The arithmetic left implicit in the speech is the part worth copying into a discount rate. If prices are up 3.7% over twelve months and the last six ran at a 4.1% annualised pace [1][2], the six months before that compounded at roughly 3.3%, which makes this an acceleration of about eight tenths of a point rather than a stale level [15]. A level can be blamed on base effects. A slope has to be argued with.
The breadth number does similar work: 54% of components rising more than 3% [3] puts the median component above 3% [16], which is the difference between a few freight and insurance lines carrying an index and a general repricing. The caveat is that 54% by count is not 54% by weight, and the Seoul Economic Daily briefing does not say which [3].
The genuinely reflexive claim is the one about conditions, not prices. Warsh's stated grounds for saying it is difficult to characterise overall financial conditions as restrictive were expanding corporate capital spending driven by AI infrastructure, solid S&P 500 profitability, and easier bank lending attitudes [5]. Read that as a financing mechanism and it inverts: every hyperscaler and memory fab writing cheques into the buildout is supplying the evidence that the rate relief which would make the next tranche cheaper is unnecessary. The capex is its own collateral against easing. Pair that with the formalised retreat from forward guidance [7] and treasurers lose the free option they have been carrying, which was a signposted path they could plan against without paying for it; the level of the policy rate matters less to a model than the variance around it, and the variance just went up.
Seoul's tape that session priced the rate, not the earnings. A combined 2.0429 trillion won of foreign and institutional net selling [10][17] took the KOSPI down 1.79% to 6788.88, about 124 points off a prior close near 6913 [9][18], with Samsung Electronics off 3.38% and SK hynix off 4.45% [12], on a day when Nvidia's second-quarter beat had partly eased peak-out worries [11]. The sell-side was moving the other way: 827 downgrades against 411 upgrades after last month's plunge, two cuts per raise, flipping in the August 22-28 week to 47 upgrades against 42 downgrades [13][19], with twelve-month forward EPS up about 6% from month-end [14].
This is probably wrong in its timing, but the expensive assumption to keep now is the cut, because the model that carries one has to argue against a stated slope and a stated read of conditions at the same time. The counter-thesis has two legs, and both are live. If the capex leg softens, Warsh can ease on financial conditions without recanting a word on inflation, which is why SK hynix president Kwak Noh-jung's claim of no clear downturn signs through the end of 2030 [8] is a monetary statement as much as a memory one. And if that 54% turns out to be a minority once weighted [3][16], the breadth argument thins and the 4.1% becomes a narrow story that decelerates mechanically. What would show me wrong on the equity side is simpler: if the revision flip holds and forward earnings keep climbing 6% a month [13][14], then a higher discount rate is absorbable and the selling was flow, not verdict.
