Now for the bad news: By your own analysis (with which I am in large agreement) we fell back in May on Manufacturing Hours Worked, to what looks like a new cycle minimum, and as such, April could not have been the trough to the recession, all else being equal.

Even more alarming, the decline in total hours worked is about the equivalent of 350,000 extra lost jobs. In other words, had we not seen the time shaved off the average hourly work week, we could have expected, perhaps, about 350k additional net job losses last month, for a grand total back in the 700,000s - Ouch!

This obviously leads one to consider: If Initial Unemployment Claims continue for the next few weeks in the range that they have been for the past several weeks -and- the average workweek does not decline yet again in the current month, might we then expect to see June's nonfarm payrolls return to the -500k and worse club.

All this talk of incipient recovery, but if too many green shoots wilt quickly on the vine, many a Jane & Joe Consumer will begin worrying that talk of recovery is just a bunch of hype.

Jun 06 04:18 AM

[Steven Hansen:]

John, i accept the unemployment 4 week moving average has peaked but the early rise in interest rates will derail our recovery or worse yet send us on another leg down.

and i am rejecting the jobs numbers yesterday. the data plainly does not correlate with other employment / unemployment data we are getting.

Jun 06 10:27 AM

[John Lounsbury:]

ciel - - -

Thanks for reading and analyzing what I wrote. It is the sincerest form of flattery to a writer to have a critical reader go into the depth that you have. Please nit-pick away.

You are correct about the tables. The two errors you pointed out (and a minor one that you missed - tsk, tsk - have been changed and the corrected tables forwarded to TheStreet.com. They may not get the corrections into the posted article until Monday.

Very interesting analysis regarding the manufacturing hours worked. I had noted the breaking news five comments above yours (reply to paintismotion) but did not appreciate the relationship between the loss in manufacturing hours worked and the total loss in employment. I will take this into account in future work.

By the way, I do not reproduce your estimate that the manufacturing hours lost would be equivalent to the loss of 350,000 jobs. Here is my calculation:

0.2 hours lost (April 39.5 hours per week average, May 39.3 hours)
x 11,9986,000 manufacturing jobs
divided by 39.5 hours per week
= 60,689 equivalent full time jobs

If the 0.2 hours had not been lost, the implication is that the total number of unemployed would have been about 60,000 larger. Perhaps you multiplied this number by 4, the approximate number of weeks in a month? That is not correct, however, because when you write out the algebra, including units, the "per week" units are self-canceling and the number of jobs lost is for the entire period covered, in this case one month.

Ciel, you are a good detail person. Please nit-pick this comment and let me know if I have it wrong.

If I mention this relationship between hours worked and total unemployment in future articles, I would like to reference my source for the idea. I propose to say: "as pointed out to me by a reader known to me by the nickname 'ciel' ".

Thanks again for taking the time to dig into all the details.

Jun 06 12:58 PM

[ciel:]

John, Thanks :)

I should probably have stated my interpretation of the drop in hours worked more clearly, as while part of my reply was in regard specifically to the manufacturing hours worked, when it came to the equivalent loss of 350,000 jobs last month, I was referencing the drop in total hours worked in the entire economy, which fell by 0.7% (Ref. MarketWatch: www.marketwatch.com/st...)

As for the estimated equivalent loss of 350k, this was not my own work, but something I had read from Joe LaVorgna, who has stated that every tenth of an hour drop in the workweek is equivalent to a loss of roughly 300,000 to 350,000 jobs (Ref. JBUSA: www.jobbankusa.com/New...) and (Ref. John Jansen post: seekingalpha.com/artic...)

Ultimately, about the only two things in my analysis above that I can personally take credit for are the nit-picking lol

Cheers! ;)

Jun 06 07:46 PM

[Erick Tippett:]

You cite impressive "data" in support of your argument in favor
of this recession showing the possibility of coming to an end
(some analysts are referring to it as a soft depression!).

Question: In all these previous cycles " nine consecutive weeks
below a peak..." was the U. S. economy several trillions of dollars
in debt and no longer a lender nation but a debtor?

There is an interesting website dealing with shadow statistics that presents a different picture than yours. Name of the website is Shadow stats.com, as I recall, check it out you might learn a thing or two.

I can see why Adolph Hitler once commented, " it is such a wonderful thing for governments that people don't think! "

EDT
Chicago, Illinois

Jun 07 12:24 AM

[John Lounsbury:]

Erick Tippett - - -

It is too bad that I can't make every article that I write complete. In an article published one day before this one I discussed what people should be thinking about when trying to review employment data and included a hyperlink to ShadowStats.com. If you have the time to read a thorough research article, it is seekingalpha.com/artic...

You may want to reconsider your concluding comment:

"There is an interesting website dealing with shadow statistics that presents a different picture than yours. Name of the website is Shadow stats.com, as I recall, check it out you might learn a thing or two.

I can see why Adolph Hitler once commented, " it is such a wonderful thing for governments that people don't think! "

I will not yield to the urge to return an insult, but if I certainly am tempted.

Jun 08 01:12 AM

[billddrummer:]

John,

Excellent analysis, and although I don't have the time right now to get into it in detail, another thing that is happening is that the 4-week moving average of initial claims peaked about 3 weeks ago, and has been trending downward since.

A portion of the downturn may be benefit expiration, as noted by Pat C., but the trend itself is more meaningful to me than outlier effects. And even if initial claims remain above 600,000 for the foreseeable future, I believe that the moving average trend will flatten or drop as time goes on.

And when there's a one-week gap downward of 10% or more in initial claims, that will signal the end of the recession (although not the end of high unemployment).

Thanks again.

Jun 19 07:53 PM

[John Lounsbury:]

billddrummer - - -

Thanks for your comment.

I assume you refer to the Continuing Unemployment Claims (called Ins. Unemployement in the DOL reports). After adjustments, that has a peak one week before the latest number (peak May 30). However, the 4-Week Moving Average still has not peaked, although the latest report (June 6) is just barely higher than the previous week. If things continue as they have over the last several weeks, that should peak very soon. A peak in Ins. Unemployment is the second early-to-coincident indicator I look at.

We still have another week to wait for the May Average Mfg Hours Worked, which is the third indicator I watch.

Jun 20 12:18 AM

[Market Sniper:]

John, I appreciate your thoughful work and thinking. It has occurred to me that perhaps we are looking at the unemployment figures through the wrong end of the telescope. I do not know by what methodology this could be approached but how about looking at job creation? Subtract the retirees from the picture (as a new person hired to replace a retired worker does not create a job) and we might get a better picture of what is going on. We just might find that very few NEW jobs are being created. Certainly no match for the jobs being lost or eliminated by several orders of magnitude. The statistics, as reported, are subject to various statistical problems at best and outright manipulation at worst. This year's graduating classes are a case in point. These graduates are NOT part of any statistics as they have not been in the job market. Anacdotal infomation suggests that the Class of '08 is having the worst time finding jobs in living memory. I do believe that the unemployment statistics hides a situation FAR worse than most imagine.

Jun 20 12:53 AM

[John Lounsbury:]

Sniper - - -

I have had several (private) discussions with people who have a similar conclusion, and one I agree with, that there are currently (and for the past several months) been about 2.5 million jobs obtained a month and from 2.9 to 3.1 (approx.) people losing jobs per month.

Your question is how many of the jobs obtained each month are really new jobs, and just not replacement of people who have resigned or been fired. You suggest that we shouldn't count retirees replaced as new jobs, but I think maybe it could be argued that it is a new job in one respect: the retiree is leaving the workforce so the net change is the same as a new job being created as far as the accounting is concerned.

The other question you ask is something that I don't know how to deal with. How many new jobs are being created that didn't exist last month? Maybe I need to look closer at the Establishment survey to see if that is covered. I would think it should be, but what SHOULD BE and what IS are not always the same thing.

Of course, we have the widely debated Birth/Death adjustment, which attempts to correct for the start-up of new businesses and the closing of old ones. This is an attempt to estimate the net new jobs created from businesses opening (new) and closing down.

Jun 20 08:19 AM

[Market Sniper:]

Thanks, John. The only reason that I brought up the question as to retirees/fired individuals being replaced is that it goes to the question of growth. No economic growth is being measured in that situation.

Jun 20 11:37 AM

[billddrummer:]

To John,

Thanks for reading my posts. The birth/death statistic has been one that I've found much fault with during the times we're currently in. In normal economic times, the birth/death statistics work. On the other hand, when some industries have clearly been devastated (construction, specialty retail, finance), the birth/death statistics will continue to show net gains in those industries despite empirical evidence to the contrary.

In my region (northern NV) the estimated unemployment rate for construction workers is 30%--higher if you factor in the impact of undocumented workers that are no longer getting hired on jobsites.

There are precious few construction jobs in progress right now, in contrast with just a few years ago. Subcontractors who were crying for workers in 2006 are closing down. General contractors can hire subs for 50% less than they could a year ago. The normally strong building season hasn't materialized, and the lack of jobs in this sector has, in my opinion, contributed to higher vacancies for apartments and houses throughout the region.

The decimation of the financial services industry is old news. Here, financial services connected to the real estate industry powered building in office complexes, increased rents for Class A space, and contributed to good sales of high end homes, expensive cars, home furnishings and all the rest.

With the collapse of the real estate market, those dollars are gone, perhaps permanently. Now, there are massive vacancies in Class A space throughout the area, luxury vehicle dealers are struggling, and higher end stores are having trouble paying the rent.

This has been a most interesting time for this area, and by extension, the entire country. It seems that no one has been spared by the depth of this recession.

And whether it officially ends this summer or later, I believe that there's a permanent change in the psyche of the average American. One that's fearful, rather than confident; focused on contraction, rather than expansion; and more concerned about making the mortgage payment than planning for a brighter future.

It's truly a shame. And I fervently hope that we as Americans can collectively snap out of it.

But I doubt it.

Jun 22 12:13 AM