Hoarding dollars through June had a price. The arithmetic of that price is why the selling cascades now: a company that sat on export proceeds when the rate touched 1,561.5 on June 5 [4] and converts at 1,334.7 [1] takes 226.8 fewer won per dollar, 14.5 percent less [1]. Each break lower pulls in more of the backlog, because the cost of waiting compounds; the rate went below 1,400 and then through 1,390 and 1,380, which Seoul traders call domino selling [14].
The bid being priced in the 1,330s rests on a slower machine than its headline capacity suggests. The National Pension Service held roughly 770 trillion won of overseas equities and bonds at the end of June, and the 15 percent strategic hedge ceiling would in theory cover about 115 trillion won of that [10]. The ratio actually in place is nearer 5 to 10 percent, which is where both the market and the Ministry of Economy and Finance put it once the new framework took effect [9], so the hedge book is something like 38.5 to 77 trillion won [2]. Cutting that ratio withdraws the dollar selling the hedges had been producing, a smaller effect than the 115 trillion won headline figure implies [11]. Barclays economist Son Beom-gi says the positions, arranged as swaps with the Bank of Korea on six- and twelve-month terms, are difficult to break early, so they come off as maturities are allowed to lapse rather than be rolled [7][8]. Run 38.5 to 77 trillion won off evenly across twelve months and it is 3.2 to 6.4 trillion won a month of dollar selling withdrawn [3] - real supply and demand, but far short of a wall. (Dollars sold forward when the rate was in the 1,500-won range [17] mark against 1,334.7 at roughly 165 won each, about 11 percent [6], a modest cost that makes letting them expire an easy call.)
The flows visibly moving the rate are exporter conversion [3] and dollar supply from strong semiconductor exports [13], plus a yen that fell into the 155 range on Bank of Japan tightening expectations and dragged the won up with it [12]. The pension leg is thinner evidence: analysts could not confirm the size of the purchases reportedly detected that day [18], and a senior foreign exchange official put it as the NPS "may have made some purchases" [15], which participants nonetheless read as a standing offer to absorb dollars in the 1,330-won range [16].
How this resolves depends on which driver is doing the work. If the exporter backlog is the driver, then 1,331.3, the level last seen on October 4, 2024 [5], sits 3.4 won away [4] and is a waypoint rather than a base. If the fund did buy in size, the 1,330s hold, and the story becomes one about a fund that was selling dollars to lean against won weakness at 1,500 only months ago and is now discussed as a buyer [17][6]. If the yen is doing most of the work, a Bank of Japan that disappoints removes the pressure and the exporter unwind reads as coincidence. The mechanism supports the first, or rather the more interesting version of the first, in which the NPS influences the pace through rollover decisions at each maturity rather than by defending a level [8]. What would argue the other way is the rate holding the 1,330s on a session when the yen weakens, which would mean something other than flow is bidding down there.
