Wednesday, August 1, 2012


- by New Deal democrat

With yesterday's release of June real income, we now have full data through midyear 2012.  Unless there are downward revisions to the critical series upon which the NBER relies, it can confidently be stated that no recession began by that time.

It is widely acknowledged that the NBER looks primarily at 4 data series in dating economic peaks and troughs:  industrial production, nonfarm payrolls, real personal income minus transfer payments, and real retail sales (some would prefer real final manufacturing and trade sales in lieu of real retail sales, but in this case it does not make any difference).

Let's look at the performance of each of these four items.  First, here is industrial production (blue) and real retail sales (red);

We can see that while real retail sales have declined in the last 3 months, similar to their decline in 2010, industrial production has continued to make new highs.

Next, here's nonfarm payrolls (blue) and real income minus transfer payments (red):

Both nonfarm payrolls and real income minus transfer payments also continued to make new highs through midyear.

Subject to revision, with only one series in decline, and the other 3 rising, as of midyear there has been no broad decline in production, income, jobs, and spending that is the hallmark of a recession.

While I continue to have respect for Prof. Geoffrey Moore's concept of long and short leading indicators with which he founded ECRI, and I appreciate that owners of proprietary models have a vital interest in not giving away their secrets, I have to say I have been dismayed by the lack of candor ECRI has displayed in the last 10 months when economic data they cited in support of their recession call turned against them.

To begin with, their original call over 10 months ago on September 30, 2011 was that a recession was "imminent," saying "It's either just begun, or it's right in front of us."  Subsequently they revised their call, saying the recession would hit "by mid year" 2012.  In support of their original call, in this interview with Tom Keene of Bloomberg television last September 30 (link is to video), Lakshman Achuthan specifically cited initial jobless claims, first at the 3:50 mark:
Tom Keene: 400,000 claims isn't happening, is it?
Lakshman Achuthan: No, forget about it! This is -- unemployment is going back up, claims are going to be rising.
Then again at the 5:15 mark:
TK: [pointing to a graph of initial jobless claims] My hand is at 400,000 --
LA: We're going over--
TK:-- and we just. can't. get. through. it.
LA: It's going back up.
Not only did initial jobless claims not "go back up" without crossing the 400,000 threshold, they have declined to as low as 350,000 without a single week breaching the 400,000 threshold to the upside this entire year.

ECRI has never acknowledged this error.

Last September they also cited Gross Domestic Income. There is some evidence that GDP tends to be revised in the direction of GDI. In another interview on Bloomberg, on December 8, 2011:
Achuthan also noted that “the other half of the GDP report,” gross domestic income or GDI (which tends to be the more accurate measure of GDP) was up just 0.3% in the most recent quarter [NDD note: Q2 2011]. The Federal Reserve has observed that when GDP and GDI differ, the GDP figure tends to be revised toward GDI, not the other way around. Achuthan warned that the GDI figures are “a big red recession signal.”
Except that GDI did not continue to decline. In fact as of revisions published just last Friday, it appears that 3Q 2011 GDI was the trough. While I don't have a graph, here's the quarterly statistics for GDI since the 1st quarter of 2011:

2011 Q1 +2.6
Q2 +0.4
Q3 -0.2
Q4 +4.5
2012 Q1 +3.5

Similarly, revisions to GDP since 2009 released last Friday by the BEA also show that YoY GDP troughed in the 3rd quarter of 2011 and has improved since then, with YoY GDP as of midyear 2012 being better than any point in 2011:

In what appeared to be a direct response to a criticism of mine, in March ECRI claimed that the proper way of looking at the 4 coincident series that mark economic peaks and troughs was their YoY performance, saying that seasonal adjustments had been affected by the deep declines during the Great Recession of 2008-09.  In so doing ECRI even abandoned the 6 month growth metrics for their Weekly Leading Index and Coincident Index that previously they had consistently touted.

As set forth above, both YoY GDP and GDI have improved since ECRI made its recession call last September.  Not once have they acknowledged that improvement.

But let's also look at the YoY performance of the 4 coincident series marking economic peaks and troughs.

First, here's industrial production:

Industrial production has been trending higher on a YoY basis since last autumn.  On a 3 month rolling average basis, it is having its best YoY growth since the beginning of 2011.

And on a longer basis, shown below, with the exception of 2010, industrial production is improving at the best YoY rate in over a decade:

Next let's turn to nonfarm payrolls.  In his recent appearances, Lakshman Achuthan has pointed out that their YoY growth rate has declined, which indeed is true, although it appears to have stabilized in the last couple of months:

But this modest 0.2% YoY decline, which stabilized in the last two months, is nowhere near the typical YoY decline in payroll growth that has preceded most recessions:

Notice that a decline of over 1% has been typical.  Even in the case of 1970 and 1974, there was a YoY decline of 0.5% in payroll growth before the recession began.

Next, let's look at real income.  On a YoY basis, the growth rate in this series bottomed at the beginning of this year, and has been rising since.  Note that the abrupt decline YoY after December 2011 was because that was the anniversary of the 1% FICA tax cut that took effect in January 2011.  I have added a second line in red to indicate what the YoY change would have been without that 1% tax cut, and you can see that the organic growth rate of real income now is actually the highest in over a year:

Finally, let's look a real retail sales.  Here there has been a very significant decline in the rate of YoY growth -- but again not nearly so much as in the case of the last 2 recessions.  In fact this level of YoY real retail sales growth is consistent with fully half of the last expansion:

In summary, not only are 3 of the 4 critical recession marker series going up, but 2 of the 4 are growing at an accelerating rate (industrial production and real income), and the growth rate of 1 of the remaining 2 has stabilized as of midyear (payrolls).  When the YoY gain in payrolls turned down several months ago, Achuthan claimed that this metric was "joining the others."  ECRI has never acknowledged that at least half of the 4 were no longer in a declining trajectory as the first half of 2012 progressed.

Beyond that, ECRI has cited the poor YoY percentage growth in real income as never having previously occurred without an ensuing recession.  That may be true, but if so, it is also true that the rebound in the real income growth rate we have seen in the first half of this year has always occurred as or after those recessions were ending:


Again, ECRI has never acknowledged this.

Finally there is the matter of ECRI's own Coincident Index.  In March Lakshman Achuthan made a big point of the fact that the YoY growth in their Coincident Index was under 2%.  He said that every time that happened, a recession followed.  In fact when Mish called ECRI out on this claim, saying their recession explanations were "disingenuous," ECRI responded directly.  Here is Mish quoting them in his follow up blog post:
The latest USCI growth rate is 1.94% (which can be rounded off to 1.9%). In January 1996, it had dropped only to 2.06% (which can be rounded off to 2.1%). This was certainly not below current readings. Of course, no recession followed.

In 1998, the USCI growth came nowhere near current readings, so the question doesn’t arise. It wasn’t until January 2001 that it fell below 2%, and the recession began two months later.

.... If you look at all the occasions in the last 50-plus years when USCI growth fell to 2.0% or below ..., it is clear that recessions began around those dates....

In sum, it is precisely accurate to claim that y-o-y USCI growth has never dropped to current readings in the past 50-plus years without a recession ensuing.

There's not the slightest bit of uncertainty in that response, is there?  YoY growth of more than 2% in their Coincident Index means no recession, while 1.94% growth means a recession is coming.

Except, just as with GDI, just as with initial jobless claims, just as with GDP, just was with industrial production, and just as with real income minus transfer payments, the data turned against them.  Here's the graph of the YoY change in the ECRI Coincident Index since January 2011:

January 3.7%2.2%
February 3.6%2.4%
March 3.6%2.1%
April 2.8%2.6%
May 2.3%2.6%
June 2.2%2.5%
July 2.1%-----
August 2.1%-----
September 2.2%-----
October 2.4%-----
November 2.5%-----
December 2.4%-----

The sub-2% reading at the beginning of this year has been completely revised away.  In fact based on present data, at no point in the last 12 months has their Coincident Index registered less than +2%.  At no point did ECRI acknowledge that revision, and at no point have they attempted to explain why that does not undercut their recession call, based on their own prior public statement that growth in this index of more than 2% did not lead to a recession.

Although I disagree with him far more often than I agree, Mish can write some great analysis and he almost always makes me think. Last week he wrote that:
Those are exactly the kinds of things that irritate me about the ECRI. The fact of the matter is Achuthan was calling for a recession in September, not December, and not June.

For details, please see my September 30, 2011 post ECRI Calls Recession Based on "Contagion in Forward Indicators"; Just How Timely is the Call?

Tom Keen: "Single Sentence, why recession now"
ECRI's Lakshman Achuthan: "Contagion in Forward-Looking Indicators"

That link clearly shows I thought a recession was imminent as well. Those are the facts. It is silly to try and hide them.

Yet, in December (after economic data firmed), Achuthan moved the date forward to June, wanting another 6 months to be proven correct.

My question in September "Just How Timely is the Call?" was a good one. The ECRI has been both very early and very late. Far from the perfect track record they claim.

That my friends is the nature of making predictions. No one is perfect, not me, not Achuthan, not anyone, and it is very foolish to pretend otherwise.

Actually, I have no problem at all with Achuthan moving the date forward. Conditions change. My problem, is revisionist history that makes it appear as if a recession call in September was a recession call for June (made in December).

All this nonsense goes away the moment Achuthan admits the ECRI does not have a perfect track record.
As Mish says, no one is perfect. I've certainly been wrong my fair share of times. If subsequent revisions to the data lead the NBER to date the onset of a recession to June 2012 or before, I will be the first to acknowledge that ECRI's recession call turned out to be correct.

But can ECRI acknowledge the reverse, based on the current data?  And how can their statements be believed in the future, when data trends upon which they explicitly rely don't pan out, or are revised away, and they never once acknowledge the problem, let alone explain it?