The emergency packages in Taiwan and Korea say less about stopping the AI rout than about where each government believes its own fragility sits. Neither measure touches the dominant flow, which is foreign investors selling stock they already own. Taiwan's short curb and Korea's margin waiver are both aimed at leverage held by locals. Read that way, Friday's announcements are not a floor under the market. They show that domestic balance sheets, not short sellers, are the weak link the authorities are trying to keep from breaking.
The headline numbers support that reading. On Friday the Taiex fell as much as 9% before closing down 4.5%, its biggest weekly loss since 2001. The Kospi dropped as much as 8.8%, tripped a 20-minute market-wide halt, and briefly traded below 5,000. Within hours, Taiwan's Financial Supervisory Commission announced short-sale curbs effective Nov. 9 and the Finance Ministry said it was ready to deploy its NT$500 billion ($15 billion) stabilization fund. Seoul suspended the margin-loan collateral rule that forces brokers to liquidate leveraged retail accounts.
The Seller Nobody Can Ban
Start with the scale of the flow, because it determines what any policy tool can do. According to Goldman Sachs flow data cited in the source reporting, foreign institutional investors pulled $3.1 billion from Taiwan on Thursday alone. That made six consecutive selling sessions totaling $19.2 billion, inside a larger run of 19 selling sessions out of 20 since Oct. 6 that totals $48 billion. Year to date, foreigners have net sold $86.7 billion of Taiwan-listed equities, and $68.9 billion of that was TSMC. Bloomberg's count is more conservative, at more than $45 billion of foreign net selling this quarter. Either figure leads to the same conclusion.
Short selling cannot account for flows of that size. Goldman's prime book reportedly showed long sales outnumbering short sales by roughly 5 to 1 across Asia in October. A pension or sovereign fund cutting its TSMC weight is not borrowing stock, so the FSC's new rules do not apply to it. Under those rules, intraday short-sale orders of borrowed securities are capped at 3% of a stock's 30-day average volume, down from 30%, and minimum short-sale margin rises to 130% from 90%.
The fund's arithmetic is just as unforgiving. NT$500 billion is about one-third of the foreign outflow since early October and about one-sixth of this year's total. Even if Taipei spent all of it, it would cover roughly a month of the current selling pace. A stabilization fund that size can smooth a session. It cannot absorb a portfolio reallocation.
There is also a possible side effect. When cash-market shorting gets harder, hedgers often move to other venues. For Taiwan, that would mean index futures and TSMC's US-listed ADR. Goldman's desk reported that futures and TSMC both touched limit down on Friday, and that the desk itself was "very busy selling and shorting intraday, probably the highest activity level in the past three months," as clients front-ran the expected ban. So the curb's first measurable effect may have been to pull short supply forward. Whether hedging pressure now moves offshore is the inferred link here, and the next few sessions will show whether the ADR trades at a persistent discount to the local line.
What the Fine Print Protects
If shorts are not the problem, why curb them? The more revealing disclosure is in the margin data. Goldman notes that Taiwan's retail margin balance fell 3.4% on Thursday to its lowest level since Aug. 31. As a percentage of market capitalization, however, it "remains near the recent highs." Prices fell faster than leverage was paid down.
That fact explains the policy. The domestic bid that has supported Taiwan is substantial: local funds bought for an 18th straight session, $11 billion over that stretch and $33.5 billion year to date, including $10.6 billion of TSMC. Part of that bid is retail money borrowed against the same stocks it is buying. A levered local buyer facing a falling tape is a forced seller waiting to happen, and short sellers make that cascade faster. Curbing shorts is therefore best read as a measure to slow the price decline enough that local margin accounts do not hit liquidation thresholds all at once. It does nothing about the foreign selling.
The currency confirms how defensive the posture is. Bloomberg reported that USD/TWD touched 33.04, the weakest level for the Taiwan dollar since April 2025, before reversing to close at 32.768, 0.3% stronger on the day. Exporters, state-owned banks and life insurers sold dollars. Those are the institutions that typically act when the central bank prefers not to intervene in its own name. Taipei is defending the stock index and the currency at the same time, so foreign selling can be kept from turning into a disorderly currency move. But the domestic buying capacity that absorbs it is being used up.
Seoul's Waiver Moves the Risk Onto Broker Balance Sheets
Korea's measure looks generous in the headline and much narrower in the details. Starting Nov. 9, brokers are exempt for three months from the requirement to maintain a 140% collateral ratio on margin loans. Seoul last granted this reprieve in July 2022, and before that in March 2020.
There are two catches. First, the waiver is optional. Brokers decide whether to grant forbearance, and they keep the credit risk if they do. The Seoul Economic Daily reported that same-day relief amounted to cutting collateral ratios by 10 to 20 points for clients who applied and delaying forced sales by one day. One securities official estimated that a 10-point cut buffers roughly another 7% decline. The Kospi fell 8.8% in a single morning on Friday. Second, critics point out that the 2020 waiver was followed by large-scale forced selling about a week later. Delayed liquidations still happen, just later.
In practice, then, Korea has moved its margin problem from the exchange to the brokers. If prices stabilize, the transfer costs nothing. If the Kospi falls another 7% to 10%, the brokers holding forborne loans either liquidate anyway or carry impaired collateral. That makes Korean broker equities and their funding spreads the place to watch for stress before it shows up in the index. That link is an inference from how the policy is built, not something already observed in the data.
The Korean leverage numbers are also less comforting than they first appear. Goldman's Seoul desk argues that the leverage flush is largely done: margin loans are about 0.6% of market cap, and the balance has fallen to roughly 28.6 trillion won, 26% below its June 24 peak. That is a real improvement. The weak point is buying power. Investor deposits have dropped 36.6% from their June peak to 88.5 trillion won. Over the past six sessions, retail bought a net 4.5 trillion won while foreigners sold 12 trillion won. Foreign net selling in Korea has now run for 12 straight weeks and reached $44.1 billion, with another $2.0 billion out on Friday. The Kospi is down 46% from its June high and only 16% up for a year in which it had been up 116%.
Korea's retail investors are less levered than they were, but they are also running low on cash. A margin waiver helps an investor who is solvent but forced to sell. It does nothing for an investor who has no money left to buy.
The Orderly-Tape Countercase
The strongest argument against this read comes from the trading desks themselves. Goldman's Sean Navin wrote that flows ran only 10% to 15% above the 20-day average and that "the flow has been orderly, and we have not seen any significant signs of panic." The Seoul team described "limited signs of aggressive selling or capitulation." If the selling is orderly, foreigners are rebalancing rather than fleeing, and rebalancing ends once target weights are reached.
Goldman's Rich Privorotsky makes the macro version of the case. He asks whether this is 1998 or March 2000, and leans toward 1998: an external liquidity shock with intact US growth that reversed once liquidity returned. There is a regional precedent too. In April 2025, Taiwan's identical short-sale package was followed by a rebound.
That precedent deserves a closer look. In April 2025 the Taiex fell a record 9.7% the morning after the fund was authorized, and the recovery came only after the US paused the tariffs that caused the selloff. Taipei's tools did not reverse that market. A policy reversal in Washington did. The 2022 episode is the more relevant comparison. Bloomberg documented then how short-sale curbs, stabilization funds and FX intervention in both Taiwan and Korea failed for months against a Fed that kept raising rates. In that case, the source of the selling stayed in place.
The current shock looks more like 2022 than 2025. The selling now is a sustained reduction of the AI trade that both markets built themselves around, and no single post from Washington can reverse it. India, which has little AI exposure, is having its best week since 2021 as funds rotate away from the theme. Korean 5-year and 10-year swaps rose about 15 basis points this week, and Goldman's EM desk called the Kospi "the dominant driver." Equity stress is starting to show up in local rates.
The "orderly" description also cuts both ways. Orderly selling at $3 billion a day for six sessions is a steady reallocation, and steady reallocations last longer than panics do.
The Flow Number That Decides It
The decision variable is a specific number, not a sentiment indicator. Watch the daily foreign net flow into Taiwan, and the gap between foreign selling and local buying that has defined the year. As long as foreigners keep selling more than $2 billion a session and local funds keep buying, the stabilization fund is mostly filling a gap the domestic bid can no longer cover alone. If local buying stops while foreign selling continues, the fund becomes the only buyer left, and at $15 billion it will run out within weeks.
In Korea, the corresponding signal is investor deposits. A further drop below 88.5 trillion won, combined with a new weekly foreign outflow, would mean the margin waiver is deferring liquidations rather than preventing them, and the one-week lag after the 2020 waiver would start to matter.
The view changes if foreign flows reverse into net buying for several consecutive sessions while both markets hold Friday's lows. That would support Privorotsky's 1998 framing and suggest the packages bought enough time. Until then, the more accurate reading of the weekend's rule changes is that two governments have identified their own levered domestic investors as the risk to contain. Neither the short curb nor the margin waiver gives a foreign holder any reason to stop selling.
