Rosenberg Explains Why Yesterday's ISM Was Likely Wrong, To Be Revised

Tyler Durden's picture

Yesterday, the market surged on an ISM number that was so stretched, and so out of out leftfield, it was higher than the top expectation by the economist panel. The government once again outdid itself in boosting numbers with the hope of surging stocks. It succeeded. Now the question is whether the imminent ISM downward revision have a comparable adverse market effect. And revised it will be: David Rosenberg explains why.

STRANGE ISM NUMBER ... DOESN’T PASS “SNIFF TEST” Here’s why:

1.Most of the regional reports were very poor in August. Either they collectively all wrong or the ISM is.

2.The share of respondents saying the experienced "growth" was 61%, the exact same as a year ago when the ISM was sitting at 52.8.

3. The ISM gain was led by employment (58.6 to 60.4 - best since December 1983) in the same month that ADP manufacturing fell 6,000 (second decline in a row - it was -11k in July when ISM employment was 58.6, so clearly the latter is proving to be, at least for now, an unreliable labour market barometer). Production also ticked up to 59.9 from 57.0 and inventories rose to 51.4 from 50.2. These are all coincident indicators, as an aside (but an important aside).

4. According to the ISM, 76% of the manufacturers surveyed said that their customer inventory levels were either “too high” or “about right". At the turn of the year, just ahead of the big inventory swing that bolstered the GDP data, this metric was sitting at 60%. As a result, it would be folly to assume that the inventory and production categories will contribute to further ISM increases in the near- and intermediate-term. Norbert Ore, who presides over the ISM survey, had this to say about inventories: “If the inventory build isn't voluntary then we have a huge issue on our hands.”

5. Meanwhile, the more forward-looking components dropped, though were hardly a disaster. But orders slipped for the third month in a row, to 53.1 from 53.5 in July, 58.5 in June and 65.7 in both April and May. That is still a sharp squeeze in the growth rate of capital goods-related order books. At 53.1, ISM orders index is down to levels last seen in June 2009 (but when they were rising in “green shooty” fashion).

6.Backlogs were down as well, to 51.5 from 54.5 in July, 57.0 in June and 59.5 in May (and peaked in February at 61.0). At 51.5, order backlogs stand at their low-water mark of the year.

7. Supplier deliveries (measure of production bottlenecks) eased for the fifth month in a row — to 56.6 from 58.3 in July and well off the March peak of 64.9.

8. Looking at five decades worth of data, the share of the time in which we see orders, backlogs and vendor deliveries all decline in tandem, and the headline ISM index rise, is the grand total of 1%. No wonder equities rallied so much — we just witnessed a 1-in-100 event! Bring your camera.

9. Export orders dipped to 55.5 from 56.5 — the lowest they have been since last December. If the overseas economy is rocking and rolling, then why onearth would this component be declining? Not only that, but it looks as though yet again, a good part of the inventory boost we still seem to be getting is being filled by imports — that sub-index jumped four points in August and does not bode well for the trade deficit, which subtracted 3.4 percentage points from headline GDP growth in Q2.

In a nutshell, ISM did smash consensus expectations in August but the composition left much to be desired. The coincident indicators firmed but the categories that actually lead manufacturing activity softened across the board.

As we said at the outset, the ISM index was at complete odds with the regional surveys. Philadelphia, New York, Milwaukee, Richmond and Kansas City were all down. Dallas and Cincinnati were up. In the past, when we had a 5-to-2 ratio to the downside, the share of the time ISM managed to eke out an advance was 4%.

It would be wise to lean against the market's initial dramatic reaction to this data. The ISM orders/inventories ratio is a decent leading indicator and it sank to 1.033x from 1.065 in July. 1.278x in Julne and 1.441x in May. The hidden nugget in today's report is that this ratio has decline to levels not seen since February 2009. And the last time it fell this fast to this type of level was in the September to December 2007 period (1.03x from 1.30x) when once again, there was tremendous confusion and intense debate over whether it was a recession/soft patch in the economy and the bear market/corrective phase in equities.

Suffice it to say that in the past 30 years, with eleven observations, ISM dropped to 47x in the three months after such a decline in the orders/inventory ratio to such a low level as is the case today. That is the average, the median, and the mode. The highest ISM reading three months hence was 51.9, so if past is prescient, today's data was likely a huge headfake.