The ISM Manufacturing Index declined slightly for March 2017, pulling back by 0.5 points after registering a multi-year high in February. The difference between the index and troubling auto sales, for example, is another reminder of what is truly a large disparity between economic statistics and sentiment. The ISM version of a PMI is considered more reliable because it asks more pointed questions, including if you are experiencing greater sales today than a year ago.

But PMI’s even in that format can be misleading because there is no further quantification. If sales drop 10% one year for 50% of respondents and then rebound 1% the next year for the same half proportion, the PMI will clearly look so much better in the latter year even if the overall economy might not truly be that much different; at least not meaningfully so.

As noted a few months ago, this disparity is pronounced especially over the past five years following the initial 2012 slowdown. It is consistent with the pattern of economy where it slows or even contracts and then only partially rebounds afterward. Sentiment thus appears more highly tuned to the relative difference rather than a more comprehensive assessment of activity and therefore its possible sustainability.

Durable goods orders, an almost perfect historical proxy for the ISM, are still averaging just 2% growth (through February 2017). That is clearly better than the -2% average of February 2017, but in reality it isn’t truly a different set of circumstances. And yet, the ISM at 57 and above suggests something more like 2010, 2004, or 1999 when new orders for durable goods were growing at 10+% in sustained fashion.

This “sentiment premium” is in my view surely related to “recession fatigue”, or at least confirmation bias about an economy that for so long refuses all opportunity:

There is almost disbelief captured in both inventory and sentiment, declared by these divergences, that the economy just couldn’t possibly be as bad as the hard(er) numbers suggest because monetary policy has been “all in” the entire time (not to mention the emphatically emphasized unemployment rate, particularly late 2014 forward). From the mainstream perspective, a higher degree of optimism in terms of sentiment actually makes sense, especially this far into an economy with no recovery since businesses are going to be longing for one (recession fatigue) every time one seems the slightest bit possible. Every small turnaround can be embraced as the turnaround, even if there is little or no contemporary basis to think that way.

Each time the economy stops slowing (as in 2013) or even slightly contracting (as in 2016) there is a rush to define that end as the terminal inflection point for the depression. The result is widespread overstating based on emotion rather than rational assessment (or the ability of sentiment surveys to distinguish more than the basics).

What’s perhaps unique to 2017 is just how widespread it has become. We see it in more than just the difference between the ISM PMI’s remarkable bounce and the small relative change of durable goods or factory orders. The Atlanta Fed produces a quarterly estimate of US GDP (GDPNow) using real time changes in the high frequency primary level inputs (and reconstructions of those inputs) that directly feed into the BEA’s matrix. This method as of today figures GDP in Q1 to be just 1.2%.

The New York Fed (out of jealousy of Atlanta?) last year began to tally and publish its own GDP primer. Their GDP Nowcast model as of Friday expects Q1 GDP to be 2.9%, a sizable170 bps difference between the models. As you might guess, the New York Fed uses different inputs than Atlanta, chiefly among them sentiment numbers like the ISM as well as non-GDP related accounts like JOLTS.

By virtue of sentiment, the first quarter should be more robust (though not actually robust, a term that used to be reserved for growth rates that didn’t start with a 2 or even a 3), yet actual “hard” data is tracking like durable goods to considerably less. It may turn out that FRBNY has it right, but that would be among the first quarters where it might be so. Again, this is not a new development, something I wrote about several years ago in the thick of 2015:

The reason the FOMC relies so heavily on confidence measures and surveys is rational expectations. The theory asserts feelings and psychology as more important than real factors, therefore it is assumed that influencing emotions is as much economically useful as even the helicopter option. The same is true of orthodox belief in the opposite, recession, as monetary policy attempts to banish “undue” pessimism as an answer to contraction.

 

With that in mind, the past year has offered what I think is a stunning rebuke to that theory. Starting in the spring of 2014, just after the nefarious cold winter, almost all the major indications jumped. The increases were questionable, and remain so, meaning that offers a useful control as far as attempting to measure the effects of pure numbers alone. Accounts like the unemployment rate, Establishment Survey and GDP surged without a matching increase in more “hard” data less subject to trend-cycle and non-seasonal seasonal adjustments. The response was striking in that the pure, number increases seemed to have worked upon at least consumer confidence (and, I would add, business confidence in the various business sentiment surveys).

Consumer confidence, in particular, surged at the end of 2014 even as accounts like retail sales had already begun their descent. In terms of what I called a grand experiment, the first part of it appeared to have worked – confidence rose sharply as the unemployment rate fell and the BLS stats overall improved in addition to two quarters of 4+% GDP in the middle of 2014. But that was, importantly, the fullest extent of “confidence”:

The second part, the more important transition, however, has been shown quite lacking. For all the immense referral to confidence and emotion, the ubiquitous TV economists talking up that 5% GDP and “best jobs market in decades” at every opportunity, it appears to have mattered so very little. In fact, as noted in May, spending very much departed at the very same time as confidence was surging.

 

The retrenchment in consumer spending started during the fall but really turned downward during the Christmas holiday period; right when confidence was at its inspired apex. It has only gotten worse since that point, with actual, nominal retail sales nearly contracting (and actually doing so ex autos). Retail sales in 2015 are worse now than they were during the whole of the dot-com recession.

It is perhaps the perfect representation of the counterfeit recovery. Actual recovery is opportunity, a condition that has been severely lacking for many years now. That is not something that can be counterfeited like confidence can. Emotion is simply not enough, though it is often modeled to be under “rational” expectations theory. The difference seems to be a small one, but it has proved to be enormous; time.

In short, the presentation of estimated job gains created broad hope in Americans not that jobs had already arrived but rather in how they thought that might signal jobs that would come. I think that is why the labor force never expanded as it should have, signaling more strongly that this was all just imaginary.

It isn’t even about monetary policy anymore so much as that incredulity over how the economy can persist in such a lowly state year after year. It seems as if we might be in the middle of a second experiment with confidence, one even more jarring in its incongruity than the first (just look at bond or money curves as a representation of perceived opportunity).  Again, every time the economy stops slowing or contracting people seem to become irrationally hopeful that means something truly radical and positive even though all experience since 2007 has demonstrated that there actually is no rational basis for that hope. The road to Japanification is surely paved with so much disbelief. It’s completely understandable in one sense, since it has been almost eight years of all the “experts” constantly claiming that things were definitely going to get better.