Disposing of the garbage from the King Report and Art Cashin

- by New Deal democrat

I dislike cherry-picking data. It almost always means that the writer started with a conclusion and reasoned backward to find support.

There is some particularly egregious cherry-picked garbage that started with the King Report; namely, that December 2011 through February 2012 non-seasonally adjusted data showed that we have lost 1.8 million jobs. Therefore, presumably, the jobs picture is actually quite bleak. This then got repeated by Art Cashin on CNBC.

Our former co-blogger Invictus eviscerated this claim over at Barry Ritholtz' Big Picture. For the record, beginning with December 2000, here is the December through February change in the non-seasonally adjusted payrolls survey:

2001 -2212
2002 -2402
2003 -2273
2004 -2050
2005 -1880
2006 -1728
2007 -2105
2008 -2519
2009 -3939
2010 -2432
2011 -2037
2012 -1801

Note that once we strip out the seasonal adjustments, the last two months of job data have been the best bar once in the last 12 years.

Apparently undeterred, when called on it, instead of just admitting he goofed (nobody's perfect after all), Cashin doubled down, claiming that he was simply following Lakshman Achuthan's lead in not accepting seasonal adjustments since the recession and looking at YoY data.

In the first place, Achuthan's rejoinder to my critism of ECRI's recent recession call defense - that recent seasonal adjustments following the 2008 recession may be overstating wintertime growth - besides being cherry-picking itself, since ECRI apparently has not historically primarily relied on this mode of measurement, overstates the conclusion that should be drawn. If we're not sure we can trust seasonal adjustments, and they conflict with YoY data, that does NOT mean we should accept the YoY data as the more reliable picture of the present condition. It means that WE DON'T KNOW which way of looking at the data is more accurate in depicting the present condition.

But let's accept Achuthan's criticism, and accept YoY nonfarm payrolls each February since 2000 as presenting the truer picture. Here's the result, again derived directly from the UN-seasonally adjusted data:

2001 1670
2002 -2028
2003 -409
2004 316
2005 2219
2006 2692
2007 1754
2008 715
2009 -5042
2010 -3568
2011 1402
2012 2016

On a YoY basis, the last 12 months have seen the 3rd best job growth in the last 12 years.

There is not a shred of credibility in the claim made by the King Report, the original Cashin claim, or the revised Cashin claim. Rather than repeating the claims at other business sites, it's time to throw out the garbage.

Read More

Expect *another* poor nonfarm payrolls report tomorrow

- by New Deal democrat

Last month, when the consensus was for a payrolls gain of about +150,000, I said to expect a report more like +50,000. I based this on the difference between how payrolls behave in the face of rising initial jobless claims vs. falling claims. For your reference, here is the scatter graph I ran last month comparing initial jobless claims vs. nonfarm payrolls, which includes the deterioration in jobs leading up to and into the recession in red vs. the improvement during the recovery in blue:



It simply makes a world of difference whether you are deteriorating into recession, or in recovery coming out of recession, leading to very different scatterplots as above (the same pattern is true for every other postwar recession).

As I have previously noted, we only have one true example of a double-dip, and that is the failed 1980 recovery turning into the 1981-82 recession. Here is the scatter graph for that double dip, showing the abortive recovery (blue) and the double-dip (red):



Given the continuing initial claims in the 410,000 - 430,000 range, the poorer (red) trend on the scatterplot graphs is the one we should be referencing. More accurately, while we don't know what a new "red line" might be developing, it is almost certain to be to the left of the recent recovery "blue" line. Even with the "summer stall" last year, where GDP only fell to +1.7%, nonfarm payrolls fell into and remained in 5-digit territory.

Predicting the nonfarm payrolls number is admittedly a mug's game - there is variability of something like +/-125,000 around any estimate. This month the consensus is still for +110,000 private payrolls, and +80,000 total (meaning 30,000 lost government jobs). Unlike last month, both of these estimates look to be "in the ballpark." At the same time, a number similar to last month's looks more likely to me, and a larger loss of government jobs (which has happened in both June 2009 and June 2010). Let me put it this way: the range of reason includes a negative overall number for the first time in over a year.

Read More

Expect a poor nonfarm payrolls report tomorrow

- by New Deal democrat

As I type this, the consensus is still for a payrolls gain of about +150,000 tomorrow. Based on the poor performance of initial jobless claims in the last month, I am expecting a considerably poorer report, on the order of only +50,000. I base this on the difference between how payrolls behave in the face of rising initial jobless claims vs. falling claims.

Below is an update of my scatter graph comparing initial jobless claims vs. nonfarm payrolls that I first ran several weeks ago. I included the deterioration in jobs leading up to and into the recession in red. Note how the red line traces a very different path than that of the recovery:



This is why I dismissed the straight line Prof. Delong drew at 400,000 jobs as the dividing point between job growth vs. losses. It simply makes a world of difference whether you are deteriorating into recession, or in recovery coming out of recession, leading to very different scatterplots as above (the same pattern is true for every other postwar recession).

But how should we expect jobs to behave in the face of seriously climbing initial claims? We only have one true example, of a double-dip, and that is the failed 1980 recovery turning into the 1981-82 recession. Here is the scatter graph for that double dip, showing the abortive recovery (blue) and the double-dip (red):



Even with the "summer stall" last year, where GDP only fell to +1.7%, nonfarm payrolls fell into and remained in 5-digit territory. Hence my expectation for tomorrow is that nonfarm payrolls retreat into 5 digits.

If that happens, watch Oil and see if it falls to a new post-April low. Only the real fear of a double-dip will cause the chokehold to loosen.

Read More

Initial Jobless Claims vs. Payrolls -- where to next?

- by New Deal democrat

Fourteen months ago, I wrote that

Some time ago, Prof. Brad DeLong of Berkeley, thinking aloud with graph, drew a line across the 1991 and 2001 recessions and recoveries, making a "note to self" that it appeared that Initial Jobless Claims post those recessions had to decline to 400,000 or less before payroll jobs were added. Thus, mused Prof. DeLong, it must be so as well, post this "Great Recession." This "note to self" was subsequently repeated by Bill McBride at Calculated Risk, from which it has now been picked up and repeated at Prof. James Hamilton's site, Econbrowser....

IT IS WRONG.
....
I am utterly confident that there will be job growth long before jobless claims fall to 400,000.

Well, we know how that turned out, enabling me to do a little victory dance in March.

With initial claims finally falling below 440,000 in three of the last four weeks, on Friday I posted this graph, showing initial jobless claims in blue, left scale, and private nonfarm payrolls, inverted, in red, right scale :



Over the weekend, I decided to take a little more sophisticated look at what might be in store if jobless claims are truly breaking out to the downside, and continue to decline to 400,000 or even less, which seems increasingly likely as real retail sales in the last four months have grown at a rate of 10% annualized.

Exactly as I did in my analysis last year, I am comparing this jobs recovery against the job recoveries from the two least severe recessions (1974 and 1982) and using that as the template.

Here is a scatter graph comparing the decline in initial jobless claims from peak to trough (y axis) following the 1974 recession with nonfarm payrolls (x axis):



Here is the same scatter graph for the jobs recovery from the peak of the 1982 recession (note I have deleted two months in 1983 where a strike caused a loss of 1 million jobs in one month and a similar gain the next when the strike ended):



Here is the same graph for our jobs recovery so far (note I am using private jobs only to eliminate distortions due to the census):



Putting the three jobs recoveries together, here is the scatter graph we get:



Focusing on that part of the graph that only shows the correlation of payrolls for jobless claims under 450,000 gives us this:



While I don't have the software to do this scientifically, the trend line averaging the data from the three jobs recoveries is probably very close to the curved trend line I have drawn in light green:



This suggests that at a monthly average of 440,000 initial jobless claims a week, the average monthly payrolls gain is about 175,000 to 200,000, and gradually increases as initial jobless claims decrease. But there is an awful lot of variance, so that - as the graph immediately above the last one shows - at until such time as jobless claims drop below 360,000, even excepting a few outliers, any given month may give a number as low as ~75,000 or as high as ~350,000.

Read More