The Labor Department reported today that seasonally adjusted new claims for unemployment insurance fell by 34,000 to 601,000 for the most recent available week, resulting in a reduction of the 4-week average for this series for the fourth consecutive week in a row.
We have been highlighting these numbers for the last four weeks ([1],
[2],
[3],
[4]), inspired by a brief statement that appeared in the Wall Street Journal on March 28. The WSJ reported that Robert Gordon, professor of economics at Northwestern University and highly regarded business cycle expert, had noticed that the 4-week average of new unemployment claims tended to peak shortly before the end of historical recessions. I was pleased to see that last week Gordon weighed in in person with his view of these data.
|
The key question of course is whether the recent declines in new claims mean that we’re really past the peak for this business cycle, or whether we could see more bad news in the weeks ahead that bring these numbers up to new highs. In my April 23 and April 30 analyses, I proposed a way to think about this question nonparametrically, that is, without making any statistical assumptions other than that the future will be like the past, simply by counting the number of weeks in previous recessions that were as favorable as we’d seen up to that point but proved to be a false positive signal. Those calculations led me to conclude that the odds were 2 to 1 that the data as of April 23 would later come to be seen as a false positive, but roughly even once we received the more favorable data of April 30.
In his May 1 write-up, Gordon performed a similar calculation to mine, though whereas I used weeks as the basic unit of observation (counting 11 out of 22 false positives), Gordon categorized the data in terms of “episodes” (with 4 out of 8 false positives). In the absence of a clear a priori definition of an “episode”, I prefer my metric, but it is interesting and reassuring that we both arrived at the same basic conclusion. Gordon went on to investigate a number of other features of a downturn, such as the length that the recession had already proceeded and the sharpness of the increases prior to the candidate peak. But it is always tricky to rely on those sorts of refinements when one is trying to draw an inference from the limited number of observations that we have available.
I’d pushed the nonparametric approach more or less as far as I was comfortable with last week, and if we want to say something useful about the way that the latest data have strengthened the signal we really need to use more parametric tools that capture some of the key dynamic properties of the new claims numbers themselves. For this purpose, I built a very simple model to forecast the seasonally adjusted new claims number (yt) in terms of its value the previous week (yt-1), the 4-week average the previous week (xt-1), and whether or not we’re currently in a recession, with st=1 if NBER says week t is part of a recession, and st=0 if expansion:
This relation was estimated by ordinary least squares, with standard errors in parentheses. The model implies a great deal of persistence in the new claims numbers, though given enough time the process would revert to historical mean values. If we were certain that the current recession is going to last forever, the forecast from here would go back up and eventually stabilize at a value of 650,000 new claims each week, 27,000 higher than the most recent average.
Imposing the certainty that the recession is going to last forever, however, hardly seems the correct approach. I’ve instead adopted the perspective that there’s a 1.56% chance of it ending in any given week, based on the observation that in the 384 recession weeks between 1969 and 2009, we saw 6 recoveries (6/384 = 0.0156). If you use that assumption from here out (i.e., that st obeys a time-homogenous Markov chain), you’d be predicting that, in the absence of any other indications to the contrary, there’s a 6% chance that we’ll be out of recession in 4 weeks, a 22% chance we’ll be out in 16 weeks, and a 50% chance we’ll be out in 48 weeks. Those seem to me like pretty reasonable numbers to use.
I then calculated dynamic forecasts of the above model with these assumed transition probabilities for st+j. Aficionados will recognize this as a Markov-switching autoregression with directly observable regimes. The diagram below plots the actual weekly data up through this point in black, the 4-week averages in blue, and forecasts of the future 4-week averages from here out as the dashed blue line. Basically the unconditional likelihood that the recession will eventually end slightly outweighs the tendency of new claims to rise if we stay in a recession, and the forecast declines slowly from its current value.
|
The above graph also plots 70% confidence intervals calculated from this dynamic simulation. The model implies that there is a 15% chance of seeing the new claims numbers go back up above their peak of 4 weeks ago. In other words, there’s an 85% chance that the worst numbers for this particular series are now behind us.
Here is how Gordon summarized the numbers that we had available a week ago:
It is always too early to make definitive conclusions, but the recent 2009 peak in new claims looks sufficiently similar to previous recession peaks to allow a conclusion that it is highly probable that the new claims peak has now occurred. The evidence provided here suggests several differences between the recent peak and previous false peaks in earlier recessions. The recent peak occurred much later in the recession than previous false peaks, and the run-up of new claims in the two months prior to the recent peak was substantially faster than in previous false peaks….
My reasoning leads me to conclude that the ultimate NBER trough of the current business cycle is likely to occur in May or June 2009, substantially earlier than is currently predicted by many professional forecasters.
Today’s data make Gordon’s prediction look even better.
Technorati Tags: employment,
recession,
macroeconomics,
economics
Very cool! To clarify though, in your count of 6 recoveries for 384 recession weeks from 1969 through 2009 (6/384 = 0.0156, to correct a typo in the original post), you are including the most recent weeks of recession in the count. How many weeks are you including for the current recession at this point?
Ho-hum. Let me repeat my predictions, unchanged since December.
1) NBER recession end date in Q309
2) 131M jobs at the bottom.
3) UR Rate crosses 10%
4) Jobless recovery into 2010..
Everyone said it was too optimistic. But it is still more pessimistic than some aspects of JDH’s and Gordon’s thesis here.
You will be surprised how right I am.
Ironman: Looks like you caught two errors, first the typo (now corrected) and second a miscount of weeks duration for the most recent recession. I was using November as a starting point but should have used December for the NBER dates. Correct count should have been 379 rather than 384 total, of which 74 are from 2007-08. I don’t think changing this would alter any of the calculations much.
Ironman: Sorry, I spoke too soon on the second point. Here’s how I’m counting: 1970(52), 1974(74), 1980(31), 1982(74), 1991(39), 2001(40), 2008(74), for a total of 384, as correctly reported the first time in the text.
1980-82 was one big 3-year recession. That most minimal of recoveries in between should not be counted as such. Many people went 3 years without working at the time.
On the other hand, can we dig out if oil starts spiking again and long Treasury rates continue to go up?
I see the corners of his mouth twitching a bit.
Simple and true. I still think the depth of this recession is largely caused by speed of which information about economic conditions is accessible to people, as unfiltered noise.
That makes every subject able to reach internet a decision maker in this cycle, similar to sound amplifier with no proper integrating high frequency filters at place- no wonder the system goes into generating mode ( recession generating) in the same way it moves into bubble generation mode during upswings.
The conclusion is, this system with well informed unfiltered elements (us) getting too much noise into their circuits is unstable in general with an unknown oscillation period ( is it 7 years or less between crisis /bubbles now? It may be getting shorter, and amplitudes correspondingly bigger) . So the next bubble may come sooner (ie. 2012) and be even bigger, and the next recession (i.e. 2014) even sharper if integrating filters are not put in right places.
Is that good or bad I have no idea.
But Murdoch’s idea that content should priced is one market based idea to reduce noise in the information consumption.
You are betting that the current recession is no different than the previous 6. Does this recession look the same as the 6 included in the analysis? Not to me. For one, none of the 6 was associated with such a disaster in the banking system.
Although you are operating non-parametrically, it seems that you are still essentially looking at past data (i.e. jobless claims) and extrapolating forward, as opposed to looking at measures which are more or less intrinsically forward looking like, say, implied volatility.
I suppose you could argue though (especially based on the reasoning in your piece about whether recessions matter) that jobless claims *are* forward looking in the sense that firms act very quickly to shed employees in the face of bearish forward-looking projects, so that the jobless claims then reflect that information.
The point of all this is the extent to which other conditioning information is going to affect how things unfold. For example, the daily (hourly) stop at CalculatedRisk will undoubtedly contain information on bad news in the pipe such as impending CRE delinquencies or jumbo mortgage defaults which will prolong the bad times.
“In other words, there’s an 85% chance that the worst numbers for this particular series are now behind us.”
No, I think you mean that 85% of the time in the past the worst numbers would have been behind us at this point.
There is a vast difference and if you don’t understand that you shouldn’t be doing statstics.
No doubt the news is positive… but since I am bearish, I had to dig to find at least one negative… the average number of weeks unemployed increased again and is the longest going back to 1948.
Obama has surprised me. His rhetoric on unemployment insurance was only that, rhetoric. His administration has not pushed unemployment insurance nor has he pushed his government jobs programs. Because of this I will have to revise my forecast of double digit unemployment by the third quarter. Maybe he is as stupid as FDR after all.
Of course he is still accumulating power in the government so recovery will be slow and painful if at all. There are still a lot of traps out there that could open up. If he does push through universal socialized medicine this summer we could see a quick return to decline. He is also going to have to face paying for his massive spending sometime. How he does that could dictate whether it will have a little negative effect over a long time or a big negative effect over a long time.
But contrary to all the econometricians it is Obama and the Democrat Congress that have the economy by the….
biofuel hits me where it hurts: the induction based on similar previous events that may not be similar enough.
Dang that Principle of Induction.
Accentuate the positive: the stimulus is working or starting to work. Can we tie this reduction in the job loss rate to these measures? Expectations of these measures? What “shovel ready” jobs are starting to work early? are quitting less quickly, now that there is a free lunch? and a beer? with live entertainment?
I see March was revised up to nearly 700k, but accentuate the positive: those were crummy jobs that did not generate any covalent jobs. No, they were swine flu jobs that contaminated other perfectly self-respecting jobs. So much depends on jobs being as independent and isolated as eggs in the carton, yes?
Ok, so what are all our friends in the RE industry doing now?
This is some serious work done in real time. If things go as predicted by this, expanding the model would be greatly appreciated
Your kidding me, right? Anyone can dump trillions in the bank and industrial sectors and cause a slowing of the recession, but the damage it is doing will be us for years! We have violated many economic priciples with this to big to fail garbage. We are setting up a very nasty situation that could end not only in a depression, but in a major world war! Hope I am wrong, but I do not think so!
Isn’t this just like the 2001 recession, where governmental intervention/stimulus forced the recession to end artificially soon, only to cause larger problems later on?
Isn’t that what is happening here?
Should we not hope that it goes a couple of quarters longer to wring out certain problems for good?
I love these generic comments: this doesn’t “look” like the previous recessions, so this one obviously will be completely different… obama is as stupid as fdr(???)… My only question is, are we here to discuss political differences, or to crunch economic numbers? Anti-Obama folks seem to want the economy to continue to fail only to be able to say “i told you so” to those of us that voted for him. If anyone has any good economic statistics to mention to rebut what JDH has to say, please feel free. But keep the rhetoric and the regurgitated beck and limbaugh vomit aside. BTW, this will be my only off-topic post. Thanks!
Sorry, Ryan, but I am in line with a number of the above negative comments. No political motive – just that the current malaise is driven by over-indebted consumers, whereas the others were not. Where is the driver for recovery? Business investment, when consumers are still over-burdened? (At some point, business investment would need to increase merely to maintain existing capacity, but my sense is that excess capacity is still growing.) International trade? (Could be – we still have a deficit of 3% of GDP – but the downturn is global, so my guess is we are more likely to pull them down than to have them pull us up.) Government borrowing is helping to hold up aggregate demand, but at the expense of future higher taxes. And much of the debt has been incurred in the mistaken attempt to maintain asset values (bailouts) instead of jobs programs.
No, I don’t see the pattern of past recessions providing much guidance for the present one.
don’s comment asks us to consider more inputs to predictions of NBER turning points than just the jobless rate. I assume the Robert Gordon’s of the world do move on from identifying an independent variable highly correlated with recession timing in the past to incorporating that variable with other stuff. True, Professor Hamilton? Or am I projecting?
I would think that, since we know a bunch of other things about the actual inputs to the NBER process, a more elaborate model would be useful if for no other reason than to show us the limits of any one indicator as a predictive tool. Say, for instance, that new jobless claims are falling because the pace of layoffs was so much more rapid in this period than in the past that layoffs can slow and still run at a very rapid pace.
Don’t auto workers who are laid off get to claim benefits? If so, then won’t we see a fresh wave of new claims very soon? What does it mean if new claims peak, then rise again? That happened in late December and early January, though only briefly. It happened last January and July, as well.
kharris: Sure, Gordon and all the rest of us look at all kinds of data. But sometimes it’s helpful to think about the signals one at a time, and ask how strong is the evidence from one variable alone? You’ll notice that in the analysis above I don’t offer any mathematical model to predict the end of the recession per se. I only ask whether a reduction in new claims of the size we’ve seen would be judged to be significant, or is it just part of the standard jagged behavior we expect in a recession?
JDH – Thanks for following up on your count of the number of weeks. I had assumed you had included all the weeks of December 2007, but its nice to have the confirmation. With it being the NBER’s peak month ending the previous period of expansion (and the peak month for non-farm payroll employment), I can see where there might be a bit of play in deciding whether to include the weeks for December 2007 in the total count or not.
From my standpoint, it makes sense to include the whole month, but I’m looking at it as being the first month after total employment peaked in November 2007.
Don’t forget about revision. March intial claims were revised to 701,000. That even caught me for a surprise.
Things fell harder and faster than we thought, then the contraction began decelerating in February. That is where we are now…….faster then we thought.
This is a bogus blog. Does Geithner fund your research?
95% of other economic blogs disagree with your conclusions. Hey, and what about revisions and the birth/death nonsense??
How are those fat tenure checks? It must be fun to get money for nothing.
My sincere gratitude to Tiny Tim, kharris, don, Heterosexual, Steve, calmo, wally, Robert Bell, biofuel, and GK, for providing perception and clear logic… I particularly wish to thank Steve for his cautions about World War as a result of America’s collosal trade deficit and government debt… My view is that America has waged almost continuous war for the last 65 years as a means of securing trade advantage and a passivated population. I expect that we will continue with the same behavior until we dissipate ourselves like the Spanish and British before us.
Count me in as a skeptic. I find it distastful that pundits, government officials, and professional economists mouth the happy talk instead of lobbying for the institutional changes required to correct the problems. Like the Romans, we’ll collapse under the overleverage and inflation of an overcirculated currency.
My personal guess is that they stay higher, around 600.000 for quite a while
But Eco, 95% of other commenters here disagree with your conclusions that this is a bogus blog…including your action that minimally thinks we are savable…if only we had your non-bogus insight…coming soon.
I’m with MarkS, but only recently…and not with any of the others…esp kharris who pencils in an extra apostrophe (“I assume the Robert Gordon’s…”) showing the flimsiness of any induction whatsoeva.
Ok, and with JDH…because, along with all the Robert Gordon’s of the world, we will soon know whether their predictions are correct –in which case we doubters can eat our hat’s …or whether they get to eat theirs.
The sophistication present in the statistical analysis contrasts mightily with the…guesses of the less sophisticated…tis the nature of bein well-trained –you dumb assess know what I’m sayin?
Sadly, there is no such distinction between the payroll job sending yet another rocket to the moon and the payroll job, maintenance engineer…making it true that cleanliness is next to godliness.
I’ll believe any green shoots when that initial claims number fals to the levels around the 1991 peak. Till then, it’s merely a stopping of the free-fall. It’s still 120,000 people filing claims every work day, and it’s still 537,000 fewer people working than in the previous month.
Until these employment numbers improve, the GDP contraction may be less, but GDP will still at levels well below what we were at in Q3 2007. And until that part changes, there’s no reason for happy talk.
It’s like how all the Wall Street douchebags have come out hiding in these last 2 months, bragging about a 30% rally in the S&P. Sounds great, until you remember that it’s still down 30% in the last 12 months, and compounded by the fact that mst real people haven’t made money in the last 10 years.
Until the average worker is still working, and making wage gains, this economy will sputter in the real world. And the hit that the last year has given to revenues at the state and local levels of government is devastating.
I’m hard pressed to see HOW we can hit the economic trough this quarter (or for several more) if, as your graphic suggests, we keep laying off a half-million people a week–even counting in the returnees, etc.
We will hit at least a temporary bottom in 4Q as employers gear up for (a) the holiday season and (b) because the stimulus package actually begins to take effect. Some of that will be lost in early 2010, however.
More relevant, however, are the possible (maybe likely, but I’ll stick with possible) layoffs from not only Chrysler, but GM, going bankrupt (including effects on both suppliers and dealers), and a re-deterioration in the financial sector (1Q was an accounting fraud for purposes of the stress test in my book).
In short, this recession is different than the one’s you and Gordon have addressed (and much more like what GK and Biofuel above suggest).
The fundamental problem I see in this analysis:
There is no definition for “recession.” See the NBER site. A recession is designated in retrospect by a group of economists standing at the taffrail, holding up their collective thumbs, and squinting at the wake. The composition of that group changes over the years and decades, as do the economic viewpoints of the individuals.
So the recession boundaries specified by the NBER are somewhat or largely arbitrary. Meaning that one of the arguments in this model, and the item it’s attempting to predict, are also somewhat or largely arbitrary.
Hence the comments above that “this one’s different.” Yeah…
The use of the metaphor “a green shoot is growing” is misleading. It implies the emergence of new private investment and the beginning of growth after recession. But what you’re describing is a slowing in the pace of contraction. A metaphor such as “the cold snap is passing” would be more appropriate. And my feeling is we’re in for more.
SteveR, this is not an arbitrary comment designed to get your attention nor one to sully your otherwise pristine reputation (for bein last to comment)…no, it is to point out to you that your criticism of NBER’s “somewhat or largely arbitrary” specification of where to draw the line on what is and what is not a “recession”…suffers the worse fate of bein plain Jane arbitrary itself. So many other criticisms to make of ‘the recession’ and you pile on…with this…academic note…about referential opacity…makin it worse, I tell you.
How many thumbs up did you give it? Was the 2 consecutive quarters of declining GDP (perhaps not sufficient and necessary conditions) an inadequate specification to refer to this “downturn” even though readers of the literature know what is being referred to by “recession”?
So for you, there can be no definitive set of recessions…such is the deplorable state of that decision procedure defining that set.
And no trucking with that family resemblance stuff…you are just going to take your marbles and go home.
Yeah…
This week, it could be as low as 580K. This would greatly improve the 4-week average.
At the same time, oil is creeping up (currently at $59), and could greatly dampen the nascent recovery. If oil crosses $70, we could have another leg of hardship to go through.
But the trough is still in Q309, no matter what.
Sounds like the same figuring that eliminated the risk of lending to subprime borrowers.
And with today’s report those green shoots are “wilting”…
There will be 2 months of ups and downs now around 8500 as people doubt the reality of recovery. In about mid July, however, stocks will take off.
You can see that from time series of stocks as they enter and exit crisis. That is visible, may be not computationally, but once You perceive the only facts about stock prices available- the historic aggregate time series price behaviour.
All the attempts to put more than one other variable behind these series are almost sure doomed. Jobless claim peeking is the only one that truly correlates with few month accuracy.
Imagine any process that correlates with stock prices and recessions over 6 points in time- what would be the probability it is NOT really correlated in time?
I do not know mathematics, but I guess such probability is so low that predictions over the 7th point almost certainly can be made based on previous 6 points.
All other variables, reports etc that does not show such historic correlation, generally, does not matter.