Consensus Wrote Two Different Februaries This Week

Take the February pair on its own terms. A 320,000 decline in openings is a 4.4% one-month drop in posted vacancies — the demand side of the labor market contracting at a pace that, sustained, changes hiring and wage arithmetic. In the same calendar month, consensus expects households to have accelerated spending by six-tenths of a percentage point off a negative base. Both can be true; February weather, timing effects and seasonal adjustment routinely separate one month's spending from the labor market it sits inside. But the two estimates are not independent guesses. They imply a specific structure — that firms trimmed hiring intentions while consumers spent through it — and that structure is what carries revision risk this week.

Why retail sales is the load-bearing number

Attention should not be distributed evenly across the four prints, because they do not carry equal information. JOLTS measures a stock of posted vacancies, is revision-prone, and reflects intentions rather than realized payrolls. Consumer confidence measures stated attitudes, and the historical relationship between attitudes and outlays has been loose enough that a 3.2-point decline can coexist with rising spending. ISM is a diffusion index describing direction, not magnitude, and at an estimate of 52.5 against a 52.4 prior, consensus effectively expects no change at all — which confines its market relevance to the tails, particularly any move toward the 50 line that separates expansion from contraction.

Retail sales is the exception. It measures money that was actually spent, it feeds first-quarter GDP tracking and corporate revenue directly, and the estimate embeds the largest required behavioral reversal of the four. The week's contradiction is not between optimists and pessimists — it sits inside consensus itself, and it resolves on a single Wednesday-morning print.

Push the chain one layer further. If February spending confirms at or above +0.5%, the falling-openings estimate can be read as labor supply and demand normalizing rather than final demand weakening — a reading that leaves cyclical revenue assumptions intact and makes the March confidence miss largely ignorable. If spending misses, the -0.1% January print stops being an outlier and becomes the first half of a two-month soft patch; at that point the JOLTS decline stops looking like normalization and starts looking like firms responding to slowing demand. Those two interpretations of the same 6.92 million number have very different implications for first-quarter consumer-facing estimates, which is precisely why the sequencing matters more than the levels.

The earnings slate happens to be a cross-check

What else could explain the pattern? The strongest competing interpretation is that there is no inconsistency at all — that vacancy counts have been normalizing from an elevated base for long enough that their decline no longer transmits to household income, and that confidence surveys have persistently understated spending. That reading is credible, and it is why this piece frames retail sales as the test rather than asserting a direction. But it is an assumption with a break condition, not a fact, and the break condition prints on April 1.

The disclosure channel adds context but no direction. Congressional filings totalled 41 trades in the window; among the largest and most recent sampled rows, the exposures being added sit on the rate-and-consumer axis the week's prints will move — four separate LGI Homes buys by Rep. Tim Moore dated March 12 through March 20, a Cracker Barrel buy on March 23 against a Krispy Kreme sale on March 17, and five Goldman Sachs buys by Sen. Dave McCormick dated February 23 through March 10. Disclosures evidence positioning, not intent, and a sample this size cannot carry a claim about the consumer. Institutional 13F moves and insider clusters registered nothing in this window, and the news sample — drawn from 20,554 articles — was dominated by securities class-action notices with no macro content. Note also a scheduled US presidential address at 01:00 GMT on April 2, a high-impact calendar item with no estimate attached and therefore no consensus to deviate from.