Signal & Noise Filter
The Last War
Like generals fighting the last war, investors get married to ideas and strategies that are no longer appropriate or applicable to current or future conflicts.
As with military strategy, what makes economic and investment planning sclerotic is entrenched political path dependency leading to a status quo of policy making.
The projection of power becomes subservient to the protection of power, fiefdoms, and ego all at the expense of the broader mission.
Thus, the most recent cyclical turning point, the Covid pandemic, shot several sacred economic cows in the head, much to the chagrin of people who made entire careers off them.
The inverted yield curve recession indicator became another noisy toy model rendered impotent by changes in fiscal policy.
Home sales and mortgage rates as indicators were replaced with housing employment and unit construction as recessionary inputs.
Soft-data survey indicators such as ISM had to be tweaked to account for the bullwhip effects of supply chain disruptions.
And on and on. The salient point here is that the economy is a dynamic system, with infinite feedback loops.
Anchoring to indicators without context is a worthless endeavor akin is to fighting the last war.
What matters, and why it matters, is time dependent and impacted by the path dependency of policy makers. Our understanding of the information surface is thus contextually driven.
The current fuss over Friday’s NFP revisions is the noise.
Quantitatively, the standard error of monthly change in NFP is 136K jobs. The combined May and June revisions of 258K jobs, while eye popping, is not unusual at this point in the cycle.
The unusual spike in state and local government hiring in the June release was widely seen as a flaw, and the revision, while bigger than expected, is not a surprise. Government is now a drag on payroll employment.
Lower NFP prints are not necessarily deleterious to the labor market given evolutions of labor market composition, such as immigration.
The 3-month moving average is now 35K/month, on a time series that is notoriously volatile and subject to extensive revision.
Make no mistake, this was a soft report, but one that needs to be taken with caveats.