The Labor Department reported today that initial claims for unemployment insurance fell by 14,000 during the most recent available week. That brings the 4-week average down for the third consecutive week and puts it 3.3% below the peak reached April 9.
|
That ongoing drop in the 4-week average is noteworthy because in each of the last 5 recessions, once the new claims number began declining from its peak value reached during the recession, the NBER subsequently dated the recovery from that recession as beginning within 8 weeks.
|
Reasoning as in my last discussion of these data, one can try to judge how meaningful the latest numbers might be as follows. If we leave out the 1970 recession, there are 230 weeks in which the NBER declared the economy to have been in recession during the 5 recessions of 1974, 1980, 1982, 1990, and 2001. In 22 of these weeks, we saw as big a drop as we’ve seen this month, namely, the 4-week average dropped by more than 3.3% over a 3-week period. Of these 22 favorable readings, 11 turned out to be part of the final move out of recession, while in the other 11, new claims turned back up to reach a subsequent higher peak. Thus, if all you had to go on was the data on new unemployment claims and its behavior in previous recessions, you might conclude that there’s a 50% chance that an economic recovery will have started by the beginning of June.
Technorati Tags: employment,
recession,
macroeconomics,
economics
Crysler bk will put a end to these ideas.
FWIW, they have already been furloughed and CAN get intial claims, but most don’t in that situation.So intial claims are higher than what the bottom line says they are for april.
I suspect another rocket shot of intial claims is coming.
If GM goes bk as well, another shot. These are death blows to any economy and will drag any recovery out.
Over all periods, the drop in claims from the peak is at only about 1 standard deviation…
Take a look at the drop in the total number of workers from the household survey… We have never experienced such a sharp drop. During the recession of 75, yes the unemployment rate was higher, but more women were entering the work force (baby boomer children turning working age), so the total employed was relatively stable… not so today.
I think such comparisons with other post-war recessions should be taken with a grain of salt. This is a different animal. I expect we will have a series of ‘recoveries’ over a number of years before this malaise it truly over.
Professor, like the analysis. However, I think there is a light flaw in your logic of comparison. You are comparing a 3.3% drop with previous drops that are greater than or equal to 3.3%. i.e. you are comparing the current drop with larger previous drops. To compare apples with apples you need to compare with drops of a similar size. e.g. those that average 3.3% or those say in a range 2.3% to 4.3%. I suspect that reduces the probability that points to an end in the recession in June.
I agree with don; given that this is already the longest recession on record since WWII, and that its genesis is overlevered household balance sheets (and not overstocked business inventories), any comparisons should be with a long, pre-WWII recession involving overlevered household balance sheets, i.e., the ’29-’33 depression. I know that much of that data is not readily or easily available, but that really is the appropriate benchmark.
Side note — Professor, we need another opinion piece from you on the latest Fed actions, e.g., Bernanke putting pressure on Lewis to close the Merrill deal, his impending decision to extend five year loans to purchasers of CMBSs, and what appears to be the lack of ‘stress’ and transparency on the stress tests.
Thanks!
“you might conclude that there’s a 50% chance that an economic recovery will have started by the beginning of June. ”
And if there were only 2 instances would you still conclude that there is a 50% chance? This is an empirically meaningless statement.
Steve van Emmerik: Please see the mathematical derivation here. The question I pose is, “how likely is it that the peak is passed given that we have observed a drop this large or larger?” The question is correctly answered by the methodology.
glimmers and green shoots: The standard deviation for an i.i.d. Bernoulli average from a sample of size 22 is 0.11, though I grant these aren’t independent. The standard deviation from a sample of size 2 is 0.35. The Bernoulli average from a sample of size 2 would not be informative. The average from a sample of size 22 is.
Aren’t there significant differences between the current recession and all previous ones when examined from the household balance sheet?
As an example in 1970, the typical household income was derived from one working male supporting 2 children and a wife; currently it is two full time m/f supporting 2.2 children. Since home prices, insurance costs, child care and medical expenses have doubled and now eat up 52 checks as opposed to 26 checks (assuming a check bi weekly pay period) can’t we argue that unemployment is less relevant in 1970 (or other previous recessions because of goods prices pre-globalization etc.) than today? Throw in the debt burden and I think even modest time frames with high levels of unemployment will have a greater rippling effect through the economy. So while you may have periods of reduced unemployment claims the reverberations created from defaults for those who cannot find jobs will force contraction since capital levels get depleted at greater rates than credit/capital is being created.
Also there was a substantial difference in savings and unemployment insurance benefits, which will magnify the effects of unemployment. In New York city (where I live) there has been a 5% increase in benefits in 18 years, rental prices have doubled in every area and in some cases (West village has quintupled).
I guess what I’m saying is that comparisions to previous recessions aren’t contextually similar since the household balance sheet reads differently from past recessions.
Thanks for your time, and I really enjoy reading the blog.
Thank you professor. I have heard many comments on this drop in unemployment claims that say it is evidence that we are coming out of the recession. As you have pointed out if we only look at the numbers statistically there is only a 50-50 chance. At best this only demonstrates that we are holding even at our much depressed economic level.
Economy is a behavioral collective phenomena. That takes care of many things- people read Internet and act along others in bursts of one type activities.
As is well known, positive news lead to bursts of positive action, negative to bursts of negative action. This alone should average out any economical argumentation based “special” status of this crisis since this has not changed, may be amplified by the most effective dissemination of information since ANY other crisis.
My guess would be that when the power law inter event time distribution spectrum between successive positive actions (e.g. like buying stock) gets more concentrated at short times than corresponding spectrum between negative (like selling stock) averaged over some period of time ( few months) crisis is largely over . It does not matter how much stock is bought or sold, what matters is how many times and with what distribution such irrational decision bursts are executed.
An interesting example of such burst of similar actions and their effect when they are overlaid on cyclical externally driven activities of humans -circadian daily and weekly cycles is given here:
“A Poisson explanation for heavy tails in e-mail communication” by R. Dean Malmgrena, Daniel B. Stouffera, Adilson E. Motterb,c, and Luís A. N. Amarala
http://amaral.northwestern.edu/Publications/Papers/Malmgren- 2008-Proc.Natl.Acad.Sci.U.S.A.-105-18153.pdf
Fine thing about this study is that there is no need to have rational decision making to produce the observed patterns, its more like going with the flow. In essence, when time to do something is limited by natural factors, people act in bursts.
I have no idea if such data about stock markets has already be analyzed from this perspective.
Authors say: human behavior is primarily driven by
external factors such as circadian and weekly cycles, which introduces a set of distinct characteristic time scales, thereby giving rise to heavy tails.
Please forgive any grammatical errors and other mistakes, it was about 1230 EST when I posted.
The benefits I refer to in the third part of my post is unemployment insurance. The conclusion of the last sentence there should read: …and in some cases risen by considerably more.
Best!
Jim-
Still no worse than the 81-82 recession, the “recession that dare not speak its name”. Wonder why?
In other news, the April ISM Index came in at 40.1.
ISM says that:
“A PMI in excess of 41.2 percent, over a period of time, generally indicates an expansion of the overall economy. Therefore, the PMI indicates contraction in both the overall economy and the manufacturing sector. Ore stated, “The past relationship between the PMI and the overall economy indicates that the average PMI for January through April (37 percent) corresponds to a 1.3 percent decrease in real gross domestic product (GDP). In addition, if the PMI for April (40.1 percent) is annualized, it corresponds to a 0.3 percent decrease in real GDP annually.””
Rich,
It’s hard to say if the current recession is better or worse from 81-82 based on initial claims. Two reasons: Back then, more people as a % of the workforce were eligible for unemployment. However, the population back then was smaller.
So you’d have to adjust the data for those two factors.
Steve-
This very point was discused in a prior post on April 23. You may want to look at that. Another poster asserted the same thing that you do and provided what he thought was evidence. I was unpersuaded – you be the judge.
My main point is that 81-82 was a bad recession, worse so far than the current one. Nevertheless, there is very little mention of it. I believe that the “solution” to that recession was tax cuts, not phony stimulus. Bringing up that recession could start a discussion that would be unhelpful to the current administration.
Steve,
Do you have any data to support your statement that more people were eligible for unemployment insurance in the 82 recession?
I think the Prof. is right, except that the trough of the recession will be in Q3 (as I have been saying since December).
Large shocks, such as a Chrysler bankruptcy, possible GM bankruptcy, and swine flu scare, have all failed to make the economy derail further. Such shocks are now absorbed.
Thus, the trough is near.
The ISM indices have to be above 50 for there to be a real expansion. Readings in the high 40s are merely a transition to a mild recession.
Steve and Anonymous,
Looking at initial claims data, the 82 recession looks worse than today’s. But today’s looks worse when you look at the rate of increase in the household survey derived unemployment rate. Just since the auto problems blew up in the fall, the rate of increase in the unemployment rate has increased.
I looked at the links in the April 23 post comments. It is too bad that the U-6 data does not go back to the 80’s. However, we can compare this recession to the 2001 recession using U-6 data.
Between 3/2000 and 6/2003, the unemployment rate increased 2.4 pps or 66%. U-6 increased by 3.4 pps or 50% to 10.3%.
Since 3/2007, the unemployment rate has increased by 4.1% or 93% to 8.5%, meanwhile, U-6 has gone from 8% to 15.6%, an increase of 7.6 pps or 95%!
i think, unfortunately, many of these posts are too cute and too optimistic by half in analyzing the initial claims turning point. The level of initial claims will be more important here, because they are so high. For example if the claims continue to descend, break 600K and go to say 500K over a years time, this will still be consistant with about a further -8% real gdp decline, totalling about half a deep as the great depression.
Some turning point! Initial claims need to get below 500K or better 400K before payroll ekes its way back to a positive
It came in this week at only 491K!
JDH will be having great fun right about now.
Sorry, I take that back, that was ADP, not initial claims.
Aha! 601,000, it turns out.
with today’s drop, the ma of initial claims for the week of May 2 was 624K, which was 5.3% below its peak of 659 for April 9. Would like to know what is the probability now — too bad Jim lives in the west coast.