The Institute of Supply Management is no stranger when it comes to seasonal adjustment fiascos [3]: recall that in June it took Stone McCarthy to tell a humiliated Bradly Holcombe that they had released the wrong data not once, but twice. Ever since then we have been particularly focused on the seasonal adjustment factor that ISM uses when fudging its unadjusted data - data which as a reminder is a survey which reflects a continuum and thus does not need to be seasonally adjusted, yet it is. This is precisely what [4]we found that last month, when the ISM reported a 55.3 print driven by a 58.9 surge in New Orders we called bullshit on the data and found that the actual, unadjusted data was the weakest since January.
Today, we find more of the same, when we learn that while the headline ISM of 57.1, seemingly the highest since April 2011 and driven by the all important New Orders print of 63.4 which was the highest since December 2013, was really a figment of seasonal adjustment.
The table below shows how the ISM takes its unadjusted, actual data, based on respondents saying whether the data is "better", "same" or "worse", and applies a seasonal adjustment factor, getting the adjusted number.
[5]
The more observant will notice that the % of respondents saying New Orders are "Better" just dropped to the lowest number since January, or 29%. And since the actual NSA number is calculated by add half the "Same" respondents to the "Better", what one gets is a New Orders number in July which was the same as June, and the lowest in 6 months.
To show what is really happening here is a chart of New Orders actual and adjusted. One is the lowest print since January. The other is the highest of 2014. Which one do you believe?
[5]
The bottom line is that the only reason why the New Orders and thus the ISM soared to the highest level in over three years, is due to that 0.907 seasonal adjustment factor.
So what does recent unadjusted data reveal if one looks at just the actual surveys for the top three key categories without the seasonal adjustment?
This:
[6]
Welcome to the seasonally-adjusted recovery, summer of 2014 edition.
