When “Positive Surprises” Are Surprisingly Meaningless

Updated on

By John P. Hussman, Ph.D.

In recent weeks, investors have abandoned all material concern about the likelihood of oncoming recession, largely because U.S. economic reports – though very tepid on an absolute basis – have come in persistently “better than expectations.” Objectively, the best showing has been in new claims for unemployment. The 4-week average has eased slightly below 400,000 in recent weeks, to a level that would no rmally be consistent with slightly positive payroll growth, though not enough to absorb normal labor force growth (the unemployment rate dropped last month partly because hundreds of thousands of people stopped looking for work). My concern here is in taking these labor numbers as predictive, in the face of an abrupt drop in federal withholding tax deposits.

As economist John Williams observes, “starting in October, a divergence developed: Whereas year-to-year change in BLS estimated payroll earnings continued at a more-or-less constant, positive level, tax receipts fell quite markedly. Where the Treasury numbers reflect full reporting, the BLS data are sampled, heavily modeled and usually heavily revised. The implication is that the BLS has overstated average earnings and payrolls meaningfully in recent months.”

The pattern in withholding tax deposits mirrors what we’ve seen in a whole ensemble of reliable leading economic measures we track – a sharp initial deterioration in early August, followed by a slight bounce into October, followed by resumed weakness that reinforces our concerns about the economy (see Have We Avoided A Recession? )

Labor figures are subject to strikingly large seasonal adjustments. The seasonal adjustment factors for non-farm payroll employment vary between 1.0157 and 0.9920, which doesn’t mean much until you put these numbers in context. Given total non-farm payroll employment of about 132 million workers, these adjustment factors mean that in any given month, the effect of seasonal adjustment on the reported payroll employment figure amounts to something between +2.1 million and -1.1 million jobs. Likewise, two-thirds of the new unemployment claims reports over the past year have been subsequently revised upward by several thousand. That’s not to diminish the importance of these figures in economic analysis, but instead to emphasize the importance of smoothing and other forms of noise-filtering that reduce the impact of short-term volatility. Suffice it to say that the improvement in new unemployment claims strikes me as a legitimately hopeful development, but there is too much short-term noise, and inconsistency with other economic evidence (reliable leading indicators, falling tax withholdings) to draw a convincing signal.

Full article here-http://www.hussmanfunds.com/wmc/wmc111219.htm

 

 

Leave a Comment