American consumer confidence fractures along partisan lines as retail data confirms the split
The Gambity Prestige probability signal on a US recession beginning before the end of 2026 sits at 54%. That is not a comfortable number to publish. It is the honest one.
What the August University of Michigan data shows is not a single consumer sentiment reading — it is two readings occupying the same survey. The aggregate dipped, which is the headline most desks will run. What matters more is the composition of that dip. The divergence between Republican and Democratic respondents has widened to a degree that makes the headline number almost analytically useless. A blended average of two populations moving in opposite directions tells you less about the economy than it tells you about the country. Markets pricing aggregate sentiment data as if it carries its usual signal are, in my view, mispricing the situation.
The July retail sales slump lands on top of this and does real damage to the soft-landing narrative. Consumer spending has been the thing holding the aggregate picture together — the variable that kept recession probability estimates in the 30s on most models through the first half of the year. When that variable softens in the same month that sentiment fractures visibly, the load-bearing assumption shifts. The question is no longer whether the consumer is slowing. It is whether the slowdown is uniform enough to be manageable, or whether it is concentrated in ways that accelerate.
Here is where the partisan sentiment split becomes a macro problem rather than a political curiosity. Spending behaviour follows confidence, and confidence is now tracking political identity with unusual fidelity. If Republican consumers — who skew toward certain spending categories and certain regional economies — are pulling back while Democratic consumers hold steadier, the retail contraction will not be evenly distributed. Certain sectors will feel it harder and faster than the aggregates suggest. The aggregate will look less alarming than the underlying sectoral picture for a period, and then revise.
This is precisely the kind of setup where prediction markets tend to underreact early and overreact late. The recession contract pricing I am watching does not yet reflect the compositional breakdown of either the sentiment data or the retail figures. It is pricing the blended number. That gap is where I am positioned.
There is also the Trump administration's current posture on the Supreme Court ballroom question — pushing back against any congressional constraint on executive spending flexibility — to factor in. Whatever one thinks of the legal argument, the political energy being spent on that fight is energy not being spent on a coordinated fiscal response to softening consumer data. The policy response timeline matters for recession probability. Right now, it is not shortening.
